&
3
3
Ai/=,.AT,+---.a,
Table 1. Phvsical and mechanical indexes of tlie soil lY Soil XUI IP e sc* 33.2% 19.5 kNin-' 14.2 0.85 Clay 31.1% 19.4 kNm-3 17.8 0.80
Soil
.K,,
A,
E.,
C'
SC" 0.71 3.61 MPa 5 kPa Clay 0.62 3.43 MPa 15 kPa (* 'SC' in the table is substitute for 'Silty Clay'.)
258
40' 25" 28"
Six tests are conducted, three for silty clay and three for clay test No 1 silty clay, the diameter of the pile is D = 30nim. test No 2 silty clay, 11 = 50mn?, test No 3 silty clay, D = 20mn?, test No 4 clay, D = 3 O m m , test No 5 clay, L) = 50mm, test No 6 clay, D = 2 0 n m The pile is jacked in flight into the soil at a rate of 2 5minIs to simulate the penetrating velocity of the practical jacking The length of the pile jacked in the soil is 25cm Soil displacements and initial excess pore pressures are measured immediately after pile jacking 3 2 Aiinlysrs of test.,
11,
E=2.6MPu,
K
I
+] /% . A,,
(4 1)
Tablc 2 Tlic \aluc for ainciidiiig factor a, ,a , ,P
a2
a1
P
7 0-7
s
0 02-0 01
2 1-2 7
6 2-6
s
0 10-0 12
18-15)
J J , =S-9
Cla!
li
=29%
I,. =7-s
K(,=l-sinpr= 0 . 5 8 , c = 8 . 3 .
Ac-
It can be seen from the table that the results ofu, are in good agreement. In the zone adjacent to the pile shaft, for instance, r=0.5m in the above case, the difference may be significant, it is so because this region is in a plastic state; while in the zone that is far away from the pile shaft, for example, r=1.5m and r=1.9m in this case, the difference is trivial. Thus the calculation can provide substantial accuracy in the elastic zone, which is very important in the prediction of horizontal soil movement during pile jacking. The agreement of the theoretical calculation and experimental measurement is also shown in the analysis of test No.4 (Chen 1999). From the above analysis, it is shown that the theoretical solutions derived from Eq.43 agree well with the experimental observation. Since parameters a, ,a2and fl can be determined from tests or field measurements, Eq.43 is a practical form in the prediction of the radial displacements of soil during pile jacking. If more tests are carried out, more detailed information about a , , a , and can be obtained. According to Eq.42 and the soil property indexes listed in tablel, the theoretical solution for initial excess pore pressure at depth z = 2.5mcan be expressed as
,!I'
sc.H=12'%.
'/2) = 2.46 ,Y,~,, = 19.5kN / nz3,
Table 3. The results of uVin test No. 1 (mm) r= I= r= I = I = z( Ill) 0.5In 0.9m 1.1111 l.5m 1.91~ 105 54 37 2,0 Theor. 10 0 43 28 6 2 Experi. 92 55 37 2,4 Theor. 106 10 0 Experi. 91 58 30 13 3 106 55 38 11 0 3,0 Theor. Exoeri. 110 18 14 17 3
in which a , , a3 and are coefficients that are related to the soil property and the pile diameter, and they can be determined by centrifugal model tests. By analyzing the results of test No 2, No.3 and No.5, No 6, a,, a? and y for some occasions are listed in table 2
so11
pr=25",p: = 2 2 " , y ' = 9 . 7 k N l m 3 ,
+
cording to table 2, a,=7.3, a2= 0.03, p = 2 . 6 5 , Then Eq.43 changes into 21,. = [(0.043 + 0.001~z)-O.O871nr] (42) Radial displacements at depths 2.0m, 2.4m and 3.0m are computed using Eq.44. Theoretical solutions and experimental measurements are listed in table 3.
arid Air of centrrfiigal niodel
(2c - K,, y gq;, R,,). 1n
tg' (45 ''
c = 15kPu,
It is shown by the results of the centrifugal inodel tests that the radial displacement 11, is continuous in the elastic and plastic zones and it decreases with the increase of I' However, it is revealed by Eq 26 that the expression of i i , at the interface of the elastic and plastic zone is different from that of Eq 11 in the elastic zone, This will certainly result in a discontinuity To avoid such a discrepancy, it can be assumed that the expression of 11, in both the plastic and elastic zones have the same form as that of Eq 27 By introducing amending coeficient a, ,a 2 ,p , and assuming that there exists an initial earth pressure q = (1 iXi,). 0,',,,/3, Eq 27 transforms into
+p
=
2.5 20 8 (43 1 r I' Theoretical solutions of A u at radius r=0.6m, 1.2m and 2 l m can be derived from Eq.47. The comparison of theoretical solutions to the experimental measurements is shown in Fig.3. = 18 11n-+-+9.8
The effectiveness of Eq 45 and a, ,a , and p can be examined by test No I and No 4 Take Test No I for example The soil is silty clay, lit,= 0 3/77 , the property indexes of the soil can be obtained from table 1 259
order to make the calculating of U , more rational and more convenient. The spatial solutions are used to analyze the results of centrifugal model tests, and the values of the theoretical prediction correspond well with those of the experimental measurements in most areas along the pile length. It is also revealed that in the upper part of the pile above the pile tip, the theoretical solutions are quite reasonable, while in the vicinity of the pile tip, both expressions of z ~ , and Aluneed to be refined. This problem, however, can be solved by considering the reversion of the principal stresses in the vicinity of the pile tip. The results of this paper can also be applied in such fields as the analysis of pile driving, cone penetration and analogous situations. The method presented in this paper is convenient for practical prediction to engineers.
Fig.3 Test No. 1. distribution of initial excess cxcess pore pressure at depth 2.5m
It can be Seen from the spatial analysis of the above tests that, the value of the initial excess pore pressure AI, and the radial displacement U , calculated fi-om the equations put forward in this paper are in good agreement with those measured in the experiment It is also shown by the spatial solutions that at a certain depth, both AIIand 11, decreases logarithmically with the increase of the radial distance from the pile axis, while at a certain horizontal radius, Air and I / , may increase linearly in the vertical direction However, it is impossible for hl and U , to increase infinitely with depth, after reaching a maximum value in the vicinity of the pile tip, hi will decrease until it disappears at a certain depth below the pile tip This phenomenon has been observed in field measuring The reason lies in the fact that, the shape of the pile tip is a cone, and soil is pushed outwards both in the horizontal and vertical directions, accompanying with a reversion of principal stresses at the pile-soil interface around the pile tip The stress boundary conditions shown in Eq 1 and Eq 2 are no longer satisfied, thus resulting in a disharmony at this location HOWever, it does not mean that the spatial solution put forward in this paper is incorrect For most areas except for those around the pile top and tip, the stress boundary conditions are satisfied, and the solution is Quite reasonable For the areas around the pile tip, t'he discrepancy can be relieved by introducing the amending factors.
REFERENCES Azzouz. A.S. & Morrison, M. J. 198s. Field measurements on pile in two clay deposits. J. Geotech. Engrg. A.X.C.E., 114(1): 104-121 Baligh, M.M. 1986. Undrained deep penetration, I : shear stresses. Geotechnique, 36(4):471-4S5 Butterfield, R. & Banerjee, P.K. 1970. The effects of pore water pressure on the ultimate bearing capacity of driven piles. Proc. 2nd South East Asian Regional ConJ on Soil Mech. and Found. Engrg.. Singapore:385-3 94 Carter, J.P. 1979. Stress and pore presure changes in clay during and after the expansion of a cylindrical cavity. Int. Jour. hhrn. andAnaly. Methods in Geomech., 31217-229 Chen, W. 1999. Pile jacked in saturated clay: mechanism of penetration and soil compaction effect. hiE. ti7e.ri.r. Hohai Univ., Nanjing, China. Cooke, R.W. 1979, Jacked piles in London clay: a study of load transfer and settlement under working conditions. Geotechnique. 29(2):113-147 Heilkel, D.J. 1959. The relationships between the strength, pore-pressure and volume change characters of saturated clays. Geotechnique. 9(3): 119-135 Huiitsinan, S.R. & Mitcliell, J.K. 1986. Lateral stress nieasurement during cone penetration. In S.P. Clemence (ed.), Use of in situ Tests in Geotech. Engrg., A.S.C.E., 617634. New York: N.Y. Kulhawy, F H 1984. Liniiting tip and side resistance fact or fallacy'? In Meyer,J.R. (ed.), Analysis and rlesign of pile foimda[ions. A.S.C.E.. 80-98 New York: N.Y. Lehane B.M. 1993, Mechanisms of shaft friction in sand from instmmented pile tests. J. Geotech. Engrg. A.S. c'.E. 119(1): 19-35 Masood, T. & Mitchell. J.K. 1993. Estimation of in situ lateral stresses in soils by cone-penetration test. J. Geotech. Engrg.. .4.SC.E.. 119(8): 1624-1639 Randolph, M.F., Carter. J.P. & Wroth, C.P. 1979. Driven piles in clay-the effects of installation and subsequent consolidation. Geotechnique. 29(4):36 1-393 Vesic, A.S. 1972. Expansion of cavity in illfinite soil iiiass, Jour. >SoilMech. Found. Div., ,4.JS.C.E., 98(3):265-289
4 CONCLUSIONS
By considering the change of internal pressure and shaft friction in the vertical direction, spatial CEM solutions for the radial displacement i i , and the initial excess pore pressure Ail generated during pile jacking are derived in this paper In comparison to the solutions obtained under the plane strain assumption, the spatial solutions can reflect the varying rule of zi, and AI/ both in the horizontal and vertical directions Centrifugal model tests in simulation of pile jacking are conducted Based on the measurements of the model tests, amending coefficients are introduced in 260
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds)0 2000 Balkema, Rotterdam, ISBN 90 5809 151 1
A rational procedure for comparing measured and calculated values in geotechnics C. Cherubini Istituto di GeologiaApplicata e Geotecnica, Politecnico di Bari, Italy
T.L.L.0rr Trinity College, Universityof Dublin, Ireland
ABSTRACT: The evaluation of any geotechnical problem (bearing capacity, settlement, etc.) is strongly affected by the presence of a number of uncertainties which may be grouped into the following separate categories: uncertainties connected with the variability of the mechanical properties of soil due to the limited number of samples tested and the natural variability of materials involved, uncertainties connected with the calculation method used, and uncertainties connected with the unavoidable differences between the design dimensions and properties. It has been shown that the reliability of calculation models may be assessed using the synthetic probabilistic approach which is based essentially on comparison between in situ measurements and calculations. Measured values Qmas can be compared with calculated values Qo1, using a factor, called If a sufficiently large number of measurements is available, the "bias factor" defined as the ratio of QcalJQmeas. the bias factor values obtained using a particular calculation method can be processed to evaluate the "accuracy" and "precision", by calculating a central tendency and a variability statistical parameter espectively from the bias factor values. Two comprehensive statistical parameters, the RI and RD Indexes, which are based on the central tendency and the variability, are shown to be useh1 for assessing the accuracy and precision of a particular calculation method. Using the calculated and measured bearing capacities of driven piles in NC clays, the accuracy and precision of the most frequently used pile driving formulae are assessed by means of these parameters.
1 INTRODUCTION I n geotechnical engineering the evaluation of bearing capacity and settlement is affected by a series of uncertainties which may be grouped into: - Uncertainties connected with the variability of the mechanical properties of soils to be investigated. This category includes uncertainties due to the limited number of samples to be tested and the natural variability of the materials involved. - Uncertainties connected with the calculation method used. - Uncertainties connected with the unavoidable differences between the design dimensions and properties of structures in contact with soil and the actual values. The fwst group of uncertainties can be minimized by means of adequate sampling and testing and by the use of probabilistic design methods based on knowledge of the variability of the mechanical properties of soils. The third group of uncertainties can be controlled by good construction methods and good supervision.
The second group of uncertainties can be managed by comparing calculated and measured values of particular geotechnical quantities. Unfortunately, at the moment such comparisons are only possible for a limited number of cases as it is difficult to find references in the literature to cases with suitable data. In this paper some fundamental concepts and definitions that are useful for comparing calculated and measured values are highlighted. To clarify the procedure, an illustrative example, based on data for the bearing capacity of driven pipe piles in cohesive soils (Ramey and Johnson, 1979), is presented.
2 COMPARISON BETWEEN MEASURED AND CALCULATEDVALUES Having available a set of n calculated values (Qol,) and the corresponding measured values (Qmeas).the so-called "bias factor", K can be evaluated as:
261
If a significant number of values is available, the set of K values acquires the characteristics of a random variable extracted fiom a population of all the possible values of that factor. Clearly, if different calculation methods are available, different sets of K values can be calculated for a specific set of Qcalc values. In this case. in order to determine which calculation method best fits the measured values, it is necessary, by analysing the different data sets, to evaluate some suitable statistical parameters for use in the comparison. This is the synthetic probabilistic approach (Cherubini and Greco, 1997). The "accuracy" of a calculation method can be associated with the central tendency of the set of data. Hence the mean or the trimean of the data can be used to provide an indicator of the accuracy. The "precision" of a calculation method can be evaluated by means of a measure of the dispersion of the examined set of data. Hence the standard deviation (SD) or the interquartile range, IQR can be used successfully to indicate the precision (Velleman and Hoaglin 1981. Kotzias et al. 1990). The use of the coefficient of variation, CV previously proposed by Cherubini et al. (1995a and 1995b) to evaluate the precision, should not be used because the CV is a function of both the standard deviation and the mean and hence the mean is considered twice. A krther definition concerns the comparison between more and less "conservative" methods. Whether one method is more conservative than another is assessed by comparing the frequency of the ratios
Qcaf
Q
< 1 and I1lca\
Qca/
Q meas
Figure 1. Contours of RI and RD plotted with respect to accuracy (mean value K) and precision (standard deviation K). An overall index that may be calculated to assess the accuracy and precision of a calculation method, which takes into account the mean and the standard deviation of K, is the so-called Ranking Index (Briaud and Tucker 1988):
> I . The first
ratio is relevant for bearing capacity situations, i.e. Qcalc < QI1ICa\, while the second ratio is relevant for settlement situations, i.e. Qcalc > Qmeas. It is possible, using the K values, to define a scale of conservatism or safety. For example a conservative method of calculating the bearing capacity could be defined as one having between 60 and 80% of the values of K < 1, a very conservative method one having between 80 and 100% of K < 1, while a method could be considered to be "neutral" when between 40 and 60% of K < 1 and unconservative or unsafe when less than 40% of K < 1. Tan and Duncan (1991) and Berardi and Lancellotta (1 994) use the term "reliable" when evaluating calculated values. The Authors consider this term to be inappropriate because it could cause conhsion with respect to more complex evaluations of the "reliability" of a geotechnical structure based on probabilistic analyses.
where: p indicates the mean value s indicates the standard deviation in is the Neperian (natural) logarithm. A new index proposed by the authors for assessing the accuracy and precision of a calculation method is the Ranking Distance, RD. As shown in the plot of central value (mean) of K versus scatter (standard deviation) of K in Figure 1, the Ranking Distance is the distance of the point representing a particular calculation method from the optimum point characterized by the mean value K = 1 and scatter SD = 0. This index can be determined fiom the modulus of the vector connecting the two points:
The RD index enables a method to be evaluated with regard to both accuracy and precision. Low RD values correspond to calculation methods that have both high accuracy and high precision while high RD values correspond to calculation methods that are either highly inaccurate or highly imprecise, or both.
262
Since the RD value is the distance of the point representing a particular calculation method from the optimum point on a graph of SD versus mean K, contours of equal RD value plot as semicircles around the optimum point, as shown in Figure 1. However, due to the fact that the RI value is the sum of the logarithm of the mean and SD of K, the contours of equal RI for mean K values less than 1 plot as approximately diagonal lines orientated at about 48" to the horizontal axis for RI = 0.25 to about 54" for RI = 1.25 as shown by the RI contours in Figure 1. Thus while the RI and RD values both enable a method to be evaluated with regard to accuracy and precision, the advantage of the RD value compared with the RI value is that, being equal to the distance from the optimum point, it is simply represented graphically and also it gives equal weighting to the accuracy and the precision. The RI value, however, being based on a logarithmic rather than a linear scale, gives a less favourable rating to calculation methods that are equally accurate and precise than does the RD value. However, for methods that are very accurate, it gives more weighting to the accuracy than the precision compared with the RD index, while for methods that are very precise, it gives more weighting to the precision than to the accuracy. The comparison between the RI and RD values is shown by the contours of RI = 0.25, 0.50, 0.75,1.00 and 1.25 and the contours of RD = 0.50 and 0.75 in Figure 1. The difference in the RD and RI values is demonstrated by examining the RD and RI values for a number of calculation methods with different accuracies and precisions. A highly accurate but imprecise method, with mean K = 1.0 and SD = 0.50, has RD = 0.5 and RI = 0.59 (Point 1 in Figure 1). A highly precise but inaccurate method, with mean K = 0.5 and SD = 0.0, has RD = 0.50 and RI = 0.69 (Point 2). The method having the same RD value of 0.50 as the equally accurate and precise method represented by Point 1 is found to be one with mean K = 0.65 and SD = 0.35 (Point 3), while the equally accurate and precise method having the same RI value of 0.59 as the method represented by Point I is found to be one with mean K = 0.77 and SD = 0.23 (Point 4), i.e. the method represented by the RI value has a higher accuracy and precision and so is closer to the optimum point than the method represented by the RD value. Using the RI value, the highly accurate (mean K = 1.0) but imprecise method that has the same RI rating as the equally accurate and precise method represented by Point 3 is found to be the method represented by Point 5 with SD = 0.80. This is very different from the method represented by Point 1
with SD = 0.5 using the RD index. Similarly the highly precise (SD = 0) but inaccurate method that has the same RI value as the equally accurate and precise method represented by Point 3 is found to be the method represented by Point 6 with mean K = 0.34, which is very different from the method represented by Point 2 with mean K = 0.50 using the RD index. These comparisons show that, when comparing two methods using the RD and RI values, the RD index gives a more favourable rating than the RI value to those methods that have equal accuracy and precision while the RI value gives a more favourable rating than the RD index to those methods that are either very accurate or very precise. A consequence of this is that, for methods with similar levels of precision and high levels of accuracy, more weight is given to the accuracy and less to the precision when calculating the RI value than when calculating the RD value. This implies that the RI index can be misleading and may be unsafe and therefore the Authors consider that the RD index is a more rational and better parameter for comparing calculation methods. The direction of the RI3 vector, which can be expressed by the angle the RD vector makes with the horizontal axis indicates the relation between the accuracy and the precision of a particular method. The region which has its centre at the optimum point in Figure 2 may be divided into the three zones indicated by the different types of shading. The zone represented by the 60' segment, with the vertical line through the mean K value of 1 as its axis. corresponds to values representing methods that are more accurate than precise, the zones within the two 30' segments above the horizontal axis correspond to values that are more precise than accurate, while the segments between 30' and 60' above the horizontal axis correspond to values representing methods that have similar accuracy and precision. 3 THE PROCEDURE FOR COMPARISON On the basis of the concepts and parameters explained in the preceding section, the Authors propose the following 5-stage procedure for comparing two or more calculation methods with respect to measured values. When n sets of data values are available for m calculation methods: 1) Calculate the n Qca,JQIneasratios for the m methods . 2) Calculate the statistical parameters of interest for each of the m sets each of n data and plot the results (histograms, box plots to identify outliers, mean and/or trimean, standard deviation and/or interquartile range). 263
Table 1. Measured pile bearing capacities and ratios between calculated and measured values I
MEN
Figure 2. Results for five pile driving methods plotted with respect to the mean and standard deviation of K and zones with different degrees of accuracy and precision. 3) Plot the results on a mean-standard deviation graph or a trimean-interquartile range graph.. 4) Calculate the values of the RI and RD indexes. 5) Compare the results obtained using the different calculation methods. After updating the calculated results by means of a Bayesian procedure (Cortellazzo and Mazzucato, 1998), using experience of the methods, return to stage 3 and then calculate improved values of the indexes.
,
118
I
,
I
I
I-IILEY
1
1311 0.2708 0.926 4 0 7 7 0.514 -0665
EN
MEN
I
GATES
1
1
0.5
HILEY
I DANISIH I
1
-0693
1
1.115 0 1 0 9
GATES
DANISH
The results in Table 2 demonstrate that: - On the basis of the K values, the MEN, Hiley and
4 EXAMPLE The data selected for an example to demonstrate the use of the Ranking Distance are those data relating to the bearing capacity of steel pipe piles in cohesive soils published by Ramey and Johnson (1979). The measured bearing capacity values (in tons) are presented in Table 1 together with the ratios between the calculated values using the EN, MEN, Hiley, Gates and Danish methods and the measured values. The values of Naperian logarithm (In) for these ratios are also given in Table 1. The following results are reported in Table 2 for each method: - mean and standard deviation of K - trimean and interquartile range of K - mean and standard deviation of InK - value of RI - value of RD - value of percent of K < 1.
264
Gates methods are very conservative, as the percentage of K values < 1 is greater than 80%, while the EN and Danish methods are neutral, as percentage of K values < 1 lies between 40% and 60%. - The EN and Danish methods are characterized by good accuracy, as demonstrated by the favourable mean values of K (> 0.85) and InK (< - 0.3), and also by the favourable trimean values of K (> 0.83). - Regarding precision, the differences between all the methods considered are small based on the standard deviation values of InK, which range from 0.522 (most precise) for the Hiley method to 0.604 (least precise) for the EN method, but are larger, and hence more significant, based on the standard deviations of K, which range from 0.255 for the Gates method to 0.476 for the EN method. The least precise method, based on the Interquartile range, however. is the Danish method with an IQR value of 0.755.
Plotting the mean and standard deviation values in Table 2 on Figure 2, it is can be seen that, from the position of these values with respect to the zones relating accuracy and precision, the EN and Danish methods are more accurate than precise, whereas the other three methods are similar with regard to accuracy and precision. When the methods are assessed on the basis of the RD index, the Danish method is clearly the best, with the lowest RD value of 0.412, followed by the MEN method with a value of 0.465 and the EN method with a value of 0.476. However, if the RI value is used, the EN is found to be the best, with a value of 0.780, followed by the Danish method with a value of 0.861. The reason why the EN method is best based on the RI value while the Danish method is best based on the RD index is because, as noted above, due to the logarithmic scale and the way the accuracy and precision are combined. For methods with similar levels of precision and high levels of accuracy, more weight is given to the accuracy and less to the precision when calculating the RI value than when calculating the RD value. In the Authors’ view, the RD index, which gives equal weight to the accuracy and the precision, is a better parameter for comparing calculation methods. 5 CONCLUSIONS A rational procedure, based on the use of both conventional statistical parameters and a new parameter called the Ranking Distance, has been proposed for comparing calculated and measured values in geotechnics. This parameter has been shown to be better than the Ranking Index. An example involving published data for the bearing capacity of driven piles in normally consolidated clays has been chosen to demonstrate the use of this procedure. From the results obtained it is concluded that, on the basis of this published data, the EN and Danish methods, and to a lesser extent the MEN method, seem to be accurate and sufficiently precise calculation methods for determining the bearing capacity of driven piles in normally consolidated clays. However, the best method based on the Ranking Distance, which takes account of both the accuracy and the precision, is the Danish method.
Proceedings of Settlements 94, College Station. Texas, pp.640-650. Briaud J.L., Tucker L.M. (1998). Measured and predicted axial load response of 98 piles.ASCE J. Geotech. Engng. 1 14(9) 984- 1001. Cherubini C., Cucchiararo L., Orr T.L.L. (1995a). Criteria to compare calculated and observed bearing capacity of piles. VIII ICASP Paris, ~01.1.pp.9-14. Cherubini C., Cucchiararo L., Orr T.L.L. (1995b). Comparison between measured and calculated values in geotechnics. FMGM 4“’ International Symposium Bergamo, pp.267-274. Cherubini C., Greco V.R. (1997). A comparison between measured and calculated values in geotechnics. An application to settlements. Probamat 2 1 Century. Probabilities and Materials Perm (Russia) G.N. Frantziskonis Ed., pp.48 1-498. Kluwer Academic Publishers. Chow F.C., Jardine R.J. (1998). Improving confidence in pile design. Workshop on Prediction and Performance in Geotechnical Engineering, Nap0 li, pp .243-278. Cortellazzo G., Mazzucato A. (1998). Safety factors with the use of pile driving formulae. Rivista Italiana di Geotecnica. Anno XXXII, n.3, pp.4862. Kotzias P.C., Stamatopoulos A.C., Kountouris P.J. (1990). Exploratory graphics and geotechnical data: some introductory remarks. Geotechnical Engineering, v01.21, pp. 127-143. Li K.S., Lee I.K., Lo S.C.R. (1993). Limit state design in geotechnics. Probabilistic Methods in Geotechnical Engineering. Canberra. Li and Lo Eds. Balkema, pp.29-42. Ramey G.E., Johnson R.C. (1979). Relative accuracy and modification of some dynamic pile capacity prediction equations. Ground Engineering, vol. 12, n.6, pp.47-52. Tan C.K., Duncan J.M. (1991). Settlement of footings on sand. Accuracy and reliability. Proc. of the Geotechnical Engineering Congress, Boulder Colorado, pp.447-455. Velleman P.F., Hoaglin D.C. (1981). Applications, Basics and Computing qf Exploratory Dala Analysis. Duxbury Press Boston Massachusetts.
REFERENCES Berardi B., Lancellotta R. (1 994). Prediction of settlements of footings on sand: accuracy and reliability. Reprinted from Vertical and horizontal deformations of foundations and embankments. 265
This Page Intentionally Left Blank
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida(eds)0 2000 Balkema, Rotterdam, ISBN 90 5809 151 1
Case study of a failed embankment with consideration of progressive failure VChoa Nanyang Technological University,Singapore
H. Hanzawa Technical Research Institute, TOA Corporation, Yokohama,Japan
ABSTRACT: This paper presents failure investigation and inverse analysis of an embankment on a clay deposit that failed along a large-scale slip plane of over l OOm while a reclamation work was in progress in Singapore. From the evidences obtained from the field it was confirmed that the slip plane was different from circular shape. The failure, which can not be explained by the analysis based on the peak shear strength, can be characterized by two factors: 1) the clay at the site indicates higher brittle behavior than most clays in the country, and 2) the project site has a natural slope in front of it. Detailed stability analyses were carried out using the peak and residual strengths determined by recompression method using direct shear test. The result of analyses suggested that a small circular failure with peak mobilized shear strength was initially developed, and then quickly the slip plane progressed all the way through the slope in a chain reaction where residual strength was mobilized. 1 INTRODUCTION
Massive reclamation projects have been progressed to facilitate development of the existing island in Singapore. Some of them are COnstructed on coral sand islands which are commonly located on slopes,
as schematically shown in Fig. 1 . A big project started in 1995 and completed in 1999 is also shown in Fig. 1. In order to make an early working base along the coral sand island, higher dike was temporarily constructed at the initial stage of the project, During construction, a large scale slip suddenly took place. A construction inspector, who was working on the top of the dike, was witnessed throwing into the sea just in a few seconds. On the other hand, an extensive series of direct shear test, DST using Mikasa’s apparatus (Mikasa, 1960) was carried out in the investigation stage of the project to evaluate shear strength characteristics of the clay through recompression method (Jamilokowski et al, 1985). The results of DST indicated that the clay at the project site was much more brittle than the clays commonly found in the region. In this paper the shear strength characteristics of the clay at the project site are interpreted, and then the detailed stability analyses considering the stressdeformation characteristics are presented.
2 SHEAR STRENGTH CHARACTERISTICS OF THE MARINE CLAY
Fig. 1. Location and topography of the project site.
Fig. 2 presents the relationship between the peak direct shear strength, Su(d)P and effective overburden stress, CY’~,, from recompression method. The samples were taken from the nearest borehole at the slip plane.
267
Hanzawa and Tanaka ( I 992) reported from comprehensive study of clays found in South East Asia and Japan that Eq. (1 ) can be used in upper part of the clay deposits:
duction is negligible for upper two samples. It can be concluded from the Figure that the soft marine clay in the site is highly brittle and strain softening is significant from the depth of -4m to -12m. Fig. 4 presents the ratio of residual (S,,,,) and peak (S,,,,) shear strengths versus depth. Shear stress at 1.5mm after displacement of failure, D, is considered as residual strength (Dam et al. 1997), except residual strength for the upper two samples being determined at the maximum displacement in the test. The average value of the strength ratio for the clay from the depth of -4m down to -12m is 0.68.
where SU,,,= shear strength in-situ, S,,,,,,~= S,,!,, at ground surface, = strength increment ratio In normally consolidated state. S,JO’\ = 0.28 was obtained from DST for the soft marine clay at the site using samples brought to normally consolidated state. Applying Eq.( 1) and S,,,,/G’\= 0.28, the relationship between Sulr,,and din can be expressed by Eq. (2) as indicated in Fig. 2. 3 ACTUALPLANE S2,,,,)= 10 + 0.280’,., (kPa)
(2)
Fig. 5 illustrates the cross section of the slipped portion before and after the failure occurred. The Figure also shows the point resistance, qr from cone penetration test (CPT) measured near the failed section along with the drilling log obtained after the failure at the center of the slipped zone. 4 STABILITY ANALYSIS WITH CONSIDERATION OF PROGRESSIVE FAILURE Minimum factor of safety, FS,,,,,,for circular slip plane, S,(mnb) from Eq. ( 3 ) , using peak shear strength, was calculated as follows:.
Fig. 2. Shear strength obtained from DST
(3)
PR
“,{(d)[P]
“~(nroh)
where pR = a correction factor for strain rate effect. Analyses were conducted with the change of lR = 0.8, 0.9 and 1 .O, and the same circular arc was found from the analysis to give FS,,,, for each value of pR. The results are shown in Fig. 6. Although the Figure shows that the FS,,,,,,of the embankment is close to 1.0, the size and shape of the arc slip plane is very much different from the actual failure as compared in Fig. 5. It is confirmed for this case study that analysis based on S,,,,,,can not be adequate to explain the actual behavior of the ground.
Fig. 3. Stress-displacement curves obtained from DST
0
Fig. 3 shows stress-displacement curves obtained from DST for the samples obtained above the assumed slip plane. Remarkable reduction in stress after the peak is observed in the curves for the samples taken from under -5m. On the other hand, re-
2
4
6
8 10 12 14 16 18 Depth (m)
Fig. 4. S,,,,IS,,,, obtained from DST versus depth
268
Fig. 5. Embankment before and after the failure.
According to the above hypothesis, the failure plane should have started from the back of the embankment (land side) and tangent to the circular arc which gives FS,,,,,,.FS,,,,, for the first stage was already presented in Fig. 6. Regarding these factors, non-circular plane shown in Fig. 7 was introduced.1t should be pointed out that the analysis of the noncircular failure plane was carried out using the peak shear strength for the upper most 2m of the marine clay, since the reduction in stress after the peak was negligible. Fig. 8 shows the result of further analyses using non-circular plane performed using Eqs. (4) and ( 5 ) . The values of FSnli,,from both average and residual strengths ranges from 0.85 to 1.2, while peak strength gives FS,,,, greater than 1.17. The result of the analyses in the Figure verifies the hypothesis presented above.
Fig. 6. FS,,, for circular slip with peak shear strength
In order to realize the actual failure, the following hypothesis on the mechanism of failure consists of two stages is introduced. This hypothesis has considerations for the brittle stress-displacement behavior of the clay and the topographic condition , i.e., existence of a natural slope in front of the embankment.
1 ) First stage: Failure was initiated under the embankment along the small circular slip plane with FS,,,,,,close to 1 .O. Shear strength mobilized along the failure plane, Su(nloh) should be Su,plfor the first stage. 2) Second stage: Due to chain reaction and topographic condition, the failure plane progressed rapidly through the mass of the clay deposit especially to the direction from under the embankment to the natural slope in front of it, and the S,,(,,,,,,,)should be somewhere between Sulpl and SUlR1for the second stage.
(4)
5 CONCLUSIONS
A large non-circular slip took place for a temporary embankment at the beginning of a reclamation project, which was constructed on coral sand islands located on slopes. Stability analysis was carried out using the peak and residual direct shear strength SUlp, and SulR,determined by the recompression method.
Fig. 7. Circular and bi-linear slip planes used in the stability analysis to calculate FS,,,,,,
269
Fig. 8. Variation of Fs,,, for non-circular slip with pR using different shear strength values
The result strongly suggested the mechanism of failure explained in the second stages as follows: 1. First stage: Failure was initiated under the embankment along the small circular slip plane with FS,,,,,,close to 1.0, and shear strength mobilized Su[nlob]at this stage should be SUlpl. 2. Second stage: Due to chain reaction and topographic condition of the site, the failure plane was progressed rapidly through the mass of the clay deposit especially to the direction from under the embankment to the natural slope in front of it, and the Su,nlobl should be somewhat between SU,,,, And SUlR1in this stage. It should also be pointed out that appropriate evaluation of the shear strength of the clay on the slope will be also an important subject. For this purpose, such a simple investigation method like cone penetration test on the slope is strongly hoped to be done in routine work. REFERENCES Dam, T. K. L., Yamane, N., Hanzawa, H. and Porbaha, A. 1997. Evaluation of progressive failure of natural clay deposits. Proceedings of International S tnposium on deformation and sive faizre in geomechanics. Elsevier 199-204. Hanzawa, H. and Tanaka, H. 1992. Normalized undrained strength of clay in the normally consolidated state and in the field. Soils und Fozind~tion.732 (1): 132-148. Jamilkowski, M., Ladd, C. C., Germain, J. T. and Lancellotta, R. 1985. New development in fielcJ and laboratory testing of soils, Proceedings of 11 IC,'LTMFE 1 57-153. Mikasa, M. 1960, Direct shear device newly developed. Proceedings of 15"' JSCE annual conference: 45-48 (in Japanese).
~
I
?
270
~
~
~
~
~
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 15I
I
On-line earthquake response tests on embankments based on clay foundation T. Fujii - Fukken Company Limited, Hiroshima, Japan M. Hyodo - Department of Civil Engineering, Yanzaguchi Universiv, Ube,Japan S.Kusakabe - TechnicalResearch Institute, Okuniura Company Limited, Tsukuba, Japan Y.Ymarnoto - TechnicalResearch Institute, Mitsui Construction Company Limited, Chiba, Japan
ABSTRACT: The present research aims at evaluating the displacement and grasping failure mechanism of embankment on a saturated clay foundation caused by earthquake. From the results of dynamic centrifbgal model tests carried out in the past, three zones can be recognized in the the failure mechanism of clay foundation around embankment. The research in this paper is focussed on the zone that is contributing greatly to the failure of embankment, and one dimensional on-line earthquake response tests were conducted for this zone interconnecting the seismic response analysis with the element tests under the boundary condition of failure pattern. The suitability of the assumed failure pattern is confirmed by comparing the response values of the online earthquake response tests with those of the centrifbgal model tests. Moreover on-line earthquake response tests were conducted using numerous input wave patterns and attempts were made to determine the cumulative horizontal deformation of the embankment toe. 1 INTRODUCTION
2 CLASSIFICATION OF FAILURE PATTERN
The embankment structures constructed on soft alluvial foundation have suffered heavy damages during Kushiro-oki Earthquake( 1993), Hokkaido-Nanseioki Earthquake (1 993) and Hyogoken-Nanbu Eartquake (1 995). Taking these experiences as opportunities, numerous research works have been carried out with regard to embankment constructed on sand foundation. The results obtained from these research works have been utilized in some design guidelines and design procedures. On the other hand, the research works on embankments constructed on clay foundation are rather few and the present state of the art is that their design procedures are almost unchanged and similar to the conventional one. However, the embankments on clay foundation have suffered damages to some extent by those large earthquakes described above. Consequently it is highly desirable to establish in the near fbture the design procedures for the embankments on clay foundation as much as those for the embankments on sand. The present research aims at conducting studies on failure mechanism of embankments constructed on soft clay foundation taking the past dynamic centrifkgal model tests as the hypothesis. In that, attempts were made to conduct on-line one dimensional earthquake response tests that interconnect the direct shear soil tests with seismic response analysis by the use of computer with a view to enhance precise studies of all failure mechanism of foundation upon which embankment structures lay.
As to the model test on embankments with clay foundation, the Public Works Research Institute of the Ministry of Construction, has once conducted a test under the centrifbgal forces of 50g (Tamoto et al., 1997). In that test, an embankment with a height of 2 m was constructed on a soft clay ground made up of 5m thick soft alluvial clay under a l g field condition. And 4 stages of excitations of the order of 100ga1, 200ga1, 300gal and 400gal were applied. Fig. 1 shows the displacement condition before and after the final excitation. From this figure it is learnt that the behavior of foundation during earthquake is different due to the relative position of foundation with embankment; they are free zone, zone directly under the embankment and zone around the embankment toe. First of all, at the free zone of foundation (Zone I), the residual displacement is hardly observed after applying excitation and the zone is in a sound state. In contrast to this, at the zone directly under the embankment (Zone 111), the residual deformations in both the vertical and horizontal directions were built up. On the other hand, a large residual horizontal displacement occurs with increasing toward the free zone along the line from top of slope to toe of slope (Zone 11). And the pattern of this deformation due to displacement is a circular shape as shown in Fig. I Failure pattern, expected to be formed at each zone, element condition, effective stress path and stress-strain relationship all are shown in Fig.2. 27 1
Original shape
Crd-ck
/
_.
c
\
...........................
, Deformed shape
...............
:
I
I
/ / / /
/
/'
.:
:-6. Fig, 1 Result of dynamic centrifugal model tests
e
......
Zone I Position Horizontal ground Failure model Cyclic shear failurc
I
i
Condition o e1ement
...... ,_.."
Ii
II
m
Under toe of slope Sliding failure
Under embankment Shake down
1 1
0 vl L ' ddJ+
T (3
I
h2
Developing ' t h e s t r a i n )eveloping the residual amplitude ;train
Fig.2 Classification of failure modes 272
X Zd
Id
From the model tests' deformation pattern shown in Fig.2, it is learnt that among all the zones have been classified above, Zone 11, having the highest residual displacement, seems to be contributing to the failure of embankment. Moreover, the authors (Hyodo et al. 1999) pointed out from the results of numerous laboratory tests that the clayey ground is prone to cyclic shear failure mainly because of a high initial shear stress of the foundation around the structure. Consequently, Zone I1 was given a hll attention and also subjected to earthquake response tests to study the failure mechanism of the foundation around the embankment structure.
Numerical m Layer
o
-
FI-P ,
On-line tseting
Accg
eCanlrol and mcarurmicni l
d
i
I
I+I :
4
T I
/I+----+
e
1 .
Base
Fig.3 Conceptual flow for on-line testing
3 SUMMARY OF ON-LINE EARTHQUhKE RESPONSE TEST In on-line earthquake response tests, a computer-run seismic response analysis and pseudo-dynamic loading test to estimate the restoration force of materials are combined by computer on-line data processing system. According to this method, a seismic response analysis was made possible to enable us to evaluate the real behavior of soil without relying on complicated structural equations of soil. Kusakabe et a1 (1999) developed an on-line testing system of 6degrees-of-freedom using a 6 series hollow torsional shear apparatus and studied the earthquake behavior of the horizontal saturated sand deposits. Moreover the group conducted tests by developing a simplified simple shear apparatus that makes the manipulation easier (Kusakabe et al., 1999). These tests did aim at investigating onedimensional behavior of horizontal ground. However the present research scrutinized the two dimensional behavior of the embankment and the foundation by carrying out on-line tests. The general concept of on-line earthquake response tests when applied to the subsoils is shown in Fig.3. First of all, the layers under analysis was transformed into a lumped-mass model, and an earthquake motion was input from its base. And the equation of motion of the lumped mass model was solved by a computer and the response displacements were determined. Afterward the shear strain forces that were equivalent to the predetermined displacements were applied to the soil specimen. And using the automatically measured shear stresses, the next step's response displacement was computed. This process of computation was repeated continuously during the period of repeated earthquake motion. This means that the nonlinear shear stresses of the soil that changes with the change of time was directly determined from the specimen of the element tests and these were inter-related with the response analysis on a line of computer. Thus this method simulates the behavior of the foundation during earthquake.
Fig.4 On-line testing model Here in this study, the total on-line tests for a multi-layered subsoil shall lead not only to a complicated test procedure but also to an increased cost of the system. Hence, in the tests conducted in this research, the shear stresses were determined for only the layer element section where a large deformation was expected, and for other layers, a substructure method was applied by obtaining the shear stresses through the use of a numerical model. 4 TESTPARAMETERS
The test section subjected to the analysis is the same section that has been used in the dynamic centrihgal model tests conducted by the Public Works Research Institute of the Ministry of Construction. The pattern of the section and the soil layer classification are shown in Fig. 4. Here in this study, the analysis was carried out by dividing the section under scrutiny into four layers S1 - S4 as shown in Fig.4. In the tests, the layer that was highly expected to undergo a large deformation during applying excitation is the upper clay layer that is close to the embankment as is obvious from the deformation pattern shown in Fig.1. Therefore this upper clay layer (S2) was taken as an on-line layer and the other layers namely the embankment layer (SI), lower clay layer (S3) and sand layer (S4) were brought under test as non-linear elastic elements. And the section subjected to close analysis was a two-dimensional one. The 273
Fig.5 Input acceleration and responding acceleration The specimen that was used for testing was an Arakawa clay (Gs:2.622, wn:51%, I,: 17, C,/p':0.52) that was remolded by a pre-consolidation pressure of o',=SOWPa and it was the same clay test sample that was used dynamic centrifkgal model tests. The other testing and analyzing parameters were set up based on the results, obtained from centrihgal model tests. Moreover the input acceleration waveforms were of three types, namely vibrational-type wave, shocktype wave and sinusoidal wave. These acceleration waveforms were used in centrifbgal model tests. And the earthquake forces were applied from the lower edge of sand layer. 5 BEHAWOR OF CLAY UNDER VARIOUS WAVE PATTERNS
tests on this section were conducted by transforming the section into one dimensional element type model as shown in Fig.4. The initial shear stress was generated on and around the toe of the slope due to the dead weight of the embankment. Taking that fact into consideration, a static slip circle analysis was conducted to find out the average shear stresses acting at around the toe of the slope, and with these stresses applied in advance at the test specimen under drained condition, on-line tests were carried out.
The input wave patterns and the response accelerated wave patterns at each element point that were obtained when on-line response tests were conducted under the maximum acceleration degree of amax =200Gal are shown in Fig. 5(a), 5(b) and 5(c) for the vibrational-type wave, the shock-type wave and the sinusoidal wave respectively. From this figure, it is confirmed that the wave period is lengthened at the upper clay layer for any type of wave pattern. When each response values are investigated, all wave patterns tend to amplifL at the embankment section (S l), but they were not amplified nor damped at sand layer (S4) and clay layers (S2, S3).
274
This tendency coincides with the results of acceleration degree that were obtained from the dynamic centrihgal tests. Fig.6 shows effective stresses paths for acceleration waveform patterns those are similar to that used in Fig.5, and Fig.7 shows the relationship between shear stress and shear strain. The Critical State Line (CSL), obtained from static tests, is also shown in Fig.6. From these figures, it is learnt that as a result of cyclic shear stresses, the effective stress path has gone up to the vicinity of the critical state line for all types of accelerated waveform pattern. And it is also confirmed that the residual shear strain tends to be largely seen in the direction where the initial shear stresses are acting. This tendency coincides with the result of occurrence of a large shear deformation at around the toe of the embankment in the dynamic centrifbgal tests. This suggested indirectly the fact that an injurious deformation took place during earthquake at around the structure where initial shear stresses are acting even if it was lying on a clay foundation.
from the on-line earthquake response tests, conducted in the present research, and the result of l g gravity conversion of the toe-of-slope's horizontal displacement after excitation in the dynamic centrifbgal model tests. From this figure, it is learnt that the online earthquake response tests produced an overestimated result larger than that obtained by the dynamic centrifbgal model tests. As to the difference of residual deformation due to the difference of waveforms pattern, the qualitative approach was quite successfbl. The following two factors were considered to be the reasons why the residual deformation quantity of the on-line earthquake response tests differed from that of the dynamic centrifbgal model tests. One factor is that the dynamic centrifbgal model tests were conducted under a high frequency excitation of 60 Hz as the latter is under the centrifbgal field of 50g and the other factor is that the strain velocity input to the test specimen was rather slow due to the limited capacity of the test apparatus, used in the on-line earthquake response tests.
6 COMPARISON OF DEFORMATION QUANTITY
7 CONCLUSION
Fig.8 shows the result of comparison of the on-line layer's horizontal displacement that was obtained
1. The acceleration waveforms pattern, obtained in the on-line earthquake response tests, were com-
Fig.6 Effective stress paths
Fig.7 Stress-strain relations 275
paratively in conformity with those, obtained from the dynamic centrifugal model tests.
REFERENCES
2. A large shear deformation that was observed around the slope toe of the embankment in the dynamic centrihgal model tests was also confirmed by the results from th-e on-line earthquake response tests. This fact suggested indirectly that an injurious deformation took place during earthquake even in the case of clay foundation. 3. The deformation that occurred in the on-line earthquake response tests was observed to be higher when compared with the result, obtained from the dynamic centrihgal model tests. Thus it was possible to grasp qualitatively the difference of deformation due to the difference of waveform pattern. From the foregoing results, it is to be concluded that the method, stated in this research, is considered to be an efficient tool to predict the duringearthquake behavior of the embankment on a soft clay foundation.
Fig.8 Comparison between results of on-line tests and results of measured centrihgal model tests of toe-of slope’s horizontal displacement
276
Tamoto, S., Matsuo, 0. & Shimazu, T 1997. Dynamic centrifugal model tests for embankment on clay ground (part2), Proc. of 32th Japanese National Soil Mechanics Research Meeting, JSSMFE: 1021- 1022(in Japanese). Hyodo, M., Hyde, A.F.L., Yamamoto, Y. & Fujii, T 1999. Cyclic shear strength of undisturbed and remoulded marine clays, Soils and Foundations, 39(2): 45-58. Kusakabe, S., Morio, S. & Arimoto, K 1990. Liquefaction phenomenon of sand layers by using on-line computer test control method, Soils and Foundations, 30(3): 174-184. Kusakabe, S., Morio, S., Okabayashi, T., Fujii T. & Hyodo, M 1999. Development of a simplified simple shear apparatus and its application to various liquefaction tests, Journal of Geotechnical Engineering, JSCE, 6 17(III-46): 299-304(in Japanese).
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 I
A numerical model for consolidation based on microscopic consideration S. Fukuhara Graduate School of Science and Engineering,Kagoshima University,Japan
H. Shikata Local Governmentof Sasebo-Ci& (FormerlyGraduate School of Science and Engineering,Kagoshima University, Japan)
R. Kitamura Department of Ocean Civil Engineering,Kagoshima Universig,Japan
ABSTRACT: A numerical model for one-dimensional consolidation is proposed based on some microscopic consideration in particle size. In the proposed model, the volume change in pore water due to the dissipation of excess pore water pressure is calculated as the primary consolidation behavior by the numerical model for voids. The numerical experiment is carried out to investigate the one dimensional consolidation behavior by the numerical model. On the other hand the one dimensional consolidation test on volcanic ash clay is carried out by the separate type’s oedometer testing apparatus, which is manufactured in our laboratory. The distribution of excess pore water pressure in the specimen and its change with time are investigated. The test results are used to examine the validity of proposed model by comparing the numerical results with those of oedometer test.
1 INTRODUCTION Kitamura et al. (1998) proposed a numerical model for the seepage behavior of unsaturated & saturated soil based on the mechanical and probabilistic consideration in particle size. Here it is called Kitamura’s model for Seepage. The unsaturated and saturatccl permeability coefficient can be calculated by Kitarnura’s model. The primary consolidation behavior is considered to be the volume change due to the dissipation of excess pore water pressure, i.e., the compression behavior due to the drain of pore water caused by the distribution of excess pore water pressure. In this paper an attempt is made to apply Kitamura’s model to analyze the primary consolidation behavior of saturated soil.
by a part of pipe and other impermeable part as shown in Fig. l(b). The diameter D and inclination angle of pipe 8, and the height DH of element are the model parameters which can express the condition of element as shown in Fig. l(a). In Kitamura’s model the diameter D and inclination angle of pipe 6 ‘ are regarded as random variables, and the probability density functions of D and 8 are used to estimate the unsaturated-saturated permeability characteristics. Then the void ratio and permeability coefficient are derived as follows.
2 NUMERICAL SIMULATION 2.1 Kitarnurn s Model for Seepage Soil is generally a multi-phased material which is composed of soil particle (solid phase), pore water (liquid phase) and pore air (gas phase). The soil structure and the void distribution in soil are random and difficult to be estimated quantitatively because the shape and size of soil particle are irregular. Consequently the soil structure and the distribution of voids in soil are random. Then the probabilistic consideration should be introduced to estimate them in soil. Fig. l(a) shows an element in which a few soil particles are included. This condition can be modeled
Fig. 1 Modeling of soil element
277
where D=m corresponds to the saturated condition for permeability coefficient. Yamaguchi et aL(1992, 1993) carried out the experiment to investigate the void distribution of clay and obtained the result that the void distribution may be expressed by two logarithmic normal distributions. Referring to this result, the following equation is used as the probability density function of p d ( D ) .
Fig.2 Probability density function of Pc( 6’)
where pd@): probability density function for macro-pore, pdz(D): probability density function for micro-pore, R mean value of p.d.f. for macro-pore, /i 2: mean value of p.d.f. for micro-pore. y 1: standard deviation of p.d.f. for macro-pore, y 2: standard deviation of p.d.f. for micro-pore.
Fig. 3 Change in excess pore water pressure with time Vni
Pc( 8)is assumed to be a shape of pentagonal as shown in Fig. 2. Then the probability density function can be expressed by the following equation.
i (m+lPayer m-la yer I r r
“m-1
Where a=l for - ~ J 2 6 5 SO, a=-I for 0 5 B 5 ~ 1 2 , [c : height of Pc( 0 ) at @ = 2 ~ 1 2 .
The height DH of element in Fig. 1 is assumed to be same as Dlo (diameter finer than 10 %) obtained from the grain size distribution curve.
(m-1)-layer
Fig. 4 Flow of pore water through adjacent layers Figure 4 shows any three adjacent layers in Fig. 3(b). The volume of pore water flowed from the mth layer to the (mt1)-th layer in the time increment At is calculated by using Darcy’s law as follows. V,,,, = k, i, S At
(7)
2.2 Modeling of otie dimensional Consolidation Figure 3(a) shows the excess pore water pressure head just after the load increment A p is applied to the specimen which is equilibrium under the load of p. The height of pressure head is same for each layer of the specimen. Figure 3(b) shows the excess pore water pressure head after a while. The pore water is drained from the top layer and thc pore water pressure head is distributed in the specimen.
Where Vmw:Volume of pore water flowed from the m-th layer to the (m+l)-th layer in the time increment At, k,: permeability coefficient of the m-th layer, i,: hydraulic gradient of tile ni-th layer, S: cross section area of specimen, A t : time increment. The volume change of the m-th layer due to thc drainage of pore water is expressed by the following equation.
Where A Vmw:Volume change of the m-th layer in time increment At. The change in pore water pressure during the time increment A t is assumed to be expressed by the following equation. Fig. 5 Separate type-oedometer testing apparatus
(9) where ( AP,),: change in pore water pressure from time ‘j’to time ‘(j+l)’, ( A Vmw),: change in volume of the m-th layer from time ‘j’to time ‘(j+l)’, (Kne),:volume of void in the ni-th layer at time ‘j’, ( AP):total stress increment. Using Eq.(9), the pore water pressure at time ‘j’ is obtained by the following equation.
Fig. 6 Grain size distribution curve of volcanic clay (10)
3.2 Material and Test Procedure The material used is a volcanic clay which was sampled at Harihara, Izumi-City Kagoshima Prefecture. The physical quantities are listed in Table 1 and the grain size distribution curve is shown in Fig.6 The material sampled as several blocks was initially soaked in water to be the water content of about 200 %. Then the slurry is poured into the preconsolidated cell, de-aired and consolidated to get the self-su orted specimen under the pressure of 19.6 kN/m . The self-supported specimens are set in three consolidation cells and one-dimensional consolidation test is started. The back pressure of 98 kN/m2 was applied to ensure saturation of specimens. Loading stage is same as that of standard oedometer test, i.e., load increment ratio ApIp=l, the time of one loading stage is 24 hours and the loading proceeds to the 4th stage of 314 kN/m2.
3 CONSOLIDATION TEST BY SEPARATE TYPE-OEDOMETER
3.1 Apparatus Figure 5 shows the separate type-oedometer testing apparatus. This apparatus was made by our laboratory, referring to that developed in the Yokohama National University (Imai and Tang, 1992). The separate type-oedometer testing apparatus is composed of three consolidation cells inter-connected with each other. In each cell a standard oedometer consolidation ring is set. The size of each specimen is G cm in diameter and 1 cm in height, i.e., the height of total specimen is 3 cm with single drainage. Three consolidation cells are named No.1, 2 and 3 as shown in Fig.5, where the upper end of No.1 cell is the drained boundary. Axial load is applied by supplying air pressure to the cell. The volume of pore water drained from the specimen is measured by an electric balance installed in a pressure chamber. The axial compression of each specimen is measured by a non-touched laser displacement device. The pore water pressure in each specimen is measured by the pressure transducer which is set under the pedestal of each cell. The data obtained by the consolidation test are automatically acquired and processed by a personal computer. Furthermore, the test procedure is controlled by the personal computer.
P
3.3 Test Results Figure 7 shows the relation between the axial strain and elapsed time obtained by the loading stage of 157 kN/m2. Consolidation initiates at the drained side cell of No.1, followed by the intermediate cell of No.2 and the undrained side cell of No.3. The final axial strain is largest at the drained side of No.1 cell, which proves the distribution of excess pore water pressure in the specimen for the standard oedometer test.
279
gest at No.1 cell, which is followed by No.2 and No.3, and the dissipation time is also the shortest at No.1 cell. Table 1
Physical quantities of volcanic clay Ps(g/cm3) [ 2.76
1 Density of soil particle
Fig. 7 Relation between axial strain and elapsed t h e (157 kN/m2)
I
Values of model parameters and experi Table 2 mental condition
Fig. 8 Relation between excess pore water pressurc rate and elapsed time (157 kN/m2) 4 NUMERICAL SIMURATION
Model for voids
Figure 9 is the flow chart of the calculation procedure in the numerical simulation. The values of model parameters and condition for simulation are listed in Table 2. The height of element in Fig.l(b) is same as Dlo obtained from Fig.6. The variances in Eqs. (4)and ( 5 ) are obtained by the assumption that the coefficient of variation are same as that of grain size distribution. The mean value 2 in Eq.(5) for micro-pore is assumed to be 1/100 of that in Eq.(4) for macro-pore. The mean value 12 in Eq.(5) is reversely obtained so that the initial void ratio for simulation is same as that of the specimen for separate type-oedometer test. According to the proposed model the numerical simulation can be carried out by using the grain size distribution curve and several physical quantities for pore water. Figure 10 shows the relation between axial strain and elapsed time obtained by numerical simulation and consolidation test. Figure 11shows the change in the distribution of excess pore water pressure with time obtained by numerical simulation and consolidation test. It is found out from Figs. 10 and 11 that the proposed numerical simulation method can follow the one dimensional conso~idation behavior of clay qualitatively.
Fig. 9 Flow chart of the calculation procedure Figure 8 shows the relation between the excess pore water pressure rate and elapsed t h e . The dissipation rate of excess pore water pressure is the lar280
ACKNOWLEDGEMENT We would like to express our sincere appreciation to Prof. Imai and the later Associate Prof. Pradhan for their valuable advice and support concerning the making of separate type-oedometer testing apparatus.
001
01
1
10
REFERENCES
1000
100
Elapsed Time (min )
Imai, G. & Tang, X.Y. 1992. A constitutive equation of one-dimensional consolidation derived from intercoiuiected tests. Soils and Foundations, 32(2):83-96. Kitamura, R., Fukuhara, S., Uemura, K. & Seyarna, M. 1998. A numerical model for seepage through unsaturated soil. Soils and Foundations, 38(4): 261-265. Yaniaguchi, H., Hashizume, Y. & Ikenaga, H. 1992. Change in pore size distribution of peat in shear processes. Soils and Foundations, 32(4): 1-16. Yamaguchi, H. and Ikenaga, H. 1993. Utilization of mercury intrusion porosimetry apparatus for evaluation of soil structure. Tsuchi-to-Kiso, 41(4): 15-20 ( in Japanese).
Fig. 10 Relation between axial strain and elapsed time obtained by numerical experiment and consolidation test 1
Loadlng Stage 7 s 45 156 9 1 kNim Elapsed Time (sec ) -=-O -@-XI -A-U -v-300 -X-3600
-+-7200
0.0
0.2
---216W
0.4
-
--+-600 --lEM) I -66400(37570)
-
0.6
0.6
1.0
U/AP
Fig. 11 Change in the distribution of excess pore water pressure with time obtained by numerical simulation and consolidation test 5 CONCLUSIONS
The consolidation test on a volcanic clay was carricd out by the separate type-oedometer testing apparatus. The numerical simulation was also carried out based on Kitamura’s model for seepage. The proposed numerical simulation method can simulate the primary one dimensional consolidation behavior of clay qualitatively. This research work may bc regarded as thc first step to establish a consolidation theory based on the discontinuous mechanics. In the proposed niodcl the unsaturated condition can casily be taken account of, which means the model is promising to bc developed for the synthetic soil mechanics.
281
This Page Intentionally Left Blank
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 157 1
Soil nailed structure in soft clay Lang Gao, Kang-He Xie & Hong-Wei Ying Department of Civil Engineering, Zhejiang University,Hangzhou, People’s Republic of China
ABSTRACT: Soil nailed structure has been proved to be a type of economical and effective excavation support system and played an important role in geotechnical engineering. But it is mostly used in such type of soil as sand, silt and stiff clay and seldom used in soft clay, especially in saturated mucky clay. In this paper, two soil nailed structures recently constructed in soft clay in China are introduced. A new method so-called Secondary Grouting is applied in one of the projects firstly. The result of the field pull-out test indicates that the new technique can increase the lateral resistance of bars efficiently. The horizontal displacements are measured during construction. It has been shown that the critical factors affecting the horizontal displacement and stability of the structure are the design parameters, the excavation sequence, the excavation depth and the construction time. It is important to set the steel bar in time after excavation to decrease the horizontal displacement. Finally, the conclusion is drawn that soil nailed structure can be used in soft clay.
1 INTRODUCTION Soil nailed structure is a relatively new type of retaining structure, which has been well developed in the last two decades and used widely in many countries (Shen et a1 1981, Bruce and Jewel1 1983, Gassler 1990, Milligan and Tei 1998). In China, the first soil nailed structure was constructed in 1982 in the slope project of Liu Wang Coal Mine in Shanxi province. Since then, soil nailed structure has been widely applied in many provinces and cities of China, such as Beijing, Guangzhou, Shanghai and Hangzhou, and produced good results not only in service but also in economy (Chen and Cui 1997, Gao 1998, Wang and Gao 1999, Li et a1 1999). But up to now, the use of soil nailed structure is almost limited in such type of soil as sand, silt and stiff clay, and seldom used in soft clay, especially in saturated mucky clay because of the low strength of the soil. It has been argued for many years by geotechnical engineers that if soil nailed structure can be applied in soft clay. Recently, two soil nailed structures were constructed in soft clay in two coastal cities of China. In one of the projects, a new technique so-called Secondary Grouting is applied firstly in China. The advantage of the new technique, according to the results from field pull-out test, is to increase the lateral resistance efficiently. As one of the important features of soil nailed structure, it is convenient to adjust design parameters during excavation to make design more rational and to safeguard the stability of construction. Field monitoring that
provides actual information relating to the stability of the structure is thus necessary. In both of the two cases, field measurements of the horizontal displacements of the soil nailed structures were carried out. To ensure the measurement faster and correct, the software JKJC (Xie et al, 1997) was used in field data processing. All above are introduced and discussed in this paper.
2 THE CASE IN SHANGHAI The construction site is located in Minhang district in Shanghai. The excavation site can be divided into three areas according to three separate excavation depth (i.e. 4.lm, 5.2m and 7.lm respectively). Steel bars are inserted along the boundaries of excavation. The parameters of physical and mechanical properties of soils are listed in Table 1. The typical section of the soil nailed structure is shown in Figure 1. The ground water level is 0.65-1.30m to the ground surface. Well point drainage is adopted during excavation to reduce water pressure. The forepoling bolts are set to avoid deep slipping. The external and internal stability of this soil nailed structure is calculated, which indicates that both of them can meet the need. The horizontal displacement of the soil nailed structure and the surface settlement are measured during excavation. The main measurement results in the area of 7.Im excavation depth are listed in Table 2. From the table, we can see that: 1. The maximum horizontal displacement at the ground surface is 18.4 mm, which is only about 0.26% of the 283
depth of excavation. At the bottom of the excavation, the relevant horizontal displacement is 1 1.1 mm. These show that the soil nailed structure is safe and the support system of excavation is effective. 2. When two soil layers were excavated but the second row of steel bars were not installed yet, the maximum lateral displacements were 5.26 mm at the ground surface and 4.79 mm at the depth of 2.2 m. This shows that the displacements are greater than those expected. Therefore, the design parameters were adjusted and some measures were taken to safeguard the soil nailed structure. The steel bar was lengthened to 13 m (the original one is 11 m in length) and installed as soon as possible. The forepoling bolts were inserted ahead of previous schedule. The final results indicate that these measures are effective. 3. The horizontal displacement is not only affected by parameters of support system but also affected by the sequence of excavation, the excavation depth of each stage and the time of installing the steel bars. The stability of each construction stage is very important to soil nailed structure because it is the most dangerous when soil is excavated to a certain depth but bars are not inserted yet at each stage. Consequently, it is necessary to calculate stability at every stage to ensure safety and special attention should be paid to the excavation sequence and the stage excavation depth.
f
otcrete 110-
3
-
.... 0
3
Figure 1. The typical section of soil nailing in Shanghai
' 1000
-0.50111
1720
3 THE CASE IN HANGZHOU The site is located in the Xiacheng district in Hangzhou. The depth of excavation is 4.3 meters. The parameters of physical and mechanical properties of soils are listed in Table 3. The ground water table is 1 meter to the natural ground. The surcharge is 20 kPa. The typical section of excavation support system is illustrated in Figure 2. Groundwater control systems are installed behind the soil nail/shotcrete wall to decrease the groundwater level during construction. In order to increase the lateral resistance between the grouting and soil, a new method so-called Secondary Grouting is used in this project firstly. This method can be described as follows: a. Drilling hole b. Grouting and cleaning the hole(exchanging the slurry for the grout ) c. Blocking the hole and performing the secondary grouting. T i check the effectiveness of the new technique, the field pull-out test was carried out. The test results indicate that the lateral resistance of bars using this method increased 20% or more. During construction, nine inclinometer guide tubes are installed to measure the horizontal displacement and referred to as I1 to I9 respectively. The measured data via inclinometer is processed and plot into diagram immediately by the software JKJC installed in notebook computer (Xie et al, 1997). The observed results from guide tube I3 are illustrated in Figure 3 to Figure 6.
-4.80111 4 -
Figure 2. The typical section of soil nailing in Hangzhou
Figure 3. The horizontal displacement from I3 (3/15/99-4/16/99)
284
Figure 5. The horizontal displacement from 13 (5/02/99-5/14/99)
Figure 4. The horizontal displacement from I3 (4/16/99-5/02/99)
Soil name Plain fill Silty clay Mucky silty clay Silty sand Muckyclay Sandy silt Mucky silt
~
uepth of excavation (m) 0 1.1 2.2 3.3 4.4 5.5 6.6 7.1
Soil type
Average thickness (m) 1.01 1.12 0.75 0.25 0.55 1.95 5.80
Unit weight y( kN/m3) 20.0 18.9 18.0 19.3 18.0 19.2 11.2
Friction angle
("1 15.5 10.7 10.3 25.0 13.0 22.1 9.4
Cohesion (kPa) 16.0 11.5 10.0 5.0 10.0 6.8 8.4
Table 2 Measured horizontal displacement Measured horizontal displacement in Finial horizontal displacement after each construction stage before installing completing construction (mm) steel bars (mm) 5.26 18.4 3.55 13.5 4.79 0.62 12.5 0.38 12.3 1.9 __ 1.8 0.3 11.1
Average thickness (m)
Unit weight y( kN/m3)
Friction angle
("1
Cohesion (@a)
Phin fill
17
18 5
15 0
150
Alluvial soil Silty clay Sandy silt
0.5 1.7 1.5
17.0 19.5 19.3
5.0 14.3 25.0
5.0 10.5 15.0
From these figures, it can be seen: 1.The horizontal displacement increases gradually during construction and the maximum displacement is 39.lmm, which is 0.91% of the excavation depth. The ratio is greater than the one of the case in Shanghai. The most important reason is that the total constructiontime of the case in Hangzhou is 89 days, while the one in
Lateral resistance (Wa) 41) 15 60 60
Shanghai is only 24days. The longer construction time result in greater displacement because of the creep of soft clay. 2. The maximum horizontal displacement occurs near the excavation bottom during construction. Accordingly, if some measures can be used to control the displacement at the bottom, the maximum horizontal dis-
placement of the whole support system will be decreased. For instance, the forepoling bolt is just an efficient method, which has been used in the case in Shanghai.
REFERENCES Bruce, D. A. and Jewell, R. A. 1983. Soil nailing: application and practise-part 1. Ground Engineering. Chen, Z. Y. and Cui, J. H. 1997. Application of Soil Nailing in Foundation Excavation. Beijing: Chinese Building Industry Publishing House (in Chinese). Gao, L. 1998. Behavior analysis of Soil Nailed Structure in Deep Excavation. MS Thesis, Dept. of Civil Eng., Zhejiang University, Hangzhou, China. (in Chinese) Gassler, G. 1990. In-situ techniques of reinforced soil. In MCGOW, A. K., Yeo, C. and Andrawes, K. Z. (eds), Perjbrmance of reinforced soil structures:185-196. London: Thomas Telford House. Li, Y . L., Li, L. and Zeng, X. M. 1999. Stability analysis and performance control of shortcrete-bolting-mesh support of foundation pit of a building in Shanghai. Chinese J of Geotech. Eng. 21(1):77-81 (in Chinese). Milligan, G. W. E. and Tei, K. 1998. The pull-out resistance model soil nails. Soils and Foundations. 38(2):179-190. Shen, C. K., Bang, S., Romstad, K. M., et al 1981. Field measurements of an earth support system. J. Geotech. Engineering Division, ASCE, 107(12):1625-1 642. Wang, Z. Q. and Gao, G. L. 1999. Application of soil layer anchor bars in soft ground deep foundation pit supporting. Comtruciii;;: Technology,28(9): 41-42 (in Chinese). Xie, K. H., Li, Q. L. and YU, Z. H. 1997. Application of objectoriented programming technique and software JKJC in deep excavation monitoring. Proc. China-Japan Joint Symposium on Recent Development of Theory and Practice in Geotechnology: 339-344, Shanghai.
4 CONCLUSION The following conclusions may be drawn fi-om the study: 1. As a type of economical and effective excavation support system, Soil nailed structure has been used widely in sand, silt and stiff clay. The two projects introduced demonstrate that soil nailed structure can also be used in soft clay. 2. It is very important to measure the displacement, adjust the design parameters, adjust excavation sequence and the excavation depth of each construction stage during construction. Soil nailed structure makes these possible and simple. 3. A new method so-called Secondary Grouting is applied firstly in China which can increase the lateral resistance of bars efficiently. 4. The horizontal displacement is not only affected by parameters of support system but also affected by the sequence of excavation, the depth of each excavation stage and the time of installing the steel bars. The shorter the time of construction, the smaller the displacement will be. 5. The stability at construction stage is very important to soil nailed structure. At each construction stage, it is the most dangerous when the soil is excavated to a certain depth but steel bars are not inserted yet. Thus, it is necessary to calculate stability at each stage to ensure safety. Special attention should be paid to the excavation sequence and the stage excavation depth. ACKNOWLEDGEMENT The financial support from the National Nature Science Foundation of China (No.59738160) is gratefully acknowledged.
Figure 6. The horizontal displacement fi-om 13 (5/14/99-6/11/99)
286
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5609 151 1
Case studies on six earth structures constructed on soft clay deposits H. Hanzawa, T. Kishida, T Fukasawa & K. Suzuki Technical Research Institute, TOA Corporation, Yokohama,Japan
ABSTRACT: This paper presents the case studies of six earth structures constructed on soft clay deposits in which three of them were failed. Stability analyses were carried out using shear strength obtained from various laboratory and field tests. The results of stability analyses demonstrate that recompression method gives superior results describing the actual behavior of the structures in the filed.
1 INTRODUCTION
2 OUTLINE OF CASE HISTORIES
Six structures were constructed on soft clay deposits under minimum possible safety. Because of a small difference between designed and constructed structures, some structures failed immediately after the completion. On the other hand, the structures were designed with the mobilized shear strength such as recompression method, RC method, unconfined compression test, UCT, field vane test, FVT and other methods. RC method with &-consolidated triaxial comapression and extension tests, &TCT and & E T were performed in the first three, while direct shear test, DST was applied in other three cases. This paper presents and compares factors of safety calculated by various mobilized shear strength, Su(mob) with actual behaviors of structures.
Locations of the projects described in this paper are shown in Fig. 1. A summary of each project is described here: 1) Fao Steel Jetty, Iraq (1976): Fao Steel Jetty was constructed at the river mouth of the Arab River. Severe stability problem took place immediately after commencement of construction when the minimum factor of safety, FS,;, was evaluated by unconfined compression strength, Su(ncr).A special property of Fao clay was brought to light through field and laboratory investigations carried out parallel to construction. RC method with both &TCT and GTET, to determine shear strength in-situ, S,, was developed through the investigation in this project. 2) Al-Zubair Embankment, Iraq (1978): Five embankments for preloading were rapidly constructed on Ai-Zubair clay. In order to shorten
Figure 1. Locations of each project site.
287
the preloading time, embankment was designed with FS,,,,,,= 1.05 using the Su(,llo,,) determined by S,,,.from RC method (with K,TCT and K,TET). Because of a slight difference between designed and constructed embankments, one embankment failed immediately after construction and two others were probably on the verge of failure. Daikokucho Dike, Yokohama, Japan (1981): A temporary dike for reclaimed land was constructed on a high-plastic marine clay. Immediately after completion, it failed and sunk into the sea. RC method (with K,TCT) was carried out together with UCT and FVT in order to investigate strength properties and the cause of the failure. Strength anisotropy was studied with K,TCT and K,TET using samples in normally consolidated state, and was applied for determining Su(moh) from RC method. Cone penetration test, CPT was as well conducted to detect the failure plane under the sunken emban kment. Banjarmasin Revetment, Kalimantan, Indonesia (1 989): The revetment was initially designed to be constructed under multi-loading with support of soil improvement by vertical drain because of inadequate Su(moh) from UCT. Immediately after the contract, Su(mnh) was newly determined by RC method (with DST) and FS,,,,,,= 1.06 was obtained without multi-loading. The revetment was then constructed in one stage. Kameda Embankment, Niigata, Japan (1 993): An expressway embankment was constructed on peaty subsoil with multi-loading. When the height of embankment was rapidly increased from 4.5m to 6.3m in three days, a large deformation took place together with tension crack and heave. Su(nloh)were determined by RC method (with DST)., UCT and FVT. Vungtau Revetment, Mekong Delta, Vietnam (1996): Vungtau revetment is a fisher port constructed in Mekong delta. About 2m reclamation is made behind the revetment, while excavation reaching 5m in maximum in front of it. FS,,,,,, values from Su~l,c-i-~ and vane shear strength, Su4v) were 0.41 and 2.85, respectively,
Project Fao Khor AI-Zubair Daikokucho Banjarmasin Kameda Vungtau
K,TCT
K,,TET
0
0
0
0
0
0
DST
0 0
0
FVT
UCT
0 0
0
0 0 0 0
0 0 0 0
at the commencement of construction. In order to evaluate more accurate FS,,,,,,value, portable CPT was carried out parallel to the construction. Its point resistance was related to S,~(nl,,h) from RC method using clay samples from a different location. The revetment was then constructed with FSl,,,,>=1.25 and safely completed. Case studies of projects 1 to 3 were described in detail by Hanzawa (1983), project 4 by Subagio ( 1 991 ), project 5 by Hanzawa, et al ( 1 994), and finally project 6 by Hanzawa (1 998).
3 CHARACTERISTIC FEATURES OF EACH CLAY Details of engineering properties of clays encountered were already described in the papers referred above. Their characteristics are briefly summarized in this section. Fao clay, with plasticity index, I, = 10-30, is characterized by indicating significant difference in consolidation yield stress, CT',when subjected to different stress increment ratio in oedometer test. Noticing this feature, a practical technique to RC method was developed. AI-Zubair clay with I, = 30-35 found about 60km west of Fao has been subjected to complicated aging effect such as desiccation (surface 5m), cementation (5m-7.5 m) and secondary compression (beneath 7m). Daikoku-cho clay with I, = 40-60 is a typical marine clay found in Tokyo Bay, but contains relatively large shells and sand seams. Banjarmasin clay with I, = 40-1 10 is divided into the upper and the lower clay bounded by the desiccated clay formed when the sea level was lowered about 10,000 years ago. Kameda cohesive soils consist of peat with w, = 100-300% and sandy clay with wN = 5070%. Vungtau clay is divided into the upper (I, = 20 -40) and the lower (I, = 20).
CPT
I
01 Highly aged 0 1 Contains sand and shell 01 ' High plastic 0 2 j Peat Low to moderate plastic
03
I
288
Clay feature
' Low plastic
Time of investigation During construction Design stage After failure After contract After contract Design stage and after contract
Field and laboratory tests carried out in each project are summarized in Table 1 together with features of each clay and the time of investigation. It should be noted that RC method was entirely adopted to determine Sll,of clays in all the projects. In this method, clear S,,,-such as Sus,)(compression), SuIld) (direct shear) and S,,,,,, (extension) are obtained from K,TCT, DST and K,TET. In the first three projects from 1976 to1981, KOTCT and KOTET, which require complicated and high quality techniques, were used. After finding that Sutld) compensates strength anisotropy (Hanzawa et al. 1992), DST developed by Mikasa (1 960) replaced the position of laboratory test since then. Correlations among various shear strengths and S,, from RC method are presented in Fig. 2.
where S,,, = shear strength in-situ measured by any method, p = a combined correction factor for strength anisotropy, strain rate and so on; a = a correction factor for the change of shear strength during construction (a>=l.O for loading and a4.0 for unloading); and p = a correction factor for progressive failure (p = 1 .O for usual cases). S,,I,, and Su(,lCT) have long time been used as Su(n,oh) based on local empirical approach with an asis sumption of p x a x @ = 1 .O. When &) or SulId) used as S,, in Eq. (l), Su(moh) must be corrected with pAand pR for S,,,,, and pR for SUS,,) as given by Eq. (2). Lyri(nrob)= sr!f (c)
4 DETERMINATION OF Sll(moh) FOR STABILITY ANALYSIS The shear strength for stability analysis, SlI(,,,,,h) is given by Eq. (1).
PA
PR
(2.1 ) (2.2)
where pA = a correction factor for strength anisotropy; and pR = a correction factor for strain rate effect.
Figure 2. Correlations among various strengths and S,,fobtained froin RC method
289
P
Project
Location
Behavior
SUwv,
SUnv,p
qJ2
/
Recompression method 1.23
SHANSEP
cc
6 CONCLUSIONS Among the case studies with various shear strengths, recompression (RC) method always gives proper shear strength as demonstrated in Table 2. Since the technique for performing DST is much easier, takes much less time and more cost-effective than triaxial test, therefore, RC method with DST is most recommendable for practical purposes. REFERENCES Figure 3. Change in shear strength during loading or unloading
Bjermm, L. 1972. Embankment on soft clay. ASCE Special conference, Performance of Earth and Earth-structures. Vol. 2. Lafayette: 1-54. Hanzawa, H. 1983. Three case studies for short term stability of soft clay deposits. Soils atid Foiiiidantions 23 (2): 140-
Based on the investigation of strain rate effect, Hanazawa (1989) proposed to use a constant pR value (= 0.85) irrespective of I, of the clay when displacement rate in DST is 0.25mm/min. On the other hand, a was evaluated with strength increment ratio in normally consolidated state, Sun/dYand with shear strength reduction ratio, SII(SLIIl, as schematically shown in Fig. 3. The values of S u n / d v and S,,(S,,,,were determined by K,,TCT for three projects (1976- 1981) and DST for other three projects (1 989- 1996). The value of p was assumed to be 1 .O in all the cases.
154.
5 RESULTS OF STABILITY ANALYSIS
Stability analyses were carried out with the modified Fellenious method using peak shear strength, SUll,,, with consideration on the change of shear strength during construction. This means that S,,, was corrected for a. The values of FS,,, obtained from various Su(mob) including corrected Su,lv,(Bjerrum 1972) and SHANSEP method (Ladd and Foott 1974) are summarized in Table 2 together with the behavior of each structure. FS,,,, values close to 1.0 for failed structures demonstrate the highest validity of the Su(,,,,,h) determined from RC method when compared with S,~(,,,,h) from any other methods. In addition, it should also be pointed out that failure planes from the analysis showed good agreement with the actual ones as reported in the referred papers.
290
Hanzawa, H. 1989. Evaluation of design parameters for soft clays as related to geological stress history. Soils atid Foinidations 29 (2):99-1 1 1. Hanzawa, H. and Tanaka, H. 1992. Normalized undrained shear strength of clay in the normally consolidated state and in the field. Soils aiidFotcr?dafions32 (1 ): 132-148. Hanzawa, H., Kishida, T., Fukasawa, T. and Asada, H. 1994. A case study of the application of direct shear and cone penetration tests to soil investigation, design and quality control for peaty soils. Soils and Fnimdatiori 34 (4): 13-22. Hanzawa, H. 1998, Application of cone penetration test for construction work (case studies in Southeast Asia). Proceediiigs qf "Workshop OH the evalimfion of growid with coiie penetratioki test: 67-77 (in Japanese). Ladd, C. C. and Foott, R. 1974. New design procedure for stability of soft clay. A Y E 100 (GT7): 763-786. Mikasa, M. 1960. Direct shear device newly developed. Proceedings of 151h./.YE awiical conference: 45-48 (in Japanese). Subagio, H. 1991. Evaluation of engineering properties of Banjarmasin clay, Indonesia. Pi*oceeditgqf Geo-Co'orrst '91. Yokohama: 93-98.
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida(eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
A field test on a new chemical grouting method to improve the liquefaction resistance of sandy layers beneath the existing structures K. Hayashi, R.Yoshikawa & N.Hayashi Penta-Ocean Construction Company Limited, Tochigi,Japan
K. Zen Department of Civil Engineering, University of Kyushu, Fukuoka, Japan
H-Yamazaki Port and Harbor Research Institute, Ministry of Transport, Yokosuka,Japan
ABSTRACT: Authors have developed a new chemical grouting method to increase the liquefaction resistance of sandy layers beneath the existing structures. The results of the experiments carried out to date indicate that this approach could prove extremely effective. On the basis of the laboratory tests, a field test was performed on the beach of Niigata city to investigate the permeability and the strength increment by improvement. The diameter of each improved area became about 4m long by %hours injection. This paper describes the results of the grouting tests at the site, together with the dynamic characteristics of the improved soils sampled after the grouting. The method was found to be very useful to improve the liquefaction resistance beneath the existing structures.
1 INTRODUCTION
After the 1995 Hyogoken Nanbu (Great Hanshin ) earthquake, design criteria for earthquake-proof structures were revised to improve and increase the resistance. The criteria are to be applied to the existing structures as well as the newly constructed ones. The problem is how to improve the foundation layers susceptible to liquefaction beneath the existing structures without any demolition of them. One possible solution to this problem would be chemical grouting directly beneath a structure, as shown in Figure 1. The chemical grouting method has been commonly used for the temporary construction, in which relatively high strength of improved soils is demanded, consequently leading to higher costs. To prevent liquefaction, however, not so high strength is required but the durability of the improved strength is requested. Also the cost reduction from that of the conventional grouting method is needed because the grouting area in practical use is veIy large. Since 1994, authors have developed a new chemical grouting method to increase the liquefaction resistance of sandy layers beneath the existing structures. In 1996, Iaboratory tests involving the injection of chemical grout into a large soil stratum resulted in the formation of cylindrical structures of improved ground with a diameter of 2.6 meters. This confirmed that the method would be highly effective as a countermeasure against liquefaction. On the basis of these results, field tests involving chemical grouting were
291
recently carried out in Niigata City. The tests clarified the application and penetration characteristics of the new chemicals. Excavations carried out after the injection process confirmed that solidification had occurred. This paper contains an outline of the field tests, a report on solidification characteristics. Through a series of tests, it is found that the new chemical grouting method is highly effective to increase the liquefaction resistance of sandy layers. Table 1 shows the differences between the conventional and new method. 2 DESCRIPTION OF TESTS 2.1 Outline of Tests The tests were carried out from October 1996 in Niigata. The aim of this tests was to verify the effects of an injection of the new grout, and to assess the environmental irnpact. Test items are as follows; (1)Injection volume of acid silica sol(amount: 9.6m3) (2) Low-speed grouting of colloidal silica (amount: 9.6m3) (3) High-speed grouting of colloidal silica (amount: 28.8m3) (4)Soundings to confirm effects (5)Excavation survey to check penetration area
A plan of the test area is shown in Figure 2, and a crosssection of the site in Figure 3. The aim of the tests was to gather data on the following two aspects of thc application.
Conventioal Method Cement or Water Glass
Grout Application of Grouting Penetration Area u
I
~
Durable
New Method Colloidal Silica
Localized
Large
about Diameter of 100cm
about Diameter of 400cm
about~200- 1,000Wa ~
about 50- l0OkPa ~ ~ ~
Temporary
Permanent
spersed with clay layers containing some silted sand at depths of 6 .- 11 meters. Apart from these layers, the injection area consisted mainly of fine sand with a fine particle content ratio (Fc) of 5% or lower. The physical characteristics of these soil layers are shown in Table 2. The relative densities were estimated on the basis of N values obtained with the standard penetration tests. The coefficient of permeability was determined by reproducing the samples with in-site relative density in d the laboratory and then conducting permeability test. The results were one order of magnitude smaller than those estimated from mean particle sizes of sand.
(1) Wide-area injection from a single pouring point (diameter of improved area: 4m) (2)High-speed implementation using rapid injection (20 literslminute) As shown in Figure 2, the aim of the test was to create a spherical improved soil with a volume of 33.5 m-?through injection from a single pouring point. The chemical grout was injected through tubes in three locations to create five improved soil. The total volume of improved soil by the chemical grouting was I67m3. 2.2 Ground Conditions Figure 3 shows the fine particle content ratio (Fc) and mean particle sizes (D5J in the area around the pouring point. The ground in the test area consisted of sandy soil inter-
2.3 Test Cases Injection was carried out under three sets of conditions using three injection bores. The results, including the quality of improved soil, were then compared. In the field test, the weights of grout 52.8 tons were injected from 5 pouring points. The volume of improved sand was 32m3 at each pouring point. The quantity of chemical grout and the volume of stabilized ground were the same in Case 1 and Case 2, while in Case 3, three sets of injections were carried out in a vertical direction. The test results for each case are shown in Table 3. Each item is explained below. (1) Types of Chemical Grout The chemical grout used for this tests were colloidal silica and acid silica sol, which were called the permanent grout. The permeability and durability of them had been ascertained through laboratory and field tests. In 1997 at Kagoshima, the durability of stabilized sand by acid silica sol was estimated by core boring samples, which was injected 16 years ago. The unconfined compressive strength of these core samples indicated almost the same strength as of 16 years ago. Both are single-solution chemicals and the viscosity of these grouts is about 2 Mpa s(=cps) at 20" C. The composition of the chemical grout was approximately the same as for those tests. The target strength of improved soil was 80kPa (after curing for seven days), which is considered to be enough for a sufficient strength , countermeasure against liquefaction. (2) Injection Speed Chemical grouting is normally injected at speeds of 8 12 liters per minute. It is known that a lower injection speed generally produces an even spherical shape. However, the injection of 9.6 rn3 of chemical grout at thc rate used for conventional methods (10 litedminute) would take 16 hours. If the injection period is too long, it becomes extremely difficult to control the gel time of the grout, and the gelation of grout will occur during the injection. Moreover,
-
Fig 1 Chemical grouting to prevent liquefaction beneath the existing structure
-
Fig 2 Plan of test area 292
, Table 2 Phvsical urouerties of soil
Depth
~ e n s i t yof soil Particle p S(s/cm3)
Maximam Density p dma(dcm3
-3- -4
2.674
1.77 1
1.366
9 . 4 0 ~1 0 4
-7- -8
2.736
1.741
1.313
4.80x 1o
1.758
1.328
GL(m)
--
Case
Grout
-1 ~
1
Asid Silica Sol
Injection Volume
'
4
Minimum Density d ~ n a ( ~ ~ ~ ~ )
Improved hjection Volume Rate
Objective S*n&(qu)
m3
P a
m3
l/min
9.6
80
33.5
20
_-.
2
Colloidal Silica
9.6
80
33.5
10
3
Colloidal Silica
28.8
80
100.5
20
because the volume of improved ground is large, the flow rate at the penetration boundary is likely to be slow. To allow for factors such as these, the injection rate for Case l and Case 3 was set at 20 liters per minute. The time required for the injection of the chemical grout in each location was, therefore, 9.6 ml/ 20 liters = 480 minutes = 8 hours. The injection period can thus be completed in about half the time required for conventional methods. (3) Injection Pressure Because a uniform injection speed was used in the tests, the injection pressure was varied from time to time. Injection pressure was measured with the passage of time.
Injection Pressure
-~
Preliminary Washing of Ground Wate none
Precedent to Grouting Speed 50 100% of Injection Volume
-
______________
(4) Preliminary Washing In Case 2 and Case 3, in which colloidal silica was used, solidification time was substantially influenced by the salt content of the pore water in the ground. For this reason, specific amounts of piped water were injected into the ground prior to the injection of the grout to remove the salt from the ground. The amount of water injected during the preliminary washing process was varied in each case, and the effects of this variation was also studied.
2.4 Application Procedures The equipment used at each stage of the tests was different. Injections were carried out according to the flowchart shown in Figure 4.
293
2.5 Equipment (1) Injection Plant The injection plant was the same as the system normally used for water glass injection procedures. At 5.5m x 15m, the plant was relatively small for use in a soil improvement method. The same plant was used for colloidal silica and acid silica sol. As the gel-time for both substances is long (8 hours), a stirring tank with a capacity of approximately 150 liters was used to provide for continuous injection using a one-shot batching method. To make up for the dilution phenomena at the penetration boundary, the concentration of the grout was progressively adjusted. (2) Injection Pump Because of the low pH (1.5 -2.0) of acid silica sol, a special acid-resistant pump was used. The maximum capacity of the pump was around 40 liters per minute, which was double the maximum injection rate. (3) Injection Plant Injection was carried out using the double-packer method via PVC injection hoses with an inside diameter of 5cm. Since the tests were implemented with an injection pressure of 5OOkPa or less and at relatively high speeds, a strainer-type injection plant with a large ground contact area was used. A cross-sectional diagram of the injection plant is shown in Figure 5. With conventional injection methods, the ground contact length of the strainer is around 1Ocm. As is apparent from Figure 5, however, the ground contact length in this case was 4 2 . 5 ~ ~ 1 .
Ca~el-3
1 Chemical Grouting
I+
Casel-3
3 TEST RESULTS 3.1 Injection Situation Injection speed and pressure were measured during the washing process and the injection process. Figure 6 shows the relationship between injection speed and injection pressure, using Case 3 as an example. As shown in the graph, it was possible to inject consistently at the required rate of 20 literdmin. During the washing process using municipal piped water, the injection pressure remained constant at around 200kPa from the outset. When the chemical grout was injected, however, the injection pressure rose gradually and reached 260kPa after eight hours. A similar trend was observed in the other cases: Because the viscosity of the chemical grout was approximately double that of water, the injection pressure increased in proportion to the permeation distance. The highest injection pressure was 450kPa (Case 3, GL-7.5m). Ground level was measured from time to time during the injection process and no rise in surface level due to injection pressure was detected. Injection pressure was determined from the original pressure inside the pump. For this reason, the figures include pressure loss within the injection tubes. 3.2 Results of Post-Injection Surveys (1) Unconfined Compressive Strength Distribution in Improved Area One month after the completion of the injection process, ground strength measurements were taken using a cone bearing test. In addition, samples were taken using a triple tube sampler. Strength distribution in the samples taken from the improved area are shown in Figure 7. The average unconfined compressive strength (4,) shown in the graph refers to the samples taken. The range over which q,, exceeds 25kPa is assumed to represent the permeation range of the chemical grout. The average of y, was calculated within this range. In Case 1 , the chemical grout used was acid silica sol. Throughout the permeation area, q, was in excess of 200kPa, which is a high level of strength for a situation in which the aim is to prevent liquefaction. In Cases 2 and 3, where the chemical grout used was colloidal silica, strength was lower overall when compared with the results of the mixture tests carried out in the laboratory. With an average q, of around SOkPa, the improved strength was low in Case 2 and in the middle range of Case 3. However, the average q,, in the upper and lower ranges of Case 3 was above the target level
of 80kPa. (2)Liquefaction Resistance of Stabilized Sand The improved area created by filling the pores with silica grout were subjected to cyclic triaxial tests.
Fig 4 Frow diagram o f test
294
Fig 5 Strainer-type grouting plant
Fig 6 Grouting speed and grouting pressure (Case3,GL-3.5)
Fig 7 Distribution diagram of unconfined compressive strength
295
Fig 8 Relationship Between Re20 and Repetitions
In contrast to the unimproved ground, there was no sudden increase of axial strain to rising pore water pressure ratios (the liquefaction phenomenon). The liquefaction is defined to occurred when the double amplitude of axial strain reached 5% under cyclic loading in subsequent cyclic traixal tests. Some of the samples obtained using the triple fined compression tests and cyclic traxial tests. To decrease the damage of sampling, all samples were tube sampler ( $ 100mm) were subjected to unconshaved to p 50mm. As shown in Figure 2, the samples tested were those taken at Point R in Case 3. The results of cyclic triaxial tests on undrained soil at various depths are shown in Figure 8. The results of cyclic testing of untreated soil samples from near the base point are also shown for reference purposes. The cyclic traixial tests yielded a B value of 95% or higher. The effective consolidation pressure ( o C ’ ) was determined by calculating the effective overburden pressure CT and then applying the formula CT c’ =(1+2*Ko)/3* o , assuming that &=OS. As is apparent from Figure 8, the results for the improved ground show considerable variation, as was the case with the unconfined compressive test results. However, liquefaction resistance in stabilized soil in which the double amplitude of axial strain (DA) reached 5 % was significantly enhanced when compared with the unimproved soil. This shows that the chemical grouting can substantially reduce liquefaction in the location where the process is applied.
Fig 9 Relationship Between Re20 and Unconfined Compressive Strength
Figure 9 shows the relationship between the unconfined compressive strength (q,) and the liquefaction resistance (Re 20) of the stabilized sand by colloidal silica, which is the force ratio at which liquefaction is reached after 20 load cycles. The graph also shows the relationship between typical q,, and Re 20 values for cementation soil. There is moderate variation in the case of stabilized sand by colloidal silica. In the case of cementation soil, however, Re 20 tends to increase as q,,rises. Under the conditions for these tests, the liquefaction resistance (Re 20) expected for a given unconfined compressive strength (q,,) is twice as high as the cementation soil. This shows that colloidal silica provides a greater strengthening effect. (3) Confirmation of Shape of Solidification through Excavation Excavations were carried out for Case 2 and Case 3 at 50 days after injection to measure the extent and form of the improved ground. Photo 1 shows an excavation in progress at GL-2.5m. The side of the improved area after excavation is shown in Photo 2. No consolidation was observed to occur above the groundwater level, since solvent-type chemical grout were used, and it was concluded on this basis that consolidation would not occur above this level. In both cases,
296
the solidified ground was in the form of a cylinder with a diameter of 4.5 5.0m around the pouring point.
-
4. CONCLUSIONS The field tests resulted in the following conclusions; (1)By using solvent-type chemical grout, it is possible to create a spherical improved soil with a diameter of at least 4m in sandy ground. (2)Where the permeation distance is around 2m, an injection rate of 20 litedminute is possible. While the shape of the improved soil was not observed to vary according to the rate of injection, the average strength of the stabilized sand was greater at 20 liters/min than at 10 liters/ min. (3)Liquefaction resistance (Re20) is greater in Stabilized soil than in normal soil, and it's value was around 0.5 -1.o. (4)At the same unconfined compressive strength (q,,),-iiquefaction strength (Re20) of stabilized sand by colloidal silica would be twice as high as in improved soil using a cement, and that shows the chemical grouting by colloidal silica would be more suitable for a liquefaction prevention method.
REFERENCES
Hayashi,K.,Miyoshi,T.,Yoneya,H.,Zen,K.,& Yamazaki,H., Y 1996. Fundamental tests on stabilized sand using acid silica
sol, Proc. of IS-TOKY0'96 : 695-700. Yonekura,R. & Miwa,M., Y 1993. Fundamental Properties of Sodium Silicate Based Grout, Eleventh Southeast Geotechnical Conference: 4-8. Zen$., Y 1994. Remedical Measures for Reclaimed Land by Premixing Method, Tsuti to Kiso, 433(2):34.
297
This Page Intentionally Left Blank
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Case history of the reclamation at Island City in Fukuoka K. Henmi Port and Harbor Bureau, Fukuoku City Japan
M. Katagiri & M.Terashi Nikken Sekkei Nakuse Geotechnical Institute, Kawasaki, Japan
K. Fukuda Coastal Development Institute of Technology,Tokyo,Japan
ABSTRACT: To predict the settlement behavior of reclaimed land by dredged materials, the soil investigation, monitoring of settlement and numerical simulation were performed as the reclamation progressed. Accuracy of the prediction has been improved gradually with the progress of reclamation. The practical use of observation method for these reclamation works was confirmed. 1 INTRODUCTION The mountain area covers more than 70 per cent of all the territory of Japan and the plain area is limited. Due to these geographical features of Japan, big cities and major ports have located and developed along the flat coastline. Cities and ports are therefore very closely connected each other. Fukuoka City having Hakata Port has been playing an important terminal base to the Asian Continent for more than 2,000 years and has brought about such a cultural climate that has enhanced the interchange of persons, commodities and information. Hakata Port has been growing as an important international trading port and now it is directly connecting to 59 ports in the world. Similarly, Fukuoka City has been developed steadily as a commercial city, and now has a population of 1,320 thousand. For the further development of Fukuoka City, a sea reclamation with 400 hectare has been planned in the eastern part of Hakata Bay as shown in Figure 1. The reclaimed land will be used as residential areas and industrial space. The objective of the former is the creation of comfortable urban life space to solve the problem of population increase. That of the latter is to provide a terminal base for the new tertiary industries which enhance the revitalization of industrial structure. This is the Island City Project. The materials for the reclamation will be as large as 28 million cubic meters; a part of which has already been and the rest will mostly be provided by the dredging of navigation channels and anchorage areas in the Hakata Bay. The dredging has been undertaken to improve harbor and channels so as to cope with larger-sized ships in recent years. Figure 2 shows an aerial photograph of the construction site taken from the east in Nov. 1998. A dredging work for the port is also seen in the center of this photograph.
Figure 1 Location of Mand City Project
Figure 2 Aerial photograph of project site
299
Figure 3 Flow of reclamation by dredged materials
Figure 4 Location of City No.1 area and measurement position of settlement o f seabed
When the existing seabed is a thick soft clay layer, the weight of reclaimed materials and extra-fill generates the consolidation settlement of seabed as well in the long term. If theJerm of construction is limited, it becomes necessary to accelerate the consolidation by vertical drainage. There are three major problems critical in this process. One is the capacity of pond. As the dredging is going on simultaneously, the capacity of pond has to be consistent with the planned total amount of the dredged materials. The others are the estimations of the amount of extra-fill and residual settlement, both of which influence the cost of reclamation. The key to solve these problems is the accurate prediction of the time dependent change of elevation of the dredged clay layer that increases initially and decreases later in the long term. As such a big reclamation project takes several years until completion, the construction program often changes with time and the prediction should be updated with the progress of the project. In this paper, the history of predictions and modifications of the time dependent settlement of reclaimed layer during reclamation and the effectiveness of observation method is described. The settlement of seabed underlying the reclaimed layer and the flatness of reclaimed land by pump-dredged materials are also mentioned.
3 RECLAMATION PLAN IN CITY NO.l AREA
Figure 5 Outline o f observation method
2 PROBLEMS ASSOCIATED WITH RECLAMATION BY DREDGED MATERIALS A sea-reclamation by use of dredged materials usually progresses with time as shown in Figure 3. When the dredged materials are poured into the pond surrounded by containment dikes, suspended soil particles settle loosely with the water content from 200 to 300 %. While the pouring continues, the surface elevation of dredged soil layer increases. The land thus created is in the unconsolidated condition and subsequently consolidates largely due to its own weight in long term. To create a reliable foundation ground at a specified elevation for the structures to be built, the placement of extra-fill over the dredged clay layer is necessary. The extra-fill also causes additional large consolidation settlement.
300
The Island City project site consists of six ponds as shown in Figure 4. Described in this paper is the City No.1 area that is about 1,100 m x 600 m in plan, and was scheduled to be reclaimed by two types of dredged materials up to DL +6.5m. A wharf and a container handling yard behind it should be brought into operation until April 2004. To keep the construction schedule, the accurate prediction of time-elevation relation of the reclaimed land became necessary. The observation method as shown in Figure 5 was employed to improve the accuracy of prediction. The observation method is a procedure of repeating the prediction in stages by new parameters modified through the back analyses of the preceding measurements of real behavior. Figure 6 shows the soil profile before the reclamation at the A-A section in Figure 4. An average depth of the existing seabed is DL-3.5 m, and the seabed is alluvial clay layer of 8 m thick. The original reclamation program (plan) and the actual record of reclamation (practice) were compared in Figure 7. The main reclamation term was 18 months since August 1997. Before the main reclamation, grab-dredged materials of about 800,000 m3 in plan (1,000,000 m3 in practice) had been poured for 4 years before August 1997. The main reclamation process was divided into two stages. The
pump-dredged materials of about 1,800,000 m3 were first poured for 6 months in plan (5 months in practice). After a little intermission, the grab-dredged materials of about 1,800,000 m3 in plan (1,600,000 m3 in practice) were followed for 7 months in plan (6 months in practice). Total volume of poured materials in practice was almost the same as that in plan, although the rate and volume of reclamation in each stage was changed.
4 HISTORY OF PREDICTION, OBSERVATION AND MODIFICATION OF PARAMETERS 4.1 Consolidationanalysis used For the consolidation analysis, CONAN proposed by Imai (1995) is used. The detailed procedure of numerical analysis is described in the companion paper by Katagiri et al. (2000). By the method, it is easy to calculate the consolidation of the layer increasing its thickness with time, such as a reclamation process. The consolidation parameters for the method are relationships of specific volume f and coefficient of consolidation c, with consolidation pressure p .
4.2 Consolidationparameters As the pump-dredged materials are hydraulically transported through the pipe and poured into the pond with a lot of seawater, their water contents in the sediment are very high. The water contents of grab-dredged materials transported by another verge to the pond, on the other hand, are not so high. Because the consolidation parameters of clay-seawater mixture in the ordinary stress level depend largely on the initial water content (Katagiri and Imai, 1994), the parameters of reclaimed layer created by pump-dredged materials may be different from those by grab-dredged ones. Therefore, it is necessary to determine the consolidation parameters of both materials. Figure 8 shows the consolidation parameters determined on a sample taken from the borrow area. The physical properties of the sample were as follows; p, = 2.668 g/cm3, wL = 71 %, Ip = 45. To determine the consolidation parameters over a wide stress range, the multi-sedimentation test (MST) proposed by Yamauchi et al. (1990) and ordinary consolidation tests (OCT) were carried out on the specimens simulating the pump- and grab-dredged materials. Based on the experience, the pumpdredged material was prepared as the clay-seawater mixture with the water content of 2,000 %, and the grab-dredged material was prepared with water content of 200 %. As shown in the figure, the compressibility is larger and the permeability is smaller for pump-dredged materials. Hereafter, the parameters are referred to “initial relations”.
301
Figure 9 Prediction and modification histories of time-elevation relation during reclamation
4.3 Prediction and modification histories of consolidation behavior of reclaimed land
Modifications of the consolidation parameters in the observation procedure were done in such a way that the inclination of the log f - log P is changed with a fixed point atp = l,ooo kPa, and that the log C, - log p relation is shifted parallel to the initial ones on the process of finding best-fit solution with the measurement.
Figure 9 shows the history of predictions of the elevation in a couple of stages together with the measured data. At the planning stage (Figure 9(a)), the prediction was carried out by the Ip-method proposed by Shinsha et al. (1990). It is the simple method in which only two soil parameters, average water content and plasticity index of the material at the.borrow site, are necessary. Although the practical use of the method has been confirmed through back analyses of case histories, it cannot reflect the difference in the initial water content of dredged materials. It will provide a rough estimation of the elevation changing with time but will neither provide any information on the distribution of water content or that of the excess water pressure. The first and preliminary prediction by CONAN was performed before the reclamation (Figure 9(b)) using the consolidation parameters of pump- and grab-dredged materials already shown in Figure 8. The reclamation history used for the calculation is that for the plan in Figure 7. The prediction indicates that the elevation at the end of first stage by pumpdredged materials far exceeds the planned maximum allowable elevation of DL +6.5 m. As the consolidation parameters are obtained based on only one sample, and the specimens were prepared only by changing their initial water content, it is uncertain that those parameters are the representative ones. Figure 9(c) shows the results of the first modification before the reclamation. The same logf- logp relations in both the pump- and grab-dredged materials were applied (Figure 8(a)). Based on the experience at new =takyushu airport (Sate et al., ~ O O O ) , two set of the log c, - l o g p relations were assumed for the pumpdredged materials; one is twice and the other is three times as large as that of the original c,values shown in Figure 8(b). The log c, - logp relation for the grab dredged materials was assumed five times as large as that of the original ones. This modi-
302
fication was based on the back analysis of the preceding area in the Island City reclaimed by grabdredged material. Modified predictions fall between two previous predictions as shown in Figure 9(c). Higher the assumed c,, the lower is the predicted elevation.
Two peak elevations predicted at the end of first stage reclamation are lower than DL +6.5 m. Although the peak elevations at the end of second stage reclamation were predicted still higher than DLt6.5 m, the reclamation has been started. Figure 9(d) shows the measured elevation of the reclaimed land monitored during the first stage reclamation. As already shown in Figure 7, the first stage reclamation was conducted at a rate faster than the original plan. Together with the back analysis of the first stage reclamation, the second modification of the prediction was carried out at the end of first stage. The monitored elevation change during the first stage reclamation by pump-dredged materials is lower than those calculated in Figure 9(c). In the improved prediction (second modification), the recorded history of the reclamation process was used for the first stage and that of the plan was used for the second stage reclamation process. In order to fit the prediction to the measured data in the first stage, the compressibility of the pump-dredged materials was reduced and the permeability was increased. Two sets of the consolidation parameters were assumed in the 2nd modification. One was a set of reduced compressibility, 80 % of the initial and the log c, - l o g p twice as large as the initialial c,-values. The other set was 90 % compressibility and three times large c,-values. The same consolidation parameters as those for the first modification were used for the grab-dredged materials. By these modifications, the peak elevation at the end of the second stage is predicted lower than the DL+6.5m as shown in Figure 9(d) and the reclamation continued. All the measured elevations during first and the second stages were plotted in the Figure 9(e). The monitored elevation and time relation falls within the range of the second modification, although the rate of the reclamation was increased again from the original plan. The latest back analysis was carried out after the second stage using all the measured data of the elevation and the actual record of reclamation history shown in Figure 7 by solid triangles. Several sets of the consolidation parameters were selected, and after trails the most suitable result expressed as a bold line with solid circles in Figure 9(e) was determined. The estimated parameters were as follows. For the pumpdredged materials the compressibility parameter was 80 % of the initial slope, the log c, - l o g p relation was twice of the initial c,-values. For the grabdredged materials the compressibility was the initial one, and the log c, - logp relation was twice as large as the initial c,-values. 4.4 Verification by water content distribution
Figure 9 focused on the time and elevation change. The prediction as well as the back analysis must fit not only with the elevation of the dredged
303
Figure 12 shows the distribution of the surface elevation measured after the first stage reclamation. Although pump-dredged materials like a liquid are poured into a huge reclaimed area, the soil particles were not homogeneously piled up. High elevation region is the location around the outlet of the transporting pipe, and is composed of the material with high sand fraction. Low elevation region, on the other hand, is far away from the outlet and is composed of the material with high clay fraction. This uneven surface elevation together with the inhomogeneity already found in the water content distribution would become major source of differential settlement in the later stage of construction.
clay layer but also with the water content distribution in the layer. Because the latter becomes the initial condition for the prediction of further consolidation settlement under the extra-fill. Actually the verification of the back analysis based on the water content distribution were conducted twice during the reclamation. Figure 10 compares the calculated and measured water content distribution as of February 1999, two months after the reclamation. The measured and predicted water content distributions share the same tendency of decreasing with depth. The predicted relation drawn as a solid curve in this figure changes at the position between layers by grab-dredged and pump-dredged materials, because the different compressibility are used. The prediction is in the center of the scattering range of measured data. Considering the inhomogeneity of the dredged layer caused by the grain size sorting during sedimentation, the combination of consolidation parameters obtained by the back analysis is confirmed to be acceptable.
5 CONCLUDING REMARKS
4.5 Settlement behavior of seabed underlying reclaimed layer Figure 11 shows the settlement of the seabed measured by means of hydraulic settlement transducers located at the positions shown in Figure 4. The two curves in the same figure are those predicted by Terzaghi’s equation with c, = 40 cm2/day under two different assumptions on the boundary conditions (bottom drainage, top and bottom drainage). Measurements were clearly responding to the first and second stage reclamation. In the first stage, measured settlement data are on the prediction by double drainage. Practically no settlement is recorded in its subsequent rest term. And the settlement of seabed during the second stage reclamation takes average of two predictions by single and double drainage. The asterisk in the figure is the settlement of the seabed directly measured by the borings after the reclamation. Thus the complicated timesettlement relation measured are thought to be a real behavior. In the earlier discussion of the prediction of the elevation of dredged clay layer, single drainage solution was incorporated with the CONAN solution. From Figure 11, however, the measured settlements are only 30 cm larger than the single drainage solution. This means that the maximum possible error caused by the prediction of seabed settlement was as small as 3% (= 0.3/10) as the thickness of the reclaimed layer reached as large as 10 m.
To predict the settlement behavior of the reclaimed land by dredged materials, the soil investigation, monitoring and repetitive numerical simulations by CONAN were performed as the reclamation progressed. Although the initial prediction using the consolidation parameters obtained from a material sampled from the borrow area is largely different from the measured value, the repetitive numerical simulations give practically useful prediction not only of the change of elevation but of the water content distribution if the appropriate parameters are determined. The appropriate parameters were determined through the observation of the real behavior and the back analyses. The prediction was improved gradually with the progress of reclamation. The reclamation was completed successfully, the site is about to be improved with plastic board drains. The water content distribution and the degree of consolidation calculated by CONAN were used as important information for the initial condition of the further settlement analysis under the extra-fill loading. ACKNOWLEDGEMENT The authors thank Prof. H. Ochiai of Kyushu University and Prof. G. Imai of Yokohama National University for their valuable advice to this project.
4.6 Surface Flatness of reclaimed land
Throughout the previous discussions, the huge City No. 1 area was considered to be reclaimed by uniform materials and to consolidate one dimensionally.
304
REFERENCES Imai, G., 1995. Analytical examinations of the foundations to formulate consolidation phenomena with inherent time-dependence. Proc. of IS-Hiroshima ‘95, 2: 89 1-935. Katagiri, M. & Imai, G., 1994. A new in-laboratory method to make homogeneous clayey samples and their mechanical properties, Soils and Foundations, 34(2): 87-93.
Katagiri, M., Terashi, M., Henmi, K. & Fukuda, K., 2000. Change of consolidation characteristics of clay due to dredging and reclamation, Proc. of ISYokohama 2000 (submitted). Sato, K., Ishinuki, K., Katagiri, M., Terashi, M. & Kitazawa, S., 2000. Reclamation control of pump-dredged clay by CONAN, Proc. of IS-Yokohama 2000 (submitted). Shinsha, H., Chiba, T., Suzuki, Y. & Yamaguchi, R., 1990. Development of the volume change prediction system for the pump-dredged clayey soils (in Japanese), Annual report of the engineering research institute, Penta-Ocean Construction, 19: 17-28. Yamauchi, H., Imai, G. & Yano, K., 1990. Effect of the coefficient of consolidation on the sedimentationconsolidation analysis for a very soft clayey soil (in Japanese), Proc. of 2ShAnnual meeting of JSSMFE: 359-362.
305
This Page Intentionally Left Blank
Coastal GeotechnicalEngineering in Practice, Nakase L? Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Change of consolidation characteristics of clay from dredging to reclamation M. Katagiri & M.Terashi Nikken Sekkei Nakuse Geotechnical Institute, Kawasaki, Japan
K. Henmi Port and Harbour Bureau, Fukuoku City,Japan
K. Fukuda Coustal Development Institute of Technologj Tokyo,Japan
ABSTRACT: In dredging and reclamation works, the seabed clays are dredged, transported and poured into a pond. In these processes, the clays are disturbed, their water contents increase largely, the grain size sorting is generated in the pond, and the consolidation parameters change. It is therefore difficult but important to determine the appropriate consolidation parameters. In this paper, the best-fit parameters back-analyzed are compared with those parameters of the specimens from the borrow site and reclaimed land. Changes of parameters are found to be influenced by the difference in the dredging procedure and original soil properties. 1 INTRODUCTION Dredging of clays are usually conducted by pumpor grab dredger. Dredged clays are then transported to a disposal pond for sea-reclamation either by a pipeline or by a barge. The structure of the clay that has originally been developed at the seabed is disturbed in the dredging and transporting processes. Figure 1 illustrates the flow from dredging to reclamation. The degree of disturbance depends on the ways of dredging and transporting. In the pump-dredged clays, the original soil structure is completely destroyed and the water content measured in the pipeline often exceeds 1,000 %. When poured into a pond, soil particles are initially suspending in the water and gradually settles to the sea bottom. Soil particles then create a sediment by re-constituting a new soil structure. The characteristics of reconstituted structure may be different from those of the original structure in the seabed at borrow area. The grab-dredged and barge-transported clays are also anticipated to experience disturbance to their structure but with much smaller degree. In addition to the disturbance, grain size sorting in the settlement process may influence the characteristics of pump-dredged clays. When pump-dredged clays are poured into the water, coarser particles are piled up near the outlet of the pipeline, whereas the finer ones are suspending in the water and transported far from the outlet until they settle down. Although the outlet is moved periodically during the reclamation, the deposit thus formed is not uniform. In the design of sea-reclamation, large volume change of dredged clays is one of the critical issues. 307
However, in the planning stage that is before the reclamation, the determination of the consolidation parameters of the reclaimed clays is difficult because of the reasons mentioned so far. In the paper, Fukuoka Island City project is taken as a case record for the examination of the change of consolidation characteristics from dredging to reclamation. This is the first step of accumulating data for establishing a procedure to determine the appropriate consolidation parameters. As it is impossible to determine the spatial distribution of soil parameters beforehand, the reclamation process is assumed as one-dimensional consolidation by a homogeneous material. The best-fit consolidation parameters backanalyzed for the pump- and grab-dredged clay layers are compared with those of samples taken from the borrow area and the completed reclaimed land. 2 CONSOLIDATION THEORY AND ANALYSIS METHOD The process from settlement to self-weight consolidation of soil particles is explained in Figure 2 (Yamauchi et al., 1990). Settlement means the condition in which single particles or flocs fall down in the water. Sedimentation means the phenomenon that sinking particles or flocs settle at the top of the other, and self-weight consolidation is the process where the sediment consolidates due to its own weight. When the soil particles in a dilute ciay-seawater mixture settle in a cylinder the particles first become cohered and form flocs, which then fall in the form of zone settling as shown in Figure 2 (Imai, 1980).
A new uniform sediment with predetermined thickness is instantaneously piled on the top surface of the sediment already formed, and just after the piling of this sediment is considered as a consolidation layer. 2.1 Consolidation theory used
The general one-dimensional consolidation equation can be expressed as follows; Mass conservation: _av_ _ - -ae az at
(1)
Darcy’s law and balance of momentum neglecting acceleration: v
k
= -(-x-+
y,
1 l+e
ae
az
y’)
Constitutive equation of soil skeleton: f(a’,e,l.) = 0
(3)
Where, e, e: void ratio and rate of void ratio, 0’:effective stress, k: coefficient of permeability, v: exit water velocity relative to soil skeleton, y7: submerged unit weight of soil, yw: unit weight of water, z:reduced coordinate. To simplify the constitutive equation, Eq. (4)that does not take viscosity into account has been used in this paper. f(a’,e) = 0
(4)
The k is determined uniformly according to e as followings : k
Subsequently, the particles that have settled become a part of sediments (t = tl). From the viewpoint of the development of effective stress there must exist a boundary between the particles still settling and those that have already became sediment, and the boundary moves upward with time. This boundary is here called as the ‘depositional surface’. With the progress of sedimentation the top surface of the group of settling particles always sinks, while the depositional surface rises. After the depositional surface reaches to the top surface of sediment, all the sediment is in the self-weight consolidation (t = tz). To perform a numerical analysis simulating the accumulation of sediment, a model shown by a steplike line in Figure 3 (Yamauchi et al., 1990) is used.
=k(e)
(5)
A numerical calculation that satisfies Eqs. (l), (2), (4)and ( 5 ) can be carried out. The most general one-dimensional consolidation theory considering no creep effect was proposed by Mikasa (1963) and Gibson (1967). Both the theories are expressed by complicated partial differential equations of the second order obtained from combining Eqs. (l), (2) and (4). In this paper, however, the coupling method “CONA”’ proposed by lmai (1995) is used. In this method, Eqs. (l), (2), (4)and (5) expressed by a single differential equation are coupled whenever the occasion demands. 2.2 Boundary condition of self-weight consolidation To analyze the consolidation of new fresh sediment piled on the top of existing sediment, the determination of the boundary conditions of the fresh sediment becomes important. In this paper, the boundary condition at the top of the fresh sedimentation is fixed at 9.8 Pa as proposed by Yamauchi et al. (1990).
3 COMPARISON BETWEEN NUMERICAL AND MEASURED RESULTS
The reclamation of the City No.1 area at Island City project in Fukuoka, Japan is selected for the case study to investigate the change of consolidation parameters. The conditions of City No.1 area are; about 600 m x 1,100 m in plan, DL-3.5 m in depth of sea, and DL+6.5 m in planned elevation at the end of reclamation. The main reclamation took place in 18 months since August 1997. About 1,800,000 m3 of pump-dredged clays were poured first for 5 months. After 7 months of intermission, about 1,600,000 m3 of grab-dredged clays were brought into the pond for 6 months. Details of the project are described in the companion paper by Henmi et al. (2000).
3.1 Consolidation parameters used and result of back analysis The initial and best-fit consolidation parameters for the pump- and grab-dredged clays are shown in Figure 4. The initial relations for the pump- and grabdredged clays were determined by the multisedimentation test proposed by Yamauchi et al. (1990) at smaller stress level, and by the ordinary oedometer test at higher stress level. The parameters in the intermediate stress level were obtained by interpolation to cover the wide stress range. Only one sample was available for the determination of initial parameters, however, the sample was not too far from the representative sample of the borrow area (see 3.3). Based on the experience, the specimens were prepared from the sample as the clay-seawater mixtures having water contents of 2,000 and 200 % for pump- and grab-dredged clays, respectively. Figure 5 shows the comparison between best-fit analyzed and measured results. In Figure 5(a), solid squares represent the maximum and minimum elevations out of about 200 measurements by means of sonic prospecting. Open circles show the range of elevations measured by leveling at three to six locations. Figure 5(b) shows the water contents distributions with elevation measured at three locations on February 1999, two months after the reclamation. Back analyses of the reclamation through the modification of the initial consolidation parameters were conducted to find the best-fit solution for the time-elevation relation and water content distribution (Henmi et al., 2000). For the compressibility, the inclination of logf- logp relation, SL (as shown in Figure 4(a)) was changed by fixing a point at p = 1,000 kPa on the initial relation. The permeability was changed by shifting the log c, - logp relation in parallel to the initial one. The best-fit solution thus obtained is confirmed to fit not only with the elevation of the reclaimed land (Figure 5(a)) but also with 309
the water content distribution in the dredged layer (Figure 5(b)). The parameters used in the best-fit solution (best-fit parameters) represent the overall consolidation characteristics of dredged clay layer that really is far from uniform. 3.2 Comparison with consolidation parameters of samples in reclaimed land Soon after the reclamation was completed, undisturbed soil samples were taken from the reclaimed layers and oedometer tests at ordinary stress level have been conducted to investigate the relation of best-fit parameters and actual variation of parameters in the reconstituted clays. Figure 6 shows the consolidation parameters of the pump-dredged clay layer. The best-fit and initial parameters are also shown in the same figures. The inclination of the best-fit log f - log p relation in Figure 6(a) corresponds to the average, and the bestfit log c, - l o g p relation in Figure 6(b) is located in the upper boundary of measured data in the normally consolidated condition. Figure 7 shows the consolidation parameters of the grab-dredged clay layer. The best-fit logf- logp relation falls in the middle of measured relations. The best-fit log c, - l o g p relation seems to be consistent with the measured relations. It is interesting to compare the consolidation parameters between pump-dredged clays and grabdredged ones. As the consolidation characteristics highly depend on the physical properties of clay, the relations between soil property and consolidation parameters are necessary. Figure 8 is a plasticity chart on which the properties of samples are plotted. Open squares correspond to the pump-dredged clays and solid rhombuses to the grab-dredged clays. The liquid limit in both the clays are between 60 and 100 96.Due to the grain size sorting, a part of the reclaimed layer is composed of extremely fine particles and contrary to this, another extreme becomes a sand heap. Those extremes are not reflected in the figure. Figure 9(a) shows the relationship between w, and SL, the inclination of log f - log p. At around 100 % of w, the SL-values of the pump- and grabdredged clays take almost the same magnitude. The SL of each clay decreases with decreasing w,. As w, decreases, the difference of SL between two clays becomes larger. At about 60 % of wL, the SL of the grab-dredged clay is a half of the pump-dredged one. The solid and broken lines in Figure 9(a) are the best-fit parameters of the grab- and pump-dredged clays respectively. The best-fit parameter of each clay is located within the range of variation and it takes the average of corresponding clay. Figure 9(b) shows the relation between c, and w,.
310
Figure 7 Consolidation characteristics of grabdredged clay layer
c,-value at 56 kPa is selected for comparison because the stress level is in the normally consolidated condition. The measured c,-values at 56 kPa are independent on w,,when it exceeds 65 % and are not influenced by the dredging method. Below 65 % of w,, the c,-value increases with decrease of w,. The best-fit and initial relations are also shown in the
same figure. The best-fit c,-value of grab-dredged clay expressed by a solid line is consistent with the measured data, while that for the pump-dredged clay layer by a broken line is higher than the measured ones. As far as the compressibility is concerned, the best-fit parameter is consistent with the actual variation of the reclaimed clays. There found an inconsistency in the coefficient of consolidation. One possible reason for the latter is that the one-dimensional behavior of overall reclaimed clay layer may not be the same as the integration of the behaviors of small parts because of two- or three-dimensional behavior caused by spatial inhomogeneity of reclaimed land.
3.3 Comparison with consolidation parameters of original clay samples In the design stage of sea-walls for the reclamation, a number of undisturbed samples had been taken from the seabed clay layer and tested by oedometer. The seabed clay at the sea-wall site is the same layer with those dredged for later reclamation. Figure 10 shows the plasticity chart for these samples. Physical properties of original clays themselves had wide variation. Solid circle in the figure corresponds to the clay sample tested to determine the initial parameters in the previous section 3.1 and it locates in the middle third of the actual wide variety of original clays. Figure 11 shows the relationship between consolidation parameters and wV In order to compare with c,-values in the normally consolidated condition, most of the c,-values are taken at 56 kPa of consolidation pressure. For the samples taken at the deep part, the values at 222 kPa in the normally consolidated condition are plotted by open triangles in Figure ll(b). The parameters, SL and c, in the original seabed clays had variation and those magnitudes depend on the wL. As the best-fit parameters are dependent on the dredging method, it is still difficult to determine the best parameters beforehand even if such information on the original parameters is at hand. It is confirmed at least, however, the best-fit parameters are within the variation range of the original parameters. The initial approximation of the best SL may be the average of original, and the best c, may be the lower boundary for the grab-dredged but higher for the pump-dredged clays. The discussion in this section applies to the parameters at ordinary to higher stress level- One should not forget that the parameters at extremely low stress level is also required for the analysis and they should be obtained by means of special tests such as the multi-sedimentation tests (Yamauchi et al., 1990) or by centrifuge test (Nishimura et al., 2000).
31 1
Figure 10 Plasticity c h a r t on samples from s e a b e d
CHANGE OF CoNSoLIDAT1oN PARAMETERS
Apart from the numerical simulation by CONAN, the actual change of parameters during dredging and reclamation processes is discussed in this chapter.
value decreases, the SL-value decreases in each case, as described earlier. The w, - SL relation of the clays from the original clay layer is located between the relations of pump- and grab-dredged clays. This means that the inclination of the logf- logp relation of the pump-dredged clay layer increases, and that of the grab-dredged clay layer decreases. Especially, for the clays with lower w,-values the change is remarkable. In the relationship between wL and c,-value in the normally consolidated state, on the other hand, the relation of the seabed samples is located above those of both the reclaimed layers. The c,-value of the seabed clays increases gradually with decrease of w,, while those of the reclaimed clays indicates a constant value in the range of w , between 65 and 100 %. 5 CONCLUDING REMARKS
To predict the consolidation settlement of reclamation by dredged clays, determination of appropriate consolidation parameters for the reclaimed layers is necessary. Back analysis by means of one-dimensional consolidation analysis, CONAN gave quite a reasonable simulation of the actual case record. However the best-fit parameters covering wide stress range are different from the parameters determined on a sample taken from the borrow area. The degree of difference is dependent on the dredging method. Compared with consolidation parameters of the samples taken from the reclaimed land, the best-fit l o g f - l o g p relations in both the grab- and pumpdredged clays are consistent with those measured for the reclaimed clays. The best-fit log c, - l o g p relations in both the cases are larger than those obtained from the samples in reclaimed land. Compared with the parameters of samples from the original clay layer, the best-fit parameters are within the range of the scattering measured data. Still imperfect but some insight was obtained for the determination of the parameters based on the data from original clay. The overall trend of the change in the consolidation parameters from dredging to reclamation was grasped based on a case record. The accumulation of similar case record will be necessary for further improvement in the determination of appropriate consolidation parameters.
Figure 12 Comparison with consolidation characteristics o f seabed and reclaimed land samples
REFERENCES
To examine the change Of paramek r s from borrow area to dredged layers, Figures and 11 are superposed into Figure l2 which shows the relationships between consolidation parameters and wP The SL-values of the clay with about 100 9% of wL is approximately 0.15 in all cases. As the w,312
Gibson, R.E., England, G.L. & Hussey, M.J.L., 1967. The theory of one dimensional consolidation of saturated clays; I Finite non-linear consolidation of thin homogeneous layers, Geotechnique, 17: 261-273. Henmi, K., Kataniri, M., Terashi, M. & Fukuda. K.. 2000. Case history of the reclamation at Island City in
Fukuoka, Proc. of IS-Yokohama 2000. (submitted). Imai, G., 1980. Settling behavior of clay suspension. Soils and Foundations, 24(2): 61-70. Imai, G., 1995. Analytical examinations of the foundations to formulate consolidation phenomena with inherent time-dependence.Proc. of IS-Hiroshima ‘95,2: 891-935. Mikasa, M., 1963. The consolidation of soft clay -A new consolidation theory and its application-, Kajima Shuppan-kai (in Japanese). Nishimura, M., Katagiri, M., Terashi, M. & Saitoh, K., 2000. A Determination method of consolidation parameters for clay sludge by centrifuge. Proc. of ISYokohama 2000. (submitted). Yamauchi, H., Imai, G. & Yono, K., 1990. Effect of the coefficient of consolidation on the sedimentation consolidation analysis for a very soft clayey soil (in Japanese), Proc. of 25IhAnnual meeting of JSSMFE, 359362.
313
This Page Intentionally Left Blank
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Limit analysis of sheet pile type retaining walls S. Kobayashi Kyoto Universio,Japan
ABSTRACT : Stability problem of a sheet pile - ground system is investigated in this papw. Both upper bound and lower bound analyses are done for this problem, although limit equilibrium met hod is usuallj used in practical ciigiiieering design. From the thcoretical point, limit equilibrium method is ambiguous in comparison with limit analysis. For better knowledge of the properties of solutions, the author shows results of both upper bound and lower bound analyses and adds some comments of these solutions. bound solution varies according to the strength ratio of reclaimed and supporting layers. However, a location assumed here may be a good approximation for many practical situations. I,et S 1x a11 angular velocity of a sheet pile, velocity of a sheet pile at each points are as
1 INTRODUCTION
Limit design becomes popular among civil engineers through recent developments of design procedures. In geotechnical engineering, it semis very natural to adopt this design concept, because almost all behaviours of soil masses are belielred to be plastic. However, the author feels that there are soriie misunderstanding of limit state designs in geotechnical problems. $’or example, limit equili brium method is comnionly used in design procedures, and sorrietimes it is thought to Ine same as limit analysis. It may be a certain ltind of approximate solutions. But there is no theoretical background. For complete understanding of limit design concept, limit analysis is inevitable. For detailed explanat ion of limit theorem, see refererices (Salenson, 1977, Shibata 67 Sekiguchi, 1995).
t 1 y =
-Sy,
(y= 0.
(1)
Considering compatibility of velocities along OH. one of kinematically admissible velocities of soil mass can be expressed as 21,
= O(tanci.s-y),
Plastic strain rates for area OAB are
Limit analpis is done for a stability problem of a self-standing vertical retaining wall. A sheet pile of which yielding moment is My installed vertically to a certain depth. A reclaimed land behind a sheet pile is filled with 4 = 0, c = CO rigid perfectly plastic material. A supporting layer beneath a reclaimed land is q5 = 0, c = c1 material. Friction between a sheet pile and a reclaimed land is assumed t o be negligible. Stabi1if.v analjwis based on upper I~onnd met lioti Assurned failure riiecha~iismis shown i n Fif. 1. A plastic hinge occurs at a foot of a sheet pile aiid its reclaimed soils are deformed with plastic shearing. A location of a plastic hinge for the best upper 2.1
315
where i:,
H
I
Figure 1: Assumed failure mechanism for upper bound analysis
Figure 2: Mohr’s stress circle and plastic strain rate circle for area OAB
Sheet pile
C J
8
I
I
H
~1
)L,
0
Figure 3: Assumed stress field for a lower bound analysis Along OB where a sheet pile and soil mass contact t o each other, it is negligible becausc friction there is assumed to be zero. A t a plast,ic hirige occurring in a sheet pile, it is calculated as
lil.’*,t,sp =
nf,@
b5)
On the other hand, since external plastic work rate is only done by self weight of soil mass, it is expressed as 316
0
lrlt”””
JH
( - p g ) d ( tan’
Q
Shear stress components
gTy=
ryzarc zero.
pyLl 5 ‘2cl (area a ) ,
. x - tan a . y)dyclx
y=xtancr
According tjo upper bound theorem, by equating internal plastic dissipation energy rate with external plastic work rate, the following equation can be derived,
pg( L
+ L 2 ) L 2c1
pg( H
+ L1 + I,’)
pgN
(area c ) ,
5 2cl (area d ) ,
5 2 q (areas both e & f ) .
Equilibria of forces and moments acting pile are then investigated.
Fo - 2 C l L1 where parameter 13 = tan CY governs failure mechanisms. By minimising height of a sheet pile N about parameter 0= tan a , the optimum mechanism /j = 1, i.e., a = n / 4 can be obtained from aH/t2/3 = 0. Thus, non-dimensional limit height of a sheet pile h = H / l can be evaluated in a closecl forin equation. 1 6
-h3
-
11’
-r
~ i= 0,
Ad
= MO
+ Fo
.,7
- c12’
+H (
2
{
- L , ) 2c1( 2
-
L,
(12) where function H ( z - L , ) is Heavkide’s step function defined as 1 for ( 2 2 L , ) H(2 - L,) = (13) 0 for ( 2 < L l )
{
Thus, a location zrn where the maxiinum bending moment occurs can be calculated as
= 0,
Because it is requested that .U7,?is lo\vei. tliaii a yield moment bly,a following inequalit!- should be satisfied.
+
My 2 AfTn = nl(:,,) = .U,
Vertical normal stress component oYycoincides with overburden self-weight, i.e. only a linear function of y. ozs
(10)
Since external forces are acting as shown in Fig. 3 , bending moment distribution in a sheet pile is in the following form,
Stability ana1.ysis based on lower hound met h o ci Lower bound analysis is done for an assumed stress field shown in Fig. 3. It is necessary to show that the assumed stress field satisfies equilibrium of stresses (or forces) and doesn’t violate the yield conditions. Mohr’s stress circles for each parts of soil masses in Fig.3 are shown in Fig. 4. For the simplicity of the analysis, the principal axes are assumed t o be coincide with vertical and horizontal directions and to satisfy the following conditions giver1 as inequalities (9) for no violation of yield conditions. This stress field is very conservativ(2, because only a few areas are in fully plastic state and the others are within a yield surface. These stress fields arc: obviously satisfying equatioiis of equilibria c ~ pb, ~ = ~ 0,, under ~ the assumptions as follows;
Horizontal normal stress component constant iIi the x-direction.
1
34 = 0, 2
where Fo and A40 are horizontal force and monients acting at a foot of a sheet pile expressed as
2.2
0
(9) a sheet
= U,
MO + CIL? - 2ClL’(L, t
where 1 = c o / ( p g ) . 771 = iWy/(co. P)are reprtseniative length aiid non-dimensional bending strength of a sheet, pile, respectively. This is a result of upper bound analysis.
0
+ 2C11,’
011
+ FOL
__ 4Cl
(15)
Noting that this inequality is only a function of length Lo, 1he maximum length of Lo call be obtained as
is
317
,
3 DISCUSSION AND ITS APPLICATION FOR DESIGN PROCEDURE where rn is non-cliineiisio~ialbending strength of a sheet pile defined previously as 772 = M g / ( c 0 . 1 2 ) , and k is a ratio of a supporting layer and a reclaimed land defined as k = c1/co. As upper part of reclaimed land is self-standing (see Figs. 3 Rr 4), non-dimensional critical height of a sheet pile 17 is expressed as
Other lengths L1 arid 1,2 are directly ohtained by Eq. (11) as
It should be noted that lengths Lo, L1 and L2 obtained by eys. (16) and (18) must satisfy inequalities (9) for the sake of statically admissible stress fields. Inequalities (9) can be summarised to one inequality in a non-dimensional form,
L L L2 'fO+L+-<.Sk, ... 1 1 I -
( 19)
where a sheet pile hight H is deleted by virtue of Eq. (16). According to the discussion above, lower bound solution can be ohtained as follows; 0
0
0
0
For a certain set of strength parameters k and m.where k is a strength ratio of rcclaimed and supporting layers, and m is a non-dimensional bending moment strength of a sheet pile, respec t ivcl y. Find a maximum non-dimensional height, of a sheet pile h. Length Lo is first calculated by Eq. (16). Then, lengths L1 and L 2 are calculated by Eqs. (18). If inequality (19) holds true, obtained values are surely one of lower solutions. Elsc. there is no feasible solution.
Calculated results are presented and discussed i n the following section.
At first, for investigation of ranges where exact solutions exist, both results of upper bound analysis Eq. (8) and results of lower bound analysis Eqs. (17) and (19) are shown in Fig. 5 . For upper bound analysis, it should be remembered that a location of a plastic hinge is fixed a t the boundary of a reclaimed and supporting layers. This assnmption may give a good estimation, but it cannot be guaranteed that results presented here are the best ones. For lower bound analysis, solutions are depending on the parameter k which is a strength ratio of two layers. For smaller k cases, feasible solutions cannot find for bigger m cases, which imply that a bending strength of a sheet pile is high. This is interpreted as follows. Let us assume that sheet piles with a high strength are used for weak ground condition cases. Even though a horizontal load due to a reclaimed land is transmitted to a supporting layer via sheet piles, a supportiug layer cannot afford t o carry it and fails. For stronger moment strength cases, the results of this study show fairly well because differences of both lower and upper solutions become smaller. This is very importhnt information which tells where the exact solutions exist. A narrow zone surrounded by two lines is the area of the exact solutions. It can be concluded that this method is expecting and gives sufficient information for practical designing process. Errors are within 20% for cases of over n z = 30 Ly doing a very siiiiplv analy si s presented here. It should be noticed that from upper bound analysis, they are overestimated especially for smaller moment strength m cases, however. Prcvious studies of a stability problem of non-supportcd vertically cut slope that is coincident to m = 0 case show that best known upper bound solution of critical height is only h = 3.83 for a circular slip mode (Shield, 1954), which is only about 60% of a solution of this study. We must recognise here agairi that velocity boundary conditions along a sheet pile are given as a linear distribution with depth in this study, which is totally different conditions for non-supported vertical cut cases. t\round m = 0 cases in this study, boundary coiiditioris along a sheet pile arc rather imaginary; although a sheet pile has a very weak yield bending strength, but velocity distribution is kept linear due to a sheet piles. This constraint has strong effects on the so111t ion.
318
Figure 4: Mohr’s stress circles for each parts shown in Fig. 3 imum installation depth of a sheet pile cl is then defined as
L1
d=-+-. 1
Figure 6: Relation between yielding moment of a sheet pile 772, soil strength ratio k and installation depth of a sheet pile d for the assumed stress field For practical interest, it is also important to cstimate an installation depth of a sheet pile. According t o this lower bound solution, an installation depth should be a t least summation of two lengths L1 Lz. Non-dimensional form of a min-
+
319
L2 1
(20)
A non-dimensional minimum installation depth of a sheet pile d can be estimated by Eq. (18) for a set of parameters ( m , k ) which satisfies Ey. (19). Relation between yielding bending moment of a sheet pile 7 n , soil strength ratio k and miriiInuni installation depth of a sheet pilc cl is shown in Fig. 6. According to Fig. 6, main stability factors of these structures seem to be both yielding moment of a sheet pile and soil strength ratio. This implies that we cannot expect more reinforcement effects by just installing sheet piles a t much more deep levels than a certain suitable level shown in Fig. 6. It may be interpreted that as failure of a sheet pile bending is dominant for this situation, deeper part of a sheet pile does nothing for the stability of these types of structures. If we expect more siipporting forces from a supporting layer, we sliould use shed piles with much more strength to transmit a horizontal load of a reclaimed land to deeper areas of a supporting layer. It should be noticed again that nothing will be expected by just installing sheet piles deeper than the levels shown here and there exists suitable combinations of moment strength 772 and soil strength ratio k . It is recommended that rational choice of a sheet pile should b~ done with considering both Figs. 6 and 5. From lower bound solutions, there exists suitable combinations of moment strength ni and soil strength ratio k . It is recommended that rational choice of a sheet pile should be done with considering both Figs. 6 and 5. The analyses in this study are very simple, however the results give important information of the solutions from both theoretical and practical interests. If more precise analysis will be in necd.
more general \.elocity fields including a circular slip mode arid the location of a plastic hinge should be investigated for upper bound analysis. More general stress fields including effects of self weight should be checked for lower bound analysis, because stress fields considered here are on very conservative and safe sides. If more general boundary value problems will be considered, limit analysis presented here can be extended directly to numerical analysis by spacial discretization which is called rigid plastic finite element analysis (Tamura et al., 1984, Kobayashi, 1999).
4
CONCLUSIONS
hilain conclusioiis obtained in this study arc as follows, e
Both upper ancl lower bound of critical heights of a sheet pile are investigated for various ranges of sheet pile strength n2 and soil strength ratio k . The results give important information of the solutions from both theoretical and practical interests.
e
From lower bound solutions, there exists suitable combinations of moment strength rn and soil strength ratio I;. There are suitable installation depths of sheet piles for this type of a structure. It is important information for bctter effects of reinforcement. The results seem to be sufficiently good for practical engineering.
e
The following points of the results should be noticed, however. From upper hound solutions, very simple velocity fields are considered. Although the results seem to be good enough for sufficiently strong sheet pile cases. the results are somewhat overestimated for less strong sheet pile cases. From lower bound solutions, the results may be conservative and on the safe side because soil mitsses are not always in fully plastic range and holding some rooms for better solutions.
e
The analyses presented in this study are very simple. However, It will give good information of solutions for both theoretical and practical points of view. if more precise analysis will be in need, limit analysis presented here can be extended to more general boundary d u e problems. For example, rigid plastic finite element method is one of them. In this sense, liniit analysis is applicable and expecting.
REFERENCES SalenSon, J. (1977) “Applications of the Theory of Plasticity in Soil Mechanics.’’ John Wiley & Sons. Shibata, T. SL H. Sekiguchi. (1995) “Bearing capacity of ground,” Kajima Press (in Japanese). Shield, R. T. (1954) “Plastic potential theory and the Prandtl bearing capacity solution,” J. Appl. Mech., 21 (2), 193-194. Tamura, T., S. Kobayashi & T. Sumi. (1984) ‘‘Limit analysis of soil structure by rigid finite element method,” Soils and Foundations, 24 ( l ) , 34-42. Kobayashi, S. (1999) “Bearing capacity & stability problems, ” in “Background of soil mechanics,” Part B, Chapter 2, Pre-summer school textbook. T ( ’ (Applied mechanics and niatliematics in geotcschnical engineering), JSG (in Japanese).
320
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Seismic retrofit design for liquefaction-induced ground displacement mitigation M. Kondoh Yokohama City,Japan
I. Tawara Port and Harbor Bureau, Yokohaina City,Japan
ABSTRACT: The authors have proposed the additional reinforcing structure which uses steel sheet piles with wedge shape heads for existing caisson type quay walls (Kondoh et al., 1998). An idea of the design is to utilize effect of the geometry brought by shape of the wedge. These reinforcing structures compact its surrounding ground due to placing the shape of the wedge. The effectiveness of this additional reinforcing structure was verified by the dynamic analysis through the site investigation after construction and the verification analysis, it is confirmed that the new seismic retrofit design is epoch-making to reduce their damage after the earthquake. This paper mainly presents the mechanism of seismic retrofit design and the results of the site investigation in field. prevents damages by liquefaction-induced large ground displacement.
1 GENERAL INSTRUCTIONS It is still fresh in our memories that quay walls of Kobe Harbor were greatly damaged by the 1995 Hyogoken-Nambu Earthquake. The major cause of this damage was liquefaction-induced large ground displacement. As an anti-liquefaction countermeasure, the method has been used to prevent liquefaction of the ground by its improvement. However, conventional design like a ground improvement is too cxpensive to construct in a wide area. This has led the necessity of new consideration for estimation and determination of liquefaction. This has also led the necessity of a cheaper and more compact earthquake-proof design with new reinforcement to protect existing quay walls from liquefaction that is applicable to these walls. This paper introduces the seismic retrofit design using steel sheet pile with wedge shape head that
2 OUTLINE OF SEISMIC RETROFIT DESIGN Figure 1 shows an outline of the seismic retrofit design using steel sheet pile with wedge shape head. It is a cross-sectional view of the example where steel sheet piles are used to front quay walls located 4.5m below sea level. These quay walls are constructed 12m below sea level by reclamation. Sands were used their reclamation, while rubbles were used to the mound and the backfill behind caisson. The stratum includes soft alluvial clay, which is offshored piled 40m deep under the reclamation layer. About 60m below sea level, there is a distribution of Kazusa group of based rocks which is a seismic bedrock for engineering. There is a worry about the occurrence of liquefaction on the reclaimed soil layers.
Figure 1. Cross section of seismic retrofit d e s i p using steel sheet pile with wedge shape head
321
The seismic retrofit design uses steel sheet piles with reverse-triangle wedge shape heads of 4.0m (width of the top of the wedge shape head) x 4.0m (height) x 6.2m (depth). These steel sheet piles are used in parallel with the perpendicular line of quay walls, with a distance of 8m from the center of these steel sheet piles. The surrounding ground was compacted by pushing wedge shape heads into a desirable depth with a vibrator, thus increasing the resistance to liquefaction. In this case, we can design using IV-type 15m-steel sheet piles to make up the quay wall. 3 METHOD OF CONSTRUCTION The method of construction of the seismic retrofit design is shown in Figure 2. First, wedge shape heads are pushed into the ground with a vibrationhammer to compact its surrounding ground. Then, steel sheet piles are constructed along their central grooves. Finally, wedge heads and steel sheet piles are combined.
4 MECHANISM OF SEISMIC RETROFIT DESIGN Figure 3 shows a conceptual view of flow force and resistance force acting on these conventional steel sheet pile wall method and steel sheet pile with wedge shape heads in case of liquefaction-induced large ground displacement. Conventional methods with steel sheet piles use a design technique to allow the subgrade reaction from the sea and sheet pile bending resistance to resist the flow force from the land. The seismic retrofit design provides the subgrade reaction and sheet pile bending resistance from the sea with resistance to the flow force from the land. It is expected that the subgrade reaction should reach its peak at the sea bottom where the heads of a steel sheet come. However, the subgrade reaction in the horizontal direction of the ground is not satisfactory at the sea bottom. This calls for the use of steel sheet piles with a larger bending resistance. The mechanism of the seismic retrofit design can be divided into two major parts. One is compaction of the surrounding ground where wedge shape heads are placed. This compaction increases thc resistance to liquefaction of the ground around wedge shape heads, providing a dircct effect to prevent liquefaction. The other lies in a geometrical brought by the wedge shape. In general the subgrade reaction is greater in the horizontal direction than in the vertical direction. This decreases necessary section performance of steel sheet piles from the viewpoint of design, shortening the length of the sheet pile. This seismic retrofit design is featured by increasing the resistance to liquefaction and the subgrade reaction, which causes a smaller ground displacement. 322
Figure 3 . Conceptual view of flow force and Resistance force
5 SITE INVESTIGATION IN FIELD Before and after the construction of the seismic retrofit design witch wedge shape heads were pushcd
into the ground with a vibration-hammer to compact its surrounding ground, in-situ test using the standard penetration test (the following, SPT) was carried out in such arrangement as shown in Figure 4. Problematic soil in the field is the part of hydraulic fills that was constructed by dredging. The fills are mainly consisted of fine sand with shell fragments and interbedded with clayey-silt thin layer of 515cm in thickness about 3m in depth. And also before the construction, N-values of SPT in the fills are less than 10 blows that show relatively loose sandy soils. These fills develop from the bed of the sea to about 6m in depth above sand mat layer underlying normal marine clay to 40m in depth. N-values of SPT after the construction are shown in Figure 5. It is clear that N-values are increased over 20 blows near the bottom of wedge shape head. N-values show a tendency to increase as closer to wedge
I
Existing quay wall
Steel sheet pile with
I
F i g r e 4. Arrangement of execution of site investigation
shape head. It was found through site investigation in field that the surrounding ground close to wedge shape head is compacted. 6 VERIFICATION OF PROPOSED NEW SEISMIC RETROFIT DESIGN The effect of the additional reinforcing structure was verified by the dynamic analysis with effective stress response called "FLIP')". The detailed conditions for analysis are referred references (for example: Kondoh et al., 1998). Figure 6 shows the view of computed residual deformations occurred around the quay wall at 20 seconds of shaking. With the no-countermeasure model, the horizontal displacement of the top of the caisson is 196cm toward the sea and the vertical displacement is 56cm toward the settlement. With the seis mic retrofit model, the horizontal displacement is 48cm to the sea and the vertical displacement is 17cm toward the settlement, which means a great decrease in deformation and a desirable result of the seismic retrofit design. As shown in Figure 7, the time history of the horizontal displacements of the top of the caisson. The horizontal displacement starts at about 4 seconds when the input acceleration of both models reaches its peak. This displacement goes toward the sea at one time, while it is pushed back to the land at another time. In this way, it gocs toward the sea gradually, and reaches its peak after shaking.
7 CONCLUSIONS To reinforce existing quay walls, the seismic retrofit design is proposed in this paper, in which the ground characteristics and structural features are used, with
Figire 5. Results of site investigation in field
323
sirable method with respect to cost. It was found through site investigation in field that the surrounding ground where wedge shape head is compacted. N-values of SPT show a tendency to increase as closer to wedge shape head. The effect of the seismic retrofit design was verified by the dynamic analysis with effective stress response. This finite element method revealed that the seismic retrofit design has the most desirable effect to reduce the deformation of quay walls after the earthquake. 8 REFERENCES
1. Iai,S. Kameoka,T., “ Finite element analysis of earthquake induced damage to anchored sheet pile quay walls ” ,Soil and Foundations Vo1.33 , No.1 , pp71-91 , March 1993. 2. TowhataJ. Ishihara,K., “ Modeling soil behavior under principal stress axes rotation ” , Proc. 5Ih International conference on numerical method in geomechanics, Nagoya , Japan , pp523-530 , April , 1985. 3. Iai,S. Matsnaga,Y. Kameoka,T., “ Analysis of undrained cyclic behavior of sand under anisotropic consolidation ” , Soil and Foundations Vo1.32 , No.2 , pp16-20 , June 1992. 4. Kondoh,M. Baba,T. Sawada,S., ” Seismic retrofit design using steel sheet pile with .wedge shape head against liquefaction-induced ground displacement (Part 1&2) ” 33‘dJapan National Conference on Soil Mechanics and Foundation Engineering , pp.959-962 , July 1998 (in Japanese). 5. Sawada,S. Kondoh,M., “ Seismic retrofit design using steel sheet pile with wedge shape head ” , Proc. Znd International conference on earthquake geotechnical engineering7pp.353-358,June1999.
Figure 6. Computed deformation of a quay wall at 20 seconds of shaking
300
I
1
2 I
I
Drnax = 196 (crn)
Toward theland
U
m
Toward the sea
e
s ,$ -300
{
Z
0.0
I 2.5
7.5
5.0
12.5
10.0
15.0
17.5
20.0
Time (sec)
(a) No-countermeasure model I
I I
Drnax = 48 (cm)
Toward the land
t Toward the sea
-3000 0.0
2.5
5.0 ’
7.5
10.0
E.50
15.0
17.5
20.0°
~
Time (sec)
@) Seismic retrofit model
Figure 7. Computed response displacement at the top of caisson
flow mechanism due to liquefaction in mind. The seismic retrofit design uses much less steels than any conventional designs, taking the seat of the most de-
324
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 15 1 1
Prediction and management of consolidation settlement with master-curve method in Tokyo International Airport D. Kozawa, S.Yamaguchi, H. Matsumoto & M.Arata The 2nd District Port ConstructionBureau, Ministry of Transport, Yokohama,Japan
H. Nakanodo & Y. Kanno Fukken Company Limited, Consulting Engineers, Yokohama,Japan
ABSTRACT: A method using master curves was used in the management of consolidation settlement in the third-phase area of the Tokyo International Airport Offshore Expansion Project. In the prediction of consolidation settlement of grounds, a predicting method capable of simulating the actual settlement with high accuracy is indispensable. On the other hand, required in quick ground-improvement work is a system which enables simplified prediction and quick feedback of predicted results to the construction works. The mastercurve method proposed in this paper is a settlement-predicting method which uses simplified models and field data to make prediction of the same accuracy level as numerically analyzing methods.
1 INTRODUCTION The Tokyo International Airport Offshore Expansion Project consists of three phases. In the first phase, New A-runway was constructed. In the second phase, terminal facilities and aprons were constructed. In the third phase, New C-runway was constructed and other facilities are under construction. (See Fig. 1). The project site was a ground reclaimed from the sea with sludge (A,,) of 250% water content which was dredged up from Tokyo Bay. Lying on the layer A,, was a layer of construction waste soil (B,) from construction sites in the Tokyo metropolitan area. Lying under the layer A,, was a 30 to 40 meter-thick soft alluvial clay layer (A,,). In the first- and secondphase areas where both the layers A,, and Ac2 were relatively thin, the ground-improving work was carried out by the vertical drain method. Plastic board drains and small-diameter fabri-packed sand drains( @=12cm) were mainly used. In the thirdphase area where the layer A,, was relatively thick and the layer A,, was unconsolidated state with its self weight, another type of vertical drain method was adopted which used supplementary intermediate drains and sand drains partially sheathed with geotextile (see Figs. 2). Because the height of construction machinery was limited in this area due to the airspace restriction of the Airport, the groundimprovement work could not be performed to the bottom of the A,, layer, leaving behind the problem of residual settlement. In the third-phase area in 325
particular, many types of technique and devices had to be developed and used to cope with lots of restrictions and problems. This paper describes the master-curve method used in the management of consolidation settlement in the third-phase area. In this area, the technique of computer-aided construction management system was adopted which fed back the results of settlement predictions based on field-observation data to the ground-improvement work. Important in predicting the future consolidation settlement is a system capable of simulating the past consolidation settlement with high accuracy. On the other hand, important in quick execution of groundimprovement work is a simple prediction system and quick feedback of predicted results. The master-curve method is a settlementprcdicting method which uses simplificd models and field data to make prediction of the same accuracy level as numerically analyzing methods.
2 MODELING OF ANALYTICAL CONDITIONS It is desirable in the master-curve method to predict the consolidation settlement of a ground with a simple and easy solution as near the actual conditions as possible. Accordingly, several models were built in the master-curve method on the complex conditions of ground behavior.
consolidation model is built based on the past settlement data of a ground to reproduce the settlement. Then, the model of consolidation is reformed, for higher predicting accuracy, based on settlement data collected in the ground-improvement work.
2.2 Modeling of settlement-analyzing method
(I) Calculation of settlement In the third-phase area of the Tokyo International Airport Offshore Expansion Project where loading on the ground surface was wide enough compared with the depth of clayey soil layers, the consolidation settlement was taking place in a onedimensional deformation pattern, without lateral movement. Therefore, the consolidation settlement was calculated with the so-called C, method.
2.1 The Definition of master-curve inethod The master-curve method is a settlement-predicting method in ground-improvement work with vertical drains. It is different from the predicting method at the design stage which predicts, without data on the past settlement, the future settlement based on soil properties determined from soil exploration and laboratory tests and such conditions for construction work as determined from general work-execution plan. In the master-curve method, the parameters of
where, S , is the final consolidation settlement; N, thickness of each clayey soil layer; P,,, initial effective overburden pressure; A P, increase in underground stress; C,, compression index; and e,, initial void ratio. (2) Calculation of consolidation rate (a) Consolidation settlement of non-improved layers: Terzaghi's one-dimensional consolidation solution for homogeneous soil was used for calculating the consolidation settlement of the non-improved layers. (b) Improved layers with vertical drains: Barron's (1948) solution is available for the consolidation theory of vertical drains. It's approximate solution is often used in the design of ground improvement due to its simplicity and easiness. However, consolidation does not necessarily progress in such an ideal way as formulated by Barron. In actual grounds, there are problems such as resistance of the materials of drains. Regarding the problem of drainage resistance of drain materials, Yoshikuni and Nakanodo( 1974) devised a strict solution for well resistance, and Yoshikuni proposed an approximate expression for easy designing and a method of applying the expression to mat resistance. In the second-phase of construction area, the pore water pressure in drains and sand mats was measured to ascertain the existence of well resistance and mat resistance, and it was confirmed that they could be well explained with the approximate expression of Yoshikuni. In the third-phase area, the consolidation process was calculated with the following equations. i
g n
yTh)='-ed--1 fln)+0.8L 1
)
326
n2 3n2-1 F(n)=-----ln(n)-___ n2-1 4n2
where U represents the degree of consolidation; T,,, time factor of clay in horizontal direction; n, drain ratio (ratio of effective drainage diameter de to pile diameter d\,,); L,,, mat resistance; L,v, well resistance; k,, coefficient of permeability of clay; k,,,, coefficient of permeability of sand mats; H,,, thickness of sand mats; B , one half of the spacing between drainage pipes arranged in sand mats; and H , length of drains.
2.3 Simple settle}?ierit-a}iuly~i}i~~ method for tli soiigli rlra ins
~ioti-
I n general, drains are buried through all layers which are subject to consolidation. Adopted in the thirdphase area, however, was a so-called non-through drain method that leave the lower part of settling clay layer unimproved due to the limitation on the height of construction machinery. The ground was improved down to the depth of Ap - 28m, which left unimproved the lower part of the layerAc2 and the whole layer D c l , allowing them to develop consolidation settlement in future. Drainage under consolidation in the soil below such drains is complex. Horizontal drainage quickens both the consolidation of the soil penetrated by such drains and the soil below such drains. However, because no complete drainage layer is formed in the top zone of the soil below such drains, the water in the zone presents complex behavior as shown in Fig. 3. A series of FEM calculations were carried out to develop the design method for non-through drains, which showed that the distribution of consolidation degree in the soil below the drains could be explained by the top and bottom drainage theory if a plain-like drainage layer was assumed above the bottom level of the drains, and that the analytical solution by Terzaghi’s top and bottom drainage theory conformed well to the results of FEM if the drainage length of the soil under the drains was regarded as H+de/2. Accordingly, adopted in the consolidation calculation of the soil below the drains was a simple design method in which the above extra thickness de/2 was added to the thickness of the soil below the drains (Hitachi et a1.,1994).
Fig.5 Time-dependent changes of excess pore pressure in drains and sand mats 2.4 Aizalytical nzetlzocl for supplementary intermediate drains The coefficients of consolidation Cv of the supersoft dredged clay layer A c l and the alluvial clay layer Ac2 were largely different from each other, the former being 40 cm7/d, the latter being 100-200 cm’/d. It is desirable in such a multilayer ground to carry out ground-improvement work to attain similar degrees of consolidation in all the layers in a prescribed time period. In our case, sand drains were 327
stage of loading by banking and allowing the ground to settle, the predicting accuracy increased. In the last stage, the calculation predicted the actual settlement very accurately.
buried through the Acl layer into theAc2 layer. With the sand drains, the supersoft Acl layer was to lag behind the Ac2 layer in consolidation, requiring supplementary vertical drains between sand drains, the latter arranged at 2.5-meter intervals. However, considering the stability of the ground and the sand drains, the availability and efficient stationing and operation of construction machinery, etc., we used plastic board drains and fabri-packed sand drains as supplementary intermediate drains, as shown in Fig. 2 (Maruyama et a1.,1992). Because the consolidation behavior under the above conditions were complex, FEM calculations were carried out to determine the arrangement of drains. In consolidation calculation for groundimprovement design and consolidation management, the effect of supplementary intermediate drains was expressed by converting their diameter and spacing into equivalent ones which showed consolidation characteristics similar to those observed in the FEM.
3.2 Verification of delay in consolidation and nonthrough drains
3 EVALUATION OF MASTER-CURVE METHOD
In this chapter, the authors will describe the evaluation of the master-curve method. 3.1 Setting of nzaster curves The master-curve method is to review and renew the predicted settlement curve of the execution design stage based on the results of field observation and soil exploration and in accordance with reclamation and filling, or banking, plans. In the ground-improvement work in the thirdphase area, the master curves were reviewed in three stages; i.e., execution design stage, stage of installing settlement gauges layerwise, and stage of loading by banking and allowing the ground to settle. Fig.4 shows the comparison of the calculated settlement and the measured settlement at each stage. In Fig.4, the square marks indicate the settlement values measured by settlement plates set on the ground surface before the ground-improvement work. The lozenge marks indicate the settlement values measured by the settlement gauges installed layerwise immediately after burying drains. Because consolidation settlement was calculated for the period after burying drains, the comparison between calculated settlement and measured settlement was also made for the period after burying drains. It can be seen in Fig.4 that as the work progressed from the execution design stage to the stage of installing settlement gauges layerwise, and to the
The features of the master-curve method used in the third-phase area are the consideration of delay in consolidation due to well and mat resistance, calculation of consolidation settlement for nonthrough drains, and calculation of consolidation settlement for supplementary intermediate drains. In the third-phase area, piezometers, or pore pressure meters, were set in drains, sand mats, and the clayey soil layers. The piezometers in drains and sand mats were to determine the consolidation delay due to the resistance. The piezometers in the clayey soil layers were to monitor the consolidation degrees of the layers. In addition, the position of the drainage layer in the soil below the drains could be estimated qualitatively from the distribution of piezometers. Fig.5 shows the time-dependent changes of the excess pore pressure in drains and sand mats, the level of the banked soil, and the underground water level. Fig.6 shows the distribution depthwise of the pore pressure and the hydrostatic pressure in drains. It can be seen from Fig.5 that the excess pore pressure in drains rose immediately after banking soil and continued to be high for some time. It can be seen from Fig.6 that an excess pore water pressure occurred because the pore water pressure was higher than the hydrostatic pressure. These observation results proved the existence of well resistance. Fig5 shows the excess pore pressure in sand mats. The slight excess pore pressure in the sand mats indicated the existence of mat resistance.
328
Fig.6 Relation between pore pressure and hydrostatic pressure in drains
Fig.8 Distribution depthwise of pore pressure in clayey soil layers Fig.7 shows the time-dependent changes of pore pressure in the clayey soil layers. Fig.8 shows the distribution depthwise of pore pressure in the clayey soil layers at the completion of ground improvement. It can be seen from Fig.7 that the pore pressure in the clayey soil layers increased rapidly as soil was banked on the ground, and the pore pressure decreased gradually while the banked soil was left as if was. Fig.8 shows the hydrostatic pressure and the overburden pressure, too. It can be seen from Fig.8 that the top drainage layer of the soil below the drains was formed above the bottom level of the drains, conforming to the conditions for drainage layer set in the master-curve method. Thus, the existence of the consolidation delay and the position of the drainage layer in the soil below the drains were ascertained from the results of measurement of the pore pressure.
Fig. 10 Comparison between reverse-calculated values and test values of consolidation parameters
3.3 Comparison between consolidation parameters determined by tests and those reverse-calculated from nzeasured settlement In the master-curve method, the consolidation parameters of a ground and its layer configuration are first determined based on the results of soil exploration. Thereafter, as the field observation data build up, such parameters and layer configuration are reviewed so that actual measured settlement can be simulated with high accuracy. The values of consolidation parameters based on soil exploration are determined through the oedometer tests of undisturbed soil specimens obtained by soil exploration. However, the review of values of the
329
parameters are necessary because the values determined by consolidation tests differ from the actual ones due to various factors such as disturbance of specimens due to the collection of specimens, difference in stress condition between test models and actual ground, and scale effect. Values of consolidation parameters (C, and C,) determined by laboratory tests of specimens obtained by check boring carried out during the installation of settlement gauges layerwise and those reverse-calculated from the data on actual settlement in the third-phase construction area will be described. Fig.9 shows the frequency distribution of the consolidation parameters. Table 1 shows the statistics for the frequency distribution. In Fig.10, the values determined by tests and the reversecalculated values are directly compared with each other. The values for the dredged clay layer Acl presented a larger dispersion as compared with the alluvial clay layer Ac2-1 layer. In the Acl layer, the reverse-calculated values of both the compression index C, and the consolidation coefficient C, were tended to be smaller than those determined by tests, suggesting the delay in consolidation in the layer. In the Ac2- 1 layer, the reverse-calculated values of C, and C, conformed well to those determined by tests, but considerable dispersion was observed. The above suggests the significance of the mastercurve method, wherein calculation conditions are reviewed based on measured settlement, in the management of consolidation settlement of reclaimed grounds. Table1 Comparison between reverse-calculated values and test values of consolidation parameters (4c c Acl layer Ac2-1 layer Reverse- test-value Reverse- test-value calculated calculated value value Sample size 143 128 172 164 1.6 1.1 1.2 Max value 1.6 Average 0.5 0.6 0.7 0.7 Min. value 0.1 0.2 0.3 0.2 S.D. 0.2 0.3 0.2 0.2 (bKv Acl layer Ac2-1 layer Reverse- test-value Reverse- test-value calculated calculated value value 172 165 128 Sample size 146 1500 1730 2250 Maxvalue 2000 170 230 260 230 Average 50 20 50 50 Min. value 150 220 340 340 S.D. \
4 CONCLUSIONS The authors described the master-curve method used in the management of consolidation settlement in the third-phase area of the Tokyo International Airport Offshore Expansion Project. This method would be effective as a method of managing the consolidation settlement of a vast reclaimed ground which contains many uncertain factors at the execution design stage and which the selection of suitable execution methods are crucial for. To predict consolidation settlement with high accuracy, i t is important to set up an analytical theory and analytical conditions fitting to the field. Besides, for the feedback of observation results to work, the simplification and systematization of analytical work is important. (1) The master-curve method is practical and capable of predicting settlement with high accuracy and dealing with fill loads which changes. (2) The master-curve method is effective in predicting the settlement of a complex ground with vertical drains. (3) Because the consolidation parameters of clayey layers, the permeability of drainage layers, etc. of a ground under work are reviewed as the banking progresses, the accuracy in predicting the settlement of the ground increases. (4) The master-curve method enables simple, easy, and rational settlement management of grounds while their work is under way. Periodical measurement of settlement is still conducted in the Tokyo International Airport to grasp the residual settlement. The authors will make further verification of the practicality of the mastercurve method with data obtained from the long-term filed observation.
REFERENCES Barron, R.A.,1948. Consolidation of Fine-Grained Soils by Drain Wells. Dans. ASCE, Vo1.113, NO.2346, pp71 8-742 IIitachi, S., H. Yamamoto, N.Ikeda, K. Oikawa, & H.Nakanodo 1994. Consolidation with nonthrough vertical drains. Proc. 2Y“ Japan National Conference on Soil Mechanics and Foundation Engineering, pp. 2107-2110.(in Japanese) Maruyama, H., Y. Kawakami, K. Watanabe, H. Nakanodo, & Y. 1n;aoka 1992. Consolidation process with vertical drains and supplemental intermediate drains. Proc. 27”’ Japan National Conference on Soil Mechanics and Foundation Engineering, pp. 21 81-2184.(in Japanese) Shiomi, M., H. Kanazawa, M. Inada, & N.Fukuda 1996. Planning and practice of ground improvement for construction of airport on
I
330
supersoft ground. Journal of Japaiz Society of Civil Engineers, Vol. 32, No. 546, pp. 23-37. (in Japanese) Yoshikuni, H. & H. Nakanodo 1974. Consolidation of soils by vertical drain wells with finite permeability. Soils and Foundatioizs, Vol. 14, No. 2, pp. 35-46.(in Japanese)
33 1
This Page Intentionally Left Blank
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Analysis of lumpy fill R. Manivannan, C. E Leung & S.A.Tan Centrefor Soft Ground Engineering, Department of Civil Engineering, National University of Singapore, Sing apore
ABSTRACT: Land reclaimed using clam-shell dredged barge deposited material is termed as lumpy fill. The computer programs SIGMA/W and SEEPNV are used to analyse the performance of lumpy fill using a field case study data obtained from Halmstad Harbour in southwestern Sweden. The results of the analysis show that high pore water pressure exits inside the clay lumps while the pore water pressure in the interlump void is small. Redistribution of load takes place as the inter-lump voids close up upon loading.
1 INTRODUCTION Land reclamation using dredged material can be carried out by various methods, including hydraulic filling of sandy or clayey materials and clam-shell dredged barge-deposited stiff clay. When stiff clay is dredged from the seabed using a clam-shell grab dredger and placed using a barge, the resulting reclamation fill consists of stiff clay lumps having large inter-lump voids. Such fill, termed as lumpy fill in the present study, may undergo unpredictable settlements of large magnitude due to closing up of inter-lump voids. In the present study, the analysis is applied to back analyse the field data obtained from a case study carried out in Halmstad Harbour located in southwestern Sweden. Material dredged from the harbour approach was deposited between two breakwaters by bottom-opening barges (Hartlen and Ingers, 1981). Total thickness of the fill was 6.4 m, of which approximately 3.0 m at the bottom was silty clay dumped by barge and approximately 3.4 m on top was hydraulically placed silty clay (Figure 1). The fill was placed on 1 to 1.5 m of stiff silty sand overlying stiff silty clay. The dredged material consisted mainly of stiff silty clay, which formed blocks of up to 1 m3 in size. Stiff silty clay, originally overconsolidated, was pumped into a part of the area to be reclaimed. The clay was dredged using a cutter suction dredger and the material was deposited from several points within the containing bund. The material leaving the dredger discharge pipe consisted of wellrounded lumps of various sizes, with diameters of up to 300 mm. Certain non-uniformity was obtained because large lumps of clay settled close to the
discharge pipe (Figure 2). The 3 m embankment corresponding to a surcharge of 58 kPa was placed after 424 days of consolidation. Figure 3 shows the monitored settlement with time. 2
CONCEPT OF ANALYSIS
The consolidation settlement analysis is conducted using a 2-dimensional finite element computer programme. In the present study, Biot’s theory is. used to analyse the problem of lumpy fill consolidation. The lumps are modelled by regular spatial arrangement of blocks representing the large discrete chunks of ciay lumps while the inter-lum voids are modelled as very soft slurry. The I-m! blocks are assumed as spherical lumps with diameter of 1.24 m. The hydraulically placed lumps are taken as 300 m diameter spherical lumps. In order to allow for passage of water flow to be continuous in the 2-dimensional simplified analysis, the interface between clay lumps and void spaces is modelled by a thin layer of highly permeable soil. The properties of this material are same as those of the inter-lump void material. The ground water table is at 1.4 m below the top of hydraulically placed silty clay. This dry soil of 1.4 m at the top is considered as surcharge for analysis of settlement and pore water pressure. The SIGMA/W and SEEPNV computer programs are employed to analyse the consolidation of the lumpy fill. Both programmes were developed by Geo-Slope International Ltd, Canada (GeoSlope, 1999). The SIGMA/W programme is a finite element software that can be used to conduct stress and deformation analysis of earth structures. The comprehensive formulation makes it possible to
333
Figure 1 Schematic profile of reclamation fill in Hamstad Harbour, Sweden
Figure 2 Shear strength profile as measured by vane test (x) and unconfined compression test ( 0 ) (after Hartlen and Ingers, 1981) both and complex problems. On the other hand, S E E P N is a finite element 'Oftware that can be wed to the movement and pore-water pressure distribution within porous
materials such as soil and rock. In addition, SIGMNW can perform a fully coupled consolidation analysis in conjunction with SEEP/W. In the present analysis, the stress-deformation analysis is solved together with a flow analysis. This makes it possible to study the generation and dissipation of pore-water pressure in response to external loads. Figure 4 shows the geometry of the finite element mesh adopted for this problem. It consists of eight-noded quadrilateral elements together with nine-point integration and irregular boundaries are modelled with six-noded triangular elements together with three-point integration. The relationship between the hydraulic conductivity and pore-water pressure is known as the conductivity function. For analysis involving saturated soils, the conductivity function must be defined for each soil Figure shows the hydraulic conductivity function for stiff clay and inter-lump void material.
Figure 3 Settlement under the centre of the test embankment (after Hartlen and Ingers, 1981) 334
stiff clay and inter-lump void material are taken as I8000 kPa and 100 kPa, respectively.
3 RESULTS AND DISCUSSION Figure 8 shows the comparison of settlement with time between the field measurement and the numerical analysis. Numerical analysis predicts the settlement of lumpy fill well except between 400 to 850 days. This field settlement behaviour might have taken place due to rearrangement/movement of clay lumps after applying surcharge. Figure 9 shows the variation of pore water pressure with time. The results indicate that high pore-water pressure exists inside the clay lumps and low pore water pressure exists at the inter-lump voids. This is due to higher permeability of inter-lump void material and lower permeability of stiff clay lump. This observation suggests that short-term settlement is controlled by the consolidation of inter-lump void material and long-term settlement is controlled by the consolidation of clay lump. The total stress profile shown in figure 10 indicates that there is a redistribution of load between stiff clay lumps and soft material. It is noted that stiff material would experience higher stresses than the soft material and this facilitates the faster consolidation of clay lump. As a result, stiff clay lumps reduce the compressibility of lumpy fill compared to hydraulic fill. Figure 11 shows the dissipation pore water pressure with time at the four locations as shown in figure 4. Pore water pressure inside the large clay lump increases rapidly and dissipates smoothly.
Figure 4 Finite element mesh for Halrnstad harbour reclamation analysis
1,OE-11' 0
I
'
40
I
I
80
I
120
'
160
'
200
pressure (kPa)
Figure 5 Hydraulic conductivity function for stiff clay and inter-lump void material The amount of water stored or retained is defined by a soil-water characteristic curve. Figures 6 and 7 show the soil-water characteristic curves for stiff clay lump and inter-lump void material, respectively. Soil properties reported by Mendoza and Hartlen (1985) are used to obtain the hydraulic conductivity functions and soil-water characteristic curves in addition to the soil parameters obtained from the site investigation carried out during the case study. Laboratory test results on slurry clay and mixture of clay balls from the clay sample dredged from the Halmstad seabed were reported by Mendoza and Hartlen (1985). Elastic modulus of
0.4oov .. ...
0
'
'
40
'
80
'
120
'
'
160
200
Pressure (kPa)
Figure 6 Soil-water characteristic curve for stiff clay lump
335
1.00
10 days - - - - .
0.80
126 days
---
445 days
---.
1377 days
0.60 W
self-weight
0.40
inside lump 0.20
surcharge 0.00
I
I
I
0
I
I
I
300
200
100
Pressure (kPa)
0
Figure 7 Soil-water characteristic curve for interlump void material
40
80 120 160 Pore water p r e w r e (kPa)
200
Figure 9 Variation of pore water pressure with depth and time
0
\
surcharge applied after 424 days
100
E E v
n
I
E
:I I
200
-
v
! 36 t
1,.
W
-fEz
E
I
II
U
n
settlement at top of dumped material
W
2 2U
6
I
300 0
I 400
I
I 800
I
I 1200
-
+> nter-lump void
Clay lum
I
1600
71me (days)
I
0
Figure 8 Comparison of field measurement with numerical analysis
I 40
I
I
I
80
120
Total stress (kPa)
Figure 10 Variation of total stress with depth Pore water pressure at the inter-lump voids near the large clay lump takes some time to peak before dissipation. It peaks when most of the inter-lump voids close up. Similar observation was made for 300mm diameter clay lump and nearby inter-lump void. Higher pore water pressure is developed inside the large clay lump compared to the small lump. During surcharge, most of the inter-lump voids have closed up due to high immediate settlement.
This observation is in line with that observed in the centrifuge model study. Manivannan et al. (1998) reported that the substantial inter-lump voids closeup occurs upon surcharge loading. Pore water pressure at the inter-lump voids also peaks immediately as most of the inter-lump voids are closed up. This indicates that surcharge is a good ground improvement method for the reclaimed land using dredged stiff materials.
336
4. The use of cutter suction dredged materials should be encouraged for land reclamation instead of hydraulic fill.
1 Inside big clay lump
2 Void near big clay lump 3 Inside small clay lump 4 Void near small clay
REFERENCES Hartlen, J. and Ingers, C. 1981. Land reclamation using fine grained dredged material, Proc. of the 1Oth International Conference on Soil Mechanics and Foundation Engineering, Stockholm, Vol. 1, NO. 24, pp. 145-148.
.->.-
a
Geo-Slope, 1999. User’s guide for SEEP/W and SIGMA/W, Version 4.22, Geo-Slope International Ltd, Canada.
Surcharge of 58 kPa applied a t 424 days 0
L
0
400
800 Time (day)
1200
1600
Figure 11 Variation of pore water pressure with time If clay lumps are used for reclamation, the compressibility can be reduced and the. consolidation time can be fast. These findings are of practical and economical importance, and the use of cutter suction dredgers should be encouraged even for relatively soft clays. 4
Manivannan, R, J.C. Wong, C.F. Leung and S. A. Tan. 1998. Consolidation Characteristics of lumpy fill, Centrifuge 98, Tokyo, pp 889-894. Mendoza, M. J and Hartlen, J. 1985, Compressibility of clayey soils in land reclamation, Proc. of the Eleventh International Conference on Soil Mechanics and Foundation. Engineering, San Francisco, Vol. 2, pp 583-586.
CONCLUSIONS
Numerical analysis was carried to examine the behaviour of lumpy fill. The following conclusions can be drawn from the analysis: The pore water pressure inside the clay lump is always higher than pore water pressure at the inter-lump voids due to lower permeability of clay lump. As the rate of dissipation of pore pressure in the inter-lump voids is faster than that in the lumps, initial settlement is controlled by consolidation of inter-lump voids. Clay lumps carry most of the applied load and it increases the rate of consolidation of clay lumps. Stiff clay lumps reduce the compressibility of lumpy fill. During loading, high settlement takes place and most of the inter-lump voids close up immediately. Pre-loading is a good ground improvement method for the reclaimed land using dredged stiff materials.
337
This Page Intentionally Left Blank
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 I
Deformation and excess pore water pressure of the Pleistocene marine deposits due to offshore reclamation M. Mimura Disaster Prevention Research Institute, Kyoto Universiv,Japan
Y. Sumikura Department of Civil Engineering, Kyoto Universiry,Japan
ABSTRACT: A series of elasto-viscoplastic finite element analyses is performed to assess the stress and deformation of the marine foundation due to offshore reclamation. The foundation is modeled following the stratification at offshore reclamation site. Attention is paid to the modeling of permeability for Pleistocene sand layers considering the sedimentation environment because the performance of excess pore water pressure is strongly dependent on the extent as well as the change in thickness of those permeable sand layers. The mechanism for the propagation of excess pore water pressure is also discussed. It is very important to know how far the excess pore water pressure generated due to reclamation propagates for assessing the effect of adjacent reclamation on the existing reclaimed land. From those findings on the performance of excess pore water pressure, the mode of advance in settlement of Pleistocene deposits is discussed.
1 INTRODUCTION The outstanding development of coastal areas has recently been accomplished in Japan. A large-scale offshore reclamation in Osaka Bay is accompanied with large and rapid settlement of deep Pleistocene clay deposits. Akai and Tanaka (1999) reported that large and rapid settlement has been preceding without significant dissipation of excess pore water pressure in the Pleistocene clay and sand deposits in the foundation ground of the offshore reclaimed site. The phenomena taking place due to the reclamation is far from the conventional concept of consolidation in which deformation advances associated with the dissipation of excess pore water pressure. Ito et al. (2000) summarized the distribution of sandy deposits in Osaka Bay using the data from elastic wave exploration and in-situ boring logs. It has been showed that the distribution of sandy deposits is different in each deposit subjected to the difference of sedimentation environment. The most serious problem originating from sandy deposits is permeability that controls the rate of consolidation of surrounding Pleistocene clays. The Pleistocene clays have distinguished structure due to long term effect of diagenesis. The compression behavior of this kind of clay has a strong resemblance to that of Canadian Clay (Leroueil et al., 1985) where no linear relationship is found between void ratio and the logarithm of applied overburden within the range of
virgin compression, though the origin of both clays is completely different. However, the effect of the above-mentioned properties such as the permeability of sandy deposits and compressibility of Pleistocene clays on the performance of stress and deformation of the marine foundation has not been rationally explained. In this paper, the performance of excess pore water pressure in the Pleistocene deposits is investigated through an elasto-viscoplastic finite element analyses for the model foundation that has been made on the basis of the subsoil data of the reclaimed site. The mass permeability of Pleistocene sandy deposits is selected as one of the parameters controlling the deformation of the foundation ground. Propagation of excess pore water pressure in the sandy deposits is also discussed based on the elastic volumetric strain of sandy deposits. It is of importance to know how far the generated excess pore water pressure will propagate in permeable sand layers because it directly influences the performance of the adjacent structures. The effect of the existence of excess pore water pressure in the permeable Pleistocene sand layers on the subsequent advance in settlement of the Pleistocene deposits is of great concern for prediction of the long-term settlement of the offshore-reclaimed marine foundations. Based on the trial calculation, the authors show how to assess the actual behavior of the foundation ground that is taking place in the case of large-scale offshore reclamation.
339
,CL
2 ELASTO VISCOPLASTIC CONSTITUTIVE MODEL AND FINITE ELEMENT FORMULATION
Surface Load
Sea Bed -30
The elasto-viscoplastic constitutive model used in this paper was proposed by Sekiguchi (1977). Sekiguchi et al. (1982) modified the model to a planestrain version. The viscoplastic flow rule for the model is generally expressed as follows:
A
E
-50
v
-70 .*
6 -90
iii
-110
I
Ma9
n-7
-130
-150
in which F is the viscoplastic potential and A is the proportional constant. Viscoplastic potential F is defined as follows:
[ Yt
F=cc.ln l+-exp
600 m
ACT^)
(3) Where {ACT'}and {AE} are the associated sets of the effective stress increments and the strain increments respectively, and [Cep] stands for the elastoviscoplastic coefficient matrix. The term { aR} represents a set of 'relaxation stress', which increases with time when the strain is held constant. The pore water flow is assumed to obey isotropic Darcy's law. In relation to this, it is further assumed that the coefficient of permeability, k, depends on the void ratio, e, in the following form: k
= k,
y)
*exp[ e - e
................>
3 PROBLEMS SET UP AND PARAMETERS
=vp
in which cx is a secondary compression index, Oois the reference volumetric strain rate, f is the function in terms of the effective stress and vp is the viscoplastic volumetric strain. The concrete form of the model is shown in the reference (Mimura and Sekiguchi, 1986). The resulting constitutive relations are implemented into the finite element analysis procedure through the following incremental form: {ACT'}= [C"] {AF} -
*
Figure 1. Model foundation for the finite element analysis
(31 -
4 ....................................................
(4)
in which ko is the initial value of k at e=eo and hk is a material constant governing the rate of change in permeability subjected to a change in the void ratio. Note that each quadrilateral element consists of four constant strain triangles and the nodal displacement increments and the element pore water pressure is taken as the primary unknowns of the problem. The finite element equations governing those unknowns are established on the basis of Biot's formulation (Christian, 1968, Akai and Tamura, 1976), and are solved numerically by using the semiband method of Gaussian elimination.
340
A series of elasto-viscoplastic finite element analysis is performed to assess the stress-deformation characteristics for the set up marine foundation consisting of alternated clay and sand deposits, which has been modeled with the boring data from the reclaimed site in Osaka Bay. The model foundation used is shown in Figure 1. Here, Ma and Ds denote a marine clay layer and sandy layer respectively. Ma 13 is the Holocene marine clay whereas others are the Pleistocene origin. The original foundation is assumed from the elevation of -17m to -150m, and in the normally consolidated Holocene clay deposit, Ma13, sand drains are assumed to be driven in a rectangular configuration with a pitch of 2.0 to 2.5 meters to promote the consolidation of this deposit. The modeling of sand drains is simulated by the macro-element method (Sekiguchi et al., 1986). Pleistocene clay layers are assumed to be lightly overconsolidated, and the value of OCR equals to about 1.3. The Pleistocene sand layers, which are expressed by Ds, are also assumed to be linear elastic material. Considering the mass permeability for these sandy deposits following the findings by Ito et al. (2000), the finite permeability coefficient, k, the constant value of 0.864m/day, is introduced for Ds 3, 4, 5, 7, and 8. On the other hand, 0.0864mlday is assumed for Ds 6 and 9 because the mass permeability for these 2 sandy layers is found to be relatively poor. The bottom boundary is assumed to be perfectly drained because the sandy layer, DslO that holds high capacity of permeability, underlies the Pleistocene clay layer, Ma7. This is regarded as a reference, called Case 1. On the other hand, the comparative analysis is performed with the assumption that all sandy layers have infinite permeability in order to evaluate the effect of excess pore water pressure in sandy layers due to the
subsequent settlement of Pleistocene clay deposits (Case 2). The parameters for the used model are rationally determined based on the procedure proposed by Mimura et al. (1990). The construction sequence is assumed that the reclamation will be completed in 960 days with a steady rate of loading. Surface load is applied to the corresponding nodal points of the finite element mesh, and as a result, the maximum effective overburden on the original seabed will become 470.4kPa at the completion of reclamation (see Figure 2).
4 RESULTS AND DISCUSSIONS The calculated contours of excess pore water pressure are shown in Figure 3 for Case 1. It is exhibited in Holocene clay layer, Ma 13 that the excess pore water pressure is not so remarkably generated even at the completion of construction (Figure 3 (a)) and dissipated after 10 years (Figure 3 (b)) which has been subjected to promotion of drainage through sand drains driven in this layer. On the contrary, as the Pleistocene sandy layers for Case 1 have finite values for the coefficient of permeability, large amount of excess pore water pressure remains even in those sandy layers as well as in Pleistocene clays. At the completion of construction, generation of excess pore water pressure is quite significant in Ma12, and Ma 10 and 9 because Ma12 is assumed to be thick without a remarkable drainage layer in it and Ds 6 existing between Ma 9 and 10 is assumed to be poorly permeable in this Case. Although it is dissipated gradually with time, excess pore water pressure of 147 kPa has been seen in Ma 9 even after 50 years (Figure 3 (c)) because of the existence of poor permeable sandy layer, Ds 6 above it. It is also very interesting how the propagation of excess pore water pressure generated due to reclamation in the sandy layers occurs to avoid the boundary effect. From the calculated results shown in Figure 3, 300 meters from the centerline of the revetment is enough to satisfy this condition, and there is a need to explain the mechanism of excess pore water pressure propagation in the permeable sandy layers. The calculated distribution of volumetric strain of sand elements, exemplified by Ds 3 is shown in Figure 4 for Case 1 and Figure 5 for Case 2. In the present research, sand is modeled as a linear elastic material with a finite permeability for Case 1 and infinite permeability for Case 2. The rigidity of the elastic material, Go is determined from ~ for the corresponding sand layers for the N s p values both cases. Mimura and Sekiguchi (1986) pointed out that the foundation subjected to surface loading
341
exhibits compression beneath the loading area and dilation in the area where no surface load is applied on it. In those figures, the same behavior can be clearly seen. It is natural that the compressive strain occurs beneath the reclamation area due to the existence of surface load. Because of delayed dissipation of excess pore water pressure in sandy deposits as shown in Figure 3, the increment of effective overburden, A o is gradually increasing with time. With the increase in effective overburden with time, the compressive strain gains accordingly. On the other hand, there is no increase in total stress on the outside of the reclaimed area. Incremental stress, A o can be generated only through the increase in excess pore water pressure propagated from the area beneath the surface load is applied in Case 1. On the other hand, we have almost one definite distribution of volumetric strain for Case 2 because there is no excess pore water pressure in Ds 3, namely, constant total stress condition irrespective of time. The ob tained relation from Figure 5 can be a so-called reference volumetric strain distribution by intrinsic deformation. Then, we can obtain the components of elastic volumetric strains due to propagated excess pore water pressure by subtracting the values shown in Figure 5 from the corresponding values of Figure 4. Figure 6 shows the distribution of the elastic volumetric strain due to propagated excess pore water pressure in Ds3. At the completion of reclamation, significant dilative deformation takes place near the revetment and it decreases gradually with the distance from the revetment. As a whole, the dilative strain is decreasing towards zero with the time, and finally it reaches the condition with no volumetric strain after 50 years. At this point we do not have any excess pore water pressure in Ds 3 at all. From these results, it is found that the generated excess pore water pressure in sandy deposits can not be propagated boundlessly because it is absorbed by the elastic dilation of sand due to expansive forces by the propagated excess pore water pressure. It is thought that the propagating distance of excess pore water pressure in sand layers can be controlled depending on the elastic rigidity of sand, but it has not been confirmed yet. This problem can directly be connected with the effect of the adjacent reclamation on the present airport fill. From the results of this particular case, the propagation distance of excess pore water pressure is limited within about 200 meters from the centerline of the revetment.
L. .3
-2
3
18000 days (50 years) 960 days Elapsed Time (days)
Figure 2. Modeling for loading sequence
Figure 7 shows the comparative profiles of excess pore water pressure with depth for Case 1 and 2. As it is assumed that the Pleistocene sand layers are perfectly drained for Case 2, the profile of excess pore water pressure for Case 2 exhibits the spike shape with a value of zero at the sand layers. On the contrary, a large amount of excess pore water pressures remains even in those sand layers for Case 1. The process of dissipation is quite similar for Ma 12 on the both sides the highly permeable sand layers exist. It is because those sand layers can work sufficiently even if they are assumed to be partially drained. After 50 years from the start of construction, the excess pore pressure is almost dissipated for case 2, but more than lOOkPa can be seen in Ma 9 and 10 for Case 1. It is obvious that the difference between both cases is significant irrespective of elapsed time. Time - settlement relations for Pleistocene deposits are compared between Case 1 and 2 for the selected clay layers in Figure 8 and 9. The condition of perfect drainage of the sand layers for Case 2 accelerates the dissipation of excess pore water pressure in the clay layers, and it is naturally found that the settlement for Case 2 precedes that of Case 1. Ma 12 is overlain by Ds 1 sand layer and underlain by Ds 3 whose ability of permeability are quite high. Then, the difference in profiles of excess pore water pressure in Ma 12 is not so significant (see Figure 7) although the finite values for the coefficient of permeability are assumed for those sand layers in Case 1 while they are assumed to be infinite in Case 2. Following the above-mentioned facts, the advance in settlement for Case 1 and Case 2 can be regarded almost the same as shown in Figure 8. It is reasonably accepted that the compression of clay layer advances equally when the process of dissipation of excess pore water pressure is same. On the contrary, the Pleistocene clay, Ma 10 is underlain by the sand layer, Ds 6 that is considered to have a poor permeability. Let us discuss the effect of permeability of sand layers on the advance in settlement of clay layers. As shown in Figure 9, the difference in settlement is larger for Ma 10 between Case 1 and Case 2, compared to Ma 12 (see Figure 8). This difference is caused by the dissipation process of excess pore water pressure in the correspond-
Figure 3. Contours of excess Pore water Pressure
ing layers. As is known from Figure 7, 100 kPa of the maximum excess pore water pressure can be seen in Ma 10 after 10 years for Case 2, whereas 300 kPa remains for Case 1. This difference in effective stress due to dissipation of excess pore water pressure induces the difference in settlement between
342
Case 1 and 2. However it is also very interesting that the difference in subsequent advance in settlement becomes less with time. It is caused by the fact that the compression of clays follows the linear relation on e-logp curve. The contribution of excess pore water pressure to the reduction in void ratio is larger in the range of low effective stress. In other words, the absolute value of settlement is larger in the primary range of consolidation than that in the later range where the effective stress becomes large due to the dissipation of excess pore water pressure even if the quantity of excess pore pressure dissipation is completely same. As far as the advance in settlement of Pleistocene clay deposits is concerned, it should be noted that the long-term settlement that can be seen in Figure 8 and 9 is mainly due to a so-called primary consolidation with the dissipation of excess pore water pressure with time. The secondary consolidation can only be seen in the Holocene clay layers where the primary consolidation is completed early due to the drainage effect of vertical drains installed. Another point to be noted is that the introduction of deterioration of clay structure due to plastic yielding into compression model (Mimura et al., 1994) can qualitatively provide the performance that the settlement advances faster without a significant dissipation of excess pore water pressure in the Pleistocene clay and sand layers.
Figure 5. Distribution of elastic volumetric strain in the Pleistocene sand layer, Ds 3 (Case 2)
Figure 7. Profiles of excess pore water pressure
343
rial with a finite permeability. Because of its elastic volume dilation, the excess pore water pressure can not propagate boundlessly. The distance of propagation is limited within 200 meters outside from the revetment in this particular case. This problem is directly connected to how the adjacent reclamation affects the existing reclaimed fill. In the sense, the precise modeling of sand in terms of elastic rigidity as well as the mass permeability is indispensable. Finally, the contribution of excess pore water pressure to the subsequent settlement is found to be more significant in the primary range of consolidation. The difference in settlement subjected to the existence of excess pore water pressure in the permeable sand layers is remarkable in the early stage of consolidation. Furthermore, in order to describe the actual performance at the site, the realistic modeling for the structural Pleistocene clays should be required, for example, non-linear compression behavior on e - logp relations which takes place associated with plastic yielding. REFERENCES Akai, K. & T. Tamura 1976.An application of nonlinear stressstrain relations to multi-dimensional consolidation problems. Annuals DPRI, Kyoto University, 21(B-2): 19-35(in Japanese). Akai. K. & Y. Tanaka 1999. Settlement behavioour of an offshore airport KIA. Proc. 12‘” ECSMGE, 2 :1041-1046. Christian, J.T. 1968.Undrained stress distribution by numerical method. Journal of Soil Mech. and Foundation Div., ASCE,
Figure 9.Time - settlement relations for Ma 10 clay layer
94 (SM6): 1333-1345. Ito Y., K. Takemura, D. Kawabata, Y. Tanaka & K. Nakaseko 2000. Sedimentation process in the Quaternary Southern Osaka Basin inferred from reflection seismic interpretation. Quaternary Research, Submitted. Leroueil, S., M. Kabbaj, F. Tavenas & R. Bouchard 1985. Stress-strain-strain rate relation for the compressibility of sensitive natural clays. Geotechnigue, 35(2) : 159-180. Mimura, M. & H. Sekiguchi 1986.Bearing capacity and plastic flow of a rate-sensitive clay under strip loading. Bulletin of DPRI, Kyoto University, 36(2) : 99-111. Mimura, M., T. Shibata, M. Nozu & M. Kitazawa 1990. Deformation analysis of a reclaimed marine foundation subjected to land construction. Soils and Foundations, 30(4) :
Assembling those aspects discussed in this paper is necessary to evaluate the actual phenomena occurring in the Pleistocene marine foundations due to offshore reclamation.
5 CONCLUSIONS A series of trial finite element analyses is performed to assess the stress and deformation of the Pleistocene marine foundation due to offshore reclamation. On the basis of the findings on the macroscopic distribution of Pleistocene sand layers, the model foundation and mass permeability of those Pleistocene sand layers have been set up for the finite element analysis in the present study. The mode of generation/dissipation of excess pore water pressure is found to depend on the modeling for the permeability of the Pleistocene sand layers. Mass permeability plays a significant role for the process of excess pore water dissipation. The mechanism of excess pore water pressure propagation in Pleistocene sandy layers is controlled by the volumetric change of sand which can expand due to the increase in inner stress increment by the propagated excess pore water pressure. In the present study, sand is modeled as a linear elastic mate-
119-133. Mimura, M., T. Shibata & K. Watanabe 1994.Post yield modeling of compression for Pleistocene clays and its application to finite element analysis. Proc. Pre-failure Deformation of Ceomaterials, 1 : 517-522. Sekiguchi, H.1977.Rheological characteristics of clays. Proc. 9th ICSMFE, 1 : 289-292. Sekiguchi, H.,Y. Nishida & F. Kanai 1982.A plane-strain viscoplastic constitutive model for clay. Proc. 37th Natl. Conf, JSCE : 181-182(in Japanese). Sekiguchi, H., T. Shibata, A. Fujimoto & H. Yamaguchi 1986. A macro-element approach to analyzing the plane-strain behaviour of soft foundation with vertical drains. Proc. 31th Symp., JSSMFE : 111-120(in Japanese).
344
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Observation and analysis of ground deformation of a road embankment on a manmade island S.0hmaki & K. Saeki National Research Institute of Fisheries Engineering, Fisheries Agency, Japan
S.Shikata Japanese Institute of Technologyon Fishing Ports and Communities,Japan
s.suzuki Chiba Institute of Technology,Japan
ABSTRACT: The sand drain method was applied as a countermeasure to a road embankment constructed on an artificial island built 011 soft seabed. In order to estimate the embankment settlement and its effects on the surrounding structures, the authors conducted field observations and carried out numerical simulations of two-dimensional consolidation. Tlie results of our investigations showed (1) the nuinerical simulations slightly underestimated the embankment settlement at the center, whereas the lateral deformation of subsoil at the end of embankment was slightly overestimated, (2) tlie residual settlement estimated from the numerical simulations five months after tlie construction of the test enibankinent was about 50 ?40 of the settlement observed in tlie field, and (3) tlie lateral deformation at a lighter’s wharf increased in the direction toward the sea immediately after the construction began but soon afterwards it became constant and was therefore considered to have been stabilized.
1 INTRODUCTION In general, fishing port structures in Japan are designed using a standard design method based on the results of a soil survey and laboratory tests. However, in some cases where embanking ccnditions are severe, or a new method is employed, or unexpected deformations occur after einbanking, field observations and numerical simulations would be typically carried out (Ohmaki, et al., 1989; Mimura, et al., 1991; Oka, et al., 1995). This paper presents the results of field observations and numerical simulations of deformation in tlie ground and at a lighter’s wharf, caused by the construction of a road embankment on an artificial island built on soft ground. 2 CONSTRUCTION OF AN ARTIFICIAL ISLAND AND TEST EMBANKMENT
Figure 1 shows the layout of tlie artificial island, the test embankment area and the area in which the soil was improved using tlie sand drain method. Tlie artificial island has an area of 6.5 ha and the volume of reclaimed soil used in its construction was 300,000m3. The alluvial clay layer under tlie DL 2.0 m and DL - 1.5 in lighter’s wharves was improved using the deep cement mixing method before laying a foundation for a lighter’s wharf. Above this improved soil, rubble mound and three-layer con345
crete blocks were placed. The area behind tlie lighter’s wharf was backfilled with stones using a mixture of cement and soil. The cement mixed soil was obtained by mixing soil dredged up from the surrounding seabed with ceinent solidification materials. The area between DL 0 in and DL + 2.0 111was reclaimed using tremie pipes froin tlie sea (the target unconfined compression strength of the cement mixed soil q,, = 198kN/m2)while that between DL + 2.0 i n and DL + 4.7 in, which corresponds to tlie maximuin height, was reclaimed using the thin spreading method (the target q, = 98 kN/in2). Table 1 Construction process Process Reclamation(1st layer) Left untouched Reclamation(2nd layer) Left untouched Excavation Left untouched Spreading of sand mat Left untouched Driving of sand drain piles Left untouched 1st filling Left untouched 2nd filling Left untouched 3rd filling Left untouched
Cumulative Period(days) 0-13 14-128 129-596 597-964 965-971 972-981 982 983-989 990-1002 1003- 1015 1016-1018 1019- 1023 1024-1026 1027 1038 1039- 1042 1043-14000
-
Period (days) 13 115 468 368 7 10 1 7 13 13
3 5 3 12 4 12958
Figure 2 shows the physical and mechanical properties of the original ground. The figure indicates that the original ground consisted of a soft clay layer as the water content ,was in the range 100% to 170%, and the plasticity index ranged from 50 to 100. The wet density was almost constant (1.327 g/cm’ on average), except around DL - 20 in. The unconfined compression strength increased almost linearly with depth, except close to the surface and the compression index Cc was 1.9 on average, except near the surface. Figure 3 shows the physical and mechanical properties of the cement mixed soil after the reclamation. The water content was around 150%, which was almost the sanie as that of the original ground, and the mean wet density was I .282 g/cm3. The mean uiicoii fined coin p re ssion strengt 11, t 110ugh d ispersing slightly, was 8 1.3 kN/in2, which was significantly larger than the one obtained close to the surface of the origiiial ground. The mean coefficient of permeability obtained fro!n a falling-head permeability test was 7.23 x 10 -3 cm/s.
Figure 1. Planar layout of the artificial island
The test embankment area was excavated to DL + 2.0 ni after the reclamation, then a sand niat (thickness = 50 cm) was placed. Sand piles (diameter = 40 cin, center-to-center distance = 1.8 m) were driven down to the bottom of the alluvial clay layer (DL 19.0 in), within the rectangular area (1 8.4 i n x 27.4 in). Using the same cement mixed soil as used in the surrounding embankment, three layers at DL + 2.5 in to DL + 3.5 m,DL + 3.5 i n to DL + 4.5 in, and DL + 4.5 111to DL +5.3 m, were reclaimed as the test embankment on the sand mat. The cross section of the embankment is shown in Figure 5 . Table 1 shows the schedule of the various tasks in the construction of the test embankment. 3 PHYSICAL AND MECHANICAL PROPERTIES OF THE GROUND
The original ground consists of a 20 in thick soft alluvial clay layer from DL 0 i n to DL - 20 in, a 7 in thick alternated layer of alluvial sand and gravel layer and alluvial sand layer froin DL - 20 in to DL - 27 i n , and a Matsushim layer with tuff below DL - 27 171.
4 FIELD OBSERVATION
Figure 4 shows the layout of the soil iniprovement area and embankment area, and locations of transducers used in the field observation. Figure 5 shows the sectional layout of embankment materials and the locations of the transducers. The section is taken along the line 1-1 in Figure 4. A layer settlement gauge (S-I) was used to calculate the settlement of each layer, based on the changes in the depth of a settlement element located at each layer boundary in the center of the soil improvement area. A bore hole inclinometer (H-5) was used to measure the horizontal displacement at each depth. A vertical and horizontal displacement gauge (HV-3) was used to observe both layer settlement and lateral displacement at each depth. Settleinent plates (V-9 to V-18) were used to obtain the distribution of surface settlement along a line at the center of the embankment. A lateral earth pressure gauge (H-6) was used to measure the lateral flow pressure occurring due to sand pile
Figure 2. Physical and mechanical properties of the alluvial clay layer
346
Water content
(a)
wet density (dcm')
Unconfined compressive strength - 9.. (kN/m') 0 200 400 1.5
Coefficient of permeability (cm/s) do' 10* 10' 10'
1.0
0.5
0
-0.5
Figure 3. Physical and mechanical properties of the cement mixed soil
were placed at equal spacing from the center of tlie sand drain area, in order to verify the drainage capability of the sand mat. Figure 6 shows tlie changes with time in the distribution of surface settlement along tlie line on which settlenient plates (V-9 to V-18) were placed, and the distribution of lateral displacement with depth at sections HV-3 and H-5. The changes were referenced from the initial values of surface settlement arid lateral displacement which were set from the measurements made 011 27 April 1998 (the 987"' day in Table 1). Figure 6 indicates that the settlement in the center of the embankment was largest and the amount of settlement decreased in accordance with the distance from tlie center. The largest horizontal displacement in sections H-5 and HV-3 was observed around DL - 5.0 In and DL 0 m, respectively. The deformation modes of the above two sections are not the same because of the difference in lateral constraints OII the two sides of tlie ernbankment.
Figure 5 , Sectional layout of the soil improvement area, embankment, aiid measuring transducers
placement aiid fill loading. An H-shaped steel was inserted into the ground and an earth pressure gauge, pore water pressure gauge, and bore hole inclinometer were installed at various depths along its length as shown in Figure 5 . Pore water pressure gauges (Pa-l to Pa-3) were installedat DL - 4.0 DL - 10.0 m, and DL - 16.0 111, in order to get a better understanding of the dissipation of excess pore water pressure in the clay layer within the sand drained area. Three water level gauges (W-1 to W-3)
Figure 6. Distribution of surface settlement and lateral displacement
347
Of
the s-oulld
Table 2. Summaiy of soil parameters in the cohesive soil layer No. o f soil layer DL(rn) Plasticity index PI
1 fOw-3 60.9
2 -3--12 114.5
3 -12%-18 92.7
Notes
sin 4
0.349
0.330
0.352
sin ’‘=OB10.23310g,~P/
Coefficient o f earth
Fig2
d
pressure at rest
0.606
0.670
0.648
Kozl-sin
Compression index Swelling index K
0.517 0.046
0.807 0.070
0.807 0.070
Fig2 Experiment
Strength parameter M
0.907
0.742
0.798
Coefficient of secondary compression
(y
Initial rate of volumetric strain vo(l/m)
1.O x 10-3
1.O x
1 . 0 10-5 ~ 1 . 0 10-5 ~
M=6sin ”’(’-
1.O x 1OW3
Assumed
1 . 0 lOP5 ~
Assumed
rl0
0.178
0.141
0.153
rlo=(lKo),(1+2Ko)
Initial void ratio eo
2.807
4.051
3.517
Experiment
Coefficient of (m/day)
permeability
ratio U Unit weight y (kN/m3)
method (Yoshikuni 1979) taking the well resistance into account; then, the degree of consolidation of an equivalent uniform ground was evaluated by Terzaghis’ one-dimensional consolidation theory. Assuming that the actual time required for 50 % coilsolidation in the sand drained area and the equivalent uniform ground agree, the following equation can be obtained:
Here, k,,,, nz,,,, and Tl,e50(= 0.197) represent the coefficient of permeability, coefficient of volume compressibility, and time factor corresponding to the time of 50 % consolidation, respectively, of the equivalent uniform ground. Similarly, k,?, in,, ( = in,& and TI7,,represent the coefficient of permeability, coefficient of volume compressibility, and time factor corresponding to the time of 50 % consolidation, respectively, of the original ground. By using the values in Table 4, k,,, was obtained to be 2.21 x 10-’ cm/s, which was about 200 times as large as the coefficient of permeability of the original ground.
6.05 x lO-’ 1.73 x 10-4 1.04 x 10-4 Experiment 0.3 13.0
0.3 13.0
0.3 13.0
Assumed Fig.2
5 NUMERICAL ANALYSIS 5.1 Material properties of the growid The physical and mechanical properties of the alluvial clay layer mentioned in section 3 were divided into three categories, in order to create a soil model based on an elasto-viscoplasticity model proposed by Sekiguchi and Ohta (Sekiguchi & Ohta, 1977; Shibata & Sekiguchi, 1980). Table 2 shows the material properties of each layer used in the numerical calculation. Table 3 shows the material properties used in the analysis for materials other than the alluvial clay layer. These materials were assumed to be linear and elastic. Their locations are shown in the sectional layout in Figure 5 . The Youngs’ modulus for the cement mixed soil, the deep cement mixed soil and the soil in the embankment was obtained from the following equation (CDM 1993): E
=
5.3 Aiialyzed section and boundary conditiori
The section shown in Figure 5 was divided into finite element meshes. The area of analysis in the horizontal direction was 120 m from the center of the sand drained area; and that in the vertical direction was about 35 m starting at a depth of DL -25111. The displacement constraints on both sides of tlie section were as follows: the horizontal displacement was fixed and the vertical displacement was free; at the lower boundary, displacement in both directions (horizontal and vertical) was fixed; and at the upper boundary, displacement in both directions (horizontal and vertical) was free. As for hydraulic boundary conditions, tlie upper surface was assumed to be drained, and the other planes were taken to be undrained. Due to tlie constraints of the software used in the analysis, the cement mixed soil area and the embankment area to be excavated and reclaimed were assumed to be drained.
0.5 4,,
Here, E is Youngs’ modulus (MN/m2) and 4L,is unconfined compression strength (kN/m1). The uiiconfined compression strength of tlie deep cement mixed soil was assumed to be 980 kN/m2. 5.2 Equivalent coeflicieiit of permeability of the sand drain area Pore water at the sand drained area was drained horizontally from the clay layer to a sand pile during consolidation, then drained vertically upward through the sand pile. As the numerical procedure used in this study cannot analyze such complicated behavior of pore water, the analysis was carried out based on the assumption that tlie sand drained area could be modeled by an equivalent uniform ground. I n other words, the degree of consolidation at the sand drain area was evaluated by Yoshikunis’
Figure 7. Numerical results of the time history of settlement
348
just before loading tlie embankment (tlie 1015‘” day in Table 1) to five months later. Regarding the maximum surface settlement around the center of tlie embankment, the numerical result was smaller than tlie observed one. This may be because the cemerit mixed soil and the embankment area were assumed to be an elastic body for which it is difficult to express localized deformations. Figure 9 shows the comparison between observed and numerical results of the lateral displacement at sections H-5 and HV-3, from just before loading of tlie embankment to five months later. At section H5 , tlie numerically calculated maximum displace ment was about 15 cm (around DL - 3.0 In), whereas the observed maximum displacement was about 9 cm, smaller by 40 %. At section HV-3, the numerically computed lateral displacement around DL - 5 m in tlie direction toward the sea is larger than tlie observed value. However, except for this area, the numerical results and tlie observed lateral displacement showed a similar tendency. Figure 10 shows tlie comparison between observed and numerically calculated horizontal displacement at tlie ibundation (DL - 2.5 in) of the 2.0 in lighter’s wharf, from the start of construction of tlie artificial island. The horizontal displacement at the lighter’s wharf increased in the direction toward the sea, during the reclamation with the cement mixed soil and the construction o f the embankment. Compared to the numerical results, tlie observed horizontal displacement showed a larger increase after around 400 to 600 elapsed days. This may be because the areas with the deep cement mixed soil, rubble mouiid, and concrete blocks were assumed to be elastic bodies.
Figure 10. Coinparison between obseived arid nuinerically coinputed lateral displacement at -2.01n lighter’s wharf
The cases with and without tlie sand drain were analyzed. The construction process in the analysis is almost identical to tlie on-site construction, shown iii Table 1. Figure 7 shows tlie numerically obtained time history of the surface settlement at the center of embankment. It is clear from this figure that with the sand drain, tlie settlement took place at an accelerated pace at the initial stage, and tlie final settlement was slightly larger although tlie difference was insignificant. 6 COMPARISON BETWEEN OBSERVED AND ANALYZED RESULTS 8 shows the compar~sollbetween observed and numerical results of the surface settlement. from
~i~~~~
Figure 1 1. Comparison between observed arid numerically colnputed layer
349
Table 3. Suininaiy of material properties used in analysis Materials
Cement mixed soil
Young’s modulus E (MN/m2) 40.7 ratio I/ 0.3 Coefficient of permeability k (m/day) 6.25 x 10-* Unit weight r(kN/m3) 12.6 Unit weight in water y (kN/m3) 2.8
Fill 40.7 0.3 6.25 x 10-‘ 12.6 2.8
,Yznt
mixed soil
490 0.3 8.64 x 13 0 3.2
Table 4. Detenniiiation of the equivalent coefficient of penneability in the sand drain area Items Notation Length o f sand drain pile (cm) H Diameter o f sand drain pile (cm) dw Central distance of sand drain piles (cm) d Diameter of equivalent effective circle (cm) d, Coefficient o f permeability of sand drain piles (cm/s) Coefficient of permeability o f clay(cm/s)
k,
Coefficient of well resistance Ratio o f diameters d , / d w
L
F=[n2/(n2-1 ) 1 1 n ~ - ( 3 ~)/(4n2) ~-1 Time factor a t 50% consolidation
n
F
T,,
Value 1900 40 180 203
1.13 x
Notes
Experiment
0.0827 5.075 0.95 0.0881
Rubble mound
19.6 0.3 8.64 x 17.6 7.8
Concrete block
Sandy gravel
Back filling
Sand mat
2.65 x 104 0.3
19.6 0.3 8.64 15.9 6.1
19.6 0.3 8.64 x 10” 17.6 9.8
19.6 0.3 8.64 x 10 17.6 7.8
22.5 12.7
3. The numerically calculated lateral deformation occurred in the direction toward the sea as was the case for the observed lateral deformation. 4. The numerically computed lateral deformation at the DL - 2.0 in lighter’s wharf was sinaller than tlie observed lateral deformation when cement mixed soil was applied in the layer DL 0 ni to DL + 2.0 m. Except for this, both observed and coinputed lateral deformations showed a similar tendency. ACKNOWLEDGMENTS:
Figures 11(a) and 1 l(b) show observed and numerically computed time history of layer settlement, respectively, obtained at section S-1 at the ceiiter of the embankment, from just before loading the embankment (tlie 1015‘” day in Table 1). The depths at which measurements of each layer settlement were made are shown in Figure 5 . The numerical results were in general smaller than the observed layer settlements. Tlie amount of numerically calculated settlement of each layer was smaller than that of observed settlement at depths of DL - 7.0 ni or deeper. It is clear from Figure 9 and Figure 1l(b) that the iiumerical results in comparison to the observations, showed a tendency to have greater vertical and horizontal deformations around DL 1.O ni to DL - 7.0 m.
7 CONCLUSIONS Tlie authors carried out field observation and numerical analysis on a test embankment and on an adjacent lighter’s wharf on an artificial island coiistructed on soft ground. The following conclusions were obtained. I . Tlie surface settlement took place at a more rapid pace in tlie initial stages and the filial settlement became slightly larger in the case with the sand drain compared to the case without the sand drain. 2. The numerically computed layer settlements aiid the ground settlement were smaller than their observed values. The amount of numerically calculated settlement of each layer was sinaller than that of observed settlement at depths of DL - 7.0 m or deeper.
The authors would like to express their gratitude to the staff inembers at Shiogama Fishing Port Office in Miyagi Prefecture aiid Pacific Consultant, Co. Ltd. for their considerable assistance in tlie field observation for this study. REFERENCES A society for the study of CDM 1993. l’lie ninnrtal,for tlesigii and construciion qf ceiiieiii deep iiiixiiig niethod 80-87 (in Japanese) Mimura, M., Shibata, S. & Ohinaki, S. (1991) Numerical investigation on the deforination of a marine foundation subjected to caisson-composite breakwater construction. PI’OC.C ‘oiiipiiier A.Tc.thodr and Atlvances iii C;eoniecliaiiic.~: 765-770. Ohmaki, S., Kanno, N. & Ikeda, T. (1989) Monitoring of composite breakwater during construction, TSUC‘HI-TOKISO JSSMFE, 37 (8): 27-32 (in Japanese). Oka, F., Yashima, A., Miura, K., Olunaki, S. R: Kamata. A. (1995) Settlement of breakwater on submarine soil due to wave-induced liquefaction, I’lac. 5”’/iileri7nliorinl C?f,sliore arid Polar Eiigiiieeriiig C ‘ O I ? ~ : 482-487. Sekiguchi, H. & Ohta. H. (1977) Induced anisotropy and time dependency in clay, I’roc. 9‘” IC’S’ILfFI!, Specialty Session 9: 229-237. Shibata, T. & Sekiguchi, H. (1980) A method of predicting failure of embankment foundation based on elasto-visco plastic analyses, Proc. JSC‘E, 30 1: 93-1 04 (in Japanese) t~o~i qf Yoshikuni, H. ( 1979) Desigii and c o i i . ~ i r ~ ~ cIiiniiageiiieiii the vertical drain ~~ieiliotl, Gihodo shuppan LTD: 40-49 (in Japanese).
350
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Effects of some parameters on braced-excavation of soft clay by numerical studies T. Pipatpongsa & H.Ohta Tokyo Institute of Technology,Japan
A. Iizuka Kobe University,Japan
M. Hashimoto Hokukokuchisui Company Limited, Ishikawa, Japan
ABSTRACT: Apart from following general outlines for input procedures and parameter findings when performing excavation analysis using FEM, some additional considerations should be taken. The influences of appropriate soil stiffness, pre-load efficiency, actual strut stiffness, soil/wall interface, crack in concrete retaining wall, and foundation piles installed before excavation have been investigated to weight the impact on analysis quality, enhancing acceptability in practice. Computed deformations and strut axial loads of a case study agreed quite well with field responses after applying these effects.
1 INTRODUCTION Accuracy of finite element analysis of excavation depends generally on the performance of soil constitutive model incorporated with its required parameters, rational idealization of soil/structure and construction processes involved, as well as how well is workmanship quality during construction. However, there are some particular points that should be considered in addition: (A) Deformation properties for elasto-plastic constitutive models are regularly based on consolidationhriaxial tests, giving virgin compression curves and sweIIing curves where stresses have been mobilized under drained condition. It is, therefore, arguable to employ these properties for handling excavation analysis of soft clay, in which stresses, on the other hand, are on unloading path under relatively undrained condition. (B) There was a vague understanding that the deformation experienced on the ground surrounded by the structure under working stress condition is controlled by properties of soil at small strain, but it was not until Burland, Simpson & St John (1979), Jardine et a1 (1985), Atkinson & Sallfors (1991) showed the stress-strain relation for the range of small strain has a significant effect on the numerical analysis and soil stiffness determination. Mair (1993) & Vaziri (1996) emphasized higher stiffness of soil at relevant low-level strains to the excavation problem should be practiced instead of value obtained from conventional laboratory. (C) No structural idealization is more difficult to cope with than bracing system and pre-load exertion in numerical modeling effort. It is not easily, in fact,
to account for amount of pre-load loss after a few days, load-transferring efficiency during exertion and actual strut stiffness in the field due to work quality and construction limitation. (D) For concrete-built retaining walls, the effective moment of inertia varies with the degree of crack undermining the sectional stiffness. The iterative procedures to modify flexural stiffness are then required, leading to more proper agreement with wall curvature and moment redistribution (Honda 1986). (E) Foundation piles installed prior to excavation can affect the wall movement. As pointed by Balasubramaniam et a1 (1994), the effect of foundation piles not only reduce the magnitude of displacement, but also change the pattern of deflection profile. Without the foundation piles, maximum lateral movements occur at the formation level, but the foundation piles shift the maximum deflection in the upper locations. (F) Having everything to do with the relative movements between soil/wall contact, interface element should model the material discontinuity at both media, allowing totally different systems assembling in the same analysis. To illustrate the above-mentioned importance and to provide parametric assessment pragmatically and economically in numerical analyses using FEM, parametric studies were therefore carried out to examine the impacts of individual effect both on wall deflections and axial forces in struts. DACSAR (Iizuka & Ohta 1987), formulated by using nonlinear incremental elasto-viscoplastic model proposed by Sekiguchi & Ohta (1977), was used to analyze a selected case study. 35 1
Figure 1. Schematization of in-situ c-log U’, curve under loading/unloading conditions, composed of recompression, virgin compression and swelling curves. Preconsolidation pressure, a’,,,is located at the end of recompression curve, indicating Y3 boundary.
Subtly enriched by natural structurization, recompression curve in state boundary surface, associated with structured soil, is superior to evaluate settlehebound ground surface than subsequent swelling curve at higher stress, which rather associated with destructured soil. In excavation, the relevant strain is far too small to mobilize the state boundary surface while water content is unlikely to change from the initial condition. The unloading behavior is, therefore, characterized by stiffness immediate to the initial state, achieved on recompression curve. Deformation moduli employed in Sekiguchi-Ohta model are obtained using triaxial test. However, the sufficiently acceptable results can be economically obtained using oedometer test. On the premise that undisturbed samples can be prepared for oedometer test and compression characteristics satisfy with those appear in the field, stiffness obtained on recompression curve by gradual loading, where O ’ V j is located, is encouraging to apply in excavation analysis.
2.2 Pre-load efficiency
2 INFLUENTIAL PARAMETERS 2.3 Soil stiffness In conventional consolidation test, the maximum stress at the end of recompression curve determines OCR, referring to Y3 sub-yield boundary in stiffness degradation curve defined by Jardine et a1 (1985) where relatively large strains are involved. For large strains, the behavior is elasto-plastic and can be described by Cam-clay-type constitutive models. In the intermediate and small strains, range of stiffness increases rapidly and cannot be characterized by the same counterpart (Atkinson & Sallfors 1991), besides implying soil parameters obtained from conventional laboratory are not applicable to this class of problem. The excavation-induced strains appear to limit between Y2-Y3 boundaries or at 0.01-1% (Mair 1993) and hence classifying the completely different classes for parameter finding based on conventional/special laboratory. Degree of strains is found corresponding to structural type (rigidhon-rigid), excavation method (top-down/bottom-up), embedment length, soil and boundary conditions. Capable in stress-induced anisotropic plasticity, efficient in time-dependent creep, Sekiguchi-Ohta model (anisotropic Cam-clay model) still has its weakness in handling stiffness degradation within Y3 involving massive loadinghnloading rules. So, this study prefers to set aside the small strain problem for future development, by instead, sustains on the scope of relatively large strain to Y3 where conventional laboratory is still employable. Figure 1 illustrates typical loadinglunloading behavior in the field for natural soils.
Pre-load efficiency including loss effect cannot be predicted numerically. Part of pre-load is resisted by connection-induced friction on king posts. Only about 50-60% of pre-load can be transferred to the soil behind the wall as reported by Finno et a1 (1991) & Ou et a1 (1996). Undetermined loss often emerged in a few days after exertion. Moreover, additional 10-40% of strut load can be fluctuated by ambient temperature observed by Fernandes (1985). For design purpose, it is recommended to consider benefit of pre-load just to remove slack from strut members. 2.3 Actual strut stiffness Due to poor member connection, the strut stiffness is observed to vary 10-60% of nominal value according to many reports (Fernandes 1985, Hata & Ohta et a1 1985, Ou et a1 1996). Mana & Clough (1981) concluded that it is difficult in knowing the actual strut stiffness and pre-loading. 2.4 Interface element Since cast-in-situ walls usually cause sufficiently rough surfaces between soils and retaining walls. No fully mobilized slip is assumed to occur at the soil/wall interface (Ou et a1 1994 & Charles et a1 1995). Soil elements locating next to walls are connected to thin layer of an elastic interface element which thickness is 3/20 of adjacent element. Shear modulus is assumed to be similar to of adjoining soil elements while bulk modulus is 10 times, proficiently ensuring only shear strain can take place between both media. Theoretically, effect of bulk modulus towards deformation is trivial in undrained condition - the greater bulk modulus develops, the
352
more Poisson’s ratio will approach 0.5. The introduction of interface element results in a proper estimation of differential movements and stress distribution when excavation gradually becomes deeper than allowing the soil directly contact with this discontinuity.
2.5 Flexurnl stiffness As long as there is no cracks (M
Figure 8. Result for a case of no interface element applied Figure 9. Result of no pile foundation in analysis
Figure 4. Results by varying irreversibility ratio Figure 5. Result when employing Karube’s correlation
where I,, , Icr = moment of inertia of the uncracked and cracked transformed section; Mcr = cracking moment; M , = maximum moment of section; f r = modulus of rupture; y , = distance from neutral axis to tension face.
353
Table 1. Primitive and modified factors Parameter
primitive
Soil stiffness* Pre-load efficiency*" Strut stiffness Flexural stiffness Interface element Foundation pile
A=l-CJCc 100% 100% 100% none none
modified
A=l-Cr/Cc 30%, 60%, 90% 40% 27%-45% apply apply * View in significance of irreversibilit ratio * * Field observation from strut 15t,2"Y and 3rd
Table 2. Comparison of irreversibility ratio obtained by various methods and references _______
Irreversibility ratio
crust
soft
medium
stiff
This study (1-CJC,)" 0.63 0.88 0.88 0.73 This study (l-Cr/Cc)* 0.92 0.90 0.91 0.92 Typical value ( l - ~ i h ) " " 0.76 0.76 0.76 Correlation (l-RR/CR)""" 0.92 0.88 0.90 0.93 Karube, 1975 (Mh.75) 0.87 0.55 0.52 0.54 Honda, 1986(Back-analysis) ----------- 0.90-0.92 --------------
* Obtained by oedometer tests Sutabutr, 1992 and Long, 1995 MRT alignment (Jeanjcw et a1 1997)
I:*
*'"'* W,-correlation from
Table 3. Quantitative impact of each effect on calculated result in terms of ratio of displacement area behind retaining wall Effect
Area ratio (A'IA) when thc effect is disregarded
AI I 1.38 Soil stiffness Prc-load efficiency Foundation pile 1.22 Flexural stiffness 0.92 Strut stiffness 0.97 Interface element 0.98 ___________ __ A: Displacement area calculated by considering all effects A'. Displacement area calculated by ignoring conccincd effect
2.6 Foundation pile and Barrette pile The stiffness of the pile element is obtained by equivalent method (Lee 1989). Finno et a1 (1991) commented that the response of the soil on the excavated side controls behavior, following that the improvement of the soil by piling and soil stabilization in the passive side is more effective than that in the active side (Ou et a1 1996). As a result, only the effect of foundations existed in the excavation zone is considered in analysis regardless of any adjacent foundations outside excavation zone.
3 CASESTUDY
Cast-in situ bored piles were arranged from level 8.6 to -57.0 m in excavation zone to support project building. The 7-staged bottom-up excavations, including strut installatiQn and pre-loading, were carried out down to -9.50 m depth. The inclinometer readings of this site were lack of reference datum points, therefore, only the differential horizontal movements relative to the bottom of supported wall can be obtained from measurements. The calculated wall deflection profile subtracted by computed lateral displacement at the bottom of the wall is the shifted curve that is used to compare with field measurements. To illustrate the impacts of concerned parameters shown in Table 1, the primitive analysis, is to compare with a modified one. The calculated results (shown as DACSAR) and their shifted profiles (shown as SHIFTED) are plotted with inclinometer readings (IC-9 and IC-10) as shown in Figure 2 and Figure 3. The large discrepancy is found when ignoring parameter modification. The study addresses the problematic soil stiffness associated with recompression characteristics in which the irreversibility ratio, A, seems to coincide with back-analysis result what was proposed by Honda (1986). By varying A as shown in Table 2, the results are given in Figure 43. The efficiency for transferring pre-loading force at exertion time is found 20-4076 regarding to trial analyses shown in Figure 6. Figure 7 shows effect of cracked/uncracked retaining wall. The cracked wall seems to be curved along well with the measured. Figure 8 shows the effect of discontinuities when curvature pattern is changed and Figure 11 shows effect of pile foundation in reducing lateral movement. Still, abrupt loss of pressure in a few days is undeteirninable. In concerning with field measureinents, most of loss occurred in the first layer. The remained pre-load learned from the measurements for the l", 2"d and 3rd strut layer are 30%, 60% and 90% respectively. By accounting on losses found, the series of trial analyses are performed to capture the efficient factor for strut stiffness (y) what is met at about 40% in accordance with Figure 10. Table 3 concludes the quantitative impact of concerned parameters on the calculated retaining wall displacement for the final stage of the case study. It is found that soil stiffness, pre-load efficiency and foundation pile are the main factors for accuracy. But flexural stiffness, actual strut stiffness and interface element modeling can be disregarded in preliminary analysis without sigriificant difference.
B-area, approximately 45x45 m2, of Sukhumvit City project located in Bangkok area was used. The retaining structure was secant pile wall, diameter 0.90 m, 20-m depth cast-in situ of primary and secondary pilc (15 and 24 Mpa grade f',), horizontally braced by 2xW35Ox35Ox [email protected] and vertically supported by 9 m-spacing secondary strip secant pilc.
4 VERIFICATION OF ANALYSIS 4.1 Exurnination of shear strain level Providing that strains emerge not less than 0.1-1% order, the soil stiffness determined by proposed mcthod is out of harm away. The excavation354
induced shear strain obtained from analysis should be verified to vouch the justification postulated. According to shear strain profiles calculated at the first and final stage of excavation (see Figure l l ) , most of soils along the retaining wall have thoroughly experienced Y3 sub-yield boundary as counted upon. But those of locating near the end tip which orders stay close to Y2 (about 0.01%), prompting to higher stiffness. However, the movement of the end tip is comparatively small so that the upturn in stiffness can be disregarded without causing significant difference, maintaining the method simplicity with applicable accuracy.
4.2 Elastic model comparison Clough and Mana (1976) concluded that if appropriate soil parameters are chosen, the type of soil model does not significantly affect predicted behavior, leaving a doubt whether a prediction by an extensively used elastic model is consistent with elastoplastic model. As the analytical results show the relatively undrained condition, p, along the wall are rather unchanged (Figure 13), the comparatively small movement during excavation is unlikely to mobilize whole soil from elastic state given by formerly initiated OCR. Figure 12 shows the comparison of analysis between both models under the same conditions. The normalized undrained soil stiffness modulus, E,/&, can be derived from Sekiguchi-Ohta model as written in Equation (4)-(6). 9(1- 2 ~ ' ) ( 1 + " , ) e x ~- A- (M - % ) M ~ ( 1 v')M ~
where = 0.434C,; (SJidea/= half of UC strength of undisturbed young clay; p = full correction factor (Ohta et a1 1985, 1989) Table 4. Normalized soil stiffness modulus at level of shear strain 0.1-17i for Bangkok clays. E"/S" crust soft medium stiff This study 230 120 220 400 Tcparaksa (19YYa, 13) 150 250 1000" Pressuremeter tests (0.1%) - 300-1500 370-1600"" (MRT project, BKK) (1%) 150-750 220-820"" Duncan correlation (1976) 400 300 450 600""" Shibuya, Tenma, Theramast (0.1%) --- 360-600 --- 800-1500 RrYarnamoto (1997-1999) (1%) --- 120-210 --- 200-400 Back-analysis on 4 sites with full instrumentation. Six self-boring pressuremeter tests. """ Initial stiffness correlated by using P I and OCR. xx
Figure 11. Calculated shear strain profiles of soil adjacent to the retaining wall. Figure 12. Comparison between prediction by elastic and Sekiguchi-Ohta models under the same conditions.
4.3 Review of normalized soil stiffness modulus
The extensive researches on soil stiffness at low strain level in Bangkok clays have been carried out by Shibuya and his colleagues (Shibuya et a1 1999) at AIT using special and multi-functional triaxial apparatus for extension and compression test, bender element laboratory and seismic cone penetration. 355
Figure 13. Calculated stress paths along front and back walls throughout all excavation stages.
Furthermore, pressuremeter tests were conducted along the MRT route by Cambridge In-situ (1997). E,& obtained at the threshold of shear strain 1% is consistent with stiffness determined by using Equation (4)-(6). The comparative values of E,& available by various methods are listed below.
5 CONCLUSION Though the method applied in the study is limited to relatively large strain problems, by all accounts, affords guidance, for rational analysis by eyeing influential effects suggested. The parametric study reveals soil stiffness is the key role for accuracy of lateral wall movement prediction while quality of axial force estimation in strut depends on details of pre-load efficiency, pre-load loss and actual strut stiffness. The performance of elastic model is enough to employ in the realm of excavation analysis when subsoil are still immobilized by excessive movement and soil stiffness at level strains concerned is well provided.
REFERENCES ACI 1992. Building code requirements for reinforced concrete (ACI 318) and commentary-AC1318R-89 (Revised 1992) Atkinson, J.H. & Sallfors, G. 1991. Experimental determination of stress-strain-time characteristics in laboratory and in situ tests. X ECSMFE (111): 915-956, Florence. Balasubramaniam, A.S.; Bergado, D.T., Chai, J.C., Sutabutr J.C. 1994. Deformation analysis of deep excavations in Bangkok subsoils, XZII ICSMFE, 1994, New Delhi. Burland: J.B., Simpson, B., 61 St John, H.D. 1979. Movements around excavations in London Clay. In Design Parameters in Geotechnical Engineering. Proceedings, 7'" Europeun Cotifetwe Soil Mechutiics ( I ) : 13-30, Brighton. Cambridge In-situ of Little Eversdcn 1997. Self-boring pressuremeter MRTA initial system project, North Contract, Bangkok.
Clough, G.W. & Mana, A.I. 1976. Lessons learned in finite element analyses of temporary excavations in soft clay, ICNMG: 496-510, Virginia. Duncan, J.M. & Buchigani, A.L. 1976. An engineering manual for settlement studies, Geotechnical Report of Civil engineering department, University of California at Berkeley. Fernandes, M.M. 1985. Performance and analysis of a deep excavation in Lisbon, X I ICSMFE (4): 2073-2078, Rotterdam: Balkema. Finno, R.J., Lawrence S.A., Allawh, N.F. & Harahap I.S. 1991. Analysis of performance of pile groups adjacent to deep excavation, JSMFD 117(6): 934-955, ASCE. Hata, S . , Ohta H., Yoshida, S., Kitamura, H. & Honda, H. 1985. A deep excavation in soft clay performance of an anchored diaphragm wall, VICNMG: 725-730, Nagoya. Honda, T. 1986. The fundamental study on application of elasto-plastic finite element analysis to the construction control of retaining wall, Dr.Etig.Thesis, Kyoto University Iizuka, A. & Ohta, H. 1987. A determination of procedure of input parameters in elasto-viscoplastic finite element analysis, Soils and Foundations 27(3): 78-87. Jardine, R.J., Fourie, A., Maswoswwe, J., & Buralnd, J.B. 1985. Field and laboratory measurements of soil stiffness, XIICSMFE (2): 551-514, San Francisco. Jeanjew, M., Vorasithsaet, M., Sookanan, N. & Teparaksa, W. 1997. Engineering soil properties along the MRT alignment, B.Eng. Chulalongkorn U., Bangkok. Lee, Y.H. 1983. Behavior of Embankments, Excavations, Piled Foundation in Soft Bangkok Clay, M.Eng. GT 82-2, AIT. Mair, R.J. 1993. Developements in geotechnica engineering research: application to tunnels and deep excavations. Proceedings, Institution of Civil Engineers and Civil Etigineering, Paper 10070: 27-41. Mana, A.I. & Clough, G.W. 1981. Prediction of movements for braced cut in clay, JSMFD (107): 759-777, ASCE. Ohta, H. & Nishihara, A. 1985. Anisotropy of undrained shear strength of clays under axi-symmetric loading conditions, Soils and Foundations 25(2): 73-86. Ohta, H., Nishihara, A., Iizuka, A., Morita, Y., Fukagawa, R., Arai, K. 1989. Unconfined compression strength of soft aged clays; XI1 ICSMFE (1): 71-74, Rio de Janeiro. Ou, C.Y., Wu. T.S., Hshieh, H.S. 1996. Analysis of deep excavation with column type of ground improvement in soft clay, JSMFD 122(9): 709-716, ASCE. Ou, Chang-Yu & Lai, Ching-Her 1994. Finite-element analysis of deep excavation in layered sandy and clayey soil deposits, Can.Geotech. J. (31): 204-214 Sekiguchi, H. & Ohta, H. 1977. Induced anisotropy and time dcpendency in clays, Proceedings of specialty session 9, IX ICSMFE, Tokyo, Constitutive equations of Soils: 229-238. Shibuya, S., Mitachi, T., Tanaka, H., Kawaguchi, T. & Lee. I. 1999. Measurement and application of Quasi-elastic properties in Geotechnical site characterization, Theme lecture for plenary discussion session 1, Proc. of 21'" Asiuti Regional Conference on Soil Mechanics atid Geotecliriicul Engineering, Seoul, Korea, August 16-20, 1999. Teparaksa, W. 1999a. Geotechnical aspects on the design and construction of the MRTA subway in Bangkok, Lecture uf British Club, Society of Professional Engineers, Bangkok. Tcparaksa, W., Thasnanipan, N.& Pornpot, T. 1999b, Analysis of lateral wall movement for deep braced excavation in Bangkok subsoils, AIT 40"' Antiiversury, Civil utid Environmentul Engineering Conference, Nov. 1999, AIT. Vaziri, H.H. 1996. Numerical study of parameters influencins the response of flexible retaining walls, Cutz.Geotec1i.J. (33): 290-308.
Charles, W.W. & Lings, M.L. 1995. Effects o f modeiiitg soil nonlinearity and wall installation on back-analysis of deep excavation in stiff clay, JSMFD 121(10): 657-695, ASCE.
356
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Modeling of the behaviour of sand drains installed at a Naval Dockyard, Thai1and I.W. Redana Department of Civil Engineering, Udayana University,Denpasar, Indonesia
B. Indraratna & W. S a l h Department of Civil, Mining and Environmental Engineering, University of Wollongong,A! S. W ,Australia
A. S. Balasubramaniam School of Civil Engineering, Asian Institute of Technology,Bangkok, Thailand
ABSTRACT: This study describes the modeling of sand drains incorporating smear and well resistance in a 2D plane strain finite element model employing the modified Cam-clay theory. In the analytical model, the sand drains system was converted into equivalent parallel drain walls by converting the coefficient of permeability of the soil. The transformed permeability coefficient was then incorporated in the finite element code, CRISP. The numerical results indicate that the inclusion of the effects of both smear and well resistance of the vertical drains improves the accuracy of predictions of the settlements, excess pore water pressures and lateral displacements. 1 INTRODUCTION Although settlements can be predicted reasonably well (Indraratna et al., 1994, 1997), the lateral deformation of embankments built on soft clay stabilised with vertical sand drains is difficult to predict accurately, even after the progress has been made in the past few years through rigorous numerical modeling. The classical solution for vertical drains (Barron, 1948; Hansbo, 198 l), can be used in the prediction of settlement along the embankment centerline. Due to the increasing popularity of the finite element method in geotechnical software, a plane strain model is employed in the current numerical analysis. Based on the work introduced by Hird et al., 1992, Indraratna and Redana (1997) extended the analysis based on the plane strain solution to include explicitly, the effects of smear and well resistance. Multi-drain analysis is pertinent to study the overall behaviour of clay foundation underneath the embankment, even though single-drain analysis is usually sufficient to model the soil behaviour along the embankment centerline. Limited case studies employing multi-drain analysis have been described in the past (Chai et al., 1995; Indraratna et al., 1997; Indraratna and Redana, 2000). In this study, a multisandwick drain analysis is conducted where the smear zone is explicitly defined on either side of the sandwick drain element. For a given drain length, the
effect of well resistance is incorporated by specifying an appropriate discharge capacity.
2 EQUIVALENT PLANE STRAIN MODEL Indraratna and Redana (1997) showed that the degree of consolidation (plane strain) can be represented by:
By explicitly defining the smear zone,
For the dimensions B, b,Tand b, defied in Fig.1, the geometric parameters a! , p and flow term 13are given by:
357
drain
The half width of drain b,, and half width of smear zone b, may be taken to be the same as their radii in axisymnnetric r,,, and r,, respectively, which gives:
b,,, = r,,, and b, = r,
smear
zone
1
(3)
In this model, at each time step, and at a given stress level, the average degree of consolidation for both axisyrnmetric ( o h ) and equivalent plane strain ( u h p ) conditions are made equal. If the radius of the axisymmetric influence zone around a single drain ( R ) is taken to be the same as the width ( B ) in plane strain (Fig.l), then the converted plane strain equation is given by:
D
a) Axisymmetric
2B
b) Plane strain
Figure 1. Conversion of an axisymmetric unit cell into plane strain (Indraratna & Redana, 1997).
3 SANDWICK DRAINS AT A NAVAL, DOCKYARD, THAILAND where, n = WrW, s = rs/rw, k, and kk are the horizontal permeabfity in the undisturbed and disturbed soil, k,, and k;, are the corresponding values in plane strain model, I is the length of drain, qwis the speclfic discharge capacity of the drain and z is the vertical depth considered.
3.1 Sub-soil and embankment conditions
For no well resistance, the permeability in the smear zone can be given by the following expression: kl:,
R
f.cj
Ignoring both the smear and well resistance, the simplified ratio of plane strain to axisymmetric horizontal permeability, k,! would be given by:
The effect of well resistance is considered in the analysis by choosing an appropriate discharge capacity (q,J of the drain. The well resistance is modelled independently, and the equivalent plane strain discharge capacity of drains (yz) is given by: 2 4, =- 4," ZB
(7)
358
This case history looks at the performance o l sandwick drains at a Naval Dockyard test embankment, Pom Prachul, Thailand. The test site is located in the Samutprakarn Province, approximately 20 km south of Bangkok, along the Chao Phraya river. Three test embankments T l , T2 and T3 (no drains) were constructed and stabilised with sandwick drains installed in a square pattern to a depth of 17 m. Prior to the construction of the embankment, eight boreholes were drdled for soil identification and laboratory tests. Specimens for oedometer and triaxial tests were obtained using 10 inch diameter tube samplers. The subsoil is relatively uniform, consisting of a thin weathered clay (0.75 m deep) overlying a soft clay approximately 17 m thick. A stiff clay layer underlies the soft clay and extends to a depth of 25 m below the ground surface. The soil retains a very high moisture content of about '75%) near the surhce to a depth of 10 m. The moisture content decreases with depth, and is about 40 %) at 20 m depth. The unit weight of the soil varies between 15 to 18 kN/m3 from the ground surface to a depth of 20 m. The field vane strength increases with depth, except at the surface where the weathered clay gives a locally increased shear strength of about 20 kN/m'. The Cam-clay parameters for each soil layer and the in-situ stress states are given in Fig. 2. In order to estimate the undisturbed soil permeability, laboratory consolidation tests were conducted on both vertically and horizontally cored specimens. The horizontal and
vertical permeability coefficients of the undisturbed soil ( k h and kv) are given in Fig. 2 and the equivalent plane strain values were estimated using Eqs. ( 5 ) and (6). The estimated values of vertical permeability (k,,) at this site given by Brenner and Prebaharan (1983) varies between 10-l' d s e c at the bottom of the stratum and 2 x 10'9 d s e c at the top. However, these values are lower than the permeability estimated by the writers as shown in Fig.2. The coefficient of horizontal permeability was taken to be 1.8 times the vertical permeability. Inside smear zone, the horizontal permeability was taken to be 1.15 times the vertical permeabllity based on current laboratory study. The drain pattern and the typical cross section of the embankment at this location are shown in Fig. 3. The sandwick drains consist of a hose made of a fibrous material of high permeability, and f i e d with dry sand. The sandwick drains were installed to a depth of 17 m to reach the stiff clay layer. The drains were installed using a 7.5 cm diameter casing with a wooden plug at its lower end and pushed into the soft ground using a 2 tonne hammer. The drains were installed in a square grid pattern at 1.5 m and 2.5 m spacing for the two embankment sections, T1 and T2, respectively. The instrumentation for monitoring settlements of these test embankments included both suri'ace and sub-surface settlement plates. Surface settlement plates consisted of a 16 mm diameter steel rod connected to a 0.40 x 0.40 m2 base plate and protected by a 19 mm diameter casing. The subsurface type consisted of a 25 cm diameter steel screw head connected to a 19 mrn diameter steel tube which was attached to a screw head by a keyway. Pore water pressures were monitored using three types of piezometers including open stand pipe, closed hydraulic and air (pneumatic) systems. Lateral deformations were monitored using inclmometers which were installed close to the embankmcnt toe. Figure 4 indicates the rate of loading (construction history) of the embankment. The embankment loading was applied in three stages. Firstly, a sand blanket of thickness 0.35 ni was placed, followed by an initial layer of i-3.l of 1.10 m, which was then raised until a total fill height of 2.35 m was attained (Fig. 4). The loading scheme of the unstabilised section of the embankment (T3: without drains) was almost the same, therefore, the performance of the embankment with and without drains could be directly compared.
Figure 3 . Cross section of embankment with drain pattern installed at Naval Dockyard, Thailand. I
60 I
501
a'
E m b a n k m a t T1 E m b a n k m e n t 1'2
50
'
(
100
'
I
150
'
(
200
'
250
Time (days)
Figure 4. Construction loading history.
3.2 Numerical analysis a i d results The finite element mesh of the embankment for multi-drain analysis is shown in Fig. 5. The foundation was discretized into linear strain quadrilateral (LSQ) elements. For the zone with sandwicks, a finer mesh was developed so that each drain element represents the sandwick containing the smear zone on either side of the drains. The locations of instrumentation such as inclinometer and 359
piezometer to monitor the performance of the embankment are placed in the mesh in such a way that their measuring points coincide with the mesh nodes. For instance, the piezometer is placed in between two drains at 0.75 m from the embankment centerline to measure the pore pressure (Fig. 5). The clay layer is characterized by drained conditions at the upper boundary only, where as lower boundary is impermeable due to the presence of a stiff clay layer below 17 m depth. In axisymmetric condition, the equivalent radius of sandwick drains and smear zone are I-,,,=0.05 m and r, = 0.3 m, respectively. For the plane strain analysis, the width of drains and smear zone were taken to be the same as their radii in axisymmetric condition, which give b, = 0.05 m and b, = 0.3 m. In the analysis incorporating both smear and well resistance, after a few trials based on single drain analysis, it was found that a discharge capacity (4,) of 50 m3/year for each drain was appropriate for numerical modeling. This discharge capacity is in the range of q,, of 25-100 m3/year as reported by Holtz et al. (1991) for drain affected by significant vertical and lateral pressures. The results of the plane strain and multi-drain analysis together with the measured settlements are plotted in Fig. 6. The analysis based on perfect drain conditions (no smear, complete pore pressure dissipation) overpredicts the settlement. The inclusion of smear effect decreases the settlements, hence improves the accuracy of the predictions. The inclusion of both smear and well resistance slightly underestimates the settlements, especially beyond 150 days. At 5 m depth where the applied stress is smaller, the effect of well resistance is hardly noticed (Fig. 7), in comparison with surface settlements (Fig. 6). Piezonrtrr T
Ir
Y
lOOm
U
Fig. 5. Finite element mesh of embankment for plane strain analysis at Naval Dockyard, Thailand.
Figure 7. Total settlements at 5 m depth below ground level along embankment centerline, Naval dockyard.
360
Figure 9. Variation of excess pore water pressure at embankment centerline, Naval dockyard.
The prediction of settlement along the ground surface from the centerlme of embankment is shown in Fig. 8. The available measured data agree well with the settlement profile, near the embankment
centerline. Heave is also predicted beyond the toe of the embankment, i.e. at about 25 m away from the centerline. The predicted and measured excess pore water pressures for the piezometer located at the embankment centerline at a depth of 7.5 m arc compared in Fig. 9. Although the inclusion of smear effect can reasonably predict the excess pore pressures up to Stage 2 of construction, the prediction of Stage 3 and post-construction pore pressures is difficult. It is believed that ‘clogging or partial clogging of drains’ is the main factor causing this retarded rate of dissipation in the long term, which is difficult to model in the numerical analysis. As shown, the inclusion of well resistance with the smear effect slightly improves the prediction. As expected, the perfect drain predictions underestimate the actual pore water pressures, for both embankments T 1 and T2. Pore pressures predicted by ‘no drains’ assumption gives the highest among all other predictions, although field data in the long term shows higher pore pressures than ‘no drains’ conditions, due to possible partial clogging of the drains. This is interesting to note that the perfect drain predictions in Fig. 9(a) and 9(b) produce very similar pore pressures. For perfect drains, 2.5 m spacing is more than sufficient. Increasing the no. of drains (i.e. reducing spacing) does not seem to reduce the pore pressure any further. Lateral deformation (predicted and measured) for the inclinometer installed at 10 m away from the centerline of the embankment is shown in Fig. 10. The lateral deformations at 230 days after loading are very well predicted when the effects of both smear and well resistance are included in the analysis. As shown in Fig. 10, the inclusion of smear effect alone underestimates the magnitude of lateral deformation. The perfect drain condition, as expected, yields the smallest lateral deformation. It is illustrated clearly in Fig. 10 that the inclusion of the drains are expected to reduce the lateral movement of the soft clay under embankment loading. The predicted lateral deformation for ‘no drains’ is also plotted for comparison. In T2, the difference in lateral displacements between ‘no drains’ situation and the field measurements is small, in comparison to T1. This also verifies that the drain spacing in T 2 (ie. 2.5 m) is too large to have much effect on both settlements and lateral displacements. In contrast, the smaller drain spacing of 1.5 m in T1 has signifkant effect on reducing lateral displacements.
361
REFEWNCES
Figure 10. Lateral displacement profiles at 10 m away from centerline of embankment, Naval dockyard.
4 CONCLUSIONS Based on the 2-D plane strain analysis and the comparison with the field measurements of test embankments stabilised with vertical sandwick drains at Naval dockyard, Bangkok, it can be concluded that the explicit m o d e h g of the effect of smear in the vicinity of the drains provides good prediction of settlement. Conventional analysis using 'perfect drains' overestimates the settlements, and underestimates the actual pore pressures. The inclusion of the effects of both smear and well resistance in the analysis improves the prediction of settlement and lateral deformation, but the contribution of the well resistance alone is small on the dissipation of excess pore water pressure. This indicates that the 17 m long drains do not suffer from well resistance to any significant extent, hence, it is the smear effect that predominantly affects the consolidation process in the vicinity of the drains.
362
Baron, R. A. 1948. Cansolidation of fine-grained soils by drain wells. Trans., ASCE, 113, 718-742. Brenner, R.P. and Prebaharan, N., (1983). Analysis of sandwick perforrnance in soft Bangkok clay. Proc. 8th European Conf. Soil Mech and Foundation Enggr., Helsinki, Vol. 1, pp. 579-586. Britto, A.M. and Gunn, M.J. 1987. Critical state soil rnechanics via finite elernents. Ellis Horwcnxl, Ltd., Chichester, England. Chai, J. C., Miura, N., Sakajo, S., and Bergado, D. 1995. Behaviour of vertical drain improved subsoil under embankment loading. J. Soil and Foundations, Japanese Geotechnical Society, 35 (4), 49-61, Hansbo, S. 1981. Consolidation of fine-grained soils by prefabricated drains. Proc. 10th ICSMFE, Stockholm, Sweden, 3, 671-682. Hird, C.C., Pyrah, I.C., and Russell, D. 1992. Finite eleiizenr rnodelling of vertical drains beneath eiizbankriients on .sofr ground. Geotechnique, 42 (3), 499-5 11. Holtz, R.D., Jamiolkowski, M., Lancellotta, R., and Pedroni, S., (1991). Prefabricated vertical drains: design uncl performance, CIRIA ground engineering report: ground irnprovenient. Butterworth-Heinemann Lid, UK, 131 p. Indraratna, B., Balasubramaniam, A. S., and Ratnayake, P. 1994. Perforrnunce of ernbunktnent stabilized with verticul drains on soft clay. J. Geotech. Eng., ASCE, 120 (2), 257273. Indraratna, B., Balasubramaniam, A. S., and Sivaneswaran, N.1991. Analysis of settlenzent and lateral defortnation o f soft clay foundation beneath two jidl-scule embanktizents. Int. J. for Numerical and Analytical Methods in Geomectianics, 21, 599-618. Indraratna, B. and Redana, I W. 1997. Plane struirz inodeling of m e u r effects associated wirh vertical drains. J. Geotech. Eng., ASCE, 123 (3,474-478 Indraratna, B. and Redana, I W. 1998. kiborutoty detertnination of smear zone due to vertical di-uin installation. J. Geotech. Eng., ASCE, 124 (2), 180-184. Indraratna, B. and Redana, I W. 2000. Numerical mtdeling of' vertical drains with smear and well resistance installed i n soft clay. J. Canadian Geotech. (in Press) hxner, J.J., Kraemer, S.R., and Smith, A.D. 1986. Prefabricated vertical drains, Vol. II: Si~tntnuiy of Research Report-Final Report. Federal H~ghway Administration, Report No. FHWA-RD-86/169, Washington D.C.
Coastal GeotechnicalEngineeringin Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Construction of vertical seawall - Prediction and performance Thiam-Soon Tan, Poh-Ling Leong & Kwet-Yew Yong Centrefor Soft Ground Engineering, National University of Singapore, Singapore
Ryuji Kamata TOA Corporation,Singapore OfJice,Singapore
John Wei, Keng-Chay Chua & Yan-Hui Loh Housing and Development Board, Singapore
ABSTRACT: Reclamation is a major construction activity in Singapore. In recent years, with greater affluence, marina development for residential purpose has become more common. This often entails the construction of a vertical seawall to form the outer boundary to facilitate the reclamation. In Singapore a frequently used technique is the use of a L-Block. However, there are concern about the movement and tilt of such relatively light blocks. An ongoing reclamation in the Northeastern part of Singapore makes use of nearly lOkm of such wall. As a requirement, prior to commencement of construction, a finite element study was carried out to analyze the possible movement and tilting of the block. Detailed instrumentation clusters were then provided in the contract, and these thus provide a good set of data to validate the accuracy of the design and also allow lessons to be learned fi-om the difference between the movement predicted and that measured. This paper is a report of the comparison between prediction and performance that was carried out at this site. This construction is still ongoing and thus the comparison exercise is not fully completed. proach is increasingly common in Singapore and has been adopted by a number of regulatory agencies. Thus, it is important that such an approach be also evaluated with hind-sight, that is comparison against actual field performance. Only through such field case study evaluation can adequate experience be developed to ensure reasonable design in the future for similar construction.
1 INTRODUCTION A major reclamation is currently ongoing in the Northeastern part of Singapore, as part of a major new town development in this area. A marina is planned for this area, and one of the requirements in this project is the construction of a long stretch of vertical seawall to provide containment for the reclamation. In this project, this entails the construction of over 5.6km of a vertical sea wall using L-shape concrete blocks. Such type of seawall has been used in a number of projects previously. An important consideration is the movement of the wall during and after construction as these are relatively light-weight walls. The layout of this reclamation project is shown in Fig. 1 and typical sections of L-Block design are shown in Fig. 2. A finite element study was required as part of the design. This study was conducted primary for the purpose of evaluating the movement and the associated tilting of the L-Block during construction. To a large degree, this study was also used as the basis for site control. In other words, if the field performance deviates significantly from the predicted values, the contractor will have to provide justifications for works to continue. It was therefore important that this study produces reasonable results. This ap-
2 SOILPROFILE At this site, the soil profile in general comprises an organic clay layer follows by a marine clay layer, usually referred to as the Upper Marine Clay layer. In some other locations, this is followed by a desiccated layer of clay and then another layer of marine clay known as the Lower Marine Clay layer. Underlying the clays is a stiff clayey silt and then a sedimentary rock. Such a profile is typical of many parts of Singapore (Chong et al., 1998). For this particular location, the soil investigation carried out revealed a soil profile as given in Fig. 3. An immediate, and probably most important, problem in any finite element study is the choice of appropriate soil models and the selection of relevant soil parameters. As the present study arises out of a construction project, the necessary soil investigation
363
was conducted in accordance with typical requirements of industry. In the present case, a number of boreholes were sunk. Standard Penetration Tests were conducted in all the boreholes. Undisturbed sampling was conducted at selected boreholes, especially those containing soft clay. For these samples, besides the usual odeometer tests, CK,UC and CK,UE, and Mikasa’s type direct shear tests were conducted. At some locations, these tests were complemented using cone penetration tests. As these are typical tests, and done by soil investigation companies, the soil parameters derived are thus partly based on the soil tests and partly based on local knowledge of these soils. Eventually, during the design, the models and associated soil parameters in Table 1 were chosen for the analysis. For design, a primary consideration in the selection is that these should represent safe parameters. The CAM-Clay model is chosen for the marine clay, mainly because this in-situ marine clay is known to be lightly overconsolidated. It is also expected to consolidate significantly during the reclamation. This clay is known to be relatively young and lightly structured, but only mildly anisotropic with shear strength values from UC and UE tests nearly similar (Chong et al., 1998), though for this particular site, the compression strength is about 15% higher than the extension strength. Thus it is expected that the CAM-Clay model will be able to model the deformation behavior reasonably. It is the behavior of this layer that will largely determine the behavior of the L-Block. For the other soils, an elastic plastic model is used, assuming a Mohr Coulomb failure criterion. For most of these soils, it is expected that the yield criterion is not a major factor, but the estimation of the stiffness is. This point will be revisited later on in the paper.
. is33
I
RLCWAION WD FILL
WLtJOO
MPL ‘A’ RiiRNlNG WALL RNtIMLNI
Fig. 2 Typical section of a vertical seawall (Figure not drawn to scale)
I L-Block30 I
-10
-20
I Bearing Layer I -30
-30m
Fig. 3 Soil Profile along L-Block 30 and ACLl
Fig. 1 Site of ongoing reclamation
Fig. 4 Finite Element Mesh
364
Om
Table 1 Properties use for FEM Analysis
weight
I Stone Fill Sand
1 Bearing Layer
Permeability
son’s Ratio
I
19.60
17
0.33
30”
18.62
18
0.33
35”
L
Marine clay layer - CAM-Clay Model ~=0.1, h=0.5 (C,=l.15), v’=0.33,y=17.0 W a
K,
=
K,= 1.0x10-~m/s
Table 2: Construction Sequence Activity
1 1 11
Dredging and sand fiiiplicement Inspection and sur-
1.25
1.75
7
0.5 0.75
8
0.75
9 10
0.75 1.5
Placement of Stone and Granite in sand key 0.25 L-Block Installation Placement of backfill stone 3 mths wait Sand-fill to 1.0m -then 0*25 Sand-fill to 5.33m at mth a distance 20m away N.A. Installation of coping concrete N.A. Sand-fill of the gap N.A. Surcharge imposed
2
6
1
vev
~
1
1
1
1 1
Long term consolidation
3 CONSTRUCTION PROCESS A schematic of the seawall is shown in Fig. 2. The sand in the sand key is placed by direct dumping and the only compaction comes from the tamping, carried out to compact the stone and granite, which is placed just under the L-Block. This compaction ef365
fort, which is not expected to be significant, is not accounted for in the analysis. For this reclamation, an aspect of concern is the movement of the vertical wall as a result of the sand filling operation, in particular the dynamic effect from the water carrying the sand during the hydraulic placement. In this project, the sand fill was placed parallel to the wall, and some distance from it. The main purpose for this is to reduce the lateral movement of the L-Block. This method had been adopted based on information gathered on other similar projects in Singapore where there were L-block movements when sand filling works were carried out in the direction of shore to sea. Trailer-hopper suction dredges were used in the hydraulic filling operations. A branch from the discharge pipe was placed about 25m behind the Lblocks in parallel to the revetment line. The sand was deposited from behind the L-blocks and progressed towards the shore. This method helps to minimize the lateral movement of the L-blocks and prevent any circular slip failure during reclamation. Considering the soil condition at this site, it would be impossible to avoid slip failure, if the filling works had been carried out from shore towards the sea with a bulldozer moving the sand forward. 4 DESIGN - FINITE ELEMENT STUDY The time dependent behavior of the L-Block retaining structure and its supporting ground was studied using the program, CRISP, which incorporates a fully coupled consolidation analysis based on Biot’s formulation (Britto and Gunn, 1987). CRISP ( W t i c a l State Program) is based on the finite element method and includes in the constitutive models available, a number of critical state models. This program can simulate the construction of the sand key, placement of L-Block and the sequence of fill placement. CRISP is a relatively well-known program and has been used to study problems of excavation and earth retaining structures, particularly those involving time-dependent behaviors (Powrie & Li, 1991; Bolton, et al., 1989 and Lee et al., 1993, and Yong et al., 1996). In the finite element analysis of the problem, the construction sequence proposed by the contractor as shown in Table 2 was used as the basic for simulation. Later on, after the construction, and in the back analysis, the actual construction sequence, also given in Table 2, was adopted. This project is still ongoing as reflected in the table. In this study, the finite element mesh used must
this area is very thick, more than 10m actually. This sand is placed by direct dumping, and thus may result in an inconsistent relative density, prevalent when a thick layer is formed. Since it is very difficult to demonstrate the dynamic effect of backfill stone placement, this omission has resulted in the difference between analysis and actual measurement. It is also believed that the elastic modulus of sand key is slightly lower in the actual condition than that used in analysis. These varied effects thus caused a large settlement in the sand in the sand key. Fortunately, the vertical wall was designed with a coping concrete to be added to the upper part in the final stage, thus allowing for some adjustment, provided this settlement resulted in a tilt that was acceptabele. In the present case the maximum tilt recorded, as shown in Fig. 6was 1: 130,and this was just acceptable. As the finite element study conducted cannot consider this dynamic effect, the predicted results were clearly off. For a fairer comparison, this settlement due to the dynamic effect of the hydraulic fill was removed from the results. The new comparison for the vertical settlement is not shown in Fig. 7. The predicted settlement is now closer to the measured value, especially in the early stages. However, the prediction in the later stages is still clearly smaller than the measured values. As a design, this degree of agreement is probably considered adequate. However, it is clear that the finite element study still has room for improvement. To be able to evaluate how to improve the analysis, especially the soil parameters selected, it is essential to evaluate the settlement of different soil layers, some distance away from the L-Block. This is to avoid complication arising from the placement of L-Block and the stone placement operations. The instrumentation cluster, ACLI, was placed 30m from the L-Block and was designed to provide this information through the extensometers attached to an installed inclinometer. The compression of different layers at the current stage is now summarized in Table 3. This comparison indicates that in the first design run, RTJN1, the soil parameters chosen have predicted incorrectly the compression of two layers, though the errors compensate each other, producing a predicted settlement of 112mm at 3 depth of -4.8m which is very close to the measured value of 123mm. in particular, two glaring aspects can be observed. First is the gross over prediction of the settlement of the stiff base layer (below a depth of 18.8m). The finite element analysis suggested a settlement of 19mm whereas in reality the movement is only I mm.
be adequate for simulating the entire construction process. CRISP allows the flexibility of removing and adding of elements, a feature that is particularly useful in this study. For the design analysis conducted, the mesh used is a shown in Fig. 4, which is not drawn to scale or proportion.
Fig. 6Tilting of L-Block
5 ANALYSIS OF RESULTS As a first comparison, the original design predictions were compared against the actual measured value. Fig. 5 shows the vertical settlement of the L-Block while Fig. 6 shows the tilt of the L-Block. It can be observed that the total settlement and tilt predicted is significantly smaller than the actual measured value. The principal cause for this disparity is the settlement caused during the placement of the backfill stone. A very significant settlement was noted during the placement of the back fill stone just behind the L-Block wall. Further the wall was also observed to tilt backwards during this phase, though the predicted results suggest that the wall should have tilted forward. One likely reason is that the sand placed in 366
order of magnitude while the compressibility parameter for the marine clay is also increased. The settlement profiles at two levels, -4.8m and -7.8m are shown in Fig. 8. As this is a back analysis, the agreement is necessarily much improved. The comparison shown in Table 3 indicates that there is still room for fine-tuning. This is not an important point in this paper. However, what is important is the fact that the changes made are reasonable, albeit, this is done in retrospect. Table 3: Compression of different layers of soils
Run No. Layer (m)
I
-4.80 to 7.80 -7.80 to -9.80 -9.80to-11.30 -11.30 to -13.30 -13.30 to -15.80 -15.80 to -18.80 Below -18.80 Sett. at -4.8
I
I
Runl/2 1 Run3 1 Actual Compression of each layer (mm) 49 101 82 10 19 38 18 I 4 I 0 4 5 -1 6 2 1 1 2 6 5 1 19 137 132 112
I
1
6 CONCLUSIONS
Fig. 8 Settlement at ACLl (after adjustment) On the other hand, the predicted compression of the top marine clay layer is too small, yielding a value of 59mm whereas the actual compression was 120mm. As was mentioned in the introduction, this was an actual design using values derived from routine soil investigation reports. Thus, little attention was paid to using a stiffness that will reflect that below a depth of 18.8m, the strain is likely to be very small and thus a much large stiffness can be used. Similarly, the marine clay at the top was known to contain traces of organic clay. However, the parameters shown in Table 1 are really suitable for marine clay only. To correct for this, a series of back analysis was carried out and soil parameters were adjusted based on these perceived behaviors in the field, that in retrospect, were not considered during design. The main adjustments were to the two layers. The stiffness of the very stiff clay was increased by nearly an
The present study offered a rare opportunity to evaluate an actual design calculation against measured field performance. The ability to perform accurate design analysis in a large and extensive project cannot be over emphasized as there are serious cost implications. Such an evaluation is only possible when compared against actual field study. The following conclusions can be drawn as a result of this study.
As is usual in construction projects, certain processes often cause major deviation from design that are unanticipated. In this case, the most significant movement occurs during the placement of the back fill stones on a thick sand key where some inconsistencies of density are likely to be observed. Without the relevant experience, such events cannot be included in the design analysis. The finite element analysis conducted nevertheless were able to produce reasonable prediction of the movement of the vertical seawall, provided it has been constructed in accordance with good engineering practice and the settlement due to events not simulated were factored out. 367
3. In the present analysis conducted, there is still significant room for improving the analysis by using appropriate adjusted soil parameters. The obvious challenge is the ability to estimate these parameters during design and not after back analysis as is the case here. ACKNOWLEDGEMENTS Part of the research reported here is funded by the National Science and Technology Board, Singapore, Grant No. MCE/99/003).
REFERENCES Bolton, M.D., A.M. Britto, W. Powrie & T.P. White 1989. Finite element analysis of a centrifuge model of a retaining wall embedded in a heavily over-consolidated clay. Computer and Geotechnics, Vol. 7, pp. 289-3 18. Britto, A. M. & M.J.Gunn 1987. Critical Stste Soil Mechanics via Finite Element. Ellis Horwood. Chong, P.T., T.S. Tan, F.H. Lee, K.Y. Yong & H. Tanaka 1998. Characterisation of Singapore Lower Marine Clay by In-situ and Laboratory Tests. Proceedings of The International Symposium on Problematric Soils, IS-Tohoku 98, Sendai. pp.641-644. Lee, F.H., T.S. Tan & K.Y. Yong 1993. Excavations in residual soils with high permeability. In Proceedings of the 1lth Southeast Asian Geotechnical Conference, 4-8 May 1993, Singapore. W. Powrie & E.S.F. Li 1991. Finite element analysis of an in-situ wall dropped at formation level. Geotechnique, Vol. 41, No. 4, pp. 499-514. Tan, T-S and Shirlaw, N.(1999). General Report on Deep Excavation - Braced Excavation. Proceedings of IS-Tokyo ’99 - International Symposium on Geotechnical Aspects of Underground Construction in Soft Ground, Tokyo, Japan. (under print) Yong, KY, Lee, FH and Liu, KX (1996). “Three dimensional finite element analysis of deep excavation in marine clay.” Proceedings of the Twelfth Southeast Asian Geotechnical Conference, 6- 10 May, 1996, Kuala Lumpur, pp. 435-440. 368
Coastal GeotechnicalEngineering in Practice,Nakase & Tsuchida(eds)02000 Balkema, Rotterdam, ISBN 90 5809 15 1 1
A proposal of method for calculating consolidation'settlement K.Terada Takenaka Corporation, Osaka, Japan
ABSTRACT: The e-log p curve of the oedometer test was able to be divided into the e,-log p curve of primary consolidation and the es-log p curve of secondary consolidation. There were the normally consolidated range and the overconsolidated range in those curves. And, the constants for calculating consolidation settlement were obtained from their curves. A method that the consolidation settlement was calculated separately for the primary consolidation portion and the secondary consolidation portion by substituting those constants was proposed. The consolidation settlements of clay layers were observed in Rokko Island. The observed values were compared with the values calculated by the proposed method and the standard method. The result was that the proposed method was almost appropriate.
1 INTRODUCTION
2 OEDOMETER TEST
The primary consolidation has been calculated based on Terzaghi theory for calculating the consolidation settlement of alluvial clay in the design of the building up to now. Recently, the necessity for calculating the consolidation settlement of diluvial clay was caused. It was necessary to calculate primary consolidation and secondary consoldiation separately for this clay. And, the consolidation settlement analysis is made by the finite element method by using the constitutive equation of Sekiguchi-Ota (1977) model, which contains the elements of the primary consolidation and the secondary consolidation. However, the compression index (C,) and consolidation yield stress (P,) obtained by the oedometer test are used for this analysis, and, the element of primary consolidation and the element of the secondary consolidation are included in them. Therefore, it was proposed a method for obtaining the compression index (C,,) and the consolidation yield stress (Pc,) concerning primary consolidation from the oedometer test, and a method for obtaining the coefficient of secondary consolidation (C,) and the time (t,) when secondary consolidation was began concerning the secondary consolidation. And, a method for calculating the consolidation settlement separately for the primary consolidation and for the secondary consolidation portion by substituting for those constants (Ccp,Pcp, C,, and t,) was proposed.
2.1 Separation of primary consolidation and secondary consolidation The oedometer tests with step loading were done with undisturbed samples which had been obtained in the diluvial clay layer in Osaka city. In this case, the sample was consolidated at each load stage for 24 hours when load increment ratio ( 4 p / p ) was 1.0, and for 12 hours when Ap/p was 0.42. The consistency of the clay was w,=112%, w,=35% and w=65%. The e-log p curves of the sample were plotted in Figure 1 by 0 for Ap/p=l.O and by A for p/p=0.42. Leonard (1962)'s result was that load increment ratio influenced the e-log p curve. And, his experiments were all for 24 hours loading. The e - log p curves of Figure 1 were almost identical unlike Leonard (1962)'s experiment. The secondary consolidation had been generated after the primary consolidation in their examinations (Terada 1999). Therefore, the variation of the void ratio ( A e) is calculated by the expression (1) when consolidation pressure is a p . The void ratio of primary consolidation (e,) is calculated by the expression (2) with the primary consolidation ratio (r), and that of secondary consolidation (e,) is calculated by the expression (3). The e,-log p curves of the primary consolidation were obtained from the data of Figure 1 by expression (2) and were plotted by and A in Figure 1. The e, - log p curves were almost identical. 369
2 Aep e,, - 2 Ae.s
e,,
= e,, -
e,
=
= =
e,, -
e,, -
2 rAe 2 (1- r b e
(3)
The oedometer tests were conducted with the block samples of the diluvial clay in the three soil conditions with difference of disturbance. Those conditions are the undisturbed (Und), the received strain of 30% at failure in unconfined compression test (Str) and the remolded (Rem). And, this method was proposed by Schmertmann(l9.53). Those e-log p curves were plotted in Figure 2. The consistency of the clay was w,=86%, w,=29% and w=59%. In the same way, the e,-log p curves of the primary consolidation and the e,-log p curves of the secondary consolidation were obtained from the data of Figure 2 by expression (2) and expression (3) and were plotted in Figure 3 and 4. The primary consolidation was not as sensitive to the sample disturbance as the secondary consolidation. From these examinations, it was obtained the idea that primary consolidation and secondary consoldiation were able to handle respectively separately. 2.2 Secoizdary consolidation The d-log t curves at each load stage of data shown in Figure 2 were overlapped with the curve ruler. The times where the former parted from the latter were the beginning time (tJ of the secondary consoldiation in the oedometer test, and the values were obtained. Next, The times for 90% consolidation (t,J were obtained by root t method from the d - root t curves of the data. The values of t, and t,,, were plotted in Figure 5. Since the values of t,,, were almost the same with the values of t, regardless of the level of sample disturbance, the t,,, was assumed to substitute the t, in oedometer test. The undisturbed samples were obtained from the diluvial clay layer under Port Island. They were consolidated in the oedometer test by the constant load for 100 days. Those results were shown in Figure 6.
Figure 5. relation between t,,, and t, in oedometer test.
Figure 6. variation of void ratio -time for long time (diluvial clay) 370
The value of load of the data shown by 0 was 1260 kPa , and the p, value was 950 kPa. The value of load of other data was 630 kPa , and the p, values was 560 -620 kPa. The figure showed that the variation of void ratio was proportional to the logarithm at time. The liquid limit was 83% 96%, the plastic limit was 21% 27%, and, the water content was 45% -51%. The variation of the void ratio ( 4e,) of secondarq consolidation is calculated by expression (4).
-
-
Figure 9. e- log p curve in oedorneter test with step loading. (diluvial clay)
(4) The curves of e-log p and e,-log p shown in Figure 1 by 0and 0 were plotted in Figure 7 by 0and @ respectively. The e-log p curve was divided into A ep and A e,, and plotted in the figure by A . The amounts of secondary consolidation for 1 day, 10 days and 1000 days were calculated by expression (4). Those values were added to the e,-log p curve, and they were plotted in this figure by X , +, and . The figure was compared with the figure to explain the method for calculating long-term compression by Bjerrum (1967). The e,-log p curve corresponded to the instant compression by Bjerrum. And, the group of dotted line in the figure corresponded to the (instant + delayed) compression by Bjerrum (Terada 1999). The thickness of the clay layer influenced secondary consoldiation (Aboshi 1981). And, the beginning time of secondary consoldiation in the
*
depth in the clay layer with large thickness was assumed to be able to calculate by expression (5) with the value oft,,, in oedometer test. ZC in the expression is the distance between the examined depth and the permeable face of the layer (m).
t,
= t9,,eZc
(5)
3 CONSTANTS FOR CALCULATING CONSOLIDATION SETTLEMENT
3.1 Consolidation yield stress The undisturbed samples were obtained at the center of the diluvial clay layer (Ma12) under Port Island and Rokko Island. The e-log p curves of these sample were plotted by 0,0, and A in Figure 8. It was reported that the p, values of 0 and were 600 kPa, and that of A was 750 kPa. Because the clay layer was in the boundary which changed from the overconsolidation into normal consolidation, the difference of the p, values made the state of the clay for consolidation misjudged. The oedometer tests with step loading were done with the sample shown in Figure 2. In this case, the sample was consolidated at each load stage for 24 hours when load increment ratio (Ap/p) was 1.0, for 12 hours when Ap/p was 0.42, and, for 6 hours when Ap/p was 0.19. Those e-log p curves were plotted in Figure 9 by 0, 0, and A.Figure 9 showed that the consolidation yield stress was measured more accurately as load increment ratio became small. The consolidation yield stresses were obtained from the e-log p curves in Figure 9 in the cases of 4 p / p for 1.0 and 0.42. And those values were plotted in Figure 10. Similarly, the tests were done with the sample obtained in other sites, and the pc values of those were plotted in the figure. The former was smaller than the latter. (AC is alluvial clay. DC is diluvial clay.)
3.2 Compression index The compression index of e-log p curves (C,) was not equal to them of e,-log p curves (Ccp)in Figurel.
Figure 8. e- log p curve for comparing pc values (diluvial clay)
371
In the same way, the values of C, and C,, were obtained from the clay samples in other sites. Those values were plotted for the liquid limits in Figure 11. The values of C, were not equal to them of C,,unlike Crawford (1964)’s experiment. The regressions of the compression index of the primary consolidation shown in the figure were obtained, and were expression (6) and expression (7).
C, C,
+ 9)
= 0.004(~,
= 0.005(~,- 5)
for alluvial clay (6) for diluvial clay (7)
3.3 Coefficient of secondury consolidation The Co-log p curves obtained from the data of Figure 4 were plotted in Figure 12. The values of C, sensitively reacted to the sample disturbance and decreased greatly immediately after the value of pc.
Figure 12. influence of sample disturbance on coefficient of secondary co~polidation.
The values of C , were obtained when the consolidation pressure was within the normal consolidation zone of Figure 1 were plotted in Figure 13. The coefficients of secondary consolidation C o were not influenced by the load increment ratio. The values of CJC, of alluvial clay were plotted in Figure 14 for pc/p by 0 in the overconsolidation zone and by 0 in the normal consolidation zone. The mean values of CJC, were calculated at pressure of 40kPa and 630 kPa, and were plotted in Figure15 by 0.And, the mean values of CJC, of the alluvial clay were calculated at other pressures and plotted by in the figure. In the same way, the mean values of CJC, for diluvial clay were plotted by 0 in the figure. The result was that the value of C,/C, was influenced by the consolidation pressure. The regression of the values of is the expression (8), and that of the value of 0is the expression (9) in the figure.
e
Figure 15. relation between mean of C,/C, and consolidation pressure.
372
The observed values of time-settlement curves of those clay layers were plotted in Figure 16 and 17 by (Terada 1997) The settlements of primary consolidation (S,) were calculated by expression (11) with the values of H, P,,, Ccp,e,, and p(=pO+A p) in the table.
These expressions are the functions of stress and compression index, and differed from the expression (10) introduced by Mesri (1973). C,
=0
C,
= 0.08p-".05C,
.08~-~)~"~C~
for alluvial clay
(8)
for diluvial clay (9) by Mesri(1973) (10)
C, = (0.05 * 0.02)CC
The degree of consolidation (U,,,) on the day to which the soil investigation was done was calculated with the c, values in Table 1. When the elapsed time was t, the degree of consolidation was U(t). The settlement of primary consolidation (S,,) was calculated by expression (12) with S,, U,, and U(t). When the elapsed time was t, the settlement of secondary consoldiation (&,) was calculated by expression (13) with the constants in Table 1. The consolidation settlements of clay layer were calculated by expression (14), and plotted in Figure 16 and 17 by 0 in 1986, and in 1995.
4 CALCULATION OF CONSOLIDATION SETTLEMENT The method for calculating consolidation settlement considering secondary consoldiation was explained based on the case history. Reclamation was completed on a site of Rokko Island in April, 1986. The day was a day when load was added. The soil investigations were done two times in May, 1986 and July, 1995. The undisturbed samples were obtained in the diluvial clay layer (Ma12) and the alluvial clay layer (Ma13). The oedometer tests were done with the samples. The constants had been obtained by these examinations were described in Table 1. And, the values of t, were calculated by expression ( 5 ) with the values of t,,, in the table and shown in the table. The differential settlement gages were set up in those clay layers on May 23, 1986.
a
"
s,, =
3
L u ( t ) (1- U,,,
s,,
1
C =H L l o g 1+eo
(I 1 -
Table 1. Constants for calculating consolidation settlement day
clay
-.-% n
160
-.^ > -
80
m
m
100
-.m
>
...
. I
3
"H: thickness, * *
H*
e,
P, C, kPa 0.96 64 2.25 4.8 0.97 67 2.40 4.5 5.8 1.03 98 2.34 0.83 1.52 847 2.5 1.08 1.70 602 1.6 1.26 618 1.82 2.1 1.17 633 1.77 2 1.15 592 1.88 1 1.06 609 1.77 1.5 1.07 81 1 1.75 1.5 1.02 657 1.83 3.5 0.80 691 1.38 4 0.47 1163 1.oo 1.5 345 0.9 1.47 2.8 0.82 191 1.91 3 214 0.87 1.90 3 0.79 349 1.76 3.7 0.96 734 1.54 1.0 1.01 834 1.61 1.5 1.02 71 1 1.67 2.6 1.12 717 1.72 3.1 0.98 844 1.72 3.1 1.05 743 1.51 2.9 0.81 1.2 4.7 1109 p: effective overburden pressure
c, crn2/d 60
m
pCp
kPa 50 55 60 950 640 640 700 660 710 1000 880 900 1100 280 210 200 280 950 1050 900 900 1100 1100 1200
373
C,,
0.54 0.54 0.56 0.39 0.52 0.61 0.58 0.58 0.56 0.61 0.58 0.47 0.30 0.53 0.52 0.56 0.48 0.46 0.42 0.43 0.43 0.46 0.40 0.38
P"* kPa 322 344 373 896 908 919 931 941 948 957 971 997 1018 295 313 330 348 818 824 837 855 874 892 911
Ca
0.048 0.054 0.050 0.037 0.034 0.035 0.038 0.033 0.037 0.038 0.033 0.029 0.024 0.044 0.040 0.041 0.039 0.035 0.035 0.036 0.038 0.041 0.039 0.032
t,
rnin 10 11 12 3 8 8 10 7 10 8 6 9 14 21 16 18 23 8 8 8 8 9 7 13
t,
day 0.07 22.92 0.15 0.01 0.17 0.53 13.18 25.65 100.48 72.37 21.16 0.072 0.02 0.06 0.84 2.25 0.14 0.01 0.03 0.2 4.25 35.53 1.5 0.16
The values calculated by Terzaghi’s expression (15) were plotted in Figure 16 and 17 by c] in 1986 and 0in 1995 with the constants in Table 1. The values of consolidation settlement calculated with the constants in 1986 were different from the values calculated with the constants in 1995. The consolidation yield stresses had been shown in the table were compared in the diluvial clay. One of causes of the difference of the settlement was measurement accuracy of the consolidation yield stress. The values calculated by expression (14) were closer to the observed values than the values calculated by expression (15) in Figure 17. It is necessary to calculate not only primary consolidation but also secondary consolidation for diluvial clay.
5 CONCLUSIONS The following conclusions were obtained by a result of the research. 1) Primary consolidation is independent to secondary consoldiation. 2) Consolidation yield stress p, is different from consolidation yield stress pcpof the primary consolidation. 3) Compression index C, is different from compression index C,, of the primary consolidation. 4) The coefficient of the secondary consolidation C is the function of the stress. And, it may be calculated by expression (8) for alluvial clay and expression (9) for diluvial clay. 5 ) The author proposed the method that the consolidation settlement was calculated separately for the primary consolidation portion and the secondary consolidation portion, and the method for obtaining the constants used for the calculation. (y
At last, but not the least, the author would like to express gratitude to Tokyo Soil Research, Co., LTD who had offered a cooperation in the soil test. REFERENCES Aboshi, H. and Matuda, H. 1981. Secondary Compression of clays and its effect on settlement analysis, TUCHI-TOKISO, No.3 : 19-24, in Japanese Bjerrum, J. 1967. Engineering Geology of Norwegian normally consolidated marine clays as related to settlements of buildings, Geoteclitiique 17, No.2: 83-1 18. Crawford, C. B. 1964. Interpretation of the Consolidation Test, Pro. A.S.C.E., Engrs., 90-SM: 5-87. Leonards, G.A. 1962. Foundation Etigitieeritig : 149:McGrawHill. Mesri, G. 1973. Coefficient of secondary compression, Proc. A.S.C.E., No.SM1: 123-135. Schmertmann, J. H. 1953. Estimating the True Consolidation Behavior of Clay from Laboratory Test Results, Proc. A.S.C.E. Etigrs., Vo1.79, No.311:1-26. Sekiguchi, H. and Ohta, H. 1977.Induced anisotropy and time dependency in clays, Proc. 9th ICSMFE, Speciulty Session 9: 229-237. Skempton, A. W. 1944. Notes on the Compressibility of Clays, Quart. Jour. Geol. Soc., Vol. C: 119-135. London. Terada, k. and Matano, H. 1997. Liquefaction and Spread Foundation Buildings during an Earthquake, Proc. of the Third Kutisai Itzteriiational Geotechtiicul Forum oti Comparative Geotedinical : 131-140. Terada, K. 1999. A proposal on Calculation Method to Predict Consolidation Settlement considering Secondary Consolidation, TUCHI-TO-KISO, N o 5 : 35-38, in Japanese.
Figure 17. comparison between calculated value and observed value for consolidation settlement (diluvial clay).
374
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Effect of the lateral resistance of coupled piles on the field loading test Kiyoshi Terauchi & Tsuneo Sat0 - The Third District Port Construction Bureau, Ministry of Transport, Kobe, Japan
Masatoshi Sawaguchi- University of Tsukuba,Japan Yoshiaki Kikuchi -Port and Harbour Research Institute, Ministry of Transport, Yobsuka, Japan Sosuke Kitazawa & Makoto Imai -Coastal Development Institute of Technology, Tokyo,Japan
ABSTRACT To develop a new design method for coupled piles that takes into considerationthe ground resistance perpendicular to the axis of piles, a field horizontal loading test of coupled piles has been conducted. From the test result, the percentage of shear is roughly 18% and is independent of loading level. This means that the pile penetration length can be shortened by about 10% as compared with previous design methods because the axial force is reduced by 18%. The subgrade reaction coefficient ratio of in-batter piles to out-batter piles with the angle of 20 degree of inclination, kuin/kHout,is 2.5. This result is nearly identical with the result obtained by other researchers. The degree of change in the coefficient of subgrade reaction by inclining piles in the reclaimed and compacted ground is same as in natural ground. 1 OUTLINE OF COUPLED PILES DESIGN METHODS
Coupled piles are widely used for anchorage of sheet pile walls in Japan. In the simplified design method, the coupled piles are considered to resist to external loads only through axial resistance of piles as shown in Figure 1 (Ministry of Transport 1989). Resistance perpendicular to the pile axis due to subgrade reaction, however, also takes part of the resistance to external loads to the coupled piles. Taking account for the effect of subgrade reaction, Yokoyama (1 977) proposed a method to calculate coupled piles resistance by in axial and perpendicular directions. Based on the results of laboratory tests, Sawaguchi et al. (1970) proposed an analytical method that takes into consideration the deformation properties of piles. Coupled piles used for anchoring sheet pile walls has been relatively close to the sheet pile walls in many Japanese ports. Subgrade reaction in coupled piles is small in such cases, because the movement of ground between the coupled piles and the sheet piles weaken t h e lateral resistance of the coupled piles. Consequently, the resistance component due to lateral subgrade reaction is less important in designing sheet pile walls. Recently, however, the distance between sheet pile walls and coupled piles are getting wider, because of availability of large size cranes and demand of the quay walls and anchorages for foundations of cranes. With anchorage positions getting widening away from the main quay wall, it is now beneficial to take into consideration the resistance perpendicular to the axis of the coupled piles.
V
V P H Pl
Figure 1. Force diagonals considered in the simplified design method. V vertical load, H:horizontal load, P,: axial resistance of out-batter pile, P,: axial resistance of in-batter pile.
In this research, a field loading test on coupled piles using on-site facilities has been conducted. The aim is to investigate the resistance perpendicular component and to estimate the coefficient of subgrade reactions in the direction perpendicular to the axis for coupled piles in anchored sheet pile quay walls.
375
2 FIELD HORIZONTAL LOADING TEST
axial force and shearing force of each pile were measured with triaxial strain gauges at 2 points on each pile and the bending strains were measured with strain gauges at 10 points on each pile (Figure 3).
2.1 Test procedure Figure 2 shows a schematic view of the ground condition, along with the coupled piles and sheet pile wall at the site where the horizontal loading test was carried out. The natural ground surface at the site was DL. -7 m. The top layer of the natural ground to a depth of 4 m was a layer of clay, overlying a gravel layer. The construction of the site was carried out as follows. First the coupled piles and the sheet pile walls were installed and the coupled piles were bound to the sheet pile walls with tie rods; sand filling was then done behind the quay wall; densified by sand compaction pile method as shown in Figure 2. The average of N-values from standard penetration tests in the improved ground was about 13. The piles used for both the out-batter piles and in-batter piles of the coupled piles were steel pipe piles with a radial thickness of 12 mm and diameter of 1000 mm. The piles were driven to an embedded depth of DL. -26 m. Two pairs of couple piles were set as one block and horizontal load was applied. The load to the pile heads was applied by pulling the tie rods with hydraulicjacks. The height of loading point was 1.1 m above ground surface. The pile head section was covered with a concrete block of 4.3 m long x 2.23 m wide x 1.6 m high to fix pile head angle. The lower surface of the concrete block was 0.5 m from the ground surface. The test was conducted with static loads using multistage, multi-cycle loads in accordance with Japanese Society of Soil Mechanics and Foundation Engineering standard (1983). Displacement of the concrete block was measured with 12 dial gauges, the
2.2 Test results The relationship between horizontal load and the horizontal displacement of each pile head is shown in Figure 4. The horizontal displacements of the pile heads are calculated from the measurement of the concrete block displacement. The relations shown in Figure 4 present less torsional displacement. In this test, the effect of torsional deformation is ignored, though the couple piles are not axi-symmetric. Relationships between horizontal load and the axial force acting on the out-batter piles and on the in-batter piles are shown in Figure 5. The axial force for each pile is determined from measurements of triaxial strain gauges. The axial force of out-batter piles represents the sum of axial force of each out-batter pile. The axial force of in-batter piles represents the sum of axial force of each in-batter pile. Both the axial compressive force on the out-batter pile and the axial extension force on the in-batter pile increase linearly with the horizontal load. The absolute values for the axial force of the inbatter piles showed a tendency to be somewhat larger than those of the out-batter piles. The in-batter piles still could not be extracted within the load range in this experiment. Figure 6 shows relationship between the horizontal load and the shear force acting on the out-batter pile heads and the shear force acting on the in-batter pile heads. The shear force acting on each pile was calculated from the measurement of triaxial gauges. Pile top section concrete
30.5m Tie rod
CDL-27.0m 0
5
10
15
20(m)
Figure 2. Cross section of the coupled piles tested. Sand layer over laid on the original groud is improved by sand compaction piles method(SCP) and gravel drains method(GD).
376
Figure 6. Relasionship between horizontal load and shear forces of pile tops. Shear forces are summed up for out-batter piles and in-batter piles.
The percentage of shear p (Sawaguchi 1970) is calculated by using the results from this experiment. The percentage of shear p mentioned here shows the ratio of the horizontal component of the shear force perpendicular to the pile axis to the horizontal load acting on the coupled pile. This is shown as the following equation:
Figure 4. Relasionship between horizontal load and horizontal displacement of pile tops. Numbers shown are the number of each pile. Piles are installed in aline as they numberd.
The shear forces shown in Figure 6 are the sum of two out-batter piles and two in-batter piles, respectively. Figure 6 shows the shearing forces acting on both the out-batter piles and the in-batter piles linearly increase with an increase in the horizontal load with few scattered results and the shearing force acting on the in-batter piles is somewhat larger than that on the outbatter piles.
percentage of shear in which, HIand H,: shearing forces at the top of the inbatter and out-batter piles; 0, and 0, : the batter angles against vertical in the in-batter and out-batter piles (20 degrees in this test case); T is the horizontal load.
377
Figure 7 shows the relation between the horizontal load and the percentage of shear. The percentage of shear is roughly 18% and it is independent from the level of the horizontal load. Sawaguchi (1970) showed that the percentage of shear of coupled piles is almost constant until in-batter piles are extracted by the model test conducted in the laboratory. The results from the present field test give the same conclusion. Moreover, Sawaguchi (1970) mentions a sharp increase in the percentage of shear when the in-batter pile begins to be extracted. Sawaguchi (1 970) also concludes that the percentage of shear varies under different experimental conditions such as loading height, angle of batter piles, and ground conditions. Assuming the percentage of shear is 20% until the in-batter pile is extracted, axial force of piles becomes 20% less than that not considering this effect when the load is completely horizontal. And the pile penetration length can be shortened by approximately 10%.
Figure 8. Relasionship between shear force and displacement perpendicular to pile axis at pile top.
Furthermore, according to test results from Kikuchi (1999a), the lateral resistawe of a coupled pile continues to increase even after the in-batter pile has begun to be extracted. It will be possible to design coupled piles even more economical if these results can be incorporated into design. 3 ESTIMATION OF COEFFICIENTS OF SUBGRADE REACTION IN THE DIRECTION PERPENDICULAR TO THE PILE AXIS Displacement of pile heads of couple piles has been simulated by considering the reaction perpendicular to the pile axes. Yokoyama (1977) proposed the pile head springs in axial direction and perpendicular to axis direction of piles. In order to appropriately evaluate pile head springs, it is necessary to properly estimate the coefficients o f subgrade reaction perpendicular to the pile axes. Relationships between displacement of pile head and the load component perpendicular to the axes of inbatter and out-batter piles that are obtained from the field test are shown in Figure 8 (Kikuchi et al. 1999b). The displacements of the pile head perpendicular to pile axes are calculated from the measured displacement, and the load perpendicular to the axes are calculated from the measured shearing force of the pile head using a triaxial strain gauge. As shown in Figure 8, it is known that the in-batter piles show very large resistance compared to the out-batter piles. This result coincides with the result shown in Figure 6 where in-batter piles have large forces perpendicular to the pile axes. What all these show is that the coefficient of subgrade reaction for the in-batter piles is larger than that of the out-batter piles. Figure 9 shows relationships between the load and displacement of the pile head perpendicular to pile axis. The relations shown in lines are calculated using Chang’s equation (p = k,?) and those in plots are the measured values in the test. According to this data, the coefficient of subgrade reaction of out-batter piles kljolt,,estimated from the measured value, decreases as displacement increases, while the coefficient of subgrade reaction of in-batter piles kljlNis almost constant and independent of displacement. Comparing the coefficient of subgrade reactions at the point where the pile head displacement perpendicular to the axis is 1 cm, the out-batter pile kHOly= 5000 kN/m3, while the in-batter kH,,,= 12000 kN/m . Figure 10 compares the measured and the calculated load versus displacement relationship as it can be expressed using the Kubo’s formula 0, = kSxyo5). In this case the estimated subgrade reaction constants for both the in-batters k,,,, and out-batter piles k$,,,,,is nearly constant independent of displacement level; for the out-batter pile k,Ol,,= 200 kN/m3 5 , and for the in-batter pile kslt1= 500 kN/m3
Kubo (1 962) showed the coefficients of subgrade reaction perpendicular to pile axis vary depending on the differences in the pile’s angle of inclination from the results of field tests and laboratory experiments (Figure 11). The vertical axis in this figure is the ratio of the coefficient of subgrade reaction for inclined piles to vertical piles. The figure shows a comparison of the coefficient of subgrade reaction between vertical and inclined piles. The dotted line shown is estimated from the field test, which was conducted by Kubo, carried out by inserting inclined piles into natural ground. The solid lines are estimated from the results of laboratory tests where the ground is thoroughly compacted at the front surface of the inclined piles. From these results, it is known that the coefficient of subgrade reaction for the in-batter piles grows larger compared to that for the vertical piles. Conversely, the subgrade reaction for the out-batter piles grows smaller than that for the vertical piles. Since the coefficient of subgrade reaction for the vertical piles is not obtained in this test, the coefficient ratio of subgrade reaction of the out-batter pile to that of the vertical pile is tentatively 0.6. Then, the coefficient ratio of subgrade reaction of in-batter pile is to be 1.5. These plots are also shown in Figure 11 for comparing the results from the research of Kubo. As mentioned in 2.1, the ground around the piles were reclaimed in this test site and soil improvement through ground compaction at the front side of the piles may appear somewhat inadequate for enough densification. This condition is different from the condition in which Kubo (1 962) tested. And the test results correspond significantly with those previously obtained by Kubo in regard to the ratio of increase in the coefficient of subgrade reaction between out-batter and in-batter piles. This result shows that it is acceptable to consider the coefficient ratio of subgrade reaction for inclined piles in reclaimed ground similar to that in natural ground.
Figure 11. Coefficient of subgrade reaction changes according to anlge of batter pile.
4 CONCLUSIONS To develop a new design method for coupled piles that takes resistance properties perpendicular to pile axis into consideration, a field horizontal loading test of coupled piles was conducted. From the test result, the
Figure 9. Comparison of caIcuIation results according to Chang’s formula and test results.
379
percentage of shear and the coefficients of subgrade reaction are discussed in this paper. The conclusions are as follows: 1) The percentage of shear is roughly 18%, and is independent of loading level in the test. This means that the pile penetration length can be shortened about 10% as compared to simplified design methods because the axial force diminishes by 18%. 2) The subgrade reaction ratio of in-batter piles to outL:. ‘:er piles, kHin/kHOul, with the angle of 20 degree of inclination is 2.5. This result is nearly identical to the result obtained by Kubo. The change in the coefficient of subgrade reactions by inclining the piles in the reclaimed and compacted ground is the same degree as in natural ground.
REFERENCES Kikuchi, Y., Oooka, S. & Taguchi, H. 1999a. Horizontal loading test results on coupled piles - Part 1 outline of tests and their results -, Proc. of 54th Annual Conf of the Japan Society of Civil Engineers, 3-(A): 836 - 837. (in Japanese) Kikuchi, Y., Abe, K. & Yuasa, K. 1999b. Change of lateral resistance of batter piles according to the ground improvement by sand compaction piles, Proc. of34th Japan National Conf on Geotechnical Engineering, 2: 1661 - 1662. (in Japanese) Kubo, K. 1962. Experimental study on the lateral resistance of piles (Part 3) - Lateral resistance of single free-head battered piles and single fixed-head vertical piles -,Monthly Reports of Transportation Technical Research Institute, 12(2): 3 1 55. (in Japanese) Ministry of Transport. 1989. Technical Standards and Explanationsfor Port and Harbor Facilities in Japan, (1): 397 - 398. (in Japanese) Sawaguchi, M. 1970. Empirical investigation on the horizontal resistance of coupled piles, Report o f the Port and Harbour Research Institute, 9( 1): 3 - 69. The Japanese Society of Soil Mechanics and Foundation Engineering. 1983. Standard Method for Lateral Loading Testfor a Pile and its explanations. Yokoyama, Y. 1977. CalculationMethods and Examplesfor Piled Structures: 147 - 152. (in Japanese)
380
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Characteristics analysis of granular fill on oil-tank soft soils Li-Zhong Wang, Xu Liang, Yuan-Qiang Cai & Shi-Ming Wu Department of Civil Engineering, Zhejiang Universig, Hangzhou, People’s Republic of China
ABSTRACT: This paper presents a granular fill behavior model on soft soil which includes a shear beam and Terzaghi consolidation. The mathematical expression of this model is a two differential-linear ordinary difference equation. Using a finite difference method, we study the influence of some parameters such as the radius, modulus and thickness of the granular fill on the maximum settlement and differential settlement of an oil tank foundation. The results which have been drawn into graphics are compared with those using traditional settlement-calculating method. 1.INTRODUCTION Soft soil is widely distributed over coastal and riverside regions. Oil-tanks are often constructed in these coastal port regions because of their special locations. Soils under large oil-tanks which are sensitive to differential settlement should bear heavy load. The compressive stress affects deep-ground soils. Usually it is the differential settlement (rather than the general settlement) that is of the concern in the design of a foundation. On the other hand, it is much more difficult to estimate differential settlement than to estimate the maximum settlement. Therefore, it is very important for the underlying foundation to be well designed and constructed. In the design of the foundation system, the critical problem is to prevent settlements from being so large as to be a danger to the structure. Granular fill is often taken into account when disposing of soft soils underlying oil-tanks. Two kinds of cases are to be considered. One is that the coastal-site level is below sea level and granular fill is paved to raise the site level and then oil-tanks are built on it. When the ratio (Rdro) of the radius of granular fill (Ro)to that of oil tank (YO) is relatively large, we may consider it to be this case. The other considers economic factors and granular fill is paved after the soft soil in a given region is dug away and Rdro is between 1.1 and 1.4.
The hnctions of granular fill may be generalized as follows: (1) to improve the bearing capacity of shallow foundations. (2) to reduce the settlement of foundations. (3) to reduce differential settlement between in the center and the edge of load region. (4) to quicken drainage and consolidation of soft soil. Settlement problem of oil-tank foundations becomes more complex because of the presence of granular fill. Study in this problem is practical to engineering for special application demand of oiltanks.
2.MODEL AND DESCRIPTION In accordance with granular fill under strip foundation, two Indian scholars Sanjay Kumar & Sarvesh Chandra( 1995) put forward a model, which was developed from Winkler foundation. The granular fill in this model is idealized into Pasternak shear layer and the soft soil is idealized into springdashpot system. Spring-constant adopts that of Winkler model. Settlement under plane-strain situation formed by strip load applied onto granular fill is analyzed, considering the primary consolidation of soft soil only. However, to oil-tank foundations, it is difficult to obtain results conforming to engineering practice when adopting Winkler model to calculate the settlement. This paper is to construct a new settlement-analysis
381
model about granular fill-soft soil system as shown in Figure 1. The model considers granular fill as Pasternak shear layer and represents saturated soft soil with the Terzaghi's one dimensional consolidation model. The spring represents the soil skeleton and the dashpot simulates dissipation of pore water pressure in the soil. Axial symmetrical load is applied onto granular fill. The equation controlling the response of the model at any instant of time can be acquired when considering the equilibrium of forces. Taking the shear layer element shown in Figure 2. into account, according that the resultant force along 2 coordinate axis is zero, the equation can be written as: 2 ~ q d r - 2 ~ c L d r - 2 7 c r H 1 x ~ ~)x2n(rtdr)H1=0 *~+~dr
a
(4) where k, =&
(5) where
&
.TV(=c"')is
is positive odd integer
the time factor for primary
H,'
According to effective stress theory, the total
(2)
(i.e. xrz=G*),
4
(1,3,5...)
(1)
stress is the sum of U, and cr: the equation can be
where q is the circular uniform load intensity, qsis the vertical stress at the interface of the shear layer and the soft soil. Since there is a relationship &
=e and m
~2
consolidation and uois the initial pore pressure.
a. rHl=O
between ,z and G*
is the
12(1+ v , )
effective stress. EOand vo are respectively the elastic modulus and Possion's ratio of soil skzleton. HZ is the thickness of the soft soil. In light of Terzaghi's one-dimensional consolidation theory, the average pore water pressure(u,) in soil may be written as:
Equation( 1) can also be simplified as: qr-gr+,H1+%
, t =E H ~ , a n dCT'
H,(I - v % )
written as: (6)
where G and With u o = q,, equation(6) can be written as:
,z are respectively the shear modulus and the shear
stress of the layer. HI is the thickness, W(T) is the vertical surface displacement, T is the distance measured from the center of the load region. Equation(1) can be written as the following form:
(7)
Then the following equation can also be obtained: o"q,(l - m=m C - - e -2M ' r ,
) (8) M2 (the average degree of consolidation in oneml
(3)
U
As to the underlying soft soil layer which is considered as spring, the strain state of elastic layer in T-z plane is studied with Vlazov model (see Figure 3). The vertical force equilibrium equation can be obtained through Langrage's principle of virtual work.
dimensional theory) equals to(1-
y+-~b) m=l
M
Equation(8)can be written as: &qsU (9) Substituing equation ( 9 ) into equation (4) , the following equation can be obtained:
Figure 1. Conceptual diagram and suggested foundation model
382
(10)
Substituting equation ( 10) into equation (3 ) , the equation can be written as:
Equation(l2) can also be applied to characteristics analysis to granular fill on soft soil disposed by prefabricated band drains or sand drains, whose main difference is the degree of consolidation at particular instant of time. The following normalization may be used to make items in equation( 11) dimensionless: 2t
-
+ GH,
Y = r lr,, W = w(r)/r,,G* =U, q* = 4 (12)
kA2 k*ro Using the above normalized items, equation( 11) becomes:
3 .FINITE DIFFERENCE EQUATIONS
The following central finite difference-scheme is used:
w (Y',)
= W(yTi+I1- WWi-1)
(14)
2h
(i=1,2,***n)
Equation (1 3) can be written as:
The degree of consolidation is taken as a hnction of time.
Hi Figure 6. Settlement-distance profiles for different shear modulus
Figure 2. Forces on the shear layer element
Hz soft s
d r
impervious b ou n d ary
4.LOADNG AND BOUNDARY CONDITIONS qo* is a uniform non-dimensional load intensity acting over a radius(2ro). Due to symmetry about the center of the load region, only radius (ro) needs to be calculated. The slope of settlement-distance profile at the center of load region is considered to be zero
Figure 3. Vlazov model
383
(i.e.
dw = 0 at dY
~ 0 )Since . the edge of granular fill
is assumed not to bear shear stress, the slope is also considered to be zero (i.e.
"I= 0 at
w=RJr,).
dY Some equations can be obtained with equation( 16) and boundary conditions as follows:
-
w,+ 4 4 .- 3w02h
3w,-4wn-, + wn-2-0 2h
(18) (19)
(i'l ,2, * *n-1) WOand W, can be substituted in equation(l7). After iteration, equation(l7) can be written as a matrix form .With a computer, W, W,, *..,W,, and then WO and W, can be obtained through equations (18) and( 19).
S.PRELTPL/m\JARY RESULTS In this part, the cases when Y is 1.2 ,1.4 and 5 are going to be discussed.
Figure 4 below shows the settlement-distance profiles for different shear modulus when the ratio of R, to r, is 5 . We may observe that the maximum settlement(S-)and the differential settlement (Sdf)between the center and the edge are larger in the case of small shear modulus (G) of the granular fill than those of large shear modulus. When U is 10%, the difference of S,, between small shear modulus and large shear modulus is not significant; however, when U is 1OO%, the difference of S,, is remarkable. The shear modulus has a more and more remarkable effect on , S , as U increases gradually under constant load, which means granular fills with larger G (compacted fills) are preferred in order to reduce both S,,,, and Sdfin engineering. Figure 5 below shows the settlement- distance profiles for different radius( tv=1.2,1.4 and 5) when G* and q* are respectively equivalent to 5 and 0.1. S, varies according to Y. When U is loo%, the S,, ratio of Y=5 to ~ 1 . 2is 0.07/0.0175 =4 .However, &$in the case of ~ 1 . whose 2 value is 0.001482 is smaller than that in the case of ~ = whose 5 value is 0.004173. In the figure , at the edge of the granular fill, the slope varies with different Y.When u/is 1.2, the slope goes smoothly, which means the granular fill with smaller Y shows higher rigidity compared with soft soil whose shear strength amounts to less than one tenth of the fill. 6.ENGINEERINGPRACTICE As Figure1 shown above, there is an oil-tank soft foundation soil ( H ~ 2 0 m )disposed by granular fill. The radius (ro) is 10m while the fill radius (Ro)is 12m. Uniform pressure qo amounts to 150kpa, and
Figure 7. Setttement-distanceprofiles for different radius
Figure.8. Settlement-distanceprofiles for different thickness ofthe granular fill
384
Table 1 shows the effects of G and U on G*. G* 1 U=10% 1 U=50% G==20MPa 1 7.357 4.159 G=lOMPa f 2.479 5.677 GSMPa 1 4.838 1.639
the elastic modulus (Eo) of the soft soil is 3Mpa.The Possion’s ratio is 0.4 and HI is 3m. The following results can be easily obtained:
I
k s 4 .1786MNld: e 3 . 57MN/m3; 4t 0 . 0 8 4
Figure 6. above shows the settlement-distance profiles when G are 20Mpa,lOMpa and 5Mpa respectively. S,,,, doesn’t vary much when U is 10%. The value is 64mm when G is ZOMpa, which is only 13mm less than that when G=lOMpa. As U increases, S,,,, becomes remarkably different. When U reaches loo%, the value is 386mm when G is ZOMpa, which is 135mm less than that when G is 10Mpa.This can be explained that when the foundation soil begins to consolidate, 2tNmay be large enough not to be neglected and the value of 2f
C *(= 5
+
U=lOO% 3.759 2.08 1.24
Table 2 shows the effects of HI on G*( G=2OMpa, Y =5\.
H1=2m HI=lm
6.237 5.12
3.04 1.92
12.64 11.52
With data above, settlements are calculated. From Figure 8. below, a conclusion can be drawn that the settlement decreases as the value of HI increases. When U is loo%, S, is 173mm for H, being 3m while it is 217mm for HI being 2m; it is 298mm for HI being lm. It is obvious that the thickness (H,) of the fill affects settlement remarkably. As H, increases at the same number, S,,,, decreases more and more rapidly. With trad~t~onal layer-wise summation method at Z//R,=3/10=0.3, the stress dispersion angle is 20’ and unit weight of soft soil is 17.8KN/m3. The designed additional stress at the bottom of the fill is
) doesn’t vary much, which accounts
k,d
for little difference in S., As U increases gradually to loo%, t can properly be omitted. At this time, the effect of variation of G on settlement is hlly displayed. Because of great difference between 2t and GHI, f can be neglected when calculating the final consol~dation settlement and then G* is (GHI)/(kfZ,). However, it must be pointed out that when the shear modulus is not very much larger than elastic modulus (Eo), the binding effect of underlying soft soil (2t) can’t be omitted. Obviously, to raise the value of G can significantfy decrease the surface settlement. Figure 7. shows the settlement-d~stanceprofiles when Y are 1.2 and 5 respectively G=20Mpa. From the figure below, we can observe that I// affects not so much on settlement when the soft soil begins to consolidate (U=lO%). As U increases, the final consolidation settlement in the center reaches 173mm for iv=5 while that reaches 586 mm for U/ =1.2. A conclusion can be drawn that the fill radius affects remarkably on when U is lOO%, which can’t be obtained through traditional stress dispersion method. Letter ‘‘8 represents the slope of the final consolidation settlement in the center and that in the edge. When 1 ~ r is 5, Sis 0.48% and when Y is 1.2, 6 is 0.16%. Therefore, another conclusion can be drawn that with the same shear modulus, the general settlement of the fill is more similar when is small at different consolidationdegree.
I
~ , = m O 2 ( p --~E ,X) 102 x 150 = 104.2Kpa do2
Z X
122
The thickness of each divided soft soil layer is 4m. At last, the result can be obtained that the final consolidation settlement in the center is 599mm while that in the edge is 305mm. Comparing the results using layer-wise summation method with those using the method provided in this paper, we can observe that the respondent results of S,, are almost the same for G=20Mpa,H=3m, 1v=1.2.But for Sdfthere is a large difference between the two results, one of which is 294mm with layer-wise summation method, and the other is 18mm. Since traditional method does not take the interaction between granular fill and soft soil into account reasonably, the result does not conform to engineering practice very well. The method given in the paper can predict and calculate the maximum settlement and the differential settlement more accurately.
385
1
7.CONCLUSIONS The generalized conclusions drawn from calculation above are summarized as follows: 1. The foundation model provided suggested above has a simple response function and can be well applied to the study of settlement response of granular fill on oil-tank soR soil. 2. It is very usekl with granular fill disposing of soft soils to reduce S,,, and S d C . The larger G is, the more remarkable the effect of the granular fill to reduce settlement is. t can be omitted when calculating the final consolidation settlement, which leads to G* =(GHI)/(X$J. However, when G is not much larger than E,, the binding effect (2t) of the underlying soft soil can’t be omitted. 3. The radius of the granular fill can affect S,, and S d f i too. The larger the radius is, the smaller S,, is. When 1 ~ / is relatively small(but still>I 1, S d f is smaller than that of larger Y. To soR soil whose shear strength is o d y one tenth of the granular fill, when Rdro is smaller, the granular fill shows more rigidity. 4. The thickness (HI) afFects the settlement remarkably. When the thickness increases evenly, S,,,, decreases rapidly. 5 . The degree of consolidation (U) aiyects S,,,, and S d f . To different H I , G , v,when U is smaller, and s d $ ; as the all three parameters don’t affect U increases, the effects can be displayed entirely through calculation, especially remarkable to S,, .
T.William Lambe; Robert V. Whitman: Soil Mechanics. SI Version (1979), John Wiley &Sons, Inc, New York. Winkler,E.( 1876): Die Lehre von der Elasticitaet und Festigkeit. Prag, Domincus.
s,,
REFERENCES Das,B .M: Advanced Soil A4echanics.Hemisphere Publishing Corporation,Washington.1983 Pasternak, P. L(1954): “On a new method of analysis of an elastic foundation by means of two foundation constants”. Gosudarstvennoe Izdatelstvo Literaturi po Stroitelstvu I Arkhitekture, Moscow, USSR.(in Russian). Sanjay Kumar & Sarvesh Chandra:“Time-dependent Settlement Response of Granular Fill on Soft Soil”. Soils And Foundations Vo13 5 Nod, 105 -108 Dec. 1995. Se1vadurai.A.P.S (1979): Elastic Analysis of SoilFoundation Interaction. Elsevier Sc. Publ. Co., Amsterdam. Terzaghi.K (1943): Theoretical Soil Mechanics. John Wiley &Sons, Inc, New York.
386
Coastal Geo~ec~~icai fn~ineer~ng in Practice, Nakase L? Tsuchida (eds)02000 Baikema, Rotterdam, lSBN 90 5809 151 1
rbation stochastic finite element method and its a ~ ~ l i c a t in ~ othe n ~eliabilityanalysis of excavation J i ~ p ~ Xu, n g Jian Zhou & Yuwei Chi ~ e ~ u r ~ofe~ne ~t t e c h n i c~anl g i n e ~ r ~Tongji n g , ~ n i ~ e~ rh ~a in~~People's ~ ~ i Republic , of China
ABSTRACT: The r e l ~ a b i ~ ~analysis ty of excavation by PSFEM (Perturbation Stochastic Finite Element Method} is carried out. The results of a numerical example demonstrate the strength of this method in solving the reliability problems of excavation. With its strict theoretical framework and accurate computational results, PSFEM is a reliable and efficient computational technique in the reliability analysis of excavation. 1 INTRODUCTION
The stability of excavation is mainly influenced by material properties of soil and external loads, which are of spatial uncertainty. Performance of the system should be considered as a kind of stochastic one, which is the function of random fields. Thus, the stress and displacement field should have characteristics of random fields. PSFEM can be used to analyze the system responses when internal and external factors change stochastically, yet there are few examples in the reliability analysis of excavation by PSFEM. Virtual works of PSFEM mainly consist of two aspects that are discretization of random field and inverse operation of stochastic operator and matrix. This paper provides an overview of PSFEM and a numerical. example on the reliabi~ityof excavation.
transformation is introduced into the local average model (Chen & Liu 1993). A 2-dimensional continuous smooth random field a(x,y; cox,+) with ensemble average rn and variance d? can be discretized into a stochastic vector ={aj,az,...,aJT, where o,,wy are sample value, n is the sum of elements. Local average of arbitrary straight quadrilateral efe~ ( x , ~)(x,y ) ~ E~.iz,), y ment e is CZ~=;(I/A~)~
6,
where A, is the area of element e, and a, is the region occupied by element e. The expectation of locally averaged random field element e and the covariance of two arbitrary elements e, e' could be given respectively as follows:
2 ANALYSIS OF STRESS FIELD BY PSFEM 2.1 Discretization of randomfield The theory of local averages of random fields (Vanmarcke 1986) is an approximate theory that functions of random fields are locally averaged into random variables in every discrete element to represent the statistic of every point in homologous element. To give the second-order statistic of solution h n c tion, the stochastic finite element method based on the theory of local averages needs only the ensemble average, variance and scale of fluctuation of original random field. Required random information inputted is decreased greatly. The method of linear coordinate 387
where p(r, s;i is the standard correlative fmction, and A
A
are the nodal coordinate value of discrete element e. NI is the shape function of element e and N,= ( ~ / ~ ~ 131 is the ~ Jacobian. ~ + ~ ~ If the crossed method of eigenvector is applied on the discrete model of random field, stochastic vector composed of stat~st~cally correlative components could be converted into one composed of statistically
~
0 < i 5 n, & 0 < j 5 n, n, < i l n & n , < j < n others
independent component (Chen & Liu 1993). The scale of calculating could be reduced greatly. Because the density of finite element mesh has nothing to do with random field but is determined by stress gradient, another mesh should be adopted to discrebe detertized random field, whose density mined by the scale of fluctuation.
(5)
After the finite element mesh is generated, the stiffness matrix K , external load vector E and nodal displacement vector _V at the average of 5- (0 ) in form of Taylor series are expanded and the expansions is truncated at the second-order. Substitution of these series into the governing equation (3) and application of the mean-centered second-order perturbation method to the result give rise to the following recursive equation group:
2.2 The second-order information of stress and displacementfields The governing equation Of displacement solution Of finite element method could be written as follows: (3) -_ KV=F Because the stiffness matrix L( depends on mechanical properties of soil material and nodal coordinates involving uncertainty, and the external load vector _F is also stochastic, equation ( 3 ) shows that the nodal displacement vector is certainly stochastic. Assume mechanical property m and external load g involve uncertainty, which can be represented by a small but random variable a and - p respectively. Hence, m = m(1+a) and g = g(1-t p) ,
m
where and are ensemble averages representing the certain parts. The random parts a and p can respectively be modeled a mean-zero continuous smooth random field. After local averages and eigenvector crosses of a and ,O respectively, stochastic vectors = (a,,a, ., anl}' and p = (p, p2,.- .,pn2}' ,a
could be obtained, in which components are statistically independent. The correlation functions of every component in g and /? are as follows:
where
-c c v ~ c o v ( t ~ , t , ) 1 "
3= 2
,=I
,=I
-
In equation (6)-(8), K Fo and Vo represent the -O ,expectation matrixes. The suffix i andj of _K ,E and V represent the partial differentials to { j and 4 at mean-zero point, which are deterministic matrixes independent on 5j and 5. Solution of aforementioned recursive equation group one by one, matrixes _ Vo, _ 6 and V, can be obtained, and the secondorder statistical expectation matrix of displacement field can be defined as follows:
And the covariance of displacement vectors of node e andfis: Therefore the synthetically stochastic vector 5= {t,,5,;. ,,{,I}T= { g , where M = ~ I + YObI~. viously p is still a mean-zero stochastic vector, -
'
which is E [-5 ] = 0.If random field a is independent on p, the elements in the covariance matrix of every component in 5- can be assumed as follows:
cov ( a , , a , > '0'
The matrix relationship between stress and displacement vector of element e is & = D"Be Y e , in which , , and ILf.are respectively the stress vector, elastic matrix, strain matrix and nodal displacement vector of element e. B" is independent of
5. The expansions of -
(P(i-nl) > P(,-nl)
0
,E and
y" in form of
Taylor series are truncated at the second-order. Substitution of the expansions into the matrix relation388
ship and transformation, give rise to the following second-order expectation of stress field of element e and the covariance of element e and f respectively:
Expansions of the main stress 0 : and o .and maximum shearing stress Z e m a of element e to random variable Ci at mean-zero point in a form of Taylor series, which are truncated at the secondorder, can give rise to the following expectations:
- -
n
n
-
- -
-2
E [ z ~ , ] =r i m = ( ( O : - O ; ) ~ / ~ + Z $ ) 1 ' 2 (17) The covariance of maximum main stress, minimum main stress and maximum shearing stress of element e are obtained respectively as follows:
Cov(a",(af)')
n
n
n
n
n
n
in which, _ Dl ,B," _ and V: - represent the expectation matrixes, and the suffix i and j represent the niatrixes of partial differentials to tiand d at mean-zero point. It is obvious that B is independent of 5. 2.3 The second-order information of main stress jeld
c0v(Zim7Z~m)
=
~ ~ Z i , , f r i m , J c O v ( t { , t J ) (20) I = 1 J=1
of eleThe partial differential of stress vector ment e to random variable Cl at mean-zero point of random vector - is E:, =(dx,f, a",,. -f,J'. On one
<
hand, the stress vector can be obtained by Now take a planar problem into account. Partial dif0' = D ' B e V e , on the other hand, by the meanferentials of equation dl, d3=(cfx+ d , ) / 2 ~ fof, ~ ~ ~ centered second-order perturbation to the equation main stress and f,,={(dx-dy)2/4+ fv2j1I2 of the vector = D," B,' y e + D,' Bf' _ V , + _-0,'B,' V," can maximum shearing stress of element e to random __-__ variable ti at mean-zero point, following equations be obtained. It is obvious that the values of cfx,l,a"y,l can be obtained: and f q , l can be obtained accordingly. Substituting them into equation (13) and (14) respectively and taking equation (15) into account, the values of dl,,,d3,, and Tema,i can be obtained. Lastly according to the equations from (1 8) to (20), the covariance of main stress and maximum shearing stress of element e and f can be obtained (Xu & Hu 1999). Let element number e =f,the variance would be obtained. d y , i and f x y i represent in which, dl,i,d3,i,f m q i , dx,,, the partial differentials of dl, d3,f m a X ",,a", and fv to random variable tiat mean-zero point respec-
-tively, and cr ," ,0 and z :y represent the average of the component d,dL.. fv of stress vector of element e. The expectation vector of stress vector = { ~ T ~ ~ , ~ can T ~be ~ obtained , z ~ ~ ~as) follows: ~ ---
Ella'l=
, ,
(0: 0; Z:y
1'
(15)
3 RELIABILITY ANALYSIS OF EXCAVATION
3.1 Local failure probability analysis of homogeneous isotropy system Assume that the shearing failure of soil coheres to Mohr-Coulomb criterion. The cohesion C and internal friction angle @ are random variables, which could be defined as C=C(lfE) and #=J(l+cv)respectively.
389
c and ? represent the averages of C and
4 respectively. The stochastic part E and ry could all be modeled to mean-zero continuous smooth random fields. Local averages and eigenvector crosses of E and I,U, stochastic vectors 4: = (cl , , . . . , E , , }' and - = (ryl ,ry2 ,. - .,ry,, } could be obtained in which ry every component is statistically independent. So the correlation functions of every component in g and iy could be defined just like equation (4). -
Synthesize stochastic vectors g , ry and the former vector
5- (obtained from PSFEM) into a new
in which Q.:,i is the partial differential of Q," to qj at the mean-zero point of 7 , and part. so: Ere,']=
Q,' is
the average
2
_ _ _ - _ = c c o s ? + ( o ; +ol)sin?/2-(of -o3/2 (23)
Because ante of QB
5- ,E_
and ry are independent, the covari-
and Ql is obtained as follows:
stochastic vector -q = = { ~ l , ~ 2 , ~={< - ~T,, g~T N,_w }T T} T , in which N=n+n3+nd. q is still a mean-zero stochastic vector. If E and ry are irrelative, the elements in the covariance matrix of every component in stochastic vector q - can be assumed as follow-s:
6- ,
The partial differentials of equation (22) to and
O
iy -
respectively at mean-zero points are:
Q.:,<,= {(a;,+ 4 , , ) s i n 8 - <4,, -4,,)}/2 (21)
Q:,E,
= cos4
- _
Q;,,, = -?sin?
Omitting the failure direction of element in homogeneous isotropy system, the failure probability of element can be obtained according to the approach extent of the stress circle to the strength curve. If r; represents the distance from the center of stress circle to the strength curve of element, and ztfiax(the maximum shearing stress of element) represents the radius of stress circle, the safety capacity of resistant shearing failure of this element is Q,"= r; -rim . According to the geometrical relationship between the main stress and strength characteristic of element, following can be obtained:
Q,'=Ccos$+(oe +o,")sin4/2-(oe -o,")/2 (22) Obviously, if Q: GO, it meant that element e would fail. Because C and cp are random variables, Q," is certainly a random variable. The linear expansion of Q: to stochastic vector
g
-
(25)
-
N
is Q,: = Q,' +cQ,:,iq in a form of Taylor series, r=l
390
+ (0; + o;)cos?/2
where meanings and algorithms of o;,,and o;,i have been given in form of equation (1 3) by PSFEM. In equation (24), if let e=f; the variances of safety capacity of element would be obtained. According to the viewpoint of ultimate equilibrium, the safety coefficient of element would be: _ _ = CcosJ+(o: +of)sin?/2 Fve= ___ - (28) E[rfnax 1 (0; -a f ) / 2
If Q,: distributes normally, the reliable index
pc
and probability of shearing failure P; of element e could be obtained according to following:
P;
= 1- q p ' )
(30)
where q) is the function of standard normal distribution.
stress of element e is shown in Figure 2. It is obviously that cf=(cfl+cf3)/2+ (cf1-cf3)cos(28')/2 and f=(cf1-cf3)sin(28')/2, in which 6=fl+d and d=tan-'(28J(cf,-cfy))/2 and fl is the mean dip angle of tangent of circle-sliding surface in element e. Thus Q: would be:
(xo, y 0)
x tan @ / 2 - (a;-
where
Figure 1. Elements cut by circle-sliding surface.
04) sin(2B') / 2
(3 1)
8' depends only on stochastic vector 6- . 8' is
determined by the stress field of element e. Q,' is also a stochastic variable. Based on the linear espan7 in a form of Taylor sion of QZ to stochastic vector series, the expectation of Q,; would be obtained: - _ E[Q,']= + ((a;+ 0 ; )+ (0; - o , " ) c o s ( ~ ~ " ) }
C,
_ _ -
x tan&/2-(a;
-
-a;)sin(2ee)/2
(32)
The covariance of Q,; and Q,/ could be obtained just like equation (24). But the partial differentials of Q,: should be as follows: Figure 2. Stress and safety capacity of element e.
k-
10"
7-
il
4
Excavation
-
-
30% <
50%--/-
50%
30% --?
10% 1
/(2c0s2 S)
(35)
where the expectation of 8' and the partial differential of 8'to tiare as follows:
Figure 3. Isograph of failure probability of excavation slope.
-
Qe
3.2 Totalfailure probability analysis of homogeneous isotropy system with a certain circle-sliding surface
__
= p e+ae - - = pe+ tan-' {22,"y/(a," - a;)}/ 2
(36)
Assume the most dangerous circle-sliding surface penetrates n, elements and the radius is r (shown in Figure 1). Shearing strength of element e on the surface r i = C +aetan@, where c f is the normal The total moment of resistant shearing MR and total moment of sliding M , and total safety capacity of resistant shearing failure Qs,~ol on the surface can be defined as follows:
stress. If 8 represents the shearing stress, the safety capacity of resistant shearing failure of element e on the surface could be defined as Q,' = ;2 - re. The relationship between the normal, shearing and main 391
the deep part of excavation, failure probabilities are approximate zero. In Figure 3, the position of the most dangerous circle-sliding surface in this excavation could be found easily. Calculations are consonant with practice.
e=I
e=l
where 6, is the length cut by sliding surface in element e. The expectation and variance of Qs,ro[are:
5 CONCLUSION The fundamental of PSFEM is elaborated in this paper. All the equations of the statistical properties of displacement field and stress field and main stress field are given. PSFEM is combined with the MohrCoulomb criterion, so it is possible to apply this method to the reliability analysis of excavation. Equations of the total safety coefficient and the expectation and variance of the safety capacity of resistant shearing failure on the most dangerous sliding surface are perfectly deduced. The reliable indices and probabilities of failure of local elements in system and the total ones on this surface are also given. By an example of the reliability analysis of excavation, it is indicated that this new method could deal with the stochastic physical and mechanical parameters in the forms of the stochastic and spatial vectors and the finite element combined with perturbation method. The calculation time required is reduced remarkably and the results are accurate.
n.
e=l
v a r ( ~ ~= j, ; ~, k ~6 e~6 )f r 2 ~ c o v ( ~ : y ~ : )(41) e=l /=I
where Cov(Q:,Q,/)is the covariance of cut element e and$ In the forms of equation (29) and (30), the total reliable index Pro1 and total failure probability P J Tcan ~ ~be obtained, if Qs,r0/distributes normally.
4 EXAMPLE An example of excavation is to be analyzed, in which clay stratum is the main stratum. The stochastic properties of physical and mechanical parameters of clay are shown in Table 1, in which the elastic module E, Poisson’s ratio v, gravity 3: cohesion C and internal fi-iction angle (b of clay stratum can all be considered as random fields. The sizes of excavation are shown in Figure 3. The reliability of excavation is calculated by PSFEM. The finite element mesh is made up of triangle discrete elements, and the random field mesh is quadrangle discrete elements. Each element of random field involves 2 to 3 finite elements. Figures of meshes of finite element and random field are passed over in this paper. The local failure probabilities of elements are calculated according to equation (30) in the form of percentage. The results are shown in Figure 3 in a form of isograph. It is shown that the failure probabilities of soil elements around the foot and the top of excavation slope are greater than that of any other positions. In
6 ACKNOWLEDGEMENT This paper is supported by the key project of National Natural Science Foundation of China (No. 5973%160). REFERENCES Vanmarcke, E. (1986), Random fields and stochastic finite elements, Structural Safety, No.3: pp. 143-166. Chen Q. & Liu X.B. (1993), Stochastic finite element method and its engineering applications, Chengdu: Press of Southwest Jiaotong University. Xu J.P. & Hu H.T. (1999), Application of perturbation stochastic finite element method in the reliability analysis of bedding rock slope, Chinese Journal of Geotechnical Engineering, V01.21, No.1: pp.71-76.
Table 1. The stochastic properties of physical and mechanical parameters of clay. Scale of Parameter Distribution Expectation Variance fluctuation
E U
C
4
Y
Normal Normal Normal Normal Normal
208(kPa) 1520(kPa2) 0 0.5 25.5(kPa) 6.2(kPa2) 4.5(02) l8.6(“) 0 1.78(g/m3)
0.31m 0.34m 0.32m 0.38m 0.30m
392
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Effects of the back-filling to the stability of a caisson Kouichi Yarnada, Shin-Ya Eguchi & Hiroshi Shinsha Penta-Ocean Construction Company Limited, Japan
Yoshiaki Kikuchi Port and Harbour Research Institute, Ministry of Transport, Yokosuka,Japan
ABSTRACT: It is well known that using the back-filling to a caisson improves the stability of the caisson. But, the extent of the improvement of the stability is affected by the interaction among the back-filling and the caisson and the foundation. To reveal the effects of the interaction the model loading tests are carried out. The shape of the back-filling, the confining condition of the mound, and the loading point are changed in each test case. The applied load, the movement of the caisson, and the deformation of both the back-filling and the mound are measured. Discussions as to the test results on the existing design method, failure mode, and the sliding resistance are made. They show that the existing design method is not enough to explain the effect of back-filling. Finally new design method is presented in this paper. Studies (Y. Ito et al. 1966) conducted in the past enable us to calculate the increase in sliding resistance force due to the presence of back-filling; however, many of the characteristics to be expected when rubble is used are yet to be clarified. In addition, to design efficiently a caisson-type composite breakwater of this kind, it may become necessary to take into consideration deformation behavior of a caisson and back-filling materials and deformation of the mound. To address these problems, it is necessary to get an idea of the behavior of a caisson and back-filling when the caisson slides and to investigate the increase in the sliding resistance force due to the backfilling of rubble. For this purpose, a static loading experiment was conducted using a large, one-tenthscale model. This paper presents the results of an experiment in which a caisson was horizontally loaded as well as the results of the stability analysis, and a new design method based on the experiment.
1 INSTRUCTION Suggested methods for improving the stability of breakwaters by reinforcement are: (1) Increasing the unit weight of filling materials; (2) Increasing the frictional resistance between a caisson and its foundation mound; and (3) Installing back-filling behind a breakwater body. A breakwater reinforced by the method (3) is referred to as a caisson breakwater with back-filling. In this method, the sliding resistance force due to the caisson and the back-filling behind it is used to secure the stability of the breakwater (Fig. 1). In construction work executed in a port, a revetment is occasionally constructed prior to a breakwater. In such construction work, the section of a caisson breakwater is often determined by the conditions of waves prevailing at the time of the execution. In such cases, the caisson width must be matched to the section required when the revetment is completed. For this purpose, part of the back-filled stone work is executed in advance in the aim of increasing the sliding resistance force by means of the back-filling.
2 EXPERIMENT USING A LARGE MODEL TEST When a phenomenon of the prototype is simulated in a scale model experiment, it is necessary to consider the effect of scaling (the rule of similarity), regardless of the type of experiment. However, in a case like the current one where the deformation behavior of crushed stone must be considered, it is desirable to conduct experiments by using a scale as close to
Fig. 1 Caisson-type composite breakwater with back-filling
393
reality as possible; the reason for this is that it is difficult to evaluate effects of the difference in scaling from reality on the behavior of particulate bodies. Due to restrictions on our experiments, we conducted experiments by using an approximately one-tenth scale model (€3. Shinsha et al. 1997). The particle diameters for foundation rubble and back-filled stone used in port construction range roughly from 10 to 100 cin. In the present experiment, however, crushed stone with particle diameters of about 10 to 40 mm shown in Figure 2, artificially crushed to roughly even size, was used as materials for both rubble mound and back-filling. In addition, we conducted a large triaxial compression test, shown in Figure 3, to determine the friction angle @ d, obtaining @ d = 49 degrees for a low confininn stress condition. Figure 4 shows the section and plan of the experimental unit. The scale of the model caisson is 1.6 m in width, 1.8 m in height and 2.0 m in length. Its frame was made of reinforced concrete, and its body is-filled with sands with the density of 2.1 kN/m3. The observation wall was constructed by fixing a 25-mm thick acrylic plate to a steel frame by bolts. Treatments of teflon coating and other means were adopted to minimize the friction between the backfilled stone and the wall surfaces on both sides (Fig. 4 (4). The rubble mound was constructed by gently placing crushed stone without letting it drop from a high position and compacting it in a manner that prevented particle breakage. h
$100 90 6 80 ‘5 70 3 60 50 40 *30 20 5 10 $ 0
I
y
8
--
I
Size Range,rnrn 37.5 19.0
9.5
5
0.001
I
26.5
% 33.1
9.5 4.75
0.1
0.01
1
100
10
Grain Size(rnrn) Fig.2 Grain-size distribution of tested back-filling materials
600
I
4
5
400
200
0
500
1000 a (@a)
1500
Fig. 3 Mohr’s stress circle Fig.7 Average horimntal displacement and inclination of the caisson
394
Fig. 8 Searing strain distribution (Case 1006)
After flattening the upper surface of the rubble mounds, the caisson model was installed. A 1 cm clearance was provided between the caisson model and each of the wall surfaces to avoid generation of friction between them. The back-filling was heaped up in the same way as rubble mounds were formed, with its back formed generally with a gradient of 1:1,2, which is roughly equal to the angle of repose of crushed stone. The average weight per unit volume of the back-filling for the experimental cases was r d= 15.1 m/m3. TO observe the displacement behavior inside the backfilling and mounds, displacement measurement targets (bolts with a diameter of 7 mm and a length of 30 cm) were arranged in a lattice formation with intervals between 10 and 20 cm inside the back-filling. In the current experiment, the caisson was loaded with a concentrated load by using hydraulic jacks (two units) that were horizontally installed at the working height for the wave pressure resultant force determined by Goda's wave force calculation formula. The standard stroke speed for the jack was 1 cm/min in both methods. The loading force was measured by the load cell fixed at the tip of the hydraulic jacks. The displacement of the caisson model was measured by using a wire type displacement meter and a dial gauge displacement meter. The internal displacement of the crushed stone was determined by photographing the displacement of the targets laterally at a fixed point and analyzing the images thus photographed. The experiment was stopped when the slope collapsed. The average horizontal displacement of the caisson at the time when the loading was stopped was about 40 to 60 cm.
3 RESULTS OF THE EXPERIMENT Table 1 lists details of the experimental cases. Each experiment was conducted with the thicker mound that is not constrained. Figure 5 outlines the section of the experimental unit. In all experimental cases, including those in which back-filling was not used, eccentric loading was adopted.
395
Table 1 List of experimental cases
lO0Ol
100
1001
100
1002
I
-
I
I
I
-
-
114
80
1:1.2
92
100
276
80
1:1.2
92
1003
100
18
160
1:1.2
92
1004
100
180
160
1:1.2
92
1005
100
18
160
1:1.823
92
1006
100
18
160
1:1.2
112
'
92
Figure 6 shows the relationship between the average horizontal displacement and the loading of the caisson. Back-filling enabled maximum loading to be increased up to a value four times that of loading for the case without back-filling. This shows that the effect of back-filling is significant. In Case 1000, the caisson tilted during sliding, with the bottom on the loaded side separated upward from the rubble mounds completely. The bottom lower end on the back-filled side was seen moving and locally destroying surface layers of the rubble mounds within a depth of about 5 cm in each case. The friction coefficient between the caisson and the rubble mounds determined on the basis of the maximum value of loading was P = 0.61. Figure 7 shows the relationship between the average horizontal displacement and the inclination of the caisson. During the initial loading period, the amount of rotation of the caisson increased with the loading; however, after the loading had reached the maximum or near-maximum value, the shape of back-filling tended to inff uence, and thereby change, the displacement mode of the caisson. The shearing strain in a triangular element consisting of three neighboring targets was calculated on the basis of coordinate displacement of each vertex. Figure 8 typically shows the shearing strain distribution for Case 1006 at the beginning of loading, near the maximum sliding resistance, and at the time of
the ultimate collapse of the back-filling. In all experimental cases in which back-filling is considered, the prevailing domain for the strain occurring in mounds was not clearly observed for an average horizontal displacement of 5 cm. However, in the vicinity of the displacement at which the loading became maximum, a prevailing domain of strain began to develop from the vicinity of the rear toe of the caisson toward the toe of the slope of the rubble mounds, leading to causing definitive sliding when the ultimate collapse occurred. As described above, calculating shearing strain distributions on the basis of the displacement of targets enables us to get an idea of the state of progressive fracture inside the back-filling and to estimate the slip plane. Figure 9 shows a slip plane assumed to exist in a domain where strain prevails.
4 DISCUSSION
Fig. 10 Predetermined slip plane in the back-filling
In designing a gravity type structure like a caisson, three modes of stability, namely, the tipping stability and sliding stability of the structure and the stability of the ground bearing capacity, are usually examined separately. Among these modes, the tipping stability and sliding stability are examined on the assumption that the structure is placed on solid ground and that consideration of deformation of the ground is not necessary. The stability of the bearing capacity is examined on the assumption that the foundation ground is subject to deformation, being considered as a problem of the bearing capacity at the time when an eccentric and inclined load acts on the foundation ground. Kobayashi et al. (1987) investigated fractures inside mounds exposed to inclined load by imposing horizontal force on the loading plate of a mound in a centrifugal field. In the current experiment, the caisson did not slide in a simple way but was displaced rotating with its rear toe somewhat stuck in the mound. It is concluded that the failure mode observed in the current experiment was due rather to insufficiency in the slide stability than to insufficiency in the bearing capacity caused by eccentric and inclined load; however, local failure of the ground that accompanies eccentric inclination cannot be ignored. This suggests the importance of examining both sliding and bearing capacity when examining the stability.
The horizontal resistance force arising from backfilling, inclusive of the wall surface friction, is given by the minimum value for the horizontal force expressed by the following equation in the conventional method (Fig. 10).
4.1 Characteristics of Resistance to Horizontal Load of a Caisson Type Breakwater with Backfilling
4.2 Examination of Bearing Capacity Stability
where PO,: Passive resistance force due to back-filling; Ws: Weight of the back-filling above the slip plane; 6 : Angle that the slip plane makes with the horizontal direction; 6 : Friction angle of the back-filling material (=49' ); and 6: Friction angle of the wall surface (=1$).
Table 2 shows the relationship between the resistance force increment Pe, due to the presence of back-filling and the passive resistance Peal. Pex values are obtained through the experiments and Pcal through calculation. The ratio Pex/Pcal of the passive resistance force obtained experimentally P,, to the result obtained by calculation Pea, ranges roughly between 0.77 and 1.34 except the case 1002 which is different from the other cases on the shape of backfilling. This result suggests the limits of calculationa1 accuracy that can be expected from the conventional method.
We conducted a stability analysis using the circular arc method for an eccentric and inclined load (Japan Ports and Harbours Association, 1999). The safety factor obtained was FS=0.94 to 1.26, which means that the results of the experiment were not reproduced satisfactorily.
We consider the resistance force in the absence of back-filling as the standard resistance force. The relationship was evaluated between the maximum horizontal load obtained in each experiment minus the standard resistance force and the passive resistance force due to back-filling as shown below.
396
Fig. 11 Load conditions for the new method Width ofcaisson : B Y
K
1.5
rn
1 .o
0.5
Case No. 1004
0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
Fig.12 Slip plane determined by the new method
4.3 New Stability Calculation Method for a Backfilled Caisson Type Breakwater Based on the Circular Arc Method In this section, we examined a new stability calculation method, the circular arc method; this method allows us to consider the stability for the bearing capacity and the stability for sliding at the same time. Figure 11 shows the load conditions in the new method. In the load conditions, the balance between the external force and the moment acting on the caisson is considered. Since the results of the experiment show that the installation width of the bottom at the time of ultimate fracture was slightly over one-half the bottom width of the breakwater body, it is assumed that the subgrade reaction is uniformly distributed in the domain of up to 0.75B of the width of the breakwater body. It is also assumed that the frictional force acting on the bottom surface covers the domain of O.lB, because the results of the ex-
periment show that the slip lines for stone materials originated in the end portions of the caisson in all the experimental cases. Table 3 shows the load conditions and the results. Figure 12 shows the positions of the circular slip planes. The safety factors calculated by the circular arc method range from FS = 1.01 to 1.10, which are closer to 1 than the values obtained by the ordinary method. It is also shown that the slip plane on which the safety factor becomes minimal is represented, in all the cases, by an arc starting from the vicinity of the lower end of the back of the caisson and passing through the toe of the slope. This shows that these results are significantly close to the slip planes estimated in the experiment.
397
Table 2 Evaluating of the passive resistance of back-filling Maxium Load
Increasement of the Resistance Force due to Back-filling
Pmax (kN/m)
Case NO.
1001
Weight of
Passive Resistance of
pex (kN/m)
Back-filling Wb (kN/m)
Back-filling (kN/m)
19.50
16.86
‘ex’
Peal
57.62
13.52
1002
93.59
49.49
38.81
17.93
2.76
1003
62.03
17.93
27.64
23.32
0.77 1.27
0.80
1004
118.58
74.48
66.64
58.51
1005
90.16
46.06
39.59
34.30
1.34
1006
68.89
24.79
27.83
23.52
1.05
REFERENCES
Table 3 The safety factors calculated by the circle arc method Case R )No. “l (
h,
W/0.75B
RI
R2
Safety
Y. Ito, M. Fujishima & T. Kitatani. 1966, On the stability of breakwaters, Report of port and habour research institute Vo1.5 No.14: 117-121, Ministry of Transport, Japan. H. Shinsha, S. Eguchi & Y. Kikuchi. 1997. ; Proc. of 32”dJapan nationarl conference on geotechnicnl engineering., Kurnamoto, 15-17 July 1996. M. Kobayashi, M. Terasi, K. Takahashi, K.Nakashima & H. Odani. 1987. A new method for calculating the bearing capacity of rubble mounds, Report of port and habour research institute Vo1.26 No.2: 371-413, Ministry of Transport, Japan Technical Standards for Port and Harbour Facilities in Japan. 1999, Japan Ports and Harbours Association
(m) (kN/m2) (kN/rn) (kN/m) Factor
1001 19.50 0.81
45.08
32.93
5.19
1.07
1002 58.51
0.89
36.36
26.56
15.68
1.03
1003 24.11
0.84
44.00
32.24
6.47
1.10
1004 85.65 0.90
30.28
22.15
22.93
1.05
1005 54.68 0.89
37.14
27.24
14.70
1.01
1006 31.65
42.34
30.97
8.53
1.01
1.25
5 CONCLUSION The results of the present study are summarized as follows: 1)The presence of back-filling enhances the stability of a caisson-type composite breakwater. The increment and the deformation behavior of the horizontal resistance force in this case are strongly affected by the shape of the back-filling. 2) In a caisson-type breakwater with back-filling, the stability of sliding, tipping, and bearing capacity are not independent, which requires that the three factors be considered together. 3) Considering the passive resistance of the backfilling allows the increment of the horizontal resistance force to be examined roughly, but does not allow the difference in the shape of back-filling to be explained sufficiently. 4)Applying the newly proposed circular arc method enables the stability of sliding and that of bearing capacity to be examined simultaneously. In the current study, the effects of back-filling for a caisson-type composite breakwater were studied by using a large-scale model experiment. In the current study, a static loading experiment was performed as a subject for a basic study. However, such study alone is not satisfactory in examining the design of actual breakwaters, and large-scale experiments using wave channels will be necessary. In addition, the current static loading experiment leaves the clarification of dynamic interactions between a caisson and its back-filling as an urgent subject. 398
Coastal Geotechnical Engineering in Practice,Nakase & Tsuchida(eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Method of probability analysis for breakwater stability Wohua Zhang, Yunmin Chen & Yi Jin Geotechnical Engineering Institute, Zhejiang University,Hangzhou, People's Republic of China
ABSTRACT: This paper presents a method of probability analysis for problems of the breakwater stability under random storm wave loads and random properties of media. The major purpose of h s study is carrying out (a) probability analysis of seepage stability for embankments; (b) reliability analysis of slope-stability of the embankment under the storm wave action; (c) statistical estimation of erosive depths due to washout sediment from the embankment. In analysis, a linerlized Rosenblatt Transformation is applied to determine the reliability index of performance functions for breakwater stability problems. The method of the analysis could provide an advanced basis on the reliability evaluation and design for the breakwater system. 1. INTRODUCTION
2.1 Generalization of safety reliability
The aim of the probabilistic safety analyses is to obtain a quantitative understanding of how the variability or uncertainties in system properties to effect the distribution of the safety performance functions of a random field (Alfiedo, Wilson 1984). The safety reliability analysis has been attempted to develop such a approximation of the effective orthogonal standard transformation for the correlated and non-normal random system using a point of view of the linerlization method respect to Rosenblatt Transformation (Alfredo, Wilson 1984). A large number of factors are known to contribute to the safety performance function for embankment stability and the breakwater failure problems (Chowdhrury et. al. 1987). This study presents an improved modeling of the reliability analysis for a general type of the safety performance function of embankments.
The level of performance of a system obviously depends on the properties of the system. In this context and for the purpose of a generalized formulation, we can define a performance function, or safety function, as g({
X I )
= g ( x , ,x*,... x,
1=0
(1)
where { x ) = {x,, x2 ...,x,) is a random basic state (or design) variable of the system, and the function g({x)) expresses the performance or safety state of the system. g({x))>O represents the safe state and, g({x))
Pf = P( g(l'x))--CLK > - ' 1 " ) = P ( Z > - P ) % ??
2. PROBABILITY APPROCH FOR SAFETY FACTORS
D
I fz(z)dz
= 1-
Fz(-P)
(3)
-8
Evaluation for safety problems of a breakwater (or embankment) stability that is subject to the failure Probability is an task in geotechical engineering. Practical statistics investigations of failed embankments have prcjvided substantial evidence to confirm that failure of embankments is a random process of a random field. Consequently, probability failure reasoning has been applied to explain reliability of the stability of a breakwater system in both drained conditions.
whel.e pgisthe mean and og is the devi,Ltionof the performance function,f,(z) and Fz(z) are the probability density function and probability distribution function of 2 respectively. 2 is the stant/ard variable, which is defined as Z = g u X I ) - Pug
(4)
and the ratio 399
p =P a
2.2 Linearlization of Rosenblatt Transformation
(5)
0- s
can be defined as the Reliabiliiy Index. When g((x)) is normally distributed, the reliability of the system can be represented by Gaussian b c t i o n (D as -P
R = P( Z < -p) =
Jb( Z)dz = O(-p)
(6)
--I)
The position of the failure surface relative to the origin of the standardized parameter space should determine the safety or reliability of the system (Alfiedo, Wilson 1984). Thus the index pas shown in Fig.Z may represent the minimum distance fiom the origin of the standardized parameter space to the limit state boundary (failure surface). Namely, the point on the failure surface with minimum distance to the origin is the most probable failure point. Actually, for the non-correlated multivariate normal space, the joint probability density function will project spherical contours on the n-dimensional plane of the standardized parameter space (Fig.Z for n=2). The symbols R, S, pp o-@ ps o-, in Fig.Z are the supply, demand and their means and standard deviations, respectively. Therefore, in some approximate senses, t h s minimum distance may be used as a measure of reliability. However, this new definition of reliability should be used with care, because this invariant format of reliability index has to be satisfied with the following assumptions €or a non-correlated multivariate normal case. The point on the failure surface, (U*), having the minimurndistance to the origin may be determined by minimizing the distance j3 = ({ U ) { U )) f , subject to the constrain g((x)) 4, that is Minimize j3 = ( { U ) ~ { U ) ) + Subject to
g({x))
=0
Figure 1. Reliability index p by probability contours in the normalized space of the supply and the demand (2,J.J
(7)
From previous section, it can be seen that if random parameters of a system are non-correlated and normal distributed, the probability integration for the reliability can be straightly carried out from minimization of the reliability index p. The most general case of a random system is the correlated and non-normal random. Thus, if we can find out a general transformation, which is able to transfer the correlated and non-normal random system into the nonCorrelated and normal system, the reliability of system will be determined strictly fiom Eqns (7) and (6). Theoretically, such a transformation can be employed based on the Rosenblatt Transformation (Alfiedo, Wilson 1984), but the difficulties to use the Rosenblatt Transformation are the nonlinearly due to the integration of conditional probability for the nonnormal distribution. This study presents a linearlhtion of Rosenblatt Transformation. Lets (x) is a correlated and non-normal random vector and (U) is a non-correlated (independence) and normal random vector. Consider a general transformation T :the vector (x) can be transformed from x-space into the u-space as {u)=r((x)), and therefore, the performance function g((u)) becomes g({x)) =g(T-’((u))) =O under the transformation. The tangential plane of the failure surface through the point (U*) is expressed as
where [ J ] = [ a ( x ) / a ( u ) ] is the Jacobian matrix. In order to linearlize the transformation we may first approximately assume the transformation r has a linear form as
cu,>
{U>= [AI-’ ( (41 or =E 4{U)+ {Px1 (9) where (A is the mean vector of (XI as (pJ=E({x))=(E(x)) , [ A ] is the matrix of the linear transformation. Consequently, the performance function becomes g((x)) =g([A](u)-{&) =O under the transformation. The tangential plane of the failure surface through the point (U*) under the transformation is
Comparison Eqn(l0) to (S), we have ( x* ) = [ A * ] ( u * ) + ( A ) and[A,*] = [ a ( x ) / a { u ) / , = [J ] . In order to define the matrix [A ] to satisfy the purpose for the transformation of K we have to consider the general case when (x) is non-normal and correlated.
400
{U") = rJI"Nx"1- {A)) Eqn(l6) becomes 4 ( u , ) = f , ( x , ) a ,
in which
Figure 2. Modeling of the embankment safety The covariance matrix [C] of {x} is defined as [ C,/] = [ Cov(x,,x)], then the covariance matrix [C 'J of the standard vector {x')={ (x,-pJ/q } can be represented
(1 1) in which a; is the standard deviation of x, , [pX,,,J is the correlation matrix of {x) Matrix [ C ' I or [ p ] is real and symmetric, thus, there is a transformation matrix [T ] to satisfy [T]T[C'/[T/=[nJ=diag[A,S,] (12) where [A] is a diagonal matrix of the eigenvalues of [C'J, The matrix flJ consists of the eigenvectors of [C 'J. Actually, the matrix [T ] is a coordinate rotation transformation, and it has [T ] -I=[ T J .' Consider the standardized transformation, {x 7= [D] -'({x)-{d), where [DJ' is the diagonal matrix as [D] -'=diag [ q,/ a ; ] . Thus, lets {U ?=[ T ] '{x ?=[ T ] '[D]-'({x}-{d) and {U)=[ A ] -'{U 7 , the correlated vector {x) can be transferred into the non-correlated standard vector {U} as {U) =
r~I-"T/"D/-"xZ
- { P x l ) (13)
When {XI is a non-normal and non-correlated, we need to define vectors CF) = {PI ( ~ F,(xj, 3 ~ ...,F, (xJ '1 and (14) If}= l,fi(XJ?L(X3,...?fn(X 4IT in which F,(x) andj;(xJ is the probability distribution function and the probability density function of x, (i=1,2...,n) respectively. When assume cq/y)=F;(xJ and Ku,)=x(x,) (i=192,....n) (15) we have
3. MODELING OF SAFETY PERFORMANCE FUNCTIONS FOR BREAKWATERS This section will present three major performance functions for the breakwater stability design under random loads due to the storm wave action, watercourse washout and effects of random properties on the embankment based on the Code for Design of Levee Project (GB50286-98, 1998). The structure and the geometrical parameters of the analyzed embankment are illustrated in Fig.2. The equations of the boundary lines for the embankment can be geometrical represented as j , ( ~ ) = m , (-x) b, i f b, S x < b 2
1
j ( x ) = j 2 ( x ) =H i f b2 1 x 5 4 (25) j3(x)=ho+-x if 4
401
3.2 Performance functions for analysis of erosive depth due to washout The formulations of erosive depth due to tilted washout and parallel washout can be expressed as hb =H, +[&
23 tan Ahn =
y' = J(x - x,)' - R 2 - y , (27) where x,, yois the coordinate of the slip circle center, R is the semi-drameter of the slip circle. Some useful triangular functions of the tangent angle 9 at the location x on the slip circle can be represented by the circle equation as dx
WO - x)
-'y
,/(x
c o d = (x-x,)2 S(x-x,)2
(28a)
- x , ) ~- R2
-P sine = -F
Yx-x0)2
-P
(
4
f
L
)
l
A, - A l
- 30d
m:g
3.3 Per$ormancekctions for slope stability of the embankment
(28b)
In order to sensitively carry out the probabilistic analysis, the all design parameters presented on the Code for Design of Levee Project (GB50286-98) should be represented (transferred to) by the essential random parameters such as the maximum and minimum tidewater level of storm wave, the stream flow speed, the grain diameter size of washout sediment from the bed of the embankment, the shear strengths of medium.
3.I
2
.-,
where h, is the local washout depth measured fkom the water table. Vcpis the average flow speed. V, is the allowed flow speed in the case of non-washout. n is the shape parameter of the cross section of the embankment. dhpis the local washout depth measured from the bed of the water depth. a is the angle between the flow direction and the bank line of the embankment. d is the equivalent diameter of particles near the slope base. Q is the designed discharge of the flow. A , is the original cross section. A , is the narrowed cross section. If Eqns(30) are considered as the performance functions, parameters, Vcp,n , d , a , Q ,A , ,A, should be considered as random variables.
where H, is the average tidewater level of storm wave. B is the width of the embankment bottom. The equations of the slip circle of the embankment can be geometrical represented as
& *= tan 6 = -
-11
Y,
The performance function for the slope stability of the embankment can be represented fiom the safety factor formulation of the slope in GB50286-98 as
K=
Performance functions for seepage analysis of the embankment
CjcsecB+[ ~ + ~j i () cjo d - ( p-~j.r,) sece/ta+~~x
Es($ +r.)s i e
(31) where c and $ are the shear strengths of the embankment material. is the unit weight of dry medium. yw is the unit weight of water. p is the pore pressure in the embankment. Ax is the width of the block bar. is the first part of the high of the slipped block bar between the water table of seepage and the slip circle in the embankment. y" is the second part of the high of the slipped block bar over the water table of seepage in the embankment as shown in Fig.3. When Ax taken small enough, Eqn(31) can be rewritten in the form of integration along the circlearc C on the slip surface as
vt
The performance h c t i o n of seepage can be carry out by the formulation to calculate the stream volume of flow out due to seepage from the per unit area on downside slope presented by (GB50286-98) as H: -h,Z g=q-qc = m:-2m, - 1 k-9, 2(B + H, +" A ) 2m: +m, (29) where q is the stream volume flowed out due to seepage from an unit area on downside slope of the embankment, k is the permeability of the embankment medium. q, is the design permissive value of q . In this analysis H,,h, and k are considered as random parameters.
402
K = (C-Ptan$)J, +y,tanql/, +Ywtan@3 (32) YxJ 4 in which
J,
=
I
J,
j*cosBdx
=
Lj*sin&
(33)
v’ =y+v* is the total slip high consisting of 1
where
”
7.
and y’ in the embankment. The intergrations in Eqn(33) should be carred out along the length of the arc C on the slip circle. Assume there are only two cross points (x, and x2) produced by the possible slip circle and boundary lines of the embankment. Obviously, there are only three possible regions for the position of the slip circle as shown in Fig.3. The x-coordinate of the two end points on the slip circle in the three cases can be represented by (x,, x2I), (x,,x, I), (x,,x2 I), which can be determined from the simple geometrical relation as presented in Fig.3. The slip circle can be determined by minimization K respect to x,, y, and R as ”
Eqn(32) with conditions Eqn(34) are considered as the performance functions, regarding to Eqn(26), parameters, c, 4, H,, h, andp should be considered as random variables. 4. FACTORS OF STORMY WAVES In above performance functions, parameters H,, yCp h, and Q randomly depend on the factors of stormy wave. The factors of storm wave can be determined regarding to the Code for Design of Levee Project (GB 50286-98, 1998 and Horikawa, 1994) as 0.0018 = 0.13th[O.7(~)”]~h{
V
V
v
} (354
0.13th[0.7(<)”] V
where H,is the average high of the storm wave. T is the average wave period. V is the wind speed. d is the average depth within the water area. g is the gravitational acceleration . t,,, is the minimum time when the storm wave becomes to stable. L is the average wave length. The length of wind area F can be calculated follows the same procedure presented in (GB 50286-98, 1998 and Horikawa, 1994). In Eqn.(35), H,,T ,V: L, d and t,,” are considered as random parameters.
5. NUMERICAL RESULTS The developed model has been applied to cany out the numerical computations of the failure probability for different performance functions of the embankment. The data of parameters necessary for the numerical analysis are calibrated fiom (GB 50286-1998, Horikawq 1994) and the analyzed embankment model is referenced from the example presented in (Zheng Zuzhen 1988).
Figure 6. Probability of the embankment-slip for different variation-coefficients of the internal frictionangle
403
In reliability analysis, the reliability index, ,8 =pg og, for the performance function g has an important operation (Chowdhrury 1987). The value of reliability index is depending on the major random variable. Thus, Fig. 7 can be used to illustrate the effect of the randomness of the friction angle o4 /p4 on the reliability index p for the slope stability of the embankment. Once determine the reliability index, the failure probability as well as the reliability of the system can be straightly carried out by the value of @(-PI in Eqn(6). Fig.8 presents the relation between the reliability and reliability index for the stability of the breakwater under random stormy wave loads and random properties of media. The reliability of a purposed design performance increases with increasing of the reliability index P of the design performance function.
6. CONCLUSION
Figure 8. Relation between the reliability and reliability index for the slope stability
Fig.4 presents the failure probability due to seepage (in logarithm scale) effected by the relative maximum water table (ratio of the maximum high of the water table and the high of the embankment. The lines A and B in Figures represent when the average tidewater level H, is in the typhoon condition and general storm condition respectively. Fig.5 shows when the upper-side slope of the embankment has been washout somewhere due to erosion , the failure probability is strongly effected by the relative mean of the eroded depth through washout sediment from the embankment (i.e. the ratio of the mean eroded depth in the embankment and the high of the embankment). It can be seen that there is significant effect of the washout depth on the erosion failure probability in the typhoon condition. Fig.6 presents the failure probability of the slope stability for the embankment effected by the randomness of the fnction angle of the material, where the randomness is expressed by the variation coefficient of the random variable such as o4/p4here. Whereas, it should be point out that the randomness of the material cohesion has no significant effect on the slip failure probability for tlus embankment.
The failure probability affected by the seepage, the eroded depth due to washout sediment, and the combined slope-stability of the embankment under the stormy wave action can be expressed by a set of performance functions involving the random stormy wave factors, random material properties and random geometry conditions. The reliability index is an important measurement of system reliability. From numerical analysis, it has observed that the erosive depth due to washout sediment has a significant effect on the failure probability especially in the typhoon condition.
ACKNOWLEDGMENTS The financial supports from The Science and Technology Development Research Fund by Zhejiang Province of China are gratefully acknowledged. REFERENCES GB 50286-98. 1998, The National Code of People’s Republic of China, Code for design of levee project, Published by the national t e c h c a l supervision authority and the constructional ministry of P. R. China. Horikawa, K, 1994, Nearshore dynamics and coastal processes, Theory, Measurement and predictive model, Edited by Korikawa, University of Tokyo press, Japan. Chowdhrury, N., Tang, H. and Sidi I, 1987, Reliability model of progressive slope failure, Geotechnique, 37, NO.4,467-48 1. .
404
Alfiedo H-S and Wilson T, 1984, Probability concepts in engineering design, Vol. 11, Secession, rescue and reliability, Hohn Wiley & Sons, New York. Zheng Zuzhen, 1988, The report of analysis for the engineering safety of the breakwater system near Qinshan Nuclear Power Station, pressed by the Management Authority of Qian-Town River of Zhejiang Province, (in Chinese).
405
This Page Intentionally Left Blank
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida(eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Numerical modelling for beach profile Wohua Zhang, Yunmin Chen & Yi Jin Geotechnical Engineering Institute, Zhejiang University,Hangzhou, People's Republic of China
ABSTRACT: This paper presents numerical modelling for analysis of coastal development problem The major focus of this paper is to develop a coupled model from the theory of cross-shore and long-shore sediment transport. The effect of erosion rate of coastal materials on the sediment transport rate has been considered as the porous evolution due to water driftthrough porous. A hybrid approach combining numerical and analytical solution have be used to analysis the development of the coast-line (long-shore) and the coastal profile (cross-shore). 2. MECHANICS OF COASTAL DEVELOPMENT
1. INTRODUCTION Flow and sediment transport near the coastal line are important in relation to several engineering topics like sedimentation and erosion around structures, backfiring of dredged channels, changes in nearshore morphology and long- and cross-shore sediment transport rates. During the last decade the development in beach sediment transport research has changed from simple phenomenological descriptions to sophisticated numerical models (Cullen 1977) in which the flow as well as the resulting sediment transport rate are described in detail of (Hanson and Nicholas 1991). The natural causes of beach erosion are due to deglaciation and greenhouse effect; sever storms, hurricanes and typhoons contribute to beach erosion by amplifying wind conditions, amplifying wave conditions as well as causing sea-level to rise; reduction in sediment supply. The coastal characteristics along the beach will also affect the rate of erosion. If the material along the coast is very hard, such as rock, they will have resistance to mechanical wave erosion and chemical weathering. However, if the material is very soft, such as soil, it will be eroded easily. These natural phenomena greatly inspired us to develop the new concept of mechanics to the area of the ocean engineering for modelling the physics and mechanism of beach erosion problem.
2.1 Sediment transport model In this study, the formulation of the sediment transport model is divided in two parts and the mechanisms of cross-shore and long-shore sediment transport will be described, for both conditions outside and inside the surface zone. The beach profile can vary considerably during a number of years or even a single storm event. The kinematics element of a shoreline change model simply ensures conservation of volume of sediment. The associated equation can be expressed in a very compact mathematical form, however, the form of greatest use is one which relates the time rate of change in sand volume V in a beach profile to spatial gradients in the components (Q, , QJ of sediment transport as
where the coefficients, A, and A, are very hard practically to measured. The cross-shore sediment transport plays an important role in the development of the beach profile and, a model which describes the morphological development can be formulated. The main assumption is that the net discharge in direction parallel to the coastline is zero. The morphological consideration of onshore/offshore sediment transport model can be taken into account by calculating the variation of sediment transport across the profile (see Fig.] and Fig.2). From the 407
sediment transport field, the development of beach profile can be analysed by continuity equation for the sediment in the cross section as aD
1
8%
=
0
%F+mF
(2)
where D = D (x , y , t ) is the water depth from the still water surface (coastal bed level ). psis the mass density of coastal bed, Qy is the sediment transport rate in the cross-shore section. a * is the effective porosity of the eroded medium, and it can be assumed by an appropriate form, wherein it may be suggested as, a* = a 1 ( 1 - y a ), where a is the porosity of the uneroded medium, y is the erosion parameter. The long-shore sediment transport is often manifest itself through the coastal erosion or accretion around coastal areas. If the beach is long enough, the accretion and erosion should continue and the coastline may move offshore on the up-drift side. The change in shore-line position Y (x , t ) can be calculated from the equation of sand conservation near the shoreline (3)
where D' is the active water depth of the beach profile near the coastal line. Q, is the sediment transport rate in the long-shore direction. It is evident that Q, and Qy are components of the sediment mass transport rate.
2.2 Modelling for the transport rate of eroded sediments The transport rate of the eroded sediment mass per unit area on the beach bed can be derived from the mass conservation based on the topographical change of beach bed as shown in Fig.2.
Figure 2. Topographical development of beach profile Lets consider an elemental volume along the shoreline with a bottom area of AY. The most interesting for the problem of coastal development is to study the topographical change of the coastal bottom near the shoreline. Therefore, the analysed elemental volume should be taken near the shoreline with thick , Ay = y - f . The size of Ay can be taken as need as possible, because of that the size along shoreline can be considered as infinite. Thus, at the time t,, the water depth from the still water level to the surface of the beach bottom is D = D (x , y , t , ), and the elemental volume of sea water based on the elemental area AYcan be expressed as V , = -I [ D + ( D - -l3D Ay ) ] A s (4) 2
dY
where the elemental base area ds equal to Ay times the length of the elemental volume along x direction. At the time t, = t, + A t , the beach depth becomes to D=D(x, y, tJ, and the elemental volume changes to + -a( D at
-
-aD Ay
dY
) A t J AS
(5)
Eqn.(S) minus Eqn.(4), we have the volume changed due to the development of the beach depth during At as
Figure 1. Illustration of sediment transport near the beach
and this changed volume should equal to that of topographical profile of the beach bed. Neglecting the higher order term , ( d'D / dyd t 1 dydsdt and substituting Ay = y - Y . Eqn.(6) can be rewritten as
408
2.3 Modelling for beach topographic development where d Y / dt, is the speed of shoreline change as presented in Eqn.(3), d D / d t is the speed of crossshore profile as presented in Eqn.(2). d y / d t actually is the drift velocity of watedsediment particle on the surface of the beach bottom near the shoreline along y direction, it should equal to the speed of sediment transported on the surface of the beach bottom near the coastal line due to stormy wave wash away. The drift velocity can be assumed to equal to V b , the time-averaging speed of water /sediment particles on the surface of the beach bed over one wave period as, dY = -
dt
vb
Substituting Eqn.( 11) into (10), we have p -a f - aD p, -ao c "( ' b )at 2 at ay
-
-- M c -
[
2-b
(La*) ( l - S z ) Y ~ ,
-11
(12)
This equation presents the model of coastal topographic development due to erosion and sedimentation of the beach medium, which is coupled with the rate of the coastline development, the average shear stress on the bottom surface and the average energy flux per unit of wave crest. On a long shoreline coast the long-shore sediment transport rate, Q, , can be determined from the wave climate, i. e. statistics for wave height and direction. If the coast is given a different orientation, the entire calculation can be carried out once more, and in this way the long-shore sediment transport rate can be determined as a function of the coastline orientation (Kamphuis 1991), such as Q,= Q,(d Y/dx).
(8)
The eroded mass transport rate per unite area on the surface of the beach bed near the shoreline can be defined as -dm - p , ( l - a * ) - - dV 1 (9) dt dt AS Substituting Eqn.(7) into (9), we have
This is a parabolic partial differential equation which can normally be solved numerically for the complex initial and boundary conditions. The establishment of the coast orientation and the long-shore sediment transport rate will require a large number of individual calculations of the longshore sediment transport rate. It will be not possible to establish any analytical solutions, instead a data base table can be established with corresponding values of Q, and d P / 8 x (Fredsoe and Deigaard 1992). One example for long-shore sediment transport rate can be assumed in the forrn as
Considering the effect of sedimentation (damage) and erosion parameters on material properties an erosion criterion can be developed from the expression presented by Dyer (1986) as -
where *b is the average frictional stress on the surface of the beach bed . ( I - 0 ) zc is the effective critical shear stress of eroded sediment medium on the surface of beach bed. The effective stress has a similar manner of that due to damage parameter presented by Zhang Wohua and Valliappan (1998) and (Valliappan, Zhang Wohua 1996). The effect due to erosion and sedimentation of coastal medium is expressed by the erosion parameter, y , and sedimentation parameter, R , respectively. The coefficient M, has an unit of the erosion-transport rate (transport of eroded mass per unit area per unit time), and it varies with other factors such as temperature and the presence of organic matter. Ariathurai and Arulanandan (1978) have investigated the relationship between M, and the action exchange capacity, sodium adsorption ratio, pore fluid concentration and temperature. Values were generally in the range 0.005 to 0.015, but varied particularly steeply with temperature, being greater at high temperatures
ak
Q, =a, (bq--,P
(m3/year)
dX
(14)
where a,, b, and /z are the coefficients to be filled from the observed data. Reference (Fredsoe and Deigaard 1992) suggested a linear expression as
Q,
a?
= 2.27xld(0.15--)
dx
(,'/year)
(15)
It should be noted that the model of the beach profile developed herein, Eqn.( 12) and (13), have provided a method which couples the cross-shore and long-shore sediment transport models and the total simulation may possibly to be carried out by the interaction between the beach profile and coastline development.
409
V
=
v
dv
dV [vdt + 7[ w d t +7 i3y 8.2
The last two terms in Eqn.(lS) can be easily evaluated by applying the linear wave theory (Fredsoe and Deigaard, 1992) for the orbital velocities giving
Figure 3. Drift velocity as seen from a Lagrangian point of view
v
3. WAVE SCOUR ON BEACH PROFILE
I = v i
2c
nH T
cosh2[k( D - z )J -Isinh2[k( D - z )J 1 (19) sinh2(kD) where H is the wave height; k is the wave number; T
The intent of this section is to provide a background of wave scour based on the hydrodynamics of water waves. Wave in the ocean actually serve as a mechanism which can abstract energy fiom wind system, store it in the form of potential and kinetic energies, and transmit it toward shoreline. The dissipation of wave energy occurs near shore in a relatively narrow zone.
is the wave period; c is the wave propagation speed. Close to the bed ( z = D ), the mean drift velocity on the surface of bed becomes v b
3.I Wave drift velocity
in which V , is the maximum horizontal orbital velocity on the bottom of the beach bed
-
v 2
v = -Hn
In relation to sediment transport, it is essential to distinguish between the mean velocity V measured at a fixed point and the dnft velocity V (masstransport speed) which is the mean velocity of a fluid particle over a wave period. The drift velocity V is always ,positive relative to the mean velocity V (Longuet-Higgins, 1970), the reason is as following: A fluid particle will stay longer below the wave crest than below the wave trough, the fluid velocity is positive below the crest and negative below the trough. The particle path is elliptic in shape, with the particle travelling forward at the upper part of the orbit and backwards at the bottom of the orbit. At the top of the orbit, the velocity is slightly higher than at the bottom, resulting in a small positive contribution to the drift. The instantaneous drift velocity v can be evaluated by Lagrangian description (LonguetHiggins 1970) as follows: consider the points P and G, where P is a point on the orbit of a particle, the mean position of which is G, (see Fig-3) and, the difference between the instantaneous velocity at P and at G is given
T tanh( kD)
3.2 Wave energy The average energy flux per unit of wave crest through a fixed vertical plane parallel to the wave crest near the shoreline (Collins 1976) is
E,=
IId~P,~hdt
P T
-?
c
(22)
D
where p is the mass density of water, 7 is the wave surface elevation from the still water level, CP is the wave potential function. In the case of a linear periodic progressive wave, an example was suggested in (Fredsoe and Deigaard 1992), and the integration of Eqn.(22) gives I 2kD E,, = - - - p g H ’ C [ I + I I6 sinh (2kD) The friction stress on the bottom surface of the beach bed can be evaluated by the drift velocity of waterlsediment particles as where cf is the friction factor. The average work done by this friction stress is
where Az and Ay are the horizontal and vertical displacements of P from G, and can be given.
Ay = Jvdt
where subscript b indicates the value on the bottom boundary. The variation of energy flux equals the rate of energy dissipation (Collins 1976),
Az= Jwdt
(17) There v, w are the velocity components of fluid particle in y and z direction, Substituting Eqn.(l7) into (16) and time-averaging over one wave period gives 41 0
Substituting Eqn(25), (26) into (24), the friction factor cf can be determined as
Thus, the friction stress can be represented by the average energy flux and driftvelocity as
4. FINITE ZIFFERENTIAL MODEL The finite difference equations for the topographic development and coast-line development can be written as
where
Hn: I {- T tanh[ k D ( x , , y j , t ) J
in which
[f(xi+l,t )-2f(xi, t )+f(xi-l,t)/ =o where i =1,2,3....(I j=1,2,3 ....M.
41 1
J
Eqn(29) to (33) consist of (N+l)*M first order non-liner ordinary differential equations respect to time. It can be solve by Runge-Kutta method under the given initial water depth fiom the still water table to the surface of the beach bed, D( x, , y, , 0 ,l , and initial position of coastal-line, f ( x, ,O,l .
From above quantifying, it can be seen that the system equations are described both in space and time domain. A wind dependent wave load as an input has to be take into account, and a special boundary treatment for the cross-shore beach profile and long-shore coastline development have to be coupled.
5. NUMERICAL RESULTS The developed model has been applied to simulate a three dimensional beach evolution. The simulated region and the initial topographic surface of the beach bottom are plotted in the form of a three dimensional graphics with a two dimensional contours on the top of the topography figure as shown in Fig.4. The analysed region is taken as 5~4(Km)'along the x-direction (shoreline or coastal line) and y-direction which is from the shoreline to sea. The necessary wave data, near-shore current date as well as the beach material data and sediment transport data for analysis in this model are referred from (Dyer 1986, Fredsoe and Deigaard 1992). The simulated results are plotted also in Fig.5 and Fig.6 similarly to Fig.4. Fig. 5 shows the developed bottom topography due to the effects of the general wave and the near-shore current after 24 hours. Fig.6 shows the developed bottom topography due to the effects of the stormy wave and stormy tide in the typhoon condition after 24 hours. From comparison, it can be seen that after a strong storm the bottom topography of the beach profile has significantly changed from that before the storm events. Therefore, the mechanism of beach profile development should be considered in two phases, one is the long-term effects due to the general wave and near-shore current, the other is the single effect due to a storm event.
6. CONCLUSIONS The developed modelling in this paper presents a feasibility for numerical simulation of development of beach profile (erosion / deposition). The advances of this modelling is the naturally coupling among the cohesive-erosion theory of porous medium, sediment transport theory and hydrodynamics of sea wave. The coastal morphological development model presented in this paper has provide a way that the cross-shore and long-shore sediment transport models have been geometrically coupled, and the simulation of coastal-topography development can be carried out by a full description of the interaction between the beach profile and coastal line development. 412
ACKNOWLEDGEMENTS The financial supports from the Science and Technology Development Research Fund by Zhejiang Province in China are gratefblly acknowledged. REFERENCES Ariathurai and Arulanandan 1978, Erosion rates of cohesive soils, J. Hydraulic. Div, ASCE, 104, HY2,279-283 Collins J. 1976, Wave modelling and hydrodynamics, Beach and Near-shore Sedimentation, Based on Symposium, (Edited by Eichard D. J. and Etbington R. L.), 55-68. Cullen P. 1977, Coastal Management in Port Phillip, Coastal Zone Management Journal, Vol. 3(3), 291-305. Dyer R. K.1986, Coastal and Estuarine Sediment Dynamics, A Wiley-Interscience Publication, New York. Hans Hanson and Nicholas K.1991, Numerical simulation shoreline change at Lorain Oho, J. of Waterway, Prot, Coastal and Ocean Engineering, Vol. 117, NO. 1 , -18. Fredsoe J. and Deigaard R.1992, Mechanics of Coastal Sediment Transport, Advanced Series Ocean Engineering - Volume 3, World Scientific Publishing Co. Pty. Ltd. 341-347. Kamphuis W. 199I , Alongshore sediment transport rate., J. of Waterway, Port, Coastal and Ocean Engg., ASCE, 117(6), 624-641). Longuet-Higgins, M. S. 1970, Long-shore currents generated by obliquely incident sea waves, l., J. Geophysics. Res. 75(33), pp 6778-6789. Zhang Wohua and Valliappan S. 1998, Continuum damage mechanics theory and application-part I: theory, part 11: application, Int. J of Damage Mech., Vol. 7,250-273,274-297. Valliappan S. and Zhang Wohua 1996, Numerical Modeling of hiethane Gas Migration in Dry Coal Seams, Int. J. for Numerical and Analytical Methods in Geomechanics, 20,571-593.
3 Improvement of soft ground by consolidation and compaction techniques
This Page Intentionally Left Blank
Coasfal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Kinking deformation of PVD under consolidation settlement of surrounding clay H.Aboshi, Y. Sutoh, T. Inoue & Y. Shimizu Fukken Company Limited, Hiroshima,Japan
ABSTRACT: Kinking deformation by buckling of PVD (Prefabricated Vertical Drain) due to consolidation settlement of clay layer around it, is discussed in this paper. A new testing device is manufactured for the purpose and deformation properties of two-kinds of PVD during consolidation of the surrounding clay are compared with each other. It is concluded that a fibredrain which is made of natural fibres only, can sustain the vertical permeability during a larger settlement strain up to 24% and on the contrary, a PVD made of plastics loses its permeability before 19% strain. The field performance of both materials is also referred.
1 INTRODUCTION Around the earliest years of the 1960’s, card-board drains invented by W.Kjellman of Sweden in 1937, were introduced to Japan. New drain-driving machines were manufactured under the guidance of the first author, which could execute drains down to -20 m deep. Paper drains, nominated by the same author, were used to stabilize a total of 10 million m2 of newly reclaimed lands for the Mazda Automobile Co. and Nippon Steel Pipe Co., both in Hiroshima in 1963. These cases were the first practical usage of PVD in the world. (Aboshi et al., 1965) (Aboshi et al., 1969) Though the method were accepted favorably in general, it was found from the earlier cases that there existed a fatal defect in PVD which could not be avoided. PVD could not reduce their length with consolidation settlement of the sorrounding clay layer. Fig.1 is a result of the check boring after consolidation at Mazda, the first case of usage of this method. It clearly showed that paper drains were effective in the dredged silt layer, and on the contrary, unsatisfactory in the natural alluvium of silty clay. Fig. 2 shows the deformation of paper drains after consolidation in both strata. In the dredged clay layer in slurry state, the paper drain can simply move laterally and undergoes deep bending, and as a result, the gutters or the water striae in the drain remained effective. In the natural clay layer, the drain could not move laterally, only reduced its length by buckling in a zig-zag form, losing its longitudinal permeability. Fibredrains invented by Prof. S.L.Lee and his associates in NUS seem to be sustainable under the
shrinking deformation to a certain degree, judged from the past experience on PVD. An experimental study was planned to investigate the consolidation properties, based on the theoretical analysis. (Aboshi, 1999(1)) (Aboshi, 1999(2)) (Hanai et al., 1999)
415
Fig.5 shows the stress-strain diagram of clay in different conditions. In natural clay layers the gradienta is always very large even in soft clays. On the contrary, a in case of slurry clay or remoulded clay is very small. As P in the above-mentioned equation equals tana , the half wave length in buckling of PVD is directly affected by a . This means that the consolidation test of PVD in a slurry state clay, frequently carried out in the laboratory has almost no meaning, from the view point of explaining deformation in natural clay ground, even if a very high consolidation pressure is applied. In our experiment using undisturbed natural clay samples, the half wave length of a plastic made PVD widely used in the world, calculated by using the measured a and EI becomes only 2.64cm. In order to make clear that this value is correct or not, the present experimental study had been performed. EI of fibredrain is very small and C/n for it is practically null.
Figure 2. Deformation of paper drain 2 DEFORMATION OF PVD IN NATURAL CLAY LAYER Fig.3 a) shows a PVD executed in a natural clay layer. When the layer thickness is shortened by consolidation, the PVD must be shortened at the same time. There is a generated force Q as the reaction. Fig.3 b) shows a cross-section at the center of PVD. As the earth pressure on both sides of PVD is always balanced, and so, whether this PVD can move laterally or not becomes the same problem with this bearing capacity problem. As is widely known, the bearing capacity q, of clay ground is 5.5 c, from Terzaghi’s equation when the load is on the surface of the ground, and about 9 or 10 times c, in very deep place such as a pile tip. This is the same case with the deep foundation. The reaction by bending of PVD can not at all exceed such bearing capacity, except the case of slurry state mud. In conclusion, the shrinkage of PVD becomes the same problem with the buckling of a column on elastic foundation shown in Fig.4. This problem had been solved in the 1920’s by Prof. Hayashi of Kyushu University and written in any textbooks of structural engineering. Here is shown the buckling load.
And the most important point is the half wavelength of sinusoidal buckling curve shown as follows.
3 PVD TESTING DEVICE AND TEST RESULTS Fig.6 shows the general outline of the newly developed testing device. The consolidation sampler, used for preparing samples for the testing device, has the inner diameter of 30 cm and height of 40 cm, and it is a two-half divisible cylindrical mould. At the time of sampling, an undisturbed specimen was taken by driving into the clay ground the cylindrical sampler, equipped with the collar at the upper edge and with the cutter at the lower edge. After the sampler was brought into the testing laboratory, its mould covers were released and the cylindrical specimen was cut into two half-cylinders. At the middle portion of the half-cylinder shaped specimen surface, a groove whose cross section is almost the same as that of the PVD to be tested was excavated using a saw. Into this groove the PVD material was inserted for testing. The specimen thus prepared was assembled as shown in Fig.6 and subjected to consolidation testing. At the surface of the specimen, an acrylic glass plate was attached to make it easier to inspect the deformation condition of PVD during consolidation settlement. From the results of consolidation tests, conducted on two similar undisturbed samples (natural moisture content:51.5%, plasticity index:42.0, consolidation yield stress: p , =78.4kN/m2) using prefabricated vertical drains namely fibredrain (hereinafter FD) and plastic board drain (hereinafter PD), e logp relationship was derived as shown in Fig. 7. The final settlement strain was E~ = 24% for FD material and E~ =19% for PD material. Furthermore the dissipation of pore pressure (with the pressure transducer, installed at 5cm apart from the drain section of lower plate) versus time is
-
l l n : half wave length EI : rigidity of PVD /3 :elastic modulus of surrounding clay ground
416
shown in Fig.8. This relationship was obtained when the consolidation pressure was increased from p =785 kN/m2 to 1422 kN/m2. For the case of FD material, the pore pressure converged to zero at the time of EOP consolidation. In contrast of the case of FD material, the case of PD material, it was found that approximately 5 m water head remained at the pore pressure transducer although the material was under the secondary compression stage. The permeability of PD at the end of testing of 19% strain, measured directly by flowing water in PD, showed almost zero.
The drain materials were taken out after the end of consolidation testing and their deformation were compared. The photograph is shown in Fig.9. The FD material was shortened and the lateral deformation was hardly observed. But instead, its thickness got bigger from approximately 9mm to 10-12mm. On the other hand, the PD material was crooked at three or four spots. It is speculated that these crooks caused the occurrence of residual pore water pressure, and no permeability at the end of the test.
Figure 5. Stress and strain curves in clay 41 7
Figure 9. Photo of PVD deformation after consolidation
4 EXAMPLES OF FIELD PERFORMANCE OF PVD METHOD The case studies of field performance of the fibredrain method, practiced in Japan are as follows. In the first case, the ground improvement works, executed at the newly reclaimed seaside land for the development of Ujina Port and its surrounding area are shown in Fig.10. For the same objective of ground improvement, two methods, the sand drain method (drain diameter: d , =40cm, drain pitch: d =2.5m and drain arrangement: rectangular shape) and the FD method ( d =l.lm,drain arrangement: rectangular) were executed and compared. The observations revealed that both methods yield settlements which are in proximity with those of the con-
solidation theory. Fig.11 shows the results from the second case study, conducted in Kanagawa Prefecture. The soil layer subjected to ground improvement was mostly made up of organic materials. The FD materials were driven into the ground in a rectangular pattern with a 1.3m pitch. This also had settlements approximately equal to those of the theoretical computation. Fig.12 shows a case of a plastic-made PVD used in the extension project of The Izumo Airport in the lake Shinji, near Matsue City. (Shimane Prefecture 1993) Though the settlement seemed to proceed as expected from the theory in the earlier period, it might have changed to the secondary consolidation after around 80% consolidation. 418
Figure 11. Comparison of computed and measured settlements (Kanagawa prefecture)
5 CONCLUSIONS Contrary to an ordinary expectation, PVD can not be bent deep in a natural clay layer during consolidation of the layer, even in case of very soft clays. From theoretical analysis, assuming the shrinking deformation of a PVD as the buckling of a column on an elastic foundation, it is found that the halfwave length of sinusoidal buckling curve of commonly-used PVD made of plastics, is only 2 or 3 cm order. In order to ascertain this phenomenon experimentally, a new testing device is developed for the purpose, and the consolidation tests using undisturbed natural clay specimens with PVD in the center are performed,
Sand drains can deform laterally without losing vertical continuity under consolidation and as a result, can sustain their function as drains in larger strain. PD cannot shrink themselves and crook or kink at certain intervals by buckling, losing their vertical permeability in larger strain deformation. There must be a certain limitation of usage in term of consolidation strain for each PD. Fibredrain has an intermediate nature between SD and PD, without any kinking deformation and is still alive as drain until 24% strain. Examples of field usage of FD and PD are shown in the last chapter. It seems that the abovementioned characteristics are shown in these field measurements. 419
Figure 12. Comparison of computed and measured settlements (Izumo airport)
REFERENCE Aboshi H. et a1.,1965. On Paper Drain Method, Soils and Foundations (in Japanese), Vo1.13, No.6, pp.3-10. Aboshi H. et a1.,1969. Stability of Soft Clay Foundation underneath Embankment Consolidated by Means of Card-Board Drains, Soils and Foundations, Vo1.9, N0.2, pp.1-14. Aboshi H, 1999. (1) On Some Problems of Consolidation and Soil Stabilization in Soft Clays, Symp. on Innovative Solutions in Structural and Geo-technical Engg. (Bangkok) pp.241-250. Aboshi H, 1999. (2) On Some Problems Concerning Soil Stabilization with Vertical Drains, Pre-Conf Syrnp. 40'"Anniversary AIT. pp.175-189. Hanai et al., 1999. On Japan's Recent Development of Soil Stabilization with Fibredrains and Deformation of PVD due to Consolidation Settlement of Surrounding Clay. Conf Symp. 4dl' Anniversary AIT. pp. - 21-E - 28. Shimane Prefecture, 1993. Record of Construction of Izumo Airport.
420
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 I
Settlement behavior of improved soils using the packed drainage procedure N.Arita, H.Takahashi & A. Shibata Kowa Company ltd, Niigata, Japan
T. Sasagawa & H.Yamada Niigata Prefecture, Japan
T. Shogalu Department of Civil Engineering, National Defense Academy, Yokosuka,Japan
ABSTRACT: The change in index, mechanical properties and settlement behavior of improved soils with 13 cm diameter packed drains is quantitatively discussed. The settlement .plates were emplaced before constructing the sand mat and packed drains. Therefore, settlement behavior throughout the whole process, including loading pre-load, can be measured. The natural water content and the compression index decreased about 20 %, the unconfined compressive strength increased about 10 % and the preconsolidation pressure increased about 70% by using driven packed drains with a sand mat. The estimated settlement increased about 15 % by taking into acount the effect of sample disturbance and the prediction accuracy of the final settlement improved. 1
INTRODUCTION
Packed drainage is one of several vertical drainage procedures used in improving soil by consolidation. It is used with a pre-loading method and hastens the consolidation of clayey layers. At first, it was thought that the consolidation and/or the change in the mechanical properties of clayey layers was not progressing satisfactorily with only vertical drainage although the consolidation and/or the change in mechanical properties of improved soils with a sand drain and/or sand compaction piles were reported by some researchers such as Akagi, 1981; Okabayashi, 1983; Chai et al, 1998; Matsuo and Honjyo, 1999, et al. However, the measurement of settlement behavior in vertical drainage construction is very difficult, due to settlement plate shifting problems caused by the construction of the packed drains. Akagi (1981) reported changes of index, strength and consolidation properties caused by making sand drains. Okabayashi ( 1983) measured the settlement behavior using a settlement plate before packed drainage construction and he reported that the amount of consolidation settlement in packed drainage construction, after building a 50 cm thick sand mat, was 50 cm. However, the changes in the mechanical properties of clayey layers were not described in his paper. In this paper, the changes of index, mechanical properties and settlement behavior of improved soils, with packed drains 13 cm in diameter, are quantitatively discussed. The settlement plates were emplaced before constructing the sand mat and packed drains. Therefore, settlement behavior
throughout the whole process, including loading preload, can be measured. The unconfined compresion and oedometer tests were performed on an undisturbed sample taken three times; First, before construction of the sand mat. Second, after driving the packed drainage piles. Third, after loading the preload. 2 OUTLINE OF GROUND CONDITIONS AND THE LOCATIONS OF SOIL SAMPLING The banking site is located at Kitakanbaragun Nakajo town in the northeast part of the Niigata plains. The Niigata plains deposits are thick clay deposits at the back of the sand hills along the coastline and consist of an upper layer and a lower layer to 40 meters with both layers separated by a sand layer ranging from 5 to 10 meters below the ground surface. The lower clayey layers are identified as Hc2, Hc3 and Hc4 from the point of view of soil properties as shown in Fig.1. The pre-loading with packed drainage procedure is used as a countermeasure for unstable soil conditions in new road construction. The layout of the packed drainage and the locations of soil sampling along the longitudinal section of the planned road are also shown in Fig.1. The elevation of the planned road increases at the bridge across the river. The depth and spacing of the packed drainage and the height of pre-loading differ according to the elevation of the planned road as shown in Fig. 1. In practice, the general procedure is to set up the settlement plates after driving the paced drainage piles. However, in this particular research,
421
the settlement plates were emplaced before driving the packed drainage piles to estimate the amount of consolidation settlement caused by the construction of the sand mat and packed drainage. To estimate the mechanical properties of clayey layers affected by the driving of packed drainage piles, the undisturbed samples were taken at the locations shown in Fig.1. The sampling was done at seven different sites before construction of the sand mat under original soils conditions (0),at one site 400 days after driving packed drainage piles ( and at another site 230 days after loading pre-load (A).Field sampling was done with a 75 mm inner diameter, fixed piston sampler (JGS 1221- 1995) to enhance sample quality. The ground surface settlement was measured automatically by a water pressure sensor at two sites (0)as shown in Fig.1 (Arita et al, 1998). Note : The water-level at these sites is located at about ground level year-round.
3
is small. Therefore, the index and mechanical properties for the horizontal direction of the clayey layers are relatively homogeneous. Figure 3 shows the test results for the undisturbed samples before sand mat construction, after driving packed drainage piles and after loading pre-load. The curved lines are determined by a cubic equation for each plot. These plots and curves show clearly that the void ratio (e) and w, values decrease and the P E 5 0 , E and 4, values increase with progressive cosolidation. Figure 4 shows the relationship between the ratios of w,,E E 5 0 and q, to the depth for the sample before sand mat construction, after driving packed drain piles and after loading pre-load. These ratios are obtained from the values at the same depth for each curve as shown in Fig.3. The ratio between these values determined after driving packed drain piles compared to those prior to sand mat construction for the Hc2, Hc3 and Hc4 are about 0.8, 1.1, 1.6 and 1.1 respectively. The Rq, value prior to sand mat construction is about 1.9 after loading pre-load. The increasing undrained strength due to 13 cm diameter packed drains is advantageous for pre-loading stability in soft clayey soil. Akagi (1981) reported that Rw,=0.9 two months after sand drain installation and Rq,=0.4 immediately after installation of 30 cm diameter sand drains and 0.65 two months later. Matsuo and Honjo (1999) also reported that Rw,=0.9 and Rq,=3.5 five months after installation of 70 cm diameter sand compaction piles. f,
CHANGE IN MECHANICAL PROPERTIES CAUSED BY DRIVING PACKED DRAIN AGE PILES AND LOADING PE-LOAD
3.I Change of strengthproperties
Unconfined Compression Tests on the undisturbed samples were performed according to the Japanese Industrial Standard (JIS A 1216-1993) for unconfined compression tests of soils. The secant modulus (EJo)is found by the equation (q,/2)/ E 50, in which q, is the unconfined compressive strength and E 5o is the strain at the value of qJ2. The results of UCT before sand mat construction are shown in Fig.2. The natural water content (w,)of clayey layers Hcl and Hc3 decreases about 50 % and those of Hc2 and Hc4 decreases about 10 %. It can be judged from Fig.:! that the effect of the different sampling sites on w,,wet density ( p J and q,
3.2 Change in consolidationproperties
The oedometer tests were done using a load increment ratio of unity and the loading duration for each load increment was one day. The values of compression index (Cc) and preconsolidation pressure ( (J 'J were determined from the e - log (J ',, curves corresponding to 24 hour compression.
Figure 1, Outline o f ground c o n d i t i o n s and t h e locations o f soil sampling. 422
Preconsolidation pressure based on Casagrande’s method has human errors concerning engineers’ opinions on the location of the maximum curvature point of the curve for e - log cr curves (Shogaki et al, 2000). Therefore, Mikasa’s method was used to identify pre-consolidation pressure on the 24 hour e - log cr curves. This method has been employed in Japan as the Japanese Industrial Standard (JIS A 1217-1990) for determining one-dimensional consolidation properties of soils. Figure 5 shows the index and consolidation properties arranged as in Fig.3. The w, and Cc decrease and the p I, E ,, and cr ’ p increase with progressive consolidation. The volumetric strain ( E ), is determined by the following equation (Shogaki, 1996);
(Schmertman, 1955; Shogaki and Kaneko, 1994) and the E ,,value becomes larger. Therefore, the E ,, value can be employed as an index in order to express the effect of sample disturbance (Shogaki, 1996; 1999). Figures 6, 7 and 8 show the relations of e - log cr log c, - log si and log mv- log 5 for the undisturbed samples prior to sand mat construction, aRer driving packed drainage piles and after loading pre-load, in which 5 is the mean value of cr . The e values decrease and the CT ’ p values increase with progressive consolidation. In the over consolidation range, the c, values decrease and the m, values increase with progressive consolidation. However, there are no certain tendencies in the normally consolidated region. The ratios of e,, E w, Cc and cr ’ p after driving packed drain piles and loading pre-load to the e,, E w , Cc and cr ’ p for the sample before sand mat construction are shown against the depth in Fig. 9. These ratios are obtained from the values at the same depth for each curve as shown in Fig. 5 . The Re, R E , RCc and R cr ’ p after driving packed drain piles to those before sand mat construction, for the Hc2, Hc3 and Hc4, are about 0.9, 1.7, 0.8 and 1.7 respectively. The mean value (R cr ’ p ) of R cr ‘p value is about 1.8 for after loading pre-load. For sand mat construction and driving
eo- e,
E = ,
- x loo(%) 1 + e,
in which the e, and e, are the initial void ratio and the void ratio under the effective overburden pressure ( cr of the specimen. In in-situ soil under the CT value, there was no sample disturbance. Therefore, the E ,,value is 0 since the e, value is equal to the e, value. The e, value decreases with sample disturbance because the void ratio decreases with sample disturbance under the same cr value
Figure 3. The results of Unconfined Compression Test.
423
Figure 4. Relationship between z and R .
Figure 9. Relationship between z and R .
packed drain piles 13 cm in diameter, the change in consdidatiQn properties can be measured quantitatively. Akagi (1981) reported the Rc, =0.24 two months after installation of 30 cm diameter sand drains and Chai et a1 (1998) reported the Rk=O. 1 for 12 cm diameter packed drains.
4
Figure 8. Relationship between m , and
7'".
CONSOLIDATION BEHAVIOR AND ITS PREDICTION
Figures 10 and 11 show (1) the relationships between the height (I?) of the sand mat and preloading, (2) consolidation settlement (S) of soil under the embankment observed at the settlement plates A and B and (3), the time (1). The solid, dashed and dotted lines in Figs. 10 and 11 represent the relationships between the estimated settlement before fill and time obtained from the six kinds of consolidation parameters as follows: 0e-log CT ', and c, obtained from the sample before sand mat construction: e(b),c,(b) @ m, and c, obtained from the sample before sand mat construction: m,(b),c,(b) @ corrected m,(b) and c,(b) for sample disturbance: m,*(b),c,* (b) @ e-log CT ', and c, obtained from the sample after driving packed drainage piles: e(d),c,(d) 424
0m, and c, obtained from the sample after driving packed drainage piles: m,(d),c,(d) @ corrected rn,(d) and c,(d) for sample disturbance: rn,*(d),c,*(d) The corrected m,* and cv* values for sample disturbance are derived from the mean curves in Figs. 7 and 8 respectively, such as the relationships between OCR (= CT ' p / CT ',J and correction values as shown in Shogaki et a1 (1998). The settlement of the sand layer, z = 5-10 meters, is ignored for consolidation of ground surface. The amount of consolidation settlement caused by the sand mat load and the construction of the packed drains is about 40 cm for settlement plate A and about 10 cm for settlement plate B, as shown in Figs. 10 and 11. The packed drains were constructed when the degree of consolidation under the sand mat load was about 50 % and then new settlement occurred. The difference in the amount of settlement between both settlement plates reflects the drain spacing. Basically, the amount of settlement of plate A with small drain spacing is greater than that of plate B. In the settlement behavior after the construction of the packed drains, the prediction of the settlement behavior using the consolidation parameters before constructing the sand mat is closer to the observed settlement than the estimated settlement after driving packed drainage piles. Furthermore, the amount of the estimated settlement, using the e-log o ',, is about 10 % less than that using the m, value. The estimated settlement for both settlement plates is almost finished after the loading of the pre-load since the c, values of the clayey layers are relatively large and the drainage distance becomes shorter. However, the observed settlement is still progressing after loading of the pre-load and the predicted settlement can not explain these behavior.
Figure 10. Relationship between H,S and
The relationship between the ratio (Rs)of the estimated settlement (Se) to the observed settlement (So) and the degree of consolidation (U) for the same data, as shown in Figs. 10 and 11, is shown in Figs. 12 and 13, where the Uvalue is defined as the degree of consolidation of the So value to the final settlement obtained from the hyperbolic method for the observed settlement. The Rs values obtained from the four kinds of consolidation parameters, namely 0, 0, @ and 8 as described above, are shown against the U value in Figs. 12 and 13. It can be seen from Figs. 12 and 13 that the prediction of the settlement behavior using the consolidation parameters before constructing the sand mat is closer to the observed settlement than the estimated settlement after driving packed drainage piles and the estimated settlement increases about 15 %, taking into account the effects of sample disturbance (@ and @). Therefore prediction accuracy of the final settlement improves. 5 CONCLUSIONS The main conclusions obtained in this study are summarized as follows: 1. The ratios of w,, E E,, and q, after driving packed drain piles compared to those prior to sand mat construction, for the Hc2, Hc3 and Hc4, were about 0.8, I . 1, 1.6 and 1.1 respectively. The Rq, value prior to sand mat construction was about 1.9 after loading pre-load. The increasing undrained strength due to 13 cm diameter packed drains is advantageous for pre-loading stability in soft clayey soil. 2. The ratios of e, E v o , Cc and o ' p aRer driving packed drain piles to those of sand mat construction, for the Hc2, Hc3 and Hc4, were about 0.9, 1.7, 0.8 and 1.7 respectively. The ratio of 0' ' p value was about 1.8 after loading pre-load.
t ( S e t t l e m e n t plate A).
425
fi
Figure 11. Relationship b e t w e e n H,S and
t ( s e t t l e m e n t plate B). sand mat was closer to the observed settlement than the estimated settlement after driving packed drainage piles and the estimated settlement increased about 15 %, taking into account the effects of sample disturbance. Therefore prediction accuracy of the final settlement improved.
REFERENCES
Figure 13. Relationship between R, and U (Settlement plate B)
3 . The amount Of the estimated using the e-log was about l0 % less than using the mv value. 4. The prediction of the settlement behavior using the consolidation parameters before constructing the
'
' V J
426
Akagi,T 1981. Effects of mandrel-driven sand drains on soft clay, Proceedings of the Tenth International Conference on Soil Mechanics and Foundation Engineering:58 1-584. Stockholm. Arita,N.,Takahashi,H.,Shibata,A. and Nagumo,M 1998. The developed of automatic settlement gauge with water level meter. Proceeding of the thirty-thirdjapan national confer ence on geothchnical engineerring. Vo1.2:2469-2470(in Japanese). Chai,J.C,Wura,N. and Shen,S.L 1998. Field performance of vertical drain installed in Ariake cray deposit. Teiheitikenkyu N0.7:64-76. Japanese Geotechnical Society 1990. Test Method for onedimensional consolidation properties of soils.(JISA 1217 -1990).The Method and Explanation of Soil test:289-294(in Japanese). Japanese Geotechnical Society 1990. Method for unconfined comoression test on soils.(JISA 1216-1993).The Method and Explanation of Soil test:320-330(in Japanese). Japanese Geotechnical Society 1998. method for obtaining undisturbed soil sample using thin-walled tube sampler with fixed piston(JGS 1221-1995).Standards of Japanese Geotechnical Societyfor Soil Sampling:1-7. Matsuo,M. and Honjo,Y 1999. Jibankankyoukougaku-noatarashii-shiten. Gihoudou publishing company:207- 22 l(in Japanese). Okabayashi,I 1983. Sekiyutankukiso-no-rron-to-jissai. Kashjma Publishing Company:105-114(in Japanese). Schmertman,J.H 1955.Theundisturbed consolidation behaviorof clay. TransactionsASCE.Vo1.120:1201-1233
Shogaki,T and Kaneko,M 1994. Effects of sample disturbance on strengh and consolidation parameter of soft cray. Soils and Foundations.Vol.34.No.3: 1-10. Shogaki,T 1996. A method for correcting consolidation parameters for sample disturbance using volumetric strain. Soils and Foundations.Vo1.36.No.3: 123-13 1. Shogaki,T.,Shirakawa,S. and Maruyama,Y 1998. Estimation of in-situ consolidation parameters using volumetric strain measured in standard oedometer tests.Proceeding of the international Symposium on Lowland Technology:137-144. Saga. Shogaki,T 1999. A method for correcting consolidation parameters for. sample disturbance using volumetric strain. Soils and Foundations.Vo1.39.No. 1: 124-125. Shogaki,T., Satamoto,T. and S.Kawata 2000. Human errors concerning preconsolidation pressure by Cassagrande’s method and their correction. Tsuchi-to-kiso.Vol.2:9-12(in Japanese).
427
This Page Intentionally Left Blank
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Recent developments of ground improvement with PVD on soft Bangkok clay D.T. Bergado & M.A.B. Patawaran School of Civil Engineering, Asian Institute of Technology,Bangkok, Thailand
ABSTRACT: The soft Bangkok clay foundation at the site of the Second Bangkok International Airport (SBIA) and the Second Bangkok Chonburi Highway Project (SBCH), were improved using PVD. At SBIA, the water content reduction Erom field measurements were in agreement with the computed values from the consolidation settlements. Employing the SHANSEP technique, there was excellent agreement between measured and predicted undrained shear strength. Moreover, the actual required discharge capacity has been successfully back-calculated. At the SBCH, a one-dimensional FEM software capable of calculating the consolidation of multi-layered soil, called PVD-SD, was also used for settlement prediction. The PVD-SD method demonstrated slight overprediction. Electro-osmotic (EO) consolidation involves application of direct current electricity through electro-conductive PVDs as electrodes. Due to E 0 consolidation, significant increase in shear strength and faster rate of settlement were achieved at shorter time. Vacuum preloading is imposed by reducing the pore pressure in the clay through the application of vacuum pressures. The vacuum preloading with PVD increased the rate of settlement by 60% and reduced the preloading period by 4 months.
1 INTRODUCTION & Long 1994, Chai et al. 1995), model tests, performance, evaluation of PVD types (Bergado 1996a), and evaluation and development of specification criteria (Bergado et al. 1996b,c). In 1992, the first major project using PVD in the Central Plain of Thailand was finally realized when PVD was utilized in the State Railway of Thailand (SRT) route Erom Klong 19 to Kaengkhoi, Saraburi. Currently, PVDs are being used for the Second Bangkok Chonburi Highway (SBCH) and the Outer Bangkok Ring Road (OBRR). In the Second Bangkok International Airport (SBIA), the PVD has been studied by full scale field embankments and the PVD ground improvement is being implemented. Moreover, vacuum assisted preloading has been studied at the SBIA site in order to reduce the preloading period, reduce the amount of sand surcharge, and eliminate embankment stability problem. In addition, the consolidation period can be further reduced by using electro-osmotic consolidation in conjunction with PVD installation. Electro-osmosis is the process wherein positively charge ions move from the anode to the cathode carrying hydrated water with them upon the application of direct electric current.
Because of its low permeability, the consolidation settlements of soft clays takes a long time to complete. To shorten the consolidation time, prefabricated vertical drains (PVD) are installed together with preloading by surcharge embankment. PVDs are artificially-created drainage paths which are inserted into the soft clay subsoil. Thus, the porewater squeezed out during the consolidation of the clay due to the hydraulic gradients created by the preloading, can flow faster in the horizontal direction towards the PVDs taking advantage of higher horizontal permeability of the clay. Subsequently, these pore water can flow freely along the PVDs vertically toward the permeable layers. Therefore, the PVD installation reduces the length of the drainage path and, consequently, accelerates the consolidation process and allows the clay to gain rapid strength increase. The PVD was first investigated for its effectiveness in improving the soft Bangkok clay in subsiding environment (Bergado et al. 1988). Later, the research direction extended into back-analyses of design parameters (Bergado et al. 1991, 1992, 1996a), numerical analysis and modeling (Bergado
429
Fig. 1 Generalized Soil Profile and Properties
of dense sand. The profiles of soil strength determined by laboratory tests are also shown. The natural water contents are reasonably uniform across the site, and lie close to the liquid limit between depths of 2 and 16 m. Most of the Atterberg values lie above the A-line on the plasticity chart, confirming the high plasticity of the Bangkok clay. The groundwater varies at depths of 0.5 to 1.O m. 3 PVDATTHESBIA
Fig. 2 Water Contents Before and After Preloading with PVD for TS3
2 SITE AND SOIL CONDITIONS The Bangkok subsoils, part of the larger Chao Phraya Plain, consist of alternate layers of sand, gravel and clay. The underlying profile of the bedrock is still undetermined, but its level is known to be between 550 to 2000 m below the ground surface. The test site is located approximately 30 km east of the capital city of Bangkok. The generalized soil profile and soil properties are shown in Figure 1. The soil profile is relatively uniform consisting of a thin weathered crust (2 m thick) overlying very soft to soft clay approximately 10 m thick. Underlying the soft clay is a medium clay layer of about 4 m thick followed by a stiff clay layer extending down to 22 m depth which is in turn underlain by a layer
Three full scale test embankments (TSI, TS2, TS3) were constructed in stages on PVD improved soft Bangkok clay at SBIA site with PVD spacing of 1.5, 1.2 and 1.0 m, respectively, in square pattern (Bergado et al, 1997). All PVDs were installed to 12 m depth. Three PVD models were installed in the test embankments, namely: Flodrain in TS1, Castle Board in TS2, and Mebra drain in TS3. After the PVD installation, the thickness of the sand drainage blanket was increased to 1.5 m. Then, clayey sand was used to raise the embankment to 4.2 m (i.e., 75 kPa of surcharge) in stages. The test embankments were 40 x 40 m in plan dimensions with 3H:lV side slopes and a finished height of 4.2 m. For TS3, a berm width of 5 m and 1.5 m high was included. Construction took 9 months to complete. The fill material was compacted to an average bulk unit weight of 18 kNm3. Surface and subsurface settlement gauges were installed near the center of the test embankment. The subsurface settlement gauges and the piezometers were installed at 2 m interval. The results show consistent patterns in settlements, pore pressures, and lateral movements. The settlement was fastest in TS3 (1.0 m PVD spacing) than in TS2 (1.2 m spacing) and TS2 was faster than TS 1
430
(1.5 m PVD spacing). The lateral deformationsettlement pattern was similar for all 3 embankments. Figure 2 illustrates the reduction of water content with depth for test embankment TS3 after 660 days of preloading (February 1996) compared to mean values measured in February 1994. Previous values from 1973 study by Moh & Woo (1987) are included as dotted lines. The reduction in water content at TS3 is more than 20% which agreed with the back-calculated values from settlement data. Fig. 5 Comparison of ChValues
The increase in undrained shear strength was predicted by SHANSEP technique (Ladd 1991). The predicted increase are indicated by solid lines in Figure 3. The corrected undrained shear strengths measured by field vane shear tests in February 1994, May 1995 and March 1996 are also plotted for comparison. There is excellent agreement between the measured and predicted increase in undrained shear strength due to preconsolidation and drainage by PVD. Assuming &/K, = 5 and using c h = 3 m2/yr, the c h versus discharge capacity, qw, can be backcalculated based on the concepts of Asaoka (1978) for the three test embankments as shown in Fi . 4. The back-calculated qwranged fiom 30 to 100 m /yr. Assuming KhK, = 5, d,/d, = 2, and qw = 30 rn3/yr, c h values were back-calculated and were in agreement with the corresponding values from piezocone tests as shown in Fig. 5. The results of the 1983 study are also plotted for comparison (Bergado et al, 1996a). The accuracy of the back-calculated parameters depends on the limitation of the Asaoka (1978) method as well as the assumption of radial consolidation.
8
4 PVD PERFORMANCE AT THE SBCH The PVD performances at Sections 2A/2 and 2B/1 of SBCH which have the thickest layer of very soft to soft clay and having maximum settlements throughout the highway, were evaluated. The field settlements as well as the fill height were compared with those proposed and predicted by the designers (Bergado et al, 1999). The settlements predicted by Asaoka’s method (Asaoka 1978) represented the settlement of the whole stratum (improved and unimproved layers), as it is based upon the monitored surface settlement record. At the final stage of loading, the subsoil at all sections has gone through about 90% or more degree of consolidation. PVD-SD is a one-
Fig. 4 Comparison of ch-q, Relations of Three Test Embankments TSI, TS2 and TS3 Using the Same Value of Smear Ratio, KhK, = 5
43 1
dimensional FEM software capable of calculating the consolidation of multi-layered soil. PVD-SD calculations indicate that most of the settlement took place at 2 to 12 m depth, corresponding to the zone of very soft to soft clay. Figure 6 shows a comparison of the settlements predicted by the different methods with the observed data for Section 2B/1. The time-settlement plot predicted on the basis of Asaoka’s method (Asaoka 1978) shows that the calculated settlements were in excellent agreement with the observed data. The PVD-SD method also yielded very good predictions whereas the one-dimensional method based on Terzaghi’s theory slightly overpredicted the settlements.
Fig. 7 Calculated and Measured Settlements for Embankment 2 (TV-2)
5 VACUUM PRELOADING A water and airtight very low density (VLDPE) geomembrane was placed on top of the drainage layer. To maintain airtightness, the ends of the liner were placed on the bottom of a perimeter trench and covered with 300 mm layer of sand-bentonite mix and water. A vacuum pump with a capacity of 100 m3/hr and pressure of -70 kPa was installed for each embankment. After applying vacuum pressure for 45 days, the embankments were raised in stages up to a height of 2.50 m. The pumps were run continuously for 5 months (Bergado et al, 1998). Figure 7 compares the FEM results with the corresponding measured data assuming no vacuum pressure, with vacuum pressure that simulated the field conditions, and with higher vacuum pressure (-60 H a ) for Embankment 2. Similar trends of results were also observed for Embankment 1. The results indicated that even with PVD instal-lation, high vacuum needs to be maintained for 4 to 5 months to achieve higher degree of consolidation.
The scarcity of sand to be used as surcharge fill has led to the proposal to explore a combined vacuum preload and surcharge technique. The idea is to reduce the fill height and shorten the time of preloading. Instead of increasing the effective stress in the soil mass by increasing the total stress by means of conventional mechanical surcharging, vacuum assisted consolidation preloads the soil by reducing the pore pressure while maintaining constant total stress. Two additional 40 x 40 m embankments were constructed at the SBIA site close to the previous PVD embankments with a platform of 0.3 and 0.8 in sandfill for Embankments 1 and 2, respectively. For Embankment 1 (TVI), 15 m long PVD was used in conjunction with hypernet and nonwoven geotextiles drainage system. For Embankment 2 (TV2), 12 in long PVD was used together with corrugated pipe and nonwoven geotextile drainage system. PVD spacing was at 1.0 m in a triangular pattern for both embankments.
Fig. 6 Time-Settlement Plot using Different Methods at Section 2 B / 1 ~SBCH Project
432
Fig. 10 Variation of Shear Strengths for Recons-tituted Sample (Small Cylinder Cell)
Fig. 8 Comparison of Maximum Surface Settlement Between Embankment 2 and Previous Embankment TS3 at SBIA Site
Fig. 9 Variation of Settlement with Time with and without Electro-osmotic Consolidation using Reconstituted Clay Samples
The final settlement of TV1 and TV2 were 0.’74 and 0.96 m, respectively. The performance of TV2 when compared to previous studies using conventional sand surcharging, as shown in Figure 8, produced an acceleration in the rate of settlement by about 60% and a reduction in the period of preloading by about 4 months. The major difficulty experienced with this type of preloading is the maintenance of vacuum pressure. Even though a vacuum preloading of 75 kPa was anticipated, the actual measured values seem to indicate an efficiency of only 40 to 50% equivalent to a surcharge pressure of 35 to 40 kPa.
6 ELECTRO-OSMOTIC CONSOLIDATION Electro-osmosis (EO) is the process wherein positively charged free water in a clay-water system moves from the anode to the cathode. Upon application of a direct current, cations in the diffused double-layer of water moves toward the cathode to gain electrons and thereby become discharged. As the cations move, they carry with them water so that there is a new movement of water toward the cathode. Consolidation will result if water is renoved at the cathode but not replaced at the anode. Studies on the effect of electro-osmotic consolilation on soft Bangkok clay were performed in the aboratory. Two types of electro-conductive drains nade from common prefabricated vertical drains PVD) were used. These consist of copper electrodes nade by inserting 2 mm diameter copper rods into he drain core, and carbon electrodes made by .vrapping the drains with carbon fibers. The samples were reconstituted and tested in a 300 mm high having 300 mm diameter small cylinder cell (Abiera et al, 1999; Bergado et al, 2000). Two holes at the top and bottom cap, spaced 200 mm apart, were provided for PVD installation. Vertical load was applied on the top cap through a loading piston. Reconstituted pressure and applied vertical stress was maintained at 5 kPa coupled with 60 and 120 V/m voltage gradients. These voltage gradients were obtained from previous investigators (Abiera et al, 1999; Bergado et al, 1998). Polarity reversal was done every 24 hours. All tests were carried out until 90% consolidation was achieved (Bergado et al, 2000). The initial water content, liquid limit and plastic limit of the soil specimen were 97%, 96% and 33%, respectively. The pH, cation exchange capacity (CEC), and total dissolved salts (TDS) were, respectively, 6.3, 46.85 meq/lOOg, and 4050 PPm. The variation of settlement against time and the variation of shear strength across the drains are 433
shown in Figures 9 and 10, respectively. The shear strengths were measured by customized miniature vane shear apparatus. Larger settlements and higher shear strength were obtained upon application of electro-osmotic consolidation compared to using ordinary drains. Moreover, the carbon electrodes displayed better results compared to copper electrodes in both 60 and 120 V/m voltage gradients. The shear strength between the anode and the cathode were almost equal indicating the effectiveness of 24-hour polarity reversal. However, the shear strengths in-between the cathode/anode locations can be lower. 7
CONCLUSIONS
Three full scale test embankments were constructed in stages on soft Bangkok clay at the Second Bangkok International Airport (SBIA) with prefabricated vertical drains (PVD) installed to 12 m depth in a square pattern. The water content reductions from field measurements were in good agreement with the computed values from consolidation settlements. The undrained shear strength with depth as measured in the field is in agreement with the values calculated from the SHANSEP technique due to preconsolidation and drainage. The back-calculated actual discharge capacity, qw, for the three test embankments ranged from 30 to 100 m3/yr. The backcalculated CI, values agreed with the corresponding results from piezocone tests. The PVD performance was also evaluated at selected sections of the Second Bangkok Chonburi Highway (SBCH) Project. The rate and amount of settlement predicted by Asaoka’s method proved to be in excellent agreement with the observed values. The settlements predicted by the one-dimensional FEM computer program PVD-SD proved to be in reasonable agreement with the measured values. Two additional embankments were constructed at SBIA to study the effect of vacuum preloading in combination with reduced amount of sand surcharging. The final settlement of TV1 and TV2 were 0.74 and 0.96 m, respectively. The performance of TV2 when compared to previous studies using conventional sand surcharging showed an acceleration in the rate of settlement by about 60% and a reduction in the period of preloading by about 4 months. Electro-osmotic consolidation under 60 and 120 V/m voltage gradient was performed on reconstituted soft Bangkok clay using drains modified by adding copper and carbon electrodes. Tests indicated larger settlements and higher shear strength for both electro-conductive drains compared to ordinary drains. However, the carbon electrodes displayed 434
better results compared to copper electrodes in both voltage gradients. Polarity reversal every 24 hours proved to be effective producing almost equal shear strength between the anode and the cathode. REFERENCES Abiera, H.O., Miura, N., Bergado, D.T. & Nomura, T. 1999. Effects of using electro-conductive PVD in the consolidation of reconsti-tuted Ariake Clay. Geotech. Eng’g. Journals, Vol. 30, No. 2, pp. 6784. Asaoka, A. 1978. Observational procedure for settlement prediction. Soils and Foundations, Vol. 18, NO. 4, pp. 87-101. Bergado, D. T., Miura, N., Singh, N. & Panichayatum, B. 1988. Improvement of soft Bangkok clay using vertical band drains based on full scale test. Proc. of the Intl. Conf on Eng’g Problems of Regional Soils, Beijing, China, pp. 379-3 84. Bergado, D. T., Asakami, H., Alfaro, M.C. & Balasubramaniam, A S . 199 1. Smear effects of vertical drains on soft Bangkok clay. ASCE Journal of Geotech. Eng’g. Div., Vol. 117, No. 10, pp. 1509-1529. Bergado, D. T. , Enriquez, A. S., Casan, L., Alfaro, M. C. & Balasubramaniam, A. S. 1992. Inverse analysis of geotechnical parameters on improved soft Bangkok clay. ASCE Journal of Geotech. Eng’g. Div., Vol. 118, No. 7, pp. 1012-1030. Bergado, D. T. & Long, P. V. 1984. Numerical analysis of embankment on subsiding ground improved by vertical drains and granular piles, Proc. 13th Intl.Conf Soil Mech. Found. Engg., New Delhi, India, Vol. 4, pp. 1361-1366. Bergado, D. T., Long, P. V. & Balasubramaniam, A. S. 1996a. Compressibility and flow parameters form PVD improved soft Bangkok clay. Geotech. Engg. Journal, Vol. 27, No. 1, pp. 1-20. Bergado, D. T., Mannivannan, R. & Balasubramaniam, A. S. 1996b. Proposed criteria for discharge capacity of prefabricated vertical drain. Geotextiles and Geomembranes, Vol. 14, pp. 48 1-505. Bergado, D.T., Mannivannan, R. & Balasubramaniam, A.S. 1996c. Filtration criteria of prefabricated vertical drains filter jackets on soft Bangkok clay. Geosynthetics Intl., Vol. 3, No. 1, pp. 63-83. Bergado, D.T., Balasubramaniam, A.S., Fannin, R.J., Anderson, L.R. & Holtz, R.D. 1997. Full scale field test of prefabricated vertical drain (PVD) on soft Bangkok clay and subsiding environment.Ground Improvement Developments
1987-97 (GeoLogan '97), Geotech. Special Publ. No. 69, ASCE, New York, U.S.A. Bergado, D.T., Chai, J.C., Miura, N. & Balasubramaniam, A.S. 1998. PVD improvement of soft Bangkok clay with combined vacuum and reduced sand embankment preloading. Geotech. Eng g. Journal, Vol. 29, No. 1, pp. 95-122. Bergado, D.T., Balasubramaniam, A.S., Chishti, I.A., Ruenkrairergsa, T. & Taesiri, Y . 1999. Evaluation of the PVD performance at the Second Bangkok Chonburi Project. Lowland Tech. Intl., Vol. 1, NO. 2, pp. 55-75. Bergado, D.T., Balasubramaniam, A.S., Patawaran, M.A.B. & Kwunpreuk, W. 2000. Electro-osmotic consolidation of soft Bangkok clay with prefabricated drains. Ground Improvement, Vol. 4 (in press). Chai, J.C., Miura, N. & Bergado, D. T. 1995. Behavior of PVD improved ground under embankment loading. Soils and Foundations, Vol. 35, NO. 4, pp. 49-61. Ladd, C. C. 1991. Stability evaluation during staged construction. ASCE Journal of Geotech. Eng g. Div., Vol. 117, No. 4, pp. 540-615. Moh, Z. C. & Woo. S. M. 1987. Preconsolidation of Bangkok clay by non-displacement sand drains and surcharge. Proc. 9th Southeast Asian Geotechnical Conference, Bangkok, Thailand, Vol. 1, pp. 8-171 to 8-184.
435
This Page Intentionally Left Blank
Coastal Geotechnical Engineering in Practice,Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 15 1 1
Improvement of hydraulic fills by using dynamic consolidation method Jing-Wen Chen & Jer-Min Liao Department of Civil Engineering, National Cheng Kung Universiy, Taiwan
ABSTRACT: Field measurements collected from literature are used to study current practice and to determine the response of the ground improvement using dynamic consolidation method. Ground conditions at these sites were all fill sands. Equations summarized from literature show that depth of influence increase with square root of product of energy per blow and energy applied on soil stratum per unit area. Another one show that SPT-N value increases with energy applied at soil stratum per unit volume. The last one show that the PMT-P,~value increases with the energy applied at soil stratum per unit area. Meantime, the difference between depth of influence and depth of improvement are also discussed.
1 INTRODUCTION The effect of the dynamic consolidation method in reinforcing soil stratum is primarily dependent on type of soil, initial conditions of soil, weight of the tamper, drop height, tamping conditions, such as; spacing, number of tamping for each layer, number of layer, time delay after each tamping and ground water level. However, the major issue in a design is whether the fill at the desired depth of improvement, applied by known energy, can meet the contract specifications. Therefore, one of the focal points in research of the method lies in the relationship between tamping energy and depth of improvement. Most of important studies in dynamic consolidation method (Menard and Broise 1975, Leonards, et al. 1980, Rollins and Kim 1994) emphasized the relationship between the square root of single tamping energy applied to fill and depth of improvement. However, a linear relationship between depth of improvement and square root of single tamping energy can not be verified theoretically. Therefore, through such energy demonstration forms as single tamping energy, energy per area, and energy per volume of applied fill, this study attempts to examine whether there exists a practical linear relationship between the applied fill and depth of improvement. It will also investigate the relationship between tamping energy and improvement of fill. The results are expected to serve as a reference in designing hydraulic method for improving the soil stratum of reclaimed land.
To meet needs of industrial development in Taiwan, there are many projects on reclamation of land going i n the seaside of western Taiwan. Normally the fill needed is made available by hydraulic extracting. However, hydraulic fill is typically handicapped not only by insufficient bearing capacity and extensive settlement but also by liquefaction. If hydraulic f i l l is intended as an industrial land, its soil stratum must be improved i n advance. Since dynamic consolidation is able to achieve the application quickly and is cost-saving, the niethod has been adopted to improve the soil stratum of reclaimed land. When a tamper impacts on a fill, air trapped i n i t will be squeezed out because of compaction. If the coinpaction is applied continually, void of soil will be decreased until it is saturated. The soil grains will realign and the soil will be more compacted when the pore water pressure rises and is dissipated and squeezed out as compaction is continually applied. Therefore, the first stage of dynamic consolidation method applied to fill is compaction, and the second stage makes fill compact because of squeezing out of water, which can be defined as the dynamic consoiidation stage. There should be difference between the two stages with respect to the effect on depth of improvement. We can determine that, according to the definition of Menard, tamping with a fixed amount of energy has fixed depth of influence applied to the soil stratum, but depth of improvement applied to the soil stratum will change as the number of tamping increases.
437
bottom and depth of improvement D,,,,. Model B is the so-called pressure bulb method in soil statics. Various ellipses indicate different equi-pressure lines. When pressure is applied to soil, soil grains shift, and the grains under the load focus shift only vertically while the side grains shift not only downward but also laterally. Therefore, regarding soil layer on the same level, the soil grains right under the plummet burden heavier pressure than those outside grains do. This primarily accounts for why the equi-pressure lines form bulb shape. Liao and Chen( 1997) found that this model was more reality. In model C, soil is regarded as an ideal elastic object. The load weight on soil is evenly burdened by the under layer of soil formed by a 30 degree line extending from the horizontal line. Therefore, two types of energy we can use to express the tamping energy applied to the fill. One is the applied energy per unit area of the fill, E(,. Another one is the energy burdened by each unit volume of improved fill, E l .
Figure . The models ofenergy transfer
2. AP 'LIED ENERGY AND DEPTH OF IMPROVEMENT
2.1 Depth of improvement and depth oj'injluence Menard and Broise ( 1975) presented
2.2 Work of Previous Research where D is the thickness of compacted fill (ni)and M is pounder weight (ton), and h represents falling height (m). Leonard, et a1.(1980) defined D as depth of influence or effective depth of compaction. Mayne (1 984) defined depth of influence as the maximun~ observable depth of fill improvement and further defined critical depth as the maximum value of depth of fill improvement. Rollins and Kim (1994) defined depth of improvement as the depth to which the dynamic cornpaction caused some improvement in a given soil property, such as; density, stiffness, penetration resistance. Chen, et al.( 1994) defined depth of improvement as that fulfilling the design requirement of SPT-N or CPT-Q,. specified in the contract. In summary, depth of influence and depth of improvement largely refer to the sanie thing. However, it will be practical when depth of improvement is regarded as the constant accumulation of depth of influence, i.e. depth of influence is the process of dynamic compaction while depth of improvement is the final result. Therefore, to facilitate analysis, this study will, in discussion, treat applied depth of fill by single tamping energy as depth of influence D,and treat applied depth of fill by energy per unit area or unit volume as depth of iniprovement D,,,, and make distinct i o 11. With respect to tamping energy applied to f i l l , we can apply the dissipation of soil pressure and classify the energy transfer model into three categories as indicated in Figure 1. I n model A, when the tamper strikes soil stratum, the energy applied to soil stratum is burdened by the volume obtained by the product of the area of the tamper
Mitchell ( 198 1 ) maintained that depth of improvement was affected not only by impact energy but also by type of soil, pattern of tamper, and existence of soft layers. In other words, when the same tamping energy is applied to sand fill and to clay fill respectively, the obtained depth of improvement is larger in the former case. Moreover, the applied energy of a free drop is more effective than that of a crane drop. Moseley (1993) indicated that under ground water level and the number of tamping can affect depth of improvement. Leonards et a1.(1980) pointed out that soft layers in soil inay interfere with the downward transfer of tamping energy and decrease depth of improvement of fill. For the present, estimation of depth of improvement is typically modeled on equation proposed by Menard and Broise( 1975) as D = ffdM . h . Value of coefficient a is usually equal to 0.5-0.6 for sandy soil. However, Menard's equation ignores the influence of the number of tamping and the size of applied energy. Thus it is not justifled to regard the equation as depth of improvement. Tai Yuan Industrial University ( 1988) presented the dcpth of improve men t as fo 11o ws :
D,,,, = 5.102+ 0.009 W '11 + 0. 009Et,
(2)
Since tamping energy per unit area, E(,is added to Equation(2), it takes consideration into the influence of number of tamping and tamping spacing. However, because the mininiuni limits of tamping energy per unit area and single tamping energy are not determined, a depth of improvement of about 5 ni is obtained when the equation is put into use, even
438
when no energy is applied to fill. This is not reasonable. In addition, Chang, et al.( 1992), based on their findings from tamping crushed rock fill, presented an equation of depth of improvement to which the number of tamping N and revised coefficient of tamping energy are added. Asahi, et al. ( 1979) inferred depth of improvement through work principle which regard dynamic as static load weight. Chen, et al. (1994) inferred from energy applied to fill per unit area, the interval of tamping and the relationship among the applied energy, the interval and depth of influence. Except Menard's equation, other equations required referring to tables and complex calculation and are inconvenient. Therefore, this study presents revisions after taking strenghts and weaknesses of each equation into consideration.
The relationships between single tamping energy E,) and depth of influence De, square root of El, and De, tamping energy per unit area E, and depth of improvement Dlnl,square root of EIl and DI,,, square root of the product of El, and EIl and D,r,l,to the 0.25 root of the product Er, and E, and Dlnlare studied. The equations of linear regression for these relationships and the corresponding coefficient of correlation are listed in Table 2. From statistical point of view, equation(3)and (4) should be the suitable equations that to present the relationship between depth of iinprovement and applied energy.
2.3 Principle and Process ojRevisions
However, for taking consideration of practical construction, it is inconvenient to use two different pounder with different weights since the depth of improvement can not be determined. Therefore, equation Dl,,,=0.667(E, )'" should be the appropriate candidate to express the relationship between depth of improvement and applied energy.
The 14 sites using dynamic consolidation method to improve the hydraulic fill are collected in this study. The results of improvement and analysis are listed in Tablel. I n the table SPT-N or PMT-P, is used to evaluate the result of the improvement. The following hypotheses are proposed to facilitate analysis of data available: 1 . To be practical, both depth of influence and depth of improvernent are designated as zero when no energy is applied to fill. 2. Depth of improvement of fill D,,,, and energy applied to f i l l form linear relationship, and so do depth of influence D ,and energy applied to fill. 3. The influence of underground water level on depth of i i n p ro veine n t is ni i n i m urn. 4. Improvement of soil stratum is conducted soon after hydraulic fill, and time effect is not considered.
D', =O. 5 95(E,,)".'
(3)
D,,,l=0.667(EIl)" '
(4)
3. ENERGY AND DEGREE OF FILL IMPROVEMENT 3.1 Examination of Inzprovernent on Fill The degree of improvement of fill which has been reinforced by dynamic consolidation method is normally judged by the measurement of Menard's compression gauge (PMT-PJ, standard penetration test (SPT-N) and cone penetration test (CPT-Qc). The Menard coinpression gauge primarily measures soil transfornlation modulus E> and soil limit
Table 1 Applied energy and depth of improvement
439
Equation Coefficient of Corre 1at i on
De=0.0282Ep De=0.595 (Ep)"j D,,,,=0.041Ea DI,,,=0.667(Ea) 0.907
0.953
0.896
0.949
D,,,=0.371(EpEa)"5 D,,,=0.63 (Ep Ea)025 0.958
0.980
r'
pressure P, . The standard penetration test is widely used; SPT-N frequently serves as the reference index to liquefaction of soil. Therefore, after fill is improved with dynamic consolidation method, an examination by using the SPT-N is quick and convenient. CPT is characterized by its ability to repeat examination of fill, but the technique has difficulty penetrating into hard f i l l . 3.2 Analysis of SPT-N Value Figure 2 shows the change in SPT-N of fill before and after energy per unit volume is applied to force and thus has lower disturbance to the fill. This technique is accurate in examining the effect of improvement on fill. We can observe the trend of positive correlation between the two factors. Figure 3 shows the relationship between energy applied to fill in terms of per unit volume t . nz /nz3 and increase of standard penetration SPT-N value. The relationship is represented in the following:
A N = 0.304 E,
(5)
in which El is energy per unit volume,
10 < E , ) <50 ( t . f i ~ / i ~ ' ) A N : increased standard penetration number Lin et a1.(1986) indicated that the sedimentary mudsand in Taiwan area, the ratio of conical penetration experimentation CPT-Qc value and standard penetration experimentation SPT-N value is about 4. Therefore, the incremental Qc value after energy is applied to f i l l can be obtained, and the ratio of the two is 1.22. Figure 4. The relationship bctwecn tamping eiicrgy pcr u n i t area and increase of S I T - N
AQc
=
1.22 El
(6)
A Q c : the increase of conical penetration CPT-Qc 3.3 Aiialysis of PMT-PL Value
Figure 2. Tamping energy per unit volume and the incrcase of
SP T-N
440
Figure 4 indicates the relationship between energy applied to per unit volume E , and the increased f i l l limit pressure value P,. From this figure, we know that data dots and line of regression are more dispersed, and r' = 0.956. Therefore, we can infer that there exists better linear relationship between the increased fill limit pressure value and energy
applied to per unit area t . m / nz'. The relationship can be express as follows:
AP , =0.038 E,
(7)
Where AP, is increased limit pressure and E(,is the applied energy per unit area. 3.4 Field Examination The refinery plant, located nearshore of Yunlin county, Mailiau as shown in Figure 5 , represents a model case history of dynamic consolidation (Shih Chiu Construction and Suivey Engineering Company, 1994). The derived equations can be calibrated by field data listed as follows: Single tamping energy, E,,= 500 t .ttz Applied energy per unit area, E(:= 489 t . i n Depth of improvement, D,,,,= I4ni Applied energy per unit volume, El = 35.5 t Initial average SPT-N = 9.4 Final SPT-N = 19.4 Initial average Q,= 40 liy/criz' Final average Q,= 85 kg/cm' C a I i brat i o n : Depth of improvement from Equation (4) Din,=0.667 (E,,)' = 0.667 (498f =14.8m = 14m (measured) SPT-N from Equation (5) A N = 0.304 ( E l ) = 0.355 (35.5) = 10.8 N (After) = N (before) t- 10.8 = 9.4 4-10.8 = 20.2 = 19.4(nieasured) Q, from Equation (6) A Q L= 1.22 E , =1.22 (35.5) , = 43.3 (kg/cm,-) Qc (after) = Qc (before) -43.3 =40+43.3 , = 83.3 = 85 (kg/cnz-)(measured)
'
,112
/ t12' Figure 5 . Location of Mailiau, Taiwan
4 CONCLUSION 1. In the implementation of dynamic consolidation method, the depth of improvement 011 fill and the product of single tamping energy and energy per unit area of the applied fill to 0.5 power form linear relationship, but the energy of reinforcement has upper limits. The depth of influence and the square root of single tamping energy form positive correlation, and the coefficient is 0.595. Depth of influence is the process of depth of improvement. In analyzing, there should be distinction between the two factors.
'
By examining the degree of improvement of hydraulic fill by using the standard penetration SPT-N value, we know that the increase of SPT-N value and energy burdened by per unit volurne El form linear relationship, and the coefficient is 0.304. J.
Froni above examination and field calibration we can clearly observe that when the depth of improvement i n a soil stratum is detemiined, Equation(3) can use to deteiiiiine the weight and height of tamping, Equation(4) for calculation of the size of energy to apply to an unit area of ground, and Equation(5), Equation(6) and Equation(7) are further used to determine whether the SPT-N value, CPT-Qc value and PMT-P/. value of fill after it has been rammed meet the specifications, capable of replacing the test tamping of pilot test and saving time and engineering cost.
For the sedimentary niudsand i n Taiwan area, the ratio of conical penetration test CPT-Qc yalue and standard penetration test SPT-N value is about 4. Therefore, the increase of Qc value after energy per unit volume is applied to f i l l can be obtained, and the ratio is 1.22.
4. By examining the degree of improvement of fill by using PMT-P,- liniit pressure value, we know that the increase of limit pressure value L I P , and energy burdened by per unit area E,,form linear relationship, and the coefficient is 0.038.
44 1
5. The empirical equations suggested in this study are calibrated by field data. The results present reasonable agreement.
Rollins, K. M., and Kim, "U.S. Experience with Dynamic Compaction of Collapsible Soils," Proceeding of In-Situ Improvement, Atlanta, pp.26-43, (1 994).
REFERENCES
Shih Chiu Construction and Survey Engineering Company, Report on the Application of Dynamic Consolidation Method to the Soil Improvement of Heavy Works Construction Foundation of Formosa Plastic Corporation, ( 1994).
Asahi, S. Hiroshi, N. Masao, H. and Nobo, S. "The Application of Dynamic Consolidation Method 011 Oil Vessel Foundation," Soils and Foundations, V01.17,NO.9,pp.5-11. (1979). Chang,Y .J. Ping,Y .C. Kong, H.F. and Chang, F. "Experimental Research on Boulder Foundation Treated by Strong Tamping," Proceeding of the Third National Seminar on Foundation Treatment, Qin Huang Dao, pp.395-400 (1992). Chen, J.W., Liao, J.M., and Wei, D. "Stress Distribution in Sand under Dynaniic Compaction," Proceedings of the 7"' Conference 011Current Development of Geotechnical Engineering, Taipei, pp.727-734 (1997). Chen, S.G. Chang, T.S. and Lee, L,Y. "EnergyEffective Depth Relationship in Dynamic Compaction for Hydraulic Fill Site", Proceeding of 16th Conference 011 Ocean Engineering, Taiwan, pp.232263 (1994). Leonards, G.A. Gutler, W.A. and Holtz, R.D., 'Dynamic Coinpaction of Granular Soils, "Journal of the Geotechnical Engineering Division. ASCE. Vol. 106, No.GT 1, pp.35-44 (1 980). Lin, P.S., Lee, F.P. and Lai, S.Y,"Correlations of Dutch Cone Test with Dynamics Characteristics," Sino-Geotechnics, No. 16, pp.5 1-62 (1 986). Tai Yuan Industrial University Strong Tamping Method" Manual of Foundation Treatment, Chapter 6, China Architecture Industry Press (1 988). 'I
Mayne, P. W., "Ground Response to Dynamic Compact i 011, J O U ~1 IofI ~the G cotechn i cal Engineering Division, ASCE, Vol. 110, No.6, pp.757-774 (1984). "
Menard, L., and Broke, Y., "Theoretical and practical Aspect of Dynamic Consolidation," Geotechnique,Vol. 25, No. 1, pp.3- 17 ( 1 975). Mitchell, J.K., "Soil Iniprovenient, state-of-the-art Report," Proceedings, 10th International Conference on Soil Mechanics, Vo1.4, Session 12, Stockholm, pp.509-52 1 , ( 198 1). Moseley, M.P., Ground Improvenient, Hayward Baker Inc. Maryland, USA, pp.2 1-39, (1 993).
442
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 15 1 1
Accelerated consolidation method in Minami-Honmoku Terminal construction project T.Chiba Minami-Honnioku Terminal Construction Ojfice, Port and Harbor Bureau, City of Yokohama,Japan
ABSTRACT: Minami-Honmoku Terminal is a container berth that has been constructed by reclaiming land, using as reclamation materials the leftover soil from inland construction works and the dredged soil obtained from dredging rivers and channels. The sea around the site is deep, having an average depth of Y.P-25m. And a soft alluvial clay stratum lies under the seabed, for which it is worried that ground subsidence will take place over a long time. Thus, as a counter-measure against residual settlement, vertical drain method was employed. This report describes the summary of ground improvement design and construction works together with the actual performance.
1 INTRODUCTION
Figure 1 shows the general plan of the MinamiHonmoku Terminal Construction Project and Table 1 the Construction Schedule. The whole project is being carried out roughly in two stages to promote efficient utilization of container terminals. The first stage construction zone that is reported in this paper shall start its services in the year 2001. The reclamation works are in progress dividing the first stage construction zone into two blocks: No.1 block (62ha) and No.3 block (27ha). The soil leftover from inland construction works and the soil obtained from maintenance dredging of rivers and channels are used as the reclamation fill materials. In October 1991 soil dumping started all from offshore side at the No.1 block using soil carriers and floating conveyor ships. The average water depth of the sea area was originally Y.P-25m, and at the seabed, the soft alluvial clay is deposited in layers, which is likely to cause ground subsidence of the reclaimed land over a long period of time even after the land is in service. The technical problem of this project is to accurately estimate the ground subsidence and to conceive effective counter-measures against subsidence in a limited time before the beginning of the container terminal service. In this report some considerations are given after making mostly the comparison between the design and the actual in connection with the ground improvement works of the first stage construction zone.
The purpose of Minami-Honmoku Terminal Construction Project is to cope with the diversified demands for commodity distribution, which resulted from the increase in container cargoes. The works consist of construction of four container terminals with water depth of approximately -15 to -16m on a man-made island, created by reclaiming from a sea area of 217ha, that is located off the coast of Minami-Honmoku roughly around the middle of the coastline of Yokohama City.
Figure 1. General plan of the Minami-Honmoku Terminal construction project.
443
2 EXISTING GROUND AND SOIL CHARACTERISTICS 2.1 Composition of the existing ground This project site was dredged in 1960’s to collect soil for reclamation of other sites. As a result the water depth before the beginning of the construction works was Y.P-20 to -27m except at the channelside revetment and the middle partition revetment. The original seabed is inclined with a gentle slope of 0.8% from the land side toward the middle portion of the bay. Figure 2 shows the contour map of mudstone (BR) layer under the seabed, which is hereinafter, called the base layer. The elevation of the base layer varies widely, ranging from Y.P-10 and Y.P-45m. Consequently, the thickness of the layer to be consolidated differs greatly with the location. Moreover, in Fig. 2, the location of subsurface soil investigation is shown. The ground investigation was carried out by boring, soil sampling and electric cone penetration tests, and it mostly aimed at grasping the thickness of the layer to be consolidated, and the accumulation of super-soft silt and confirming the soil strata that will function as a drainage layer.
The reclaimed and existing ground of the first stage construction zone is rather comply on generally comprises from the upper layer down to the bottom in the order of the leftover soil from construction works, including mainly gravel (Bg), a non-uniform reclamation soil stratum comprising alternately intermixed dredged soil (Bc) and construction leftover soil (Bs), an alluvial sandy soil stratum (As), an alluvial clay stratum (Ac), and diluvial clay deposits (Dc) in the rolling hill of the mudstone (BR) layer of the No.1 block. Since the height of the reclaimed land is Y.P+S.Om, the thickness of soil stratum to be consolidated is approximately 50m at the maximum.
2.2 Soil properties of existing and reclaimed ground Figure 3 shows the profile of soil properties that are the results of soil tests conducted on the existing and reclaimed soils. The summarized properties of each stratum are as below. Bg Stratum: Non-uniform stratum with N-values widely varied, ranging from 0 to 13. It is widely distributed over the ground surface with thickness of about 10m. Bc Stratum: The reclaimed soil stratum including mainly cohesive soil with N-values ranging from 0 to 3. Some parts of the area include sludge with natural water content of 100%. Bs Stratum: The reclaimed soil stratum including mainly sand with N-values ranging from 1 to 14. It is difficult to distinguish the boundary of the stratum from that of the Bc stratum. As Stratum: It is widely distributed around the original seabed of the No.1 block with N-values ranging from 4 to 11. This stratum has a function of drainage. Ac Stratum: Existing unconsolidated clay layer, rela-
tively homogenous with N-values ranging from 1 to 7. It is considered that the consolidation of the stratum will be speeded up with As layers existing above and below.
Figurc 2. Contour map of mudstonc (RR) layer under the seabed.
Dc Stratum: It forms an alternate stratum with sand layers and it has N-values ranging from 5 to 16 and the thickness of the stratum is approximately lm. Since the stratum is sparsely distributed and hardly distinguished from the Ac stratum, it is assumed as a part of the Ac stratum in the design work.
444
Figure 3. Profile of soil properties
3 ACCELERATED WORKS
CONSOLIDATION
3.1 Generul policy The present reclaimed land has to begin its service as a container terminal wharf within approximately three years after the land reclamation together with the infrastructure development including roads and supply lines. Hence, with an aim to minimize the ground subsidence after starting the services, the ground improvement was performed on 59hu of land, approximately 65% of the total area. The general policy of the ground improvement includes the following. To up-keep the residual settlement within 30cm at the starting time of the infrastructure development (one year before the beginning of the terminal service). To employ the accelerated consolidation method that is called the vertical drain method (sand drains, cardboard drains). It is used in combination with the embankment loading method considering from the viewpoints of workability and economic viability. To conduct the improvement of the ground that includes the alluvial and earth-fill reclaimed soil strata above the base layer (BR). To use repeatedly the loading materials for the economic viability and finally use them as roadbed materials for the terminal wharf road construction.
3.2 Design of accelerated consolidation works Figure 3 shows the profile of soil properties that are widely scattered. Hence, in case the soil constants were taken directly from the results of soil tests, it is predicted that the computed settlement and computed settlement speed would differ greatly from those of the measurement. Thus, prior to the design works, a test embankment was constructed on a test ground that was provided within the reclaimed land of the No.1 block, where the field observations, such as settlement, horizontal displacement, pore water pressure, ground water table were conducted and these results were used in the design computation. The consolidation settlement was computed by the one-dimensional consolidation method. The degree of consolidation was computed by Terzaghi’s onedimensional theory and also by Barron’s approximate solution that takes into consideration the well resistance. Based on the analysis of results from instrumented measurements and also based on the results from pore water pressure dissipation tests, the ground surface, the As stratum and the strata above the base layer were assumed to be acting as the drainage layers. The consolidation settlement was computed for 80 observation points that were located on the first stage construction zone, considering the reclamation history, ground conditions, landuse configuration and ground preparation schedule. The studies on accelerated consolidation settlement began from the most economical embankment loading method. The vertical drain ground improvement method was scrutinized only when the embankment loading method could not yield the results that satisfy the objectives. The embankment loading
445
ground improvement method was designed under the upper-limit design conditions: loading height = 6m, left-over period = 6 months. The so-called vertical drain methods, such as sand drain method, sacked sand drain method and board drain method were examined by conducting comparative studies. As a result of these studies, the sand drain method was found to have the following advantages. Consequently it was adopted for the No.1 block. 1) The method employs a big diameter drainage pipe and the effect due to the consolidation time lag is small. 2) There is no problem of percolation of the hard gravel layers that lie between the soil layers of the reclaimed ground. 3) It is possible to secure the required number of long-size sand-drain driving machines. Moreover, as to the No.3 block, the sand drain method and the most economical board drain method were employed.
3.3 Execution of works for accelerated consolidation settlement Figure 4 shows the outline of works for the accelerated consolidation settlement. The area where reclamation was completed most early in the No.1 block was improved by the embankment loading method (PL : loading thickness = 2.5-3.5m), and other areas were improved by the sand drain method (SD : pile diameter = OSm, layout pattern = square pattern, spacing = 3.5~2).As to the area out of the operation range of the machine that is the maximum piling range equal to 35m, a part of the alluvial clay stratum remains non-improved. As to the No.3 block, the whole area of the container yard was subjected to the sand drain ground improvement method (SD : pile diameter = OSm, square layout arrangement, pile spacing = 4 . 0 ~ ) . Regarding the area adjacent to the channel side, sand-drain piles were driven into the ground by changing the pile spacing gradually from 6 to 10m with an aim to control the differential settlement after the beginning of the service. And finally, regarding the area where land reclamation was completed, the ground was so soft that the continuity of the sand drain pile was to be worried about. Consequently, the board drain ground improvement method was adopted (BD : drain material = natural fiber, 9mm thickness by 90mm width; square layout arrangement, drain spacing = 2.0, 2.5m). The sand mat thickness is taken 0.5m and the under-drainage pipes were provided at 20m intervals in the sand mat to make up for the drainage efficiency. 446
Figure 6. Contour map of measured settlement values.
4 FEATURES OF IMPROVED GROUND 4.1 Characteristics of consolidation settlement After conducting accelerated consolidation works, the measurements were made to find out settlement, pore water pressure and ground water table of the reclaimed land. The embankment for loading was removed only after it was confirmed that the residual settlement was within the allowance value. The review of the predicted (computed) settlement was made on the basis of the results from measurement.
Figure 10. Comparison of unconfined strengths (4") before and after ground improvement.
This is due to the fact that at these points, exceedingly soft clays were accumulated under all aspects of reclamation process and soil properties of these clay layers were not taken into account in the calcu1ation. Figure 8 indicates the time - settlement curve of point E, shown in Figure 7 (and 5). This point is the deepest point in the first stage construction zone with the foundation elevation Y.P-45m, and the lower 15m thick clay layers out of the alluvial strata were not improved. From this it is learnt that the consolidation of the reclaimed layers was almost completed until the beginning of the terminal services and the residual settlement was taking place at the alluvial clay strata. Moreover, in the design caIculation, a comparatively precise prediction was possibly made with respect to the settlement of the alluvial clay strata, but with respect to the soils of
Figure 8. Time -settlement curve (point E).
The contour map of computed settlement values and the contour map of measured settlement values at the time of completion of one dimensional consolidation are shown in Figure 5 and Figure 6 respectively. And Figure 7 shows the comparison of the computed and measured settlements. According to these figures, both settlements were inclined to be roughly in conformity with each other. Regarding the settlement values, the measured settlements were mostly larger than the computed values in the case of the No.1 block, but they are slightly smaller in the case of the No.3 block. The results show that the measured settlement values at points from A to E in Figure 7 (and 5 ) are larger than the computed values. 447
the reclaimed ground, the results were fairly larger than the predictions due to the above mentioned reason. Thus, to verify the effect of the accelerated consolidation works, the computed time-settlement curve was modified to fit in with the measured settlements. Accordingly, it was confirmed that the requirement of the ground improvement to keep the residual settlement within the allowance value of 30cm was fulfilled. The soil constants, used in the calculation and the soil constants, modified by fitting in with the measured settlements are shown in Table 2. The apparent consolidation coefficients (ch) of the reclamation soil, obtained from backward calculation by Monden Method using measured settlements are shown in Figure 9. The apparent ch of the reclamation soil is 1000 2000 cm2/d and it is almost equal to the value at the lower range of ch shown in Table 2. These ch values in Table 2 were derived by the fitting analysis.
-
4.2 Mechanical characteristics The changes of unconfined compression strengths (qu) before and after ground improvement by sand drain method are shown in Figure 10. The post improvement results were obtained three months after the end of the 10-months' pre-loading period. Major changes of strength were not confirmed at the Y.P 0 Y.P-lOm Bg layer because it has a high content of sand and gravel, but the strength increase, proportionate to the embankment load, was confirmed at the Bc layer around Y.P-10 Y.P-15m.
-
-
Table 2. Calculated soil constants and modified soil constants Designed value Property Umtwei t r, Void ratio e Consohdabon yeild stress
P, WW
Compression index CC Coemcienr cb(cm'/d) Of consol'dabon Coemcient of well resistance
L,
M d f i e d value
Reclaimed Alluwal Reclaimed Alluvial layer strata layer strata Above ground water level . 18 6 Underground wafer level 9 8 (r)
1.20 5r,H
180
120
Cr,H
010/020
050
5000
300
180
5r,H 015/020
050/060
5000 / 1000 200 / 300
I 06
The reclamation works will be continued toward the second stage construction zone having a larger depth of water in the near future. And it is the authors' desire to scrupulously carry out future works while attention will be focussed on the foundation behavior of the first stage construction zone, doing continuously analysis and observation of data. Acknowledgements : My warmest gratitude and grateful appreciation go to Mr. K.Yamada and Fukken Co., Ltd. who gave helpful advice to complete this report. REFERENCES Monden, H., Y1963. Memoir, Faculty of Engineering. Hiroshima University (eds), A New Time Fitting Mefhodfor the Settlemetit Analysis o f Foundallon on Sofi Clays, 2( 1):21-29.
5 CONCLUSION Since this reclamation site made use of the leftover soils from construction works as reclamation materials, the characteristics of soils were widely scattered with the origin of soil. Sand seams (thin layers of sand) exist irregularly. Hence the soils were largely non-uniform and the evaluation of their characteristics was difficult. Thus the results from soil investigation together with those from instrumented measurement on the test embankment were effectively used to reflect them to the analysis, which made it possible to achieve a reasonably precise settlement prediction using a simplified one-dimensional consolidation computation. Moreover, in carrying out the accelerated consolidation works, the economic viability was given top priority, and the embankment loading method and the sand drain method with a wider spacing of sand piles were adopted. The objectives laid down at the beginning of works were fulfilled.
448
~ e l a ~ o nbetween s ~ ~ s~ ~ ~ ~ eand m elateral n t d ~ ~ ~ l a ~ofe soft m ~~n t ~ under embankment
~
1H.f.Chung & Y.S*Lee Korea institute of Construction TechmlogxSeotd, Korea
K. H. Kinn & K. N. Jin Kurea Land C ~ ~ ~ Seoul, ~ ~Korea a ~ u n ~
ABSTRACT : ~ e # ~ e m e nand t lateral disp~acementin clay f o ~ d a t i o ntreated with pack drain under embankment during construction and consolidation have been analysed. This paper described the relationship between settlement and lateral displacement of the soft ground at the test site. It presents an analysis of the settlement and lateral displacement data recorded, with concIusions regarding the practical effectiveness of the vertical pack drains installed. The vertical settlement and lateral displacements were evaluated using the field test data from i n s ~ e n t a t i o nof settlement plates, extensometers and inclinometers. The correlation between settlement and lateral displacement was obtained. These correlation can be used effectively for prediction of the rates and magnitudes of the behaviour of the soft clay treated by pack drains under embankment on clay foundation. pack drain. This study provides a means of evaluating the effectiveness of pack drains in improving soft ground subjected to embankment loading. The depth of soft g r o ~ is d 30m, and the depth of pack drains installed in soft ground is 25m. The pack drains were not installed from 25m below the ground level to the bottom of the soft clay layer. Thus, the depth of upper 5m at the bottom of the soft clay layer was untreated by pack drains. The e m b ~ e n fit1 t consisted of a granitic residual soil compacted to a unit weight of 17 kN/m3.The drain spacing are 0.8m, I .2m, 1.6m, 2.0m respectively. In order to obtain continuous settlement and lateral displacement distribution curves across the e m b a ~ e n tand to prevent disturbance of the construction work, the profile settlement gauge and inclinometer were used in this site. The magnitude and distribution with depth of the iateraf deformation 6, are fictions of the position of the inclinometer with respect to the e m b ~ e n t sIn. the present study, the lateral displacements observed at 5m apart from the toe of the e m b ~ e n slopes t are used. To investigate the variations of the magnitude of settlement and lateral displacements with both e ~ b a ~ e load n t and time, it has been found appropriate to refer to the maximum lateral displacement 6, observed along a vertical profile and to compare its variations to those of the maximum settlement S observed under the embankment.
It has been suggested that settlement and lateral displacement can be considered a good indicator of the stability of soft foundation under embankment. Marche and Chapuis( 1974)have suggested the use of observations of lateral displacement as a means of controlling the stability of embankments during construction. Their approach is based on the assumption that increased lateral displacements are a sign of incipient failure. A field test site was constructed to evaluate the performance of vertical pack drains for soft ground improvement. The test site was heavily instrumented, and good performance data were obtained. Vertical pack drains were installed with different type and spacing at this test site. The key objectives of the test site were the development of specific information regarding the behaviour characteristics of the marine soft clay layer under embankment. Works are being conducted for three years from 1996.8 to the present time. This field test site is located Yangsan area near Pusan in Korea. This study is concerned with the soft ground i ~ ~ r o v e ~ by e npack t drains at fufi-scale triai field test site. A comprehensive array of 4 types was buiIt in the same field test site to assess the relative efficiency of various pack drain schemes. A typical plan view of test site is shown in Figure 1. The test site is divided into 4 sections for different type of
449
2
G E O T E ~ ~ I C APROPERTIES L OF SOFT CLAY
The soft soil of field test site constitutes of marine clay up to 30m thick. The subsudace geology data at the site reveal the existence of a filled layer of about 1.5m above a 28.5m thick layer of alluvial soft silt clay. The sandy gravel layer is appeared below the bottom of clay layer beyond 30m below ground level. The artesian pressure of 33kPa is existed in the sandy gravel layer below the clay layer, and applied upward to the clay layer. The variations of density, void ratio, and consolidation parameters with depth for clay of test site are presented in Figure 2. 3 ANALYSES AND DISCUSSIONS OF RESULTS
3.1 Settlements and lateral d~splace~~ents with c ~ n s t r ~ stages c~~~n When an embankment construction begins on a clay deposit, a significant consolidation and shear deformation should be expected to occur in all stages of construction. Thus, settlement and lateral displacement are developed in the clay deposit under the embankment. The magnitude of settlement and lateral displacement during construction and at the end of construction is extremely difficult to predict. Because it depends essentially on the relative importance of the drained and undrained phases of response ofthe foundation to the conshction such as the rates of load application and of construction, the geometrical characteristics, and the mechanical properties of the clay. To investigate the variations of the magnitude of settlement and lateraf displacements with construction stages, settlement and lateral displacement under the embankment are analysed according to the elapsed time. Representative example of vertical settlement and lateral displacement measurements with construction sequence is shown in Figure 3. The magnitude of vertical settlement and lateral displacement are increased with construction sequence and the elapsed time (Indraratna et al., 1994). By examining Figure 3, it may be seen that the shape of 6=f(z), which corresponds to a homogeneous foundation, is more or less stable after the embankment construction. The magnitude and the depth developing maximum lateral displacement are presented in Table 1. Where, 6,,?.z,,, and H represent the maximum lateral displacement, the depth developing maximum lateral displacement, and the thickness of soft ground, respectively. An analysis of the data presented in Table 1 shows that average 6,, is about 36.98cm and average z, is about 0.2551-r.
450
displacement 6, remains small at the initial stage, on the other hand the lateral displacement increments are very large from the middle stage. 3.4 Correlation between settlement and lateral displacement with construction stages The behaviour of clay foundations under stage constructed embankments has been analysed in detail by Tavenas et al. (1978). Irrespective of the depth at which it occurs, the maximum lateral displacement developing during construction and consolidation can best expressed as a function of the settlement of the embankment. To investigate the variations of lateral displacement 6, with consolidation settlement S during one stage of construction and consolidation, the variations of 6, with S observed under embankments have been calculated from the field monitoring.
Figure 5. Lateral embankment loads
displacement
shapes
with
0.255H means that the depth developing maximum lateral displacement is 25.5% of total depth of soft ground below ground surface. 3.2 Shapes of lateral displacement with depth of soft ground
The magnitude and distribution with depth of the lateral deformation 6, after 8m height of fill are presented in Figure 4. In this figure, the normalised curves on 6/6, and z/H corresponding to the different observations are drawn. They confirm the stability of the distribution of lateral displacements with depth. Large lateral displacements develop in the upper layer. The lateral displacement with depth is decreased. 3.3 Lateral displacements with embankment loads
The variations of the maximum lateral displacement 6, with the embankment load q11yh observed during the construction in this test site are presented in Figure 5. In this figure the embankment load means the height of embankment fill. The average construction velocity of fill is 5.8dday in this study site. In all cases the lateral displacements are increased with embankment loads. The lateral
(d) pack -4 Figure 6. Variations of the maximum lateral displacement 6, with the settlements S
451
construction and 6, =O.O 1S+2 1.82 during the periods of consolidation after embankment. From these COrelating equations, we can recognise that t$e lateral displacement 6, increases linearly with settlement S during the periods of construction and consolidation.
Table 1. The ma nitude and the depth developing maximum lateral fiisplacement
I
Magnitude
1
6,(crn)
1
H(m)
1
&,(in)
I
z,/H
1
Remark
1
Pack- 1 Pack-2
27.14
30
9.0
0.300
Pack-3 Pack-4 Average
54.27 29.54 36.98
30
5.5 8.5 7.7
0.183 0.283 0.255
30
-
4 CONSLUSIONS
Table 2. Summary of the relationships of settlement and lateral displacement
Methods 1
2 Pack drain
3 4 Ave
I
During 1” Em bankrnent 6,=0.183-4.44 6,=0.19S-2.52 6,=0.208+0.43 6,=0.26S- 1.78 6,=0.193-2.08
During Consolidation 6,=0.01 S+18.08 6,=0.0 1s+2 1.77 6,=0.01 S+2 1.60
I
6,=0.0 I s+22.17 6,=0.01 S+21.82
Figure 6 presents the variations of the maximum lateral displacement 6, with the settlements S for the typical case of pack drains. There is a scatter of the data but a least square regression analysis indicates that correlation is established in terms of 6, = a -S@. In the construction and consolidation stages, the lateral displacements are much smaller than the settlements. And the lateral displacements are larger in the construction loading stage than in the consolidation stage. Generally the lateral displacements develop much more slowly than the settlements (Hartlen et al., 1996). This conclusion can also be applied to stage-constructed embankments. The relationships of settlement and lateral displacement can be expressed in terms of 6, = a .S+p, since the settlements S and lateral displacement 6,, are also variably affected by the duration of construction (Tavenas et al., 1980). The summary of the relationships of settlement and lateral displacement for all cases is presented in Table 2. The lateral displacements in these cases are ralated to the settlements by a linear relation of the type 6, = a .S$. During the initial construction phase, a has values varying between +0.18 and +0.26 and p has values varying between -4.44 and +0.43. During the later consolidation phase, a has value +0.01 and p has values varying between +18.08 and +21.77. It seems that a is reduced with time for the very long term, possibly as a result of the increasing importance of secondary consolidation phenomena. The total average 6, = a -S@ for prefabricated band drains are 6, =0.19S-2.08 during the periods of embankment 452
In this study, the improvement efficiency of soft ground below embankment stabilised with pack drains is analysed. The effectiveness of vertical drains could be evaluated by considering the surface settlement and lateral displacement changes in relation to the consolidation behaviour of the foundation clay layer below embankment. The settlement and horizontal displacements developing in clay foundations treated different pack drains during construction and consolidation have been analysed. The development of a significant consolidation at the beginning of any embankment construction has been confirmed. An analysis of the data presented in this study shows that average 6, is about 36.98cm and average is about 0.255H. The relationships of settlement and lateral displacement for pack drains are 6,=0.19S-2.08 during the periods of embankment construction and 6,=0.0 1S+21.82 during the periods of consolidation after embankment. This data indicate a linear increase of the maximum lateral displacement with the maximum settlement. These relationships can be used effectively for prediction of the rates and magnitudes of the behaviour of the foundation clay treated by vertical drains under embankment on marine soft soil. REFERENCES Korea Land Corporation (1999), “A study on the consolidation of soft ground”. Hartlen J, and Wolski W (1996), “Embankments on Organic Soils”, Elsevier, pp. 181-233 Indraratna ByBalasubramaniam AS, and Ratnayake P ( 1994), “Performance of Embankment Stabilised with Vertical Drains on Soft Clay”, J o f Geotech Engng, Vol 120, No 2, pp. 257-273. Leroueil S, Magnan J-P, and Tavenas F (1990), “Embankments on Soft Clays”, Ellis Horwood, pp. 47- 176 Tavenas F, and Leroueil S (1 980), “ The Behaviour of Embankments on Clay Foundations”, Can Geotech J, Vol. 17, pp. 236-260. Marche, R, and Chapuis R (1974), Controle de la stabilite des remblais par la mesure des deplacements horizontaux, Can Geotech J, Vol. 11(1), pp. 182-201
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Influence of the void ratio in soils treated with air foam and cement Y. Hayashi & A. Suzuki Department of Civil and Environmental Engineering, Kumamoto Universig, Japan
ABSTRACT: The properties of soil treated with air foam and cement for different void ratios were studied. The mechanical properties of the treated soils were greatly influenced by the void ratio. In this study, unconfined compression strength. which has previously been shown to have a profound relation to the void ratio, was found to be an important indicator of the stress level in the shear behavior of treated soil. The study aBo showed that the treated soils could be regarded as over-consolidated soils. The triaxial compression behavior depended on consolidation pressure divided by the unconfined compressive strength; however the failure strength did not depend on it. Table 1 Soil Property
1 INTRODUCTION Soils treated with air foam and cement are popular lightweight geomaterials which possess properties that make them very useful. The foremost property is the lightness of the soil. Ground constructed with a lighter soil intrinsicallytransmits loads a lesserload to the lower ground. Other merits include strength and density, which can be easily controlled, and soil, even if not suitable for construction, can be used as the raw material for treated soil. The treated soils can be also used in narrow space fields because of their high liquidity. Quality management of improved soils is often based on the unconfined compression strength. Hayashi and Suzuki (1998) indicated that the unconfined compression strength of soils treated with air foam and cement decreased as the void ratio (e)increased. In this study, the authors examined treated soLs with a microscope and conducted unconfined compression, triaxial compression. consolidation and permeabilitytests on the treated soilswith differente values Through the experimental results, this paper describes the importance of e and the unconfined compression test as indices in the evaluation of the treated soil properties. 2 EXPERIMENTAL METHOD 2.1 Raw Soil Dredged soil takes up about 45% by volume of dumped wastes in Japanese port (Tsuchida et al., 1996). The utilization of such soil is desirable, hence the soil for this study was taken from the Kumamoto port. The properties of the soil, classified as silty sand, are shown in Table 1.
wet densit v
liquid limit plastic limit clay contents silt contents sand contents
I
1.69 dcm3 30% 26 % 14%
2.2 Preparation Specimen The air foam was prepared from an animal-proteinfoaming material with a density of 0.031g/cm3 (Details in Hayashi et al., 1998). The air foam was mixed into a cement slurry with a 100% cement-water ratio. The slurry was then mixed into the prepared soil, with a 59% in water content. to maintain a homogenous mixture. The cement ratio was 0.2, and the ratio of air foam to dry soil was varied from 0.006 to 0.03 by mass, respectively. The specimens were prepared by tamping the mixture into steelmolds (150mm in diameter and 150mm in height) and were then cured for several days in a thermostatic room (20+3T). The specimens were removed from the molds and trimmed for each experiment. They were wrapped in polyethylene film to protect them from drying and were then cured in the same room. 2.3 Experiment Measurement of the air distribution in the treated soil Each specimen was divided into sections and photomicrographs were taken of several cross-sections 453
PermeabilitvTest
of each sample. Air voids were identified from the photomicrograph, and the area, number and shape of each air void were measured using a computer with imaging software.
The specimens measured 50mm in diameter and 100mm in length and curing times were greater than 56 days. The specimens were set in a chamber and loaded under a confined pressure of 0.03MPa and a back pressure of 0.025MPa through the lower pedestal into the specimen. The amount of water running from the top cap was measured.
Unconfined Compression Test The specimens measured 50mm in diameter and 100mm in length and the curing times were 4,7,14,28,56 and 112 days. The compression tests were conducted for an axial strain rate of l%/min. The load and displacement were measured.
3 RESULTS AND DISCUSSION
One-dimensionalConso lidation Test
3.1 Air-void Distribution
The specimens measured 60mm in diameter and 20mm in height and the curing time was 7 days. The loads of the experiment were varied from 0.0196 to 1.923 MPa and the loading period for each stage was 1 hour. The displacement was measured at fixed times.
Photograph 1shows three photomicrographs of the treated soil. The void ratio (e) and scales are also given. e is obtained the followingequation:
Consolidated-undrained Triaxial Compression Test As the permeability of the treated soil was very low, it sheared without drainage for ordinary loading. The triaxial compression test should be examined for unsaturated soil because the treated soil always exists in unsaturated condition.The experiment for unsaturated soil is not general method and the matrix strength of the unsaturated soil can be also obtained from effective stress for a saturated soil. The experiment was performed using following method. The specimens measured 50mm in diameter and 100mm in length and the curing times were 64 to 83 days. Each sample was saturated using vacuum procedure (Rad and Clough, 1984). After confirmation of saturation by determining that the Skempton B value was over 0.95, a fixed isotropic consolidation pressure, chosen from 0.03, 0.06, 0.09, 0.12 and 0.15 MPa, was applied to each specimen until primary consolidation was completed. The axial load was then applied with an axial strain rate of O.OS%/min under undrained conditions. The load, displacement, and pore pressure were measured under a compression of up to 15% axial strain.
,where pnJis the perticle density of matrix composed of soil and cement and pdis the dry density of specimens. In the photomicrograph of e=2.0, although very small, each air void can be seen clearly. As e increases, the air voids become greater and are transformed. The distribution of the air voids is shown in Figure 1.The diameters were calculated by assuming that each void was a circle and, because the small air voids were hard to discern, the small size air voids were adjusted by a calculated air-void content ratio. The frequency of air void not more than 0.05mm in diameters were over 60%, and it was higher against e. rhese distributions may be approximated by the following zxponential function:
, wherex is the air-void diameter and d is the mean and :he standard deviation of diameters. d is 0.065mm, 3.057mm and 0.042mm of 3.3, 2.6 and 2.0 in e , respectively. 3.2 Unconfined Compression Property The air voids in porous materials affects the strength, and the unconfined compression strength (4,) and modulus of deformation (E,,) decrease exponentially as e ofthe treated soil increases. This can be expressed by the Equation 3:
Figure 1 Air void diameter distribution
The coefficients a a n d pare soil constants. Hayashi and Suzuki (1998) have shown that the strength increases with age up to 56 days. Figure 2 shows the q, and E,, at 7,56 and 112days. The values for the treated soil aged for 7 days is easily distinguishable from the 56 and 112 day old specimens. However, the 56 and 112 day old specimens
454
could not be distinguished from each other. The determined constants a,/3, and multiple correlation coefficient R2are shown in Table 2. The qu value is affected more than the Esobecause the /3of q, is greater than that ofEso.
Table 2 Coefficients
Q and
56
Table 3 Experimental conditions Code 1-30 1-60
pr 1.02 1.02
e
3.31 3.34
o’, q u l o J c 30 0.052 60 0.101
3.3 One-dimensiona L Compression Property Figure 3 shows the e to Logp relationship of the treated soil. Aclear yield stress @,> can be appeared. The e value decreased a little with increased stress belowp,. However, abovep, e decreased profoundly as the stress increased. The compression index Cc varied from 1.5 to 3.0, independent ofe, and is much greater than for alluvialclays. The p, value is also related to e shown in Figure 4,and the q, line calculated from Equation 3 is also included in the figure. Thep, point falls somewhere between l.Oxqu to 1 . 5 4,. ~ 3.4 Undrained Triaxia L Compression Property The wet density (g), e value and consolidation pressure (( of the i, specimen ) submitted for this experiment are shown in Table 3. The far left column shows the experiment code used in the legend of the following figures. The d) q, ratio is considered the index denoting the reciprocal of the over-consolidation ratio.
Figure 3 e -1ogp relationshq
Photo I u-oss-section ot the treated soils
455
Figure 4 e - p
relationshq
Figures 5(a)-5(c) show deviator stress (4) to axial strain and excess pore pressure (Au) to axial strain (E,) relationships. In Figure 5(a), dcis 0.03MPa, 0.09MPa and 0.15MPa for the treated soilwithconstant e.Allvalues of q represented a clear peak value - failure strength (qJat small E,. After the peak they decreased as E, increased. The modulus of elasticity (E=dqld~,) at low strain was almost constant of all the treated soils. The tendency for q, to became smaller for larger dcis shown: the reason for this is discussed later. The behavior of du shows a dilation of the soil, and this behavior became profound as dcincreased. Such behavior is equivalent to that obtained in over-consolidated clays. Figure 5(b) shows the results of the treated soils with different e values but the same dc. q, increased as e decreased - a tendency similar to that observed in the unconfined compression test. The behavior of du became apparent as e increased. When q, decreases as e increases, (E,)
a) e k 3 . 3
it means that the dcto q, lessens, and so it shows the sameresult as that represent in FigureS(a). From the results shown in Figures 5(a) and 5(b), the authors found that both e and dcaffected the shear behavior. As e is related to q,, a stress parameter (dJq,)was defined, as shown in Table 3. The q and du values are normalized by 4,. The shear behaviors with constant d J q , are shown in Figure 5(c). Because each behavior ofq/q, and du/q, exhibited a similar trend among the three treated soils, q, seems to represent a stress parameter for the shear behavior of the treated soils with different e and dc. q, and theE are plotted withdc in Figures 6 and 7. Both values depended on e but not on dcand show some scattering. The q,and the E divided by q, or Esoare shown against dlq, in Figures 8 and 9. q was nearly equal to q,, whereas E was about 40% of ks0. The deformation property shown by the triaxial compression test appears to show some difference to that shown by the unconfined compression test. 3.5 Coefficient of Permeability The relationship between k and e is shown in Figure 10. Though the treated soil has much void, the coefficient of permeability ( k ) may increase as e increases, from 10-sto 10-6(cm/s) and the soil could be regarded as barely permeable geomaterial because of the independence of the air voids (see Photograph 1). Because the degree of saturation changed slightly after the experiment, the existence of a water route could be surmised, and the coefEicient of permeability k may be an apparent index.
b) o =0.09MP a
Figure 5 Stress - strain and excess pore pressure -strain behaviors 456
Figure 10 k to e
(3)Aclear yield stress could be obtained through the onedimensional consolidation test, and it was analogous to the unconfinedstrength. (4)The confiningpressure levelto unconfined compression strength (d,/q,) provided the shear behavior in the hiaxialcompressiontest. The failurestrengthwas almost equivalentto the unconjjnedcompressionstrength,while the modulus of elasticity was much smaller than the modulusof deformation. (5) The coefficientof permeabilitywas comparativelylow in spite of the high void ratio.
4 CONCLUDING REMARKS Asoil treated with air foam and cement is a singular,highstrength geomaterial in spite of the high void ratio. The. mechanical properties of treated soil with different void
The void ratio affected many properties of the treated soil and the unconfined compression strength was an important index of stress level. Inhomogeneity of the qualityoftreated ground is a serious problem in the fields. Density change due to defoaming of air foam can be considered as a factor of inhomogeneity, the authors hope this study will give some help to establish for evaluatingthe grounds.
ratioswere studied and the followingresultswere obtained.
ACKNOWLEDGEMENT
(1)The air-void diameterin the treated soils became bigger as the air-void ratio increasedand the distributioncould be supposed an exponentialdistribution. (2) The unconfined compressionstrength and the modulus of deformationdecreasedexponentially as the void ratio increased.
This studywas performed as a part ofjoint research project undertaken by Kumamoto University, Fut aba-Komuten Inc. and Nippon-Hod0 Inc. The authors are grateful to these organizationsfor financialsupport. The authors are also indebted to Ms. Y. Ujiyama and Mr. S . Hirano for assistancein carrying out these experiments.
457
REFERENCES 1)Hayashi,Y., Suzuki,A. and Kitazono, Y. (1998), "Effect of Soil Properties on the Improvement with Foam and Cement Milk", Environmental Geotechnics, A.A. Balkema, pp .637-642. 2) Tsuchida, T., Takeuchi, D., Okumura, T. and Kishida, T. (1996), "Development of light-weight fill from dredgings", EnvironmentalGeotechnics, Balkema, pp.4 15-420. 3) Rad, N. S., Cloughand G. W., (1984), "New procedure for saturating sand specimens", Journal of GeotechnicalEngineeringDivision, Proceedings of ASCE, V01.110, NO.GT9, pp.1205-1218.
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 15 1 1
Design of prefabricated vertical drain method on reclaimed marine clay H. Irnanishi, D. Zhang & S. Suwa Geo-ReseurchInstitute, Fukuoku, Japan
ABSTRACT: In the analysis of consolidation of soft ground improved by means of prefabricated vertical drain (PD), coefficient of consolidation with horizontal drainage (ch) is one of the important parameters. There are many laboratory tests and some measurement-based analysis had been carried out. However, the ch used to the PD method is still unsolved by these studies. Then, a series of field tests was conducted on an observation site of reclaimed ground. By changing the interval of drain spacing while keeping the loading and soil condition approximately same, the consolidation settlements of the reclaimed ground were measured. Based on the observed results and numerical analysis results, this paper examines ch of reclaimed ground. And then, the effectiveness and the cost of ground improvement were examined from both drains spacing and pre-loading.
1
However, it is not sufficient to find the mechanism and there is no established estimated technique. Design engineers and field engineers are always worried about the determination of coefficient of consolidation to estimate the construction period. Both c, and drain spacing (Ds) are very important parameters to design the PD method to fix the construction cost and the period, moreover the numerical analysis is important as well. We got an opportunity to measure the soft ground settlement improved by PD at test fields. This paper presents the knowledge from not only measured data by observational method but also calculated results by Finite Differences Method (FDM).
INTRODUCTION
Prefabricated Vertical Drains (PDs) have been carried out in several sites and several conditions in soft ground. The actual observed consolidation speed in the field is different from calculated consolidation speed on design. Their differences are caused by thin sand layers in the soft clay, equivalent diameter of PD; smear effect and well resistance (Yoshimi, 1979, Mizukami et al. 1996). They are concerned with drain distance (H) and coefficient of consolidation with vertical drainage (c,) of soft ground. The determination of c, is one of the most important things when vertical drain method is carried out. Many researchers have been studying about a consolidation speed, which is caused by PD due to the horizontal drainage in soft clay. The c, should be considered theoretically. Therefore many investigations about c, were carried out in laboratories. It was found that ch was larger than c, obtained from laborat07 test. On the other hand, it makes clear that the calculated consolidation speed used by c h in laboratory is faster than the observed speed. However, it is reported that the calculated consolidation speed used by c, in laboratory is well closed to the observed speed, The reason is "smear effect" which is produced around the drain materials in soft clay ground. There are a few studies about smear effect.
GEoTECHNICAL OBSERVATIONAL METHOD
AND
Figure 1 shows a ground plan of test fields. Test works were carried out on three cases, which are l.Om, 1.2m and 1.6m intervals of spacing respectively. Figure 2 shows the geologic column and soil properties by boring and laboratory tests. The seabed consists of Holocene marine clay in 7 meters thick and Pleistocene coarse sand in 10 meters thick on the Tertiary sand stone and shale. Dredged clay has been reclaimed in 10 meters thick on these strata. It is found that both reclaimed clay and 459
Holocene clay are in the state of unconsolidation and the mean c, is around 50 cm2/day by laboratory test. Sand seams can be recognized among the reclaimed clay by electrical cone penetration test, through which penetration resistance, pore water pressure and skin friction can be measured. Measuring devices in the soft clay consist of pore pressure meters and differential settlement gauges.
Differential settlement gauges positioned on thi border not only between reclaimed clay and marin1 clay but also between marine clay and sand. Afte covering the reclaimed clay with sand, we set U] those instruments in the clay.
460
Fig. 5 Measured Settlement Curve and
Fig. 3 Measured Settlement Curve and
Calculated Curve (Drain Spacing 1.6m)
Calculated Curve (Drain Spacing 1.0m)
Fig. 6 Drain Spacing and Reduction Ratio ch/cvO When Ds is 1.0 m, Chequals 0.5 times cv0;while for case of Ds being 1.2 m, c h equals 0.6 to 0.8 times cv0, and for the case of Ds being 1.6 in, c h equals 0.8 to 1.1 times cv0. By using these parameters, it could be found that the calculated consolidation curves by FDM showed good consistency with the observed data. Figure 6 shows the relation between Ds and the reduction ratio (c/cv0). The results suggested us as follows. (1) Based on the assumption that three testing fields have the same ground condition, it is found that c h is getting larger in proportion to increase of Ds. (2) Using c,, of the laboratory test, the observed consolidation speed is slower than the calculated speed by using Barron's equation. ( 3 ) cIJc,o of reclaimed clay is smaller than that of alluvial clay.
Fig. 4 Measured Settlement Curve and Calculated Curve (Drain Spacing 1.2m)
3 DRAIN SPACING AND COEFFICIENT OF CONSOLIDATION The calculation method was FDM, which consisted of both Terzaghi's consolidation equation and Barron's vertical drain equation. Figure 3 to Figure 5 show the comparison between calculated results and observed data. In these figures, ch was backcalculated with help of measured settlement result, while supposing the original c, obtained from laboratory test to be cvo. Relations of chand cvowere obtained as follows.
46 1
Fig. 7 Measured Settlement Curve of Field Test (Drain Spacing 1.0m)
Fig. 8 Compare of Measured Settlement Curve and Calculated Curve (Drain Spacing 1.Om)
It is considered that the reasons of these phenomena are caused by not only smear effect but also disturbed area, which occurred during penetration of mandrel to soft clay.
and alluvial clay by reclamation works. Figure 8 shows the relation between degree of consolidation and elapsed time. There are three curves by FDM. The curve A is calculated with ch taking account of the settlement before pre-loading. It can be fitted to the observed data. The curve B is calculated with c,, which is obtained from laboratory test. It is found that the curve B is overestimated compared with observed data. The curve C is calculated with c,, that did not consider the settlement before pre-loading, which showed an underestimated tendency. Therefore, it is very important to estimate the settlement before pre-loading and determination of ch.
4 IMPORTANT NOTICES FOR RECLAIMED CLAY Figure 7 shows one of the consolidation curves, which was obtained from test works. These observed data show that settlement started as soon as PD was driven to the reclaimed clay. It amounted to one third of total settlement before pre-loading. It was considered due to dissipation of excess pore water pressure, which occurred in reclaimed clay 462
Table 1 . Example unit construction costs.
PD c o s t Material
I
Land Fill cost Working
* Length of board =18.0m
Fig. 9 Examining on the Economy of PD Design
5 ECONOMY OF PD DESIGN From the results of these test works, it has been made clear that the consolidation settlement could not be speeded up, even when the distance between drain spacing was made small. This showed good consistency with the computing results by means of Barron's theory. The smaller the distance between drain spacing is, the higher the construction cost will be. Namely, using ,c, obtained from the preinvestigation borings to predict the consolidation settlement in field would probably result in uneconomic design. Therefore, it is concluded that, if the design was based on relationship of ch and drain spacing, the construction cost could be made cheaper.
Generally, in the design of drain spacing, to shorten the construction period, two aspects of filling and drain spacing are mostly considered. Here, as an example, basing on the ground condition of these tests works, the construction cost of ground improvement for each 10,000m2 is estimated by using PD method. Calculation of the construction costs is as follows; suppose the design load (P) being 2.5 m (assuming the residual settlement being zero), and the target degree of consolidation being 90%; the period of construction is from the start of PD construction to the end of target degree of consolidation. Trial calculations on the construction period and construction cost were made for five kinds of drain spacing cases: 0.8 m, 1.0 m, 1.2 m, 1.4 m, and 1.6 m, respectively. Table 1 shows the parameters used in the calculation. Figure 9 shows the calculation results. From these trial calculation results, it is made clear that; (1) To shorten the construction period, it is effective by shortening the distance of drain spacing. (2) Relationships between the height of filling and construction period are not always linear. When the height is greater than 3.5 m, the effect of the filling in shortening the construction period becomes weaker. ( 3 ) The most economic drain spacing is 1.0 m for the construction period of 160-220 days, 1.2 In for the construction period of 220-260 days and 1.4 m for the construction period of 320-360 days respectively, in this example. 6 CONCLUSIONS Based on the field test data of settlement and the analytical results, the following conclusions are derived. (1) For estimating the consolidation speed, it is necessary to make a modification of c, obtained from laboratory tests. The smaller the interval of drain spacing is, the smaller c h becomes. cdc,o for Holocene clay is larger than that for reclaimed clay.
463
(2) It is necessary to consider the settlement before pre-loading for giving a correct estimation of construction period. FDM is more useful than the conventional method. (3) The economies of PD designs were examined, basing on the costs of different test works. And then it was made clear that if the design was based on relationship of c h and drain spacing, the construction cost could be made cheaper. (4) When the construction period is given previously, it is possible to economize the cost of ground improvement by means of both drains spacing and pre-loading.
REFERENCES Imanishi H., Zhang D., 1999. Test Works concerning with Spacing of Prefabricated Vertical Drains, Proc. Second China-Japan Joint Symposium on Recent Developments of Theory & Practice in Geotechnology, pp. 185- 190, 1999. Kamon M., Pradhan T.B.S., Suwa S., 1991. Valuation of Design Factors of Prefabricated Band-Shaped Drains”, Geo-coast ‘91, pp.369-374.sa Mizukami J., Kobayashi M. and Tsuchida T., 1991. The horizontal coefficient of consolidation c/, Journal of JSCE, No.535/III-34, pp. 1-12. (in Japanese) Tanaka H., Ohta K. and Maruyama T., 1991. Performance of Vertical Drains for Soft and Uniform Soils (in Japanese), Report of the port and harbor research institute, Vo1.30, No.2, pp.212-227. Tanaka H., 1990. Settlement of ground improved by band drains at the development of Tokyo international Airport, 10th S.E.Asian Geotech. Conf., pp. 133-138. Yoshikuni H., 1979. Design and management of works in Vertical drain method (in Japanese), Gihodo.
464
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds)1( 2000 Balkema, Rotterdam, ISBN 90 5809 151 1
Fibredrain development, design and performance S.L. Lee, G. I? Karunaratne, M. A. Aziz & K.Y.Yong National University of Singapore, Singapore
ABSTRACT: Fibredrain is a prefabricated vertical drain made of natural fibres which are biodegradable, hence ecologically harmonious and environmentally friendly. The manufacturing process entails low energy consumption. The tensile strength of Fibredrain is such that it can withstand the stresses associated with deep installation as well as the densification of granular soil layers at the surface by heavy tamping. The apparent coefficient of consolidation Ch is a function of the permeability and compressibility of the clay and the field performance of the drain. The latter is influenced by the deformation of the core and the filter, the clogging of the filter and the kinking of the drain in the consolidation process. The design of soil improvement using Fibredrain is discussed. Finally the satisfactory performance of Fibredrain in several selected projects in East and Southeast Asia is reviewed. of the PVD between the top of the mandrel and the pulley at the top of the installation rig is known to course the tearing of the filter sleeve of some PVD. Adequate tensile strength in the PVD core and filter to withstand installation stresses is therefore beneficial. In many soil improvement projects, to minimise ground settlement, recently placed granular fill need to be densified while the PVDs are employed for subsoil consolidation under the fill, as shown schematically in Figure 1. Fill soils, even with a high ground water table, can be effectively treated with the application of high energy tamping. The PVD passing through the fill soils should be able to withstand the above high energy tamping without significant damage to its drainage performance. In addition, adequate tensile strength and flexibility are important for a PVD to withstand the sliding and heaving of soil layers in such treatment projects.
1 INTRODUCTION Infrastructure construction on soft clay leads to long term settlement problems. which are best dealt with by pre-consolidating the clay until the anticipated settlement under the design load is enforced prior to superstructure construction. Prefabricated vertical drains (PVD) are normally used to achieve the desired degree of consolidation within the project duration under a predetermined surcharge intensity, which should take the post-construction settlement due to secondary compression into account. An account of the development of PVD in Japan was given by Aboshi (1999). A PVD should facilitate permeation of water into the PVD from clay subjected to excess pore pressure and to convey the water axially to the drainage boundary. Both axial and filter permeability of the PVD are important in this regard. The filter cover should satisfy two requirements simultaneously, i.e. cross-plane water transmission and soil retention. The reduction of axial discharge flow capacity caused by the deformation of the core and the filter, the clogging of the filter and the kinking of the drain in the consolidation process must be taken into account. The cross-plane filter permeability. Should be large eneough to take advantage of the occasional pervious inclusions, such as lenses, seams and layers of sand and silt that exist in natural soil deposits. During installation in deep deposits of soft clay PVDs are subject to tension as the mandrel is withdrawn. Furthermore, the fluttering of the portion
Figure 1 . Sequence of ground treatment with Fibredrain and heavy tamping
465
Fibredrain reported in this paper is made of organic jute fibre and coconut coir. The core consists of four axial coir strands enveloped within the filter comprising two layers of jute burlap to form a rectangular strip measuring 80- lOOmm by 8- 10 mm. Three continuous stitches running longitudinally prevent the folding of the drain (Lee et al. 1995, Karunaratne et al. 1999). Figure 2 shows a bale of Fibredrain delivered on site. It has been used in many projects in Southeast Asia and Japan involving treatment of soft marine and fluvial clays as well as peaty clays in conjunction with surface densification using high energy tamping with satisfactory performance. Fibredrain is a green product. It is biodegradable, ecologically harmonious and environmentally friendly. From the environmental point of view, the energy consumption of jute production is only 15% of that needed for synthetic drains. The solid waste generated during production is basically organic, and biodegradable. Some of the more important laboratory tests conducted to examine the properties of Fibredrain are reported in the following, detail of which can be found in Karunaratne et a1 (1999). 2 AXIAL PERMEABILITY The discharge capacity of Fibredrain determined in accordance with ASTM D47 16-87 has a range of 200 m’lyr for a lateral pressure of 10 kPa and 20-30 m’/yr at 300 kPa (Lee et al. 1989a) as shown Figure 3. The discharge flow capacity decreases with increasing lateral pressure corresponding to the depth of installation which causes the deformation of the core and the filter. It is of interest to note that the field performance of PVD, for one way drainage, which is usually the case, requires maximum axial flow capacity at the surface where the lateral pressure tends to be small. 3 FILTER PERMEABILITY AND CLOGGING POTENTIAL Apparent opening size (AOS) of the two burlaplayer filter design of Fibredrain is in the range of 200-600 pm. The large AOS serves to tap pervious layers and lenses in clay deposits and the double filter layer intercept the clay slurry generated immediately after installation of Fibredrain in the clay as the mandrel is withdrawn. The cross-plane permeability of the filter was measured under a constant head across a stack of four identical burlaps fastened across an open end of a circular perspex cylinder (Lee et al. 1989a). Edges of the burlaps were sealed off with rubber membranes and a reinforcing coarse wire mesh was placed below the fourth burlap and clamped to the cylinder. The 466
cross-plane filter permeability determined in this way for clean water was better than 10-5d s . This is equivalent to the coefficient of permeability in fine sand which helps in tapping natural drainage layers such as sand and silt seams, lenses and other pervious paths. The clay slurry of water content ranging from 65% to 600% was passed through four burlap layers mentioned above under a pressure head of 0.5 m of water. About 2 litres of marine clay slurry at water content of 260% and 600% flowed out within 15 minutes. For Singapore marine clay at water content of 65% (LL = 70%), the passage of water was slower but clear almost from the beginning. The four burlap filter layers were then removed carefully and cleaned in an ultra-sonic agitator to extract the clay embedded in each burlap layer separately. Gravimetric analysis showed that soil particles retained on the outermost burlap layer and the immediately next inner layer, but no particles were detected in the third and fourth burlap layers (Figure 4a). The water passing through the four burlap layers was examined similarly to be free of any soil particles (Lee et al. 1989a). The retention of soil particles on the two filter layers under axial permeability test with marine clay at 65% water content is shown in Figure 4b. This study shows that clay of near liquid limit will not enter the drain core during the installation process as well as the consolidation process but will be retained by the two burlap layers. 4 KINKING Miura et al., 1995b compared the apparent coefficient of consolidation, Ch, of Ariake clay (W, = 86-97 %, Ip = 48-54%, Wn = 105-134%, Gs = 2.592.62) obtained with Fibredrain and another PVD, identified by PD, by installing both separately in 500 mm diameter and 500 mm high consolidation cells. On back analysis, the clay was found to have a Ch of 1.35 m’lyr and 2.19 m’lyr with PD and 9.2 m’lyr and 10.7m’Iyr with Fibredrain at consolidation pressures of 98 kPa and 294 kPa respectively. Figure 5 shows the deformed shapes of the two drains after carefully removing the clay around the drains at the end of consolidation. PD has deformed significantly resulting in kinking and decrease in axial drainage capacity. The deformation of Fibredrain, on the other hand, was largely confined to increase in thickness and longitudinal compression of the drain without kinking. Axial compression in the Fibredrain is largely manifested as an increase in cross sectional area due mainly to the unwinding characteristics of coir fibres in the filter layer as well as the core. The resulting increase in cross-sectional area and the decrease in filter opening size enhance the unclogged water flow into the drain.
Figure 2. Bale of Fibredrain and anchor shoes Figure 5 . Deformed shape of PD and Fibredrain (Miura et al. 1995a)
Figure 3. Discharge flow capacity of Fibredrain
Figure 6. Tensile test and biodegradability (Miura et al. 1995a) M Parbcle
50
5 TENSILE STRENGTH
S i z e Imrcron) (a 1
Jr
1
Particte Size (micron) Ib) Figure 4. Distribution of particles retained on filter layers: (a) filter permeability test, (b) axial permeability tcst
It should be observed from the above consolidation experiment that the apparent Ch for clay tested under lateral pressure is influenced by the reduction on flow capacity of the two drains due to kinking, clogging and deformation of the core and the filter. 467
Figure 6 shows the tensile strength of the Fibredrain under air-dry, after 4-day soaking in water, after 50day outdoor exposure and after consolidation in Ariake clay for 126 days (Miura et al. 1995a). The tensile strength varied from 9 16kg, 860 kg, 77 1 kg to 208 kg respectively for the cases considered. The strength deterioration with time in the Ariake clay shows the biodegradability of the fibres. It should be pointed out that the deterioration in strength does not imply a reduction in drainage capacity as observed in Pantai Mutiara reclamation project where Fibredrain performed satisfactorily for more than two years (Lee et al. 1988). Because of the high tensile strength, the robustness and the flexibility of the jute filter layers, Fibredrain can withstand high energy tamping and still retain its drainage functions unimpaired. It has the highest tensile strength (6.8 kN with 8.7% strain at rupture) of all PVDs in the market today, which helps in unimpeded and concurrent application with high energy tamping.
Table I . Recommended reduction factors for PVD (After Cause for reduction Deformation of filter into core space Deformation of corc and intrusion of‘ filter into core space Chemical clogging of filcr or core space Biological clogging of filter or core space Kinking due to settlement of clay
Notation Fi
Magnitude 1.5 - 2.5
Fd
1 .0 - 2.5
Fc
1.0- 1.2
Fb
1.0 - 1.2
where the average degree of consolidation U is defined by
Fk
1 .0 - 4.0
U = 1 - exp (-8Ch t /D2 p)
Table 2. Required discharge flow capacity, QT(m’/yr) Ch D Sf(m) (m’/yr) (In) 0.5 1.0 1.5 2.0 1.46 2.92 4.38 5.84 1 I 3.69 1.23 2.26 4.92 1.5 1.10 2.21 3.31 4.42 2
10.23 20.45 30.68 1 17.21 25.81 1.5 8.60 2 7.73 15.47 23.20 13.15 26.30 39.44 9 1 11.06 22.13 33.19 1.5 9.94 19.89 29.83 2 Note: Value of d = 0.0551n is assumed 7
40.91 34.41 30.95 52.59 44.25 39.77
2.5 7.31 6.15 5.52
51.13 43.02 38.69 65.74 55.31 49.72
6 DISCHARGE FLOW CAPACITY
The field performance of PVD depends on the interaction between the PVD and the soil. To account for reduced field discharge Row capacity of PVDs, the discharge flow capacity determined in the laboratory should be reduced by suitable reduction factors as shown in Table 1. Koerner (1997) suggested that the field flow rate Qf should be determined from the laboratory flow rate QIobtained from short-term laboratory tests in accordance with ASTM D4716, at lateral pressure of 200 kPa under a hydraulic gradient of 1.0 by a reduction factor R such that
R = (Fi x Fd x Fc x Fb x Fk)
(3) where Qf is determined by Equation (1) and Qr is estimated by
(5)
In Equations (4) and (3,A = 7cD2/4, SOthe initial slope of settlement-time curve, Ch the coefficient of consolidation in horizontal flow, D the influence diameter of a drain, t the time, p s ln(D/d)-% , d the equivalent drain diameter (Hansbo, 1979) and U the ratio of current settlement to the final settlement Sf. It is of interest to note that multiplying So from field settlement-time curve by A yields a good estimate of the drain performance and U predicted by Equation ( 5 ) using Ch obtained from laboratory test will yield accurate result only if Qr given by Equation (4) is smaller than Qf given by Equation (1). The value of Qr defined by Equation (4) are given in Table 2 for arguments of Sf, Ch and D. Referring to Table 1, kinking is not a problem for Fibredrain (Miura et al. 1995b) and hence Fk for Fibredrain is close to 1.0. For short duration projects Fc is low or close to 1.0 (Koerner, 1997). Mlynarek (1998) reported that biological clogging is not a problem if AOS is large. AOS of Fibredrain is in the range of 200-600 pm, hence Fb for Fibredrain is 1 .O. For Fibredrain the value of QI as defined by Koerner (1997) ranges from 40 to 60 m3/yr (Lee et al. 1999) at lateral pressure of 200 kPa under a hydraulic gradient of 1.O. The flow capacity at lateral pressure of 300 kPa ranges from 20 to 30 m’/yr, hence F, x Fd ranges from 1.5 to 3.0. The reduction factor R for Fibredrain is therefore in the range of 1.5 to 3.0.
(2)
The reduction factor R as suggested in Table 1 ranges from 1.5 to 36. The large range of values of R illustrates why some types of PVDs were only partially successful in some projects. If the required flow rate for a given project is Qr then the factor of safety FS is given by
Figure 7. Piezometric variation in clay and within Fibredrain, Hiroshima, Japan (After Yoshida et al, 1995)
468
7 DESIGN OF SOIL IMPROVEMENT WITH FIBREDRAIN
9 RECLAMATION AT PANTAI MUTIARA, JAKATA. INDONESIA
In the design of soil improvement work using Fibredrain, Sf and Ch can be estimated from the results obtained from soil investigation and the design load. The value of Qr can be obtained from Table 2 as a function of the adopted spacing and compared with the discharge flow capacity given in Figure 3 depending on the lateral pressure corresponding to the depth installed and the design surcharge. The latter is determined by the prescribed consolidation time. It should be observed that the discharge flow capacity shown in Figure 3 has taken into account the reduction factor discussed earlier as, for Fibredrain, Fc = Fb = Fk = 1 and Fi x Fd reflects the effect of lateral pressure which is taken into account in Figure 3. The design of the spacing to achieve a certain degree of primary consolidation, say 80%, under surcharge, which is equivalent to 100% primary consolidation under design load, within a prescribed time is determined as usual by means of Equation ( 5 ) . If Qf
For soil improvement in a reclamation for a housing project in Pluit area of Jakarta Fibredrain with surcharge was used to consolidate 16m to 18m of very soft (Cu = 2-14 kPa) to medium stiff clay (Cu= 20-40 kPa). The upper 12m of the deposit was of highly plastic silts and organic clay with w, = 5070%, WL= 60-130%, I p = 30-70%, void ratio e = 1.53.5 and C, = 0.7- 1.2. Fibredrain was installed at 2m spacing and the maximum flow through a drain due to the compression of clay volume within the tributary area yielded a rate of 4.7 m3/yr for the single drainage condition. In phase I1 of the project, for Ch of about 9 m2/yr, a drain spacing of 1.4 m was adopted to achieve the desired degree of consolidation in 3 months with an appropriate surcharge. Subsequently increase in undrained shear strength and ground settlement was observed to be satisfactory (Lee et al, 1988). In this reclamation, excavation for drainage channels in an area where Fibredrain was installed about eight years earlier, found no trace of its existence, verifying the biodegradability and hence ecological friendliness of Fibredrain.
QdQr.
In this regard, the discharge flow capacity of Fibredrain given in Figure 3 corresponding to the design lateral pressure can be used as QF, as the reduction effect has already been taken into account. For PVDs in general the reduction effect caused by the deformation, clogging and kinking of the drain caused by the lateral pressure, intrusion of clay and axial compression in the consolidation process must be taken into account.
10 TREATMENT OF EX-MINING LAND NEAR KUALA LUMPUR, MALAYSIA Fibredrain was used in conjunction with high energy tamping for the treatment of ex-mining land for a housing project (Lee et al. 1989b) for which a safe bearing pressure of 60 kN/m2 with a factor of safety of 2.5 was required. The site, composed of loose mine tailing soils, had N values less than 2 in clays and between 2 and 10 in sand. To treat the deep seated clay varying from 9 m to 21 m, Fibredrains were installed first at 1.5 m square spacing, followed by high energy tamping with a 15 tonne pounder falling freely from heights between 10 m to 25 m with 6 to 12 blows per pass for three passes over a 6m x 6m grid, together with a fourth ironing pass imparting an energy of 225 to 250 tm/m2 for dynamic compaction in the sandy areas. The energy intensity was increased to 270 tm/m2 - 335 tm/m2 for dynamic replacement (DR) in the clayey deposits. A surcharge fill of 4.5 m was placed and the resulting settlement was observed to taper off in 60 to 70 days. Figure 1 illustrates the concept of this treatment. Drains and surcharge combined with high energy tamping enforced anticipated total settlement of the order of 1 m. This case record shows that using Fibredrains, surcharge and high energy tamping application, a safe bearing capacity adequate for 5-storey residential houses can be easily achieved. For highway embankment construction, this treatment method can be easily adopted.
8 FIELD MONITORING IN HIROSHLMA,JAPAN Yoshida et al. (1995) reported the installation of Fibredrain in one project site in Hiroshima under the jurisdiction of Hiroshima Prefecture Government. The site was underlain by about 15-18 m of Hiroshima Clay. Fibredrain was installed at a spacing of 1.1m and sand drains at 2.5m spacing. A considerable excess pore pressure remained in the clay at the end of eight months in the sand drain area. Figure 7 shows the monitoring of piezometers installed in clay in the Fibredrain area at El -9.2,m, 12.5m and -19m, where the clay occurs between El 8.5m and -25m. Piezometers were also installed at El -10.6m, -13.6m, -16.6m, -19.6m and -22.6m within Fibredrain core to investigate the well resistance in the drain. Within the same eight-month period, the piezometers in the Fibredrain core showed only an excess pore pressure not exceeding 10 kPa indicating that well resistance is insignificant and the axial flow capacity is more than adequate. 469
advice rendered by Professor H. Aboshi, honourable adviser to Fukken CO Ltd., Hiroshima and adviser to Amano Corporation, Onomichi, and the support given by Dr T. Inoue, Fukken CO Ltd., Hiroshima, Japan as well as the contribution of Mr Ludi Bone, P.T. Indonesia Nihon Seima, Jakarta, in the manufacture and supply in strict accordance with the specifications of Fibredrain are deeply appreciated.
1 1 PEATY SOIL TREATMENT IN SINGAPORE
This site was a waterlogged land with peaty clay and fluvial deposits. In a series of field trials conducted in 1983 to evaluate the feasibility of stabilising some 7.8 m of peaty clay underlain by 5.6 m of fluvial clay (Lee et al. 1984), Fibredrains were installed at 2.2 m square spacing from a 1 m thick sand blanket. Subsequently, dynamic replacement and mixing (DRM) was employed in treating the upper layer of the peaty clay. DRM consisted of dropping a 15 tonne pounder through a height of 15 m to 20 m in six passes. A surcharge equivalent to 3.7 m of well rolled clayey sand fill was placed subsequently for verification. The settlement in the area treated with Fibredrain and DRM levelled off within 6 months compared with areas using drains and surcharge method and surcharge only where the settlement continued at the same rate.
REFERENCES Aboshi, H. 1999. On some problems of consolidation and soil stabilisation in soft clays. Proc. the Seng-Lip Lee symp. on innovative solutions in structural and geotechnical engineering, Bangkok, May 14- 15, 1999, 24 1-250 Hansbo S. 1979. Consolidation of clay by band-shaped prefabricated drains. Ground Etigimeritg, 12 (5), 2 1-25. Karunaratne G.P., Lee S. L., Aziz M.A. Yong, K.Y & Soehoed A. R. 1999. Fibredrain for soil improvement. Proc. the Seng-Lip Lee symp.on innovative solutions in structural and geotechnical engineering, Bangkok, May 14-15, 1999, 26 1-275 Koerner, R.M. 1997. Designing vt3irl7 gc.o.sytitheric.s, NJ: Prentice Hall. Lee S.L., Karunaratne G.P., Aziz M.A., Yong K.Y, Aboshi H. & Inoue T. 1999. An environmentally friendly PVD for soil improvement works. Proc. 11"' Asian regional conf on soil mechanics and geotechnical engineering, Korea, 1 : 43 1-434 Lee S.L.. Karunaratne C.P., Aziz M.A. & Inoue T. 1995. An environmentally friendly prefabricated drain for soil improvement. Proc. B. B. Broms symp. on geotechnical engineering, Singapore, 13- I5 December, 1995, 1-9 Lee S.L., Karunaratne G.P., Das Gupta N. C., Ramaswamy, & Aziz M. A. 1989a. Laboratory testing and field behaviour of Fibredrain. Proc. Symp. on application of gcosynthetic and geofibre in South Asia, Petaling Jaya, 1 - 17 to 1-25, Lee S.L., Yong K.Y., Tham K.W., Singh J. & Chen W.P. 1989b. Treatment of' ex-mining land by Fibredrains, surcharging and high energy impact. Proc. symp on application of geosynthetic and geofibrc in South Asia, Petaling Jaya, 5- 18 to 5-22. x e S.L., Karunaratne G.P. & Yong K.Y. 1988. Perforniance of Fibredrain in Pantai Mutiara. Proc. seminar on ground improvement application to Indonesian soft soils, Cawang, Indonesia x e S.L., Lo, K.W., Karunaratne, G.P. & Ooi, J . 1984. Improvement of peaty clay by dynamic replacement and mixing. Proc. Seminar on soil improvement and construction techniques in soft ground, Singapore, 208-2 14. .vIiura T., Tou M., Murota H. & Bono M. 1995a. The basic experiment on permeability characteristics of Fibredrain, (In Japanese). Proc. of annual regional meeting, Kyushu, JSCE Miura T., Tou M., Murota H. and Bono M. 1995b. Large Scale Consolidation test on drainage characteristics of Fibredrain, ( I n Japanese). Proc of annual meeting of JSCE Mlynarek, J. 1998. Panel discussion on filtration and drainage, 5'h Intern. conf. on geotextilcs, geomembranes and related products, Singapore, 4. 1383- 1385 '{oshida, Y.. Hamada. K.,Sakimori, H and Coto, H. 1995. Effectiveness of soil stabilisation of Fibredrain method. (In Japanese), Proc. of annual meeting of JSCE
12 CONCLUSIONS From the forgoing discussion the following conclusions can be made. 1. The apparent coefficient of consolidation Ch in field performance of PVD is a function of the permeability and compressibility of the soil as well as the property of the PVD, i.e., the deformation of the core and the filter, the clogging and kinking of the drain. 2. The clogging and kinking potential of Fibredrain is negligible and the reduction of axial discharge flow capacity of Fibredrain due to the deformation of the coe and the filter due to lateral pressure is taken into account in Figure 3. 3. The design of the spacing is governed as usual by Equation 5 and the required field discharge flow rate Qr by Equation 4 or Table 3 with Sf and Ch estimated from soil investigation results and the design load. If Qr is larger than the value given by Figure 3 corresponding to the design lateral pressure, the apparent Ch in Equation 5 should be scaled down in proportion to the ratio of the discharge flow capacity. 4. The high tensile strength and robustness of Fibredrain is beneficial to withstand installation stresses and application of high energy tamping commonly employed in soil improvement projects. 5. Fibredrain is biodegradable and requires low energy consumption in production from natural fibres, hence ecologically harmonious and environmentally friendly. ACKNOWLEDGEMENTS The co-operation of Drs S.D. Ramaswamy and N.C. Das Gupta in the development of Fibredrain and the
470
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Ground behavior during the consolidation by vacuum preloading of soft clay layer Kouki Matsumoto - Hazamu Corporufion,Tokyo,Jupun Kazuyoshi Nakakuma - Muruyama Industry ConzpanyLimited, Suitamu, Jupan Hiroyasu Shima - Kujitani Engineering Company Limited, Tokyo, Japan Hiroshi Ichikawa - Tokyo Consultant ConzpaizyLimited, Japan Goro Imai - Yokohuma Nationul UniversiQ,Japan
ABSTRACT: This paper describes ground behaviors during the consolidation by vacuum preloading of soft clay layer with vertical drains beforehand installed. Based on four practical cases carried out in the last two years, it was confirmed that the reduction in pore pressure within the ground caused by vacuum pressure occurs down to the bottom of the vertical drains, and that the influenced area of ground surface deformation is limited only to a distance nearly equal to the drain length from the edge of improvement area. Settlement measurements at the center of the preloading areas showed a fact that the traditional one-dimensional prediction method far overestimates final settlement, and it was suggested that the traditional prediction method should be revised from the viewpoint of isotropic consolidation characteristic in the case of consolidation by vacuumed water pressure.
1 INTRODUCTION Vacuum preloading method is one of the ground improvement methods for soft clay, in which vacuum pressure is applied to a ground area covered with an airtight membrane to pump out the pore water from the ground. Principles of this method were first introduced by W.Kjellman in the early 1950’s (Kjellman 1952). This method has been recognized to be effective for improvement of very soft soil, especially when surcharge material is deficient. A recent case history of the improvement by vacuum preloading at reclaimed land in Xingang Port at Tianjing in China (480,000 square meters) (Shang 1998) is reported. In Japan, this method was frequently used in actual constructions from 1960’s to 1980’s. Actual performance in their constructions were, however, not well because of the following reasons, then vacuum preloading method had not been used in Japan up to the time several years ago. 1) It is hard to keep air tightness. 2) There are limits in improvement depth. 3) Effectiveness is low for the ground where sand layers with high permeability are sandwiched. 4) High cost resulting from the sheet piles used to surround the improvement area aiming at the increase the vacuum degree. Taking these shortcomings into consideration, Nakakuma, one of the authors, has improved the materials used in the traditional vacuum preloading method, especially for airtight membrane and vertical drain. A schematic view of the vacuum preloading system improved by him is shown in Figure 1. 471
It is most essential for the design and execution of the vacuum preloading method that settlement and lateral displacement induced by the vacuum consolidation could be correctly evaluated. Lateral displacement of the ground must occur toward the inner side of the improvement area during vacuum preloading, because the reduction in pore water pressure must isotropically consolidate clay. This behavior is entirely different from the cases of the consolidation by conventional surcharging. In this paper, firstly, displacement behaviors of the ground surface which were observed at four sites in which the vacuum preloading method was used. (Ninomiya et al. 1998, Sandanbata et al. 1998, Matsumoto et al. 1998). Subsequently, the applicability of the conventional prediction method for the settlement during vacuum preloading is discussed by comparing the prediction with actual measurements.
2 CONSTRUCTION METHOD AND FIELD MEASUREMENT Construction procedures used in the vacuum preloading method were as fol1ows:l) Sand mat work, 2) vertical drains installing, 3) lateral drain tubes laying, 4) collecting tubes laying, 5) protection sheet laying, 6) airtight sheet laying, 7) connection of collecting tubes to a vacuum instrument, 8) loading by vacuum pumping , and 9) the termination of vacuum pumping.
Figure 1. Overview of vacuum preloading system.
Figure 2 . Soil profiles at each site.
Observational measurements done in the field work were the change in vacuum pressure between the ground surface and the airtight sheet, the pore water pressure within the ground along the center line among the vertical draind, and displacement of the ground surface.
Figure 3. Vacuum performance at each site.
3 FIELD CONDITIONS
4 OBSERVATION RESULTS 4.1 Vacuum pressure in the vacuum tank and
3.1 Improvement specijkation Table 1 shows the improvement specification for the four sites which are dealt with in this paper. The main purpose of the improvement was to accelerate consolidation and increase the strength of the ground.
beneth the airtight sheet Figure 3 shows the vacuum performance during the vacuum preloading. The vertical axis indicates the positive vacuum pressure measured from the atmospheric pressure. The pressure reduction under the airtight sheet was 10 to 25% smaller than that in the vacuum tank. And the absolute value of the pressure reduction at ground surface maintains a nearly constant at each site, and was about 60 kN/m2 at the site A & C and about 80 kN/m2 at about other sites; the average was approximately 60 kN/m2. These results are summarized in Table 2.
3.2 Ground conditions Figure 2 shows soil profiles at each site. The depth of installed vertical drains is 10 to 27 m. Any of the grounds consists of the layers of soft clay, organic clay and/or peat. A sand layer with the thickness of about 2m is sandwiched at the site A and D.
472
Table 1. Improvement specification and consolidation degree attainted by vacuum preloading. Vertical drain Vacuum driving Site
Improvement area (m2)
Depth
(m)
Spacing (m)
400 10,500 7,525 1.025
27 10 11.6 20
0.8 0.7 0.8,l.O 0.7
Tern (days)
Final degree ofconso'idation U(%)
36 50 30 42
Above80 Above 80
~~
A B C D
80
70
Table 2. The pressure reduction under the airtight sheet and inside of the vacuum tank by vacuum pumping. Specification per one vacuum um
Site
A
B C D
hea D ~ ~ : (m') (m)
400 2,182 1,818 512
27 10 11.6 20
pvolume (m')
10,800 21,820 21,095 10,250
Pressure of tank
Pressure of ground surface
(kN/m2) 80 40-80 80-90 80-90
(kN/m2) 40-60 40-80 50-60 70
vacuum
4.2 Pore pressure reduction in the ground Pore water pressure gauges were set up in a hole bored at the center of the adjacent four vertical drains at specified depths to monitor the pore pressure change in the ground. Figure 4 shows the reduction of pore pressure due to vacuum pumping at the final stage- The reduction of Pore pressme value from the atmospheric pressure is expressed in the abscissa; its origin means no reduction and 98 W / m 2 means the complete reduction by vacuum.
Figure 5. Settlement and lateral displacement of ground surface at the final stage of vacuum pumping.
Figure 4. Reduction of pore pressure due to vacuum pumping at the final stage. Figure 6 . Movement of the ground surface outside of the improvement area.
473
as positive value. Figure 6 shows the ground surface movement especially at the outside of the improvement area. The vertical axis shows their values normalized by the maximum settlement observed in the improved area. The horizontal axis is the distance from the border normalized by the length of vertical drain. From these figures, the following findings could be gotten. 1) Settlement at the border of the improvement area reaches 50 to 90% of the maximum settlement. 2) Inward lateral displacement of the ground surface is not so large outside the area apart from the border by the distance about 100 to 150% of the length of installed vertical drains. 3) The maximum inward lateral displacement is found at a distance within 40% of the drain length and reaches about 30 to 40% of the maximum settlement.
Lateral displacement (cm)
Lateral displacement (cm) 5
GL (m) 0
5
-15 -10 -
-20
-IS
-2 0
i I:
Figure 7. Lateral displacement within the ground at the border of improvement area.
4.4 Lateral displacement within the ground Figure 7 shows inward lateral displacement within the ground at the improvement area border. Its value is almost zero near the bottom of vertical drain, and gets larger toward the ground surface reaching its the maximum at the ground surface. Figure 8 shows its behavior normalized by the maximum settlement. The maximum inward lateral displacement at the ground surface is about 40% of the maximum settlement. In all cases, tension cracks caused by the lateral deformation were observed. They surrounded the improvement area with the maximum opening of about 5 cm.
..-
0.0
0.20 0.40
0.60 0.80
5 DISCUSSION ON SETTLEMENT PREDICTION
1.0
Normalized displacement by maximum settlement Figure 8. Lateral displacement at the border normalized by the maximum settlement.
The following results could be concluded from this figure: 1) The reduction of pore pressure in the ground is about 40 to 80 kN/m2, and their average is about 60 kN/m2. 2) The degree of pore pressure reduction is even at any depth from the top surface to the bottom of vertical drain. 4.3 Movement of the ground surface Figure 5 shows the shape of the ground surface at the final stage of vacuum pumping. In the vertical axis settlement and lateral displacement normalized by the installed length of vertical drain is plotted, and in the lateral axis the distance from the border of the improvement area. In the figure, the lateral movement toward the improvement area is defined 474
In the vacuum preloading, any soil element in the ground to be improved reduces its volume not only in the vertical direction but also in lateral direction, because vacuum pumping consolidates clay isotropically. In the conventional design method for settlement, however, one-dimensional deformation in vertical direction is assumed. The difference of the deformation mode was examined by using a twodimensional soil-water coupling FEM analysis (Matsumoto et al. 1999). Figure 9 shows a typical example of the analytical results for uniform ground. Calculated settlement in this case was only 62% of the one-dimensional settlement. 5.1 Comparison observations
of
analytical
results
with
The applicability of the conventional onedimensional settlement calculation method was examined by the use of FEM analysis for the site A, and its results were compared with the data observed at the site. “A” site is a fresh land reclaimed one and half year ago. Under the freshly filled layer soft soil
underlays with the thickness of about 35m. Vertical drains were driven down to the depth of 27 m below the ground surface.
Figure 12. Stress state before improvement.
Figure 9.Two-dimensional deformation analyzed by FEM. GL Log. 0
-2 -4
-6
(a) Schematic of desigh load
-8
$
-1 0
: ; r l
C O
-1 2 -
401d\I/m2 10
16 20
30
35 (days)
(b) Change of the pressure shown in Figure 13(a)
-42
Figure 13. Pressure change assumed in the analysis.
(m)
Figure 10. Ground condition at the site A.
Figure 14. Calculated settlement vs. observed results.
(1) Ground condition Ground conditions at the site A before improvement are shown in Figure 10. Figure 11 shows the relationship between the consolidation pressure p and the coefficient of consolidation c, obtained by the standard consolidation test. The minimum value of cvis about 250cm2/day.
Figure 11. Consolidation pressure p vs. coefficient of consolidation c,.
475
(2) Initial stress state in the ground The value of the consolidation yield stress pc before the improvement is smaller than that of the effective over burden pressure (initial stress) as shown in Figure 12. The initial stress value was calculated by assuming the static water pressure distribution. Based on the comparison shown in Figure 12, it was confirmed that excess pore water pressure accompanying the filling remained within the ground before the improvement by the present vacuum preloading. Settlement due to the dissipation of the residual excess pore water pressure was evaluated by the use of two-dimensional FEM analysis as well as the conventional one-dimensioned one. (3) Calculation method for settlement The C, (compression index) method was used to calculate consolidation settlement. Consolidation pressure by vacuum pumping was assumed to be uniformly loaded from the ground surface down to the bottom of vertical drain as shown in Figure 13(a). Furthermore, its value was assumed to change as shown in Figure 13(b). The parameters (C, and c,) were determined based on the results of the standard consolidation testing. The minimum cv value of 250cm2/daywas used for calculations. (4) Comparison between calculated settlement and measured results Figure 14 shows the comparison between the calculation results and the observed settlement. The calculated settlement well agrees with the measurement results due to the weight of the sand mat, which took place before vacuum pumping. This shows that the values of soil constants were correctly determined. However, regarding the settlement during vacuum pumping, calculated final settlement was 140 cm and the measured settlement was 60 cm. The measured settlement is only 43% of the calculated one. The value of difference is larger than the FEM results as shown in Figure 9. 6 SUMMARY Following results have been obtained: 1) It was confirmed that the reduction of pore water pressure within the ground caused by vacuum occurs down to the vertical drains, and their average is about 60 kN/m2. 2) The influenced area of ground surface deformation induced by the vacuum consolidation is limited only to a distance nearly equal to the drain length from the edge of improvement area. 3) The traditional one-dimensional prediction method far overestimates final settlement measured at the center of improvement area, and it was suggested that the traditional prediction method should be revised from the viewpoint of isotropic consolidation characteristic in the case of consolidation by vacuumed water pressure. 476
REFERENCES Barron,R.A. 1948. Consolidation of fine-grained soils by drain wells, Trans. ASCE, 113: pp.718754. Kjellman,W. 1952. Consolidation of clay soil by means of atmospheric pressure, Proceedings of Conference on Soil Stabilization, Massachusetts Institute of Technology, Boston, pp.258-263. Matsumoto,K. Ohno,M., Nakakuma,K., Shima,H., Ichikawa,H. & Imai,G. 1998. Study on applicability of vacuum consolidation method for deep soft clay ground, Proceeding of the International Symposium at Lowland Technology, pp.287-294. Matsumoto,K., Ohno,M., Koga,T. & Nakakuma,K. 1999. Comparison of deformation behavior between vacuum consolidation and surcharging with vertical drains due to finite element method, Proc. of 54* annual Conf. of Japan Society of Civil Engineers, 111, pp.520-52 1 (in Japanese). Ninomiya,H., Honda,K., Akashi,K., Hayashi,H., Umezaki,T. & Shiono,T. 1998. Improvement effects of soft clay ground due to newly developed vacuum consolidation method (Ver. 1 to Ver.3), Proc. of 33'd Japan National Conf. on Geotech. Engrg., pp.2 139-2144 (in Japanese). SandanbataJ., Koga,T., Ishihara,K.., Kato,T. & Nakakuma,K. 1998. Ground behavior during ground improvement due to vacuum consolidation (Vol.1 to V01.2). Proc. of 33'd Japan National Conf. on Geotech. Engrg., pp.1093-1096 (in Japanese). Shang,J.Q. et al. 1998. Vacuum preloading consolidation of reclaimed land: a case study, Can. Ge0tech.J. 35: pp.740-749.
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Simplified prediction of the shape of up-heaved ground caused by SCP N.Mori, T. Ito & T. San0 Hakodute Port Branch Ofjce, Hakodate Developnleizt and Construction Department, Hokkuido Development Agency, Japan
ABSTRACT: Sand compaction pile (SCP) method is a soil improvement method, where sands are compulsorily driven into the ground with vibration, replacing the soft clays with the dense sand columns. Hakodate Port Branch Office is constructing 14 m depth quay-wall and 12 m depth quay-wall in Hakodate Port, where soft siltclay layers up to approximately 30 m depth are being improved by the SCP method of 78.5% improvement ratio. Because the present depth at the construction site is shallow, the workability of construction vessels is highly affected by the lack of draft margin due to the upheaval by SCP driving. This paper reports a new method for predicting the shape and the height of upheaval due to the SCP driving. The method was actually used to the SCP work in Hakodate Port, and its applicability was evaluated to be appropriate in practice.
2
1 INTRODUCTION
Sand compaction pile (SCP) method is a soil improvement method, where sands are compulsorily driven into the ground with vibration, replacing the soft clays with the dense sand columns. The degree of improvement by SCP is usually presented by the improvement ratio or the replacement ratio, which is the ratio of volume of driven sand to the total volume of improved ground. When the improvement ratio is larger than 50%, the upheaval of ground cased by the driven sands must be taken into consideration in the design SCP method. At present, Hakodate Port Branch Office is constructing 14 m depth quay-wall and 12 in depth quay-wall in Hakodate Port, where soft silt-clay layer up to approximately 30 m depth is being improved by the SCP method of 78.5% improvement ratio. Because the present depth at the construction site is shallow, the workability of construction vessels is highly affected by the lack of draft margin due to the upheaval by SCP driving. This paper reports a new method for predicting the shape and the height of upheaval due to the SCP driving. The method was actually used to the SCP work in Hakodate Port, and its applicability was evaluated to be appropriate in practice.
477
PREVIOUS RESEARCH PREDICTION
ON
UPHEAVAL
The prediction of the shape of upheaval had been carried out by calculating the upheaval ratio, the maximum height of upheaval, and the average height of upheaval by using empirical equations with the parameters of improvement ratio and insitu shear strength. Fukute et al. (1988) presented a method to predict upheaval caused by SCP, in which the shape after the improvement is predicted by superimposing the standard shape for a row of improved ground. Fukute et a1 applied the method to the construction works at Kansai International Airport. Hirao et al. (1996) revised the method by comparing the measured shapes of actual upheavals with those predicted. In this study, the prediction by Hirao et al. is compared with the measured in Hakodate Port, and a new method is proposed.
3 PREDICTION METHOD Fig.1 shows the standard cross section of 12 m depth quay-wall. As the water depth before construction was shallow as 7.4 m, dredging of the seabed was necessary before SCP work in order to keep the draught of construction vessels after upheaval took place by SCP driving. The prediction of the shape of
Fig. 2 shows the flow diagram of the prediction. 4 METHOD TO PREDICT A BASIC SHAPE UPHEAVED GROUND
OF
4.1 Upheaval rate Fig.3 is the plane view of construction site, where Area A, B and C had been already improved by SCP method. Using the cquation provided by Hirao et al. (1996), the upheaval rates of Area A, B, and C, were calculated and compared with the measured values in Table 1. The prediction was made on the Area D where the improvement of SCP was to be newly carried out. Hirao et a1 (1996) proposed the following equation for SCP of 2 m diameter: y =2.117 / L + 0.718 a, + 0.056 (1)
Figure 1 Standard section of 12 m depth quay-wall heaved ground were carried out with considering the following points: 1. The upheaval rate y (the ratio of the total volumeof heaved ground to the total volume of driven sand) was determined based on the past experiences at Hakodate Port. 2. Considering the effect of dredging before SCP driving. 3. Determining a single row of upheaval caused by SCP as a basic shape, based on the past construction work in Hakodate Port 4. Predicting the overall shape of up-heaved ground by superimposing basic shapes
To determine the upheaval rate
where
,U :upheaval rate L: depth to be improved (m) a,: improvement rate
As shown in Table 1, it is found that the measured values are approximately 20% larger than those predicted by Eq.(l). Since this 20% difference cannot be ignored for the construction work, the upheaval rate was set at p=0.96, which is obtained by the actual measurement. To determine areas under influence of upheaval by considering: . direction in which SCP are driven . inclination of the ground caused by pre-
by reviewing the results from areas which have already been improved
excavation for the foundation
I
To predict the basic shape of upheaved ground
I
. to predict the amount of upheaved ground . to predict the height of upheaved ground
* To predict the shape of upheaved ground using calculation of superposition To predict the shape of upheaved ground in areas which have already been improved (calculating the coefficients of the shape)
I To predict the shape of upheaved ground in areas to be improved this year
1 To compare the predicted shape of upheaved ground o f areas to be improved this year with the shape acquired by actual measurement Figure 2. The flow diagram for prediction of the shape of upheaved ground
470
.. __--_--_--,.
Outside the port
Quaywall(-14rn) I Quaywall(-12m)
'
I I
1 1
' I
'Area&' New1 improved area (Area U) AreaA [AreaC I
Inside the port
Figure 3. Each improvement area
Figure 7 Predicted areas of upheaval (cross section )
-KIEVThhe drrccnon 10 whlch rand pdcs are dnvcn
0 =4j3
Figure 4. Areas e =60' under influence of upheaval according to the direction in which sand piles are dnven
.------;1-rI 0 ~ 6 0 +" n
a
=45'
- a
Figure 5. Areas under influence of upheaval according to the ground inclination
When the flat ground is up-heaved in the cross sectional direction, it is reported that the angle of the upheaval can be assumed to be 60 (National Fishing Port Association,1994). Based on this assumption and considering that the upheaval caused by SCP takes place with pushing the cohesive soil into the direction of a row of SCP, Hirao et al. (1996) determined the angle of upheaval as shown in Fig.4, which have been observed in the past construction works. When the ground is declined, it is said that the areas influenced by upheaval change due to the inclination as shown in Fig.5 (National Fishing Port Association, 1994). It is therefore reasonable to propose that the angle of the upheaval will change due to the ground inclination, when SCP are driven in the ground after the excavation. Accordingly, during the present construction work, predicted areas of upheaval on the cross section were considered as shown in Fig.6, by combining Figs. 4 and 5. On the area of upheaval on the longitudinal section, it has been found that the upheaval width in the area, which has been already improved, tends to be narrower than that in the area which is newly improved. Considering this fact and also the effects of ground inclination caused by upheaval in the area which has already been improved, we assumed areas under influence as shown in Fig. 7.
I-:::: 8 160'
e =4j'
- U
- a
Figure 6. Predicted areas under influence of upheaval on the cross section
4.2 Areas of" upheaval '
The areas of upheaval, that is the area influenced by the upheaval, should be determined both in the cross sectional and the longitudinal directions against the line of quay-wall.
Values acquired by actual measurement and predicted values Amount of upheaved Amount of sand to be Upheaval rate ground(m3) dumped(m3) Area A Area B Area C Average
30,005 75,480 35,670
29,155 8 1,997 38,037
1.03 0.92 0.94 0.96
479
Predicted values Upheaval rate 0.76 0.77 0.76 0.76
Calculation conditions a
47.9m
a1
18.0~1
b 53.1111
Calculation results
I 16.9~1
b, 34.0~1
V
h
7794m3
5.27m
4.3 Prediction on the basic shape
5.2 Shape of up-heaved ground on the cross section
Using the upheaval rate p=0.96 determined in 4.1, the volume of up-heaved ground in a block V will be calculated. The "block" defined here refers to each driven row of sand piles. In this construction work, three sand piles of 2 m diameter were driven simultaneously with a pitch of 6m and this process is repeated three times, and the width improved in one block is 18 m. The equation to calculate V is as follows:
We consider that thc height of up-heaved ground on the cross section means the average height of upheaved ground on the longitudinal section. Hirao et al. (1 996) used a standard shape of the cross section as shown in Fig. 9, which are determined by shape parameters HIIH,,,,, H2/H,,, and the location of H,,,,, showed by the ratio X /(B/2). In the present study, the same shape parameters were measured in the SCP works carried out in 1996 and 1997 at Hakodate Port, and obtained as follows:
V=(Design volume of driven sand ) X 0.96 (2) (Design volume of driven sand) = a1 X 61 X 1 X a,
HI = 0.89 H,,,,
HI= 0.34 HI,,
where ul = improvement length bl = improvement width 1 = improvement depth a, = improvement rate (78.5%).
According to the data by Hirao et a1.(1996), the location of H,,,, was assumed to be X=(B/2) X 0.3 (See Fig.9). After correcting the shape as mentioned above, the final shape of up-heaved ground in the cross section was predicted.
To calculate the upheaval height h, it is assumed that the shape of the area of upheaval will be a trapezoid. And h is given by using Obelisk's formula as follows:
where, a=length of areas of upheaval in longitudinal section b = length of areas of upheaval in cross section. For Area D in Fig.3, prediction was made with the conditions in Table 2,where calculated values of V and h are also presented.
Figure 8 Predicted shape of up-heaved ground (longitudinal section)
5 PREDICTION OF OVERALL SHAPE OF UPHEAVED GROUND 5.1 Shape of up-heaved ground on the longitudinal section
By considering that the upheaval height h will take place in the standard longitudinal section, the shape of up-heaved ground on the longitudinal section is predicted by superimposing the blocks of standard shape in their height (See Fig.8).
480
\ I
t -
The direction to which sand pilcs are drivcn
I /
Figure 9 Predicted shape of up-heaved ground (cross section)
Measuredshape
Block 6
.-
Block 5
Block 4
Block 3
Block 2
Block 1
- - - --I
I
I
Figure 10 Predicted shape of up-heaved ground (longitudinal section) The shape throughactual mmurcmcnt
1
XBm
I
Figure 13 Predicted shape of up-heaved ground ( cross section)
i
15m
'
18m
6 COMPARIZON BETWEEN PREDICTED AND MEASURED SHAPE ~
Figure 11 Relationship between draught of construction vessel and predicted upheaval ( in case with dredging)
i
1Sm
I
Comparisons were made between the predicted shape and the measured shape of up-heaved ground after the SCP driving in Area D. Fig.10 is the comparison of the shape in the longitudinal direction. As shown in Figure, the shape of the measured ground had a gentler slope than the predicted one. The reason for this is probably that, as the ground is very soft, the steep inclination of the ground immediately after the SCP driving may become gentler by the effect of the waves and currents. However, comparing the predicted and the measured, the accuracy of the prediction is satisfactory for the purpose of the construction control. Fig.10 is the comparison of the shape in the longitudinal direction. As shown in Figure, the shape of the measured ground had a gentler slope than the predicted one. The reason for this is probably that, as the ground is very soft, the steep inclination of the ground immediately after the SCP driving may become gentler by the effect of the waves and currents. However, comparing the predicted and the measured, the accuracy of the prediction is satisfactory for the purpose of the construction control. Figs.11 and 12 show the relation of the draught of construction vessels and the predicted upheaval by SCP. Fig.11 is the case that the dredging is carried,
18m
Figure 12 Relationship between draught of construction vessel and predicted upheaval (in case without dredging)
481
while Fig12 is the case without the dredging. From these figures, it is found that the draught of the vessel is enough in the case with the dredging, while the draught was not enough in the case without dredging. Fig.13 shows the comparison in the cross section where H,,,,, was observed. As shown in this figure, the prediction seems not to be good. This is because the directions of driving piles in the actual construction work were not strictly same as those of the prediction. For accurate prediction, the direction of SCP driving in the actual construction work should be taken into consideration.
7 CONCLUSION In this paper, the prediction of ground upheaval after SCP improvement work was made based on the previous researches and the experiences in Hakodate Port. The results are summarized as follows: 1. In the prediction method, the effects of dredging before SCP driving were newly taken into consideration for determining areas of upheaval. The shape of upheaved ground was drawn simply by superimposing standard shapes. 2. New shape factors were introduced to ensure accurate prediction of the shape of up-heaved ground based on the experiences in Hakodate Port. 3. Comparing the prediction with the measured shape of upheaval, it is found that the prediction was satisfactory. And the necessity of dredging was confirmed.
REFERENCES T. Fukute, Y. Higuchi, M. Furuichi & H. Tsuboi (1988): Prediction on the Shape of Up-heaved Ground Under the Sea Caused by iarge-scale Driving of Sand Compaction Piles. 33rd Symposium on Geotechnical Engineering, Japanese Socicty of Soil Mechanics and Foundation Engineering, pp23 -28. (in Japanese) T. Hirao, H. Tsuboi. H. Taga & M. Matsuo (1996): Prediction on thc Shape of Up-heaved Viscous Ground Under the Sea Caused by Driving of Sand Compaction Piles. 31st Meeting for the Study of Geotechnology: pp83 - S4. (in Japanese) National Fishing Port Association (1994): Guide to Designing Buildings in Fishing Ports: pp88 - 89. (in Japanese)
482
Modelling the effects of surcharge to reduce long term settlement of reclamations over soft clays D. ET. Nash Department of Civil Engineering, University ojBristol, UK
S.J. Ryde Richard Davies Associates, Bradford on Avon, U K (Formerly Bristol University)
ABSTRACT: Long-term settlement of reclamations constructed over soft soils may be reduced by use of surcharge, although there is often uncertainty over how long the surcharge should be maintained. A onedimensional finite difference consolidation analysis is outlined showing that vertical and radial drainage of a multi-layer soil profile in the zone of influence of a vertical drain may be modelled. The analysis allows inclusion of a zone of peripheral smear around the drain, permeabilities which vary with void ratio, and creep both during and after primary consolidation using an elastic visco-plastic constitutive model. Drawing on data from the Bothkennar soft clay research site, the model is used to predict settlements beneath a hypothetical test fill with time, including the secondary settlement behaviour when surcharge is removed. The potential application of the model in assessing surcharging strategy for reclamation schemes is discussed.
1 INTRODUCTION
more features of a structured clay. In this paper the application of the procedures to reclamation projects is discussed and illustrated by predictions of the behaviour of the soft clay beneath fill hypothetically placed at the Bothkennar soft clay research site.
Construction of reclamation over compressible soils frequently necessitates the use of ground improvement techniques to minimise postconstruction settlements. Vertical drains are often installed to accelerate consolidation. and sometimes a temporary surcharge is used to pre-load the ground. The principles for compensating for primary consolidation are well-established (Johnson 1970), but partial compensation for secondary consolidation is harder to achieve with confidence (Bjerrum 1972). Procedures for determining the rate and magnitude of secondary compression after surcharge removal are not well established and generally involve empirical relationships between the coefficient of C, for nornially secondary compression consolidated clay and that for the over-consolidated clay (Jamiolkowski et al. 1983, Mesri et al. 1994). Furthermore, it is observed that C, post-surcharge frequently increases with time. Similar problems are encountered in designing surcharge to embankments built on soft clays. The authors recently developed numerical procedures for analysing the consolidation of a soft soil adjacent to a vertical drain to explore the effects of creep, applied them to the back-analysis of the performance of some embankments built over estuarine alluvium in the UK @ash and Ryde, 1999). A simple isotache constitutive model originally proposed by Yin and Graham (1989, 1996) was adopted, and subsequently this has been developed to include
2 MODELLING CONSOLIDATION ADJACENT TO VERTICAL DRAINS 2.1 Background The consolidation of clay in the vicinity of vertical drains is the subject of several closed form solutions (eg Barron 1948. Hansbo 1981). but such solutions are quite restrictive. Numerical methods pro1 ide more flexibility and permit the inclusion of a multilayer profile, but while several such analyses have been developed recently (eg Mesri and Choi 1985). none was readily available which could model creep during primary consolidation. Accordingly the finite difference procedure BRISCON. was developed b! Ryde ( 1997) to model one-dimensional strains arising from vertical and radial flom? permitting the effects of non- linear stiffness, creep. and smear around a vertical drain, and anisotropic permeability varying with void ratio to be considered.
2.2 Finite diference procedure The one-dimensional consolidation equation for a soil element with vertical and radial flow may be expressed as:
483
where k, and k, are permeabilities in vertical and radial directions, U is excess pore pressure, m, is the elastic coefficient of volume compressibility and o, is the total vertical stress. The soil may be modelled as linear elastic (using a constant m,) or non-linear. with or without creep; the last term expresses the creep strain rate as described in the section below. Equation (1) has been expressed in implicit finite difference form (Ryde 1997). By dividing the soil into a series of sublayers and considering nodes at the centre of each, a solution is obtained in a similar manner to that adopted by Reece (1986) in the analysis of heat flow through a metal bar. Free drainage is assumed at the vertical drain and drainage boundaries, although the procedure may be extended to include the effects of drain resistance. As the analysis proceeds the coordinates are updated, with the values of soil permeability, stiffness and creep rate being those applicable to the current soil state. To simulate an equal strain condition the total stresses applied at ground level may be redistributed using an iterative procedure. During development, BRISCON was used to analyse some simple problems, and comparison made with closed form solutions (Barron, 1948) and solutions obtained using CRISP. Implementation of the creep model was checked by making comparison with problems analysed by Yin and Graham (1996). In all cases satisfactory agreement was obtained. 2.3 Simple isotache model for one-dimensional compression
(RTL). Subsequent creep is determined from a set of isotaches through the introduction of the concept of “equivalent time” te which is the time taken to creep under constant effective stress from the RTL (where fe is zero) to the present state. The total strain at any time is given by:
The equivalent time and creep rate are given by:
in which ryis the slope of the void ratio vs In([,) plot (similar to the conventional coefficient of secondary consolidation C,), vo is the initial specific volume, and f, defines the creep rate on the RTL. Thus 1, isochrones are isotaches or lines of constant creep rate. The parameters dpoand dp0’ which locate the RTL and to are curve fitting parameters which may be obtained from high quality oedometer tests. Figure 1 illustrates such a model with a linear reference time line. An element of soil initially at state A is subjected to an increment of total vertical stress Ao. During consolidation the state moves along a path such as AB, crossing various isotaches, and at point B the excess pore pressure has dissipated. Thereafter the clay creeps from B to D at constant effective stress. If the loading is reduced with a consequential reduction of effective stress say from C to E the creep rate is reduced to that of the corresponding isotache at E. Thereafter creep continues at a reduced rate towards point F.
A model for the creep behaviour of soft clays was outlined by Bjerrum (1972), who showed that onedimensional strain at constant effective stress increases linearly with logarithm of time. On a diagram of strain versus logarithm of effective Vertical Effective Stress (log scale) stress, this gives rise to a series of parallel lines each showing the strain at constant time; Bjerrum also stated that these were lines of constant creep rate Reference Time Line (isotaches). These ideas have been incorporated in equivalent time = 0 several models. but there has always been a difficulty in selecting the appropriate time origin. This difficulty is avoided by using a constitutive model in which creep rate is determined directly using isotaches, such as the elastic visco-plastic model proposed by Yin and Graham (1989, 1996). In this model, creep occurs throughout consolidation, which is different to that proposed by Mesri and Choi ( 1985) who argued that compression strains at the end of primary consolidation are unique. Yin and Graham adopted the A-K model used in critical state soil mechanics to define “instant” elastic-plastic behaviour, with the normal Figure 1. Void ratio vs. effective stress in EVP model showing consolidation line replaced by a reference time line path during consolidation and subsequent creep behaviour.
-r-----
I
484
2.4 Modelling creep behaviour ojstructured soils
3 CASESTUDY
The simple isotache model outlined above (herein denoted model 1) is able to predict many aspects of the behaviour of soft clays pertinent to reclamation projects. Since the creep rate is uniquely defined by the current state of the soil, the model may be used to predict the behaviour in unloading as well as loading. However the model also has several limitations. Firstly, the isotaches are linear whereas structured clays frequently exhibit curved normal consolidation lines. Secondly the separation of the isotaches is constant whereas tests show that at high stresses the value of v, which defines the separation of the isotaches. decreases. Thirdly there is no lower limit to creep and under small applied loading the model may predict unrealistically high creep rates. These problems may be partially overcome if natural strain E" is considered rather than engineering strain E (following Butterfield, 1979) where natural strain is defined as:
The models outlined above have been used to predict the behaviour of the clay at the Bothkennar soft clay research site in Scotland if a wide fill were placed. This site was chosen because the clay was the subject of an extensive collaborative research programme undertaken in the early 1990s. The ground conditions at Bothkennar are well described in a series of papers (Geotechnique 1992).
3.1 Ground conditions at Bothkennar The post-glacial organic estuarine silty clay at Bothkennar is of high plasticity. and its sensitivity from vane tests averages about 5. A desiccated crust 2m thick overlies about 16m of lightly overconsolidated soft clay (OCR about 1.6), beneath which is a layer of dense sand and gravel. During the characterisation study a large number of careful oedometer tests were undertaken on samples obtained with the Lava1 sampler (Nash et al. 1992), comprising 45 incremental load (IL), 24 constant rate of displacement and 13 restricted flow tests. Each IL test involved 20 to 30 load increments, which were generally applied daily. A typical loading sequence utilised four equal load increments up to the in-situ vertical effective stress. Then small load increments of around 10 kPa were used to define the yield stress, after which larger increments with a load-increment ratio of 1 were applied to a maximum stress of around 2000 kPa. Some tests were run with extended load increments, and many had unload-reload loops. Recently the data were reexamined to determine compressibility and creep parameters. An example of the test data is given in Figure 2, which shows data from a sample taken from half way down the profile. On a plot of engineering strain versus logarithm of effective stress. a line was carefully drawn tangent to the first part of the curved
(4) in which v is the specific volume as before. Butterfield showed that for many natural clays a plot of natural strain or logarithm of specific volume against logarithm of vertical effective stress is more linear than the usual semi-logarithmic plot. A second model has been developed in which equations (2) and (3) are formulated in terms of natural strain instead of engineering strain. Further curvature of the isoraches has been introduced in model 3 by using a power formulation as proposed by Den Haan ( 1992). In developing these models, the assumption (similar to Ca/Cc) has been made that the ratio remains constant along the isotaches, which leads to convergence of the isotaches at high stresses. The use of these models in assessing the effectiveness of surcharge in reducing long-term settlements of reclamation is illustrated in the next section.
Figure 2. Data from IL test 16B-3: a) experimental results and results predicted by BRISCON and b) experimental creep data.
485
TABLE I : Soil parameters derived from IL oedometer tests on Bothkennar clay. Layer
Thickness
y kN/m’
e,
RTL strain rate %/hr
Oi)epror q , C P ‘ I q)’
Orus1 1.5in Soft sitty clay 2.5m Soft silty clay 3.0m Soft silry clay 3.0m Soft silty clay 4.0m Soft silty claq 2.0m Soft silty cla) 2.0m Dense sand and gravel
16.87 16.38 15.60 15.77 16.21 16.73 17.I0
1.100 1.750 1.950 I.80O 1.700 1.550 1.350
150 1.6 1.5 I .5
I .55 I .4 1.3
1%
0.137 0.358 0.448 0.486 0.504 0.421 0.306
1%
1% 1% I% 1%
I%
0.014 0.024 0.030 0.032 0.033 0.028 0.021
0.03 0.04 0.04 0.04 0.04 0.04 0.04
0.03 0.03 0.03 0.03 0.03 0.03 0.03
kh twsec
1.OE-09* 1 .OE-09
1.OE-09*
I . 1E-09 I .9E-09 1.1E-09 6.4E-10 3.6E-10 8.6E-10 2.0E-10* 4.7E-10* denotes assumed valites.
1.2E-09 9.3E-I0 5.7E-I0
*
normal. consolidation line. Here the origin used for strain was taken after recompression to the in-situ stress. This line was chosen as the RTE for model 1, and its position was fixed by determining its slope 2.3A/v,, (or GJv,,), and the stress at 1% strain. Creep parameters were derived from plots of strain against logar~thmof strain rate, whose slope is equal to 2.3t,v/v,,. Figure 2b shows the data from test 168-3 for the three increments either side of yield plotted in this way. It may be seen that the slopes vary (increasing to yield and then decreasing), but that the creep rates at the end of each increment after yield were generally similar at about 0.03%/hour. This uniformity of strain rate justifies selecting a RTL equal to the 24-hour normal consolidation line. Using the data obtained from this test a check was made that the observed behaviour was satisfactor~~y predicted using BRISCON. The results are shown alongside the experimental data in Figure 2a where it may be seen that for stresses around yield there is good agreement. For clarity the unload-reload loop is only shown for the simple isotache model 1 (Yin and Graham, 19961, and while the model correctly predicts the swelling on unloading and creep when the previous maximum stress is approached, it does not predict the hysteresis observed in practice. At high stresses this model over-predicts the changes of
k,, ndsec
void ratio. Better agreement is obtained using models 2 and 3 based on natural strain which have a curved RTL, whose gradients were matched to that of the linear RTL at the reference point. It should be noted that creep occurs throughout consolidation, so during each increment the effective stress path is generally above the RTL. Load increments of longer duration would of course result in larger strains. 3.2 Numerical study The data from all the IL tests have been examined and soil parameters obtained for the whole profile. These are summarised in Table 1. The permeability data is based on results reported by Hight et al. (1992) in their table 6. The initial effective stress profile was based on hydrosta~ic groundwater conditions below 0.8m depth. which were also taken as the boundary conditions for consolidation. The effects of placing fill equivalent to 100 kPa on the ground surface with and without vertical drains and surcharge has been simulated using BRISCON. The profile of vertical effective stress before filling is shown in Figure 3, together with the final effective stress profile and the values of stress that were used to locate the RTL. It may be seen that, throughout the soft clay below the desiccated crust, the final effective stress is significantly larger than the yield stress ensuring that the clay will then be normally consolidated. First. benchmark analyses were carried out to predict the long-term consolidation behaviour with vertical drainage alone under 100 kPa, placed initially at 1 kPa/day. The three creep models outlined above were used, as well as three non-creep models in which the normal consolidation line was substituted for the RTL. and the long-term settlements are given in table 2. The time-settlement
TABLE 2: Settlements after 2 5 0 0 0 ~days under 100 kPa Model I . (Yin & Graham 1996) 2. (based on ~ u ~ e r ~ 1979) eld 3. (based on Den Haan 1992)
Figure 3. Profiles of vertical effective stresses.
486
Without creep 1.44m I .40m 1.29m
With creep 2.56m 2.4im 2.211~1
Figure 4. Predicted time-settlement behaviour under 100 kPa loading using models of type 2 (after Butterfield 1979).
behaviour for the analyses made using models of type 2 based on natural strain (Butterfield, 1979), are shown in Figure 4. This shows that without vertical drains, large settlements would occur over a period of around 35000 days (100 years), the long consolidation time being due to the long drainage path as well as the combination of low permeability and high conipressibility of the clay. The creep increases both the magnitude of the long-term settlement and the time for primary consolidation by about 70%. Next the effect of installing vertical drains prior to filling was considered. Analyses were carried out assuming a drain of effective diameter lOOmni to be at the centre of a unit cell 2 metres in diameter, with a smear zone of diameter 200mm. In the smear zone the horizontal permeability was reduced to be equal to the vertical permeability of the undisturbed clay. Figure 4 shows that the times for consolidation are reduced to around 1000 days (3 years), but that in the long term, the settlement increases to equal that calculated for the clay without vertical drains. Finally, the effect of placing a temporary 50 kPa surcharge in conjunction with the vertical drains was considered. After first analysing the effect of an indefinite 150 kPa loading, analyses were undertaken for the surcharge left in place for 1, 2 and 3 years (see Figure 4). It was found that when the surcharge was removed after 1 year (point A in Figure 4) there were negligible excess pore pressures at mid-depth, but that after a short period of swelling, significant creep settlements resumed (at a rate of 70mm/year). The profile of effective stress midway between the drains just before unloading is shown in Figure 3. which confirms that the stresses at that time exceeded the final stresses (after 487
surcharge removal) at all depths. The varying creep parameters result in a profile rather different from the normal isochrone shape. Leaving the surcharge in place for 2 or 3 years (points B and C in Figure 4) almost eliminated further creep movements reducing the rates of settlement to 3 and 0.3 mm/year respectively. 4 DISCUSSION
The EVP isotache models used here are able to predict many aspects of consolidation behaviour relevant to reclamation projects. In particular they are useful in predicting how long surcharge should be maintained to reduce creep after unloading to acceptable rates. The study above shows clearly that surcharging should be continued long enough to achieve the settlement that would occur in the longterm without surcharge, a criterion that may be hard to fulfil in practice. Parametric studies may readily be carried out with procedures such as BRISCON to explore different scenarios. However it must also be recognised that the models assume that the creep rate is only dependent on the current stress and void ratio, and that the value of ty (or C,) remains constant at large times, assumptions whose validity are uncertain (Yin, 1999). Daily load increments in IL consolidation tests can only provide useful creep data under sustained loading for relatively short times, and special extended testing would be needed to establish behaviour on unloading definitively. The simple isotache model can actually be used to make an estimate of the effect of removing a
surcharge without resort to a numerical procedure. With reference to Figure 1, at a given strain. reduction of effective stress from dl to dl (say from point C to E), results in a reduced creep rate and increased equivalent times which for t, )) to are approximated by:
(5) Equation ( 5 ) may be used in the field to provide an estimate of the reduction of creep rate when a surcharge is removed using data from field instrumentation. This may be illustrated using the data from the study above, for example by considering the effect of removing surcharge after I year (point A in figure 4). The profile of effective stress before surcharge removal is first compared with that in the long term (Figure 3). Dividing the soft clay into layers, the ratios of effective stresses are computed for the centre of each layer. and Equation ( 5 ) is used to calculate the reduction in average creep rate for each layer. and thus for the whole profile. Such calculations have shown that when the surcharge is removed, the average creep rate after pore pressure equilibration would reduce to around 30% of its previous value. This compares with a reduction of settlement rates to 9% of its previous value calculated using BRISCON. but some of the settlement occurring prior to unloading arises from pore pressure dissipation rather than creep. A better comparison is obtained when surcharge is removed after 2 and 3 years; equation ( 5 ) implies that creep rates are reduced to 1.1TOand 0.5% of their previous values, which compare favourably with 1.1% and 0.6% obtained from BRISCON. Such agreement between BRISCON and equation ( 5 ) for surcharge removal after 2 and 3 years is expected, since by then the excess pore pressures have almost completely dissipated. At earlier times the simple approach using equation ( 5 ) underestimates the reduction of settlement rate achieved, which in practice would generally be conservative.
5 CONCLUSIONS The consolidation of soft soils around vertical drains frequently presents difficulties to designers of reclamation schemes if there is significant creep. The elastic visco-plastic models utilised here enable simple calculations to be made to assess surcharge effectiveness where primary consolidation occurs quickly, and their incorporation in the finite difference procedure BRISCON facilitates design to reduce secondary settlements even if primary consolidation is not complete.
6 ACKNOWLEDGEMENTS The authors are grateful to the Engineering and Physical Sciences Research Council for sponsoring the initial part of research. REFERENCES Barron, R.A. 1948. Consolidation of fine-grained soils by Drain Wells. Trans. ASCE, 113:718-754. Bjerrum, E. 1972. Embankments on soft ground. State of the Art report. Proc. Spec. ConJ on Performance of Earth and Earth-supported structures, Purdue University. 1 : 1-54. ASCE. Butterfield, R. 1979. A natural compression law for soils (an advance on e-log p ' ) . Geotechnique 29(4):469-480. Den Haan, E.J. The formulation of virgin compression in soils. Geotechnique 42(3):465-484. Geotechnique (1992). Bothkennar soft clay test site: characterisation and lessons learned. Symposium in Print. Geotechnique 42(2): 161-378. Hansbo, S. 1981. Consolidation of fine-grained soils by prefabricated drains. Proc. lUth lnt. Cons on S M F E , StockhoO?1.VOI 31677-682. Jamiolkowski, M., Lancellotta, R. & Wolski, W. 1983. Precompression and speeding up consolidation. General Report to Spec. Session 6. 8'" Eur. ConJ on SMFE. Helsinki. Vol. 3:1201-1226. Johnson, S.J. 1970. Precompression for improving foundation soils. Proc. ASCE, 96(SM1):1 11-144. Hight, D.W., Bond, A.J. and Legge, J.D. 1992. Characterisation of the Bothkennar clay: an overview. Geotechnique 42(2):303-348. Mesri, G. & Choi. Y.K. 1985. Settlement analysis of embankments on soft clays. Proc. ASCE. 1 1 I(GT4): 441464. Mesri G., Lo, D.O.K. & Feng, T.W. 1994. Settlement of embankments on soft clays. Vertical and Horrzontd dejormations of foundations and embankments. Geot. Spec. Pub. NO 4018-55.ASCE. Murray R.T. 197 1, Embankments constructed on soft foundations: settlement study at Avonmouth. Road Research Laboratory, Report LR 4 19. Nash, D.F.T., Sills, G.C. and Davison. L.R. 1992. Onedimensional consolidation testing of soft clay from Bothkennar. Geotechnique 42(2):24 1-256. Nash, D.F.T. and Ryde, S.J. 1999. Modelling the effects of surcharge to reduce long term settlement of an embankment on soft alluvium. Proc. 12"' Eur. Cons on SMGE, Atnsterdatn, Vol. 3: 1555- 156 1. Reece G. 1986. Microcomputer modelling by finite differences. Macm illan. London. Ryde, S.J. 1997. The performance and back-analysis of embankments on soft estuarine clay. PhD Thesis. University of Bristol. Yin, J-H. & Graham J. 1989. Viscous-elastic-plastic modelling of one-dimensional time-dependent behaviour. Canadian Geot. Jnl. 26(2): 199-209. Yin, J-H. & Graharn, J. 1996. Elastic visco-plastic modelling of one-dimensional consolidation. Geotechnique. 46(3):5 I 5 527. Yin, J-H. 1999. Non-linear creep of soils in oedometer tests. Ge'otechniqtre. 49(5):699-707.
CoastalGeotechnicalEngineeringin Practice,Nakase & Tsuchida (eds) 0 2000 Balkema, Rotterdam, ISBN 90 5809 151 1
Stability of soft clay improved by SCP with low replacement ratios under backfilled caisson loading Md. Z. Rahman, J.Takemura & T.Mizuno Tokyo Institute of Technologx Japan
M. Koda Railway Technical Research Institute, Tokyo,Japan
ABSTRACT Stability of clay improved by SCP with low replacement ratio under backfilled caisson loading is investigated. The influence of various factors especially weight of the caisson is revealed by series of centrifuge model tests. The test results show that increase in caisson weight can decrease the lateral displacement of the caisson under backfill loading. Stability analysis shows that there is an optimum caisson weight for a specific backfill load and improvement condition beyond which the factor of safety becomes smaller. So it is confirmed that increasing the caisson weight to some extent can increase the stability during backfilling. 1 INTRODUCTION
2 CENTRIFUGE MODEL TEST
Sand Compaction Pile (SCP) Method with low replacement ratios is one of the alternatives to reduce the construction cost of SCP improved soft seabed in harbor construction works (quay wall, sea wall etc.). But relatively larger settlement and lower stability of low replacement ratio SCP are impeding the practical use of this method. In recent years, several researches through model to full scale tests (Yagyu et al., 1989, Nakase & Takemura, 1989, Terashi et al. 1991, Rahman et al., 1999b) have been conducted on SCP with low replacement ratios to study the bearing capacity, settlement, short and long term stability and deformations. The previous studies show that behavior of SCP improved ground is not clearly known and depends on several factors. Since the shear strength of the sand is directly proportional to the normal effective stress on the shear plane, the caisson weight is one of the important factors to be investigated especially for the low replacement ratio SCP ground. This study focussed on the stability of soft clay improved by SCP with low replacement ratios of %=30%-50% subjected to gravity caisson and backfill loading. The investigation is carried out by centrifuge model tests and stability analysis. The effects of different parameters have been investigated especially the weight of the caisson. From the investigation it is confirmed that the increase of the caisson weight to some extent for a specific backfill pressure and improvement condition can increase the stability of the low replacement ratio SCP ground.
Ariake clay was used for model soft ground, for SCP and sand drain Toyoura sand was used whereas Zircon sand was used for backfill material. The clay has the following major properties; specific gravity, G, =2.67, liquid limit =71 .l%, plastic limit =38%, compression index, C, =0.52, strength increase ratio cJp0.42. The properties of Toyoura sand are; specific gravity, G, = 2.64, maximum and minimum void ratios are 0.974 and 0.609 respectively. The relative densities of SCP and SD are 80% and 30% respectively. The angle of internal friction of the Toyoura sand with Dr=80% under triaxial compression condition is about 40". The properties of Zircon sand are; specific yavity, G, =4.6, submerged unit weight y'=20 kN/m and the angle of internal friction is about 40". To simulate the construction of the caisson, the empty caisson box was filled in-flight by water and ZnClz solution in two different series of tests respectively. The specific gravity of ZnCl2 solution used is 1.77. The observed tilting of the caisson was less than 3", the effect of the tilting on the position of the center of gravity of the caisson can be negligible. The model preparation and test procedure is presented here in brief. Details of the centrifuge model test is given by Rahman et al. (1998). Tests were performed using two strong boxes in two series. The small box was 500nlm in length, 350mm in height and 150 nun in width and the large box was 700mni in length, 450mm height and 150mm in width. Fig1 shows the schematic illustration of test setup showing different parameters and dimensions adopted in the small box. The clay was remolded and poured 489
into the strong box. A drainage layer of 20mm in thickness was prepared on the base of the strongbox prior to the clay pouring. Preconsolidation pressure of 5 kPa was applied to the clay layer using bellofram cylinder on the lab floor and after completion of the preconsolidation, the specimen was consolidated under 90kPa seepage pressure (Takemura et. a1 1991). On completion of consolidation, markers were placed on the front side of the clay block.
Figure 1 Test setup (all dimensions in mm)
Figure 2 Loading process with time
490
Thickness of the model ground was 150mm which correspond to 15m in prototype under the centrifugal acceleration of lOOg employed in the tests. The SCP and SD were installed in a triangular arrangement. The diameter of the model SCP and SD were 15 mm and 7 mm respectively. After preparing the model ground the earth pressure cells and pore pressure transducers were inserted to the predetermined positions. The model caisson with width and height of 75mm (7.5m in prototype) was made of the steel wall box with frictional base and instnunented with two pressure cells on the inside bottom to measure the pressure of the liquid filled. It was placed on SCP ground and Toyoura sand was spread on the both side of the caisson to provide surcharge pressure of 5kPa under lOOg field. Potential meters and laser displacement sensors were installed. The model was then taken on the centrifuge and sand hopper and liquid tank were mounted on the strong box. Water was then poured into the both side of the caisson to a height of 6Omm above the ground level and maintained constant throughout the test. The average contact pressure from the empty caisson in partially submerged condition under 100g is 20ltPa which was 15kPa higher than the surcharge pressure. Centrihigal acceleration is increased stepwise by log upto 1OOg. On completion of the dissipation of excess pore water pressure, the caisson was filled with water in series 1 and with ZnC12 solution in series 2 in-flight at a constant rate from the liquid tank on the strong box. In series 2 care was paid so that the caisson was not overflowed. The increment of the caisson load with time is shown in Fig.2(a). Average increments of caisson pressure are 70 and 115 kPa in series 1 and 2 respectively. After filling the caisson, consolidation of model ground was allowed for 10 min equivalent to about 10 weeks in prototype scale. Zircon sand was then rained down on the SD improved part of the model ground to simulate stage construction of backfill by using the sand hopper. Due to manual operation of the sand hopper, there were some differences in loading process between test cases as shown in Fig. 2(b). Lateral placement a ! c i se;*lmentof paisscii was mmitored by ~ * : n gi:ldisplacemeni sensors and potential niztyrs respzctively. Centrifugation has continued tijl tile consolidation of model ground under fill exceeded 90%, which ~ i a sconfirmed by the pore water pressure transducers. Photographs were taken to measure the displacement of the ground. Water content at different depths and locations were measured after tests. Replacement ratio (aJ, width of SCP portion (W) and caisson weight (W,) are the parameters studied in the tests. 30% and 50% replacement ratios, the width of the SCP portion of 1.2B and 2B and the caisson weight of 85 and 130 kPa were employed in the tests. Test conditions are given in Table 1.
Table 1: Test conditions Series Test a,
W
(%I -
casel 30 1.2B case2 30 2.0B 50 1.2B case3 case8 30 1.2B L case9 30 2,OB case11 50 1.2B B (width of caisson): 75nm (7.5m 1
?-r
Backfill load, Wbf (@a’) 173 I40 140 140 140 140 in prototype)
Net caisson load, W, (Wa) 85
130
3 TESTRESULTS All the test results are given in prototype scale. The settlement of the caisson during caisson loading is plotted against the increment of caisson load as shown in Fig.3. The settlement of the caisson in case 1 was not measured accurately. The settlement increases with the load and no marked change in the slope of the settlement-load curve even upto the increment of the caisson load of 115 kPa except of case8. In case8, a large settlement was observed at load increment of 65kPa without increasing the load. This was due to consolidation of the SCP ground as the load from caisson was kept constant for about 100 days as seen in Fig.2, which allowed the ground to be consolidated. The slope of this curve before and after the load increment of 65 kPa is same which meaiis no failure was occurred. The deformation of SCP ground due to the caisson loading was mainly vertical. whereas the deforniations of the SCP ground and the caisson due to backfilling were mainly horizontal. The lateral displaceineiit during backfilling with backfill loads is shown in Fig. 4. The lateral displaceineiit of the caisson increases with the backGll load and similar trend is observed in all cases irrespective of the iniproveinent and loading conditions. In the previous study (Rahman et al. 1999b) it was confirmed that the lateral displacement of the caisson could be reduced by both increasing replacement ratio and widening the SCP improved portion toward the fill, when the improvement width is two times the caisson width and the SCPs are installed under the caisson and extended toward the fill, the effect of increasing the replacement ratio from 30% to 50% is not significant. The lateral diyplacement of the caisson is also dependent on the rate of backfilling rate. In this study, although there were some differences in backfilling process among the test cases. the investigation is made ignoring that effect. Figure 5 shows the effects of the improvement width (W) and the net caisson weight (W,) on the lateral displacement of the caisson due to backfilling. The net caisson weight is the total load intensity of the caisson minus the surcharge pressure (5kPa). The lateral displacement is the one measured at the backfill load of
140 kPa, as the backfill pressure in casel is higher than other cases. The figure shows that increasing the net caisson load from 85 kPa to 130 kPa is effective in reducing the lateral displacement of the caisson under the conditions in the tests. Widening SCP portion has significant effect on the reduction of the lateral displacement of the caisson. It is intereating to see that lateral displacement of the caisson is almost the same ur,Cir the both light and heavy caisson when the improvement width is two times the width of the caisson. So, for this backfilling condition the improvement width two times the caisson width is enough to ensure the short term stability.
Figure 3 Vertical settlement of the caisson vs increment of the caisson load
Figure 4 Lateral displacement of the caisson with backfill load
Figure 5 Effect of lateral improvement width and caisson load on the lateral displacement of the caisson
491
the caisson 15 months after the construction are plotted against W and W,. The long term behavior of the SCP ground under this type of backfilled caisson loading is relatively complicated. The settlement of the caisson after the construction can be divided into two phases. One is immediately after the construction the settlement of the caisson mainly caused by the consolidation at SCP portion and the other is the settlement mainly caused by the differential settlement between the SCP and SD portion after finishing the consolidation at SCP portion. In the later phase, the settlement depends on the stiffness of the SCP ground under the caisson as well as the load applied to that due to differential settlement. The widening the SCP area toward fill reduces the load due to this differential settlement on the SCP area under the caisson, consequently the settlement of the caisson reduces. The increase of the caisson load increases stiffness of the SCP portion under the caisson. As a result the settlement is also reduced. So, the wider improvement area toward the fill and the increase of the caisson load to some extent can reduce the long term settlement thus increase the long term stability.
Figure 6 Settlement of the caisson after backfilling with time
4 STABILITY ANALYSIS Factors of safety of the iinproved ground was calculated by using Modified Felleiiius Method. The shear strength of the composite ground is obtained by calculating the shear strength for sand and clay part separately along the sliding surface as shown below. for sand pai-t:
Figure 7 Effect of improvement width (W) arid net caisson load (W,) on long term settlement
When the overall stability is considered, the long term stability should be investigated from the deformation point of view especially for the low replacement ratio SCP ground where the deformation after the construction is relatively large. The settlement of the caisson with time after the construction is shown in Fig. 6. The figure shows that the settlement of the caisson increases with time. The long term settlement after the construction of the caisson was studied by Rahman et al. (1999a, b). Those studies showed that the settlement of the caisson had continued after the completion of consolidation at SCP portion. They pointed out that the long term settlement of the caisson was caused by both the consolidation settlement of SCP portion and differential settlement between the SCP and SD portion under the fill behind the SCP portion. They concluded that the long term settlement could be effectively controlled by adopting higher replacement ratio and widening the SCP portion toward the fill. Similar response is also observed in this study. It can be seen from Fig.6 that both the widening the improvement width (W) and the increase of the net caisson pressure (W,) from 85 to 130 kPa reduce the long term settlement. Effects of W and W, can be confirmed in Fig.7. In these figures the long term settlements of
T = (y',z + Ao,p,)lm?(p, cos2a and for clay part:
(]a>
T = c0 + kz + Ac, (1 b) where, T is shearing streiigth of composite ground at depth z of sliding surface, Y ' ~is effective unit weight of sand pile, Ao, is the average increment of pressure on iinproved surface at depth z. p, is the ratio of the stress in the sand pile to the average stress respectively, CO is the initial shear strength of clay at depth z=O, k is the rate of increase of undraiiied strength of the original clay with depth, Ac, is the increase of the strength of clay due to consolidation at time t. The increment of strength of clay due to consolidation is estimated by the following equation:
(2) Ac, = A T k,/ P ) U , Where AD, is vertical stress increment in the claj part, c,/p is ratio of increase of clay strength to the increase of effective vertical stress and Ut is the degree of consolidation due to vertical and radial flow.
492
Table 2: Factor of Safety (FS) case no
1
2
3
8
>
I1
FS (after caisson loadirg)
1.82
1.82
2.03
1.41
1.41
1.61
FS (after
1.06
~
m A
1.35
1.28
1.29
1.44
1.38 .
backfilling)
The details of the calculation of stability analysis and selection of parameters are presented by Rahman et al. (1999b). -In this analysis the angle of internal friction of SCP was taken as 40' and the stress concentration ratio in the SCP was assumed as 3. The calculated factor of safety after caisson loading and caisson and backfill loading is shown in Table 2. Under caisson loading higher factor of safety was obtained in case3 in series1 and case11 in series2, with a,=50% than the other cases where %=30% was used. Widening of SCP portion does not increase the factor of safety significantly under caisson loading. In case2 and 3 the number of SCPs are almost same 32 and 33 respectively. It was found that widening the SCP portion under the fill could increase the factor of safety than increasing the replacement ratio under only the caisson with the same number of SCPs. As the load from fill was higher than that from caisson, mobilized strength of SCP under fill was higher than that under the caisson. The observed lateral displacement of the caisson is plotted against the calculated factor of safety immediately after the backfilling in Fig.8. The unique relationship between the lateral displacement of the caisson and the factor of safety can be seen. This relationship reveals that irrespective of the loading and improvement condition the lateral displacement increases at the lower factor of safety and large lateral displacement is found at the factor of safety lower than 1.2. Similar phenomena was also observed in the previous study (Takemura et al., 1991).
LConsolidation*
Time (days) Figure Ger,eralizedloadingprocess
Figure 10 Relationship between the net caisson weight and the f;ictor of safety
Factor ot'safety (FS) Figure8 Relationship between factor of safety and lateral displacenient of the caisson after backfilling
Figure 11 Relationship between the factor of safety and the net caisson load (W,) under different backfill loads (Wbf)
493
&=30%and a,=50%. So it can be concluded that the factor of safety increases with caisson load to an optimum value and then again decreases for a specific backfill load and improvement condition. Fig. I 2 shows relationships between the factor of safety and W, with different W. The figures show that when the caisson load is smaller than the backfill load, the wider improvement width toward the fill is effective in increasing the factor of safety but the effectiveness decreases with increasing the improvement width. But for higher caisson load than backfill load, the widening of the improvement width is not significant in increasing the factor of safety.
5 CONCLUSIONS
Figure 12 Relationship between the factor of safety and the net caisson load (W,) under different improvement widths ( W)
Figure5 and Table:! show that increasing the caisson load can increase the stability of the SCP ground under the applied backfill loading. However, there may be several combination of net caisson load (W,) and backfill load (Wbf) to get the optimum stability in the construction of the backfilled gravity type caisson with a specific improvement condition. In order to investigate the effects of the caisson and backfill load on the stability, a loading process is assumed as shown in Fig.9. Under this loading process, the caisson is constructed in 21 days followed by 70 days consolidation and then the backfilling is made in 70 days followed by the consolidation. The calculated factor of safety (FS) for the backfill load of 140 kPa is plotted with the net caisson load (W,) is shown in Fig.10. The factor of safety immediately after the caisson construction decreases with the net caisson load. But immediately after the backfilling the response is quite different. The factor of safety depends on the net caisson load, backfill load and improvement width. Figure1 1 shows the factor of safety with net caisson load under different backfill loads. When the width of improvement is 1.2 times the caisson width, the factor of safety increases with the net caisson load to an optimum value and then again decreases. The position of the optimum factor of safety. denoted by an arrow shifts to the right (toward higher net caisson load) with the increase of the backfill load and left with widening the SCP portion toward the fill side. However, when the improvement width is 2 times the caisson width, the factor of safety decreases with increasing the caisson load without showing the optimum W,. under this range of loads. This response is observed in both
From this study the following conclusions can be drawn. Widening SCP area toward fill is effective to increase the stability of SCP improved soft clay under backfilled caisson loading. increasing the caisson weight to some extent can also increase both the short and long term stability. The factor-of safety increases with caisson load to an optimum value and then again decreases for a specific backfill load and improvement condition. When the caisson load is smaller than the backfill load the factor of safety increases by widening the improvement toward the fill but for the caisson load higher than the fill load, the widening of SCP portion toward the fill does not increase the factor of safety significantly.
6 REFERENCES Nakase, A., & Takemura, J. 1989. Stability of clays improved by sand compaction piles, Techical report. No. 40, Dept. of Civil Engg., Tokyo Institute of Technology. 1-1 8. Rahman, Md. Z., Takemura, J., Kouda, M., Yasumoto, K. 1998. Stability and deformation of soft clay improved by SCP with low replacement ratios, Proc. 13"' SEAGSC, Vol. 1: Taipei, Taiwan. pp.393-398. Rahman, Md. Z., Mizuno, T., Kouda, M & Takemura, J. 1999a. Long term settlement of soft clay improved by low replacement ratio SCP under backfilled caisson loading, Proc. 51"' ./SCE iinnziul Conventio~,Hiroshima, September 2724. Rahman, Md. Z., Takemura, J.. Kouda, M. and Yasumoto, K. 1999b. Experimental study on deforination of soft clay iniproved by low replacement ratio SCP iinder backfilled caisson loading, Soils u17d Foiir7ckution (Submitted) Takemura, J., Watabe. Y . , Suemasa. N., Hirooka. A . & Kiinura. T. 199 1. Stability of soft clay improved with sand compaclion piles, Proc. 9"' h i m Regiomd Confirence, Vol. I , pp. 543-546. Terashi, M. & Kitazume, M. and Minagawa, S. 1991. Bearing capacity of improved ground by sand compaction piles, Deep Foioidutioii Improvements: De.sig17 constrzictior? mtl testing, ASTM STP 1089, pp 47-6 1 Yagyu. T. Endo, H., Takahashi. K., Yukita. Y . , Uniehara, Y. 199 1. Strength characteristics of soil improved by low replacenient sand coinpaction piles. GEO-CO,-IST ' 9 I , Y o kohama. Japan. pp 423-428.
494
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) (c) 2000 Balkema, Rotterdam, ISBN 90 5809 151 1
A case of vacuum preloading in combination with filling I. Sandanbata & K. Matsumoto - Hazunza Corporatioi?,Tokyo,J u p n K. Nakakuma - Muruyama Iizdustiy Company Limited, Suitania, Japan M. Kubo - Shiniizu Coi-poration, Tokyo,Jupun TYoshida - Konoike Coizstruction Conipany Limited, Tokyo,Jupuiz K.Yamaguchi - Asia Air Survey Conipany Limited, Tokyo,J a p m R. Arita - Hokknido Developnzent Bureau, Jayur?
ABSTRACT: In a highway project, an embankment, the maximum thickness of which was about 13 meters, was constructed on a soft ground (about 24 meters thick) consisting of an organic clay layer (about 7 meters thick). In spite of the fact the initial strength of the soft ground indicated the critical height of 2 meters for staged filling, the concurrent adoption of vacuum preloading enabled to accelerate the filling. The embankment of 13 meters thickness was rapidly filled up in about 70 days, hence the average filling rate was 18.5cdday. This value is about ten times larger than that of conventional surcharging. The filling method using vacuum preloading is found to be most appropriate for stable construction of an embankment on a soft ground in a short period of time.
1 INTRODUCTION
consolidation in combination with filling are summarized.
When an embankment is constructed on a soft ground, the conventional surcharge method is often executed. The filling operation is usually performed step by step after the required strength of the ground is generated in each phase to prevent the embankment failure. Thus the period of construction will be generally extended. In case shortening of filling duration is required, a ground improvement method is applied additionally to strengthen the soft ground. Vacuum preloading method is to accelerate consolidation of a ground by applying vacuum pressure to the ground covered with an airtight membrane, and by pumping out forcibly the pore water and the gas within the ground through vertical drains driven into the ground. By vacuum application beneath membrane, the pressure in the drains is reduced. Therefore the generation of excessive pore water pressure during filling process is suppressed by accelerating the drainage. This leads to an immediate increase of the ground stability during filling process. The construction method to apply vacuum preloading method in filling process is most appropriate for stable construction of an embankment on a soft ground in a shortperiod of time. This paper describes the subjects in the following order; the plan of the project, the vacuum operation and observation results during the vacuum consolidation and during construction of the embankment. Finally, the technical problems remained concerning the method to execute vacuum
2 OUTLINE OF THE PROJECT In the highway project in this paper, the maximum thickness of the embankment is about 13 meters. It was constructed on a soft ground (about 24 meters thick) including an organic layer (about 7 meters thick) near the ground surface. The soft ground is consolidated to eliminate or reduce the settlement after the construction. The conventional surcharge method requires a long period of time. So, the vacuum preloading method was adopted to improve the soft ground by conducting a feasibility study. The project was executed over three years. A construction of test embankment with vacuum preloading (1,036m2) was executed precedently in the first year (Ninomiya et al. 1998). The main construction was perfomed in the second and the last year. This paper mainly describes the summary of the construction (18,289m2) in the second year.
Figure 1. Plan of vacuum preloading.
495
Table 1. Specification of each block. Embankment Spacing of thickness(m) drains(m)
Treated area(m’)
Figure 2. Construction process.
A
1,710
6.3
B
2,278
9.1
0.9 0.8
C
1,466
13.1
0.7
D
2,097
12.2
0.7
E
2,911
10.7
F
2,578
9.1
G
2,633
7.4
H
2,616
6
0.8 0.8 (0.9) 0.8 (0.9) 0.9
Total
18,289
Table 2. Average of the measured vacuum pressure. Ground surface Block Vacuum pump (kN/m’) (kN/m’) A
B C D E
F G H
77 86 81 82 79 77 83 87
58 53 30 40 39 22 43 32
Figure 3. Soil parameters.
The construction area was located at a connection area of a bridge as shown in Figure 1. The improvement area was divided into eight blocks in order to execute vacuum consolidation. Each block had an area of 1700-2900m2. The specifications of each block are summarized in Table 1. The vertical drains were driven into the ground to 20 meters depth. Spacing of the drains was chosen from three types such as 0.7, 0.8 and 0.9 meter according to the difference of the embankment structure (main road, byway and height of embankment). Major measurement was performed in two blocks (“C” and “H” block). Take note that embankment thickness and spacing of drains in these blocks are different from each other. Figure 2 shows the construction process. From the technical viewpoint, for example, in order to reduce the residual settlement, the determination of the period for the vacuum operation before and after filling is a very significant problem. These problems, however, have not been solved and there is no authorized design method because the construction records have not been accumulated. So, consolidation rates and settlement values were evaluated by using the conventional design method i.e. Terzaghi’s theory and Barron’s equation. The vacuum operation started at 10 to 28 days period before the filling, and terminated at about 40 days period after completion of filling. The total period for the vacuum operation is about 4 to 6 months. 496
3 GROUND CONDITION
The soft peat layer (about 7 meters thick) with natural water content of 100-700% was deposited on the ground before improvement. Under the layer, clayey silt layer (about 16 meters thick) was deposited. Fine sand layer was inter bedded from GL-7 m to -9 m. Typical soil parameters are indicated in Figure 3. It is inferred that the initial stress state before construction is to be in a normal consolidation state as shown in Figure 4. The values of the coefficient of consolidation (corresponding to the effective overburden pressure consideringL the load increment by embankment) were 20cm /day and 70cm2/dayfor the organic clay and the clay layer, respectively. 4 OBSERVATIONAL RESULTS 4.1 Overview of measurement Figure 5 shows the measurement points in “C” block and “ H y block. To monitor the ground behavior during construction, pore water pressure gauges, differential settlement gauges and inclinometers were set up under the ground. A pore water pressure gauge was set up at each hole bored in the middle of four drains arrayed in a square at the specified depth.
planned to start vacuum consolidation before the arrival of a cold wave. However, we had to execute vacuum preloading in the severe winter because of the delay of the process in the early construction stage before the process of vacuum preloading. There is a possibility that the gauges used to monitor negative pressure could not indicate the differential pressure between the pressure under the membrane and the atmospheric pressure correctly when outside air temperature fell far below zero degree. Therefore, it may be impossible to correctly measure the pressure reduction generated on the ground surface in the second construction year. 4.3 Ground surface settlement
Figclre 4. Stress condition before the construction.
Figure 6 shows the settlement of each block. In “C” block with the thickest embankment (13.1 meter thickness), the settlement exceeding 4 meters was recorded until the end of the vacuum consolidation. The observed settlement value corresponds with the designed one calculated fiom the vacuum pressure and the weight of the embankment. The settlements before filling were 33 cm and 34 cm in “C” and “H” block respectively. 4.4 Differential settlement of each layer Figure 7 shows the compressive strain in the ground calculated from the observational results by using the differential settlement gauges at the final stage. Major ground compression took place in the peat layer and the organic clay layer near the ground surface. The values of compression in the peat layer and the organic clay layer were found to be 65-75% (for “C” block) and 7844% (for “H” block) of those at the end of vacuum consolidation.
Figure 5. Measurement plan (“C” & “H’ block).
4.2 Vacuum consolidation work Table 2 shows the average of the measured pressure reduction values in the vacuum pump and under the airtight sheet on the ground surface. It indicates that pore pressure reduction of 70 kN/rn2 or more was generated in the vacuum pump. Under the airtight sheet, the reduction values were about 30 to 40 kN/m2 except “A” and “B” block, which were smaller than that measured in the vacuum pump. The loss of vacuum pressure under the membrane is larger than the values of other construction sites. In case records carried out for recent two years, the average of vacuum pressure under the airtight sheet indicated about 60kN/m2 (Sandanbata et.al. 1998, Matsumoto et.al. 1998). It is presumed that the gauges under the membrane did not function properly under severely low temperature in winter season. This construction site is in Hokkaido, the northern district of Japan. Outside air temperature reaches -20 degree centigrade in winter. At first, we
Figure 6. Settlement ofeach-block.
497
about 40cm was measured on the ground surface. In “H” block with lower embankment, the same tendency in ground behavior was observed, though the values were smaller than those of “C” block. In the first year of test embankment of this project, slight lateral deformation toward the outside of the improvement area was observed during filling period as shown in Figure 9. This behavior is entirely different from that observed in the second year shown in Figure 8. One of the reasons why the results in the first year were largely different from the behavior in the second year may be due to the fact that the period of vacuum operation before starting of filling was longer than that of the second year. In the first year test embankment, the vacuum operation period before filling was 43 days. In the second year construction presented in this paper, the period was 10 and 21 days for “C” and “H” block, respectively. Figure 7. Compressive strain in each stratum.
4.6 Pore pressure under the ground Figure 10 shows the pore water pressure under the ground in “H” block. The changes of pore pressure in the ground after vacuum operation are shown in Figure 11. If the intermediate sand layer exists in the ground, it is anticipated that the vacuum pressure may be decreased by the inflow of the pore water from the surrounding area. However, in this case, the rise of pore water pressure in the sand layer was not observed even if the water pressure in other stratum was increased by the filling operation. It is clear that the intermediate sand layer improved the ground stability effectively by suppressing the rise of the excess pore pressure.
Figure 8. Lateral displacement in the ground. (Main embankment in the second year)
4.5 Lateral displacement in the ground Figure 8 shows the lateral displacements before filling and at the end of filling observed within the distance of one meter from the border of the improvement area. The positive values indicate the movement toward the outside of the embankment. The ground was displaced toward the inner side of the improvement area due to the vacuum preloading. The lateral displacement toward the outside of the improvement area began to take place from the start of the filling. The maximum value of lateral displacement toward the outer side was about 70cm in the organic clay layer at “C” block with the thickest embankment. The lateral displacement of
Figure 9. Lateral displacement in the ground. (Test mh-mkment in the first year)
498
Figure 12. Pore pressure vs. time after the filling
Figure 10. Pore pressure in the ground.
Figure 13. Normalized settlement by the maximum settlement.
4.8 Effect of drain spacing
Figure 1 1. The change value of pore pressure.
The ground behavior due to the difference of the spacing of drains was examined through the observed results. Figure 12 shows the change of pore water pressure in the ground from the start of filling. With regard to the dissipation rate of the excess pore pressure after the completion of filling, the rate in “C” block (the spacing of drains was 0.7 m) was 2.4 times faster than “H” block (the spacing of drains was 0.9 m). The observed settlement rates are shown in Figure 13. In order to compare the settlement rates of each block, the settlement was normalized by the final settlement predicted from the fitting method by using hyperbolic curve. As expected, it is clear that the smaller spacing of the vertical drain accelerates the consolidation rate.
4.7 Stability control of embankment In the second year of this project, the ground deformation showed the tendency to deform largely toward the outside during filling process. However, the embankment failure did not occur during the constructicn period. The stability control method to prevent the ground failure under the filling in combination with vacuum preloading is not still enough studied. We have not obtained enough knowledge on ground failure under the vacuum preloading. That is because there is no example of ground failure where filling is executed concurrently with vacuum preloading. Therefore, we adopted the conventional stability control method for filling as in the conventional surcharge in Japan. We controlled filling rate so as to set the maximum horizontal displacement rate to 1.5cdday or less near the top of the slope. In case the rate was about to exceed this limit, we stopped the filling and observed the behavior of the ground carefully.
5 CONCLUSION The following results can be concluded. 1) In case an embankment is constructed on a soft ground, concurrent adoption of vacuum preloading 499
method with filling process makes it possible to execute a rapid earth filling. 2) In the reported case, the embankment of 13 meters thickness on the organic soft clay ground was rapidly filled up in about 70 days, the average filling rate of 18.5 c d d a y , which indicates about ten times faster than the conventional surcharge method. 3) The maximum lateral displacement recorded in the soft ground due to the filling load was about 70cm. However, the failure of the ground was not occurred in this project. It is considered that the increase of excess pore pressure due to the filling operation was restrained by the combination with the vacuum preloading. 4) Evaluation methods have to be established for the following items. - The stability control for filling in combination with vacuum preloading. - The period of vacuum operation before and after the filling to prevent excess residual settlement. REFFERENCES Matsumoto,K., Ohno,M. PhiOanh,T., Nakakuma,K. and Nakakuma,.K 1998. Ground behavior during ground improvement due to vacuum consolidation (V01.2). Proc. of 33'd Japan National Conf. on Geotech. Engrg., pp. 10951096 (in Japanese). Ninomiya,H., Honda, K., Akashi, K., Hayashi, H., Umezaki, T. and Shiono, T 1998. Improvement effects of soft clay ground due to newly developed vacuum consolidation method (Ver. 1 to Ver.3), Proc. of 33'd Japan National Conf. on Geotech. Engrg., pp.2139-2144 (in Japanese). Sandanbata,I., Koga,T., Ishihara,K., Kato,T., and Nakakuma,K 1998. Ground behavior during ground improvement due to vacuum consolidation (Vol.1). Proc. of 33'd Japan National Conf. on Geotech. Engrg., pp. 10931094 (in Japanese).
500
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, lS5N 90 5809 151 1
Self-weight consolidation of dredged clay with plastic board drain IS.Sat0 & N.Yoshida Department of Civil Engineering, Fukuoka Universiv, Japan
T. Nomura Kinjo Rubber Company Limited, Osaka, Japan
ABSTRACT: The Plastic board drain (PD) method for self-weight consolidation has been proposed by the authors as one of soil improvement methods in reclamation works using high water content dredged soils. This method is the vertical drain method for accelerating self-weight consolidation using a PD and floating driving machine to dehydrate such very soft grounds in a short period. Therefore, the effectiveness of the PD for self-weight and loading consolidation was confirmed by laboratory tests using a large size cylindrical container (H=2lOcm, D=30cm) filled with dredged clay slurry of high water content (w=lOOO%). 2 TEST PROCEDURES
1 INTRODUCTION
In recent years, for maintenance of navigational channels or purification of the environment, it is often planned to dredge the clayey deposits on the surface of sea, and utilize the dredged spoils as landfill and manmade island materials. Typically, the dredged spoils are temporarily stored in a nearby sedimentation pond and then very soft ground is formed. Nevertheless, since an artificial ground filled with the spoils has a very high water content such as 200 to 300%, it is necessary to dehydrate it properly for the utilization to landfill and manmade island. Kamon et a1.(1991), Shinsha et. al. (1991) and Yoshikuni et.al. (1994,1995) have researched the vertical drain method using plastic board and horizontal drain method as one of the soil improvement methods in reclamation works for dredged clay. Therefore, the authors have developed the vertical drain method for accelerating self-weight consolidation using plastic board drains as a method of dehydration such very soft grounds in a short period. This method is not only possible to pour the large quantity dredged spoils into the landfill pond but also is possible to utilize a landfill ground as soon as possible after the end of reclamation. In this paper the effectiveness of the plastic board drain (PD) for self-weight consolidation was confirmed by the laboratory tests using a large scale cylindrical container (H=210cm,D=30cm). Moreover, the results from the loading consolidation tests after the self-weight consolidation tests is also reported. 501
2.1 Clay sample, PD and apparatus The clay sample used in this study was an alluvial marine clay called as Kanda clay, which was taken from the port of Kanda bay in Fuku0k.a prefecture, Japan. Pieces of shells were taken out from clay slurry by using a sieve of 2.0mm diameter. The particle size distribution curve of Kanda clay is shown in Figure 1. The physical properties of this clay are given as: G,=2.525, w,=140%, w,=84.2%, w,=36.9% and Ip=47.3. The clay slurry was completely remolded in a mixer with the 3% density salty water. At this stage, the water content of prepared clay slurry is about 1000%. A PD sample used in this study was 50mm width, 3.6mm thickness, which has the composite boards with filter sleeve attached to the profiled core as shown in Figure 2. The water contained in clay slurry is drained through the inside of a core, when the PD sample is inserted into very soft clay ground. The testing apparatus used for investigation of selfweight consolidation characteristics of slurry dredged clay consisted of 7 parts of large scaled cylindrical container (H=210cm) as shown in Figure 3. This each container has D=30cm inner radius and H=30cm height. The inner surface of the cylindrical container is made of acrylic with lcm in thickness.
were measured. The pore water pressure in the clay sample was measured by the transducer installed at the bottom of the cylindrical container as shown in Figure 4. In order to investigate the mechanism of the selfweight consolidation with PD, two types of the consolidation tests were carried out : 1) self-weight consolidation test (SW-test) in which pore water drains from the upper surface only, 2) self-weight consolidation test in which the PD is set up at the center of sample (PD-test), in other words, pore water drains from the PD and the upper surface.
2)L,ouding consolidution test Figure 5 illustrated that the loading and the perineability tests apparatus. This loading system is controlled by air cylinder. A series of tests are carried out as follows: After self-weight consolidation, a stage loading consolidation tests for SW and PD-test did with the each vertical load ov=19.6kPa and 49.0kPa. The tests were performed in step by step way under the consolidation pressure. After the each loading consolidation test, the constant head permeability test carried out to investigate the permeability performance of PD samples in the clay specimen under constant loading. After the 2nd stage loading test, in order to investigate the strength properties of clay specimen for each tests, the cone penetration tests were carried out at the top of clay sample. In the case of PD-test, after the final consolidation step in the model test is finished, the mold was disassembled carefully so as to see the deformations of the prefabricated drains inside the clay. After the consolidation test, the local water The points were chosen with the radiation from the center at thc middle of PD. It was measured to a depth direction 5 cm each.
contents of the clay was measured.
Figure 3 Test apparatus
2.2 Test procedure
3. TEST RESULTS AND DISCUSSIONS
1 ) Selj-weight consolidation test
3.1 Self-weight consolidation test
The clay samples prepared by the above-mentioned method were poured into the cylindrical container up to ;I height of about 1C)Omm per day for each layer. The total layers are five, finally leading to 250cm height using the top cap mold. The pouring method wa$ used by the water pump. The PD samples was inserted at center of the clay sample after 24 hours since pouring of 5 layers. After pouring for each layer, thc surface settlement and excess pore water pressure 502
Figure h shows the time versus height of clay surface rclations for cach layer as clay slurry is poured. The height of clay slurry after 24hr. starting since pouring clay slurry has risen at the constant rate. The rising rate of slurry clay surface is 37.3cm/day and thus the amount of settlement for each layer is almost the same .
Figure 9 Relationships between elapsed time and consolidation velocity
Figure 6 Relationships between elapsed time and height of clay surface for each layer
U (96)
90
Conyolidalion
PD-test t 24550
I
1 I
Time (min)
SW-test t 281530
The clfecl of acceleration
t 2/ t
I
t ,/ t
11.5
j
0.09
LJ
503
0Y
(c,;)
(k~",)
50 50
19.6 40.0
Conwlidatirm
PD-lest t 7500 1330
,
Time (min)
SW-test t 17190 34430
The effect of acceleration
t
J
t
6.9 25.9
,
t ,/ t 0.15 0.04
Figure 7 shows the time-settlement relations of the slurry clay surface for SW-tests and PD-tests. These results indicate that the settlement data of PDtests was terminated in 6ldays after start of mcasurenient. On the other hand, the results of SWtests was terminated in 310 days. The relation of time-settlement markedly that indicates the consolidation is accelerated by PD. In this Figure, the final settlement, S,, is estimated by using a hyperbolic fitting method for PD-test data. Based on the abovemcntioncd proccdure, the final settlements (&) is about C)O.lcm, which is calculated by the each consolidation curve. The degree of consolidation at SW-tests and PD-tests was calculated from the final settlements (S,). The effect for accelerating consolidation by means of PD is estimated by comparing the consolidation time at U=c)Oo/O.The results are shown in Table-1. The self-weight consolidation in which the PD is set u p at the center of sample has been accelerated at as the 11.5 times as the case without PD. This result indicates that it is very effective for accelerating consolidation to make use of PD during self-weight consolidation. The relationships between the pore water pressure and elapsed time for SW-tests and PD-tests are shown in Figure 8. This Figure indicates that the pore water pressure dissipates for the case with PD with the progress of self-weight consolidation. The rate of pore water pressure dissipation is higher than that in the case without PD. Figure 9 shows the relationship between consolidation velocity and elapsed time. The consolidation rate of PD-tests is larger than that in SW-tests until U=87%. Afterwards, the consolidation velocity of each test becomes the same steady rate. The above-mentioned results from the experiment indicate that it is very effective for accelerating consolidation to make use of PD during self-weight consolidation.
3.2LoircEing consolidation test The settlement versus elapsed time curves for selfweight and loading consolidation of SW-test and PDtest are shown in Figure 10. In this Figure, the relation of time-settlement markedly indicates that the consolidation is accelerated by PD for each loading step. The rate of initial settlement of PD-test is greater than the case without PD for each loading steps. Comparing the elapsed time for the end of consolidation, the SW-test requires about three times as longer time as PD test. This result indicates that the PD after high deformation by self-weight consolidation maintains a high permeability performance.
504
Figure 13 Results of grain size analysis
Therefore, the effect for accelerating consolidation using PD is estimated by comparing the consolidation time at U=50% for each loading step. The results are shown in Table-2. The loading consolidation in which the PD is set up at the center of sample has been accelerated at as the 11.5 and 25.9 times for each loading steps as the case without PD. This result indicates that it is very effective for accelerating consolidation to make use of PD during loading consolidation. After the final consolidation step in the model test is finished, the mold was disassembled carefully so as to see the deformations of the prefabricated drains inside the clay. Figure 11 shows the deformations of PD inside the clay. A PD sample was deformed and bent with large curvature by consolidation and received the strain up to 70% to the axial direction. The photograph shows that the PD sample undergoes large deformation following the settlement of a model ground. A sharp local kinking which is one of the major reason for the reduction of the permeability performance of PD, was not observed. In order to investigate the permeability characteristics of the PD after loading consolidation the permeability tests were performed under the constant head as shown in Figure 12. The result of the permeability test of PD sample for the triaxial cell is also simultaneously shown in this figure. These experimental results were obtained with a hydrauric gradient i = 1.0 condition. In this research, the PD samples with B=Scm width and L=20cm length were used. In the triaxial cell tests, samples were consolidated by the isotropically under the stress of 29.4kPa. The permeability of PD sample tends to decrease with the increase of the strain. On the other hand, the permeability of PD under the vertical consolidation is about 1/9 as much as the results of PD sample. This is caused by the length and bending of PD. And, the cross section of PD sample becomes narrower by transformation of the non-woven filter for the confining effect of consolidation. However, this coefficient of permeability is estimated from the amount of drained water using the Darcy’s low, it will be set as k= 4.3 cm/sec, and it turns out that it fully has the drainage function as compared with the permeability of model clay foundations. In order to place the PD sample in the slurry of high water content, the clogging of the non-woven fabric filter by the fine clay particle would be one of the ploblem. Therefore, the grain size analysis was performed for the clay samples after the loading consolidation test. Figure 13 shows the average grain size D,,, according to distance from PD sample. It shows the tendency that the average grain size becomes smaller, 505
Figure 16 Results of cone penetration test
as i t is closer to the PD sample. It is shown that the small size grains are accumulated near the PD sample under self-weight consolidation stage. However, the particle diameter of clay specimen is almost uniform, which may not be one of the major reason to reduce the discharge capacity. The relation of the water content distribution in model foundations and the distance from PD sample is shown in Figure 14. It is understood that variation in a water content distribution of clay specimens is large. It turns out that water content gets a little
higher as it goes further from the PD sample. The average water content distributions of the center and edge of clay specimen for S W and PD-test are shown in Figure 15. It can be observed from this figure that the average water content distribution in the depth direction is uniform with a maximum variation of about 20%. This result may be interpreted as that the consolidation occurs faster at the upper layer of specimen and there is a lag in the depth direction. However, the water content distribution of SW and PD-test after loading consolidation indicate the same tendency. This result indicates that the consolidation of clay is efficiently performed by PD. After loading consolidation test, the local undraincd shear strength of the clay for SW and PDlest was measured by cone penetration test. The average undrained shear strengths at edge of each layer in the depth direction are shown in Figure 16. The shear strength of the clay for each test slightly decrease in the depth direction. However, the strength for SW and PD-test indicate almost same tendency.
4.CONCLUS IONS The conclusions obtained in this study are summarized as follows: (1) The PD material has high permeability under large dcforrnation due to self-weight consolida t ion. (2) It is very effective for accelerating consolidation to make use of PD during self-weight consolidation. The consolidation by means of PD has been accelerated at as the 6.5 times as the case without PD at U=50%. (3) The PD sample after high deformation by selfweight consolidation maintains a high permeability performance. Comparing the elapsed time for the end of consolidation, the SW-test requires about three times as longer time as PD test. (4) The water content and undraind shear strength distribution in the depth direction of clay specimen for SW and PD-test indicate the almost same tendency. These results seem that the consolidation of clay specimen was efficiently performed h y PD sample.
REFERENCE Kamon, M., Pradhan, T.B.S. and Suwa 1991 : Evaluation of design factors of prefabricated bandshaped drains, Proc. of International Conference
506
of the Geotechnical Engineering for Coastal Development (Geo-Coast '91),Vol. 1, pp.329-334. Shinsha, H., Watari, Y . and Kurumada, Y., I991 : Improvement of very soft ground by vacuum consolidation using horizontal drains, Proc, of International Conference of the Geotechnical Engineering for Coastal Dewlopment (Geo-Coast '91), Vol.1, pp.387-392. Yoshikuni, H., Kim, H., Hirokane, S., Moriwaki, T. and Kusakabe, 0. 1991 , Consolidation behaviour of dredged clay ground improved by horizontal drain method, Proc, of International Conference of the Geotechnical Engineering for Coastal Development (Geo-Coast '91), Vol.1, pp.99103.
Yoshilwni, H., Hirokane, S., Moriwaki, T. and Kusakabe, 0. 1994, A study on the effectiveness of horizontal drain method (in Japanese), Journal of Geotechnical Engineering (JSCE), No.499 , pp.87-96.
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 I
Reclamation control of pump-dredged clay by CONAN T. Sat0 & K. Ishinuki Fourth District Port Construction Bureau, Ministry of Transport, Shimonoseki, Japan
M. Katagiri & M.Terashi Nikken Sekkei Nakase Geotechnical Institute, Kawasaki, Japan
S. Kitazawa Coastal Development Institute of Technology,Tokyo,Japan
ABSTRACT: Along with the progress of reclamation by pump-dredged marine clays, the detailed soil investigation and monitoring were performed together with the numerical simulation by “CONAN” for predicting the consolidation behavior of dredged clay layer. Although the dredged clay layer is not uniform due to the grain size sorting during the sedimentation process, one dimensional consolidation analysis gives practically useful prediction not only of the change of elevation but of water content distribution and pore water pressure distribution, if the appropriate parameters are determined.
1 INTRODUCTION Two projects are going on simultaneously on the sea northeast of Kyushu Island as shown in Figurel. One is the maintenance dredging of sea bottom sediment to keep the required depth of navigation channels and anchorage area. The Strait of Kanmon between the Honshu and Kyushu islands is one of the busiest sea routes in Japan and its maintenance is of vital importance. For the protection of marine environment, dredged materials, mostly clays have been and will have to be discharged into a pond surrounded by containment dikes on the sea. This in turn continuously creates a new artificial island of extremely soft soil condition. The other is a New Kitakyushu Airport construction project which is a relocation of existing one inland, thereby reducing the air noise to the residential area and increasing the capacity of airport. The site for the airport was selected on the artificial islands created by the maintenance dredging. The merge of two projects had a merit of reducing the amount of hill-cut materials substantially that might otherwise be enormous and cause a destruction of the environment at borrow area. However, the project posed a challenging problem of predicting huge consolidation settlement of soft dredged clay during the construction of airport. The water contents of marine clays of high plasticity easily reach as high as 2,000 % after pumpdredged. When such a dredged clay slurry is discharged into the pond, suspended soil particles settles loosely with the water content between 200 to 300 %.
Figure 2 Design o f New Kitakyushu airport and monitoring sites in K1-area
507
Elevation (DL+m) I
I
amount of extra-fill, and to keep the residual settlement within an allowable magnitude. In this paper, an outline of the new Kitakyushu airport, the results of field observation and settlement analysis, and the evaluation of consolidation parameters for dredged clays will be described.
ground improvement
2 IMPORTANCE OF PREDICTION OF SETTLEMENT 1979
1999
2605
Time
Figure 3 Reclamation history and construction plan
Figure 4 Ground conditions o f reclaimed lands
The dredged clay layer thus created is in the unconsolidated condition and subsequently consolidates largely due to its own weight in the long term. To create a reliable foundation ground for the airport at a specified elevation the placement of extra-fill over the dredged clay layer is necessary. As the extra-fill causes additional large consolidation settlement, it is necessary to accelerate the consolidation by vertical drainage if the term of construction is limited. The accurate prediction of the time dependent settlement is important in such a project in order to plan the capacity of containment dikes, to estimate the
The project site consists of four disposal ponds as shown in Figure 2 and airport facilities such as a runway, taxiway and apron will be constructed across three ponds KO, K1 and K2. The KO area had been reclaimed from the sea by dredged materials for about 18 years from July 1979 to March 1997. The K1 and K2 areas are planned to be reclaimed in a shorter period of about three years each. The reclamation of the K1 area has started in October 1996 and ended in June 1999 and that of K2 area followed since December 1998 and will end in March 2001. Both the total and residual settlements vs time relation of each area will be different due to these differences in the history of reclamation as shown in Figure 3. The vertical axis shows the elevation from the datum line (DL). The airport should be in operation in 2005. Soil profile of the KO area was investigated in September 1995, 18 months before the end of reclamation. Open circles in the Figure 4 show the measured data of the KO area. Figure 4(a) indicates that the water content of top 2 m layer is .around 200 %, that of the layer between DL+7m and DL-2m ranges from 50 to 100 %, and that of the layer beneath DL2m exceeds 100 %. It is interesting to see the variation of the preconsolidation pressure, p o with the elevation in Figure 4(b). The broken line in the figure is the effective over-burden stress, 0'".The magnitude of po in the top 2 m layer are nearly the same with that of dV.In contrast to this,po and d Vare different in the deeper layers. The layer between DL+Sm and DL-3m exhibits over-consolidated condition and the layer beneath DL-3m exhibits the unconsolidated condition. The middle layer, together with its lower water content, will show the lower compressibility under the extra-fill loading. The reclamation of KO area was slow taking about 16 years, and in each year there was an intermission of approximately six months. These high preconsolidation pressures in the middle layer are thought to be caused by the desiccation. Similarly the soil profile of the K1 area was investigated in November 1998 before the end of the reclamation. The solid triangles in the Figure 4(a) show the water contents. Whole the layer here has high water content and seems to be the unconsolidated condition. Much larger consolidation settlement is
508
anticipated in the K1 area under the subsequent extra-filling. The same is anticipated for K2 area. Throughout the reclamation of KO to K3 areas, similar sea bottom sediments from nearby navigation channels have been and will be used. When, however, the reclamation process is different, the progress of consolidation differs, the soil profile of the completed ground differs, and hence the subsequent consolidation settlement will be much different. The runway sensitive to the change of gradient has to be constructed on three different areas having different reclamation histories. Therefore, it is important to predict the settlement behavior with high accuracy, and to examine the construction conditions carefully. The error in the prediction will affects seriously the cost and term of construction. In the following sections, new reclaimed area, K1 with large anticipated settlement will be discussed in details.
3 CONDITIONS OF K1-AREA The K1-area is 900 x 940 m in plan and the average depth of seabed and thickness of the alluvial clay layer are DL-7.7m and 5 m before the reclamation, respectively. A planned reclamation elevation by dredged materials is DL+7m. The materials are mainly marine clays dredged from three navigation channels located near the construction site as shown in Figure 1. Figure 5 shows the consolidation parameters of three samples obtained from these borrow areas. The parameters are determined by the hydraulic consolidation tests (HCT) at smaller stress level and by ordinary consolidation tests (OCT) with incremental loading at the higher stress level. The parameters in the intermediate stress level are obtained by interpolation to cover the wide stress range. The physical properties of the materials are shown in the same figure. Although the numbers of test data are quite limited, three materials have shown almost the same characteristics. The consolidation parameters for the preliminary prediction are decided as average relations among these data as shown in Figure 5. The reclamation history of the K1-area is shown in Figure 6. Total amount of dredged soil at their borrow area is about 10 million cubic meters. For the numerical analysis, true mass of solid part of the reclaimed materials is needed. The mean water content of 101 % obtained from the field investigation was used for the calculation of true mass.
4 METHOD AND APPLICATION OF ONEDIMENSIONAL CONSOLIDATION ANALYSIS 4.1 Numerical method used
The numerical method used in this paper was known as ‘CONA”, which was developed based on a generalized consolidation theory (Imai, 1995). The numerical code was developed by considering the accumulated layer based on the technique proposed by Yamauchi et a1.(1991). The detailed procedure of this numerical method will be described in a companion paper by Katagiri et al. (2000). 4.2 Consolidation parameters
Figure 6 Reclamation history of K l area
Figure 7 shows the range of consolidation parameters of the materials sampled from the K1-area after the reclamation (after the grain size sorting). Individual data will be shown later in Figure 12. To determine the consolidation parameters over a wide stress range, the multi-sedimentation tests (MST) proposed by Yamauchi et al. (1990) and the ordinary consolidation tests (OCT) were carried out. Thef-p and c,-p relations of all samples range widely according to their physical properties. The average relation in Figure 5 on original samples before the grain size sorting falls within this wide range of actual variations. 509
Figure 9 Water content distributions in K1 area
4.3 Application of one-dimensional consolidation
Figure 7 Range of consolidation parameters of samples obtained from K1 area and assumed consolidation parameters
Figure 8 Time records of elevation of K l area
In the back analyses, the consolidation parameters are assumed based on the average relations shown in Figure 5. For the compressibility, the inclination of log f - logp relation was changed by a fixing point at p = 1,000 kPa on the average relation, as case 2 - 4. The permeability was changed by shifting the log c, - logp relation in parallel to the average one, as case B and C. These assumed relations are within the range obtained from the samples from the K1-area.
When pump-dredged materials are poured into the water, soil particles settle freely with grain size sorting. Coarse particles are piled up near the outlet of discharging pipe, and finer ones will be transported and will be accumulated far from the outlet. Although the outlet is moved periodically during the reclamation process, the deposit thus formed is not uniform. However, it is impossible to determine the spatial distribution of soil parameters beforehand, the reclamation is assumed as one-dimensional consolidation by a homogeneous material in the numerical analysis. To fill this gap in the assumption and reality and to improve the accuracy of prediction, monitoring of the dredged layer and the back analyses are conducted along with the progress of reclamation and the consolidation parameters are modified accordingly. For the monitoring, the height and water content distribution of dredged layer have been measured. In addition set of pore pressure transducers are installed in the different elevation within the dredged clay layer and alluvial clay layer underneath on November 1998 (24 months after the start of reclamation) in three locations (Figure 2). If this simple approach is successful, this will be a practical tool to evaluate the average behavior of the reclaimed land, to obtain an overall volume of extrafill, and to provide sufficient information for the basic design of the ground improvement.
5 EVALUATION OF NUMERICAL ANALYSIS BY MONITORING RESULTS 5.1 Comparison between monitoring and numerical results Along with the progress of reclamation, back analyses are carried out in several stages and parameters are modified for the prediction of next stage.
510
Figure 10 Excess pore water pressure distributions in the middle of reclamation by dredged materials
Figure 11 Excess pore water pressure distributions after reclamation by dredged materials
Figure 8 shows the change of elevation of top surface of dredged clay layer with the progress of reclamation. The calculations by four different sets of parameters are exemplified in the figure. The calculation must fit not only with the height of the dredged clay layer but also with the water content and pore water pressure distributions in the dredged clay layer. Finally selected parameters which simulates the overall behavior are those for the Case-3C. Here, the case-3C means a set of #3 logf- logp relation and #C log c, - logp relation shown in Figure 7. Figure 9 shows the water content distribution obtained on a number of samples taken at the K1 area on October 1998. As the grain size sorting occurs during the reclamation process, the grain size distribution and Atterberg limits are also investigated on all the samples taken. The data plotted on the upper right of the figure with water content above 200 % and enclosed with a circle are obtained from the
51 f
samples with higher liquid limit and with larger clay fraction content in comparison with original soils. The data with lower water content on the lower left enclosed with another circle are from the samples which contains sand fraction more than 25 %. The original characteristics of the dredged materials at the borrow area were Ip = 59-69, Fc = 43-50 % and Fs = 4-8 %. When ignoring the data enclosed with two circles, predicted water content distribution by the Case 3C agrees with the measured ones. Figure10 compares the measured and calculated distributions of excess pore pressure at the same time instances of December 2 1998 (Figure 10(a)) and January 6 1999 (Figure lO(b)), when the water levels in the reclaimed land were almost the same as the seawater levels outside the reclaimed land. Three measurement locations, No. 1-1, 1-2 and 1-3, are indicated by open circles in Figure 2. Here, excess pore pressure is defined as the measured pore pressure minus hydrostatic pressure. The grain size distribution and the water content were investigated on the samples taken during the installation of transducers. The data enclosed in a circle are those obtained from the transducers embedded in the soil with a large sand fraction. When these data are ignored the calculation by Case 3C agree well with measured ones. Figure 11 shows the comparison between the measured and calculated distributions of excess pore pressure at January 2000, 7 months after the reclamation. The measured points are indicated by solid squares in Figure 2. The pore water pressures in the reclaimed land were measured by the pore pressure dissipation tests using a piezometer cone, and were obtained from the assumption that the ground water table was consistent with the ground surface. The numerical results of case-3C under 3 and 5 meters of thickness of alluvial clay layer were drawn in the same figure. In the reclaimed layer, every measured data are located in a relatively narrow range, and its distribution has the maxim~mvalue at the bottom of reclaimed layer. The numerical results also show the same tendency. The analyzed distributions are located in two third in the range of measured distributions and are recognized to be consistent with the measured ones. 5.2 Assessment of consolidationparameters The range of consolidation parameters obtained for the reclaimed soil samples of K l area has been shown in the Figure 7. The every test results are shown again in the Figure 12 together with the physical properties of samples for further consideration. Figure 12(a) shows the relation between specific volume, f and the consolidation stress,p. Numbers in the figure represents the clay fraction Fc, plasticity index Ip and sand fraction Fs. Specific volume f is higher and f-log p relation becomes steeper with in-
creasing Fc and Ip. The consolidation coefficient c, in the Figure 12(b) become lower with increasing Fc and Ip. These trends are agreeable because lower plasticity materials show the lower compressibility and higher permeability. The best-fit f-p and c,-p relations obtained from the back analysis (Case 3C) are also shown in the same figures by the solid broken line. This relation falls within the range obtained from the sampled materials and is corresponding to the materials having Ip of 40 to 50, Fs of 5 to 10 % and Fc of 40 to 50 %. Compared with the mean original three materials, the material evaluated by the back analysis was with the lower plasticity.
6 CONCLUDING REMARKS
(1) Time history of reclamation affects drastically the soil condition and changes the compressibility of dredged clay layer. (2) Dredged clay layer is not uniform due to the grain size sorting during the sedimentation process. (3) Although the spatial variation of consolidation parameters is large in the dredged clay layer, one dimensional consolidation analysis gives practically useful prediction not only of the change of elevation but of water content distribution and pore water pressure distribution, if the appropriate parameters are determined. (4) The best-fit consolidation parameters falls in the range of those obtained from the samples sorted in the sedimentation process. ACKNOWLEGEMENT
Along with the progress of reclamation by pumpdredged marine clay, the detailed soil investigation and monitoring were performed together with the numerical simulation by CONAN for predicting the behavior of dredged clay layer. From these studies, the following conclusions can be drawn:
The authors would like to thank Prof. H. Ochiai of Kyushu University for their valuable advice. The authors also thank Prof. G. Imai of Yokohama National University for technical advice of CONAN.
Figure 12 Consolidation parameters and physical properties
512
REFERENCES Imai, G., 1995. Analytical examinations of the foundations to formulate consolidation phenomena with inherent time-dependence. Proc. of IS-Hiroshima '95, 2: 891-935. Katagiri, M., Terashi, M., Henmi, K. & Fukuda, K., 2000. Change of consolidation characteristics of d a y from dredging to reclamation, Proc. of IS-Yokohama 2000. (to be submitted). Yamauchi, H., Imai, G. & Yano, K., 1990. Effect of the coefficient of consolidation on the sedimentation consolidation analysis for a very soft clayey soil (in Japanese), Proc. of 25'" Annual meeting of JSSMFE: 359352. Yamauchi, H., Imai, G., Watanabe, IS.& Ogata, K., 1991. S~dimen~ation-conso~i~ation analysis of pump-~redged cohesive soils, Proc. of Geo-coast '91: 129-134.
513
This Page Intentionally Left Blank
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 I
Field performance of PVD improved soft clay under embankment S.L. Shen Institute of Lowland Technology, Saga Universig, Japan
c.w.Yang China Academy of Railway Science, Beijing, People’s Republic of China
N.Miura & J.C.Chai Department of Civil Engineering, Saga University,Japan
ABSTRACT: This paper presents a case history of the performance of prefabricated vertical drain (PVD) improved soft clay deposit under embankment loading in eastern coastal region of China. The thickness of soft deposit is 16 to 20 m. The PVDs were installed to a depth of 19 m with spacing of 1.5 m in a triangular pattern. Field performance of embankments constructed on natural subsoil and PVD improved subsoil is analyzed using FEM. A simple approach for estimating the equivalent vertical hydraulic conductivity of PVD improved subsoil is used. Following influential factors, (1) well resistance (discharge capacity), (2) smear effect, and (3) drainage boundary conditions are investigated. Furthermore, the back calculated field performances of PVD are compared with these of other case histories reported in the literatures. Suggestions are made on determining proper design parameters related to the PVD behavior on soft clay deposit. 1 INTRODUCTION
2 GEOTECHNICAL PROFILE
In China, many lines of expressway have been constructed within recent 10 years in order to improve the infrastructures. Hongzhou-Ningbo (HN) expressway locates on the southern coast of Hongzhou bay, eastern coastal region of China. It starts from Hongzhou, the Capital of Zhejiang Province to Ningbo, the biggest harbor city of the same Province. The total length of HN expressway is 145km, in which about 92km is passing through soft clay deposit (in China it is called mucky clay). In order to get the reliable data and experiences to guide the design and construction, 12 field hll-scale test embankments with total 3.15km long were constructed and investigated (Wang et al., 1998). Ground treatment methods used were preloading plus: (a) gravel sand mat on natural subsoil, (b) two layer geotextiles reinforced gravel sand mat on natural subsoil, (c) vertical drain (VD), including sand drain (SD) and plastic vertical drain (PVD). The soft ground can be divided into two typical sections based on its thickness. The first one is relatively thicker with a thickness of soft soil 38 to 51m. The thickness of the other section is much thinner than the first section and the thickness of soft clay is 16 to 20m. The behavior of two test embankments constructed on PVD improved subsoil and on natural subsoil at the thinner deposit section is investigated in this paper.
The generalized soil profile and soil properties of soft deposit at the thinner section are shown in Figure 1. The soil profile is consisting of a thin weathered crust (TC) from 1 to 1.5m thick overlying a silty clay (SCl) approximately 4m thick. The third layer is very soft mucky clay (MC) with a thickness of approximately 1Om. Underlying the mucky clay is a soft clay layer called mucky-silty clay (MSC) approximately 4m thick. Below it is a medium to stiff silty clay layer (SC2) from 3 to 5m thick extending down to 21.3 to 23m depth which is in turn underlain by a layer of loose clayey sand (CS). The soft silt clay and mucky clay have the water content greater than their liquid limit, low hydraulic conductivity and lower shear strength. 3 CONSTRUCTION OF TEST EMBANKMENTS Figure 2 plots the cross section and plan of the test embankments and instrumentation. As seen in the figure, a 0.5m thick sand mat (hydraulic conductivity: >O.OOlm/s) was placed on soft ground at first. Then, the decomposed granite was filled and compacted to a unit weight of 20kN/m3. The height of embankment on improved subsoil is 5.88m and that on natural subsoil is 4.66m. For the improved case, prefabricated vertical drains (PVDs) were installed in a triangular pattern with the spacing of 1.5m and 19m deep. Table 1 lists the properties of PVDs used in the field. The discharge capacity Q,,, provided by manufacturer is greater than 1500m3/year. 515
Figure 1. Soil profile and soil properties at the test site of thinner deposit section.
4 MODELING OF THE PVD IMPROVED
SUBSOIL
Figure 2. Cross section and plan of embankment and field instrumentation.
Field behavior of soft ground under embankment loading were observed through monitoring surface settlement, sub-surface settlement, lateral displacernents of the subsoil under the middle point of embankment slope and about lm from embankment toe, and pore pressure variation by piezometers at different soil layers, The field instrumentation points are also indicated in Figure 2. Table 1. Size and physical properties of PVDs. Thickness Width Unit 0,". Material (mm) (mm) weight (m3/yr) Filter Core (g/m) 6 100 108 1580 Nonwoven Polyolefine polyethylene * Provided by manufacturer.
In order to analyze the behavior of PVD improved soft subsoil, finite element method (FEM) is used. There are three existing approaches for modeling the PVD improved subsoil in FEM. The first method employs a 1-D drainage element (Hird et al., 1992; Chai et al., 1995). The second method adopts a special formulation of FEM program (Sekiguchi et al., 1986). The third method is a simple approximate way, in which an equivalent value of vertical hydraulic conductivity (kVJ of PVD improved subsoil is estimated. Then, the behavior of PVD improved subsoil can be analyzed as the same way as that of natural subsoil case through using this equivalent hydraulic conductivity value (Chai & Miura, 1999a). It is proved theoretically that the maximum error for the average degree of consolidation using the third method is less than 10%. When considering the vertical drainage of deposit and well resistance, it is confirmed that the error is much smaller compared using the aforementioned first method in practical engineering case (Chai & Miura, 1999a). The third approach is used in this study. 4.1 Epivalenf hydraulic conductivity
The equivalent vertical hydraulic conductivity of PVD improved subsoil, k,,,, can be expressed as (Chai and Miura, 1999a):
where !=drainage length; Ddiameter of unit cell; and kh, k,,=horizontal and vertical hydraulic conductivity of soil layer, respectively. Parameter p can be expressed as follows:
n k 3 21'kh p =In-+~ln(s)--+z----s k, 4 3q,, 516
where n =D/d,, (&,=diameter of vertical drain); s=d,/d,,, (d,=diameter of smear zone); k,=horizontal hydraulic conductivity in the smear zone; and q,,,=dischargecapacity of PVD in field. 4.2 ~ o d e i j ofdrain n~ effect 4.2.1 E ~ ~ ~ i ~ ~ drain a i ediameter nf The modified approach for equivalent drain diameter (d,,,)of a band-shaped PVD drain by considering the corner effect is used in this study. It can be calculated using the following equation (Rixner et al.,
Table 2. Cfvalue for a few clay deposits. Deposit CJ References Bangkok clay (1OOkm from sea) 25 Chai et al. (1995) Bangkok cIay (close to sea) 4 Chai et al. (1996) Malaysia Muar clay deposit 2 Chai and Bergado( 1993) Ariake clay (close to sea) 4 Chai and Miura( I999b) Louiseviile (Canada) I Tavenas et al. (1986) St-Alban (Canada) 3 Tavenas ct al. (1986) Mucky clay in China 6 Present study (close to sea)
Many uncertainty factors affect the value of kdk, in field. A simple approach has been proposed to evaluate this value (Chai and Miura, 1999b).
1986):
d,,
w -I-f
zz
2
where w-Tidth of a band-shaped PVD and t=thickness of the PVD. where C'hydraulic conductivity ratio between field and laboratory values; khrub=horizontal hydraulic conductivity from laboratory test; k,~~b=hydrauIic conductivity in smear zone from laboratory test. Cf is greater than unity and varies with the clay type and sediment environments. The Cfvalues of a few clay deposits are listed in Table 2. The value of khlQb/kslub is varied from 1 to 5, based on laboratory tests (Hansbo, 1987; Madhav et al., 1993).
4.2.2 Discharge capacity of PVD The discharge capacity of P W in field is much lower than the value reported by the manufacturer. The main factors for the well resistance may be as followings: 1. Con~ningthe drain by clay. 2. Effect of air bubbled trapped in the drainage path. 3. Effect of folding of the drain. 4. Clogging by clay particles. 5. Creep of drain filter. All above factors make the discharge capacity reduction in field. Chai and Miura (1999b) showed that long term confined in clay discharge capacity of several PVDs is only 5% to 10% of that for short term confined in rubber membrane.
4.2.4 EJ^fecfof sand maf Part or all of the water coliected by the drain flows to the ground surface first, and then drains out by the outlet system, the sand mat. The hydraulic conductivity of sand mat affects the analytical results. Chai and Miura (1999b) showed numericaIly that if the hydraulic conductiv~tyof sand mat is greater than 10-4m/sec,the free drainage path assumption for the sand mat is acceptable. In this case, the hydraulic conductivity of sand mat is 10"3mfsec,which is larger than 10-4mlsec,so that the free drainage assumption is used in this study. 4.3 ~ o conditi~ns ~ and ~no ~ e i ~~a r a m e~tin ers FEM analysis
4.2.3 Smear eJ^fect When a PVD is ~ n s t a ~ ~ine soft d clay g r o u n ~by using a mandrel, a disturbed zone cailed smear zone around the mandrel is created, in which the hydraulic conductivity is reduced significantly. There are two parameters needed to evaluate the smear effect, namely, the diameter of the smear zone (d,) and the hydraulic conductivity ratio (kdk,), i.e., the value in the undisturbed zone (4th) over that in the smear zone
In the finite element analysis, the plain strain condition was assumed. The modeled range was 25m deep from ground surface, and horizontally 120m away from the embankment centerline. The displacement boundary conditions were as follows: at bottom, both vertical and horizontal displacements were fixed, and for left and right vertical boundaries, the horizontal displacement was fixed. The adopted drainage boundary conditions were as follows: the ground surface and bottom line (weathered rock) were drained. The left and right boundaries were undrained. Figure 3 shows the FEM mesh for the embankment on natural subsoil.
(ks). Based on several field and laboratory test results, the diameter of the smear zone dsis estimated from Equation 4 (Jamiolkski et al. 1983; Hansbo 1987; Miura et al. 1993): ds=(2 to 3)d,,
(4)
where d,~~=equivalent diameter of the cross-sectiona1 area of mandrel. In practice, if there are no test data for evaluating the smear zone size, it is suggested that the value of dS=3d,, can be used (Chai & Miura 1999b).
517
Table 3. Model parameter for natural subsoil. clay p* eo h K M V kh k,, layer (m/m3) ( 10-8m/s) TC 19.3 0.81 0.08 0.001 1.0 0.3 3.0 3.0 SC 1
18.5
1.07
0.17
0.017
1.0 0.35 0.64 0.26
1.26 0.23 0.023 0.8 0.35 2.58 1.72 MC 14.3 1.10 0.18 0.018 0.8 0.35 1.90 0.96 SMC 17.9 S C 2 19.3 0.81 0.08 0.001 1.0 0.35 0.45 0.21 * p =unit weight; eFinitial void ratio; h=virgin loading slope in e h @ plot (p is effective mean stress); K=reloading/ unloading slope in e-In@ plot; and Wslope of failure line in p versus q plot (q is deviator stress); iFPoisson‘s ratio.
120m
Figure 1. Finite element analysis mesh. Table 4. Parameters related PVD behavior.
Table 5. Equivalent values of hydraulic conductivity of PVD improved subsoil. Clay layer Depth khk, keh ke, kelf%h=k,Jk, (m> (x 10-81ds) Top crust 0-1.2 1.0 39.8 39.8 13.3 Silty clay 1 1.2-4.8 2.5 19.9 8.1 31.3 Mucky clay 4.8-14.3 1.5 50.3 33.4 19.5 Mucky Siltyclay 14.3-18.3 1.9 15.1 8.0 7.9 Silty clay 2 18.3-19.0 2.2 12.7 5.8 28.3
The mechanical behavior of the soft clay layers was represented by the Modified Cam-Clay model and the clayey sand layer, the sand mat as well as decomposed granite fill material were assumed to be elastic. The determined model parameters for subsoil are listed in Table 3. For the clay layers, the parameter were determined from a laboratory standard oedometer test and triaxial test results on the undisturbed samples (Sun et al. 1988), except for Poisson’s ratio and hydraulic conductivity. Poisson’s ratio, v, was assumed empirically. For the values of hydraulic conductivity, first, the representative laboratory values of khlab and kvlab are selected as shown in Figure 1. From Figure 1, the ratio of khniv was determined as fiom 1.5 to 2.5 for the soft clay layers. Then, the values of vertical hydraulic conductivity were adjusted to fit the observed field data of the embankment on natural subsoil, which are about six times the laboratory values ( C F ~ ) The . values of hydraulic conductivity listed in Table 3 were initial values with respect to initial void ratio (eo); during consolidation, they vary with void ratio according to Taylor’s equation (Taylor, 1948). Parameter Ck in the equation was set up as 0.45eo in this study. The subsoils are in lightly overconsolidated to normally consolidated states with a maximum overconsolidation ratio (OCR) of about 5 for the top crust. The lateral earth pressures were calculated using the equation proposed by Mayne and Kulhawy (1982).
The ground-water level was about 1.5m below ground surface. The parameters for clayey sand were assumed as Young’s modulus E=25,000kPa and Poisson’s ratio 1-0.25. The unit weight of clayey sand was 19.5kN/m3. The mechanical properties of the fill material were assumed as follows: E=30,000kPa and ~ 0 . 2 5The . unit weight of the fill material was 2 0 1 ~ ~ ~ . The parameters for PVD are listed in Table 4. The equivalent drain diameter for PVD was calculated using Equation 3, the diameter of the smear zone was estimated 3 times the equivalent mandrel diameter (Chai & Miura 1999b). The average value of khladksrclbwas from 2 to 2.5 (Madhav et al. 1993). Since the ratio of C ~ i s6 in this case, according to Equation 5, kdkS is fiom 12 to 15. For the value of discharge capacity in the field (qw$, the backevaluated value was from 79 to 158m /yr to fit the measured settlement of the PVD improved case. The equivalent hydraulic conductivity values for the PVD improved subsoil are calculated by Equation l, using the aforementioned parameters. The results are listed in Table 5. The hydraulic conductivity of PVD improved subsoil is increased from 8 to 31 times. 5 BACK ANALYSIS OF TEST EMBANKMENTS 5.1 Analysis of the embankment on natural subsoil The purpose of analyzing the embankment on natural subsoil is to verify the model parameters as well as the numerical procedure, as proposed by Chai and Miura (1999b). The initial values of hydraulic conductivity in Table 3 were adjusted by comparing the numerical results with the measured data of the embankment on natural subsoil. The simulated results for settlement and excess pore pressure are compared with the measured data depicted in Figure 4. Figure 5 shows the simulated and measured values of lateral displacement under the middle of embankment slope for the case on natural subsoil. As seen fiom Figure 4, the analysis simulated the settlement curve well. If checking it carefully, we can 518
find that in the initial period, FE analysis is a little over-estimated the settlement, however, after 600 days, FE analysis underestimates the settlement. This is because the secondly consolidation effect and this effect cannot be considered in the Modified Cam-Clay model.
Figure 6. Embankment construction procedure. settlement-time and excess pore pressure-time curves for embankment on PVD improved subsoil.
Figure 4. Comparison of the simulated and measured values of settlenient and excess pore pressure for e K ~ b ~ one natural n ~ subsoil.
Figure 7 . Simulated and measured lateral displacement profile for embankment on PVD improved subsoil.
5.2 Analysis of enibariknienfson PVD improved su bsoif
5.2.1 Back-calculated valzie of discharge capaciy In order to fit the measured data, the discharge capacity Of the pm was varied in the The numerical result shows that the discharge capacity is about 158rn3/yrand 79rn3/yr with respect to the hy-
Figure 5 . Simulated and measured lateral displacement profile for eItibankme~ton natural subsoil.
519
draulic conductivity ratio kdkS=15 and 12. The adopted values are kdks= 13 and qH,=1OOm3/yr. These back-analyzed values agree well with the laboratory and field data in soft Ariake clay as reported by Chai and Miura (1999b). 5.2.2 Comparison between measured and simulated values Figure 6 depicts the simulated and field measured data of the settlement and excess pore pressure for the embankment on PVD improved subsoil. The simulated and measured values for the lateral displacement are shown in Figure 7. The results given in Figures 6 and 7 are under the conditions of discharge capacity q,,,=100m3/yr and smear effects of kh/ks=13 and ds=3d,,,. As seen in the figure, based on these parameters, the simulated settlement agreed well with the field data. However, if checking carehlly, we can notice that the analysis predicted well during the construction period and slightly overestimated the settlement during the consolidation period after construction until to about 600 days (finished of primary construction). One possible reason is the continuous variation (reduction) of discharge capacity of PVD with elapsed time. In the analysis, it was assumed that the field discharge capacity of PVD is constant. Chai and Miura's (1999b) laboratory test results showed that the discharge capacity reduced with the elapsed time. Another reason for the FEM under-prediction of the settlement after 600 days, is that after 600 days, the primary consolidation finished and the secondly compression could not be simulated by the Modified Cam-Clay model. The simulated excess pore pressure is lower than the measured one during consolidation period. The reason is not clear. The analysis shows that, for the same embankment height 4.66m, although the absolute lateral displacement for the PVD improved case is about twice that of the embankment on natural subsoil (as shown in Figure 5 and 7 ) , the ratio between maximum lateral displacement and settlement is 0.37 for improved case and 0.48 for unimproved case. PVD improvement reduced the lateral to vertical displacement ratio. 6 CONCLUSIONS From the field measured data and the FEM simulated results, the following conclusions are given: 1. The field hydraulic conductivity is about 6 times the laboratory test values of soft clay deposit at the test site in eastern coastal area of China. 2. The field discharge capacity of PVDs in soft mucky clay is about 100m3/yr.This value agree the laboratory results on discharge capacity tests with clay confinement (Chai and Miura, 1999b). 3. The hydraulic conductivity in smear zone is about 1/13 of the horizontal hydraulic conductivity
of the field value. 4. The procedure for analyzing the PVD improved subsoil proposed by Chai and Miura (Chai and Miura, 1999b) is applicable to the soft clay deposit in China. 5 . Installation of PVD with the spacing of 1.5m makes the vertical mass hydraulic conductivity of soft clay deposit increasing about 20 times. REFERENCES Chai, J.C., Bergado, D.T., Miura, N., & Sakajo, S. 1996. Back calculated field effect of vertical drain. Proc. 2nd Int. Conf Soft Soil Engrg. Nanjing: Hohai University, Voll:270-275. Chai, J.C. & Miura, N. 1999a. A simple method for analyzing consolidation of PVD improved subsoil, Proc. 8th Australia-New Zealand Conference on Geomechanics, Hobart, Vol. 1: 243-249. Chai, J.C. & Miura, N. 1999b. Investigation of factors affecting vertical drain behavior. Journal of Geotechnical and Geoenvironinental Eiigineering, ASCE 125(3): 2 16-226. Chai, J.C., Miura, N., Sakajo, S., & Bergado, D.T. 1995. Behavior of vertical drain improved subsoil under embankment loading. Soils and Fouiidations 35(4): 49-6 1. Hansbo, S. 1987. Design aspects of vertical drains and lime column installations. Proc. 9" Southeast Asian Geotech. Conf., Bangkok: Southeast Asian Geotechnical Society, Vol.2: 8-1-8-12. Hird, C.C., Pyrah, I.C. & Russell, D. 1992. Finite element modelling of vertical drains beneath embankments on soft ground. Geotechnique 42(3): 499-51 1. Jamiolkowski, M., Lancellotta, R., & Wolski, W. 1983. Precompression and speeding up consolidation, general report. Spec. Session 6, Proc. 8* Eur. Cod. SMFE, Rotterdam: Balkema, 1201-1226. Miura, N. & Park, YM. &Madhav, M.R. 1993. Fundamental study on the discharge capacity of plastic board drain. J. Geotech. Engrg., JSCE, 35(III): 3 1-40 (in Japanese). Madhav, R., Park, Y.M., & Miura, N. 1993. Modelling and study of smear zones around band shaped drains. Soils aiid Foundations 33(4): 135- 147. Rixner, J.J., Kraemer, S.R., & Smith, A.D. 1986. Prefabricated vertical drains. Engnrg. Guidelines, FWHA/RD-86/168, Washington, D.C. : Federal Highway Administration, Vol. 1. Sekiguchi, H., Shibata, T., Fujimoto, A. & Yamaguchi, H. 1986. A macro-element approach to analyzing the plain strain behavior of soft foundation with vertical drains. Proc. 3 1st Synp. JGS: 111-1 16 (in Japanese). Sun, J. & Wang, B.J. 1988. Finite Eleriient Analysis in Geotechiiical Engineering. Shanghai: Tongji University Press. (in Chinese). Tavenas, F., Tremblay, M., Larouche, G. & Leroieil, S. 1986. In-situ measurement of permeability in soft clays. ASCE Spec. Conf. on Use of In-situ Tests in Geotech. Engrg., ASCE, New York, 1034-1048. Taylor, D.W. 1948. Fundanientals of soil mechanics. New York: Wiley. Wang, 2. M. et al. 1998. Field experimental study on the soft ground treatment of Hongzhou-Niingbo (HN) expressway foundation. In Ti-leng Cai (ed.), Soft Ground Treatment in HAT Expressway; Hongzhou: Hongzhou Press. (in Chinese).
520
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Strength and deformation characteristics of cement-mixed soft clay M. Sugai Maeda Corporation, Tokyo, Japan
ETatsuoka, M. Kuwabara & K. Sugo Universip of Tokyo, Japan
ABSTRACT The strength and deformation characteristics of a dredged Holocene soft marine clay mixed with a low content of cement were evaluated by unconfined and consolidated drained triaxiaI compression tests and oedometer tests. The strength and deformation characteristics of the original untreated clay were also evaluated. The stiffness of the cement-mixed clay becomes very low as the yielding starts in both shearing and compression. At pressures lower than a certain level, the cement-mixed clay is stronger than the untreated clay, while the reverse is true at higher pressures. The smaller peak strength of the cementmixed clay is due to large void ratios, close to the initial value, maintained even at high pressures, unlike the untreated clay. The strength of the cement-mixed clay becomes similar as the untreated clay only at very high pressures after having exhibited large compression with substantial damage to the original cementation. conditions for a long term during and after construction.
1 INTRODUCTION Very high water content soft clay dredged from the sea bottom was often dumped into surrounding sea areas, which is now basically not allowed for environmental conservation. Dredged clay is often used as the backfill material for reclamation projects along sea shores. In such projects, however, due to very low shear strength and rigidity of dredged clay, before commencing subsequent construction work, reclaimed land should be, for example, reinforced by using geosynthetic sheets, or improved with means of sand drains or sand compaction piles. Yet, the residual deformation of such reclaimed land could be too large. As a method to construct stable on-shore landfills with small residual deformation using dredged clay, the clay pre-mixed with a relatively small amount of cement before filling. This construction method, without compaction, is often employed because of its high construction speed. In such cases, it is important to find the minimum amount of cement that is sufficient and necessary for the stability and small residual deformation of constructed landfills. Although the strength and deformation characteristics of such a material is evaluated usually by unconfined compression tests, the behaviour at relatively high pressures, which may be encountered in some field cases, such as high embankments to a height of, for example, 50 m, cannot be evaluated by this conventional test method. In some cases, it is necessary to evaluate the deformation of fill that takes place under drained
The objective of the present study is therefore to evaluate the drained strength and deformation characteristics of dredged marine soft clay with a high water content (about 50%) mixed with a relatively low content of cement. To this end, a series of unconfined and consolidated drained triaxial compression (TC) tests were performed for a wide range of consolidation pressure up to 1.96 MPa. A dredged clay with from a actual construction project was used (Fig. 1 and Table 1). For this clay, it had been required to determine the cement content so that an about 50 m high embankment can be constructed safely and with small residual deformation, but at the minimum construction cost. To evaluate the yield pressure and compressibility under one-dimensional (1D) conditions, a series of oedometer tests were also performed. For a comparison purpose, similar tests on normally consolidated untreated clay (without cement mixing) were also performed. oI3
2 TESTING PROCEDURES 2.1 Test material The material described in Figure 1 was used to prepare 10 cm-diameter specimens, and it was seived to remove soil particles larger than 2mm for prepare 5 cm-diameter specimens. The natural water 521
content of the tested dredge clay was higher by 6.6 % than the liquid limit. As a typical field value, a cement content of 3.5 % of clay in dry weight was selected. After mixing with cement, slurry of the cement-mixed clay was poured into a split mould having the inner dimensions that were the same as the dimensions of the specimens for unconfined and TC tests, and then cured for seven days under sealed constant water conditions before using for each compression test. To prepare specimens of untreated clay, the original clay was mixed again, and the slurry was poured into a 1D consolidation cylinder with an inner diameter of 20cm. After thoroughly de-aired, the slurry was re-consolidated at a vertical stress of 98 kPa for several days. Specimens for TC tests were trimmed from a large consolidated clay cake. 2.2 Drained triaxial compression tests
4 Cement-mixed clay consolidated to
CT = 0.049 0.588 MPa: The diameter and height of specimen were 10 cm and 20 cm. The TC specimens had the same initial water content and degree of saturation as prepared. The specimens were aged for 24 hours at the final consolidation state and the axial strain rate was 0.03 %/min in all the unconfined and TC tests in the present study. The effects of bedding error at the top and bottom ends of specimen could be significant in such compression tests on a cement-mixed soil at relatively low C T ' ~ (Tatsuoka and Kohata, 1995). Therefore, axial strains were measured locally with a pair of local deformation transducers (LDTs; Goto et al., 1991), set at the opposites ends of specimen diameter (Figure 2). Axial strains were measured also externally for monitoring. Changes in the specimen diameter were measured by using three clip gauges set around the specimen diameters (Tatsuoka et al., 1999). To ensure uniform deformation of specimen, which is essential for reliable local strain measurements, the top and bottom ends of specimen were capped with gypsum on site in the triaxial cell. l3
b) C e m t-e n- e d clay consolidated tQ CT',= 1.B and 1.96 MPa: The diameter and height of specimen were 5 cm and 10 cm. Axial strains were measured only externally, as the effects of bedding errors become negligible after consolidated up to such high pressures. As the clip gauges were not used, the test at CT = 19.6 MPa was performed on a saturated specimen, measuring the volumetric strain by the amount of pore water expelled from the specimen. The volumetric deformation characteristics from this test was used to estimate that in the test at CT = 11.6 MPa on an unsaturated specimen. Differences in the drained strength and deformation characteristics between unsaturated specimens as prepared and saturated ones under otherwise the same conditions were found negligible with this cement-mixed clay. c) Untreated clay consol idated to o',= 0,098 0.49 MPa: The diameter and height of specimen were 5 cm and 10 cm. Axial strains were measured only externally. Ic
IC
Table 1 Physical properties of the tested dredge clay
1.381
IVoid ratio -
-
~_
-
I I
Plastcity index
I
I
Figure 2. Triaxial testing method.
4%1 22.4% 20.7
522
2.3 Constant-rate-of-strain oedometer tests on cement-mixed and original clays (Figure 3)
kPa. To ensure drained conditions, a low axial strain rate (0.0055 %/min and 0.015 %/min for cement-mixed and untreated clays) was used. As the axial strains were measured externally, the effects of bedding errors would not be negligible, in particular with the cement-mixed clay at relatively low stresses.
Specimens had a diameter Of cm and a height Of cm. The cement-treated specimens were prepared the Oedometer and Pouring the curing in it, while the untreated clay specimens were by trimming the consolidated cake. The -prepared specimen was saturated with a back pressure of 98
3 TEST RESULTS
3.1 Behaviour at
CJ, 5
0.59 MPa
Figures 4 and 5 show the relationships among the deviator stress q = o1- 0, , the axial strain E , and the volumetric strain E from the drained TC tests on cement-mixed and original clays (and one unconfined compression test on the cement-mixed clay). A short broken curve seen in the stress-strain curve for CJ ', = 589 kPa denotes the range without data due to a malfunction of the data recording system. Figure 6 shows the stress paths together with the peak stress points. The following trends of behaviour may be seen from these results. 1) The stiffness and peak strength of the normally
Figure 3. Oedometer tests in t h e triaxial cell.
U
,'=294.3 kf'a U
\
100
.'=I962 kPa
-
c \
2
2 250 :350
v
b 300 (
0
-
a . . . +.J
* L 2 .-
6>
0
1
2
3
1
200 -
1.io
100
-
-
-2.0
5
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
Volumetric strain, E (%) froin unconfined and drained TC tests (d3= 0.049 - 0.588 MPa)
Axial strain, ( E (%) Figure 4. Relationships between; a) q and E , ; and b) q and on cement-mixed clay .
A
5
v
P
Figure 5. Relationships between; a) 4 and clay.
E , ; and
b) q and
E
froni tlie drained TC tests (d3= 0.09s - 0.49 MPa) on untreated
523
consolidated untreated clay increases essentially proportionally with the confining pressure o , as expected. The specimens exhibit very similar volume contraction until the peak stress state, irrespective of the consolidation pressure o 2) On the other hand, the pre-peak stiffness of the cement-mixed clay first decreases with o up to o ,' = 98 kPa, and then increases with the further increase in The shear yield point was defined as the point where the curvature along each log q and log E , curve is maximum and plotted in Figure 6. It is seen that corresponding to the trend described above, with the increase in o ,' , the shear yield stress (qy)shear first decreases, and as loaded in TC beyond the shear yield point, the stiffness decreases substantially and the volume contraction becomes more noticeable. The decrease in (qy)shear with the increase in o', may be due to the damage by isotropic compression to the cementation that has developed during the initial curing. As oI3 increases more, (qy)shear starts increasing gradually, which may be due to effects of restructuration, or re-bonding at inter-particle points (Barbosa Cruz and Tatsuoka, 1999) developing at the final consolidation stress state (i.e., ageing effects in general terms). A similar result with another type of cement-mixed saturated clay has been reported (Tatsuoka and Kobayashi, 1983). Note however that the (qy)shear values of the cement-mixed clay at these o', values are comparable with those of the untreated clay developed by drained creep under similar consolidation stress states. The behaviour of the cement-mixed clay in drained TC at o', larger than 192 kPa is less contractant than that at lower o',s, which may be due to relatively large isotropic compression that has taken place before these TC tests. 3) As the peak strength of the cement-mixed clay increases only very gradually with the ot3up to OS9 MPa, the peak strength becomes nearly the same with that of the untreated clay at o', of around 0.2 MPa, or p ' = (a + 20 / 3 at failure of around 0.3 MPa. At values of o', (or p ' ) higher than that respective value, the cementmixed clay becomes weaker than the untreated clay, with the difference increasing with oI3 in the range of o', shown in Figure 6. As shown later, as oI3 (or p ' ) increases further, the peak strength of the cement-mixed clay starts increasing, approaching to that of the untreated clay. l3
13.
I3
o I 3 .
I,
are shown in Figure 8. The following trends of behaviour may be noted: 1) In this range of o ' , (or p ' ) , the volume contraction of the cement-mixed clay in both isotropic and TC compression is significantly smaller than that of the untreated clay, due likely to the resistance against compression by cementation. Therefore, the void ratio of cement-mixed clay at the ultimate failure is substantially larger than that of the untreated clay. It is likely that in this pressure range, the peak strength of the cement-mixed clay is controlled mainly by the cementation that has developed during the initial curing, while this initial cementation could be somehow damaged by isotropic compression. At the same time, as the total volume contraction that has taken place until the ultimate failure is very small, it could not contribute noticeably to the increase in the peak strength. For these reasons, the strength of the cement-mixed clay becomes significantly lower than that of the untreated clay. 2) The cement-treated clay starts yielding at an isotropic compression pressure o =(p7y)iso.comp of around 0.15 MPa, which is denoted by a open triangle in Figure 6. Although it is much smaller than that of the untreated clay, the cement-mixed clay starts exhibiting noticeable volume contraction by isotropic yielding at p' values exceeding (pyy)iso~comp . l3
3.2 Behaviour at very high pressures Figure 9 shows results similar to Fig. 4, obtained from the two drained TC tests at o t 3 = 1.18 MPa and 1.96 MPa on the cement-mixed clay, together with those of two tests at lower o*,s. At these high o', values, the peak strength is substantially larger than that at oI3 equal to, or lower than, OS9 MPa. The effective stress paths together with the peak stress points from all the compression tests are presented in Figure 10. It may be seen that the peak strength of the cement-mixed clay has become
13)
The relationships between the void ratio and the effective mean principal stress p' during isotropic and triaxial compression of the cement-mixed and untreated clays are presented in Figure 7. The zoomed-up relationships for the cement-mixed clay 524
Figure 6. Stress paths with the peak stress points.
close to that of the untreated clay at these high pressures, while exhibiting a large volume contraction until the ultimate failure state, similarly to the untreated clay (see Fig. 5b).
Figure 7. Relationships between void ratio and p ’ during isotropic and TC compression of the cement-mixed and untreated clays.
Thin solid curves in Figure 11 represent the relationships between the void ratio and p ’ (obtained by assuming that a ’ ,= a ’ , / 2 )from one oedometer test on the cement-mixed clay and other two on the untreated clay. The relationships during isotropic and subsequent TC compression are also shown. The isotropic yield pressure of the two triaxial specimens is larger than the value seen in Figure 8, which is considered more reliable due to a large specimen size (10 cm in dia. and 20 cm in height) and a larger number of tested specimens. For both cement-mixed and untreated clays, the relationships from the two different types of tests are consistent to each other. It is likely that the difference in the void ratio between the oedometer test and the isotropic compression tests for the triaxial specimens of cement-mixed clay would be due to relatively large effects of bedding errors in the oedometer test. The compressibility in 1D compression of the cement-treated clay at p ’ exceeding (P’y)iso.comp is noticeably larger than that of the untreated clay. A large decrease in the void ratio during the isotropic compression up to o = 1.18 MPa and 1.96 MPa and that during drained TC loading would explain the increase in the peak strength in the TC tests at these oI3values. It may be seen from Figures 10 and 11 that the difference in the peak strength between the cement-mixed and untreated clays is largest when the difference in the void ratio is largest. l3
Figure 8. Zoomed-up e-p’ relationships in isotropic and TC compression for the cement-mixed clay.
It should be noted however that the axial strain at the peak stress state is very large and the overall prepeak stiffness is very low in these TC tests at very high pressures, similarly to the normally consolidated untreated clay (Figure 5). This feature can also be seen from Figure 12, in which the contours of same axial strains and the shear yield points from the drained TC tests on the cementmixed clay are summarised. It is likely that as the void ratio decreases largely in isotropic compression 3500
3500
3
-
3000 2500
v
0.
-
2000
m
2 1500 i
5 1000 .> 500 3
10
Axial strain,
20
15 E
0
25
-
2
-
, (%)
Figure 9. Effects of c f 3 on the relationships between q and mixed clay for a wide rang of d3
1
0
1
2
Volumetric strain, E,
and those between q and
&
3 E vo,
4
5
6
7
(%)
in drained TC tests on the cement-
noticeably larger than those of the untreated clays at these very high pressures, the peak strength of the cement-mixed clay is still somewhat smaller than that of the untreated clay.
4 CONCLUSIONS 1) When the pressure level is below the yield pressure, the void ratio of the cement-mixed clay does not decrease noticeably in isotropic or ID compression, the void ratio being higher than that of untreated clay. In this range of pressure, the peak strength is nearly constant, controlled by the cementation that developed during the initial curing. 2) As the consolidation pressure becomes larger than the yield pressure and as the shear stress becomes larger than the shear yield stress in drained TC, the cement-treated clay becomes much softer, exhibiting large volume contraction. 3) The peak strength of cement-mixed clay could be much smaller than that of the untreated clay in some pressure range where the void ratio is substantially larger than that of the untreated clay. It seems that the strength difference is largest when the difference in the void ratio is largest. 4) After having exhibited a large compression at very high pressures, the strength of cement-mixed clay increases with pressure, approaching that of the untreated clay. However, due to large damage to the initial cementation by large compression that takes place until the ultimate failure, the pre-peak stress-strain behaviour becomes very soft, exhibiting large axial strain and volume contraction by the peak stress state.
Figure 10. Effective stress paths with the peak stress points from all the compression tests.
Figure 11. Relationships between e a n d p ’ in isotropic and 1D compression and T C for cement-mixed and untreated clays.
Figure 12. Contours of same axial strains and shear yield points from the drained T C tests on the cement-mixed clay.
and subsequent drained TC, the original interparticle bonding by cementation have been nearly totally damaged, while the strength gain by ageing at the final consolidation pressure is not significant compared with the high consolidation pressures. It seems that due to the void ratios that are still 526
It can be suggested therefore that for constructing a high landfill with using a cement-mixed soft clay, the cement content should be determined so that the working stress in each lift is always lower than the instantaneous yield stress. That is, the cementmixed clay should not yield at any moment after having been cast-in-place. Otherwise, the fill may exhibit large deformation with cracking at some locations. It is unsafe to assume that the strength of the cement-mixed clay is always larger than that of the untreated clay, but at pressures exceeding the yield pressure, the peak strength of the cementmixed clay could become substantially smaller than that of the untreated clay. Further study will be necessary to generalise the conclusions, in particular for more plastic clays.
REFERENCES Barbosa-Cruz,E.R. and Tatsuoka,F. 1999. Effects of stress state during curing on stress-strain behaviour of cement-mixed sand. Proc. Second Int. Conf: on Pre-Failure Deformation Characteristics of Geomaterials, IS Torino '99 (Lo Presti eds.) 1:509-516. Rotterdam: Balkema. Goto.S, Tatsuoka,F., Shibuya,S., Kim,Y.-S. and Sato,T. 1991. A simple gauge for local small strain measurements in the laboratory. Soils and Foundations, 31(1): 169-180. Tatsuoka,F. and Kobayashi,A. 1983. Triaxial strength characteristics of cement-treated clay. Proc. the 8th European Conf: on SMFE, S(1): 421-426. Tatsuoka,F. and Kohata,Y. 1995. Stiffness of hard soils and soft rocks in engineering applications. Keynote Lecture, Proc. of Int. Symposium Pre-Failure Deformation of Geomaterials, IS-Hokkaido (Shibuya et al., eds.) 2: 947-1063. Rotterdam:Balkema. Tatsuoka,F., Modoni,G., Jiang,G.L., A n h Dan,L.Q., Flora,A., Matsushita,M., and Koseki,J. 1999. Stress-Strain Behaviour at Small Strains of Unbound Granular Materials and its Laboratory Tests. Keynote Lecture, Workshop on Modelling and Advanced testing for Unbound Granular Materials, January, 1999, Lisboa (Correia eds.): 17-61. Rotterdam: Balkema.
527
This Page Intentionally Left Blank
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida(eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 I
Compaction pile method utilizing coal ash as filled material H.T&, Y. Nishio & H. Suzuki Chubu Electric Power Company Incorporated, Japan
M. Higo, K. Harada & H. Nitao Fudo Construction Company Limited, Tokio,Japan
ABSTRACT: Clean sand is normally used as the filled material in the sand compaction pile (SCP) method of soft ground improvement, the most widely used method in Japan. But environmental concerns have prompted as an alternative use of coal ash, a by-product of coal-fired thermal power stations, and site tests have been carried out using pile materials composed mostly of coal ash. This paper reports on this experimental ground strengthening work undertaken on a coastal coal ash-filled landfill site liable to liquefaction, to confirm the implementation characteristics and improvement effect of the pile materials. The results showed that the SCP method can be used as a countermeasure (increased density) to liquefaction in coal ash ground, and that coal ash makes a suitable filled material. 2 REQUIRED PERFORMANCE AND MATERIAL PROPERTLES OF FILLED MATERIAL
1 INTRODUCTION An important environniental technology issue for Japan’s construction industry is the effective use of construction-displacedsoil from construction sites, and coal ash produced by coal-fired thermal power stations. The quantity of fly ash and clinker ash produced by power stations amounted to 5.15 million tons in 1995, and is expected to increase considerably, to about 9 million tons in 2000 (Matsuo et al. 1997). At that same time, the quantity of sand available for use as filled material for the SCP method of ground improvement has been declining year by year and it may be extremely difficult to secure a supply in the future. As a result, a ground improvement method that makes effective use of waste soil has been developed for practical applications (Matsuo et al. 1997). It uses waste soil as the filled material in a compaction filled method. Because quality stipulations for the filled material require a fine content of less than 15%, coal ash has not been considered suitable for this purpose, due to its fine content of 80-90%. Thus, an analysis of the material properties and performance required for coal ash to be used as a filled material, and a clarification of its engineering characteristics, will throw important light on an assessment of its suitability for recycling in this way. This paper reports on the results of laboratory tests undertaken on substitute materials based on coal ash, and also on site tests at a coal ash landfill site (Taki et al. 1999) prone of liquefaction carried out to confirm the implementation characteristics and improvement effect of coal ash materials used as the filled material.
2.1 Material properties required of pile material Figures 1-2 show the required functions of the pile material used in compaction methods of ground improvement such as the SCP, and reference values for the purposes of design and implementation to accomplish those functions (Tsuboi et al. 1993). Figure 1 shows that in the case of clay ground, a composite ground is formed of piles and clay ground surrounding them, and so the function required of the piles is not only pile strength but also drainage for the surrounding ground. On the other hand for sandy ground, the only function required of the filled material is the strength as a pile, and as a countermeasure to liquefaction the improvement effect relies comparatively little on the filled material. Thus, in a case like this, in coal ash landfill ground prone to liquefaction, what is important is whether the material can be discharged through the casing pipe and formed into well compacted piles in the ground. When using a material such as coal ash with a high fine content, problems such as discharge of material from the casing pipe or clogging, or longer cycle times for the implementation (loss of implementation efficiency) can be envisaged. 2.2 Material properties For coal ash to be used as the filled material in compaction pile improvement work in ground liable to liq529
I Pile function I
I
Required of pile material
Required reference value
Clay ground (Mixed ground)
Sandy ground
6 d
: Drainage compression shear strength; Fc: Fine content; k : Coefficient of permeability; DIS:15% grain size; Dx5: 8S% grain size
Fig. 1 Functions of piles and reference values for implementation design
I
II
Properties for pile formation
,
I
Implementation requirements
IInI
Conveyance through casing pile (discharge efficiency) .
I I
Reference values
1
1\1
Fc: Fine content
Fig. 2 Material properties for pile formation and reference values for implementation design sistance. When the degree of compaction exceeds 90%, the angle of shear resistance exceeds 30" for all the specimens, indicating sufficient strength.
uefaction, its implementation characteristics and improvement effect need to be confirmed. Ahead of the site tests, laboratory tests were performed on the following materials to determine their properties.
1 Cleansand 2 Site soil (fine content about 25%) 3 Clinkerash 4 Fly ash 5 Cement mixture cfzy ash with cement added at 30kg/m3) 6 mixture offly and clinker ash (mixing ratio 2 : 1) The results of the tests on these substitute materials are shown in Table 1, and grain size distribution curves are shown in Figure 3. Figure 4 shows the results of drained triaxial compression tests with differFig. 3 Grain size distribution curves for substitute ent compaction degree to confirm strength. In the figmaterials ure the degree of compaction Dc (dry density of the prepared specimen p d max, Table 1 Physical properties of substitute materials dry density in compaction tests pd max) is taken as the density index. Although the data is somewhat scattered, with increasing fine content of the specimens, the lower the degree of compaction, the smaller the angle of shear re- *proportions in ash waste delivered from power stations 530
Fig. 6 Implementation times for substitute materials
Fig. 4 Strength characteristics of substitute materials 3 OUTLINE OF SITE TESTS AND RESULTS
Site tests were carried out to study the suitability of the compaction method using waste soil (site soil) or power station-generatedcoal ash waste, as a countermeasure to liquefaction in coal ash landfill ground. The results are given below: Fig. 7 Relation between fine content and relative implementation times
3.1 Site test outline Test implementation was carried out with two objectives: to confirm the implementation characteristics of such materials as coal ash used as filled material, and to check the improvement effect in coal ash ground.
is possible, taking 1.6 times that with sand. Coal ash (fly ash) is believed to require more time for compaction than sand because its fine content exceeds 80%. The relation between fine content and relative implementation times is shown in Figure 7 (Tsuboi et aI. 1993). The figure indicates that as the fine content of the material increases, implementation efficiencytends to fall.
3.2 Test to confirm implementation characteristics The usual equipment for SCP implementation work was used, with some modifications (Matsuo et al. 1997).Also, a 500mm diameter butterfly point was used at the tip of the casing pipe, as shown in Figure 5, as suitable for discharge of the materials, and also to prevent coal ash flowing back into the pipe during compaction due to the low specific gravity of fly ash compared with clean sand. Implementation characteristicsof various substitute materials were drawn from these site tests and are shown in Figure 6. They confirm that implementation using fly ash
Table 2 Substitute material and improvement ratio specifications r
Improvement ratio (pitch) .-;;i
-'E E
Clean sand Site soil Coal ash (fly ash)
21%
15%
(0 1.2m) (U1.6m) 0
0 -
0 0
10% (U2.0rn)
-
0 -
Fig. 8 Test conditions to c o n f m improvement effect
Fig. 5 Shape of tip of casing pipe
531
.......
Before improvement
-After improvement Pitch: 01.6m
Coal ash Clean sand
.~
L L 0.2
0.0 0.0
,
~
..__ -.__ 0.5
10
15
2.0
25
0.0
0.5
Void ratio, e
1.0
15
2.0
2.5
Void ratio, e
Fig. 9 Changes in void ratio before and after improvement and according to pile material
Fig. 10 Changes in void ratio before and after improvement and according to improvement ratio
Table 3 Comparison of theoretical and measured viod ratio after improvement
I
I
1
Average void ratio after improvement void ratio Improvement ratio (%) Pitch (m) before improvement Theoretical values Measured values Measured decreasein void ratiollheoretical decrease in void ralio 0 1.2 0.73 1 (0.63 1) 0.829 (0.533) 84.5 % 27 0 1.6 1.362 1.008 (0.354) I .I06 (0.256) 72.3 % 15 1.135 (0.227) 1.189 (0.173) 0 2.0 76.2 % 10 ) decrease in void ratio (
I
I
3.3 Tests to confirm improvement eflect
4 CONCLUSION
Tests as detailed in Table 2 and Figure 8 were undertaken with the objective of confirming the improvement effect with different substitute materials, and at different improvement ratios (pitches). As coal ash landfill ground shows the same tendency to liquefaction as loose sandy ground (Taki et al. 1999), the improvement effect was confirmed through an evaluation of the decrease in void ratio, as in sandy ground. Figure 9 shows porosity before and after improvement and for different substitute materials. The figure indicates no significant difference in improvement effect due to the filled material. This is thought to be because well compacted piles of 700mm diameter were formed of all the materials, with no great difference from the degree of compaction of the surrounding ground. Figure 10 shows void ratio before and after improvement and for different improvement ratios (pitch). The figure shows that as the improvement ratio increases, porosity after improvement decreases. When the average measured void ratio before and after improvement was compared with theoretical values, whereas it is known that the void ratio in sandy ground decreases after improvement almost to that of the theoretical value (Table 3), for coal ash landfill ground, the decrease in void ratio was only about 80% of that to the theoretical value, as shown in Table 3. This may be because coal ash landfill ground has a finer content than sandy ground and doesn’t compact so easily.
This paper studies the suitability of substitute filled materials based on coal ash (clinker ash, fly ashwement, fly ashi-clinker ash, etc.) in site tests using the compaction pile method of ground improvement at a coastal site of coal ash landfill ground. Their material properties, and their implementationefficiency and improvement effect in use, were examined. The results confirmed that the compaction pile method can be used as a countermeasure to liquefaction in coal ash landfill ground, that implementation is possible with fly ash and clinker ash, as types of coal ash, used as the pile material in new ground material, and that in design terms they produce the same improvement effect as sand when used as the pile material. REFERENCE Matsuo M., Honjo Y.(eds), [ 19991: New Viewpoints on Soil and Environmental Engineering - Effective Use of Waste Soils, Ch. 2 Types of Construction-Displaced Soils and Their Background, Gihodo Shuppan. (in Japanese) Matsuo M., Kimura M., Nishio R., Ando H., [ 19971: ‘Development of Soft Ground Improvement Method Using Waste Soil’, Civil Engineering Society Papers, 35 (567): 237-248. (in Japanese) Taki E., Hayashi Y., Nakajima H., Suzuki E, Nitao H., [ 19991: ‘PhysicalProperties and Tendency to Liquefaction of Coal Ash Ground’, Civil Engineering Society54th Annual Lecture Series lII-B261: 522-523. (in Japanese) Tsuboi H., Hongo T., Okuda S., MatsumotoJ., Nakasurni I., [ 19931: ‘Effect of Fine Content on Soil Characteristics in Sand Used for Ground Improvement’,Symposium on Ground Discrimination and Engineering Classification, 129-134. (in Japanese)
532
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
CPT investigation to the ground treated by deep mixing method using Flyash Gypsum Cement M.Tanaka & H.Tanaka Port and Harbour Research Institute, Ministry of Transport,Yokosuka,Japan
J.Asano & K.Azuma Electric Power Development Company Limited, Tokyo,Japan
ABSTRACT: The ground was treated by Flyash- Gypsum- Cement (FGC) using Deep Mixing Method, for deep excavation work. The treated ground was investigated by Cone Penetration Test (CPT) as well as conventional laboratory test for specimens collected by sampling. This paper discusses the applicability of CPT, comparing with test results from laboratory test, for measuring the engineering properties of the treated ground. It is found that in combination with laboratory testing, CPT testing is capable of evaluating properly the strength of the treated soil. 1 INTRODUCTION It is conventional that the ground treated by deep mixing (DM) method using cement or lime is evaluated based on uniaxial compression strength (4u). However, 41(values of the treated ground are considerably scattered, and it has been a controversial subject as whether this large variety in 4u value is due to non-uniformity of the treated ground or sample disturbance caused by sampling. Several attempts for employing CPT were made in the past to the investigation for the treated ground. However, it has been usual that the capacity of the penetration is not large enough for CPT to penetrate to a required depth due to large stiffness of the treated ground. The authors had an opportunity to conduct a CPT investigation in the treated ground by DM method using flyash gypsum cement (hereinafter referred to as the FGC-DM method), which was produced relatively lower strength treated soil than conventional DM method dose. This improvement method was applied to excavation work with great depth. This paper presents characteristics of the treated ground by the FGC-DM method, using CPT.
such as a jet column method without mixing blade. However, the cost for treatment of such these methods is considerably higher than that of the DM method. In addition to high cost, a jet column method often deform sheet piles to the outside as a result of its high pressure of injection, when the original ground is very soft (Tanaka et al. 1992). Hence, it has been desired to develop a DM method with moderate strength of the treated ground. It may be possible for the treated ground to get low strength by reduction of amount of cement. However, it inevitably leads to a wide scatter in strength for the treated ground because of reduction in efficiency of mixing due to small amount of cement. The FGC-DM method was developed to solve this problem. By adding fly ash gypsum to cement, the volume of the stabilizer is increased and hence the strength after the improvement becomes more uniform (Asano et al. 1996).
2 FGC-DM METHOD It is in general difficult to drive sheet piles into the ground treated by the conventional DM methods using Portland cement or lime, because of high stiffness of the ground after treatment. Thus, when the DM method is used as a supplemental measure for excavation, the ground is left untreated about 1 m from the sheet pile. This untreated ground, then, is treated by other soil improvement techniques,
Photo 1 Investigation i n progress
533
Fig. 1 Soil Profile
from the present ground surface. Below this fill layer, a clay layer reclaimed by the dredged soil is found. The original sea bottom is located at a depth of about 15 m. In this site, a designed ql, value for the treated ground was set at 400 kPa. The component of stabilizer to get this strength was: 62 kg of cement, 155 kg of flyash gypsum, and 217 kg of water'per 1 m3 to the original ground (Tabei et al. 1999). Fig.2 shows a typical cross-section of the braced excavation. Two layers were treated: the ground surface for reducing the deformation of the sheet pile wall, and the bottom part for increasing passive earth pressure and for preventing the bottom heaving. 4 TEST RESULTS AND DISCUSSION
Fig. 2 Cross-section of the ground improvement
To evaluate strength of the treated ground by the FGC-DM method, laboratory tests as well as CPT were carried out. Samples were collected by using a rotary type of a triple tube sampler for the treated ground and a fixed piston type of a thin wall sampler (hydraulic type) for the untreated ground. To directly compare test results from the laboratory and the CPT investigations, sampling and CPT were carried out at the same treated column. These points were approximately one meter apart from each other. Strictly speaking, both tests should be conducted at the same time since strength and stiffness of the treated soil increase with curing time. However, when a certain time elapse, effect of curing time on strength of the treated soil becomes insignificant. Therefore, these influences in this investigation may be negligible, although the date for conducting the laboratory tests and CPT investigation were somewhat different.
The CPT investigation was conducted at a location for the untreated ground and at 10 locations for the treated ground. A penetration machine with large capacity (the maximum penetration force is 196 kN) was used to penetrate the cone into the ground. Photo 1 shows the vehicle mounting the penetration machine used in this investigation.
3 DESCRIPTION OF TREATED GROUND The site of this investigation is located in a reclaimed area faced on the Tokyo Bay. Fig. 1 shows a soil profile of the ground before the treatment. The land was reclaimed with sand fill to the depth of 3 rn
534
Fig. 3 CPT results of the FGC ground
Fig. 4 Shear strength obtained from CPT, and qu/2
in mind that the increase infs is in particular significant. This fact gives a reason for a large penetration force is required to CPT investigation. It is generally said that high qr values indicate existence of a sandy layer, but in case if the ground improved by the FGC-DM method, both qr and U values become large. This behavior is quite different from that of sandy ground or soft clayey soil ground.
4.1 CPT performance in treated and untreated ground Fig.3 compares the CPT data obtained at typical two locations after and before the treatment. The cone resistance corrected by the effective area (qr), skin friction V;.) and the pore water pressure ( U ) are presented by solid line for the original ground and dotted line for the treated ground in this figure. It is seen from these measurements that the original ground consists of the following layers: one cohesive layer is found at depths between 5 and 8 m and another cohesive soil is between 13 and 23 m from the ground level. A relatively dense sandy layer exists at a depth of 5 m. And an alternating layer of sandy and cohesive soil is found from 8 m to 13 m depth. It is seen in the figure that qf, U andJ. measured at the location B1 increase by as much as 4 to 6 times, 2 to 3 times and 15 to 30 times that before treatment, respectively. The same increases in qr, U andJ. were observed at the location B3. It should be more
4.2 Shear strength The upper part of present ground was reclaimed with gravelly or sandy soils, so it is difficult that to get the samples from the ground. The q, value from this part would be scattered. Therefore, shear strength of the samples which were taken from lower part of the ground will be discussed. This section will compare the shear strength of the treated ground measured by the cone penetration test (s, (CPT)) and UC test, i.e., qJ2. To calculate s, from CPT, the cone factor (NkJ in the equation (1) should be assumed:
535
5 CONCLUSIONS where, s,, (CPT): Undrained shear strength from CPT
ql: Corrected cone resistance C T , , ~ :Overburden pressure in total stress Nkl: Cone factor First, let us consider the shear strength of the original ground, which lies below the original sea bottom. As shown in Fig. 4(a), s, from CPT and q,/2 laboratory test coincide well with each other when Nkr = 14. It is reported that the Nkc varies between 8 to 16 for marine cohesive soils in Japan (Tanaka et al. 1992). The Nkt factor for the present investigation is in this range. Figs. 4(b) and 4(c) show the relationship between s,, (CPT) and q,,/2 for the ground after the treatment. It is found that s, (CPT) using Nkr = 14 significantly overestimate qJ2 of the treated ground. Ifs, (CPT) is calculated using Nkl = 20, then it becomes closer to the q,/2 values measured from the laboratory test at the location B1 as well as B3.
It turns out from this study that CPT investigation is able to evaluate the ground improved by the deep mixing method using flyash gypsum cement. The findings obtained from this study are summarized below: 1. Investigation using CPT for the ground improved by the FGC-DM method reveals a different behavior from that of usual cohesive ground or sandy ground. Response of CPT for the treated ground is characterized by large point resistance, large pore water pressure, and large skin friction. 2. The cone factor (NkJ for the treated ground by the FGC-DM method is greater than that of usual Japanese marine clays. However, it is not clear whether this large Nkr factor is caused by underevaluation of q,/2 due to sample disturbance, or by the inherent characteristics of improved ground. REFERENCES
It may be concluded that Nkl values for the treated soils are much larger than those of ordinary cohesive soils. However, there is a possibility that q,/2 value may under-evaluate the true strength due to sample disturbance caused during the sampling process. In fact, many horizontal cracks were observed in the samples collected by rotating the sampler in this investigation. On the other hands, it is well known that the treated soil presents a smaller strain at failure and smaller residual strength after peak than ordinary cohesive soil; i.e., it is a typical brittle material (Tanaka & Terashi 1986). These different characteristics of the treated soil may be attributed to relatively large Nkt factor. Further studies are required to know which is main factor for large Nkr. One of the most advantages of CPT over conventional laboratory testing may be that testing data are available nearly continuously to depth. For example, let us consider a situation where CPT data are not available but only q,,/2 values are plotted with depth in Fig.4(c). According to the specification stipulating the ground improvement, the treatment of the bottom section is to be started at a depth of 13 m and ended at a depth of 20 m. However, the q,,/2 values considerably varies in this section and it is not clear whether this variation is due to errors in testing including the sample disturbance, or mishandle of the treatment work. However, with help of CPT data, it is able to correctly evaluate laboratory test data test.
Asano, J., Ban, K., Azuma, K. & Takahashi, K. 1996. Deep Mixing Method of soil stabilization using coal ash. Proc. of IS-Tokyo'96, The 2ndInternational Conference on Ground Improvement Geosystems: 393-398: Rotterdam: Balkema. Tabei, F., Mouri, T. & Enomoto, T. 1999. Deep Mixing Method using flyash gypsum cement for Construction of the Isogo Thermal Power Plant. Electric Power Civil Engineering No. 279: 57-6 1. (in Japanese). Tanaka, M. & Tanaka, H. 1992. Effect oftjet grout method on retaining walls. Proc. of 2 7 Annual Meeting of JGS : 1753-1754. (in Japanese). Tanaka, H., Sakakibara, M., Goto, K., Suzuki, K. & Fukazawa, T. 1992. Properties of Japanese normally consolidated marine clays obtained from static piezocone penetration test. Report of Port and Harbour Research Institute, Vol. 31, No. 4: 6 1-91. (in Japanese). Tanaka, H . & Terashi, M., 1986. Properties of treated soils formed insitu by Deep Mixing Method. Report of Port and Harbour Research Institute, Vol. 25, No. 2: 90- 119. (in Japanese).
536
Coastal GeotechnicalEngineering in Practice,Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 7
Use of stone column for improvement of very soft clay in the marine swarnp area Wanchai Teparaksa Center of Soft Ground Engineering, Department of Civil Engineering, Chulalongkorn University,Bangkok, Thailand
ABSTRACT: The application of stone column technique was firstly adapted to improve the very soft clay in the marine swamp for constructing flood protection dyke. This dyke is about 2.5-3.0 m high and 11 km. long constructed on the 14-18 m. thick soft clay aims to protect the rising up of the seawater and create additional flood problem to Bangkok City in the rainy season. The construction was started by dumping the crush stone to prepare the platform for truck hauling and the heavy stone column equipment of about 40 tons weight. The crush stone backfill was sunk and laterally flow into the very soft clay to about 10m depth. After completion of platform, the stone column soil improvement technique was started. The original 18 m. length of the stone column was not successful due to its collapse in the very soft clay beneath the platform. Therefore, the length of stone column (1 .Om in diameter) was changed to 10 m. and installed inside only its own platform. The platform of about 10m depth and the dyke embankment were acted as a surcharge to induced a huge of long-term settlement in the future. Due to the limitation of construction technique, the stone column seems to be not suitable to improve the very soft clay in the marine swamp area.
1 INTRODUCTION
behavior (Xanthakos et al, 1994, Koemer, 1985 and NAVFAC DM-7.3, 1983). Densification technique is generally suitable for cohesionless soils or granular materials by improving the bearing capacity of soils such as Dynamic Compaction Method, Vibroflotation or Vibrocompaction Method, Sand Compaction Pile and Stone Column technique. Consolidation behavior aims to accelerate the settlement of soft cohesive soils such as Preloading with and without PVD technique. The other soil improvement technique is to from the pile foundation such as the Cement Column technique, Jet Grouting technique, and Lime Column.
Soil improvement has been widely used in Thailand, mostly to improve the soft Bangkok Clay. The general soil condition of Bangkok area consists of 12-15 in. thick soft to medium dark gray clay. The water contents is about 70-90%. This soft clay is very sensitive and has anisotropic and time dependent undrained stress-strain-strength properties and is a creep susceptible materials. The prefabricated vertical drain (PVD) technique has been used to improve the soft clay of the New Bangkok International Airport as well as the outer ring road connecting to the airport. Recently, cement column is also widely used to stabilize the soft ground as a road foundation. The application of stone column is rarely used in Bangkok especially for soft clay due to limitation of construction technique. This paper presents the possibility of using the stone column to improve the very soft ground in the marine swamp area. The construction technique is discussed concerning the lateral soil movement. The FEM analysis based on the field performance is also presented.
3. PROJECT DESCRIPTION AND SOIL CONDITIONS In rainy season, flood problem is very serious for Bangkok city, a capital of Thailand, due to a huge of water be push from the Northern and Northeast part of Thailand to Bangkok area as a lowest basin and as an outlet delivered to the sea. In rainy season, around September to October month, normally the sea water level is also very high and causes difficulty in pushing the entire flood water to the sea. All floodwater, therefore, has to be pumped out to the sea by using the temporary dyke or existing road along the seacoast. The research project is the construction of the permanent flood protection dyke along the sea coast
2. SOIL IMPROVEMENT TECHNIQUES The soil improvement techniques were divided according to the performance and-soil behavior into 2 groups as Densification behavior and Consolidation 537
in the marine swamp area parallel to the Sukhumvit road (temporary dyke for flood protection) in the eastern zone of Bangkok city so call Samutprakarn Province as shown in Figure 1. This flood protection dyke aims to protect the rising up of the seawater and create additional flood problem as well as accumulated to normal rainfall flood into Bangkok City in the rainy season. The flood protection dyke of about -2.5-3.0 m. high and 11 km. long was constructed in the marine swamp area on the worst Bangkok very soft dark gray clay. The stone column technique was adapted to improve the very soft marine swamp to act as the foundation of the dyke embankment. The dyke was about 4m. width having the side slope of 1:1.5. The stone column was initially designed as a pile foundation to support the embankment by penetrated into the stiff silty clay layer at about 18 m. depth below the ground surface in order to minimize the long-term settlement. The stone column was design to use 4 stone columns for each row with longitudinal spacing of 2.50 m. and diameter about 1.0 m. as shown in Figure 2. The soil condition consists of 14-17 m. thick very soft dark gray clay. The water content is over than 100% and the shear strength is about 8 kN/m2. This soft clay is very sensitive and has anisotropic and time dependent undrained srtess-strain-strength properties and is a creep susceptible materials. The medium stiff gray clay about 4-6 m. thick and very
Figure 3 Soil profile along the route stiff silty clay is found below the soft clay. Figure 3 shows the soil profile carried out during stage of feasibility study along the Sukhumvit road. The general and engineering properties of soils was presented in Table 1. Figure 1 Project location 538
Table 1 Engineering Soil Properties Depth (m)
Soil Type
Undrained shear strength
Water Contents (%)
(kN/m3)
Sensitivity
yt
(kN/m*) 0.0-
Verysoft
15.0 15 0-
18 0 18 0-24 >24
Medium Clay Stiff sllb' Clay
8
> 100
14.0-54.3
7-8
25-35
60-70
16-16 5
5-6
130
30-35
1"Sand
19 20
4. METHOD OF CONSTRUCTION The flood protection dyke was proposed to construct in the swamp area on the very soft clay by constructing the embankment of 2.5-3.0 m. high on the very soft ground that improved by means of stone column. For constructing the stone column in soft ground, the platform is really necessary for supporting the stone column equipment of about 40tons weight. For constructing the platform, the hydraulic fill method was selected. In order to create a platform, a huge of crush stone was dump into the very soft clay. The crush stone was mixed and fastly sunk into the very soft ground. The dump crush stone was stable when its own stress distribution was equal to the undrained shear strength of the very soft clay.
Figure 4 Results of FEM analysis for ideal case
Figure 5 Results of FEM analysis for case of embankment with platform According to the record of used crush stone backfill volume, the soft ground was replaced by crush stone up to about 10 m. depth (Teparaksa 1999, and Pansereewong, 1996). After completion of the platform, the stone column was carried out on this platform by penetrated the stone column through the platform and very soft clay until reaching the stiff silty clay layer at about 18 m. depth. The finite element (FEM) analysis was carried out to study the soil behavior and plastic flow of soil before and after completion of platform with out stone column by using 2D program namely "PLAXIS" (BrinkgrevebtBrand, 1996). The FEM analysis was based on the Mohr Coulomb or bi-linear soil modeling by using the soil stiffness of very soft clay according to the result of self boring pressuremeter test in Bangkok clay (Teparaksa, 1999) of Eu = 100 Su. The FEM analysis for case of non-soil improvement without platform could not be completed, because the soft soil was flow and absolutely failed after loading the embankment. Figure 4 presents only the schematic of the mesh deformation and soil movement of the ideal case but not the actual results of the analysis. In case of constructed platform by means of dumping the crush stone into the soft soil and form the bulb of crush stone in the ground about 10m depth with out stone column, the result of FEM analysis was presented in Figure 5. Figure 6 shows the result of the FEM analysis for soil movement after installation of the stone column inside its own platform. After. comparing between 539
case of with and without stone column (Figure 5b and 6), it is clear that the stone column did not improve any stability of the dyke. It also can be seen that the soft clay was push and heaved above the water surface in the wide area up to about 40 m. away from center of the dyke. This evident was also proved by the photograph during construction on the landside and seaside as shown in Figure 7a and 7b
Figure 8 Stone column installation
Figure 6 Result of FEM analysis for case of embankment with platform and stone column
Figure 9 Platform during stone column installation
Figure 7(a) Platform (land side)
Figure 10 Heaving soils are removed after completion of works After completion of the platform the stone column was started by using vibro-replacement technique with the vibrator of about 300 mm. in diameter by vibrating and rotating and filling the small size of crush rock through the holes of the vibrating rod as shown in Figure 8.
Figure 7(b) Platform (sea side)
540
The initial stone column of 18 m. long was not successful due to the crush rock be collapsed after penetration through the platform and spread out in the very soft clay. This evidence can be explained by common basic soil mechanics that the very soft soil is easily failed under heavy loading with very low confined stress. After trial and failed, the length of the stone column was changed from 18 m. to be only 10 in. long. This means that the stone column was performed only inside the area of filled platform and did not improved any soft clay beneath the platform. The diameter of the stone column was about 1.0 m. which was estimated according to the used volume of the stone per one column. This means that it was not necessary to use the stone column technique to stabilize the very thick soft clay. Figure 9 shows the photograph during performing the stone column and it can be seen that the filled platform is already strong enough to carry the bearing stress from the heavy equipment and the truck. The embankment of about 2.5-3.0 m. height was later constructed on this platform to acting as the flood protection dyke. After completion of the dyke, some heaving soft soil materials were removed to give a good view of the final dyke as shown in Figure 10.
dark gray clay. The water content is over than 100% and the shear strength is about 8 kN/m'. During preparing the platform for constructing the stone column, a huge of crush stone was dump and sunk into the very soft clay. The backfill platform was sunk to about 10m. depth and laterally flow out in the wide area. The original length (18 m) of the stone column was not successful due to its collapse in the very soft clay beneath the platform at about 10 m. depth. The stone column, therefore, finally carried out only 10 m. long inside its own platform area. The long-term settlement was estimated about 860 mm. due to the embankment and excessive surcharge of platform. Because of the very thick soft ground conditions, the stone column technique seems to be not suitable for improving the very soft clay in the marine swamp area. ACKNOWLEDGEMENT The author express their appreciation to Mr. Narong Thasnanipan and Mr. Pornpot Tanseng of Seafco Co.Ltd for their assistances in the preparation of this paper.
5. LONG TERM BEHAVIORS OF DYKE
REFERENCES
The stone column of about 1.0 m. in diameter was carried out on its own platform without improves any soft ground. The crush stone of the platform will be formed a surcharge to the very soft ground and will induce the long-term settlement. Unfortunately that only one set of instrumentation consists of settlement plate and inclinometer was failed during the construction of the platform. The estimating of the long-term settlement based on 1D consolidation theory was about 860 mm. (Teparaksa, 1998). This long-term settlement is very high and the height of the dyke has to be risen up in the future. The study on the long-term settlement of this flood protection dyke constructed on the very thick soft clay using the stone column is very interesting. Unfortunately that no any instrumentation was provided to measure the long-term settlement of the dyke as well as the performance of the stone column and the backfill platform.
Brinkgreve, R. and Brand, P.A.( 1996): Application of PLAXIS for soil and rock plasticity, Short Course on Numerical Analysis in Geotechnical Engineering, AIT. Koerner R.M. ( 1985): Construction and Geotechnical Methods in Foundation Engineering, Mc.Graw Hill Book Company. NAVFAC DM-7.3 ( 1983): Soil Dynamics, Deep Stabilization, and Special Geoteclinical Construction, Design Manual 7.3, Dept. of the NAVY. Pansereewong, W.( 1996): Study of Stone Column for Improvement the Stability And Settlement in Soft Ground, Master Thesis, Chulalongkorn University ( In Thai). Teparaksa W.( 1998): Possibility in using stone column for soft soil improvement, Seminar on Ground Improvement, EIT, pp. 109-126 ( in Thai). Teparaksa, W. (1 999): Principle and application of instrumentation for the first MRTA subway project in Bangkok, 5th International Symposium on Field Measurements in Geomechanics, Singapore, November. Xathakos, PP, Abrainson, LW, and Bruce, D.A.( 1994): Ground Control and Improvement, John Wiley & Sons Inc.
6. CONCLUSIONS The stone column technique was firstly adapted to improve the very soft clay in the marine swamp for constructing flood protection dyke. This dyke is 2.53.0 m high and 11 km. long aims to protect the rising up of the seawater into Bangkok area and create much more flood problem in the rainy season. The soil condition consists of 14-17 m. thick very soft 541
This Page Intentionally Left Blank
Coastal Geotechnica/~n~ineering in Practice, Nakase & Tsuchi~a(eds)02000 ~ ~ / k e~ ~ ~a , ~ ISBN e f 90~5809 a 151 ~ 1,
New design method of short vertical drains to improve a soft clayey gound in the Mekong Delta Le Ba Vinh & Goro Irnai Department of Civil Engineering, Yokohama National University,Japan
ABSTRACT: This paper presents a new method to design vertical drains. Their length has long been determined based on the depth of consolidation zone within which the effectiveness of consolidation process can be expected. An attempt to minimize the length of vertical drains has been made. The vertical drain of varied lengths across the section of an embankment has been proposed. Based on a relationship between the water content and compactness of soft clay during the consolidation process, a new method to predict the settlement, consolidation time and load-bearing capacity relationship of soft clayey ground underneath the embankrnent has been introduced in this paper. where, y = unit weight of soft soil, c = cohesion of soft soil, # = internal friction angle of soft soil. c and # are determined in total stress.
1 INTRODUCTION
In the South Vietnam, Mekong Delta is one of areas with very thick layer of soft soil (Brand and Premchitt 1989), in which areas the method of 2. For load djstributing by an isosceles triangle: vertical drains are widely used to improve the 2c. cos q5 + yb. sin d, surface zone underneath cbnstructions, 1; that case 4s the vertical drains cannot reach the firm base, and a0 the total system of the construction and the 'One 'Oats On the s u ~ o ~ d i n gwhere, b = half width of loaded area, the values of soft soil. It is required, therefore, to analyze the a, are determined as shown in Table stability of the system in combination with a When dealing with the Mekong Delta clays, it is Process Of the 'One to be required to &eat them as (c, #) material, even though This paper presents a new design approach to this they are fully saturated. Problem 'Pecial natures Of the Mekong When a method for consolidation acceleration is Delta in mind. used in the step loading fill, for example, the vacuum preloading method with vertical drains, the 2 VALUE OF FILL P ~ L O A FOR~ ~ values ~ of c and # in equations (1) and (2) should be replaced by new value increased by consolidation. VERTICAL DRAINS In order to prevent an unimproved soft ground from failure, a fill preloading process must be done by step loading. The plastic zone caused by fill preloading should be generated at only one point in the ground (VN MOC 1997). Therefore, the value of allowable fill preloading for vertical drains can be determined as follows: 3.. For uniform load: 4s =
xc cot fp
);
3 CALCULATION OF VERTICAL DRAIN^ (Dw)
THE
DEPTH
OF
Based on the results of a research (Lareal et al. 1989) into drainage of soft ground, consolidation Table 1. Value of a,.
(1)
cot 0) 4- (0) -
543
Figure 2. Determination of initial hydraulic gradient, io. Figure 3. Relationship between initial hydraulic gradient and liquidity index at various plasticity index of soil. 1- Clay No 1, Ip = 33.8, 2- Clay No 2, Ip = 29.9, 3- Clay No 8, Ip = 25.1,4- Clay No 7, Ip = 18.2, 5- Clay No 4, Ip = 12.4, 6- Sandy Clay No 6, Ip = 7.2 (Khanh et a1.1995).
due to embankment solidifies the soft ground down to a definite depth. Therefore, the bottom of vertical drains to be installed should reach that definite depth. The basic concept to determine that limit depth of drainage is as follow: i 2 io
(31
where, i = hydraulic gradient caused by embankment load, io = initial hydraulic gradient of soft clayey soil. Scheme of the conventional experiment to determine the initial hydraulic gradient is shown in figure 2. At the stable state of water level with no water flow through the soil specimen Z, we get:
(4) where, ~0 = io.yw.z
(5)
Therefore, the initial hydraulic gradient is defined as follows:
Experimental results on the initial hydraulic gradient of Mekong Delta soft clay (Khanh et al.1995) are shown in figure 3. In order to minimize the length of vertical drains, varied lengths along the cross section of an embankment has been proposed. Figure 4 shows an example of real embankment on soft ground. Based on the fact that the embankment load causes its effect only within a finite area of soft ground, the
544
Layer No2: Road base: sandy clay, y2 = 18.6 KN/m3. Layer No4: Soft clayey ground. y4 = 15.8 KN/m3.
Figure 4. Cross section of an embankment on soft ground.
2
12
22
32
42 Stress
0 5
Calculated stress due to
10 15
20 Depth (m>
U
I
phe location of vertical drains (From center axis of embankment)
Figure 5. The determination of vertical drains’ length at various locations underneath the embankment. length of vertical drains should be a finite one, which may be as equal as the depth of the influential areas due to embankment. It is here determined by the following equation: 0, =
...
...
...
...
...
.. .
. ..
. ..
. ..
._
...
.._
0.2Yz
where CT, = total stress in the soft ground due to the embankment load, Y = unit weight of soft ground, z = depth from the bottom of embankment. Figure 5 shows a graph to determine the length of vertical drains at various locations underneath an embankment. The bottom of vertical drains are determined by intersection points between the straight line of 0.2 Yz and the stress curves oZ z calculated at various axes underneath the embankment.
Figure 6. The determination of vertical drains’ length: 1- Vertical drains, 2- An approximate boundary of the bottom of vertical drains, 3- The calculated boundary of the bottom of vertical drains.
-
545
4 STABILITY ANALYSIS OF A TOTAL SYSTEM OF EMBANKMENT AND SOFT GROUND I ~ P R O V E DBY VERTICAL DRAINS In the case of soft ground, modes of potential failure include general failure and local failure as shown in the following analyses: 4.1 Slope stability analysis
Fellenius’ method (Leonard 1962) has been widely applied in Vietnam to analyze an embankment’ s slope on soft ground improved by a preloading process. In the analysis, the shape of a failure surface has been assumed to be an arc. A computer program named SOIL-SAP for automatically analyzing slope stability was made by Khanh (1998). The input data used in the slope stability analysis is shown in table 2. The cohesion and internal frictional angle of soft ground have been increased due to water content’ s decreasing after the preloading process as shown in figure 7. The analysis results have presented the ten most dangerous slip arcs and the safety factor as shown in figure 8.
Table 2. Input data used in the slope stability analysis Name ThickY C 6 ness(m) (kNim’) (kN/m2) Coating: 1.o 21.2 22.4 17.6 sandy clay Road base: 1.o 18.6 20.0 15.2 sandy clay Sand mat I .o 17.2 0.0 28.0 Soft ground 20.0 15.8 6.0 5.0 4.2 S t a b ~ l analysis i~ of soft ground unde~neat~ the embankment. Load-bearing capacity of soft ground improved by vertical drains is checked by the use of the following equation: R = F.q
where, F is safety factor. It must be grater than 1.5; q is design load of a construction acting on the ground at the bottom of the foundation. R is allowable loadbearing capacity of the ground at the end of preloading process of soft ground (VN MOC 1997):
Figure 7. Change of water content, characteristicsand settlement of soil with time during two stages of fill preloading.
Figure 8. Slope stability analysis. Safety factor by Fellenius’ s method.
546
(8)
R - m1*m2 [Aby + Bhy’+ D . c ~ , ~y’h] K
(9)
where ml and m2 are the coefficients of working conditions of ground and construction, respectively. K is a reliable coefficient depending on the exactness of soil experiments. A, B and D are parameters depending on the angle of internal fkiction of the soil at the end of a preloading process, +f,J: A = f (+f,2>; B = f (4+,2);D = f (4+,2);Y and Y’are the unit weight of soil under and above the foundation bottom, respectively. The reinforcement effect of drain material is taken into account by using the unit weight, y, cohesion, c, and internal friction angle, +, which have been increased due to both drain material and water content’ s decreasing after the preloading process.
beginning and the end of loading step number i, respectively, A = density of soil, D = depth of the active zone of consolidation in the ground defined in Figure 6. Settlement which takes place during an operating process of construction project can be predicted as follows:
so=-
wi2- w
1 A
.D
(13)
- +w,2
where So = settlement of soft ground during construction project, wi2 and wfare water contents at the beginning and end of the construction project, respectively.
6 THE SETTLEMENT TIME OF SOFT GROUND To calculate required times, t~ and th, for the same degree of consolidation of two soil layers having different depths, H and h, respectively, Maslov (196 1) proposed the following equation:
5 SETTLEMENT CALCULATION OF SOFT GROUND IMPROVED BY VERTICAL DRAINS The settlement of a soft ground to be improved by preloading with vertical drains can be divided into the two, one of which preloading and the other takes place by construction. As mentioned above, preloading should be usually done by many steps of fill loading. After each step of loading, the soft ground consolidates resulting in water content decrease of soft soiI followed by settlement as shown in figure 7. The settlement of a soil layer having depth H i s as follows:
Based on this equation, the following equation can be here proposed to predict a process of consolidation by vertical drains. tf
=,(yj”
where
time required for a test specimen to its primary consolidation, L = m:$;spacing between vertical drains), r = radius of vertical drain (figure l), h = height of specimen, n = consolidation parameter of soil, and can be determined by the oedometer test. If the Terzaghi’ s theory can be completely applied to Mekong Delta clay, n should be equal to 2. We cannot determine the value of n by the use of equation (14) because we couldn’t get filed data, therefore, the following equation is here proposed to determine n-value in laboratory. Based on the equation (141, we get:
if S, = loo%, we get:
[
e = A.w; A = -):o;1:
From the two equations (10) and (1 l), the final settlement of soft ground in each loading step can be determined as follows based on the water content change between the beginning and the end of each loading step:
- + Wil 6
where Si= settlement of soft ground after loading step number i, wil and wi2are water contents at the 547
th =
where, tl and t 2 are required times for the same degree of consolidation of two soil specimens having different height, h, and h2, respectively. Experimental results on the consolidation parameter of Mekong Delta soft clay is shown in figure 9.
7 SUMMARY A new method to design short vertical drains to improve a soft clayey ground is proposed based on the conventional methods which has been widely practiced for the Mekong Delta soft clays, new concepts being added to the conventional one. Main differences between the Mekong Delta soft clayey ground and alluvial clay layers in Japan are as follows: 1) To deep soft clay layer in the Mekong Delta: Full improvement by vertical drain could not be done. An area improved by short vertical drains floats on the surrounding soft ground. 2) Mekong Delta soft clay has apparent cohesion c and friction angle 4 in the term of total stress. It is (c, 4) material, not c material (4 = 0). Based on the knowledge above stated, several proposals for designing a floating ground solidified by short vertical drains are presented in this paper. 1. In the condition of very thick and soft ground, the designing of short vertical drains can be done based on the determining the depth of active zone of consolidation. 2. A new setting of vertical drains’ length that decreases from the center axis to the edge of embankment has been proposed based on the equation (7) not only for this project but also the general. 3. Both the slope stability and load bearing capacity analysis has been carried out considering the cohesion and internal fiictional angle of soft ground have been increased due to water content’ s decreasing. Also, the settlement and settlement’s time can be predicted as equation (12), (13) and (15) from the experimental studies on water content of soil during the consolidation process. ACKNOWLEDGMENT: The author would like to thank Prof. Kazuo Tani for his helphl comments on this paper. REFERENCES Brand, E. W. & Premchitt, J. 1989. Comparison of the predicted and observed performance of Muar test embankment on Malaysian Clays. Proc. Symp. on trial embankments on Malaysian
Figure 9. Relationship between consolidation parameter, n, and plasticity index, Ip, at various state of soil (Khanh et al. 1995). Table 3. Consolidation parameter, n. n Ip
27.7 21.0 16.0 6.5
I L = ~ .I~=0.88 I~=0.62 I~=0.37 I~=0.12 2.00 2.00 2.00 2.00
1.83 1.80 1.72 1.64
1.54 1.36 1.18 0.94
1.24 0.68 0.30 0.00
0.56 0.00 0.00 0.00
Marine clays, Kuala Lumpur, Vol. 2: 10-1/8. Khanh, L.B., Ly T.V. & Vinh, L.B. 1995. An investigation on soft clayey ground in the Mekong Delta. Hochiminh City University of Technology (HUT). Khanh, L.B. 1998. Stability of road base on soft ground during the consolidation process. Proc. Conf. on New Technology in Construction, HoChiMinh City, 19-21 February, 1998: 49-52. Lareal, P., Nguyen Thanh Long & Le Ba Luong 1989. Remblais routiers sur sols compressibles dans les conditions du Vietnam: 24-169, Paris, France. Leonard, D.A. 1962. Foundation and Fundamental: 81-1 18, New York, U.S.A. Maslov, N.N. 1961. Sur le problem de la resistance au cisaillement des sols argileux plastiques a consolidation incomplete. Proc. 5‘h Int. Conf, on Soil Mechanics and Foundation Eng., Paris, 1722 July, 1961: 243-248. Vietnam Ministry of Construction (VN MOC) 1997. Specification for design of traffic construction. Proceeding of Vietnam Construction Standards, Hanoi: TCVN No 4054-5729. 548
Coastal Geotechnical EngineeriRgin Practice, Nakase & Tsuchida(eds) 02000 Balkema, Rotterdam, ISBN 90 5809 751 1
Bo Myint Win & R.Bawajee SPECS Consultants Pte Limited, Singapore
VChoa ~ a n ~ ~a ~~~ g~ n o ~l on i~~ie ~r ~sa iil~~, ~ u p o r ~
ABSTRACT: Implementation of mega soil improvement works required systematic approach and planning. Firstly, design concept for the various areas, which required various loading conditions and various allowable duration of soil improvement is necessary to be correctly adopted. During the course of implementation, determination of penetration length, selection of equipment, material and quality control of material are important. Due to the different ground conditions with various thickness of soil profile, different types of vertical drain rigs with various heights and various principles of penetrating mechanism were required. Selection of mandrel and anchor is equally important since its control the degree of disturbance to the ground. Finally systematic planning of i n s t ~ e n t a t i o nand monitoring, method of assessing degree of consolidation are essential. This paper described procedure and process of implementation of soil improvement works in Changi East Reclamation projects in Singapore. 1 INTRODUCTION Changi East Rec~amation Projects include 2000 hectares of reclamation. The reclamation is being carried out under four phases since vast quantities of fill material and prefabricated vertical drains required. As the reclamation covers quite a large area extent, the soil profiles and characteristic of soils are different from one area to another. In addition to that the fbture loading and timing of land usage are different within the area. Therefore design of soil improvement works were considered accordingly based on existing soil profile, future land use and allowable duration for soil improvement works. In addition to that acceptance criteria of the soil ~mprovementworks were also varied since design criteria and future land use were varied. The prefabricated vertical drains were designed to install with various spacings in order to accelerate the consolidation of soft compressible underlying soil. Due to the thick layer of soft clay and large area required for soil improvement works, a total of 140 over million meters length of vertical drains are being installed. As such, efficient system of implementation is required for such a large Soil improvement projects. The process of implementation includes selection of materials, quality control, selection of installation rigs and accessories and monitoring of improvement together with performance assessment. This paper described the implementation of soil improvement works at Changi East Reclamation prqjects.
Figure 1. Site location and reclamation Area. 2 PROJECTA~A Project area is located eastern part of Singapore (Fig.1). The reclamation is being carried out foreshore of eastern coastal Iine of Singapore for various future land usages. Majority of area is being reclaimed for expansion of future airport and the remaining areas are for industrial and other usage. As such various end-users called for various requirement and allowed for various durations of implementation. In addition to that nature of underlying soils are varied through out the area. Therefore implementation of soil improvement works is necessary to plan to suit the natural soil conditions as well as end-user’s requirements.
549
Figure 2. Geological profile at the project site along south to north. Generally soft marine clay layer of thickness varying between 10 to 50 metres underlies the majority of the area. This quaternary deposit of soft clay layer is locally known as Kallang formation and which is underlain by old alluvial formation of medium to dense silty clay. Marine clay was deposited during the rise and fall of sea level. As such the desiccated layer of lower marine clay was found in most of the area as 2 to 3 meters intermediate thin layer in between upper and lower marine clays. Sometimes intermediate sand layer could be found instead of desiccated lower marine clay. The parameters of upper, lower and intermediate clay layers are shown in the Table 1 and details characteristic of Singapore Marine Clay could be found in Bo Myint Win et a1 (1 998a ). In general thick layer of Marine Clay could be found at the northern part of the area and the thickness could be as thick as 50 metres and the profile is found to be uniform. However at the southern part of the area thickness of soft clay layer was varying between 10 to 30 metres thickness and Lower Marine Clay was deposited in the deep and steep valley cut. This makes the significant variation of total thickness of clay layers. In some locations at Southern part, Upper Marine clay was missing and Lower Marine Clay was found as overconsolidated clay. At the Southern most area, soft marine clay deposit was rarely found and localized pocket of Marine Clay was found as overconsolidated clay. The soil profile along North - South line is shown in Fig. 2. In addition to the natural soft clay
Table 1. Range of Physical and consolidation characteristics of Singapore Marine Clay at Changi. UMC
ISC
18.6414.23 19.6 15.7 10-35 70-88 50 80-95 18-20 20-28 0.7-0.9 1.8-2.2 2.68-2.76 2.6-2.72 0.2-0.3 0.6-1.5 0.012 0.0043 0.023 0.025 0.0076 0.008 0.042 0.1 15 0.09-0.16 0.05-0.15 0.47-0.6 1-4.5 10-30 3-7 5-10 2-3 1.5-2.5 3-4
-
OCR
LMC
SPS
15.716.67 40-60 65-90 20-30 1.1-1.5 2.7-2.75 0.6-1 .O 0.012 - 0.023
12.26 15.7 140 - 180 6 0 - 115 22 45 2 4.5 2.68 0.5 - 1.7
0.012 0.0383 0.14-0.2 0.8-1.5 4-10 3-5 2
-
-
-
0.3 - 1.0
0.2 - 0.75
* UMC = Upper Marine Clay, ISC = Intermediate Stiff Clay LMC = Lower Marine Clay, C, = Recompression Index SPS = Siltpond Slurry, C, = C, for Recompression
formation, there was a big slurry pond inside the reclamation area which was earlier formed to contain the mine tailing waste from the sand quarry located 10 Km from the present location. The fine materials from sand quarry were transported to the foreshore with high water content and contained in the bunded
550
Table 2. Comparison on design of soil improvement works with pre-fabricated vertical drains. Area Year of Thickness Type Future Design of clay (m) of clay Land use Marine Clay Runway A 1992 20 - 40
Design Design Surcharge Specified Acceptance Criteria Spacing (sq:) Surcharge El. period 1.5m + 10 mCD 18 Months 90% degree of consolidation equivalent to fill & surcharge load
B
1992
10 - 35
Marine Clay Taxiway
C
1995
20 - 30
Marine Clay Infrastructure Area 1.8m
D
1995
30 - 45
Marine Clay Infrastructure Area 1.8m
E
1995
30 - 45
Marine Clay Others
1.8m
F G
1995
20 - 40
Marine Clay Roads
1.5m
1998
40
Marine Clay Future Material
1.5m
+ 8.5 mCD + 8.5 mCD + 9.5 mCD + 8.5 mCD + 8.5 mCD + 12 mCD
H
1998
40
Stockpile Area Marine Clay Infrastructure
1.8m
+ 10 mCD
1.8m
24 Months 24 Months 12 Months 12 Months
- do - do - do - do - do - do - do -
18 Months 90% degree of consolidation to finished level to + 5.5 mCD + Future load 20 kPa 12 Months - do -
I
I998
40
Marine Clay Infrastructure
1.5m
J
1998
10 - 35
Marine Clay Infrastructure
1.8m
+ 10 mCD + 9 mCD
18 Months
K
1992
10 - 20
Soft Slurry
2.0m
+ 9mCD
36Months 90% degree of
Infrastructure
3 passes
area. The thickness of slurry in the pond varying between 1 to 20 meters and water content of slurry is varying between 140 to 180 %. The elevation of top of slurry is about -4 mCD. The details of this slurry pond could be found in Bo Myint Win et al. (1998b) and parameters of slurry-like soil is also shown in Table 1. As such this slurry-like soil would produce significant large strain and takes longer time to consolidate. Therefore special effort were put to implement the systematic soil improvement works for such soil.
3
18 Months 24 Months
- do consolidation equivalent to fill & surcharge load
were known, surcharge load equivalent to future loading were applied. Vertical drain spacings were designed based on thickness, parameter of underlying clay and duration allowed for surcharge period. Details of design for each area are shown in Table 2. 4 DETERMINATION OF VERTICAL DRAIN LENGTH Vertical drains are required to penetrate throughout the compressible clay layer. On the other hand the installation of vertical drain to the clay layer of less than of equal to 2 meters thickness would be wasted since it could be improved by preloading alone. Therefore the area has to be profiled in order to determine the penetration length of vertical drain. The preliminary profiling was carried out with seismic reflection survey followed by confirmatory boreholes and preliminary estimation of vertical drain length was carried out. During the course of the installation, panels were sub-divided into 50 metres by 100 metres rectangular panels and one each reference Cone Penetration Test (CPT) was carried out to determine the penetration length. Some additional CPTs were carried out whenever discrepancy between actual penetration length and reference CPT was encountered in the certain panel.
BASIC DESIGN CONCEPT
Generally prefabricated vertical drains were designed to install to improve the drainage system of compressible soil. Spacings of vertical drain were selected to achieve the 90% degree of consolidation with fill and surcharge load after taken into consideration of submergence effect caused by sinking of fill below groundwater level within surcharge period. For such cases, at least minimum preloading pressure of equivalent to two to three meters thickness of sand (35 to 50 kPa) were achieved. Some locations where no special treatment were required for future load, soil improvement works were carried out to achieve 90 % degree of consolidation due to fill load plus general future load of 20 kPa. At certain locations where future loads
551
Table 3. Specification of prefabricated vertical drain ( Requirement versus Supplied materials). Property
Unit CORE
Material
Dimension ofdrain
Specified Requirements Continous plastic drain core
Colbond MD 7007 MD 7007 MD 7007 Flexi CX 1000 Holland Korea Malaysia FD 767 Polyester Corrugated Corrugated Corrugated Corrugated
wrapped in non- woven
Filament 38 groves 40 groves 37 groves
39 groves
100 f 2 3 to 4
Polyester Nonwoven 100 5
100h2 3.5 f 0.2
FILTER
geotextile material.
Width mm Thickness mm
PP PP PP PP non-woven non-woven non-woven non-woven 103 3.1
103 3.4
Darcy Permeability
m/s
>5 x 1O4
15x 104 1 x 104
1 x 104
1.02~ 104* 1 x 104
Discharge capacity of drain
m3/s
>25 x 10" At 350 kN/m soil pressure after 4 weeks
90
95
100
69*
52
Discharge capacity of drain under deformation
m3/s
>10 x 10" At 25% relative compression
80
71
77
45*
45
Soil retention capacity
Microns
AOS Og5 <75
<75
73
73
40*
75
kg/lOcm width kg/lOcm width
>I00
Atelongation 210 Min-2%, Max- 10 % Atelongation 232* Min-2%, Max- 10%
220
250
147*
294*
212*
275*
167*
3 10*
Tested at 1% strairdmin after saturation in H 2 0 at 10°C for 48 hrs
11
40.8
40.8
89*
80*
1 1*
49.7*
68.4
97*
95*
At 100kg/lOcm width At3 kg/cm
6*
3%at 50 kg/cm 4*
2*
4*
4*
1.4*
Tensile strength ofentire drain
Dry
Tensile strength of filter
Dry
kg/cm
>3
Wet
kg/cm
>3
Wet
Elongation Dry YO of entire % drain Wet * Tested at third party laboratory
>100
4 0 <10
5 SELECTION OF TYPES OF VERTICAL DRAIN Selection of vertical drain was solely based on the specification requirement. However now a day more or less all types of vertical drains are being manufactured to meet the specification and parameters of drain are becoming standardized. In the selection of vertical drain, more emphasis was put On to the parameters such as discharge capacity and parameters such as tensile strength of entire drain. The required specification and parameters of supplied materials are shown in the Table. 3.
2*
2*
Due to the requirement of vast quantity of vertical drain it was difficult to rely on the supply of drains from one source. Therefore parallel usage on another type of vertical drain was required. In Changi East Reclamation Project several types of vertical drain were used since soil improvement works were carried out under various Phases. In each phase, at least two types of prefabricated vertical drains were used. The types of vertical drain used were Mebra MD 7007 (Manufactured in Holland, Korea Malaysia) Colbond CX 1000 and Flexi FD 767. However all types of vertical drain meet specification requirement. 552
Table 4. Types of rig used in Changi East Reclamation. Type of rig
Type of base Machine
Cofra
0 & K Excavator 70 - 110 RH 30, RH 40 0 & K Excavator 70 - 120 RH 30, RH 40 (01) Hitachi Excavator EX 1100 Crane 7 7 - 100
Econ
Yuyang
B+B
Daewoo - Solar 450 - Excavator Daewoo - Solar 450 - Excavator Excavator
B+B
Excavator
Chosuk Dae Yang
Capacity of Penetration Height of base machine Power (tons) Rig (m) (tons)
Maximum Mechanism penetration of penetration depth (m)
20 - 30
36 - 55.5
50.5
20-30
3 6 - 56.1
51.2
25 - 30
52.5 - 55.8 52.8
45
25
54.6
51
33 - 55
20-34
42-56
51
3 1 - 47
29 - 45
43 - 50
41 - 48
Hydraulic motor Motor, wire & Pulley - do -
Hydraulic motor sprocket & chain Hydraulic motor multi pulley system Hydraulic motor Push in roller & clamp Hydraulic sprocket & chain Vibro pushin
Maximum penetration length I days (m 114 hrs.) 30,000 27,000
10,000
12,000 13,500 19,200 8,600
Figure 3. Types of mandrel use in Changi project. 6
SELECTION OF INSTALLATION MANDREL AND ANCHOR
RIG,
Therefore suitable types of rig are required to be selected based on penetration depth required, type of soil and profile. In addition to the common type of vertical drain rig, the special types of rig with special features are required for difficult area. There are also vertical drain rig with hydraulic balancing, which is normally used in soft ground area, and vertical drain rig with vibratory penetration, which is generally used for hard penetration. In addition to the type of rig, selection of type of mandrel and anchor is also important. Fig. 3 shows various types of mandrel and anchor used in the Changi East soil improvement works. It is common practice to use the smaller dimension of mandrel in order to reduce the disturbance to the ground.
For prefabricated vertical drain works carried out in the large area where soil profile and properties are varied, required more than one type of Rig. For example shallow area may require shorter rig with lower capacity whereas deep area may require taller rig with higher capacity. In some places powerful installation rig with heavy base machine is required to be able to penetrate the hard intermediate soil layer interbeded with soft clay. The common types of vertical drain rigs with height, weight of base machine, penetration mechanism and maximum thrust force available are shown in the Table 4.
553
soft soil around the vertical drain throughout the fill until the ground surface when installing in the very soft soil or highly disturbed soil. At such location smaller mandrel with tippered shoe and anchor bar combined with hydraulic balancing may reduce the formation of extruded soil column. 7 QUALITY CONTROL OF VERTICAL DRAIN MATERIALS
Figure 4. Quality control of vertical drains. However flexibility of smaller mandrel would deter the penetrability and verticality of mandrel penetration. Therefore sometimes bigger dimension with stronger stiffness mandrels were bound to be used for deep penetration and penetration through intermediate hard layer. In such case significant disturbance could be encountered and it has to be taken into account in the design phase. Size and shape of the anchor must be strong enough to be able to anchor and able to protect the ingression of soil into the mandrel. Smaller anchor with tippered mandrel reduces the disturbance during installation. However the hydraulic balancing inside mandrel may require for penetration in the very soft clay where excess pore pressure could force the soil to ingress into the mandrel. The large mandrel with oversized anchor would create column of extruded
Since large quantity and several types of vertical drains are being used, good quality control system is essential. The procedure of quality control involves evaluation of manufacturer specification against required specification. Third party and on-site laboratories carried out the laboratory tests and monitored the consistency of supplied materials. Hydraulic parameters of prefabricated vertical drain such as permeability, Apparent Opening Size(A0S) and discharge capacity are important for better performance of vertical drain whereas mechanical property such as tensile strength of entire drain is important to be able to withstand the stress caused during mandrel penetration. In Changi East Reclamation Project, third party testings were called for full scale tests on every one million metres length of vertical drain and discharge capacity and tensile strength test on every 20,000 metres length of prefabricated vertical drain are required to carry out by on-site laboratory for checking the consistency of supplied material. One research laboratory was also established to carry out the special tests to simulate the in-situ condition in order to check the possible performance of vertical drain under field condition. Details of quality management on prefabricated vertical drains could be found elsewhere (Bo Myint Win et al. 2000). Some of the test results from on-site laboratory are shown in Fig. 4. With these three adopted laboratories, quality and consistency of supplied prefabricated vertical drains were managed to control.
8
INSTRUMENTATION MONITORING AND ASSESSMENT OF PERFORMANCE
For a soil improvement works with prefabricated vertical drains, the systematic instrumentation scheme is essential to monitor the performance of vertical drain and to assess the degree of consolidation. Two major instruments such as settlement plate and piezometer are necessary to monitor the consolidation process. Deep screw settlement gauges and multi-level magnetic settlement gauges were designed to monitor the settlement of various sub-layers whereas piezometers were installed at the center of each sublayer .to monitor the degree of consolidation or extent 554
of effective stress gain of sub layers. Some of the monitoring results are shown in the Fig. 5. With these instruments not only process of consolidation could be monitored but also degree of consolidation could be assessed. In addition to the soil instruments, in-situ testing is the alternative method to assess the degree of consolidation. The various types of in-situ testing equipment, which could be used for determining improvement, are as follows:
(1) Field vane Shear Test (FVT) (2) Cone Penetration Test (CPTU) (3) Dilatometer Test (DMT) (4) Self Boring Pressuremeter Test (SBPT) The another way of measuring the improved soil parameter is laboratory testing on the collected soil samples from post improvement borehole. The parameters indicate degree of improvement are moisture content, void ratio, undrained shear strength and apparent preconsolidation pressure. From those values, degree of improvement could be determined. Details of assessment of the degree of consolidation were described by Bo Myint Win et al. (1997). Some examples of assessed degree of consolidation from various methods are shown in Fig 6 to 8.
9 CONCLUSION Implementation of Mega soil improvement works required systematic approach and planning. Correct selection of prefabricated vertical drain material
Figure 5. Monitoring data from soil instruments.
Figure 6. Comparison of degree of consolidation from different instruments at vertical drain location (A2S-6). 555
In addition to that systematic instrumentation scheme together with regular monitoring program is required to monitor degree of soil improvement. Insitu testing equipment, and laboratory testing are also helpful in assessing the degree of improvement and consolidation of improving soil. 10 ACKNOWLEDGEMENT The authors would like to express their thanks to Mr. Ador Y. Franco and Mr. Rodolfo S. Arellano of Hyundai Engineering and Construction Company Limited for their help in Computer graphical works. 11 REFERENCES Bo Myint Win, A. Arulrajah & V. Choa 1997. “Assessment of degree of consolidation in soil improvement Project”. Proceedings of the International Conference on Ground Improvement Techniques: Macau : 71-80. Figure 7. Comparison of in-situ over consolidation ratio after soil improvement at vertical drain location (A2S-6).
Bo MYint Win, A. hirajah bi v. Choa l99ga. ”Site characterization for a land reclamtion project at Changi in Singapore”. Geotechnical Site charcterization, Robertson & Mayne (eds) Balkema, Rotterdam : 3 3 3 -3 3 8.
Bo Myint Win, A. Arulrajah & V. Choa 1998b. ”Land reclamation on slurry-like soil foundation”. Problematic soils, Yanagisawa, Moroto & Mitachi (eds): Balkema, Rotterdam : 763 - 766. Bo Myint Win, J. Chu & V. Choa 2000. “Quality Management on Prefabricated Vertical Drain Works in Land Reclamation Project”. An International Conference on Geotechnical & Geological Engineering, November 2000. (Submitted).
Figure 8. Comparison of laborato results fi-om prior to and after improvement borehxs. could be made only if comprehensive specification is adopted. Selected and deployed resources such as installation rigs and accessories (mandrel and anchor) should suit the type of soil and ground condition at improvement area. Full scale quality control laboratory is essential to maintain the consistency of parameter of supplied vertical drains. 556
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Design and construction method of compaction grouting as a ground-improving technique against liquefaction S.Yamaguchi, D. Kozawa, M.Arata & H. Matsumoto The 2nd District Port Construction Bureau, Ministry of Transport, Yokohama,Japan
M.Taki & Y. Kanno Fukken Company Limited, Consulting Engineers, Yokohama,Japan
ABSTRACT: The authors have been researching into the method of applying compaction grouting method (CPG) to ground-improvement work against liquefaction of sandy soil under earthquakes. The CPG has been used mainly for correcting the differential settlement of buildings and structures and protecting culverts by compacting the soil around them. There are not many cases of the technique used for the prevention of liquefaction of sandy soil. Therefore, before applying the technique to actual grounds for that purpose, subjects such as establishment of a design and construction method, evaluation of improving effect, influence on surrounding soil, etc. have to be dealt with. The authors applied the technique to an actual ground and propose a design and construction method, which proved practical. "Compaction Grouting Denver's Systems." In this test work, the spacing between injection bores was changed to grasp the relation between the displacement ratio (the volume ratio of grout columns to the soil to be compacted) and the improving effect. Besides, two systems of forming grout columns were used to grasp the relation between the work-executing procedure and the displacement in the adjacent soil zone in the ground. The ground was investigated before and after the test work to ascertain the improving effect and grasp the deforming behavior of the adjacent zone.
1 INTRODUCTION Compaction Grouting involves injection of stiff, mortar-like grout into previously drilled holes in the soil in a closely controlled manner. As the mass of grout increases under pressure, the soil is densified through compaction (e.g., Warner 1982, and Tamura et al. 1995). The principle of the CPG is to inject grout of very low liquidity into a ground to form grout columns, which exert compacting effect upon their surrounding soil zone in the ground. As grout is pressurized and injected with a pump, the CPG is called ''static compaction method." Accordingly, the CPG is distinguished from dynamic compaction methods such as the sand compaction pile method (SCP). In Japan, mainly used so far for the prevention of liquefaction in sandy grounds have been the SCP and the cement deep mixing method of soil stabilization. These methods present problems such as heavy vibration during execution of work and large displacement in adjacent soil zones in the ground. If there are important structures around an area of ground-improvement work, those methods cannot be adopted. Needed under such circumstances is a method which does not affect adjacent area and structures. The CPG is suitable to use in such circumstances if it is effective for the prevention of liquefaction in sandy grounds. The authors performed full-scale compaction grouting in a ground at an old taxiway of the Tokyo International Airport on a testing basis by using the
2 EXPERIMENTAL WORK 12.1 Strata and soil profiles of ground The experimental liquefaction-preventive work was czarried out in a plot on an old taxiway of the Tokyo Ihternational Airport. The stratigraphy in this area iInd the soil properties of the stratum to be improved 'were as follows. 1) Stratigraphy: The 2.3 to 2.8 meter-thick top layer c-onsisted of filled soil to raise the ground level and 1 pavement (asphalt, and compacted subbase course ;snd subgrade). Below the top layer was a 7.0 to 8.2 I meter-thick sandy layer consisting of dredged sludge, ?which was the main layer to be improved and called ,A,,. Lying below it was a clayey layer Acl. The ;groundwater level was in the fill layer. Accordingly ;grout columns were formed in the length of 9 m, ifrom 2.8 m below the ground level. 557
2) Soil properties of Aso: Table 1 shows the soil properties of the natural layer Aso. As shown in Figure 1, its fine-grained fraction (finer than 75 p m ) content Fc was 10-50%, revealing the sandy soil considerably inhomogeneous. Table 1. Physical properties of Aso.
2.71
0.2
7.5
11.8 0.90 1.53 0.80
2,2 Exper irnenta 1 Cases The experimental liquefaction-preventive work was carried out in four cases as shown in Figure 2. Cases 1, 2, and 3 were to see the relation between the displacement ratio (a,) and the improving effect. Cases 2 and 4 were to compare the ground behavior between two different grout-column forming systems. One system was to form the grout column from the bottom to the top (called “bottom-up method”). The other was to form first the top 2meter portion and then form the lower portion from its bottom to top (called “top-down/bottom-up method”). The former method was applied to Cases 1, 2, and 3; the latter, to Case 4. 2.3 Soil tests
Figure 1. Grain size distribution curves of Aso.
Figure 2. Details of experimental work.
The following in-situ tests and soil-property tests were conducted before and after the experimental work. See Figure 2 for the test locations. The parameters involved in the tests are design parameters used for the evaluation of liquefaction in Japan and important criteria to estimate the improving effect by the CPG. 1) Standard penetration test (SPT): SPT N-values were measured every one-meter depth. 2) Soil property test: Bulk density P t, grain-size distribution curves, fine-grained fraction content Fc, plasticity indices Ip, etc. were determined. 3) Density log in borehole: Density was continuously measured depthwise with a radioisotope. 4) Selfboring pressuremeter test (SBP) in soil: Coefficients of earth pressure at rest KO were determined in situ to evaluate the confining effect of grout column’s lateral pressure on the lateral movement of soil. 5 ) Cyclic undrained triaxial test (liquefaction test) of soil: The cyclic stress ratio ( a d / a ;)20 was determined with undisturbed specimens of soil. ( ad/a: )zo is the value of ad/2a: when liquefaction occurs at the 20th loading under the condition that the double amplitude of axial strain DA is 5% ( o is the amplitude of cyclic deviatoric stress; a: , isotropic confining pressure).
2.4
Field measurement of deformation of ground
Regarding the behavior of the ground, the following items were measured. See Figure 2 for the points of measurement. 1) Leveling: The vertical displacement of the ground surface, that is, the elevation of pavement by the CPG was measured with level. 2) Horizontal displacement: The horizontal displacement of soil, that is, the lateral propagation distance of displacement by the CPG was measured with inclinometer. 3) Pore water pressure: The pore water pressure of the layer A,, was measured with electrical pore pressure meters. The dissipation process of the excess pore water pressure around grout columns was monitored, and the time for the post-grouting survey was determined. The excess pore water pressure decreased to almost zero in two to three days.
3 RESULTS OF EXPERIMENTAL WORK To evaluate quantitatively the improving effect of the CPG in the inhomogeneous layer A,,,, it was necessary to relate the evaluation method to the design method described later. The basic data for the evaluation, which directly concern the design criteria against liquefaction in Japan, are as follows.
Figure 3. N-values before and after CPG.
3.1 N-values Figure 3 shows the pre- and post-CPG vertical distribution of N-values for each Case. F c values are also plotted in the Figure. It is apparent that the Nvalues were increased by the CPG. It is suggested that as the degree of compaction and the displacement ratio heighten, the increment in the N-value increases. However, the effect of Fc on the compaction degree, or N-value, is unclear from these data. Anyway, it was ascertained that the CPG compacted the sandy soil. The N-value is the most important index in discussing the design criteria, as described later. 3.2 Soil parameters Figure 4 shows the relations between p I , KO,and ( od/a: )20 on one side and the depth on the other. 1)Bulk density of soil: 0 I was increased by the CPG in every Case. As there was dispersion depthwise in the measured values of soil density and the layer A,, was inhomogeneous, it was difficult to evaluate the increment in the soil density quantitatively. 2) Coefficient of earth pressure at rest: The pre-CPG coefficient ranged from 0.45 to 0.60; the post-CPG coefficient, from 0.65 to 2.95. As the mean value of the
Figure 4. Soil properties before and after CPG.
559
natural layer A,, was 0.5, a common value, these measured values appeared accurate enough. Therefore, the increase of KOwas due to the increase of lateral pressure caused by the CPG. The dispersion depthwise taken into account, the increment in KO due to the CPG can safely be expected to be in the range of 0.5 to 1.5. 3) Cyclic stress ratio: ( ad/a: )zo was apparently increased by the CPG although its degree was not so salient as those of the N-values and KO.Its cause may be the effects of the isotropic consolidation (K,=l .O) in the cyclic undrained triaxial test. It is generally known that as the KO increases, the ( a d / o :)20 increases. Therefore, the (a, /a: )20 shown in Figure 4 (c) may be underestimated. 3.3 Behavior of soil around CPG Figure 5 shows the distribution of displacement of the soil around the grout columns at the cross section A-A’(see Figure 2) in Case 2. The injection pressure of grout basically acts radially, or laterally. However, when the displacement ratio is large, the injection pressure acts not only laterally but also upward, causing both lateral and upward displacement. 1) Elevation of ground surface (Case 2): The maximum elevation of 52 mm was recorded at the central part of the improvement area. The maximum elevation decreased to 48 mm in 10 to 20 days because as the grout columns hardened, they contracted slightly. As the measuring point moved
Figure 5. Behavior of soil around CPG in Case 2. 560
outward away from the boundary of the improvement area, the elevation decreased to become zero at a distance of about 15 m. However, although Case 4 had the same displacement ratio as Case 2, it presented a smaller maximum elevation of 31 mm, the final value being 28 mm. Accordingly, the top-downbottom-up method can effectively be used when it is desirable to minimize the elevation of ground surfaces. The above elevation values are smaller than those usually observed, and its reason would be the resistance of the pavement lying above the grout columns. 2) Horizontal displacement (Case 2): The horizontal displacement of the pavement, A,,, and at the outward distance of 2 m from the boundary of the improvement area was -8 mm, +32 mm, and +67 mm, respectively. As the measuring point moves away from the boundary, the horizontal displacement decreased to become zero at a distance of about 14 m, as in the case of the elevation. 4 DESIGN METHOD The authors examined and considered the data from the experimental work to reach the conclusion that the design method for SCP (e.g., Mizuno et al. 1987) could be applied to CPG with some modification. In short, the post-CPG N-value can be formulated through the relative density D,and the void ratio e of
=0.010Fc+1.130 [see Fig.7(a)]
=0.010Fc~~.130 [see Fig.7fa)l
density of natural ground; a:,, effective overburden pressure; Hi,N-value after improvement without consideration of F,; N,, N-value after improvement with consideration of F,; emu, maximum void ratio; ernill, minimum void ratio; and e,, void ratio equivalent to N I . The calculation procedure of Figure 6 (a) will be described below. 1) Relation among N , a:, and Dr (Gibbs and Holtz 1957, and Meyerhof 1957) D,
=2
1 ~ ~ ~ /+~0.010:. 0.7
)
(1)
where Dr is in 5%; a,‘,,in kN/m’. 2) Defining formula of e and D, ~r
*N-valoe of plan value after CPG (a)Calculation of N, from a
(b)Calculation of a from N,
Figure 6. Outline of the proposed design method.
=
(2)
(ern,, -el/(ernax -emin 1x100
The values of ernnsand erniri can be determined by maximum and minimum densities tests of sands. In the above e x p e r i m e ~ t awork, ~ they were found from Figure 7 as follow:
(3)
emax= 0.010Fc +1.130 emin= 0.006Fc + 0.671
(4) 3) Calculation of NI without considering effects of fine-grained fraction The equation (1) gives the following equation for the relative density of the natural ground:
1.
2 1. 1. 1.
‘
Fc (%I
(a) emaxNFc
1.3
The following equation for the void ratio of the naturaf ground is derived from the equation (2):
1.1
2 0.9
0’50
10
20
30 40 50 60
Fc (%I
(b) e,,,-Fc Figure 7. Relationship between ernmand emlnand Fc.
sand. Omitting the description of the process of reaching the conclusion, we will describe the outline of our proposed design method and its verification result. 4.1
Outline of design method
Figure 6 shows the calculation flow of the design method. The calculation procedure of (a) is to calculate the N-values between grout columns after CPG, and the calculation procedure (b) is to calculate the necessary displacement ratio in the stage of design. In other words, these procedures are just opposite to each other, and hence they reach one and the same result. The signs in the Fig. and their meanings are as follows: No, N-value of natural ground; e,, void ratio of natural ground; Dro,relative
)/loo
(6) The displacement ratio a, is given by the following equation: ‘ 0 =ernin
0.7
-DrO(ernax -ernin
a , =(eo -e,)/h+e*> (7) From the equation (7), el = eo - a , (1 +eo ) is derived. By substituting it for e in the equation (2), the following equation is given for the relative density between grout columns after improvement:
Dr, = (ern,, )/(ernu -emin 1x100 (8) The following equation is derived from the equations (1) and (8): N,
= (0.7
+ 0.010:
,@,I
/21)2
(9)
4) Reduction of improving effect by fine-grained fraction (Calculation of N2) The relation between NI and N2 was derived by comparing N I given by the equation (9) and measured N2 (affected by the fine-grained fraction) to introduce a reduction rate P shown below, and an experimental equation was obtained from the plotted values in Figure 8 as follows: 561
Figure 8. Relationship between
0
the N-value, soil density, KO,and ( D~ /LT: )20,proving itself applicable enough to the measures against liquefaction. 2) The CPG caused displacement in the soil around the grouting area. The maximum elevation of the ground surface and the maximum lateral underground displacement (in Aso) were 52 mm and 38 mrn. The ground-surface elevation was smaller than those usually observed. Its reason would be the resistance by the pavement lying on the grout columns. 3) The design method for liquefaction-preventive work with CPG proposed in this paper is based on a principle similar to that of the design method for SCP. The proposed method is simple and practical. Besides, the fine-grained fraction content is considered in the proposed method, it enables us to make reasonable design which reflects the soil properties of actual grounds. The authors are planning to perform further detailed study to evolve the design theory and enhance the design method.
and Fc.
z*o 1.5
;r"
----. 1.
c
0. 0. No
ACKNOWLEDGMENTS
Figure 9. Relationship between No and W2/W2.
The present study was made under the guidance of the investigating committee for liquefactionpreventive methods for the ground under the existing pavement in the Tokyo International Airport. The authors would like to thank Prof. K. Zen of Kyushu University and the members of the committee.
P=",-No)/tN2-No) = 1.05 - 0.51Log F,
Therefore, N,is given as follows: N , =No +P@, - N o )
As the proposed design method is very simple, it can conveniently be used in designing CPG for the prevention of liquefaction. 4.2
REFERENCES
Verification of precision of proposed design method
The proposed design method described above was applied to the experimental work. Figure 9 shows the relation between No and N ; / N , , Nh representing the measured N-value after improvement, N ; / N , representing the order of approximation between the measured value and the estimated value (calculated value by the proposed design method). As shown in Figure 9, N ; / N , ranges from 0.5 to 1.5, but its mean is about 1.0. Thus, the proposed design method proved to have sufficient accuracy for actual field application.
5 CONCLUSIONS The conclusions of the present study are summarized below: 1) The CPG is effective in compacting sandy grounds. In our experimental work, CPG increased
Gibbs, H.J. & W.G. Holtz 1957. Research on determining the density of sand by spoon penetration test. Proc. 4th Int. Con$ on Soil Mechanics and Foundation Engineering, London, 1:35-39. Meyerhof, G.G. 1957. Discussion of Session 1. Proc. 4th Int. Con$ on Soil Mechanics and Foundation Engineering, London, 3. Mizuno, Y., N. Suematsu & K. Okuyama 1987. Design method of sand compaction piles for sandy soils containing fines. Tsuchi-to-Kiso, Japanese Geotechnical Society, 35(5):21-26. (in Japanese) Tamura, M., H. Sibata, T. Satoh & K. Ohsawa 1995. Strengthening of building foundation by compaction grouting. Tsuchi-to-Kiso, Japanese Geotechnical Society, 43( 11):31-34. (in Japanese) Warner, J. 1982. Compaction Grouting-The first thirty years. Proc. Conf. on Grouting in Geotechnical Engineering, ASCE, 694-707.
562
Coastal Geotechnical Engineering in Practice,Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Effects on environmental aspect of new sand compaction pile method for soft soil M.Yamamoto & M. Nozu Fudo Construction Company Limited, Tokyo,Japan
ABSTRACT: The conventional sand compaction pile (SCP) method, which is a widely applied ground improvement method, aims to increase the bearing capacity of soft ground by reinforcing it as compacted ground, or by increasing the density of loose sandy ground. However, the use of a vibro-hammer generates noise and vibration, which may adversely affect the surrounding environment. To reduce noise and vibration, the authors developed a “non-vibratory SCP method”, that is based on a rotary penetration system that uses a forced liftinddriving device and a rotary drive motor. And as an another new SCP method, “SCP with excavated soil method”, in which the excavated soil will be recycled as the material of the pile instead of sand, has developed in recent years. Because it has been increasingly difficult to obtain highly permeable materials required for the SCP method and sand drain (SD) method. This paper sets out the development objectives and development history of above two methods, which are both applicable for environmental point of view.
1 INTRODUCTION
2 NON-VIBRATORY SCP METHOD
The sand compaction pile (SCP) method is a way to improve soft ground through the installation of well compacted sand piles. It conforms with fundamental principles of compaction and consolidation drainage. As a result, it has been used successfully on many occasions for the improvement of all types of soil, since 1950s in Japan. In soil improvement field, latest key environmental technology issues include “noise-free and vibration-free construction in urban areas” and “recycle of soil obtained during the excavation for building foundations and tunnels and of coal ash produced at coal-fired thermal power plants as the pile material”. This paper, first of all, explains the operational equipment and procedure for the “non-vibratory SCP method”, as well as its improvement effect, which is almost same as the conventional SCP method. In order to clarify the mechanism of non-vibratory compaction of loose sand layer, authors tried to conduct the laboratory test. (Nozu, et al. 1998.) Secondary, as for the “SCP with excavated soil method”, this paper describes the results of a field test conducted on the landfill at the Nagoya port. As shown in Figure 1, above two methods might be applied simultaneously in urban area.
2.1 Outline In the case of sandy ground, SCP method is widely used as a remedial measure against liquefaction. It is one of the most reliable methods for this, and its effectiveness in compacting the ground to control liquefaction has been verified in several strong earthquakes. However, the use of a vibro-hammer generates noise and vibration, adversely affecting surrounding areas, and so it is not suitable for use on urban sites or those very close to existing structures. In an attempt to overcome these problems, a new, non-vibratory compaction method, called the nonvibratory compaction method, has been developed (Yamada,T. and Nozu,M. 1996). This allows cost savings in operations close to existing structures, and minimizes noise and vibration to eliminate adverse effects on the surrounding environment. 2.2 Equipment, operation procedure
563
The non-vibratory compaction assembly is shown in Figure 2. The major difference between this and the conventional SCP method lies in the means of penetratiodwithdrawal of the casing pipe. In the conventional SCP, the casing pipe is penetrated by means of vibromotive force supplied by a vibrohammer, and is withdrawn by winding a cable. Instead of a vibro-hammer, the non-vibratory compaction method uses a forced liftinddriving device to raise or lower the casing pipe as it is rotated.
Vibration free
Recycle(use excavated soil)
Non-vibra~orySCP SCP with excavated soil method method Figure 1. Required sand compaction pile method from environmental viewDoint
The operation assembly consists of an pile driving device as its base machine, an electric rotary drive motor to rotate the casing pipe, and liftinddriving device with a rack-pinion type geared motor. The total weight of the assembly provides the necessary force, and an oil pressure motor drives the rotation of the pinion gears. The main characteristic of the non-vibratory compaction operation is the wave action with which the sand piles are installed, as shown in Figure 3. 2.3 Confirmationof Reduced Vibration Figure 4 plots the relationship between recorded level of vibration (dB) and distance from operation equipment (m), for several work sites. Japanese regulations stipulate that “vibration must not exceed 75dB at the perimeter of the site of specified construction work”. This level and background vibration data measured at Site A, are also marked in the figure. Vibration measurements conform to JIS Z 8735, a Japanese standard for recording vibration levels. The figure shows a reduction in vibration of 25 to 30dB in non-vibratory compaction operations over convent~onalSCP methods, with vibration hardly perceptible at distances as little as 10m away from the operation equipment. 2.4 ~ ~ n f i r m a ~ofi oCompaction n Eflect
Figure 2. Non-vibratory compaction assembly
Figure 4.Decrease in vibration with distance
Comparison of the results of several trial sites were made to determine whether or not the compaction effect of the non-vibratory compaction method matched that of SCP methods. As an example, the results of standard penetration test made before and after improvement work at Site A are given (Yamada, et al. 1996). At this trial site, both SCP and non-vibratory compaction method were used for two different improvement ratios, a, = 10% and 20%, as shown in Figure 5, and N values for the two methods were compared. Figure 6 shows comparisons of the distributions with depth of SPT N-values measured before and after improvement, for ~mprovementratios of 10% and 20%, together with the soil profile and the fine grain content. As the fine grain content distribution shows, the top 4m of the ground was clay and was therefore not subject to improvement. In the case of the 10% improvement, for the sand layer at depths of between 4m and 10m, with a level of fine grain content of about 20%, SPT N-values measured before improvement of between 10 to 14 increased after improvement to levels of between 15 and 25 for both non-vibratory compaction method (marked 6 and A) and SCP (marked 1 and +) improvement. In the case of the 20% improvement, SPT N-values after improvement increased to around 20 to 25, the improvement effect of the non-vibratory compaction and SCP methods again being virtually the same.
564
Improvement as = 10% *
Improvement as = 2070
location of standard penetration lest after improvement
Figure 5. Site A; improvement specification (unit: mm)
At Site B, the non-vibratory compaction method was used for ground improvement to prevent liquefaction in the ground beneath sea walls (Suganuma, et al. 1997). It was decided that SCP should be implemented to ensure the stability of newly constructed embankments after the removal of the existing ones. Because of concern over the adverse effect of vibration on nearby structures, the non-vibratory compaction method was selected. Figure 7 shows a cross section of the site ground and specifications of the sand piles. Figure 8 shows the relationship between the revised N- values (N1=1.7N/( 0 .'/98+0.7)) before and after improvement and the fine content, taken from standard penetration test undertaken to confirm the improvement effect. The revised N values before and after improvement are relatively smaller as the fine content increases, but increased revised N values after improvement shows the improvement effect.
3 SCP METHOD WITH EXCAVATED SOIL
Figure 6. Site A; comparison of improvement effect
Non-vibratory SCP method 0 700, as = 15% -12.5
23.1 m (unit: m)
(unit: mm)
Figure 7. Site B; cross section, SCP specification
3.1 Outline In recent years, it has been increasingly difficult to obtain highly permeable materials required for the sand compaction pile (SCP) method and sand drain (SD) method. Continued acquisition of the good materials in the future will undoubtedly be difficult. In order to cope with such a situation, .research and investigation and pilot constructions have been carried out. As a result, a "soil improvement method using excavated soil from construction as the materials instead of sand" has been developed (Matsuo,et.al. 1997). If excavated soil can be used as soil improvement material at the site, in particular, construction cost can be frequently reduced. 3.2 Construction procedure (1) Classification by material The "soil improvement method using excavated soil construction as the materials" is classified into two types according to the material property. The basis for classification is whether or not it is possible to pull out a casing pipe in the soil to discharge the material and then insert the pipe again to expand the pile diameter for making a compacted pile. Classification here is based on the fine content Fc, the sum of cohesive component and silty component in the excavated soil. a. Material for SCP (Type 1: 15%
Figure 8. Site B; relation between revised N-value (NI) and fine content
The construction methods are outlined in Figure 9. 565
that when the Fc of the material exceeds 15%, the permeability coefficient rapidly drops to 10-'m/sec or below. Drainage of the pile is, therefore, carried out not by the pile itself but by additional drain pipes such as plastic board drain (PD). With Type 2, 4 pieces of PD are attached to a felt bag (PD-attached bag), in which excavated soil is placed. Required number of PD can be estimated by 2-D FE-analysis. There is no limit on the fine content of the excavated soil used for Type 2. The water content of the soil should be 25% or less for this case. That limit is imposed for the convenience of construction. If the water content is higher than 25%, it should be reduced to 25% or less by using additives like quicklime. Here, a phenomenon of "volume decrease and strength increase owing to the consolidation of the cohesive soil due to installation of the piles, without embankment" is considered as a function of the SCP (Figure 10). The conventional SCP design assumes that the surrounding clay loses its strength after pile driving, but later its strength recover to the same level of the natural ground. It has, however, come to be known that consolidation for an adequate time enables the clay strength to increase over that of the natural ground even without any surcharge such as embankment. Because the clay can be consolidated by dissipating (with artificial drains) the excess porewater pressure that is generated in the soil by the cylindrical (horizontal) cavity expansion during the driving of sand piles (Asaoka, et.al. 1994). The SCP with excavated soil method makes an active use of the above feature to dispose of excavated soil. Specifically, the driving of piles of excavated soil enables the ground that has risen due to pile driving to settle in the subsequent consolidation period, after all the excavated soil can be returned in the ground. When the excavated soil is used for compacting sandy soils, the soil that has risen can be scooped up to use it as the material again.
Figure 9. SCP with excavated soil method
Figure 10. Consolidation due to sand pile installation
Figure 11. Construction procedure for Type 1
(2) Construction method The construction machinery is almost same as that for ordinary SCP. It is characterized by some features applied to enable soils containing a large amount of fine content to pass through the machine inside smoothly, using special hopper with low-friction inner wall such as stainless and/or nylon resin, constant diameter casing pipe, and PD-attached bag for the pile. The construction is carried out by the following procedure.(Figure 11,12.) Figurc 12. Construction procedure for Type 2
3.3 Construction work for field verification of soil improvement effect
Type 1 can be used for compacting sandy soils, and/or stabilizing and reducing the settlement of COhesive soil ground. Type 2 is used to accelerate the consolidation of cohesive soil ground. It is known
(1) Outline of construction work for field test The objectives of the field test were to verify the increase and dissipation of excess pore-water pressure during construction, and related strength increase and ground surface settlement in cohesive soils.
566
upper part consists of filled clay layers and the lower part at a depth of 12 m or below consists of alluvial sandy layers. The pile was 15 m long. (2) Field test results and discussion Figure 17 shows vertical distributions of the unconfined compression strength qu of the cohesive soils improved by the piles. For all types of method,
Figure 13. Site condition and instruments arrangement
Figure 16. Soil profile of the natural ground Figure 14. Grain size distributions of the materials
Figure 17. Vertical distribution of the clay strength Figure 15. Site view
Figure 13 shows the construction condition and instruments arrangement. The grain size distributions of the materials used are given in Figure 14. For Type 1, PD were applied after the driving of piles of excavated soil was completed. Type 3 is the same construction method as Type 2, although the material for the former type has a higher fine content (50%
Figure 18. Change with time in pore-water pressure
567
the cohesive soils improved by the piles due to consolidation. Figure 19 shows settlement immediately following the installation of piles to improve clay layers on the landfill. The settlement shown is purely for the clay layers on the landfill because the readings of differential settlement gauges installed at the bottom of the clay layers are subtracted from the settlement of the ground surface. According to the dissipation of excess pore-water pressure, settlement of 300 to 400 mm occurred at each point without any embankment load. Figure 20 is a vertical distribution of the N-value at the pile core (after the lapse of one and three months) for Type 1. It shows that the pile strength in sand layer is very high and it is kept for long time. Figure 21 shows an increase of N-value at the point of center of four piles in the improved alluvial sandy layer deposited under the reclaimed layers (Type 1). After one month, incremental N-value was about 2-10. Thus, this method was also proved to be fully effective for compacting loose sandy soils.
Figure 19. Time-settlement relationship N-value
4 CONCLUSIONS This paper has described about the non-vibratory compaction method, and SCP with excavated soil method. As for the SCP with excavated soil method, the number of vehicles for removing excavated soil and transporting sand for pile material is also reduced. These two methods, therefore, have a favorable impact on the environment and are expected to be applied more widely in the future.
1 -month 3 -inon t hs
Figure 20. Vertical distribution of N-value at the pile core
REFERENCE
Figure 21. Increase of N-value in the improved alluvial sandy layers
strength increase was outstanding. Strength increased more at a greater depth. Figure 18 presents changes with time in pore-water pressure during and after the installation of piles for improving cohesive soils at a depth of 6 m below the ground level. The excess pore-water pressure that increased rapidly with the progress of construction work returned to the level of hydrostatic pressure within a few months. The change corresponded to the strength increase A q,, of 568
Asaoka, A., Kodaka, T. and NOZU, M. 1994, Undrained shear strength of clay improved by sand compaction piles, SOILS AND FOUNDATIONS, Vo1.34, No.4. Nozu, M., Matsunaga,Y. and Ohbayashi,J. 1998, Application of the static sand compaction pile method to loose sandy soil, Proc. of the Int. Symposium on Problematic Soils, BALKEMA, 751-755, IS-TOHOKU,Sendai, Japan. Matsuo, M., Kimura, M., Nishio, R. and Ando, H. 1997, Development of soil improvement method using construction waste soil, Journal of geotechnical engineering, Japan society of civil engineers, No.567, U-35, 238-248. (in Japanese) Suganuma, S., Fukada, H. and Nakai, N. 1997. Case Studies of Ground Improvement using a NonVibratory SCP Method, Proceedings of 52nd Conference of Japan Society of Civil Engineering, m-412-413. (in Japanese) Yamada, T. and Nozu, M. 1996. Compaction Effect on Sandy Ground of Non-Vibratory Compaction Method, Proc. of 31st Conference for Japanese Geotechnical Society, 49-50. (in Japanese)
4 Engineered geo-materials made from solid wastes with/ without chemical treatment
This Page Intentionally Left Blank
Coastal GeotechnicalEngineering in Practice,Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Properties and application of expanded glass waste at lightweight ground material H. Abe & E. Fukazawa Kajima TechnicalResearch Institute, Tokyo, Japan
H. Mizutani & Y. Kato Kajima Corporation, Tokyo,Japan
ABSTRACT Authors have been carrying out researches toward an effective utilization of material generated by melting, expanding and consolidating waste glass bottles as lightweight ground materials. Various laboratory material tests have been performed, and the properties of expanded glass as lightweight ground material have been revealed. In addition, through field trial construction, it has been found that the expanded glass is fully applicable as lightweight ground material in the field of civil engineering.
2 PROPERTIES AND FIELD APPLICABILITY OF EXPANDED GLASS AS LIGHTWEIGHT GROUND MATERIAL
1 INTRODUCTION Recycling and effective utilization of various kinds of waste have drawn attention recently from the viewpoint of global environmental conservation. The authors have been carrying out researches toward an effective utilization of material generated by melting, expanding and consolidating waste glass bottles as lightweight ground materials (hereafter referred to as expanded glass lightweight ground material or simply as expanded glass) in the field of civil engineering’-’). Various laboratory material tests have been performed, and the properties of expanded glass as lightweight ground material have been revealed. In addition, the construction and quality control methods have been examined through field trial construction, and based on the results, the material has been applied in actual construction site for the back filling of structure. It has been found that the expanded glass, which is the product of an effective recycling technology of used glass bottles, is fully applicable as lightweight ground material in the field of civil engineering, such as filling material of embankment, refilling material and back filling material. This paper presents the results of the above investigations.
Various laboratory material tests were carried out in order to clarify the properties of expanded glass (see Photo 1) as lightweight ground material, considering the application of the material for embankment filling, refilling and back filling as shown in Figure 1. Construction and quality control methods were also examined through field trial construction.
Photo 1. Expanded glass waste
571
Figure 1. Application of expanded glass waste as lightweight ground material
2.1 Laboratory tests
Table 2. Specific gravity and Water absorption
Laboratory tests were conducted on the items listed in Table 1. The test method adopted in this study is according to the Japanese Industrial Standard, and the standards of the Japanese Geotechinical Society and the Japan Highway Public Corporation, which are standards for soil mechanics and concrete test.
Specific gravity in absolute
Water absorption
dry condition
(%I
0.41
18.3
Table 3. Compaction energy
-
P
d-
Grain size
Compaction Dry density Maxmum Grain size Grain size Table 1. Test Item Test Item
energy
Specific gravity,Water absorption
JIS A 1110
Water content of submersion sampl
JIS A 1203
Grain size analysis
JIS A 1204
Compact ion
JIS A 1210
Triaxial compression Loading of constant load
(t/m3) grain size D50(mm) D20(mm)
JGS 0524
JF
0
0.20
Dmax (mm) 37.5
24.2
18.5
1Ec*
0.30
37.5
18.6
8.8
2 Ec
0.36
37.5
14.6
2.0
3 Ec
0.41
37.5
11.6
0.5
1 Ec=54.9N*cm/cm3
p~ax=137.2kN/m2
Permeability
JIS A 1218
Slaking
JHS 110
Stability to sodium sulfate
JIS A 1122
Field rolling compaction
p
Test Method
lton Vibrating roller
J1S:Japan Industrial Standard JGS:Standard of Japanese Geotechnical Society JHS:Standard of Japan Highway Public Corporation
The test results are shown in Tables 2 to 5, and Figures 2 to 5. Major observations from the test results are as follows. a. The dry specific gravity was 0.41 and the absorption ratio was 18.3%(Table 2). b. The material was classified as gravel according to the grain size. Compaction caused slight refining.
572
Table 4. Test result of triaxial compression Dry density
-p d (t/rn3) 0.2 0.3
Sample condition Dry Submerge Dry Submerge
Cohesion Angle of shear CU (m/m2) resistance,4 d 0
29
0
27
0 0
33 31
0.3
Dry Submerge
0 0
34 33
0.4
Dry Submerge
0 0
37
The compressive strain was about 1.8% under the loading of 98 kN/m2 when the compacted material density was 0.3 t/m3 (Figure 5). g. The permeabi~ity coefficient of compacted expanded glass k was 1.2~10'to 2.9x10-' cm/s at a density of compacted material of 0.3 to 0.4 t/m3. This indicates that the permeability coefficient of expanded glass is at the same level as those of sand or gravel (Table 5). h. The slaking ratio of expanded glass was obtained by a test based on JHS 110, a Japan Highway Public Corporation standard (a method of testing rock slaking ratio). The slaking ratio was approximately 0.1%, and there was very little change in the grain size composition of expanded glass due to alternate wetting and drying (Table 5).
35
Table 5. Test results of permeability and stability Dry density Coef. permea- Slaking p
d
(t/m3) bility k (cm/s)
Stability sodium sulfate
(%)
(mass lost,%)
0.30
1.2~10'
0.1
3.7
0.41
2.9~10~
-
-
i. A stability test using sodium sulfate revealed that the percentage of mass lost of expanded glass was about 3.7%, which was lower than 12% the number specified in the Standard Specifications for Design and Construction of Concrete Structures. This indicates that the expanded glass has a high resistance to weathering (Table 5).
Compaction of 1 Ec reduced the mean grain size D,, from its original 24 mm to about 20 mm (Figure 2).
2.2 Field applicability
c. The features of compaction was that with the increase of energy of compaction, the dry density p d increased. The dry density p d increased from 0.21 t/m3 at a compaction energy of zero Ec to 0.30 t/m3 at 1 Ec and to 0.36 t/m3 at 2 Ec (Table 3).
A field trial construction was carried out to collect basic data for construction and quality control of expanded glass (Photo 2). For compactio~,a l-ton vibrating roller and a 10-ton swamp bulldozer were used. T h e test result s h o w e d a t e n d e n c y of increasing density of compacted material with the increasing number of passes of the roller. With more than eight times of compactions, the density converged to a constant level. It was also found that four times of compactions with a l-ton vibrating in pd=0.30 tim37 and four times
d. Absorption tests were carried out on the assumption of the underwater use of the material at a depth of 3m. Water pressure during immersion Pu was set at 29.4 kN/m2. The results of the tests showed that the immersion caused the water content of expanded glass to increase, but the value tended to reach a constant level during about 30 days of immersion. The water content was about 40% under the condition of water pressure p,=29.4 kN/m2 and days of immersion period (Figure 3). e. As for the shear behavior of compacted expanded glass, was clarified. The cohesion was zero and the friction angle Q>d was around 27-37 degrees when the density of compacted material p d=0.20-0.40t/m3.The increment of p d leaded to larger @ (Table 4,Figure 4). f. Compressive deformation characteristics of expanded glass were investigated under constant loads for a long time. The compressive ratio reached a certain value after about 400 minutes of loading.
Figure 2. Grain size accumulation curve of Compacted expandedglass
573
45
0.55
A
water pressure p,=29.4kN/m2
-
0.50
lOton swamp bulldozer
m h
5
0.45
v
3 *
40
-j
0.40
Q
E E
*
.-
3
L
8 35
-0
5 0.25
dry density p d=0.41t/m3
Y
s
0.35 0.30
'D
j
0.20 0.15
n -0
10
20
30 40 50 elapsed time (days)
60
four times compactions I
I
I
I
5 10 15 20 number of rolling compaction (times)
70
25
Figur 6. Relationship between dry density and number of rolling compaction
Figure 3. Water content of immersion sample UU"
/
back filling
P v1
400-
200
0
400 0
600 800 (kN/m2)
1000
1200 (unit ;m)
Figure 4. Mohr's stress circle of compacted expanded glass Figure 7. Real application 4.0 I
1-
96 hour loading 3.0
box culvert
1 (7) expandedglass
10ton truck 1.2m3/bag. 25bagIload
2.0
(Photo 3.) 0.0
0
20
40
60 load
80 100 p (kN/m2)
120
140 160 lton vibrating roller
four times rolling compaction
Figure 5. Loading test results (Relationship between compressive ratio and load)
(Photo 4.) control
I
cornpactions with a 10-ton swamp bulldozer resulted in p d =0.37 t/m3 (Figure 6 ) . From a series of the laboratory tests and field cornpaction tests, the basic data were obtained on the characteristics of expanded g l a s s and o n the applicability to the actual construction site. It was confirmed that expanded glass could be used forfilling, refilling or backfilling (Figure 1) as being expected.
Figure 8. Construction flow
574
p d 2 0.30 t/m3 in-situ density test
3 APPLICATION EXAMPLE As shown in Figure 7, expanded glass was used for backfilling For a box culvert (with an inner crosssection of 7.0 m x 4.0 m and a length of about 90 m), which was a crossing structure under a road within an athletic facility. The construction was carried out by the procedure shown in Figure 8. Photo 3 shows and Photo 4 shows compaction work. About 1,000 m3 of expanded glass was used. Four times ofcompactions were conducted using a 1-ton vibrating roller. For quality control, density tests were carried out for every 100 m3 of expanded glass b y replacing local water so that the target density of compacted material p could be equal to or larger than 0.30 t/m3. The mean of measured values was 0.32 t/m3, which reached to the satisfactory level.
575
Because the density of the expanded glass waste it floats in the water. Therefore, when used below the ground water level, filling must be doue over the expanded glass waste.
4 CONCLUDING REMARKS Reported above are the results of laboratory tests, field trial constructions and real appli~tionsof expanded glass. Through the series of studies, it was found that expanded glass was fully applicable as material for embankment filling, refilling or backfilling. Expanded glass is a safe recycled
products from the viewpoint of environmental conservation since it causes no concern about toxic substance emission. The authors are currently studying the feasibility of "a combined use of expanded glass and excavated earth" in order to make an effective use of excavated earth, and to make contribution to a further use of recycled products through establishing above mentioned technologies. REFERENCE Ueda, T., Fujisaki, K., Mizutani, H 1997. Properties of lightweight soil mixture made of recycled scrapped glass, Proc. of 25th con$ of Kanto Branch of JSCE, (inJapanese): 986-987. Iwama,K.,Ukita,T.,Mizutani, H 1998. Applications of new lightweight fill material, Proc. of 26th. Con$ of Kanto Branch of JSCE, (in Japanese): 528-529
576
CoasfafGeoiec~nicalEng;neefi~gin Practice, Nakase & ~suc~;da (eds)02000 8alke~a,f f o ~ e f d /S8N a ~ , 90 5809 151 7
A study on the soil ~ p r o v ~properties ~ ~ n t of FGC ~ ~ d ~ nagent ing B.S.Chun ~ e ~ a rof ~Civil ~ ~ngin~ering, n t ~ u n ~ a ~n ngi v e r s iKorea ~?
J.C. Kim Korea Institute of Geo TechnologyIncorporated,Seoul, Korea
ABSTRACT: Increase in the traffic volume requires construction of new roads. In Korea, most of the pavements are covered with an asphalt concrete or Portland cement concrete. Therefore a new pavement system is highly demanded to protect the natural environments around the roads. The purpose of the study is to investigate properties and applications of soilcrete using soil mixed with Fly ash, Gypsum and Cement(FGC) stabilizer and then to evaluate the benefits to be gained by soilcrete pavement when the existing pavements are replaced by soilcrete.
1 INTRODUCTION Soil-cement is a mixture of pulverized soil material and measured amounts of Portland cement based soil stabilizer and water, compacted to high density. As the cement hydrates, the soil-cement mixture becomes harder and more durable. The soil material in soil-cement can be almost any combination of sand, silt, clay and gravel or crushed sand. Local granular material (such as slag, caliche, limestone, scoria) plus a wide variety of waste(such as cinders, fly ash, and screenings &om quarries and gravel pits) can be used to make soilcement. Also, old granular-base roads, with or without their bituminous surfaces, can be recycled to make good soil-cement (ASTM 1995). Soil-cement is sometimes called cement-treated base or cement-stabilized-aggregate base. Regardless of what it is called, the principles governing its co~positionand construction are the same. Like in this study, engineering properties of FGC-treated base is similar to that of the ordinary lean concrete, so we will intend to call the FGC-soil mixture as ‘soilcrete’ from now on. Soilcrete is a traditional material for the above mentio~edground works. But it is not widely used because its quality is not homogeneous. The properties of soil materials, kinds and amounts of soil stabilizer uncertainly influence the quality of soilcrete. So there are some restrictions on permanent structures. With increasing emphasis on nature conservation, the application cases of soilcrete are rapidly increased at the light traffic road such as park, farm, forest, golf course etc.. Because the leaving of weathered soilcrete can be returned to the surrounding nature without pollution.
FGC hardening agent properly mixed with Fly ash, Gypsum and Cement was used for early stabilization of unconsolidated soil. The treated soil is clay, which is spread nationwide. And experimental design was applied to make the reference mixtures table of soilcrete. The design of experiment is the system by which one can efficiently and reliably evaluate all possible methods considered for a particular objective, that is, this system consists of the method of laying out calculations and experiments, data analysis method, and rationali~ationof the characteristic values. In this study, we will use the method of carrying out efficient response analysis by using central composite orthogonal arrays for numerical analyses in, for example, design calculations and planning calculations. As long as the conditions are given, numerical values, computer calculations are possible no matter how complicated the problem is, but if one wishes to find the output changes when the initial conditions, boundary conditions, system parameters, etc., are varied. The number of conditions becomes enormous and the calculation is almost impossible. Therefore, orthogonal arrays can be used for more variables. Through the response analysis of the relation between content of FGC hardening agent and water content of treated soil, surface value of workability index or strength trend can be plotted in 2 or 3 dimensional. From this statistical method we could make the reference mixtures table for the strength of 50 1 50kg/cm2 soilcrete (Japan Cement Association 1984).
-
577
2 EXPERIMENTAL DESIGN 2. I Ma~erials There are two typical hardening agents used in the shallow and deep soil mixing method, cement and lime. In the past, quicklime as dry hardening agent was used to improve soil, but ordinary Portland cement, as slurry, is now the primary agent. Additives, which have a solidifying admixture or liquefying effect, tailored to the soii or cement blended with pozzolanic materials like blast furnace slag, fly ash, gypsum, etc. can be used. In particular, organic soils are very difficult to treat because soil acidity can affect the hydration reaction of ordinary Portland cement. In this case, special blended improved cements should be used. The mixture of FGC hardening agent consists of Fly ash, Gypsum and Cement. It makes the fibrous ettringite hydrates, which is especially effective for the hardening of organic and cohesive soil. The typical properties of FGC hardening agent as shown in table 1.
admixture selection. The most commonly used dosage of hardening agent lies between 100 and 200kg/rn3 of native soil and can reach, in some cases, 300kg/m3 over. The.. achieved unconfined compressive strength (qtJ varies between 10 and 50kg /cm2. But because the target of compressive strength is 50kg/cm2 over in this study, the dosage of FGC stabilizer lay between 100 and 300kgim'. In this study, we selected the yellowish clay that is found nationwide. The typical properties of the selected sample were shown as in table 2. Tabie 2. The typical properties of soil Chemical Composition Ratio(%)
SO3 Ignition loss
Table 1. The typical properties of FGC hardening
7.5
Physical Properties
Gravity Class
2.45 SP-SM
Chemical
CaO
1 2 3
31.7
Si02
I
44.3
MgO
2.3
NazO
1.1
KlO
0.8
SO3
8.5
Ignition
1.5
1
4 6 8 10
12 16 20 24 30 40 60 80 100 D5o
1 I
12 21 29 37 47 57 65 72 79 85 87 94 99 100 100 100 6.9
2.1.1 Soil materials Achieved unconfined compressive strength of improved soil by in situ mixing with a hardening agent depends on several parameters such as soil characteristics and type, quantity and mixing ratio of hardening agent. Detailed soil investigation and laboratory tests to evaluate the water content and sand, silt, clay and huinus content of the subsoil should be done before
2.1.2 Flowing soil-cement u~mjxture In concrete technology, flowing concrete admixtures called superplasticizers have come into common use to improve the workability of fresh concrete and reduce water content. Recently, chemical admixtures like superplasticizers and sand are also used for improving the consistency for soil-cement. In this study, as flowing agent of soil cement, standard fine sand for mortar and lignin sulfonic acid-based dispersion agent for cement was used . 2.2 Experiment ~ e t ~ o ~ 2.2.1 Mixing Soilcrete Hardening agent, clay, water, fine sand and superplasticizer are put into the mortar mixer at the same time to make soilcrete. It takes 30 seconds to mix soilcrete with low speed, and it continues mixing soilcrete with high speed for 30 seconds. After leaving the mixed fresh soilcrete for 90 seconds, it continues mixing soilcrete again for 90 seconds. 2.2.2 Makirzg and measurir?gspecimens After mixing soilcrete, slump should be immediately measured using mortar slump cone to estimate the consistency of soilcrete. And the 50 X 50 X 50mm cubic specimens are made. These specimens are placed into the moisture chamber set with the 23 k 1 "C and 95%RH over. The compressive strengths of the cured specimens are measured at the 3, 7 , and 28day aging. 578
Table 3. Design matrix and test results Exp.No. (n)
.
Superplasticizer
Fine Sand
x,
ratio
(so11 " I ) 0.5
c
x2
FGC agent
X;
x %,
Slump
kg/m3
(cm)
Compressive Strength (kgicm') .
3days
7days
28days
1
-1
-1
1.0
-1
100
2.5
8
24
2
-1
0.5
-1
1.0
1
300
1.8
46
88
129
3
-1
0.5
1
2.0
-1
100
4.5
10
25
35
4
-1
0.5
1
2.0
1
300
1.9
53
111
158 55
35
5
1
1.5
-1
1.0
-1
100
3.5
25
4
6
1
1.5
-1
1.0
1
300
4.8
61
114
173
7
1
1.5
1
2.0
-1
100
6.5
24
38
52
8
1
1.5
1
2.0
1
300
5.1
65
124
186
9
0
1
0
1.5
0
200
4.9
21
34
79
10
0
1
0
1.5
0
200
6.0
18
31
67
11
-0.5
0.75
0
1.5
0
200
3.3
21
35
85
12
0.5
1.25
0
1.5
0
200
4.8
27
48
108
13
0
1
-0.5
1.25
0
200
2.9
31
57
115
14
0
1
0.5
1.75
0
200
4.1
25
49
95
15
0
1
0
1.5
-0.5
I50
3.2
23
41
65
16
0
1
0
1.5
0.5
250
2.9
43
63
141
2.3 Experiment and results The objective of experiment is to determine the reference mixtures table of FGC treated soil. In order to be more economical and effective, factors and levels are arrayed by the central composite design. The fine sand lay between 0.5 and 1.5 weight ratio to unit soil, superplaticizer lay between 1.0 and 2.0 weight percent to cement, FGC hardening agent lay between 100 and 300kg/m3, and the levels are 5 respectively (Park 1995). As a result, the number of design matrix consist of 8 control points, 2 center points and 6 axial points, that is, come into 16 mixes. The mixture and test result of each mix are shown as in table 3. 3 ANALYSIS OF RESULTS 3. I Correlation a i d regression analysis
3.1. I Correlation analysis In order to find the correlation between dependent variables and properties acquired, we analyzed the data of table 3. The results were shown as in table 4. From table 4, the slump of workability index increases in some degree with the increasing content of fine sand, superplasticizer, but it decreases a little with the increasing content of FGC agent. While the coefficient between the compressive strength of treated soil and the content of FGC agent is correlative in plus. But the compressive strength of treated soil was decreased a little with the increas-
Table 4. Matrix of correlation coefficient Slump Fine sand Superplasticizer FGC agent Slump Compressive 3days 7days strength 28days
0.63 0.38 -0.23
Comp. Strength 3days 7days I28days 0.31 0.20 0.22 0.05 0.07 0.05 0.86 0.85 0.93 -0.13 -0.18 -0.17 0.98 0.96 0.93
I
ing slump. Because it is difficult to get the optimum compaction at treated highly workable soil. 3. I .2 Tlie regression model Response surface analysis is the multiple regression models that can explain several independent variables on the 2-dimensional or 3-dimensional coordinates through statistical testing and estimation. Then both the independent and dependent variables are quantitative. The response relationship between the dependent variable y and independent variables X I , x2, ..*'.., xk is the 2nd order polynomial regression model, and the equation is shown as in formula (3.1).
579
The possible number of independent variables is 10. Each variable is independent and come into E-I? (0, 2).Therefore the fitted response surface througli the least square method is expressed as shown in formula (3.2). k
y = PO 4 - y p , x i 2f
-yp,xx,
From table 5 , the coefficient of determination R2 for the fitted slump regression equation is high in some degree. While the testing results for the coefficient of regression (0)- p (33) are given in table 6.
(3-2)
OI
r=O
From the equation (3.2), if design matrix, data vector, regression coefficient vector are specified as X , y , ,L? respectively, matrices can be formed. And then regression coefficient ,L? can be calculated through the least square method by the formula (3*3).-
x=
1 1 1 .. .
XI1 x21
f
Xln X 2 n
XI2 X 2 2
XI3 X 2 3
.. .
. ..
Estimated Coefficient
Standard eFiOr
Critical value for T test
/B0,:constant
4.05
0.39
10.51
PW)
-0.58
0.36
- 1.59
l m
-2.02
2.37
-0.85
Xkl ... XkZ
. ..
XX-3
. ..
Xkn
p = (XIX)-’x’y
Y=
(3.3)
X’is transpose matrix, (x‘Y)“ is inverse matrix that design matrix Xmultiplies transpose matrix X‘. From the table 4, the effective factors of slump are determined as fine sand, superplasticizer, and FGC agent. And the effective factors of compressive strength are determined as fine sand and FGC agent. The regression coefficients were calculated by equations (3. l ) - (3.3), and the accuracy of regression coefficients were tested as follows (Park 1987). 3.1.3 Testing the accuracy for the slump regression equation The results of the variance analysis for the slump regression equation are given in the table 5.
Table 5. Analysis of variance table for the slump re-
s q m es
Coefficient of regression
frcedom
2 53
From table 6, the effective factors of T test are ,8 (,)(fine sand), ,8 (z)(superplaticizer), ,8 ( 2 3 ) (interaction between superplasticizer and FGC agent), 19(3) (FGC agent), ,8 (13) (interact~onbetween fine sand and FGC agent) in order. Therefore, the relationship between output y and factors is given by the following equation. y
=
4.06 4 1.17 X (Sand) + 0.71 X (SP) 0.42(FGC) + 1.98 x (Sand)2 + 0.15 X [(Sand).(SP)] + 0.40 X [(Sand).(FGC)] 0.22 X $SP)2-0.58 x [(SP)*(FGC)J 2.02 X (FGC) (3.4)
3.1.4 Testirzg the accumcjifor tlie compressive strength regression e ~ ~ ~ ~ ~ t i ~ From table 4, superplasticizer does not have an influence on the compressive strength. Therefore fine sand and FGC agent were only selected as effective factors for the regression model of compressive strength. And the results of the variance analysis for the compressive strength regression equation are given in table 7. From table 7, the coefficient of determination R2 for the fitted compressive strength regression equation is relatively high. While the testing results for the coefficient of regression 13 (0) - p (33) are given in table 8. From the table 8, the effective factors of T test are p (2) (FGC agent), p (1) (fine sand) in order. Therefore, the relationship between output y and factors is given by the following equation.
580
~
Table 7. Analysis of the variance table for the com-
Table 8. Testing table for the coefficient of compres-
Figure l(a). Slump contour lines of relationship between sand ratio and FGC content
05
17.9
29.74
0.60
Table 9. Reference mixtures table with regard to the FGC hardening agent treated clay Design properties
10
1 5
Sand Ratio (XSOI I)
Figure l(b). Slump contour lines of relationship between sand ratio and SP content 2
Unit weight I
(C*
wt%)
1
05
10
1 5
Sand Rat io(XSoi I )
Figure 2. Compressive strength contour lines of relationship between sand ratio and FGC content 3.3 Suggestion Reference Mixtures f o r the FGC Treated Clay
y
=
4
260
1.2
1.3
6
260
1.5
1.5
93.4 + 14.2X(Sand) + 59.6x(FGC) 8.1 X(Sand)' + 4.4 x[(Sand) (FGC)] + 17.9 x(FGC)' (3.5)
3.2 Contour lines on the response surface The contour line for the slump between sand ratio by soil and the content of FGC agent or superplaticizer is given in the figure 1. And the contour line for the cornpressive strength between sand ratio by soil and the content of FGC agent is given in the figure 2.
If the figure 2 would be superimposed on the figure 1(a), from the combinations between compressive strength and slump, the content of FGC agent and sand ratio pairs could be determined by the superimposing diagram. And if the figure 2 would be superimposed on the figure l(b), the content of FGC agent, sand ratio and superplaticizer ratio pairs could be determined as the same way. As the results, the combinations of mixtures with regard to the design strength of the FGC treated clay are suggested as table 9. Table 9, with respect to the pairs of the mixing strength 50kg/cm2 combined with the slump 6cm(hereafter, stands for 50-6) and 100-2, the mix581
After FGC hardening agent was added 200kg/m3 to the testing clay, the hardened soilcrete 3days aged was taken by SEM instrument for observation of the hydrated textures. And to compare the FGC treated soil, ordinary Portland cement treated soil specimens were also molded. The magnification of SEM photographs is 10K.
Because FGC hardening agent forms much the fibrous ettringite hydrates from the early ages, we can conclude that FGC agent will be more effective to speed up the consolidation of the high water or high organic soil like marine clay. When FGC hardening agent was agent was mixed with weathered soil, the unconfined strength by curing time was increased in proportion to the amount of hardening agent used. In addition, the unconfined strength showed its peak at 20% of water content. It means that the unconfined strength is influenced by compaction. When FGC hardening agent was agent was mixed with yellowish clay. The flow value was more influenced by fine aggregate than by superplasticizer when it was not hardened. The unconfined strength increased in proportion to the unit amount of hardening agent. According to the results of the in-situ application, the unconfined strength on site was higher that the required strength. It is assumed that adjustment of mixing ratio and application method can make highstrength design possible.
4 CASESTUDY
REFERENCES
Project name: Pavement of trail near the Suwon Stream. Modification of design: Change concrete pavement to earth-friendly soilcrete pavement Purpose: To provide the citizens with a convenient and cozy rest area, the Suwon Stream and its surrounding area are renovated into earth-friendly, clean and beautiful ones. Required Equipment: Roller, compact, dump trucks, backhoe loader and soilcrete mixing system Unconfined strength of soilcrete on site: 9 specimens were formed on site to measure 3, 7, and 28-days strength. According to the results from the unconfined tests, 3-day strength was 52kg/cm2, 7-day strength was 103kg/cm2 and 28-day strength was 187kg/cm2. 28-day strength was measured7 to be higher than the required strength of 180kg/cni-
ASTM PS23 1995. Provisional Sturzclurd Guide for Use of Coal Combustion FIJ) Ash in Structural Fills: 1-10. Japan Cement Association 1984. Ground iniproveinen t man ual by cernent tjpe sta hilizer Park, S.H. 1987. Regiwsion anulysis: 52 1-57 1 : Daeyoimgsa.
tures were determined by assumption because they could not be determined on the response surface. As the slump increases, sand ratio and the content of superplasticizer increase. The content of FGC hardening agent is controversial. That is, as the slump increases, the content of FGC agent increases at the low strength of the about 50kg/cm2, but the content of FGC agent decreases slightly at the high strength of the about 1OOkg/cm2 inversely. These controversial results are concluded to cause to the low accuracy of the estimated slump regression model because the experimental scope is relatively wide. 3 . 4 .Examination of Hydrated Textures
5 CONCLUSIONS Through the statistical analysis method, we found the order of having an influence on the workability of FGC hardening agent treated clay was sand ratio, superplaticizer, interaction of superplasticizer and FGC agent, FGC agent in sequence. And the order of having an influence on the hardened soilcrete was FGC, sand ratio in sequence. Using the response surface regression analysis in some degree could make the practical reference mixtures table, and if the experimental scope was divided into more detail, we could conclude to get the more accurate reference mixtures table.
582
Coastal GeotechnicalEngineering in Practice, Nakase (I: Tsuchida (eds)02000 Baikema, Rotterdam, ISBN 90 5809 151 I
A study on the ground improvement of year 2002 World Cup stadium site using hardening agent B. S.Chun & H. S.Kwon Department of Civil Engineering, Hanyang University,Korea
K. H. Kim Department of Civil Engineering, Suwon Science College, Korea
ABSTRACT: Football stadiums for 2002 World Cup are being constructed at several places in Korea. One of the stadiums under construction has some problems of soft ground and irregular soil strata. It is considered that shallow mixing method using hardening agent has the advantage of securing trafficability, reducing the working time, solving the problems of lack of replacing soils, etc. In this paper, three different types of representative hardening agent in Korea were used to achieve the optimum. The execution on shallow depth with hardening agent and the property of laboratory improve men^ strength with ground soil were analyzed by the type of hardening agent, curing days and water content. The unconfined strengths on site with various mixing ratios were estimated from the results of the laboratory tests. It was estimated that strength increase on site was higher than the required strength.
1 ~TRODUCTIO~ As the industry develops recently, the population gets larger and the economy grows, nationwide developments have become necessary in many parts such as housing, tr~sportation,and industrial complex. The development in relatively difficult ground conditions like landfill area became unavoidable. Therefore, the treatment of soft grounds emerged as a significant matter. Effective and economic ways of treating soft ground were highly required more than ever before (KICT 1988). Replacement, dewatering, hardening and compaction are the major principles of soft ground treatmenl. The importance of ground improvement method using the hardening agent is emphasized recently because it provides advantages of shortening the working time, securing the required strength, solving the problems of lack of replacing soils, etc (Bergado 1996). As the necessity of construction on extremely soft grounds increases, it becomes difficult to gain the required strength of the ground only with the cement and lime. Hardening agents are widely used to stabilize the ground on which relatively lightweight structures such as retaining walls, tanks and houses are placed (Housing Research Institute 1998).
Recently, const~ctionson soft ground have increased. It causes lots of problems such as poor trafficability, high cost, environmental problems, etc. Therefore, the shallow improvement methods using hardening agents have been applied to the improvement of landfills and highways to solve those drawbacks. In Korea, football stadiums are being constructed nationwide for World Cup. This paper is a study on a case history of the soft ground improvement on one of the stadiums. The ground condition is weak and with irregular soil strata. Prolonged working time, lack of replacing soils and dumping problem of poor soils are foreseen on this site. Three representative harden~ngagents applied to verify the effectiveness of the soft ground improvement. Unconfined compression tests were executed on specimens with various mixing ratios and curing times of hardening agent to assess the stabilizing ability. The effectiveness of the hardening agent was verified after the stability methods were applied to the construction site. 2 SHALLOW MIXING METHOD Shallow improvement method with hardening agent improves the soft ground by mixing agent with the soil to the depth o f 1-2m. Then the ground gets
583
hardened. This method can be used for both temporary and pernianent purposes such as securing trafficability, increasing bearing capacity and iniproving durability, permeability and strength. The proposed method can help the soft ground acquire the required strength by adjusting the mixing ratio of hardening agent, and has following advantages; shortening the working time, cost reduction, efficiency in small or middle scale constructions. Compared to replacement method, the amounts of excavated soils are reduced. This method can also be applied where replacing soils are not easy to achieve (Housing research institute 1998). 3 MIX PROPERTIES AND LABORATORY TESTS 3.1 Ground condition Sieve analyses were carried out on the soils from World Cup stadium site and a few representative soft alluvial soils, which were collected from U1San, Mok-PO, Bee-In, Jin-Hae and Kun-San. The particle-size distribution curves for those six kinds of soils are shown in Figure 1. The soil strata of the ground under consideration have irregular and various soil compositions down to lOmeters from the planning ground level. Laboratory tests were performed to obtain the properties of the representative soft clayey soil collected from the construction site. The test results are shown in Figure 1 and Table 1.
Figure I . Particle-size distribution curve in ground yt=
weight of (hardening agent + water +soil) volume of niold
(1)
The weight of hardening agent and soil is dry weight that doesn’t consider natural water content. A, B and C type hardening agent were used to tests and then mixing ratios were 0, 3, 5 , 7, 9, 10, 15,20%. The Equation(2) determines hardening agent mixing ratio. Mixing ratio =
weight of hardening agent XI00 weight of (hardening agent + water + soil)
(2) In case the degree of saturation is loo%, water content is determined by Equation(3). In case wet unit weight is 1.7 t/m3, water content was calculated to 51%, and wet unit weight 1.9 t/m3 to 31%. Wnter content
3.2 Applied hardening agent The three different hardening agents named A, B and C were used as admixtures and their chemical composition are shown in Table 2. Quantity measurement was carried out by means of XRF(X-ray Fluorescence Spectrometer). Materials such as H 2 0 , C O 2 and SO3 were counted as the loss of ignition.
=
weight of water weight of (hardening agent
+ soii)
XI00
(3)
Cylindrical specimens were made of 5 x 10 cm (diameterxheight). Curing periods were 3,7 and 28 days, curing temperature was 2 0 f 3°C in ordinary temperature, and specimens were cured in humidity oven. Table 2. Properties analysis by XRF
3.3 UnconJined compression test Percent o f total weight(%)
3.3.1 Preparation of specimens Specimens were made to have the unit weight of 1.7t/m3 and 1.9t/m3, which are made with dried hardening agent with disturbed soils from the field under the assumption of 100% saturation.
Hard-
ening type
A
Soil Classification
Specific
Specific
Si02
A120,
Fez03
TiOz
ity
3.06
MnO Igni-
grav-
MgO
K20
20.72 3.36
CaO
Total sum
Na2O
PzOj
4.75
2.G9
0.21
0.05
46.74
0.98
0.02
0.19
20.89
100.00
tion loss
44.20
584
4.1 Required laboratory unconfined strength The improvement depth of stadium ground was estimated to 1.5m by Bousinesq method, and the required expecting strength was 10 kg/cm2 for acquiring long term allowable bearing capacity considering that the strength reduction ratio (field strength/laboratory strength) is 0.5 and safety factor is 2(Cement Association 1994). 4.2 Imp rovem eizt strength clz ara cter ist ics Figure 2. The unconfined strength after 3-day curing
Figure 3. The unconfined strength after 7-day cur-
Figure 4. The unconfined strength after 28-day curing 3.3.2 Testing method The unconfined compression test using constant strain method (1 .O mm/min) was performed by Korean Standard(KS F 1996).
4 RESULTS AND ANALYSES Considering surcharge load and section property in the main stadium ground, the required improvement strength was estimated. According to types of hardening agent, curing period and water content, the property of compressive strength and the effect of ground improvement were analyzed.
The effect of improvement in field(7-day curing) and laboratory(3, 7, 28-day curing) respectively was analyzed by unconfined compression test. 4.2.1 Compressive strength-types of hardening agent According to the mixing ratios, improvement effects, which had been acquired in a laboratorymixing test, appear Figure 2-4. In case water content was 51%, C type hardening agent’s property was better than other hardening agent’s, and in case water content was 31%, A type hardening agent was somewhat better. In case water content of C type hardening agent was 51% and that of A type hardening agent was 3 1%, required improvement strength was satisfied with 10 kg/cm2 in mixing ratio 5%. 4.2.2 Compressive strength - curing time A and C type hardening agent showed excellent performance among three hardening agents. According to the specimens formed with A and C type hardening agent, the longer the curing time is, the higher the compressive strength is. As shown in Figure 3, 4, comparing the 7-day strength and 28day strength, the increment of the unconfined strength of A type hardening agent is appeared to be larger than that of C type hardening agent. 4.3 Improved strength increase on site
A type hardening agent was applied to the site at the mixing ratio of 5%. The treated ground was well mixed and compacted. The 7-day strength was 5.7, 6.3, 7.1 and 6.0 kg/cm2 . The maximum dried unit weight was 1.89 t /m3. The required strength on site was 5kg/cm2. Even though the strength is expected to decrease by 0.5kg/cm2, the mixing ratio of 5% is enough to satisfy the required strength. 5 CONCLUSION The effectiveness of shallow mixing method by hardening agent and the results of the analysis are as follows.
585
The advantages of trafficability, increase of durability, shortening the working time, financial benefits and solving the problems of replacing soils shortage and environment, can be acquired when the proposed method is applied. According to the laboratory tests on the specimens of three representative hardening agents at 51% water content, C type hardening agent showed better performance compared to other hardening agents. At 31% water content A type hardening agent showed a bit better performance. The 7-day strength of C type hardening agent was 10.04 kg/cm2 at mixing ratio of 5% hardening agent and 51% water content. The 7-day strength of A type hardening agent with 31% water content showed 1 1.43 kg/cm2at the same mixing ratio. Both results satisfy the required strength of 10.0 kg/cm2 . The longer the curing time is, the higher the strength of the improved soils by hardening agent. Comparing 7-day strength and 28-day strength, A type hardening agent showed higher strength than C type hardening agent. A type hardening agent showed higher strength at 3 1 '?A0 water content than at 5 1 % water content. C type hardening agent showed higher strength at 5 1% water content than at 3 1% water content. In-situ installation was carried out based upon the laboratory tests. The treated ground was well mixed and compacted. The 7-day strength was 5.7, 6.3, 7.1 and 6.0 kg/cm2. The mixing ratio of 5% is enough to satisfy the required strength of 5.0 kg/cm2 .
REFERENCES Bell, F.G. 1993. Engineering Treatmeiit of Soils: 240-267. E&FN SPON Bergado, D.T. et al. 1996. Soft G ~ O UImprovernent II~ i n ~ O U ~ I C ~ Jand I C I otlier emiroiimeiits: 234-269, ASCE Press Housing Research Institute, Korea national housing corp. 1998. A study or1 the upplying stardw-cl of softground iniprovenient ~izethotl.3- 142 Japan Cement Association 1994. GTOZLI~CE improvement ~ ~ i ~ i iby t acement l t-ype stabilizer: 1-87, 381-395 Korea Institute of Construction Technology 1988. A study on the softground wit11 shallow stabilizatiorz niethod:33-65 KS F 2314. 1996. Testing method for- tinconfined compressive strength of cohesive soils: 1-5 Noble, D. F. 1968. Reactions and Strength Development in Portland Cement-Clay Mixtures:39-56. H.R.R NO. 198
586
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Compression characteristics of sludge cake H. Fujii & S.Nishimura Faculty of Environmental Science and Technology, Okayama UniversiQ,Japan
YWakitani & T Inoue Graduate School of Natural Science and Technology,Okayama University,Japan
0.Seino & N. Nishino Chugoku-ShikokuAdministration Office of Ministry of Agriculture, Forestry and Fisheries, Okayama, Japan
ABSTRACT: The laboratory consolidation tests were performed to investigate the compression characteristics of the dewatered sludge as a construction material. The sludge is dredged from the bottom of a lake. After the dehydration, it is broken into fragments. The fragments shows different compression behavior from that of ordinary undisturbed clay. The remarkable point in our research is: The e-logp curves of fragments have two kinds of compression index in the normally consolidation region. The first compression index has linear relation with initial water content, while the second one is almost constant.
1 INTRODUCTION Dredged sludge has been used as banking material at Kojima Lake of Okayama Prefecture, which is located western Japan. Kojima Lake was created nearly forty years ago by enclosing a part of Kojima Bay by a dyke. For the past forty years, a large amount of sludge has accumulated. The Ministry of Agriculture, Forestry, and Fisheries planed a project to clean up the lake, and has been dredging the bottom sludge. The dredged sludge is mixed with a coagulant, then is forcefully dewatered to yield what is called “sludge cake”. Thus obtained sludge cake is used as banking material (Oda, 1997). Embankment using sludge cake has following problems. Since the sludge cake immeddiately after banking keeps a large granular shape, the continuum theory is not applicable. With increase of compressive load and the passage of time, a granular shape begins to break. Finally, the sludge cake changes into continuous substance. Embankment using sludge cake is saturated below the water table and unsaturated above that. As above mentioned, two characteristics of deformation behavior in sludge cake are clarified in this study, (1) behavior from granular to continuous state, (2) behavior in saturated or unsaturated state. In order to study compression characteristics of the sludge cake, two kinds of consolidation tests were performed. They are the standard oeudometer test and the large-scale consolidation test. The tests were carried out using three kinds of samples. One was undisturbed original sludge cake. The others were miniature sludge cakes; (a) strained samples made of the sludge cake using a sieve, and (b) com-
pressed samples prepared by cutting after remolding and recompression. The miniature sludge cakes made for experiments to simulate compression behavior of granular sludge cake used as a banking material. Unsubmerged tests and submerged tests were performed for comparison between compression behavior of sludge cake in unsaturated state and that in saturated states.
2 SAMPLE AND TEST METHOD 2.1 Dredged sludge and sludge cake The physical properties of sludge differ in each dredging zone in the lake. Their average properties are shown in Table l.(Fujii, 1997). The sludge in Kojima Lake is dredged at two zones, the left shore area and the right shore area, using respective high concentration dredgers. The dredged sludge is sent to a dewatering yard by sludge pumps, where the sludge is held for two days to complete sedimentation. After removing the supernatant, a coagulant is added to the concentrated mixture to T a b l e 1. Physical properties of sludge for dredging in t w o regions of t h e lake Properties
Bottom of lake Waterway
Specific gravity of soil particle G s Water content w (5%) Grain size distribution (96) Sand fraction Silt and clay fraction Liquid limit Wf. (%) Plastic lim w/’ (%) Plasticitv Index 1,.
587
2.58 200
2.67 100
5 95 120 40 80
45 55 70 30 40
Table 2. Physical properties of sludge cake ProDerties
Cake A Cake B Cake C Cake D
2.67 41 N,P. N,P. N.P.
Specific pvity of soil particle Gs
Water content Liquid limit Plastic limit Plasticitv Index
w (5) w /(%) w~(Ft)
ZP
2.75 75 N,P. N,P. N,P.
2.64 78 103 61 42
2.61 110 154 66 88
agitate the two. Then, the mixture is dewatered by a filter press machine to obtain sludge cake (Oda, 1997). The standard amount of additive coagulant is 60 kg of poly-aluminum chloride and 12 kg of slaked lime per 1000 kg of dried sludge. The filter press is operated under pressures from 600 to 700 kPa for 100 minutes. The sludge cake immediately after dewatering by the filter press has dimensions of 2.0 m in diameter and 10 cm in thickness. The sludge cake has broken into small blocks from 5 to 15 cm in diameter during transportation (Fujii, 1997).
having 2 mm and 0.425 mm of opening, respectively. (b) The compressed sample; a sludge cake was recompressed after remolding, then was cut into a piece of 5 mm square using a wire saw. A large-scale consolidation apparatus made the recompressed sample. The apparatus comprises an acrylic cylinder having 18 cm in inner diameter and 60 cm in height, and the load is applied by weight or compressed air (Fujii et al. 1999). The samples were recompressed under three levels of pressures (denoted asps), 160, 320, and 640 kPa, for 60 minutes of compression time. The amount of each sample was adjusted to form about 2 cm in thickness after the compression. Cake A and Cake B were used as the specimens (1) sludge cake. The specimens (a), strained samples, were prepared from Cake C and Cake D, separately. The specimens (b), compressed samples, were prepared from Cake C. 2.3 The method of experiment Six kinds of experiment were carried out as Series 1, 2, 3, 3w, 4 and 4w. Series 1 : The standard consolidation tests were performed for sludge cakes. Consolidation pressures (p) were 10, 20, 40, 80, 160, 320, 640, and 1280 kPa. Series 2 : The standard consolidation tests were performed for strained samples. Since the primary consolidation of strained samples end in a short time, the consolidation test was carried out with 1 hour of intervals for stepwise loading. Series 3 and 3w : The consolidation tests for the compressed samples were conducted using the
2.2 Samples Four kinds of sludge cake were sampled for testing at the site. These cakes were dewatered from the sludge dredged at three points in the left shore area and one point in the right shore area, each of which is named as Cake A, Cake B, Cake C, and Cake D, respectively. Table 2 shows average physical properties for each cake. Their water contents are 41 to 110%, and the liquid limits are 103 to 154%. Each cake has different values of water content and liquid limit (Fujii, 1999). The three kinds of samples were prepared for test. Namely,one is original sludge cake, which is undisturbed sample, and another is miniature sludge cake, which were made from disturbed sludge cakes. (1) The sludge cake: a specimen was prepared by cutting the cake into a piece of 6 cm in diameter and 2cm in height. (2) The miniature sludge cake: (a) The strained sample; a sludge cake was remolded, then it was strained using two kinds of sieves successively, each
Cake
Specimen
Test
A - Sludge cake
- Series 1 B - Sludge cake - Series 1 { Strained sample Series 2 Compressed sample Series 3 and 4 Series 3w and 4w D - Strained sample - Series 2 ~
Figure 1. Combination of Cakes, Specimens and Tests Table 3. Results of consolidation tests Series Number of specimen
Initial water content wo
Initial void ratio er,
%
YO
1 1 2 2 2 3 3w 4 4w
5
40- 42 62- 78 127-149 115-121 205-208 56- 73 56- 73 55- 73 54- 76
Degree of saturation Src,
1.19-1.23 1.83-2.08 3.35-3.87 2.99-3.19 5.50-5.52 3.77-5.53 4.24-5.58 4.16-4.63 4.03-5.62
*: Stress at first large curvature point,
90-100 91-100 96-100 100 100 31- 43 31- 41 33- 44 30- 36
Compression index
G
Gl
0.23-0.28 0.48-0.80 0.67-0.90 0.54-0.64 1.46-1.63
Pc*
Pd**
kPa
kPa
50- 55 80-1 10
1.54-1.82 1.80-2.53
**: Stress at second large curvature point
588
c 2
0.87-1.12 0.82-0.94 0.96-1.15 0.87-1.02
7- 13 5
<
65-200 10
above-described large-scale consolidation apparatus. In that case, the values o f p were 10, 20,40, 80, 160, 320, and 640 kPa with 1 hour of intervals for stepwise loading. Comparison was given between an unsubmerged test (with Series 3) and a submerged test (with Series 3w). Series 4 and 4w : The casep = 5 kPa and 7 kPa were added to those of Series 3 and 3w. The compressed samples were lightly filled in the
consolidation cell to reach the initial sample height to about 2 cm. Figure 1 shows the combination of cakes, specimens and tests. The miniature sludge cakes were prepared for simulating the behavior of granular sludge cake used in situ as banking. After all, in Series 3, 3w, 4 and 4w, a granular shape of the specimen affected consolidation characteristics. In Series 2, behaviors of the specimens was like continuum substance.
3 TEST RESULTS 3.1 Relation between void ratio (e) and consolidation pressure (p) Table 3 shows outline of the test results. Figure 2 (a) shows typical e-logp curve for each series. In Series 1 for sludge cake, the overconsolidated region and the normally consolidated region are clearly identified. The consolidation yield stress (p,) ranges from 50 to 110 kPa. For both miniature sludge cakes such as strained samples (Series 2) and compressed samples (Series 3), the e-logp curves become straight lines over the whole region. While for the compressed samples, Series 4 shows two large curvature points. Figure 2 (b) shows typical e - l o g p curve for eachps value of compressed samples. In Series 3, the curve becomes straight line over the whole region for everyPS. In Series 4, there are two large curvature points for everyps. Figure 2 (c) shows typical e-logp curves of compressed samples for unsubmerged test and submerged test. In Series 3w, the e-logp curve becomes straight line over the whole region as in the case of Series 3, and the lines of both series are almost in parallel to each other. In Series 4w, there are one large curvature point like a second curvature point in the case of Series 4. The first curvature point may be smaller than 5kPa which is the first consolidation pressure. In Series 4 and 4w, the e-logp curves have large curvature points. As shown in Figure 3, the consolidation pressure at the first point is defined as p,,
Figure 2. e
+
Figure 3. Definition o f p candp,
logp curves
589
and that at the second point is defined asp,. Compression index (Cc) is gradient of e-logp relation. In the region o f p >p,, C, is divided into two sections with the boundary of pd.The compression index in the region o f p
Figure 5. Relationship between stress at second large curvature point Pc and initial water content w oin Series 4
smaller than 10 kPa which is the first consolidation pressure. When first consolidation pressure is too large, a granular shape of specimen may break at first loading. The state that Series 2 gives straight line over the whole region resembles a case of ordinary normally consolidated clay. The results of Series 4 and 4w tests correspond to compression characteristics in the case that sludge cake is used for embankment. That is, in a region of p
Figure 6. Relationship between compression index C, and initial water content w,
590
from 65 to 200 kPa. Both p , andp, increase with the increase of p , value. In Series 4w, p , and pd are smaller than those in Series 4. The value of p , in Series 4w is not found and it might be smaller than 5 kPa, which is the first consolidation pressure. The water content of compressed samples decreases whileps increases. In Series 4, as shown in Figure 4 and Figure 5, p , and p,, have linear correlation with the initial water content (w,)of the specimen. Thus the phenomenon is approximated by equation (1) and equation (2),giving correlation coefficients of 0.902 and 0.972, respectively.
p , = -0.284 w,+ 28.5 pd = -7.47 w,+ 606.9 In Series 4w, p , is smaller than 5 kPa and pdis 10 kPa. 3.3 Compression index (CJ Figure 6 (a) and (b) show the relation between Cc and wo.Figure 7 (a) and (b) show the relation between C, and initial void ratio (e,), for individual Series.
Table 4. G-w, and Cc-eolinear relationshiDs for each series Series
Specimen
1 1 2 2 4 4w
Sludge cake Sludge cake Strained sample Strained sample Compressed sample Comoressed s a m d e
Correlation
Coefficient of correlation
G =0.014~o-0.312 C =0.533&,-0.368 C =0.01 lwo-0.707 G ~0.389ffi-0.615 Cl =0.014w0+0.798 G1=0.031wo+0.148
0.990 0.954 0.986 0.987 0.951 0.989
C, in Series 1 and 2, and Ccl in Series 4 and 4w have linear correlation with w, or e,, respectively. There are many reports on clay about the correlation between compression index and physical properties (e.g. Ogawa et al. 1978), and a similar correlation is observed in the sludge cake. Table 4 gives approximation equations and correlation coefficients. For the sludge cake of Series 1, Cc of Cake A and Cake B ranges from 0.23 to 0.80. Although the test results were obtained from different kinds of cakes, the correlation coefficients between C, and w,,and between C, and e,, are as large as 0.990 and 0.954, respectively. For the strained samples of Series 2, C, of Cake C and Cake D ranges from 0.54 to
Figure 9. Relationship between compression index C, and initial void ratio e,,
Figure 7. Relationship between compression index Cc and initial void ratio e,
591
sludge cake, there observed compressibility close to that of ordinary normally consolidated clay. The compression index (Cc)has a linear correlation with the initial water content (w,)or with the initial void ratio (e,). (3) For the compressed samples, another kind of miniature sludge cake, there observed two large curvature points on e-logp curve. The two points @, andp,) have a linear correlation with the initial water content (w,). (4) For the compressed samples, the first compression index (Ccl)afterpchas a linear correlation with the initial water content (w,),and the second compression index (Cc2) shows almost constant value. The phenomenon presumably comes from a significant influence that the samples are in a granular shape during the initial stage of consolidation. ( 5 ) The tests using the compressed samples showed that the submerged test gives very small values of p , a n d p , compared with the case of unsubmerged test.
Table 5. Compression index G of compressed sample
PI *
Submerged or Unsubmerged
plSPc** Unsubmerged plSP,** Submerged p l >>P, ** Unsubmerged pl>>P,** Submerged
Compressibility index
n&,,*** G1=0.014w,+0.798 Gl=O.O31w,+0.l48
D)D'i***
62=0.95 G2=0.95 62=0.95 C2=0.95
* : Consolidation pressure at the first loading ** : Stress at first large curvature point of e-logp curve *** : Stress at second large curvature point of e-log p curve
1.63. For the strained samples, the correlation coefficients between Cc and w,,and between Cc and e,, which were derived by overlaying the results of tests with different kinds of cakes, show as large as 0.986 and 0.987, respectively. C,l for compressed samples of Cake C ranges of from 1.54 to 2.53. Cc2 of the samples ranges from 0.82 to 1.15. For the compressed sample, the correlation between C,l and w, of only Series 4 and 4w give linear relationship. The correlation coefficient for that case is large, 0.951 for Series 4 and 0.989 for Series 4w. Regarding Cc2of Series 3, 3w, 4, and 4w, the value is almost constant with change of w, and e, . The average value is 0.95. Figure 8 shows the relation between C, and w, for all series. Figure 9 shows the relation between C, and e, for all the series. C c l and C,2 of the compressed sample (Series 3,3w, 4 and 4w) are greater than Cc of the sludge cake (Series 1) within a similar w, range. The reason is that e, of the compressed sample is greater than that of the sludge cake. The w, value of the strained sample (Series 2) is very great in comparison with those of other samples. However, in the case that e, is roughly 4.0, Cc of strained sample coincides with Cc2 of the compressed sample. The compressed sample is considered to simulate the in-situ compression characteristics of the sludge cake as banking material. The compression characteristics of the compressed sample are summarized in Table 5. When the actual settlement prediction of the sludge cake is performed, two kinds of the compression index C,l and Cc2 need to be used.
REFERENCES Fujii, H. 1997. Compression and Consolidation on Compressed Dehydrated Sludge. Report of Research: Okayama University (in Japanese) Fujii, H. 1999. Compression and Consolidation on Compressed Dehydrated Sludge. Report on Research Project, Grant-in-Aid for Scientific Research (in Japanese) Fujii, H., Nishimura,S., Inoue,T., Shimada,K., Wakitani,Y. & Nishino,N. 1999. Deformed Properties of Dehydrated Dreged Sludge in Kojima Lake. Proc. of the Third Japan National Symposium on Environmental Geotechnology: 23-26 (in Japanese) Oda, T. 1997. The Works of Dredging of Muddy Sediment, Water Removal and Filling-up Reclamation i n Lake Kojima. Jour.JSIDRE, 65 (5) : 35-40 (in Japanese) Ogawa, Y. and Matsumoto, K. 1978. Correlation of the Mechanical and Index Properties of Soils in Harbour Districts. Report of the port and harbour research institute, 1 7 (3) : 3-89 (in Japanese)
4 CONCLUSION The consolidation tests of sludge cake samples and miniature sludge cake samples revealed the followings. (1) For the sludge cake samples, there exists consolidation yield stress point on e-logp curve, showing compressibility of overconsolidated clay ordinarily observed. The compression index (Cc)has a linear correlation with the initial water content (w,) or with the initial void ratio (e,). (2) For the strained samples, one kind of miniature
592
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, lS5N 90 5809 151 1
Effect of alkdine-earth metal cations on stabilization of loessial and clayey soil using alkalizing L. N-Gera & E E.Volkov BashNIIsfi-oy,Ufa,Russia
ABSTRACT: Results of laboratory investigations are presented to understand the effect of alkaline-earth metal cations on the interaction between loessial and clayey soils with natrium hydroxide. It is shown that the admixtures of 2.5 N and 5 N NaOH solutions influence the crystallization of new formations and increase in soil strength. The introduced admixtures decrease the natrium hydroxide concentrationtwice without changing the strength characteristicsof the stabilized soil. 1 INTRODUCTION Increase in human economic activities results in development of a large number of different technical solutions, that can change the environmental geoecological situation. Knowledge how such solutions influence the dispersed soils is actual and allows to evaluate in time the extent of risk from an ecological point of view. The influence of natrium hydroxide on the dispersed soils has been under investigation for some years in the Institute “BashNIIstroy”. The investigations show that the alkali solutions of 1 N concentration promote the swelling of the clayey soils and the solutions of more than 1 N concentration contribute to soil strengthening (Volkov 1988). The strength of soil samples increases with the natrium hydroxide solution concen~ration increase. The strength increase involves the accumulation of natrium silicates and aluminates in the interpore space, the concentration of which leads to a new phase formation - natrium hydroaluminosilicate of the sodalite type, cementing soil particles. As a result of above process, a method of alkalization is worked out that helps to solve the practical problems of strengthe~ngclayey soils as structural foundations. During alkalization, a focal zone is formed where the short term transformations of soils occur as a result of physical-chemical and t h e r m o d y n ~ cconditions. As a rule, the quantity of the injected alkali is less than the value of soil sorption. As the tests show, the alkali interacts with the soil, meaning the alkali solution does not spread beyond the local zone (Zlochevskaya 1990). The average concentrat~onof the alkali solution used for soil strengthening is a 5 N solution. From the technical and economical point of view it is more advantageous to decrease this consumption of alkali.
One method of decreasing the alkali consumption is the introduction of inorganic compounds of the alkaline-earth elements and the silicates, which have the binding properties.
Two soil types of very different geological origin can be found in Bashkortostan Republic - alluvial loam aQa with the index (number) of plasticity (difference of hurnidities on the boundary of fluidity and the boundary of rolling-out) 0.08, light silt clay dQm with the index (number) of plasticity 0.24 - have been chosen for the laboratory investigations. Also, loess-like loam pdQa with the index (number) of plasticity 0.09 was tested. Soils were unsalted, carbon-bearing and in the undi~turbedstate - macroporous. In the granulometric loam composition parts of silt prevailed. For clays the amounts of silt and clayey fractions were equal. When watering the soil samples until the state of full saturation, the significant decrease in strength was noted. Natrium hydroxide solutions of 5 and 2.5 N concentration and electrovacuum soda were used as the chemical agents. The electrovacuum soda is a product of a sodium salt electrolysis with natrium hydroxide obtained with the electrochemical method and contains 1 10 g/l of NaOH and 140 g/1 of NaCl. Twelve kinds of admixtures of the alkaline-earth elements and aluminium inorganic compounds of soluble and insoluble compounds were tested. The main criteria of the admixtures contribution eva~~ation was the ~ t r e n index ~ h of the dkalized cylindrical specimew with the dimension 40x40 m. The samples were hardened in conditions of capillary inleakage in excess content of alkali agent of
593
the definite concentration (5 N, 2.5 N of an electrovacuum soda). After 28 days of hardening the samples were uniaxial compressive strength tested, therrnographically and X-ray phase analysed, plasticity tested and particle-size analysed. 3 RESULTS AND DISCUSSION
The results of uniaxial compressive strength tests of samples alkalized with 5 N NaOH solution with an optimum quantity of admixtures were 3-5 times higher compared to results of tests of samples alkalized with the same solution of NaOH without admixtures (table 1). To define the effect of admixtures upon structure formation when alkalizing of loesslike loam with 5 N NaOH solution, the samples were thermographically and X-ray phase analized. The results of thermo-graphical and X-ray analyses showed the admixtures actively influenced the phase composition of new formation (Fig. 1a, b).
Table 1. Strength of the alkalized samples with the inorganic compound admixtures, MFa Concentration of NaOH solution, N Admixtures 2.5 N 5N loess-like clay Loessclay loam dQm likeloam dQm PdQm ~dQm Strength without admixtures 0.07 0.07 0.27 0.11 Ca(OH)2 0.50 0.15 0.9 0.28 CaC12 0.38 0.19 0.84 0.30 CaS04 0.26 0.57 Phosphogypsum 0.42 0.16 0.33 0.32 CaC03 0.17 0.57 d(OH)3 0.42 0.15 0.86 0.13 0.39 0.08 1.05 0.30 MgO 0.18 0.10 0.66 0.14 MgCh MgS04 0.23 0.66 Ba(OH)2 0.48 0.12 0.77 0.21 BaC12 0.88 0.23 2.03 2.14
Figure 1. X-ray difiactograms (a) and therrnograms (b) of loess-like loam: 1 - initial loam; 2 -loam alkalized with 5 N NaOH solution; 3 -loam alkalized with 5 N NaOH solution with Ca(OH)2; 4 -loam alkalized with 5 N NaOH solution with CaC12; 5 -loam alkalized with 5 N NaOH solution with phosphogypsum; 6 -loam alkalized with 5 N NaOH solution with MgO; 7 -loam alkalized with 5 N NaOH solution with MgCl2; 8 loam alkalized with 5N NaOH solution with Al(OH)3; 9 -loam alkalized with 5 N NaOH solution with
594
Table 2. Plasticity of loess-like loam samples with inorganic admixtures alkalized with 5N NaOH solution after 28 days of hardening
Admixtures Primary Without admixtures Ca(OH)2 CaC12 CaS04 Phosphogypsum CaC03 d(OH)3 dcl3 MgO MgCb MgSO4 Ba(OH)2 BaC12
The upper plasticity limit,unit fractions 0.29 0.43 0.37 0.47 0.40 0.48 0.47 0.42 0.44 0.44 0.42 0.44 0.44 0.39
The low plasticity limit,unit fractions 0.18 not rolled-out not rolled-out not rolled-out not rolled-out not rolled-out not rolled-out not rolled-out not rolled-out not rolled-out not rolled-out not rolled-out not rolled-out not rolled-out
Index (number) of plasticity, unit fractions 0.11 no no no no no no no no no no no no no
Table 3. Dispersion of loess-like loam samples with admixtures of inorganic compounds alkalized with 5N NaOH solution after 28 days of hardening Granulometric composition, %, particle size in mm
Admixtures Primary
>o. 1 0.1 1.6 0.4 1.9 3.1 1.3 0.2 0.5 0.5 0.1 2.4 3.9 0.9 2.9
0.1-0.05 3.7 32.8 42.8 27.6 23.9 29.3 29.1 12.0 12.0 32.9 13.8 11.9 15.1 12.1
0.05-0.01 65.9 49.5 36.2 48.0 46.2 46.5 51.0 67.2 67.2 45.8 62.7 57.1 53.3 61.8
0.01-0.005 0.005-0.001 6.0 9.6 7.4 4.4 7.5 6.8 8.0 7.3 10.5 8.3 6.6 8.5 10.1 6.1 6.9 6.5 6.9 6.5 9.5 5.5 7.8 8.9 13.3 7.0 12.7 9.7 6.2 8.1
<0.001 14.7 4.3 6.3 7.2 7.9 7.8 3.5 6.9 6.9 6.1 4.4 6.8 8.3 8.9
Table 4.Plasticity of loess-like loam samples with effective admixtures of inorganic compounds alkalized with 2.5N NaOH solution after 28 days of hardening
Admixtures Primary Without admixtures Ca(OHj2 CaCI2 Phosphogypsum d(oH)3 MgO Ba(OH)2 BaC12
The upper plasticity limit,unit fractions 0.29 0.3 1 0.33 0.41 0.35 0.33 0.32 0.38 0.44
The low plasticity limit,unit fractions 0.18 0.24 not rolled-out not rolled-out not rolled-out not rolled-out not rolled-out not rolled-out not rolled-out
595
Plasticity number,unit fractions 0.11 0.07 no no no no no no no
Table 5. Dispersion of loess-like loam samples with effective admixtures of inorganic compounds alkalized with 2.5N NaOH solution after 28 days of hardening Granulometric composition, %, particle size in mm Admixtures >o. 1 0.1-0.05 0.05-0.01 0.01-0.005 0.005 <0.001 0.001 6.0 9.7 14.7 Primary 0.1 3.7 65.9 32.8 49.5 7.4 4.4 4.4 Without admixtures 1.6 8.7 8.9 7.7 Ca(OH)2 0.7 9.1 64.9 46.1 30.3 6.9 7.8 5.2 CaCl2 3.7 5.1 9.6 7.4 Phosphogypsum 0.7 10.8 66.5 8.5 9.4 8.4 d(OH)3 0.2 9.0 64.5 11.8 6.9 4.4 0.1 5.6 71.2 MgO 14.0 9.2 3.8 0.3 3.2 69.5 Ba(OH)2 BaC12 0.1 7.8 62.6 12.8 9.5 7.3
The weak endothermic effects are seen in thermograms in the range of 545-585", this is bound up with the transformations of bonds Si-0-Me (where Me is Al, Mg, Ca), i.e. with the transformations of Al, Mg, Ca-hydrates (Cerkinsky 1998). The thermograms are practically similar to each other with the exception of those of the alkalized samples in MgO and phosphogypsum presence. In MgO presence the endothermic effect in the range of 160" in the thermogram is widening and in phosphogypsum presence the deep endothermic effect in the range of 120" splits into 2 not deep endoeffects at 70" and 260". This indicates the appearance of the new formation different from the previous ones. X-ray difiactograms confirm this assumption: the alkalized sample with the phosphogypsum admixture clearly shows responses of significant intensity. New responses a pear with the interplanar distances: 4.41, 4.10, 3.53 , etc., which approximately can be refered to the alkali phosphate of the Al203.P20 kind. As can be supposed, phosphogypsum which contains from 2.5 to 3% of P205, promotes the formation of phosphate compounds with binding properties. All compounds effectively influence the alkalized samples strength similar to berlinite. The X-ray diffractograms of the sample alkalized in presence of MgO, show the splitting and widening of the responses because of structudes disorder, for instance in the range of 4.10-3.20 A. This is apparently due to formation not only of calcium hydrosilicates but magnesium hydrosilicates as well. Alkalizing of soil with 5 N NaOH solution leads to loss of plasticity, soil is not "rolled-out into a cord" (table 2) and the granulometric composition of soil shows an increase of content of coarse dust-like and sandy fractions (table 3). Introduction of admixtures into the 5 N NaOH solution has no practical effect upon the change in plasticity and dispersion of alkalized samples.
g:
The benefit of the admixtures introduced during alkalization and some strength "reserve" allow decrease in concentration of the alkali agent and to make the effective concentration of 2.5 N working. It is also possible to use some intermediate products, for instance, electrovacuum soda (NaOH+NaCl) with the initially low alkali concentration. If the ultimate uniaxial compression strength of the sample after the 5 N NaOH solution influence is taken as the main criterion of the alakalization efficiency, with the lower concentration of natrium hydroxide, all the admixtures can be divided into 3 groups according to our investigation results: ineffective - AlC13; low-effective - CaC03, CaS04, MgS04, MgC12; effective - Ca(OH)2, phosphogypsum, CaC12, Al(OH)3, MgO, Ba(OH)2, BaC12. For instance, looking at the alkalized samples, the influence of the effective admixtures on the samples strengthening with 2.5 N NaOH solution is decreased in the series: for loess-like loams: Ca(OH)2> Ba(OH)2> phosphogypsum, Al(OH)3> MgO>CaCl2>BaCl2; for clays: CaC12 > phosphogypsum > Ca(OH)2, Al(OH)3,BaC12 > Ba(OH)2. Similar results have been obtained when testing the samples alkalized in the electrovacuum soda in presence of the above admixtures. X-ray diffractograms and thermograms of loesslike loam alkalized with 2.5 N NaOH solution with admixtures are similar to X-ray difiactograms and thermograms in figure 1.
596
The introduction of inorganic compounds when alkalizing the sample changed their plasticity properties. If, when exposed to 2.5 N solution, the soil did not lose its plasticity, the admixtures introduction resulted in plasticity loss (Table 4). The results of the sieve- and microaggregate analysis confirm the increase of sandy and coarse dusty-like fractions owing to the clayey fraction decrease (Table 5).
4 CONCLUSIONS 1 Introduction of cations of alkaline-earth metal into 5 N NaOH solution when alkalizing clayey and loess soil leads to 3-5 times increase of uniaxial compressive strength of samples compared to strength of samples alkalized with 5 N NaOH solution without admixtures. 2. Cations of earth-alkaline metal actively aaect structure formation when alkalizing clayey and loess soil. It is confirmed by X-ray diffractograms and thermograms of samples. 3. Samples of clayey and loess soil alkalized with 5 N NaOH solution with- and without admixtures are not plastic, the effect of dusty and sandy fractions in their granulometrk composition increases. 4.The increase of strength of clayey and loess soil samples alkalized with 5 N NaOH solution with admixtures allows decrease in concentration of alkali agent when alkalizing samples up to 2.5 N concentration. 5. Compressive strength of samples of clayey and loess soil alkalized with 2.5 N NaOH solution with admixtures is comparable with compressive strength of samples alkalized with 5 N NaOH solution, 6. Samples of clayey and loess soil alkalized with admixtures lose their plasticity, the effect of dustylike and sandy particles in granulometric composition increases compared to samples alkalized with 2.5 N NaOH solution. 7. Admixtures of alkaline-earth metal can be arranged in the following series according to their active influence on structure of new formation when alkalizing clayey soil; for loess-like loam Ca(OH)22 Ba(OH)2> phosphogypsum, N(oH)3 > MgO > CaCl2 > BaC12; for clay CaClz > phosphogypsum > Ca(OH)2, A(oH)3, Bach > Ba(OH)2. 8.The regularities of an effect of inorganic compounds admixtures with the content of alkaline-earth elements allow control of process of clayey and loess soil alkalizing at the specific stages of chemical trasformation. In practical use the above regularities allow decrease of alkali consumption without change of given strength and deformation properties of the alkalized soil. 597
REFERENCES Cherkinsky Y.S. 1998. Chemistry of the polymer inorganic binding agents, monograph,L.:Chemistry, p.223. Volkov F.E. and Zlochevskaya R.I. 1988. A new chemical way of water-saturated loessial and clayey soils strengthening - “alkalization”. Engineering Geology. No. 1, p. 15. Zlochevskaya R.I., Volkov F.E., Makeeva T.G., Samarin E.N.,Shlykov V.G. 1990. Interaction of clayey and loessial soils with the concentrated alkali solution. Engineering Geology. No.2.pp.3351.
This Page Intentionally Left Blank
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Development of light-weight soil using excavated sand and its application for harbor structures in cold regions M.Hirasawa Civil Engineering Research Institute, Hokkaido Development Bureau, Sapporo, Japan
S.Saeki, S. Kodarna & T.Yakuwa Otaru Port Construction Ofjce, Hokkaido Development Bureau, Japan
T.Tsuchida Port and Harbor Research Institute, Ministry of Transport, Yokosuka,Japan
ABSTRACT The light-weight soil construction method was developed to meet demands for reducing relative and uneven settlement, decrease earth pressure, prevent lateral displacement and improve earthquake resistance of harbor and airport structures. It was decided to use this method for Ishikari Bay New Port in Japan, primarily to reduce the earth pressure behind existing quaywalls. However, the viscosity of the material soil was extremely low. Therefore, it was necessary to adjust the grain size distribution of the soil material. Because this method is expected to be used even in winter, when temperature falls below zero, it is also necessary to examine physical properties of this soil before it is placed and cured at low temperatures. This study summarizes the mixing methods for light-weight soil, the results of laboratory experiments on its physical properties at low temperatures corresponding the construction conditions at Ishikari Bay New Port, and the results of the on-site construction during winter.
1 INTRODUCTION For the construction of harbor structures and airports in Japan, demand recently has been increased for light materials to reduce relative and uneven settlement of structures, decrease earth pressure, prevent lateral displacement and improve earthquake resistance. Light-weight soil (LWS) is made by mixing air foam, expanded polystyrol (EPS) beads, or other lightening materials, and cement, with clayey soil whose water content exceeds its liquid limit. This type of light-weight soil in which air foam is used as the lightening material is referred to foam-treated soil (ITS) in this paper. It is possible to place FTS underwater. Examples of its application in harbor and airport structures in Japan include restoration of earthquake-damaged Port Island in Kobe and revetments at Tokyo International Airport. Test application using this method has been conducted nationwide in anticipation of future uses. At Ishikari Bay New Port in Japan, it was decided to use the FTS construction method primarily to reduce earth pressure behind the existing quaywalls. Sand with an extremely low viscosity, collected from seabed behind the quaywall at the time of construction work, was used for this project. For previous construction, soil with a higher fine-grain content (e.g., with more silt and a higher viscosity), such as dredged soil, had mainly been used. There-
fore, it has become necessary to adjust the grain size distribution of the material soil. Because it was expected to use this material in winter, when temperatures fall below zero, it was necessary to examine the properties of FTS at low temperatures in advance. Regarding the influence of low temperatures on FTS, the effects of freezing and thawing on solidified materials had been examined. However, few studies have been conducted on physical properties of FTS at low temperatures (OOC or lower).
Figure 1 Location of Ishikari Bay New Port
599
Table 1 Soil properties
Maximum main size
( m m > l 4.75
Table 2 Target physical properties Figure 2 Cross-section of the quaywall
Note:* target values for material at the age of 28 days I
I Mixing and agitatinp 1 I
Kneading I
Table 3 Mixing conditions A i r foam ( urn3)
1 B-3 B-4
1
I
Density measurement
]
440
426 386 364 344 326 370 368
w*
0 45 60 70 80 90 75
80
Note:* w=water content
The sand compaction pile and gravel drain in the figure were used in conjunction with the FTS as countermeasures against liquefaction.
Figure 3 Experiment procedure
This study reports on the methods of mixing FTS by using sand as material soil, and the results of laboratory experiments on its physical properties when it is placed and cured at low temperatures. The results of on-site application of FTS during winter are outlined below.
3 MIXDESIGN
3.1 Experiment
2 SUMMARY OF QUAYWALL IMPROVEMENT Ishikari Bay New Port is a major port near Hokkaido's capital, Sapporo (Figure 1). The quaywall investigated in this study was constructed of steel sheet-pile in 1982 and was the first to be used at the port. Because of the progress of steel corrosion, it was decided to conduct improvement to reduce earth pressure behind the quaywall and take countermeasures against ,liquefaction during earthquakes. The FTS construction method mainly contributes to reducing earth pressure. For the improvement, a section behind the steel sheet-pile quaywall (5.25 m in vertical depth, crown height +0.25 m, 16.00 m in horizontal depth) was replaced with FTS whose unit weight was 1.3 tf/m3 (=12.7 kN/m3) (Figure 2).
600
(1) Objectives and procedure A mixing experiment was conducted to investigate the suitability of using sand excavated from behind the quaywall (Table 1) as material soil and of adding bentonite to keep the target physical property values to the level shown in Table 2. This experiment was conducted following the procedures shown in Figure 3. The generated quantity of FTS, its wet density, unconfined compressive strength, flow values and water content were measured. The flow values were measured using a cylinder 8 cm in diameter and 8 cm in height. The underwater placement was conducted using an underwater separation resistance tester. The force-feed speed was 20 cmls. The cement in the figure was portland blastfurnace slag cement type-B. Air foam, a surfactant foaming agent, was diluted 20-fold with tap water and foamed 25-fold. (2) Mixing conditions
Table 3 shows the mixing conditions for the experiments. The bentonite replaced 10 to 20 % by dry weight of the material soil. 3.2 Effectiveness of bentonite
(1) Flow property Figure 4 shows the relationship between water content (w) and flow value for both atmospheric and underwater placements. The flow value can be used as a reference for uniform placement of FTS. In atmospheric placement, the flow value increased as the water content increased, regardless of the amount of cement per unit volume. B-3, B-6 and B-7 satisfied the target values (15 - 20 cm). In underwater placement, the difference in flow values depending on the change of w was smaller than in atmospheric placement. The flow values were small, 10 cm or less. (2) Density and strength property From cases in which the flow values exceeded the standard, B-7 was chosen as a model of FTS, and its wet density and unconfined compressive strength were examined. Figure 5 shows the change of the wet density ( p,) and the unconfined compressive strength (q,,)with the material age. The target value of p , (1.3 g/cm3 or less) was achieved when the material age reached 28 days. In the case of underwater placement, although p , increased as the material age increased, the rate of increase slowed as the material age increased further. The 28-day material satisfied target values of q,, both in atmospheric and underwater placements (at least 590 and 200 kN/m2, respectively). (3) Appropriateness of the addition of bentonite B-7 FTTS satisfied all the target physical values, including the target density and strength in underwater placement. This proved that the addition of bentonite was appropriate when the grain size distribution of the material soil was adjusted. Three patterns of mixing, in which the added bentonite quantities varied by 225% of the standard (B-7 mixing conditions), were examined under onsite application conditions. It was found that separation tended to occur underwater as the quantity of bentonite was decreased. Meanwhile, as the quantity of bentonite was increased, a lump of bentonite formed within the FTS. From these results, it was concluded that B-7 mixing conditions were roughly appropriate in terms of the quantity of added bentonite.
Figure 4 Relationship between water content and flow value
Figure 5 Change of wet density and unconfined compressive strength with material age
4 LOW TEMPERATURE EXPERIMENT 4.1 Experiment (1)Testing procedure To reproduce temperatures expected on-site, specimens were prepared in a low-temperature observation room. Wet curing was conducted in a thermostatic chamber and flow value, wet density and unconfined compressive strength were measured. Details of the procedures were the same as in the case of atmospheric placement shown in Figure 3. Flow values were measured immediately after kneading and five minutes later to examine the influence of low temperatures on fluidity. (2) Mixing conditions Experiments were conducted under three mixing conditions (Table 4). L-1 had the same mixing conditions as those for B-7, mentioned earlier. (3) Temperature conditions As shown in Table 5, five combination patterns were established according to outside air temperature on site (corresponding to room temperature in this experiment) and seawater temperature at the FTS placement site (corresponding to curing temperature in this experiment).
601
Table 4 Mixing: conditions Case
L-1 L-2 L-3
Cement Bentonite ( kg/m3) (kg/m') 200 98 220 96 240 94
Soil ( kg/m3) 393 384 375
Seawater Air foam w ( kg/m3) (h') (%) 80 368 393 384 375 80 375 381 80
4.4 Intensity property
(1)Effect of the kneading temperature Figure 8 shows the relationship between the material age and the unconfined compressive strength (4") for L-1, at a curing temperature (TJ of 5°C. Even in the case of T-4 in which T, dropped to -2.5"C, no influence by initial freezing was observed.
Table 5 Temperature :onditions Material Soil Cement Bentonite Air foam Diluted water Force-feed air Kneading water Room temperature Curing temperature
temperature (T,). Here, A p t tended to increase when T, c 0°C. This may be due to the change in bubble volume following the shrinkage of the JTS.
I
10 10 10 10 0 0 -5 10 5 5 5 10 5 5 5 10 0 0 5 10 5 5 5 5 0 0 0 -5 10 10 5 0 5 5
--
9 16
The temperature of injected air was considered to be equivalent to the outside air temperature. When the outside temperature was -5"C, it was anticipated from the experiment results on foaming properties that foaming would be difficult because the foaming nozzle would freeze. It was concluded that temperature control would be necessary on-site, and the temperature was set at 5°C.
I
i
LL
15
-5
I
I
5
0
10
Tm ("C)
Figure 6 Relationship between kneading temperature and flow value 0.10 A
0
6 2
4.2 Fluidity
0.05 0
v
Figure 6 shows the relationship between the internal temperature of LWS immediately after kneading, and the flow values immediately after kneading and five minutes later. The internal temperatures (T,) of T-3, T-4 and T-5 just after the kneading were 1.2, 2.5 and 8.7"c, respectively. The average flow values of L-1, L-2 and L-3 are shown in the figure. The target flow values (15 - 20 cm) were obtained for each temperature condition. However, the flow value for T-4 just after kneading was smaller than those for T-3 and T-5; the flow value for T-4 five minutes after kneading decreased by 1.4 cm. This may be attributed to the freezing of water inside the FTS. Given these facts, the temperature of FI'S force fed within the grout hose might be affected by outside air temperature in which the placement site might be far from the pump. This might reduce the fluidity and should be considered at the time of placement.
c 0.00
Q
L-1 I
@
L-21
A
L-3
'00
2 -0.05
-5
0
10
5
T m ("C)
Figure 7 Relationship between h e a d i n g temperature and density variation
loo0 800 "E \
600
Z
5 400
6
200
7
4.3 Wet density Wet density ( p t ) of both the 7-day and 28-day materials was f 5 % of the target value (1.1g/cm3). Figure 7 shows the relationship between the density variation ( A p , = ptzs- P , ~ and ) the kneading
14
28
days
Figure 8 Effect of kneading temperature on uncodined compressive strength of FTS
602
Tc C'C) 22 10
5 0
7
14
28
days
Figure 9 Effect of curing temperature on unconfiued compressive strength of FTS
2000
E
1500
\
5 1000
v
CO
2 500 0 0
500 1000 15002000 qU7 (kN/m2)
Figwe 10 Unconfined compressive strength of the 7- and 28-day FTS
Also, the same level of qu as for T-3 and T-5 (in which T, 2 0°C) was obtained, and the target value of 590 kN/m2 was realized when the material age was 28 days. Similar tendencies were observed for other mixing conditions. The Freezing temperature of concrete is considered to be between -0.5 and -2.O"C. The developed strength significantly decreases when early-age frost damage occurs. In such cases, the concrete tends to have problems of durability and watertightness.
Therefore, AE (air-entrained) concrete is generally used in Japan when the temperature at the time of construction is -3.O"C or lower. When AE concrete is used, non-frozen water inside the concrete is pressurized as a result of freezing of excess water, and as a result of-entering bubbles that abound within the concrete. This reduces the water pressure and prevents destruction of the concrete structure. FT'S has a larger quantity of bubbles per unit volume and a higher coefficient of permeability than does AE concrete (Kikuchi et al., 1998). It was considered that these factors would make it easier for the pressurized water to enter neighboring bubbles and prevent the influence of early-age frost damage even under the temperature conditions of T-4. (2) Influence of curing temperature Figure 9 shows the relationship between the material age and unconfjned compressive strength (4") when T, = 1.2"c in the case of L-1. The higher is T,, the greater is 4". Similar tendencies were observed for specimens under other mixing conditions. The figure also shows qu when T, = 22°C (expressed as T-0), indicating that greater qu was generated than for T-1 (T, = 10°C). As described in the above-mentioned results, the lower was T,, the smaller was qu when the material age was less than 28 days. Therefore, when FTS is used sooner at construction sites, not only regular mixing tests but also low-temperature experiments should be conducted in advance to determine expected yield strength. (3) Change of strength according to the material age In Figure 10, qu of the 7- and 28-day FTS were compared for all cases of mixing conditions. The strength of the 28-day FTS was approximately three times that of the 7-day FTS, regardless of mixing and temperature conditions.
1-1
In a Cold Protection Cover
Figure 11 Layout of machinery and equipment
603
I
5.2 Quality of lightweight soil Figure 12 shows the effect of the elevation on the wet density ( p t ) and the unconf-ined compressive strength (qu)of samples collected by boring from the FTS application site 28 days after placement. All the samples collected from each elevation satisfied the target value of p t(1.3g/m3 or less). However, as the water depth increased, p t tended to increase due to the influence of water pressure. qu greatly exceeded 200 kN/m2, the target value for underwater placement. This may be attributed to the fact that the underwater placement was conducted successfully, without much separation of the material, and that the density increased due to the influence of water pressure (Tsuchida et al., 1996a). Figure 13 shows the relationship between qu and the secant modulus (Eso), which can be approximated by the formula described below. The values were generally dispersed in areas lower than those of the on-site experiment results using the dredged soil of Tokyo Bay (Tsuchida et al., 1996b).
Figure 12 Effect of elevation on wet density and unconfined compressive strength for FTS
E,, = (50-200)
qu
6 CONCLUSIONS
1. When sand was used as material for LWS, underwater separation resistance was improved and underwater placement became possible by adding an adequate amount of bentonite. 2. Even at low temperatures, it was possible to apply FTS by conducting appropriate temperature control, in addition to normal execution management. 3. At low temperatures, the lower is the curing temperature, the lower is the unconfined compressive strength.
Figure 13 Relationship between unconfined compressive strength and secant modulus for FTS
5 ON-SITE APPLICATION 5.1 Outline of the application method The on-site application of FTS was conducted at Ishikari Bay New Port during the period between January 27 and February 16, 1999. The placement of FTS was conducted by dividing the total placement length of 31 m into three sections using steel plates, and the design depth of 5.25 m into four layers. The total quantity of placement was 2,550 m'. Figure 11 shows the layout of machinery and equipment used. Based on the low-temperature experiment results, the entire plant, excluding the cement and bentonite silos, was wrapped with a coldprotection cover to keep the internal temperature at or higher. When seawater was used, its temperature was raised to 5°C by a heater. The outside air temperature was below zero through most of the placement period, and the lowest temperature was -9"c, observed on February 3.
REFERENCES Kikuchi, Y & H.Yoshino 1998. Permeability of LightWeight Soil Made of Dredged Slurry Mixed with Air Foam. Report of Port and Harbour Research Institute, Y o k o ~ ~ k37(1): a , 34-56
Tsuchida, T., Kasai, J., Mizukami, J., Yokoyama, Y & K.Tsuchida 1996a. Effect of Curing Condition on Mechanical Properties of Light-Weight Soils. Technical Note of the Port and Harbour Research Institute, Yokosuka, 834
Tsuchida, T., Yokoyama, Y, Mizukami, J., Shimizu, K. & J.Kasai 1996b. Field Test of Light-Weight Geomaterials for Harbour Structures. Technical Note of the Port and Hurbour Research Institute, Yokosuka, 833
604
CoastalGeotechnicalEngineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
A new method for predicting strength of cement stabilized clays S. Horpibulsuk & N. Miura Department of Civil Engineering, Saga University,Japan
T. S. Nagaraj Department of Civil Engineering, Indian Institute of Science, Bangalore, India
ABSTRACT: Coastal regions with deep deposits of soft clays cause many problems associated with the engineering construction since they have low strength and high compressibility. Chemical admixture improvement is one of the effective techniques. This technique can be applied for the deep and shallow foundations. Surplus clay, such as waste clays from construction works, can also be improved by this technique to be an engineered geo-material. Based on the oedometer and triaxial test results, here it is revealed that the clay water content/cement ratio, wc/c is the dominant parameter, which governs deformation and strength characteristics of the stabilized clays. As a result, two simple equations related to wc/c for predicting strength and combining the relation among strength, curing time, water content of clay and cement content are presented. cavated at about 2-m depth is generally very soft gray silty clay (about 66% clay, 26% &.and 8% fine sand). The clay is highly plastic (CH) with natural water content of about 110-120%. The liquid and plastic limits are in the order of 106% and 53%, respectively. The salinity is about 1.6 g/l.
1 INTRODUCTION Chemical admixture stabilization has been extensively used both in deep and shallow foundations in order to improve inherent properties of soil such as strength and deformation behavior. Moreover, the new geo-material of waste clays being released from the construction works can be made by mixing with cement. It is of interest to understand and be able to predict the strength development of stabilized materials. There are very few researches on this area. Nagaraj et al. (1995) have examined the behavior of compressibility and permeability of cement stabilized soft clays. Also, the prediction of strength development of cement stabilized clays has been presented by Nagaraj et al. 1996, 1997 and 1998. But the use of this prediction is limited for the clays at their liquid limit states. To overcome this limitation, a factor-wc/c is proposed as the influential parameter for predicting strength and combining the relation among strength, curing time, clay water content and cement content. With this relation, one can estimate the strength at any curing times and conditions of clay water content and cement content from only one set of strength data.
2.2 Methodology of testing The clay paste was passed in 2-mm sieve for taking off the pieces of shell and the bigger size material. Its water content was decreased and increased until the requirement was realized. The water content of clay for this study covered a wide range starting from liquidity index of 1.0 and extending up to 2.0. The clay was mixed with ordinary Portland cement powder at three levels, namely 10, 15 and 20% of dry mass of the clay by a soil mixer for 10 minutes and put into cylinder containers (50-mm diameter and 100-rnm height). After one day, the specimens were taken out from the containers and wrapped in vinyl bags. The specimens were placed in a constant temperature and humidity room until testing. Unconfined compression (UC) and oedometer tests were carried out after 7, 14 and 28 days of curing. The rate of vertical displacement in UC tests was 1 mdmin. Consolidated undrained triaxial (CU) tests at effective cell pressure, dcof 100, 200 and 400 kPa were conducted after 28 days of curing. A back pressure of 190 kPa was applied to ensure that degree of saturation of the specimen was higher than 95% by checking B-value. The rate of vertical displacement was 0.125 mdmin.
2 EXPERIMENTAL INVESTIGATION 2.1 Soil sample The soil sample used in this paper is leached marine Ariake clay collected at Saga, Japan. The sample ex605
3 EFFECT OF CLAY WATER CONTENT
/CEMENT RATIO ON DEFROMATION CHARACTERISTIC The effect of clay water content/cement ratio, wc/c on the increase of yield stress, 0 ; is illustrated in Figure 1 by a set of curves for wc/c values of 8, 10.6 and 16. The clay water contentlcement ratio, wc/c is defined as the ratio of clay water content to cement content. The cement content, A , is defined as the ratio of the weight of cement powder to dry weight of soil. It is apparent that the stabilized samples exhibit very low or negligible compressibility up to their yield stresses. Beyond their yield stresses, the markedly high compressibility is realized. An observation is worthy of discussion here that the lower value of w d c , the greater is the enh-ancement of the yield stress. It is, moreover, revealed that, the parameter- wc/c is the influential factor governing the deformation characteristic of clay-cement mixtures as illustrated by Figures 2 and 3. They demonstrate the consolidation behavior of two combinations, which have the same wc/c of 10.6. The first one is the stabilized sample at initial clay water content of 106%. The second One is made from the ‘lay at its water content Of I6O%. It is noticed that the strain-vertical pressure relations and yield stresses of both samples are practically the same. But the initial void ratios are different for all curing times. Thus, it is concluded that for clay water content ranging from LI = 1.0 to LI = 2.0, the mixtures having the same wc/c of 10.6 develop the same value of yield stresses, o’, and the deformation characteristics are identical. The stabilized clays with high water content are stable at higher void ratios.
Figure 2, Relationships between void ratio and vertical pressure of cement stabilized samples having the same wc/c value of 10.6
Figure 1. Effect of clay water contentkement ratio on compression behavior of cement stabilized Ariake clay Figure 3. Relationships between axial strain and vertical pressure of cement stabilized samples having the same wdc value of 10.6
606
4 EFFECT OF CLAY WATER CONTENT /CEMENT RATIO ON STRENGTH The effect of clay water contentkement ratio on strength characteristic of cement stabilized clay is clearly manifested by the results of consolidated undrained triaxial (CU) test. Figures 4 and 5 show the deviator stress, q and shear strain, E, relationships of stabilized samples at wc/c values of 16 and 10.6, respectively. The characteristic shapes of the (4, E,) curves of samples are found that the deviator stress increases to a peak value and then strain softens to a lower value of q. With the incremental values of dc, the enhancement of shear stress is evident in these figures. Moreover, the lower the wck, the larger is the amount Of strain softening showing defined and distinct peak. The excess pore pressure development with shear strain is shown in Figure 6 for the samples cured for 1 month with high wc/c value of 16. The reduction of excess pore pressure after peak deviator stress is insignificant. In contrast, the behavior of samples with low wc/c value of 10.6 is similar to that of overconsolidated clay. The (Au, E,) curves show distinct peaks that occur at low values of shear strain and the negative pore pressure is found for the sample subjected to low dCvalue of 100 kPa as illustrated in Figure 7. The role of dcis very clear that the higher value of dC, the greater is the pore pressure generation, It is, moreover, noted that the cement treatment modifies the pore pressure response behavior by reducing considerably the strain at peak excess pore pressure, Au,,. The Aumaxoccurs at very low strain for the sample of low effective cell pressure, d,but at larger strain for the sample of higher effective cell pressure, d,.
Figure 4. Relationship between deviator stress and shear strain of cement stabilized samples at wc/c value = 16 and 28 days of curing
Figure 5 . Relationship between deviator stress and shear strain of cement stabilized samples at wc/c value = 10.6 and 28 days of curing
Figure 6. Relationship between excess pore pressure and shear strain of cement stabilized samples at wc/c value = 16 and 28 days of curing
Figure 7. Relationshp between excess pore pressure and shear strain of cement stabilized samples at wc/c value = 10.6 and 28 days of curing
607
Figures 8 and 9 show the deviator stress-shear strain and excess pore pressure-shear strain behavior of the mixtures having the same wc/c value of 10.6. These figures are obtained from the results at 28 days of curing and effective cell pressure of 200 kPa. Both mixtures show practically the same feature of (4, E,) and (du, E,) relations. From the above results, it is of interest to conclude that the clay water contentlcement ratio is a salient parameter controlling the engineering behavior such as strength and deformation characteristics. As a result, the strength prediction can be proposed in terms of wc/c. Figure 8. Relationship between deviator stress and shear strain of cement stabilized samples having the same wc/c value of 10.6 at coilfining pressure = 200 kPa
5 STRENGTH PREDICTION OF STABILIZED CLAYS Now, strength prediction of cement stabilized clays is discussed based on the test results of marine clays (Ariake clay, Hiroshima clay and Bangkok clay) as well as inland clays (Brown soil, Black cotton and Illite). Details of these clays are shown in Table l . The prediction referred to the clay water content/cement ratio’s concept is proposed as follows: “. . .for given clay-cement mixtures, the strength at any curing time depends on one factor-clay water content/cement ratio, wc/c” . The observed relationship between unconfined compressive strength aRer a certain period of curing can be expressed by a formula having the following
Figure 9. Relationship between excess pore pressure and shear strain of cement stabilized samples having the same wc/c value of 10.6 at confining pressure = 200 kPa
Figure 10. Strength prediction based on clay water contentkement ratio’s concept of cement stabilized Ariake and I-hroshima clays
where qu is the unconfned compressive strength of cement stabilized clay at a stated age, (wc/c) is the clay water content/cement ratio, A and B are constant depending on the characteristic of clay, type of cement and curing time. Figure 10 shows the strength prediction of cement stabilized Ariake and Hiroshima clays, which agrees well with this proposed method indicated by the coefficients of correlation are higher than 0.97 for all cases. For the clays having their water contents are equal to and higher than liquid limit and wc/c ranging from 4 to 16, the A-values are 5364, 6586, 7803 Wa for 7, 14, 28 day-curing time, respectively and the Hvalue is I .24 for all cases of 7, 14 and 28 day-curing time which are derived from equation (1). The relationship between qu and wc/c of Bangkok clay is also performed well with high coefficients of correlation (higher than 0.99 for all cases) as shown in Figure 11. The A-values are 969, 1130, 1739 kPa for 7, 14 and 28 day-curing time, respectively. The B-value is 1.24 for all cases of curing time.
608
Since B-value is identical for both marine and inland clays, the same strength ratio equation of cement stabilized marine and inland clays at a particular curing time can be obtained in terms of clay water content/cement ratio as follows
Clay water content/cement ratio Figure 11. Strength prediction based on clay water content/cement ratio's concept of cement stabilized Bangkok clay
where S(wc,c,, is the strength to be estimated at clay water content/cement ratio of (wc/c)I and S(wc/c)z is the strength value at clay water content/cement ratio of (wc/c)z From this prediction, it reveals that the wc/c does not play any role on the strength development with time. As a result, the strength normalization of stabilized clays by the 14 day-strength can be performed by the linear regression analysis in terms of curing time Only as fOllOWS = a + b In D
(3)
s 1 4
where SDis the strength after D days of curing, Sl4is the 14 day-strength, D is the curing time, a and b are constant depending upon the type of'clay. It found here that a = 0.12 and b = 0.32 for marine clays. The a and b for inland clays have investigated by Nagaraj et al. 1995 as -0.18 and 0.46, respectively. The interrelationship among strength, curing time and clay water content/cement ratio, wc,% for predicting strength development of cement stabilized clays is expressed by combination of equations (2) and (3):
Clay water content/cement ratio
(4)
Figure 12. Strength prediction based on clay water contenucernent ratio's Concept Of Cement stabilized inland clays
where ,"(wc,c)l,D is strength of stabilized clay to be estimated at clay water content/cement ratio of (wc/'c), at D day-curing time and S(wc/c),14is strength of stabilized Clay at clay water content/cement ratio of (wc'c) a*er l 4 day-curing time.
Similarly, the strength prediction of cement stabilized inland clays is presented with coefficients of correlation being higher than 0.96 as shown in Figure 12. The A-values are 2383, 3296 and 4515 for 7, 14 and 28 days of curing, respectively. The B-value is 1.24for all cases of curing time. From above prediction, the B-value can be taken as 1.24 for the both cases of marine and inland clays but the A-values are different. For every curing time, the A-values of stabilized marine clays in Japan are highest. Among the clays presented here, such the marine clays provide the highest strength at the same conditions of curing time, clay water content and cement content.
6 CONCLUDING REMARKS
Based on the results of oedometer, triaxial and unconfined compressive tests, it is demonstrated, here that the clay water content/cement ratio is the dominant parameter influencing the yield stress and strength development. The prediction of strength is finally proposed based on relation to the results from marine and inland clays. The following conclusions are drawn from this study:
609
Table 1. Basic properties of marine clays for stabilization. Sample Bangkok clay Hiroshma clay h a k e clay (1) Ariake clay (2) Ariake clay (3) Liquid Limit 103 122 80 100 106 43 47 32 45 53 Plastic Limit 60 75 48 55 53 Plasticity Index 2.75 2.60 2.62 2.63 2.56 Specific gravity Uddin (1995) Yamadera (1999) Yamadera (1999) Yamadera (1999) Horpibulsuk (1999) Table 2. Basic properties of inland clays for stabilization. Sample Liquid Limit
Brown clay Black cotton (1) Black cotton (2) Illite 62 72 86 118 Department of Civil Engineering, Indian Institute of Science (1995)
1) The salient effect of clay water contentkement ratio, wc/c on the stabilized clay is found that the wc/c is the influential parameter governing the strength and deformation characteristics. The mixtures having the same wc/c develop practically the same yield stress and strength. Lower the wdc, higher is the yield stress and strength. 2) Since the behavior of stabilized clays is remarkably governed by w c k , the strength prediction in terms of wc/c as well as the interrelationship involving strength, curing time and wc/c are proposed as presented in equations (1) and (4), respectively. The implementation of the presented method is to simplify the task of arriving at the cement content and curing time by laboratory investigations to realize the target values. 3) This paper reveals that the strength of the stabilized clays strongly depends upon only wc/c. Thus to obtain uniform strength of stabilized clays for engineering works such as deep foundation, shallow foundation, improvement of the surplus clays for engineering purposes etc. at construction sites in which water content of clay varies with depth and location, the wc/c of mixture must be controlled. The clay water contendcement ratio's concept, thus, overcomes not only the engineering requirement but also the economical viewpoint.
REFERENCES Brom, B.B. & Boman, P. 1975. Lime stabilized column. Pr0c.5'~Asian Regional Col6 Indian Institute of Science, Bangalore. India. 1: 227-234 Horpibulsuk; S., Mura, N.and Nagaraj, T.S. 1999. Prediction of strength and consolidation parameters of cement stabilized clays. Report of the Faculty of Science and Engineering, Saga Univeristy, Japan. Vol. 28, No. 2: 27-38. Department of Civil Engineering 1995, Generalized approach to predict strength development by cement admixture. Research Report, Indian Institute ofscience, India. Nagaraj, T.S., Miura, N., Yaligar, P. and Yamadera, A. 1995. Analysis of compressibility behavior of soils with induced cementation. Indian Geotechnical Conference. IGC -95. Bangalore. India. 1: 53-56.
610
Nagaraj, T.S., Yarigar, P., Miura, N.and Yarnadera, A. 1996. Prediction strength development by cement admixture based on water content. The Second International C'oiference on Ground improvement Geosystems, Grouting and Deep Mixing, Tokyo. 14-1 7May 1996. 1: 43 1-436. Nagaraj, T.S., Miura, N., Yamadera, A. andYaligar, P. 1997. Strength assessment of cement admixtured soft clays - Parametric study. International Conference on Grour7d Improvement Techniques, 7-8May 1997. Macau: 379-386. Nagaraj, T.S., Miura, N. and Yamadera. A. 1998. Induced cementation of soft clays - Analysis and assesment. International Symposium on Lowland Technologyy,Institute of Lowland Technology, Saga University, Saga, .Japan: 267278 Treval C. 1968. The properties offlesh concrete. New York: John Willey&Sons. Uddin, K. 1995. Strength and deformation behavior of cement-treated Bangkok clay. D.Eng Thesis. AIT.. Bangkok, Thailand. Yamadera, A., Nagaraj, T.S. and Miura, N. 1998. Prediction of strength development in cement stabilized marine clay. Improvement of Soft Ground, Design, Analysis and Current Research, Institute of Lowland Technology and Asian Center for Soil Improvement and Geosynthetics, Saga and Bangkok: 141-153. Yamadera, A 1999. Microstructural study of geotechnical characteristics of marine clay. Ph.D. Thesis, Saga University, Saga, Japan.
Tire shreds as lightweight fill for construction on weak marine clay D. N.Humphrey University of Maine, Orono, Maine, USA
N.Whetten Haley and Aldrich, Hartford, Conn., USA
J.Weaver & K. Recker Haley aid Aldrich, South Portland, Maine, USA
ABSTRACT: Tire shreds were used as lightweight fill to improve slope stability for two highway projects Constructed on weak marine clay. Tire shreds, which are scrap tires that have been cut into 50 to 300 mm pieces, were chosen because they have an in-place density of 0.80 to 0.93 Mg/m3, which is less than half the density of conventional soil fill. In the first project, a 4.3-m thick zone of tire shreds was used as lightweight fill to improve global stability of a bridge approach fill founded on weak marine clay. In addition, the tire shreds reduced horizontal pressure on the bridge abutment. Some 400,000 tires were used in the project. In the second project, two layers of tire shreds, each up to 3.05 m thick, were used as lightweight fill for a highway embankment founded on 9 m of weak marine clay. This project used about 1.2 million tires. The full guidelines are contained in ASTM D627098, Standard Practice for Use of Scrap Tires in Civil Engineering Applications (ASTM 1998). Field studies have shown that tire shreds have a negligible effect on water quality (Downs et al. 1997; Humphrey et al. 1997, 1999a,b; Humphrey & Katz, 2000). This paper presents two case histories using tire shreds as lightweight fill for highway embankments constructed on weak marine clay foundations. In the first project tire shreds were used as lightweight fill and bridge abutment backfill for the approach embankment for the North Abutment of the Merrymeeting Bridge in Topsham, Maine. In the second project tire shreds were used for two approach embankments for a new bridge over the Maine Turnpike in Portland, Maine. From the information presented below, it will be clear that tire shreds are a viable material for use as lightweight fill for highway construction on weak marine clays and similar applications.
1 INTRODUCTION Tire shreds are scrap tires that have been cut into pieces with a maximum size ranging from 50 to 300 mm (ASTM 1998). This material is lightweight, free draining, and compressible. Moreover they have a thermal resistivity that is about seven times higher than soil and they produce low earth pressure. Because of their special properties tire shreds are increasingly being used as lightweight fill for embankments constructed on weak foundation soils (Humphrey et al. 1998; Whetten et al. 1997), lightweight backfill for retaining walls and bridge abutments (Tweedie et al. 1997, 1998a,b; Humphrey et al. 1998), compressible inclusions behind integral abutment and rigid frame bridges (Humphrey et al. 1998; Reid & Soupier 1998), thermal insulation to limit frost penetration beneath roads (Humphrey & Eaton 1995; Lawrence et al. 1999), and drainage layers for road (Lawrence et al. 1999) and landfill applications (Jesionek et al. 1998). In 1998 approximately 18 million tires were used in these applications in the United States. Civil engineering applications of tire shreds had a significant setback in 1995 and early 1996 when three thick tire shred fills (each greater than 8 m thick) experienced a serious self-heating reaction, however, guidelines to limit self-heating are now available (Ad Hoc Civil Engineering Committee 1997; ASTM 1998). Key features of the guidelines are to use larger size tire shreds, limit the amount of fine material in the shreds, limit layer thicknesses to 3 m, and limit the access of the fill to water and air.
2 NORTH ABUTMENT APPROACH FILL The key element of the Topsham Brunswick Bypass Project was the 300-m long Merrymeeting Bridge over the Androscoggin River. The subsurface profile at the location of the North Abutment consisted of 3 to 6 m of marine silty sand overlying 14 to I5 m of marine silty clay. The clay is underlain by glacial till and then bedrock. The existing riverbank had a factor of safety against a deep seated dope failure 61 1
that was near one. Moreover, the design called for an approach fill leading up to the bridge abutment that would have further lowered the factor of safety. Thus, it was necessary to devise a strategy to both improve the existing factor of safety and allow construction of the approach fill. The best solution was to excavate some of the existing riverbank and replace it with a 4.3-m thick layer of tire shreds. Tire shreds had the added advantage of reducing lateral pressures against the abutment wall. Other types of lightweight fill were considered including geofoam and expanded shale aggregate. However, tire shreds proved to be the lowest cost solution. The project used some 400,000 scrap tires (Whetten et al. 1997). 2.1 Project Layout and Construction The surficial marine sand was excavated to elevation 5.2 m and then the H-pile supported abutment wall was constructed. A 4.3-m thick zone of tire shreds was placed from station 53+50.6 m to the face of the abutment wall at station 53+72.0 m. The fill tapers from a thickness of 4.3 m at station 53+50.6 m to zero thickness at station 53+35.4 m thus providing a gradual transition between the tire shred layer and the conventional fill. It was estimated that the tire shred layer would compress 460 mm due to the weight of overlying soil layers. As a result, the layer was built up an additional 460 mm so that the final compressed thickness would be 4.3 m. The tire shred layer was enclosed in a woven geotextile (Niolon Mirafi 500X) to prevent infiltration of surrounding soil. The tire shreds were spread with front end loaders and bulldozers and then compacted by six passes of a smooth drum vibratory roller with a static weight of 9.4 metric tons. The thickness of a compacted lift was limited to 305 mm. Tire shred
placement began on September 25, 1996 and was completed on October 3, 1996. A longitudinal section of the completed abutment and embankment is shown in Figure 1. This project was designed and built prior to development of the guidelines to limit self-heating of tire shred fills. However, the project did include some design features to limit self-heating. The first was to use larger size shreds (called Type B shreds) in the lower portion of the fill from elevation 5.2 m to elevation 8.2 m. The Type B shreds were specified to have a maximum dimension, measured in any direction, of 305 mm; a minimum of 75% (by weight) passing the 203-mm square mesh sieve, a maximum of 25% (by weight) passing the 38-mm square mesh sieve, and a maximum of 5% (by weight) passing the No. 4 (4.75-mm) sieve. Gradation tests showed that the shreds generally had a maximum din;ension smaller than 150 mm. Type A shreds, with a maximum size of 75 mm, were placed from elevation 8.2 m to the top of the tire shred fill. It would have been preferable to use the larger Type B shreds for the entire thickness, however, a significant quantity of Type A shreds had already been stockpiled near the project prior to the decision to use larger shreds. It was judged that it would be acceptable to use the smaller Type A shreds in the upper portion of the fill. Moreover, it would have been preferable to limit the total thickness of the tire shred layer to 3 m as recommended by the guidelines to limit self-heating. As an additional step to reduce the possibility of self-heating, the tire shreds are overlain by a layer of compacted clayey soil with a minimum of 30% passing the No. 200 (0.075 mm) sieve. The purpose of the clay layer is to minimize the flow of water and air though the tire shreds. The clay layer is ap-
Figure 1. Cross section through North Abutment tire shred fill.
612
proximately 0.6 m thick and is built up in the center to promote drainage toward the side slopes, A 0.6rn thick layer of common borrow was placed over the clay layer. Overlying the common borrow is 0.76 m of aggregate subbase. Tire shreds undergo a small amount of time dependent settlement. For this project a thick tire shred fill adjoined a pile supported bridge abutment. This led to concerns that there could be differential settlement at the junction with the abutment. However, Tweedie et al. (1997) showed that most of the time dependent settlement occurs within the first 60 days. To accommodate the time dependant settlement prior to paving, the contractor was required to place an additional 0.3 m of subbase aggregate as a surcharge to be left in place for a minimum of 60 days. In fact, the overall construction schedule allowed the contractor to leave the surcharge in place from October, 1996 through October, 1997. The surcharge was removed in October, 1997 and the roadway was topped with 229 mm of bituminous pavement. The highway was opened to traffic on November I1 , 1997. Additional construction information is given in Cosgrove and Humphrey (1999).
2.2 Instrumentation Four types of instruments were installed: vibrating wire settlement gauges, settlement plates and temperature sensors placed in the tire shred fill; and pressure cells cast into the back face of the abutment wall. The vibrating wire settlement gages gave no useful readings and will not be discussed. Six settlement plates were installed. SPI-I and SP2-1 were at the base of the tire shred layer. SPI-2 and SP2-2 were at the mid-height of the tire shred layer. SPI-3 and SP2-3 were at the top of tire shred layer. Vibrating wire pressure cells were installed to monitor lateral earth pressure against the abutment wall. Three Roctest model TPC pressure cells (PCI-1, PC1-2, PC1-3) were installed on the face of the abutment wall 4 m right of centerline at elevations 6.7, 7.8, and 8.8 m. Three Roctest model EPC pressure cells (PC2-1, PC2-2, PC2-3) were installed 4 m left of centerline at the same elevations. Tire shreds were placed against all the cells.
ings are consistent with at-rest conditions measured on an earlier project (Tweedie et al. 1997; 1998a). Cells PC2-1, PC2-2, and PC2-3 showed different behavior. On 1019196, cell PC2-2 showed a pressure of 30.2 kPa while cell PC2-1, located only 1.07 m lower, was 20.0 kPa and cell PC2-3, located 1.07 m above PC2-2, was 12.31 kPa. These cells were the Iess stiff EPC cells. Large scatter has been observed with EPC cells on an earlier tire shred project (Tweedie et al. 1997, 1998a,b). This is thought to be due, at least in part, to large tire shreds creating a nonuniform stress distribution on the face of the pressure cell. The average pressure recorded by the three PC2 cells was 20.9 kPa, which is slightly higher than the PCl cells. Between 1019196 and 10/31/96 the lateral pressure increased by 1 to 2 kPa. The pressures have been approximately constant since that time. Table 1. Summary of lateral pressures on abutment wall. PCl-1 PC2-1 PC1-2 PC2-2 PCI-3 PC2-3 Date Cell el.=6.70m Cell el.=7.77m Cell el.=8.84m 10/3/962 7.84' 7.41 6.04 2.62 1.41 7.27 10/9/9G3 17.04 20.04 19.61 30.22 17.05 10.91 10/31/96 18.27 21.05 20.98 32.84 20.24 12.31 'Horizontal pressure in @a. 2Datetire shred placement completed, 3Date soil cover and surcharge placement completed.
Settlement during construction is shown in Figure 2. By the end of fill placement the tire shred fill compressed about 370 mm as indicated by plates SP1-3 and SP2-3. Post construction settlement is shown in Figure 3. In the first 60 days after the end of fill placement, the top of the tire shred layer settled an additional I35 mm. Between December 15, 1996 and December 31, 1997 the fill underwent an additional 15 mm of time dependent settlement. By late 1997, the settlement of the plates at the midheight of the tire shred layer (SPI-2 and SP2-2) and top of the tire shred layer (SPI-3 and SP2-3) had essentially stopped The total compression of the tire shred fill was 520 mm, which was 13% greater than the 460 mm that was anticipated based on laboratory compression tests. The difference is due, at least in part, to 2.3 Measwed Horizontal Pressure and ~ e t t ~ e ~ e ~time t dependent settlement that is not accounted for in the short term Iaboratory tests. The final comThe lateral pressure at the completion o f tire shred pressed density of the tire shreds was about 0.9 placement (1 O/3/96) and completion of soil cover Mg1m3. and surcharge placement (1019196) is summarized in Table 1. Lateral pressures on 10131196 are also 2.4 Tem~eratureof Tire Shred Layer shown. It is seen that at completion of tire shred placement, the pressures increased with depth. A small amount of self heating of the tire shreds ocHowever, at completion of soil cover and surcharge curred. Five out of the 12 thermistors in Type A placement, the pressures recorded by cells PC 1-1, shreds experienced a peak te~peratureof between PC1-2, and PCI-3 were nearly constant with depth 30 and 40°C. In contrast, only two of the 18 therand ranged between 17.1 and 19.6 kPa. These findmistors in the larger Type B shreds experienced a 613
peak in this range and these two sensors may have been influenced by warmer overlying Type A shreds. This suggests that larger shreds are less susceptible to heating. In any case, the peak temperatures were too low to be of concern. Since early 1997, the overall trend has been one of decreasing temperature, however, the temperature of the shreds do appear to be slightly influenced by seasonal temperature changes. 3 PORTLAND JETPORT INTERCHANGE Tire shreds were used as lightweight fill for construction of two 9.8-m high highway embankments in Portland, Maine (Humphrey et al. 1998). These embankments were the approach fills to a new bridge over the Maine Turnpike. The bridge is part of a new interchange that will provide better access to the Portland Jetport and Congress Street. This site was underlain by about 12 m of weak marine clay. Test results indicated that the clay is an overconsolidated, moderately sensitive, inorganic clay of low plasticity. Undrained shear strength varied from approximately 72 kPa near the top to 19 kPa near the center of the layer. The designers for the project (the Maine offices of HNTB, Inc. and Haley and Aldrich, Inc. and the University of Maine) found that embankments built of conventional soil were too heavy resulting in an unacceptably low factor of safety against slope instability. They looked at several ways to strengthen the foundation soils but these were too costly. Constructing the embankments of lightweight fill was chosen as the lowest cost alternative. They considered several types of lightweight fill including tire shreds, expanded polystyrene insulation boards, and expanded shale. Tire shreds were chosen because they were $300,000 (US) cheaper than the other alternatives. Moreover, the project would put some 1.2 million tires to a beneficial end use. Wick drains were also used to accelerate consolidation of the foundation soils.
Figure 3. Post-construction settlement of the North Abutment fill.
3.1 Project Layout and Construction Several steps were taken to comply with the guidelines to limit heating of thick tire shred fills (Ad Hoc Civil Engineering Committee 1997; ASTM 1998). The guidelines required that a single tire shred layer be no thicker than 3 m, so the tire shred layer was broken up into two layers, each up to 3 m thick, separated by 0.9 m of soil as shown in Figure 4. Low-permeability soil with a minimum of 30% passing the No. 200 (0.075 mm) sieve was placed on the outside and top of the fill to limit inflow of air and water. The final precaution to limit heating was to use large shreds with a minimum of fines. The shreds had less than 25% passing the 38-mm sieve and less than 1% passing the No. 4 sieve (4.75-mm).
The shreds had a maximum size measured in any direction of 300 mm to ensure that they could be easily placed with conventional construction equipment. The embankment was topped with 1.22 m of granular soil plus 1.22 m of temporary surcharge. The purpose of the surcharge was to increase the rate of consolidation of the soft clay foundation soils and was unrelated to the tire shred fill. The tire shreds were placed with conventional construction techniques. First geotextile was placed on the prepared base. Then the shreds were spread in 300-mm lifts using a Caterpillar D-4. Each lift was compacted with six passes of a vibratory roller with a minimum 9.1-metric ton operating weight.
614
Table 2. Measured compressibility on centerline of tire shred layer for Portland Jetport Interchange project. Lower layer Upper layer Plate Station Measured Predicted Measured Predicted SW1 7+62 12.6% 22% 8.3% 14% 11.2% 14% 21% 13.4% SW4 7+92 10.9% 14% 19.1% 22% SE1 9+14 23% 9.3% 14% 17.3% SE4 9+45 Average 15.6% 22% 9.9% 14%
<-
r S U R C H A R G E MATERIAL -SUBBASE
GRAVEL
Figure 4. Cross section through embankment for the Portland Jetport Interchange (Humphrey et al. 1998).
After placing the shreds, the contractor placed a geotextile separator on the sides and top of the tire shred zone and then the surrounding soil cover.
Embankment 2 Upper Tire Shred Layer TH36 -8- TH37 TH38 ----4--- TH39
* *
3.2 Construction Settlement and Unit Weight Settlement plates were installed at the top and bottom of each tire shred layer to monitor settlement. Compression of each tire shred layer at the end of fill placement is summarized in Table 2. The compression predicted based on laboratory compression tests on 75-mm maximum size tire shreds is also shown. It is seen that the predicted compression is significantly greater than the measured value. Thus, the compressibility of shreds with a 300-mm maximum size appears to be less than for 75-mm maximum size shreds. This was one factor that led to overprediction of the final in-place unit weight. The final in-place unit weight was predicted to be 0.93 Mg/m3 compared to an actual value of 0.79 Mg/m3, a difference of 18%. This difference cannot be entirely accounted for by the difference in compressibility. Thus, it is likely that the initial (uncompressed) unit weight of these larger shreds is less than for 75-mm maximum size shreds.
-
-
-
15
-
in G71197
I
I
a129197 12/27/97 412619a
a12419a 1212219a 4/21/99 8/19/99
Figure 6. Temperatures in upper tire shed layer of lightweight embankment fill at Portland Jetport Interchange.
3.3 Temperature Measurements Monitoring the temperatures of the tire shred fill was of great interest because of past problems with heating of thick tire shred fills (Humphrey 1996). The warmest temperatures were measured at the time of placement when the black tire shreds were heated by exposure to direct sunlight. Initial temperatures ranged from 24 to 38°C. After being covered with the first few lifts of fill, the temperatures began dropping with time. Temperatures had stabilized between l l and 15°C when monitoring was discontinued in July 1999. Typical temperature measurements are shown on Figures 5 and 6. 4 SUMMARY
The low unit weight, widespread availability, and low cost of tire shreds has led to their being used as lightweight fill for embankments constructed on weak foundation soils. The engineering properties of tire shreds are known including gradation, unit weight, compressibility and shear strength. When the special properties of tire shreds are needed for a
Figure 5 . Temperatures in lower tire shed layer of lightweight embankment fill at Portland Jetport Interchange.
615
Humphrey, D.N. & L.E. Katz 2000. Five-Year Field Study of the Effect of Tire shreds Placed Above the Water Table on Groundwater Quality. Preprint No. 00-0892, Washington, DC: Transportation Research Board. Humphrey, .D.N., N. Whetten, J. Weaver, K. Recker & T.A. Cosgrove 1998. Tire Shreds as Lightweight Fill for Embankments and Retaining Walls. Proc. of the Con5 on Recycled Materials in Geotechnical Applications, New York: ASCE, 51-65. Jesionek, K.S., D.N. Humphrey & R.J. Dunn 1998. Overview of Shredded Tire Applications in Landfills. Proc. of the Tire Industry Coizjkrence, South Carolina: Clemson Univ. Lawrence, B., D. Hurnphrey & L.-H. Clien 1999. Field Trial of Tire Shreds as Insulation for Paved Roads. Proc. of the Tenth Int. Conf on Cold Regions Engineering: Putting Research into Practice. New York: ASCE, 428-439. Reid, R.A. & S.P. Soupir 1998. Mitigation of Void Development Under Bridge Approach Slabs Using Rubber Tire Chips. Proc. of the Conf on Recycled Materials in Geotechnical Applications, New York: ASCE, 37-50. Tweedie, J.J., D.N. Humphrey & T.C. Sandford 1997. Tire Chips as Lightweight Backfill for Retaining Walls - Phase 11,” Storrs, Conn.: New England Transportation Consortium Tweedie, J.J., D.N. Humphrey & T.C. Sandford 1998a. Full Scale Field Trials of Tire Chips as Lightweight Retaining Wall Backfill, At-Rest Conditions. Transportation Research Record No. 1619,64-7 1. Tweedie, J.J., D.N. Humphrey & T.C. Sandford 1998b. Tire Shreds as Retaining Wall Backfill, Active Conditions. J. of Geotechnical and Geoenvironmental Engineering, ASCE, 124:10, 1061-1070. Whetten, N., J. Weaver, D. Humphrey & T. Sandford 1997. Rubber Meets the Road in Maine. Civil Engineering, ASCE, 67:9, 60-63.
project they are often the lowest cost alternative. Thus, civil engineers are choosing tire shreds because they offer both the properties needed to solve special problems and lower costs to satisfy the demands of their clients for the most economical project possible. In the next few years, major increases in the number of scrap tires used for civil engineering applications is possible because of their growing record of successful performance combined with guidelines to limit self-heating of thick fills, recently published ASTM guideline specifications, and groundwater data showing that they have a negligible environmental impact. ACKNOWLEDGEMENTS Jeff McEwen was a key player of in the North Abutment project when he worked for the Portland office of T.Y. Lin International and more recently for the Portland Jetport Project with his new employer, HNTB, Inc. of Portland, Maine. His leadership and insight on these projects is greatly appreciated. University of Maine graduate student Tricia Cosgrove is thanked for her hard work on the North Abutment project. The Maine Department of Transportation and Maine Turnpike Authority are thanked for funding these projects. REFERENCES Ad Hoc Civil Engineering Committee 1997. Design Guidelines to Minimize Internal Heating of Tire Shred Fills. Washington, DC: Scrap Tire Management Council. ASTM 1998. Standard Practice for Use of Scrap Tires in Civil Engineering Applications. ASTM 06270-98. W. Conshohocken, PA: Am. Soc. Testing & Mat. Cosgrove, T.A. & D.N. Humphrey 1999. Field Performance of Two Tire Shred Fills in Topsham, Maine. Augusta, Maine: Maine Department of Transportation. Downs, L.A., D.N. Humphrey., L.E. Katz & C.A. Rock 1996. Water Quality Effects of Using Tire Chips Below the Groundwater Table. Augusta, Maine: Maine Department of Transportation. Humphrey, D.N. 1996. Investigation of Exothermic Reaction in Tire Shred Fill Located on SR 100 in llwaco, Washington. Washington, DC: FHWA. Humphrey, D.N. 1999a. Water Quality Results for Whitter Farm Road Tire Shred Field Trial. Orono, Maine: Dept. of Civil and Env. Eng., University of Maine. Humphrey, D.N. 1999b. Water Quality Effects of Using Tire Cliips Below the Groundwater Table - Final Report. Augusta, Maine: Maine Department of Transportation. Humphrey, D.N. & R.A. Eaton 1995. Field Performance of Tire Chips as Subgrade Insulation for Rural Roads. Proceedings of the Sixth International Conference on LowVolume Roads. Washington, DC: Transportation Research Board, 2:77-86. Humphrey, D.N., L.E. Katz & M. Blumenthal 1997. Water Quality Effects of Tire Chip Fills Placed Above the Groundwater Table. Testing Soil Mixed with Waste or Recycled Materials, ASTM STP 1275, W. Conshohocken, PA: Am. Soc. for Testing & Mat., 299-313.
616
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchjda(eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Consolidation property of a highly compressible construction waste sludge M. Ibaralu, T.Sasahara & H.Akagi Waseda University,Tokyo,Japan
Y. Sugawara Tokyo Electric Power Company,Ibaraki, Japan
ABSTRACT: The slurry produced as a by-product of construction works has a high water content and contains a large amount of fine clay particles, thus necessitating its flocculation and dehydration for the sake of volume and water content reduction. The purpose of this study is to investigate the consolidation properties of flocs of different sizes, made by the application of flocculant to sludge, under lower consolidation pressures. Three types of experimental studies were carried out using clay flocs of different sizes. Based on the results of dewatering and consolidation tests, it can be concluded that the floes' compressibility decreases with the increase in their size under consolidation pressures below 10 kPa, and that the hydraulic consolidation test could reliably predict initial deformation under lower consolidation pressures.
1 INTRODUCTION
2 SAMPLES, FLOCCULANT AND SPECIMENS
In construction works such as shield tunnels or diaphragm walls, a great deal of slurry, called “construction waste sludge”, is produced. It has a high water content and contains a large amount of fine clay particles, thus necessitating its flocculation and dehydration for the sake of volume and water content reduction. For the dehydration of construction waste sludge, dehydrators such as filter-presses or belt-presses are commonly used. At the intermediate treatment plant selected for this study (Ichikawacity, Chiba), a special dehydrator called a “Dehydrum”(Figure 1) is used. The Dehydrum is a large size cylindrical rotor (@=3,340 mm and L=10,230 mm), which rotates in a single direction at a slow speed. The consolidation pressure of the Dehydrum is assumed to be much lower than that of other dehydrators, which are estimated to be approximately 500 kPa to 700kPa. It has been observed at the present plant that the consolidation properties of flocs, made by the application of flocculant to sludge, change according to their sizes. The purpose of this study is to investigate the consolidation properties of flocs of different sizes under lower consolidation pressure. Three types of experimental study were carried out using clay flocs: l)a gravity dewatering experiment to confirm the changes in the dewatering property of the flocs, 2)a constant rate of strain consolidation (CRS) test to investigate consolidation properties under a wide pressure range, and 3)a hydralic consolidation test to investigate consolidation properties under a lower pressure range.
In the plant, the quality of dehydrated sludge varies greatly due to the floes' size. It has been known that the size of the flocs varies in terms of additive volume, additive order and the combination of flocculants (J.H.Liu et al. (1996), Kawaguchi et al. (1998)). Therefore, additive volumes of flocculants have provided the key method of managing dewatering treatments. The authors tried to control the size of flocs by additive volumes of flocculant using bentonite slurry. The physical properties of the bentonite used are shown in Table 1. Distilled water was added to powdered bentonite to prepare sample slurry, and the slurry was settled for 24 hours to allow for sufficient swelling. Flocculants were then added to slurry, and the flocs obtained are used for experiments. In this study, we changed additive volumes and orders of organic flocculant (polyacrylamide solutions, 0.1% by weight) and inorganic floculants (LAC, a kind of pol yaluminium chloride) in order to prepare three sizes of flocs: large, medium, and small(Tab1e 2, Figure. 2).
3 GRAVITY DEWATERING EXPERIMENT 3.1 Introduction The gravity dewatering experiment was performed on three types of flocs of different sizes in order to confirm the changes in the dewatering property of each floc.
617
Figure 1. Dehydrum installed at plant.
3.2 Experimental Procedure Three types of floc, the mass of which had been known, was poured into a 75pm sieve and the mass of discharged water passing through the sieve was measured. The water content at the beginning of the experiment was determined 3600 % for each sample. From these measurements, estimates was made regarding the relationships between water content and elapsed time.
3.3 Results and Discussion Figure 3 shows the relationships between water content and elapsed time. This figure demonstrates that the time required to reach a steady state in volume change becomes short and the void ratio reduction of flocs becomes large as the flocs increase in size. Thus, this experiment demonstrates the influence of the flocs’ size on their dewatering properties, a characteristic that has been previously observed in the treatment plant. Table 1. Physical properties of Bentonite. 3578 Water content w (%) 2.58 Density ps (Mg/m3) 406.7 Liquid limit wL(%) 28.3 Plastic limit wp (%) 378.4 Plasticity index Ip 10.1 PH Sand (2mm-75mm) 0 Silt (75mm-5 ,U m) 12 88 Clay (5 ,U m-)
Table 2. Additive volume of flocculant and grain size of flocs. Floc size
LAC
large medium small
0.09% 0.19% 0.28%
Polyacrylamide 65PPm 65PPm 65PPm
Figure 2. Several type of flocs to floccurants (a) small size, (b) medium size, (c) large size, and (d) without additive.
Grain Size @ 10mm@ 3-7mm @ 1-2mm
4 CONSTANT RATE OF STRAIN CONSOLIDATION TEST 4.1 Introduction Using gravity dewatering experiments, the changes in consolidation properties due to size of flocs was 618
confirmed, which had been observed previously at the treatment plant. The constant rate of strain consolidation test (CRS) was then performed on the three flocs of different sizes to investigate each consolidation property under a wide pressure range.
4.2 CRS Test Procedure The CRS test apparatus consisted of the loading apparatus, the consolidation cell, and loading piston. A guide ring of 3.5 cm was fixed on the 0-ring in order to obtain a specimen of sufficient height, since the flocs were highly compressible. The inside surface of the consolidation ring was greased to minimize friction during loading. Each of three types of flocs prepared in Chapter 2 was poured into a consolidation ring up to a depth of 30 to 40mm and settled. The axial load was monitored using the pressure transducers, the pore pressure at the impervious base was measured using a pore water transducer and the deformation of the specimen was monitored using dial gauge. These data were monitored and recorded using a data logging system. The strain rate was set at a rate of about OS(%/min), which was larger relative to the 0.01-O.l(%/min) recommended, because the excess pore pressure could not be measured until the deformation had proceeded considerably at the recommended strain rate. The applied back pressure was 10kPa. The CRS test conditions are listed in Table 3.
I-
2
350
c
0
U IO
300
:
h
c250
'D
2
'
$200
150
e,
loo%
5
10
15
Elapsed Time
20
25
30
t (min)
Figure 3. Measured water content versus elapsed time.
Figure 4. e versus log
U
,' for CRS tests.
4.3 Test Results and Discussion This large deformation at the initial state of the dewatering process would cause a loss in consolidation energy. Furthermore, the amount of flocs fed into a dehydrator at the beginning of the dewatering process would determine the amount of dewatered sludge obtained (George Tchobanoglous. & Franklin L. Burton. (1993)). Therefore, either the method for making large size flocs or the reliable prediction of this initial deformation under lower consolidation pressure would be indispensable technology for the successful operation of the Dehydrum.
The CRS test results are summarized in the form of e-log 0 v' in Figure 4. Figure 4 shows that compressibility under a consolidation pressure of lower than 10 kPa varies greatly due to the size of the floc. The Compression Index of each ~ ~ O C ' Cc S under such consolidation pressures are about 39.6 for the small size, about 21.8 for the medium size, and about 10.2 for the large size. However, the remarkable differences in compressibility among the three flocs were not observed under a consolidation pressure of higher than 10 kPa. As a result, the flocs in the Dehydrum would be dewatered under consolidation pressures of lower than 10 kPa, where the remarkable differences in compressibility due to size of floc could be observed during a series of CRS tests. From the e-log o v' curves in Figure 4, it was found that the compressibility of large size flocs maintains an approximately constant value under a wide consolidation pressure range and firm flocs have already been formed during the flocculation for large flocs. It is clear that the large deformation ofthe tToc aggregates occurred with a decrease in floc size. For example, the void ratio of small flocs decreased to approximately one-fourth from that at the beginning of the test to a consolidation pressure up to 100 kPa.
Table 3. CRS test conditions. Flock size large Rate of strain E (.%/min) 0.50 Initial height 1.99 Ho (cm) Final height H n (cm) 0.25 Initial water content wn (%I 105.2
619
medium
small
0.57
0.50
1.75
1.96
0.29
0.17
80.0
36.7
5 HYDRAULIC CONSOLIDATION TEST 5.1 Introduction The gravity dewatering experiment (Chapter 3) and a series of CRS tests (Chapter 4) shows that both the method for making large flocs and the reliable prediction of initial deformation during a dewatering treatment under lower consolidation pressure would be effective for improvement of the dewatering efficiency of Dehydrum. It is possible to manage the size of flocs by means of managing the additive volume of flocculant to some extent, as shown in Chapter 2. However, since construction waste sludge generally contains some chemicals, the flocculation of the sludge at the plant site would become much complicated. In this Chapter, the hydraulic consolidation test proposed by Imai (1979) was performed on flocs of three different sizes in order to confirm the test’s effectiveness for the reliable prediction of floc deformation under low consolidation pressure.
5.2 Hydraulic Consolidation Test Imai (1979) proposed a new type of consolidation test called the “hydraulic consolidation test” for the measurement of the compressibility and hydraulic conductivity relationships of soft soils. In principle, the hydraulic consolidation test is performed by applying a downward hydraulic gradient across a soil specimen in a rigid-wall consolidometer. Seepage forces consolidate the soil and produce a nonuniform effective stress distribution within the specimen. Once steady flow conditions are reached, local pore pressures are measured using needles that are inserted into the specimen from underneath. The distribution of the void ratio at a steady state is determined based on local water content measurements that are obtained by slicing the specimen after it has been removed from the cell. From these measurements, relationships for void ratio, vertical hydraulic conductivity, coefficient of volume change, and coefficient of consolidation can be obtained as a function of vertical effective stress.
On the other hand, Fox and Baxter (1997) assumed that the logarithms of vertical hydraulic con ductivity (k,) varies linearly with the logarithms of vertical effective stress (a’,).They proposed an analytical method that eliminated the need for measurements of local pore pressure using needles, which had been problematic in the original hydraulic consolidation test proposed by Imai (1979). Assuming the logarithms of vertical hydraulic conductivity (k,) varies linearly with the logarithms of vertical effective stress (a’,)following equation can be written:
where A = -d log kvl d log a’,,and kv,= vertical hydraulic conductivity corresponds to an arbitrary reference stress aya. For this assumption, the results of constant rate strain of consolidation test for bentonite flocs in Chapter 4 indicate that a linear log k, versus log d,is a good approximation (Fig. 5). Therefore, the author adopted the method of data analysis proposed by Fox and Baxter (1997), thus the measurement of local pore pressure was eliminated in this study.
5.3 Hydraulic Consolidation Test Procedure The hydraulic consolidation test was performed on flocs of three different sizes, in accordance with the method proposed by Imai (1979). The hydraulic consolidation test apparatus consists of a consolidometer, air compressor, air regulators, pressure gauge, head tanks, and double-tube flow meters. The consolidometer used in this study has dimensions of 10 cm in diameter and 15 cm height. In this study, the compressibility and hydraulic conductivity of a bentonite floc specimen were measured using a two-stage flow procedure (Fox and Baxter (1997)). To begin with, the flocculants were added to bentonite slurry to prepare a bentonite floc specimen and then poured into the consolidomter, paying careful attention to avoid entrapping air bubbles or rupturing the flocs. 80 70
60 50
.E! 40 c. cd L-
2
30
0 20
10 0 0 001
Figure 5. Measured hydraulic conductivity versus vertical effective stress for 6 types of bentonite flocs.
0 1
001
1
Vertical effective stress
10
(kPa)
0 V
Figure 6. e versus log
620
0
for hydraulic consolidation tests.
1o2
6 CONCLUSIONS
Figure 7. Comparison of hydraulic consolidation test with CRS test in terms of e versus log U form.
Same back pressure (10 kPa) was applied to the two head tanks. For the first flow stage of the test, a downward hydraulic gradient was applied across the specimen only due to local water head to make (ufuh) (kPa) approximately 5.0 kPa, where U,, U, = the pore pressures at the top and bottom of the specimen. Once the steady flow was reached for flow stage 1, the specimen height H(m), discharge velocity v (m/sec), and (ut- uh)(kPa) value were recorded. The hydraulic gradient was then increased up to twice for the second flow stage of the test by adjusting the air regulator. Specimen height, discharge velocity and ( u t - uJ value were again recorded at a steady state for flow stage 2. Then, sampling was performed by inserting a sampling tube vertically into the specimen (a tube of 35 mm in inside diameter, 120 mm length and 1 mm thick). After the tube was removed from the specimen, a piston was inserted into the tube from its bottom to push the specimen out of the tube gradually. Then, water content distribution was obtained by measurement of water content of sliced sample column.
5.4 Test Results and Discussion The test measurements at steady flow state for each flow stage are given in Table 4 to Table 6. The hydraulic consolidation test results are summarized in the form of e-log o v fin , Figure 6. From the e-log o "'curves in Figure 4,the tendency for the compressibility of flocs to decrease with increases in size of flocs was recognized under a consolidation pressure of lower than 10 kPa, which was the same tendency as observed during CRS tests. The results of CRS tests and hydraulic consolidation tests were demonstrated in Figure 7 which indicates that the hydraulic consolidation test is a suitable method for estimation of the consolidation characteristics of flocs under much lower consolidation pressures. Therefore, the hydraulic consolidation test could reliably predict initial deformation under lower consolidation pressures, thus improving the dewatering efficiency in the plant.
The above types of the dewatering experiment and the consolidation tests for bentonite flocs demonstrate the following conclusions: (1)The gravity dewatering experiment shows the influence of flocs size on their dewatering properties, which had been observed previously in the treatment plant. (2)The CRS test results revealed that the flocs' compressibility differs greatly according their size under a consolidation pressure of lOkPa and the smaller the size of flocs become, the greater the flocs' initial deformation. (3)The hydraulic consolidation test shows that the flocs' compressibility tends to decrease with the increase in floc size under consolidation pressures lower than 10 kPa, and the same tendency was observed in CRS tests. (4)A reliable prediction of initial deformation under lower consolidation pressure would be accomplished using the hydraulic consolidation test. Table 4. Summary of test results for two stage hydraulic consolidation test of small floc. Test measurement U, - u b
(kPa) o',,(kPa) O1"b ( k W H (m) v (m/sec)
Flow stage 1 4.9 0.001 5.50 0.0607 8.1 x 10-fi
Flow stage 2 9.8 0.001 10.4 0.0575 1.1x lo's
Table 5. Summary of test result for two stage hydraulic consolidation test of medium floc. Test measurement ub (kPa) a'- (kPa) u'vb (kPa) H (m) v (m/sec)
U,-
Flow stage 1 4.9 0.001 5.46 0.0568 1.5x 10.~
Flow stage 2 9.8 0.001 10.3 0.0578 1.6 x 10-5
Table 6. Summary of test result for two stage hydraulic consolidation test of large floc. Test measurement U, - U,, (kPa)
(Ha) O1"b ( W H (m) v (m/sec) o'vt
621
Flow stage 1 4.9 0.001 5.46 0.0568 1.5 x 10-5
Flow stage 2 9.8 0.001 10.3 0.0558 1.6 x 10-5
7 REFERENCES Fox, P.J. & Baxter, D.P. 1997. Consolidation Properties of Soil Slurries From Hydraulic Consolidation Test, Journal of Geotechnical and Geoenvironmental Engineering, vo1.123, No.8, August, 1997 Imai, G. 1979. Development of a new consolidation test procedure using seepage force, Soils and Foundations, ~01.19,NO.3: 45-60 Kawaguchi, M., Horiuchi, S . & Asada, M., Horio, M. 1998. Dewatering and fusion treatment of waste bento nite slurry. Proc. 3rd Environmental Geotechnics, Sec0 e Pinto (ed.), vol. 2: 673-678 Liu, J.H., Wu, L.J. & Yang, G.S., Li, S.K. & Liu, L.L., Chen, M.M. & Zhu, R.F. 1996. A method to improve the settlement velocity of dredged soil particles. Proc. 2nd Environmental Geotechnics, Kamon (ed.), vol. 1: 373-378 Tchobanoglous, G. & Burton, EL. 1993. Waste water engineering -Treatment, Disposal, and Reuse- Ttird Edition. METCALF & EDDY, INC., McGraw-Hill Publishing Company, 621-630
622
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Application of shredded tires as lightweight backfill J. Lee, R. Salgado & C.W. Love11 Purdue University,West Lafayette, Ind., USA
ABSTRACT: The growing interest in utilizing waste materials in civil engineering applications has opened the possibility of constructing reinforced soil structures with unconventional backfills. Scrap tires are a highprofile waste material for which several uses have been studied, including the use of shredded tires as backfill. A triaxial testing program was conducted to investigate the stress-strain relationship and strength of tire chips and a mixture of sand and tire chips. The test results and additional information from the literature were used in the numerical modeling of wall backfills, both unreinforced and reinforced with geosynthetics. The numerical modeling results suggest tire shreds, particularly when mixed with sand, may be effectively used as backfill. The triaxial samples were 150 mm in diameter by 300 mm high. The samples were’tested using an MTS Soil Testing System connected to a data acquisition system. Dry unit weights of rubber-sand and tire-chip samples were approximately 12.5 and 6.3 kN/m3,respectively. Triaxial tests were conducted at confining pressures equal to 28, 97 and 193 kPa under consolidated-drained condition. Two or more tests were conducted at each confining pressure to ensure repeatability of results.
1 INTRODUCTION It is estimated that 1 to 2 billion scrap tires have been disposed in huge piles across the United States. An additional 250 million tires are discarded every year. Almost 30% of these scrap tires wind up in overcrowded landfills and thousands more are left in empty lots and illegal tire dumps. Since rubber tires do not easily decompose, economically feasible and environmentally sound alternatives for scrap tire disposal must be found. One possible application consists of using shredded tires, alone or mixed with soils, as fills and backfills, which requires that the mechanical properties of such materials and their response under realistic loadings be studied. The product of tire shredding is usually referred to as “tire chips” when they are generally between 12 and 50 mm in size and have most of the steel belting removed. The terms tire shreds or rough shreds are used for larger particles. This terminology will be adopted herein. In the present paper, a laboratory study is used for a preliminary assessment of the mechanical properties of tire chips and rubber-sand (a mixture of sand and tire chips or shreds). The results are compared with previous work on similar materials. The properties are then used in the finite element modeling of a full-scale test wall with a tire-shred backfill.
2.1 Tire chips Figure 1 shows the deviatoric stress and the volumetric strain versus axial strain for the tire chips under the three confining pressures. The general shape of the stress-strain curves shows a nearly linear behavior with increasing deviatoric stress under increasing axial strain. The material did not reach a peak deviatoric stress under the different confining pressures. Tire chips show an almost linear decrease in volume with increasing axial strains. The volume change at o3= 28 kPa confining pressure varies linearly with axial strain up to 5 percent strain and stabilizes at higher strain levels. The volume change vs. axial strain relationship at o3 = 97 kPa is linear up to 15 percent strain and shows a declining rate at higher axial strains. At o3 = 193 kPa, volume change is linear on axial strain practically throughout the test. The voids within the tire chip aggregate are reduced as the axial strain increases. Under low confining pressures, the tire chip sample began to bulge at low strains. For higher confining pressures, the
2 TRIAXIAL TESTS ON TIRE CHIPS AND RUBBER-SAND Triaxial tests were performed using tire chips with a 30-mm minus size and no exposed steel belts, and rubber-sand, a mixture of Ottawa sand and tire chips. 623
been extended to greater axial strains. The differences in strength values obtained can be attributed to the different sizes of the testing materials. 3 STRESS-STRAINMODEL FOR ANALYSIS 3.1 Hyperbolic model The nonlinear stress-strain behavior of soil has a significant influence on the stresses and displacements developed within a geotechnical structure. Nonlinear elastic models can be expected to provide acceptable prediction of the soil behavior at relatively low shear stress levels. The soil stiffness modeled in this manner increases with increasing confining pressure and decreases with increasing shear stress level. A very low stiffness is assigned to a soil element with stress condition at failure. According to the hyperbolic model proposed by Duncan et al. (1980), the deviatoric stress ol - ojis related to axial strain E through:
where (U, - 03),,1~ = ultimate (asymptotic) value of deviatoric stress; Ei = initial tangent Young's modulus. The ultimate deviatoric stress is related to the deviatoric stress at "failure" (i.e., the soil strength) through where Rf = failure ratio, a value typically between 0.5 and 0.9. The initial tangent modulus increases with confining stress according to Figure 1. Results from triaxial compression tests for tire chips.
sample initially deforms vertically, and, at around 10 YOstrain, bulging is apparent. Ahmed (1 993) and Masad et al. (1995) have also studied the shear behavior of tire chips and granulated rubber, respectively, under triaxial conditions. The tests conducted by Ahmed (1993) did not measure the volumetric change during shear. 2.2 Rubber-sand
According 'to triaxial test results for rubber-sand (Bernal et al. 1997), the deviatoric stress tends to stabilize at increasing levels of axial strain for all confining pressures. Volumetric strain indicates an initial decrease in volume, and varying levels of dilation have been observed for the three confining pressures. Masad et al. (1995) also tested a mixture of 50% granulated rubber and 50% Ottawa sand by weight. Their results are again generally comparable to the present results, although a peak is better defined in the tests by Masad et al. (1995), which have also
where K = modulus number; n = modulus exponent; P, = reference stress (numerically equal to the atmospheric pressure) in the desired units for E,. Complete description of the mechanical response of an elastic material requires another parameter in addition to Young's modulus. The hyperbolic model permits use of the bulk modulus B, which also varies with confining stress:
?)
B = &fa(
(4)
where K, and m are dimensionless parameters. The original model was modified by Seed and Duncan (1984) with a limiting lower bound for bulk modulus as follows:
624
3.2 Determination oj'hyperbolic parameters jor tire shreds and rubber-sand Hyperbolic parameters for 50-mm tire shreds were calculated by Gharegrat (1 993) after analyzing the results of compressibility and direct shear testing on tire shreds from various suppliers. The results of Gharegrat (1993) are used in the subsequent analyses, because the 50-mm tire shreds resemble more closely the material used in the retaining wall tests described later. The parameters obtained were ycomp = 6.3 kN/m3, c = 7.6 kPa, $ = 21", K = 33.5, Kb = 24.8, n = 1.10, m = 1.09, KO= 0.33, and R f = 0.61 The hyperbolic parameters for rubber-sand were calculated from the triaxial test data presented earlier by following the procedure presented by Duncan et al. (1980). The volumetric and deviatoric stress vs. axial strain triaxial testing results were used for this purpose. The hyperbolic parameters used for the numerical analysis of the rubber-sand fills are ycomp= 12.5 kN/m3, c =17.5 kPa, 4 = 42", K = 83.7, Kb = 120,n = 0.47, m = 0.33, KO=0.7, and R f = 0.7 The coefficient of lateral earth pressure at-rest KO for tire shreds was determined to be 0.33 from the average value observed in the field wall test by Tweedie et al. (1998). The value of KOfor rubbersand was obtained as 0.7 through compressibility testing (Bernal 1996). 4 RETAINING WALL TEST BY THE UNIVERISTY OF MAINE 4.1 Wall tests Tweedie et al. (1998) used a full-scale wall at the University of Maine to measure the pressure exerted by a tire shred fill under both at-rest and active conditions. The testing facility had a height equal to 4.8 m and a length equal to 4.5 m. Three different types of tire shreds were tested. The first two types, Pine State Recycling and Palmer Shredding, consisted of 76-mm maximum size pieces and a mixture of steel and glass belted tires. The third type, produced by F&B Enterprises, has 50-mm maximum size pieces with most of the belts removed. The wall facing was not allowed to move during construction and compaction of the backfill, nor while the loads were applied, to simulate at-rest lateral pressure conditions. Horizontal pressures were measured under the surcharges equal to zero, 12, 24 and 36 kPa using three pressure cells installed on the front wall panel at the elevations equal to 0.51 m, 2.08 m, and 3.86 m from the bottom. After applying full surcharge, the active condition was simulated by allowing the front wall to rotate around the bottom support. The maximum surcharge was also removed and then reapplied two to three times, to see the effects of repeated reloading. 4.2 Numerical modeling of wall tests The finite element analysis of the tire shred fill was done using ABAQUS with specifically written sub-
routine for the hy_perbolic midel previously described. The finite element meshes were prepared with the same dimensions as the field-test facility, including the 10 cm of wall thickness. Both the front and back walls were modeled as made of a relatively rigid material (with high stiffness compared with the backfill) following a linear elastic model. No displacement was allowed at the front and back walls for the at-rest condition. Two different meshes were used: one for modeling the unreinforced and the other for modeling the reinforced backfill using geotextile. According to Huesker Inc. (providers of the geotextile used), the ultimate widewidth tensile strength achieved at 9% strain (elongation at break) is 16.4 kN/m, the cross sectional area is 0.00064 m2, and the weight is 0.44 kg/m2. These basic properties of the geotextile were used in the analyses. Interface elements were used between geotextile reinforcements and the backfill material, as well as between the backfill and the wall facing. The friction angle between the front wall and the tire shred backfill was 32", based on field test results by Tweedie et al. (1998). Because the back wall was made of timber cribbing supported by a steel frame with 2-in gap between each row of cribbing, the frictionless interface was assumed between the back wall and the tire shred during surcharge for the atrest condition. Six layers of geotextiles were used between fill lifts 1 and 2 , 3 and 4 and so on up to the last geotextile layer placed between fill lifts 11 and 12. For the at-rest condition, the walls were assumed to be fixed throughout the analysis. The active condition was simulated in such a way that the front wall was hinged at its base and horizontally fixed by a highstiffness bar at its top until the surcharge was fully applied. After the full surcharge of 36 kPa was in place, horizontal displacements equal to 6.7 cm and 14.2 cm, corresponding to the 0.8" and 1.7" rotations applied in the field test, were applied at the top of the front wall. Further numerical analysis was done of a tire shred fill, a geotextile-reinforced tire shred fill, a rubber-sand fill, a geotextile reinforced rubbersand fill, and a sandy gravel backfill to check the validity of the results and compare the performance of these different backfills. The program was run under the same loading and boundary conditions for all cases and the results were compared. The parameters for the sandy gravel backfill were based on typical values given by Duncan et a1 (1 980): ycomp = 20 kN/m3, c = 0, 4 = 32", K = 500, Kb = 350, n = 0.4, m = 0.25,Ko = 0.5, and R f = 0.7 Figure 2 shows the deformed meshes used to model the wall test with tire-chip backfill for the atrest and active conditions, where the darkened areas of the meshes represent undeformed mesh shapes before application of surcharge. The deformations of the mesh shown in Figure 2 were exaggerated for better visualization. Because different friction angle values were used for the front and back wall, the deformed mesh for the at-rest condition shows asymmetric shape. The settlement of the backfill surface and the horizontal pressure acting on the wall
625
Figure 3. Measured and F.E. results with 36-kPa surcharge: (a) average settlement vs. applied pressure and (b) horizontal pressure on wall under at-rest condition.
Figure 2. Deformed finite-element meshes for tire-chip backfill with 36-kPa surcharge under (a) at-rest condition and (b) active condition.
under 36-kPa surcharge can be seen for the at-rest The pressure-settlement condition in Figure 3. curves in Figure 3(a) were initialized at 6-kPa surcharge, as was done in the field test. The measured and predicted average settlements plotted in Figure 3(a) were obtained from the average values of the settlement observed fiom the settlement grid in the wall test, and corresponding places in the finite element model, respectively. The predicted settlements for the unreinforced case in Figure 3(a) agree reasonably well with measured values, but underpredictions (of as much as 15%) are observed for the larger surcharges. One possible reason for that difference may be the time-dependent changes in stress
and settlement, which were indeed observed in the field test. It can also be seen that the reinforcement of the backfill has very little effect on the vertical settlement. The predicted values of horizontal pressure on the wall in Figure 3(b) show overestimated results by average 20 - 30% compared to the measured horizontal pressure. The degree of overestimation was relatively small at the top of the fill and becomes larger with increasing depth. The effect of reinforcement on horizontal pressure on the wall is observed to be quite significant. Although the wall facings were fixed, allowing no displacement during surcharge, the frictional resistance between the reinforcement and the backfill tended to cause elongation of reinforcement, which in turn caused the decrease of horizontal pressure on the wall. Approximately 20 - 40% of decrease in horizontal pressure resulted by using reinforcement. Figure 4 represents the distribution of horizontal pressure on the wall under the active condition for two wall rotation angles, 0.8" (Figure 4(a)) and 1.7" (Figure 4(b)). The measured horizontal pressure was obtained 1 day after the rotation was applied. 626
Figure 4. Measured and F.E. results with 36-kPa surcharge: (a) horizontal pressure on wall with 0.8" rotation and (b) horizontal pressure on wall with 1.7" rotation.
Overall, predicted horizontal pressure values were higher than measured values, most significantly so at the bottom of the wall. Both the field test and the predicted results for the tire shred backfill indicate that larger rotation angles (or horizontal displacements) may be required for the full development of the active condition. The effect of reinforcement on horizontal pressure under an active condition was also significant. The relatively uniform distribution of horizontal pressure of reinforced backfill down to an elevation of about 2 m, seen in Figure 4(b), was due to the local separation between backfill and wall. The effect of the reinforcement in reducing the lateral pressure on the wall is minimal for the at-rest condition. This result was expected, as the displacements required to mobilize the pullout force do not develop under this condition. The active case shows large deformations of the tire shred fill near the wall surface; the deformations are reduced as the distance from the wall increases. In this case, the reinforcement significantly reduces the horizontal pressure on the wall.
Figure 5. Finite-element results with 36-kPa surcharge: (a) settlement of backfill surface and (b) horizontal pressure on wall with different backfill materials under at-rest condition.
The test results by Tweedie et al. (1998) and the numerical analysis presented in this study indicate the economic advantage of using tire shreds as a backfill material, leading to lower pressure on the walls and smaller wall thickness when compared with conventional backfills. One potential problem of the use of tire shreds as a backfill material, however, is the considerable amount of settlements that may be caused by surcharge. The use of rubbersand can minimize the settlement problem. Figure 5 shows the settlement of the backfill surface and horizontal pressure on the wall for a rubber-sand and a gravel backfill for 36-kPa surcharge. The backfill surface shown in Figure 5(a) is symmetric because the same fiction angle was used for both the front and back wall facings. The settlement of the rubber-sand backfill was about 1.6 cm at the middle of the backfill surface, which is much smaller than that observed for a tire-shred backfill (about 33-cm average settlement in the field wall test). The results of the compressibility tests on rubber sand using a large PVC tube (as described earlier) would suggest a settlement much larger than the 1.6 cm obtained using the numerical analysis with the hyperbolic model.
627
This underprediction may be due to the different values of interface friction angle in the laboratory test and the one used in the numerical analysis. The compressibility test results showed that the interface friction angle between rubber sand and the PVC pipe varied in the range of 25" - 3 1O for stress levels (0 35 kPa) comparable with those observed in the numerical analysis of the rubber-sand backfill. In the numerical analysis, however, constant value of interface friction angle equal to about 32" was assumed. The other possible reason for the underprediction is the neglect in the numerical analysis of self-weight compression, which occurs prior to the placement of the surcharge (Lee et al. 1999). Although the settlement observed for rubber-sand is still larger than for a conventional gravel backfill, the horizontal pressure on the wall was observed to be smaller for rubber-sand than for the gravel backfill. Further reduction is observed if reinforcement is added.
5 SUMMARY AND CONCLUSIONS A large number of used tires are disposed of every year. A more productive, environmentally desirable use of these tires would be the construction of embankments and backfills with tire shreds or mixtures of tire shreds and sand (rubber-sand). Such fills are lighter than traditional soil fills. Additionally, the present study shows that the strength of these materials are usually adequate for such applications. Stiffness and strength properties for tire shreds and rubber-sand were established with basis on a laboratory testing program. It was observed that tire chips and shreds show a nearly linear relationship between deviatoric stress and axial strain at the confining stresses used in this study. The volume change relationship was also nearly linear, except at the lowest confining stress, for which a constant volume condition was reached at large axial strains. Mixtures of sand and tire chips (rubber-sand) present a response intermediate between those of pure sand and pure tire chips. Rubber-sand has an initial tendency to contract, followed by dilation. This is the typical response of a sand, but the range of strains for which there is contraction is wider than for sands, and dilatancy is much less. The observed results are in general agreement with the results from other authors. The laboratory test results were used to establish the parameters required for the hyperbolic modeling of these materials. A finite element analysis was conducted to.mode1 the performance of tire shred and rubber-sand backfills, and geotextile-reinforced backfills under at-rest and active conditions. The results were compared with field tests performed by Tweedie et al. (1998) at the University of Maine. The finite element analyses produced reasonably good estimates of deformations and stresses for a tire shred backfill under at-rest condition, while showing overestimation for the active condition. In addition, the analyses indicated that the performance of rubber-sand, being both lightweight and reasonably strong, compared well with a sandy gravel, as a backfill material. 628
REFERENCES Ahmed, I. 1993. Laboratory study on properties of rubber roils. Report No. FHWA/IN/JHRP-93/4, School of Civil Engineering, Purdue University, West Lafayette, Indiana. Bernal, A. 1996. Laboratory study on the use of tire shreds and rubber-sand in backfills and reinforced soil application. Ph.D. thesis, Purdue University. West Lafayette. Indiana. Bernal, A., Salgado, R., Swan Jr., R.H. and Lovell, C.W. 1997. Interaction between tire shreds, rubber-sand and geosynthetics. Geosynthetics International. 4(6): 623-643. Duncan, J. M., Byrne, P., Wong, K. S. and Mabry, P. 1980. Strength, stress-strain and bulk modulus parameters for finite element analyses of stresses and movements in soil masses. Geotechnical Engineering Research Report No. UCB/GT/80-01. University of California. Berkeley. California. Gharegrat, H. 1993. Finite element analyses of pavements underlain by a tire chip layer and of retaining walls with tire chip backfill. A4SCE Thesis. University of Maine. Orono. Maine. Lee, J.H., Salgado, R., Bernal, A., and Lovell, C.W. 1998. Shredded tires and rubber-sand as lightweight backfill. J. of Geotech. And Geoenviron. Enging. ASCE. 125(2):132- 14 1. Masad, E., Taha, R., Ho, C. and Papagiannakis, T. 1995. Engineering properties of tire/soil mixtures as a lightweight fill material. Geotechnical Testing Journal. ASTM. 19(3): 293-304. Seed, R.B. and Duncan, J.M. 1984. SSCOMP: a finite element analysis program for evaluation of soil-structure interaction and compaction effects. Geotechnical Engineering Research Report No. UCB/GT/84-02. Univ. of California. Berkeley. California. Tweedie, J., Humphrey, D.N. and Sandford, T. 1998. Tire chips as lightweight backfill for retaining walls - phase 11." Report to the New England Transportation Consortium. Univ. of Maine. Orono. Maine.
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Minimization of heavy metal leaching effect from dredged sediments involving reclamation Masashi Kamon -Disaster Prevention Research Institute (DPRI),Kyoto UniversiQ Japan Takeshi Katsumi - Department of Civil Engineering, Ritsumeikan University,Kusatsu, (Formerly Disaster Prevention Research Institute (DPRI),Kyoto Universiq, Japan)
Naoki Sawa - JGC Corporation, Yokohama (Formerly Department of Civil Engineering, Kyoto University, Japan)
Keijiro It0 - Department of Civil Engineering, Kyoto University,Japan
ABSTRACT: A concept to minimize the negative effect of heavy metal leaching on geo-environment from bottom sediments due to dredging/reclamation works is discussed, based on the experimentally evaluated leaching characteristics of Zn from the clay slurry. Batch leaching tests are conducted on Fukakusa Clay (LL = 47.4%, PL = 26.0%, CEC = 23.6 meq/lOOg) suspension doped with Zn (Zn(N0,)I.6H,0) as an index heavy metal. Also, the consolidation leaching test is proposed to examine the leaching mechanism in the dehydration process. These experimental results show that (1) the relationship between Zn adsorption per 1 g soil and equilibrated concentration of clay suspension varies with extracting volume ratios (US),(2) Zn adsorption per 1 g soil is dependent on the doped mass of Zn per 1 g soil regardless of US, and (3) consolidation has no effect on the leaching of Zn. Based on the above results, leaching mass of heavy metals associated with dredging, dehydration, and reclamation is parametrically predicted to assess the effects of water contents befordafter dredging and dehydration. 1 INTRODUCTION Bottom sediments in harbors and coastal sites are dredged in large quantities (approximately 3x 1O7 m3/y) to maintain the navigable waterway and purify the sea water quality in Japan. In view of the limited capacity of landfill sites to dispose dredged sediments, the effective reuse in geotechnical applications such as reclamation is strongly encouraged. Prior to reuse of these dredged sediments, the adverse environmental impact should be assessed since the sediments might contain toxic substances such as heavy metals and leaching of such substances from the reclaimed sediments affects the geo-environment. In particular, the bottom sediments from highly industrialized coastal area may contain lead, chromium, copper, mercury, cadmium, and arsenic which originate from the industrial activities. In considering geo-environmental impact due to the leaching from high water content geo-materials, the effect of change in water content before and after the reclamation has to be taken into account (Kamon et al. 2000). Because the sediments are usually dredged at a extremely high water content (500-lOOO%), dehydration is an effective way prior to the reclamation (Kawachi et al. 1996; Kamon et al. 1998a), otherwise significant consolidation will occur after reclamation. Thus, the chemical transport associated with consolidation has been an important geotechnical issue for sludge-state waste reclamation (e.g. Carrier et al. 1983; Gibson et al. 1995).
In particular evaluating whether the pore-water squeezed due to consolidation contain a significant amount of heavy metals is essential. Also, the chemical condition, such as redox, has to be considered, since dredging and reclamation processes above water level may not be able to maintain the reduction condition under which the metals are rather immobilized (Yoshinaga 1995; Kamon et al. 1998b). Thus, the goal of this study is to minimize the adverse environmental impact of heavy metal leaching from dredged sediments involving dehydration, reclamation, and consequent consolidation based on the experimental results of batch and consolidation leaching tests. Table 1. Properties of Fuk rkusa Clay (under 75 Km) Particle density 2.72 &m3 Liquid limit 47.4% Plastic limit 26.0% Silt fraction (5-75 pm) 38.2% Clay fraction (< 5 pm) 61.8% CEC 23.6 meq/100g Exchangeable cations Na 0.01 meq/100g 13.5 meq/lOOg Mg Al 1.4 meq/l OOg K 0.06 meq/lOOg 8.7 meq/100g Ca
629
BLT was conducted basically in accordance to the JLT 46 method prescribed by the Environmental Agency, Japan. Zn(NO,), 6H,O solutions at the concentrations of approximately 100, 500, and 1000 ppm were mixed with Fukakusa Clay at US (liquid per solid ratios) of 1, 3, 5, 10, and SO. The obtained clay suspensions/sludges containing Zn simulate dredged sediments. After shaking the sludge for 6 hours, Zn concentration of the filtrate was determined using ICP (Inductively Coupled Plasma) at equilibrium concentration. The mass of Zn adsorbed on the clay can be calculated from the equilibrium concentration determined. CLT was proposed to reveal the leaching mechanism in the dehydration and consolidation processes. The oedometer used for CLT was newly developed to obtain the squeezed water during consolidation (Figure 1). The clay mixed with the equal weight to water (which means 100% water content) containing 100, 500, and 1000 ppm Zn(N03);6H,0 was poured into the oedometer. The vertical pressure of range 1.48 kPa to 90.16 kPa is being increased incrementally after the 90% of degree of consolidation is achieved for each pressure. Squeezed/drained water was collected separately for the first and last half, and Zn concentration was determined using ICP. 3 EXPERIMENTAL RESULTS
3.1 Batch leaching test (BLT)
2 EXPERIMENTAL PROCEDURE 2.1 Muterials Fukakusa Clay, a dry powder processed clay, was used for the leaching tests. Only the fraction smaller than 0.075 mm in particle diameter was used. The main properties are listed in Table I , and the dominant clay minerals present in the soil system are illite and kaolinite. The pH values of the clay exhibited acid (pH = 3.0-3.5). Although the reduction-oxidation state is an important factor affecting the leaching of metals, only the adsorption properties against zinc was evaluated on a dry processed clay but redox effect was not evaluated in this study. Zinc was used as an index heavy metal. Zn(NO,)? 6 H 2 0 solutions at several concentrations were used. The pH value of the solutions exhibits a weak acid. 2.2 Leaching tests Two different types of leaching test were conducted; batch leaching test (BLT) and consolidation leaching test (CLT). 630
The results of BLT shown in Figure 2 illustrate that the mass of Zn adsorbed on a unit mass of soil ranges from 0. I mg/g to 5.5 mg/g, which depend on the initial and equilibrium concentration of solutions and US (liquid per solid ratios). As the initial concentration of Zn solution increases from 100 ppm to 1000 ppm, the Zn adsorption also increases by a factor of 3-5. The increase in WS also results in an increase with the amount of Zn adsorption. In comparing the series of tests with the similar values of initial and consequent equilibrium concentrations, the increase in L/S from 1 to SO resulted in the 7times increase in the adsorbed Zn (from 0.17 mglg to 1.25 mg/g) when the initial concentration is 100 ppm, and increased by 13 times (from 0.41 mg/g to 5.52 mg/g) with initial concentration of 1000 ppm. There are several mechanisms discussed on the heavy metals immobilization in the soil-water system (e.g. Yong et al. 1992; McBride 1994). Yong et al. (1 992) summarized that the immobilization of metals are attributed to exchange to other cations on clay surface, formation of insoluble compounds such as carbonate and/or hydroxide, adsorption onto organic matters, and residual fraction. Under the experimental conditions conducted in this study, it is considered that hydroxide and carbonate did not form because
the clay-water system exhibited a weak acid. Also, the Fukakusa clay is a pure processed clay which does not contain organic fractions. Therefore, the cation exchange behavior is considered as the dominant mechanism to immobilize Zn. Calcium and magnesium ions, which initially exist on the particles surface due to the attraction by the negative charge, have been replaced by Zn++in water. Thus, the adsorption depends upon the L/S values and initial Concentration of solutions is consistent with the basic theory on ion exchange on clay surface. Figure 3 summarizes the Zn adsorption on clay in relation to the Zn doped with the clay-water system per 1 g soil. The plots show that the Zn adsorption is simply dependent on the mass of Zn doped per 1 g soil, and the regression can be expressed as:
Figure 5. Zn adsorption in CLT and BLT
MA= 0.403 M,0.649 (R = 0.988)
3.2 Consolidation leaching tests (CLT)
(1)
Figures 4 and 5 show the results from CLT for different initial concentration levels of 100, 500, and 1000 ppm. The results from BLT under US = 1 are also plotted with an equilibrium concentration to compare the BLT and CLT results, since the initial water content of the sample subjected to CLT was 100% (L/S = 1). The slurry used for CLT finally achieved water content of 30% after the vertical pressure of 90.16 kPa was applied. Although the CLT has a different extraction procedure from the BLT, the leaching concentration in CLT is almost identical to the equilibrium concentration in BLT, and consequently the Zn adsorption in CLT is equal to the one in BLT. This result might be considered susceptible, and not consistent with the results obtained from BLT. From BLT, the mass of Zn adsorbed on a unit mass of soil simply depends on the mass doped into soil-water system per a unit mass of soil. In CLT, unlike BLT, the total mass of Zn in soil-water system decreases with the proceed of consolidation, because the drainedkqueezed water contains a certain amount of Zn, but the mass adsorbed after consolidation process is equal to the value from BLT for US = 1. The reason for this might be that competing cations such as Cat+ and Mg++are also extracted from the soil-water system with the drained water. The equilibrium constant, K, which has to be taken into account here is:
where MA is the Zn adsorption on 1 g soil (mg/g), and MD is the mass of Zn doped per 1 g soil (mg/g). In the range of concentrations for which the experimental work was conducted, it can also be roughly concluded that 10% of dozed Zn is adsorbed on clay while 90% remain in the water.
6 535 3
W
4
2
e 3
3
2
2
81 0 0
10 20 30 40 50 Zn added to soil (mg/g)
60
Figure 3. Zn adsorption versus mass of Zn added to the sediments
K = [X(S)] [Zn++(R)][Zn++(S)].' [X(R)] - I
(2)
where [Zn++(S)] and [X(S)] are the concentrations of Zn and other exchangeable cations in liquid (mg/gsoil), and [Zn++(R)]and [X(R)] are the mass of Zn and other cations adsorbed on soil (mg/g-soil). If the other competing cations are drained with the squeezed water, the Zn adsorption will increase to keep K constant. As a result, CLT achieved the almost equal leached concentration and adsorbed mass to BLT. Figure 4. Equilibrium concentration in BLT to leaching concentration in CLT 631
-00
80
....... L .......
I.
.......
.......
.......
......
60
1 ;
,
40
/
I
.
*
5
,
.
.
..,
...,.......-.
...... .,..
....... ...... ? . . . . . . , . . . . . . . . . . . . . , . . . . . . 1 . . . . . .
20
I
..
,
.
10 15 20 25 30 Axial displacement (mm)
.............
......
~
45
0 20
40
60
80
5004
100 120 140
Water content achieved by dehydration (%)
Figure 6. Concentration of leached Zn in CLT Figure 6 shows the concentrations of leached Zn in the drained water which was separately collected. The initially drained water has exactly the same Zn concentration as the last half portion of squeezed water from the sample for CLT, which means that the progress on consolidation does not affect the leaching. It might be because the adsorbed Zn on soil surface is attracted by a negative charge, which is strong enough so that the consolidation pressure equal to or smaller than 90 kPa could not separate Zn+*from the clay surface. In conclusion, leachedhdsorbed amount of Zn from consolidation process can be roughly estimated from the batch leaching test with the same value of US as in the initial condition prior to the consolidation.
Figure 8. Calculated results of mass of heavy metal leached during post-reclamation Initially, the in-situ bottom sediments have a water content of w,, concentration of the target heavy metal contained of c,, and mass of the heavy metal contained of Dredging process increases water content to w, (>wo). This increase in water content is due to the fact that the grab dredger would excavate the bottom sediments with a amount of sea water which does not contain heavy metal. Thus, the heavy metal concentration after dredging, c1,is:
4 PARAMETRIC ANALYSIS 4.1 Basis
The mass of leached heavy metals involving dredging, dehydration, reclamation, and consequent consolidation was parametrically calculated under the following assumptions, using above experimental results.
w.
c, = W O c,
IW,
(3)
Before and after the dredging process, the heavy metal mass does not change (M, = MJ. Due to the dehydration treatment process, the water content is reduced from w , to w?. According to the experimental results mentioned above, the concentration of the leached heavy metals is not affected by the degree of dehydration. The squeezed water generated by the dehydration should, therefore, have the same heavy metal concentration as the pore water in dredged sediments (c2 = c,). The mass of heavy metal contained in the sediments decreased (M, < M,) because some amount of heavy metals were drained out to the squeezed water.
Figure 7. Mass of heavy metal existingin the sediments involvingdredging dehydration, and reclamation
632
After the reclamation of dehydrated sludge, consolidation will occur to attain the water content of w3 (< w?). Similar to the dehydration process, the consolidation has no effect on the adsorption of heavy metals on clay. Therefore, the concentration of heavy metals remains in the sediments, c,, is equal to c, and cl, and the mass of heavy metal in the sediments decreases (M, < M,) with a decrease in water content. Thus, the leached mass during the consolidation process (after reclamation), M,, is:
The overall process for above calculation is illustrated in Figure 7. The leached mass, instead of concentration, is taken into account in order to compare the cases with different water content with the absolute value of mass.
be advantageous for large scale of dredging and reclmation projects. The second option might be considered rougher and less technical, however, will bring out less cost investment and shorter execution period. For the above discussion, several factors were omitted; for example, treatment cost of dehydrated water containing heavy metal. Also, the discussion was made based on the experimental results on Zn adsorption test for Fukakusa Clay. Therefore, further research is needed, for example taking into account the effect of chemical conditions (redox, pH), constituents of the sediments (organic matters), and chemical interactions between heavy metals, since the experimental work presented herein was conducted under the limited conditions only considering the cation exchange as the heavy metal immobilization.
4.2 Discussion und practical implications
5
An example of the calculated results from the above mentioned method is shown in Figure 8. The leached masses after reclamation were calculated with relation to the water content achieved by dehydration process, and three different water content levels after dredging process (w, = 100, 300, and 500%) were considered. The initial condition of the bottom sediments are 100 ppm of heavy metal concentration, 100% of water content. The achieved water content of reclaimed sediments after the completion of consolidation is assumed 30 % . The mass of leached heavy metal during the consolidation is linearly related to the water content achieved by dehydration, and also related to the dredged water content. In the case that the dredged water content (w,) is low (loo%), the decrease of water content in dehydration (w,) has more significant effect on the leached mass compared to the case of high dredged water content (500%). To achieve the same level of mass leached, for exaniple 10 mg per 1 g dry soil, the dredged sediments have to be dehydrated to low water content (40%) if the effort is made to dredge the sediments at low water content (loo%), while the sediments only have to be dehydrated to 80% water content in the case that the sediments are dredged at 500% water content. Therefore, in order to reduce the mass of leached metal, there might be two possible options; ( 1 ) to make an effort in dredging sediments with a low water content and also achieving the low water content through dehydration, and (2) to dredge the sediments with a comparatively high water content. For the first option, several types of dredging equipment which have been recently developed to minimize the dredged volume, that is water content, can be used. Since the water content is low, the important advantage is that the volume of the dredged sediments treated and reclaimed is small, which might
The following conclusions were obtained: 1. From the batch leaching test on Fukakusa Clay, the relationship between Zn adsorption per 1 g soil and the equilibrated concentration of clay suspension varies with the liquid and solid ratios .(L/S). However, Zn adsorption per lg soil simply depends on the mass of Zn doped per I g soil regardless of L/S. 2. Consolidation of the vertical pressure smaller than 90 kPa has no effect on the leaching of Zn. Therefore, the leaching mass can be estimated from the results of batch leaching test. 3, From the parametric analysis, achieving the low water content through dehydration from sediments dredged with low water content is an effective way to minimize the mass of leached metal involving postreclamation. To dredge and dehydrate the sediments with a comparatively high water contents is also another possible way. 4. In this study, only the cation exchange is considered as the heavy metal immobilization mechanism. Therefore, further research is needed considering other factors affecting the leaching behavior, such as the effect of redox, pH, and organic matters, as well as the overall project cost including the treatment of the water drained through dehydration.
CONCLUSIONS
ACKNOWLEDGEMENTS
Dr. G. Rajasekaran (Kyoto University) provided helpful comments in preparing this manuscript. REFERENCES Carrier, W.D., Bromwell, L.G., and Somogyi, F . (1983): Design capacity of slurried mineral waste
633
ponds, J o u m l of GeotechnicnlEngineering, ASCE, Vol. 109, NO.5, pp.699-7 16. Gibson, R.E., Potter, L.J., Savvidou, C., and Shiffman, R.L. (1995): Some aspects of onedimensional consolidation and contaminant transport in wastes, Compression and Consolidution of Cluyey Soils, H. Yoshikuni and 0. Kusakabe (eds.), Balkema, Rotterdam, pp.8 15-832. Kamon, M., Katsumi, T., and Inui, T. (1998a): Dehydration-solidificationtreatment and geotechnical utilization of waste sludge from construction works, Environmental Geotechnics, P.S. Sec0 e Pinto (ed.), Balkema, Rotterdam, pp.603-608. Kamon, M., van Roekel, G., and Blumel, W. ( 1998b): Assessment of geo-environmental hazards from dredged materials, Environmentul Geotechnics, P.S. Sec0 e Pinto (ed.), Balkema, Rotterdam, pp. 1057- 1074. Kamon, M., Katsumi, T., and Watanabe, K. (2000): Heavy-metal leaching from cement stabilized waste sludge, Geotechnics of High Wuter Content Materiuls, ASTM STP 1374, T.B . Edil and P.J. Fox (eds.), ASTM (in press). Kawachi, T., Katsumi, T., Tran Duc, P.O., and Yamada, M. (1996): Treatment and utilization of waste sludge/slurry from construction works in Japan, Environmentd Geotechnics, M. Kamon (ed.), Balkema, Rotterdam, pp.75 1-756. McBride, M.B. (1994): Environmentul Chemistry o f Soils, Oxford University Press, New York. Yong, R.N., Mohamed, A.M.O., and Warkentin, B.P. ( 1 992): Principles of Contaminant Trunsport in Soils, Elsevier, Amsterdam. Yoshinaga, K. ( 1995): Mercury-contaminated sludge treatment by dredging in Minamata Bay, Dredging, Reclamation, and Contuinment of Contaminated Sediments, ASTM STP 1293, K ,R . Demars, G.N. Richardson, R.N. Yong, and R .C. Chaney (eds.), ASTM, pp. 182- 19 1 .
634
Coastal Geotechnical Engineeringin Practice, Nakase 8, Tsuchida {eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Permeability of light-weight soil made of dredged slurry mixed with air foam and cement Y. Kikuchi Port and Harbour Research Institute, Ministry of Transport, Yokosuku,Japan
H.Yoshino YuchiyoEngineering Company Limited, Japan
ABSTRACT: One of the effective utilization of dredged slurry is by means of forming new geo-material by mixing it with air foam and cement. The characteristics of such materials can be influenced by the sea water. In severe cases, where the permeability is high, pollution of ground and sea water remains a possibility. To check the permeability of such a material is the most important part to utilize it. In this paper, several permeability experiments and numerical calculation are carried out to discuss the permeability of light-weight soil (LWS) made-up of dredged slurry mixed with air foam and cement . The pore fluid in LWS is compressive, because of existing of air foam. To estimate ‘true’ permeability, compressibility of the pore fluid is taken in consideration in this study.
1 INTRODUCTION Light-weight soils (light-weight super geo-materials, LWS) are geotechnical materials made of high water content clays mixed with lightening materials and cementing materials. They are characterized that the unit weight is light and their shearing strength is strong. Such characteristics are favorable to backfilling materials for reducing the earth pressure to the structure. In this study, air foam is used for the lightening material. As LWS is intended to be used under water table in port area, a part of air included in the material can be exchanged for water during the long period. This kind of problem will be caused not only by water pressure but by its high permeability. The problems as the strength reduction of the material by contact with sea water or the ground water pollution through the material alkaline may occur if we use LWS. These kinds of phenomena is mainly related to the permeability of LWS. As LWS is a new material composed from saturated clay, cement, and air foam, less information has been given. The air included in LWS is in bubble state. Each bubble stays its position after the LWS solidified. Though LWS is in unsaturated state, permeability mechanism of it is different from ordinal unsaturated soil. Purpose of this study is to clarify the permeability characteristics of LWS. Permeability tests by triaxial apparatus and constant consolidation test apparatus are performed and applicability of consolidation equation of geo-material considering compressive pore
fluid is checked. Then discussion as to the relation between ‘true’ and ‘apparent’ coefficient of permeability of LWS is made.
2 PERMEABILITY THEORY FOR SOIL INCLUDING AIR LWS includes air foam in its pore. The air foam is considered to be isolated each other and its volume can be changed according to the change of the pore water pressure surrounding the air foam. In such a meaning, LWS can be treated as a material including compressive pore fluid. The consolidation of soil including compressive fluid is expressed as follows;
v=
k au
(2)
Pg az
in which p : density of the pore fluid, z: distance, t : time , 1’ :apparent velocity of the fluid , n : porosity (=
e-) , l+e
k : coefficient of permeability, U : absolute pressure of the pore fluid, g: gravity acceleration.
635
Considering the condition CJ = d +( U - 1) = const. and !?.? >>
az2
a-au =-&
k
Table 3. List of Test condition (triaxial test).
($1
case
, following equation is derived.
A-1 A-2 A-3 A-4 A-5 A-6 A-7
a2u
mvpg az2
void ratio 4.17 4.18 4.16 4.13 2.75 3.02 3.45
initial water content 108.2 107.5 107.7 105.7 98.6 109.3 126.1
gradient
22 50 99 20 20 20 21
air foam with with with with without without without
in which
a = 1+
G ,C
I,,,:
compressibility of pore fluid, m,,:
mV coefficient of volume change for saturated soil. Verruijt( 1969) showed C is expressed as follows if the compressible fluid can be modeled as the mixture of water and air.
in which C’,y:compressibility of water, S,: rate of saturation. It is reasonable that LWS can be treated as the material mentioned above to consider the permeability of LWS. 3 PERMEABILITY EXPERIMENT 3. I Specimen Preparation Marine clay dredged in Kawasaki Port is used for main material in this experiment. The physical properties are shown in Table 1. The grain size accumulation curve is shown in Figure 1. The mixing condition for making LWS is shown in Table 2. The target unconfined strength at 28 days cured qU2* is 200kN/m2. The mixture is filled up to the mold. It is put into the pressure cell filled with water before solidified for cured in the pressure simulating the under water condition. The pressure in the cell is 0,50, 100 kN/m2. Curing duration is between 14 days and 28 days. Table 1. Physical properties of Kawasaki clay
P,
density liquid limit plastic limit
2.678 g/cm3 52.1 OO/ 23.0 ?a‘
WI. WP
Table 2. Mixing condition for LWS. dry mass water cement air foam Total
Mass (kg/m’) 442 574 75 9 1100
Volumetric fraction (l/m’) 164 574 24 238 1000
3.2 Permzabllity Experiment with Triaxial Apparatus The experiment procedure is as follows. At first, the specimen cured in 100 mm high and 50mm diameter mold with 0 kN/m2 curing pressure is set into the triaxial cell without trimmed. Then the cell pressure is applied to the certain level and start consolidation. The consolidation duration is 3 days. After consolidation ended, the pressure is applied to the volume change burette connected to the bottom of the specimen. Then the water is let into the specimen. Absorption and expulsion volume of the water is measured by using volume change burettes connected to both the bottom and top of the specimen during the water flow. Water flow continues for a week. Table 3 shows the test conditions in the series. Figure 2 shows the relation between time and volume of water absorption. Numbers shown in the figure mean the applied pore pressure of the bottom of the specimen. Figure 2 shows the volume of water absorption is different from the applied pressure. Each relation curve is convex in the beginning, then it goes to linear. The result means that absorption velocity is rather high in the beginning but is getting constant later. Figure 3 shows the relation between time and volume of expulsion from the specimen. In this case, the relation curve is concave in the beginning, then it goes to linear. Figure 4 shows the volume change of the water included in the specimen. In this series, the volume change by the compression of the skeleton can be ignored. Each result shows the water included in the specimen increases in the first thousand minutes from the start of the water flow, and stays almost constant later . The amount of the volume change of the water differs from the pressure applied in the bottom of the specimen. From these findings, we can conclude that the volume of the air foam decrease according to the change of pore water pressure and it takes long time to be stable condition. The specimen made of the same amount of cement and water content without air foam is conducted water flow test for checking the effect of air foam. Figure 5 shows the relation between duration and water absorption and expulsion. The difference of the two specimen condition is air foam is included or not. The water content of the both specimen is almost the same. 636
absorption is very fast and the water content increases in the beginning,but the water content goes to constant later. These two test results show that the permeability decided from the water absorption and expulsion ‘apparent’ permeability -in the case without air foam is larger than that in the case with air foam even the smaller void ratio in the case without air foam. Figure 6 shows the change of ‘apparent’ permeability according to time. The ‘apparent’ permeability shown in this figure is decided by the water absorption to the specimen and the total cross section. The cases from A-1 to A-4 include air foam. The cases from A-5 to A-7 exclude air foam. The ‘apparent’ permeability of the cases including air foam change accordingto time in the first thousand minutes but later it goes to constant. On the other hand, the ‘apparent’ permeability of the cases without air foam is almost constant.
The specimen includes air foam in the case A-4 and does not include air foam in case A-6. Figure 5 shows that when the specimen does not include air foam, the amount of absorption and expulsion of the water is almost the same and absorption and expulsion velocities are almost constant. The other hand, if the specimen includes air foam, the velocity of water
Figure 5. Difference of water absorption and expulsion by existingOf air foam*
637
Figure 6. Change of ‘apparent’ coefficient of permeability according to time. consol i d a t i o n pressure p
(kN/m2)
4.5
4
3.5
c u r i n g pressure 100 kN/m2
a 0
x
,A!
3
\\
E
‘\
0
\\
2.5
cured 28 days 2
‘\ ,
1.5
Figure 7. e - log p curve of LWS in CRC.
.+-
0 c
c 0
Lc .+-
m 0
1.5
2
2.5
3
3.5
4
4.5
mean void r a t i o e
Figure S. Coefficient of permeability in CRC according to consolidation theory for saturated soil.
Figure
638
Coefficientof permeabilityof Lws.
3.3 Permeability Experiment with Constant Rate Consolidution Test (CRC) The mixing condition of the specimen is the same in the case of the experiment using triaxial apparatus. The curing pressures selected are 0, 50, 100 kN/m2. The mixtures are cured in 20mm high and 60mm diameter stainless rings. The back pressure in the constant rate consolidation test is selected to 100 kN/m2.This pressure is selected for the precise measurement of excess pore water pressure in the bottom of the specimen. Strain rate is 0.05 %/min. Maximum consolidation pressure applied is high enough to recognize to decide the consolidation yield stress. Figure 7 shows consolidation curves obtained by CRC tests. As shown in Figure 7, consolidationcurves are affected by the curing pressure as the higher curing pressure makes the smaller initial void ratio and the somewhat smaller compression index Cc. But the consolidation yield stresspcdoes not affected by curing pressure and the range of it is between 600 and 1000 kN/m2. Figure 8 shows the estimated relation between void ratio e and the coefficient of permeability k,, applying the consolidation theory for saturated soil. Applying the consolidation theory given for saturated soil to LWS gives bad estimation of the coefficient of volume change m V and the coefficient of consolidation cv, because pore fluid of LWS is compressive. As a result, the estimated coefficient of permeability k,,has some uncertainties. Therefore the relation shown in Figure 8 need some correction.
4 DISCUSSION In order to estimate the permeability of LWS, the simulation on the flow test and CRC is conducted. The equation used for this simulation is equation (3) shown in 2. In the test mentioned in 3.2, we observe the higher ‘apparent’ permeability in the beginning of the test in LWS with air foam. The reason of such a result is the change of pore pressure and the shrinkage of air foam before reaching steady state. Figure 9 shows the several simulation results of the flow test in the condition with different initial pore pressure. Figure 9 shows the change of amount of water absorption to the specimen against time. The result of this simulationshows that the coefficient a in the equation (3) should be suitably selected. The coefficient a is a parameter of mVand pore pressure U. The simulation results shown in Figure 9 have different initial pore pressure, because the selection of initial pore pressure has the large effect for simulation compare to mvand the bubble included in LWS might be pressured (Kikuchi et al. 1995). By the comparison between simulation results and test data, the model used here can simulate the flow test phenomena except the precise estimationof the absolute volume of the water absorption.
Figure 10 shows the simulation and test result of the ‘apparent’ permeability. The simulation results shown in Figure 10 are made iin the case that initial pore pressure is assumed 0 kN/m2. The point where ‘apparent’ permeability becomes almost constant is almost the same. This result also shows this model can explain the phenomena. The simulation is also done for CRC results. The result shows that the coefficient of permeability calculated by the theory for saturated soil in the early stage of consolidation is too high to the coefficient of permeability calculated by equation (3). On the other hand, the permeability by the theory for saturated soil is getting near to the permeability by equation (3) according to consolidation progress and finally the permeability by the theory for saturated soil is lower than the permeability by equation (3) . The permeability and void ratio are known to have a strong relation in saturated soil. We can consider that air foam has no role in water flow in LWS with air foam as mentioned in 2.2. Figure 11 shows the corrected relation between the permeability and void ratio. The permeability of LWS shown in Figure 11 is calculated with equation (3) and with considering that the cross sectional area of air foam is excluded for calculating the effective cross sectional area for water flow. Here, We define this permeability is ‘true’ permeability. Also, the calculation is done in the condition that air foam is ignored for calculating the void ratio. Figure 11 shows that the relations between coefficient of permeability and void ratio of LWS with air foam and of LWS without air foam estimated by triaxial apparatus agree well. This result shows that this kind of correction can represent the permeability of LWS with air foam in a macro point of view. Figure 11 also shows that the relation between permeability and void ratio of the Kawasaki clay used for main material in LWS and the corrected relation estimated by CRC. As Kobayashi et al. (1990) show that there is a linear relation between the permeability in logarithm and void ratio, the permeability of Kawasaki clay is extrapolated in the figure. The estimated relation by CRC is almost parallel to the relation in Kawasaki clay. And from the relation in LWS shown in Figure 11, the permeability of LWS is rather small because of mixing cement. Terashi et al. (1983) show that the reduction of permeability is affected not only by grain distribution but also by void ratio. Addition of 75 kg/m3 cement to clay make one or two order reduction of permeability (Terashi et al. 1983). The result shown in this study shows that the reduction of permeability ranges from 1/30 to 114.
5 CONCLUSION In this study, water flow test by triaxial apparatus and CRC are conducted to estimate the permeability of LWS. And we consider the water flow mechanism of LWS with air foam and make simulations on water flow. Finally, Comparisons of the test results with 639
simulation results are made. Following conclusions are drawn from this study: 1) ‘True’ permeability in LWS is consistent with the permeability of the clay with cement. 2) Apparent permeability in LWS is affected because of compressibilityof the air included in the material. 3) To estimate ‘true’ permeability of LWS by using consolidation test, it is necessary to take in consideration the compressibility of the pore fluid. REFERENCES Kikuchi, Y. and Takahashi, K. 1995. Geotechnical Topics Related on Maintenance of Port Facilities, Proc. ofAnnual Seminar ofport and Harbour Research Institute: 19 - 20. (in Japanese) Kobayashi, M. , Mizukami, J., and Tsuchida, T. 1990. Determination of the Horizontal Coefficient of Consolidation c,,, Report of Port and Harbour Research Institute, 29(2): 63 - 83. (in Japanese) Terashi, M. ,Tanaka, H. , Mitsumoto, T., Honda, S., and Ohhashi, T. 1983. Fundamental Properties of Lime and Cement Treated Soils (3rd Report). Report of the Port and Harbour Research Institute, 22( I): 69 - 96. (in Japanese) Verruijt, A. 1969. Elastic storage of aquifers, Flow Through Porous Media: Chapter 8, Academic Press.
640
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)@ 2000 Balkema, Rotterdam, ISBN 90 5809 151 1
Development of a construction method for revetment using fly ash and cement treated sand S.Kitahara, M.Okazaki & Y.Wakayama TomakomaiPort Construction OfJice,Hokkaido Development Bureau, Japan
H. Kobayashi & N.Tanaka Hokkaido Electric Power Company Incorporated, Ebetsu, Japan
ABSTRACT: A construction method of revetments using cement and fly ash treated sand was developed in this study for the purpose of re-utilization of industrial waste and reduction of construction cost. I n this method, silty dredged sand is mixed with fly ash, cement, and seawater. With this method, it was found that the construction cost could be reduced by about 30% compared to general construction method using rubble stones. Because the treated sand is placed underwater before hardening, it is necessary for the materials not to be segregated in under-water condition and to reveal the required strength after placement. The long-term strength of the treated sand is investigated with the core sample collected after one year.
1 INTRODUCTION To dispose of dredged sand, a dumping site of about 24 ha was built in the West port District of Tonlakomai port. A settling basin was needed at this dumping site to prevent water pollution caused by sand landed by pump dredging. Usually, such a settling basin was constructed using a rubble-stonerevetment to divide the basin from the dumping site. However, in this study, the revetinent in the basin of the West Port District was constructed using treated sand. This was a mixture of silty dredged sand, fly ash, cement, and seawater. In the construction, it was placed underwater before hardening. Therefore, it is necessary to keep un-segregated in under-water condition and to reveal the required strength. The mix proportion was examined by trial mixture in the laboratory. In Japan. it is predicted that fly ash as an industrial waste will be produced more than 10,000,000 t in AD 2000.The effective utilization of fly ash is demanded, from the viewpoint of resoiirce recycling. The fly ash used ill this study was produced in the coal power plant of Nokkaido Electric Power Co., Inc. in the East Port District of Tomakoniai Port. The method proposed in this study is effective in reducing construction cost, and in saving rubble stone and other natural resources.
dard cross section, which also indicates construction A sand mound Of - ” in thickness was constructed with dredged sand taking account of a Safety factor Of the Circular failure, fT0111 Which1 the revetrnent height was limited to 51n. In order to achieve a slope stability, the revetinent was surrounded wit11 perl1leable sheets.
2 OUTLINE OF CONSTRUCTION The construction was carried out froin September to October, 1998. The revetment was extended to 21 4m. Fig. 1 shows the construction site and a stan-
Figure1 Construction site and standard cross section (unit: m)
64 1
Dredged Sand 2807
Fly Ash
4
0
81
8
I0
82
5
I0
14 2
19 0
1770
1486
Specific Surface by Blaine Test (crn2/g)
-
2.760
Ignition Loss (%)
-
1.500
Density of Soil Particle (g/cm') Gradation Test
Gravel Fraction over 2mm (%I Sand Fraction 0 075mm -., 2mm (94) Silt Fraction 0.005n~ni 0 075mm (%) Clay Fraction under 0 005nim (%) Optin~umMoisture Content
-
Figure:! Stepped revetment construction Compaction Test
Since efficient niixing of the materials and uniformity of treated sand are required, a plant systeni was adopted for the mixing, taking into account the large daily mixing and adopting a mechanism for ad-jiisting the water content of the mixture. This plant had a material feeder that could adjust the proportion of mixture components by monitoring the properties of the material. It also had a 2-shaft mixer with high mixing efliciency. The prod~ictivityof this plant was 125 in3/hr. For the revetnient construction, the treated sand was placed in 3 steps. The procedure of the Is' and ZiId step enibankments is shown in Fig.2. The treated sand was transported by dump truck from the tnixing plant. However, the shear strength to allow 2 dump truck traffic was not achieved until 6 days after construction. (The required shear strength was calculated by circular failure was 1OSkN/d.) Therefore, the treated sand was transported by dump truck front the plant to a section of the revetment, which was constructed 6 days before. Then, the treated sand was forwarded to the Backhoe 1 by the Backhoe 2, as shown in Fig.2. The 1'' step embankment was constructed of treated sand placed by the Backhoe 1. The Backhoe 1 was stepped on the revetnient constructed 1 day before and placed the treated sand into the sea. The width of the 1'' step embankment was decided considering that the Backhoe 1 could be stepped on it one day after the co~is~ruction. Its crown elevation was + l . 6 nt, which is higher than the high water level. The 2'Id step etiibaIikinent was constructed to the required width (12 in) by widening the 1'' step embankment. Finally, the 3rd step embankment was constructed to the designed crown height (1-2.0 m).
2364
(%)
Maximum Dry Density
I (g/cm3)
T MIX ~ O PNR O P O R T ~ OBY ~ 3 D ~ T ~ R ~ I ~ AOF TRIAL MIXTURE IN THE LABORATORY EXAMINATION 3.1 Reqzrired q ~ i a ~ i ~
In order to construct the revetinent of the treated sand, it should have the following quality. (1) The treated sand should be tin-segregative in the water. (2) A designed strength should be revealed. The designed strengths o f the Is' and 2Itdstep embankments were calculated by the stability of the revetment under construction. The required shear strength was determined by circular failure analysis. However, in the execution control, the unconfined compression strength was used based on the relation as follow. The results are shown in Tablel. c = qu/2 where c = Shear Strength: qu strength
(1) =
tinconfined compression
In the calculation, the load of l.0-ni3-class backhoe was used for the calculation of the strength at 1 day; 1Ot dump truck load was used for the calculation of the strength at 6 days. 3.2 Examiiiafion and consideration of proportioiiing
Table 1 Standard design strength
3.2. I h 4 a f e ~ ~ a ~ The sand used for the proportioning examination was dredged in 1995 or 1996, and was stored. The fine-grain fraction less than 0 . 0 7 j m ~of~the dredged
6OkNi 1'
642
2o
1
0
y =
1
1 . 6 3 4 5 ~+ 11.936
1
2
M I n i -p
3
4
5
lump (em)
Figire3 Relation of table flow and mini-slump
3.2.2 Examination method and f o m m g method of test piece (1) Fluidity test The fluid it^^ of treated sand is usually measured by flow test (JIS A6201). However, the mini-slump value (JIS AlI73, slump corn height H = 15 cm), which can easily evaluate the fluidity in the field, was used in this study. The flow value and minislump value are related as shown in Fig.3. (2) Under~~jater segregation test There is no standardized test to examine the degree of segregation u n d e ~ a t e r .In this study, the degree is defined the turbidity. The lower turbidity, the higher perfo~naiiceon un-s~gregationis obtained. That test was carried out as shown in Fig.4. A sample was put into a cylinder of transparent vinyl chloride, in 6 steps. The sample in one step was about 33 c1n3. The water was taken after 2, 4 and 6 steps each, and turbidity was measured based on JIS KOIOI. Turbidity was defined by the strength of transmitted light. (3) Specimen fbr unconfined Compression Strength Test The specimen made in air was 5 cm in diameter and 10 cni in height, based on JGS T82 1. The specimen made underwater was also tested to examine underwater strength. The specimen (12.5 cm in diameter, 25 cni in height) was placed in a mold. Both the specimens made in air and underwater were cured in 20°C water until the age of the test. The compression test was carried out based on JGS T5 1 I . 3.2.3 Relation between the amount offly ash, undenvates ~ n - s e g ~ e ~ a tand i o ns ~ r e ~ ~ g t h The mini-slump value of 2.5 cm and the amount of cement of 80 kgim' were set. The relation between the amount of fly ash and both turbidity and strengtli when varying the amount of fly ash from 0 to 300 k g h 3 is shown in Fig.5. ~ement=80kg/m3,F I y ~ ~ h = 3 0 0 k g6steps /~~,
sand was I%, and it was classified into silty sand. It was impossible to secure trafficability of construction machinery. The basic physical properties of the dredged sand arid the fly ash are shown in TabIe2. Portland cement and seawater were used as the materials.
9,
2OO
1
2 3 4 5 6 Mrni-slump value just rnixed(crn)
Figure6 Relation between mini-slump and turbidity
643
7
Table3 Setting up of target strength
L
2
500.
E
U 3
a 1ver;s:gn Factor Associated with Mixing Efficiency of Plant E) Laboratory Target Strength kN/In2) F=C*D*E)
0.0 500.0 1000.0 1500.0 StrengthtCreated in the Air)kN/mZ
Figwe7 Relation of strength in air and underwater strength
Turbidity decreases with increase in the amount of fly ash. (Refer to 3.2.2(2) for turbidity.) Strength increases with increase in the amount of fly ash, and the longer the curing period is, the greater the measured strength is. These results show that the addition of fly ash to the dredged sand is effective in improving underwater un-segregation and uncoii~iied compression strength. The optimum amount of fly ash was determiiied to be 30Okg/iii3, considering the workability of the treated sand and economic efficiency.
~
~~~
~
~~~~
~
3.2.6 ~ z x t ~ design re
The unconfined compression strength of treated sand is confirmed in the f o l I o ~ ~ norder: g laboratory strength, in-situ strength, and core strength sampled from the revetment. The core sample strength must satisfy the design strength (Tablel). Therefore, an overdesign factor is needed on each steps. The overdesign factor for in-situ target strength was determined to be 2.0 based on variation of core sample 3.2.4 ~ n ~ ~ ~o e ~~ i c ; eof treated ~ ~ sand ~ and ~ d ~ strength ~ previously investigated. The l a b o r a t o ~tarwaiting time arntil placing, on underwater unget strength can be obtained by two overdesign facsegregation and strength tors to in-situ target strength. One is a factor associated with underwater placement, and the other is a The anioutit of cement and fly ash used were 80 factor associated with mixing efficiency of the plant. kg/ni3 and 300 kg/m3 respectively. The relationships The former factor was deterrnined to be 0.6 based on between the mini-slump value just after mixing and Fig.7 and the latter factor was set as 0.7 by the trial the turbidity are shown in Fig.6. constructio~i of the previous year. The design The turbidity for the treated sand placed within parameters are summarized in Table3. 0.5 hours after the mixing iiicreases with increase of The design mixture was determined as follows. mini-slump value. However, the turbidity of treated sand placed more than 1 hour after the mixing is at(1) The amount of fly ash in the treated sand was most constant with mini-slump value. 300 kg/m3 This iniplies that underwater un-scgregation was (2) The mini-slump value of treated sand was 2.5* influenced by the fluidity of just mixed material and 1.0 cin. by the waiting time until placing. It was found that (3) The amount of cement sliauld be adjusted so that the treated sand did not readily segregate in lower the target strength is obtained. fluidity and longer waiting time before placing. The mix proportion is shotvn in Table4. As mentioned above, the mini-siutnp value was detertniIied to be 2.5 I I .0 cm, the waiting time until placing was determined to be 0.5 hours after mixing. 4 CONSTRUCTION RESULTS 3.2.5 ~ t r e n g t specimen ~~: ~ ~ ~ ian air d evs. s ~ e ~ ~ ~ ~ e i i made uriderwater 4. I Final proportioniiig in the consfrtiction The strength of the treated sand placed underwater is Tile initial amount of cement was set at I 10 kg/m3 impo~ant.For this reason, the ratio of strength in air in consideration of the fly ash type. However, the to strength undcrwater was examined. The relation- amount of cement was finally decreased to 60 ship betweeri the strength in air and underwater kgim’step by step based on the core sample strength. strength in the same proportioning is shown in {Refer to Table5.) Fig.7. When ~~ini-slLi~np value was 2.5 cm, there was The strength of the treated sand placed underwater almost no slump-loss by construction, and a revetwas found to be 0.6 to I . I times the strength of the inent slope could not be secured because of the large treated sand in air. This strength is reduction caused fluidity. In case of small nii~ii-sIunipvalues, there is by the material segregation under water. 644
little underwater segregation and revetment slope construction is easy. Therefore, the mini-slump value was revised to be 0.5 cni 3-0.5 cm.
Table6 Release metal test results Dredged sand + LL’BA (300kg/in3 ) 4- cement (80 kg/rn3) ND
4.2 Strength Strength at some field points are shown in TableS. In TableS, (E) was computed by the following equation. All the average strengths (D) satisfied the n i i n i ~ u n irequired core sample strength (E). Therefore, construction was found to be successful.
(E), quf = Fc + t* CT
ND
(2)
niiuin Arsenic
where Fe = Design Srsengli ( = 2 1 O k N ~ t ~= )Deviation ~ ( = 1 282 as 10% Risk Ratio), CT = Standard Deviation
ND
I mg/I
ND
ND
0.11 0.05
Table4 Standard proportioning (kdni’)
1
TreatedSand
1
1 170- 1220
Amount of I Amount of Fly Ash Ceinent 300
I
Seawater
1998,
Sep
SP (Distance froit1the start point) (Refer to Fig 1) Fi! Ash T>pe(*)
10 2511%
LIlBA 80
1
Under 0.1 Under 0.1 Under 0.5 Under ~0.1 _ Under 15 Undes 02 Under 0. I
where ND = Non Detectable
I998. Oct. 15 202111
.
ND
I
Table5 Unconfined ~ ~ i n i ) ~ e s s test i o n results (6-day-old samples)
Construction Datc
ND
Standard value
UPBA
UPBA
80
60
Amount of FI? Ash (hgjii;’)
I10
Design Strength (hN/ni’) (A)(Refer toTablel) Laborator! Strength (hNlin’) (B) Strength at Plant (hN/in’)(C Ill-situ Nuiiibcr of Core Data Saznplc Atcragc Strength Streiigtli (hN/ni’) (D) Standard De\ iatioii Coc~cleill of Variation
2 10
2 10
f 3 70
400
1210
370
12
I2
15311
490
0.6.5
0.09
42
18
770
4.3 Long-term strength The unconfined compression test was carried out for field core sample after one year, in October 1999. The field core sample was taken at SP70m, 147m, 202m (Distance from the start point. Refer to TabIe5.) of the lStbank. The relationsflip between age and compressive strength is shown in Fig.8 for each proportioning. The strength at one year is higher for each p r o p o ~ ~ o n ~increased ng in one year, and it is thought that the stability of the revetrnent is improved. For each proportioning, the increase in longterm strength (28 - 365 days oid) is more gradual than the increase in short-terni strength (6 - 28 days old). Moreover, for long-term strength, the greater the amount of cement of the proportioning is, the larger the increase in strength obtained is. It is considered that strength improvement was achieved gradually by the ljme-pozzolan reaction of fly ash and cement.
940
(‘%I)
Miiuinuili
Required Core Sauiplc Strcriqth Dctcnriilied Statisticall?
iquf)(E)
where (*) = These age coal names LI=Lithgou, B.4=BlanAthol, MS=MLiswellbroo~,(Australia IS the producer for LI, BA and MS ) UP=Uinta Premtum, (U S A is the producer ) Blend Ratio. LI/BA=LI 60% / BA 40%- MS=MS loo%, UP/BA=UP 60% I BA 40%
Figure8 Relation between age and unconfiiied coinpression strength
645
_
4.4 Influence on the environment Since fly ash is highly alkaline, it was considered about influence on the environment by material separation. Therefore, pH was measured in the sea surrounding the revetment and the dumping site through the construction period. Although the pH rose to 8.5 from 8.3 in the dumping site, it fell quickly after the construction. Moreover, examination of metals released by the treated sand found all items to meet the standards. Those results are shown in Table 6. Therefore, it is thought that there was almost no influence on the environment.
5 CONCLUSION The construction method for revetment using fly ash may be summarized as follows. (1) Fly ash increases both underwater un-segregation and strength. (2) Underwater degree of un-segregation and strength increases when the fluidity of the treated sand decreases. (3) The strength increased in one year, and the longterm strength was satisfactory. (4) The construction does not influence on the environment. ( 5 ) Construction costs can be reduced by about 30% compared to general construction method using rubble stone. REFERENCES Hase,K., Kitahara,S., & Nakayama,T., Y1999. Long-term characteristic of treated sand and subbase materials using fly ash. Technical Research Presentation of Hokkaido Deve Iopment Bureau, In Japanese. Kobayashi,H., Tanaka,N., & Takahashi,M.., Y 1999. Develo ment of the underwater hardening object usin ash. Electric-power engineering works No284, In Japanese.
8,
Wakayama,Y., .Suzuki,K., & Okazaki,M., Y 1998. The application as the revetment material of the treated sand mixed fly ash. Technical Research Presentation of Hokkuido Development Bureau. In Japanese.
646
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Field test on pneumatic flow mixing method for sea reclamation M.Kitazume Port and Harbour Research Institute, Ministry of Transport, Yokosuka, Japan
N.Yoshino The 5th District Port Construction Bureau, Ministry of Transport, Yokosuka, Japan
H. Shinsha, R. Horii & Y. Fujio Japan Dredging Reclamation Engineering Association, Japan
ABSTRACT: Recently, Pneumatic Flow Mixing Method is developed for construction of sca reclamation and land development using soft dredged clay, in which the dredged clay is mixed with relatively small amount of stabilize agent in transfer pipe. The clay and agent mixture forms many separated plugs in the pipe, and is thoroughly mixed by turbulent flow in the plug. The mixture has relatively large strength so that no additional soil improvement is required. This technique is cxpected to providc an economical and rapid construction method for sea reclamation. The authors startcd a research project which includes laboratory and field tests. In the field test, execution technique and mechanical properties of the trcated soil were investigated at Nagoya Port Island. In this report, detail of the method and strength properties of the treated soil are dcscribed to show its high applicability for sea reclamation.
1 INTRODUCTION Many man-made islands have been frequently constructed in Japan to obtain enough plain area for airport, electric power plant, manufacture plant and so on. These islands require huge amount of soil for sea reclamation, which is often obtained in mountainous areas. However in recent year it becomes more difficult to obtain cnough amount of soil because of economical and/or environmental limitations. Meanwhile it also becomes difficult to find and construct disposal site for soft clay dredged from sea-route construction or sea berths. These circumstances promote to use the dredged soft clay as reclamation material. The island constructed with thc dredged clay slurry is so week and has high compressibility that no structures can be constructcd without any soil improvcmcnt. Vertical drain method is one of most frequently used soil improvement method to treat such a clay layer. However the method requires relatively long time to complete consolidation. Recently, a ncw sea reclamation technique has been developed in Japan, named as Pneumatic Flow Mixing Method, in which the drcdgcd soft clay is rnixcd with small amount of stabilize agent in thc pipe during transfer by cornpressed air pressure and is deposited for sea reclamation. Thc method requires only stabilize agent supplier facility to thc cxisting pneumatic facilities. And it is wcll known that thc soft clay treated with stabilize agent has rapid incrcase in strength. As no mixing blade is rcquircd and soil improvement can be pcrformcd
continuously in the method, this method is expected for construction of large island in relatively short time and more economically. Authors started a research project to study the applicability of the method for sea reclamation through the investigations of properties of treated soil, construction techniqucs, executing ability and so on. In the project, laboratory mixing tests, ccntrifuge model tests and field tests were performed (Kitazume, 1997, Kitazume, 1998, Kitazume, e t al, 1999 and Makibuchi, et al. 1999). This paper describes field construction tests in detail.
2 PNEUMATIC FLOW MIXING METHOD 2.1 Meclzanism ojthe method
Transferring of soft clay in the pipe requires large prcssure due to the friction generated on the inncr surface of the pipe. When relatively large amount of compressed air is injected into the pipe together, the clay is divided into small blocks by air blocks in the pipe, as schematically shown in Fig. 1. Thc scparated clay blocks are thus forwarded to the outlet by the compressed air. The formation of thc plug flow, which is composed of the clay and the air block, can function to reduce the friction on thc pipe surfacc and i n tcrn can considerably reduce thc air pressure required to transfer. Generally, the plug flow can best be generated at an air-to-clay ratio of 20 to 400 (Akagawa, 1980, Iwatsuki, et al., 1998). 647
type. In the former type, the stabilize agent is put into the clay before thc compressed air is injected into thc pipe. In the later type, on the other hand, the stabilize agent is put into after the air injection, as shown in the figure. The soil mixture is allowed to dump to reclamation site through the cyclone on the hammer-set vessel, which functions to reduce the air pressure transferring the clay plugs. There are scvcral variations available in Japan, in which some equipment are installed in the pipeline to improve the mixture of the clay and agent (Yamane, et al., 1998, Porbaha, et al., 1999, Horii, et al., 1999).
Figure 2. Image of mixing process in the pipe
3 FIELD TEST
The injected air pressure is dependent upon many factors such as the properties of clay, the volume of injected air, the pipe diameter, the pipe length and so on. In thc current practice, the air pressure of 400 to 500 kN/m’ is frequently used after consideration of the pressure capacity of the pipeline.
3.1 Purpose of the test
2.2 Mixing effects ofpneumatic transfer Figure 2 shows imagine of mixing process with soft clay and stabilize agent in the pipe during the transferring. As the clay plugs are transferred at very high speed of order of about 10 m/sec, the turbulent flow is generated within the plug due to the friction on the inner surface of the pipe. The turbulent flow provides to mix the clay and the stabilize agent. Previous research efforts found that thorough mixing can be obtained in the condition of Reynolds number, Re = uD/v of 500 to 3,000, whcre U is the plug speed, D is the pipe diameter and v is the viscosity of the plug. It is also known that at least 50 m to 100 m of transferring length is necessary to ensure satisfactory mixing. 2.3 Current facility Figurc 3 schematically shows one kind of the Pncumatic Flow Mixing Method available in Japan. In the figure, the dredged clay in the barge is loaded into the hopper on the pneumatic vcssel at first, and is transferred by the compresscd air to the reclamation site. Stabilizc agcnt is thcn injected to the clay on the stabilize agent supplier vessel and thcy are thoroughly mixed during the transferring. There are two types of the method according to where thc stabilize agcnt is put into; compressor addition type and line addition
Figure 3. Schematic flow of pneumatic transfer.
648
Air pressure required at the inlet is depend upon many factors already mentioned before. Series of field executing tests were performed at Nagoya Port Island in Aichi Prefecture as shown in Fig. 4 to investigate the effect of the properties of the soil mixture on the transferring. In the tests, the soil mixture having several combinations of water content and amount of cement and clay, werc transferred. The air pressure changes along the pipe were measured in the tests. And also carried out the unconfined compression tests in order to investigate the average strength and its deviation of the treated soil manufactured in-situ. 3.2 Facility Figure 5 shows a group of vessels for the field test which includes the pneumatic vessel, the stabilize agent supplier vessel and the hammer-set vessel. The facility used in the test is one type of the method named as Pipe Mixing Method. In the method a special designed plug detection system is installed in the pipeline as schematically shown in Fig. 6 to monitor the plug movement and to improve the mixture of the clay and thc stabilize agent (Makibuchi, et al., 1999). Major capacities of the facility used in the test are summarized in Tablc. 1 . 3.3 Materials Soft clays used in the test were marine clay dredged in Nagoya Port. Although the marine clay was drcdgcd at almost samc site at Nagoya Port, its propertics were found to be different in each test case. Average physical properties of the clay are summarized in Table 2.
Tablel. Maior capacity of facility. Facilitv
Cauacitv
Pneumatic vessel max of main power max of transporting capacity niax of stability supplier StabillLe agent supplier vessel niax capacity Hammer-set Vescel diameter of cyclone Pipe line diameter length
2.000-ps 150 ni’/h. two lines 30 t o n h 300-m3/hr
$ 1500. two sets $ 350 nini L = 180to373m
Table 2 . Major physical properties of clay. Property
valuc
soil grain density soil particle analysis sand content silt content clay content consistency limits liquid limit,W, plastic liniit,W,, plasticity index,$ ignition loss,Li
2.677 g/cni3 7.7 YO 39.0 % 53.4 % 80.8 % 34.1 % 46.7 6.9 %
Table 3 Test results on plug characteristics. Test condition transferred soil water content Liquid limit cement added Tcst results plug speed plug volume plug length plug interval
Case 1
Case 2
Case 3
2 10 mVhr 132.5 Y” 70.5 3 8 kg/nii
296 mi/hr 117.2% 77.6 78 kgini’
170 m’ihr 96.2 % 81.5 52 k g h ’
I I .9 n h c c
12.8 ndscc (1.9 - 25.0) 0.30 nii (0.18-0.36) 3.1 ni (1.9 - 3.8) 6.4 sec (0.6 - 29.0)
10.9 d s e c (1.6- 25.0) 0.41 ni’ (0.23-0.52) 4.3 ni (0.23- 5.4) 7.1 sec (1.3- 30.3)
( 1 .5 - 25.0)
0.36 ni’ (0.25-0.45) 3.1 m (2.6 - 4 7) 4.4 SCC (0.5 - 18.2)
4 TEST RESULTS 4.1 Characteristics of the clay plug
Thc dredged clays were diluted with seawater in the barge to obtain the prescribed water contcnt of 1.3 * W1, where WI is thc liquid limit of the clay. Stabilizc agent used was typc-B slag cemcnt, which is one of thc most common stabilize agcnt in Japan. Seawatcr was added to the cement in advance to make cement slurry with the water and cement ratio of 100 %, and then the cement slurry was injected directly to thc clay plug by the help of the plug dctection system mentioned before.
Three tests were performed to monitor thc plug charactcristics by thc plug dctection system. In the test, thc clay mixturc having several combinations of thc water contcnt and the amount of ccment were transferrcd by the air pressurc of about 400 kN/m’, as summarized in Table 3. Although the mcasured data have large scatter throughout the tests, it can be found that thc clay plugs with an average volume of 0.36 m3 arc transferred at average speed of about 12 m/s and an avcragc interval of about 6 seconds.
4.2Air pressure distribution in the pipe Another series of tests were performed to investigate the air pressure changes during the transferring, in 649
which the air pressure was measured at five locations along the pipe. In the tests, the water content, the transferred clay volume and the amount of cement were changed 98 % to 128 %, 182 m3kr and 300 m3hr and 0 to 80 kg/m3,respectively. The measured air pressure changes are plotted in Fig. 7 along the transfer distance from the inlet. In the figure, it is found that a relatively large decrease occurred in the pressure between at the pneumatic vessel (PO)and at the stabilize agent supplier vessel (PI) irrespective of the test condition. These large decreases are probably because the clay in the pipe is stiII unstable to form the plug flow. After at the point PI, the air pressures decrease almost linearly to zero at the outlet (P4) with increase of the transfer distance. Thesc decreases in the air pressure are thought to be attributable to the wall friction on the pipe. The effect of clay volume transferred on the air pressure decrease is also shown in the figure. It is found that the air pressure at the inlet (PO)becomes large with increase of the clay transferred. As far as the test condition, the required air pressure should be increased about 100 kNim’ when the clay volume transferred increases from 200 m3kr to 300 m3/hr.It is also found that the addition of cement causes increasing the pressure about 100 kN/m’ at the inlet (PO),because the cohesion of the mixture becomes large.
Figure 7. Air pressure decrease in the pipe. Figure 15.U ~ c o ~ ~ compressive ned strength against the amount of cement.
650
4.3 Strength of treated soil To invcstigate the strength profilc of the treated soil, anothcr field tests wcre performcd in which the amount of cement added was changed to 38 kg/m3, 57 kg/m', 68 kg/m7 and 78 kg/m3. For these tests, the clay slurry mixed with thc cement was reclaimed into the small ponds cxcavatcd in advance, as shown in Fig. 8. The unconfincd compression tests were conducted on the treated soil manufactured in-situ. Total of 120 spccimcns with 5 cm and 10 cm in diameter and height respcctivcly were sampled in the field and were tested to confirm the average and deviation of the strength of the treated soil manufactured in-situ. Figurc 9 shows the strength deviation of the 28 days cured treated soil, which was manufactured with the initial watcr content of 125 % and the cement of 57 kg/mi. In the figurc, the strength deviation shows almost similar shape of the standard deviation profile. The coefficient of deviation of the strength is 32.3 %, which is almost samc order to that of Deep Mixing Method irrespectivc of the mixing procedure (Hosomi et al., 1996). Figure 10 shows test summary of the unconfincd compressive strength, in which the average strengths are plotted against the amount of stabilized agent. It can be seen from the figure that the average strengths of the treated soil increase almost linearly with increasc of the amount of cement. Another unconfined compression tests wcre also performed on the field-manufacturcd specimens whose diameter and height were 50 cm and 100 crn respectively. Their average strengths are also plotted in the figure. It is found that the compressive strength on the large specimen also increases with increase of the amount of cement, but the avcragc strength on the large sized specimen is smaller than that of small sized specimen. The strength ratio of the large specimen against the small sized specimen is about 0.7, which is almost samc as the previous rescareh on the treated soil manufacturcd by the Deep Mixing Method (Futaki, ct al., 1996). In the figure, test results on the trcatcd soil manufactured in a laboratory are also plotted. It is also found that the laboratory samples show the largest strength among three test spccimcns. It is well known that the strcngth dcviation is much dependent upon the cemcnt ratio, the manufacturc tcchniques and the size of specimen. Figurc 11 summarizcs the relationship between thc coefficicnt of dcviation of unconfined compressivc strcngth against the amount of cement. It is found that thc coefficient of deviation on the laboratory samples is relatively small of 15 %, and almost constant irrcspectivc of the amount of ccmcnt. This means that quite uniform mixing can be obtaincd in thc mixer in a laboratory. The field-manufactured spccimcns, on the other hand, indicate relatively largcr deviation of strength comparing the laboratory spccimen. It is found that the field specimcn with small sizc is about 35 % irrespcc-
tive of the amount of stabilize agent. But the specimen with large size shows almost samc coefficient as the small sized specimen as far as the amount of cement exceeds about 50 kg/m' but increases in the coefficient to about 60 % when the amount of agent decreases to 38 kg/m3. 5 PRACTICAL IMPLICATIONS Thc utilization of treated dredged clay as replacement or fill material could be one of the most desirable applications. Figure 12 indicates various types of expected applications of the Pneumatic Flow Mixing Method. The advantage of this newly developed mixing method is rapid stabilization togethcr with rclatively economical. The settlement of the reclamation fill and induced horizontal earth pressure imposed to the retaining system will considerably be decreased.
Figure 11. Strength deviation of field manufactured treated soil against the amount of cement. [backfill behind caisson]
[backfill behind sheet wall]
PTP [surface improvement] bank, treated soil
U
[underwater backfill] v
soft foundation
Figure 12. Expected application of the Pneumatic Flow Mixing Method.
65 1
Kitazume, M. 1997. Centrifuge model tcsts on stability of embankmcnt improved by ccmcnt, Pr-oc. of 32th Annual Conference of Japanese Geotechnical Society: 2429-2430. (in Japanese). Kitazume, M. 1998. Centrifuge modcl tests on stability of improved embankmcnt, Proc. of 33th Annual Corzf e ren ce of Japanese Geotech n ica I Society : 2257-2258. (in Japanese). Kitazume, M., Matsubara, Y., Matsuura, T., Hayashi, K., Shinohara, K., Oomori, K., Kaneshiro, T., Hoshi, H. & Kojima, T. 1999. Field test on soil admixture stabilization with suppressed pH agent, Proc. of 34th Annual Conference of Japanese GeoteclzizicalSociety:805-806. (in Japanese). Makibuchi. M., Yoshino, N., Kitazume, M. & Okano, K. 1999. Strength characteristics of the clay ground improved with a small amount of cement, Proc. of 34th Annual Conference of Japanese Geotechnical Society: 807-808. (in Japanese). Porbaha, A., Yaname, N., Asada, H., Kishida, T. & Sakamoto, A. 1999. Air-transported stabilized dredged fill. Proc. of 34th Annual Conference of Japanese Geotechnical Society: 809-8 10. Yamane, N., Taguchi, H., Fukaya, T., Dam, K.L., Kishida, T. & Iwatsuki, T. 1998. Strcngth characteristics of cement-treated soil using comprcssed air-mixture pipeline. Proc. of 34th Annual Conference of Japanese Geotechnical Society: 2253-2254. (in Japanese).
6 CONCLUSIONS The research reported herein forms a part of an ongoing effort to investigate the applicability of the Pneumatic Flow Mixing Method for sea reclamation and back filling watcrfront retaining structures. The specific conclusions derived from this study are as following: (1) As far as the ficld test condition studicd, thc clay plugs with an average volume of 0.36 m3arc transferred in the pipe at average specd of about 12 m/ s and an average interval of about 6 seconds. (2) The air pressure decreases almost linearly with increases of the transfer distance. Thc air pressure required increases with increase of the amount of clay and addition of cement. (3) The averagc of unconfined comprcssive strcngth increases with increase of the amount of stabilize agent. The treated soil manufactured in laboratory has highest strcngth rather than the field manufactured treated soil. (4) The large sized specimen shows smaller strength than thc small sized specimen, its ratio obtained in the field tests is about 0.7. ( 5 ) The coefficient of strength deviation of thc fieldmanufacturcd treatcd soil is less than 35% in the case of ccmcnt volume of 50 kg/m3 but increases about 60 ‘70 whcn the ccmcnt volumc dccrcases to about 40 kg/m3. As a final remark, it is concluded that thc Pncumatic Flow Mixing Method has rclativcly high applicability for construction of man-made island, sea rcclamation and back filling. REFERENCES Akagawa, K. 1980. Gas-liquid two-phase flow. Mechanical Engineering 11: 15. (in Japanese). Futaki, M., Nakano, K. & Hagino, Y. 1996. Design strcngth of soil-cement columns as foundation ground for structures, Proc. of the 2nd International Corzfereizce 011 Ground Improvement Geosystems: 48 1-484. Horii, R., Shinsha, H. & Fujio, Y. 1999. Plant for the pneumatic flow mixing mcthod. Journal of Kensetsu no Kikaika: 30-35. (in Japancsc). Hosomi, H, Nishioka, S 8L Takci, S. 1996. Method of deep mixing at Tianjin Port, People’s Republic of China, Proc. of the 2nd Inter-national Conference on Ground Improvement Geosystems: 49 1 -494. Iwatsuki, T., Kamiyama, Y., Hashimoto, F., Yanai, E. & Masuyama, T. 1998. Effectivc ccmcnt-mixing mcthod for mud transport using a comprcsscd-air mixture pipelinc, Annual Journal of Hydraulic Engineering, Japan Society of Civil Engineereia, Vol. 42: 655-660. (in Japanese). 652
Coastal Geotechnical Engineeringin Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Mixing design of liquefied stabilization soil with sand M. Koda, S.Tanamura, 0.Murata & S.Takizawa Railway TechnicalResearch Institute, Tokyo,Japan
M. Ichihara Nittoh-Daito Industrial Company Limited, Tokyo,Japan
G.L.Jiang Integral Geotechnology Institute Company Limited, Tokyo,Japan
ABSTRACT: In this study, the mass of mixed sand W, for a constant volume of slurry at a constant density pf was changed to select the fine fraction ratio F, of Liquefied Stabilization Soil (LSS) as a test parameter. Firstly, a number of unconfined compression tests of LSS samples were carried out to investigate the relationships between the density of LSS pus, mass of water W in lm3 of LSS, mass of cement C in l m 3 of LSS, mass ratio of water to cement W/C, flow value F, unconfined strength qUz8and fine fraction ratio F,. Secondly, a series of consolidated - undrained triaxial compression tests of LSS samples were conducted to investigate the effect of F, on q-E, relations. 2
1 INTRODUCTION
As the volume ratio of excavated soil to industrial wastes is high, recycling has been expected. However, this material needs soil stabilization because it generally has a high water content and fine soil particles. Recently, Liquefied Stabilization Soil (US) has been used extensively as a back filling soil for cut and covered tunnels or invert part of shield tunnels from the viewpoint of recycling the excavated soil from underground construction sites in urban areas. LSS is made from slurry of excavated soil, sand, cement and water, which can be installed without cornpaction in the field. The excavated soil occupies one - third the volume of US. In this study, the mass of mixed sand W, for a constant volume of slurry at a constant density pf was changed to select the fine fraction ratio F, of LSS as a test parameter. Then the mass of mixed cement was kept at 80kg with lm3 of slurry. Firstly, a number of unconfined compression tests of LSS samples were carried out to investigate the relationships between the density of LSS pLss,mass of water W in Im3 of U S , mass of cement C in Im3 of LSS, mass ratio of water to cement W/C, flow value F, unconfined strength qUz8and fine fraction ratio F,. Based on these results, the relationships between F, and the properties of LSS were shown. Secondly, a series of consolidated - undrained triaxial compression tests of LSS samples were conducted to investigate the effect of F, on q-E, relations.
PHYSICAL PROPERTIES OF EXCAVATED SOIL, SAND AND CEMENT
This chapter explains the materials of LSS; excavated soil, sand and cement. The excavated soil (ps=2.66g/cm3, I,=64) was collected in the Yokohama MM21 area and the sand (pS=2.71g/cm3) in Kimitsu city, Chiba Prefecture. In this study, grain size distributions of the excavated soil and sand were adjusted to adopt F, as a test parameter. Namely, from the excavated soil and sand, particles over 75pm and under 75pm, respectively, were removed. Therefore, F, can be calculated from the mass of each material, each water content w and each particle density p,. Figure 1 shows their grain size distribution curves. The mixed cement is the general cement stabilizer (TAIHEIY 0 CEMENT Co. Ltd., GS10) at a cement particle density of 3.04 g/cm3.
3 MIXING PROCEDURES FOR LIQUIFIED STABILIZATION SOIL The LSS samples are classified into TYPE A and TYPE B, whose density is 1.14 g/cm3 and 1.10 g/cm3, respectively. After 80kg of cement was added to l m 3 of each slurry (mass of cement added to lm3 of slurry C’=80kg), the slurry was mixed by a fast hand mixer for about two minutes and then predetermined mass of sand W, was added before the slurry was mixed again for about two minutes. TYPE A is classified into three kinds of F, (TA-1, TA-2 and TA3) and TYPE B also into two kinds of F, (TB-1 and 653
Table 1. Mixture proportion of LSS
TB-2). These mixture proportions are shown in Table 1. It is expected that the unconfined compression strength qu2g of TYPE A is greater than that of TYPE B because the density of TYPE A slurry is larger than that of TYPE B. This means that the volume of water in TYPE A is smaller than that in TYPE B. After LSS slurry was prepared, the flow value F of the slurry including sand and cement was measured. At first, the slurry was inserted into a mold with diameter $=80mm and depth D=80mm. The mold was then set on a glass plate with length L=SOOmm and width W=SOOmm and pulled up from the plate quickly. After that, the diameter of LSS on the plate, F (in mm), was measured. Then, the density of LSS pus and water content of LSS w were measured. Before LSS was inserted into the mold with diameter $=50mm and height H=100mm, LSS was kept in a container at -750mmHg for 10 minutes to remove the entrained air. After LSS in the mold were kept for 28 days at 25°C and 70% moisture, unconfined compression tests and consolidated - undrained triaxial compression tests (ICUC test) were conducted.
pr: density of slurry, V,: mixed slurry volume, w: water content, Ws: mixed sand mass, C’: mixed cement mass, LSS=V,+W,+C’
5
CONSOLIDATED - UNDRAINED TRIAXAL COMPRESSION TESTS
A series of consolidated - undrained triaxal compression tests were conducted to investigate the effect of F, on the relationship between q-E, relations.
4 RELATIONSHIPS BETWEEN qu28 AND F, The relationships between qUZ8 and the conditions of mixture proportion are shown in Figure 2 and Figure 3. Horizontal coordinates are F, and vertical coordinates are pus, mass of water W in lm3 of LSS, mass of cement C in Im3 of LSS, qu2g, w/C, flow value F. In these Figures, pLss increases and F decreases as F, decreases. The decreasing rate of F with F, becomes larger when F, is below 15% in Figure 2. Because of the interlocking effect of sand particles in LSS, F decreases. It is considered that qu2g is closely connected with W/c. The relationship between W/C and qu2g is shown in Figure 4. From the Figure, qu28 is in inverse proportion to W/C. These results could give the following new mixture proportion of LSS. At first slurry at a certain density is made by mixing water and excavated soil. And then the density of the mixed slurry is measured. Secondly, the mass of cement is calculated from the relationship between qu28 and W/C. Here, the mass of water W in lm3 of LSS is calculated from the density and mass of the slurry. Finally the mass of sand is calculated from the necessary density of LSS. Therefore, it is not necessary to control the density of slurry if this LSS mixing procedure is used. Figure 5 shows the relationship between F, and the coefficient of variation in qu2g cov(qu2g). When F, decreases, cov(qu28) decreases too. It seems that the scatter of qu28 becomes smaller, when sand is inserted into LSS. Figure 6 shows the relationship between the void ratio e and F,. From Figure 5 and Figure 6, it is known that cov(qu,g) decreases because the void ratio of LSS decreases and the samples become dense with decreasing F,. 654
Figure 2. Relationships between mixture proportion and qu28(TYPEA)
Figure 7. Relationships between q and E,(TYPE A)
Figure 5. Relationship between C0V(qu28)and F,
The U S specimens were set in a triaxal cell and isotropically consolidated at o,’=49kPa or 98kPa. Then undrained compression tests were conducted at a loading rate of O.OS%/min. Before the consolidation, B values were measured and B=0.8 was obtained.
655
The relationship between qr/qmaxand F, is shown in Figure 9. It is known that the q-E, relation shows that it is the hardening type with increasing E, if qdq,,, is 1, and the relation shows it is the softening type with increasing E, if qdq,,, is below 1. From the Figures, hardening of q with increasing E, can be found for F, less than 10%. However, as described in the discussion about Figure 2 and Figure 3, the liquidity, an advantage of LSS could be failed if F, becomes less than 15%. So it is important to satisfy the conditions F>150 and F,<10% to obtain LSS with high performance and execution.
6 CONCLUSIONS l.qu2Band F of LSS decrease and pLssincreases with decreasing F,. The rates of decreasing qu28and F with F, increase when F, is less than 15%. However, C0V(qu,,) becomes smaller, when sand inserted to LSS. 2.From the above conclusion, the new mixture proportion method without controlling the density of LSS is proposed. 3.It is found that the q-E, relation obtained from ICUC tests of LSS samples becomes the hardening type relation from the softening type relation with increasing E, for F, less than 10%. From decreasing F,; increasing sand in LSS, it is considered that the interlocking effect of sand particles is obtained in q-E, relations. REFERENCE Tatsuoka, F. 1976. Some problems and evaluation for soil testing -some soil mechanical problems and soil testing in lab. on planing of Tokyo-Aqua-Line-. Proc. of symp. on some problems of resent soil mechanics and foundations: 55-103. (in Japanese)
Figure 9. Relationship between qrlqmaxand F,
Figure 7 (a) and (b) and Figure 8 (a) and (b) show the q-E, relations obtained from the tests. The deviator stress q increases as the axial strain E, increases and the excess pore water pressure Au in the sample also increases. However, Au keeps the value equal to the confined pressure qO’with increasing E, after having reached the maximum Au. Consequently, the effective confined stress or7 (=arO7-Au) becomes zero. This type of behaviour of Au in LSS is almost the same as that of cement treated soil as shown by Tatsuoka (1976). The difference between the maximum deviator stress qmax and residual deviator stress qr decreases with increasing F, as shown in Figure 7 and Figure 8. 656
Coastal Geotechnical Engineering in Practice, Nakase L? Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Construction of immersed undersea tube tunnel (TokyoWest Fairway Tunnel) on the very soft ground Tadaaki Masuda, Eizo Sasaki & Toshio Muranaka Port a i d Harbor Construction Department, Bureau of Port and Harbor, Tokyo Metropolitan Government, Japan
ABSTRACT: The ground under the port of Tokyo, as seen from the aspect of the tunnel construction, consists of a thick soft alluvium deposit flown from rivers. Ground settlement is still in progress with all the reclaimed lands. On the reclaimed land of Jonanjima, in which the tunnel starting shaft has been constructed, ground settlement of 5-10 c d y e a r stilI continues. This report describes the most modem design and construction of an undersea-immersed tube tunnel applicable to soft grounds. 1 INTRODUCTION
girder could not be achieved due to the limitation of the height as required by the Tokyo International Airport, which is closely located. For constructing the tunnel, two plans -the immersed tube tunnel and the shield tunnel - were studied. The latter offered advantages: lower construction cost per unit length of tunnel and fewer disturbances to the vessel traffic than the former method. However, by the shield tunnel construction method, the tunnel would be about 2,100 m long, because the earth coverage on the tunnel below the vessel route has to be as thick as 20 m and slope of the approach roads to the tunnel should be 3 - 4 %. On the other hand, by the immersed tube method, the tunnel becomes about 1100 m long, because the immersed tube tunnel needs 2 m thick counterweight fill for preventing the tubes from floating. Considering the economy of the construction, the immersed tube tunnel method was adopted. Figure 1 shows the cross sectional image of the tunnel.
As the southeastern part of the City of Tokyo faces the Tokyo Bay, many reclaimed lands have been constructed in the bay for reinforcing the harbor function of the Tokyo port. Complete with the extensive modern harbor facilities, the port is now one of the leading trade ports in Japan. The Tokyo Port West Tunnel is a 1.4 km long undersea tunnel of immersed tube construction laid underneath the Tokyo West Fairway, the main waterway for vessels incoming and outgoing from the port of Tokyo. This tunnel was intended for connecting the reclaimed lands separated by the waterways for reinforcing the distribution function of the port and completing the ring highway of 20 km diameter prepared in the periphery of Tokyo. For crossing the Tokyo West Fairway, two plans building of a bridge and a tunnel - were studied. The required clearance of AP + 52.4 m under the bridge
Figure 1. Cross section of Tokyo Fair Way Tunnel - Immersed tunnel method.
657
Figure 2. Profile of the ground at the sites.
2 SITUATION WITH THE IMMERSED TUBE TUNNEL IN JAPAN 2 .1 Immersed tube tunnels near the port of Tokyo The immersed tube tunnel method has a 100-years history. About 130 tunnels have been built by this method all over the world. Inclusive of ones presently under construction, we have 14 tunnels (eight for roads, five for railways and one for belt conveyer transportation) around Tokyo. The reasons for such overwhelming number of applications are as follows. 1) Compared with the shield tunneling method, the immersed tube tunnel method requires higher construction cost per unit length. However, the shorter length of the approach roads offsets this disadvantage in the economy. In large cities where land is still very costly, the possibility of utilizing hinterland effectively still offers an excellent advantage.
2) The area around the Tokyo port has an advantage of being able to secure tunnel construction yards (210,000 m’) by sealing off canals (construction of canal cutoff walls and dry docks) for producing caissons to be immersed. This offers a favorable condition in view of the ease of the construction and cost reduction.
2.2 Countermeasures to scope with large scale earthquakes A committee consisting of specialists was established in 1997 to study the earthquake-proof construction in order to cope with big earthquakes that are as severe as the Hanshin Awaji class epicenter type earthquake. The study of designs was done by analyzing techniques using a new non linear response, that takes into account the plastic behavior of the ground and structure, instead of by conventional analysis on the premise of the elastic response of the structure. In the study two types of
earthquake waves are used; 1) the earthquake wave adjusted to the maximum acceleration 375 gal for the earthquake-proof quay wall design and; 2) the earthquake wave based on the recorded earthquake vibration of the Kobe port island in the southern part of the Hyogo prefecture.
increase of the depth appears not to have caused the increase of p,, which remained almost constant. The reason of this distribution o f p c seems to be the absence of a drainage layer between the Yc and Nac layers. It seems to have caused the extreme delay of the consolidation in the lower part of the Yc layer. Due to the same reason, the consolidation was delayed in the upper area of the Nac layer. However, the effect of consolidation is increased with the depth in the Nac layer. The rate of the consolidation tends to increase toward the normal level with the increase of the depth in the layer. The consolidation is still in process in the construction area. The overall settlement is assumed to be 1- 2 m. Therefore, when a foundation type was selected o r ground improvement was to be conducted, this consolidation must be taken into consideration. 2) Central breakwater side: A relatively solid To layer (diluvial deposit) is dominant from the middle of the passage area. T h e distribution o f the consolidation yield stress p, is in the state not fully consolidated in the reclaimed U layer, while the state in the Yc layer is normally consolidated. The Tosl layer located at the bottom had been over consolidation by about 4.0 - 6.0 kg/cm2. The presumed subsidence about 0.5 m. It was assumed that the settlement accompanying the construction of the shafts and land tunnels would be small.
3 GROUND OUTLINE 3.1 Stratum organization The ground of the area where the immersed tube tunnels are placed consists of several layers as shown in Fiwre 2. The features of these layers are as follows. reclaimed soil (U): It mainly consists of clayey soil, which has 80 YOof fine grained soil. As the natural water content is a little above the liquid limit, it is unstable and easy to liquefy. the Yurakucho clay layer (Yc): The width of the natural water content fluctuation was very wide. As the natural water content is a little below the liquid limit, soil in this layer is also unstable and easy to liquefy. This condition is similar to the layer 1. 3) the Yurakucho gravel layer (Yg): It is a gravel layer consisting 50 % of sand. 4) No.7 clay layer (Nac): The grain constituent was almost the same as that of Yc. Compared with that o f Yc layer, the natural water content fluctuates much less and its value is also low. As the natural water content is very much below the liquid limit, it is not so unstable compare with the Yc layer. 5) No.7 sand layer (Nas): The natural water content is relatively high, 44 - 53 %. Due to the presence of humus, the specific gravity is low compared with those of other layers. 6) the Tokyo clay layer (Tocl-3): The layer mainly consists of silt, and is relatively stable. 7 ) the Tokyo sandy layer (Tos 1-2): It contains 26 YOof fine grained soil, but the natural water content is as low as about 3 1 %. 8) the Edogawa sandy layer (EdS1): The grain constituent is similar to that of Tos 1-2, but the natural water content is below than that of Tosl-2.
4 DESIGN AND CONSTRUCTION OF IMMERSED TUNNEL
The immersed tube tunnel is a construction consisting of immersed tunnel section (located at the sea bottom or below the underground water level consisting of box formed immersed caissons) and vertical ventilation tower (vertical construction for connecting caissons and land tunnel sector, for conducting the ventilation of both tunnels) and the approaching roads (tunnel sector and open cut). 4.1 Ventilation tower
4.1.1 Foundation structure style design
3.2 Assumption of the consolidation settlement
The ventilation tower is located at the joining points of the caissons and land tunnels. They must never sink, move nor be deformed by earthquakes for a long time in the future. They have to be installed accurately, because they must be finally connected with the immersed and land tunnels accurately. Accordingly, the following three structure styles are studied and compared. 1) Caisson spread foundation - pneumatic caisson and others. 2) Piled foundation - caisson & foundation piles; foundation piles are categorized in ready made piles (steel pipe piles) and cast-in-place piling (reverse piles, etc.). 3) Special foundation - caisson & steel pipe sheet pile.
The grounds on the Jonanjima side passage area and outside of the central breakwater have various features as follows. 1) Jonunjirna side: An extremely soft Yc layer accumulation of thickness around 35 m and N value roughly below 2 overlays on a Nac layer of thickness roughly 15 m. Nac layer is a diluvial clay layer and its mechanical properties are similar to those of the alluvial clay. The distribution of consolidation yield stress p , is smaller than the effective overburden pressure line in the Yc and Nac layers, indicating that both layers were in the state not having been fully consolidated yet. Especially in the Yc layer,
659
Type 1 foundations with deep supporting layer are difficult to construct. Type 2 and type 3 foundations are considered to be suitable compound foundation structure for the operation, because the pile foundation shall share the load in the large depth, where the pneumatic caisson structure will lave weakness. The type 2 with cast-in-pile necessitated maintaining provisional piled jetties in the sea for a long time and silt-protecting films became necessary for preventing water pollution. Accordingly, the type 2 with cast-inpiling is decided not be suitable for the construction at this site. On the other hand, ready made pile system allowed the use of steel pipe piles whose upper part above the caisson installation level could be cut and removed and could be reused as upper piles for the main installation. As the type 3 involves an economic problem of having to dispose of cut and removed steel pipe piles, the type 2 with ready made pile system was eventually adopted for the construction. 4.1.2 Countermeasures against soft ground As the pile top displacement exceeded the allowable value of 15 mm (the minimum value of 1 % of the pile diameter and 15 mm) due to insufficient lateral resistance of the foundation pile driven into the existing ground in the ventilation tower on the Jonanjima side, the ground is reinforced by foundation improvement. As the present ground is of soft clay layer, the sand compaction pile method (80 % improvement, the diameter of sand pile is 1500 mm) is employed for the foundation improvement . The improvement depth is AP -32.00 m that allowed to secure 5.5 m or over I / (the pile length) from the bottom of the caisson (AP 26.0 m). The scope of the improvement is determined over 10 m from the front face of the shafts, since the vessel passive earth slope failure gradient is 25.3 degree. The ground shallower than AP -26 m is improved, as the lateral resistance for the temporary piled jetties have to be increased. The design target value for the improvement is set N=15 at the center of the sand pile, so that the coefficient of the horizontal subgrade reaction (kh=30 kg/cm3 for sandy soil, kh=10 kg/cm3 for clay) can be secured. The average N value of the original ground is 2; however, the average N value at the start of the construction in 1994 was 23. The increase of the strength by aging was conspicuous, as the average N value at the same spot check bore increased to 38 in 1997. The N values were over 50 at many spots checked. The reason for the increase of the strength is the promotion of the drainage of the ground inclusive of those of the surrounding area, by the forced replacement of the unconsolidated Yc layer with sand by conducting sand compaction that allowed to drive in sand holes into the 5 m thick Ys layer that has been sandwiched between the Yc and Nac layers to consolidate the layers (Figure 3).
Figure 3. Ventilation tower at Jonanjirna site.
4.1.3 Method of construction The ventilation tower consists of the base of box formed concrete construction and the building above. The form of the completgd base is B45xL45xH32 m. A steel shell caisson , B45xL45xH24.5 is adopted for the base for insuring a trouble free operation conducted on the bottom of the sea. In the Jonanjima side, the construction is started with the ground improvement by sand compaction pile method. After that, 121 steel pipe piles of 1,500 mm diameter are driven down as deep as the bearing stratum. Then the insides of the piles are cleared down to the depth of -24.5 m and piles are cut off at the depth, intermediate depth between the cutting edge and the bottom of the caisson. The steel shell caisson was towed from the factory to the site. After settling the caisson to the sea bottom, the repetition of the caisson arrangement, digging and sinking works, continued to reach the base of the caisson the depth of -26 m. The operator in a surface control room remote-controlled the excavator using radio. The caisson was stabilized after about 5,000 m3 concrete having been poured into the digging workroom to unite the caisson with the foundation piles. 4.2 Land tunnel 4.2.1 The standard shape of a span of the land tunnel It was a box-formed construction of the size of B24.60xH8.1 OxL25.0 m. The ground under the installation site is (1) soft and assumed ultimate settlement of 0.50 - 2 m, and (2) requiring design technologies for excavating the ground down to a great depth. Countermeasure must be designed from two aspects, i.e. subsidence and failure as follows; (1) use of supporting piles for preventing the subsidence and 660
(2) foundation improvement to reinforce the bottom of the excavation.
shear strength of the excavated bottom as the counter measure to the heaving in the Jonaniima side, while the ground improvement is required for increasing the water tightness of the excavated bottom in the central breakwater side. Considering the scale, term and the easiness of the operation, the following measures are studied and compared for the ground improvement. a: Deep mixing method -this method is categorized in Cement deep mixing method (CDM), Lime deep mixing method (DLM), Dry jet mixing method (DJM), and Jet grout method (Column jet grout method (CJG), Jumbo jet special grout method (JSG)) b: Quick lime pile method c: Chemical grouting method d: Freezing method The improved ground strength by method (b) is smaller than by other methods and is not large enough to keep the stability during all the construction process in this site. However, as the present state of the soil of the reclaimed land at the site is too weak to support the heavy machinery, it was decided to use the method (b) for getting the initial stability and keeping the trafficability. The method (c) had a problem on the reliability, because the method had not been used often for large scale ground improvement. The method (d) is not suitable for the work of this site in view of the term and cost. The rising of the bottom may cause the maximum damage among three phenomena including the rising of bottom, heaving and boiling that may cause the failure of the excavated bottom. The thickness of the improved ground required for preventing from this problem should be designed to keep the ratio of total of the ground weight and adhesion of the earth retaining wall and adhesion of the foundation piles to the artesian head difference larger than 1.50. This means that the ground improvement and the foundation piles are required for dealing with the consolidation settlement and the rising of the bottom in the Jonanjima side. The thickness of the improved ground shall be 5.0 - 9.50 m. As the artesian head difference in central breakwater side is smaller than that in the Jonanjima side, the thickness of the improved ground should be 3.50 7.40 m in central breakwater side. I
4.2.2 Countermeasures against soft ground (1) Measures against settlement (use of steel pipe piles): As the consolidation settlement was relatively large in the Jonanjima side, the countenneasures using the foundation improvement and foundation piles are compared. As the required improvement depth for the foundation improvement is as large as about 60 m, the use of foundation piles is judged advantageous in view of the ease of the operation and the construction cost. The amount of the consolidation settlement is assumed to be 1 - 2 m. It is decided that the pile support is to be applied to the entire length. It is decided that the foundation piles are to be applied to the sector of the 30 m land tunnel connected with the ventilation tower to deal with the differential settlement in central breakwater side. As the aim of the use of the foundation piles is to prevent the consolidation settlement, it is necessary to drive the piles to the bearing stratum, the Edsl layer, which is the diluvial deposit with over 50 of N value. The total length of the piles is about 70 m. According to the experience, this pile length could be achieved only by using steel pipe piles or reverse piles. In case of using reverse piles, there are several problems, for example, borehole wall failure due to groundwater level fluctuation by the leakage of water in the Nas layer with fine sand, or the stability of borehole wall in the intermediate soft Yc layer. The steel pipe piles have less problem and is decided to be adopted for the pile foundation of the land tunnel. (2) Countermeasures against failures (heaving of the ground bottom or boiling): A risk of the following failure will be present on the bottom of the deep open cut excavation in the ground according to the result of the soil investigations. In the Jonanjima side, Yc layer exists from -5 m to -42 m and N value of the layer is around 0 to 2. The weight of the earth behind the sheathing or the ground surface load near the sheathing might cause slipping surfaces that may result in heaving as the rising of the excavated bottom or the sheathing wall. The ground at central breakwater site consisted of, from the top, U layer, Yc layer, Ys layer, and Tocl and Tosl layers. The Ys layer of the thickness 3 m, as it is a continuous layer to the sea, an upward permeability current may be caused by the difference o f the water levels. When the seepage pressure exceeded the effective weight of the earth, the boiling of earth might be expected. Also, when the excavation is going on in the layer Toc 1, as this layer is continuous into the sea, an upward seawater pressure will act on the bottom surface of the Tocl layer. It is presumed that, when this upward force exceeded the effective weight of the layer above, a failure of the bottom of the excavation might have been caused by the floating (heaving of the ground bottom) of the bottom. (3) Ground improvement and measure of heaving: The ground improvement is required for increasing the
4.2.3 Construction work (1) Excavation work by means of earth retaining The method of open cut with earth retaining wall is used for the land tunnel sector, while a reclaimed land is constructed for the work sector on the sea for the ventilation tower construction. SMW, slurry congeal wall, and steel pipe piles are considered construction methods for the earth retaining wall. These methods are compared in accordance with the soil condition and penetration depth. Accordingly, the two methods were used, namely SMW for the sectors of the excavation depth 10-15 m, and steel pipe piles for the depth over 15 m to the deepest.
661
(2) Foundation improvement work Quick lime pile method: It is adopted for the entire length outside the central breakwater reclaimed land, and also for the excavated area of the 30 m land tunnel sector to be connected with the ventilation tower except for the bottom of the excavated area - ground layer lower than the ultimate floor surface. Deep mixing soil stabilization method It is employed for improving the ground under the layer improved by the quick lime pile method. As the improved ground required to be highly watertight and uniform so as to deal with the ground rising problem, the piles for the improvement are of block type structure so that they overlap with each other. However, as the grounds prepared by the dry jet mixing method and cement or lime deep mixing method contain many concrete bricks in the upper area of the reclaimed soil all over the application area, it is estimated to cause a difficulty in driving the digging blades at precise spots. Furthermore, the preparation of reliable ground was considered difficult by these methods, because the foundation piles, which are to be driven in beforehand for dealing with the consolidation settlement problem, may disturb the operation. Jet grout method (CJG method, JSG method): Although they are the methods partly different from one another, the methods primarily used mineralizer solution that are spread into the ground in a form of a rotatingjet. While cutting the ground, the method produced cylinder formed foundation. It is the method that is capable of improving the ground without having been disturbed by obstacles. The CJG method fully substitutes improved coiurrms for original ground and is more costly than the JSG method which substitutes partly. In such a reason, the use of CJG method is limited to boundary areas to soil retaining walls, where high accuracy is required for the water tightness.
Figure 4.Standard section of land tunnel.
5 CONCLUSION From the start of the operation in 1994 to October 99, six of the eleven caissons have been installed in a longitudinal accuracy of 20 mm. The ventilation tower has been completed except for the upper building. The works are in smooth progress toward the targeted opening period of the tunnel in 2002.
(3) Effect of the ground improvement Figure 5 shows the improved area in land tunnel. Lime piles were struck in a grid formation of 1 . 6 0 1.60 ~ m pitch. The check boring is conducted after one month of the curing period. The strength of the piles, q, have a level of 89 - 150 kN/m2,which is twice the targeted value. Samples collected indicates thin layers that seem to be of concreted lime. The reaction of the quicklime seems to lead the reinforcement of the ground. The effect of ground improvement is as anticipated. The confirmation o f the ground strength improvement of the JSG and CJG methods was to be conducted at every 3,000 m3 as specified in the specification. Inspection was conducted with 40 JSG piles and 20 CJG piles on the entire Jonanjima site. Result shows the strengths which are larger than 1.5 3 times of the design value.
662
CoastalGeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Engineering properties of granulated blast furnace slag H. Matsuda & T. Koreishi - Yamaguchi University,Ube,Japan N.Kitayama - Fukken Company Limited, Hiroshima,Japan Y. Ando - Fujita Company Limited, Tokyo,Japan Y. Nakano - Nagamune Kosan Limited, Tokuyama,Japan
ABSTRACT: For the purpose of using granulated BF (Blast Furnace) slag as an alternative material to marine sand which has so far been conventionally used in the foundation engineering, the static shear-strength properties of granulated BF slag were examined and compared with those of the natural marine sand. As a result, it was clarified that the grain-size distribution and the density of granulated BF slag were very similar to those of marine sand, and the void ratio of granulated BF slag was larger due to a lot of air bubbles inside the granulated BF slag grain and consequently its maximum dry density after compaction was smaller, as compared with those of marine sand. Although there occurred grain crush in granulated BF slag in a relatively low confining-pressure range, the friction angle of granulated BF slag in the critical state is higher than that of marine sand. and numerous points have yet to be clarified. To make good use of granulated BF slag in the ground improvement, the authors have studied and reported its physical and mechanical properties including compaction characteristics, permeability and simple shear strength (Matsuda et al. 1998). In the present study, triaxial compression tests on granulated BF slag and marine sand have been carried out to ascertain basic differences in their shear characteristics.
1 INTRODUCTION The sand compaction pile method is a soft-ground improvement method that excels in securing high stability and earthquake resistance of ground and has consequently been used for the foundation grounds of important structures. However, as this ground improvement method requires a large quantity of uniform-quality sand having high strength and permeability, gathering marine sand has recently become increasingly difficult. On the other hand, granulated BF slag is being produced from blast-furnace slag in large quantities. Sixty-nine percent of blast-furnace slag is shipped out from ironworks in Japan as granulated BF slag, with its annual quantity amounting to about 22 million tons in 1998 (Nippon Slag Association 1999). A large part of granulated BF slag is used as a material for the production of cement, and it is also used as a construction material for the roads subbase courses and also as a back-filling material for structures (Nagasaka et al. 1997). Besides, its applicability to the port and harbor structures has recently been scrutinized (Investigation committee 1989). Due to the fact that granulated BF slag is an artificial material and uniform in quality, and is available in large quantities, it may be considerable to use granulated BF slag as a substitute to marine sand for ground improvement by the sand compaction pile method that requires a lot of uniform marine sand. So far, there have been few cases of applying granulated BF slag to the method,
2 CHEMICAL COMPONENTS OF GRANULATED BLAST FURNACE SLAG Granulated BF slag is produced by quickly cooling and thereby granulating melted blast-furnace slag with pressurized water (100-350 kPa). As shown in Table 1, granulated BF slag is composed of calcium oxide: CaO (41.O%), silica: SiO?, (33.4%), alumina: A1203 (14.5%), magnesium oxide: MgO (6.0%), etc. It also contains small amount of sulfur S (1.0%), manganese Mn (0.7%), and iron oxide Fe0 (0.4%) (Investigation committee 1989). It mostly consists of amorphous vitreous grains, of which the grading is uniform. Each grain has numerous air bubbles both inside and out on its surface, taking a squarish shape (Investigation committee 1989) and having a latent hydraulic property.
663
Table 1. Comparisons between chemical properties of materials. W) SiOs M 2 0 3 CaO MgO S MnO F e 0 BF slag
Z:?zd
33.4 14.5 41.0 22.0
6.0
1.0
0.7
0.4
5.5
65.0
1.4
1.0
-
3.0
Natural sand 60.0 22.0
0.5
0.8
0.1
0.1
-
Andesite
60.0
17.0
6.0
3.0
0.2
1.0
-
Weathered soil
59.6 22.0
0.4
0.8
0.01
0.1
-
Table 2. Density, maximum void ratio and minimum void ratio. (dcm3)
P,
emax
emin
Granulated BF slag(1)
2.695
1.302
0.844
Granulated BF slag(I1)
2.624
1.521
1.043
Genkai sand
2.661
0.802
0.439
3 PHYSICAL, PROPERTIES OF GRANULATED BLAST FURNACE SLAG Physical properties, compaction characteristics and permeability of granulated BF slag were examined and compared with those of the natural sand. The materials used were two kinds; they are granulated BF slag and the natural sand from the Sea of Genkai. Figure 1 shows their grain-size distribution curves. It is reported that granulated BF slag has the uniformity coefficient in the range of 2.5 to 4.2 (Investigation committee 1989), and the sample of Genkai sand used in the present study is also in the same range of uniformity coefficient. Table 2 shows the densities and the minimum and maximum void ratios of the granulated BF slags and Genkai sand. The granulated BF slags had almost the same unit weight of soil grains as Genkai sand, but the minimum and maximum void ratios of the former were larger than those of the latter. It is because numerous air bubbles are formed in grains when the melted blast furnace slag is quickly cooled down with pressurized water. The porosity of granulated BF slag grains is evident from numerous pores observed on their surfaces as shown in Photo1. Figure 2 shows the results of non-cyclic compaction tests of granulated BF slag (11) and Genkai sand. The dry density of Genkai sand was influenced by its water content and it is approximately 1.8gicm3 at the maximum, in contrast to this, the dry density of granul_ated BF slag (11) was fairly constant at about 1.5g/cm3. It is confirmed that the dry density of granulated BF slag is not influenced by the water content.
Figure 3. e-log curyes.
664
Table 3. Coefficients of permeability.
g) BF slag(1) Genkai sand
65
g)
k (cds)
2.5-3.2X
62 6.1-7.3
10-l
k (cds)
82
1.7-2.2x
10-'
X 10-2 83
3.3-3.9X
10'2
4 SHEAR STRENGTH CHARACTERISTICS OF GRANULATED BLAST FURNACE SLAG
4.1 Drained shear strength
Figure 4(b). q/p' -
relationship on Genkai sand.
The grain-size distribution curves of the granulated BF slags obtained before and after the compaction tests were compared and no significant changes between them were observed, which indicated that grain crush hardly occurred in the granulated BF slag. On the other hand, it is reported that if granulated BF slag is compacted more than the number of compaction times made in ordinary compaction tests, its grains are crushed and its grainsize distribution changes (Investigation committee 1989). To ascertain the compression characteristics of the granulated BF slag, consolidation tests were carried out using a standard oedometer test device. Figure 3 shows the relationships between void ratio and consolidation pressure (e - log p) of granulated BF slag (I) and Genkai sand. The granulated BF slag had a large void ratio over a range of relative density as mentioned above, and their compressibility increased under increasing consolidation pressure than that of Genkai sand. Table 3 shows the coefficients of permeability of granulated BF slag (I) and Genkai sand, which were determined by constant-head permeability tests. The coefficient of permeability of granulated BF slag (I) (k = 1.7 - 3.2 x 10.' cm/s) was larger than that of Genkai sand (k = 3.3 7.3 x 10-2cm/s) at both the relative densities of about 60% and 80%. On the other hand, it is known that the coefficient of permeability of granulated BF slag decreases due to its latent hydraulic property. It is reported, however, that granulated BF slag retains a permeability coefficient of the order of 10-3cm/s even after it has been fully solidified (Matsuda et al. 1998).
-
665
Consolidated-drained tri-axial compression tests on granulated BF slag (11) and Genkai sand were carried out to compare their shear strengths. Densely compacted specimens (relative density being about 80%) were tested at confining pressure 50, 100, and 150 kPa. Granulated BF slag (11) was also tested at confining pressure of 300 kPa. Figure 4(a) and Figure 4(b) show the relationships between axial strain and stress ratio 4/p7, where q is deviator stress andp' is effective mean principal stress. They are given as follows: q = a,-a3 and p' = \al + k 3 )/3. As shown in Figure 4(a), the stress ratio of granulated BF slag (11) decreased as the confining pressure increased. On the other hand, the stress ratio of Genkai sand remained constant regardless of the variation of confining pressure. Grain crush induced by the shear may be the main factor of such decrease in the stress ratio of granulated BF slag. It seemed that grains of granulated BF slag (11) were crushed at the confining pressure of 50 150 kPa (and 300 kPa) during the tests, whereas the grains of Genkai sand remained almost intact at the same confining pressure 50 - 150 kPa. Figure 5(a) and Figure 5(b) show the relationships between the volumetric strain E, and the axial strain & a on granulated BF slag (11) and Genkai sand. As the confining pressure increased, it is significant that granulated BF slag (11) tends to contract its volume with the increase in axial strain, whereas the relationships between axial strain and volumetric strain of Genkai sand remained unchanged regardless of the different confining pressure. It seemed that as the confining pressure increased, grain crush was in progress, and the progressing grain crush changed the expanding tendency of granulated BF slag (11) to a contracting one. On the other hand, it seemed that since the confining pressure for Genkai sand was relatively low in this study, the grain crush by shear was small. Hence it is considered that the volume increased at every level of confining pressure. Figure 6(a) and Figure 6(b) show the dE,/dE, q / p ' relationships for granulated BF slag (11) and Genkai sand. Where E is shear strain, which is defined by E , = 2 / 3 ( ~ ,- E , ) , E a represents the axial
-
strain, and E ,. is the radial strain. Accordingly, the following equation can be derived: (1)
q / p " A4 -dE,,/dE,
where M stands for the critical stress ratio. If d e , , / d e , is zero, A4 equals q l p ' , which is the strength ratio without the dilatancy. As shown in Figure 6 (a) and Figure 6 (b), granulated BF slag (11) had the value of M=1.5, which was larger than that of Genkai sand. From the equation sin$, = 3M/(6 + M , the critical angle of internal friction & for granulated BF slag (11) was found to be 36.9". On the other hand, Genkai sand had the value of M =1.3, and its critical friction angle was found to be 32.3". Thus, granulated BF slag (11) had a larger critical friction angle than Genkai sand.
4.2 Undrained shear strength To ascertain the differences between the basic shear strength of granulated BF slag and those of natural sand based on their effective stress paths, consolidated-drained tri-axial compression tests were carried out. Densely compacted specimens of granulated BF slag (11) and Genkai sand were tested at the confining pressure of 50, 100, and 150 kPa. Figure 7(a) and Figure 7(b) show the relationships between axial strain E , and stress ratio q / p o for granulated BF slag (11) and Genkai sand. Where p o is the consolidation pressure. Both granulated BF slag (11) and Genkai sand have a large tangent modulus.
Figure 7(b). q/po-
Figure 5(b).
E
-
E
,
relationship on Genkai sand.
E
,
relationship on Genkai sand.
As the axial strain increased, the strain hardening occurred in Genkai sand, whereas granulated BF slag (11) developed only slight strain hardening. Both granulated BF slag (11) and Genkai sand showed the
666
tendency that as the consolidation pressure increases, the stress ratio decreases, but the decrease in the stress ratio of granulated BF slag (11) under the stress change from 50 kPa to 100 kPa was larger than that under the stress change from 100 kPa to 150 kPa. It is because the increasing confining pressure crushed grains or made the contacting surfaces smoothness and reduced the interlocking force among them. Figure 8(a) and Figure 8(b) show the relationships between axial strain and pore water pressure ratio ulpo for granulated BF slag (11) and Genkai sand. Under the consolidation pressure of 50 kPa, both granulated BF slag (11) and Genkai sand developed relatively large negative excess pore water pressure due to dilation and under this condition the magnitude of the negative excess pore pressure increases with the increase in the axial strain. In both cases, as the consolidation pressure increased, the excess pore water pressure changed from the negative to the positive. This means the change from expansion to contraction of the material. Similar tendency is also seen in the results of drained shear tests. It is considered that this tendency is due to the fact that the expansion is controlled by the crush of grains under increasing consolidation pressure. Figure 9(a) and Figure 9(b) show the p’/p0-q/po relationships of granulated BF slag (11) and Genkai sand. The effective stress path of Genkai sand was hardly affected by the variation of confining pressure, whereas that of granulated BF slag (11) is significantly affected. But, when the stress in granulated BF slag (11) reached its critical state, the stress paths under all confining pressures converged to a single straight line.
5 CONCLUSIONS For the purpose of using granulated BF slag as an alternative material to marine sand which has so far been used in foundation engineering, the authors examined various properties of granulated BF slag and compared them with those of marine sand. In conclusion, following results were obtained. 1)Granulated BF slag is crushed at the low confining pressure as 50 150 kPa (and 300 kPa) whereas the grains of Genkai sand remained almost intact at the same confining pressure 50 150 kPa. 2)Granulated BF slag has a larger critical friction angle than Genkai sand. 3)The effective stress path of Genkai sand was hardly affected by the variation of confining pressure, whereas that of granulated BF slag is significantly affected. 4)When the stress in granulated BF slag reached its critical state, the stress paths under all confining pressures converged to a single straight line.
-
-
Figure 9(b). Effective stress path for Genkai sand.
667
ACKNOWLEDGEMENT The authors would like to express their thanks to the Steel Industry Foundation for the Advancement of Environmental Protection Technology for its financial support to the research works performed in this paper, and also to Mr. Aung Swe of Fukken Co. Ltd., who has supported the English translation. REFERENCES Investigation committee on the utilization of blast furnace slag 1989. Application manual of the blast furnace slag in port and harbor construction. Port and Harbour Research Institute Ministry of Transport Japan, and Nippon Slag Association. (in Japanese) Matsuda, H., N. Kitayama, Y. Ando, & Y. Nakano 1998. Effective utilization of granulated slag in geotechnical engineering. Ground Engineering, 16: 33-40. (in Japanese) Nagasaka, Y., H. Sakai, K. Ohmori, & T. Hirata 1997. Disposal and utilization of surplus soil and waste from a geotechnical engineering point of view. Tsuchi-to-Kiso, Japanese Geotechnical Society, 45(5): 55-60. (in Japanese) Nippon Slag Association 1999. Statistical yearbook on iron and steel slag, 1998. (in Japanese)
668
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida (eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Estimating the dimensions of lightweight fill behind a seawall using the slice method Y. Mitarai, T. Amino & Y.Yamamura TechnicalResearch Institute, TOA Corporation,Yokohama,Japan
ABSTRACT: This paper presents the results of a parametric study to estimate the optimum dimensions of a lightweight fill used to reduce the lateral pressure behind a seawall. Three modes of failure were assumed and analyses were carried out using the conventional slice method. The effect of bottom width, thickness and shape of the lightweight fill on development of lateral pressure behind a quay wall were investigated. In addition, the parametric study was extended to estimate the effect of using rubble stone as the interface with the lightweight fill.
1 INTRODUCTION
2 ESTIMATION OF LATERAL, PRESSURE
Super Geo-Material (SGM) is a lightweight slurry fill developed for backfilling seawalls in the coastal areas. It consists of low quality soil or dredging, cement and a lightweight material, such as air foam or EPS (Expanded PolyStyrol) beads. This slurry fill has many advantages including reduction of lateral earth pressure behind the retaining walls, prevention of ground subsistence, as a countermeasure against liquefaction, and also recycling of low-quality soils in the construction site. It is also possible to control the density (ranging from 6 to 15 kN/m3) and the undrained shear strength, S, (50 to 150 kPa). This newly developed lightweight slurry fill was firstly adopted for reconstruction of the damaged quay walls at Kobe Port, which was damaged by the Great Hanshin-Awaji Earthquake of 1995 in Japan (Matsuo, 1996; Tani, 1996). In order to determine the optimum dimensions of the improved body behind the seawall, it is necessary to estimate the lateral earth pressure. From this point of view, the earth pressure for design is estimated using the conventional Slice Method, commonly used in slope stability analysis. This method was adopted in the design manual for lightweight treated soil in Japanese Standard manual for port and harbour facilities. This paper presents the results of a parametric study to estimate the optimum dimensions (such as width, depth, and shape) of a lightweight fill used behind a seawall to reduce the lateral pressure.
In conventional slice method, the lateral earth pressure is calculated using Eq.(l) (Tsuchida, 1999; Amino et al., 2000). A straight sliding plane acting from the backward of the structure is assumed in this method; and then the soil block enclosed by the sliding surface and the wall is divided into regular slices, as schematically shown in Fig.1.
Figure 1. Calculation of lateral pressure using the slice method
l+tanatan@ .
669
The parameters are as follows: P= resultant lateral earth pressure, a= angle of sliding plane, @ = internal friction angle of back ground fill, 6= friction angle of wall, kh= horizontal seismic coefficient, I= sliding length of a slice, WI=total weight of a slice, WI=effective weight of a slice, Ti= shear force of a slice, Vi= vertical force at right side of a slice, and Ei= horizontal force at right side of a slice. The maximum earth pressures are calculated for 3 modes of failure as schematically shown in Fig.2 when SGM treated lightweight fill is used. These assumed modes of failure are as follows: Mode 1: Earth pressure indicates the value when sliding plane is outside of the treated zone. This case will be caused by too high shear strength or inadequate width of SGM treated materials. Mode 2: This mode takes place when the width of SGM treated zone is wide enough and the failure plane is located inside of the treated mass can be treated as an infinite surcharge material in this case. Mode 3: Maximum earth pressure zone is 10cated along the SGM treated zone. Shape of the sliding plane is compounded from ABCD, and secondary sliding plane is assumed at point C, which enables to evaluate the earth pressure is the maximum in this mode among the three when the design seismic coefficient and or the treatment depth is great.
Figure 3. Base section used for parametric study 3 OUTLINE OF A PARAMETRIC STUDY
This section summarizes a case study to determine the dimensions of the SGM treated ground for a quay wall. Cross section of quay wall is presented in Fig.3 along with various conditions employed both in design and SGM treated mass. The final objective is to obtain the width and the depth of the improved zone. The results of the parametric study are presented in the following section. The initial conditions of this study is as follows: (a) Material parameters & surcharge: 0 wet unit weight of SGM =12kN/m3 0 apparent cohesion c=10kN/m2 0 0 0
0
wet unit weight of the backfill=10kN/m3 internal friction angle 6 =30° Friction coefficient between SGM and the backfill material ,U =tan3O0=0.60 surcharge pressure: w=lOkPa
(b) Dimensions: SGM bottom width L=5.0 to [email protected] thickness h=3.0m, 6.0m, 9.0m 0 Caisson bottom level = -9.0m, -12.0m, -14.5m 0 Caisson top level =+3.0m 0
0
Earth pressure ratio in the parametric study using slice method is given by Eq. (2).
where PO= resultant earth pressure of non-treated ground (kN/m), P= resultant earth pressure conesponding to the three modes shown in Fig.2 (kN/m).
4 RESULTS OF THE PARAMETRIC STUDY 4.1 Effect of the bottom width of SGM zone
Figure 2. modes
Slip surface assumed for different failure
670
Figure 4 shows the Rp obtained from the study versus L values (bottom width of SGM-treated zone). In case of H (height of quay wall)=17.5m, there is a trend that Rp values decrease with increasing L
values under constant H and k,. However, the Rp values indicated a constant value when L and k, are relatively small values. On the other hand, as shown in Fig. 5 (a and b), correlation exists among Rp, Lmin and failure modes illustrated in Fig.2. The constant value of Rp pointed out earlier is caused by failure mode2 such that Pmax and SGM mass can be treated as an infinite mass when L is beyond Lmin, as indicated in FigSa.
SGM mass is therefore available to treat as a surcharge ~ M S S in this case. On the other hand, Rp again increases when h and k, exceed 9.0m and 0.25, respectively as shown in FigSb. This was caused by failure mode3 giving P m a because of the small value of p = 0.6.
Figure 4. Effect of varying width of SGM on lateral earth pressure ratio
RP h<9. Om &
1
MODE-3
1
r”
kh SO.20
h 2 9 .Om &
MODE-1
MODE-2
I
k, =O. 2 5
MODE-3
,L
+ L Figure 5. Effect of various modes of failure on lateral earth pressure ratio
671
Figure 6. Effect of h/H on lateral earth pressure ratio 4.2 Effect of the thickness of SGM zone Fig.6 shows the values of Rp versus hlH ratios. Rp values proportionally decrease with increasing h/H ratios when kh=o to 0.2, because failure mode 2 prevails, in which SGM can be treated as an infinite mass, gives Pmax. However, Rp indicate approximately a constant value of 0.7 to 0.8 when kh=0.25 and h/H>0.5. This was caused by the small ,U value as pointed out in section 4.1. 4.3 Effect of the shape of SGM zone Fig. 7 shows the Rp values versus the area of SGMtreated zone, in case of H= 17.5m and kh=0.25. The shapes of the SGM treated zones are reversetrapezoid, trapezoid and rectangular. The effect of shape could be negligible when h is relatively small like h d m , while this effect should become greater the larger the h values (hr9m). In case of equal area of treated zones and different shapes, the rank of Rp values is as follows: reverse-trapezoid < trapezoid < rectangle. The reason is described as follows. The earth pressure, calculated based on the slice method, is evaluated with the assumption that the secondary failure plane at Mode 3 (see Fig.2), gives the smallest P value (= resultant earth pressure acting from back of treated area) for reverse-trapezoid, and the same P value for trapezoid & rectangular shapes. The Rp values of rectangle are larger than those of trapezoid, and this is caused by the differences in (a) length of shear plane being resistant to earth pressure and (b) weight of soil mass above the shear plane.
672
Figure 7. Effect of change in cross sectional area of SGM on Rp
in which the sand fill ( 6 =30° ) is replaced by rubble stone (6 Evidently, appropriate evaluation of cb value of backfill is also an important subject. Shape effects on Rp are presented in Fig.10. It is clear that reverse-trapezoid effect is higher than rectangular and trapezoid.
Figure 8. Investigation of interface of SGM with rubble stone for 3 cases 5 EFFECT OF INTERFACE OF SGM & THE RUBBLE STONE It should be pointed out that the effect of lateral earth pressure ratio was not so pronounced when khd.25 and hr9m. This was caused by the small ,u value adopted in the analysis. In order to overcome this problem, the use of the material with higher p value should be considered. Three cases were investigated in this study as schematically shown in Fig.8. 1) rubble stone at the bottom end of SGM-treated zone (Case-A), 2) use of rubble stone as back fill (Case-B), and 3) combined Cases A and B (Case-C). Results of analyses are summarized in Fig 9, and the use of stone is very effective in the Case B and C
~i~~~~ 9. Variation o f ~ withA p when rubble stone is used as the interface 673
2) The Rp values decrease with increasing L values under constant H and k h , but the values are constant when L and kh are relatively small (i.e. in case of h s 6m and khs 0.20). On the other hand, Rp again increases when h and kh exceed 9.0m and 0.25, respectively. 3) The Rp values decrease with increase of hlH ratios when kll=O to 0.2, because mode2, in which SGM can be treated as an infinite mass gives Pmax. However, the Rp value indicate approximately a constant value of 0.7 to 0.8 when kh=0.25 and hlHA.5. This was produced by small p value of 0.6 (i.e. in case of using sand with (b =30° for backfill).
4) The effect of shape could be negligible when h is relatively small, for example h s 6m, while this effect should become greater when h is larger (hr9m). In terms of equal treated zones and different shapes, the rank of Rp values is as follows: reverse-trapezoid
5) It is possible to get the value of Rp to a level of 0.5 to 0.6 with the combined use of SGM and rubble stone when khd.25 and h29m. This can be done by using granular fill or stone with high ,U values. REFERENCES
Figure 10. Effect of shape of SGM on Rp From the analyses, it can be expected to get the value of Rp to a level of 0.5 to 0.6 with the combined use of SGM and rubble stone when kh20.25. One problem in using stone is penetration of the SGM into the stone. Therefore, it is necessary to develop a cost-effective technique to avoid penetration of the SGM into the rubble stone. 6 CONCLUSIONS
Amino, T., Mitarai, Y., and Tsuchida,T. (2000) Design method for improvement dimensions of the lightweight cement treated mass material by slice method, JGS Symposium, Japanese Geotechnical Society, Tokyo (in Japanese) Matsuo, 0. (1996) Damage to river dikes, Soils and Foundations, Special issue, January, Tokyo : 235-240 Tani, S. (1996) Damage to earth dams, Soils and Foundations, Special issue, January, Tokyo : 263-272 Tsuchida,T.(1999) Slice Method for Pressure Analysis and its Application to Lightweight Fill, Technical Note of the Port and Harbor Research Institute Ministry of Transport, May, Yokosuka
A parametric study is presented to determine the dimensions of SGM behind a seawall using the conventional slice method. The following conclusions and suggestions are derived from this study: 1)It is possible to estimate the lateral pressure ratio (Rp) using SGM at backfill of quay wall using the slice method.
674
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida(eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Geo-material properties of wasted oyster shell-sand mixture and its application as material for sand compaction pile Y. Miyaji Shiogama Port Construction OfJie, Ministry of Transport,Japan
T.Okumura Faculty of Environmental Science and Technology,Okayama University,Japan
ABSTRACT: The recycling, reclamation and material reuse activities become important issues for social responsibilities in recent years and for the future. construction industry also takes action for its waste, and is requested to reuse the miscellaneous industrial waste, too. Wasted oyster shell makes major environmental pollution in Ishinomaki City, Miyagi prefecture, where oyster harvest is major parts of fishing product. This paper reports the recycling of a huge amount of oyster-shell from fisheries waste, being environmental problem. The report contains geotechnical study and a case record of reusing oyster-shell for soil improvement under the break water at Ishinomaki Port. 1 INTRODUCTION
200 species of oyster inhabit on the Earth and 30 species live in Japanese waters. 4 species of them have been harvested and used for food. Japanese eat oyster from ancient times, the oyster shells were wasted with their unused daily necessities, and such shell heaped up as shell mound which is the subject of archaeology now. 80 years ago, aquaculture of MAGAKZ :giant pacific oyster (Crussostrelu gigas) has begun in western Japan. In Ishinomaki, oyster are hatched in June and July, continues to be hung 1 -2 years, then is landed for market. Soft meat in shells is sold as useful food and hard shell is wasted as useless. Fishery enterprises of Miyagi prefecture produce 3.7 % of Japanese annual haul. Ishinomaki city takes 9.7 % part of Miyagi pref.. Fishery household of oyster aquaculture in Ishinomaki city is 27.9 % of that in Miyagi pref. and Miyagi pref. is 19.0 % in Japan. Fishery product of oyster aquaculture in Ishinomaki city is 11.2 % of fishery product in that city. This product has being increased. Landed quantity of oyster in Ishinomaki is 1,700 ton to 1,800 ton per year, recently getting increased to 1,900 ton. Quantity of oyster-shell follows the product of oyster and has became 13,000 ton per year. 25 % of oyster shell is utilized for fertilizer in agriculture, feed for livestock and medical supplies, but the remaining 75 % is either disposed of or stocked without any eventual use in mind in Ishinomaki city which is a tourist resort.
Figure
Harvesed Oyster
befor 1991
Figure1 shows quantity of oyster shell produced annually by 8 fishery associations in Ishinomaki city.
675
There was no place to stock oyster-shell in 1991, and scenic area was occupied by such useless oyster shell which was source of sanitary pollution. Photograph1 shows one of the stock yard. In 1991, study on utilization of oyster shell was begun for port and harbor constructions. Soil laboratory tests were taken place to confirm possibility of use for sand compaction pile (SCP) for soft soil improvement. 2 ENGNEERING PROPERTIES OF OYSTER SHELL, CRUSEHD SHELL AND CRUSHED SHELL-SAND MIXTURE
In 1991 and 1992, laboratory tests, for uncrushed oyster shell, crushed shell with various grain size and crushed shell-sand mixture with various ratio were carried out (Hashidate et al., 1994). Tests for crude and crushed oyster shell were performed by large scale apparatus. Oyster shell has various shapes and sizes, 8-14 cm in length, 4-7 cm in width, 2-4 cm in thickness. Oyster has 2 pieces of shell. Right side shell is flat and heavier than left side shell and left shell is rising and lighter than right. Oyster shell is said to be composed mainly of calcium carbonate, inside covered by hard skin, outside covered with soft thin plate layer and has void for habitat of sea organism. Oyster shell is rather thin, but well graded when crushed. Properties of crushed shell are: a) Grain density and bulk density are smaller than those of sand. b) Permeability is greater than that of sand. c) Angle of internal friction (shear strength)is comparable to sand. d)CBR is comparable to sand. e) Compressibility is larger than that of sand. f) Toxic substances are not included. g) Mixability with sand is good.
Figure 3 Model ground for SPT
Table 1 Laboratory tests Soil test
Grain density 3
Material Shell Crude oyster shell Crushed shell( q5 rnax3flmm) Sand Crude sand incl. Impurities washed sand Mixture E=B+C+D(in volume)
A B C
D E
Soil test
(dcm) 1.74 2.70 2.70 2.70 2.70
Max. density
Min. density
Grain size
(g/cm3) 0.25 1.08 1.55 1.69 1.45
(RIern3)
(mm) 59 2.50 0.56 0.61 0.63
50%
0.18 0.80 1.26 1.38 1.11
Permeability Optimum moisture content
Material Shell ICrushed shell( q5 max3flmm) B
I
t%)
(cm/s) 3 . 1 81~Oa
34.5
676
Moisture content
Bulk density
Compression index
(%)
33.9 24.0
(pJcm3) 0.34 0.91
7.0 0.77
15.2 9.2 14.6
1.28 1.31 1.16
o.044 0.078
Modified CBR Max. dry Degree of Degree of density compaction compaction (dcm3) 90% 95% 1.13 14.8 16.5
Samples and test results are tabulated in Table1 and Figure2. In 1992, trial usage of oyster shell-sand mixtures to 81 sand compaction piles for improving soft clay layer under the breakwater in Ishinomaki port, and some relevant laboratory tests and field tests were performed. In 1993, practical usage to 420 sand compaction piles and relevant investigations were performed.
3 STRENGTH OF INSITU SCP MATERIALS Strength of driven SCP materials is requested to coincide with the required strength by design. Standard penetration test (SPT) for SCP is common in Japan, to confirm in-situ strength. In case of sand, relation between SPT N-value and strength (internal friction angle) is fairly established. However, in case of shell-sand mixture, it is unknown, and requested to establish relations. In 1992, SPT on the model ground and large scale direct shear test were carried out for mixture with various densities. Strong correlation was observed between SPT N-value, dry density and internal friction angle of shell-sand mixture (Okumura et al., 1996). Then it became possible to estimate strength (bulk density and internal friction angle) of in-situ material from SPT. Figure3 shows model ground where SPT were carried out for shell-sand mixture E of various densities. Test results are indicated graphically in Figure4 and FigureS.
Figure 5 Direct shear test result
4 APPLICATION FOR SCP IN CONSTRUCTION OF PORT AND HARBOR
Making use of such good properties, large amount of oyster shell can be utilized in port and harbor constructions, for back filling of revetments and quay walls, sand or gravel drains and sand compaction piles for quay walls, littoral nourishment, sea bottom covering, etc. For the purpose of effective usage, however, such engineering properties of oyster shells and shell-sand mixtures as bulk density, shear strength and compressibility should be clarified.
In 1992, 81 SCP with mixture E were driven to soft clay experimentally. Mixing ratio of the mixture was sand: crushed oyster shell=2:l in volume. Mixing was made naturally on the way of transportation from the stock yard to the working vessel (SCP barge). Sand loaded by 2 dumping trucks and the crushed shell by 1 truck were dumped alternatively to the cargo ship equipped with grab bucket. Each dumping truck carries same volume of material, to the hold, and sand and shell were mixed and moved to level for keeping cargo ship stability by its grab bucket. Further more, the materials were unloaded from the hold and loaded to hopper on the SCP barge. Then the mixture was transferred to 3 belt conveyers and supplied to driving pipes. These actions of works made good mixing of the materials. Photograph2 and Photograph3 show such mixing works. Homogeneity was confirmed by examining grain size distribution of SPT samples taken from the driven SCP.
In Situ laboratory of cargo
jUlibviXdcliiyi110:j Silt content /yean grain size
(mm)
~
0.774
Uniformity coefficient
4.692
Coefficient of curvature
1.069
+ 0.711
6.469
1.281
0.654
I
I
4.002
1.067
677
SPT result showed that N-values were more than 15 through whole length and their average was 16.2, and almost equal to 16.3 of sand alone. 3 types of crushing machines were tried and impact type was adopted. From 1993, same mixture was applied to SCP for soil improvement under the breakwater and from 1997, applied to that under quay wall in the same way of experimental driving. Homogeneity of samples taken from the belt conveyers was inspected continuously for each 5,000 m3 by grain size distribution test. The test indicated almost same results with laboratory test. Table2 shows comparison of the results made by such works. Figure6 shows grain size distribution of crushed oyster shell taken from 4 types of the crushing machines. Figure7 shows grain size distribution of the mixture sampled on the belt conveyers. Figure8 shows SPT result of driven SCP with the mixture as a typical case. As index of material strength of SCP, 144 Nvalues in 16 SCP made of pure sand and 158 Nvalues in 18 SCP made of the mixture were collected in 7 years. Average N-value of sand was 17.9, and of the mixture was 18.8. It showed the mixture has equivalent or more strength to sand.
Figure 7 Grain size distribution of shell-sand mixture
678
A cycle time of driving SCP was 56.2 minutes for both sand and mixture. Initial settlements of soil improved by SCP by weight of caissons and cap concrete placed on were observed, and the settlement on sand was 26-23 cm and on the mixture was only 24 cm. Calculated settlement which was estimated from the value of material taken from SPT of experimental SCP, was 21 cm for sand and 42 cm for the mixture. Observation shows settlement for the sand coincides with the designed and for the mixture is not different from sand, i.e. only 60 % of its design. Figure9 shows imagined SCP method. Figure10 shows plan layout of SCP in Figurell. Figurell shows cross section of application of mixture SCP under the breakwater when construction completed. 5 UTILIZATION OF OYSTER SHELL TO SCP AND ITS EFFECT TO ENVIRONMENT In 8 years from 1992 to 1999, annually average 9,000 ton (2,000 ton min., 20,000 ton max.) of oyster shell were changed to total 91,441 m3 of crushed-shell and mixed with sand.
Figure 10 Details of shell-sand compaction piles(unit:rnm)
Figure12 shows quantity of oyster shell utilized for SCP, which is about 70 % of harvested oystershell in Ishinomaki city Above utilization made contribution to the environment as follows: 1. Reduced quantity of useless stock of oyster shell and number of stock yard (remained one yard only). 2. Improved landscape of Ishinomaki which is scenic area and tourist resort in a quasinational park. 3. Stopped offensive odor and flies by utilizing of fresh oyster shell to SCP without any stock. 4. Maintained and increased the yield of oyster, without any problem of waste disposal. 5. Contributed to conservation of the natural environment and resource by deducting consumption of sand. 6. Reduced cost of soil improvement by adopting simple mixing method of sand and shell.
6 SUPPLIMENTARY REMARKS Utilization of oyster-shell for SCP induced various use of oyster-shell in Ishinomaki city; remediation of small river and town drainage, materials for road constructions and sports court, etc. This project adopted the sand-shell mixture with mix-ratio of sand : shell =2 : 1 in volume. Laboratory tests for mixture with mix-ratio of sand : shell =1 : 1 were carried out for increase of consumption of oyster-shell. Test results showed such mixture has enough strength for SCP. It is said that SCP materials including high percentage of silt size grain does not show good compaction effect, and in crushing work of oyster-shell, small size grains (silt size) are produced naturally.
Figure 9 Imagined SCP Methord
3,140 s c p (length 18 m each) of sand-shel1 mixture were driven to soft soil under the breakwater and the quay in Ishinomaki port.
679
Figure 11 Shell-sand compaction piles under a composite breakwater (unit:m)
REFERENCES Hashidate, H., Fukuda, T., Okumura, T. & Kobayashi, M. 1994. Engineering Properties of oyster shell-sand mixture and its application for imProc. proving soft se,a bottom sediments. Techno-Ocean 94 International Symposium, 2:607-6 12. Okumura, T. & Kobayashi, M. 1996.0yster ShallSand mixture as material of sand compaction piles. Proc. of The Second International Congress on Environmental Geotechnics: 863-868
Figure 12 Utilized Oyster Shell after 1992 However, sand-shell mixture including high percentage of silt size was proved by the vibration test to have no noticeable difference from the one with low content . ACKNOWLEDGEMENTS: These investigations were carried out under the instruction of “Investigation Committee for Recycling Construction Materials in Ishinomaki Port (Chairman :Prof. Yanagisawa, E.)” and with cooperation of Ishinomaki City, Miyagi Prefecture, and Miyagi Federation of Fisheries Cooperative Associations. The authors are much obliged to the above organizations.
680
Coastal GeotechnicalEngineeringin Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISEN 90 5809 751 7
Evaluation of engineering properties of in-situ light-weight soil with air foam using X-ray CT technique T. Mukunoki & T. Nagatome Graduate School of Science and Technology,Kumamoto University,Japan
J.Otani Department of Civil and Environmental Engineering, Kumamoto University,Japan
Y. Kikuchi Port and Harbour Research Institute, Ministry of Transport, Yolwsuku,Japan
ABSTRACT: The purpose of this study is to investigate both the material property and the behavior of stabilized light-weight soil due to unconfined compression using industrial X-ray CT scanner. Here, the distribution of the density, the condition of mixing, and the distribution of air foam in in-situ specimen were investigated quantitatively using image processing analysis. Furthermore, a series of unconfined compression test were conducted for these specimens and the density evolution was investigated by scanning of the specimen during unconfined compression. Finally, the effectiveness of the industrial X-ray CT scanner to geotechnical engineering was confirmed based on all the results in this research. 1 INTRODUCTION
method. And recently, this machine for industrial use has been developed, in which the power of X-ray is much higher than that of medical one. The authors have started the research on the application of the industrial X-ray CT scanner to geotechnical engineering since 1997(0tani et al. 1997). As far as the research on the engineering property of SGM is concerned, the authors have conducted the nondestructive testing on SGM made in laboratory using industrial X-ray CT scanner and both the physical and mechanical properties were evaluated quantitatively (Otani et al. 1999). The purpose of this study is to evaluate the engineering property of in-situ SGM using industrial X-ray CT scanner. In this paper, the distributions of the density and the air foam in the in-situ specimens are evaluated quantitatively, and the density evolution and the failure mechanism due to unconfined compression are also investigated based on the results of nondestructive testing with the image processing analysis. The detailed introduction of the system of industrial X-ray CT scanner has been described in the reference (Otani et al. 1997).
“Light-weight soil” which has originally been developed for the purpose of reduction of applied forces such as embankment and reclamation on soft ground. A stabilized light-weight soil which is composed of any soils, cement, and light materials such as air foam or Expanded PolyStirol (EPS) beads is one of this technique. Recently, this material has been widely used in Japan. One of the applications for this material is the use of backfill materials for quay walls at the seaport for the purpose of soil improvement technique on soft ground. Super Geo-Material which is known by the name of SGM (Tuchida 1995, Tuchida et al. 1996) has been proposed by the research group including the Port and Harbor Research Institute (PHRI), Ministry of Transport, Japan, the Coastal Development Institute of Technology (CDIT) and 23 private corporations. This material is made of mixing dredged slurry with cement and air foam or EPS beads. Because of the composite feature for this material, the evaluation of mixing condition seems to be a key issue. However, it is very difficult to investigate the condition in the soil without any destruction, and the quantitative discussion has also not been done so far. Meanwhile, an X-ray Computed Tomography (CT) scanner, which is known by the name of medical diagnostic methods, has been used even for engineering purposes as a nondestructive testing
2 X-RAYCTDATA The CT image data are evaluated quantitatively with following so called ‘CT-value’: CT-value = ( p , - p , ) ~ pw /
681
(1)
where pf: coefficient of absorption at scanning point; p,,,: coefficient of absorption for water; and K: material constant. It is noted that the coefficient of absorption for air is zero for the condition of K= 1000 and then, the CT-value of the air is -1000. Figure 1 shows the relation between CT-value and unit weight of the soil specimen of SGM. Those results show a linear relation among all the results, so that the CT-value can be a parameter for evaluating the distribution of the density. Thus this CT-value makes quantitative discussion possible under this relation with CT images. It is promised from these results that the change of the density with strain localization in the soil could also be evaluated quantitatively. It is noted here that the CT images are drawn with black color for low density and white color for high density and the total number of gray level is 256.
where P a v : average value of density for the specimen; C T a v : average value of CT-value for the specimen; and CTl: average value of CT-value for each cross section. In order to obtain the contents of the air foam in each cross-section, the CT-value of air foams in the soil is determined using thresholding technique. The basic idea of thresholding technique is illustrated in Fig.6. The process of obtaining the thresholding value for CT-value is shown as follows: Based on the mixing condition shown in Table 1, it is assumed that the change of the density between at the time of mixing and after sampling is caused by the decrease of air foams due to overburden pressure with water in the sea, and the volume of air foams in the SGM after sampling is calculated; Then, a CT-value for the thresholding between air foams and soil with cement is assumed and the volume of air foams is calculated in each cross section using this value; If the average volume of the air foams in the specimen after calculation is not equal to the one obtained for the process (a), the trial thresholding value is changed. Then, the process is restarted from the process (b) until the calculated average value of air foams is close enough to the value obtained at the process (a).
3 RESULTS OF IMAGE PROCESSING ANALYSIS The specimens used in this study are in-situ SGM constructed in the caisson at new Kumamoto seaport, Japan and were sampled at the depth of 0.9-1.0m in the ground. Table 1 shows the mixing condition of the specimen used in this study. The dredging soil was Ariake clay and the liquid to make air foam is an interfacial active agent. The size of the specimen is the diameter of 50mm with the height of 100mm. The scanning of the specimen was conducted from the top to the bottom of the specimen with every lmm thickness, so that 100 images were obtained.
3.1 Evaluation of material property In order to investigate the distribution of air foam in SGM, the soil stabilized by cement without air foam was also scanned. Figure 2 (a) is one of the cross-sectional images for this soil and Fig. 2(b) shows the distribution of CT-value for the direction shown by arrows in Fig.2 (a). Figure 3 shows the same results for the case of in-situ SGM. According to these results, although the distribution of density for the soil without air foam is relatively smooth as shown in Fig.2 (b), there is a variation of the density for the SGM as realized by not only the gray level distribution in Fig.3 (a) but also the distribution of CT-values in Fig.3 (b). Figures 4 and 5 show the distributions of both wet density and the contents of air foam in the soil specimen. The wet density in each cross-section, pl is obtained by following equation:
Finally, the thresholding value for the air foams in this case is determined and this value was the CT-value of 185. As shown in Figs.4 and 5, it is easily realized that both wet density and volume of air foams are distributed randomly in the soil. Especially, there is much air foam around lower part of the specimen. Thus, the physical property of the SGM is capable of investigating quantitatively using industrial X-ray CT scanner.
Density of the SGM specimen(t/m3)
Fig. 1 CT-value -density of SGM specimen relationship
682
the initial stage to the pre-peak stage according to Fig. 8(a) and (b). However, there is a large shear band at the post-peak stage as shown in Fig.8(c). 90
80
-
70
5
60
2
50
E
2 M '-
-
Average 1.16t/rn3
40 30
0.6
0.8
1.0
1.2
1.4
Wet density (t/m3)
Fig.4 The distribution of wet density 90 80 70
-
60
E
E 50 3 M 40
W
. I
2
30
20 10 0
10
20
Contents of air foam (%)
30
Fig.5 The distribution of air foam
Fig.3 Two-dimensional image and the CT-value distribution
3.2 Failure of the SGM In this study, the unconfined compression test and nondestructive testing using X-ray CT scanner were conducted separately. The nondestructive testing was conducted at the stages of before compression, pre-peak and post-peak. Figure 7 shows the stress-strain relationship for the unconfined compression test. The locations (a), (b) and (c) in the figure show the scanning points. Figure 8 shows the images at the vertical section for the three stages. It is noted that there is a slight change from
Fig.6 Thresholding technique
2000
-
0 Scanning - point _
1500
(a) Initial stage (b) Pre-peak stage (c) Post-peak stage
I
24 & W
2 1000 2
;ij
500
0
2.5
Strain (%) Fig. 7 Stress-strainrelationship
In order to visualize the surface condition of shear zone, Figure 11 shows a graphical presentation of failure surface in the specimen for the post-peak stage. The complicated shape and surface condition can be realized and thus the effectiveness of this graphical presentation is confirmed. Figure 12 shows the distribution of average density at each stage due to unconfined compression using Eq. (2). It is pointed out that the average density of the specimen decreases during unconfined compression but the shapes of distribution curve for these stages are not much changed. Especially, it is considered that the decrease of the density at the process from initial stage to pre-peak stage around the bottom part of the specimen is related to the initial condition of the distribution of air voids around the lower part of the specimen as shown Fig.5. Figure 13 shows the value of density subtraction due to unconfined compression between the pre-peak condition to the initial condition and the post-peak condition to the pre-peak condition. It is realized that the value of ppost-p&-ppre-peak is large around the height where the value of Ppre-peak-Pinitil is relatively small. Based on those discussion, it is concluded that that mechanical property of SGM can be also visualized and evaluated quantitatively using X-ray CT scanner.
4 CONCLUSIONS The industrial X-ray CT scanner was applied to the field of geotechnical engineering for the purpose of investigating the engineering property of the light-weight soil with air foam. The conclusions drawn from this study are shown as follows;
Fig.9 Two-dimensional images of each location from the bottom at the post-peak stage Figure 9 shows the cross-sectional images at three different heights of the specimen at the post-peak stage. It is realized from these figures that the shear band is not the plane surface but the rounding and complicated one. Figures 10 shows the vertical cross sectional images for post-peak condition. Nine images of equally spaced slice are presented for each condition, in which the location of each vertical cross section is indicated in the horizontal cross sectional image as shown Fig. lO(a). As shown these figures, shear zone was observed in the each cross section. Furthermore, it was pointed that the bigger air voids were not always connected during the process of progressing the shear zone.
(1) The distributions of both wet density and air void ratio were investigated visually and these were also evaluated quantitatively; and (2) The change of density due to unconfined compression is caused by the effect of shear band growth but not the change of air voids and the failure of SGM is affected by the existing of relatively large air voids. It is also concluded that the X-ray CT scanner for industrial use may be a useful tool for geomaterials not only for visualization of inside behavior but also the quantitative evaluation of the engineering property in the soil.
684
Fig. 10 Visualization of vertical cross sectional images to the direction of right angle .-'
0.9
1
J
1.1 1.2 1.3 1.4
Wet density (t/m3)
Fig.12 Distribution of average density at each stage due to unconfined compression Fig. 11 Three-dimensional graphic of slip surface Table 1.Condition of mixing Mass(kg/m3) Volumetric fraction (L) Dry mass 367.0 136.0 Water 624.0 606.0 Cement 100.0 32.8 Air foam 9.3 225 .O Total 1100.3 999.8
ACKNOWLEDGEMENT X-ray CT scanner apparatus used in this study is a property of the group of Rock Engineering in Department of Civil & Environmental Engineering, Kumamoto University, Japan. The author would like to express their gratitude to the head of this group, Professor Sugawara, K. and Obara, Y. who gave us the opportunity to use this valuable apparatus.
685
1
90 80
70 n
60
E E 50
W
E .-M
'
40
30 20 10
-0.1 -0.05 0 0.05 Value of density evolution (t/m3) Fig. 13 Density evolution at each stage due to unconfined compression
REFERENCES Otani J., Obara Y. and Mukunoki T. 1997. Characterizationof failure in soils using X-ray C T Scanner. Proc. of International Symposium on Deformation and progressivefailure in Geomehanics, IS Nagoya97, Pergamon: 311-314. Otani J., Mukunoki T. and Kikuchi Y. 1999. Evaluation of engineering property on light-weight soil with air foam using X-ray CT scanner. Proc. of Eleventh Asian Regional Conference on Soil Mechanics and Geotechnical Engineering in Korea, Balkema: 57-60. Tsuchida T. 1995. Super Geo-Material Project in coastal zone. Proc. of International Symposium on Coastal Ocean Space Utilization COSU'95, Yokohama : 22-31. Tsuchida T, Takeuchi D., Okumura T., and Kishida T., 1996. Development of light-weight fill from dredgings. Proc. of Enviromental Geotechnics, Kamon (ed.), Balkema: 415 -420.
686
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Geotechnical characteristics of stone-rnuck for construction material of port and harbor structures H. Nakajima & E. Hamada - Kiso-Jiban Consultants Company Limited, Tokyo,Japan K. Shinohara & T. Kondo - The 2nd District Port Construction Bureau, Ministry of Transport, Yokohama, Japan
KUW % U ~- Service Center of Port Engineering, Japan S.Furuya - Tetra Company Limited, Japan (Formerly: Service Center of Port Engineering)
ABSTRACT: Stone-muck had been wasted in many quarries in Japan after rubbles or rock(stone) materials were produced. However, the huge amount of stone-muck that remained in quarries is recently spotlighted as construction material especially for port and harbor structures because of its low cost performance and environmental preservation. The authors made field investigations on main quarries in Tohoku district (northern area of Honshu island) to study the production process and piled condition of the stone-muck. A series of large scale triaxial compression tests and some physical tests were carried out on stone-muck samples collected from typical quarries taking into account its use as construction material for the port and harbor structures, for instance as mound foundation of breakwater or as backfill behind revetment. It is found from these studies that the geotechnical characteristics of the stone-muck are related to the production process of rubbles(rock materials), its geological background and the degree of weathering of rock mass. It is also proved that the strength characteristics are strongly related to uniformity coefficient, particle breakage under confining pressure and dilatancy at failure. It is concluded in this paper that if the stone-muck is placed on appropriate part according to its characteristics, it can be used effectively as construction materials for the port and harbor structures.
2 CLASSIFICATION OF STONE-MUCK
1 INTRODUCTION High quality rock materials, for instance, thoseproduced from fresh granite or andesite rock mass with high compressive strength, high unit weight and low water absorption, have been normally used for port and harbor structures. However, under recent trend of reconsideration of construction cost for public facility and of environmental preservation, stonemuck is re-valued as construction material especially for the port and harbor structures. Quite recently the stone-muck material of weathered sandstone was used as foundation mound of a part of revetment at deep seawater area in Yokohama port"). It was the first case showing an application of stone-muck material for port and harbor structures in Japan and its movement is still being monitored. Stone-muck remains in many quarries in Japan. However, since the usable amount, cost performance or geotechnical characteristics vary in each quarry, the piled condition, production method, and geological background were investigated. From such investigation, the stone-muck materials are ranked into three class taking into account its regular use.
2.1 Production process of rock materials Forty quarries in Tohoku region were investigated to study the production process of rock materials and piled condition of the stone-muck. The geological background in these areas are granite, andesite and sedimentary rock of sandstone and mudstone of Mesozoic and Paleozoic period. Figure 1 shows the production process of rock materials at a typical quarry in Japan. Each investigated quarry does not produce only large sized rock materials(rubbles), for instance those facing stone of mounds or backfill behind revetments, but also crushed materials for road construction or others, because hewed rock pieces should be used effectively according to its size or quality. As shown in this figure, after the surface soil and the highly weathered zone are excavated by excavator, the moderately weathered rock is hewed by drilling machine. The fresh rock mass is hewed by blasting and drilling, and hewed rock pieces are broken into adequate size by breakers. After the rock materials mainly for port and harbor structures are selected, fresh rock pieces are crushed by crusher plant. While these quality controlled materials are shipped to ap687
propriate construction sites, the excavated soil or the weathered rock pieces which failed quality tests for the rock material are dumped at stock yard in the quarries. These wasted rock pieces including the excavated soil are called “Stone-muck”.
2.2 Classification of the Stone-muck Since the quality of stone-muck left in quarries is not homogeneous, it is needs an appropriate classification for using it as construction material. Figure 2 shows the classifications of stone-muck materials based on the ground condition where they were hewed from. The regularly used rock materials are produced mainly from the fresh rock mass zone shown in Figure 1. Therefore, surface soil, highly to moderately weathered rock pieces and screenings of the fresh rock pieces remain as wastes in the quarries. As the surface soil mainly consists of clay, silt and volcanic ash, it cannot be used for marine structures because of water contamination by mud leakage. Then, the stone-muck taken from three different parts of the rock ground can be used and are ranked as three classes accordingly as follows (see Figure 1). -Class I : Good quality stone-muck(S-M), Mainly consists of fresh rock pieces -Class 11 : Fair quality S-M, Mainly consists of moderately weathered rock pieces -ClassIII: Poor quality S-M, Mainly consists of moderately weathered rock pieces including soils Table 1 Classifications of the stone-mucks and tests conducted Class No. (Geology)
Geological Background
Fresh tertiary sandstone
I (sed-t)
Weathered andesite
II
Fresh sandstone with surface soil
Physical Tests
Triaxial comp. test
A(6)
(an>
0
0
(sed)
0
no data
Fig. 2 Classification ________________________________________------------------------------------
of stone-muck(S-M) based on the rock ground condition
(Remarks: 0conducted on the collected samples,
a referred
to past studies by SDPCB,
an; andesite, sed-t; tertiary sedimentary rock, gr; granite)
688
3 GEOTECHNICAL CHARACTERISTICS OF STONE-MUCK 3.1 Material characteristics Table 1 indicates the studied stone-mucks and tests carried out on various samples. They are classified into appropriate class No. according to the above mentioned classifications. Figure 3 shows the relationship between dry specific gravity and water absorption of rock pieces in the stone-mucks. The data corresponding to sandstone and granite rubbles used for mound type breakwater are also plotted in this figure"). As shown in this figure, stone-muck pieces can be divided into two groups, A and B. The water absorption Q of group A in which I (sed-t) and II (an) are included is more than 10%. Q of group B in which I (an) , I (gr) and III (sed) are included is less than 7%. The water absorption can be an index of quality of rock pieces, because it indicates the bulk of void or presence of hair cracks. Therefore, the quality of the rock pieces of tertiary sandstone or weathered andesite is inferior than that of fresh andesite, fresh granite or fresh sandstone. Figure 4 show the grain size distribution curves of each type of stone-muck and rubble. Ranges of uniformity coefficient of each stone-muck type are Uc -4.15-37.4 for I (an), 14.9-354 for I (sed-t), 30.7300 for I (gr) , 620 for II (an) and 900 for Ill (sed) respectively. On the other hand, the uniformity coefficient of the rubble used as mound is around Uc = 4. Therefore, it seems that the stone-muck is better graded than the rubbles. The stone-muck with the highest Uc is ID bed) , second is II (an) and the lowest is I (an).
Fig. 4 Typical gradation curves of each type of stone-muck and rubble 3.2 Strength characteristics A sample of II (an) was chosen for the large scale triaxial compression test and related tests, because the strength characteristics of this class of stone-muck have not been obtained in past studies(see Table 1). Since the strength characteristics of granular material are related to its density, a series of maximum & minimum dry density tests were carried out prior to the triaxial compression test. The minimum and the maximum dry density of the stone muck sample were obtained by pouring statically in a mold and vibrating it on a vibration compactor respectively. Specimens for the relative density test were made by pouring underwater to take into consideration the fact that the materials are used as marine structures under seawater. Maximum & minimum dry densities of 1.711 and 1.295 g/cm3, dry density of the poured sample of 1.469 g/cm3 were obtained from the tests. Accordingly, a relative density of Dr =70.3% is calculated. Because of well-graded with high uniformity coefficient (Uc=620), relatively high density condition is given on IT (an) specimen in spite of the underwater pouring.
Fig. 3 Relationship of water absorption to dry specific gravity in rock pieces 689
A series of large scale triaxial compression test was carried out on the II (an) sample to study the strength characteristics of the stone-muck. The size of specimens was 30cm diameter and 60cm height. Grain size of the specimens were adjusted such that the maximum grain size is 53-63mm diameter and considering similarity in the grain size distribution. The specimens were made by underwater pouring with air dried samples. Conditions of the triaxial compression test were isotropic consolidation before shearing and drained condition during shear under constant confining pressures ranging firm 49-589 kN/m2. The strain rate during shear was O.l%/min and maximum axial strain was 15%. Internal friction angle a dobtained from the triaxial compression tests, including past studies by SDPCB, are as follows(see Table 1); - I (gr) - I (an) - I bed-t) - 11 (an) -
@d
Rubble of sandstone
- Rubble of granite
=41-42' 31 - 39 38:39' 37 26 - 30' 29 - 33-
As indicated above, it can be seen that the shear strength of the stone-mucks is relatively high compare to the rubbles in spite of the fact that the stonemuck is composed of poor quality rock pieces. This is caused by particle breakage and dilatancy as mentioned below. Figure 5 shows the relationship of internal friction angle with cohesion ignored to confining pressure obtained from the triaxial compression tests. 6"is the internal friction angle with cohesion ignored defined on each Mohr's stress circle at maximum compressive stress. As shown in this figure, 4, of the stone-mucks is not related to the confining pressure greater than 196 kN/m2. On the other hand, 6 o of the rubbles continues to decrease with increase in confining pressure. Therefore, @ of the rubbles is more strongly related to confining pressure than that of the stone-muck. Figure 6 shows the relationship of Marsal's particle breakage B, (7) to confining pressure obtained from the triaxial compression tests. It seems that the rate of increase in the particle breakage of the stonemucks is not much affected by increase in the confining pressure as compared to the rubble. This is explained from the point of view of the grain size distribution(gradati0n) as shown in Figure 7. Well- graded materials such as the stone-muck are relatively dense and have more contact points between particles. Poorly-graded materials such like as the rubble are loose and have less contact points. 690
Fig. 6
Relationship of particle breakage to confining pressure
Though the quality of each particle(rock pieces) of the stone-muck is poor, the particle breakage caused by the stress concentration does not occur much. Therefore, dilatancy contributes to the shear strength effectively on the stone-mucks and the shear strength of it is higher than the rubbles. It means that the strength of the stone-muck materials (or rock materials) is strongly related to its particle distribution. Figure 8 shows the relationship of internal friction angle to uniformity coefficient obtained from the triaxial compression tests. The internal friction angle is
strongly related to the uniformity coefficient and is distributed in the zone shown in this figure. A sample with relatively high coefficient of curvature Uc’=17.4 is out of this zone but if Uc is a high value, it is not affected by Uc’ as in a sample with Uc’=36.3.
/
Mound foundation
Class I1 in Box caissons Class I1 for Back fill
[IA\\ --
-
Class ;I
/A
As shown in the zone of b d distribution in Fig. 8, the range of b d tends to be large with smaller Uc value. This means that the strength of sample with poorly-graded is more affected by the strength of each particle. For example, the range of b d when Uc=2 is 7’ (25-32’ ) and the uniaxial compressive strength of rock pieces ranges from 0 c= 9.7 to 110.7 MN/m2.
Class I for Replacement
Class I for the core of Mound foundations
Fig. 9 Application of the stone-muck materials for a composite type breakwater Based on these results, it can be seen that the bettergraded (high Uc value) has more contribution to the strength( b J of the stone-muck than the quality of each rock piece. Furthermore, uniformity coefficient indicates the amount of contact points between particles and the internal friction angle can be determined assumed from the uniformity coefficient. 4 USE OF THE STONE-MUCK When the stone-muck is used for port and harbor structures, the quality of the rock pieces and strength characteristics of the stone-muck material should be taken into consideration. Figure 9 illustrates an example of an application of the stone-muck materials for a composite type breakwater. Since the material with high strength should be used for the core of the foundation mound to bear the load of the caisson and the backfill, and the forces from the wave, stonemuck of class- I (an) is applicable. Stone-muck of class- I (an) and D[ (an) can be used as the backfill behind the caisson wall, because they have enough strength to reduce the earth pressure. All kinds of the stone-muck whose density can be controlled is applicable as contents of the caisson boxes and the reclamation fill.
Fig. 7 Differences in physical characteristics related to particle composition (Gradation) Well-graded
Poor1y-graded
(Density) Dense (Point of contact) More (Stress Concentration) Large
Loose Small Small
5 CONCLUSIONS (1) Stone-muck can be classified into three classes based on the geological background of the rock mass and the production process of the rock materials. (2) Specific gravity and water absorption, which indicate the quaIity of the rock pieces in the stone-muck, are related to the classification of the stone muck. (3) Shear strength parameter @ in the stonemuck is more strongly related to the gradation than the quality of the individual rock piece. The better-graded stone-muck has the higher shear strength it has. (4) Shear strength parameter of the stone-muck can be determined from uniformity coefficient .
Fig. 8 Relationship of internal friction angle to uniformity coefficient
69 1
( 5 ) In using the stone-muck effectively for port and harbor structures, both the quality of the individual particle(rock pieces) and the strength characteristics should be considered. Especially the classifications and the gradation are important. Utilization of the stone-muck considering the above mentioned results should contribute to the effective and economical construction of port and harbor structures. However, proper construction method in placing the stone-muck under the seawater should be established. Finally, its gradation and strength characteristics after placing under seawater are also important subjects which need to be studied. REFERENCES (1) 2nd District Port Construction Bureau (1994) Report on Geotechnical Characteristics of Sand stone-muck as Construction Material for Proposed Revetment at Honmoku Area in Yokohama Port (in Japanese). (2) SDPCB (1994) Report on Geotechnical Characteristics of Stone-muck as Construction Material for Proposed No. 5 M o . 6 Berth in Onahama Port (in Japanese). (3) SDPCB (1997) Report on Geotechnical Characteristics of Stone-Muck as Construction Material for Proposed No.3 Berth in Soma Port (in Japanese) (4) SDPCB (1998) Report on Geotechnical Characteristics of Stone-muck as Construction Material for Proposed Shiogama Port (in Japanese). (5) Mizukami,J. and Kobayashi,M. (1991) Strength Characteristics of Rubble by Large Scale Triaxial Compression Test, Technical Note of The Port and Harbour Research Institute, No.699 (in Japanese).
(6)Shiomi,M.,Tobiki,I.,Kinouchi,M.,Mizukami,J., Fujiwara,Y.,Nakajima,H.,and Nakashima,K. (1997) Strength Characteristics of Sandstone-Muck by Large Scale Triaxial Compression Test, Proc.30th JNCSMFE, Kanazawa, Vol. 1, 294-295(in Japanese). (7) Hirsxhfeid,R.C. and Poulos,S.J.(1979) Embankment-Dam Engineering, Casagrande Volume, John Wiley&Sons,Inc., 130-132. (8) Shoji,Y. (1983) Study on Shearing Properties of Rubbles with Large Scale Triaxial Compression Test, Report of The Port and Harbour Research Institute, V01.22, No.4, 59-120 (in Japanese).
(9)Kobayashi,M.,Terashi,M.,Takahashi,K.,Nakashi ma,K. and Odani,H. (1987)A New Method for Calculating the Bearing Capacity of Rubble Mounds, Report of The Port and Harbour Research Institute, Vo1.26, N0.2~371-411(in Japanese).
692
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida(eds)02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Applications of fly ash as material for coastal works T.Okumura Faculty of Environmental Science and Technology, Okayama University,Japan
ABSTRACT: Fly ash is equivalent to silt size soil except that the grain density is small, and not suitable as construction material. However, reactivity of fly ash to cement is rather high, and hence treated fly ash, high strength and light weight, is applicable as materials for coastal works. 1 INTRODUCTION
(22) plants in Japan, and 12 times from two representative plants. Fresh clinker ashes were sampled 6 times from the same two plants. 76 samples in total from the above sampling were tested in the laboratory.
Population explosion and energy crisis on the earth lead us to a trend to utilize coal more than oil. Coal reserves are much longer, 220 years, than that of oil, 45 years, in the world. Therefore, constructions of electric power plants for coal are in higher pitch than for oil in Japan. Electricity production will become 44GW in 2007, being 22GW in 1997 (MITI statistics 1998). Percentage of coal power plants in whole electricity will become 16 % in 2007, being 10 % in 1997. Coal ash from all industries including electricity and iron was 7.3 million tons in 1997, and will be 12 million tons in 2006 in Japan (CCUJ statistics 1998). 68 % ( 5 million tons) of coal ash was effectively utilized mainly for cement in 1997, but 32 % (2.3 million tons) was disposed to reclaimed land. However, from the view point of global environment, the rest of 32 % should be utilized effectively to reduce reclamation of the sea. This paper reviews engineering properties of fresh and aged untreated coal ash and cement treated fly ash, and reviews applications of treated fly ash to coastal constructions.
(2) Physical properties of fresh fly ash Grain density of fly ash is 2.19kO.ll(standard deviation) (g/cm3) as a whole, which is much smaller than that of common soils. There is rather large difference (0.17) in the mean values of grain density between two representative plants. This difference is much larger than the standard deviation of each plant (0.08, 0.07). It means that there exists clear difference between two plants owing to kind of coal, method and temperature of burning, sampling place, etc. Coefficient of variation of grain density is 4.8 % in among all plants but smaller value of 3.5 % in time dependent variation at one plant. This tendency of variation is common in most properties as described later. The variation of grain density of fly ash is much larger than that of soils in coastal area of 1.9 %. Fine content of fly ash is 87 % in average, and silt content is 73 k 11 (“A).Therefore, it is classified into silt (M) of fine (F), according to the Japanese Unified Soil Classification System. Maximum density of fly ash, according to JGS T 161-1990, is 1.13 L 0.18 (g/cm3), and minimum density is 0.78k0.14 (g/cm3). Consistency tests clarified that fly ash is non-plastic.
2 ENGINEERING PROPERTIES AND THEIR VARIATIONS OF COAL ASH 2.1 Investigations onjkesh coal ashes (1) Samples For the purpose of recycling coal ash from thermal power plants as material for constructions, geotechnical properties and their variations were investigated (Takahashi et al. 1995; Okumura et a1 1996). Fresh fly ashes were sampled twice from all
( 3 ) Chemical Ch~acteristicsof fresh fly ash Chemical components of fly ash are si02: 63 2 6 (?h),Af203: 2 4 k 4 (?A), Fe203: 4.122.3 (%), CaO:
693
2.2 Investigations on agedjly ash
2.8k2.7 (%), MgO: 1.0k0.6 (“A), etc. Unburned component (carbon, etc.) is 3 . 2 k 2 . 3 (%) and pH is 10.9k2.2. (4) Mechanical properties of fresh fly ash Maximum dry density by compaction test is 1 . 2 5 k 0.13 (g/cm3) and optimum moisture content is 25.8 k 5 . 8 (“A). For the consolidation and triaxial test, fly ash was prepared to be the moisture content of 10 %, compacted statically (JGS T 812-1990) to be the relative density of 95 %, soaked in water for 28 days, and then supplied for the tests. Coefficient of consolidation of fly ash is 3.25 -f 1.73 (x103cm2/day), equivalent to that of silt. Compression index is 0.16 k 0.09. Coefficient of volume compressibility is 4.3 k 8 . 8 (~lO-~/kPa) at the pressure of 100 kPa and 3 . 4 2 1.9 (xlO-’/kPa) at 1 MPa. Coefficient of permeability is 8 . 9 k 6 . 8 (~lO-~cm/s) at 100 kPa and 1.92k2.73 (~lO-~cm/s) at 1 MPa. As a whole, consolidation characteristics of fly ash are equivalent to those of silt. Cohesion by the isotropically consolidated undrained triaxial test is 0.64k 1.12 (x 102kPa), and angle of internal friction is 19.9k6.0 ( ), and hence, triaxial compression characteristics of fly ash are equivalent to those of silt. ( 5 ) Engineering properties of fresh clinker ash Grain density of clinker ash is 2.26k0.12 (g/cm3), which is a little larger than that of fly ash. Average gradation of clinker ash is fine content 8 %, sand content 58 %, gravel content 34 %, and it is classified into sand (S). Uniformity coefficient is 14.1k 7.8, and is well graded. Maximum density of clinker ash is 0.83k 0 . 1 7 (g/cm3), and minimum density is 0.67 k 0.13 (g/cm3>. Chemical components of clinker ash are Si02: 6 1 2 5 (%), A1203: 2 1 k 5 (%), Fe203: 5.022.6 (%), CaO: 2.6&2.0(%), MgO: 1 . 0 2 0 . 5 (%), unburned component: 5.1 k 5.8 (%), pH: 9.5 k 1.3 (%), and not much different from those of fly ash. Fresh clinker ash prepared as in the case of consolidation test of fly ash was supplied for direct shear test. Initial void ratio is 2.06k0.51 (relative density 65 %). Shearing resistance of clinker ash does not increase proportionally with consolidation pressure, and angle of internal friction decreases with pressure. It will be effect of grain crushing of clinker ash tested. At the consolidation pressure of 100 kPa, cohesion is 2.06k0.51 (x102kPa) and angle of internal friction is 46.6k8.5 ( ).
Two sites of fly ash deposits were investigated (Takahashi et al., 1996a). Summary of the results are as follows. Grain density ranges from 2.1 to 2.3 (g/cm3). Bulk density is about 1.5 (g/cm3), equivalent to marine clay layers. Moisture content of the deposits is 50 to 100 %. Grains are silt size predominant and non-plastic. N values by Standard Penetration Test is less than 1 very often below the ground water level. However, static cone penetration tests show qlof 1 to 2 MPa, suggesting that static strength of the deposit is not so small. Isotropically consolidated undrained triaxial comprssion test of undisturbed samples from the deposit showed the following strength as the lower limit,
Compression index C, ranges from 0.2 to 0.8. Coefficient of consolidation C, ranges from 1,000 to 10,000 (cm2/day). Liquefaction resistance of the deposit is equivalent to that of Toyoura Sand of relative density of 75 %, which means that liquefaction resistance of aged fly ash is relatively low. 2.3 Investigations on cement reactivity offleshfly ash Okumura and Ohhashi (1996) and Okumura and Kai (1997) analized the correlations between the unconfined compressive strength, q,, (x 1OOkPa), of cement treated fly ash with several factors, and selected 3 major factors as follows, Fly ash from electric industry (3 15 data, R2=0.689) log qu=0.O92C+0.43110gTC-0.26110g~+0.06 1 (2) Fly ash from general industry (270 data, R2=0.868) log q,=0.092 C +0.32910gTc-1.5491ogw+2.410 (3) Fly ash from general industry (52 data, R2=0.858) by fluidized bed boiler log qu=O. 049Ca+O.28 810g T C - 1.93710g~t-4.03 5 (4) where C: cement content (“A), Ca: CaO content (“A), Tc:curing time (day), w :moisture content (%). Figure 1 shows tested data vs. calculated by equations (2) to (4).
694
Figure 1(c) Test data vs. calculated strength (general industry by fluidized bed boiler) Making use of these advantages, cement treated fly ash can be used as material for such civil structures as embankment, land fill, back filling of quay wall and revetment, etc. Sand size grains made from solidified fly ash by cement are utilized instead of sand as materials for sand drains, sand compaction piles, etc. Fly ash is also used as sub-material for solidifying soft clayey soils in Deep Mixing Method, where the strength of mixture of clay, cement and fly ash is not too high nor too low, making it easy to excavate the treated ground. 4 APPLICATIONS FOR COASTAL WORKS
Figure l(b) Test data vs. calculated strength (general industry) Reactivity of fly ash to lime is generally lower than that to cement (Okumura et al., 1998; Okumura et al., 1999). 3 GENERAL APPLICATIONS
As described in section 2, fly ash is equivalent to silt, and not suitable as material for embankment or fill. Therefore, it is necessary for fly ash to treat with cement or lime. In this case the material has the merit that it is lighter than the treated clay or silt because of its small grain density.
Several test projects and practical applications of utilization of cement treated fly ash have been commenced in coastal area (Takahashi, 1996). Cement treated fly ash is used as materials for subgrade of pavement (Takahashi et al., 1996b). Cement treated fly ash slurry is utilized for surface covering of very soft clay fill (Takahashi et al., 1996c) and back filling of revetment (Shoji et al., 1999). In these applications the strength of treated fly ash is 500 to 1,000 kPa, much better quality than sand but lower cost. Therefore, treated fly ash will be more widely used in coastal area. 5 CONCLUDING REMARKS In order to increase the ratio of effective usage of fly ash from electric and other industries, cement reactivity and its variation of fly ash, rational
695
method of design and execution of cement treatment and filling of fly ash should be studied more. Diffusion of toxic substances from the treated fly ash to the surrounding environment should be also cared. Especially the influence of high pH and sometimes suspended solid become problematic during and after construction works. REFERENCES Okumura, T. & T. Ohhashi 1996. Factors affecting chemical reactions of coal ash and cement, Proc. 3 1' Annual Meeting of Japanese Geotechnical Society, pp. 925-926 (in Japanese) Okumura, T., K. Takahashi, M. Tezuka, Y. Umehara & M. Kobayashi 1996. Geotechnical properties of fresh coal ashes from thermal power plants, Proc. 2"d Int. Cong. Environmental Geotechnics, Vol. 2,pp. 869-874 Okumura, T. & H. Kai 1997. Effect of cement and lime treatment on coal ash, Proc. 52"d Annual Meeting of Japan Society of Civil Engineers, Vol. 3-B, pp.436-437 (in Japanese) Okumura, T., H. Kai & R. Watabe 1998. Experimental study on the effect of cement and lime treatment of coal ash, Proc. 9h Annual Meeting of Japanese Society for Waste, Vol. 2,pp. 794-795 (in Japanese) Okumura, T., K. Ueda, H. Kai & A. Yasui 1999. Experimental study on stabilization of soft soil with stabilizer and coal ash, Proc. 34' Annual Meeting of Japanese Geotechnical Society, pp. 2 11-2 12 (in Japanese) Shoji, Y., K. Takahashi, T. Asai & T. Sumino 1999. Utilization of fly-ash cementmixture for backfills of quay, Jour. Const.Manage. & Eng., JSCE, No. 637, pp. 137-148 (in Japanese) Takahashi, K. 1996. Coal ash utilization for port and harbor construction, Int. Clean Coal Tech. Symp. on Coal Ash Utilization, pp. 69-80 Takahashi, K., Y. Umehara, M. Tezuka, T. Okumura & S. Ono 1995. Soil engineering properties of coal ash from thermal power plant (1) - Physical and chemical characteristics of fresh coal ash, 30' Annual Meeting of Japanese Geotechnical Society, pp. 757-760 (in Japanese) Takahashi, K., K. Suzuki, K. Ozasa, T. Okumura & M. Shima 1996a. Geotechnical properties of aged fly ash in reclaimed land of thermalpower plant, Proc. 2"d Int. Cong. Environmental Geotechnics, Vol. 2, pp.679-684
696
Takahashi, K., K. Ozasa, H. Miyagawa, T. Okumura & K. Suzuki 1996b. Strength properties of subgrade made of mixture of fly ash, sand and cement, Proc. 31' Annual Meeting of Japanese Geotechnical Society, pp. 237 1-2372 (in Japanese) Takahashi, K., K. Ozasa, H. Imashio, T. Okumura, K. Suzuki & H. Asada 1996c. A test project of surface treatment with cement added fly ash slurry, Proc. Symp. on Cement Treated Soil, pp. 179-186 (in Japanese)
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
New ground material made of dredged soil for port and a q o r t reclamation projects T.Okumura Faculty of Environmental Science and Technology,Okayama University,Japan
S.Noda, S. Kitazawa & K.Wada Coastal Development Institute of Technology,Tokyo,Japan
ABSTRACT A large volume of quality soil is needed for the construction of ports, harbors, and also for the reclaimed land to build airports and industrial facilities. In addition, the disposal of dredged soil from rivers and harbors, as well as construction wastes, is a serious problem. Therefore, to responding to these needs a new ground material is developed by mixing dredged soil or sludge with cement to form stable ground material. The method involves adding water to the dredged soil or construction soil to transform it to slurry. Then by adding a hardener (such as Portland cement) and a lightening agent (such as bubble or beads), the final product can be used as fill material for harbor and airport reclamation work or as a backfill for quay walls. The treated soil reduces settlement of soft ground and lowers earth pressure behind the quay walls. This paper presents considerations for design and construction of the new ground material to be used as a guide for engineers involved in harbor and airport construction, coupled with an example of its application to a practical project. 1 INTRODUCTION An average of Y30 billion was spent on the
construction of bulkheads used to store the wastes every year. Finding a suitable disposal sites has become a serious problem due to shortage of land. Figure 1 shows the capacity and breakdown of the wastes stored in these bulkheads. This figure reveals that between 1986 and 1995, approximately 18 million m3 of the wastes were disposed every year. 45% of these was dredged soil produced by harbor work, 21% was surplus soil from project sites in the cities, and 8% was the industrial wastes. In addition, harbor construction projects also use large quantities of good quality soil, as fill material, as backfill for quay walls, and also to improve soft ground. Therefore, finding methods to treat wastes to use as alternate construction material would be extremely effective in both extending the lifetime of the costly waste disposal sites and recycling limited resources. Figure 2 shows changes in the prices of sand, crushed stone, and pit sand per cubic meter and also the price of ordinary Portland cement per ton since 1965 in the Tokyo metropolitan area. The figure indicates that while the price of soil has been
697
increased steadily, the price of cement has fallen since 1980. This means that the gap between the price of ordinary fill and cement treated soil has been reduced sharply. In response to these circumstances, in recent years, several research projects have been carried out to develop the technology needed to process dredged soil or construction wastes to recycle it as highquality ground material. For example, the Bureau of Ports and Harbors and the Port and Harbor Research Institute of the Ministry of Transport and private research institutes have been involved in developing a new ground material known as Super Geo Material (SGM) since 1992. The SGM that is a form of lightweight treated soil was developed jointly by the Port and Harbor Research Institute, 23 private companies, and the Coastal Development Institute of Technology. Based on the results of their work, fullscale utilization of the method has been carried out at the Kobe port and at the Tokyo International Airport, among others. This paper presents the considerations for design and construction of the new ground material to be used as a guide for engineers involved in the design and construction of harbors and airports and also to promote the use of this technology.
soil has fluidity immediately after it is mixed, but as it reacts with the hardener, it is transformed into a hardened soil. This improved material has a strength equal to or exceeding that of soil typically used for such projects. It is possible to adjust the mix proportion to control the strength and density of the lightweight treated soil. The wet density can be adjusted between 1.0 to 1.2 gf/cm3 and the shear strength from 1.0 to 2.0 kgf/cm2. 2.3 Characteristics of the technique The main characteristics of the lightweight treated soil method are as follows: 1. Unlike natural ground material, lightweight treated soil is homogenous with density and strength that can adjusted to appropriate values. Therefore, it is an effective way to reduce the consolidation settlement of the ground. 2. Because its mix proportion can be designed so that it will not be separated under water, it can be used with little impact on the surrounding ocean water environment. 3. Since it has good fluidity, it can be poured in any desired shape by pump feeding without the need for compaction. 4. This method is an effective way to directly use dredged soil which has a high moisture content. Overview - Application range - Properties of treated soil - Properties and uses of construction method
Figure 2. Changes in the Prices of Fill Materials and Cement in the Tokyo Metropolitan Area
Design - Design procedure and soil constants - Load, external force, and earth pressure - Bearing capacity and slope stability - Liquefaction and consolidation settlement
2 OUTLINE OF THE TECHNIQUE
Mix Proportion Air bubble mix - Styrene beads mix -
2.1 Scope of the application This technique applies to two types of lightweight treated soil: (a) "bubb1e"treated soil and (b)"beads" treated soil. Bubble-treated soil is made by mixing air bubbles, and beads-treated soil is made by mixing expanded polystyrene beads of 1 to 3 mm in diameter with soil and cement in a slurry form.
Execution - Method and control - Confirmation testing
I
Annexed Documents - Properties of light-weight Submerged separation resistance properties test
2.2 Properties of the lightweight treated soil
-
The lightweight treated soil has a density between 0.6 and 1.5 g/cm3 made by mixing seawater, hardener (Portland cement), and a lightening agent (air or styrene beads) with dredged materials. The 698
I
- Sample design calculation - Sample mix proportion test -
Sample application
________
Figure 3. Configuration of the Manual
(1) Use as landfill or backfill behind a newly constructed quay wall
Figure 4-5. Additional embankment
Figure 4-1. Gravity type quay wall
(4) Use as submerged backfill and embanking on soft ground
Figure 4-2. Sheet pile quay wall
Figure 4-6. Submerged backfill
(2) Use to retrofit or improve functions of an existing quay wall
Figure 4-7. Embanking work on soft ground
Figures 4. Examples of Applications Figure 4-3. Seismic retrofitting, raising, or increasing the overburden load on a quay wall
2.4Applications of the method The lightweight treated soil method is used to reduce unequal settlement and earth pressure, and to restrict lateral displacement. Figure 4 shows various applications.
2.5 Design procedure Figure 5 outlines the procedure to be used for design of lightweight treated soil. A detailed study of using lightweight treated soil demonstrates the inadequacy of conventional formula of the lateral backfill pressure based on the semi-infinite theory, which is not applicable for the case of a finite backfill. This effect should be taken into consideration while calculating the active earth pressure of the treated ground.
Figure 4-4. Deepening a seawall
(3) Use for embanking of adjacent work
699
1
Preliminary site investigation and testing of raw materials Testing the mix proportions of lightweight treated soil
1
Setting the strength and density of the lightweight treated soil
3 Setting the range of ground improvement with lightweight treated soil
+
Testing the stability of facility including lightweight treated soil [11 Load and external force [2]Earth pressure [3] Bearing capacity [4] Slope stability [S] Liquefaction of nearby ground 161 Settlement
Determining the strength, density, and area of the improved ground
Figure 5. Design Procedure (Dredging )
The dredging and transporting work involve obtaining the raw material that will be the base material for lightweight treated soil and transporting it to the construction site. Methods of transporting the material are categorized as continuous transport by belt conveyor or pipeline, and as discontinuous transport by dump truck or soil barge. When the raw material is dredged soil, it is common to transport it by a barge. When soil is from the construction sites, it is often transported by belt conveyors or dump trucks. To prepare the slurry water is added to the raw material while agitating by a vibrating sieve to remove the extraneous matters. Moisture content and density are also adjusted to the required levels. The slurry is fed into a storage tank before feeding to the kneading mixer. Mixing is performed by adding the lightening agent (air bubbles or styrene beads) and the hardener to the adjusted slurry. Placing refers to a process that includes feeding the material with a pump and placing it using a tremie pipe. Depending on the placing site, it is categorized as either submerged placing or atmospheric placing.
3 APPLICATION FOR THE EARTHQUAKE RESTORATION IN KOBE PORT
(Transporting)
(Mixing) Beads treated soil
(Placing)
(Curing) Figure 6. Execution Procedure
2.6 Outline of the execution Lightweight treated soil is manufactured based on the following steps: dredging, transporting, mixing, placing, and curing, as shown in Figure 6.
The full-scale application of lightweight treated soil was the use of 21,610 m3 of bubble treated soil as the fill material behind a quay wall of 7.5 m high in the Port Island that was damaged due to the Hanshin-Awaji Earthquake disaster. At the time of the earthquake, this 180 m long facility was under construction, i.e., the caissons had been installed and backfill was completed. The earthquake displaced the normal line of the quay wall between 0.8 and 3.8 m to the ocean side. It also lowered the top of the caissons between 1.2 and 2.5 m, but the rotation of the main vertical axis was only between 0" and 3" and the face line of the quay wall remained relatively linear after the earthquake. Since there was an urgent need to complete the restoration work promptly, and it was necessary to align it with the face line of adjacent quay walls, it was decided that the best method would be to restore it using the existing caissons, that had been shifted forward by the earthquake. This method however, would require measures to deal with the increase in lateral earth pressure that is caused by the increase of the design seismic coefficient (from 0.18 before the earthquake to 0.20 after the earthquake). For this reason, it was decided to increase the passive resistance of foundation riprap on the front of the 700
caissons and at the same time to reduce the earth pressure during an earthquake by replacing part of the ground behind the quay wall with lightweight treated soil. As a result, design and execution problems were resolved and, as shown in Figure 7, it was possible to reduce the project costs to less than 1/3 of the cost of other alternate methods. Bubble treated soil was used as the lightweight treated soil for this project. The design wet density was 1.0 t/m3 above water and 1.2 t/m3 for the submerged condition. The design shear strength was set at 1.0 kgf/cm2 in order to obtain the necessary bearing capacity, because apron paving was placed above it and a container crane foundation was supposed to be constructed, as part of the project. The raw material used to prepare the bubble treated soil was dredged clay obtained from channel dredging work performed inside the Kobe port at the time of construction. It was highly plastic soil with a liquid limit of 97%. The dredged material was mixed with seawater and the hardener was Grade B Portland blast furnace cement. The bubbles used were formed by diluting it by 10 times with a foaming agent whose main constituent was a synthetic surface active agent, and foamed 25 times before use. The target unconfined compressive strength after 28 days of curing was set at 6.0 kgf/cm2 by the laboratory test. That was 3 times the design strength. The mix proportion was 140 kg of cement and 279 (in the atmosphere) and 196 (submerged) of bubbles, respectively, per cubic meter of bubble treated soil. On-site pouring tests had demonstrated that part of the bubbles de-foamed during the mixing, feeding, and pouring stages. Therefore, the quantities of bubbles specified above were increased by about 15% and 20% for the actual construction work.
Figure 8. Wet Density of Samples Taken from Cores
Figure 8 shows the wet densities obtained from the core samples taken 1 to 10 months after execution. In the atmosphere, the density was about 0.03 tf/m3 higher than the target value, but this was a result of contraction accompanying its initial hardening. In parts placed underwater, however, the values were a little lower than the target values. This result shows overall that the weight reduction required by the design was satisfactory. In addition to the application for the Kobe port, Table 1 shows major applications of the lightweight treated soil method for different projects.
4 CONCLUDING REMARKS Lightweight treated soil is an alternate fill material for use in the construction of ports and airports. Lightweight fill is effective since the facilities constructed along the coastlines are usually on soft and unstable ground that prone to settlement and lateral deformation and inadequate bearing capacity.
Figure 7. Cost Comparison of Different Construction Methods
70 1
Table 1. Major Applications of the Lightweight Treated Soil Method Project Quay wall (- 10 m) in Port of Fushiki-Toyama
I
Quay wall (- 7.5m) in Port Island, Port of Kobe Seawall and Ground improvement in Tokyo International Airport Quay wall in Ishikari Bay New Port Ground improvement in Oi wharf, Port of Tokvo Quay wall (- 7.5 m) in Port of Yokohama
1
Execution Period
1992’11 -93.1 1995.12 -96.4 1996.5 -99.11 1998.11 -99.11 1999.10 -2000.4
1
Quantity Placed (m3) 900
Special Features
I
21’610
847610 7,110
1
2000.5-
11,200
70,000
In addition to the advantage of being light that reduces the dead load, the following three characteristics make lightweight treated soil attractive to be used along the coastlines: 1. It can be applied above water or underwater 2. It can be applied economically in large quantities in a short period of time, 3. It can be used to improve seismic resistance and resistance to wave action and to reduce settlement . The lightweight treated soil method was prepared based on the results of the abovementioned applications, and is expected to promote the use of the lightweight treated soil in future port and airport construction projects.
ACKNOWLEDGMENTS The authors would like to express their sincere gratitude to the members of the Technical Committee of the Port and Airport Lightweight treated soil method for their deliberations and collaborative research work.
REFERENCE Coastal Development Institute of Technplogy. 1999. Technical Mannual of Light-weight Treated Soil Method in Port and Airport Projects. Tsuchida, T., Ali Porbaha and Kishida, T. 1999. Restoration of A Wharf Structure, Journal of Performance of Constructed Facilities. ASCE, Vol.l3,No.l, pp.1-6. Tsuchida,T. 1999. Development and Use of Foamed Treated Soil in Port and Airport Projects. Report of the Port and Harbour Research Institute, Vo1.38, No.2, pp.131-167. (in Japanese)
702
1
FIRST APPLICATION IN JAPAN First large-scale application in Japan Use of dredged soil Used of sandy soil on site. Applied in air. Feeding pipe length 150 and 300m Use of mixture of soil on site and bentonite Winter application in cold region Submerged application at - 10 m Use of dredged organic soil Use of protein active agent
Tsuchida,T., Okumura,T. Takeuchi, D. and Kishida,T. 1996. Use of Artificial Light-weight Materials for Backfilling of Quaywall. 2nd International Conference on Soft Soil Engineering,Vol.2,pp.807-812,Nangyn. Tsuchida,T., Okumura,T. Takeuchi, D. and Kishida,T. 1996. Development of Light-weight Fill from Dredgings. 2nd International Congress on Environmental Geotechnics, Vol.1, pp.415- 420,Osaka. Wako, T., Tsuchida,T., Matsunaga,Y., Hamamoto K, Kishida,T. and Fukasawa,T.1998. Use of Artificial Light Weight Materials for Port Facilities. Journal of J.S.C.E.,No.602,VI-40, pp.35-52. (in Japanese).
I
Coastal Geotechnical Engineering in Practice, Nakase & Tsuchida (eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Estimating the stochastic variability of geomaterials Ali Porbaha & Takao Kishida TechnicalResearch Institute, TOA Corporation, Yokohama,Japan
Takashi Tsuchida Port and Harbour Research Institute, Ministry of Transport, Yolwsuka,Japan
ABSTRACT: This paper presents a series of techniques employed to assess the stochastic variability of geomaterials. The analyses include: estimation of spatial continuity, modeling of the variogram, calculation of error variance, incorporation of an unbiased interpolation technique, and finally the testing of cross validation. A summary of some relevant case histories of the projects is presented.
1 INTRODUCTION
geomaterial. In the methodology presented here, the spatial continuity of the geomaterial is modeled and the error variance is minimized in an interpolation technique to provide the means for making the best linear unbiased estimates of the property. A summary of the techniques applied in this process is described here.
Development of new geomaterials are gaining popularity for a variety of reasons. The main rationales for these developments are, for instance, the need for supply of quality construction materials in the highly developed urban and/or coastal areas; implementation of the enacted environmental ordinances to use low quality or recycled materials in the construction industry; development of engineered materials with controlled properties, and so alike. According, it is necessary to develop procedures for quality control and characterization of these geomaterials. The techniques presented here employs the results of cone penetration tests (CPT) in the analyses. Although any cone may be used, the FRICON (Frictionless cone) has been mainly specifically developed for this purpose. Since most geomaterials have an artificial cementing reagent in which penetration by the conventional CPT may pose difficulties due to heavy load required during the test. Usually limited data is available from a project site due to economic reasons. Therefore, a series of analyses is necessary to extend the results of limited measured data to the whole continuum of the geomaterials. Since the properties of the geomaterial vary on the micro level, the properties must be viewed as random variables, with the value of a property assumed stationary over the treated zone. In a broad perspective spatial continuity provides the tools for describing the stochastic structure of a random variable, such as a property of the ground treated with a
2 ANALYSIS OF SPATIAL CONTINUITY The spatial continuity is usually described in terms of either a variogram or a covariance function, as schematically shown in Figure 1. The property of a geomaterial at any location x along the depth will be denoted as z(x). The same property has a value of z(x+h) at a distance h from the initial point measured at x. For relatively small separation distances, the values z(x) and z(x+h) will probably be autocorrelated; for large separation distances, the autocorrelation will be zero, i.e., the values z(x) and z(x+h) will be independent. For very small h, the autocorrelation should be large, and it should decrease to zero as h increases.
(a)
h
(b)
h
Figure 1: Spatial continuity (a) Variogram, (b) Covariance function 703
At some point, the value z(x) will be independent of z(x+h); this point is called the range of influence and is denoted as r. Of interest in assessing the stochastic structure of a property is the variability between the two values separated by distance h. The variogram, which is denoted as 2y(h), characterizes the variability of the property z between the two points:
1 " 2 y (h) = - [z( xi)( I xi + h)I2 n ;=I
C
in which n is the number of measurements made at separation distance h; and xi is the location of a point with respect to some reference point. Eq. 1 has the form of the expected value and is actually the expected value of the random variable [z(x) - z(x
+WJ2
Figure 2. Several variogram models: (a) Spherical model, (b) Exponential model, (c) Gaussian model (d) Hole effect model (e) Linear model, and (f) Quadratic model
(2 1
2y(h) = E[z(x) - z(x +h)]'
In order to quantie the variogram, realizations of the property must be available. In application of Eqs. 1 and 2, we assume that the intrinsic hypothesis is valid; this hypothesis states that the value of the variogram depends only on the separation distance h and not the location x of the sample. points. In other words, Eqs. 1 and 2 assume that the difference z(x) z(x + h) is a random variable with second-order stationarity. Eqs. 1 and 2 define the variogram. Dividing these values by 2 yields the semivariogram y(h). The semivariogram is used later on for the estimation problem with the interpolation technique. 3 VARIOGRAM MODELING
The spatial continuity should be represented by a model. Just as probability functions are fit using sample data that may be presented as a histogram, a sample semivariogram computed with Eq. 1 can be used to fit a semivariogram function or model. The most frequently used semivariogram model is called a spherical model and has the form:
YO= Yr Figure 4: Effect of anisotropy described as rose diagram
when
h>r
(3b)
in which r is the range of influence; and yr is a semivariogram model parameter called the sill. yr is often 704
quantified using the variance of the sample measurements z(x). The spherical model is just one of many models used to represent a semivariogram; it is widely used because its properties are easily computed and it has the shape and scale properties that characterize many data measurements. Figure 2 shows several other models, and Figure 3 illustrates a typical analysis with spherical model for a geomaterial. The calculation of omnidirectional variogram does not imply that the spatial continuity is the same in all directions. Therefore, one should proceed to explore the pattern of anisotropy with various directional variograms (if possible). One practical approach is to calculate several directional variograms and plot a rose diagram (see Figure 4) described as a function of direction. The major and minor axes of the ellipse represent the axes of anisotropy.
4 ERROR VARIANCE The ultimate objective of the analysis is to provide a means of estimating the property of the geomaterial at any point x along the depth. In addition to the best estimate of the property, we must also be interested in the accuracy of the estimate. If we have a value of the property z(x) measured at a single point x along the depth, then assuming other information is not available, our best estimate of the property at a point (x+h) is z(x). The variogram defines the accuracy of the estimate. That is, if we have a single point estimate of the property z(x) at a point, then our best estimate of the property at any other point (x+h) is z(x) and the accuracy of z(x+h) is the error variance 2y(h). The standard error of estimate S, would be the square root of the error variance. If instead of a single point sample, we collect a sample of n measurements along the depth of the treated zone, then our best estimate of the property would be a weighted mean value of the individual points:
To develop an expression for the error variance when the sample consists of n measurements, with each sample point having a weight w,, both the error variance associated with each sample point and the point to be estimated and the error variance among the sample points must be assessed. The first source of the error variance would be the weighted average variogram value between sample point i and the point to be estimated, 2Cw,y(h,), where h, is the separation distance between sample point i and the point to be estimated. As the sample size increases, the first part of the error variance will decrease because of the greater level of confidence associated with larger samples. Therefore, the within sample variation must be subtracted from the point sample variation because it reflects variation that is not part of the total error variation. The within sample variation is the weighted average semivariogram value between each point in the sample. Therefore, the error variance, S,, is given by:
in which hi, is the separation distance between sample points i and j. The second term on the right-hand side of Eq. 5 is the variance that is internal to the sample points.
(4) in which wi is a weight for Z(x,) that reflects the importance of measurement Z(xJ. The error variance of Z is no longer 2y(h) because the larger sample size, i.e., n rather than 1, should be expected to reduce the error variance. The reduction in the error variance depends on the number of points in the sample and the relative independence of the sample points. 705
5 INTERPOLATION There are numerous interpolation procedures for modeling spatial data, including triangulation, moving averages, trend surface analysis, objective analysis, spline, and kriging. Analysis of spatial continuity is intended to be used as part of the estimation process when kriging is applied. However, estimation requires us to decide which of the sample points to use for estimation and what weight should be given to each sample point. Given that the data analysis yields values for the range of influence and sill for Eq. 3, it seems reasonable that only sample points located within the range of influence of the unknown point should be used to make estimates with Eq. 4, and the weight given to each sample point should be inversely proportional to the ordinate of the semivariogram corresponding to the distance separating the sample point and the unknown point. To formulate a solution, we need to satisfy several requirements for modeling: ( I ) an estimation model, which is given by Eq. 6; (2) an objective function that defines the best fit; ( 3 ) constraints, when neces-
sary, that place limitations on the solution; and (4) a data base. The data base consists of the sample points Z(xi), which are used with the linear estimation model of Eq. 4. The best model is often taken to imply that the error, or estimation, variance is a minimum. Thus, we have as the objective to minimize the error variance. But for the kriging solution, if we want an unbiased model, we must impose the constraint that the sum of the weights, wi, equals 1:
;=I
The estimation variance, which will be denoted as S, for sample (similar to oe2for the population), depends on the values of the unknown weights, the structure and magnitude of the semivariogram, the location and magnitude of the sample points, and the type of estimation to be made (i.e., point, core length, or volumetric). We can minimize the error variance by taking derivatives of the objective function with respect to each unknown and setting the derivatives equal to zero; this provides a set of n equations with n unknowns. While the solution of these n normal equations would produce a minimum error variance, the resulting model would not be unbiased. For this, the constraint of Eq. 6 must be included in the system of normal equations. Thus the objective function is to minimize:
in which h is an unknown. It should be apparent that the solution procedure for this interpolation technique is an example of Lagrangian optimization, with h being the Lagrangian multiplier. There are n+l unknowns (i.e., the n values of w; and A), and there are nt-1 equations (i.e., Eq. 6 and the n derivatives of Eq. 7 with respect to each w;).Thus, we only need an expression for estimating oe2to find the solution. To obtain a solution to the estimation of the value at a point, whether the sample points are distributed linearly in either space or time, an estimate of the error variance can be made by:
in which S; is the ith sample, element; Y indicates the value of the criterion variable (e.g., the variation 706
of strength where the estimate is needed), and yax,&CI,C,)is the average semivariogram value between all combinations of C, and C,, where C, and Cj are dummy variables and may be either S, or Y . In Eq. 8, S, and Y would represent the ith measurement for the sample and the unknown value of the strength at any point. The summations of Eq. 8 include only the n sample points within the range of influence since for the points beyond the range of influence w, is zero. Eq. 8 indicates that the error variance consists of three parts. The first term represents the variation associated with differences between the sample point measurement S, and the criterion Y for which a value is needed. The second term reflects the variation within the sample; that is, the average semivariogram value for all elements of the sample. The third term, i.e., ya,&Y, Y ) , reflects variation that is not error variation, so it must be subtracted from the total expected variation between the sample and the unknown value of the criterion. The third term is similar to the second term in that it represents variation that is not error variation, yet it contributes to the total variation between the sample elements and the unknown value of the criterion.
Difference in estimate
Figure 5: An example of cross validation analysis (vertical axis represents the Krigged estimate) For a system in which there is a single point of interest, the average semivariogram value for a separation distance of zero must also be zero. The subtraction of the two terms indicates that we must reduce the error variation because we are interested in a mean value (i.e., the mean of all estimates). Having formulated the objective function (Eq. 8) the optimal values of the w, and h can be obtained by Lagrangian optimization, as follows. The normal equations are obtained by algebraic manipulation:
(9b)
As an example, if the sample consists of three points, Eqs. 9a-d reduce to :
Figure 6: Analysis of the geomaterials for several sites in Fushiki (top), Ishinomaki (middle), and Kobe (bottom) 707
REFERENCES
Eqs. 9a-d represent a set of n+l simultaneous equations with n+l unknowns, which can be solved either analytically or numerically. The solution provides the weights that yield the minimum error variance as defined by Eq. 8.
6 CROSS VALIDATION In a cross validation exercise, the estimation method is tested at the locations of existing samples. The sample value at a particular location is temporarily discarded from the sample data set; the value at the same location is then estimated using the remaining samples. Once the estimate is calculated we can compare it to the true sample value that was initially removed from the sample data set. This procedure is repeated for all available samples. The resulting true and estimated values can then be compared. Figure 5 shows a typical output of the cross validation analysis 7 CASE HISTORIES The concepts described in the preceding sections have been applied for several projects. The results are shown in Figure 6, however, due to page limitations, the details of the analyses will be presented separately. The estimated value of the desired property at any point (i.e., distribution of strength with depth for these projects) may be used to delineate the isostrength contours. Obviously, the accuracy of the results depend on the estimated values at various stages of the analysis. However, the main contribution is derived from modeling the spatial continuity of the data that has a significant impact on the accuracy of the results.
8 SUMMARY There are several elements in establishing the guidelines for estimation of variability in geomateria1 modeling. The methodology presented here is based on the consideration of spatial continuity of the geomaterial and minimization of the error variance. Further references may be consulted for specific details in various staEes of the analvses.
708
Bames M. G. (1 978) Statistical design and analysis in the cleanup of environmental radionuclide contamination, Desert Research Institute, University of Nevada System, W O 1253-12. Bames, M. G. (1 980) The use of Kriging for estimating the spatial distribution of radionuclides and other spatial phenomena, Trans-Stat, Battelle Memorial Institute, Pacific Northwest Labs, Richland, WA. Clark T. L., Dennis R. L., Voldner E. D., Olson M. P., Seilkop S, and Alvo M. (1 986) The international sulfur deposition model evaluation. Preprints from Fifth Joint Conference on Applications of Air Pollution Meteorology. I 8-2 1 November 1986, pp. 57-60. Durham J., Dennis R., Laulainen N., Renne D., Pennell B., Barchett R, and Hales J. (1 986) Regional Eulerian model field study and evaluations. ASRL Rep., available from Office of Research and Development, U.S. EPA, Research Triangle Park, NC. Endlich R. E., Ferek R., Eynon B. P., Maxwell C, and Valdes A. D. ( 1 986) Statistical analysis of precipitation chemistry measurements over the eastern United States. EPRI Rep No. VAPSP 1 12. Owosina, A., Lall, U., Sanyogomi, T., and Bosorth, K. ( I 992) Methods for assessing the space and time varaiability of groundwater data, Utah Research Laboratory, State of Utah, USA. McBratney, A.B., and Webster, R. ( I 986) Choosing function for semi-variograms of soil properties and fitting them to the sampling estimates, J. of Soils Science, 37, 61 7-628. Wahba, G. (1 979) How to smooth curves and surfaces with splines and cross validation, Technical Report No. 555, Department of Statistics, University of Wisconsin. Yates, S.R., Warrick, A.W., and Myers, D.E. (1986) Disjunctive kriging, I. overview of estimation and conditional probability, Water Resources, Research, 22,615-623.
Coastal GeotechnicalEngineering in Practice, Nakase & Tsuchida(eds) 02000 Balkema, Rotterdam, ISBN 90 5809 151 1
Underwater casting test of light-weight treated soil made of waste soils T. Satoh, N.Ueno, K. Mitsukuri & K. Kawano The 4th District Port ConstructionBureau, Ministry of Transport,Shimonoseki,Japan
TTsuchida Port and Harbour Research Institute, Ministry of Transport, Yokosuka,Japan
ABSTRACT: Kumamoto Port is located in Kumamoto Prefecture, Kyusyu Island of Japan. The ground condition is poor with the ultra-soft clays of about 40m. The difference in tidal level is large and the water is shallow. New technological developments are required for economical and efficient port construction in such conditions. Utilizing light-weight treated soil has been turned out to be an effective work for quay wall constructions. The construction works utilizing light-weight treated soil are generally performed at the water depth shallower than -3m. However, large water depth construction below -3m is designed for Kumamoto Port. In this study, underwater casting tests of light-weight treated soil were performed up to the water depth of about .-10m at Kumamoto Port. Several laboratory tests were also performed on the samples of one year after the casting. 1 KUMAMOTOPORT
1.2 Quay wall design
1.1 Location of Kumarnoto Port
Since the Ariake clays are ultra-soft clays, soil improvement is inevitable for the construction of port and harbour structures. As the scale of soil improvement work is large, the cost of the improvement is usually more than 50% of the total cost of construction. The 4th District Port Construction Bureau, Ministry of Transport has investigated the possible methods for reducing construction cost in this area, and suggested that the light-weight treated soil shall be used as backfilling of the quay wall. The light-weight treated soil is a material in which the dredged slurry is mixed with cement and the light-weight agent such as air foam or expanded Polystirol (EPS) beads, and has been developed by Port and Harbour Research Institute (Tsuchida et.
Kumamoto Port is located in Ariake Bay. The Ariake Bay is a typical inland sea and the depth of water is very shallow with the average of about 20m. A vast tideland appears at the low tide due to the large difference in tidal level of 4.5m. The location of Kumamoto Port and the field test in this study is shown in Fig.1. In this area, extremely soft clay layer, that is called Ariake clay, of 30 to 40m thickness is deposited at the seabed. Fig. 2 shows natural water content w,, liquid limit wL and unconfined compression strength qu with depth of Kumamoto Port. As shown in Figure, the water content is larger than the liquid limit and seems to be normally consolidated state.
*j*5v\. . Gfw
Ariake Bay
I
&*
krn
K
u
m
i
F
I
JKumamoto
oeo c. .
-~ 1. v
South Breakwater
D
Figure 1. Location of Kumamoto Port. 709
el, 1996, Coastal Development Institute of Technology, 1999). As the wet density of light-weight soil is 1.0 -1.2 and the undrained shear strength is 200400 kPa, the back-filling of the light-weight soil makes the small earth pressure and the reduction of the foundation improvement. Fig.3(a) and Fig.3(b) show the comparison of quay wall in the conventional design and the case of back-filling of light-weight treated soil. As shown in the figures, the width of the concrete caisson is reduced by using lightweight treated soil and the scale of the ground improvement work becomes smaller due to the reduction of the weight of the whole construction structure. Consequently, the construction cost can be reduced by 20-25% using light-weight treated soil.
v V
20.000
-
H . W . L +4.50m V 4 0 m L . W . L +O.OOm
V +5.?Om I
I
i :
,
I
,
I ,
* I
I
Sand Compaction (Improvement Ratio= 80%)
15.000
20,000
,
, I
I ’
.
GF:gnt .
,3.000
2 UNDERWATER CASTING TEST 2.1 Properties and mix proportion of treated soil As light-weight agents, air foam type agents and expanded PolyStirol (EPS) beads, the diameters of which are 1-3mm, (Fig. 4) were utilized. Before performing casting test, the experiments including physical and chemical properties tests of dredged soils, selection test for the kind of air foam agents, stability test of air foam, quantitative test of cementing materials, washout resistant test for light-weight treated soil cast underwater, and curing test under pressure were performed to determine the mix proportion of the light-weight treated soil. The materials utilized for casting tests are shown in Table 1. The target values of underwater casting test are shown in Table 2. The mix proportion for the lightweight treated soil is shown in Table 3.
V -13.Om
=
Figure 3(b) Cross section of quay wall (back-filling of light-weight treated soil)
Figure 4 Expanded PolyStirol (EPS) beads. 2.2 Outline of casting test Underwater casting test of light-weight treated soil was performed inside the concrete caisson being under construction at Kumamoto Port to prevent the floating EPS beads from spreading into the sea. The inside of the concrete caisson was divided into 4 sections and the casting tests of No.1 to No.9 were carried out at the sites as shown in Figure 5. Total casting quantity of light-weight treated soil for 5 days was 860m3. For casting light-weight treated soil using EPS beads, about 2% volume of EPS beads were floating. But they did not spread into the sea.
Figure 2 w,,, wLand quwith depth.
710
of core samples collected 28 days after casting was slightly larger than that immediately after mixing. Table 4 shows the increase rate in wet density of light-weight treated soils of different types . In the case of utilizing the surface active agent type air foam, the increase rate in wet density in the deep layer was larger than that in the upper layer. Such a phenomenon is explained due to the decrease of air volume caused by the water pressure at the site. When the light-weight agent of protein type was used, the wet density became much larger than the target value. It seemed that the air foam made by this agent was weaker that that made by the surface active agent type, and the most of the air foam was disappeared during the transportation and the casting.
Table 1.Utilized materials Material
Water Cementing material Light-weight agent
Dredged soils Unit weight: 2.689 t/m3 Natural water content: 84.4% Liquid limit: 63.8% Ignition loss: 6.9% Seawater B-type Portland blast-furnace slag cement Air foam Surface active agent type Protein type EPS beads Mean grain size: 2mm
Table 4 The increase rate of wet density
Table 2. Target values
1
compressive Unconfmed streneth Wet density (at casting) Wet density (28 days after casting) Flow value
1
Light-weight agent S (surface active agent) P (protein) B (EPS beads)
Approximately 200 kPa Approximately 1.1t/m3 Approximately 1.2 t/m3 150-200 mm
Upper layer 4.7% No.6 3.6% No.9 14.8% No.2 17.2% No.8
Casting site Middle layer 7.2% No.4
Deep layer 10.2% No.1 10.3% No.7
15.1%No.3 5.0% No.5
Table 3. Mix proportion of light-weight treated soil. Table 5. Soil tests
I
I
Air foam
1
Beads
1
Unconfined compression test Direct shear test of Mikasa type (CU)
1 Total
I
1000
1
Above result showed that when light-weight treated soil is cast at large depth, it is necessary to add extra volume of air foam to compensate for the possible decrease in air volume. The reduction of air volume and the increase of density depends on the casting depth and the type of the air foam agent. In the case of Ariake Clay in Kumamoto Port, the air foam made by protein type agent was too weak comparing with that made by surface active type agent.
1000
The layout of casting is shown in Fig. 6. The dredged soils were moved into the mixing plant from the barge by a backhoe. Then, cementing material and light-weight agent were added and mixed. The treated soils were cast through a tremie pipe by squeeze type concrete pump. During the casting test, about 2-3 % of mixed EPS beads were segregated and floated in the site. But as they did not spread into the sea, all of the floating beads were collected easily.
3 RESULTS OF SOIL TESTS OF CORES COLLECTED 1 YEAR AFTER CASTING
3.1 Outline of soil test In 1999, just 1 year after casting, in order to investigate whether the quality of light-weight treated soil changes with curing 1 year after casting, soil sampling of SGM at Kumamoto Port and several soil tests were performed as shown in Table 5.
2.3 Control of wet density
The wet density is an important index to control the quality of the light-weight treated soil in casting test. The investigation indicated that the wet density 711
Cell-2 No.9 (S, 100)
cell-4