Retrofit of Reinforced Concrete Structures to Resist Blast Effects
[ I I by John E. Crawford, L. Javier Malvar, James W. Wesevich, Joseph Valancius, and Aaron D. Reynolds A na l y s es w er e con du ct ed t o demo ns t r at e t he eff ect i v enes s of j ack eti ng col umns umns of existi ng reinforced conc concrete rete multi multi story buildings to i mprove mprove their their surviv abili abili ty to attacks attacks by explosives. xplosives. standoff distances, charge si zes, zes, and and steel steel and and compo composi si te jackets jackets were con consi si dered dered.. T wo bui bui lding designs were analyzed: one in which the building members were designed pri mari mari ly for gr avi ty loads seismic I ) and one one i n whi ch the the memb membe ers were desi gned to resist resi st sei sei smic smi c (UBC seismic zone 4). Structural Structural respon response se predictions predictions with the three-dimensional, element code using a concrete material model especi specially ally desi desi gned to predi predi ct nonli near near concrete concrete response responsess t o explosi explosi ve loads. The results indicate that jacketing can be an effective means to retro fit an existi ng facili facili ty to lesse lessenn its vulnerab vulnerabili ili ty to blast blast loads. loads.
Keywords blast effects; reinforced concrete; steel jacket; composites;
INTRODUCTION Preliminary numerical analyses of conventional reinforced concrete multistory buildings confirmed their vulnerability to attacks by explosives. The focus of these analyses was the response of the perimeter columns of a typical building to blast loads. Two building designs were analyzed: one in which building members were designed mainly for gravity loads (UBC seismic zone 1) and one in which the members were designed to resist seismic loads (UBC seismic zone 4). Both cases were considered since the zone 4 design includes significantly higher lateral reinforcement (for confinement purposes) which enhances the resistance to shear. The numerical analyses showed that structural lapse of the building as a whole was typically started by the failure of perimeter columns on the first and second floors. Retrofit techniques consisting of strengthening the columns with round steel jackets or composite wraps were assessed. Structural response predictions were completed with the three-dimensional Lagrangian finite element code using a concrete material model especially designed to predict nonlinear concrete responses to explosive loads. The study also includes the use of different charge sizes and standoffs so that architectural considerations that limit the threat (e.g., increasing standoff) can be evaluated.
RESEARCH SIGNIFICANCE Recent events have emphasized the vulnerability of conventional multistory buildings to blast loads. This paper assesses simple retrofit techniques in typical multistory reinforced concrete buildings to improve their blast load resistance. The retrofit designs considered focus on jacketing concepts which have been widely applied in the mitigation of earthquake hazards for highway bridges, mostly by the California Department of For seismic zone 4 these retrofit techniques could then have dual application. Also, the effect of standoff is evaluated as a way of reducing the blast load on the structural elements. MODELING OF BUILDINGS FOR BLAST ANALYSIS To illustrate the issues involved, the response of a multistory building composed of reinforced concrete columns and floor slabs is shown in Figure 1. For this situation, a 2000 po pound (909 Kg) ANFO bomb was placed at a standoff of 20 feet (6.1 m), a burst height of 6 feet (1.83 m), and centered on an exterior building column. To simplify the problem, the ‘loading applied consists of only (i.e., fragment and debris effects are ignored). The was generated separately (e.g. see reference 5). The pressures reflected off the exterior surfaces are predicted with relatively high fidelity (in contrast to, the pressure field inside building, which is complex and difficult to predict). The results indicate that the exterior perimeter columns on the first and second floors are bl blown out at an early time, failing in shear near the supports. Failure of just a few perimeter columns results in a partial or complete collapse of the structure (Figure 1). the retrofit techniques analyzed concentrated on the upgrade of the perimeter columns on the first and second floors.
Structural Journal, 94, No. 4, July-August 1997. Journal, Received September 29, 1995, and reviewed under Institute publication policies. Copyright 6 1997, Concrete Institute. All rights reserved, including making of copies unless permission is obtained from tbe copyright proprietors. Pertinent discussion will be published in May-June 1998 ACI Structur Structural al Journa Journall if received by January 1.1998. ACI
John is a principal engineer at and Case, Engineers. His is to the analysis and for structural systems with particular emphasis on the effects of blast and shock on such interests include and theoretical structural analysis techniques pertaining to nonlinearity and failure, response at high of loading, and particularly related to contact surfaces and models.
some failure mechanisms, such as direct or punching shear failure, can only be approximated.
Modeling with elements In contrast, continuum models may be very general and can idealize much of the actual physics. However, they are
member is a senior engineer with and Case, Structural E ngi neers. He has conducted research on analysis, testing, and numerical modeling of concrete and reinforced concrete structures subjected to static, dynamic, and blast loads. is a senior engineer with Kamgozian and Case, Structural Engineers. He dynamics research primarily involving the nonlinear prediction of weapons effects on various com ponents. A of his research which now includes a more robust formulation of concrete material a of this research. Joseph is president of responsible for the design of buildings as well as blast Aaron Reynolds was an engineer at
and Case,
and Case,
Engineers. He is and institutional Engineers.
The modeling of buildings for predicting response to blast loads may include an array of simple single-degree-of-freedom to full 3-D nonlinear continuum models, some of which are described this paper, An extended discussion of the various tradeoffs involved in model selection is beyond the scope of this paper (e.g. see reference however some of the significant features which distinguish these types of first principle calculations are the importance of including the effect of confinement on the concrete strength and ductility, the effect of strain rate (i.e., apparent material strengthening due to rapid loading), the possibility of direct shear responses (i.e., dynamic shear and the difficulty of determining the loading for many of the structural members. A major issue in computing a finite element response is related to the disparity between the modeling required to capture localized shear and bending failure mechanisms immediately after the detonation, which usually occur over tens of milliseconds, and the modeling needed to capture the global failure mechanisms associated with the collapse of the structure, which occur over times on the order of seconds. The requirement for these relatively long calculation times (i.e., for global collapse) is largely caused by the time needed to include gravity effects, as the structure attempts to redistribute gravity load, following any localized failure of the structural elements. To represent the structure within a finite element context, two basic modeling approaches are available, based on the use of either structural or continuum elements.
with structural elements Models composed of structural elements, e.g., beam and shell elements, are very efficient. The model shown in Figure 1 used beams to represent the columns and shells to represent the floors. This allowed the model to be run for several seconds. However, these models require some sim plifications in the physical component idealization, and some a priori knowledge from the analyst of the structural response. In the case of beam and shell element models, 372
disp.
scal e
Fig. I-Example of building modeled with structural elements at 0.5
Fig. ments
of bunker modeled with continuum ele-
ACI Structural Journal
July-August 1997
Table l-Characteristics of to the first story Load case number
loads applied
Peakreflected pressure, psi
Peak reflected impulse, psi-s
(6.2)
Charge size,
Standoff,
1500
10 (3.05) 20 (6.10) 40 (12.2)
8100 (55.9) 2500 (17.2) 420 (2.9)
(3.05) 20 (6.10) 40 (12.2)
12,000 (82.7) 4400 (30.3) 840 (5.8)
1
4
6.9 (47.6) 3.2 (22.1) 1.6 (11.0)
Height of burst was 6 ft (I
much more difficult to generate, more costly to run, and require a level of expertise not readily available, particularly when they involve nonlinear computations. As an example, Figure 2 depicts a continuum model response of a portion of a munitions bunker wall to the detonation of a nearby munition.”
Modeling with continuum and structural elements Hybrid models having both continuum and structural elements provide an opportunity to develop accurate numerical predictions in the regions of most importance and efficient calculations in the remainder of the structure. A hybrid model was used in this study, where the perimeter columns use continuum elements for the concrete (to capture any shear response) and beam elements for the steel reinforcing bars, while in areas of less importance structural shell elements were used to model the floor joists and slabs. VALIDATION OF RESPONSE PREDICTIONS There is not much experimental data in the open literature by which to evaluate the accuracy and applicability of the models used to predict the effects of blasts on structures. Most test data is either compromised because of its incom pleteness (e.g., lack of complete material characterization), ill-defined boundary conditions (e.g., as often occurs in tests involving single structural members, such as slabs and beams), or is derived from weapons effects programs and is not widely disseminated. One validation study that is available for models similar to the ones shown in paper is presented in reference 10. That paper studied the response of substantial dividing walls to close-in charges (Figure 2). The metric for validation was the velocity of the debris, which were predicted within 10 percent of the test data.
Fig.
of multistory building used in evaluation
ANALYTICAL PROCEDURE USED TO EVALUATE THE RETROFITS Conventional multistory design To measure the effectiveness of the various retrofit designs, a baseline design for a multistory building was generated, as shown in Figures 3 and 4. Two designs were developed: one in which the columns and beams were designed mainly for gravity loads (i.e., consistent with UBC seismic zone 1) and one in which the columns and beams were designed to resist seismic loads (i.e., consistent with UBC seismic zone 4). This allows the evaluation to include the effects of the increased ductility and ultimate strengths associated with a building designed for a highly active seismic zone. Portion of building used for analysis To reduce computational demands, only a single bay from the bottom three stories of the building is used for the response predictions (Figure 5). Symmetry is assumed along the east-west edges of this section. While this is an approximation, it does produce a model of reasonable accuracy and size for evaluating the effects of jacketing. To keep the model simple with little compromise to the column response, the
Table 2-Maximum displacement for first floor perimeter column Maximum Zone 1
TNT yield,
Standoff,
No jacket
(6.10)
40 (12.2)
Zone 4
I
Steel jacket
jacket
No jacket
3000 (1364) .
Steel jacket A ?
1
1500 (682)
10 (3.05) 20
displacement, in (cm)
5.2 (13.2) 0 . 3 (0.7)
1500 (682)
1.0
(2.5)
Ill
failure
failure
safe
14 (35.3)
0 . 7 (1.8)
0.96 (2.4)
0 . 3 (0.7)
0 . 5 (1.3)
1.1 (2.8)
2 . 9 (7.4)
3000 (1364)
failure
1500 (682)
0.17 (0.4)
safe
safe
safe
safe
safe
3000 (1364)
0.79 (2.0)
safe
safe
safe
safe
safe
Note: 1 ft 0.3048 m; 1 in. 2.54 cm; 1 lb = 0.454 kg For the cases where similar calculations indicated that failure would occur
ACI Structural Journal
July-August 1997
3.’
column
safe.
373
GRAVITY LOAD FROM UPPER
GRAVITY LOAD FROM UPPER STORIES
stirrups
TOP
Fig. S-Sections modeled in the analysis (a) Section for
1 b u i l d i n g
Section for
4 building
Fig. I-Typical details associated with the building for a north-south section south edge of the bay floor and girders are fixed at the location of the first interior column. Loading Airblasts at three different ranges were calculated for two different ANFO charge sizes. The peak reflected pressure and impulse at the mid-height of the first floor column are given in Table 1. The gravity load is applied to each finite element within the mesh; the gravity load from the upper stories is applied as a pressure load over the top of the column, as shown in Figure 5. Separate pressure histories are applied to the exterior faces of the first and second story columns. Jacket concepts The main benefit of jacketing is gained from the effect that increased confinement has on the strength and ductility of concrete, as shown in Figure 6. As a secondary benefit, the jacket offers protection from fragment damage and a shape that can more readily deflect fragments. In this application the jacket will be most useful in mitigating direct shear but it can also provide increased axial and bending The steel jacket seams are typically welded, and the gap between the jacket and the existing column is filled grout. Composite or fiber reinforced plastic (FRP) jackets can also be used. Figure 7 depicts the jacket designs used in this study for the columns on the first and second floors of the building shown in Figure 3. This type of column jacketing has been shown to significantly increase the column ductility, typically from a ductility of 1.5 to As a consequence this type of column retrofit (using either steel or FRP jackets) has been extensively applied in California for highway bridge columns.* Material models The material models for the concrete and steel reinforcement include elastic-plastic behavior, rate effects, and fracture. The new concrete material model implemented in includes softening together with a
7500
5000
Unconfined
2500
M O D E L
0
Fig.
3
4
5
6
7
of materi al model to experi mental data for concrete
energy based localization limiter to prevent any spurious mesh sensitivity. The concrete material model includes a radial strain rate enhancement in the principal stress difference versus pressure plane which is valid for uniaxial, biaxial and triaxial tension, as well as uniaxial and biaxial compression. For the analyses, an ASTM A 615 Grade 60 steel was used for reinforcement, a rupture strain of 13 percent. The concrete had a nominal strength of 5000 psi (34.5 For this particular study a relatively weak carbon wrap was used with a thickness of 0.019 inch (0.5 mm) per layer, a strength of 54 ksi (372 and a stiffness of 7600 ksi (52 Only six layers of the composite were applied (additional layers would further stiffen the structural member and
1.4,
1 WRAPS WRAPS 0 WRAPS
-2
0
0.1
0.2
0 .3
STRAIN
I VERT.
Fig. compression
tests
0.4
TESTS TEST MODEL
MODEL
0.5
(%)
model versus test results of cylinder
Fig. 7-Jacket designs for zones 1 and 4
1.0 0. 8
0. 8 0.4 3
0.2
tn 0.0
-0.002 -0.001 TENSION
Fig.
0.000 STRAIN
0.001
0.002
0.003
COMPRESSION
Fig.
for the porti on of the building studied
model strains for a uniaxial compression test 1.0 0.8 0.6 0.4 0.2
0.0 0
0 .1
0.2 0.3 0.4 POISSON RATIO
0.5
Fig. 9-Poisson’s ratio variation in the numerical model
0.6
Fig. of 1 foot standoff and 3000 lb charge
column for a
2.0
30” SQUARE COLUMN
1.8
44”
ROUND COLUMN
0.8 - - - -
COMPOSITE
0. 40.2
Fig.
of zone 3000 lb charge
column for a 20
the deflections). Figure 6 depicts some of the behaviors modeled for the concrete. Material data and details for the material models are given in references 11 through 15. The column jacketing system is dependent on the lateral dilation of the concrete for development of the confining action. Concrete in uniaxial unconfined compression exhibits a constant Poisson ratio of about 0.2 until approximately 75 percent of the compressive strength, corresponding to a volumetric compression phase. At that point extensive internal cracking starts developing and the apparent Poisson ratio starts increasing to 0.5, where there is no further volume variation. For increasing compression the apparent Poisson ratio keeps increasing until the overall volumetric strain becomes zero, then becomes positive (net volume The ability of the numerical material model to reproduce the volumetric expansion phase is the key to the proper representation of the jacketing confinement effect. Figure 8 shows the corresponding output from the new concrete material model for a single concrete element in uniaxial unconfined compression. The predicted variation of apparent Poisson ratio as a function of the load is shown in Figure 9. ASTM C39 compression tests carried out on 6-inch (15.2 cm) diameter concrete cylinders jacketed with two layers of a carbon composite resulted in a strength increase of 20 percent at a peak strain of about 0.005. Figure 10 shows the test results for plain and jacketed concrete cylinders. Figure 10 also shows the predictions for both cases. It is ap parent that the material model is able to properly represent the jacketing effects. With respect to the composite material, although carbon (or graphite) and glass fibers have typically been used for column retrofits,* aramid fibers (e.g. Kevlar) would actually be more appropriate for blast loads, due to their impact resistance. Mesh
The concrete portions of the columns and girders are modeled with three-dimensional eight-node brick elements; the reinforcement is explicitly modeled with truss elements. Shell elements, which replicate the nonlinear
WRAP
JACKET
0.10
0.11
Fig. in. = 2.54 cm)
0. 12
0. 13
0. 14
column types (I
are used to model the floors and floor joists. The mesh for the unjacketed columns is shown in Figure 11. EFFECT OF RETROFITS
The response of the building section for the steel jacket retrofit design is illustrated by plots of the deformed shape of the first floor perimeter column, which are shown in Figures 12 and 13 for various charge sizes, standoffs, and design configurations. The corresponding response for an un jacketed column is also included for comparison. As can be seen from the results, a jacket can have a substantial beneficial effect on the performance of the columns and prevent structural failure of the building as a whole. Figure 14 compares the response time history of four types of columns: the original 30-inch (76 cm) square column, a (112 cm) diameter circular column (resulting from rounding the square column), and the circular column with either a steel jacket or a composite wrap. This comparison is for the case of a (6.1 m) standoff and a (682 Kg) charge. As the confinement on the original square column increases the peak deflection decreases. In the case of the composite wrap the response could be further decreased by increasing the number of wraps. Table 2 provides a summary of the deflection for the various column designs considered. For a small standoff of 10 feet (3.05 the unjacketed column fails for both charges, but a steel or composite jacket can prevent this failure. For a standoff of 40 feet (12.2 no failure is predicted. Zone 4 columns are somewhat more resistant to shearing. This is more apparent when the FRP jacket displacements are compared. The relative thickness used in the composite wrap made it less effective than the steel jacket, but similar improvements could be obtained by increasing the number of wraps. It should be noted, however, that while structural collapse may be prevented, this is only a partial solution for
ing occupants as it does not prevent the propagation of high pressures or debris within the building. CONCLUSIONS The effects of standoff and column jacketing on enhancing the blast resistance. of conventional reinforced concrete structures were analyzed. It was shown first that structural collapse appears to result from the shearing of first and second floor perimeter columns. Jacketing the columns with a steel or composite jacket prevents collapse for most of the cases studied. Although a weak composite wrap system was used in this analysis, similar responses can be obtained for steel and FRP jackets provided that enough wraps are ap plied. Increasing the standoff distance, whenever possible, appears as a simple solution. Further work should be performed to develop additional retrofit designs to enhance the floors, reduce debris production, and channel high pressures away from building occupants.
tural Journal, Sept.-Oct. 1994, pp. 5. Major Hazards Assessment Panel Working Party, Explo sions in the Industries, Major Hazards Monograph, Institution of
Chemical Engineers, Rugby, U.K., pp. 6. Biggs, J. M., Introduction to Structural Dynamics, McGraw-Hill, 1964. 7. Crawford, J. E., and Karagozian, J., “Behavior and Design for Structural Components to Resist Blast Loadings,” Technical Report TR-95-25, Karagozian Case, Glendale, CA, August 1995. 8. Crawford, J. E.; Holland, J.; Mendoza P. J.; and Murtha Failure Methodology Based on Shear Deformation,” Fourth ASCE Engineering Mechanics Division Specialty Conference, Purdue University, Lafayette, May 1983. 9. Slawson, R., “Dynamic Shear Failure of Shallow-Buried Roofed Reinforced Concrete Structures Subjected to Blast Loading,” Technical Report SL-84-7, USAE Waterways Experiment Station, Vicksburg, MS, Apr. 1984. 10. Bogosian, D., “Parametric Analysis of Substantial Dividing Walls,” Technical Report TR-94-20, Karagozian Case, Glendale, Oct. 1994.
ACKNOWLEDGMENT While the results presented in this paper privately funded, the authors would like to acknowledge the support of the Defense Nuclear Agency in developing the basic work associated with predicting and validating the response of reinforced concrete structures loadings.
REFERENCES 1. Whirley, R. G., and Engelmann, B. E., A Nonlinear Explicit Three-Dimensional Finite Element Code for Solid and Structural Mechanics,” User Manual, Report UCRL-MA-107254 Rev. 1, Lawrence Livermore National Laboratory, Livermore, CA, November 1993. 2. “GRP Wraps up Bridge Repairs,” Reinforced Plastics, Vol. 39, No. 78, July-August 1995, pp. 30-32. 3. Priestley, M. J. N.; Seible, and Fyfe, E., “Column Seismic Retrofit Using Fibreglass Epoxy Jackets,” Advanced Composite Materials in Bridges and Structures, First International Conference, Sherbrooke, Que bec, Canada, 1992, pp. 287-298. 4. Priestley, M. J. N.; Seible, F.; Xiao, Y.; and Verma, R., “Steel Jacket Retrofitting of Reinforced Concrete Bridge Columns for Enhanced Shear Strength--Part 2: Test Results and Comparison with Theory,” Struc-
11. Malvar, L. J.; Crawford J. E.; and Wesevich J. “A Concrete MateProceedings of the 10th ASCE Engineering rial Model for Mechanics Conference, Boulder, May 1995. 12. Malvar, L. J.; Crawford J. E.; and Wesevich J. W., “A New Concrete Technical Reporr TR-94-14, Karagozian Material Model for Case, Glendale, Dec. 1994. 13. U.S. Navy Naval Facilities Engineering Command, “Structures to Resist the Effects of Accidental Explosions,” NAVFAC P-397 Design Manual, Alexandria, 1991 (also Army and Air Force AFM 88-22). 14. Ross, C. A.; Thompson, Y.; and Tedesco, J. W., “Split-Hopkinson Pressure-Bar Tests on Concrete and Mortar in Tension and Compression,” Materials Journal, 86, No. 5, Sept.-Oct. 1989, pp. 15. Malvar, L. J., and Simons, D., “Concrete Materi al Modeling in Explicit Computations,” Proceedings, Workshop on Recent Advances in Computational Structural Dynamics and High Performance Computing, USAE Waterways Experiment Station, Vicksburg, Apr. 1996. 16. Park, R., and Paulay, T., Reinforced Concrete Structures, John Wiley Sons, NY, 1975,769 pp.
