Powder Technology, 72 (1992) 31-37
Scaling considerations G. S. Patience*,
31
for circulating fluidized bed risers
J. Chaouki
Dkpatiement de Gt!nie Chimique, Ecole Polytechnique de Monttial, Monheal, Que., H3C 3A7 (Canada)
F. Berruti
and R. Wong
Department of Chemical and Petroleum Engineering, University of Calgay, Calgary, Alta., T2N IN4 (Canada) (Received
June 10, 1991; in revised form December
27, 1991)
Abstract The ratio between actual gas velocity to particle velocity in the hydrodynamically fully developed region of Circulating Fluidized Bed risers (CFB) may be approximated by cp= 1 + 5.6/Fr + 0.47Frp’ = U&VP. This ratio, termed the slip factor, is about 2 at operating conditions characteristic of industrial risers several meters in diameter and agrees with observations of J. M. Matsen (in D. L. Keairns (ed.), Fluidization Technology, Vol. 2, Hemisphere, 1976, p. 135). The proposed relationship between the gas and solids velocity is an adequate first approximation to estimate gas and solids residence times, blower capacity and standpipe length.
Introduction Circulating Fluidized Beds (CFB) are being considered as alternatives to more conventional Fluidized Bed processes because of their apparent intrinsic advantages, including short and controllable residence times for the gas and solids, high turn down ratios and flexibility. Industrial scale plants for coal combustion, aluminum oxide calcination, catalytic cracking, Fischer-Tropsch synthesis successfully employ this technology. Contractor and Chaouki [l] and Gianetto et al. [2] have discussed a number of potential catalytic processes that are likely candidates for CFB technology. The main difference between bubbling, turbulent beds and CFB risers is gas velocity. Typical gas velocities in CFBs range from 2-10 m s-l. At these velocities, solids are readily entrained by the gas and are carried to the top of the vessel. Cyclones and rough cut separators separate the solids from the gas phase. The solids are returned to the riser bottom by a standpipe. The longitudinal solids hold-up in the riser, discussed by Yerushalmi et al. [3] and by Li and Kwauk [4], exhibits a relatively dense region at the solids entry point and a dilute phase above it. A number of models have been proposed to characterize the riser hydrodynamics. Rhodes and Geldart [5] used the entrainment model, developed for fluidized beds by Wen and Chen [6], to describe the dilute phase and adapted a classical *Presently at E.I. du Pont de Nemours DE 19880. USA.
0032-5910/92/$5.00
& Co., Wilmington,
two phase model of a bubbling fluidized bed to the dense phase. Kunii and Levenspiel[7] adopted a similar approach and correlated decay constants based on a number of experimental investigations to model the decrease in solids hold-up along the riser. Also, measurements of the internal flow structure of the riser by Hartge et al. [ES],Bader et al. [9], and Bolton and Davidson [lo] indicated that large radial gradients exist, with significantly higher concentrations of solids near the wall. A complete description of the hydrodynamics of such a flow structure is difficult. The gas-solid flow is typically characterized by large relative velocities between the two phases. Two mechanisms have been proposed to account for the difference in velocity: Yerushalmi et al. [3] suggested that particles agglomerate into clusters whose void fraction approaches E,.,,~,whereas Rhodes et al. [ll] and Berruti and Kalogerakis [12] postulated that particles ascend in the core in a dilute phase and descend along the wall as a dense annulus. Ishii et aE. [13] have recently incorporated both mechanisms into a clustering annular flow model. Most of the studies on the hydrodynamics of CFB systems reported in the literature have been conducted using laboratory scale units (i.e. relatively short and narrow). Scale-up to industrial reactors several meters in diameter and tens of meters in height is uncertain at best. Experimental rigs. are not only limited by diameter and height constraints but also by the maximum circulation rates attainable. Matsen [14] reported that typical industrial FCC units operate at solids fluxes
0 1992 - Elsevier Sequoia.
All rights reserved
32
between 500 and 1500 kg m-’ s-l. The majority of experimental rigs employ non-mechanical devices and Geldart group B particles, which facilitate circulation rate control but may ultimately limit the maximum solids fluxes attainable. Although CFB boilers generally operate at solids circulation rates less than 100 kg m-’ S -I, catalytic reactors require different operating conditions. Despite the growing body of literature, more fundamental information on the hydrodynamics of large scale CFB reactors is needed to assess the potential of this technology and to establish design criteria. Scaleup parameters are useful for the design of industrial CFB units. These parameters should not only estimate the overall pressure drop for a given gas velocity and circulation rate, necessary to size compressors and the standpipe, but also adequately predict reactor performance including gas-solids contact efficiency and heat transfer characteristics. The internal flow structure of small experimental units is well understood, but Dry and La Nauze [15] suggest that the symmetry of the radial solids distribution measured in small units may not apply to large units. However, experiments conducted in a commercial FCC riser by Saxton and Worley [16] using a radiation attenuation technique, indicate that a two-phase type flow pattern might adequately describe the flow phenomenon. In addition to the uncertainties in scaling-up the riser diameter, few studies address the effect of height on the hydrodynamics. In tall risers, differences in the gas velocity between the top and bottom of 50% are conceivable at high circulation rates. Grace [17] indicates that further complications may arise from the exit and entrance effects, wall intrusions or roughness and the coefficient of restitution of the particles. Glicksman et al. [18] documented the increase in the solids void fraction by changing the geometry of the exit from a smooth elbow to a sharp one and presented data suggesting that objects intruding into the riser may significantly influence the local solids hold-up. Glicksman et al. [18], Ishii et al. [13] and Ishii and Murakami [19] proposed scaling laws to predict the behaviour in large scale units. Scale-up criteria were derived based on the principles of fluid-particle systems. The criteria were then verified in geometrically similar small scale lab units. Axial solids hold-up and pressure fluctuations were generally used as the basis for comparison. Despite the differences in derivations, Glicksman et al. [18] maintain that the scaling laws proposed in the literature are not dissimilar. They examined two units with a 34 mm and 152 mm square cross section. Ishii et al. [13] developed scaling parameters based on the Clustering Annular Flow Model and validated the theory experimentally in two geometrically similar units 200 mm and 50 mm in diameter. However, only very
low circulation rates and gas velocities were considered (U,<2 m s-l, G,<20 kg mm2 s-l). Jazayeri [36] developed a graph, based on the data of Van Swaaij et al. [20], that predicts the suspension density at various gas velocities and circulation rates. In the present work, we consider a large pool of data, measured in laboratory and pilot scale risers, to develop a generalized scaling criterion.
Slip factor as a scaling law The slip factor is the ratio of the actual gas velocity to particle velocity, 4p= U&V
P
(1)
The average particle velocity, VP, is evaluated based on the solids circulation rate, VP= G,/p,(l - 6)
(2)
and the void fraction, E, is calculated assuming that the time average pressure drop is attributable only to the hydrostatic head of the solids, E= 1 - dpl(p,g dz)
(3)
Clearly, neglecting the particle acceleration contribution to the pressure drop restricts the analysis to the hydrodynamically fully developed region of the riser. Moreover, eqn. (3) ignores the wall shear stress contribution to the pressure drop, which may be significant under certain operating conditions as shown by Van Swaaij et al. [20] using y-ray absorption. Their results indicated that at low riser density conditions the measured pressure drop was systematically higher than the hydrostatic head of solids and at high densities it was lower. They showed that the wall shear stress in the fully developed region is 20-40% of the total pressure drop at gas velocities greater than 13 m s-l and mass fluxes greater than 180 kg me2 s-‘. At gas velocities below 6 m s-l and circulation rates up to 350 kg m-’ S -’ the contribution of the shear to the measured pressure drop was negative and less than 25% of the total pressure drop. The slip factor is not commonly reported in the scientific literature. However, Bolton and Davidson [lo], Yang [22], and Kunii and Levenspiel [7] have used the slip velocity as a parameter to model the hydrodynamics of experimental risers, v,, = U,lE - VP
(4)
In addition, the slip velocity, VsI, has been used as a parameter for heat and mass transfer correlations. Rhodes and Geldart [5] and Patience and Chaouki [21] have assumed that Vsl is equal to the particle terminal velocity from which the particle velocity is calculated.
33
However, work by Patience et al. [23] clearly shows that the latter assumption is not true and better criteria are required to evaluate the average solids velocity. In large industrial scale FCC risers Matsen [14] reported that the slip factor cp, is approximately equal to 2 and hence the particle velocity equals Ug/2e. Comparisons between large scale industrial units and experimental units is complicated not only because of the differences in geometry but also because of the differences in operating conditions; high circulation rates, high temperatures and pressures. Sternerding [24] showed that the riser pressure drop was independent of the transport gas properties in the range, 1.4 lo-’
5
I V
I 0
V 4 -
v
I
I
1
oe
Us=
6 m/s
0 Gs= l Gs=
wcm
102 108
u,=
2 !z
2
kg/n+ kg/m
s
6 m/s
V G,= 26 kg/n+ ‘I Gs= 07 kg/m
0
.M
I
l T
s
l
0
100
200
and
400
300
500
700
600
P .u,p (kdm3)
Fig. 2. Longitudinal
0.33 < pg < 5.0kg rnm3. However, Findlay and Knowlton [32] suggest that the solids mass fraction is inversely proportional to the gas density to the 0.6 power. More study on the effect of gas properties on suspension density is required. Van Swaaij et al. [20] reported data with FCC catalyst in the developed region of an 0.18 m diameter riser at circulation rates up to 500 kg mm2 s-l and their experimental circulation rates are compared with the predicted values assuming a slip factor of 2 in Fig. 1. Reasonable agreement between the predictions and the experimental values is evident despite the wide range of operating conditions. The gas velocity varied between 4.3 to 15.1 m s-’ and the mass flux from 133 to 514 kg m-’ s-‘. Figure 2 illustrates the longitudinal solids density at gas velocities of 6 and 8 m s-l and various solids circulation rates in a riser 83 mm in diameter and 5 600
I
I
r
I
I
op=z l p = 1+5.6/Fr+0.47Fr,0’41 ” E
500
\ r” r E:
400
: 4 e
300
2
200
2 a ”
100
0 0
100
200
Experimental
300 Mass
400 Flux
500
600
(kg/m’s)
Fig. 1. Data of Van Swaaij et al. [20] compared with predictions assuming a slip factor of 2, and a slip factor = 1 + 5.6/Fr + 0.47Fr,“4’.
suspension
density, Patience
[25].
W
U,=
6 m/s
0 G,= l G,=
102 kg/n+ 196 kg/m
s
WI= 6 m/s V
Gs= 26 kg/m’s
v
G,=
67 kg/m’s
l
I 02
,
I
10 Slip
Fig. 3. Variation
w
20 Factor,
t
C
30
40
(p
of the slip factor along the riser length.
m tall as reported by Patience [25]. Sand with a mean diameter of 275 ,um and a density of 2630 kg rnF3 was used in all experiments. The suspension density in the riser decays exponentially, reaches a steady value around the middle of the column and eventually increases toward the top. The exit configuration was an abrupt reducing right angle. Brereton [26] attributes the increase in the density at the exit to internal refluxing of solids. Experimental data, shown in Fig. 2, are replotted in terms of slip factor, instead of suspension density, in Fig. 3. At the entrance, the calculated slip factors are greater than 20 and at the top of the riser they are greater than 7. However, in the centre of the riser, above the acceleration region, the flow profile is flat and perhaps hydrodynamically ‘fully developed’. The slip factor is close to 2 in this region, which agrees with measurements in risers up to 1.5 m in diameter
34
that operate at elevated circulation rates with Group A powders as reported by Matsen [14]. At the entrance, the slip factor increases with mass flux, which may be attributable, in part, to the overprediction of the solids fraction because the acceleration contribution to the pressure drop was neglected. Also, the slip factor appears to be greater at higher gas velocities at the top of the riser. In Fig. 4, data measured by Wong [27] in a 3 m tall riser 50 mm in diameter is shown. The apparent slip factor in the acceleration region is plotted together with the ‘actual’ slip factor in which the acceleration contribution of the particles is taken into account. The contribution of particle acceleration to pressure drop, hence density, was calculated based on the work by Weinstein and Li [30]. The figure indicates that, although the ‘actual’ slip factor in the acceleration zone is greater than 2, ignoring the acceleration effect greatly overestimates the slip factor, hence total solids hold-up. The slip factor calculated in the hydrodynamically developed region, based on data reported by a number of researchers, is plotted against the gas velocity in Fig. 5. Table 1 summarizes the particle characteristics and riser geometry of each study. Both Geldart A and B powders were used in the experiments for which the particle terminal velocities vary between 0.2 m s-l and 2 m s-l. A slip factor of 2 correlates the data reasonably at gas velocities between 6 and 12 m s-l. This agrees with the value reported by Matsen [14] for industrial risers, which typically operate at velocities greater than 8 m s-l. To account for the increase in q with decreasing gas velocity, as shown in the figure, the following relationship is proposed: cp= 1 + 5.6jFr + 0.47Frp4* 6-
5-
E0
3-
.? Ci
U
=
7.9
GE
=
57
m/s kg/m's
174
d; =
/.un
0 (p experimental 0 0 corrected for
0 4-
I
I
I
0
9. i 9
(5)
acceleration
0 0
0
0 2
D
”
CJ
Height
0
00
1 .o (m)
Fig. 4. Slip factors in the hydrodynamically (acceleration) of a riser, Wong [27].
developing
region
10
I
I _--
I (P =
1
I
I
I
1+5~3/Fr+0.47Fr,~.~~
06 0
2
4
6
8
10
12
14
16
up b/s) Fig. 5. Slip factors in the hydrodynamically fully developed region at different gas velocities. Data referenced in Table 1.
Agreement between predicted and experimental circulation rates of Van Swaaij’s data [20] using this correlation is good, as shown in Figs. 1 and 5. The fit is superior compared to the single parameter estimate of rp=2. Equation (5) suggests that, at gas velocities much greater than single particle terminal velocities, the solids hold-up increases with diameter, i.e. 4(1-
l) = (1 +5.6(Dg)o.5/U,
+ 0.47Fr,0-41)cG,lUg
(6)
The effect of riser diameter on suspension density has not been fully explored. Arena et al. [31] studied two risers 41 mm and 120 mm in diameter and concluded that the density increased with diameter at the same operating conditions. Kato et al. [34] reported that, for small tubes, the density increased with diameter to the 0.4 power, whereas a power of 0.2 fit data collected by Findlay and Knowlton [38] better. Larger riser diameters were used in the latter study. The correlation predicts that the solids hold-up is relatively independent of particle characteristics as long as the superficial gas velocity is much greater than the particle terminal velocity. Moreover, it suggests that the solids hold-up is relatively insensitive to gas properties. At high gas velocities, typical of pneumatic conveying, 5.6/Fr tends to zero and cp approaches 1 +0.47Frp41. Typically, FCC risers operate at slip factors near two; Govier and Aziz [37] suggest that in pneumatic conveying 1 < cp< 2, which agrees with the proposed correlation. Brereton [26] reports large differences in the slip factor between a smooth exit and an abrupt geometry. The slip varies between 1.88 and 2.32 for sand particles (open squares in Fig. 5) in a CFB with a smooth exit. The slip factor for sand in the same unit with an abrupt exit geometry (filled squares) varies from 8.2 at low gas velocities to 3.6. Most of the data reported in the
35 TABLE
Key
V 0
1. References
and experimental
Study
Arena et al. [LB] Brereton [26]
n
A
0 + v 0 A 0
Patience [25] Rhodes and Geldart [5] Stemerding [24] Van Swaaij et al. [20] Wong [27]
conditions
for the data in Fig. 5
Riser geometry
Particle properties
D
Height
Type
Cm)
Cm)
0.041 0.152 0.152
6.4 9.3 9.3
0.152 0.083 0.152 0.051 0.18 0.05 0.05
9.3 5 5 2-10 3 3
Remarks
dP (pm)
Zg mm3)
(“m SC‘)
Glass Sand Sand
88 148 148
2600 2650 2650
0.46 0.99 0.99
Alumina Sand FCC FCC FCC Sand Sand
65 275 64 65 65” 93 174
3500 2630 1800 1600 1400” 2500 2500
0.36 1.9 0.2 0.18 0.16 0.48 1.2
Smooth exit Smooth exit Abrupt exit
Abrupt exit Abrupt exit, 20 < T< 250 Abrupt exit , Developed region, 15 < T < 500 Developed region Abrupt exit Abrupt exit
‘Estimates.
CFB with the abrupt exit geometry varies between 4 and 6. The slip factor with the alumina particles and an abrupt
exit geometry
(open
triangles)
lies between
2.1 and 4.7. The large slip velocities reported by Brereton [36] may be a result of the contribution of two different phenomena. Firstly, the abrupt exit configuration, in this unit, might affect the solids behaviour significantly. Secondly, secondary gas was supplied to the column which extends the acceleration region of the riser. The combination of the abrupt exit and extended acceleration zone may prevent the establishment of a fully developed flow region. However, experiments by Patience [25] and Wong [27], conducted in short risers with abrupt right angle exits, for which the acceleration zone would presumably affect the solids loading more than in taller units, gave much lower slip factors. Whereas values of rp greater than those predicted by the proposed correlation are evident in the work of Brereton [26], smaller values of cp are calculated from data obtained in the experimental unit of Arena et al. [28] as reported by Louge and Chang [29] and Yang [22]. In this case, the slip velocity approaches the single particle terminal velocity. At gas velocities of 5 m s-l and circulation rates ranging from 80 to 390 kg rnp2 s-l the slip varies between 1.15 and 1.055. Inaccuracies in the measurement of the circulation rate could explain the differences in slip factor as reported by Brereton [26] and Arena et al. [28]. The slip factor is inversely proportional to the mass flux; hence, underestimating the circulation rate will result in large calculated slip factors and overestimating the rate gives low values of q. Patience and Chaouki [33] show that using the particle velocity along the downcomer wall, the technique used by Brereton [26] in the experiments with sand, can underestimate the circulation rate by up to 40%. Brereton [26] used a butterfly valve technique
in the experiments with alumina and eqn. (5) fits these data well compared to the experiments with sand where the solids circulation rates were made by tracking the wall velocity in the downcomer. The general agreement between slip factors reported for industrial FCC reactors and experimental risers suggests that pressure drop predictions may be possible without having to develop large and costly pilot plants. That is, given the desired residence time of the gas and solids, the blower requirements and height and diameter of the reactor may be calculated. The insistence of geometrical similarity, as suggested by Ishii et al. [13] and Ishii and Murakami [19], appears to be too restrictive. In the present work, the proposed slip factor model is shown to provide a reasonable estimate of the average gas and solids residence times. Particle characteristics do not affect the slip factor as long as the operating gas velocity is significantly greater than the particle terminal velocity (in the fully developed flow regime of the riser). Van Swaaij et al. [20] and Stemerding [24] used Geldart Group A powders, whereas Patience [25], Brereton [26] and Wong [27] used Geldart Group B particles. Zhang et al. [35] suggest that particle density and particle size distribution of Geldart A materials do not affect the radial voidage profile when comparing systems operating at the same suspension density.
Conclusions Correlations available in the literature do not seem to predict the relationship between the gas and solids velocity adequately. An examination of a large pool of data from both experimental laboratory scale CFBs and industrial units indicates that the ratio of interstitial
36
gas velocity to particle velocity, or slip factor, is approximately equal to 2 in the hydrodynamically fully developed flow regime of risers at superficial gas velocities greater than 6 m s-‘. However, an improved relationship is also proposed,
which better describes the slip factor dependence on gas velocity, riser diameter, and particle properties. The suspension density increases with diameter and decreases with increasing gas velocity. At high gas velocities cpapproaches 1 + 0.47Frto.41.This relationship applies to the region, above the acceleration zone at the entrance and below the deceleration zone at the exit where values of C+J are greater. Therefore, to estimate gas and solids residence times, blower capacity, and standpipe length requirements the entrance and exit effects must be considered. However, entrance and exit lengths are typically much shorter than the developed region, so errors in ignoring the pressure drop contribution in a 40 m tall riser would be less than 10%. The slip factor, cp, is reported to be independent of gas properties and particle characteristics (at gas velocities much greater than the single particle terminal velocity).
References
1 R. M. Contractor
10
11
12 13 14 15
List of symbols riser diameter D particle diameter d, pressure gradient dPld.z Froude number, U&D)‘.’ Fr Froude number, V&‘gD)“~5 Fr, solids flux in riser GS gravitational constant g time-averaged pressure P temperature T superficial gas velocity utx particle velocity v, slip velocity KI particle terminal velocity v, vertical co-ordinate z
16 17 18
19
20 21 22 23
24 25 26
Greek letters void fraction E void fraction at minimum fluidization Emf gas density Ps particle density 4 slip factor cp
27 28
29 30
and J. Chaouki, in P. Basu, M. Horio and M. Hasatani (eds.), Circulating Fluidized Bed Technology III, Pergamon, Oxford, 1991, p. 39. A. Gianetto, S. Pagliolico, G. Rover0 and B. Ruggeri, Chem Eng. Sk., 45 (8) (1990) 2219. J. Yerushalmi, M. Cankurt, D. Geldart and B. Liss, AIChE Symp. Ser., 74 (76) (1978) 1. Y. Li and M. Kwauk, in J. R. Grace and J. Matsen (eds.), Fluidization ZZJ Plenum, New York, 1980, p. 537. M. J. Rhodes and D. Geldart, Chem. Eng. Res. Des., 67 (1989) 20. C. Y. Wen and L. H. Chen, AZChE J., 28 (1982) 117. K. Kunii and 0. Levenspiel, Powder Technol., 61 (1990) 193. E.-U. Hartge, Y. Li and J. Werther, in P. Basu (ed.), Circulating Fluidized Bed Technology, Pergamon, Toronto, 1986, p. 153. R. Bader, J. Findlay and T. M. Knowlton, in P. Basu and J. F. Large (eds.), Circulating Fluidized Bed Technology II, Pergamon, Oxford, 1988, p. 123. L. W. Bolton and J. F. Davidson, in P. Basu and J. F. Large (eds.), Circulating Fluidized Bed Technology II, Pergamon, Oxford, 1988, p. 139. M. 3. Rhodes, P. Laussmann, F. Villain and G. Geldart, in P. Basu and J. F. Large (eds.), Circulating Fiuidized Bed Technology ZZ, Pergamon, Oxford, 1988, p. 20. F. Berruti and N. Kalogerakis, Can. J. Chem. Eng., 67 (1989) 1010. H. Ishii, T. Nakajima and M. Horio, J. Chem. Eng. Jpn., 22 (5) (1989) 484. J. M. Matsen, in D. L. Keairns (ed.), Fluidization Technology, Vol. 2, Hemisphere, 1976, p. 135. R. J. Dry and R. D. La Nauze, Chem. Eng. Prog., 86 (7) (1990) 31. A. L. Sazton and A. C. Worley, Oil Gas J., 68 (20) (1970) 84. J. R. Grace, Chem. Eng. Sci., 45 (8) (1990) 1953. L. R. Glicksman, D. Westphalen, C. Brereton and J. Grace, in P. Basu, M. Horio and M. Hasatani (eds.), Circulating Fluidized Bed Technology IIZ, Pergamon, Oxford, 1991, p. 119. H. Ishii and I. Murakami, in P. Basu, M. Horio and M. Hasatani (eds.), Circulating Fluidized Bed Technology III, Pergamon, Oxford, 1991, p. 125. W. P. M. Van Swaaij, C. Buurman and J. W. Van Breugel, Chem. Eng. Sci., 25 (1970) 1818. G. S. Patience and J. Chaouki, Chem. Eng. Res. Des., 68 (A) (1990) 301. W.-C. Yang, in P. Basu and J. F. Large (eds.), Circulating Fluidized Bed Technology II, Pergamon, Oxford, 1988, p. 181. G. S. Patience, J. Chaouki and G. Kennedy, in P. Basu, M. Horio and M. Hasatani (eds.), CircuZatingFZuidized Bed Technology III, Pergamon, Oxford, 1991, p. 599. S. Sternerding, Chem. Eng. Sci., 17 (1962) 599. G. S. Patience, Ph.D. Dissertation, Ecole Polytechnique de Montreal, Canada, 1990. C. M. H. Brereton, Ph.D. Dissertation, University of British Columbia, Vancouver, Canada, 1987. R. Wong, M.Sc. Thesis, University of Calgary, Canada, in preparation. U. Arena, A. Cammarota and L. Pistone, in P. Basu (ed.), Circulating Flutdized Bed Technology, Pergamon, Toronto, 1986, p. 119. M. Louge and H. Chang, Powder TechnoZ., 60 (1990) 197. H. Weinstein and J. Li, Powder Technoi., 57 (1989) 77.
37 31 U. Arena, A. Cammarota, L. Massimilla and D. Pirozzi, in P. Basu and J. F. Large (eds.), Circulating Fluidized Bed Technology II, Pergamon, Oxford, 1988, p. 223. 32 J. Findlay and T. M. Knowlton, Final Report: Pipeline Gas from Coal (IGT Hydrogasification Process), (1980) 306. 33 G. S. Patience and J. Chaouki, in P. Basu, M. Horio and M. Hasatani (eds.), Circulating Fluidized Bed Technology III, Pergamon, Oxford, 1991, p. 627.
34 K. Kato, Y. Ozawa and H. Endo, in K. Ostergaard and A. Sorenson (eds.), Fluidization, Engineering Foundation, New York, 1986. 35 W. Zhang, Y. Tung and F. Johnsson, Chem. Eng. Sci., 46 (12) (1991) 3045. 36 B. Jazayeri, Hydrocarbon Process., 70 (5) (1991) 93. 37 G. W. Govier and K. A&, Z7zeFlow of Complex Mixtzues in Pipes, Van Nostrand Reinhold, Toronto, 1972.
