Ind. Eng. Chem. Res. 1993,32, 2714-2720
2714
SEPARATIONS Concentration and R Adsorption
very of C02 from Flue Gas by Pressure Swing
E. S. Kikkinides and R. T. Yang’ Department of Chemical Engineering, State University of New York at Buffalo, Buffalo, New York 14260
S. H.Cho Korea Institute of Energy Research, Taedok Science Town, Taejon, Korea
The feasibility of recovering C02 from flue gas to generate a high-purity product by pressure swing adsorption (PSA)is studied by simulation using a proven predictive PSA model. Two promising sorbents, activated carbon and carbon molecular sieve, are examined. A nonconventional PSA cycle is designed for this purpose. With activated carbon as the sorbent, C02 can be concentrated from 17% in the flue gas to a product of 99.997 % (at a C02 recovery of 68.4% and at a reasonably high feed throughput). Despite the favorable kinetic selectivity for C02 in the carbon molecular sieve, separation results using this sorbent are not as good, because the kinetic selectivity (of COdN2) is not high enough and consequently the equilibrium selectivity dominates the separation.
Introduction The emission of C02 from power plants that burn fossil fuels is the major cause for the accumulation of C02 in the atmosphere, which causes long-range environmental problems. Separation can play a key role in alleviating this problem, and research on membrane separation to recover C02 from flue gas is already in progress (King, 1992). In its purified form, C02 has found many uses in the chemical industry. It can be used in solid form (dry ice), liquid form (e.g. refrigeration equipment) or gaseous form (e.g., for carbonated beverages, fire extinguishing equipment). It can also be used as a reactant in many important chemical reactions or as an inert blanketing gas to prevent oxidation (e.g., for food packaging). Today C02 is produced as a by-product of fermentation and lime-kiln operations and by separation from flue gases by gas-liquid absorption processes. Typical flue gases contain around 17% C02, the balance being N2 (79% ) and 0 2 (4 % 1. Trace amounts of SO2 and NO, can also be found, but they are usually much less than 1%in total. The objective of this research is to see if it is feasible to use pressure swing adsorption to recover and concentrate C02 from flue gas into a concentrated form and thus a useful product. The first and most important issue in any adsorption process is to find the appropriate sorbent for the proposed separation. In the present case this sorbent must possess a high selectivity for C02 over both N2 and 02, with fast uptake rates in order to achieve a reasonably high throughput. The second important issue is to design a proper PSA cycle with which one can concentrate the desirable component from the mixture, which in the present case is COZ. The adsorbent that was chosen in this study was activated carbon. It can be found from earlier studies (Knoblauch, 1978;Carrubba et al., 1984;Tsai, et al., 1985;
* Author to whom correspondence should be addressed.
Ritter, 1985; Kapoor and Yang, 1989) that carbon has a much higher adsorption capacity for C02 than for Nz and 0 2 . Non-carbon sorbents, such as zeolites, also have high selectivities for C02. However, COSadsorption on these sorbents (especially zeolites) is too strong, which makes desorption difficult. A further consideration on selection of sorbent concerns possible effects of the “impurities” in flue gas, such as water, S02, and NO,. Again, the strengths of adsorption of these impurities on non-carbon sorbents are unfavorable for desorption. In general, activated carbons (AC) have large pores so that diffusional resistances are negligible for the gases in consideration. On the other hand, carbon molecular sieves (CMS) are microporous sorbents so that diffusion in the micropores is dominant. Earlier studies reveal a high kinetic selectivity of CO2 over N2 at room temperature, which is decreased at higher temperatures. The choice between AC and CMS in a PSA process depends on the nature of the PSA cycle. In conventional PSA cycles it is the light component (i.e., the least adsorbing species) which can be recovered at high concentrations in the adsorption product. However, in the present case the major interest is in the concentration of the strongly adsorbed species (C02) in the desorption product while keeping the adsorption product clean from COz contamination. As a result, a nontraditional PSA cycle is designed. It has been pointed out earlier that a purge step with the strongly adsorbed component can be employed after the adsorption step and result in a significant increase of the purity of that species in the product (Tamura, 1970; Sircar and Zondlo, 1977;Cen and Yang, 1986;Baksh et al., 1990; Suh and Wankat, 1989). In all previous cases the separations were equilibrium controlled and the diffusional resistances were negligible. This suggests that activated carbon should be the better sorbent (than CMS) to be used in the present work, since more CO2 will be stored in the pores of the sorbent during the adsorption steps and consequently more C02 will be recovered from the desorption step.
OSSS-5885/93/2632-2714$04.~0lQ 0 1993 American Chemical Society
Ind. Eng. Chem. Res., Vol. 32, No. 11, 1993 2715
A V T 1
1
t2
Part of the concentrated C02 product from the blowdown step is used as the feed to another bed during its purge step. The volumetric feed flow rate during the purge step is the same with the total volumetric feed flow rate during the adsorption step; i.e., the volumetric purge to feed ratio is fixed at 1.
Mathematical Model I1
AL
111
IV
V
1
A complete adiabatic model is used in this study. The mass balance equation for component i in the bed is ayi
a2yi
ayi
R,T
-- D, -+ u -+ -(p,,/e)[ dz P at az2
dqi --yi dt
E-dqjdt 1 = 0 N
]-I
(1)
The overall mass balance is
Figure 1. Schematic diagram showing the sequence and basic steps of the present PSA cycle. I1 is adsorption step; 111 is purge with concentrated CO2 step; IV is countercurrent blowdown; I is pressurization with adsorption product step.
The scope of the present study is primarily to examine whether it is possible to recover high-purity C02 from flue gas using PSA. The sorbent selected for this separation was a BPL activated carbon (from Calgon Corporation), and equilibrium isotherms of C02, N2, and 0 2 were measured on this sorbent. The results were used as input parameters to an equilibrium PSA mathematical model which was solved numerically to predict if, and under which conditions, high-purity C02 can be recovered. Results were also compared with a detailed kinetic PSA model with input parameters obtained from adsorption and diffusion of the above gases on a Bergbau-Forschung CMS.
Experimental Section The thermogravimeter analysis (TGA) technique, employing a Cahn system 113 recording microbalance with programmed temperature control, was used to measure uptake curves as well as equilibrium isotherms of C02,02, and N2 at different temperatures. High-purity He was used as the inert carrier gas and the gas for regeneration. The sorbents used were BPL activated carbon, supplied by Calgon Corp., and the Bergbau-Forschung carbon molecular sieve. Results of the above measurements were compared and supplemented with independent measurements performed in this laboratory using the differential adsorption bed (DAB) technique a t high as well as low pressures and for different partial pressures of the adsorbative gases. Details on the DAB technique and results can be found elsewhere (Chen et al., 1993).
and the heat balance is
The way the particle fluxes dgi/dt are calculated depends on the nature of the sorbent selected. For the case of a CMS being the sorbent, a complete intraparticle diffusion equation should be solved for each species. Assuming spherical shape and uniform radius, R, for each microcrystal, the intraparticle diffusion equation for each species, i, becomes (4)
with boundary conditions (assuming crystal diffusion control) aqi/dr = 0 at r = 0
(5)
qi = qi* at r = R
(6) where qi* is the equilibrium amount adsorbed at the surface of the crystal and can be calculated using the extended Langmuir isotherm:
PSA Cycle Description A PSA cycle similar to the one used by Baksh et al. (1989) is considered. The four steps (with equal time length) of this cycle are (1)pressurization with adsorption product, (2) high-pressure adsorption, (3)high-pressure purge with concentrated C02, and (4) countercurrent blowdown to subatmospheric pressure. A schematic of the PSA cycle is shown in Figure 1. In the present work, the effluent from the strong adsorptive (C02) purge step is recycled and mixed with the feed line in the adsorption step. Note that in the mixing process the volumetric feed flow rate of the original mixture was varied in each cycle in order to keep the same total volumetric feed flow rate in the adsorption step.
(7)
where q m and b are the Langmuir parameters and are functions of temperature: qm = k, - k,T; b = k, exp[kJT]
(8) Although there is evidence that diffusivity, De,varies with surface coverage, we assume De to be independent of surface coverage since at low pressures this dependence is generally very weak. The temperature dependence for De follows an Arrhenius-type form:
2716 Ind. Eng. Chem. Res., Vol. 32, No. 11, 1993
where E, is the activation energy of diffusion for each species and D,,o is the preexponential factor. For the case of activated carbon being the selected sorbent, the much simpler equilibrium model holds, where eqs 4-6 are replaced by
The bed boundary conditions are of the following form:
at z k = 0, and
at z k = L. Note that the index k corresponds to the step number in the PSA cycle and for each step we have the following. Step II. High-pressure adsorption: 211 = 2,
YII,~= Qf,i; ~ 1 =1
P = PH
where 9f,iis the average mole fraction of species i from the feed line and the product line from step 111. Step III. High-pressure purge with concentrated COz: 2111 = 2, YIII,i = YpIv,i; ~ I I = I uf; P = PH where ypIv,i is the time-averaged effluent mole fraction of species i , during step IV from the previous cycle. For the first cycle a stream of pure COZwas used to initiate the process. Step IV. Countercurrent blowdown:
= L - 2 , UIV = 0; P = P ( t ) Step I. Pressurization with adsorption product: ZIV
q=L-z
P =P(t) The initial conditions for each step are the conditions at the end of the preceding step. Note that the pressure history is an input in PSA and in the present work is represented by an exponential change with time, from PH to PL during blowdown and from PL to PH during pressurization: dPldt = a p e + ' (13) where AP is the difference from PH to PL during the blowdown step and vice versa during the pressurization step. In the above equations the basic assumptions are ideal gas behavior, constant viscosity and heat capacity of the gas phase, negligible variation of all dependent variables in the radial direction, and negligible pressure drop along the bed. Note that duldz does not appear in eq 1, not because of any assumptions but because it was eliminated by proper combination of the species and the overall balance (Doong and Yang, 1986).
Table I. Adsorption Bed Characteristics and Operating Conditions for Standard PSA Simulation Adsorbent Bed bed length 500 cm 100 cm bed diameter 0.48g/cmg bed density (AC) 0.708g/cms bed density (CMS) heat capacity of the bed 0.26 cal/(g K) Operating Conditions feed composition, % N2/02/C02 79/4/17 range of feed superficial velocities 8-14 cm/s 0.4 void fraction PH 1.2 atm PL 0.1 atm 7 calI(mo1K) heat capacity of gas mixture 298 K ambient temperature, T cycle time 16 min (4min/step)
The above equations can be made dimensionless by defining several dimensionless quantities as described elsewhere (Kikkinides and Yang, 1993). The adsorber characteristics together with the operating conditions for the standard case are listed in Table I. In addition, the dimensionless Peclet numbers for mass and heat dispersion, Pel and Pez, respectively (Kikkinides and Yang, 1993), were both set equal to 100 since in an industrial adsorber the effect of axial dispersion is negligible. Numerical Solution of the Model. A numerical scheme already introduced previously (Kikkinides and Yang, 1993)is used in this work, too. The only difference of the present work is that it was found that step I1resulted in very steep wave fronts for C02 and a fully implicit Euler method had to be used throughout the simulation to ensure stability. This resulted in somewhat higher CPU requirements. Note that this modification does not affect the accuracy of the solution since the integration scheme uses a variable time step controlled by a prespecified tolerance. Typical simulations require 3-4 min of CPU time per cycle in a VAX-8800. It took generally 35-50 cycles to reach cyclic steady state.
Results and Discussion Adsorption Isotherms and Uptake Rates. The pure component equilibrium isotherms for COZ,0 2 , and N2 were fitted to the Langmuir equation, and the resulting Langmuir parameters are listed in Table 11. The isotherms of the above gases on either AC or CMS were almost identical, so for the sake of simplicity the equilibrium input parameters were the same for each sorbent. (Note, however, that the bulk densities were different.) The equilibrium isotherms for COZ,02, and N2 on either AC or CMS, at 25 "C are plotted in Figure 2. The results as expected show a very high selectivity for C02 over both 02 and N2 at 25 "C. The selectivity ratio decreases as temperature increases, because of the stronger physical bonds of COPwith carbon, which result in a much higher heat of adsorption for this species, compared to the ones for 0 2 and Nz. However, even at 80 "C the selectivity ratio of 17% COZover 79% NZis still high (around 2.5). In addition, uptake curves of the above gases on CMS were fitted to the solution of the diffusion equation for different temperatures, and the resulting diffusivity parameters as functions of temperature are also listed in Table 11. Typical uptake curves of these gases on CMS at 25 OC are ploted in Figure 3. A similar behavior was observed for the case of diffusivity selectivity. At 25 "C COZdiffuses slightly faster than 0 2 and muchfaster than Nz. This result is in agreement with previous results (Carrubba et al.,1984). However, at higher temperatures
Ind. Eng. Chem. Res., Vol. 32,No. 11,1993 2717 Table 11. Langmuir Parameters, Heats of Adsorption, and Diffusivities of CO;, 0 2 , and Na,on Bergbau-Forschung CMS ki, k2, Q, component mmol/g mmol/ (g K) ks,atm-1 kd, K kcal/mol D,,dR2, s-l EJRg, K COP 02
17.075 4.437 4.199
NP 2.5 I
I
I
- con
I
h
0.0467 0.0085 0.0091
100
I
I
I
346,3 1515.9 1331.6
0.60968 7.82 X lo-" 1.94 X
7.3
0.1945
1808.6
3.8 3.8
4.0880
2810.6 3447.7
I
1.0372 I
I
I
I
I
ti
\ e c 2.0
-1
E E
6\"
90
v
0
w
'\
1.5
[rl
1 _._.__._.-.-.---.
,_._._._.7-.--
0.0
0.0
0.2
I
--
___._._.-. I
I
0.4 0.6 0.8 P R E S S U R E (atm)
1.0
Figure 2. Equilibrium isotherms of COz, 0 2 , and N2 on either AC or CMS at 25 O C . 1 .o
........
w
5
0.8
Na
-
0.6
........ ............. ................. ........ ......
,,
I
-
...... ................ ....... ...... ..........
....
........
........ ,._."
.....
d
.......
0.0
1
I
I
0
5
10 TIME
I
8oi
I
15 20 (min)
I
25
30
Figure 3. Uptake curves of COz, 02 and Nz on CMS at 25 "C.
it was found that the kinetic selectivity decreases due to the lower activation energy of CO2 compared to those of 0 2 and N2, respectively. Note though that higher temperatures affect much more the equilibrium than the kinetic selectivity. PSA Simulation. The main objective of the simulation study is to design a PSA cycle which can produce highpurity C02 in the desorption product while removing a substantial amount of that gas from the adsorption product. The main parameters that describe the performance of the PSA process are the C02 product purity and recovery as well as the purity of the adsorption product.
30
I
I
40
50 CO,
1
I
1
60
70
80
RECOVERY,
90
100
%
Figure 4. Comparison between AC and CMS as selected sorbents for the present separation.
Definitions of the above parameters can be found elsewhere (Baksh et al., 1990). The proposed PSA cycle is different from previous ones in the sense that it recycles the product from the purge step back to the adsorption bed. This increases the C02 recovery and makes the process more realistic. The first step of the theoretical part of this work was to screen sorbent. Therefore, a series of simulations were performed to compare the two different sorbents. The simulations were performed under the standard conditions listed in Table I and for a variety of flow rates to construct the C02 purity vs C02 recovery curves in each case. The results are presented in Figure 4. It is evident that activated carbon gives much higher C02 purity in the desorption product, for a wide range of C02 recoveries, than carbon molecular sieve does. The reason for this difference is due to the lack of any diffusional resistances in the case of activated carbon. For carbon molecular sieve, the benefit of kinetic selectivity is not large enough to offset the diffusional resistances for C02. In general, the stronger the diffusional resistances the closer the particles follow the frozen solid assumption (Le,, not enough time for the adsorbed phase to diffuse out or into the pores, during the pressure-changing steps). On the other hand, the weaker these resistances, the closer the particles follow the instantaneous equilibrium assumption in any step (gas diffuses in and out of the pores instantaneously). As a result, activated carbon is expected to store higher amounts of CO2 during the adsorption and purge with concentrated C02 steps, and then desorb most of it during the subsequent blowdown step. Thus the performance of the PSA process should be much better using a sorbent free of diffusional resistances than using a molecular sieve. In fact, additional simulations showed that for an inverse diffusion time constant of the order of 0.1 s-l, for each component in the gas mixture, the kinetic and the equilibrium models give almost identical results
2718 Ind. Eng. Chem. Res., Vol. 32, No. 11, 1993 100
90
p:
3
a
80
r
I
I
I
1.o
I
I
c
I
0.8
0.6 h
0.4
0.2
1
0.0
. " t
1.0
u
0.8
A
To=2SaC T =BOT T:=80°C I
60
0.6 h
0.4 I
1
70
I
80
I
I
90
1
0.2
100
0.0
CO, RECOVERY, 7%
Figure 5. Effect of ambient temperatureon the performance of the PSA process.
(provided that all equilibrium parameters are the same). This also means that in the limit of long adsorption times carbon molecular sieve behaves more like an activated carbon, as was also pointed out elsewhere (Shirley and LaCava, 1993). Thus the first conclusion of this work is that activated carbon is the more appropriate sorbent than CMS, for the separation of C02 from flue gas and the subsequent concentration of C02. The next step is to investigate how the performance of the PSA process will be affected by varying operating conditions such as ambient temperature and desorption pressure, which are both very important in terms of energy demand for this process.
Cyclic PSA Behavior Simulations are performed using different residence times while keeping the same cycle time. For the case of an equilibrium model, it is the ratio of cycle time over residence time that matters and not the two individual time parameters. This is verified by running two simulations where in the second one, the residence time is half of that in the first case and, at the same time, the cycle time is half of that in the first case. The results were identical up to the fourth significant digit. Note that this is not the case for a kinetic model, where, in the present work, lower cycle times with accordingly lower residence times will result in a worse performance of the PSA process.
Effect of Ambient Temperature Since flue gases are usually at high temperatures, it will be of importance if the PSA process could be operated at as high a temperature as possible. Therefore simulations were performed with initial and feed temperatures of the beds a t 60 "C as well as at 80 "C. The resulting C02 purity vs C02 recovery curves are plotted in Figure 5 for three different ambient temperatures. It is evident that the three curves do not differ significantly, indicating that the PSA system can be operated at stack gas temperatures. The only disadvantage of using higher temperatures is that significantly lower feed flow rates have to be used, in order to achieve a high performance. However this can be improved by decreasing the cycle time while increasing
Figure 6. Bed concentration profiles at the end of the step I1 at cyclic steady state. (a) T. = 26 O C and (b) T, = 80 OC.
0.8 0.6 h
0.4 0.2 0.0
1.0
1
'
t
k
h
j
_ _ - 00
02
06
04
08
10
Z/L
Figure 7. Bed concentration profiles at the end of step I11 at cyclic steady state. (a) T, = 25 O C and (b) T, = 80 OC.
the feed flow rate of the process accordingly, thus maintaining high enough flow rates to keep the productivity of the process high. I t is interesting to compare the dynamics of the process a t cyclic steady state for two cases which gave almost the same C02 purity and recovery, which were run at different ambient temperatures and feed flow rates. The bed concentration profiles are shown in Figures 6-9. In Figure 6 the concentration profiles at the end of the adsorption step are presented. It is seen that the C02 wave front has penetrated deeper in the bed at the high ambient temperature than at the low one. Also a rollup appears at the high ambient temperature. Note that in both cases 0 2 has already broken through the bed, and that CO2 exit concentration is almost the same for both cases. In Figure 7 the corresponding concentration profiles for the purge
Ind. Eng. Chem. Res., Vol. 32, No. 11, 1993 2719 1.0
1
1.10
1
.>
. . . . . . . . . . ......... ...............
0.8
0.6 h
0.4
0.2
0.98 . *. _ . _ . _ . r . . . _ . _ . - . - .
0.0 0.2
0.0
0.4
0.6
0.0
0.8
.................................. bo
0.8
-2 .............
0.6
I
h
0.4
1.01 :.-... f
1.00 "
0.99
'
0.0
0.2
4
0.0
0.2
0.4
0.6
0.8
1.0
z/L Figure 8. Bed concentration profiies at the end of step IV at cyclic steady state. (a) T, = 25 "C and (b) T, = 80 "C.
t. t
'
'
~
0.2
'
'
.
"
04
06
"
'
0.8
'
1.0
Z/L
0.0 t
0.6
1.0
0.8
1.02
1 .o
0.8
0.6
0.4 Z/L
Z/L
1.o
0.2
1.0
.
............. NE
0.4
loo
1 A
PL=O.l
atm
P L = 0 . 3 atm
I
- .- .................................
h
Figure 10. Bed temperatureprofilesat the end of each step at cyclic steady state. (a) T, = 25 "C and (b) T, = 80 "C.
1
0.2 0.0 0.0
1.o
0.2
0.4
0.8
0.6
1.0
.........................
0.8
0.6
60
.............
h
20
0.4 0.2
40 CO,
. 60
80
100
RECOVERY, %
Figure 11. Effect of desorption pressure on the performance of the PSA process.
0.0 0.0
0.2
0.4
0.6
0.8
1.0
z/L Figure 9, Bed concentration profiles at the end of step I at cyclic steady state. (a) T, = 25 "C and (b) T. = 80 "C.
with concentrated C02 step are presented. Again it is seen that at the end of this step more C02 is stored in the bed at 80 "C than at 25 "C ambient temperature. The same is true at the end of the subsequent blowdown step (Figure 8). Note, however, that the concentration of C02 at the exit of the bed is the same at both ambient temperatures. Finally, at the end of pressurization with product step the concentration profiles look quite similar, especially at locations closer to the open end of the bed (Figure 9). This implies that the average flue gas product purities are the same for both cases. In Figure 10 the temperature profiles at the end of each step are shown for the cases of 80 and 25 "C ambient temperatures. As expected, much higher temperature rises occur at the low ambient temperature since the capacity
of the bed for all speciesis much higher a t that temperature. A maximum temperature rise occurs during the purge with concentrated C02 step since C02 adsorbs more strongly on carbon than NOor 02. This rise is of the order of 30 "C when the ambient temperature is 25 "C and decreases to about 5 "C when the ambient temperature is 80 "C.
Effect of Desorption Pressure A series of simulations are performed with PL = 0.3 atm. The other conditions are the same with the standard ones shown in Table I, except for the feed flow rate which was varied in order to construct the COQpurity vs C02 recovery curve. The resulting curve is compared with the corresponding one with PL = 0.1 atm in Figure 11. From this figure it can be seen that a small elevation of the desorption pressure from 0.1 to 0.3 atm makes the performance of the process significantly worse. Conclusions Pressure swing adsorption is considered for the first time as an alternative method to concentrate up to around
2720 Ind. Eng. Chem. Res., Vol. 32, No. 11, 1993
99.997% C02 from flue gas, while the amount of CO2 contaminant in the adsorption product is around 6.3 % (C02 recovery is 68.4% 1. Activated carbon is used as the sorbent. A nonconventional PSA cycle is used, where a purge step with the desorption product is introduced after the adsorption step. The effluent from that step is recycled and mixed with the feed mixture. This PSA cycle is effective even at relatively high temperatures for adsorption, which makes the process attractive from an energy consumption point of view. Comparison between a kinetic PSA process utilizing a carbon molecular sieve sorbent and an equilibrium PSA process utilizing an activated carbon sorbent shows that the latter process is superior, resulting in a much more concentrated C02 in the desorption product for the same adsorption product purity. Thus diffusional resistances can only limit the performance of this particular PSA process and the kinetic selectivity (for COz over N2 in CMS) is of secondary importance as compared to the equilibrium selectivity (of CO2 over N2 in activated carbon).
Acknowledgment This work was supported by NSF under Grant CTS9212279 and by the donors to the Petroleum Research Fund administered by the American Chemical Society.
Nomenclature a = parameter in the pressure history curve, l/s b = Langmuir parameter, defined by eq 8, atm-I cp,g= heat capacity of the gas phase, cal/(g K) c,,,~ = heat capacity of the solid phase, cal/(g K) De = effective intraparticle diffusivity, cm2/s DL = mass axial dispersion constant, cm2/s L = length of the bed, cm N = total number of components in the mixture P = pressure, atm q = adsorbed amount, mol/g = volume-averaged adsorbed amount, mol/g q m = saturated amount adsorbed, defined by eq 8, mol/g Q = (-AH) = heat of adsorption, cal/mol r = radial distance from the center of the microsphere, cm R = radius of the microsphere, cm R, = gas constant, atm cm3/(molK) To= initial and ambient temperature, K t = time, s u = interstitial velocity, cm/s y = mole fraction in the gas phase z = axial position in the bed, cm
Subscripts b = bed f = feed H = high i = species i j = species j k = step number in PSA cycle L = low p = product Superscript * = equilibrium
Literature Cited Baksh, M. S. A.; Kapoor, A,; Yang, R. T. A New Composite Sorbent for Methane-Nitrogen Separation by Adsorption. Sep. Sci. Technol. 1990,25(7-8),845. Carrubba, R. V.; Urbanic, J. E.; Wagner, N. J.; Zanitsch, R. H. Perspectives of Activated Carbon-Past, Present and Future. AIChE Symp. Ser. 1984,80,76. Cen, P. L.;Yang, R. T. Separation of Binary Gas Mixture into Two High-Purity Products by a New Pressure SwingAdsorption Cycle. Sep. Sci. Technol. 1986,21 (9),845. Chen, Y. D.; Yang, R. T.; Uawithya, P. Diffusion of Oxygen,Nitrogen and their Mixtures in Carbon Molecular Sieve. AIChE J. 1993, in press. Doong, S. J.; Yang, R. T. Bulk Separation of Multicomponent Gae Mixtures by Pressure Swing Adsorption: Pore/Surface Diffusion and Equilibrium Models. AIChE J. 1986,32,397. Kapoor, A.; Yang, R. T. Kinetic Separation of Methane-Carbon Mixture By Adsorption on Molecular Sieve Carbon. Chem. Eng. Sci. 1989,44 (8),1723. Kikkinides, E. S.;Yang, R. T. Effects of Bed Pressure Drop on Isothermal and Adiabatic Adsorber Dynamics. Chem. Enn. - Sci. 1993,48 (9),1545. Kikkinides, E. S.;Yang, R. T. Gas Separation and Purification by PolymericAdsorbenta: Flue Gas Desulfurizationand SO2 Recovery with Styrenic Polymers. Znd. Eng. Chem. Res. 1993,-32,2365. King, C. J. Separations in Japan. AIChE Annual Meeting, Miami, 1992;Paper la. Knoblauch, K. Pressure Swing Adsorption: geared for smallvolume users. Chem. Eng. 1978,85,87. Ritter, J. A. Investigation on the Adsorption of C&, C02, CO, H2 and H2S and their Mixtures on Activated Carbon. M.S. Thesis, S.U.N.Y. at Buffalo, 1985. Sircar, S.; Zondlo, J. W. Fractionation of Air by Adsorption. U.S. Patent 4,013,429,1977. Suh, S. S.; Wankat, P. C. A New Pressure Swing Adsorption Process for High Enrichment and Recovery. Chem. Eng. Sci. 1989,44, 567. Tsai, M. C.; Chen, W. N.; Cen, P. L.; Yang, R. T.; Kornosky, R. M.; Holcombe, N. T.; Strakey, J. P. Adsorption of Gas Mixture on Activated Carbon. Carbon 1985,23 (2),167. Yang, R. T. Gas Separation by Adsorption Processes;Butterworths: Boston, 1987.
Receiued for reuiew June 1, 1993 Accepted August 9,1993.
Greek Letters e
= fractional void in the bed
XL = mass axial dispersion constant, cmVs Pb = density of the bed, g/cm3
e Abstract published
1993.
in Advance ACS Abstracts, October 1,
