Pergamon PII:

them. Engng Vol. 22. Suppl., pp. S637-S640, 1998 8 1998 Elsevier Science Ltd. All rights reserved Printed in Great Britain 0098~1354/98 $19.00 +O.OO SOO98-1354(9?3)00113-6

Computers

Short-cut evaluation of pressure swing adsorption systems

Yonsoo Chung , Byung-Ki Na, and Hyung Keun Song
Korea Institute of Science and Technology 39-l Hawolkok-dong, Sungbuk-gu, Seoul 136-791, KOREA Abstract A short-cut method is proposed to evaluate the performance of pressure swing adsorption (PSA) systems. The basic steps in PSA systems are categorized into the adsorption and desorption steps. Each step is mathematically modeled under the assumption of adsorption equilibrium in a batch mode. It easily generates the relation between compositions of the feed and the product of the whole PSA system which can be used just like the x-y diagram in liquid-vapor equilibrium processes. It is also useful in the design of bed couplings through recycle and/or cascade of its product. It would be used as a screening tool in adsorbent selections and process comparisons at the early design stage for PSA systems. 0 1998 Elsevier Science Ltd. All rights reserved. Keywords: PSA systems, a batch equilibrium model, short-cut evaluation, x-y diagram, separation factor Introduction Pressure swing adsorption has been considered as an economical way to separate gas mixtures. The performance of PSA system has been measured by product purity and recovery in parallel with its economy. To obtain high purity and recovery, most PSA systems use some extra steps involving bed couplings through the recycle and cascade of its product in addition to the most basic PSA steps of pressurization, adsorption, depressurization and desorption (Yang, 1987). This interconnection of steps complicates the design of the PSA system and its operating procedures. It is thus important to develop a good sequence of steps and to evaluate its performance quickly and accurately in the early stage of design for PSA systems. To describe PSA systems in mathematical language, a set of partial differential equations has been usually used. The PDE model has been employed to describe the changes of compositions along the position inside beds and the operating time (Kaghavan et al., 1985). To describe the system rigorously, it requires various physical properties such as adsorption isotherms, heat and mass transfer coefficients, heat capacity, heat of adsorption, porosity and void diction of adsorbent, and many others. However, it is very difficult and impractical to measure and estimate all of these properties. This limits the use of the rigorous modeling and simulation of PSA systems especially in the early stage of design at which quick evaluation of the performance is required. To evaluate the performance of PSA systems quickly, Sorial et al. (1983) adopted the notion of the separation factor which was defined as the ratio of the mole fractions of two components on the surface of the adsorbent divided by the ratio in the bulk-gas. The separation factor was used to estimate the selectivity and to evaluate the performance of different adsorbents, but it could not describe the effect of different operating pressures on the system performance. This paper proposes a short-cut model which can easily describe characteristics of the PSA system including the effect of different operating pressures and quickly evaluate and compare their performances especially in the early stage of design. A batch equilibrium model To make a simple and relevant mathematical model for PSA systems which guarantees quick and easy solution without serious loss of accuracy, a batch system is assumed. The batch system consists of a vessel with fixed volume and an adsorbent fully packed inside the vessel. Only batch operations such as feeding gas mixture and adjusting operating pressures are involved. Equilibrium between two homogeneous phases, the bulk-gas and adsorbent phases, is always assumed. The operating conditions for all steps and the corresponding adsorption isotherms are given. In building the batch equilibrium model, the four basic steps of the PSA system are categorized into two different groups. The steps of pressurization and adsorption are merged into one group: the adsorption step. The other two steps of depressurization and desorption are grouped to another: the desorption step. Each group is modeled under the above assumptions.

??

To whom correspondence should be addressed.

S631

S638

European Symposium

on Computer

AIded Process Engineering--X vessel packed with the adsorbent; R the universal gas constant; T the operating temperature. a,, 6, and p are known constants. (1) and (2) show component mass balances and adsorption isotherms, respectively. Here, (2) has the form of the extended Langmuir isotherm, but any kind of equilibrium relations can be used in (2) if it reveals relationship between q, and Pod. (3) and (4) are compositions in the bulk-gas phase and feed. (5) shows the relation between the amount and the void volume inside the vessel packed with the adsorbent and (6) represents the concentration in the bulk-gas phase. From the above set of equations, all of the variables can be calculated with given values of POd T and W.

Fig. 1. Modeling procedure. Desorption Step The modeling procedure is schematically shown in Fig. 1. The left-top diagram in Fig. I shows the initial state of the batch. It is followed by the right-top which shows the feeding (pressurization) step. The right-bottom is the state of adsorption equilibrium resulted Tom pressurization. The adsorption step in the batch equilibrium model consists of these two With diagrams on the right hand side. depressurization of the vessel, a new state of equilibrium is reached as shown in the left-bottom diagram. The vessel is evacuated and its state is returned to the initial. This procedure is carried out repeatedly in the PSA system. Adsorption Step When a gas mixture of c components of which mole fractions are a priori known is fed into the vessel, each component is distributed into the bulk-gas and adsorbent phases. According to the amount of the gas mixture fed into the vessel, a corresponding equilibrium state is reached between the two phases. If the initial state of the batch is clean, the resulting equilibrium state can be mathematically described by the following set of equations. f, = If the pressure of the vessel is reduced Tom the above equilibrium state and kept at somewhat lower value, for example by adjusting a valve, a part of the gas mixture will be expelled from the vessel and another state of equilibrium will be reached at that lower pressure. The new equilibrium state can be modeled by the following set of equations.

m,+ 4,
r, =

= n, f r, + s,

i = l,...,c

(7) (8)

a,4

l+Cb,f:

i = l,...,c 1

<= -!L P& i = I.....cCn, P cm, LzL= Pde cfl,

4 -=St

(9)

i = l,...,c-

Cn,

cs,

(11)

qr =l+Cb,f
4 =

m,+ 9, a,4 w

i = I,...,c i = l,...,c i = l,...,ci = l,...,c1 1

(1) (2) (3) (4) (5) (6)

-!3- P,d
Cm, f,

zt =cI; w=pv Pod V = Cm, RT

where ni and r, are the amounts of component i in the bulk-gas and adsorbent phases at the new equilibrium The amount of the expelled state, respectively. component i and the new pressure are presented by s, and P+, respectively. Other variables are the same as in (1) to (6). The above set of equations is almost the same as (1) to (6) of the adsorption step. (7) to (9) are the same as (1) to (3) and ( 10) has the same meaning as (5) and (6). (11) shows that the expelled gas and the bulk-gas have the same concentration. In the above set of equations, m,, q,, Wand Pad are known variables from the set of (1) to (6) and the value of Pde is given. This allows solution of (7) to (11) for variables n,, r,, s, and P,. Note that if all of the remaining components inside the vessel can be obtained through evacuation, it will be the fmal product of the batch PSA system. x-y diagram of the PSA system The batch equilibrium model generates detail information on equilibrium states in the PSA system. With given pair of operating pressures, the amounts of all components in every phase and step can be

where m, and q, are the amounts of component i in the bulk-gas and adsorbent phases at equilibrium, respectively, andf; is the amount of component i in the feed. P, is the partial pressure; P,d the pressure at adsorption equilibrium; z, the feed composition; W the amount of adsorbents; V the void volume inside the

European Symposium on Computer Aided Process Engineering-8

S639

Especially the relation between calculated. compositions of the feed and the product is one of the most important information in the design of the PSA system just as the x-y diagram in liquid-vapor equilibrium processes. With the batch equilibrium model, the mole fraction of the feed is given as z, and the amount of the product is calculated as ni+r,. The x-y type diagram of the PSA system can be represented by the relation between zi and (n, + r, ) / C (n, + ri) . One example is shown in the diagram of Fig. 2. Fig. 2 shows the relation between mole fractions of the feed and the product of the PSA system in which binary gas mixtures of carbon dioxide and nitrogen is adsorbed into molecular sieve 13X at 1.1 bar and desorbed at 0.1 bar and operating temperature is always 298K. It has the same shape as the x-y diagram of liquid-vapor equilibrium processes. Each point in this diagram represents a pair of compositions of the feed and the product which can be a real design condition. If x-y diagrams are available for two or more adsorbents under the same operating conditions, it will be very helpful in selecting adsorbents and optimizing operating conditions. The x-y diagram of the PSA system is also usefitl in its design of bed couplings through recycle and/or cascade of its product. Fig. 3 shows the purity enhancement through the cascade of its product. The first feed of composition of zl produces the product of pl. And, if this product is used as a feed for the second PSA system in the cascade mode, i.e., 22 = pl, its product will have the composition of p2, and so on. This way allows prediction of the product purity. In the case of the product recycle, a similar procedure can be used to predict the product purity with the aid of the lever rule. This kind of analysis can generate very useful information on the optimality of the PSA system even at the early stage of design.

Purity vs. recovery In addition to the product purity, the product recovery is also an important factor characterizing the effectiveness of a PSA system in separating gas mixtures. This information is also obtained from solutions of the batch equilibrium model. Fig. 4 shows both changes of the product purity and recovery of a PSA system separating carbon dioxide and nitrogen mixtures according to the number of cascade beds. In Fig. 4, it is apparent that the molecular sieve 13X is better than activated carbon in separating carbon dioxide corn nitrogen. This information is very useful as a screening tool in adsorbent selections at the early design stage for PSA systems. Separation factors The separation factor used in Serial et. al. (1983) can be obtained from the solution of modeling equations for the adsorption step as follows
4,

a,/=-.mi

mJ qj

(12)

where au is the separation factor of component i referred to on i. (12) shows that this separation factor is calculated at a fixed pressure, that is, it cannot describe the effect of different operating pressures. Because the pressure swing is the major proposition for the separation in the PSA system, it is apparent that the genuine characteristics of the PSA system cannot be revealed correctly without proper consideration of pressure effects. This problem can be resolved by developing a methodology which achieves the relationship between compositions of the feed and product in the PSA

0.0

.2

.4

.6

.8

1.0

22
mole fraction of feed

mole fraction of CO2 in feed Fig. 2. An example of the x-y diagram.

Fig. 3. Design of the cascade process.

S640

European Symposium on Computer Aided Process Engineering-8

g? 90

120/
100 i

1
defined by (12)

; s E

80 70

1 {
I I f

8 60 50 $ 4o

(
20 i
0 defined by (13)

I
.2

0.0

.4

.6

.8

10

number of cascades

mole fraction of feed, CO2

Fig. 4. Effectiveness of the cascade design. system. As discussed in the above, this relationship can be easily formulated with the batch equilibrium model and it can be used as a real separation factor for the PSA system. If a new separation factor for the PSA system is defined as the ratio of the mole fractions of two components in the product divided by the ratio in the feed, it can be represented as follows

Fig. 5. a, vs. ZU .

model, the batch equilibrium model is much simpler and easier to apply to any PSA system and its results are accurate enough to characterize the system. Even though it cannot predict the performance of the kinetically-controlled system, it can be used as a screening tool in adsorbent selections and process comparisons at the early design stage for PSA systems. Notation a,, bi, p = constants [mole/Kg/bar, l/bar, Kg/m ] A = amount of component i in the feed [mole] m,, n, = amount of component i in the bulk-gas [mole] pl, p2 = mole fkaction of the product [0] P& Pd, = adsorption and desorption pressure [bar] P, = partial pressure [bar] q,, r, = amount of component i in the adsorbent [mole] R = universal gas constant [J/mol/K] s, = amount of the expelled component i [mole] T = temperature [K] Y = void volume of adsorbent inside the vessel [m ] W = amount of adsorbent [Kg] zl, 22, z, = mole fraction of the feed [0] aV, zq = separation factor [0]

(13)

where z,, is a newly defined separation factor.

This

separation factor reveals selectivity of the whole PSA system consisting of a pair of the adsorption and desorption steps while the separation factor of (12) just shows the equilibrium state at the adsorption step which is only a half of the process. Fig. 5 shows changes of separation factors defined in ( 12) and ( 13) according to the feed composition based on the same PSA system as in Fig. 1. It shows that a, overestimates selectivity compared to Z!, because a, is calculated only for

the adsorption step while ?Z,, is estimated for the whole PSA system including the desorption step.

References Conclusions Based on a batch system in equilibrium, a short-cut method has been developed to evaluate the performance of the PSA system. In the model, the basic steps of the PSA system are grouped into two simple ones. The batch equilibrium model provides detail information on the relation between compositions of the feed and the product which can be used in the design of PSA systems just like the x-y diagram in liquid-vapor equilibrium processes. It can also be used in the design of bed couplings through recycle and/or cascade of its product. Compared to the PDE Raghavan, N.S., Hassan, M.M., and Ruthven, D.M., 1985, Numerical simulation of a PSA system, Part I: Isothermal trace component system with linear equilibrium and fmite mass transfer resistance. AIChE J. 31(3), 385-392. Sorial, G.A., Granville, W.H., and Daly, W.O., 1983, Adsorption equilibria for oxygen and nitrogen gas mixtures on 5A molecular sieves. Chem. Eng. Sci. 38(9), 1517-1523. Yang, R.T., 1987, Gas Separation by Aakorption Processes. Butterworths, Boston, MA.