Ckmical~~~ scirnrr Vd. %. pp IVIZB P-on Press Ud.. 1911.

Printedin Great Britain

AN ANALYSIS SORPTION

OF A NONISOTHERMAL ONE-COMPONENT IN A SINGLE ADSORBENT PARTICLEA SIMPLIFIED MODEL

A. BRUNOVSKA and 3. ILAVSKk

Department

of Organic Technology, Slovak Technical University, 880 37 Bratislava, Jgnska Street I, Czechoslovakia

and
V.

HLAVACEK
166 28 Prague 6, SuchbLtarova Street

Department of Chemical Engineering,

Institute of Chemical Technology, 1903, Czechoslovakia

(Received 21 December 1979; in revised form 24 April 1980; accepfed 19 May 1980)
Abstract-h

this paper on approximation is described making it possible to calculate temporal temperature and adsorbed amount profiles during an adsorption process in a single pellet. The suggested model includes heat transfer in an external film and mass transfer inside a porous structure. The profiles calculated from the approximation and rigorous model are compared and it is shown that for a majority of operating systems the simplified model can be used and approximates very well the results obtained from an exact model.

INTRODUCTION

The model developed by us recently[ 1,21 takes into consideration both external and internal gradients of temperature. A detailed numerical analysis of the mode1[1,2] as well as experimental observations[3] indicated that the internal temperature gradients are low and may be neglected, i.e. the pellet is supposed to operate isothermally. Moreover, the complete model results in a very high computer time expenditure. We wish to present in the following section a practical simpler model, which may be used for describing of most sorption processes occurring in a single pellet. The suggested model takes into consideration the heat transfer resistance in the external film and mass transfer resistance in the porous structure. The goal of this paper is twofold (i) to study the applicability of this model and (ii) to demarcate domains of its validity in important practical cases.
2. MATHEMATICAL MODEL

j nRpC,

= hR=D

) x-R

(-AH.,)

-47rPh,(TThe initial and boundary conditions are:

t=O,O~xaR:a=a~,T=~, r>O,x=R

To).

(2)

: a = a*(co, T),
:$o.

(31

x=0

For a general equiIibrium isotherm c = c(a, T) eqn (1) may be rewritten:

(4) Equations (2)-(4) may be rendered dimensionless:

Following approximations are introduced (1) Intraparticle mass transfer may be described by means of the Ficks law, the diffusivity is assumed concentration and temperature independent. (2) One component system is considered. (3) Heat transfer resistance is concentrated in the external fihn. (4) Constant average heat capacities characterize the system, heat of adsorption is assumed independent of concentration and temperature. (5) There is an equilibrium between the gas phase and solid phase concentrations everywhere in the particle. (6) Only the gas phase diffusion is considered. For the assumptions mentioned above we can write:

(6)
and since

Q = p(q)Jl(f+ - K*

g=

3&9_,

-0e

(8)

(1)
123

7=o,o<[Sl:

q=O,B=O,

124

A.

BRUNOVSKA et

al. The non-

rO,~=: q=~+K,(l-++K*-K*~)

It K*-KZ$

cedure was used as proposed by us earlier[l]. linear mass balance was rewritten as (9)

f=O:$pO.
Here we have denoted

Here the coefficients A3 and A4 are defined as follows:

A, = cpll

dco- cf) -,Le=

nf-a;

A -,Bi=++
PW

3LeBie 6 .

For the Langmuir isotherm

To solve the linearized parabolic equation (14) the Crank-Nicholson finite-difference scheme was used. After approximation of eqn (14) by finite-differences following equations result: [$-$I *i-l.,+, + [++$I - [$+$I &.,+1
qitl.j+l+

ai (a, - a8 B =(-A&) K, =Iat-a. Tipc, a, - a, K2= a,(08 - a,)

(4 - al), a_ (-AH& Rx
+~ej+,=[~-~]4i-l.j+[~-~]4..j

+ [g&q4j+.i++%j.

(15)

Evidently, the solution of this model depends only on the value of the product LeBi and not on the individual values of Le and Bi
3. NUMERICAL SOLUTION OF GOVERNING EQUATIONS

For the center of the particle eqn (14) is A d4+A

a7

de=,, dr

& 3 ap results in:

(16)

Equations (7)-(9) represent a coupled set of a nonlinear parabolic partial differential equations and an ordinary differential equation. The proposed finitedifference scheme, which is similar to that described for a complete model[l], is based on a combination of explicit and implicit procedure. After denoiing the r.h.s. in eqns (7) and (8) rl and r2 respectively the time derivatives may be replaced by forward differences
A, %.i+ y

The finite-difference
[$+%I

approximation

q*.j+l-$$q*.j+l+$Oj+l

[gqq,,,++&2.,++.

(17) of (8) yields:

The finite-difference

approximation

4i.i + **

ej+, kr2.

0, = r,

(0)

@j+l

Oi=

(11) Here we have denoted As = 3cp& In the implicit finite-difference scheme the coefficients AI-A5 are evaluated at the point (i, j t l/2) and are iteratively improved. The finite-difference equations may be written in a concise form: xx I- xxx
xxx

Here we have denoted Al = 1 + &#& At = &#. The set of explicit finite-difference equations may be easily solved for qi./+, and O,,,
qi.j+1 = qi.i +

(rl - A& +

(2)
(13)

9,+, = 8, + kr,.

x
X

XX X

The nonlinear right-hand sides r, and rz and coefficients A, and AZ are evaluated at the point (i, j). The explicit finite-difference formulas (12) and (13) are used to calculate a first guess of a new profile j + I. The calculated values must be improved by an implicit finite-difference procedure. To construct a feasible implicit finitedifference scheme the same type of linearization pro-

.*.. . . .. . . . . .. *. . . . . .. .. . . . . .

xxxx

X X

xxx
xxx

x-

An analysis of a nonisothermal one-component sorption The

125 AND APPROXIMATION

MODEZ

occurrence matrix can be rearranged to

xxx

4. COMPARISON OF
In

RIGOROUS

Figs. l(a)-3(b)

a comparison

of

temporal

tem-

perature and adsorbed amount dependences is shown.

The values of the parameters are the same as in our former paper [ I]. Since in the approximation model the parameters Le and Bi do not enter in the description individually but as a product Z_Q X Bi we can easily follow the effect of both these parameters on the quality of the approximation if the value of the product J+ex Bi is constant. Our calculation have been performed for cL.e X Bi = 0.05. Based on the numerical calculations and for the temperature of the ambient gas 70C the over-

xxxx

xxxx .... .. ........ ..... ..... ......... XX-XX xxx xxx -. A special algorithm for band matrices was used to solve this set of linear equations[4].

0.7s

06. e

02-

O.l-

0+ 0

Olr

0.2

03

OL

05

08

o.7

Od

0e9

11

Fig. I(a). Adsorption in a spherical particle (cLe = I, Bi = 0.05, u = 10, p = 0.3, K, = 0.7, K?= 0, 6 = 0.05). Temperature dependence. approximation model; - - - - -, integral mean value calculated from rigorous model:-.: -, rigorous model 15 = 0); - . - ..., rigorous model (f = I).

04

03

0.2

03

OC

05

0.6

0.7

0.8

OS

1.r

Fig. l(b). Adsorption in a spherical particle. Adsorbed amount dependence. p, - -, rigorous model.

approximation model:

126
0.7

A. BRUNOVSK.~ et al.

04.

Fii.

2(a). Adsorption

Temperature

in a spherical particle (cLe=O.S, Bi= approximation model; - - dependence. -

0.1,

a = 10, @= 0.3,

K, = 0.7,

K? = 0, 8 = 0.05).

- -, integral mean value calculated from rigorous model; - - I -, rigorous model (6 = 0); - .. - .., rigorous model (5 = 1).

0.1 Adsorption

02

03 in a spherical

OA particle. -----.

05

0.6

0.7

0.8
-,

0.9

1.l
approximation

12
model;

Fig. 2(b).

Adsorbed amount dependence. rigorous model.

heating of the particle surface is -50C. From the figures may be inferred that the agreement between the rigorous and approximation model is very good. For a strongly nonisothermal particle (see Fig. 3a) the approximation model is capable of describing the temperature dependences with accuracy better than 10% with respect to the rigorous model. In this particular case the maximum temperature difference in the particle is (for T, = 70C)

AT = 30C. The agreement between the adsorbed amount dependences calculated from both models is also very good. A comparisoi of the computer time expenditure indicates that the approximation model results in a lower computer time. In the approximation model a longer time step may be used. In the example discussed the rigorous model requires a time step f I x lo- while for the ap-

An analysis of a nonisothermal one-component sorption

127

OR

Fig. 3(a). Adsorption in a spherical particle (cL0 = 0.05, Bi = 1, a = 10, fi =0.3, K, = 0.7, K* = 0, 6 = 0.05). approximation model; - - - - -, integral mean value calculated from rigorous Temperature dependence. model:-~-~-,rigorousmodel(~=O):-~~~-~~~,rigorousmodel(~=l).

0.1

02

0.3

Olr

05

0.6
I-

07

6.6

6.9

1.1

Fig. 3(b). Adsorption in a

spherical particle.

Adsorbed amount dependence. -, - .., - .... rigorous model (f = I).

approximation model:

proximation model a time step % I x lo- was sufficient. As a result the computer time for the approximation model is lower by a factor ten. 5. CONCLUSIONS For nonisothermal sorption processes occurring in a single adsorbent particle a simplified model was developed which approximates the temperature profiles within the particle by an integral mean value. This approximation yields very good results and can be used for calculation of nonisothermal sorption processes.
CES Vol. 36. No 1-l

Moreover the computer lower by a factor ~10.

time for a simplified model is

NOTATION

a8
a. Bi c

sorbate concentration in particle initial sorbate concentration in particle equilibrium sorbate concentration in particle monolayer capacity in the Longmun equation Biot number gas phase concentration

128

A.

BRUNOVSK~

et nl.

gas phase concentration in the bulk flow specific heat of sorbent diffusion coefficient heat of adsorption space step length, gas-to-solid heat transfer coefficient k time step length Le Lewis number Q dimensionless gas phase concentration 4 dimensionless sorbate concentration in particle R radius of particle T temperature Ti initial temperature t time X space coordinate
Greek symbols

dimensionless adiabatic temperature rise dimensionless parameter porosity of particle dimensionless temperature rise dimensionless parameters in adsorption isotherm effective thermal conducting of sorbent particle density of sorbent particle dimensionless space coordinate dimensionless time

REFERENCES

dimensionless adsorption energy

[I] Brunovska A. et al., Chem. Engng Sci. 1978 33 1385. [2] Brunovska A. et at., C/tern. Engng Sci. 1980 35 757. [3J llavskq J., Brunovskb A. and HIavQEek V.. Chem. EngngSci. in press. [4] NEmec J., Personal communication (Prague 1977).