Chemtcal En&eertng Scknce, Vol. 47, No. Printed in Gnat Britain.

17/l&

pp. 43154321,

1992.

tmos2509p2 ss.00 + 0.00 Q1992Per$amonPmuLtd

REACTOR REVERSAL

PERFORMANCE WITH PERIODIC FLOW FOR A MULTISTEP COMPLEX REACTION+

A. N. ZAGORUIKO and YU. SH. MATROS and

Institute of Catalysis, Prospekt Akademika Lavrentieva,5, Novosibirsk, 630090, Russia

V. RAVI KUMAR and B. D. KULKARNI* Chemical Engineering Division, National Chemical Laboratory, Fune 411008,
(Received

India

10 Jnnuary 1992; acceptedfor

publication

1 April

1992)

Ahstrati-Analysis of a triangular network reaction scheme taking place in a packed-bed reactor is carried out under conditions of periodic flow reversal. The analysis indicates that better conversions and selectivities for lower gas inlet temperatures are possible compared to the conventional unidirectional flow operation. The effects on reactor performanceas a function of cycling time for periodic flow reversal show the existence of an optimum value.

INTRODUCTION

Based on the results of steady-state and transient analysis of reactor models, many novel and intuitive ideas which could improve the overall reactor performance have been suggested and implemented in practice with considerable success [see e.g. McKarmin et al. (1988); Hoffman and Schadlich (1986)-j. The traditional mode of reactor operation has been to run it at a steady state, which yields the best conversion and selectivity. The optimal conditions can be obtained by evaluating the reaction kinetics and then optimizing for the main system parameters in the reactor model. Subsequently, a detailed steady-state analysis and a parametric sensitivity analysis demarcates safe regions for operation, where control will not be a problem. However, with the advent of technological improvements in reactor construction as well as control equipment, the new trend has been to operate reactors under transient ,conditions using suitable control strategies. These control strategies, which have been logically arrived at from the detailed analysis of reactor models, suggest that operation of reactors under transient conditions can not only improve the reactor performance greatly, but also reduce the costs of operation, energy and construction. Unlike the case of irreversible exothermic reactions, reactions following reversible and nonlinear mechanisms are best conducted at points away from equilibrium and high product concentrations so as to suppress the reverse rates. The fixed-bed reactor analysis of realistic mechanisms indicates that an optimal temperature profile over the bed 1engXh exists. To maintain this steady-state temperature profile, the reactor design would then involve a complicated
tNCL Communication No. 5349. *Author to whom correspondence

should be addressed. 4315

setup of heat exchangers, with the catalyst loading in the reactor brought down for the sake of control. These disadvantages can easily be eliminated by operating in an unsteady-state mode which allows for periodic flow reversal of feed. When the cold feed is introduced into a reactor heated up to reaction conditions, a temperature wavefront propagates towards the bed exit at a velocity much slower than the fluid. The rate of propagation of the temperature field depends on system parameters such as its reaction kinetics, activation energy, heat capacity, thermal conductivity ofthe bed, external operating conditions, etc. At an intermediate time, the temperatures at the two ends of the bed can be lower than in the middle section, establishing a temperature profile in the bed, which at a particular time, will correspond closely to an optimum temperature profile. As the wavefront crosses over this optimum profile, reversal of the direction of fluid flow causes the wavefront now to progress in the opposite direction where it crosses the optimum once again; whence the cycle is repeated. After the initial transients are over, a definite cyclic regime is established. Such strategies have been successfully implemented for a commercial plant for SO2 oxidation and some other processes (Boreskov and Matros, 1984; Matros et al., 1988; Matros, 1988). They are seen to be very cost-effective. The aim in the work presented here has been to study the performance of a multistep reacting system with periodic flow reversals. To do so, we study the behavioral patterns of a representative mathematical model formulated for the specific case of a complex reaction taking place in a packed-bed reactor under conditions of periodic flow reversals. To facilitate the evaluation of the system performance, the analysis focuses its attention on system conversions and selectivities and compares them with those obtained employing conventional steady-state operation. The

4316

A.N.

ZAGORUIKO

et al.

results do seem to indicate that this operation strategy can be utilized to optimize the overall efficiency, with possible improvements in conversion-selectivity characteristics. This paper is organized as follows. The following section details the mathematical model and an appropriate formulation for analysis. Subsequently, the results obtained on analysis of the model are described. The final concluding section summarizes the scope of the work carried out.
MODEL FORMULATION

r3 = k, exp (-

EJRt?)

.i?,eiz3 - e

epc?
P >

(6)

Typically, mechanisms for industrial catalytic reactions are written as a sequence of possible process steps. To study the behavioral patterns for a reactor operating under periodic feed reversal conditions, we shall, therefore, begin by considering a representative complex reaction network which has features required for studying the conversion and selectivity behavior. Let the mechanism be A+v,,BI:S+D A+v,,Bi:E+D A + v,,E%,,S + D (-AH (-AH (-AH = = =

In eqs (4)-(6), k,, j = 1, 2, 3, are the rate constants, e,, i = 1, 2, . . . , 5, represent the concentrations of the specified species at the surface of the catalyst and fI is the temperature of the catalyst. The other constants have been defined in the notation. The model equations employing the above rate eqs (4)-(6) for the reaction taking place in a catalyst bed can now be appropriately formulated. Retaining simplicity so as to facilitate interpretation of results, we shall make the following assumptions:

(1) The flow pro8le in the reactor corresponds to a

Q1) Qd Qd

(1) (2)
(3)

where A and B are reactants and S, D and E the products. It may be noted that step (3) is considered to be a reversible step and in the analysis to follow we shall assume that the forward-reverse rates follow equilibrium kinetics. Also, all three steps (l)-(3) are considered to be exothermic with Qt representing the heats of reaction of the ith step. If S is the desired product, then, the formation of unwanted product E in steps (2) and (3) lead to a decrease in selectivity to S. Therefore, it would be desirable to formulate strategies which can enhance both the conversion and selectivity to S. It may be noted that the gross features of the above steps (l)-(3) may be viewed as a mechanism where the products E, S of two competitive parallel reactions from B are linked by reversible reactions of the type shown below: E

plug-flow reactor with the presence of mass and heat transfer resistances between the gas phase and the catalyst. The rate eqs (4)-(6) qualitatively describe the overall performance of the reaction and, hence, no other features like intraparticle effects have been considered. (2) Concentrations of reactants in the inlet mixture are maintained low to minimize the presence of unusual heat effects. In such a case volume change effects due to the reaction may also be neglected and have, therefore, not been considered. (3) Since periodic feed-reversal strategies alleviate the necessity for heat exchangers, the reactor is adiabatic. With the above assumptions, the mass and heat balance equations for the packed-bed reactor are: Mass balance: ac, u -ji + BIL#(CI /VL(cr &) = 0 (i = 1, 2, . . . , 5) (7)

&I) = C vijrj (8)

(i = 1, 2, . . . , 5; j = 1, 2, 3) Heat balance: CPU g + cxS,,,(T e) = 0

(9)

/ii B\
-Li Let us for the sake of notational convenience denote the concentrations of various species in steps (l)-(3) by A = cr, B = c2, S = cj, D = cq, and E = cs. It is assumed that the rates of the individual process steps, ri, can be described by the following equations:
r1 = kl

(i = 1, 2, 3) Initial and boundary conditions: At At For I=& T= Thp,

Ci = Cpp,m/ai=O

(IO)

(11)

I = L: &%/al

= 0

(12) (13)

r = 0: e(Z) = etitr(Z).

To allow for periodic reversal of gas flow direction at time intervals of 7 = z,, we may reinitialize the process states for 1o [O, L] : At ~=nz,:8+(1)=8-(L-Z) (n=1,2,3 ,...) (14)

exp(-

E,/R@e,e~;t E,/RB)C,t?,

(4) (5)

r2 = k, exp (-

Reactor performancewith periodic flow reversal where the superscripts + and - refer to the two different directions of feed flow. If Iz = 0, the effect of axial heat conductivity can be neglected and eq. (7) can be recast as (1 -e)r~=crS,(r-8)+CQ,~~ (j=I,2,3) (15) with the boundary being simplified to: At conditions [eqs (11) and (12) J c i=Ct
imp

4317

400

la

I=O:T=T P

(16)

To study the system behavior the coupled set of partial differential equations [eqs (4)-(g) and (15)] with conditions (13), (14) and (16) can be solved numerically to yield the ci, &, T and 6 profiles as a function of axial distance I and time z for varying operating conditions. To quantify the characteristic conversion and selectivity of the process, we also define the following quantities for specified values of the axial distance, namely, Overall conversion of A X,(z) = 1[cp;/cP] (17)
x

SO!
000 O-M

0.40

O-60

o-00

l-00

l/L

o-soo-00-070O-60-

-_--

____

---

_____

--___--Tsc~---

..-

where X,(z) identifies the instantaneous conversion of substance A. The time-averaged conversion of A, namely XA, within a flow reversal time-cycle period, z,, is then given by X, = [l/r,] = X,(z)dr s0 (18)
O-00 o-20 040 O-60 0.80 t-00

Overall conversion of A to S Xs(7) = cs*/c ip (1%

l/L Fig. 1. Calculated (a) catalyst temperature(0) and @) conversionto S (X,) as a function of axial distancein the reactor for r, = 475 s and Tbp = 70%. (-)forr=O;c-.)forr = OJr,; (- . -) for 7 = s,; and (- - -) cycle-averagedconversion to S. included varying the cycle duration r, in the range of 300-800 s with a view to studying the effects of cycle duration z, on system behavior. The conventional steady-state solutions of the reactor operating in the absence of feed reversal (i.e. r, + 00) have also been calculated and superimposed on the figures. Thus, a comparison of the results with respect to cycle-aver.. aged conversion of A [eq. (18)], A conversion to S [eq. (20)] and selectivity to S [eq. (22)] in both the presence and absence of feed reversal becomes possible. Reactor pro&s of catalyst temperature, 0, and conversion, Xs, along the length of the bed for an inlet temperature of 70C and r= = 475 s is shown in Fig. l(a) and (b). We observe from Fig. l(a) that the temperature profiles at the beginning as well as at the end of the cycle are symmetric suggesting that the operation has stabilized. The location of the developed hot spot at the middle of the cycle (0.5~~) is seen in Fig l(a), while Fig. l(b) shows the A to S

where Xs denotes the instantaneous conversion of species A to S. The corresponding expression for the cycle-averaged conversion to S is then given by & Selectivity The instantaneous selectivity to product S can be defined as LP(T) = X,(r)/X,(t) and the average selectivity as s = X,/X*.
RESULTS The AND DISCUSSION

= [l/r,]

*=X,(z) dr s0

(20)

(21)

(22)

results obtained on solving the model equations after eliminating the initial transients are presented in Figs 1-5. Table 1 summarizes realistic values of the parameters and the rea$or operating conditions used in the simulation studies. Calculations were carried out for inlet gas temperatures (Tm) ranging from 40 to 1SOC. Studies for each Timp

4318

A. N.

ZAWRUIKO

et al.

400.9 _ B 300-

200-

O-60

--~----_---__~--___
100

i
350 G 400 9 UC 450 500

0.40

J-oao-

0.30

! 300

1_----__--___-_____1
350 400 Tc , set 450 500
I

0.60

1
350 400 -r c ,sec 450 I so0

o-75 , 300

Fig. 2. Calculated (a) catalyst maximum temperature (P), (b) cycle-averaged conversion of A (z), (c) cycle-averaged conversion to S (xs), and (d) selectivity to S (pd) as a function of 7, for an inlet temperature T = 40C. () for unsteady-state mode; (- - - -) steady-state mode, TV-, co.

Table

1. Operating

conditions

and parameter

values for simulation

Inlet gas composition 3% (vol) A; 1.5% B; 95.5%

of inert gas (for example,

nitrogen) d,, = 5 mm; E = 0.4; y = 293 kcal mW3 K- ;

Catalyst bed parameters Residence time = 2 s; catalyst particles: S, = 6(1 - .z)/dp Gas flow velocity: II = 0.5 m s-l

spheres; diameter

Kinetics kI = 1.0 x 103; k, = 300.0 x 103; k, = 0.5 x 10; 2, = 0.5; vz2 = 1.5; vs2 = 0.5; VJ3 = 1.5;
El/K = 2000 K; E,/K

= 5000 K; E,/K a (Aerov

= 1850 K

Heat transfer coefficient.t a = N&./d&,;

et al., 1979)

Nu = 0.395Rt~~. =Pr~-~~; Re = ud~po/epo & (Aerov er aI., 1979) Sq = &p,D, kcal m- )

Mass transfer coefficients,T j$ = Sh,D,fd,;

Sh, = 0.395Re0.64Scf.33; (at normal

Heat effects of reaction

conditions,

Q1 = 2200; Q2 = 5500; Q3 = 550 Equilibrium K,,(o) constant* 1.038 In @

= 5910/8 + I.1310 x lo- 0

tCoefficients 01and #?[were calculated for a base temperature of 3OOC, with gas parameters &, pi, &, Pr) taken for pure nitrogen. *The temperatures employed to evaluate the constants for gas parameters, beat elects of reaction and equilibrium constant are in K.

Reactor performance

with periodic flow reversal

4319

Ix4
~+___,

300

400

500

600

700

600

..7J_____

400

500 Tc

600 ,sec

700

600

Tc , set

______---------

--_-

0.65 0.60 0.55

0.65

3&

400

500

600

700

800

0.75

300

4&

500 Tc,

6th set

7bO

600

Tc , set

cycle-averaged

Fig. 3. Calculated (a) catalyst maximum temperature(@m X ), (b) cycle-averaged conversion of A (xA), (c) conversion to S (x&, and (d) selectivityto S (sse ) as a function of r, for an inlet temperature for unsteady-statemode; (- - - -) steady-state mode, 2, + cc. q = 100C. (-)

conversion profiles at the beginning, middle and end of the cycle. A plot of the average conversion to S during the cycle is also shown in Fig. l(b). Figures 2-4 show the results obtained on varying the inlet gas temperatures (namely, 40, 100 and 15OC). For each plot the results are graphically portrayed in terms of the cycle maximum catalyst temperature (P =), cycle-averaged conversions of A (XA), A to S (r3,) and selectivity to S (s = ). From Fig. 2(b) we note that in the conventional steady-state operation for an inlet temperature of 40C the residence time is not sufficient enough to obtain chemical equilibrium at the outlet of the bed. Thus, the maximal outlet conversion is low, although the selectivity to S is nearly 100% CFig. 2(d)]. In unsteady-state operation since heat is pumped to the bed by periodic flow reversals, the temperature profile of the catalyst in the bed is higher [deduced from Fig. 2(a)]. The reaction rates, therefore, correspondingly increase compared to those of steady-state operation. Thus, periodic-reversal of feed allows the system to approach thermodynamic equilibrium and yield higher conversion levels [Fig. 2(b) and (c)] for this inlet gas temperature. The best approach to cquilibrium however, is seen to depend on the cycle time, xc. With a cyclic feed reversal strategy we observe that the hot-spot temperature decreases with increase in

cycling time z, [Fig. 2(a)] and for a critical value of r, there exists an optimum B which allows the system to perform close to equilibrium rates. Thus, in the lower range of r= values both conversion and selectivity rise with decrease of fPnx. For cycles longer than 430 s conversions become controlled more by kinetics than by thermodynamics and xA, xs decrease [Fig, 2(b) and (c)] even though P increases consistently [Fig. 2(d)]. Similar results were obtained for T*p = 70C and are therefore, not presented. For a higher Tinp = lOOC, the achievable temperature profile during steady-state operation is closer to the value which provides high enough reaction rates for reaching equilibrium conditions [comparing Figs 2(b) and 3(b)]. The gain in xs for unsteady-state operation when compared to the steady-state mode is not substantial and could be realized in a limited range of 530 < zc < 780 s [Fig. 3(c)]. The B for unsteady-state operation is once again seen to be higher than that for the steady-state process. On increasing Tinp to 150C we observe that this temperature is high enough to achieve near-equilibrium conversions for the chosen residence time even under steady-state conditions. Therefore, the conversions achieved become thermodynamically controlled [Fig. 4(a)-(d)]. Due to this fact, at high Tinp, conversion and selectivity in steady-state regime are higher

4320
620,

A. N. ZAGORUIKO et al.
0.88 -p____-_-_-_-____ -_____

b O-6444oc!! i; 360-m -______ -___________ O-72 -I

zy:f
O-85

c ,-=

0.60

__----_--__-__---_F
=

0.92-

0.84-

300

400 r,,

ezoo S6C

eoo

700

Fig. 4. Calculated (a) catalyst m_aximum temperature (P ), (b) cycle-averaged eonversion of A (xA), (c) cycle-averaged conversion to S (Xs), and (d) selectivity to S (s ) as a function of 7= for an inlet tempeiature ) for unsteady-state mode; (- - - -) steady-state mode, r, + co. & = 15o C. (-

o-so

1
_L---__

conditions operation steady-state tice. Finally, unsteadyfor a range For time.

with closer

lower to

inlet

temperatures desirable zs

facilitates by in pracfor both

equilibrium feature the

conditions than

0.60 / /

operation-a Fig. 5 shows and of each

O-70

achievable operation

steady-state

in the reactor

TbP values
Tiwp the

and the chosen residence maximum conversion at-

tainable in the unsteady mode were obtained by using an optimum value of T,. It can be seen that in the region unsteady loaded Higher periodic of low inlet gas temperature a more (-c 100C) the process provides e&lent use of the

catalyst compared to steady-state operation. A conversion to S is possible for an identical pumping by the flow-reversal strategy. CONCLUSIONS

residence time due to the heat

Fig. 5. Maximum conversion to S (xs) as a function of Thp for unsteady-state mode ( -) and steady-state mode (- - - -) of operation.

The

present

work

studies

the

properties

of

the

model catalytic reaction under the influence of periodic flow reversal. Unsteady-state mode of operation at low inlet temperatures can improve tainable conversion. From the practical point this result suggests used for producing the atof view,

than

in unsteady-state

mode,

suggesting be

that

operaat

tion in the unsteady mode may higher inlet gas temperatures.

unwarranted

that unsteady-state process can be S from A-containing gases with

This set of results with varying inlet temperatures under unsteady-state allows us to infer that operating

better results than a conventional steady-state process, with advantages both in conversion to the de-

Reactor performance with periodic flow reversal

4321

sired product as well as energy consumption in heating of the inlet gas. For high inlet gas temperatures the process becomes thermodynamically controlled and strictly from a performance viewpoint steady-state mode of operation may be preferred over periodic feed-reversal strategies. The trends in the results are, however, encouraging and show that significant advantages may be realized by implementing this strategy in the operation of reactors.

1.4

gas flow velocity in the bed under normal conditions, m s- 1

Acknowledgements-This work has been carried out under the auspices of the Indo-USSR cooperation programme sponsored by the State Committee of Science and Technology, USSR, and the Department of Science and Technology, India, on Mathematical Modelling of Catalytic Reactors and Process Design. The collaborators of this work wish to acknowledge the funding support of the individual departments in carrying out this work.

NOTATION
=i 1RP

COU
1 =i

ci
CP

dP

Di
I L

K*

Q, i

Ld
T T P

ith-component concentration in the gas phase at the inlet of the reactor ith-component concentration in the gas phase at the outlet of the reactor ith-component concentration on the catalyst surface (mole fraction) ith-component concentration in the gas phase gas-phase volume heat capacity at constant pressure, kcal m - 3 K - 1 catalyst particle diameter, m diffusivity of species i, m2 s- equilibrium constant for step (3) axial distance, m length of bed, m heat effect of jth reaction, kcal mv3 observed jth reaction rate, s- outer surface of catalyst per unit volume,
m -1

Greek letters heat transfer coefficient, kcal mm2 s- K- ith-component mass transfer coefficient, ms- catalyst volume heat capacity, kcal mW3K- 1 Y & bed porosity catalyst temperature, K e e- maximum catalyst temperature r,, K A. effective heat conductivity coefficient of the catalyst bed frame, kcal m- 1 s- 1 K- 1 heat conductivity coefficient of the reaction gas, kcalm- s- K-l stoichiometric coefficient of ith component ii in jth reaction gas-phase density, kg mm3 PO 7 time, s cycling time for periodic feed-reversal direc=, tion, s
REFERENCES

Aerov, M. E., Todes, 0. M. and Narinsky, D. A., 1979, Avvaratus with Fixed Bed of Grain Material. Hvdraulic a2 Heat Basis of Work. Khimiya, Leningrad. _ Boreskov. G. K. and Matros, Yu. Sh.. 1984. Fixed catalvst bed rea&ors operated in steady- and-unsteady-state c&itions, in Recent Advances in the Engineering Analysis of Chemically Reacting Systems (Edited by L. K. Doraiswamy), pp. 142-155. Wiley Eastern, New Delhi. Hoffman, U. and Schadlich, H. K., 1986, The influence of reaction orders and of changes in the total number of moles on the conversion in a periodically operated CSTR. Chem. Engng Sci. 41,2733-2738. _ -

Matros, Yu. Sh., Noskov, A. S., Chumachenko, V. A. and Goldman, 0. V., 1988, Theory and application of unsteady catalytic detoxication of efHuentgases from sulfur dioxide, nitrogen oxides and organic compounds. Chem.
Engng Sci. 43,2061-2066. Matros, Yu. Sh., 1988, Catalytic Processes under Unsteadystate Conditions. Elsevier, Amsterdam. McKarmin, M. A., Schmidt, L. D. and Aris, R., 1988, Forced oscillations of a self oscillating bimolecular surface reaction model. Proc. Roy. Sot. Lmui. A417, 363-388.

gas temperature, K inlet gas temperature, K