Influence of some phenomena occurring on the surface and in the active phase of the vanadium catalyst on the reactor dynamics

9 1997 Elsevier Science B.V. All rights reserved. Dynamics of Surfaces and Reaction Kinetics in Heterogeneous Catalysis G.F. Froment and K.C. Waugh, editors

511

Influence of S o m e P h e n o m e n a O c c u r r i n g on the Surface and in the Active Phase of the V a n a d i u m Catalyst on the R e a c t o r D y n a m i c s Krzysztof Gosiewski Institute of Inorganic Chemistry, ul Sowiflskiego 11 44-101, Gliwice, Poland 1. INTRODUCTION During the simulation of the process of SO 2 oxidation, carried out by the present author for a number of years, it became apparent that certain modifications of the Boreskov equations [1,2], introduced following the procedure described in ref [3], lead to a representation of the effective reaction rate that is sufficiently accurate from the practical standpoint. It might seem, therefore, that there is no need to look deeper into the processes occurring in the active layer of a catalyst. This would indeed be the case if the problem could be reduced to calculating the oxidation rate in the mathematical model. The studies on dynamic models conducted by Gosiewski (e.g. [4, 5]) have revealed that the phenomena taking place on the surface and in the active phase of a catalyst, independently of the kinetics itself, can have a profound influence on the accumulation of heat in the catalyst bed. This conclusion has also been corroborated by experimental observations. In reference [5] a review has been presented of the dynamic models employed to describe the stationary catalyst bed. Independently of the degree of their mathematical complexity, the models used so far reduce the thermal capacity of the bed to a term that contains only the specific heat of the catalyst, with all other ways of accumulating heat omitted from the analysis. The present study is concerned with certain phenomena that take place on both the external and internal surface of a pellet of the industrial vanadium catalyst during SO2 oxidation. These are phenomena specific to this type of catalyst and reaction. We cannot exclude, however, the possibility of similar phenomena manifesting themselves, albeit on a different scale, in other catalytic processes. The oxidation of SO 2 over the vanadium catalyst takes place in the liquid melt phase of active species (mainly V20 5 and K2SO4). This observation, described extensively in the literature (e.g. [6, 7, 8]) seems to have been proved beyond any doubt, and will be discussed here only with regard to consequences it may have in the description of the bed dynamics. In recent years two papers have appeared [11,12] which support an earlier observation made by the present author, namely, that the phenomena occurring in the active phase of a catalyst can have a profound effect upon the dynamic parameters of the model. In this study the available information concerning this problem is reviewed and substantiated by the author's own measurements.

2. ACCUMULATION OF HEAT IN THE VANADIUM CATALYST BED EFFECTIVE THERMAL CAPACITY OF THE BED The absorption and desorption of gases from a porous catalyst surface, along with the dissolution of gaseous species in the liquid phase of the melt of active components, have a considerable impact on the effective thermal capacity (of. [4,5]). These phenomena affect the

512 measured accumulation of heat in the bed, as they lead to the increased uptake of heat from the heating gas when the bed is warmed up, and the enhanced release of heat when the bed temperature drops. In order to obtain an agreement between the time scales of the actual and simulated transients, a simple way of including these phenomena into the mathematical model has been proposed in ref. [4] instead of the specific heat of the catalyst, C k, a certain effective value C k eft is introduced that can be identified by comparing the actual dynamic transients in a reactor with those simulated using the model. It can be easily shown that, in practical situations, in the models discussed the time scale is affected only by the product (Pu Ck)- The suggestion that the time scale can also be influenced by the thermal capacity of the reactor body was dropped based on appropriate calculations which included this parameter (cf. Appendix A in ref. [4]). The ratio Ck eft / Ck at which the time scale predicted by the model becomes equal to that corresponding to the actual start-ups of industrial catalytic oxidation plants is Ck eft / Ck ~ 1 - for the warming up of a reactor with the hot air free of SO 2, at a temperature of up to 400 oC; Ck eft / Ck ~ 2 - for the warming up with the SO2-containing process gas, at a temperature above 400 ~ following the initial warming up with hot air; Ck eft / Ck ~ 3 to 5 - for the warming up with the SO2-containing process gas, during the start-up following short breaks in the reactor operation, before which no purge was carried out. This approach, which is formally equivalent to identifying an unknown (or difficult to determine) parameter has raised some doubts. It seems therefore reasonable to try to study this question in more detail. A short discussion of this problem has been presented in ref. [4]. It has to be noted that the heat of physical adsorption of SO 2 over silica gel at 273 K is about 34-103 kJ/kmol; for other gases it is in the range 5-100.103 kJ/kmol. The heats of chemisorption for gases are of the order 50-500 kJ/krnol. In ref [5] the heat effects that accompany these phenomena have been estimated as equal to at least 100 kJ/(1 mol) of the gas desorbing from the surface. The thermogravimetric studies carried out by the present author (cf. [5]) over the range 300 - 600 oC for several catalyst samples led to the change in the mass of a sample from 1 to 5% of the initial mass, and to the value of Ck eft from 2.33 to 4.8 J / (g. K). Similar studies of Br6tz will be discussed in section 3.1 3. ADSORPTION OF GASEOUS SPECIES IN THE LIQUID MELT PHASE OF THE ACTIVE CATALYTIC COMPONENTS 3.1 Thermogravimetric studies of Bri~tz et aL [10] Some interesting results of thermogravimetric studies were published by Br6tz et al. in 1979 [ 10]. It follows from these studies that samples of the vanadium catalyst pellets, heated up to above 400 oc and treated with gases of varying compositions (from the pure inert gas, N2, to a mixture of the inert with the active components, SO2/O2/N2) show the change in mass from 2 to 11% of the initial mass (depending on the composition of the gas used). The results presented in [ 10] provide a valuable source of information on the nature of the phenomenon. The change in mass of the activated catalyst can be observed even in these cases when the sample is treated with gases which do not react chemically (N 2 or the mixture O2/N2). An increase in mass at temperatures of 440 and 520 ~ was 2% for the pure N 2 and 5% for the mixture O2/N 2. These changes are, however, much more pronounced for the mixture SO2/N 2 (7 - 8%), and reach a maximum for the mixture SO2/O2/N2, i.e. the one which enables the oxidation to SO 3 to occur. The increase in mass observed in this case was 10 -11% of the initial mass of the sample. The studies were preceded by careful activation of

513 the catalyst, which consisted in the sulphatation of the activator K20 to K2SO 4. Thus, it seems obvious that the changes in mass do not result from the activation process itself. Of particular interest are the transients of change in mass of the samples recorded in the study of Br0tz et al.. [ 10], after the samples were subjected to a cyclic saw-tooth temperature variation from 375 to 575 ~ Such a variation leads to the cyclic changes in the catalyst mass, of the amplitude up to 4.6% depending on the composition of the gas mixture. These transiens reveal that the variation in mass is a totally reversible process and that it becomes significant only above the melting point of the active species, i.e. above 390 - 400oc. It seems certain that the phenomena found by Brotz et al. [10] cannot be explained otherwise than by the absorption (and desorption) of the gaseous species into a catalyst sample. The authors explain these phenomena by the dissolution of the gaseous components in the liquid melt phase. Although, in the context of the results presented in ref. [ 10] this is only a hypothesis, the possibility that this is indeed the case remains very high. 3.2

Discussion of the model of Bunimovich et al.

[11]

In ref. [ 11 ], Bunimovich et al. have made an attempt to include the process occurring in the liquid phase into the mathematical description of the bed. They stress an "extremely high capacity" of this phase with respect to SO 3, which leads to a high time constant for the accumulation of this component in the bed. The authors relate these phenomena especially with the absorption of SO 3 into the liquid phase of the catalyst, and support this idea by Gas

Surface reaction

SO [g/l~m2] 0.4 ,, " 9

b

, <35 ~

-\ 0,3 r \ X

07_

0,1S ' - ~ .

Fig. 1 Schematicdiagram of the reaction zone

:i

i

:

0,25,\

0,1

I

,

~

,

i

i

i

i ~

I

i

: I

i :

: '

; .

1

t i at the outlet!from pass iii ,,,y . . ~

.~N,

~ ' " , d. ~

!. ~.

a~ the 9udet frorri pass iV ~ i~/

Oi , I I 0 10 20 30 Chemical analysis results: 9

,~

I

III pass

9

! _ ~9

i

I

IA

40

50

Z, .

== 60 Time

70 [hours]

IV pass

Fig. 2 Contentof Sulphur oxides expressed as SO cor~centration at the reactor outlet during the purge of the bed. pointing to a very high difference between Henry's law constant for SO 3 and for the other gaseous components. In reference [11] the equations of a mathematical model are also presented; the model assumes, however, the absorption of all the gaseous species into the liquid phase. Also, the computer simulation results are Nven, with emphasis placed upon the accumulation of the mass of SO 3 in the melt. Bunimovich et al. do not discuss the associated heat phenomena although the model proposed in [11] includes the description of such phenomena. Despite some formal objections which can be raised with regard to this model, both the reasoning and results presented in [11] seem interesting, even if they contradict to some extent the expenmental results of Brrtz et al. [10]: if the absorption of the gaseous species in the liquid melt concerns only SO 3, then the change in mass of the catalyst, found in

514 ref. [ 10] even when SO 3 could not possibly be produced, becomes difficult to explain. It may be that some other surface phenomena come into play. The model does, however, require some discussion. If we assume that it is indeed SO 3 (i.e. the reaction product) which is mainly absorbed in the liquid phase, then the natural consequence of such an assumption should be the occurrence of the reaction mostly on the surface of the liquid phase (cf. Fig. 1) If, however, all the gaseous species are assumed to dissolve in the melt, then the reaction can take place in the bulk of the melt. 4. DESORPTION OF SULPHUR OXIDES DURING THE PURGE OF THE REACTOR BEFORE A ROUTINE BREAK IN OPERATION

Before switching off a catalytic SO 2 oxidation reactor to carry out routine overhauls, the reactor is purged with the hot dry air to remove sulphur oxides from the bed. The large mass (or molar) capacity of the catalyst bed is well illustrated by the fact that, in order to obtain the outlet concentration of sulphur oxides below 0.001~m 3, the reactor has to be purged for several dozen hours. To estimate this capacity the measurements were carried out of the outlet concentration of sulphur oxides during the purge. Typical concentration of sulphur oxides at the outlet from passes III and IV of an industrial reactor during the purging procedure are shown in Fig. 2. The first three passes were purged independently of the parallel pur~ng of pass IV. The reactor was purged with a total of about 40,000 m3(STP)/h of hot air (inlet temperature--- 440 ~ It is difficult to determine the actual molar capacity of the catalyst based solely on the measurements performed since, due to the heat losses during the purge, only pass I remained at a temperature above the melting point of the melt (about 400 oc). Moreover, during the flow of the gas through cold heat exchangers a large part of SO 3 blown out of the bed condenses to form H2SO 4 or even oleum and undergoes corrosive reactions. It follows, however, from the measurements performed that during the purge more than 250 kg of SO 3 is removed from the reactor, the major part of which probably originates from pass I which contains abut 13,000 kg of the catalyst; these amounts are therefore quite considerable. 5.

TIME DELAY OF THE MAXIMUM EMISSION OF SO 2 DURING THE

START-UP OF THE REACTOR

As Mishra and Sebastian [12] have pointed out, during the start-up of the reactor, following the heating of the bed from an external heat source (the so-called "cold" start-up) a well marked time delay (of about 45 - 60 minutes) is observed in the appearance of the maximum SO 2 emission at the outlet, compared with the transient predicted by mathematical models that neglect the absorption in the active liquid phase. The studies carried out by the present author [4] on the simulation of the "cold" start-up of an industrial installation also reveal a delay in the appearance of the maximum SO 2 emission by almost an hour. Although this phenomenon has been explained by the slow attainment by the sulphur burner of its full operational capacity, it is also possible that the conclusions presented in ref. [ 12] may, at least in part, be correct. It is suggested in [12] that delay is associated with the solubility of the gaseous species in the liquid melt (cf. Brotz et al. [10]) Mishra and Sebastian regard the dissolution of SO 2 (rather than SO 3, as suggested by Bunimovich [ 11 ]) as being of primary importance. If the melt absorbed only the product of the reaction, the delay in the appearance of the maximum SO 2 emission would be difficult to explain. In order to estimate the "molar capacity" of the bed that might produce an almost one hour's delay a number of simulations were carried out. The simulations were based on a dynamic model of a reactor bed, in which, together with the already mentioned effective

515 accumulation of heat, a term C m has been introduced which artificially allows for the accumulation of SO 2. The equations describing model of the bed are therefore

Pu Ck eff

~ Tk ~0 2 Tk + a S ( Tg - Tk ) + ( - AH ) r Pu & =2eft 0l 2

Cm

~Xsoo Ot -

= - ng

OXsoo Yl _ _ r . P u

(2)

(3)

Since the delay found in references [4] and [12] concerned the start-up of the whole oxidation plant that included a multi-pass catalytic reactor, the simulations were carried out using the model of the whole plant comprising not only the reactor itself, but also the installations for the recovery of the heat of reaction. For the assumed values of C m the startup was simulated from the state in which the installation was thoroughly cooled down. The SO 2 concentration transients at the outlet of the reactor are shown in Fig. 3. 6.

CONCLUSIONS The phenomena occurring on the surface and in the liquid active 0.0351 =0 phase of the vanadium catalyst have undoubtedly a major effect on the 0 . 0 3 0 ~ C m= 1 [ transients calculated by the dynamic models of catalytic reactors. The physicochemistry of these phenomena has not, so far, been satisfactorily elucidated. The introduction of an effective specific 0.010[ ~\ l___ [ heat and effective molar capacity of the bed as parameters to be identified by comparing the actual transients with those predicted by 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 Time [ hours ] the model is only a temporary measure. It cannot be expected that Fig. 3 Simulation of the SO 2 concentration at the the phenomena occurring on the outlet from an oxidation plant with a reactor packed with 82.4 m 3 of the vanadium catalyst, for the different values catalyst surface can be thoroughly of the mass accumulation, Cm and for explained without expensive and tedious studies. The results of the Ck eft = 2.1 kJ/(kg.K) present investigations can be summarized as follows: (1) The heat effects associated with the phenomena taking place in the active phase of the catalyst can lead to severalfold prolongation of the time scale of the dynamic mathematical model of the bed. (2) If the absorption (and desorption) of gases in the amount of about 5 10% of the catalyst mass (about 1 to 2 kmoles of gases / m 3 of catalyst) results from the change in temperature by 200 K, the accumulation of heat in the bed will change by several kJ / (kg- K). For instance, the heat effect of these phenomena expressed in SCb molefraction [- ]

~176176l/f >--l~ /

I t

516 terms of the effective specific heat will be 2.6 to 6.8 kJ / (kg 9 K) for the physical adsorption and 7.6 to 20 kJ / (kg- K) for chemisorption, with the physical specific heat of the catalyst as low as about 1 kJ / (kg- K) ! (3) Based on the thermogravimetric studies [10] it can be concluded that all the gaseous species are absorbed into the catalyst to a similar extent, although it is not clear whether this is due only to the dissolution of the gases in the liquid melt phase. However, it follows from these studies that this process is fully reversible. (4) About 4 0 - 60 minutes' delay in the appearance of the maximum emission during the "cold" start-up of an SO 2 oxidation reactor cannot be plausibly explained if only the absorption in the liquid melt phase is taken into account. This would require the "molar capacity" of the bed, C m, of about 10 kmol / m 3. At such an assumption, the amount of SO 2 retained in the bed during the first 40 minutes of the start-up would correspond to the change in mass of the catalyst of 8 - 10%. Therefore, it seems little probable that it is solely the absorption of SO 2 in the bed which is responsible for such a considerable shift in the maximum emission in the unsteady state. Nomenclature C~, specific heat of the gas [kJ/(krnol .K)] Ckeff -effective thermal capacity of the catalyst [kJ/(kg.K)]

c~ (-a~) l-

~g

-

r-

xi -

-

effective molar capaci~ for SO?

[kmol/m 3 ] heat ofreaction [kJ/kmol] distance coordinate along the bed [m]

S -

extemal specific surface area of the catalyst [m2/m 3] Tg, Tk gas and catalyst temperatures, respectively [K] t-

time [s]

or-

Greek letters individual heat transfer coefficient [kJ/(m 2- K-s)]

molar gas flux [kmol/(m2. s)]

~eff-

effective heat conductivity. K)]

effective rate of oxidation of SO9. to SO 3 [kmol/(kg 9s)]

Pu -

bulk density, of the catalyst [kg/m 3]

[W/(m-

mole fraction of component i

References 1. G.K. Boreskov., R.A. Buyanov, A.A. Ivanov, Kmet. Katal. 8 (1967) 153 - 159. 2. G.K. Boreskow, M.G. Slinko, V.S. Beskow, Khim Prom. 3 (1968) 13-18. 3. R. Sztaba, K. Gosiewski, Ir~. i Ap. Chem. No 6 (1981) 3 - 7. 4. K. Gosiewski, Chem. Engn and Proc. 32 (1993) 111 - 129. 5. K. Gosiewski, ~eria Chem. i Proc. 3, (1994) 393 -413. 6. H. Livbjerg, J. Villadsen, Chem. Engn. Sci. 27 (1972) 21 - 38. 7. G.H. Tandy, J. Appl. Chem. 6 (1956) 68-74. 8. G.K. Boreskow. W.W. Illaronov, R.P. Ozierov, E.W.Kildisheva, Zumat Ob. Chim. vol. XXXIV (1964) 23-29. 9. S.J. Gregg, The Surface Chemist~ of Solids Chapman & Hall Ltd London, 1951 10. W. Br6tz, B. SchOnbucher, H. Issler, Ger. Chem. Eng. 2 (1979) 108. 11. G.A. Bunimovich, N.V. Vemikovskaya, V.O. Strots. B.S. Balzh aev. Yu.Sh. Matros, Chem. Eng. Sci. 50 (1995) 565 - 580. 12. J.C. Mishra, M.T. Sebastian, Proc. Int. Conf. Unsteady State Processes in Catalysis USPC-1 Novosibirsk (1990) 659 - 664