Catalyst Deactivation 1999 B. Delmon and G.F. Froment (Editors) 9 1999 Elsevier Science B.V. All rights reserved.
447
Effect of catalyst deactivation on the process of oxidation of o-xylene to phthalic anhydride in an industrial multitubular reactor W. Krajewski a and M. Galantowicz b a Polish Academy of Sciences, Institute of Chemical Engineering, ul. Battycka 5, 44-100 Gliwice, Poland b "K~dzierzyn" Nitrogen Works PLC, Central Research Laboratory, 47-223 K~dzierzyn-Ko~le, Poland Abstract
Kinetic parameters and deactivation constants are determined for a catalyst employed in the process of oxidation of o-xylene. Numerical results are presented for a multitubular reactor; these results have been obtained over several years of the catalyst operation. 1. INTRODUCTION Phthalic anhydride is produced on an industrial scale in the process of oxidation o-xylene with atmospheric oxygen over vanadia-titania catalysts. Modem catalysts are characterized by high activity and selectivity. However, the process itself is strongly exothermal, leading to an excessive temperature increase in the reactor and enhanced catalyst deactivation at the high temperatures [ 1,2]. The aim of the present study is to determine the kinetic parameters and the catalyst deactivation constants, and to describe the oxidation of o-xylene in a mutitubular reactor with the catalyst which has operated for a long period of time. 2. DETERMINATION OF KINETIC DEACTIVATION CONSTANTS
PARAMETERS
AND
CATALYST
The catalyst contains vanadium pentoxide, titanium dioxide in the form of antase and promoters. The catalyst is supported on ceramic semi-rings 10x6x2 mm. The catalyst studied is similar to the KVT 286 catalyst described in [3] and to the commercial catalyst presented in ref. [4]. Since all these catalysts display similar characteristics, the process occurring over the one studied in this work is described using the Langmuir-Hinshelwood kinetics. The introduction of the power or the Mars-Krevelen kinetics leads to less satisfactory results. The reaction scheme is as follows: where:XH- o-xylene TA 4 Pl P A - phthalic anhydride MA - maleic anhydride P C - products of complete combustion XH , PA TA - o-tolualdehyde P I - phthalide Into the reaction rate equations catalyst activity coefficients are introduced that describe the formation of phthalic anhydride and intermediate PC products, S l , maleic anhydride, s 2, and the
448 products of complete combustion, s 3. The reaction rate equations have the following form: r, =
xGL,
r, = s , k , , G , L
= s k ,GL,
r, = s, k 4 G L
rs = SlksX'ejL,
r6 = s, k6x ~ L
$
(1)
$
r7 = slkTxr~L,
r8 = siksxt, AL
where: k i = k ~ exp(-E)/RRT) , x s - mole fraction of the species at the catalyst surface, L - adsorption term defined by eq.(8) in ref. [4]. The mass and heat balance equations for a unit tubular reactor are the same as eqs (10)(15) in ref. [4]; the adsorption constants are given in Table 4 of ref. [4]. The catalyst activity coefficients are described by the following formulae sj = s~. exp(-kaat mj), j=1,2,3 (2) where the time t is raised to the power mj, and the deactivation rate constants are given by the Arrhenius equation kaa = k$~ exp(- Eda//RT , ~ )
j= 1,2,3
(3)
Formula (2) is similar to Erofeev's equation quoted in ref. [5]. The kinetic parameters of the process and the catalyst deactivation constants are determined similarly as in ref. [3]. The kinetic and deactivation constants are presented in Tables 1 and 2. As can be seen the numerical values are in part similar to those for the catalyst described in ref. [3]. Table 1. Kinetic constants for the individual reactions Reaction Reaction E~ ki~ No. lO/kmol kmol/(m3 h (0.1MPa)2 } 1
2
1 2 3 4 5 6 7 8
3
XH---~TA XH ---~MA XH ---~PC " TA ---~PI PI ---~PA XH ~ P A TAMPA PA---~PC
4
0.18980 0.22031 0.99018 0.55480 0.52560 0.12264 0.81760 0.10281
108443 96720 85415 85415 93789 108862 93301 85415
* * * * * * * *
1013 l0 ll 10 l~ 10 It 1012 1013 1012 1011
T a b l e 2. Deactivation constant s fo r the vanadia-titania catalyst
.
.
.
.
.
.
.
.
.
l~/kmol .
.
.
.
.
h-~ .
.
.
.
1
2
3
4
5
1 2 3
1.3634 1.4579 1.4398
219391.5 192990.6 201350.6
0.24742 *1016 0.53456 '1014 0.17834 *1015
0.350 0.220 0.270
449 3. THE PROCESS OF OXIDATION IN A MULTITUBULAR REACTOR The process is quite complex, as the cooling medium (a mixture of molten sodium nitrite and potassium nitrate) flows through a system of baffles and is gradually heated up. Consequently, the individual tubes of the reactor operate under different thermal conditions. The multitubular reactor studied (Fig. 1) has a diameter of 4 m and is 4 m high. It includes a central tube of diameter 0.96 m which contains a coolant circulation pump and a steam generator to remove the heat of reaction [6,7]. The reactor is equipped with 3 disc and doughnut baffles located at 0.5 m, 2 m and 3.5 m from the upper tube plate. The tubes 0.026 m I.D. are arranged in a hexagonal pattern with the tube pitch 0.04 m. The height of the cata;ytic bed is 2.35 m. The uppermost layer of the bed is located 1.55 m below the upper tube plate. The heat transfer in the intertubular space is described by the model put forward by Pignotti [8]. According to this model, between the baffles the coolant flows at right angles to the tube bundle, and is completely mixed in the window of the baffles. The heat balance for the cooling medium is given by equation (3)in ref. [7]: 8r~//~ = y r u
. x. ( ~ - ~ )
(4)
where: TA, TB - temperature of the reactants and coolant, respectively, X - dimensionless radius of the reactor. The process in the reactor tubes is described by equations analogous to those for a unit tubular reactor (see the preceding section and ref. [4]). The calculations were done for the coolant flow rate 1540 m3/h. The numerical methods described in refs. [3,7,9] were employed. The coolant flows countercurrently to the reacting mixture. Figs. 2a, b and c show the results calculated after one month of the catalyst operation. During that time the catalyst was still undergoing activation pretreatment; also, a low mole fraction of o-xylene in the feed, 0.0055, was used. The coolant inlet temperature was 651 K. The process occurring in the tube bank situated next to the central tube is denoted with 1, whereas that taking place in the row of tubes neighbouring the jacket of the reactor - with 2. Fig. 2a shows the profiles of the conversion degree of o-xylene (XH) and the yields of the individual products: phthalic anhydride (PA), maleic anhydride (MA), products of complete combustion (PC), o-tolualdehyde (TA) and phthalide (PI). The most vigorous process is observed in the initial section of the bed before the second concentric baffle. In the middle section of the bed the reaction proceeds further, and the conversion degree for o-xylene at the end of the bed is as high as 0.9999. There are minor differences in the course of the process between the individual tube banks. Fig. 2b presents the temperature profiles at the catalyst surface (squares), for the reacting mixture (solid squares) and coolant (solid line) close to the central tube (1) and the external jacket (2). The temperature maximum appears in the upper section of the bed, whereupon the temperature decreased slowly. There is also a small maximum of the coolant temperature close to the external jacket of the reactor. Fig. 2c shows the catalyst activity towards phthalic anhydride for the consecutive tube banks, starting from the central tube of the reactor. A strong decrease in the catalyst activity can be observed in the initial section of the bed. In the next rows the drop in activity is even higher. Similar activity profiles are obtained for the formation of maleic anhydride and the products of complete combustion. After four months of catalyst operation the mole fraction of o-xylene in the feed was gradually raised to 0.008447 (40 g/m 3 (STP)), and the coolant temperature was decreased to 636 K. Further operation was carried out at a constant o-xylene concentration in the feed, while the inlet coolant temperature was slowly increased so as to
450 maintain a constant o-xylene conversion degree of 0.9999 despite the progressive catalyst deactivation. After 73 months the inlet coolant temperature was 650 K. Fig. 3 shows the process after this period of time. As can be seen, the profiles of conversion degree, yields and temperatures have changed. Close to the front of the bed the process has been slowed down (Fig. 3a), while further down the bed it has again been speeded up to give high conversion degrees at the end of the bed. At the same time slightly larger differences are observed in the course of the process between the first and the last tube banks. Due to the increase in the coolant inlet temperature the initial section of the bed as well as that below the second baffle starts to operate vigorously. Well-marked peaks of temperature appear in these sections (Fig. 3b). This can be explained by the catalyst deactivation. It follows from Fig. 2c that, in the front section of the bed, the catalyst underwent strong deactivation. Over this area the process has been slowed down, and both the reacting mixture and coolant temperatures have decreased. As the time elapses, the zone of lowered catalyst activity spreads further along the bed. The calculations reveal a good agreement with the results obtained in an industrial reactor. In the process of oxidation of o-xylene in multitubular reactors the change in the catalyst activity with time has to be taken into account. Simultaneously, the operating parameters of the reactor have to be continually altered. A not too high flow rate of the cooling medium leads to irregular operation of the individual parts of the catalytic bed. 4. REFERENCES 1 V. Nikolov, D. Klissurski, A. Anastasov, Catal.Rev.-Sci.Eng., 33 (1991) 319. 2 M.Galantowicz, W.Krajewski, B.Wielowifiska, S.Karpifiski, in Catalyst Deactivation 1994, Studies in Surface and Catalysis, Vol.88, (eds. B.Delmon, G.F.Froment), p.591, Elsevier, Antwerp 1994. 3 M. Galantowicz, W. Krajewski, B. Wielowifiska, S. Karpifiski, Chemik, 47 (1994)409. 4 J. Skrzypek, M.Crrzesik, M.Galantowicz, J. Solifiski, Chem. Eng. Sci., 40 (1988) 611. 5 A. Bielafiski et al., in Vanadia catalysts for process of oxidation of aromatic hydrocarbons, (eds. B. Grzybowska-Swierkosz, J. Haber), p.33, PWN Warszawa 1984. 6 V.A. Nikolov, A.I. Anastasov, Ind. Eng. Chem. Res., 31 (1992) 80. 7 W. Krajewski, in 12th National Conference of Chemical and Proces Engineering, p.229, Poznafi 1986 8 A. Pignotti, Trans. ASME, J. Heat Transfer, 106(1984) 361. 9 W.Krajewski, W.Wielowifiska, M.Galantowiez,
t~
rt ~p0uA
Report, KBN Research Project No 312759101 (in Polish), IIChPAN, Gliwice 1994
Fig. 1. Multitubular reactor.
451
XT
a)
0.9 0.8 0.7 ' ~ o.6
~ .~ ~" 0 > = 0 r~
PA
O.S O.4 0.3
0.2 o.1 P[
o ~ o
0.5
I
1.5
PC MA TA
2
2.5
=-~;. . . . . 2
2.5
750
b)
700
~
650
~600 550 500 ~ 0
c)
-
~ 0.5
1 1.5 Bed length, m
1.5
o
<
0.5
~ ~ "
Figure 2. The process of oxidation of o-xylene in a multitubular reactor after a period of I month, a) profiles of conversion degree and yields, b) temperature profiles, c) profiles of catalyst activity
452 1 0.9
a) r~
0.8
I// 2
91=) 0.7
9g,
0.6
~ .~
o.5
....... ~,,,,,,/,>,,,,>>~,PA
#.4,
0.4
.....
~ 0.3 > 0.2 o 0.1 r,.) 0
0
0.5
1
1.5
2
JL"I
PC MA TA 2.5
800
b)
750 700 650
~600 O
550 500
9
0
!
;
!
:
0.5
1
1.5
2
=
2.5
Bed l e n g t h , m
c)
1.5 <
1.o .~ 0.5 < '
%
,
Figure 3. The process of oxidation of o-xylene in a multitubular reactor after a period of 73 months, a) profiles of conversion degree and yields, b) temperature profiles, c) profiles of catalyst activity