Ind. Eng. Chem. Res. 1988,27,415-420
415
Intraparticle Diffusional Effects and Kinetics of Desulfurization Reactions and Asphaltenes Cracking during Catalytic Hydrotreatment of a Residue Constantine Philippopoulos and Nickos P a p a y a n n a k o s * Department of Chemical Engineering, Laboratory of Chemical Process Engineering, National Technical University of Athens, Patission Street, 42, GR-106 82 Athens, Greece
T h e cracking of asphaltenes and the desulfurization of asphaltenic and non-asphaltenic fractions of a residue were studied in a bench-scale, tubular, trickle bed reactor. Reaction conditions were 50 bar and 350-430 O C . Two types of commercial Co-Mo/A1203 catalysts were used. T h e experimental intrinsic rate data were described by the power-law rate equation for both cracking of asphaltenes and desulfurization reactions of the asphaltenic and non-asphaltenic fractions. T h e values of the effective diffusivity for each type of reaction have been determined from data derived from experiments carried out by using catalyst extrudates. T h e ratios of the effective diffusivities for the bimodal catalyst t o those for the unimodal catalyst were equal t o 2.9 f 0.1 for both cracking and desulfurization reactions. The effective diffusivities for cracking and desulfurization reactions of asphaltenes were identical for both catalysts used. The use of heavy crude oils and residues is increasing today due to continued depletion of petroleum resources. This tendency does not seem to alter in the near future, although substitute energy sources are being actively developed worldwide as a result of the foreseen shortage of liquid hydrocarbon resources. Heavy residues are usually characterized by the large content of sulfur and asphaltenes. The removal of sulfur from residues is necessary because of the emission restrictions of SO,. Asphaltenes are mainly responsible for the catalyst deactivation during the catalytic cracking process of heavy distilates or residues, as they greatly contribute to residue Conradson carbon. The use of deasphalted oils for feedstock to catalytic cracker has been practiced today. However, asphaltenes conversion to non-asphaltenic molecules during hydrotreatment will enable the efficient use of residues to produce lighter hydrocarbons. Consequently, the technology of upgrading heavy residues will be influenced by how asphaltenes are treated (Takeuchi et al., 1983). The knowledge of kinetics and effective diffusivities as well as of the influence of the catalyst pore structure on diffusional limitations permits the design of hydrotreaters and the selection of the proper catalyst. Furthermore, catalyst selectivity does not only depend on kinetics of each reaction but also on the relative diffusivities of the reacting molecules, when diffusional limitations exist. In this work we report on kinetic data for asphaltenes cracking and desulfurization of asphaltenic and non-asphaltenic fractions of the Greek atmospheric residue, using an integral trickle bed reactor, determine the effective diffusivities for both cracking and desulfurizationreactions, and investigate the effect of the different catalyst pore structure on the effective diffusivities of the different types of molecules present in the residue. Theoretical Treatment The differential mass balance equation for the trickle bed reactor, assuming bed isothermality and plug-flow pattern, is R dVR = -F dC (1) where R represents the reaction rate per reactor unit volume. When R is substituted for the intrinsic rate per catalyst unit volume R ', the mass balance equation becomes 0888-5885/88/2627-0415$01.50/0
€(I- [b)qR'dVR = -F dC
(2)
where C stands for asphaltenes and asphaltenic and nonasphaltenic sulfur concentrations in the residue and q for the effectiveness factor. The reaction rates for cracking of aephaltenes and desulfurization of asphaltenic and non-asphaltenic fractions are described by the power-law kinetic equation (Mosby et al., 1973; Sudoh et al., 1984; Papayannakos, 1986):
R' = k,Cu
(3)
In this model, the apparent rate constant k , includes the term of the hydrogen partial pressure, when this pressure is constant ( k , = k,'pH,B). The effect of the hydrogen sulfide presence in the gas phase on the rate constant due to H2S adsorption on the catalyst active sites is not distinguishable when hydrogen sulfide partial pressure is kept low ( -1. Equation 4 can, thus, be integrated by a numerical method and give C, for given values of k,, De, fb, V,, F , and V,/S,. When diffusional limitations are absent, 7 = 1,eq 4 gives
Cil-m- Col-a = €(CY - 1)(1- fb)k,(LHSV)-'
(6)
where (LHSV)-' = VR/F. Experimental Section A flow scheme of the experimental setup is shown in Figure 1. An integral trickle bed reactor was utilized, constructed from a stainless steel tube, 2.5 cm in inside 0 1988 American Chemical Society
416 Ind. Eng. Chem. Res., Vol. 27, No. 3, 1988 PR
i i
4Nv
Figure 1. Flow diagram of the apparatus used in this study.
diameter. The reactor was heated and controlled by means of four electric resistances placed outside the reactor tube, and the temperature over the reaction zone was kept constant within *15 "C. The reaction temperature was monitored with five thermocouples located along the length of a thermowell. The catalyst beds were supported on inert particles which had the same dimensions as the catalyst particles to avoid exit effects. Entrance effects were avoided by using the same inert particles a t the top of the beds (Levenspiel, 1972). A high-pressure metering pump forced the oil from the tank through the reactor to the gas-liquid separator. The liquid samples were withdrawn from the separator in constant time intervals. System pressure was maintained by the hydrogen cylinder regulator. Hydrogen in the reactor flowed downstream cocurrently with the liquid reactant, and its flow rate was controlled by a valve after the liquid separator. The reactor was operated isothermally, until steady-state conditions were attained. An experimental run consisted of determining the conversion of asphaltenes to non-asphaltenic molecules and the conversion of sulfur in asphaltenic and non-asphaltenic fractions of the residue. Sulfur concentration in residue, asphaltenic, and non-asphaltenic fractions was measured by an X-ray fluorescence analyzer. Asphaltenes were precipitated from the residue sample with 15 parts npentane (by volume) to 1part sample. The asphaltenes were filtered from the slurry after digestion for 4 h with stirring (Drushel, 1972). For all runs the pressure in the system was kept at 50 bar, the reaction temperature varied within the range 350-425 "C, and the liquid flow rates varied from 20 to 380 cm3/h. Hydrogen flow rates were maintained a t 40 f 3 nL/h, sufficient to keep the HzS content of the exit gas low in the reactor and thus to minimize inhibition effects by H2S. The decline of the catalyst activity, due to carbon and metals deposition on the active sites of the catalyst surface, was determined by carrying out periodically experiments at standard conditions. The operating parameters of these experiments were T = 350 "C, P = 50 bar, and (LHSV)-' = 1.0 h. The catalyst activity was considered as one, t = 1.0, after the first 2 h of operation of the catalyst. Kinetic experiments have been carried out after the first 150 h of operation of the catalyst. In the first 150 h of operation, a fast decline of catalyst activity was observed, followed by a period a t relatively constant activity beyond that, 0.11 It 5 0.15, for both catalysts. This can be attributed to a rapid coke deposition on the catalyst surface in the first period to a steady level of coke. Further deactivation is caused mainly by low-rate demetalization reactions.
Table I. Typical Properties of Catalysts and Catalytic Beds catalyst property CH (HT 400E) CG (G-51) coo, wt % 3.0 3.5 MOO,, wt % 15.0 10.0 A1203, wt % 82.0 86.5 specific surface area, m2/g 230 235 86 109 mean pore diameter, 8, total pore vol, cm3/g 0.50 0.63 pore size distribution monodisperse bidisperse micropore pore vo1," cm2/g 0.49 0.50 micropore predominant diameter, 8, 80 80 macropore predominant diameter, A 10000 bed of crushed particles mean particle diameter, mm 0.34 0.34 catalyst wt, g 101.9 102.7 catalyst vol, cm3 139.7 142.2 bed of extrudates particle diameter, mm 1.59 3.00 catalyst wt, g 121.9 102.4 catalyst vol, cm3 156.6 152.5 "The catalyst pores with radii of 15-150 A, arbitary termed micropores, were determined by Nz absorption.
The atmospheric residue (+315 "C) used in this study was obtained from the Greek petroleum deposits in the Aegean Sea. Its sulfur content was 5.0 f 0.2 wt 90, the metal (Ni + V) content was less than 20 ppm, and the asphaltenes content was 25 w t % (Papayannakos, 1986). The sulfur content in asphaltenes was 7.6-8.0 wt %. Thus, the asphaltenic sulfur concentration in the oil was C, = 1.6 X g/cm3, and the non-asphaltenic sulfur concentration was C , = 2.4 X g/cm3. The catalysts used were commercial ones and have been selected on the basis of their physical properties. The properties of the catalysts and of the catalyst beds are shown in Table I. Prior to experimentation, the catalysts were first reduced with hydrogen for 4 h a t 400 "C and 10 bar and then sulfided with 1.8% H2S in Hz for 4 h a t 350 "C and 50 bar. Measurements of pore size distribution indicated that crushed particles had retained the structure of catalyst extrudates. The oil density a t the reaction conditions has been calculated to be 0.8 g/cm3.
Discussion of Results (i) Kinetic Expressions. The obtained data for asphaltenes cracking are plotted according to eq 6 in Figures 2 and 3. As shown in these figures, the second-order kinetics with respect to asphaltenes concentration in oil, a = 2 in eq 3 and eq 6, was found to fit the calculation of the cracking reaction rates of asphaltenes for both catalysts. Sudoh et al. (1984) have also used the second-order kinetics for asphaltenes cracking. The feedstock used in their study was similar in properties to the one used in this study, but the properties of the catalyst were not given. In Figures 4 and 5, the correlation of the kinetic data for asphaltenic fraction desulfurization is attempted according to eq 6. The third-order kinetic equation with respect to asphaltenic sulfur concentration in oil, a = 3, was found to fit the desulfurization rates for both catalysts well. The same reaction order was determined for asphaltenic fraction desulfurization of the same residue when a differential trickle bed reactor and the CH catalyst were used (Papayannakos, 1986). The data for non-asphaltenic fraction desulfurization were correlated with the second-order kinetic equation with respect to the non-asphaltenic sulfur concentration in oil, CY = 2, as shown in Figures 6 and 7 according to eq 6, for both catalysts. The second-order reaction for the non-
Ind. Eng. Chem. Res., Vol. 27, No. 3, 1988 417
Loo
30-
LOO 'C
Y
h
-
a 350T
0
01
02
03
I LHSV?
Figure 2. Plot of kinetic data for asphaltenes cracking (catalyst CH; mean particle diameter, 0.34 mm).
Figure 3. Plot of kinetic data for asphaltenes cracking (catalyst CG; mean particle diameter, 0.34 mm).
asphaltenic fraction desulfurization of the same residue was determined by using a differential trickle bed reactor and the CH catalyst in experimentation (Papayannakos, 1986). The concentration of the asphaltenic and non-asphaltenic sulfur in oil was calculated as the product of the concentration of the fraction in the oil and the sulfur content in the fraction. The application of the Arrhenius expression k, = A exp(-E/RT) (7) where R is the gas constant, enables k, to be calculated at
04
0.5
06
07
08
E ,h
Figure 4. Plot of kinetic data for asphaltenic fraction desulfurization (catalyst CH; mean particle diameter, 0.34 mm).
Figure 5. Plot of kinetic data for asphaltenic fraction desulfurization (catalyst CG; mean particle diameter, 0.34 mm). Table 11. Intrinsic Rate Expressions for Cracking and Desulfurization Reactions Obtained from Experimentation with Crushed Particles reaction catalyst CH catalyst CG asphaltenes cracking k, = 1.3 X lo7 X k, = 9 x 1011 x exp(-14000/ T)" exp(-21000/ T)o asphaltenic fraction k , = 1.6 X 10l8X k, = 1.4 x 10l6 x desulfurization exp(-24500 TIb exp(-20oc10/ T ) b non-asphaltenic fraction k , = 1.2 X 10e! X k , = 1.2 x 1010 x desulfurization exp(-13500/ T ) c exp(-16500/ 7') In (cm3 of oil)2.(g of asphaltene)-'.(cm3 of catalyst)-'.s-'. In (cm3 of ~ i l ) ~ . of ( g asphaltenic s u l f ~ r ) - ~ . ( cof m ~catalyst)-'.s-'. in (cm3 of oil)*.(g of non-asphaltenic sulfur)-'.(cm3 of catalyst)%-'.
418 Ind. Eng. Chem. Res., Vol. 27, No. 3, 1988 0.a
0 0 as A A nas
7.0-
\ >
$ 0.0-
-7.0
-
1.5
lL
16
1 i ~ . 1 !OK-') 0~
Figure 8. Arrhenius plot of rate constants (k, in (cm3of oil)2-g-'. (cm-3of catalyat)d for asphaltenes cracking, a, and non-asphaltenic c mcata~ fraction desulfurization, nas; k , in (cm3 of ~ i l ) ~ . g - ~ . (of lyst)-W for asphaltenic fraction desulfurization, as). ~
3
21
22
03
,
OL
25 ,
06 ,
07
~
O,B
,
(LHSV/-'.E,h
0
Figure 6. Plot of kinetic data for non-asphaltenic fraction desulfurization (catalyst CH; mean particle diameter, 0.34mm).
0
375 LOO 625
301
375 'C 10.
5t
/t/J
0
5
10
15
Figure 9. Comparison of experimental and predicted values of asphaltenes concentration at the reactor exit for catalyst CH. Table 111. Determined Values of D e
reaction asphaltenes cracking asphaltenic fraction desulfurization non-asphaltenic fraction desulfurization
Figure 7. Plot of kinetic data for non-asphaltenic fraction desulfurization (catalyst CG; mean particle diameter, 0.34mm).
any temperature within the temperature range considered in this study. In Figure 8, k , values are plotted according to eq 7. Table I1 gives the rate expressions for both catalysts. From the data presented in Figure 8, it is observed that reaction rates of the same type do not appreciably differ by using either catalyst CH or catalyst CG, although
D. (cm2/s) for catalyst catalyst CH CG 4.8 X lo4 1.45 X 4.8 X 1.4 X 2.9 X 8X
in all cases catalyst CH appears to have a slightly higher activity. (ii) Diffusional Limitations. To investigate the effect of the intraparticle diffusional limitations on the overall rates, the effective diffusivity for each type of reaction was calculated by using extrudatea of the catalysts CH and CG. The step-by-step integration of eq 4 along the length of the reactor allowed calculation of the exit concentrations for each run. A nonlinear regression method was used for the determination of the best De values for each run and type of reaction.
Ind. Eng. Chem. Res., Vol. 27, No. 3, 1988 419
/ A
Temperature i 0 C l
15
10 m
P
Y V
5
cexp 10’
Figure 10. Comparison of experimental and predicted values of asphaltenes concentration at the reactor exit for catalyst CG.
5
0
IO
Cexp lo3
;
Figure 13. Comparison of the experimental and predicted values of non-asphaltenic sulfur concentration at the reactor exit for catalyst CH.
Temperature i “2 I
Temperature i°C)
A 350
15N
0
7i
u 10-
5
10 Cexp l o 3
Figure 11. Comparison of the experimental and predicted values of asphaltenic sulfur concentration at the reactor exit for catalyst CH.
0
L25
Figure 12. Comparison of the experimental and predicted values of asphaltenic sulfur concentration at the reactor exit for catalyst CG.
Table I11 gives the calculated values of De for both catalysts used. As shown in Figures 9-14, a very good agreement of experimental and calculated values was achieved. The values of effective diffusivity for asphaltenes cracking and asphaltenes desulfurization were the same. This indicates that sulfur is concentrated neither in the large asphaltenic micelles, in which case De,, < De,,,, nor in the small asphaltenes, in which case De,, > De,,,,but it
0
05
10
15
20
Cexp 10‘
Figure 14. Comparison of the experimental and predicted values of non-asphaltenic sulfur concentration at the reactor exit, for catalyst CG.
is equally distributed in the asphaltenes. The effective diffusivity values for the CG catalyst have been calculated to be 2.9 f 0.1 times higher that those for the CH catalyst for both cracking and desulfurization reactions. This difference can be attributed to the different catalyst pore structure. Catalyst CG, with a bimodal pore size distribution, contains a number of large pores which provide large channels for easy access into the catalyst of the reacting molecules. The small pores are required to offer the catalytic surface for reaction, but the diffusion through them is limited. Catalyst CH contains only small pores which do not allow easy penetration to the inferior of the catalyst particles. The tortuosity of catalyst CH is calculated to be 2.5 times higher than the tortuosity of catalyst CG by using the formula
De = Dmi-c’,/~
(8)
(Satterfield, 1980) where D, stands for the bulk diffusion coefficient. The same value of the ratio of the tortuosities is calculated using the data for cracking and desulfurization reactions. For both catalysts the values of the effective diffusivity for the non-asphaltenic fraction desulfurization are 5.8 f 0.1 times higher than those for asphaltenic fraction desulfurization. The different size of the asphaltenic and non-asphaltenic molecules accounts for the small effective diffusivity values of asphaltenes compared to those of non-asphaltenes.
420 Ind. Eng. Chem. Res., Vol. 27, No. 3, 1988
In all experiments of asphaltenes cracking and non-asphaltenic fraction desulfurization using crushed catalyst particles, diffusion does not appreciably affect the calculation of kinetic parameters since the mean effectiveness factor of the catalyst bed, 9, is calculated to be 9 2 0.75. This also happens to asphaltenic fraction desulfurization experiments, except for those conducted at 400 OC for the CH catalyst and those conducted at 425 "C for the CG catalyst where 0.50 5 9 50.75 and diffusional limitations begin to become appreciable.
Conclusions The kinetic data were obtained by using an integral trickle bed reactor and the Greek atmospheric residue as a feedstock. Results indicate that the power-law kinetic equation can describe very well the intrinsic rate equations for asphaltenes cracking, when a = 2; asphaltenic fraction desulfurization, when a = 3; and non-asphaltenic fraction desulfurization, when a = 2. Effective diffusivity values determined for the bimodal catalyst were 2.9 f 0.1 times higher than those determined for the unimodal catalyst for both cracking and desulfurization reactions, although the value of mean pore diameter of the bimodal catalyst particles was only 30% higher than that of the unimodal ones. Effective diffusivity values for asphaltenes cracking were identical with those for asphaltenes desulfurization, for both catalysts. For the non-asphaltenic sulfur-bearing molecules, the determined values of effective diffusivity were 5.8 f 0.1 times higher than the values of the effective diffusivity of the large-size asphaltenes. This is an expected difference, attributed to the higher molecular weights of asphaltenes compared to those of the non-asphaltenic molecules. From a practical viewpoint, the analysis discussed in this study could be attractive for industrial applications in catalyst selection or hydrotreaters design. After the effective diffusivities and intrinsic reaction rates are determined, the performance of the catalyst is evaluated and the proper catalyst selection can be achieved according to demands for high asphaltenes conversion and/or deep desulfurization. Nomenclature A = frequency factor
C = concentration, g/cm3 of oil De = effective diffusivity, cm2/s E = activation energy, cal/mol F = oil volumetric flow rate, cm3/s k, = rate constant LHSV = liquid hourly space velocity, h-l R = reaction rate per reactor unit volume, g/(cm3of bedas) R'= reaction rate per catalyst unit volume, g/(cm3of cata1yst.s) S = external surface of the catalyst particle, cm2 ip= temperature, K V = volume of catalyst particle, cm3 v", = reactor volume, cm3 Greek Symbols a = reaction order t = catalyst activity
si, = catalyst bed void fraction
lC = catalyst particles void fraction 7 = effectiveness factor T
= tortuosity
Thiele modulus Subscripts a = asphaltene as = asphaltenic sulfur i = input nas = non-asphaltenic sulfur 0 = output Registry No. Co, 7440-48-4; Mo, 7439-98-7 @ =
Literature Cited Drushel, H. V. Am. Chem. SOC.,Symp. Ser. 1972,17, 181. Froment, G. F.; Bischoff, K. B. Chemical Reactor Analysis and Design; Wiley: New York, 1979; pp 184-188. Levenspiel, 0. Chemical Reaction Engineering, 2nd ed.; Wiley: New York, 1972; pp 275-283. Mosby, J. F.; Hoekstra, G. B.; Kleinhenz, T. A.; Stroka, J. M. Hydrocarbon Process. 1973, 52, 93. Papayannakos, N. Appl. Catal. 1986,24, 99. Papayannakos, N.; Marangozis, J. J. Chem. Eng. Sci. 1984,39,1051. Satterfield, C . N. Heterogeneous Catalysis in Practice; McGraw-Hik New York, 1980; pp 259-265. Sudoh, J.; Shiroto, Y.; Fukul, Y.; Takeuchi, C. Ind. Eng. Chem. Process Des. Dev. 1984, 23, 641. Takeuchi, C.; Fukul, Y.; Nakamura, M.; Shiroto, Y . Ind. Eng. Chem. Process Des. Deu. 1983, 22, 236.
Received for review February 18, 1987 Revised manuscript received October 19, 1987 Accepted November 4, 1987