Vanadium Mobility in Fluid Catalytic Cracking - ACS Symposium

Chapter 21

Vanadium Mobility in Fluid Catalytic Cracking Deactivation and Testing of Hydrocarbon-Processing Catalysts Downloaded from pubs.acs.org by UNIV OF TEXAS AT EL PASO on 11/03/18. For personal use only.

1

1

1

Richard F. Wormsbecher , Wu-Cheng Cheng , Gwan Kim , and Robert H. Harding 2

1

2

Grace Davison and Research Division, W. R. Grace and Company—Conn., 7500 Grace Drive, Columbia, MD 21044

Previous work has shown that the poison precursor for vanadium poisoning of fluid cracking catalysts is a volatile species, vanadic acid, H V O , formed in the regenerator from the interaction of H O vapor and oxides of vanadium in the V oxidation state. The vapor pressure of the vanadic acid is very low, and may not be high enough to explain the mobility of vanadium in the regenerator. Some authors have suggested that the vanadium is transferred throughout the catalyst inventory by inter-particle transfer, through an unspecified collision mechanism. In this work, the mobility of vanadium is studied by measuring the rate of vanadium transfer to a vanadium trap. By varying the particle size distribution, and consequently the number density of particles, the collision frequency can be changed. In these experiments, the rate of vanadium transfer was shown to be independent of collision frequency. Mass transfer calculations for a volatile species in a fluid bed show that even very low vapor pressures are sufficient to support mass transfer in the bed. The experimental data are best fit by the model that a vanadium species undergoes volatile transfer, where the vapor pressure fits a second order Freundlich isotherm. Because the catalyst surface is inhomogeneous, these results suggest that the vapor pressure of the vanadic acid is governed by a coverage dependent heat of adsorption. 3

4

+ 5

2

The poisoning of fluid cracking catalysts (FCC) by vanadium is well known (7, 2). In general, vanadium is deposited on the cracking catalyst as coke by vanadyl porphyrins in the feed. During regeneration, the coke is burned off, and vanadium can be oxidized to the V oxidation state. Woolery et al. (5) have shown that the oxidation state of the vanadium can alternate between +4 to +5 in + 5

0097-6156/96/0634-0283$15.00/0 © 1996 American Chemical Society

284

DEACTIVATION AND TESTING OF HYDROCARBON-PROCESSING CATALYSTS

the cracking regeneration cycle, but that the predominant state is +5 in the FCC regenerator. These oxides of vanadium then react with the H 2 O vapor in the regenerator to form vanadic acid, H 3 V O 4 , which is volatile (4). Zeolites are known to be sensitive to acid-catalyzed hydrolysis, which destroys the zeolite framework. The vanadic acid apparently catalyzes this reaction in the presence of steam in the regenerator (4). Pine (5) has shown that this reaction is a solid state transformation that is catalyzed by a vapor phase species. Vanadium is also know to be very mobile throughout the catalyst inventory (1). Other work (4) has shown that steam is necessary for intra-particle transfer, and for the poisoning of the cracking activity by vanadic acid. Although the poison precursor of vanadium poisoning was shown to be volatile vanadic acid, because of it's very low vapor pressure, it may not explain the observed the inter-particle mobility of vanadium. Kugler and Leta (6) have shown that the distribution of vanadium in equilibrium catalysts is consistent with either a volatile or inter-particle transfer mechanism, von Ballmoos et al. (7) have suggested that the mobility of vanadium is due to particle-to-particle collisions. What is currently known about the mobility of vanadium is that it is dependent on the vanadium being in the V oxidation state, and that water vapor is present at the conditions of regeneration. Also, V 2 O 5 melts at 943 K, which is below the operating temperature of most regenerators, 970 - 1020 K. Presumably, if particle-to-particle collisions are responsible for the transport, then the water vapor must lower the surface tension of the liquid species sufficiently to facilitate the transport. It is the purpose of this paper to study which mechanism, particle-toparticle or vapor phase transport, is responsible for the mobility of vanadium in the FCC unit. The approach used in this work is to measure the rate of vanadium transport to a basic oxide "vanadium trap" in fluid bed experiments. By varying the particle size distribution, the collision frequency can be changed and the rate of transport deterrnined. Also, calculations of the mass transfer of a vapor species in a fluid are performed. + 5

Experimental The catalyst chosen for this study is a 40 wt% USY catalyst bound in an alurnina sol/kaolin matrix, similar to GRACE Davison's DA catalyst family. It contains no rare earth, and is therefore considered to be an octane catalyst. The trap used in this experiments is a 40 wt% MgO based trap bound in a La203/Al203 matrix (8). The trap has been shown to have a high affinity for vanadic acid. Fluid bed steaming experiments were performed in temperature controlled sand bath reactors using the conditions given in Table I. Staged bed steaming experiments were performed in a fixed bed reactor, using the conditions given in Table I. Density separations of catalyst from trap in the fluid bed experiments were performed by a procedure similar to that in earlier work (9, 10). Particle size measurements were performed using a light scattering technique from a Malvern 3600E particle size analyzer. Chemical analysis was performed by ion coupled plasma analysis, standardized to NIST standards.

21. WORMSBECHER ET AL.

Vanadium Mobility in FCC

285

Table I. Fluid and Staged Bed Experiments Conditions

45 gr. catalyst/.5 gr. trap 5hr.@ 1088K95%Steam 8.5 cc/minN 7.5 cc/hr. liquid H 0 2

2

Trap

Catalyst Before Steam wt%V After Steam wt% V % V transferred

0.5 Float (Catalyst) 0.186 64

0 Sink (Trap) 2.58

Staged Bed Experiment Conditions

9 cc/hr. liquid H 0 10 cc/minN 5 hrs. @ 1088 K Catalyst

Trap

0.509

0.02

0.511

0.02

2

2

Before Steam wt% V After Steam wt% V Results and Discussion

The mobility of vanadium in a fluidized bed is illustrated by afluidbed steaming experiment using a "vanadium trap" which will react with the mobilized vanadium and render it immobile (4). Traps which are effective for reacting with vanadic acid are basic oxides, such as MgO based materials. The trap used in these experiments contain 40% MgO and is very effective for reacting with vanadic acid. In this experiment, 90 wt% catalyst/ 10 wt% trap blend is steamed for 4 hours at 1088 K, 95% vol. H 0 vapor, in a fluid bed. Only the catalyst is impregnated to 0.5 wt% V prior to blending and steaming. After steaming, the catalyst and trap fractions, which have different densities, are separated by the sink/float technique, and analyzed for vanadium. The results (Table I) show that the catalyst retained 0.186 wt% V, while the trap gained 2.58 wt% V. Essentially 64% of the vanadium was transferredfromthe catalyst to the trap. A staged bed transpiration experiment was also performed. In this case, the catalyst was again impregnated with 0.5 wt% V and loaded into a fixed bed reactor. A layer of glass wool was placed over the catalyst, and the trap was layered on top of the glass wool. In this way, there was no contact between the beds. Steam vapor was flowed through the catalyst bed then up to the trap bed at 9 cm /h liquid, 10 cm /min. N . Results (Table I) show that the catalyst did not lose vanadium, nor did the trap gain vanadium so that there was no 2

3

3

2

286


transpiration of vanadium between the beds. The equilibrium vapor pressure of H 3 V O 4 over pure V2O5 at 1088 K and 1 arm water vapor pressure is 2.9 x 10" atm (77). Assuming the gas phase is saturated with H 3 V O 4 vapor, in the staged bed experiment, the vanadium on the catalyst should have decreased from 5000 ppm to 4750 ppm, a change which is well within the detection limit of analysis. There is an apparent paradoxfromthese experiments: vanadium moves readily from catalyst to trap in a fluid bed, yet vanadium transpiration is negligible under the same conditions. This result would suggest that the mechanism of transport is by particle-to-particle collisions in the bed. Still, an alternate question arises, what are the requirements for mass transfer of vapor phase vanadium in a fluid bed? Following the correlation of Richardson and Szekely (72) , the mass transfer coefficient, km, in afluidbed can be calculated from the Sherwood number, Sh, and a knowledge of the particle Reynolds number, Re, in the bed by, 5

8

0.374 Re" Sh = k

m

Re namely that the vapor pressure of vanadic acid is very low at the surface of the trap due to the chemistry of the interaction, then CH3VO4, gas can be calculatedfromthe experimental data given in Table I, trap

< < :



CH3V04, gas = 1.05

287

3

x lQr mol cnr , 11

(5)

7

PH3V04, gas = 9.4 x 10' atm.

(5a)

The calculated vapor pressure of vanadic acid is a factor of 30 lower than the equilibrium vapor pressure of vanadic acid over pure V2O5. At this vapor pressure of vanadic acid, the transfer of V to trap is rapid while the removal of V by transpiration is negligible. Approximately 64% would be transferred from catalyst to trap, and about 0.05% removed by transpiration. This is rationalized by noting that the velocity of the vanadic acid vapor (calculated from the kinetic theory of gases), 4.4 x 10 cm s , is four orders of magnitude higher than the superficial velocity of the fluidizing gas, Equation 3a. From these calculations it is clear that mass transfer of vapor phase vanadic acid in a fluid bed is sufficient to account for the transport of vanadiumfromcatalyst to trap. Although vapor phase mass transfer of vanadium is consistent with all our data, further experiments were performed to test the particle-to-particle mechanism. In this case, the rate of vanadium transfer is expected to be proportional to the collisionfrequencyin the bed, 4

_1

V transfer rate = k

coUision

N

catalyst

N

trap

y

(6)

where N i and N are the number densities of the catalyst and trap particles, respectively and y is the amount of vanadium transferred per collision, which is a function of the concentration of the vanadium on the catalyst. The number densities can be varied by varying the particle size distribution. In these experiments, the catalyst was split into twofractionsby screen classification. Table II gives the particle size distributions for both the coarse and fine catalyst, as well as the trap distribution. By assuming spherical particles of uniform particle density, the number density can be determined, and is shown in Table II. The ratio of the particle densities for the fine and coarsefractionsis cata

yst

trap

T

Afine /A coarse = 6.5 .

(7)

Catalyst and trap blends were steamed in a fluid bed using the same conditions as in Table I. Weight blend ratios with 95/5, 90/10, 85/15 of catalyst trap for both the fine and the coarsefractionswere steamed, and the sink/float technique was again used to measure the rate of vanadium transfer. The results (Table III) show that there is not a strong dependence of the vanadium transfer rate on particle size. The rate enhancement, which is the rate of the fine catalyst / rate of the coarse catalyst, only varies by 10%, whereas the number density varied by a factor of 6.5. This experiment shows that the rate of vanadium transferfromcatalyst to trap is only weakly dependent on the particle density, and it implies that particle-to-particle collisions are not the dominant mode of vanadium transfer.

288


Table II. Particle Size Distributions of Fine and Coarse Fraction of Catalyst Catalyst Coarse

Catalyst Fine

Trap

Volume % in 0-20 microns 0-40 0-60 0-80 0-105 0-125 0-150 0-200 0-320

0 0.7 1.8 15.5 41.6 59.6 75.6 93.7 99.5

0.3 10.8 57.2 87.4 96.6 98.5 99.5 99.8 100

1.8 14.9 29.6 56.9 80.6 89.6 94.9 99 100

APS/microns

113

57

75

1.70E+06

1.05E+07

2.30E+07

3

# Particles/cm

1.1

1.1

0.90 0.17

6.1

Rate Enhancement: rate fine/rate coarse

0.06 4.13

0.90 0.21

6.7

4.5

0.06 3.83

sink float (trap) (cat.)

0.10 2.90

0.10 2.75

0.84 0.14

0.83 0.14

1.0

7.3

7.2

0.16 1.98

0.17 1.80

sink (trap)

85/15 Blend

sink float (trap) (cat.)

90/10 Blend

5.1

0.94 0.25

0.94 0.29

float (cat.)

95/5 Blend

V transfer rate (g V h-1 g cat-1) X 10-4

Fine Catalyst/Trap Blend wt. fraction wt%V normalized

V transfer rate (gVh-1 gcat-l)X10-4

Coarse Catalyst/Trap Blend wt. fraction wt%V normalized

Separated Fraction

Nominal wt. catalyst/wt. trap blend

Table III. Coarse and Fine Catalyst/Trap Blends 5000 PPM V on Catalyst/5 hrs. @ 815 °C/95% Steam/Fluid Bed

290


On the contrary these results support the interpretation that gas phase vanadic acid is in pseudo-equilibrium with vanadium on the FCC catalyst, and the rate limiting step in vanadium transfer is the adsorption of vanadic acid vapor onto the trap. Since the rate of vanadium transfer is independent of catalyst particle size, intraparticle vanadium transfer must be very fast compared with interparticle vanadium transfer. The vanadium transfer experiments can provide information on the vapor pressure of vanadic acid, using the same techniques as outlined above. Equation 4 can be rewritten as, V transfer rate = k' PH3V04 Qrap-

(8)

where Ctrap is the concentration of the trap in the blend. If one assumes that the equilibrium between H 3 V O 4 vapor and surface vanadium follows the Langmuir isotherm, then at low vanadium levels, the vapor pressure of vanadium is approximately linearly dependent on the vanadium concentration, then, PH3V04

=

K VcatalysU

(9)

where V i is the concentration of V on the catalyst and K is the equilibrium constant for adsorption. By substituting Equation 9 into Equation 8 and integrating over time gives the concentration dependence of vanadium on the catalyst as a function of the ratio of trap concentration to the catalyst concentration in the blend, cata

yst

Vcatalyst

^catalyst

GX

P

r \

(10)

^catalyst J

where rate constant k\ the adsorption constant K and time, which are constant for all the experiments, have been lumped into the new constant kj. It is also possible that, because the surface of the catalyst is inhomogeneous, the equilibrium between H 3 V O 4 vapor and surface vanadium may not follow the Langmuir isotherm. In this case, the vapor pressure can be modeled by a Freundlich isotherm, PHWOA

=

^catalyst '

Substituting Equation 11 into Equation 8 and integrating over time gives,

where time has again been lumped into the constant


291


The best fit curves of both models to the vapor pressure data extracted from the coarse blends (Table III) are shown in Figure 1. In the case of the Freundlich isotherm, an exponent value of 2 gave the best fit. As can be seen, the second order Freundlich isotherm is a closer fit to the observed data than the linear model. Similar plots are obtained for the fine catalyst experiments. This fact indicates that the heat of adsorption of vanadium in the V oxidation state is coverage dependent. This result is more clearly seen by the plot of "vanadium pick-up factor", Vtrap/V aiyst> versus trap/catalyst concentration (Figure 2). This plot shows the strong dependence on the second order model, again verifying the inhomogeneous nature of the catalyst surface for vanadium. A second set of experiments was performed to determine the rate of transfer as a function of vanadium loading. In these experiments the entire blend, with 10% trap, was impregnated at different vanadium levels. The blend was then fluid steamed using a cyclic procedure, which switches between oxidizing and reducing conditions using propylene, in much the same spirit as occurs in normal FCCU operation. The stearning procedure is described in detail in (75). The data are shown in Table IV. Because of pore volume differences between the catalyst and trap, the initial % V on the catalyst is corrected. Figure 3 shows the vanadium concentration on the catalyst after stearning versus the initial % V on catalyst before steaming. As before, the data were fitted to the two models and again the second order Freundlich isotherm best fits the data. Figure 4 shows the vanadium pickup factor, which as before show a strong dependence for the second order model. The rate of vanadium mobility in FCCU's is dependent on many operating factors. These might be whether the unit is operated in full or partial combustion, which will effect the average oxidation state of the vanadium. Other factors will be "freshness" of the vanadium, presumably older vanadium has had more of an opportunity to react with the catalyst matrix and become immobile. Steam concentration in the regenerator, catalyst make-up rate, temperature, and two-stage regeneration will all effect the mobility of vanadium. + 5

cat

Conclusion The results of this work show that even though the vapor pressure of vanadium is low, the transfer velocity of vanadium vapor is high and the rate of mass transfer in a fluidized bed is high. A high rate of vanadium transport to traps and a low rate of vanadium transport by transpiration are consistent with the vapor phase transport model. The vapor pressure of the vanadic acid follows a second order Freundlich isotherm, which reflects a coverage dependent heat of adsorption. The rate of vanadium transferfromcatalyst to trap is only weakly dependent on the number density of the catalyst or trap particles. This lack of dependence suggests that inter-particle collisions are not the dominant mechanism for vanadium transfer. Vanadium mobility in FCCU's is a complex issue dependent on many operating variables.


0.50•

observed 1" order: Langmuir isotherm

0.40-

\V

—

~~ " 2 order: Freundlich isotherm nd

\\

0.30-

•

v\

0.20-

0.10. . . . . . . . .

0.000

0.1 0.2 wt. of trap/ wt. of catalyst

Figure 1. Vanadium Concentration on the Catalyst a Function of Trap Content. Blends of Coarse Catalyst with Trap.

Figure 2. Vanadium Pickup Factor as a Function of Trap Content. Blends of Coarse Catalyst with Trap

0.3


293


initial % V on catalyst, before steaming Figure 3. Vanadium Concentration on the Catalyst as a Function of Vanadium Loading.

Table IV. 90% Catalyst/10% Trap Blends for Co-Impregnated Samples At Different Vanadium Loadings Initial Blend

Initial wt. % V

Float

Sink

Pickup Factor

wt% V

corrected on

(cat.)

(trap)

% V Trap/% V Cat.

catalyst

wt% V

wt%V

.172

.135

.109

.715

6.6

.343

.270

.203

1.47

7.2

.503

.396

.269

2.59

9.6

After cyclic propylene steam: 30 cycles with 60% volume H2O cycled between balance gas of 5% wt. C H in N , and air with 4000 ppm S 0 , total time 20 hrs.,T = 771°C, Fluid Bed 3

6

2

2

294


10.00-

observed ' 1" order. Langmuir isotherm " 2" order Freundlich isotherm 1

9.008.00-

V s 3

7.00-

/

/

6.005.00-

1

i • *' 1

111

•

1 1 1 1 1 1 1

0.1 0.2 0.3 0.4 0.5 initial % V on catalyst, before steaming Figure 4. Vanadium Pickup Factor as a Function of Vanadium Loading.

0.6



295

Literature Cited 1.

2. 3.

4. 5. 6. 7. 8. 9. 10.

11. 12. 13. 14. 15. 16.

Nielsen, R. H; Doolin, P. K. in Fluid Catalytic Cracking: Science and Tech., Magee, J. S. and Mitchell, M. M., Ed.; Elsevier Science Publishers B.V.: New York, 1993, pp 339-384. Ritter, R. E.; Rheaume, L.; Welsh, W. A. and Magee, J. S., Oil & Gas J., 1981, July 6, pp. 103-110. Woolery, G. L.; Chin, A. A.; Kirker, G. W. and Huss, A. in Fluid Catalytic Cracking - Role in Modern Refining; Occelli, M. L., Ed.; ACS Symposium Series 375; American Chemical Society: Washington, DC, 1988, pp. 215228. Wormsbecher, R. F., Peters, A. W. and Maselli, J. M., J. Catal., 1986, Vol. 100, pp. 130-137. Pine, L. A., J Catal., 1990, Vol. 125, pp. 514-524. Kugler, E. L. and Leta, D. P. J. Catal., 1988, Vol. 109, pp. 387-395. von Ballmoos, R.; Deeba, M.; Macaoay, J. M. and Murphy, M. A., Annual NPRA Mtg., San Antonio, TX, 1993, Vol. AM-93-30. Kim, G. U.S. Patent 5,407,878 (1995). Palmer, J. L. and Cornelius, E. B., Appl. Catalysis, 1987, Vol. 35, pp. 217235. Beyerlein, R. A.; Tamborski, G. A.; Marshall, C. A.; Meyers, B. L.; Hall, J. B. and Huggins, B. J. in Fluid Catalytic Cracking II - Concepts in Catalyst Design; Occelli, M. L., Ed.; ACS Symposium Series 452; American Chemical Society: Washington, DC, 1991, pp. 109-143. Yannopoulos, L.N., J. Phys. Chem., 1968, Vol. 72 (9), pp. 3293-3296. Fluidization Engineering; Kunii, D. and Levenspiel, O., Ed.; Robert E. Krieger Publishing Company: Huntington, New York, 1977, Vol. 534. Richardson, J. F. and Szekely, J., Trans. Inst. Chem. Engrs., 1961, Vol. 39, pp. 212. Perry's Chemical Engineers'Handbook;Perry, R. H. and Green, D. W., Ed., 6th Ed., McGraw-Hill, Inc., 1984. Fuller, E. N.; Schettler, P. D. and Giddings, J. C., Ind. Eng. Chem, 1966, Vol. 58 (5), pp. 18. (Boock, L. T., Petti, T. F., and Rudesill, J. A., International Symposium on Deactivation and Testing of Hydrocarbon Conversion Catalysts, ACS Annual Meeting, Chicago, IL, August 1995)

Vanadium Mobility in Fluid Catalytic Cracking - ACS Symposium

Recommend Documents