AM-88-51

 

EFFECTS OF FEEDSTOCK TYPE ON FCC OVERCRACKING

By

Ernest L. Leuenberger, Executive Technical Service Engineer
Eric L. Moorehead, Manager of Petroleum Catalyst Evaluation
Dennis F. Newell, Technical Service Engineer
BASF Corporation
Edison, New Jersey

 

Presented at the

1988 NPRA
ANNUAL MEETING
March 20-22, 1988
Convention Center
San Antonio, Texas

 

Introduction

Fluid catalytic cracking (FCC) is a petroleum refinery process that involves the breaking of carbon-carbon bonds in a heavy hydrocarbon to produce more valuable products of lower molecular weight. Usually FCC units are operated to maximize gasoline production. Since the invention of the current generation of high activity zeolitic cracking catalysts, it has been possible to operate a cracker at severities so high that gasoline is broken down into C3 and C4 hydrocarbons. This "overcracking" of gasoline to gases was minimized by reducing the contact time between catalyst and hydrocarbons. The riser FCC unit(Ref. 1, 2) replaced the fluid bed unit in the 1970's as a result of the industry's desire to minimize overcracking by reducing contact time.

As catalyst manufacturers continued to improve the activity and stability of zeolitic cracking catalysts, refiners came to realize that excessive activity could also cause overcracking in riser units. This phenomena was characterized by excessive regenerator temperatures in at least two published articles(Ref. 3, 4). The experimental work in this report was undertaken to help explain and predict overcracking in riser FCC units.

Experimental Method

In order to determine the relationships between catalyst activity, feedstock type, and gasoline overcracking in a riser FCC unit, an experimental program was designed using five levels of catalyst activity and three different feedstocks. One large sample of a commercially manufactured FCC catalyst was divided into five equal portions. Each portion was artificially aged to a different activity level bzy steaming at 1450F and atmospheric pressure. Steaming times were varied from 0.5 hours to 24 hours to give microactivity test results ranging from 84 to 61 weight percent conversion. Properties of the fresh and hydrothermally deactivated catalyst samples are summarized in Table I.

The properties of the three feedstocks used in the experimental program are listed in Table II. The Altamont feed is one of the most paraffinic feedstocks cracked in the continental USA. It has a high API gravity, a high aniline point, and a low nitrogen content. The Mid Continent feed is more typical of USA feedstocks. It is less paraffinic, as is demonstrated by its PNA analysis as well as its lower aniline point and lower API gravity. The third feed is a mixture of hydrotreated West Coast and imported Minas gas oils. Its higher nitrogen content makes it more difficult to crack than the other two feedstocks.

Catalyst and feedstocks were introduced into a laboratory riser circulating pilot unit. Contact time between catalyst and oil was estimated to be 2 seconds. Catalyst circulation rates were varied while other operating conditions were held constant. Table III summarizes the laboratory reactor operating conditions.

Varying catalyst circulation allows the gasoline yield and other products of the cracking reactions to be compared at constant conversion for different catalyst activities. When increasing activity reduces gasoline yield at constant conversion, the higher activity is said to be overcracking the gasoline.

Overcracking Due to Excessive Activity in a Laboratory FCC Unit

Overcracking occurred on all three feedstocks for the higher activity catalysts. The yield shifts that were characteristic of this Overcraking are shown in Figures 1 through 4. Figure 1 indicates the highest activity catalyst increased coke yield by 1 weight percent at all levels of conversion. The gas yields in Figure 2 show that excessive activity increases C3and C4 saturate gases but does not affect olefin yields or lighter gas yields. This tendency of overactive catalysts to produce high levels of LPG saturates is apparent in the high isobutane to butylene ratio for the 84 activity catalyst in Figure 3. The high coke and LPG saturate yields reduce gasoline yield at constant conversion, as demonstrated by Figure 4.

The laboratory overcracking data indicates that the loss of gasoline yield and the increase in coke and LPG saturate yields are all interrelated phenomena. The strong correlation between the three variables allows gasoline selectivity to be predicted from either the coke make or the isobutane to butylene ratio. Figure 5 demonstrates that both high and low activity catalysts have the same selectivity at constant coke make.

Increasing activity has a similar effect on coke and isobutane yields for all three feedstocks, but gasoline overcracking appears to occur at different activity levels for each feed. Figures 6 and 7 show that coke yield and isobutane to butylene ratio increase smoothly as activity is increased and cat-to-oil is reduced to hold conversion constant. However, Figure 8 shows that the three feedstocks can each tolerate a different activity before gasoline selectivity begins to decline. The paraffinic Altamont feedstock shows signs of overcracking above 65 activity, while the more aromatic Mid continent feedstock has no significant loss in gasoline yield until the activity exceeds 71. The higher nitrogen feedstock shows a broad maximum in its gasoline selectivity curve between 68 and 77 activity.

A Model of Overcracking

Recent investigations of zeolite properties and their effects on FCC selectivities help to explain the reason high activity catalysts can cause overcracking. Pine et al(Ref. 5) demonstrated that zeolite unit cell size has a strong effect on the selectivity of the catalyst. Rajagopalan and co-workers reported that coke yield is higher for catalysts with large unit cell sizes(Ref. 6)  and that the effective particle size of the zeolite crystals affects the catalyst selectivity(Ref. 7). Corma et al(Ref. 8) concluded that effective zeolite crystal size decreases as zeolite is aged, while it has been observed by all that cell size is lower for aged catalysts. These investigations suggest active catalysts have larger unit cell sizes, larger effective zeolite crystal sizes, higher coke selectivities, and higher isobutane to butylene ratios than more deactivated catalysts.

The properties of the catalyst used in this study confirm that the active catalysts have higher unit cell sizes. Examination of the active catalysts by scanning electron microscope shows they have fewer defects in their zeolite crystals, which indicates they also have a larger effective crystal size. The higher unit cell size is responsible for the higher coke make observed during overcracking. This higher coke make and the larger effective crystal size can also result in a drop in gasoline selectivity. As more coke is deposited on the zeolite, the diffusional resistance within the crystal increases. As Rajagopalan stated in his model for the poor gasoline selectivity of large zeolite crystals(Ref. 7), the increased diffusional resistance favors gasoline overcracking. The larger effective crystal size of the more active catalyst also increases the diffusional resistance and favors overcracking.

Increases in activity need not always result in gasoline overcracking. The overcracking occurs only when the reaction rate of gas oil in the zeolite pore structure is equal to or greater than the rate of gas oil diffusion to the cracking sites in the pore. In this situation the pore diffusion controls the rate of cracking, so longer pores and more coke blockage in these pores will slow down the rate of gasoline production. Gasoline recracking becomes more prevalent in the zeolite as gas oil penetration is limited by pore diffusion. The pore diffusion effectively eliminates gas oil molecules from the ends of the pores and allows the cracking sites at pore ends to recrack gasoline without interference.

The square root of the ratio of gas oil cracking rate to gas oil diffusion rate is the Thiele Modulus. For a first order reaction, it can be defined as:

where k is a first order rate constant, L is pore length, and D is the diffusivity in the pore. When the Thiele Modulus approaches one, any further increase in catalyst activity can cause overcracking. As long as the Thiele Modulus is well below 1, an increase in activity will not promote overcracking. The Thiele Modulus will increase towards 1 as the pore lengths increase, as carbon deposits reduce gas oil diffusivity in the pores, and as reaction rates increase due to higher activity. All these changes that bring on overcracking occur as catalyst activity increases.

Feedstock Effects on Overcracking

Feedstocks which have lower cracking rate constants or higher pore diffusion rates will have a lower Thiele modulus, and thus will be less likely to overcrack at a given level of catalyst activity. If diffusivity effects are constant, it should be possible to correlate the activity at which overcracking begins with an estimate of the cracking rate constant. This correlation is expected to show that the feedstocks with the lower cracking rates require higher activity to overcrack.

Mobil Research and Development has published a detailed model in which cracking rates are related to feedstock properties(Ref. 9, 10). Increases in feed nitrogen and aromatic contents are said to decrease reaction rates because these feed components strongly adsorb on the cracking sites. The detailed analysis of aromatic hydrocarbons needed to use the model equations are not available on these feedstocks, but the increase in aromatics content of the Mid continent feedstock could decrease the reaction rate substantially. If half of the 14 percent increase in aromatics were in Mobil's heavy aromatic ring classification, the Mobil paper states the rate would be reduced by 33 percent. A 33 percent drop in the rate would allow the activity to increase from 66 to 74 without increasing the Thiele modulus. This effect of aromatics is sufficient to explain the higher catalyst activity required to overcrack the Mid continent feed.

The equation that relates nitrogen poisoning to cracking rates in the Mobil model also explains the fact that the gasoline selectivity curve in Figure 8 has a different shape for the higher nitrogen feedstock. The model equation states reaction rates decrease as nitrogen increases and as cat-to-oil ratio decreases:

The higher nitrogen decreases the Thiele modulus by decreasing the reaction rate and allows the feedstock to withstand higher catalyst activity. In addition, the cat-to-oil is lowered to hold conversion constant in Figure 8, and this increases the effectiveness of the nitrogen and further reduces the Thiele modulus. As a result, gasoline selectivity does not decline as rapidly on the West Coast feed as it does on the lower nitrogen feedstocks.

Application of the Model to Commercial FCC Units

The higher delta coke yields measured for increased catalyst activity were found to cause significant rises in regenerator temperature when the laboratory data was used to model a commercial FCC unit. Table IV shows the simulation of cat cracker performance based on West Coast feed experimental results for an activity increase from 71 to 76. The heat balanced model predicts a 32F rise in regenerator temperature. If a unit is limited by regenerator metallurgy, compensating changes will need to be made in the unit operation to prevent this predicted temperature rise. These compensating changes will often result in a loss of conversion. A second model simulation is included in Table IV to show the effects of reducing reactor temperature to keep regenerator temperature constant. Lower reactor temperature results in a loss in conversion despite the gain in activity. Gasoline yield and octane losses are also predicted. These changes would normally reduce product value.

Our simulation of the effects of catalyst activity indicates that catalyst makeup should not be increased in temperature limited units unless the type of catalyst is changed to lower its coke making characteristics. The laboratory results on high activity catalysts demonstrate that activity increases coke make due to an increase in zeolite cell size. Since fresher catalyst always has a higher unit cell size than equilibrium, increasing fresh catalyst additions should have the same effect in a commercial unit as higher activity does in the laboratory. If catalyst additions must be increased in a temperature constrained unit due to loss problems, then it is best to add lower unit cell size equilibrium catalyst to provide the additional makeup.

If a commercial FCCU is not constrained by regenerator temperature, there is a maximum catalyst activity above which gasoline selectivity declines. This optimum activity level increases as the feed becomes more difficult to crack. If a unit is processing a paraffinic low nitrogen feedstock, the maximum gasoline selectivity is achieved near 65 activity. If catalyst circulation can be increased enough to operate at other unit constraints such as the product recovery limits, it is best not to increase the activity above 65 for these feeds. As the aromatic content and/or the nitrogen content of the feed increases, the maximum activity also increases. A 70 activity is optimum for many common feedstocks, while activities as high as 77 may yield the best gasoline selectivities for high nitrogen feeds. Of course it may not be desirable to operate at an activity as high as 77 if the unit reaches a regenerator temperature limit at a lower activity.

The majority of commercial FCC units operate near the maximum equilibrium activity that is consistent with good gasoline selectivity. Figures 9 and 10 are plots of the activities maintained by 33 commercial FCC units. All the units surveyed hold activities between 62 and 74. In both figures, the highest activities correspond to difficult to crack feeds - those with high nitrogen and low aniline point (high aromatic content). The plots also show that units with high metals on catalyst are operated at lower activities than units with clean catalyst. High metals on catalyst indicate the presence of residual gas oil, which cracks at a faster rate and deposits more coke than lower molecular weight, clean gas oils. Therefore, the Thiele modulus is higher for resids unless the activity is maintained at a lower level. The added coke make of resid also tends to increase regenerator temperature, so in cases of regenerator temperature limited units the optimum activity is further lowered. These three effects - higher cracking rates, higher coke make, and lower feed pore diffusion rates - lower the optimum activity for resid feeds.

While the conclusions of this report provide a general guideline for determining catalyst activity targets, individual units may have circumstances that are exceptions to these guidelines. For example, bed crackers have been shown to overcrack at lower activities than riser crackers. In addition, some refineries may find that operating the unit to produce some overcracking may improve the cat cracked gasoline octane or profitably provide the isobutane needed for an alkylation unit. The actual activity target for each unit should thus consider unit design factors and refinery economics as well as the guidelines in this report.

References

1. J.R. Murphy, Oil & Gas Journal, November 23, 1970, Page 72.

2. C.W. Strather, W.L. Vermillion, and A.J. Conner, Oil & Gas Journal, May 15, 1972, Page 102.

3. G. Wilson, NPRA Ouestion and Answer Session, 1977, Page 54.

4. W.R. Wichers and L.L. Upson, Oil & Gas Journal, March 19, 1984, Page 157.

5. L.A. Pine, P.J. Maher, and W.A. Wachter, Journal Of Catalysis 85 (1984), Page 466.

6. K. Rajagopalan and A.W. Peters, Journal Of Catalysis 106 (1987) Page 410.

7. K. Rajagopalan, A.W. Peters and G.C. Edwards, Applied Catalysis 23 (1986) Page 69.

8. A. Corma, E. Herrero, A Martinez, and J. Prieto, paper presented at A.C.S. Symposium On Advances In FCC, New Orleans Meeting, August 30 - September 4, 1987.

9. Jacob, Gross, Voltz, and Weekman, A I Ch E Journal Vol. 22, No. 4 (1976) Page 701.

10. Gross, Jacob, Nace, and Weekman, U.S. Patent 3960707, June 1, 1976.