John F. Hohnholt
Valero Refining Co.
Houston
Woodrow K. Shiflett Arthur J. Suchanek
Criterion Catalyst Co. LP
Houston
Increasing the operating severity of the atmospheric resid hydrodesulfurization (Arhds) unit at Valero Refining Co.'s Corpus Christi, Tex., refinery boosted yields of diesel and distillate products. After an initial run of two cycles on the original process licensed catalyst, catalysts and operating conditions were changed, resulting in a subsequent 30-month run.
With a catalyst reload of four different silica and alumina hydrotreating catalysts in the reactors of the Arhds unit, the unit was operated for about 10 months in a normal hydrotreating (HDT) mode. Successful operation in that mode prompted Valero to increase severity to a hydroconversion mode, with the total run of 30 months ending in July 1989.
Reloading the four Arhds reactors with the silca and alumina catalysts, designed for heavy, contaminated feeds, was a key factor in boosting capacity of the unit to 60,000 b/sd from the original design capacity of 46,100 b/sd and allowing the increase in severity.
Feed contaminant removal and other unit performance parameters were monitored regularly during both modes of operation. Details of the runs in both modes are presented here.
VALERA'S REFINERY
The present plant flow scheme at Valero is shown in Fig. 1. Products from the Arhds unit are further converted by the heavy oil catalytic cracking unit (HOC).
The Arhds unit was originally designed for desulfurization of 46,100 b/sd Arabian heavy atmospheric tower bottoms (ATB) with a 1-year cycle life. Table 1 shows the design parameters for this high-pressure unit, designed by Gulf Research & Development Corp.
Fig. 2, a flow scheme of the Arhds unit, shows that the unit is comprised of two trains, four reactors per train, with a catalyst ratio of 1-2-33. The first reactor is a guard reactor and both trains feed a common separation and fractionation system.
The refinery was commissioned in mid-1983. The Arhds unit initially used two cycles of the design basis licensed catalyst.
In 1986, a demetallation catalyst was loaded into the guard reactor of the Arhds unit. The catalyst was originally a Shell Oil Co. brand. Subsequently, Shell and American Cyanamid Co. formed the joint venture company, Criterion Catalyst Co. LP.
Because the performance of the demetallation catalyst was deemed acceptable, Valero selected a complete unit fill of Shell/Criterion 967-317-917-447 catalysts for the next cycle that began in January 1987.
The 967 and 917 catalysts are silica catalysts. The 317 catalyst is nickel-molybdenum (NiMo) on alumina for deep hydrodesulfurization (HDS), and the 447 is cobalt-molybdenum (CoMo) on alumina for deep HDS.
The new catalyst loading and modified operating conditions boosted the Arhds capacity about 30% to 60,000 b/sd.
CATALYST SELECTION
Prior to the cycle with the Shell/Criterion catalysts, the Valero operating philosophy was to achieve optimum-quality, heavy oil cracker feed. The Arhds unit objective was to achieve maximum removal of contaminants (sulfur, nitrogen, metals, and Conradson carbon residue).
Metals, in particular, have a large economic effect on the HOC operation. Part of the objective, of course, was the need to achieve maximum cycle life with the least costly feedstock available.
These feedstocks, at the time of the start-up of the catalyst system, were seen as mixtures of atmospheric tower bottoms (ATB's) and vacuum tower bottoms (VTB's) from the Middle East, mainly Arabian, crudes. A second objective was to achieve target contaminant removals at a feed rate 30% above design, thus minimizing the purchase of low-sulfur ATB for HOC charge.
Implementation of debottlenecking in the HOC operation created an imbalance in feed rates requiring the purchase of low sulfur ATB'S. Normal Arhds hydrotreatment of the feed to no more than 10 ppm total metals and 0.5 wt % sulfur in the 650 F. + product was seen as the prime unit objective.
Prior to the selection of the catalyst system, extensive pilot studies were conducted at Shell's Westhollow research center in Houston to help select the optimal catalyst system for the Valero operation at high liquid hourly space velocity (LHSV). The studies included numerous runs to determine the effects of crude type, viscosity, and asphaltene molecular size on catalyst performance.
This pilot work studied the catalytic reactions from a finite catalyst particle approach. A technical paper was recently published which discussed this resid hydroprocessing philosophy.1
The combination of moderate length catalyst aging studies and finite element analyses resulted in the selection of the 967-317-917-447 catalyst system for the run which started in January 1987.
HYDROTREATING VS. HYDROCONVERSION
Resid hydrotreaters are designed to operate with catalysts which precisely remove the contaminant heteroatoms (sulfur, nitrogen, oxygen, metals) from the ATB feed, whether the molecules that contain these are in the vacuum gas oil (VGO) or asphaltene portions of the oil. Hydrogen is used in the process to replace the contaminant in the reacting molecule and form H2S, NH3, or H2O.
In addition, the highly carbonaceous asphaltene molecule is partially converted to products lighter than the feed boiling range while the heteroatoms are being removed.
Conversion to lighter products by these mechanisms is a function of the degree of contaminant removal, catalyst condition, temperature, and position in the catalyst life cycle. Conversion is normally highest at end of run when the temperature is high and both thermal conversion and hydroconversion are significant.
If the unit is operated at high temperatures throughout the run to force maximum heteroatom removal, and thus increased conversion, this is called mild hydrocracking or mild hydroconversion (the hydroconversion mode discussed in this article). Catalyst life and product quality are significantly affected by coke deposition resulting from operation in this mode.
HYDROTREATING OPERATING MODE
The Valero Arhds unit from January through November 1987 was operated in the normal hydrotreating mode on feeds which were generally blends of Kuwait and Arabian ATB'S, but with some VTB. Fig. 3 shows the contaminant removal capability of the process in the hydrotreating mode for the period.
By November 1987, the weighted average bed temperatures (WABT's) had increased from the start of run at 680 F. to about 695 F. at end of run (Fig. 4). Fig. 4 also shows the feed rate variations during the hydrotreating mode.
The feed to this point, compared to the design feed, was about half as difficult to process. However, through optimization of catalyst shape, the space velocity was maximized 30% above design, offsetting some of the benefits derived from improved feedstock reactivity. This feed rate was not achieved on the first two catalyst cycles.
One other interesting note is that target feed rates and contaminant removals were obtained at a catalyst WABT of 10-15 F. below design.
The lower catalyst WABT optimized hydrogen consumption reactivity while minimizing yield of low valued off-gases (C1-C4). A secondary benefit of lower off-gas yields is increased hydrogen partial pressure, which directly affects coke deposition and cycle life.
HYDROCONVERSION MODE
Valero's operating philosophy shifted as the economic climate changed; better quality feeds became cheaper, and it became profitable to process the better-quality feeds in the hydroconversion mode at increased feed rates.
With this objective, additional pilot studies were conducted at the Shell Westhollow research center to determine process conditions for hydroconversion using the commercial catalyst system, which by this time had 10-months age in the normal hydrotreating mode.
Based on the commercial performance and the pilot studies, proper feed blends were determined for at least another year's catalyst life. Operating temperature was raised from the normal hydrotreating mode of about 695 F.
WABT to about 735 F. WABT.
While the overall quality of the feed from a metals and sulfur standpoint was better than the design Arabian heavy, the inclusion of up to 10% VTB significantly increased the feed viscosity, and thus the severity of operation, specifically on the guard reactor. Fig. 5 shows feed and product analyses for the unit's operation in the hydroconversion mode from Jan. 29, 1987, through July 3, 1989.
During the initial part of the operation, with the Arhds unit in the hydrotreating mode, the conversion to lighter-than-feed-boiling-range products was about 4-5 vol % combined naphtha and distillate (Fig. 6). Conversion was increased to the 15-17 vol % range with the hydroprocessing mode of operation by raising unit operating temperatures (Fig. 7).
During operation in the hydroconversion mode, the first reactor, or guard catalyst (917, 317), plugged when processing a highly viscous immiscible blend. These catalysts had to be replaced with fresh catalyst.
There are various thoughts as to what caused the reactor plugging. Catalysts in resid processing are layered according to the types of reactions that are most desired in each section of the unit.
The initial layer of catalyst in a resid unit should be a catalyst which is active for contaminant removal, but most importantly can remove and retain large amounts of metals. Catalyst particle size should be large enough to give a sizable void volume and allow deep penetration of particulate matter into the catalyst bed.
The catalyst should also encourage the asphaltene molecule to take the most profitable reaction route, that is to lighter oils, not coke. Therefore, the catalyst should also have a mild hydrogenation function.
The ultimate holding capacity of the catalyst for deposited metals and coke is related to the porosity of the catalyst and, specifically, to the ability of the oil to enter the catalyst particle and react efficiently until the pores are filled. The ultimate capacity of the pores is thus equal to the sum of the coke and the metal sulfides present when the catalyst has lost its effectiveness.
Normal hydrotreating produces less coke than does hydroconversion, while metals removal is nearly the same for both modes of operation. Therefore, the coke-to-metals ratio in the pores is higher for catalysts operated in the conversion mode.
The combined effect of coke forming and demetallation reactions on the catalyst life is thus evident. Higher viscosity feeds due to VTB addition only compound the problem by increasing diffusion resistance into the catalyst particle.
However, the ultimate holding capacity of the catalyst may not be achieved if the void space between catalyst particles becomes plugged with material such as precipitated asphaltenes or tramp iron caused by unstable, dirty feed. Analysis of the Valero guard reactor catalyst indicated that this was the most likely cause of the guard reactor plugging.
The inner pores of the catalyst were rendered ineffective by blocking off the very important catalyst feeder pores. The guard reactor catalyst was subsequently replaced with fresh, higher pore volume, 317 catalyst to net higher hydrodemetallation (HDM) combined with higher metals deposition capacity.
The run proceeded in the hydroconversion mode smoothly until temperatures reached maximum operating limits. As the catalyst reached its coke and metals deposition capacity, performance started to decline as expected.
Overall run length was nearly 30 months elapsed time, with actual time on oil of 28 months.
Feed rate was sustained at maximum levels throughout the cycle despite the increase in catalyst WABT and hydroconversion (Fig. 7). Surprisingly, reactor temperature profile was also very stable throughout the cycle, enabling operations to increase catalyst WABT to 765 F. at end-of-run.
CATALYST ANALYSES
An important part of understanding resid hydroprocessing is to study the spent catalyst samples.
Samples of the spent catalyst were taken from the top and bottom of the original guard reactor, the second guard reactor charge, and the top, middle, and bottom of Reactors 2 through 4.
Overall, as mentioned, the function of the catalyst is to take up metals and carbonaceous materials.
Because the overall operation was predominantly in the hydroconversion mode, and thus more coking was taking place, the percentage of total pore volume occupied by metal sulfides should have been less than if operated in the normal hydrotreating mode.
However, actual demetallation was a very acceptable 70-90% of the feed metals during the 30-month operating period.
Fig. 8 shows scanning electron microscope (SEM) profiles of the spent catalyst from the top of the guard reactor.
Each plot, from left to right, represents a scan across a catalyst pellet, with the relative contaminant level shown on the vertical axis.
From these, it is obvious that metals completely penetrated the catalyst, and the catalyst was fully utilized.
REFERENCE
- Adams, C.T., Del Paggio, A.A., Schaper, H., Stork, W.H.J., and Shiflett, W.K., "The Tailoring and Selection of Catalyst Systems for Fixed-Bed Residue Hydroprocessing," Paper presented at the Catalysts In Petroleum Refining Conference, Kuwait, Mar. 4-8, 1989.
Copyright 1990 Oil & Gas Journal. All Rights Reserved.