Charles F. LeRoy, Michael J. Hanshaw, Steven M. Fischer, Tariq Malik, Randall R. Kooiman
Champlin Refining & Chemicals Inc.
Corpus Christi, Tex.
Champlin Refining & Chemicals Inc. is systematically improving its hydrotreating operation through a program of research and commercial unit testing.
The program is helping Champlin learn to optimize catalyst deactivation rates and run lengths by altering heavy vacuum gas oil (HVGO) cutpoint, unit severities, temperature management approach, and other parameters.
This first of two articles reports significant observations from full scale runs in which catalysts and other variables were changed. A second article that will appear next week describes the approach to optimization of the hydrotreating catalyst system and the economic impact of the project.
The medium-sized, fully integrated refinery located in Corpus Christi, Tex., uses the Unibon (UOP) hydrotreating process to process vacuum and coker gas oils from heavy Venezuelan crude.
The Champlin facility is wholly owned by Petroleos de Venezuela SA and currently is running 100% Venezuelan crude at an average rate of 130,000 b/d. The blend that is processed is a combination of several different Venezuelan crudes. Table 1 shows the analysis of a typical crude blend processed at the refinery.
Table 2 shows the crude slate data for 1983 through 1990. Fig. 1 shows a simplified process flow diagram of the heavy oils complex with typical unit rates. Table 3 shows the Unibon feedstock characteristics typical for this operation.
The period covered in this article is Apr. 10, 1986, to Dec. 31, 1990. This covers six catalyst changes, which are referenced as follows:
- Run 10, 4/10/86-11/i/86, 205 days
- Run 11, 11/12/86-5/9/87, 178 days
- Run 12, 5/23/87-2/14/88, 267 days
- Run 13, 3/25/88-2/11/89, 323 days
- Run 14, 2/27/89-3/11/90, 377 days
- Run 15, 4/4/90-2/23/91, 325 days (and continuing).
A summary of the performance data on each run is shown in Table 4. The type of catalyst used for each run is presented in Table 5.
RUN SUMMARIES
It is useful to provide a short summary of significant events that occurred during each run.
RUN 10
The analysis for the report begins with Run 10. This run suffered a short length--only 205 days (19.1 bbl/lb)--and rapid catalyst deactivation at the end of the run. These problems were caused by the introduction of a higher percentage of Venezuelan crude and the onset of deep-cut operation in the vacuum unit.
Deep-cut operation allows the vacuum unit to recover additional HVGO and increases the HVGO cutpoint from 1,000 F. to 1,100 F. These two factors caused high metals levels in the Unibon feed--10-15 ppm (Ni+V)--and premature catalyst deactivation.
RUN 11
Run 11 was the shortest run in the series. It lasted only 178 days. The catalyst life was a little higher for this run at 24 bbl/lb because feeds rates were higher. Severity had been reduced to try to extend the run length.
Toward the end of this run, it was suspected that the metals were high in the Unibon feed and that metals deactivation was causing the short run lengths. The vacuum unit and coker unit operations were reviewed to find the source of the metals. The wash oil rate in the vacuum unit was increased, which reduced the Ramsbottom carbon from 0.9 wt % to 0.5 wt %
RUN 12
Demetallization catalyst was first loaded into the top section of the first (or lead) reactor for Run 12. This catalyst acted as a guard bed and absorbed the metals in the feed. The severity was again reduced to extend the run length.
The metals level in the HVGO was controlled by adjusting vacuum wash oil rates and HVGO cutpoint. Metals were maintained in the 3-5 ppm (Ni+V) range, which translated into a 1,060-1,070 F. endpoint for HVGO. It was calculated that this metals level would balance the deactivation rate attributable to metals with the deactivation rate attributable to coking.
There were problems in the vacuum unit in December 1987 that caused packing damage. This produced high metals contamination in the HVGO until the vacuum unit was repaired during the scheduled turnaround in February 1988. This upset increased metals contamination in Run 12. Run 12 lasted 267 days, or 38.5 bbl/lb, which is a 30-40% improvement over Runs 10 and 11.
RUN 13
Run 13 was a continuation of the efforts started in Run 12. Additional thought was put into the demetallization catalyst selection. It was decided that two different suppliers' systems would be tried in the two parallel reactors.
Champlin's Unibon unit is unique in that it has two independent parallel reactor systems. This allows side-by-side catalyst trials in a commercial unit for separate evaluation.
The feedstock processed during this run was of slightly better quality than earlier runs. Therefore, the demetallization catalyst was not pushed as hard as the original plans designated.
One of the catalyst systems did show superior stability during the run, but the other supplier's system showed superior performance in desulfurization.
The run lasted longer than any of the previous runs; 323 days, or 44.9 bbl/lb. This was a 15% increase over the previous run and was accomplished at a higher severity.
There was life left in the catalyst at end of run (EOR), as is shown by the EOR weighted average bed temperature (WABT) of 710 F. The maximum EOR outlet temperature is in the 780-800 F. range, with the mechanical limits of the reactors being 850 F.
RUN 14
Run 14 was a transitional run. Both reactors were loaded with the catalyst system that showed the better desulfurization performance in Run 13. During this run, the emphasis shifted from stability concerns to increased severity, to improve the quality of the FCCU feed.
The demetallization catalyst appeared to be controlling the metals that had caused short run lengths.
During the previous two runs (12 and 13), severity had been reduced to ensure run lengths. Both runs met the scheduled turnaround dates, with catalyst life remaining.
During this run, the temperatures were raised more aggressively and the philosophy of temperature-management was reviewed and modified.
This run was the longest run to date; 377 days, or 54.7 bbl/lb. This is an additional increase of 15% over the last run. There was catalyst life remaining after this run, as indicated by its EOR temperature of 730 F.
RUN 15
Run 15 utilized the more stable catalyst system from the Run 13 evaluation. The temperature management approach was changed from an equal average bed temperature scheme to a more severe second reactor catalyst operation. This established a 20-25 F. delta WABT between the two reactors.
This was done to increase severity and better utilize the second reactor catalyst.
Run 15 has been operated with a 5% increase in desulfurization and a 10-12% increase in denitrification. The catalyst stability is still good, with a projected catalyst life of 50 bbl/lb at EOR.
This temperature management approach is a better balance of deactivation rates between the first and second reactors.
This discussion of Runs 10-15 was intended to show the basic progress achieved during the past 5 years. Fig. 2 shows WABT vs. catalyst life for the five runs.
METALS ANALYSIS
During Run 11 in early 1987, Champlin discovered a problem with the analytical test method used to test vanadium in HVGO from Venezuelan crudes: vanadium results were off by a factor of ten. Instead of being 0.5-1.0 ppm, they were 5-10 ppm. This explained the rapid deactivation observed in Run 10.
Champlin began investigating the analytical methods for testing HVGO for metals such as vanadium and nickel in the ppm range. At that time, the refinery was using the Graphite Furnace Atomic Absorption (GFAA) method for determining the metals content of hydrocarbon samples. The investigation showed that certain compounds called refractory monoxides could be formed in the furnace. These compounds caused the vanadium atoms to be organically bonded.
As the furnace went through its heating stages, some vanadium was burned off with the organic material, producing a low recovery when reaching atomization.
Conversations with other refineries and vendors revealed that this problem could be minimized by adding 30% sulfuric acid with 0.1% Triton X-100 to the sample. This prevents refractory monoxides from forming. This was done and the measured vanadium levels were higher.
Additional work on retained HVGO from other crude materials showed that this problem was peculiar to the Venezuelan feedstock. The types of compounds that form these refractory monoxides in the analyzer are present in Venezuelan HVGO material.
During this period, the GFAA procedures were reviewed and modified. An analysis of the reproducibility and accuracy of this method for the analysis of vanadium in the 0-20 ppm range in HVGO revealed that this procedure had an accuracy of 30% by the statistical process control (SPC) techniques used.
This was not acceptable, so a study was done to look for alternative methods and equipment to replace the GFAA. This study recommended the purchase of an Inductively Coupled Plasma Atomic Emission Spectroscopy (Icpaes) analyzer.
This analyzer can measure the elements Ni, V, Fe, Cu, Na, and Cr in oil and water samples. It has a wide linear range of 0.1-2,000 ppm and an accuracy of 1.6%.
Matrix effects are minimal with Icpaes. It is easy to use and can be preprogrammed to run sample analyses. Icpaes samples require minimal preparation.
The GFAA method was very time-consuming, taking six times longer to run samples than the Icpaes method.
All these factors led to the purchase of the Icpaes for the refinery.
This was a significant event in the troubleshooting effort. Later analyses revealed that the difference between 3-5 ppm (Ni+V) and 7-9 ppm (Ni+V) was significant, and that the accuracy of this analysis was very important.
We have established an SPC program on this test, and many others, in the laboratory.
A standard sample is injected into the analyzer at regular intervals and the results are collected and plotted on a Shewhart control chart (X-bar and R-chart). If the results fall out of the control limits, the analyzer is checked for problems by the lab technical support staff. Fig. 3 shows an SPC chart for the vanadium test.
METALS IN UNIBON FEEDS
After it was discovered that high metals in the Unibon feed were causing the rapid catalyst deactivation rates, a program was started to define the sources of the metals. This program involved four areas of activity:
- The study of the metals distribution in HVGO according to boiling range
- The survey of other refineries for input on metals levels they observed
- The adjustment of vacuum unit operations
- The adjustment of coker unit operations.
Champlin decided to analyze the HVGO fraction from 900 to 1,100 F. for metals and determined the metals distribution in the base crude (BCF-22). BCF-22 was distilled under vacuum in the control laboratory to provide a 900 F. residue comprising approximately 41.5 vol % of the crude.
Three gal of this material were sent to an outside laboratory where they were further distilled using a short path, high vacuum, wiped-film evaporator (Distact). The overhead and residue samples were analyzed in an Icpaes to determine nickel and vanadium content.
Compounds of both metals were found to be volatile. Metal compound volatility was nonlinear with respect to percent distilled. Vanadium compounds are more volatile than nickel compounds.
The data were regressed to develop distribution curves of metals as a function of percent distilled. These curves are presented in Figs. 4, 5, and 6.
The curves follow somewhat of an "S" shape. The change in metals as a function of distillation is quite sharp in the lift range in which the vacuum unit is typically run. The conclusion of this work was that metals are present in the boiling range of 900-1,100 F. and increase nonlinearly between 1,000 and 1,100 F.
A telephone survey of several refineries was taken to collect reference data on other refineries' operations. There was a wide variety of comments and observations.
Several had observed metals (Ni+V) in HVGO samples in the 1-20 ppm range, Many supported the idea that metals varied dramatically with HVGO cutpoint. Several had suffered metals deactivation in HVGO hydrotreaters and had made cutpoint reductions in the vacuum unit to reduce metals. The metals levels of heavy coker gas oil (HCGO) was also discussed.
It appeared that typical metals levels of HCGO were lower than HVGO. Metals were in the range of 0.5-5 ppm total metals (Ni+V) in HCGO.
The introduction of deepcut operations in late 1985 included three major changes to the vacuum unit: a change of column internals to high efficiency packing, the injection of steam into heater coils, and a revamp of the heater, resulting in a change from four passes to eight passes. This project was completed in March 1986 and put into service in August 1986, which is about the time the deactivation rates accelerated.
We believed that the 100 F. increase in the cutpoint of HVGO increased the metals dramatically. This was proved in 1987.
Several vacuum unit adjustments were made to ensure no metals or Conradson carbon carry-over was occurring in the tower. The grid wash was optimized to minimize carry-over of heavy material. This optimization was limited by spray header pressure differential limits and packing efficiency.
The flash zone vacuum was maintained at 30-36 mm Hg with cold HVGO pumparound. The transfer temperature was set at 780 F. with velocity steam. Table 6 presents a set of typical operating data for the vacuum unit.
Since the deep-cut project was put into operation and the metals deactivation problem developed in the Unibon, the vacuum unit HVGO cutpoint has been controlled to keep the metals content (Ni+V) of the HVGO in the 36 ppm range.
The calculated total metals accumulated on the catalyst at EOR for Runs 12-14 averaged 5-6 wt %. The catalyst suppliers indicate that you can have as much as 9-10 wt % metals and still have acceptable activity.
Fig. 7 presents data on metals accumulation (Ni+V) for Run 15, in wt % vs. bbl/lb, through Oct. 31, 1990. It shows a family of curves based on total feed metals (Ni+V) ranging from 3 to 10 ppm, and the impact of feed metals on total accumulated metals at EOR.
It was decided that the last 3-4 months of operation would be run at 6-7 ppm (Ni+V), which will produce 7-8% total accumulated metals on the catalyst. This is a 30-40% increase over previous runs.
It should be noted that these total accumulative metals numbers are calculated by subtracting the total weight of metals in the product from the total weight of metals charged to the unit. This accumulation is then divided by the total weight of catalyst loaded.
This is an indication of overall metals loading, but it does not indicate metals distribution in the reactor. Most of the metals laydown is in the first reactor, as indicated by spent catalyst analysis.
The coker unit operation was also reviewed. The HCGO averaged about 4 ppm vanadium. This is on the high side, compared to the other refineries surveyed. Adjustments were made to the coker unit fractionator to clean up the HCGO fraction. Below-pan wash was utilized on drum switches as well. Advanced control techniques were implemented in 1989, which have improved the operation, reduced the impact of drum switches on the fractionator, and improved the quality of the HCGO. These adjustments have reduced the vanadium levels to the 2-3 ppm range in the HCGO. Table 7 shows typical coker unit operating data.
Copyright 1991 Oil & Gas Journal. All Rights Reserved.
