Search

BROOKS (BP Refining)

I would like to start with a little background on bottoms coke to present some clarity on the basis for my answer to this question. Bottoms coke sources could come from multiple places. But to address the question with respect to residence time, our impression was that we are mostly looking at the thermal cracking and decomposition reactions in the main fractionator bottoms as opposed to asphaltene precipitation or reactor transfer line coke generation and carryover.

BP has done a bit of internal research based on a 1988 paper by TOTAL that was presented in an AKZO catalyst seminar. We looked at the pool residence time, pool temperature, and asphaltene content. Based on that paper, we took those three pieces and combined them with some internal data to regress an equation. We then backed out a coking point temperature (based on about 0.45 wt% coke) as our coke formation point. We ended up with a coking point temperature, and that is what BP targets when we watch our main fractionator for bottoms coking. We do not focus as much on the residence time as we do on this coking point temperature. We have noticed that in this equation, when you look at the different variables of temperature, feed paraffinicity, and residence time, feed paraffinicity and temperature have much greater effects in our equation than residence time, which is why we focus on coke point temperature.

For most of our units, we use the same coke point temperature equation and track a certain temperature for the main fractionator bottoms pool. We will try to keep it about 10°F to 30°F below the calculated coke point temperature. That will then be one of our KPIs, or monitoring points, for our bottoms to help prevent coking in that circuit. Basically, to help control this, we focus on removing heat through the pumparound system. Also, 10 of our 11 BPoperated FCCs have an internal slurry quench system, which includes a slurry pumparound distribution ring near the top of the pool and a stripping steam ring near the bottom of the pool to provide extra agitation. Both of those rings will provide cooling and help us maintain the main fractionator bottoms’ pool temperature below the coking point temperature that I mentioned earlier.

To go back and address the question, though, since it was not really about temperature, BP calculates our residence times based on the main fractionator bottoms’ head volume, and then we divide it by the DCO product flow rate out of the unit. By that definition, BP’s residence times for most of our units are in the hour to hour and a half range. We feel that this is a fairly good proxy for the residence time in the circuit without having to go develop detailed calculations around the volume in the piping or in the exchangers.

If you are considering reducing the residence time, then it is conceivable for you to do this by dropping the liquid level in your main fractionator bottoms, which would then lower the residence time in your main fractionator. And if you are doing any major modifications, you can adjust the volume of the main fractionator head, although that is typically not cost-effective.

When looking at residence time as opposed to temperature, consider that as you drop the time in this pool, you will have a greater chance of building up coke into the bottom of the pool, which will result in extra clogging problems in the main fractionator bottoms circuit. You should always consider your pumparound rates and the minimum pump NPSH (net positive suction head) on your bottoms pumparound circuit. Lowering your residence time could reduce the amount of the NPSH available. But typically, BP focuses more on the temperature and less on the residence time.

PIMENTEL (CITGO Petroleum Corporation)

I have a little to add to what Halle said. CITGO’s basis for residence time calculation is the total volume from the normal fractionator operating level all the way to the inlet passes of the slurry feed heat exchanger divided by both the slurry and recycle. The sum of these two flows represents what is being purged out of the system. Again, there are not many options to reduce residence time without sacrificing the cycle oil yield than operating at your lowest possible fractionator level.

As for temperature control, we do find significant temperature variances between different points in the fractionator bottoms compartment after injecting the quench oil. So, I would recommend that you monitor temperature not only by the temperature of the product leaving the fractionator, but you should also take some readings around the bottoms to find the significant temperature variances.

WILLIAMS (KBR)

KBR designs for a one-minute holdup between the maximum and minimum levels based on the total slurry draw-off rate. Operating the tower at a lower level will reduce residence time within the column.

AVERY (Albemarle Corporation)

I agree with the comments made by Halle, Sergio, and Jesse. My detailed answer is in the Answer Book.

NAVEEN DIMRI (Reliance Industries Limited)

What is the typical CSO (clarified slurry oil) asphaltene content? Is there any limit for the fouling control?

BROOKS (BP Refining)

The equation we use does not include the asphaltene level specifically. The feed quality parameter is around the feed paraffinicity, which is obviously not really the asphaltene. We have that built in to account for feed quality. I have seen some other general references for the industry around feed quality. A lot of people will look at their minimum bottoms temperature and then subtract 20ºF or 25ºF off of that as an extra cushion around feed quality differences, if they have particularly nasty feed.

PIMENTEL (CITGO Petroleum Corporation)

We do not measure asphaltenes, but we do control the API. We have found that when we operate below -3 API or -4 API, the coking tendency increases significantly. So, we have set that as a limit for a minimum API of your slurry oil.

NAVEEN DIMRI (Reliance Industries Limited)

At BP, what is the asphaltene content you normally create?

BILL WILSON (BP Products North America Inc.)

Just to clarify, in this coking calculation, we actually looked at the paraffinicity of the feed. So that is the parameter. Normally, we do not actually measure the asphaltenes in the slurry or, in fact, even in the feed.

BROOKS (BP Refining)

Coke in the main fractionator bottoms section can originate from multiple sources. The key typical sources are coke formed in the reactor transfer lines, thermal cracking and decomposition in the main fractionator, and precipitation of asphaltenes. High catalyst fines in the bottoms can act as nucleation sites for the asphaltene precipitation and thus result in higher bottoms coke formation. Catalyst and coke entrainment can be considered as determined by factors outside the main fractionator. Thus, when focusing on controlling main fractionator bottoms coking, we focus on the key factors influencing thermal cracking and asphaltene precipitation: bottoms pool residence time, pool temperature, and asphaltene content. To avoid coking in the main fractionator bottoms section, we typically do not focus on the residence time as much as we focus on maintaining lower temperatures.

BP has done a significant amount of work internally around the factors that influence main fractionator bottoms coking. Based on a 1988 paper by Total at an AKZO catalyst seminar, the thermal decomposition of DCO is directly related to both residence time and pool temperature. BP used data from this study and additional internal data to regress a relationship between weight percent coke formation in the bottoms, residence time, and pool temperature. Additional studies followed which indicated that FCC feed paraffinicity also has a significant effect on these main fractionator coking reactions. Thus this factor was also added to our regression equation. For our purposes, BP assumes the critical coking point in the main fractionator bottoms at 0.45 wt% coke. The regression equation is solved for temperature at the 0.45 wt% coke level. This temperature is our “coking temperature.” The BP units regularly track the coking point temperature and work to maintain main fractionator pool temperatures 10°F to 30°F below this coking point temperature to prevent main fractionator bottoms coking. Typically, refiners have addressed this by specifying a maximum allowable bottoms temperature which is reduced by 20°F to 25°F for highly paraffinic feeds.

A holistic look at this equation suggests that the effects of temperature and feed paraffinicity on bottoms coking is much greater than the effect of residence time. Typically our main controls around temperature are to install slurry pool quench systems and focus on heat removal in the pump around circuit. Ten of our eleven FCCUs in the BP system have slurry pool quench systems to maintain low pool temperatures. These quench systems typically include a distribution system to deliver cooled slurry pumparound into the base liquid as evenly as possible. This stream should be submerged in the liquid, near the top of the pool. The system also typically includes a steam distributor near the base of the slurry pool to cool the bottom of the pool and aid in mixing the cooled slurry into the pool. This steam ring can have multiple benefits including:

- improving the stripping of the DCO and thus may also improve the DCO flashpoint,

- cooling the slurry in the base of the fractionator and thus reducing coking tendency, and

- providing agitation to help mix the quench and liquid in the fractionator base.

To focus on the base question around residence times, BP calculates the residence time by calculating the main fractionator bottom head volume based on liquid levels as the pool volume and dividing this by the slurry product rate. BP’s typical residence times are between one and 1.5 hours for this system. Some of our units run with residence times over two hours due to large main fractionator bottoms volumes and minimum bottoms make operations.

It is important to note that reducing the main fractionator bottoms volume in an effort to reduce residence time can have the potential to cause issues around build-up of coke shed from the transfer line. Without major equipment modifications, a site could conceivably reduce residence time by minimizing bottoms liquid levels and maximizing the bottoms pump around rates consistent with the bottoms heat removal required to maintain the desired fractionator heat balance. If major modifications or revamps are made to the tower (including a full tower replacement), and a goal is to reduce the residence time in the bottoms pool, modifications can be made to the shape of the bottom head of the fractionator. Changing the shape to a cone or other lower volume head will reduce the residence time. This is typically not a cost-effective modification in most situations which is why we focus on maintaining lower slurry pool temperatures (as mentioned before). If residence time is reduced, it is also important to review the pumps and revamp as necessary for lower NPSH required due to lower liquid head volumes.

PIMENTEL (CITGO Petroleum Corporation)

The basis for the residence time calculation are: the slurry oil flow (as produced from the unit plus flow recycled to riser) and the volume of the main fractionator bottoms compartment from the normal operating level plus the volume of the piping from the bottom of the tower to the inlet of the first set of heat exchangers. There are not many options to reduce residence time other than lower the main column bottoms level. We have found that maintaining a continuous flow of cold slurry quench into the bottom head of the main column reduces coke formation and subsequent plugging of the slurry pump screens or pumparound exchangers. This practice minimizes hot spots in the slurry liquid in the bottom of the tower where coke can form even if the bottom outlet temperature is at target (690°F max). If the bottom temperature drops below our target when the quench flow is at minimum, we reduce slurry pumparound duty at the steam generator. We have experienced slurry exchanger fouling due to the presence of excessive PNAs (polynuclear aromatics) and find that this often correlates with very low slurry API. This can be mitigated somewhat by sacrificing LCO recovery and dropping LCO into slurry.

AVERY (Albemarle Corporation)

Including the composition of the main column bottoms oil and the liquid temperature, an important variable impacting coking in the main column bottoms circuit is the hydrocarbon residence time in the bottom of the fractionator. Excessive time and temperature can lead to coke formation. The oil residence time can be determined from the volume of oil in the bottom of the tower and the bottoms flow rate leaving the tower. The residence time should be minimized by maintaining a minimum bottoms liquid level and maximizing the bottoms pumparound rate consistent with the bottoms heat removal requirement necessary to maintain the desired fractionator heat balance.

SUBHASH SINGHAL (Kuwait National Petroleum Company)

Lower fractionation, bottoms level, and optimum temperature (less than 365°C) are a few considerations to avoid slurry circuit fouling.

WILLIAMS (KBR)

In general, reducing fractionator bottoms coking starts with an optimum slurry system design to minimize the column residence time and bottoms temperature. We typically target around 680°F in the bottoms. This design should include the fractionator soft area, as well as the entire slurry exchangers train.

As far as the bottom of the fractionator, KBR takes two steps to mitigate the impact of coke within the bottom of the tower itself. A coke strainer is installed in the bottom outlet nozzle to retain the larger coke particles while allowing the smaller particles to exit the cat tower into the external circuit. This strainer is a vertical cylinder, slightly larger than the outlet nozzle, with holes on a square pitch extending the full height of the cylinder itself.

To enhance the coke strainer’s performance, KBR specifies a perforated hollow ring steam sparger around the coke strainer cylinder. This sparger fluffs the bottom of the columns stump to keep the smaller coke particles in suspension until they exit with the slurry liquid mixture. As far as the circulating circuit itself and its design, the slurry exchanger operations can finally occur at the velocities held below at threshold. Therefore, KBR recommends an exchanger tube design that allows the operator to obtain a slurry velocity range between 4 fps to 10 fps. To focus on the lower end of that range at two velocities below 4 fps, solids or catalysts can fall out of the solution and, over a period of time, plug the exchanger.

In addition, all exchangers within the train itself, including the steam generator or feed exchangers, should contain at least a one-inch tube diameter to prevent plugging. In some cases, we have to address the coking conditions within the exchanger train. KBR recommends two 60% parallel design configurations for each exchanger within the train. This additional capacity provides flexibility within the system in the event that one exchanger is taken out for cleaning while maintaining feed rate only requiring a slight reduction of feed rates while the cleaning process takes place.

To monitor the slurry performance, KBR installs differential pressure measurements across the duplex filters and tube side of the exchanger train. Both indications are routed to the DCS with a calculated heat transfer coefficient for each exchanger to signal the operator that the system is potentially starting plug. For cleaning and periodic removal of the coke itself removal, KBR recommends the use of the duplex filters I mentioned earlier to remove the large particles that exit the fractionator bottom outlet nozzle. As mentioned, the DPIs (differential pressure indicators) are the key indicators to advise Operations when to clean these filters. Here, instruments the filters have to be isolated, flushed, steamed, depressured, and cooled prior to coke removal. To address the fouling exchanger, one must isolate flush and typically hydroblast the tubes for coke removal.

SCHOEPE (Phillips 66)

I decided to add this example because it was unique. One of our units replaced a TCC (thermo for catalytic cracking) unit. Based on TCC catalyst carryover incidences, they developed a very unique coke trap. Our coke trap is close to 10 feet long. During the last two runs, we accumulated anywhere from four to six feet of coke in this main fractionator, but we never knew about it until we actually shut down. It almost seems to be a self-limiting problem because as you build up the coke, you decrease your slurry pool residence time; and then, you stop coking. In any case, in my opinion, there is really no drawback to designing a very long coke trap.

.

LALL (UOP, A Honeywell Company)

Jesse covered this very well. My comment is that UOP has now made it a practice to take the next slurry product from the bottom of the pump discharge header at the end of that header in an attempt to clear out the bottom circuit of coke and fines and aid in the preferential removal of bits out of the system.

AVERY (Albemarle Corporation)

Coke traps are used on the outlet nozzle at the bottom of the main column to prevent large pieces of debris or coke that may have spalled from the tower during an unplanned short feed outage/shutdown from plugging the main column bottoms piping, valves, or pump strainers. Coke traps can be slotted pipes or birdcages designed such that the open area is equivalent to the outlet nozzle cross-sectional area (no pressure drop). All FCC main columns should have some kind of device in place as coke formation will occur in the bottom part of the tower.

LALL (UOP, A Honeywell Company)

The tendency for slurry oil in the main fractionator bottoms to form excessive amounts of coke is a function of residence time, composition, and temperature. Minimizing coke in the main fractionator bottoms requires proper design and operation of the process variables.

The higher the residence time of the slurry oil in the system, the greater the potential for coke formation. The residence time of slurry oil in the bottom of the main fractionator is simply the volumetric inventory of the bottoms circuit divided by the net slurry product flow rate. The bottoms inventory includes the volume of the fractionator bottoms, slurry exchangers and all associated piping. However, once the unit is constructed the inventory volume is essentially fixed and the only variable available to reduce residence time is lowering of the liquid level in the fractionator bottom which is typically a small fraction of the overall volume.

Keep in mind that temperature affects the rate of coke formation. The hottest section of the system is the counter-current contacting region of the disc and donut trays where the cooled slurry pumparound returned to the main column, de-superheats the entering reactor vapors. Within this zone, it is important to supply enough cooled slurry pumparound flow to provide proper de-superheating while maintaining a wetted surface on all of the trays. Improper contacting and localized drying of the trays can result in hot spots which increase fouling and coke deposition. To ensure proper tray wetting and de-superheating of the inlet reactor vapors, the total slurry pumparound circulation rate is designed to be the greater of either 150% of the design feed rate or 6.0 gpm/ft2 (the equivalent metric unit in cubic meters per hour per square meter equals 14.7 m3 /hr/m2 ) based on the column cross-sectional area.

Increased circulation also reduces the amount of time that the slurry oil sits inside of the fractionators which is the hottest environment of the slurry oil system. Increasing the pumparound rate decreases the mean average temperature of the slurry oil providing some decrease in coke deposition. The less time that material spends in the hottest section of the system, the lower the tendency to form coke.

In general, UOP recommends an initial maximum bottoms temperature of 655°F (345°C) for highly paraffinic feedstocks and 690°F (366°C) for highly aromatic feedstocks. Highly paraffinic feedstocks are typically characterized as having UOP K values ≥12.0 and highly aromatic feedstocks are typically characterized as having UOP K values ≤11.5. These general values are very empirical. The bottoms temperature limit for the main fractionator, however, needs to be confirmed through actual operating experience. Note that these temperature values are without quench recycle. Quench will sub-cool the bottom of the column directionally reducing the tendency to form coke by providing a cooler environment. It is important to note here that the quench is injected directly to the bottom of the column via a distributor to ensure good distribution of cooled slurry and mixing with the hot bottoms to avoid localized pockets of hot liquid.

When using quench, it is very important to constantly monitor the composition of the material. Each refiner needs to identify through trial and error, the maximum bubble point of the material prior to excessive coking. In monitoring the composition of the bottoms material through temperature, it is also important to consider the operating pressure of the main fractionator flash zone. The bottoms temperature will increase with pressure for a bubble point liquid of constant composition. Any temperature excursion in the bottom circuit has the potential to cause thermal cracking of the bottoms material and formation of coke deposits. This risk is compounded if the bottoms material is already near the stability limit for coking.

The second part of the question deals with the removal of coke from the bottoms system. A coke trap with typically 150 mm (millimeters) x 100 mm openings is located over the bottom outlet to prevent large pieces of coke or other debris from entering the bottom line and causing blockages in downstream circuit. Parallel coke strainers are provided at the inlet of the slurry circulation pumps allowing smaller coke chips and catalyst fines to pass through while, very importantly, retaining larger pieces of coke which have passed through the coke trap in the fractionator bottom and can be damaging to the circulation pumps. The coke strainer is a basket offset-type with side entry and bottom exit to avoid accumulation of catalyst fines in the basket strainer body. Finally, UOP has made it a practice to take the net slurry product from the bottom of the pump discharge header at the end in an attempt to clear the bottoms circuit of coke or fines aiding preferential removal of bits out of the system.

WILLIAMS (KBR)

In general, minimizing fractionator bottoms coking starts with an optimum slurry system designed to minimize column residence time and bottoms temperatures (typically below 680ºF). This design should include the fractionator sump area, as well as the entire slurry exchanger train.

KBR takes two steps to mitigate the impact of coke within the bottom of the fractionator. A “coke strainer” is installed on the bottoms outlet nozzle to retain the larger coke particles while allowing smaller particles to enter the external circuit. This strainer is a vertical cylinder slightly larger than the outlet nozzle with holes on a square pitch extending the full height of the cylinder. To enhance the coke strainer performance, KBR specifies a perforated halo ring steam sparger around the coke strainer cylinder. The sparger fluffs the bottom of the column sump to keep the smaller coke particles in suspension until they exit with the slurry liquid mixture.

As for slurry exchangers operations, fouling can occur if velocities are held below a threshold. KBR recommends an exchanger tube design that allows the operator to obtain a slurry velocity range between 4.0 fps and 10.0 fps. At tube velocities below 4.0 fps, solids (catalyst) fall out of solution and over a period of time will plug the exchangers. All exchangers (steam generator and feed/bottom exchangers) within the system should contain at least a one-inch tube diameter to prevent plugging. Measured tube side differential pressures and calculated heat transfer coefficients over time are the best methods for identifying potential coking within the heat exchanger train.

Also, to address coking conditions within the exchanger train, KBR recommends two 60% parallel design configurations for exchanger trains. This additional capacity provides flexibility within the system in the event one exchanger is taken out of service for cleaning.

For periodical coke removal, KBR recommends the use of duplex filters to remove large coke particles that exit the fractionator bottom outlet nozzle. Differential pressure indicators are key indicators to advise operations when to clean each filter. To address a fouling exchanger one must isolate, flush and hydroblast tubes for coke removal.

AVERY (Albemarle Corporation)

I put distillation methods into two different categories. One is simple distillation or acts of distillation, which is either a D86 or D1160. D86 is at atmospheric conditions; the D1160 would be at a vacuum. More people are switching to simulated distillations or using a GC (gas chromatography) to do that. The most common number is a D2887. There are other numbers, depending on if there is oxygen in the stream, if you want light ends, or if you are really focusing on heavy ends.

The GC methods are becoming more common because they require a smaller sample size and because an operator can do far more methods that way. As an example, in our laboratories, we used the D1160s but have now switched to D2887 for most types of streams. We also use high temperature simulated distillation for some heavy resid types. We could run about 12 in a day if we did a D1160. Since we switched to simulated distillation with GC methods, one operator can do a hundred samples in a day by himself. So, we saved quite a bit a time when we switched to using GC methods.

I took a survey of over 60 FCC units worldwide. The slide shows a list of the naphtha streams, LCO streams, and slurry in feed. As you can see, D86 is more commonly used for the light boiling point ranges. This survey also ranged from 2000 to 2012. If I did a poll through that period of time, the ones close to 2000 used more of the simple distillations, like a D86, D1160; most common are all of the GC methods. More people are using what I listed as HTSD, which is high temperature simulated distillation. HTSD is a GC method that helps you measure the really heavy components that boil up to 1350°F.

I spoke with our R&D (Research & Development) guys and asked them to create a graph showing some examples of a D86, D1160, and simulated distillation. The yellow stream represents D1160 as the standard done in a vacuum. If you run a D86, you are going to see more thermal decomposition occurring above 650ºF. I forgot to mention that a D86 is generally not used for heavier boiling points; because somewhere around 600°F to 650°F, you will see thermal decomposition. You can see the D86 represented by the pink line. Once it gets into a higher temperature, about 700ºF, it will have a different slope due. When you start going to GC methods like the D2887, you will see larger tails. So, if you look at D1160 and at the simulated D2887, you will see a larger tail here in green and also a larger tail on the backend.

KEVIN PROOPS (Solomon Associates)

To the panel members in operating companies, as you have changed your distillation methods over time, have you gone back and updated all of your unit monitoring calculations to now calculate standard conversion at a 430 cutpoint using your new methods, or did you ignore the change in distillation method?

AVERY (Albemarle Corporation)

Data from refineries is often generated by using kinetic modeling. Within that process of collecting data, they will have inputs for the method. Those systems will automatically take care of the corrections.

BROOKS (BP Refining)

Our sites use models similar to those Cliff just mentioned.

SCHOEPE (Phillips 66)

I second that for Phillips 66.

AVERY (Albemarle Corporation)

Basically, distillation methods are in two categories: simple batch distillations (D86/D1160), and simulated distillations by gas chromatography. Initially refiners used ASTM D86 to determine oil boiling points (BP). The D86 simple batch distillation, or low efficiency distillation, is performed at atmospheric conditions. Decomposition or thermal cracking of the material can occur at temperatures greater than 650°F (344°C). Due to this high temperature decomposition, the ASTM D1160 method was developed. The D1160 is performed at reduced pressures [typically around 10 mm Hg (millimeters of mercury)]. At this pressure, oil fractions up to 1,000°F (538°C) can be accurately analyzed.

Simulated distillations (SimDists) are performed by GC. The most popular method is ASTM D2887. D2887 determines the boiling point distribution by injecting the oil sample into a GC that separates the hydrocarbons in a boiling point order. The retention time in the GC is related to the BP through a calibration curve. Recently, newer GC methods have been used. High temperature simulated distillation (HTSD) can accurately measure BPs greater than 1,000°F (538°C).

Simulated distillations save a significant amount of time, utilize less manpower, and require a smaller sample size: all advantages over simple distillations. At Albemarle, we run a standard D2887 for lighter BP fractions. For heavier oils such as FCC feed, we utilize a customized HTSD to assure we accurately measure the heaviest BP compounds. Our D1160 capacity was approximately twelve samples/day using one liter of oil and at least one technician. Our D2887 system has a capacity of 100 samples per day using less than one milliliter per sample and one technician.

Data from over sixty FCCUs were reviewed. The data runs from 2000 to 2012 and comes from every WW region. The general observation is that the method analyzed on feed is also performed for the slurry. For naphtha and LCO, many refiners are staying with the D86 method. More recent data and data from larger and/or major oil companies tend to utilize GC methods. Second-generation GC methods (HTSD) are showing up more often.

SCHOEPE (Phillips 66)

All Phillips 66 sites except one use simulated distillation methods for gasoline LCO and slurry distillation. D7096 is used for gasoline, D2887 is used for LCO, and high temperature D7169 or D6352 is used for slurry. One Phillips 66 site still uses D86 for gasoline and LCO.

LALL (UOP, A Honeywell Company)

The distillation overlap of LCO bottoms product varies somewhat. In our experience, it is in the range of 20ºF to 60ºF. We also have a few units that report distillation overlaps. Most main fractionators are not designed with the objective of providing good fractionation between the LCO and bottoms section due to the existence of insufficient trays. The bottoms design is determined by the heat recovery considerations of the fractionator.

Fractionation qualities are a function of the reflux between the LCO and bottoms section. If the reflux is excessive, then less bottoms heat recovery is achieved.

LCO recovery can be optimized by the use of a bottoms quench. The quench is directly injected into the bottom of the fractionator to prevent coking. As the LCO product draw is increased, the bubble point of the bottom's material increases. The LCO draw rate should never be too high to allow the bottoms material to become unstable.

Now I am going to refer to the slide. For a typical FCC main column operation, the LCO boiling range material in the slurry product is roughly 8 vol% (volume percent) to 10 vol% by distillation. If the LCO boiling range material is greater than 10 vol% based on distillation, then an adjustment should be made to the main column operation to achieve higher LCO recovery. Having less than 5% of the LCO material in the slurry product would be considered excellent. Other options utilized by UOP are bottom sidecut strippers using steam and vacuum flushers to maximize LCO recovery.

PIMENTEL (CITGO Petroleum Corporation)

In one of our refineries, we put together an operating scheme to maximize LCO recovery, as shown on the slide. First, we have a sidecut, from the fractionator, of heavy cycle oil that is produced with a cutpoint of 800ºF as hydrocracker feed. Even so, our slurry still has more than 10% diesel-type material.

To enhance recovery, we have reprocessed the slurry back into a vacuum column along with the resid. Slurry oil has a very high boil-up rate in the vacuum column; about 80% to 90% goes out with the distillates. Most of the LCO is recovered as light vacuum gas oil and goes straight to our distillate hydrotreater to produce ULSD.

There is some internal recycle of the 800ºF to 900ºF type of material that will go back to the FCC as medium vacuum gas oil. Of course, the heavy portion of the slurry will go to the delayed coker. With this operating scheme, we can achieve a very high LCO recovery at the expense of some vacuum and FCC capacity. We do not recommend going to 100% recovery because of the possibility of accumulating PNAs in the recycle loop, so we still recommend having some slipstream of slurry as a product.

BROOKS (BP Refining)

Our experience is similar to that of the panel. We have seen a wide range of LCO to bottoms overlaps. Our recommendations fall in line with what Jag mentioned around improving your separation or using the external stripper to recover some of the LCO and reprocess.

PIMENTEL (CITGO Petroleum Corporation)

I know you do not have heavy cycle oil in Solomon.

KEVIN PROOPS (Solomon Associates)

Sergio, how long have you been doing this scheme?

PIMENTEL (CITGO Petroleum Corporation)

For several years.

KEVIN PROOPS (Solomon Associates)

Have you seen any erosion increase in the vacuum unit, especially in the transfer line to the vacuum tower?

PIMENTEL (CITGO Petroleum Corporation)

Good question. No. In fact, the main problem we have had with this occurred at the beginning when we used some flushing oil pumps to perform the recycle. They were not rated for the erosive catalyst fines of the slurry, so we damaged a couple of pumps. We had to replace them with a properly designed pump for the service.

KEVIN PROOPS (Solomon Associates) Do you see any issues in the vacuum and delayed coker?

PIMENTEL (CITGO Petroleum Corporation) We do have some solids accumulation in one tank that is in between; but in the vacuum column and the coker, we have not yet found solids accumulation.

LALL (UOP, A Honeywell Company)

The LCO bottoms distillation overlap (typically 95% to 5% points) varies widely and is in the range of 20°F to 60°F. We also have some units that report distillation gaps. Most main fractionators are not designed with the objective of providing good fractionation between LCO and bottoms section as insufficient trays exist, rather heat recovery considerations from the fractionator bottoms system dictate. Typically, five trays below the LCO draw exist to the HCO pumparound section and below three further cleanup trays excluding the disc and donut section. Fractionation quality is a function of the reflux between the LCO and bottoms sections and if the reflux is excessive, less bottoms heat recovery is achieved. LCO recovery can be optimized by the use of bottoms quench. The quench is directly injected into the bottom of the fractionator from the outlet of the slurry steam generators to prevent coking and polymerization. As LCO product draw is increased, the bubble point temperature of the bottom's material increases. The LCO draw rate should never be too high to avoid the bottom materials becoming unstable and promoting asphaltene precipitation.

For a typical, good operation of FCC main columns, the main column bottoms distillation: IBP (initial boiling point) to 8% to 10% is normally LCO boiling range material. If LCO boiling range material is greater than 10 vol% from the distillation curve, then operation should be adjusted to achieve higher LCO recovery. If LCO is less than 5% in the slurry product, this can be considered as an excellent operation.

To improve separation, a reduction of slurry steam generation or bottoms pumparound heat removal is required. This will increase the column internal reflux and improve separation. Testing and impact on economics [i.e., loss of HP (high pressure) steam versus improvement in separation], however, need to be investigated. With the reduction of steam make, quench needs to be used so that the bottoms temperature does not exceed 680°F to 685°F (360°C to 363C).

In some installations, the separation may not be critical for the refiner. In these installations, LCO is added to the slurry material as cutter stock. If the refiner does not use cutter stock for the slurry, then minimizing LCO in the slurry material should be investigated as above.

A bottoms side stripper using steam and vacuum flashers have been two types of design utilized by UOP to maximize LCO recovery. Note that minimizing LCO in the slurry material will increase coking in the bottom of the fractionator and column bottoms’ heat exchangers. Therefore, good recordkeeping and monitoring trends of unit operation is critical.

PIMENTEL (CITGO Petroleum Corporation)

In our refinery, we produce HCO as a sidecut for hydrocracker feed (800°F cutpoint) which helps increase distillate recovery. Still about 25% of our slurry oil is in the diesel range (700°F minus). In order to enhance LCO recovery, we have re-processed part of the slurry oil in the vacuum column along with the ATBs (atmospheric tower bottoms) drawing. This way the LCO is recovered along with the LVGO (light vacuum gas oil) as hydrotreater feed while the middle cut (800°F to 900°F) is recycled to the FCC along with the MVGO (medium vacuum gas oil). With this operating scheme, very high distillate recovery from the slurry oil can be achieved at the expense of some capacity in the vacuum unit and the FCC.

BROOKS (BP Refining)

BP’s experiences are similar to others on the panel. We have a wide range of LCO/bottoms overlaps, some as high as 80°F. Our recommendations on options to improve this would be to consider packing the beds to improve separation and/or using an external DCO stripper to recover additional high value LCO product.

AVERY (Albemarle Corporation)

There are large numbers of chemical and physical properties that can be measured. I am really just going to focus on the ones that are most typical. First of all, in the metals’ category will be the sodium, alumina, and rare earth. As you know, rare earth is not a compound; it is a group of elements in the Periodic Table. The most common would be lanthanum, cerium, praseodymium, and neodymium, but there are others that have been measured and/or reported. We use XRF (X-ray fluorescence) to measure those components. Other properties that people will look at and monitor on a regular basis are surface area, apparent bulk density, and particle size: either zero to 20 or zero and 40. Average particle size is around 80+. Also, physical testing, such as attrition resistance, is very common. This plot shows typical standard levels in manufacturing variation.

Every FCC unit can have an area of greater concern. For example, a large inventory FCC requires more fines to circulate properly. One FCC unit design may require a certain attrition resistance. More recently, many refiners have been focusing on rare earth variation due to associated surcharges. Consult your corporate group or catalyst supplier regarding the areas where your unit needs tighter specifications.

BROOKS (BP Refining)

At BP, we are cognizant of the fact that vendors have different manufacturing controls and tend to have control ranges for the catalyst. We also understand that these can vary somewhat between vendors. At BP, we typically use the vendor’s supply of fresh catalyst manufacturing ranges. We will check those against catalyst parameters that could cause issues on our units to make sure that variation ranges are still within acceptable operability for us. We are comfortable with the fresh catalyst properties varying in those ranges.

We will periodically review compiled Certificate of Analysis data that we receive for the shipments, and we will track some of the properties to see if we have gone off-spec. We do not typically go all the way off-spec, but we will use this data as the beginning of a discussion if we start to see a parameter that is or could cause issues.

In the past, we have used some of these data to have discussions with our vendors about certain properties. What you see on the slide is one example. At first you might say, “My, that is a high percentage for zero to 40.” But this is on a unit that has a very long standpipe, so it actually helps our circulation. The blue lines on the slide represent the upper specification limits we got from the vendor; the red line shows the lower limits; and, the target is shown in green. We compiled all of the Certificates of Analysis data for the zero to 40, which we would do for other properties also.

We started to notice that the samples tended to trend on the bottom of the specification range, as opposed to the middle. So the request we made was, “Is there any way you can look at your manufacturing to bring us back up near the middle of the range instead of the bottom?” You can see that we changed catalyst formulations because the site and vendor were both able to do that. There is now more variation across the targets rather than at all of the points concentrating below the target line. That is typically what we look at when reviewing fresh catalyst property data and tracking from the vendors.

You can also track your e-cat property data. For certain properties, you may be able to see a starting indication of issues that could be caused by your fresh catalyst. For examples like those that Cliff mentioned regarding tracking your rare earth, you should be able to see reasonable responses in your e-cat that follow what you are getting from your fresh catalyst.

MICHAEL TEDERS (Valero Energy Corporation)

Halle, you said you look at the rare earth content for the fresh catalyst. Have you ever done your own independent analysis for that fresh catalyst to understand if you are really getting what you are buying?

BROOKS (BP Refining)

We have had third parties look at it on occasion, but we do not do that on a regular basis. It is more of a situation that we start to see a problem on some of the data or a property that does not look right, like being unable to find the rare earth we think is supposed to be there. Then we will take it to the third-party.

MICHAEL GRECZEK (BASF Catalysts LLC)

I agree with Cliff and what he showed. I want to add that I think it is basically a question of specification. Let’s just say you have a fresh catalyst surface area specification of 280 m2 /g (meters square per gram) minimum, a 300 m2 /g target, a 320 m2 /g maximum, and also a rare earth specification of 1.8 wt% minimum, 2.0 wt% target, and 2.2 wt% maximum. These are very typical specifications. If you deliver within those specifications, the delivered fresh catalyst is all in spec. But if you have a catalyst delivered with the minimum surface area and maximum rare earth versus a catalyst delivered with the maximum surface area and minimum rare earth, then I want to make sure everyone knows that these catalysts are catalytically different, although both are “in spec”. You can be in spec, but what you really want to be is around the target. That way you ensure constant catalytic performance. BASF spends a lot of money to make sure we deliver fresh catalysts that are always around the target.

AVERY (Albemarle Corporation)

There are a large number of chemical and physical properties that can be measured on fresh FCC catalyst. I will first answer this question by mentioning the chemical and physical properties most commonly reported. The most common chemical properties are Na, Al2O3 and REO (rare earth oxides). RE, or rare earth, is not an element; but rather, a group of elements. The most common RE elements are La2O3, CeO2, Pr6O11 and Nd2O3, which are the oxides of lanthanum, cerium, praseodymium, and neodymium. The most common physical properties are surface area (SA), apparent bulk density (ABD), particle size distribution [PSD, 0 to 40 and APS (average particle size)], and attrition index (AI). The specifications of these properties are a balance between the catalyst plant capabilities and the FCCU requirements. A common two-sigma variation is represented in the slide.

Every FCCU can have an area where there is greater concern. As an example, a large inventory FCCU may require more fines to circulate properly; a certain FCCU design may require a certain attrition resistance. More recently, many refineries are focusing on the REO variation due to associated surcharges. Consult with your corporate group or a catalyst supplier regarding the areas where your unit needs tighter specifications.

BROOKS (BP Refining)

We are cognizant of the limits mentioned before around vendor manufacturing controls and also note the control ranges for different vendors can vary somewhat. Since the individual vendors know their ability to meet certain parameters in the fresh catalyst, BP typically uses the vendor supplied fresh catalyst formulation specifications to determine an acceptable range of variation for received catalyst properties. In each case, the limits provided by the vendor are reviewed to assure that the stated variability will not significantly impact the performance of the unit if the fresh catalyst properties move between the given limits. We periodically review compiled certificate of analysis data on received shipments to determine if the product is meeting the base specifications. It is important to note when reviewing this data that the variation in certain fresh catalyst properties, such as volatiles, can increase when the vendor moves between manufacturing plants or becomes capacity limited. We have also used statistical process control rules on this data to watch for trends consistently above or below the target property value, even if they are inside the specification range. Based on these analyses, we have made requests of our vendors to target tighter property specifications if necessary.

This is an example of a review graph we have used for fresh catalyst data. In this example, we noticed that the fresh catalyst wt% 0 to 40 micron was trending regularly within the upper and lower manufacturing limits, around the specified target of 16. However, it was noted that we were receiving shipments of Catalyst B and C that were typically trending on the low end of the target range. Since this is a critical property for this unit due to circulation limits, we made a request that the vendor look into their manufacturing process to see if they could shift the property back toward center. As can be seen in the Catalyst D data above, the vendor was able to make slight adjustments to bring the range back to historical patterns, with values both above and below the target.

It is also recommended to consider your fresh catalyst properties when reviewing your vendor e-cat sample results. Some shifts in e-cat property trends, such as rare earth levels, can sometimes result from shifts in fresh catalyst quality (among other possible causes).