AFPM Q&A—2: Annual conference highlights FCC problems, solutions

Sept. 3, 2012
Fluid catalytic cracking was the focus of extensive questions and discussion at the 2011 American Fuel and Petrochemical Manufacturers's Q&A and Technology Forum, Oct. 9-12, San Antonio.

Fluid catalytic cracking was the focus of extensive questions and discussion at the 2011 American Fuel and Petrochemical Manufacturers's Q&A and Technology Forum, Oct. 9-12, San Antonio.

In a discussion among panelists and attendees, this annual meeting addresses problems and issues refiners face in their plants and attempts to help them sort through potential solutions.

This is the second of three installments based on edited transcripts from the 2011 event. Part 1 in the series (Aug. 6, 2012) focused on high-temperature hydrogen attack. The final installment (Oct. 1, 2012) will focus on crude, vacuum distillation, and coking.

This FCC session employed six panelists (see accompanying box). The only disclaimer for the panelists was that they discuss their own experiences, their own views, and the views of their companies. What has worked for them in their plants or refineries might not be applicable to every situation, but it can provide sound guidelines for what would work to address specific issues.


What can cause exothermic reactions in propylene driers and guard beds and how can these reactions be prevented?

Lorsbach: By way of introduction, propylene-recovery units are located downstream of FCC units and consist of a C3/C4 splitter, a de-ethanizer, and a C3 splitter in series.

Propylene from the C3 splitter overhead goes through a series of treaters for final propylene product polishing: usually regenerable molecular sieve driers for moisture removal and then one or more guard-bed adsorbent vessels for removal of carbonyl sulfide (COS), phosphinearsine, and other trace contaminants. Typically, the guard-bed adsorbents are metal oxide bound on alumina.

Occasionally, multipurpose hybrid absorbents are used that will take out both moisture and COS, for example. So, there are a variety of final treatment adsorbents that can be used in propylene-recovery units to remove trace contaminants to produce polymer-grade propylene.

With regard to molecular sieve driers, there is a small exotherm that occurs when you initially charge liquid propylene to the drier, typically on the order of 10° F. or 15° F. This heat is dissipated by the flow of liquid propylene through the drier. Exothermic runaways are unlikely to occur in moisture driers if they are operated correctly with proper liquid flow through the drier adsorbent. We will see a bit later where you can have trouble when it is operated incorrectly.

There are a number of risk factors that can cause temperature excursions in these drier and guard-bed systems. Introduction of flammable or reactive fluids into a vessel containing air; introduction of a high concentration of a reactive, strongly adsorbed material into fresh; or regenerated absorbent are some examples. If this risk is present, then you can use a low-reactivity absorbent material to mitigate the risk of a thermal excursion.

The third risk factor is use of a highly reactive fluid to heat or cool the absorbent bed. Examples of such fluids are ethylene, propylene, and other olefins. As mentioned, when the equipment and absorbent operating guidelines are followed closely, there is usually only a small absorption exotherm that is dissipated by normal flow through the equipment.

There have been two incidents that UOP is aware of in propylene-recovery units downstream of FCC units, and there are more incidents of this type in the petrochemical industry. These kinds of guard beds and absorbent beds are used widely in petrochemical processes. The two examples I have for FCC units are the following:

• In one case, a board man changed the temperature permissive on a drier outlet temperature after the drier was regenerated. It had been 50° C., and he changed it to 170° C. This change in the permissive set point allowed the regenerated drier vessel to be put back on stream in the lag position where it discharged hot propylene into a downstream metal oxide guard-bed vessel.

Because the metal oxide guard-bed absorbent is sensitive to temperature, the high propylene inlet temperature initiated a spontaneous exothermic of the metal oxide component of the adsorbent to the corresponding elemental metal. Essentially, it quantitatively reduced all of the metal oxide in those two guard beds, releasing a lot of heat that fused the absorbent and blistered the paint on the vessels and the piping, but the vessel did not lose containment.

• The second incident was more serious in that the vessel lost containment. In this case, the guard-bed vessel was loaded with two adsorbent materials. The first adsorbent layer was for removal of moisture and trace oxygenates, and the second adsorbent layer was for COS removal.

In this case, rather than charging liquid propylene, the refiner charged vapor-phase propylene. The adsorbent beds channeled and were not able to dissipate the heat from the exotherm. Basically, the undissipated heat caused the carbon steel vessel shell to overheat and the vessel to rupture at the interface between the drier adsorbent bed and the adjacent COS removal adsorbent bed.

Agnello-Dean: As Tom said, the number of incidents ended up being relatively few; so I was only able to find a couple within our history files. The incidents did not really appear to have any common causes associated with propylene as the reactive material.

The propylene does have the potential to have an exotherm. Usually, it is not noticeable, except for a few degrees. If the material is selected properly for the propylene, as far as the right sieve size, then you can help minimize that exotherm.

So you really need to make sure you have the right material selected for propylene service and also that your start-up procedures really allow for monitoring the temperature and looking for the exotherm, recognizing that the formation of the exotherm is a potential problem, and then reacting to it if you start to see the temperatures take off too quickly.


What are the potential impacts on FCC LPG and FCC gasoline properties from processing coker off-gas into the FCC gas plant and from processing coker naphtha in the FCC riser?

Teders: This question is about coker products in an FCC. These products are known to contain more sulfur H2S, diolefins, and sometimes mercaptans than the typical FCC products. Because this may overburden an LPG sulfur extraction unit, care needs to be taken.

We did experience excessive sulfur and light cat naphtha when a coker light end stream was routed to the FCC gas plant. The coker stream contained high levels of H2S but did not contain an excessive amount of gasoline boiling range mercaptans when it was measured in the lab.

However, when the coker stream was removed from the FCC gas plant, the mercaptan content in the light cat naphtha dropped to normal levels. We think there was a recombination reaction with H2S and olefins that was the mechanism responsible for the excessive amount of mercaptans in the FCC light cat naphtha.

At another location, coker naphtha sent to the FCC gas plant resulted in excessive fouling in the debutanizer thermosiphon-style reboiler. No coker naphtha is sent to the riser except on occasion when there is coker naphtha in the slops when that happens.

English: The coker LPG will also tend to be less olefinic than the FCC LPG. That will dilute the FCC LPG and have an effect on downstream units, like the alkylation unit, or possibly propane propylene splitters. There will not really be much difference if it goes through the FCC or blends in downstream. The coker naphtha will tend to be a lower octane than the FCC gasoline as well. Again, if you put it in the riser or if you blend it in some place downstream, you will see the same kind of effect.

We have observed that if you put the coker naphtha into the FCC riser, although there would be very little conversion of it to lighter material, about 20% to 30% of the sulfur will be converted to H2S. On the other hand, if you inject the coker naphtha through a separate injector below the FCC feed, you could get a significant amount of conversion to LPG with additional improvements in octane and conversion of sulfur to H2S.

That said, there is also going to be a heat-balance effect of putting any light material, such as coker naphtha, into the FCC riser. You are going to get a higher cat/oil ratio. It may appear that there is conversion of the coker naphtha when really what you are seeing is more conversion of the gas oil because of a higher circulation rate.

Dalip Soni (Lummus Technology): We have experienced self-injecting coker naphtha into the FCC riser and seen the effects already mentioned by the panelists. Self-injecting coker naphtha removes H2S, upgrades the octane number of the coker naphtha, and produces more LPG. That is quite a good way to upgrade coker naphtha into the FCC and saturate the diolefin.

Dan Webb (Western Refining Co.): When you add coker naphtha to the riser, what is the effect on the other product yields, like distillate yields and gasoline?

English: If you just put it in there and do not make other processing changes, I would expect there really will be no direct effect. The indirect effect is that the cat circulation rate will increase and the yield of light cycle oil (LCO) will decrease. So, if you keep riser temperature and preheat constant, you will see a reduction in the LCO yield because of an increase in cat circulation rate and perhaps a higher specific gravity and lower cetane index.

Typically, though, people do not have the freedom to do that because they are running against some kind of constraints. If you put in coker naphtha, then you might have to drop the riser of the temperature to stay within a wet gas, for example. So, in any given unit, the end result is going to look a lot different.

Phillip Niccum (KBR): Just as an additional point of reference: KBR offers a Superflex process that has a very high temperature reaction on ZSM-5 type additive. Given the traditional FCC type of residence time, we can get a very high conversion of olefinic naphtha, such as the coker naphtha, in it.

Going back to Mr. English's comment, if you inject the coker naphtha low in the riser, then you may also be doing something to compromise the cracking of your fresh feed that is coming in higher and may see a cooler temperature or less severe cracking.

So really, if your objective is actually to upgrade the naphtha into something light, such as propylene or BTX, then you really should have a second riser and some ZSM-5 additive.

Michael Wardinsky (ConocoPhillips Co.): Before anyone runs back to your plant and says, "I have a great idea. Let's run coker naphtha to the FCC," make sure you understand what is going to happen to the FCC gasoline sulfur content because that coker naphtha typically has several thousand ppm sulfur in it. You could wipe out your entire corporate gasoline sulfur average.

Tesoro Corp.'s 120,000-b/d Anacortes refinery lies about 70 miles north of Seattle on the Puget Sound. It primarily supplies gasoline, jet fuel, and diesel to Washington and Oregon, and manufactures heavy fuel oils, LPG, and asphalt. Its crude feedstock arrives via pipeline from Canada and by tanker from Alaska and foreign sources. Photo from Tesoro Corp.

With the Tier 3 gasoline sulfur standards coming in the US, I doubt that processing coker naphtha in the FCC is going to be possible, unless you have a really stout cat gasoline hydrotreater on the backend.


What conditions or contaminants will deactivate ZSM-5 additive? What is the half-life of ZSM-5 in clean feed operation? Will contaminants such as vanadium, sodium or other metals adversely affect the propylene selectivity of ZSM-5? What is the best way to monitor the effectiveness of the ZSM-5?

Fletcher: The deactivation mechanism for FCC catalyst is primarily related to unit cell size reduction and, eventually, collapse or sintering of the zeolite crystal. The mechanism of ZSM-5 deactivation is quite different.

The deactivation mechanism is simply the dealumination of the ZSM-5 crystal, and activity is lost through the loss of active aluminum sites. The crystal structure does not collapse. The activity retention and half-life of the ZSM-5 additive in the circulating inventory is strongly affected by hydrothermal conditions within the regenerator with temperature being the dominant variable.

ZSM-5 additive activity is less affected by contaminant metals than is FCC catalyst due to the fact that heavy feed molecules containing contaminant metals, such as vanadium, are less likely to crack on ZSM-5. ZSM-5 will therefore maintain its activity longer than will the FCC catalyst.

It is worth pointing out that a unit experiencing high equilibrium vanadium levels will likely experience a loss in conversion, which will reduce LPG yields. This loss may give the appearance of a ZSM-5 effect. The propylene selectivity will likely remain unaffected.

The activity retention difference between ZSM-5 and FCC catalyst will increase as the equilibrium metals level increases. Intercat has evaluated ZSM-5 additive half-lives for several units and found a typical half-life of about 18 days, with a minimum of 2 days and a maximum of 36 days.

ZSM-5 additive activity in an operating unit is strongly affected by the catalyst replacement rate. Units having a very high replacement rate present a higher average ZSM-5 activity than units with very low changeout rate. A paper presented at the 2000 American Chemical Society conference investigated the subject of LPG selectivity differences in detail. This study reviewed additives having different ZSM-5 crystal content, different levels of additive additions in the FCC, additives from different manufacturers, additives with different silica-to-alumina ratios, and additives steamed at different severities.

The results of the study can be plotted on one chart. These data demonstrate that if one additive were more selective than another, the propylene yield would fall on a different line, which did not occur. All additives tested at all concentrations fell on the same line. We also found that the propylene yield increases faster than butylene yield and that higher delta LPG yield leads to higher propylene yields.

Additive zeolite content, type, method of manufacture, and steaming severity have no effect on the selectivity of the final LPG product. The ratio of propylene to butylene in the final product depends only on how much LPG is made. The conclusion is that ZSM-5 additive selectivity is determined by the zeolite structure alone. Therefore, measuring activity differences is more important than looking for selectivity differences with a standard ZSM-5 additive. (Please note that these results apply only to standard ZSM-5 technology.)

While propylene selectivities are determined by the ZSM-5 crystal structure, the activity and stability of the various additives are determined by the crystal stabilization technology employed plus the interaction of the crystal with the matrix. An additive containing properly stabilized ZSM-5 crystal, combined with a strong matrix, will result in excellent activity retention with superior propylene yields when compared to other technologies.

Intercat and other suppliers have invested significantly in development of ZSM-5 additive technology, which is reflected in our broad product portfolio. Intercat possesses an extensive range of ZSM-5 additives maximizing propylene, butylene, and octanes. Additionally, Intercat produces ZSM-5 additives that minimize LPG increase for wet gas compressor-limited operations.

There are several selectivity-based ratios that can be used in monitoring ZSM-5 performance. These include: propane olefinicity, propylene yield vs. LPG, propylene vs. conversion, propylene vs. butylene, and propylene vs. gasoline. The most important of these ratios are the propane olefinicity and the propylene-to-LPG ratio.

Meyers: Our planning and economics team monitors LPG yield shifts at varying ZSM addition rates. The goal is to maximize the alkylation unit at reduced FCC feed rates.

So, the vectors of C3 and C4 olefinicity per percent ZSM 5 per feed rate are used in the LP model. Typical results would be maybe 3 vol % C3/bbl/ZSM loading and 2 vol % for C4/bbl/ZSM loading. Then, our half-life and clean vacuum gas oil (VGO) operation was about 18 days.

Joseph McLean (BASF Corp.): We had a unique opportunity about a year ago. We had a customer operating with a very high level, I would say, a record high level of ZSM-5. The customer went from a very clean hydrotreated VGO operation to running some resid that was really loaded with nickel. The e-cat nickel went from a couple hundred ppm to above 3,000 ppm in a period of just a few weeks. Because that is not something we see very often, we were able to study it in some detail.

Then following that, the customer stayed at the high nickel level for a while and started antimony injection. So, we got to see all three periods: the clean period, the high nickel without antimony, and the high nickel with antimony. We were able to analyze the e-cat microscopically and spectroscopically.

We saw essentially no nickel on the ZSM-5 particles. There were a lot of them there; so it was not just a geometric effect. Vanadium may be different because it moves around a lot more than nickel. This was really a quick effect. But some of the other numbers we have seen indicate that this may be half the level. We had a hard time seeing any on the ZSM-5 particles. It was essentially all on the active catalyst particles.

John Zoller (Albemarle Corp.): When we analyze the refiners' e-cat, we get a complete yield structure. This is a wonderful way to follow how ZSM-5 is working in the commercial unit by observing the changes in the C3 and C4 olefins in the LPG yield from the laboratory test unit.

Benchmarking FCC units

What are your recommended practices and categories for benchmarking FCC units? Include, as appropriate, process performance, reliability, capital efficiency and operations.

English: For most different types of process units, we recommend several basic categories of benchmarking. These include hardware, reliability and maintenance practices, energy efficiency, process performance, safety and process management, operator training, and now greenhouse gas production.

The key in each category is to remember that the benchmark is not used to tell you directly what you should change. It is used to show you what areas to look at. The classic example: Look at the maintenance manpower benchmark. If you are above typical, cut maintenance people. That is the wrong response.

The right response is to benchmark your work practices, determine the deficiency, correct the work practices, and only then consider reassigning people who are no longer needed. Reliability is often benchmarked with various measures such as on stream factors and maintenance cost/bbl.

HSB Solomon Associates covers this area fairly well, and lots of people use their techniques. As I said, the next step beyond that is to track or benchmark your work processes and look at things like individual equipment and system performance, costs, failure modes, and time-to-failure statistics.

In energy, we find that specific benchmarks are not that useful because the tradeoff between yields and profit and energy consumption is very different at different refineries. For instance, a high-conversion FCC unit is going to use a lot more energy than a low-conversion FCC unit, but you do not just want to drop all your units down to low conversion to save energy.

What we look for are gaps in expected performance. The areas are the efficiency of the flue-gas train, shaft work, steam consumption, and heat integration. Once gaps are identified, we then look deeper to see where issues arise. Typical benchmarks for process performance are already published. Again, much the same logic as energy, instead of benchmarking yourself against others, we prefer to use a process model or the planning linear program to determine the economic operation for your unit.

We then use routine monitoring to track whether we are really at that optimum. Of course, for safety, over the past several decades, the industry has done a tremendous job of looking at things like lost-time statistics and has made dramatic improvements.

Now we are recommending that you compare yourself to either industry norms or regulatory requirements with regard to management practices, training systems, mechanical integrity parameters, and the adequacy of process safety information like relief system design documentation, management of change implementation procedures training, incident investigation, resolution audits, and adherence to compliance audits.

So, in other words, instead of benchmarking the top level that people have done historically, now we are looking deeper into the processes that lead to the performance and making sure that they are up to snuff.

Teders: If you were able to catch Travis Capp's keynote address this morning, you would note that Valero is very serious about the Solomon Associates survey. We used that survey to benchmark our refineries and FCC units for energy and reliability.

Travis also mentioned that we have a separate energy stewardship program that they developed. It is a very successful program. We have also used a constant-pricing model to benchmark our FCCs where we take the yields and apply constant pricing. The trick is to get the feed priced right, based on its quality. We use this constant-price benchmarking model to identify underperforming units for potential capital projects.

The latest thing we have been doing is something called unit health monitoring in which we take the data from all the various data historians and compile them into a single source on the company's internet site. In this case, we can use these key process indicators to measure performance and reliability. We set limits: If the unit is going beyond a certain limit, we will catch it in time before it ends up as a reliability or performance problem.

Another tier of benchmarking is from catalyst suppliers. They tend to give us more third-party data so that we can see the unit from a different view.

Regenerator, cyclone replacement

We are aware of at least one refiner who replaces the regenerator head, including cyclones, at every turnaround to eliminate the time required for inspection and repair. What are the key factors you consider when assessing the economics of replacing the regenerator and-or reactor head(s) vs. the traditional procedure of internal inspection and repair? How does the size of the regenerator and-or reactor play into the decision?

English: Complete regenerator or reactor head replacement has become fairly common in cases where the entire cyclone set is being replaced either due to age or design change. The cost of the additional vessel head and heavy-lift equipment is offset by the shortened turnaround time that results from doing most of the cyclone installation external to the regenerator and outside of the shutdown period.

There are also some very small FCCs in which internal work is almost impossible, and there is no other option to replace or repair cyclones. This question extends this philosophy to routine turnarounds where complete cyclone replacement is not anticipated. If we assume that a redundant vessel head is available from a previous turnaround where cyclones were replaced, then the additional costs of replacing the head at every turnaround would be the carrying cost of the capital value of the spare head, the capital and carrying cost of the spare cyclone set, and the cost of heavy-lift equipment required for each head swap.

Additionally, significant degradation of the "spare" equipment and its insulation may take place during the 3-5-year period that it is not in use. The offsetting savings will be reduced turnaround time because the need for preparing the vessel for entry, scaffolding, and inspections will be reduced or eliminated.

Perhaps of greater value will be the elimination of the possibility that unexpected damage will be revealed during the internal inspection that could result in an unplanned extension of the turnaround.

The factors that could affect the tradeoff between these costs and benefits would therefore be interest rates, which will determine the carrying cost of the extra vessel head and cyclone set, unit downtime cost, availability of a skilled workforce experienced in conducting internal inspections and repairs in a refinery environment, availability and cost of heavy-lift equipment, planned unit run length, unit design as it affects the ease of removing the vessel heads, plot space both for placement of lifting equipment and staging of internal work and unit location.

The size of the unit will also play a major role in the feasibility of this approach. For small units, located in industrial regions, lift equipment of sufficient size will be available. For the largest units, acquiring cranes of sufficient capacity will likely be infeasible. Even in this situation, a contractor specializing in this type of lift and long lead times are required.

This appears to be a fairly dramatic way to remove the uncertainty from a turnaround. Any time the unit is opened for inspection there is a likelihood that unexpected damage will be found but proper records and monitoring can minimize this risk. Among those things that should be monitored are operational history since last repair, whether the vessel has run inside of normal operating parameters. If it has been outside, how many times, how long, to what extremes, etc.

Also key here is feed quality coupled with rates vs. design and the quality of inspection data—this is the most important variable. Newer regenerators with fewer inspections, along with sites whose inspection programs may not possess Best Practices (quality narratives, pictures, adequate inspection points/areas, etc.) will be at a disadvantage.

Also important here is the confidence in the quality of the last repairs with regularly scheduled, consistent on stream thermal imaging data evaluation from initial star-tup following the last repair. We have seen decisions to replace when it should have been to repair and repair when a better decision would have been to replace. In each instance, incomplete or inadequate data was the cause. One of the biggest factors for refiners that consistently achieve upper quartile turnaround performance is not performing work that is unnecessary.

In summary, a good inspection program with an adequate history that includes operational performance data allows for accurate work scope determination to drive estimates on cost and schedule. Only with this as the starting point should the pros and cons of an alternative be evaluated.

Teders: As Al mentioned, this is an economic question.

At Valero, we do have experience with replacing the head as a way of economically replacing cyclones in reactors and regenerators. This technique reduces field work during the turnaround. Because cyclone replacement is typically a critical path to the FCC turnaround, replacing the head can be a great time-saver.

The cut-the-head-off option can pay for itself by reducing a turnaround time. So, if time is critical for your FCC and you are losing quite a bit of money for downtime, then the cut-the-head-off option could be quite attractive economically. We have not found it to be economical, however, to cut off the regenerator head or replace cyclones at each turnaround.

As Al mentioned, we very closely monitor the cyclone velocities at Valero with a unit-health monitoring program. We also have good inspection programs; our philosophy is to run the equipment reliably instead of running it hard and then replacing it at each turnaround.