NPRA Q&A—1: Safety, process questions addressed at annual conference
Safety and process operations received considerable attention during the 2010 National Petrochemical and Refiners Association Q&A and Technology Forum (Oct. 10-13; Baltimore).
This annual meeting addresses real problems and issues refiners face in their plants and attempts to help them sort through potential solutions in a discussion with panelists and other attendees.
The panel…
• Emerson Domingo, process design engineer,
• Steve Shimoda, senior process design specialist,
Sunoco Inc. Shaw Energy & Chemicals Group
• Pat Dennler, process engineering manager,
• Conrad Jenson, engineering manager,
This is the first of three installments based on edited transcripts from the 2010 event. Part 2 in the series (OGJ, Aug. 1, 2011) will focus on crude and vacuum distillation and coking with an emphasis on safety, coking, and corrosion control. The final installment (OGJ, Sept. 5, 2011) will highlight gasoline processes, especially dealing with safety, alkylation, and naphtha hydrotreating.
The session employed five 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.
Safety
With advance controls on the fluid catalytic cracking unit and gas plants, what are refiners doing to train new operators and keep their experienced operators ready to handle FCCU upsets and emergencies? Are refiners using simulators to help with the training and retraining?
Domingo: Sunoco does have dynamic matrix control (DMC) in all the FCC gas plants and also the reactor regenerator system. Only one of our FCCs actually has a complete process simulator.
For the FCC with a simulator, the initial board training is typically a total of 6 weeks, which includes classroom training, process simulator training, and one-on-one special training with an experienced board operator on shift.
Process simulators typically have 20 training scenarios that range from simple ones, like responding to a high or low liquid level, to a complex one, like responding to a wet gas compressor trip. We also train on FCC start-up and shutdown, and that is part of the scenarios in the process simulator.
For DMC training, we initially train the new operators with the DMC off. They get used to manually operating the unit with the DMC off; and then after a while, we turn on the DMC and have them get used to operating with the DMC on as well. We also explain to them what is happening with the DMC and how is it being controlled at the point.
As far as keeping them fresh, we try to have the operator taken off shift and spend a day on a process simulator. We then go through some of the important process scenarios again, as well as—we hope—new scenario updates to the simulator, which are based on past FCC upsets and how to better respond to them.
For DMCs, there are typically times during the normal run where the DMC is turned off purposely for operators to handle operational issues; so we use that as a nonscheduled training for the operators. And then if it has been operating well for a while and the DMC continues to be on, we ask the board operators to turn off the DMC and have them run for a shift with the DMC off to refamiliarize with manual unit operation.
And finally, in addition to all this, we conduct "What if?" drills in the refinery. We especially do these when there is a high-risk maintenance in the unit or in the refinery. We review the emergency procedures that might be used just in case they are needed.
Shimoda: As I started researching this, I focused just on the use of the simulators for training. I found that in the 1999 Q&A, there was a very similar question. If you go back to that transcript, you will read six very positive responses from different refiners that were very enthusiastic about using simulators to retrain their operators.
I would contrast that with the response we heard at the [NPRA Cat Cracker Seminar, Aug. 24-25, 2010, Houston] where four other refiners said they had varying degrees of success with simulators.
It is my understanding that for grassroots units, the use of simulators was a good tool to train operators, but several of the refiners found that there was an issue with maintaining the simulators.
This is similar to feedback we have heard from our clients. There is enthusiasm for grassroots units where operators have not had FCC operating experience. But for clients with existing units, there is less interest because of the costs associated with maintaining a simulator.
We know of one refiner that has multiple sites. It has switched to using a generic web-based simulator and was able to save some money while still having the training tool.
Dennler: Because we have a rather long response, I'm going to hit only the high points, and then make one additional comment.
Shell Global Solutions has a formal training program related to operator competency for recognizing hazards and learning proper responses to abnormal situations. It is important that these lessons from the past are not lost, and to this effect, Shell Global Solutions has a global program called Learning from Incidents.
The training program is equally important not only to the cat cracking operators, but also to new process engineers and control systems engineers as well. Refresher training is also given to experienced operators. Drills with simulated scenarios are used on shift to practice responses.
These are known as red tag drills because the idea is for the shift to use red and green tags to identify equipment valves that need to be stopped, closed, or started and opened as a response to abnormal situation. The shift's responses are used to determine if the emergency procedures can be improved. They can also be used to assess where the follow-up training would be beneficial.
Simulators are another tool to improve the ability for the operator to recognize an abnormal situation while practicing responses. We are aware that companies use tailor-made simulators for board operators to remain certified. We are aware of a case in which simulator training prepared the operators to make the right move during an abnormal situation. This experience would allow quantification of the benefits of such a simulator.
It is important to ensure that advanced process controls for normal operations do not introduce the wrong response during abnormal situations. Such potential scenarios need to be thought through by management change review for proposed modifications to the advanced process control system.
I think we are all aware of the data indicating that the industry is approaching another round of retirements. In the next couple of years, there are going to be a lot of experienced people leaving the industry. So, it is even more pertinent for us to find training programs that can supplement and maintain the expertise so we do not lose it.
Process
What have refiners done to mitigate or eliminate coke buildup in reactors? How do you monitor and vary feed quality, reactor severity, catalyst formulation, and other variables to impact coke formation. How does feed distributor operation and design impact reactor coke buildup?
Domingo: A lot of coke can form during start-ups, shutdowns, and unit upsets. During start-up, the temperature inside the reactor is too low and the hydrocarbon inside the reactor will condense and form coke.
So, on start-up, it is important to make sure that all the internals are hot and that you are at operating temperature before oil-in. If you can, it is desirable to start up the unit in the first few hours without resid, so that there is less chance of heavy hydrocarbons condensing in cold spots.
Similarly, during unit upsets and shutdowns, feed needs to be pulled before the riser gets too cold. For Sunoco, this is typically at 900° F. Once feed is pulled, the riser and reactor should be purged completely with emergency steam so that all hydrocarbon vapor is swept out of the system and pushed towards the main fractionator.
Also during an emergency shutdown, it is just as important to make sure that the oil stays out of the riser. There have been times when the feed has been diverted at the right temperature; but by accident, the oil is put right back into the riser without anyone knowing about it.
This can be caused by valve misalignment, auto diverter valve malfunction, or just total confusion during a shutdown, or maybe having an inexperienced operator on the board for his first shutdown. Once feed is pulled, it is a good practice to close a manual valve in the feed system as close to the feed nozzle system as possible. The divert system typically has isolation valves for maintenance purposes. Choose that valve and make sure to close it when you divert the feed, as the divert system might get reopened by itself accidentally.
During normal operation, maintain riser and reactor temperature well above the minimum. It is important that the oil be well-atomized and mixed well with the catalyst in the feed injection zone. Monitor the performance of the feed nozzle and make sure you have the right differential pressure across the feed nozzles.
The only time you really have a chance to troubleshoot these feed nozzles is before you start up. So, at the end of a major turnaround, before the manways are closed and any feed nozzles are taken down, place scaffolding inside the riser and check that each nozzle is clear. Make sure the nozzles have not been plugged with refractory or other debris. Then, just before start-up, go through the entire feed nozzle check procedure nozzle by nozzle and reconfirm none is plugged.
In the reactor, there should be a steam ring purge at the reactor dome just above the cyclones to prevent coke buildup. The steam used should be superheated. In one of our reactors, in lieu of a reactor dome steam, the riser has a tee inertial separator and a shotgun pipe that extends all the way towards the reactor dome.
This "shotgun extension" diverts riser effluent right up the reactor dome to keep that area from being stagnant. And, finally, the stripping section should be well designed to allow for net upflow of steam towards the reactor cyclones. This helps keep velocities up in the reactor and prevent dead space there.
Jenson: Holly has one FCC that had significant reactor coking. After initial start-up in 1996, run lengths were limited to about 11⁄2 years due to severe coking in the reactor transfer line. During shutdowns, there was also found to be significant coke accumulation in the reactor dome and the external area of the cyclones, as well as in the cyclones and sections of the riser.
Feed nozzle steam subsequent to this first shutdown was then maximized, and riser temperature increased to about 990° F. to improve atomization and vaporization of the feed. The result was somewhat successful. It increased unit run length by 25% but was still not satisfactory.
The long-term solution was to change out the feed distributor design and increase the residence time on the riser, and then we modified the cyclone vent tubes. The sum of all of these changes has been increased run lengths of over 4 years. The last time we ran the reactor, we just had a cubic foot of coke in the dome and cleaned it up easily.
Dennler: It is best to understand how coking occurs.
Hot vapors condense in stagnant areas of the vessel and piping due to several reasons: feed quality or unvaporized feed, feed injectors, poor orientation and atomization, insufficient reactor stripping steam, low reactor temperature, reactor head and overhead vapor line heat loss, and inadequate dome steam. I agree with Emerson completely that it needs to be superheated.
Shell monitors feed parameters that reflect feed crackability and coke formation. Many of our units run with a rigorous FCC simulator in the background as part of our APC [advanced process control] system for that unit. Catalyst formulation is determined by our Shell Global Solutions FCC R&D group. Our sites send samples to R&D and our catalyst vendors to track catalyst performance.
Phillip Niccum (KBR): I noticed that Emerson Domingo's slides talked about making sure that the net stripping steam flow is up through the stripper. I have an example I can mention in which we saw a problem such as this.
The unit had a coking problem in the bottom part of the disengager, and the stripper was running at a very high mass flux. Surveys done with gamma rays and other tests confirmed that we were actually entraining a lot of this stripping steam down into the spent catalyst standpipe into the regenerator. When you do that, the net flow can, in fact, be down, and you might be able to strip something in the stripper, but if you carry those vapors into the regenerator, then you have not actually accomplished anything.
The point of mentioning this is that flux tube technology was later installed in the stripper, which is a baffle modification designed to allow the stripper to run at higher mass flux without entraining the steam down into the regenerator. Following installation of these modified stripper baffles at the next turnaround, the coking problems had been eliminated.
Jack Olesen (Praxair Inc.): Conrad, you mentioned the coking problem in the transfer line. Did that go away when the severity on the unit was increased along with the reduction in coke in the reactor?
Jenson: The transfer line coking decreased with the higher severity, but it did not completely go away. We went through a round of insulating flanges and tightening up everything we could to retain heat but we had limited success.
The feedstock is about 50% DAO [deasphalted oil] and has a 1,270° F. to 1,280° F. end point; so it has a heavy fraction. The ultimate solution was modification of the feed nozzles and an increase in riser time.
Ray Fletcher (Intercat Inc.): I would like to add one small point to the answer that Conrad gave. As process engineers, it is worthwhile for us to monitor the coking tendency within the unit. The simple way to do this is to monitor the 90% or the 95% distillation point on the feed and compare that to the 90% or the 95% point on the slurry. For example, if the slurry 95% point exceeds that at the feed, then you know you are polymerizing and that there is a coking tendency within the unit.
To be able to use this technique for the reactor side, however, we need to make sure that on the fractionator tower bottoms we are operating that part of the tower properly, which is a time-temperature relationship. So, assuming the fractionator is operating well and that you see a coking tendency within the unit, many times the recommendation—as Conrad indicated—is to increase the conversion on the unit. In our experience, the coking tends to go along with lower conversion levels.
A second response would be to work closely with your catalyst supplier to ensure that you are utilizing best-available catalyst technology for deep bottoms cracking.
John Sultzbaugh (Buell Division of Fisher-Klosterman Inc.): A question for Conrad: Can you comment on the cyclone gas tube modifications you made?
Jenson: Originally, the vent tubes were very short between the primary and secondary stages, probably a foot or two. We ran the tubes to about 6 or 7 ft above the bed at the location of our coke line. The last time we inspected them, the coke line was right at the end of those tubes.
Dennler: I have one last comment. I am a firm believer in giving credit where credit is due. At the 2006 Cat Cracking Conference [Aug. 1-2, Houston], KBR and Exxon put together a paper that was actually entitled "Operations Troubleshooting." They did a really nice job talking about this particular phenomenon.
Domingo: A lot of coke formation can occur during start-ups, shutdowns, and upsets of the unit. During start-up, it is important to make sure that all the reactor internals are hot and at operating temperature before introducing oil. If the temperature is too low, then the hydrocarbon will condense and form coke. It is also desirable to start up during the first few hours without resid in the feed so that there are fewer chances for heavy hydrocarbon to condense in cold spots.
Similarly, during a unit upset, feed needs to be pulled before the riser gets too cold. This is typically done at 900° F. Once feed is pulled, the riser and reactor should be completely purged with emergency riser steam so that the hydrocarbon is swept out of the system and pushed to the main fractionator.
It is also important to make sure that oil stays out of the riser by manually closing a main valve going to the feed nozzles. There have been instances in which feed has been diverted at the proper temperature but, by accident, is re-introduced into the riser without anyone knowing about it. This can be caused by valve misalignments or an automatic divert-valve malfunction. Once feed is diverted, it is a good practice to close a manual block valve in the feed system. Do not rely on the automatic divert system because it may get reopened.
During normal operation, maintain riser and reactor temperature well above minimum at all times to avoid condensation of hydrocarbons. It is very important that the oil be well atomized and mixed well with the catalyst in the feed injection zone. Monitor the performance of the feed nozzles to make sure you have the right ΔP [differential pressure] across the feed nozzles.
At the end of a major turnaround, you should check each feed nozzle to make sure all are clear. Go through the entire feed nozzle check procedure before oil-in to confirm that there are no plugged nozzles.
In the reactor, there should be dome steam purge just above the cyclones to prevent coke buildup outside the cyclones and in the reactor head. The steam should be superheated and is normally controlled with an orifice plate to ensure adequate flow. One of our risers has a T-inertial separator with a "shotgun" at the top of the riser, which directs some of the riser effluent to the reactor dome to keep it from becoming stagnant. This "shotgun" extension is in lieu of reactor dome steam. Also, the stripper section should be well designed to allow for a net upflow of steam towards the reactor cyclones and to keep up velocities in the reactor and prevent dead space.
Refiners operating FCCUs that produce high levels of propylene have seen different or excessive product contaminants when compared to a less severe operation. In your experience, how has this impacted gasoline or LPG treating unit? What specific contaminants have you identified? What impact have you seen in amine color, consumption, or foaming tendency? What actions have you taken that have mitigated or prevented treating unit issues?
Shimoda: Regarding the first question about high levels of propylene and how they are achieved: If you are just elevating your propylene levels from your gasoline operation by adding ZSM-5, then you would not really expect to see any more contaminants in your system than you would normally. If you are getting the propylene production through high severity and then increasing your ROT [reactor operating temperature] through bottoms cracking, then you are certainly going to see more contaminants.
When I first brought this question to the tech service guys out in the plants, their first reaction was that our biggest problem is butadiene. Sure, there is HCN [hydrogen cyanide] that will end up in the amine system, but as long as you have a good program to take care of your amine, then you might have to replenish it a little sooner than normal.
On some of these units, [the client] wants to produce polymer-grade propylene. You will then start seeing some other contaminants, like COS [carbonyl sulfide], mercury, arsine, and some oxygenates. These must be removed to get to the levels you need for the propylene.
Shaw has a DCC [deep catalytic cracking] technology unit that has been operating for more than 12 years. As far as LPG treating, we really have not seen any particular issues there. Standard amines have been used successfully. Mercaptans extraction is accomplished through a general caustic.
On the gasoline treating side, there needs to be special attention paid to reboiler design, which is what the tech service guys really brought up. Chemical additives have been used to suppress gum formation.
As you go to the polymer-grade side, I will just reiterate that mercaptans removal, COS removal, arsine, and mercury removal are necessary to reduce to the purity levels needed for your polymer-grade propylene.
Klinghoffer: Over the course of about 21⁄2 years, we made a conscious effort to drop riser temperature and have the upstream units—more specifically the vacuum units—make better quality gas oil. When we ran higher riser temperature, besides making butadienes that hurt the alkylation unit, we noticed plugging of the extractor at our Merox LPG treater. When we dropped the riser temperature, the fouling of our Merox extractor was a lot less. The extractor had plugged two to three times a year, sometimes, three to four. Now, it is typically once a year.
Dennler: High riser temperatures, with and without catalyst additives, are two methods to increase propylene make. As a result, the amount of hydrogen cyanide formed tends to increase. Corrosion control methods, such as waterwash and ammonium polysulfides, are used to protect the unit against cyanide-related corrosion.
When cyanides are contacted with amines in the downstream treating units, the result is going to be the formation of higher heat-stable amine salts. Monitoring and controlling heat-stable salt levels in the amine will minimize corrosion and fouling in the amine system.
High riser temperatures can also increase production of butadiene, arsine, and COS. Butadiene that ends up in the feed to the alky unit can have a significant impact on acid consumption, both in HF [hydrofluoric] and sulfuric acid units. Selective hydrogenation converts butadiene to butenes or butane and reduces the butadiene make from the FCC.
In a 2009 AIChE Conference [AIChE Spring National Meeting, Tampa, Apr. 26-30], UOP presented a paper detailing arsine removal that typically requires an absorbent guard bed using promoted aluminum catalyst. COS can also be removed by raising the amine temperature. One has to be careful, though, when raising the amine temperature so that you do not end up having a negative impact on the H2S removal, which favors low temperature in your amine system.
For gasoline, mercaptan levels may actually be reduced by high riser temperatures. The LPG treaters typically have enough excess capacity for mercaptans; sometimes impacts are not incurred.
Dave Smith (UOP LLC): One recent trend on the treating of LPG streams is to utilize adsorbents as a trim to the Merox unit or other caustic extraction processes to reduce some of the sulfur species to even lower levels, thereby meeting new, more stringent specifications on LPG sulfur content. As a trim to the scrubbing process, adsorbents are particularly effective for trace removal of COS, H2S, CS2, mercaptans, dimethyl disulfide, and dimethyl disulfide.
Shimoda: The answer to this question depends on what is considered high levels of propylene and what is the method to achieve it. If the additional propylene is produced by the addition of ZSM-5 to a standard operation, then I would not expect to see additional effects of contaminants. However, if the propylene production is increased through higher severity, ROT, bottoms cracking, etc., then there will be additional contaminant issues.
At Shaw, we have had several recent designs with the goal of maximizing production of propylene. The highest level of polymer-grade propylene (PGP) is achieved using the Shaw DCC process. This process uses high reactor temperature and postriser bed cracking to complete conversion of naphtha to LPG. Selectivity is maintained by minimizing hydrocarbon partial pressure.
Butadiene contamination may be more pronounced at the higher reaction severity. Special attention to reboiler design is important in mitigating issues of reboiler fouling. Note that downstream C4 alkylation feed may contain oxygenates, such as acetone, as well as higher butadienes.
Impact on gasoline treating: Hydrotreated feed gives low mercaptan in gasoline. Diolefins are slightly higher but can be handled with standard antioxidant chemical injection. Special attention to reboiler design is important in mitigating issues of reboiler fouling. Note that DCC technology includes nonhydrotreated feeds as well. Several DCC units have operated more than 12 years on nonhydrotreated feedstocks and use only chemical additives to suppress gum formation.
Impact on LPG treating: There are no particular issues in LPG treating. Standard amines have been successful in more than 12 years on DCC operation. There is no foaming or degradation of amine that requires special attention. Note that normal operating procedures keep the amine clean by filtering and carbon adsorption. Reasonable acid-gas loadings on the rich amine prevent corrosion in the system. In the referenced DCC unit, DEA was used as the amine, since CO2rejection was important to offgas treating for ethylene recovery.
RSH extraction from LPG by regenerable caustic has also been successful in more than 12 years of operation.
Other specific contaminants include methanol and acetaldehyde. Trace methanol is easily removed from C3 stream using a regenerable adsorbent bed. Note that some crudes contain trace arsine or mercury. These can be removed with nonregenerable adsorbent beds.
More Oil & Gas Journal Current Issue Articles
More Oil & Gas Journal Archives Issue Articles
View Oil and Gas Articles on PennEnergy.com
