During the 2017 American Fuel and Petrochemical Manufacturers Operations & Process Technology Summit (formerly Q&A and Technology Forum), Oct. 2-4, 2017, US domestic and international refiners discussed gasoline processing operations, with an extended focus on topics of safety, catalytic polymerization, alkylation, and reforming.
This annual meeting addresses real problems and issues refiners face in their plants and provides an opportunity for members to sort through potential solutions in a discussion with panelists and other attendees.
This is the second of three installments based on edited transcripts from the 2017 event. Part 1 in the series (OGJ, Sept. 3, 2018, p. 70) highlighted discussion surrounding hydroprocessing operations. The final installment (OGJ, Nov. 5, 2018) will focus on fluid catalytic cracking (FCC).
The session included a four-person panel comprised of industry experts from refining companies and other technology specialists responding to selected questions and then engaging attendees in discussion of the relevant issues (see accompanying box).
The only disclaimer for panelists and attendees 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
For the isomerization unit, what is your treatment for streams containing high concentrations of hydrochloric acid (HCl) in the reactor emergency depressurization system?
Sugg: Currently, UOP doesn’t include treatment of gas from the reactor effluent as it enters the flare header using the emergency depressuring system in our Penex or Butamer process unit designs. So, depressuring to flare with a highly concentrated hydrogen chloride stream isn’t to be taken lightly due to possible corrosion in the flare system, because the flare headers are typically wet. Before depressuring the reactors to flare, steps should be taken by operations personnel to minimize the volume of gas to flare. One of these steps is to decrease the reactor-section pressure to just above the net-gas scrubber pressure through the stabilizer scrubber so that the amount of acidic gas sent to the flare is minimized.
Further, upon completing the depressuring step, UOP recommends following with the dry gas purge—preferably nitrogen—to sweep the acid gases from the flare system. During or after purging with nitrogen, it’s recommended to drain low-point drains in the section of the flare header where acidic-free water can collect and cause corrosion. For Penex and Butamer units, experience has shown that extensive corrosion hasn’t occurred in the flare system with the procedures and guidelines provided. UOP is aware of an isolated incident in which a customer reported substantial corrosion to its common flare-header section. But after changes in the customer’s procedures and changes to the physical layout of their flare-header system, the corrosion was no longer reported.
Sabitov: At Phillips 66, we have two types of isomerization units with respect to the hydrogen circuit configuration: a recycle gas unit and a hydrogen once-through unit. The recycle gas units will depressure the reactors from the separator either through the off-gas scrubber or directly to the flare header. The hydrogen once-through units will depressure through the off-gas scrubber via the stabilizer. They also have an option of depressuring directly to the flare header through the emergency depressuring line at the reactor outlet. For those units where we use chloride alumina catalyst, you obviously have a very high-HCl concentration in the reaction zone. The following considerations would apply: The initial partial-reactor depressuring should be performed through the off-gas scrubber down to the pressure of the off-gas destination, which could be set either by the fuel-gas header or by a saturated gas plant. This depressuring can be accompanied with the reactor hydrogen sweep. Overall, this step enables neutralization of the peak-HCl levels in the scrubber. If full-reactor depressuring is required due to the continuing runaway situation, then reactors should be depressured to the flare header. In this case, some ingress of the HCl to the flare header might be possible. I should say that those are rare events in our units; typically, the initial partial-reactor depressuring is efficient enough to stop a runaway situation. Having said that, we do have one C4-isomerization unit that was built with the provision of the reactor depressure to the dedicated acid-relief header and which is equipped with its own scrubber. Also, if necessary, our refineries can protect their flare headers with ammonia injection during isomerization-reactor depressuring events.
Dubin: We don’t have much to add. Those were two very good answers from Patrick and Alex. Axens’s approach is to go through the scrubber system to depressurize directly in there and then into the caustic scrubbing. The one item that we do want to point out is to make sure that the valve systems are protected from an acidic environment so that they are functional when the emergency does arrive.
Romero: My question is about the potential soil deposits that occur when you are neutralizing with ammonia in a header going into the flare system. That is typically trying to keep that as clean as possible to ensure a free passing to the flare.
Sabitov: Yes. I guess formation of the ammonium chloride would probably be possible at the right sublimation conditions. We haven’t really seen any big issues with salt formation. Again, the events of the full depressuring are extremely rare, so that doesn’t seem to be presenting any practical problem.
Is the presence of pyrophoric compounds common in feed-effluent exchangers? What neutralization methods do you employ before exposing the equipment to the atmosphere?
Sabitov: I’d like to start by saying that reforming units operate with very low-sulfur levels. Typically, reformer feed sulfur is less than 1 ppm; therefore, reformers would fall under the area of low-risk units with respect to the presence of pyrophoric compounds, compared to, say, hydrotreaters. If we see any issues with pyrophoric compounds, they’d be typically encountered more when handling spent catalyst or spent-catalyst dust such as the continuous catalytic reforming (CCR) dust collector product.
We did a survey of our sites, and in general, our reformers don’t really have issues with pyrophoric compounds in their combined feed-effluent exchangers during turnarounds. The same survey also showed that we seem to have two groups of units. The so-called clean units just go ahead and open their combined-feed exchangers (CFEs) after conventional nitrogen sweep; so, really no issues. And then, we have the other group of units where they do see black sludge on the effluent side of the CFEs. That material typically represents some mix of polynuclear aromatics (PNAs), ammonium chloride salts, catalyst dust, iron scale, and potentially iron sulfide. The key for handling those cases is to be aware of the pyrophoric component presence. If a preventative action is needed, wetting those deposits with water would be a basic mitigation technique.
Those units with a history of the exchanger fouling would open their exchangers for turnaround only after performing the cleaning step, greatly reducing the risk of pyrophoric material. We use various cleaning techniques for the exchangers. For those exchangers that can potentially be sensitive to the chloride-corrosion attack (such as a Packinox stainless-steel bundle), we do a waterwash with the soda-ash solution to neutralize the chlorides before those exchangers can be open to the atmosphere. Low-alloy or carbon-steel CFEs have been steamed out or cleaned using a combination of steam and the chemical injection, which is basically a mix of enzymes, surfactants, and the oxidizers. I should note that after those cleaning steps, there have been many cases where we’ve still observed PNAs on the effluent side. Typically, a separate aromatic solvent circulation would be required to remove those PNAs. Our semiregenerative (SR) units typically open up CFEs after the carbon-burn step of the catalyst regeneration and really don’t make any special cleaning provisions for opening.
Sharon: Alex answered this question well. Our experience is similar, so I just want to add a few points. As Alex said, you should always assume pyrophoric material is present. We concur with Alex’s statement that you can also primarily see these materials in hydroprocessing units. Also, the higher the severity, the higher the risk of heavy PNAs being present.
Sugg: This question could also be about the naphtha hydrotreater unit’s feed-effluent exchanger. On this topic, it’s generally sufficient to waterwash the hydrotreater feed-effluent exchanger before exposure to air. If there’s a concern or a previous history with iron-sulfide fires, however, we know of others in the industry who have used methods that include strong oxidizers such as potassium permanganate, percarbonate, and peroxymonosulfate. A qualified cleaning contractor experienced in this work could provide more details. I think the rest of it has been covered.
Philoon: In 2015, UOP led an AFPM Principles & Practices session on the topic of naphtha hydrotreater (NHT) and reforming unit feed-effluent exchanger fouling and cleaning.
In Honeywell UOP Platforming units of both CCR and SR types, the feed-effluent exchanger or CFE is generally a clean service. There are occasionally problems with feed-side fouling due to peroxide gums, coke, or iron scale, or reactor effluent-side fouling due to ammonium chloride salts or polynuclear aromatic compounds. It’s uncommon to have problems with pyrophoric compounds (most typically iron sulfide) in CFEs. In principle, it’s possible for corrosion products and iron-sulfide scale to collect in low-flow areas of the CFE and be a hazard when exposed to air during maintenance. In many units, however, the CFE is washed with water to remove salts before the exchanger is opened to the air. Water doesn’t prevent iron sulfide from reacting with oxygen, but it does reduce the rate of mass transfer so that the rate of reaction is slowed to the point that the exchanger can be opened and the material removed before a problem occurs. Also, in SR units, a coke burn is generally done before the unit is opened for maintenance. While this does not remove all the iron sulfide, it may reduce its concentration.
Note that the most common location of problems with pyrophoric material in CCR units is with the dust removed from the fines-removal system. Both SR and CCR-type units may experience problems with pyrophoric material during catalyst unloading. That said, UOP is aware of one instance where a fire occurred as the bottom head of a vertical combined-feed exchanger was dropped. The refinery reported that hydrocarbon was present due to plugged drain lines, and it suspected that ignition was provided by pyrophoric material. The fire was quickly extinguished without any injuries or major equipment damage.
NHT units, like all other hydrotreating units, are units where exchanger-fouling issues are more common. The feed side of the feed-effluent exchanger may become fouled with peroxide gums or corrosion products from tankage or upstream units. The reactor-effluent side may become fouled with ammonium salts, particularly if feeds high in nitrogen are being processed. Because of the presence of sulfur and iron, there’s always a concern about pyrophoric iron. When a feed-effluent exchanger train is to be opened for cleaning or other service, it’s generally sufficient to waterwash the exchanger before changing the environment to air. The wash procedure will easily remove salts and other debris, and the water will provide a layer that reduces the rate of oxygen diffusion, therefore limiting the rate of the iron-sulfide oxidation reaction. If there’s a concern or previous history with iron-sulfide fires, over the years the refining and petrochemical industries have used many methods to treat pyrophoric iron sulfide with chemicals, including strong oxidizers such as potassium permanganate, percarbonate, and peroxymonosulfate. The heat-exchanger vendor and the exchanger-cleaning contractor may be able to provide a detailed procedure.
Located on 722 acres just outside Ardmore, Okla., about 100 miles south of Oklahoma City, Valero Energy Corp.’s 90,000-b/d Ardmore refinery processes medium-sour and sweet crude oils to produce about 37,000-b/d of gasoline and 20,000 b/d of distillates. The refinery receives crude and feedstock via four third-party pipelines, with petroleum products transported to market via rail, trucks, and Magellan Midstream Partners LP’s pipeline system. Photo from Valero.
Catalytic polymerization alkylation
What operating conditions in a catalytic polymerization unit are conducive to the formation of esters? What is the main effect that esters have on unit performance?
Kinderman: The mechanism by which oligomerization takes place over the solid phosphoric acid (SPA) catalyst necessitates that a phosphate ester is formed between the catalyst and the olefin during the reaction. The temperature required to disassociate the phosphate ester and convert the olefin depends on the olefin in question. Our research has shown that with the mixture of isobutene and normal butene, the oligomerization takes place at about 250° F. If pure propene is used, that temperature is at least 275° F. In case the esters leave the reactor, they’re decomposed into phosphoric acid, and you may see corrosion downstream.
Sugg: Our response is similar to Bryan’s, but there’re a couple of differences. The concern for ester formation is limited to the C3-only feeds when you have reactor temperatures < 275° F. As a little bit of history, this concern with low-temperature operation with SPA catalyst was raised by UOP scientists back in the 1930s. Our early experiments show that the C3 oligomerization occurred readily at 275° F. or higher. Once the temperature was lowered to 257° F., however, the stable esters were formed, which blocked the catalyst acid sites.
Because the presence of even a small amount of C4 olefins prevents stable ester formation, it’s only the C3 units that must be concerned. UOP has a guideline to maintain C3-olefin catalytic polymerization operation above 300° F. We feel this is reasonable because it provides a margin of safety to ensure that no part of the catalyst bed operates below 275° F. (considering the recycle and quench-gas flows, which impact the inlet and bed temperatures). If you get a low-temperature ester formation situation, you’re going to get a subsequent loss in the conversion, which will require higher temperatures to maintain your desired product yields. As Bryan said, once the esters are formed, they’re very corrosive; so, you have to be very careful handling them after that.
Buchan: Ester formation in a catalytic polymerization (cat poly) unit is an issue for units with a C3 olefin-only feed, as the presence of even a small amount of C4 olefins prevents stable ester formation. With a C3-olefin feed, low-reaction temperature is conducive to ester formation. The Honeywell UOP guideline is to avoid reactor operation below about 300° F. (150° C.). The concern with low-temperature operation causing the loss of stable phosphoric acid ester from the SPA catalyst was reported back in early UOP work in the 1930s. C3-olefin oligomerization was reported at 275° F. (135° C.) or higher reactor temperature; but at the slightly lower temperature of 257° F. (125° C.), the isopropyl ester will not decompose. The result was that C3-olefin oligomerization wouldn’t proceed at all; the stable esters will occupy all catalyst acid sites. This phenomenon was demonstrated in an autoclave reactor using propylene and 90% phosphoric acid. No polymer layer was formed at 257° F. (125° C.). The critical decomposition temperature for the phosphoric acid ester in the presence of propylene is somewhere between 257-275° F. (125-135° C.). If just a small amount of the stable phosphoric acid ester desorbs to the hydrocarbon, there could be a major corrosion problem.
Cat poly units with a C4 olefin-only feed don’t form the stable phosphoric acid esters at such high temperatures as occurs with the C3-olefin feed; so, a C4-olefin feed will show reactivity over SPA, even at relatively low temperatures. Presence of just a small amount of C4 olefin makes apparent C3-olefin reactivity look higher for a C3-C4 feed because presence of C4 olefin (competitive adsorption) prevents the stable C3-olefin ester from occupying all the acid sites as it would in 100% C3-olefin feed.
The UOP Guideline to avoid C3-olefin cat poly operation below 300° F. (150° C.) is a reasonable general guideline to make sure that no part of the catalyst is below 275° F. (135° C.), considering there are recycle and quench flows impacting the inlet temperature and the exotherm in each bed. This guideline provides some safety margin from the anticipated condition where acid loss and resulting corrosion will occur.
Although UOP has no confirmed experience with ester formation in a C3-only cat poly unit, the expected impact on the process would be removal or blocking of active catalyst sites, reducing the effective catalyst activity. The reduction in conversion would require higher temperatures to maintain desired conversion levels. In addition, the stable phosphoric-acid ester is highly corrosive due to the presence of water, which will potentially create a major corrosion problem.
Reforming
What is the importance of water content in reformer feed and recycle gas on the performance of the catalyst? What are your desired water concentrations in each of these streams?
Sabitov: Water levels in the reformer are an important variable impacting water-chlorine balance and should be maintained in the reactors within optimum range. I’m not going to talk about water impact on chlorine on the catalyst because I’d rather focus on water impact on the metal function of the catalyst. We consider water as a metal-function attenuator. While on one hand, a little bit of water helps to take the edge off platinum-cracking activity; in general, an excessively wet unit would always be a bad performer from a yields-and-stability standpoint. A little less water would be better than too much water.
Different types of units are getting their water in different ways. All of them, however, should be running with dry hydrotreated feed containing not more than 1-3 ppmw of water. The SR unit will inject 3-5 ppmw of water on feed, resulting in about 15-25 ppmw of water in the recycle gas. Cyclic units and CCR units receive sufficient moisture from catalyst coming out of regeneration and don’t inject water to the feed. Cyclic unit recycle-gas moisture may vary within the range of 5-50 ppm, sometimes higher right after bringing the reactor in service after regeneration. CCR unit recycle-gas moisture typically stays between 10-20 ppm.
High-water upsets are not uncommon in the industry, so I want to spend some time on the case where the unit isn’t getting enough water. Phillips 66 operates SR reformers where we don’t inject any moisture into the feed. So, ultimately, as cycle progresses after regeneration, it becomes very dry with recycle-gas moisture falling below 5 ppm. To demonstrate the shift in metal-function activity vs. acid-function activity in time, we use a local environmental plan (LEP) ratio trend. For those unfamiliar with that terminology, the LEP ratio is calculated as product moles of C1+C2 divided by product moles of C3+C4, reflecting balance between cracking on metal function over cracking on acid function. An increase in the LEP ratio will indicate an increase in metal-function activity over acid-function activity. On a graph, the LEP ratio starts not too far away from predictions for an ideal start-of-run (SOR) catalyst when the system still has reasonable levels of moisture after regeneration. As the system dries out, the LEP ratio increases, evidencing an increase in metal-function activity over acid-function activity. At some point, as catalyst builds up the coke, impacting catalyst metal-function activity, the LEP ratio trend turns around and starts falling; however, it remains above the SOR catalyst prediction.
Dubin: Managing the water content in the reactive environment is critical for the overall performance of the reformer unit and catalyst. The content of water must be controlled to allow for a sufficient quantity of hydroxyl (aluminum hydroxide) sites while simultaneously controlling the HCl levels to have the right balance of water-HCl to set the catalyst HCl content (aluminum-chloride). The overall level of hydroxide and chloride sites determines the overall acidity level of the catalyst, which needs to be in balance with the metal activity to maximize catalyst performances. By maintaining the appropriate balance, yields of desired products (reformate, aromatics, and hydrogen) are maximized. Additionally, the optimization of the catalyst performance will extend catalyst life and minimize potential corrosion and fouling in equipment.
As the water content of the reactive environment is the combination of the water content in the hydrotreated feed and the unit recycle gas, the measurement and follow-up of each is necessary for the overall health of the unit. For most unit operations (both feed qualities and product octane targets), a target concentration of 20±5 ppmv water in the recycle gas is desired. For more paraffinic feeds at higher-octane targets, a slightly lower recycle-gas water content can improve yields.
For reformer feed, the moisture content can depend on the quality of stripping in the upstream hydrotreater; but often, the reformer feed-water content is in the 3-4 ppmw range. Typical rules of thumb say that a 1 ppmw increase in reformer feed-water content will increase the water content of the recycle gas by 2-3 ppmv. The need to further increase the water content can depend on the type of reformer in operation. With cyclic and moving bed-type reformers, the frequent regeneration of the catalyst generally introduces enough moisture into the system to maintain the recycle-gas water levels at the targeted levels. Should a cyclic or moving-bed reformer operate in a noncontinuous, infrequent regeneration operation, a reformer feed-water injection may be required to maintain the proper acidity level on the catalyst. For fixed-bed reformers, water injection to maintain the targeted recycle-gas moisture level is common.
Sankalia: For the CCR Platforming unit, Honeywell UOP seeks to minimize the water content in the reformer feed as low as possible. Due to CCR, an equilibrium moisture level will be established between the catalyst-circulation loop and the recycle-gas loop. Since the moisture level is already established, there’s no need for extra moisture in the platforming feed, affecting the moisture-chloride balance in the CCR unit. The typical value of feed moisture for a CCR unit is < 2 ppmw due to upstream fractionation. The typical value of recycle-gas moisture in a CCR unit is 15-20 ppmv and shouldn’t exceed 30 ppmv.
In a semiregenerative platforming unit, there’s no CCR, and as a result, continuous water injection into the reformer feed is required to maintain the proper water-chloride balance on the catalyst. The water injection varies based on catalyst type, but a typical value of water injection is 4 ppmw (assuming 1 ppmw feed water coming from the upstream unit). Typical values of the recycle-gas moisture for an SR unit are in the range of 10-20 ppmv.
Quick increases in recycle-gas moisture above acceptable ranges tend to act as a metal-function poison, enhancing the acid function of the catalyst and facilitating the formation of coke. If the high-moisture content continues for an extended period, chloride is stripped off the catalyst, which reduces the acid function. Underchlorided catalyst will reduce the overall activity of the reactors. Some of the observable effects on the system include:
• Decreased hydrogen production.
• Decreased recycle-gas hydrogen purity.
• Increased LPG make.
• Lower C5+ yields.
• Increased HCl in recycle gas.
• Increased rate of coke production.
• Reduced reactor temperature differentials (∆Ts).
For CCR units, increased coke make will affect the CCR burn-zone operating parameters and, if not adjusted, increases the likelihood of coke slipping to lower sections of the regeneration tower, leading to catalyst and-or equipment damage. For fixed-bed units, increased coking will lead to shorter cycle lengths. Regenerations of highly coked catalyst can also damage reactor internals of the SR units.
The panel
Geoff Dubin, olefins and light oil hydroprocessing technology manager, Axens North America Inc.
Bryan Kinderman, Americas sales director, Clariant Corp.
Alex Sabitov, senior reforming and isom engineer, Phillips 66 Co.
Kyle Sharon, Ardmore refinery technical director, Valero Energy Corp.
Patrick Sugg, regional service manager, US Gulf Coast refineries, Honeywell UOP LLC
The respondents
Martha Buchan, Honeywell UOP
Steve Philoon, Honeywell UOP
Raúl Romero, NALCO Energy Services LP
Vikas Sankalia, Honeywell UOP
