AFPM Q&A-2 Discussion expands to include hydroprocessing

Sept. 7, 2015
This second of three articles presenting selections from the 2014 American Fuel and Petrochemical Manufacturers Q&A and Technology Forum continues a discussion of safety and also addresses issues of mechanical integrity and profitability related to hydroprocessing.

This second of three articles presenting selections from the 2014 American Fuel and Petrochemical Manufacturers Q&A and Technology Forum (Oct. 6-8, Denver) continues a discussion of safety and also addresses issues of mechanical integrity and profitability related to hydroprocessing.

The first installment, based on edited transcripts from the 2014 event (OGJ, Aug. 3, 2015, p. 52), addressed gasoline processing operations, with a focus on safety, blending, and reforming issues. The final installment (OGJ, Oct. 5, 2015) will highlight discussion surrounding processes associated with fluid catalytic cracking.

The session included four panelists 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

As more refiners consider installing zeolite catalyst in their hydrotreating units, what are your recommendations for a depressuring system?

Blackwell I am going to discuss both the emergency depressurization system (EDS), which is a manual depressurization system, and the automated depressurization system (ADS). I will start with the EDS since the ADS piggybacks on this manual system.

All hydroprocessing units have an EDS. The EDS valve(s) must be capable of activation by a dedicated button in two locations: one on the control panel and one in a safe field location. There is little difference in EDS sizing criteria between Chevron Lummus Global (CLG) and Chevron Corp. (CVX). For CLG, we provide two equally-sized valves in parallel, with each valve sized to reduce the pressure to 25% of normal in 20-30 min. The CVX standard is depressurization to 50% of normal in 15 min and depressurization to nitrogen header pressure in 30 min.

The CLG standard is two redundant and equally-sized valves installed in parallel. The primary activation is one valve, but the operator has control of the second valve. So you can effectively double the depressurization rate if your relief system is sized accordingly and if the situation warrants it. When both valves are utilized, the CLG system will depressurize at a rate similar to the CVX standard. Finally, the fail-safe design condition is for the EDS valves to "Fail Open" with loss of signal.

With EDS activation, the makeup hydrogen flow is reduced to 25% of normal to ensure that depressurization occurs as rapidly as expected. CLG will trip the heater to pilots. CVX trips the heater to a full chop.

Now I will discuss the ADS system. I mentioned that EDS is required for all hydroprocessing units. The current practice for ADS is that it is not required for hydrotreating service using hydrotreating catalysts only, but it is required for any unit utilizing hydrocracking catalyst, including mild hydrocrackers.

We are currently reviewing the newest generation of high-activity hydrotreating catalysts. There are some extremely active hydrotreating catalyst options available now, and we are considering adding these catalysts to the list for ADS inclusion. No decision has been rendered yet, but you might keep this in mind if you think you may use these high-activity catalysts.

A California superior court recently cleared the way for Chevron Corp. to proceed with its long-planned, $1-billion project to modernize operations at its 257,000-b/d refinery in Richmond, Calif. (OGJ Online, Apr. 22, 2015). The project will replace some of the refinery's oldest processing equipment with technology meeting the broad spectrum of US air quality standards without changing the basic operation or amount of crude the refinery can process. A main project component is replacement of the refinery's existing 1960s-era hydrogen plant to enable more efficient and reliable processing of higher-quality hydrogen. Photograph from Chevron.

We currently apply one exception to our ADS guidelines. The decision process for ADS inclusion is really based around the activation energy, as well as the operating temperature and its proximity to the onset of thermal cracking, which is 850° F. We currently utilize a very mild, amorphous-like hydrocracking catalyst system to increase volume swell in diesel hydrotreating service. The activation energy of this system is extremely low. In this case, ADS is optional as long as the expected operating temperature is also low, relative to 850° F.

Next I will discuss ADS activation. ADS triggers are designed to provide the console operator sufficient time and notification to gain control of the unit or manually initiate depressurization. Time, in this case, is not a fixed clock time. Rather, it is the effective time created between the alarm and trip set points. Based on experience with early implementations, we have a concern with false trips, so we utilize redundant voting. One of our ADS triggers is loss of recycle gas flow. For this, we use two-out-of-three voting. High outlet-line temperature is another trigger. This is based on two-out-of-two voting. For catalyst bed temperatures, we only trigger the ADS from catalyst beds containing hydrocracking catalyst and only from the bed outlet temperatures. For these locations, any two high temperature indicators (TIs) at the same bed outlet will trigger the ADS.

Finally, ADS functionality: When the ADS trips, it will depressurize the unit until all permissives are clear. For CLG, the permissive for pressure is reducing the system to 50% of design. The CVX permissive for pressure is reducing to 25% of design. We use a 50° F. bandwidth on high-temperature trips, so TIs have to drop 50° F. below their trip temperature before the permissive will clear. Finally, 5 min of continuous recycle compressor operation is required to clear the associated permissive. If all of the permissives are clear, then the ADS will stop depressurization. If all permissives do not clear, depressurization will continue down to the nitrogen header pressure.

My last point on ADS: The operator is generally in control of the depressurization process. Once all permissives are clear, the operator can continue or discontinue depressurization at his or her option. It is also possible to discontinue depressurization with the system pressure above 50% of design as long as all other permissives are clear.

Wu Derek had an extensive answer on this question, so I will just add a few points regarding what we do at Shell. All of Shell's hydrotreating units are equipped with both low-rate and high-rate depressuring systems, and both systems are manually activated. The low-rate depressuring systems are sized so they depressure from a normal operating pressure to around 100 lb over 1 hr. The low-rate depression valve closes when the high-rate depression valve opens.

The high-rate depression valve is normally used only when there is a major unit upset or an emergency situation, such as a big fire or a large leak. The valve is normally sized to depressure from a normal operating pressure to 100 lb in 15 min. When the high-rate depression valve is opened, it normally trips the feed pumps, charge heaters, makeup gas, recycled gas compressors, and wash oil or washwater pumps. The unit is essentially shut down.

For the hydrocracking unit, the depression system is designed slightly differently compared with the hydrotreating unit. For the hydrocracking unit, the low-rate depression valve can be either manually or automatically activated, and it can be tripped losing the recycle gas compressor. Also for the cracking bed, if there is a 50° F. above-grade change in the per-minute average temperature for the two TIs in the bottom bed, the low-rate depression valve will be automatically opened. It can also be tripped on the top bed TIs if the temperature is anywhere from 15-25° F. above the rate of temperature change per minute. The low-rate depression valve can also trip any single TI exceeding the designed temperature.

The high-rate depression system in the hydrocracker is designed very similarly to the hydrotreaters. It is manually activated when there is both a major emergency on the unit and when temperature discretion occurs, which is often in the hydrocracker and not the hydrotreaters. At Shell, if the zeolite catalyst is utilized in the hydrotreating unit, we would require the hydrotreater to follow the same depressuring system as the hydrocrackers.

Robledo Derek and Kathy covered a lot, so I will now give Haldor Topsoe's view on this scenario. Just recognize that any zeolite-containing material is going to possess an inherent ability to crack hydrocarbons; therefore, you will have the potential for a temperature excursion to take place, especially in the loss of recycle gas flow.

When Haldor Topsoe designs a hydrotreater, we specify a fast-depressurization system based on the American Petroleum Institute (API) Recommended Practice (RP) 521. We usually set that at 100 psi/min. The goal is to reduce the pressure to 50% in a matter of 15 min. That is usually achievable when you have a unit operating below 800 lb. Once you get above an operating pressure of 800 lb, the depressurization rate is increased above 100 psi/min to fall within the recommended practice of reducing pressure to 50% in 15 min.

For a hydrocracker, we specify two emergency depressurization systems: a low-rate and a high-rate system. These systems are designed to depressure at rates of 100 psi/min and 300 psi/min, respectively.

The low-rate depressurization system opens on a loss of recycle gas. Since the recycle gas flow provides the means for temperature control in the reactors, Haldor Topsoe's philosophy requires that depressurization be initiated at a low rate upon stoppage of the recycle gas compressor. This moves the reactor in a safe direction, but you can still stop the depressurization if the compressor is restarted and the catalyst temperatures are normal.

The high-rate depressurization system opens up at the onset of a temperature runaway. A temperature runaway is classified as any reactor skin temperature exceeding the reactor design temperature. If the temperature rise is 55° F. above the normal operating temperature level, then the bed is severely upset, and the unit should be fully depressured. A rate of 300 psi/min will usually exceed the 50% reduction requirement.

Part of the question was about endpoint or cold-flow improvement. Refiners are often considering adding these catalysts into their hydrotreaters. The operating procedures must be revised to reflect the differences in operation, and the operators should be trained for this new operation. At a minimum, Haldor Topsoe recommends that the unit meets API RP 521 of reducing the pressure 50% in 15 min. This is lower than what we would specify for a grassroots design. Haldor Topsoe believes, however, that this is a reasonable compromise for a revamp if:

1. A careful analysis is made of the new operation.

2. The operating procedures are reviewed and properly modified.

3. The operators are properly trained for the new operation.

4. Confirmation is made that the unit instrumentation is adequate to identify the scenarios requiring depressurization.

Again, when you are using catalysts which contain zeolite, a review of the unit's control and shutdown systems is imperative. When considering any of these catalysts for use in your hydrotreater, consult your experienced licensor or catalyst vendor, both of whom should be able to provide guidelines.

Bull Kathy, can you define what you meant by decaying average temperature?

Wu Basically, it just means looking at individual TIs and comparing their rates of change per minute. The damped decaying average value for a set of radial thermocouples is calculated each minute as a moving weighted average and incorporates the value each second by adding one sixtieth of the latest value to the weighted average over the previous 59 sec.

Bull If that rate of change is 50° F. or above in 1 min, then you trigger?

Wu Yes.

Adkins I am wondering if you have any specific recommendations on the type of instrumentation for thermometry inside the reactors, both from the hydrotreating and hydrocracking standpoints.

Blackwell There are differences, but we have had good experience with high-temperature, high-pressure thermometry provided by both of the vendors represented and showing exhibits here today.

Mechanical Integrity

What are your criteria for retiring a hydroprocessing reactor? What kind of failures have you seen? What are the inspection techniques you use and your frequency of inspection?

Wu This is a complicated question. In general, there is no need to retire a reactor unless there is an active degradation mechanism identified. So if there is one identified, inspection plans need to be in place to verify the degree of degradation and identify the mechanisms in the corrosion review documenting it.

At Shell, we use a risk-based inspection technology to establish the required extent for inspection and frequency of the inspection, as well as the assessment of the reactor's remaining life.

Here is one example. There is a soon-to-be retired reactor in naphtha hydrotreating service that was made in 1958, so it is quite old. The reactor was made of C-½ moly material with stainless-steel 304 overlays. There are two corrosion concerns with this reactor. One is the high-temperature hydrogen attack which has been managed through the inspection program and Ensure Safe Production (ESP) practices with regard to operating temperature and hydrogen partial pressure. With the recent industry experience on high-temperature hydrogen attack and the expected new API curves for this type of metal service, it is possible the Shell inspection program and the ESP practices may not be acceptable for this type of service. As a result, the refinery is considering replacing the reactor.

The second corrosion concern on this reactor is the sulfidation. Because if the stainless-steel 304 is detached from the reactor wall, there will be a concern of sulfidation corrosion. The corrosion rate can be as high as 10-15 mils/year. This type of corrosion can only be inspected through video. Doing repairs is also an option, but only for the small areas, as this would also require extensive shutdown and cost. Based on all these factors, the refinery has decided to retire this reactor.

The fitness-for-service (FFS) assessment can be used also to evaluate the structure integrity of in-service reactors. If inspection reveals issues with a reactor's mechanical integrity and the reactor is not retired, you can use API 579 guidelines to make assessments on its suitability for continued service. API 579 was developed for equipment in the refining industry for quantitative engineering evaluations on structural integrity. It also can be applied to make run, repair, and replacement decisions.

Is your reactor still safe in service today? The FFS assessment can be used for projecting the remaining life of the equipment. It also can be applied for certain damages like general localized corrosion and the presence of cracks or creep and fire damages, as well the other items on the list.

If the FFS assessment results show the reactor is still suitable for current operating conditions, an appropriate monitoring and inspection program will still need to be in place to ensure its mechanical integrity in service. If the assessment result shows that it is not suitable for service, you could consider rerating the equipment if you do not want to retire the reactor; or, you could consider retiring the reactor.

What kind of reactor failures have we seen in the industry? We have seen hydrotreaters, hydrocrackers, and even a pyrolysis gasoline (pygas) hydrotreater's reactor fail where a reactor was damaged with a hole. Many times, reactor failure was caused by a temperature excursion where the reactor was operated above the designed temperature. There are also other types of failures that are caused by corrosion, such as high-temperature hydrogen attack or cracking when reactors are exposed to too low a temperature while still under pressure. The reactor failure can also be caused by hydrogen embrittlement when the reactors are cooled down too quickly and the hydrogen dissolved into the reactor wall could not get out right away. Temper embrittlement was affected mostly on primarily 2.25Cr-1Mo material. Failures were also observed due to poor toughness of the material and mechanical or thermal fatigue.

The inspection techniques commonly used for reactor inspection are the nondestructive examinations (NDEs). Some of them have already been mentioned by Derek for their plant use, including UT, magnetic particle, penetrant, and radiographic. The infrared thermometry was actually used mostly for the cold-wall reactor design. There are also advanced nondestructive examinations such as automatic ultrasonic testing, advanced ultrasonic backscattering technique, time-of-flight diffraction, and angle beam spectral analysis for walls. Some of the tests have required a sample be extracted into a boat or scoop for the test.

The inspection frequency normally is determined by the risk-based assessment. You can reference the API guidelines for this type of inspection methodology. For the risk-based inspection (RBI) methodology, one can establish the failure probabilities and risks and then rank the equipment. Based on the ranking, the inspection frequency can be determined. It is normally recommended to focus inspection on the high-risk equipment. The cost associated with that can be high but can be offset by reducing the inspection effort on low-risk equipment.

Blackwell Our view on reactor replacement is that we have experienced very few failures. We do not expect reactors to fail. We expect to be able to repair them as needed and operate them indefinitely. We are operating reactors that we have had in service since the mid 1960s. In one case, we are considering a replacement, but this would be an economic replacement. Generally, even for older reactors, we do not manage them as though they have a fixed end of life. We replaced two reactors that were affected by adjacent fires. We do not have much of a case history where a reactor incident or defect caused us to remove it from service. In fact, we have repaired significant cracks, bulges, and other defects. While issues are uncommon, they are generally with older reactors. We have no expectation of issues with newer reactors.

Reactor age is not just calendar age. Cycles can significantly affect a reactor's condition. We have some ongoing issues associated with high-cycle reactors. In addition, older calendar-age reactors can be less reliable and may require ongoing repairs to keep them in service due to material quality, design, and fabrication issues. However, improvements in each of these areas have significantly improved reliability for newer reactors.

There have also been design improvements. For example, nozzle design and location has improved. On older reactors, nozzles are considered to be a potential weak point and are closely monitored. We seem to have fewer concerns with newer nozzles.

There have been significant improvements in materials technology such as cladding, for example. Bonding methods and the degree to which cladding is bonded to the base metal is significantly superior in newer reactors.

Fabrication methods have also improved. Welds are an example. Historically, the heat-affected zone around the weld has been an area of concern. Now, welds are implemented and geometrically designed to minimize the heat-affected zone, which has improved reliability.

For full-disclosure purposes, I mentioned that I am not a trained materials engineer. I also have no specific training in pressure vessel mechanical design, so treat my comments for this next slide as an informed third party. Unfortunately, for those who are responsible for pressure vessel inspection and reliability, there is no industry standard that specifically addresses reactor retirement. There are standards around high-pressure equipment; most notably, the FFS engineering assessment. However, there is not a one-stop checklist that you can use to determine whether a reactor should be repaired or retired.

There may be some reprieve on the horizon, but not as near-term as you might hope. I understand that API 934-I and API 934-H covering an inspection standard and a repair standard are actively in the works. While these documents are under development, I understand that they are not imminent, and the timeframe suggested to me may be another 2 or 3 years. Also, they will likely be issued as a technical report rather than a recommended practice. We do not expect failure, but reactors can degrade if not operated and maintained correctly. On my next slide, I introduce the concept of reactor life management. Through life management, we are trying to accomplish two things. First, we are trying to limit thermal cycles, which is a fatigue issue. The intent is to maximize plant reliability and extend cycles, where possible, in order to extend the life of reactors and other pressure vessels. If you have to pull a unit offline, you should try to minimize the thermal cycling.

The second issue to be managed is minimizing post-weld heat-treating cycles because this can lead to derating. Minimizing post-weld heat-treating is accomplished by managing crack and defect repairs. We use the following process to manage repairs. Cracks and defects are evaluated by highly trained specialists who make a determination about whether these cracks can be left and monitored or require immediate repair. I understand that cracks within the cladding do not generally represent a serious risk and can usually be left and monitored for propagation.

Cracks and defects in the base metal will raise a higher level of concern. However, even these can sometimes be left in place. Usually they are not left as a crack; rather, they are ground out to transform them into a local thin area rather than a high-stress crack. This is how we approach reactor life management.

While we have not had many failure or forced retirement issues, we are currently evaluating the economic replacement of one reactor. We have a 1960s-vintage reactor that has recurring cracks in the closure ring grooves and in the internal attachment welds, and these are all time-consuming to repair. They have added work and extended every recent shutdown. So there is an ongoing cost associated with the condition of this reactor. However, this cost is not sufficient to justify replacement. We can repair each of the defects so that the overall condition is still good enough for it to pass an FFS engineering assessment.

We are now in the process of developing additional economic incentives to determine if replacement is justified. For example, the reactor we are evaluating is old, and older reactors tend to have a high minimum pressurizing temperature (MPT). High MPT translates into startup-shutdown risk and time, so there is an economic value that can be assigned to replacement with a lower-MPT vessel. The design of the reactor internals is dated, and we know that performance can be significantly improved, in terms of catalyst utilization, yield, and cycle length. Finally, there are some configuration and throughput limitations. We do not yet know whether this will provide sufficient justification, but our experience suggests that economic replacement is more likely than forced retirement.

I will make a final point about derating. While derating avoids forced retirement, it may provide sufficient justification for economic replacement, especially if the derated reactor can no longer meet the minimum process requirements for the service.

Zaritsky If there is a crack in the cladding which is considered to be a nonissue, does that not leave the reactor more susceptible to corrosion on the base metal?

Wu I believe so, yes. That particular reactor is one of the concerns for our Shell refinery. They were worried that if there is a crack or if the overlay detaches from the base material, there might be additional corrosion caused by the crack or detachment.

Proops Adding to Derek's comments, I am surprised that you did not add, "Put in a bigger reactor" to increase catalyst volume and lower your start-of-run temperature to extend life.

Hansen I just want to make sure I understand the changes in the Nelson curves years back; specifically, the removal of the carbon half-moly line. It sounds like Shell is saying that they are evaluating our reactors based on carbon-½ moly, and Chevron is saying that they are looking at the whole picture. However, we feel like we will not replace reactors that are carbon-½ moly based on the Nelson curves. My question is more to Kathy from Shell. Regarding the part of your naphtha reactor that you reviewed, was the replacement based on being in the zone of a carbon-½ moly?

Wu Yes. Nelson Curves will be coming out for the carbon-½ moly type of reactors' metallurgy. Based on that information, we are evaluating those reactors which have high risk and considering replacing those reactors with improved metallurgy.

Hansen Okay, because we are looking at the same situation. We have a carbon-½ moly naphtha reactor which we are considering making bigger and replacing metallurgy with one constructed of 2.25% chrome -½ moly.

Driving Profitability

What strategies have you utilized to balance available catalyst life in hydroprocessing units with scheduled turnaround times, and how can this be optimized to increase profitability?

Wu To balance the hydroprocessing catalyst life, we set turnaround timing where we can consider two scenarios. One is the instance where excess activity is available for the catalyst before the turnaround timing is reached; the other is when there is not enough catalyst activity available, causing the unit to have a hard time reaching the projected turnaround schedule. In this case, if there is excessive pressure drop built in the reactor, it also could impact the turnaround schedule. For the scenarios where there is excess activity available before you reach the catalyst cycle life, you could minimize unused catalyst activity by processing more barrels of difficult feed; for example, processing more light cycle oil or coker gas oil to allow for the utilization of additional catalyst activity with more difficult feed. This type of feed may also create a higher deactivation rate, thereby allowing you to reach end-of-run faster.

Another way is to look at increasing the distillation endpoint of your feed, which will allow you to upgrade the lower value of feed to a higher value product and increase the profitability of the unit. This is particularly applicable for an ultralow-sulfur diesel (ULSD) unit based on the feed type.

For the FCC pretreating unit, you can consider processing more barrels of feed containing higher metal contaminants since the higher metal contaminants will poison the catalyst and then cause more catalyst deactivation while utilizing the activity. Also, you can consider changing the operating mode of the FCC pretreating unit by changing from hydrodesulfurization to hydrodearomatic [and then] to a maximum aromatic saturation mode, improving the feed quality to the FCC unit, which leads to yield improvement.

For instance, where there is not sufficient activity left to meet the projected turnaround schedule, several options exist to extend the catalyst cycle life. These options may have economic penalties that need to be weighed against the cost of changing the turnaround date. For a hydrotreating unit, if your end-of-run temperature was constrained by the feed heater and the reactors were operating in the flat temperature profile, you could switch over to ascending temperature profile to get more wraparound heat in the front of the reactor in order to run the reactor hotter to extend the cycle life. If you are already operating the unit at the end-of-run temperature and want to extend the run cycle, then consider diverting some of the difficult feed to other units, or even reducing the feed rate to still be able to meet the product specification.

Another way to look at this is to relax the product qualities specification. For instance, for the ULSD unit, you might consider increasing your sulfur specification in diesel product and then rebalancing the refinery's overall sulfur specification by looking at other ULSD producers and allowing them to run a more severe operation and produce a diesel product lower in sulfur.

Overall, there are many options that can be considered. The best solutions could be different for every refinery and each particular situation, depending on the refinery configuration, capability, and economics.

Sharpe We have used metal dispersant chemical when we had high delta-P (pressure differential) at end-of-run conditions in our hydrocracker. It is very expensive in some of our units, but it will help you get to the turnaround. So that is just one other option to consider that is not listed on the slide.

The panelists
Derek Blackwell, senior staff engineer for hydroprocessing technology, Chevron Lummus Global LLC
Sergio Robledo, hydroprocessing catalysts technical manager, Haldor Topsoe Inc.
Kathy Wu, senior hydroprocessing technologist, Shell Global Solutions (US) Inc.
Rick Manner, hydroprocessing specialist, KBC Advanced Technologies Inc.
The respondents
Jeff Bull, Valero Energy Corp.
Michael Adkins, KP Engineering LP
Steven Zaritsky, Axens North America Inc.
Kevin Proops, Solomon Associates Inc.
Ray Hansen, Sinclair Wyoming Refining Co.
Danna Sharpe, Flint Hills Resources LLC