Search

SCHOELLKOPF (Advanced Refining Technologies) 

Generally speaking, the catalyst is above 500°F. In the presence of hydrogen without the presence of H2S (hydrogen sulfide), the rate in which the sulfur can be stripped off the metal sulfide is very quick. Note that the rate of reduction reaction is determined by the hydrogen partial pressure, temperature, and time, with temperature having the strongest impact. There are a multitude of scenarios that can cause H2S to not be present in the reactor and, therefore, strip sulfur from the catalyst. Typically, during sulfiding, your sulfiding agent pump trips, which is the origin of your H2S, or the H2S is purged for the open valve misalignment. Another possibility is that the aiming system is accidentally lined up. Depending on during which stage of the startup this occurs, you will have a relatively minor inconvenience or a more complicated issue. In the case where DMDS (dimethyl disulfide) is lost and you are above 500°F and at pressure, the reduction reaction will occur quickly. The same is true for mechanical failures of equipment such as recycle compressors or feed pump trips. 

Figure 1 on the slide shows the reactions associated with sulfiding. The reversible reaction there on the left actually has to be conducted outside of the reactor. So, once you have stripped the sulfides off the catalyst, you will basically have to start all over again. If H2S is lost for any reason during the startup, ART) recommends immediately beginning to reduce the temperature to get the catalyst below 350°F. If possible, continue circulating hydrogen until the unit can restore the operating conditions. If the reason H2S was lost is expected to continue longer than 24 hours, then cooling to below 250°F is recommended to minimize any reduction opportunities. You just do not want to have any hot spots within the reactor.

 

 

Hydrogen stripping can also be very useful for removing soft coke and restoring some lost catalyst activity. However, that process can also lead to removal of sulfur from the catalyst. A minimum of a 1,000 ppm H2S is recommended to be maintained within the recycle loop during the hydrogen stripping procedure to avoid any accidental reduction. We also suggest working with your catalyst supplier for any additional specific guidelines. 

For a temporary shutdown, which can be defined as less than 24 hours or up to 48 hours, it is preferred that you shut down the scrubbing device, of course, and purge with nitrogen. However, depending on the H2S content, recycle gas can be at some acceptable temperatures to hold the unit: the cooler, the better. You continue circulating hydrogen, assuming you have plenty of H2S in the recycle loop: 1,000 ppm or better; 2,000 ppm or better is recommended. Then of course, it is always recommended to work with your catalyst vendor for any specific guidelines in unit and anything specific to your unit. 

 

LONG (HollyFrontier Corp - Navajo) 

At Navajo of HollyFrontier, we have not had any experience – in recent history – of stripping sulfur from the catalyst. Most incidental or inadvertent stripping of sulfur from hydrotreating catalyst occurs during the startup of the unit. We mitigate that potential within our startup procedures which state that the temperatures are not to exceed 400°F during the hydrogen once-through heat-up purge before introducing the feed. The biggest impact would be with regard to catalyst run-length. The highest risk pertains to if all sulfur is removed and the catalyst is stripped to base metal. At that point, the catalyst is completely deactivated. 

 

AL-FUDHAIL (Saudi Aramco) 

Just to add to what has been mentioned already: Pilot plant testing has verified that cobalt-based catalysts require higher H2S concentration than nickel in order to remain sulfided. The tendency to reduce, again as mentioned, will be higher for a clean catalyst system in startup cycle or even clean feed service. That is basically it for me. Everything else has already been mentioned. 

 

UNIDENTIFIED SPEAKER 

As you said, a common catalyst requires a higher H2S concentration. Do you have any idea of how much higher? Is it 20% or 50% higher?  

 

AL-FUDHAIL (Saudi Aramco) 

I do not really have a quantitative number. However, you would typically want it to be in the 100 to 500 ppm range. Imagine a round-up higher number – 500 ppm or close to 1,000 ppm – of H2S concentration to keep it in a sulfided state. 

 

SERGIO ROBLEDO (Delek US) 

I think the point of the question was that we all understand where you are in oxidic state. You are going to reduce the catalyst very fast, right? Often, the question comes up when you are up in operation and have been running for a year or two. My question to the audience is: How many of you have run a hydrotreater at 600°F with no H2S present and for how long of a window activity penalty? 

 

AGGUS (Becht Engineering Co., Inc.) 

Can I answer that for those who do not wish to be identified? I do know of one instance, during a hot hydrogen strip, where it had not been previously carried out on the unit. The procedure that was utilized had been previously been employed for another unit without an amine absorber in the recycle circuit. In this case, the H2S absorber is left in operation and the temperature run up to 700°F; so, eight hours and no loss of activity. At least the way that it was monitored against the budget curve, for the remainder of that run, there was no decrease in overall catalyst activity. Therefore, the amount of energy required to get stripping of sulfur done may be higher than some may think. 

 

ANDREW MORELAND (Valero) 

We had a similar incident where we had a unit on standby circulation. Per procedure, we cooled the reactor down to 450°F. The amine scrubber was off, so we monitored the H2S build in the circulating gas. It had built up to between 50 and 100 ppm, at which point we have sufficient concentration of H2S to keep the catalyst sulfided. So, I agree with what was said: There is no activity effect, but sulfur will come off the catalyst. I will say that some of the sulfur that comes off the catalyst can be put back on as soon as feed is introduced without much of an activity penalty.  

 

AGGUS (Becht Engineering Co., Inc.) 

Just turning into an AA meeting. [Laughter] 

 

AL-FUDHAIL (Saudi Aramco) 

Actually, I have worked with a hydrocracker, and my experience was that it was quite a troublesome unit. We had a lot of shutdowns and many ups and downs. Really, there were times when we ran the recycle gas with fresh catalyst after a shutdown and after a purge. There is not really much H2S in the system. Again, this experience is similar to what has already been mentioned. Once you get the feed into the unit, whatever sulfur that has come off will soon come back and be sulfided in the catalyst. Really, not much penalty or any reduction in activity has been noticed. 

Now coming back to when we are done with the unit or getting ready to shut it down for a catalyst changeout, we do the hydrogen hot strip. And boy, that catalyst system still emits sulfur H2S, even after we have done the 750°F hydrogen strip with no H2S, clean hydrogen. You take it offline and try to slip that blind to get out a lot of H2S from that reactor. So, I do not know, because I have never experienced sulfur stripping out of a catalyst system. 

 

PATRICK GRIPKA (Criterion Catalyst & Technologies L.P.)  

I am going to respond to Brant’s comment and the other comments made after his. There will be some H2S coming off because you have coke on the catalyst. There is some sulfur inside that coke which you put on the catalyst, so a lot of it depends on how heavy the feed you are operating, how long you have operated, etc. I think the recommendation will still be to err on the safest side, like Lyle suggested. If you are doing a hydrogen strip, 1,000 ppm H2S is a good place to be. Like Andy said, “It will build up some.” Therefore, 1,000 ppm H2S is essentially a safety margin for you to operate under. 

 

AGGUS (Becht Engineering Co., Inc.) 

Yes. We are not trying to encourage bad behavior. [Laughter] 

 

BRANT AGGUS (Becht Engineering) 

I saw an incident where a recycle hydrogen absorber in a gasoil hydrotreater circuit was inadvertently left in service during a hot hydrogen strip. The recycle hydrogen had essentially a 0 ppm H2S concentration, as reactor inlet temperature was held at 700°F for nearly eight hours. After the unit was brought back online, no deviation from the pre-strip required WABT (weighted average bed temperature) was observed nor did the required WABT overtime deviate from the pre-strip deactivation trend. Stripping sulfided catalyst might require more energy than many think. 

 

SUHEIL ABDO (Honeywell UOP) 

Base-metal hydrotreating and hydrocracking catalysts are susceptible to sulfur stripping at high temperature in the presence of flowing hydrogen and in the absence of H2S. For that reason, we advise unit operators to avoid prolonged exposure of sulfided catalysts to flowing hydrogen at temperatures exceeding 500°F if the H2S content of the hydrogen is below 50 ppmv (parts per million by volume). It is recommended to maintain between 50 and 100 ppm H2S in the recycle gas to ensure adequate sulfur availability. Other factors, such are exposure time, may impact the severity of stripping, but keeping the temperature below the 500°F limit should protect the catalyst for a prolonged period. 

  

PATRICK GRIPKA (Criterion Catalysts & Technologies) 

There are three operating scenarios in which hydrotreating/hydrocracking catalysts can be reduced: 1) upon startup, 2) upon abnormal operations in a very low H2S environment that is usually a result of a unit upset, and 3) during normal operation in a very low H2S environment. 

 

 

 

2. Likewise, in instances of a unit upset in which recycle gas is lost, our primary recommendation is to cool the catalyst beds to 450°F or below as soon as it is safe to do so. Upon restart, establish at least 1,000 ppm H2S in the recycle gas before heating above 450°F. Depending on the nature of the shutdown and cooldown, an LCO/diesel flush or a hot H2 strip may be recommended to minimize the loss of catalyst activity to the shutdown. 

3. During normal operation in a low H2S environment [e.g., a very low feed sulfur NHT (naphtha hydrotreater) unit], it is recommended to inject some sulfur source [e.g., DMDS (dimethyl disulfide)] in order to keep a pre-determined level of H2S in the recycle gas.  

In reality, only 50 to 100 ppm H2S is needed to keep metals sulfided. Some will argue that even a trace of H2S is sufficient. But the reality is that the lower the H2S content, the better the distribution that is required; otherwise, you may have pockets where no H2S is present. Also, reduction requires time and temperature; so, it is not an immediate process. Having a conservative guideline of 1,000 ppm H2S to prevent sulfur stripping/metals reductions provides a comfortable safety factor. For specific details on your operation consult your Criterion representative.

 

 

 

PAPPAL (Valero)

This question is a handful to cover in three minutes, but here we go. This is a generic refinery processing sour crude. First of all, we add a coker to this generic refinery.  

 

 

The heavy vacuum gasoil (HVGO) produced will be high in sulfur and nitrogen. The HVGO and the coker heavy gasoil (CHGO) are upgraded through the FCC. The CHGO, in this case, can be 14 gravities with 4,500 ppm of nitrogen. I want to acknowledge Jeff Bull, my colleague in San Antonio, who modelled these generic refinery configuration cases. 

 

 

In an FCC/coking refinery, running this HVGO/CHGO blend results in a very low FCC conversion level of about 63%. The FCC produces a great deal of cycle oil and slurry. In addition, the volume swell for this case is not very high. 

 

 

Now, we add a hydrocracker to this refinery configuration. Recall that we have the same crude slate, HVGO, and CHGO streams to upgrade. We add a reasonably powerful hydrocracker operating at 40 or 50% conversion. 

 

 

At this hydrocracker conversion level, many of the aromatics are converted and the sulfur and nitrogen are essentially completely removed. The unconverted oil from that hydrocracker operation is much improved at 35 gravity with virtually no sulfur or nitrogen, compared to the HVGO/CHGO feed to the hydrocracker. To stay in gasoil balance, the refinery would have to import HVGO or increase crude run. This scheme adds refinery G/D (gasoil/distillation) production ratio. 

 

 

The FCC operation in this configuration is much improved as the cat cracker conversion is over 80% instead of the low 60s without the hydrocracker. There is significant light olefin yield to produce alkylate and lots of gasoline for blending. FCC cycle oil yields are significantly lower, and C3+ volume swell increases by 10%. Likely, this is a more economic operation than the previous operation. The message here is that adding a partial conversion hydrocracker to a coking refinery can have a large impact on the operation and can improve the overall economics of this refinery. 

 

 

To summarize, the FCC and coker operation can be decoupled as the hydrocracker converts aromatics. Potentially, the coker could produce a more difficult coker heavy gasoil; the hydrocracker would protect the FCC from that difficult feed. Our volumetrics in the coker and cat are decoupled a step further. If we were to add a deasphalter, our ability to run heavy crude would increase while staying within coking and FCC capacity limits. Adjustments to FCC and coking operation are likely needed. There is further opportunity to optimize crude. 

 

 

What is the impact of the hydrocracker on the lube block? 

 

 

The UCO, as feed to the conventional base oil plant, increases the VI (viscosity index) potential of the product mix. The VI boost is decoupled from crude source. The hydrocracker is a great homogenizer of feedstock. Hydrocracking selectively cracks out aromatics, concentrating paraffins. Paraffins are the preferred component in the lube block. The conventional lube oil block would benefit from increased base oil yield from that plant. In addition, non-traditional crudes could be upgraded to produce premium lubricants. 

 

 

If a catalytic lube configuration is producing Group 2 and Group 3 lubes, using the fuel’s hydrocracker UCO as feed increases the high-quality base oil yield. The benefit is reduction of the lube hydrocracker conversion target required to meet the given VI target. Directionally, lube oil yield increases at lower tube hydrocracker conversion. Overall, the yield of premium quality base oil from that plant will increase. 

 

 

AL-FUDHAIL (Saudi Aramco) 

As you can see, and as Gordon has mentioned quite eloquently, this is quite a hefty question; so, bear with me to get through the answer.  

A lot of my answers are based on actual experience with one of our existing units. It is a vintage unit designed in 1977 and commissioned in 1982. It is a single-stage unit with liquid recycle, two parallel chain configurations, and four reactors in series in each train. The reactors have now been in service for over 35 years. Due to aging of the reactor, in 2009 we discovered signs of overlay disbanding that eventually propagated to cracks in a few locations and required a patch repair. 

Now the first question is: What are common mechanical defects that occur to the weld overlay material in hydroprocessing reactors? First, microstructure changes or sigma-phase embrittlement. This damage can occur either during original fabrication and post-weld heat treatment or during operation. Therefore, the weld overlay conditions during a fabrication govern the susceptibility to hydrogen disbonding. Higher traveling speeds and welding currents result in good resistance as they generate fine microstructure without, of course, planter grains at the interface. Therefore, the fabrication stage is a crucial step and requires dedicated attention to avoid inherent damage. 

  

Second, low cycle fatigue cracking thermal grading during startup and shutdown cycles: Thermal stresses and residual hydrogen concentration drive weld overlay cracking, as well as the disbonding mechanism, i.e., cracking along weld overlay, which is a chrome-moly-based metal interface. During heating and cooling cycles, the coefficient of thermal expansion mismatch between the stainless-steel weld overlay and base metal generates excessive stresses at the interface for an embrittled transition layer. This cycling of temperatures aggravates the situation, as you can see shown on the slide. 

 

 

Third, weld overlay disbonding: High concentration of hydrogen remains in the reactor even after shutdown. Such hydrogen – in excess of 4 ppm – at ambient temperature is considered beyond equilibrium concentration and can cause hydrogen embrittlement of transition zone, which is the more sensitive zone. Hydrogen disbonding is purely a hydrogen distribution issue. The best way to slow down this problem is to control the amount of hydrogen dissolved in the weld overlay and base metal. This resolution can be achieved by utilizing a lower operating temperature, lower operating pressure, and smooth shutdown procedure. Also, the microstructure at the interface – being a primary factor – can be improved to govern the resistance to disbonding. 

 

 

Fourth and not least is the synthetization of non-stabilized dust in stainless steel. Almost all non-stabilized stainless steel gets sensitized – either during fabrication or during service, leading to loss of its corrosion resistance to hydrogen and H2S attack. This loss of sensitivity will also make overlay susceptible to polythionic acid or chloride stress corrosion cracking. Therefore, NACE requires washing all stainless-steel surfaces in reactors – especially the ones that are non-stabilized or sensitized in reactor circuit – with an alkaline solution to neutralize against such hazards during outages. 

The second question is: What are the most common locations for defects, and does the location play a factor in the mechanical integrity of the equipment? As you can see on the slide, the most critical areas where damage can occur are at structure welds. These areas require periodic inspection of specific parts of the equipment such as hands-to-shell junction, nozzle details, internal supports, and skirted vessel joints. Finding a defect in these areas can affect the mechanical integrity of the equipment. 

 

 

 

Third: How are the defects detected and repaired? Defects are detected via periodic inspection using several techniques. Acoustic emission testing (AET) is a qualitative technique that can detect small-scale damage during pressurization, depressurization, and/or plant operation. The benefit of AET is that it can optimize subsequent inspection with NDE (non-destructive examination) focused on locations of high signal intensities that are identified by the acoustic emission prior to shut down. Second, automatic ultrasonic testing techniques include external C-scan and mapping of the shell to detect internal damage. Third, the time-of-flight diffraction technique is used to detect and measure macrocracks, preferably UT (ultrasonic testing) if crack sizes are needed for Fit For Service assessment or remaining life calculations. Fourth is your basic institute of metallographic replication to detect microcracks. Fifth, the conventional technique: straight-beam UT (ultrasonic thickness), MT (magnetic particle inspection), or PT (dye penetrant testing) for internal/external surface inspection. These tests are all used to detect macrocracks. As for the repairs, in the past we have utilized the original reactor manufacturer to develop patch repairs for the defects. 

Last but not least: How often do refiners conduct Remaining Life Analysis and/or Fit For Service on critical equipment, Remaining Life Analysis or Fit For Service Assessment are normally based on inspection findings about the state of the reactors, such as cracks or base metal corrosion. The assessment is done by calculation based on inspection data, material of construction, data of components, and historical operating data of the equipment. For aged or vintage designs, CRIP (Research Center in Pulp/Paper Engineering) evaluation may be required if operating above 800°F for chrome-moly steels. Thank you.  

 

LONG (HollyFrontier Corp - Navajo) 

Noaman did a good job answering this question, so I will just try to highlight some of our experiences. We have not experienced these mechanical defects on weld overlay material in hydroprocessing reactors in recent history. We have experienced more issues with regard to fabrication defects.  

 

MARK MUCEK (Honeywell UOP) 

The common mechanical defects that occur are disbonding and cracking. Non-vanadium-modified 2¼ Cr-1 Mo material is more susceptible to disbonding than vanadium-modified 2¼ Cr-1 Mo-V material. V-modified 2¼ Cr-1 Mo has a much higher solubility of hydrogen compared to conventional 2¼ Cr -1 Mo.  

Disbonding occurs at the interface between the weld overlay and the base metal substrate. There have also been instances of cracking between the weld overlay beads due to incomplete slag removal. Another potential situation would be if the cracks spread or propagate into what are called “spider cracks”. If the cracks are down to the overlay/base metal interface, the concern is that the process fluid can come in contact with the Cr-Mo (chrome-molybdenum) base metal and cause significant corrosion, which will undermine the weld overlay and create a larger problem. If the cracks propagate through the overlay and into the Cr-Mo base metal, which is not necessarily common, a situation may develop that could cause the crack to propagate into and through the base metal, resulting in a leak to the atmosphere. 

Disbonding is usually found visually when a slight bulge is noticed in the overlay. Cracks may also be determined visually or with Penetrating Testing if there is a suspect area. Disbonding is typically monitored rather than repaired. If the disbonded area or the bulge in the overlay get too large, or if there are multiple disbanded areas, the reactor is often replaced. Disbonded overlays have been repaired, but these repairs do require a local post-weld heat treatment (PWHT). If the affected area is substantial, conducting a PWHT becomes impractical. 

Some of the industry Best Practices and repair techniques are:  

1. If the cracks are very shallow, they can be ground to extinction and the area monitored at each unit turnaround. 

2. If the cracks are slightly deeper but do not penetrate to the base metal interface, it may be possible to do a weld repair without a PWHT. In order to accomplish this repair, a full-size weld mockup needs to be made and a weld procedure developed. A cross-section of the mockup must be polished and examined. Optical metallography and hardness testing of the Cr-Mo base metal must demonstrate that the weld procedure results in a repair that does not cause a heat-affected zone in the Cr-Mo base metal. 

3. If the cracks propagate to the interface with the base metal after grinding the cracks out, one of two methods are typically employed. 

1. A stainless-steel patch is placed over the ground area and fillet-welded to the intact weld overlay. The patch may or may not have refractory material placed in the gap. This type of repair must be inspected at every turnaround to ensure that the fillet welds are not cracked. 

2. The weld overlay is restored, and the area is given a local PWHT. 

4. If the cracks propagate into the base metal, these cracks must be ground to extinction. The excavation in the Cr-Mo (chrome-molybdenum) base metal is restored with matching filler metal, followed by restoration of the weld overlay. The repair must then receive post-weld heat treatment. Alternatively, the reactor can be replaced. 

 

W. BRYANT DUNCAN (Honeywell UOP) 

How often you conduct Remaining Life Analysis (RLA) and Fitness For Service (FFS) evaluations is a hard question to answer for any one specific refiner. The equipment design is based on expected operating conditions such as pressure, temperature, fatigue cycles, and corrosion rates. The reality is that this critical equipment is not always operated per design. Operating circumstances, operating performance, feed makeup, unit excursions, and manufacturer quality will all have an impact on the remaining life and Fitness For Service details associated with all pressure-containing equipment. 

The type of inspections performed during the life of the equipment, either onstream and/or during planned maintenance events, will dictate the next inspection intervals, the remaining life of the equipment, and whether the equipment is fit for service. The information gathered during these inspections/testing events will be utilized in any considerations related to useful life of critical equipment. Critical data collected over time should be utilized to either validate the design life of the equipment or correct the projected damage progression rates in the design of the equipment. This information could also be utilized to give you corrosion rates (short- or long-term), next required inspection dates, and/or anticipated repair or replacement dates, or the data could be used when making repair or operational decisions to safely operate the equipment. 

The following are only a few of the codes or recommended industry Best Practices that could be considered in the type of inspection, testing, and methods considered when determining or affecting remaining life and/or fitness for service. Keep in mind that there may be requirements, above and beyond those listed here, which are driven by jurisdictional considerations, owner/operator internal procedures, or the owner/operator insurer.  

• API 510 Pressure Vessel Inspection Code: In-Service Inspection, Rating, Repair, and Alteration 

• API 570 Piping Inspection Code: In-Service Inspection, Rating, Repair, and Alteration of Piping Systems 

• API 653 Tank Inspection, Repair, Alteration, and Reconstruction 

• API 571 Damage Mechanisms Affecting Fixed Equipment in the Refining Industry 

• API RP 572 Inspection Practices for Pressure Vessels 

• API RP 573 Inspection of Fired Boilers and Heaters 

• API RP 574 Inspection Practices for Piping System Components 

• API RP 579 Fitness-for-Service 

• API RP 580 Risk-Based Inspection 

• API RP 581 Risk-Based Inspection Methodology 

• API RP 584 Integrity Operating Windows 

JOHNSON (Motiva Enterprises LLC) 

What we have modeled here for you today is both cases. So, obviously, in a general sense, increasing heavy coker gasoil content will increase Conradson carbon and nitrogen. Then, the coking tendency in deactivation rate will also increase.  

To walk you through the curves, the top case shows when we are looking at the treating and cracking reactors relative to increasing the endpoint only. As you can see, in the beginning, the curve is fairly ‘flat’, in terms of the deactivation rate. However, depending on the ‘severity’ of the endpoint shift, you will get more of an exponential curve increase in deactivation. In the other case where we hold cutpoint constant and increase the percentage of heavy coker gasoil within the feed pool, we get more of the linear relationship, in terms of deactivation rate on cracking and treating reactors. 

  

Obviously, the more ‘difficult’ nitrogen species will increase contaminants such as nickel, vanadium, and arsenic, which will also cause more carbon buildup on the catalyst and increase catalyst plugging and aging. With heavy coker being as olefinic as it is, we are also increasing the polynuclear aromatic precursors and then, therefore, increasing the buildup of our polynuclear aromatics. It is interesting to see what truly happens at cutpoint versus just barrels on qualities. 

 

 

RAMACHANDRAN (Bharat Petroleum Corporation Limited) 

From a practical standpoint, the impact of higher feed rate and endpoint increases 

• Fouling of heat exchangers, 

• Plugging of high-pressure vapor coolers, 

• Pressure drop on the catalyst bed, 

• Deactivation of catalyst, and 

• The formation of coke, and also causes coloration and instability in the product. 

The impacts of this higher feed rate and end point can be reduced by ensuring that we have an amount of drag or purge stream – sometimes at the cost of conversion – which will split unconverted oil into light and heavy fractions. Only the light stream is recycled; the heavier stream is taken away as rejects. What helps is selective HPNA absorption from the recycled stream, as well as maintaining the feed endpoint typically around 1022°F instead of 1100+°F in case of once-through hydrocrackers with about 85 to 90% conversion. Maintaining a high recycle ratio in the coker also helps, although it will result in more coke formation and will crack into lighters at the cost of distillates. 

  

PAPPAL (Valero) 

In an extinction recycle hydrocracker with high conversion targets, HPNA (heavy polynuclear aromatics) production likely increases as CHGO is added to the feed. Generally, you have to reduce conversion and increase unconverted oil purge rate to compensate for the increased HPNA production. To judge the viability of CHGO upgrading through an extinction recycle hydrocracker, an economic analysis is likely required. The overall value of processing CHGO through the hydrocracker would be compared to the next best non-hydrocracking alternative for processing the CHGO. In general, the higher the pressure of the hydrocracker, the more improved the capability for upgrading CHGO at high conversion levels will be. 

 

SALVATORE TORRISI (Criterion Catalysts & Technologies, L.P.) 

I have a response and a question. In general, my experience is that coker fractionators perform differently than vacuum towers. A10ºF increase in endpoint for a coker gasoil adds more difficult species than when you increase HVGO endpoint by 10°F. The actual fractionation quality is not quite as good in a coker, so that seems to drag in a lot or all of the elements you mentioned.  

I heard you speak about many of the impurity implications except for asphaltenes. Do you have any general rules for asphaltenes, in terms of increasing endpoint by 10ºF? Is there a maximum value you recommend for color, asphaltene content, or metals level, from a coker feed quality standpoint? Do any of the panelists have an opinion? When is an endpoint too high? 

 

PAPPAL (Valero)  

CHGO in a properly functioning unit should be very low in asphaltene content. A typical number quoted is 300 to 500 ppm as the limit of asphaltene content to a hydrocracker. So, if a blended feed starts exceeding that value, deactivation will likely increase. Online measurement of HPNA in unconverted oil is currently unavailable. What is needed, from an operating standpoint, is a quick-and-dirty workaround that relates the laboratory-measured HPNA values with an easily measured property. An example is using the color of the UCO as a proxy for the HPNA content of that stream. The objective is to limit the HPNA within targeted levels. As more difficult feed is added to the hydrocracker, optimization within limits is required. 

 

JOHNSON (Motiva Enterprises LLC) 

In terms of Motiva’s experience with processing heavy coker gasoil, typical limitations you are up against are the coke fines themselves, especially if we have drum switching issues, just like David was saying with the color as a ‘go-by’ and not directly measuring the PNAs. However, the color is an indicative point of where you stand, in terms of how much higher you can push conversion. We have never had an experience when we were able to maintain conversion targets in the 96 to 98% range. But when pushing the percentage of the pool, we do start to see some color changes and other limitations on filtration and pressure drops. 

  

HEIDI DALEY (Haldor Topsoe, Inc.) 

HCGO will contain very high nitrogen content and can also have high concentration of polyaromatics and carbon residue as the endpoint is increased. Higher HCGO rate and/or endpoint will, therefore, increase the catalyst deactivation rate in a hydrocracker. Obviously, from a catalyst vendor’s perspective, we recommend that you utilize a catalyst with maximum HDN (hydrodenitrification) activity in this application to deal with the higher nitrogen feedstocks. In addition, a tailored guard bed will help to minimize contamination of those high-activity catalysts that are designed specifically for some of the contaminants that will be present. 

For a full conversion unit with recycle, HPNA formation will increase as you increase the endpoint and/or the amount of coker gasoil in the feedstock. HPNAs will have to be managed to avoid buildup in the recycle oil. Haldor Topsoe’s solution is our HPNA Trim™ technology which can be added to an existing hydrocracking unit. It basically involves removing the HPNAs with a simple stripping operation added to the backend on the recycle oil – as was mentioned previously – and separating HPNAs out based on their higher boiling point, thereby greatly reducing the likelihood of deactivation of the catalyst due to the condensation of the heavy PNAs. 

 

ANDREW MORELAND (Valero) 

Waterwash is another limit you will hit. A lot of our units cannot handle increasing heavy coker gasoil or any increase in feed nitrogen content based on waterwash limitations. 

 

RAJESH SIVADASAN and SIMERJEET SINGH (Honeywell UOP) 

Processing heavier and cracked feedstocks poses many challenges to the hydrocracking unit. Thermally-cracked feedstock such as HCGO, apart from being unsaturated, has relatively lower API, higher sulfur, and nitrogen content, higher proportion of C7 insoluble, and Conradson carbon residue (CCR). An increase in the HCGO distillation endpoint results in a significant increase in the proportion of polynuclear aromatics (PNA) and asphaltenes, both of which are coke precursors, which results in an exponential increase in catalyst deactivation rates. 

Analysis of a commercial HCGO feed sample indicates that the proportion of 4- to 5-ring aromatic compounds, which are HPNA precursors, begins to appear in the feed as the HCGO 95% distillation point approaches 775°F (410°C) and increases exponentially with each incremental step change. Figure 1 below shows a typical change in HCGO feed PNA concentration with the increase in the tail-end distillation. 

 

Figure 1.4+-Ring Aromatics 

Figure 2 below shows four different feed streams with their PNA type breakup. Red dots are the distillation endpoint of each of the feed blend components. It can be seen that even though the HCGO endpoint is much lower than the SR VGO, relative concentration of 5- and 7-ring+ compounds or PNA precursors is much higher. Increased proportion of these precursors in the feed may result in HPNA buildup in the unit for the same conversion. 

 

Figure 2.PNA-Type Breakup 

 

Figure 3. Increased HCGO Endpoint 

 

Increased quantities of HCGO processing also increases the pounds of silica entering into the unit. If the guard bed catalyst does not have sufficient margin to remove this silica from the feed, it may lead to premature breakthrough into the main treating catalyst leading to shortened life; so, the guard bed catalyst also needs to be customized for processing HCGO in the unit.  

Other issues with HCGO processing in a hydrocracker unit is the inability to or difficulty in maintaining the design HCGO endpoint due to transient nature of delayed coker unit operation. Increasing the sampling frequency, Advanced Process Control implementation on coker fractionator, and use of SIMDIST test methods D2887/D7169 (instead of D1160) often prove useful in monitoring and controlling the quality of HCGO feed. Advanced characterization methods like high-resolution mass spectroscopy (HRMS) can be also be utilized, especially during the design stage, to identify the feed contaminants at the molecular level and customize the catalyst system and unit design. 

Both new and existing units with the capability of processing these types of feeds should have high operating pressure (preferable greater than 2,000 psi), larger reactor volume, and sufficient capacity margin in the existing hydrogen supply to cater to the increased demand. Depending on the unit operating severity, recycle gas scrubber, hot separator, and some sort of HPNA management option may also need to be included in the unit process flow scheme to improve the unit’s reliability and maintain or exceed the existing catalyst cycle lengths. 

 

RICHARD TODD (Norton Engineering Consultants, Inc.) 

Heavier feeds will definitely have an impact on both the pretreating catalyst and the cracking catalyst. Pretreat deactivation may increase significantly and affect cycle life. The potential of fouling exchanger surfaces also increases due to HPNA formation, which may also limit cycle lengths. In an HC without recycle or with the bottoms going to an FCC, catalyst aging may be the primary effect. 

LONG (HollyFrontier Corp - Navajo) 

Hydrogen plants have several areas to target, when it comes to efficiency. There are several factors that contribute to energy efficiency, and all process variables vary greatly from plant to plant. 

I am going to start with the pressure swing absorber. The main target production efficiency of a hydrogen plant is the PSA (pressure swing adsorber) efficiency. This is calculated as a ratio of PSA product hydrogen to inlet PSA hydrogen. A PSA efficiency greater than 85% is considered to be adequate. Another ratio to consider is efficiency of conversion, which is the unit feedstock- to-hydrogen production ratio. This efficiency of conversion has an impact on total plant operations. Typical efficiency of conversion is 2.1 to 2.4, and variations in the plant average value can indicate operational upset. Operational impacts on PSA efficiency consist of valve switching failures and PSA feed gas. As carbon monoxide concentrations increase in feed gas, the efficiency of the PSA is reduced. Another way to measure energy efficiency is by evaluating the energy conserved per unit of hydrogen produced. 

Now we move on to the reformer. The factors that can impact energy efficiency in a steam methane reformer (SMR) indicate catalyst activity, burner operations, heat lost to atmosphere, furnace operation, heating values in BTU (British thermal unit), tube life, shift equilibrium or steam-to-carbon ratio S/C, and potentially ambient temperature. A direct-monitored target for SMR consists of methane slip and outlet target temperature, which varies from plant to plant. Methane slip consists of 1.5 to 5 dry mol% and impacts heating values, in terms of BTU. Methane slip is controlled in the reformer by shifting the equilibrium or manipulating the steam-to-carbon ratio and SMR outlet temperature. Typical steam-to-carbon ratio is 2 to 3.5, but this ratio can greatly vary; again, from plant to plant.  

If the reformer reaction equilibrium was shifted to increase hydrogen make, steam, or temperature, this modification will reduce methane content in the PSA off-gas in the reformer heater. As BTU values of the off-gas decrease, the secondary burners may have to be fired harder, which will require an increased amount of purchased natural gas. Increasing methane slip correlates to an increase of heating efficiency and – potentially – an increase of reformer tube life. Increasing methane slip is achieved by decreasing the steam-to-carbon ratio or decreasing reformer outlet temperature. To increase hydrogen, make, the steam-to-carbon ratio can be increased; however, impacts, heating value, and tube life should be considered. It should also be noted that operating at higher reformer temperatures will directly decrease tube life and catalyst life.  

In the high-temperature shift converter, the factors that impact energy efficiency in the shift converter are the inlet temperature and catalyst activity. The shift converter should target constant inlet temperatures as temperature swings impact catalyst activity. 

The exotherm across the shift converter should be monitored, as well as CO slip. Inlet temperature can be increased to maintain constant exotherm as catalyst deactivates. The local startup temperature differential is 100ºF, and a target startup run temperature is below normal. It is common to have temperature step changes that occur every six months. If a plant is short on hydrogen, the inlet temperature can be increased to promote CO conversion. The target shift converter CO slip consists of 1.5 to 3 dry mol% and indicates catalyst deactivation. This percentage depends on each facility, as well as start-of-run conditions. The operational factors that impact efficiency of carbon monoxide slip are inlet temperature and the steam-to-carbon ratio. CO slip can be decreased by increasing inlet temperature or increasing the steam-to-carbon ratio. The temperature differentials will be indicated by catalyst activity. Inlet temperature can also be increased to achieve the target temperature differential, which changes from plant to plant. 

Lastly, the sulfur guard: Frontend sulfur removal has an impact on catalyst efficiency in the hydrogen plant. The sulfur component being removed is H2S (hydrogen sulfide) and variations of mercaptans. The desulfurization previously consisted of activated carbon at ambient temperatures. Desulfurizers have used several types of media, from zinc oxide to carbon beds. The media type can impact efficiency, depending on the plant design and temperature parameters. 

 

AGGUS (Becht Engineering Co., Inc.) 

All I will add about efficiency is that a lot of the people who designed the plant – CB&I, in particular – use a plant efficiency term. It is not just the amount of natural gas you convert to hydrogen; you have to throw the steam in there as well. I do not know how many hydrogen plant unit engineers we have here. But to those who are in the audience, I want to say that it is a good exercise to add this efficiency calculation to your daily monitoring. You want to take the amount of feed on a heating value basis, the fuel that is going into the furnace, and then the difference between your energy and boiler feed water to steam, and then divide that by the hydrogen product. If you monitor those values every day, you will be doing a good job of taking care of your SMR. If only it was that simple, right? 

The heavy hitter, as far as energy usage and efficiency in your SMR, is the furnace. It is just like any furnace: You want to be able to monitor your excess oxygen and excess air. So, try to maintain 2.5 to 3% excess O2 in the furnace. What also really affects the efficiency of hydrogen recovery, as Sarah said, is the PSA. The unit temperature is easy to monitor. Make sure you are staying below 110ºF. It is your PSA. You will have best absorption efficiency if you do that. Then if you are really nice to your process control engineers, you can also play with the cycle time on the PSA. So, longer absorption times will give you less hydrogen loss and blowdown repressurization and should increase your recovery. 

 

LEWIS LUDWIG (UNICAT Catalyst Technologies, Inc.) 

 I will just say that the typical minimum for steam-to-carbon ratio is 2.8. I think the panelist said 2. We think you would run into real problems running at 2. The other important point is pressure: the lower the pressure you can run on the outlet of the SMR, the better the equilibrium. However, that is usually a design consideration and not something that the operator can do with an existing SMR. 

 

KEN CHLAPIK (Johnson Matthey) 

I have a few comments. One, in tomorrow’s P&P session, we will have Air Products talk to us about some of their experiences in dealing with efficiency through the decades that they have been producing hydrogen. It will be an operator’s view, so you are welcome to experience that at 8:00 tomorrow morning. 

We put a lengthy answer in the Answer Book, and I want to add two comments about efficiency now. We are hearing that a lot more hydrogen plant operators want to address and improve on efficiency. As you start to change some of these variables, monitoring your unit will become even more important. The steam methane reformer, as has been said by the panelists, is a critical part of that hydrogen production unit. We have been working with Daily Thermetrics over the last decade and have developed and established the application of their CatTracker® thermometry for in-tube reformer thermometry. CatTracker® for reformers gives you a “sight glass” in that reformer, with respect to the reactions that are going on in the tube. When running the reformer at these more efficient conditions, the operator can see the quick response of operating changes as they are made. This enables the operator to maintain reliability at these new conditions, which is so important in hydrogen production. 

My other comment is that there are operators who are trying to look at low capital ways to address efficiency and production. Usually when you start looking at improving efficiency at low capital, this limits the number of options. Johnson Matthey has a step out reforming technology in our CATACELJMSSR, a stackable structure reactor that replaces the catalyst pellets in the reformer tube. SSR enables an operator to make a step change in efficiency through improved heat transfer, activity, and pressure drop in the reformer tube allowing the same production with 5 to 10% less firing or 5 to 10% more production prior to plant modifications. 

 

ROBERTSON (AFPM) 

Since there are no other questions, we will now conclude this Question & Answer session. Thank you, again, to all of the panelists for their informative presentations and responses today. We appreciate all of their efforts and contributions. And, thanks to all of you here today for your participation, as well. 

 

BRANT AGGUS (Becht Engineering) 

When discussing efficiencies, it is important to define the plant efficiency term. In most cases, hydrogen plant efficiency is measured by calculating the energy [BTU/scf (British thermal unit/standard cubic foot)] required to generate product hydrogen. This calculation involves adding input streams on an energy basis (feed and fuel), subtracting the output streams (export steam, other export streams, etc.), and then dividing the result by the product hydrogen flow (see equation below).  

Efficiency (BTU/scf) LHV: (Feed + Fuel – Steam)/Hydrogen Product 

The export steam term is based on the energy difference between the export steam conditions and the incoming boiler feedwater conditions. This simple formula is used by technology licensors, like CB&I, to benchmark unit performance. It is a good idea to include it in daily unit monitoring and long-term trending. 

Plant configuration, particularly the addition of combustion air preheat, will affect efficiency; so, it is important to compare like-to-like. 

In addition to the items Sarah covered about operational factors that impact efficiency, I will add that furnace-side operation has a large impact on overall plant efficiency. Excess air should be minimized (typically to 10%, or 3% excess O2). 

For the PSA unit, the cycle time can be maximized to decrease the hydrogen loss associated with the blowdown and repressurization steps. In addition, the inlet temperature should be maintained below 110°F for optimal performance of the unit. 

 

SARAH LONG (HollyFrontier Corp - Navajo) 

Hydrogen plants have several areas to target when it comes to efficiency. There are several factors that contribute to energy efficiency, and all process variables vary greatly from plant to plant.  

 Pressure Swing Adsorber (PSA) 

The main target production efficiency of a hydrogen plant is the PSA efficiency. PSA efficiency is calculated as a ratio of PSA product hydrogen to inlet PSA hydrogen. A PSA efficiency greater than 85% is considered to be adequate. Another ratio to consider is efficiency of conversion, which is the unit feedstock-to-hydrogen production ratio. The efficiency of conversion has an impact on total plant operations. Typical efficiency of conversion is from 2.1 to 2.4, and variations in the plant average value can indicate operational upset. Operational impacts on PSA efficiency consists of valve switching failures and PSA feed gas. As carbon monoxide (CO) concentrations increase in the feed gas, the PSA efficiency reduces.  

Another way to measure energy efficiency is by evaluating the energy consumed per unit of hydrogen produced. 

  

Reformer 

The factors that impact energy efficiency in the steam methane reformer (SMR) include catalyst activity, burner operation, heat loss to atmosphere, furnace operation, heating values (BTU), tube life, shift equilibrium or steam-to-carbon ratio (S-C), and potentially ambient temperatures. A direct monitoring target for the SMR consists of methane slip and outlet temperature, which vary from plant to plant. Methane slip consists of 1.5 to 5 mol% dry and impacts heating values, in terms of BTU. Methane slip is controlled in the reformer by shifting the equilibrium or by manipulating S-C and SMR outlet temperature. Typical S-C is 2 to 3.5, but it can greatly vary from plant to plant.  

If reformer reaction equilibrium was shifted to increase hydrogen make, the result will also be in a reduction in methane content in the reformer heater PSA off gas. As BTU value of off gas decreases, the secondary burners may have to be fired harder. This shift to increase hydrogen make requires an increased amount of purchased natural gas. Increasing methane slip correlates to an increase of heating efficiency and potentially an increase of reformer tube life. Increasing methane slip is achieved by decreasing S-C ratio or decreasing reformer outlet temperature (temperature is in range). To increase hydrogen, make, S-C can be increased, but increasing hydrogen can have an impact on heating values and tube life should be considered. It should be noted that operating at higher reformer temperatures directionally decreases tube life and catalyst life.  

High-Temperature Shift Converter (HTSC) 

The factors that impact energy efficiency in the shift converter are the inlet temperature and catalyst activity. Shift converters should target constant inlet temperatures as temperature swings impact catalyst activity. The exotherm across the shift converter should be monitored, as well as the CO slip. Inlet temperature can be increased to maintain a constant exotherm as catalyst deactivates. The local startup dT (delta T; temperature differential) is 100°F, and a target SOR (start-of-run) temperature is below normal. It is common to have temperature step changes occur every six months. If the plant is short on H2, the inlet temperature can be increased to promote CO conversion. The target shift converter CO slip consists of 1.5 to 3 dry mol% and indicates catalyst deactivation. These process conditions depend on each facility and on SOR conditions. The operational factors that impact this efficiency or CO slip are inlet temperature and S-C ratio. CO slip can be decreased by increasing inlet temperature or increasing S-C ratio. The temperature differentials will indicate catalyst activity. Inlet temperature can also be increased to achieve the target temperature differential, which changes from plant to plant.  

Sulfur Guard  

Front-end sulfur removal has an impact on catalyst efficiency in the hydrogen plants. The sulfur component being removed is H2S and variations of mercaptans. The desulfurizers previously consisted of activated carbon beds at ambient temperatures. Desulfurizers have used several types of media from zinc oxide to carbon beds. The media type can impact efficiency depending on plant design and temperature parameters. Desulfurizers have gone into the direction of a variation of zinc oxide catalyst. Zinc oxide catalyst can increase carbonyl sulfide removal, along with increase efficiency by decreasing heat requirements, with regard to CSO. To increase sulfur removal efficiency, a layer of HDS catalyst can provide increased removal of organic sulfur compounds. 

 

ABIGAIL SUP (Johnson Matthey Inc.) 

Part A: Modern hydrogen plants are around 5 to10% more efficient than those built in the 1990s. These improvements have been achieved through flowsheet modifications such as pre- and post-reforming [for example, a GHR (gas-heated reformer), MTS (medium-temperature shift), pressure swing absorption (PSA), and the use of combustion air preheat, as well as advancements in catalyst technology (e.g., Johnson Matthey’s CATACEL SSR). Modern flowsheets can be very efficient with estimates approaching the theoretical minimum amount of energy required to produce a unit of hydrogen with values of just over 300 BTU/scf of hydrogen (taking credit for steam export) being quoted for some plants. Though for typical hydrogen plants in operation today, energy efficiency values generally are in the range of 350 to 425 BTU/scf of hydrogen.  

The thermal efficiency of a hydrogen plant will depend on the: 

• Quantity of heat recovered from the process gas. 

• Amount of heat recovered from the flue gas. 

• Total heat loss to the environment (function of size, design, condition of reformer). 

• CH4 (methane) slip from the reforming section (impacted by steam-to-carbon (S-C) ratio, catalyst selection and age, reformer balancing, reformer design, material limitations, etc.). 

• CO slip from the WGS (wet gas scrubber section) (catalyst selection and age, configuration: HTS, MTS, HTS+LTS, WHB size); and/or, 

• Downstream purification design (PSA versus CO2 removal/methanation, PSA efficiency). 

For new plant designs, improvements in energy efficiency are generally evaluated against any increases in capex (capital expense) and/or opex (operating expense) required. For example, an MTS flowsheet may achieve a lower CO slip compared to an HTS, but the value of the additional hydrogen may not be able to off-set the additional capex required, such as the cost of a larger waste heat boiler (WHB), a more expensive catalyst, larger catalyst volumes, greater susceptibility to poisoning, and a reduction system.  

The optimal plant efficiency for any plant will vary due to factors such as the: 

• Cost of feed, 

• Cost of fuel, 

• Cost of power, and 

• Value of steam. 

There are also numerous other factors that impact plant design, which then affect the efficiency of the plant. Some of these factors include plant scale, feedstock flexibility, turnaround schedule, payback targets, compression requirements, pressure drop versus vessel cost. 

Part B: A hydrogen plant is designed with a specified arrangement for heat integration which sets the theoretical limit for the plant’s efficiency. The operational factors that provide the largest impact in moving a plant towards this limit include the following: 

• Plant rate: Generally, the plant is less efficient at lower rates due to the relatively higher heat losses and difficulty in maintaining good distribution, which can result in a higher CH4 slip or require a higher S-C. 

• Reformer balance: maintenance of burners, adjustment of air dampers and fuel pressure, condition of tunnels, etc. 

• Ability to properly identify and measure plant bottlenecks; e.g., accuracy of tubewall temperatures (TWTs), sampling analysis, instrumentation, etc. 

• Determining the optimum steam-to-carbon ratio: reducing steam requirement without compromising catalyst life/performance. 

• Minimizing excess oxygen: decreasing fuel usage while still maintaining reliable operation. 

• Optimization of inlet temperatures to the HTS/MTS/LTS bed for minimum CO slip. 

• Monitoring of purification section to prevent poisoning of downstream units, which can significantly impair the ability to run at optimal operating conditions. 

• Startup and shutdown procedures, which can affect catalyst life and performance. 

• Heat loss: refractory condition, insulation, wind shield, etc. 

• Catalyst selection, which affects CH4 slip over life of charge, heat transfer, pressure drop, carbon formation, etc. 

• Routine cleaning and maintenance of heat exchangers. 

When pursuing improvements in energy efficiency, it is important to consider the impact of any proposed changes to ensure continued safe and reliable operation. For example, reducing the steam-to-carbon ratio too far could result in carbon formation on the reforming catalyst or over-reduction of the HTS catalyst. This change to the process could adversely impact the plant’s long-term efficiency or its ability to achieve maximum rates prior to the next scheduled changeout. 

For this reason, daily plant monitoring becomes more critical when pushing a plant towards optimal efficiency because the plant is most efficient when it is running closer to its limits. To safely and reliably maintain operation at optimal conditions, operators need to be able to respond to any changes or deviations that could move the plant outside of designated limits (e.g., TWTs, carbon forming conditions, minimum excess oxygen, etc.)  

One monitoring tool that can help in this area is Daily Thermetric’s CatTracker® technology, which can be placed directly inside a reformer tube with catalyst loaded around it is using Johnson Matthey’s proprietary loading method. CatTrackers® can measure the process gas down the length of the tube, giving operators continuous feedback on the reformer’s condition. This information can enable operators to detect poisoning, carbon formation, or other issues before they significantly affect performance.  

As noted in the list above, catalyst selection also provides an opportunity to relieve operational constraints and improve a plant’s efficiency with minimal capital investment. For example, Johnson Matthey’s promoted reforming catalyst, KATALCO 25-4Q, allows operators to run at lower steam-to-carbon ratios – compared to traditional catalysts – without the risk of carbon formation at more severe conditions. Also, Johnson Matthey’s CATACELSSR can provide a step change in performance by lowering fuel usage for the same capacity, decreasing TWTs allowing the plant to run more aggressively, reducing pressure drop across the reformer, and even allowing an increase in capacity, if needed. 

When searching for efficiency improvement opportunities, it is recommended to seek input from a catalyst vendor, such as Johnson Matthey. Johnson Matthey has the tools, modelling capabilities, and knowhow to help operators identify improvement opportunities, evaluate their relative impact/benefit, and provide recommendations that enable customers to improve their overall efficiency. This expertise includes onsite reformer surveys and data analysis, as well as kinetic modelling of the unit operations to predict the impact of operational changes. For more capital-intensive improvements, such as looking for step changes in capacity and efficiency, Johnson Matthey conducts revamp studies. 

THEISS (Marathon Petroleum Corporation) 

As far as interlocks being tied to fired heaters, at Marathon, we rely heavily on API (American Petroleum Institute) 556 to develop our internal practices. Inside our internal practices, we have two main potential problems we are trying to combat: preventing an explosion due to uncombusted fuel and preventing a tube rupture that can lead to a fire or an explosion. On all of our fired heaters, we have interlocks for low- and high-fuel gas pressure and loss of process flow. For our balanced draft heaters, we include loss of airflow, high furnace pressure, and loss of flame. For forced draft and natural draft, we include all of those, the exception being furnace pressure. 

We do allow shutdown time delays on several components. It is really a matter of looking at each one on an individual basis. We developed a lot of our guidance on time delays based on API 556. The current version of API 556 no longer has that specific guidance, but I think they are looking to try and include it again. Regarding time delays, we allow time delays on fuel gas pressure, firebox pressure, flame detection, process flow, process temperature, and level, if applicable. 

As far as bypassing interlocks, we generally do not allow bypassing of shutdown systems. There are some exceptions. Obviously, performing maintenance is a big one that we look to allow. Some of our facilities actually have an automated report from the DCS (distributed control system). If there is something bypassed on a shutdown system, a notification via the report gets pushed out to the Operations Group. With this report, Operations can tell if something was placed in bypassed that should not have been. We do have certain operating procedures that allow bypassing of ESD (emergency shutdown) systems; but in general, we do not allow any other bypassing. 

We do have some startup and shutdown provisions, particularly on heater interlocks. We have arming logic that initiates once you have a specific condition established. The logic is armed and ready to shut down the heater if we have greater than 50% of the burners lit and achieve a loss of flame signal. There are also turndown scenarios when the unit is turned down. The logic will arm to “all out of all” trip until greater than 50% of the burners are lit. 

 

LÉGARÉ (Andeavor Martinez Refinery) 

Andeavor’s procedures are similar to Marathon’s. We have a series of standards that we created based on API 556 and NFPA (National Fire Protection Association) 85 guidelines for heaters and fired boilers. The standard goes into a lot of detail around levels of instrumentation required on both the process side and the fuel side of the heater. Being in California, we also have requirements for CEM (Continuous Emissions Monitoring). The State outlines the requirements for CEMs as well. Chris McDowell of Andeavor is in the audience today. She can speak to all we need to do to comply with the regulators in the San Francisco Bay Area. The protective systems are also clearly defined in our standard; the recommended instrumentation levels are also there. 

 

 

The next slide shows some of the time delays we outlined in our standard as well. I will not go through each one, but you can see that they are all specified with ranges. In addition, there are also valve travel allowances that vary depending on the size of the valve. The point I do want to make, though, is that a lot of these standards apply to new construction. When you are dealing with a retrofit, the approach can change. The proverbial “it depends” comes into play where you need to look at your physical layout of your heater or furnace and see what you actually can accomplish. So, it is really best to work with your technical personnel or technology suppliers and SMEs (subject matter experts) to figure out what exactly you can achieve to get to the inherently safest solution. 

 

 

 

As far as bypasses, we operate in a way that is very similar to what Jeremy said. I will add that Andeavor also mandates that you have a procedure in place to allow for the bypass. In the event you do not have a procedure, the MOC (management of change) policy kicks in. The requirement for that MOC is that you have a clearly defined mitigation plan to deal with the bypass. The point I do want to make, which is what I have seen, is that that mitigation plan does need to involve the right level of personnel in your organization. You want to engage your instrument engineers, furnace SMEs, and process engineers, not just try to bypass the whole system to come up with the mitigation plan that is achieved at the end of shift, but which may not have the required integrity around the technical review. It is important to make sure that the mitigation plan is done properly. 

In addition, the policy we have also outlines various levels of organizational approvals depending on the amount of time that the system will be bypassed. For example, if you are looking at a three-day bypass, an Operations Superintendent has to buy off on it. If it is a three-week bypass, then the Ops Manager is the one who gets involved. Lastly, what we do with these mitigation plans is table them. We try to keep them in a file so that in the future, if we have to do a similar bypass, we can at least use that mitigation plan as a solid starting point. 

 

 

 

MIKE ADKINS (KP Engineering, LP) 

One of these subsets that you guys touched on that, from a design aspect, KPE would get involved in a lot is purging. Of course, end user always want to try to increase the purge rate to get through that purging process as quickly as possible. Some of the heaters KPE has seen have steam eductors on them, purge air blowers, or just straight steam into the heater. My question is to you guys who are refiners. What do you typically prefer and like to use during that purging sequence? 

 

LÉGARÉ (Andeavor Martinez Refinery) 

We are not typically using steam for purges. We get the fans started and use them to purge the system. 

 

TARIQ MALIK (CITGO Petroleum Corporation) 

Eric, I think you had the time delay reflected on the screen, right? 

 

LÉGARÉ (Andeavor Martinez Refinery) 

Yes. 

 

TARIQ MALIK (CITGO Petroleum Corporation) 

What is the purpose of having a four-second time delay? That was about the maximum in one of them. So, in four seconds, what are you going to accomplish? You cannot react to these alarms. 

 

LÉGARÉ (Andeavor Martinez Refinery)  

No. I think, like Jeremy said earlier, the API standards specify these time delays. I think this is like a legacy system that we still have in our standard.  

 

TARIQ MALIK (CITGO Petroleum Corporation) 

I thought the purpose was to give the board operator a chance to react or do something to make sure they are not spurious or are actually happening. I know some controls are touchy over there. But a four-second delay? You might as well not have a time delay. 

 

LÉGARÉ (Andeavor Martinez Refinery) 

Yes. I do not think we are looking at really four seconds for an operator response. It may just be four seconds to deal with the blip in the instrumentation and get a balancing out of the signal. 

 

TARIQ MALIK (CITGO Petroleum Corporation) 

You have a warning system on this? You have multiple instrumentations – two out of three – at the SIL 3 (safety-instrumented level 3) or SIL 2 level? 

 

LÉGARÉ (Andeavor Martinez Refinery)  

That is correct. 

 

TARIQ MALIK (CITGO Petroleum Corporation) 

Okay. My follow-up question for the panel actually has to do with the heaters. Do your heaters have explosion doors, or have you done away with them, sealed them shut, or thrown them away? 

 

LÉGARÉ (Andeavor Martinez Refinery) 

We still have some of them in our furnaces. 

 

THEISS (Marathon Petroleum Corporation) 

Yes, I am sure we still have some, but probably not all. We have some new construction that may exclude them. 

 

GAMBOA-ARIZPE (CITGO Refining & Chemicals, L.P.) 

Yes, we still have them. 

 

CHRIS STEVES (Norton Engineering Consultants, Inc.) 

Jeremy, you mentioned flame detection. Are you doing that on all of your heaters and all burners individually, or are you looking at trying to just verify if there is any flame in the firebox? How does that work? 

 

THEISS (Marathon Petroleum Corporation) 

 Most of our heaters, especially the new designs, have flame detection. I would say there are very few within Marathon that do not have flame detection within the system. 

 

BILL CATES (Hunt Refining Company) 

Are you doing the flame on the main flame or are you doing a pilot? 

 

THEISS (Marathon Petroleum Corporation) 

In some applications, both. 

 

EREMY THEISS (Marathon Petroleum Corporation) 

Shutdown Interlocks 

Marathon Petroleum Corporation (MPC) standard practices rely heavily on the guidance recommended by API 556. Most of our heater shutdown interlocks are derived from this API Recommended Practice. Our internal practices are intended to prevent a heater explosion due to uncombusted fuel in the firebox or a tube rupture that can lead to an explosion or uncontrolled fire. Specific interlocks that result in a fired heater shutdown, as defined in our standard practice, include low/high fuel gas pressure and loss of process flow. Further guidance on alternate heater configuration is also given. Balanced draft heaters shutdowns include loss of air flow, high furnace pressure, and loss of flame. Although forced draft and natural draft heaters’ interlocks do not include high furnace pressure, they do include loss of flame. 

Shutdown Time Delays 

We do allow certain time delays within the SIS logic. These delays were derived from original API 556 guidance. The length of the time delay is based on acceptable risk tolerance evaluated independently by MPC subject matter experts. MPC has time delays on the following shutdowns: fuel gas pressure, firebox pressure, flame detection, process flow, process temperature, and level (where applicable). Currently, API 556 does not include time delays, but including guidance on time delays is under consideration for the next revision. 

Interlock Bypass Philosophy 

In general, we do not allow bypassing of shutdown interlocks during operations. We do provide guidance to bypass under specific instances of maintenance. During periods of maintenance, alternate monitoring plans are established with Operations to ensure that the intent of the shutdown system is intact. There also may be special circumstances conducted that involve bypassing, but these instances will only be executed under a specified procedure. When possible, we recommend that these special procedures be implemented through operation mode selectors that an operator can select to automate the logic. If deviations to the procedure are required, an MOC is necessary to execute the deviation. 

For periods of startup and shutdown, we develop arming logic for flame detection to prevent unnecessary trips, which can lead to unsafe conditions. For startup conditions, this logic is armed once the first burner detects flame for “all out of all” voting; meaning, a loss of flame on all burners is a vote to trip. Once greater than 50% of the burners detect flame, the logic reverts to the normal shutdown logic, which is typically that less than 50% of the total burners detect flame is a vote to trip. To manage process turndown scenarios, we also have low-fire mode which will revert back to “all out of all” voting logic.  

 

ERIC LÉGARÉ (Andeavor Martinez Refinery) 

Andeavor utilizes a series of internal Engineering Standards to address fired heater instrumentation, control, and protective systems. These standards were developed using the content of API 556 and NFPA 85. Andeavor’s standards define the required and recommended instrumentation for the fuel and process sides of gas fired heaters. The control and protective systems are based on instrumentation mandated by the standard. Examples of required instrumentation include fuel gas pressure; combustion air flow; firebox pressure and temperature; excess oxygen; draft; and, where applicable, flue gas analysis via CEMS for regulatory compliance. 

Protective systems and allowances for overrides, bypasses, and permissives are also defined in the standard to allow for safe and effective design and operation of fired heaters. Andeavor’s standards protect against the accumulation of combustibles in the firebox, overheating of heater tubes, high/low draft and flameout. Instrumentation linked to the protective system [Safety Instrumented System (SIS)] should be independent of the control instrumentation.  

The design of the protective system does allow for time delays with allowable ranges provided as follows: 

PALL Fuel Gas Pressure 1-4 sec  

PAHH Fuel Gas Pressure 1-2 sec  

FALL Comb Air Flow 5-10 sec  

Dropout Doors Fail to Operate 1-2 sec  

PAHH Firebox Pressure 5 sec  

Failure of Stack Damper to Open 1-2 sec  

PALL Pilot Gas Pressure 1-4 sec  

PAHH Pilot Gas Pressure 1-4 sec 

Note that the above information corresponds to new heater designs. For retrofit projects, it is recommended to work with your project team and subject matter experts to implement the design that best satisfies the standard with which you are trying to comply. 

Startup overrides are required in the protective functions of the control system to allow for the startup of fired heaters. The operator’s display will include a notification that the protective function is overridden during startup conditions. These overrides will allow for startup steps such as furnace purges and burner light off. Once the startup conditions are cleared, the protective system is engaged automatically by the DCS. 

Bypasses on input devices and/or protective systems for maintenance, calibration, and testing are permitted in accordance with the site’s operating and emergency response procedures. Sites can manage these bypasses via Operations or Maintenance procedures or MOC, if procedures do not exist. A mitigation plan should be part of the procedure or MOC being followed, and the plan must be communicated to all affected personnel. Note that the plan is only as robust as the quality of review that went into its development. Ensure controls and SIS experts are consulted when developing a mitigation plan. 

The plan should also be kept in a location accessible to the board operator. Management approval of the mitigation plan is required with escalating levels of responsibility defined as a function of the bypass period. As an example, a three-day bypass period requires the approval of an operations superintendent. A three-week bypass period requires the approval of the Operations Manager. Bypassed alarm status should be visible to the board operator. The protective system and/or input device should be put into service immediately following completion of the work. The mitigation plan should be logged for reference in the future. 

 

RICHARD TODD (Norton Engineering Consultants, Inc.) 

All fired heaters should be equipped with safety instrumented systems (SIS) that take the heater to a “safe state” upon detection of a potentially unsafe condition. Recommendations for the implementation of these systems can be found in API-556 “Instrumentation, Control, and Protective Systems for Gas Fired Heaters”. Typically, most heaters should be equipped with instrumentation and logic to remove fuel gas from the heater on the following conditions: 

• Low fuel gas burner pressure, 

• High fuel gas burner pressure, 

• Low process flow, 

• Low combustion air flow (if a heater with FD fans), and 

• Loss of flue gas removal (if a heater with ID fans). 

Other process conditions may require automated heater shutdowns as well. A HAZOP with LOPA should be conducted to determine if additional safeguards are required. 

Time delays are usually built into the logic of the SIS logic solver to prevent spurious trips due to instrument noise. A thorough review of the system and the calculation of process safety times (the time to reach an unsafe state from the start of a process upset) should be conducted to be sure that the chosen time delays do not exceed the process safety time. Multiple instruments with voting logic (i.e., two out of three voting) are also used to improve SIS reliability and to decrease the frequency of spurious trips. 

Instruments that are used in SIS should be equipped with bypasses to allow for maintenance to be conducted with the heater in service. The use of bypasses should be managed with a safety device bypass procedure that requires appropriate reviews and approvals so that the instrument can be bypassed without impacting the safety of the equipment. Typically, bypasses on safety instrumentation should not be utilized during process upsets or due to unusual operating conditions. Startup of heaters may require the high and low fuel gas pressure trips to be bypassed for a short amount of time as burners are being initially lit, but this bypass can be safely managed in a well-designed SIS that may add additional safeguards and that will automatically remove the bypass after a prescribed period of time. 

LOGEROT (Prosys Inc.) 

The question asked about the protocols. But rather than naming the protocols, I think it would just be best to describe what the installation looks like. With wireless transmitters that are in use, you are usually installing a mesh network and you have a Modbus gateway connected to the DCS. You have a separate wireless gateway that is connected to that as well, and the wireless gateway connects to all the transmitters. Each one of the transmitters can actually act as a hub and is able to receive and transmit data with the other transmitters. That way, transmitters can find other transmitters close by and multiple pass-backs to the gateways. So, if a transmitter is out of service for any reason, the other transmitters will basically find pathways around it to get communication back to the wireless hubs. 

Note that in this kind of arrangement, you need to have a robust gateway; because basically, the wireless gateway represents a single point of failure. Therefore, most installations use redundancy there, sometimes triple redundancy, to make sure that communication stays open. 

Where are the main uses of wireless in refineries today? In our experience, they are usually remote areas of the refinery where a signal and power wiring are not easily run out. The big advantages, obviously, are cost savings, conduit wiring, and cable trays. Inside the battery limits of, say, a crude unit, wireless is not as common. Where used, it is usually for auxiliary type of measurements such as corrosion monitors, vibration monitors, additional temperatures, and pressures that are not central to the process. You might also consider wireless technology, as sometimes we have outside operators who have handheld devices that are wirelessly connected back to the control room. For that one-ring device, there will also be a network of gateways available for the wireless device to communicate back to the control room. 

What are the concerns associated with wireless technology? The first real concern is cybersecurity. I mean, everyone is concerned about security these days. Basically, every wireless device and transmitter in your plant represents a potential entry point for intruders. So, you have to be very careful to put in strong security protocols to make sure that intruders will not get into your network. What can happen from an attacker? An attacker can jam your signals. You could lose proprietary data, and – worst of all – an attacker could end up gaining control over part of your process. You really do not want that to happen, which is why security is a big concern when using wireless. Second is overall reliability. Basically, we have been using hardwired signals for decades. Wireless signals are just not as robust in today’s technology as are the hardwired signals. For example, how often do you have to go reset the Wi-Fi in your house? That is an example of when wireless is not as reliable as it could be. 

The last part of the question had to do with whether wireless is acceptable for control or if it is just for monitoring purposes. When I say ‘control’, I am talking about closed-loop control where you have a wireless transmitter communicating to the control room and there is a control action. They then signal out to the final element, usually a valve. It is probably also wireless, but it might be hardwired; but there is at least some component in that closed-loop control that is wireless. Our typical answer is that it is just not used very often for closed-loop control, and it is usually not recommended. One of the problems is battery life, because the transmitters you are using in the field are battery-operated. If you have a very high refresh rate – like, typically, a five-second refresh rate, then your batteries will die too quickly. That is one reason why you would not want to be using wireless for controls. So, generally speaking, it is not recommended or used. However, that is not to say that wireless control will not, sometime in the future, be relatively common. 

  

THEISS (Marathon Petroleum Corporation) 

The chart on the slide is really the internal guidance we use at Marathon. You can see that for what we call Class 0 and Class 1, we do not allow wireless communication, which basically inputs to an SIS (safety instrumented system) or some control point that is detrimental to the process. An example of a detrimental control point would be an FCC (fluid catalytic cracking) reactor/regenerator pressure differential transmitter for which we would not allow wireless control. 

There are a couple of applications in which we allow control. I do not think they are widely used within Marathon; but with some corporate guidance and corporate technologist approval, we can use them wireless for control. These would be considered Class 2. An example would be a tray tower control for a temperature.  

As you move further down the chart to Class 3, you can see an example of where you have the overhead water boot. You have a remote signal that goes into the board, but it relies on the operator to go out and make the move to drain that water boot and start to pump or open up a valve. Class 4 and Class 5 are really for informational purposes. My example for a Class 4 would be a secondary alarm on a tank where you have a primary alarm that is hardwired in and a secondary level or a backup level that could be used remotely. Class 5 would be temperature indication on heat exchangers just to gather data for fouling. 

 

 

JEREMY THEISS (Marathon Petroleum Corporation) 

Technology continues to progress in this field. Since 2011, we have had guidance that allows some usage of wireless instrumentation, but this technology is limited based on application. The table below identifies our stance on certain applications. 

 

CATEGORY	CLASS	APPLICATION	DESCRIPTION	WIRELESS ALLOWED?
Safety	0	Emergency Action	Always Critical: fundamentally necessary for reliable SIS functionality	No
Control	1	Closed-loop regulatory	Often Critical: required for continuous stable unit operation	No
2	Closed-loop supervisory	Usually Non-Critical: not required for continuous stable unit operation	Yes, with approval from corporate technologists and site engineering
3	Open-loop control	Human in Loop: Signal processed by human, human manipulates field device	Yes, with approval from corporate technologists and site engineering
Monitoring	4	Alerting/alarming	Potential short-term operations consequence	Yes, with approval from site engineering
5	Logging	No immediate operations consequence	Yes, with approval from site engineering

 

Examples: 

Class 0: Inputs to a Safety Instrumented System 

Class 1: FCC Reactor/Regenerator pressure differential transmitter (used to manipulate flue gas stack valve) 

Class 2: Tower tray temperature 

Class 3: Water boot high/low level where control or field operator starts/stops a pump or opens/closes valve 

Class 4: Storage tank secondary level alarm 

In most of the approved applications, redundant wireless gateways are required to minimize disruptions to a failed gateway. Other points to consider for determining if wireless is acceptable include the required scan rate of the application, wireless distance limitations, and potential for wireless interference. Guidelines should be made to ensure battery life or alternate power to the wireless device is sustained and has monitoring capabilities. 

 

DARWIN LOGEROT (ProSys Inc.) 

Wireless Protocols 

Rather than naming the protocols in use, it is probably better to describe the installation. Where wireless transmitters are in use, they are often installed in a mesh network similar to cellular towers or a Wi-Fi network with multiple hubs. A Modbus gateway is connected to the DCS and to the wireless gateway that is connected to all the transmitters. Each wireless transmitter acts as an individual hub and is able to receive and transmit data with others. This way, a transmitter can find other transmitters close by and have multiple paths to the wireless gateway. If one or two transmitters are out of service, the remainder will adjust to provide continuous communication.  

In this arrangement, a robust wireless gateway is important. If only one wireless gateway is provided, it can represent a single point of failure, potentially losing view of all instruments using that path to the DCS. Users will typically install redundant gateways to mitigate this. 

So, where are the main uses of wireless technologies in refineries today? The locations are usually remote where the signal and power wiring are not readily available. The big advantage is cost (savings in conduit, wiring, cable trays, power distribution, etc.) and the ability to monitor remote data, such as in a large, spread-out tank farm. 

Inside a refinery process battery limit, use of wireless is not so common. Where it is used, some of the more common wireless applications are in corrosion monitors, vibration monitors, and additional temperature and pressure monitoring on vessels and exchangers (auxiliary to the hardwired temperatures and pressures). 

Another use of wireless technology is for hand-held devices used by field operators. With this arrangement, another mesh network is employed to connect the wandering device to the DCS. The field operator can use the device to monitor operating conditions, execute periodic rounds, and take notes regarding observations. Major DCS manufacturers are offering this technology as an extension of the control system, but the control itself is done with hardwiring; only monitoring and setpoint adjustment are done through wireless. 

What are the concerns associated with wireless technology?  

First and foremost is cyber security. Every wireless device represents a potential entry point for an intruder. Security protocols are better and stronger now than ever, but many potential users are still reluctant to install extensive wireless devices. Security concerns include the possibility of wireless signals being jammed by an attacker, potential loss of proprietary data, or, worst of all, an outside intruder gaining control of part of a process. 

The second concern is reliability. Wireless communication generally is less robust than hard-wired connections.  

A third concern is the data refresh rate and its connection to battery life. For example, a one-minute update rate on the transmitters was tied to a life of about 10 years, whereas an update rate of four seconds reduced that life to two years. The relationship between refresh rate and battery life, of course, impacts how wireless can be used for basic control and impacts wireless maintenance costs. 

So, is wireless communication acceptable for process control?  

Allowing for a slightly wider definition of “wireless control”, it is in widespread use today – the plant radio. For example, the console operator can contact the outside operator: “Hey, go open/close the bypass valve around the control valve that is not working.” But more seriously, purely wireless communication for process control in a refinery is seldom used or recommended, especially in a unit that is tightly connected geographically, such as a crude unit or FCC. Where wireless communication is in use, it is almost exclusively for monitoring only, primarily due the problems outlined above. The closest approach to wireless control is using the handheld devices to adjust setpoints. The control itself is still through hardwire communications from the transmitters to DCS controllers and to the valves. 

That said, there is no reason to expect that as technology improves, the current problems will not be overcome, at least in part. Perhaps future refineries will include widespread use of wireless process control.