Search

KOONTZ (HollyFrontier)

First, a little background: API 941 discusses high temperature hydrogen attack. At low temperatures, less than about 430°F, carbon steel has been used successfully up to 10,000 psi. But with elevated temperatures, the molecular hydrogen will dissociate into atomic hydrogen, which can readily enter and diffuse into the steel. The hydrogen reacts with the carbide in the steel to form methane, which is trapped inside the steel and will eventually form a crack or blister. The addition of carbide stabilizers to the steels – such as chromium, molybdenum, tungsten, and vanadium – can resist the decarburization reaction within the steel.

Beginning in the 1940s, G.A. Nelson collected and published empirical curve data to demonstrate the conditions at which high temperature hydrogen attack is expected for specific metallurgy and operating conditions. These data are known as the Nelson Curves and have been continuously updated over the years to include additional failures due to HTHA. The Nelson chart originally included a curve for carbon-0.5 moly (molybdenum), which was midway between the carbon steel curve and the 0.25 chrome-0.5 moly curve. Ever since 1970, a series of unfavorable service experiences with carbon-0.5 moly steels have reduced confidence in the position of its curve on the chart. Data indicate that how the metal is fabricated does have a strong correlation to its susceptibility to HTHA. In 1990, API removed the carbon-0.5 moly curve completely from the Nelson chart.

HollyFrontier generally uses the carbon steel curve on the Nelson chart to evaluate carbon-half moly steels in its process units. Some process equipment has utilized stainless steel cladding or weld overlay to mitigate the concern with HTHA of the base metal. However, this does not completely eliminate the risk. Hydrogen will still diffuse through the cladding and affect the base metal. The partial pressure of the hydrogen at the base metal will be lower than without the cladding, but it must be carefully evaluated to assure that HTHA would not be expected. Real-world experiences have also demonstrated that no cladding or weld overlay is perfect. It only takes one small imperfection for the hydrogen to find the base metal and attack it. API says that it is not advisable to take credit for the presence of a stainless-steel cladding overlay when selecting the base metal for a new vessel.

HollyFrontier has performed a full review of its process units to identify the risk for HTHA, especially for carbon-0.5 moly steels. The review indicated concerns in the reaction section of some of the units. The general strategy for HollyFrontier for carbon-0.5 moly steel operating above the carbon steel Nelson Curve is to remove it from service. If the carbon-0.5 
moly steel is clad or overlaid with 300 series stainless steel, a Fitness for Service Evaluation is performed to ensure that the equipment is safe for operation.

One example is of a naphtha hydrotreater that was operating slightly above the carbon steel Nelson Curves. Based on a detailed analysis of the cladding, it was deemed acceptable for continued operation. We are, though, in the process of a project to replace that reactor.

In another example, a kerosene hydrotreater operates in two distinct regimes. It makes ULSK (ultra-low sulfur kerosene) part of the time and jet the rest of the time. While operating in the jet regime, the operating point is below the carbon steel Nelson Curve and is, therefore, not a problem. However, when we operated in the high severity mode, it did go above the carbon steel curve. Since the analysis was performed, we no longer run the high severity operation, and there is a project underway to replace that reactor.

The original question asked about reformers. Older semi-regen reformers are especially of concern since they operate at much higher hydrogen partial pressure than does a modern CCR. HollyFrontier operates two semi-regen reformers in El Dorado and Woods Cross. These reformers have unusually high reactor pressures for reducing coking. Of course, there is concern about both the feed effluent and the reactor circuits. The review did not indicate any carbon-0.5 moly steel in the reformers. However, in one case, it was found that at end-of-run conditions, the operating points were touching the 1.0 chrome-0.5 moly curve for a particular vessel. Historical data was used to determine how much time had been spent touching the “curve.” This data was then used to perform a detailed Fitness for Service Evaluation, and the reactor was deemed “fit for service”. Future inspections will utilize advanced nondestructive evaluation (NDE) on these areas to increase our confidence that HTHA is not impacting the integrity of the vessel.

KOONTZ (HollyFrontier Corporation)

High temperature hydrogen attack (HTHA) is discussed extensively in API Recommended Practice 941. At low temperatures (less than ~430ºF), carbon steel has been used successfully at pressures up to 10,000 psi. However, at elevated temperatures, molecular hydrogen will dissociate into atomic hydrogen which can readily enter and diffuse through steel. The hydrogen reacts with the carbide in the steel to form methane, a process termed decarburization. The methane is too large to diffuse out of the steel and eventually the internal pressure is high enough to cause a blister or fissure in the steel. These cracks eventually result in a significant deterioration of mechanical properties which can cause a loss of containment. The addition of carbide stabilizers to the steel such as chromium, molybdenum, tungsten, vanadium 
and titanium resist the decarburization reaction within the steel.

G.A. Nelson collected and published empirical data for API beginning in the 1940s to demonstrate the conditions at which HTHA is expected for specific metallurgy and operating conditions (temperature and hydrogen partial pressure). These data are known as the Nelson curves and have been continually updated over the years to include additional failures due to HTHA. The Nelson chart originally included a curve for C-0.5 Mo (carbon-0.5 moly) that was midway between the CS (carbon steel) curve and the 1.25 Cr (chrome)-0.5 Mo curve. Since1970, a series of unfavorable service experiences with C-0.5 Mo steels has reduced confidence in the position of its curve on the chart. Data indicate that how the metal is fabricated has a strong correlation to its susceptibility to HTHA. In 1990, API removed the C-0.5 Mo curve completely from the Nelson chart. HollyFrontier generally uses the CS curve on the Nelson chart to evaluate C-0.5 Mo steel in its process units.

Some process equipment has utilized stainless steel cladding or weld overlay to mitigate the concern with HTHA of the base metal. However, this does not eliminate the risk completely. Hydrogen will still diffuse through the cladding and affect the base metal. The partial pressure of hydrogen at the base metal will be lower than without the cladding, but it must be carefully evaluated to assure that HTHA would not be expected. Real world experience has also demonstrated that no cladding or weld overlay is perfect; it just takes one small imperfection in the cladding/overlay to allow full hydrogen pressure to impact the base metal. Furthermore, inspection of the base metal is difficult as it is hard to find the damage below the cladding/overlay before enough damage has been done to decrease the mechanical strength. API says that it is not advisable to take credit for the presence of a stainless-steel cladding/overlay when selecting the base metal for a new vessel (i.e., overlay/cladding would be used to resist other corrosion attacks, but not HTHA).

HF has performed a full review of its process units to identify risks for HTHA (especially for C-0.5 Mo steel). The review indicated concerns in the reaction section of some of the units. The general strategy of HF for C-0.5 Mo steel operating above the CS Nelson curve is to remove it from service. If the C-0.5 Mo steel is clad overlaid with 300 series stainless, a fitness-for-service evaluation is performed to ensure that the equipment is safe for operation. In one example, a naphtha hydrotreater is operating slightly above the CS Nelson curve. Based on a detailed analysis of the cladding it was deemed acceptable to continue operation. However, HF is in the process of replacing the reactor and part of the feed/effluent circuit to assure long term reliability.

In another example, a kerosene hydrotreater operates in two distinct regimes when its product switches between jet and ULSK/ULSD (ultra-low sulfur diesel). It is below the CS Nelson curve when producing jet, but it passes above the curve when producing product at 10 ppm sulfur. Since the HTHA analysis has been completed HF no longer runs the unit in high severity mode. A project is underway to restore the ability to operate in high severity mode. The original question asked about reformers. Older semi-regeneration reformers are especially of concern since they operate at much higher hydrogen partial pressures than a modern CCR. HF operates two semi-regeneration reformers in El Dorado and Woods Cross that have unusually high reactor pressures to reduce coking. Of course the main areas of concern are from 
the feed/effluent exchanger(s) through the reactors and furnaces. The review did not indicate any C-0.5 Mo steel in HF reformers. However, in one case it was found that at end-of-run conditions the operating points were “touching” the 1 Cr-0.5 Mo Nelson curve. Historical data was used to determine how much time has been spent “touching” the curve. This data was used to perform a detailed fitness-for-service evaluation. Future inspection will utilize advanced NDE on these areas to increase confidence that HTHA is not impacting mechanical integrity.

PIZZINI (Phillips 66)

In our cyclic units, just based on the air consumption, we can measure the coke each time a reactor comes out for a regen. We are not grabbing samples. Our experience with the cyclics is that if you get up around 8% coke on catalyst, the unit will be pushed a little too hard. You will then need to think about backing it down on feed rate octane, finding a better-quality feed, or possibly increasing the hydrogen/oil ratio.

In one of our studies, we found that 5% to 6% coke on catalyst for a cyclic was just about optimal between the activity and the effect on yields. Every time we switch in a new reactor, we have a bump in moisture, resulting in more C4- until the unit dries out.

Where we could regen faster, we discovered that the optimum cycle was to allow 5% or 6% coke on catalyst. Some of our semi-regen vessels do have catalyst sampling options, but we actually discourage taking those samples frequently. The concern, of course, with a radial reactor is that you never want to uncover the top part because it could allow gases to bypass into the center downpipe.
The octane control on a semi-regen is just a matter of sending samples to the lab and tracking the octane. The purposes of the other components are to figure out the length of your run and manage feed rate and severity to make the run.
On CCRs, we do sample the spent catalyst directly for coke on catalyst. That number is compared with the calculated coke number, which is determined based on the air demand. Again, that number is used to track whether or not we are pushing the unit too hard. If it will be necessary to make less coke, then you could reduce the octane feed or increase the hydrogen-oil ratio. The goal is to maintain a maximum circulation of catalyst.

STEVES (Norton Engineering Consultants, Inc.)

My experience and comments are about a moving-bed reformer or CCR. We track coke in two ways. As Paul mentioned, we take a sample of the catalyst and measure the coke in the lab. We also then calculate, by oxygen balance, the coke in the regenerator that is being combusted. We do not use the coke results to adjust the operating conditions for octane, but we do use it to ensure that the amount of coke entering is within the coke burning capability of that
regenerator.

I have seen coke prediction spreadsheets used to estimate the coke being made on the catalyst when the regenerator section is taken out of service for maintenance or screen cleaning. With that spreadsheet tool, we evaluate multiple cases to determine the impact of feed rate octane and feed quality on the coke levels to ensure that the coke stays below the maximum target levels before the regenerator is returned to service.

PATRICK BULLEN (UOP, A Honeywell Company)

For CCRs, we recommend that you do weekly sampling of the coke in your unit’s local lab to make sure that you are at the right point for your calculations and that you then make adjustments as needed.

PIZZINI (Phillips 66)

On fixed-bed cyclic units, the coke on catalyst is quantified on each reactor when it cycles out for regeneration. This information is mainly used to identify if the unit is being pushed too hard with respect to rates, octane, and H2/oil ratio. Experience has shown that coke on catalyst greater than 8% may indicate a coke-imbalance which will require a rate cut to hold constant octane. A study on one of our cyclic reformers showed that a regeneration frequency that averaged 5% to 6% coke on catalyst provided the optimal yields and activity. For that unit, regenerating too often caused a drop in average C5+ and H2 yields due to the moisture spike that results when each reactor comes back into the process from regeneration. Otherwise, RITS are adjusted “as needed”, based on daily octane results.

On one of our CCRs, we measure actual coke on spent catalyst to validate the air-demand coke calculation. Both readings are used to adjust reactor operation if necessary to make less coke (e.g., feed quality, severity, feed rate, H2/oil ratio) with a goal of maintaining maximum catalyst circulation. Catalyst samples are also used for chloride adjustment. 

On our semi-regeneration reformers, we discourage frequent catalyst samples to test for coke content because over time the loss of catalyst increases slump and causes bypassing at the top of the reactor. This can affect reactor performance and lead to catalyst fluidization and attrition.

SUBHASH SINGHAL (Kuwait National Petroleum Company)

In CCR units, the coke in spent catalyst is analyzed, and regenerator conditions are adjusted to achieve the target coke level in the regenerated catalyst. Constant octane is maintained by adjusting reactor temperature, preferably flat profile.

DUBIN (Axens North America)

We have seen that the maximum allowable iron on catalyst cannot be reduced to a simple number. Historically, about 3,000 wppm is the level at which we see yield start to suffer, but not every wppm of iron has the same impact on the unit. Iron deposited on the surface of the catalyst, usually from corrosion-related byproducts, tends to have less of an impact on the overall performance. Some units have tolerated quite a high level of iron as long as the iron stayed on the surface of the catalyst bead. However, if the iron is able to migrate into the center of the catalyst bead, then the quantity needed to hurt yields could be much less than the 3,000 wppm I noted earlier.

As the iron migrates, we see the larger iron species hindering redispersion during regeneration. The yield penalty seen from this loss of redispersion mirrors that of a decreased metallic function for a given octane. Higher inlet temperatures are needed to meet a given octane, and subsequent loss of reformate is observed.

This slide shows a microprobe analysis of catalyst moving inward from the outer surface of the catalyst. You can see that the alumina is distributed uniformly throughout the catalyst particle. The iron concentrates on the outer surface of the bead, but we see that very little of it has migrated to the core. Under what we would consider low chlorine conditions, or sufficiently moist conditions, the migration of iron is mitigated. However, under higher chlorine content or drier reaction zones, the iron movement can increase and migrate into the center of the bead, leading to reduced yields. We consider a high chlorine environment to be 1.2 wt% to 1.3 wt% chloride on catalyst and dry being less than 10 wppm moisture in the recycle gas. Note that moisture tends to inhibit the migration of iron into the center of the bead.

Getting to the second aspect of the question about the allowable limit of iron on catalyst, we use the same limit on iron for fixed-bed and moving-bed reactors; however, the impact is not quite the same. For moving-bed reactors, the iron is distributed evenly across the catalyst due to the nature of the catalyst circulation, whereas in fixed-beds units, all of the iron deposit is in the first reactor. If the first reactor catalyst is sufficiently poisoned, the reforming reactions will move into the second reactor, and so on. However, it is difficult to play catch-up with the reforming reactions due to the endothermic nature of the reaction. Long term, it is hard to catch up on octane with a poisoned first reactor. For cyclic units, we have seen that the poisoning of the first reactor is essentially the same as a fixed-bed unit. However, with the swing reactor present, you have the ability to change out the first reactor catalyst if the catalyst underperforms
for whatever reason.

PIZZINI (Phillips 66)

It is really just a practical consideration regarding iron. We had a semi-regen unit that had to shut down, drop screen, and reload the first reactor without the benefit of a coke burn. We had removed most of the catalyst and were down to just the dust and fines. We then experienced a dust collector event on the vacuum truck. The dust contained pyrophoric iron since it had not been through a burn. In this case, the vacuum truck was set up with a nitrogen purge. However, the investigation showed that the truck did not have enough nitrogen purge. So just keep in mind that the iron can be pyrophoric, even on a reformer.

PATRICK BULLEN (UOP, A Honeywell Company)

UOP agrees with Axens that the situation is complicated.

DUBIN (Axens North America)

The maximum allowable iron content on catalyst cannot be reduced to a simple number. Historically, we have seen that at approximately 3,000 wppm, the impact on yields from iron starts to become significant. However, every ppm of iron on the catalyst is not equal in its effect on the unit. Often, iron that has deposited on the surface of the catalyst, typically from corrosion-related byproducts, has a limited impact on the overall performance of the catalyst. In several 
instances, we have seen that the catalyst was able to tolerate high levels of iron while still performing at or near the expected level of conversion. 

The quantity of iron needed to cause deterioration in the performance in certain circumstances can be well less than the 3,000 wppm mentioned earlier if the iron migrates to center of the bead from the outer surface. We have experience with a client who saw a decrease in performance at a lower content of iron on the catalyst. However, a significant portion of the iron in this situation had migrated to the center of the bead. As the iron migrates to the center of 
the bead, platinum becomes less accessible as the larger iron molecules – e.g., FeS (iron sulfide)– plug the pores of the catalyst. The yield penalty observed during an iron poisoning situation mirror that of loss of metallic function. Higher average bed inlet temperatures would be required to compensate for the lost metallic function leading to decreased C5+ yields.

In Axens’ view, the iron contamination limit for semi-regeneration and continuously regenerative catalyst are the same, but the long-term operation with high iron catalyst is not. As the iron in the feed is distributed across the continuously regenerated catalyst evenly, the impact can be lessened or even delayed due to the catalyst circulation. For a semi-regeneration unit, the sulfur will accumulate on the first bed, potentially killing the activity of the first bed before the iron would theoretically migrate to the second bed; ‘theoretically’, because not many refiners would be able to maintain reformate production at an acceptable level with the long-term poisoning of the first reactor. Even moderate poisoning of the first bed in a semi-regeneration unit is a concern as it is unrealistic that a refiner can play ‘catch-up’ in subsequent reactors to meet the desired conversion due to the nature of the reforming reactions. Cyclic operations suffer from the same overall concerns as a semi-regeneration unit, but with a swing reactor present, the unit has the ability to replace catalyst on the fly when the yield penalty from poisoning is no longer acceptable.

PIZZINI (Phillips 66)

P66 also experienced a catalyst unloading issue related to pyrophoric iron on a lead reactor which had to be dropped and screened in an unburned condition. The presence of iron led to a vacuum truck dust collector event when vacuuming fines from the lead reactor. This was found to be the result of insufficient N2 purge on the vacuum truck.

KOONTZ (HollyFrontier)

I will start with a bit of review of some reactions, and then I will get into a couple of examples of what we have done at HollyFrontier. Of course, the downside to reforming is that the liquid product has less volume than the feed to the unit due to physical laws inherent to the chemical reactions. First, the high-octane product will have a higher density than the feed; and second, some portion of the feed will be cracked to LPGs and fuel gas in the process.
There are three main classes of reactions in the reformer: dehydrogenation, dehydrocyclization, and isomerization. An example of dehydrogenation is the conversion of methylcyclohexane to toluene. The octane is increased by 46, but the downside is that the liquid volume decreases about 17%. This reaction happens very quickly in the reformer.

The second main reaction is dehydrocyclization, which is the process of a paraffin going to a naphthene and then ultimately to an aromatic. The example shows normal heptane (n-heptane) going to toluene, which has a large octane increase of 120. However, it also has a massive liquid volume decrease of 27%. A few refiners might be tempted to try to avoid this reaction; but unfortunately, it is necessary because you cannot really get the octane of a straight-run naphtha much above about 70 (R+M)/2 without this reaction.

The third major reaction is isomerization, which is the process of changing the carbon structure of the molecule. The example shows normal hexane going to 2,2-dimethylbutane. The octane increase is about 67. The beauty of this reaction is that the liquid volume is essentially unchanged. So, in many respects, this is the preferred reaction. Of course, not all isomerizations have this big of an octane improvement.

Finally, hydrocracking is generally undesired in a reformer. The example on the slide shows normal hexane going to two propane molecules. In this case, the liquid volume has increased significantly; however, the value of the propane is dramatically less than the value of the gasoline that it was when it started. Of course, that is even more so the case today with the high price of crude relative to natural gas.

So specifically, HollyFrontier has several older semi-regen reformers that operate at relatively high pressures to reduce the coking with the goal of achieving an acceptable run-length between regenerations. The reality of the current large differential between crude and natural gas has forced HF to look hard at the operation of these reformers. The price differential between crude and natural gas always drives us to minimize the severity of octane of these reformers, but 
we also have to deal with our day-to-day constraints. The price differential would tend to push us to regenerate more quickly, but that is often not advantageous. The reason is that the outage time, due to the additional regeneration, is usually not enough to make up for the higher liquid yield.

As mentioned earlier, being able to produce E10 gasoline is, of course, of benefit for lowering severity. This is especially valuable for a high-pressure reformer. Some refiners separate aromatics for sale as chemicals. At one of the HollyFrontier sites, we deal with the particular process of extracting benzene. We have reviewed this quite a bit. For example, the conversion of normal hexane or methylcyclopentane to benzene results in a liquid volume
reduction of 32% or 21%, respectively. Initially, you see that that benzene has a much higher price than gasoline, but its higher price is almost wiped away when you consider the reduction in volume that you get for making it.

An alternative for some refiners might be to isomerize that C6 portion of the gasoline, instead of trying to make it into benzene, and then recover the benzene. We are evaluating this option at one of our sites. Of course, I want to mention that C5s should always be routed away from a reformer because they have no chance of making an aromatic. Unfortunately, they do have a good chance of cracking, especially in a high-pressure unit. C5s should generally be directed to an isomerization unit.

At one of our units, we have occasionally reduced the feed rate to a semi-regen reformer, which allows us to significantly increase the volume yield of the overall refinery. At our El Dorado plant, we have taken advantage of the lower feed rate by lowering the pressure on this particular reformer by about 75 psi. As a result, we see our liquid volume yield go up by approximately one percent. The unit does not have a net gas compressor. In order to lower the 
pressure, the net gas must be compressed through the recycle compressor to “get into” the hydrogen header. Since the feed rates are reduced, the hydrogen-hydrocarbon ratio stays about the same for the lower recycle gas rate.

At another plant that also has a fairly high pressure semi-regen reactor, we have a unit with a compressor between the low-pressure separator and the high-pressure separator. This unit was designed to take the net gas off of the high-pressure separator. However, excess hydrotreater 
makeup compression at downstream units now allows the net gas to be taken from the low-pressure separator. This yields a larger hydrogen-hydrocarbon ratio because the recycle compressor does not have to process that net gas, so the pressure in the unit is run a little bit lower. Of course, this depends on being able to have a compressor downstream that can handle lower pressure net gas. One downside is that the net gas purity is lower since it came from the
lower pressure separator.

The original question asked about chloride on the catalyst. HollyFrontier has not attempted to modify the chloride on catalyst, so I do not have any comment about that scenario.

MUEHLBAUER (Valero Energy Corporation – Benicia Refinery)

Similar to what Mark just said, we believe that operating the reformer is all about meeting the octane pool demand while maximizing the liquid volume yield. That has been the same strategy we have used regardless of natural gas prices changes. It has always been the most economic strategy. The exception is that if the refinery is hydrogen-limited, then you will have to factor the hydrogen economics into that decision as well. But more specifically to the chloride on 
catalyst, we have actually found that the chloride drives the acid function on the catalyst in the ring-closing reactions.

Reducing the amount of chloride on catalyst can directionally improve liquid volume yield. In some cases, we have under-chlorided up to about 10% of the manufacturer’s base recommendations. We have done that in about a third of our reformers. On the other two-thirds, we actually operate about 5% lower than the recommendations. Some units operate the last reactor above the manufacturer-recommended chloride levels because we are heater-limited in those particular units. So, if you do not have the furnace, then the chloride will help you.

Even at the 10% level, these changes are very subtle: no more than up to 0.3 vol%. Therefore, it is directionally helpful. If you go too far, you could get catalyst agglomeration issues. We found that it is advisable to calibrate your local lab with either the licenser’s lab or some standard lab that you trust, just to be able to see some of these changes.

KOONTZ (HollyFrontier Corporation)

Naphtha reformers are critical units for U.S. refiners to increase the octane of straight-run naphtha for gasoline blending and to produce GT-BTX® for the chemical industry. However, the downside of reforming is that the product C5+ liquid volume is significantly lower than that of the feed. This results from two consequences inherent to the chemical reactions. The high-octane product is higher density than the feed, and some of the feed is cracked to LPG and fuel gas.

There are three main classes of reactions that increase the octane of the product: dehydrogenation, dehydrocyclization, and isomerization. Dehydrogenation is primarily the process of a naphthene producing an aromatic and hydrogen. As an example, methylcyclohexane could be converted to toluene and hydrogen. This process would increase the RON (research octane number) of the MCH (methylcyclohexane) by ~46, but it would also reduce its liquid volume by ~17%. This reaction happens quickly and is primarily catalyzed by the metal function of the catalyst.

Dehydrocylization is the process during which a paraffin produces a naphthene and hydrogen. The naphthene will generally proceed to an aromatic via dehydrogenation. As an example, n-heptane could be converted to toluene and hydrogen. This process would increase the RON of the nC7 (normal heptane) by ~120, but it would also reduce its liquid volume by ~27%. This reaction is the slowest of the primary reactions in a reformer. The ring-forming is primarily 
catalyzed by the acid function of the catalyst and the dehydrogenation is due to the metal function. It results in a large reduction of liquid volume; but without it, the octane of a typical full-range naphtha can get no higher than 60 to 70.

Isomerization is the process of a paraffin or a naphthene rearranging its carbon structure; for example, n-hexane (normal hexane; nC6) could be converted to 2,2-dimethylbutane. This process would increase the RON of nC6 by ~67 with essentially no change in liquid volume. The reaction rate falls somewhere between dehydrogenation and dehydrocyclization and is primarily catalyzed by the acid function of the catalyst. For a fuel refiner this is generally the ideal 
reaction; however, most other isomerizations do not result in such a large RON improvement.

Hydrocracking is the fourth main class of reactions in a reformer and is generally undesired. As an example, n-hexane plus hydrogen could be converted to two propane molecules. The reaction is favored by high temperature and high pressure and is primarily catalyzed by the acid function of the catalyst. This process would increase the liquid volume by ~34%, but the value of propane is tied more closely to that of natural gas than to crude oil. With 
the price of crude near historical highs and the price of natural gas near historical lows, the volume increase is nowhere near enough to make up for the decrease in value.
Older semi-regeneration reformers were designed for higher pressures to reduce coking and enable the unit to run for an acceptable time between outages to regenerate. The impact of the large differential between crude and natural gas more greatly affects a reformer operating at 
higher reactor pressures due to increased hydrocracking. This would generally drive the daily optimization to minimize reformer octane (i.e., severity or temperature). It would also push a refiner to regenerate catalyst in semi-regeneration reformers more often to avoid the EOR (end-of-run) conditions when reformer temperatures and hydrocracking are higher. However, the increased yield from the regeneration must be balanced with the lost opportunity due to the outage.

Producing sub-grade gasoline for blending with 10% ethanol can be a significant benefit for a refiner having a reformer with low liquid product yield. The lower octane of the sub-grade (~3 points lower) allows the refiner to run the reformer at lower severity or bypass sweet naphtha around the reformer directly to gasoline.

Some refiners separate aromatics from the reformate for sale to the chemical industry. One particular aromatic that HF (Editor’s note: HF stands for HollyFrontier in this response) has evaluated is benzene. The conversion of n-hexane or methylcyclopentane to benzene results in a liquid volume reduction of 32% or 21% respectively. The increased price of benzene over gasoline primarily reflects this reality. An alternative for a refiner would be to send the C6 
portion of the naphtha to an isomerization unit to avoid the volume loss inherent to a reformer.

C5s should always be routed away from a reformer and generally to an isomerization unit. C5s will not form an aromatic, but they could hydrocrack to LPG and fuel gas in a reformer (especially in a high-pressure unit).

Reformer yield can be improved by lowering the feed rate, especially for a semi-regenerator. Reducing the feed rate can be accomplished by fractionation to route light naphtha to isomerization and heavy naphtha to distillate, or via the crude slate to reduce the total straight-run naphtha. The HF ElD (Holly Frontier, El Dorado, KS) reformer takes advantage of this at times and lowers the reactor pressure by as much as 75 psi. This increases the product C5+ liquid 
yield by as much as 1 LV%. The unit does not have a net gas compressor due to the high separator pressure. At low rates the pressure is reduced and the recycle compressor also compresses the net gas (lowers recycle rate). However, due to the lower feed rate, the H2: HC 
ratio and coking rate both remain about the same.

HF Cheyenne has reduced its semi-regeneration reactor pressure in a different way. The unit has a low pressure and a high-pressure separator with a compressor in between. The unit was designed to take the net gas off of the HPS (high pressure separator). However, excess hydrotreater makeup compression allows net gas to be taken from the LPS (low pressure separator) and yields a larger H2: HC ratio. The reactor pressure can now be run lower (higher liquid yield) and achieve the same coking rate due to the higher H2: HC. One downside is that the net gas is lower purity having originated from the LPS.

HF has not attempted to modify the chloride on catalyst in its reformers to alter the acid function of the catalyst. It is recommended that you consult your catalyst supplier before attempting a change such as this.

SUBHASH SINGHAL (Kuwait National Petroleum Company)

Chloride on catalyst is maintained at a desired level (1 wt% to 1.2 wt %) for optimized catalyst performance and to ensure that balance of metal and acid activity. At KNPC, we never alter chloride on catalyst to shift yields to gas/LPG.

KOONTZ (HollyFrontier)

HollyFrontier operates five semi-regen reformers and two CCRs. There has not been a specific effort to replace catalyst in order to take advantage of the higher liquid product yield that is possible with newer catalyst. However, the spread between natural gas and liquids certainly impacts the decision when looking to upgrade. For two of the semi-regen units, the most recent catalyst replacements were installed after approximately 20 regen cycles, which included multiple dumping and screening events. HollyFrontier would generally only consider replacement during a turnaround, which is about every five years, because it would not be worth shutting down the unit.

Tracking liquid product yield as a function of octane over time is the most important factor used to justify new catalyst. Of course, today, with the large spread between gasoline and fuel gas, that breakeven point would probably be sooner, especially on high pressure units. Replacing relatively new catalyst with more advanced ones is also considered; however, we would be cautious about doing that because of the difficulty of comparing pilot plant data with operating data from our plants.

DUBIN (Axens North America)

We consider typical replacement levels to be six to eight years for moving-bed, eight to 10 years for fixed-bed, and three to seven years for cyclic units. Those ranges are based on typical regeneration frequencies and good quality regeneration, as well as proper hydrotreatment of the feed to the reformer. We know that surface area is lost with each regeneration; so, whatever your regeneration, you are going to lose yields relative to your initial yields. But as Mark 
mentioned, replacement on lost yields relative to initial yields should not really be the only basis.

The most current catalysts on the market may offer improved economics. Due to a changing feed composition or product severity required, a newer generation catalyst may offer an economic incentive over your current load. A detailed evaluation should be conducted to determine the economic drivers at your particular site, in terms of hydrogen, its need, and its value. The same is true for the liquid products. This will help you determine if it makes economic sense to send the existing load to reclamation and purchase a new load of catalyst.

On the valuation of natural gas, we have seen that the reformer hydrogen is generally being devalued. Natural gas prices are so low that new steam methane reforming units are being brought on stream both for ULS (ultra-low sulfur) fuels production, as well as general upgrading. The hydrogen from hydrogen plants is preferred in both quantity and quality. In terms of total quantity available, you can build your hydrogen plant for whatever your need is going forward. The quality of the hydrogen from the hydrogen plants is near pure leading to improved partial pressures and potentially reduced equipment sizes.

RON MARRELLI (HollyFrontier)

Both of you mentioned a specific time between regenerations. Is time the main factor in determining frequency of regeneration or catalyst surface area? What is the best way to really determine when the catalyst needs to be changed?

DUBIN (Axens North America)

For a moving-bed, Axens recommends tracking surface area. In general, you should see that surface area will trend with yields. Tracking surface area can be done quite easily for moving-bed applications. For fixed-beds, it is not as easy on a continuous basis; but during any regeneration or dump-and-screen, you can track and note the surface area. Ultimately, it will be the yields that justify changing out the catalyst.

KOONTZ (HollyFrontier)

I agree with his statement.

MUEHLBAUER (Valero Energy Corporation – Benicia Refinery)

UOP’s experience with units of similar age to ours has indicated that the key to maintaining long catalyst life and good performance is ensuring good platinum dispersion. If we can maintain good platinum dispersion, then product yields from the CCR Platformer unit, reformate production, and hydrogen production will remain flat over time. In some of the pressurized regenerators, CycleMax regenerators, and even atmospheric units, we have had up to a thousand regeneration cycles with no real loss and C5 plus yield or total aromatics yield. It is not as simple. Like the question before, there is not really one simple answer as there was with iron. If you want to talk more about it, feel free to stop by at the UOP suite. We can have a discussion.

KOONTZ (HollyFrontier Corporation)

HollyFrontier operates five semi-regeneration reformers and two CCRs. There has not yet been a specific effort to replace catalyst to take advantage of higher liquid product yield; however, this certainly impacts the decision process when looking to upgrade to a better catalyst. For two of the semi-regeneration units, the most recent catalyst replacements were after approximately 20 regeneration cycles (also included multiple dump and screen events). HF would generally only consider a catalyst replacement during a turnaround (about every five years). Tracking liquid product yield as a function of octane over time is the most important factor used to justify new catalyst. With today’s large price spread between gasoline and fuel gas, the break-even point to replace catalyst is certainly sooner than in the past. Replacing relatively new catalyst with a more advanced catalyst is also considered. However, care must be taken to assure that a yield improvement prediction for a new catalyst is not due to different feed properties or operating conditions.

DUBIN (Axens North America)

Typical replacement levels are six to eight years for moving-bed applications, eight to 10 years for semi-regeneration units, and three to seven years for cyclic. The ranges are based on typical regeneration frequencies, good quality regenerations, as well as proper hydrotreatment of the reformer feed. A reduced catalyst life could be expected with increasing regeneration frequency as surface area is lost on the catalyst with each regeneration leading to a reduction in 
reformate yield relative to the initial yield. 

Changing the reformer catalyst solely because the yields have decreased relative to the start of the run should not be the only basis for replacement. Replacing the existing catalyst due to advances in reforming catalyst technology, changes in unit severity, or changes in the unit feed composition, may still make economic sense even if the current catalyst load still has life left. A new load of catalyst could bring improvements in any number of economic drivers, 
reformate yield, hydrogen yield, increased cycle length, etc. If the combined improvements in the economics of the reformer provide a desirable rate of return, then it makes sense to send the existing load to reclamation. Axens has developed a tool with just this idea in mind. Looking at the key economic drivers in the reformer to help refiners understand whether a newer catalyst load will offer a greater profit than what their current operation is providing.

However, it must be noted that the very low prices of natural gas are leading to a de-valuing of the reformer hydrogen production. New steam methane reformers (SMR) are being built at a number of sites to help meet both ULS fuel production, as well as upgraders. The increased hydrogen production by SMR devalues the reformer hydrogen not just because of the increased quantity often available, but also because of the increased quality. The high purity 
hydrogen available by SMR can significantly improve hydrogen partial pressure for high pressure units, reducing costs, and further decreasing the value of reformer hydrogen production.

DUBIN (Axens North America)

The most current catalysts on the market are multi-promoted using a number of different promoters beyond the base platinum-rhenium or platinum-tin. It is not a one-size-fits-all market, so there are tailored designs for different needs. For CCRs, the current drive is for improved yields. Units are often octane-long. The goal is now to maximize barrels as best as we can. We can then try to contrast to the early 2000s where gasoline was high in demand and high activity.
Moving-bed catalysts were desired to keep coke make down or within the design parameters of the existing unit. For fixed-bed units, we are now seeing the drive towards increased stability. Refiners are trying to push the time between regenerations as long as possible. The economic incentive to stay onstream is bigger than the extra octane barrels or liquid volume product derived from a different type of catalyst.

JOE ZMICH (UOP, A Honeywell Company)

UOP is always looking to improve the catalyst performance; not only activity, meaning lower reactor inlet temperatures for desired octane, but also higher reformate production. In the North American market, it becomes a little more complicated with reformers intending to operate at much lower octane. As octane is decreased, the paraffin conversion in the reactor system will go down. You will then be relying on differentiation of naphthene, specifically C5-ring naphthene conversion. It is more difficult to differentiate catalysts at low paraffin conversion.

DUBIN (Axens North America)

The newest generation of reforming catalysts is multi-promoted using a number of different metals beyond the base platinum and rhenium or tin. These promoters are being used to tailor the operation to fit exactly what the refiner needs, as opposed to a one size fits all market.

 The current market drive for continuously regenerated catalyst is for increased yields. As the overall gasoline market stagnates, refiners have been looking to recover as many barrels as 
possible from their units while operating at reduced severity. Many refiners are octane long, reducing the need for high activity catalysts. This is a big change from the early 2000s when refiners were looking to maximize gasoline octane out of their continuously regenerated reformers, requiring high activity catalyst to keep coke within the design range on their unit. 

For the semi-regeneration market, the driver has been towards increased stability. 
Refiners are trying to maximize their time on stream, using the reduced coke make in the current generation of semi-regenerative catalysts to stretch the time between regenerations. The savings, obtained by staying on stream, are often a bigger driver than extra octane barrels or hydrogen. 

SUBHASH SINGHAL (Kuwait National Petroleum Company)

There are continuous advancements in catalyst systems for increased cycle length and product selectivity. Performance improvement – in terms of high octane, high H2 rich gas, high LPG, and less fuel gas – are important to researchers.

STEVES (Norton Engineering Consultants, Inc.)

Flexible thermocouples, as I understand, are now a part of the standard design for regenerators of moving-bed units. At a refinery in which I worked; we replaced the original slider thermocouples with multipoint thermocouples. In this particular case, there were three sets of thermocouples installed on the outside of the regeneration screen in order to obtain temperature readings in the bed. Each thermowell contained eight thermocouples spaced at one- foot intervals down a length of the screen. When specifying these types of thermocouples, it is important to determine if a representative cross-section of temperatures is being collected for the bed.] You can easily obtain many temperature readings. As mentioned in my answer to Question 31, proper temperature monitoring can be critical to ensure that the burn is progressing properly and that there are no temperature excursions. In my mind, it is better to have too many temperatures than not enough and potentially be running blind.

DUBIN (Axens North America)

What Chris said is true for Axens. The standard design would include flexible thermocouples. They do offer the ideal ability to locate the temperature measurement points exactly where you desire them. Also, the number of points can be significantly increased with a minimum number of nozzle projections off the side of your regenerator.

PATRICK BULLEN (UOP, A Honeywell Company)

Our standard offering for the CycleMax is to use the flexible thermocouples. Most units are either Gayesco or Daily Thermetrics. Daily Thermetrics is very popular right now. The main issue is replacement in old atmospheric units. It is difficult to use the Gayesco or Daily Thermetrics systems due to space considerations and the logistics of getting the thermocouples into the regenerator area.

R.K. (RICK) GRUBB (Chevron Products Company)

Do you have much experience with thermocouples in a fixed-bed reformer?

DUBIN (Axens North America)

At Axens, we have completed revamps of fixed-bed reactors with the addition of thermocouples. On older units, avoiding welding on the shell is often desired. Ideally, it is preferable for you to make use of an existing nozzle.