Search

FERNANDEZ (Jacobs Consultancy Group)

This question is quite related to the previous one: the same type of system—the platinum on aluminum-based catalyst. The problem now comes from the oxygen ingress coming in with makeup hydrogen, with the culprit typically being CO and CO2. The problem is slightly different here. The amount of oxygen that you bring is not as high, obviously, because of lower mass flow of hydrogen in comparison to the feed; but it’s much more pernicious. It’s typically a problem when you have nothing and there’s not much you can do in the isomerization unit. So what you really have to look at here is the alternatives to solving the problem. Obviously, methanation is an alternative; but, as always, we would always recommend looking at where the problem is and seeing if you can address it there, rather than spending capital on additional equipment.

It’s important that we look at what may be the sources of CO and CO2 in the refinery hydrogen systems. Everyone traditionally thinks that CO and CO2 are formed in the steam reformer; that is true. Older type hydrogen plants have solvent extractions and a methanation unit to eliminate CO and CO2 in the hydrogen. So there you will see, at most, 10ppm CO plus CO2.

Instead of having this methanation system, modern hydrogen plants have PSA units that are pretty good at removing both CO and CO2; but there’s always a balance between the purity that you get in the PSA unit versus the recovery. So in some PSA units where the units are being pushed either in terms of capacity or on hydrogen recovery, you do have the potential of having CO leakage coming out into your hydrogen system.

A third source—and it’s sometimes forgotten about—is that there is CO coming out with the net gas from reforming units. Some of it might be residue from regeneration operations, but some of the CO actually comes by formation of CO in the unit, particularly in what we would call wet units or units that have a high moisture content in the recycle gas. In those units, CO is formed on a reverse water shift reaction. So this is an area that you have to look at because there will be a source of CO that you may be able to manage.

And lastly, we see there are many refiners today that try to recover hydrogen to the max from all their other services, including FCC units, Coker off gas, and off gas from the hydrotreaters. Those streams tend to be contaminated with all kinds of contaminants, including CO and CO2.

Regarding the solutions to the problem, unfortunately it’s not a very easy problem to track down. The reason is that many refineries have several sources of hydrogen coming together and mixing in a big hydrogen header system. Trying to find out where the CO is coming from is pretty difficult.

I recall that years ago, we were helping a customer that had a Penex unit in which they were having a very high rate of deactivation of the catalyst. This refinery definitely had what we would call a wet reformer. They were injecting large amounts of water and chlorides to try to keep catalyst activity. You definitely knew, and could measure, the amount of CO that was coming in that hydrogen stream.

Unfortunately, that was not the only hydrogen stream. When we tried to correlate the rate of the activation of the catalyst with the amount of CO that was in the hydrogen, the rate of deactivation was higher. One idea was to put in a methanation unit, but you’re only going to resolve part of the problem. And really, the problem is in the bad operation of the reformer. So methanation didn’t seem to be a very good solution there.

We’ve also talked with other refiners that actually had a major problem with CO and CO2, but these guys are talking in the range of 100 to 1000 ppm contaminants of the makeup hydrogen. They know very well what the source is. This refinery recovers hydrogen from FCC, cokers, and hydrotreaters, and they do that through a cryogenic unit. So for them, there really is no alternative. They had to put in a methanation unit and they’re very happy with it. Once they put in that methanation unit, their operation became very stable. They completely eliminated the catalyst deactivation that was being caused by their makeup gas.

In summary: If you cannot prevent or remove it, you’re going to have to methanate it. Methanation is the ultimate solution. It works. There are simple systems. It’s well known how they operate. It’s a relatively low capital cost solution, probably three-quarters to a million-dollar installation for a small hydrogen stream. We would recommend that before anyone commits to the methanation unit, they do two things: One, make sure you know the sources of the CO and confirm that you cannot solve them at the source, which is obviously always the cheapest solution. Secondly, make sure that it is the only problem with contamination in your isomerization unit, because you may be calculating a payback on that methanation unit based on large catalyst deactivation which might not be all caused by the CO.

HAZLE (NPRA)

Jay.

ROSS (Axens)

Yes, thank you, Pedro. We agree with all those points, but I would also like to take a minor issue with the CO contention. We naturally limit all oxygen. That is going to the Isom process because of the potential for H2O formation. This occurs with alcohol as a direct formation of water and with the CO, CO2 in the possibility of methanation. That’s typically limited to about 10 ppm in the hydrogen gas. However, these units do operate at very low temperatures and the conditions in the Isom—in our view—significantly promote methanation. A reversible poison by carbonyl formation with the platinum is probably a more significant concern, and thereby affects the metal activity of aromatic saturation preventing coking, etc. We’ll go to the next slide.

In one particular unit, it was a little different than the unit that Pedro showed. This is a two-reactor traditional chlorinated aluminum Isom unit, but with the lead benzene reactor, because there was a relatively high benzene in the unit. Here, they had a problem on a methanator on the hydrogen plant, so a fairly large concentration of CO passed through the Isom unit as a wave. The benzene reactor experienced a sharp increase in delta P and the exotherm from the saturation of the benzene was temporarily inhibited almost completely as the CO passed through.

And then there were some thermal effects causing the lag to be a little different, as you might expect; but effectively, the benzene was pushed downstream and had to be picked up, in this case, by the lag reactor. Over the course of about four hours, everything came back to normal. That’s not necessarily proof positive that there was no catalyst damage; but in our view, there was limited permanent catalyst damage due to methanation and water formation, and, rather principally in this case with CO, a temporary poison of the platinum metal effects.

HAZLE (NPRA)

Clever.

HATZEL (Tesoro)

Just a quick history of the Mandan methanator: We had a C5/C6 Isom unit installed in early 1981. After multiple catalyst poisoning episodes and catalysts regeneration or replacement, the methanator was installed in 1985 and did seem to work. Looking back to the files, we see that a CO-related poisoning pretty much ceased. That doesn’t mean that we didn’t find other ways to kill the catalyst in the years after that. We did, with alcohols and other things that got into the feed. As late as 1995, they partially bypassed the methanator inadvertently. They went out and found the bypass valve cracked open, but it was too late. Within just a few days, the catalysts had been poisoned. So that has kind of been our experience.

When talking to our Isom experts, I think we give pretty serious consideration, especially with the advent of some of the benzene regulations, to having methanators on future Isom units.

HAZLE (NPRA)

Those are the panel responses. The last question is on isomerization. There is one right here. Questions? Comments? I think I saw a hand over here.

OYEKAN (Marathon Petroleum)

Soni Oyekan, Marathon. I just have questions for the panel. In the case of the example for Mandan: Was the hydrogen recycled or once-through to Isom unit initially? Secondly, for Jay: In terms of the levels of concentration of CO and CO2 that you’ve suggested, I believe the upper limit is 10. We might have temporary relief after the carbonyl is formed initially. Should we worry if it’s above that or basically be satisfied that we’ll have this passivation of this carbonyl? So, there are two questions. One is about the effect of this carbonyl with hydrogen once-through units, and the second one will be max carbonyl and CO and CO2 level that we’ll be looking at in these streams.

HAZLE (NPRA)

Let’s start with Jay.

ROSS (Axens)

Wait until Soni gets his card out there. As with other, we traditionally recommend less than 10 ppm CO plus CO2. And as with all these things, even with the dryers, you recommend as dry as you can, but you recognize that there will, in fact, be some breakthrough and slow degradation. The example that I showed was a rather extreme one, where with the methanator failure, they had 3 or 4% CO. So it made a wonderful lesson and example, but it was perhaps a bit extreme.

HAZLE (NPRA)

Clever, once-through or recycle?

HATZEL (Tesoro)

You’re testing my memory a little bit. We actually shut the Isom unit down in 2000, but I believe that was once-through. The source of the hydrogen unit was from a cyclic reformer.

HAZLE (NPRA)

Other questions?

PROOPS (Solomon Associates)

Kevin Proops with Solomon. I was going to ask Clever where his hydrogen came from, because my experience has been with dedicated hydrogen from a cyclic reformer, not significant deactivation due to CO. Pedro gave a good overview of managing your hydrogen system; but I guess I would suggest that if you’ve got a complex hydro system, especially with hydrogen plants, you should consider taking the lowest CO sweet hydrogen stream direct to the Isom unit rather than trying to deal with any problems after you’ve blended your hydrogen streams together.

Refiners also sometimes have issues with having a high enough pressure hydrogen stream when you don’t really want to end up with another compressor if you don’t have to. So it’s nice if you have the pressure. One of the disadvantages a lot of times in older reformers is their higher pressure, because the yields [tend] to be worse. Maybe one of the advantages is that you can then take the hydrogen stream off of that unit when that unit is available directly to an Isom unit.

HAZLE (NPRA)

Other questions or comments? There would be another one back here.

DETRICK (UOP)

Kurt Detrick, UOP. The example Jay showed us is a good example of what we would expect with a high breakthrough of CO: a severe depression of the platinum activity. And certainly, once you take away the platinum activity, the catalyst does neither benzene saturation nor isomerization. Eventually, though, that effect will go away.

However, our experience in many commercial situations has been that the CO does also act as a permanent poison, probably through a methanation reaction in the reactor leading to water formation. We’ve seen this with low level CO, constant low level CO, and steady deactivation. Once the CO is removed, that deactivation goes away.

There is clear commercial evidence of permanent damage to catalysts due to CO, and with CO2 as well, although there does seem to be a difference. The CO does seem to be a little more potent poison. I think it’s because we’ve seen evidence that CO2 can actually break through the reactor. It doesn’t all get taken up in the reactor. We’ve seen a couple of suggestions of that. The data that we’ve seen is also consistent in that maybe the CO2 isn’t as strongly held by the catalyst or picked up as completely. I think you can pretty much figure that all of the CO is going to go to water. It may not be right on the top of the bed, but it’s going to all lay down on that bed and it’s all going to make water.

And again, Kevin’s commented that historically, the older reformers often had much lower levels of CO. Just the operating conditions led to lower CO levels and those have been very reliable sources of hydrogen for Isom units for years. Some of the newer units are operating at conditions that are going to generate 10 ppm CO—and maybe even a little bit more at times— and those are the ones that are concerning in the current situation.

HAZLE (NPRA)

Kurt.

DETRICK (UOP)

Sorry. Jay.

ROSS (Axens)

I think it’s clearly been reported previously that the CO will methanate more readily than the CO2. I believe it’s the first one to go. And certainly, we recognize that even though we are at relatively or very low temperature comparatively, there is certainly some methanation occurring. It’s just that others, in the past, have tended to treat it as quantitative. I just wanted to make the point that it wasn’t perhaps quantitative.

OYEKAN (Marathon Petroleum)

Soni Oyekan. This is just a question for most of the people in here. Is anyone working with dryers that have mixed absorbance in them that want to help you with moving to CO and others to dry down the water?

HAZLE (NPRA)

Panel? Pedro.

FERNANDEZ (Jacobs Consultancy Group)

One of the ways that people have said dryers in makeup hydrogen service can be made more productive is by adding multiple types of absorbance; but typically you can get something that can remove water and CO2. I don’t think that there is any good experience with absorbance that can effectively remove the CO. I would look for anyone that has done that.

HAZLE (NPRA)

Anyone else on the panel? Clever.

HATZEL (Tesoro)

Again, it’s been a while. I seem to remember, though, that we had some on the liquid feed belt system—not on the hydrogen coming in—and that we had some mixed absorbance on that same unit: sulfur on the top and water on the bottom.

HAZLE

(NPRA) Anyone in the audience? Fred.

HILL

Kurt, on the effect of the CO and CO2 going through the unit, did you see any difference between a once-through and a recycle unit?

DETRICK (UOP)

Kurt Detrick, UOP. I think with the CO2, possibly. Again, to get back with Jay, certainly I agree that the conditions are pretty mild. And with CO2 anyway, quantitative methanation is probably not happening. It doesn’t all get methanated, at least not the first time through. But in a recycle gas unit, a lot of it gets another chance to go back through. So I think you’ll see, with the CO2, that it could be more damaging to a recycle gas unit than it would be to a hydrogen once-through unit. Like I said, we did see some evidence of CO2 coming out of a hydrogen once-through reactor, even the second reactor. We’ve never seen that with CO. It may not be 100% of the CO gets converted to water, but it’s pretty close. So I don’t think there’s going to be much difference there between a recycle gas unit and a hydrogen once-through unit. In theory, that follows, too, because CO is much more easily methanated than CO2.

HAZLE (NPRA)

Other questions? Alright. I owe you a break. Emerson has bought you coffee this morning. It’s waiting for you in the hall. We are going to resume at about 10:40 and we will start with Naphtha Hydrotreating.

BHARGAVA (KBC Advanced Technologies, Inc.) 

In summary, it is all about turbine blade deposition in the expander on the flue gas. First, I will talk about the causes, catalyst loading being the number one and the only cause for most of the turbine blade deposition. The catalyst loading on the flue gas inlet is what determines the amount of deposition. From a mechanical standpoint, the blade deposition increases as the performance of a third-stage separator goes down – for whatever reason – or the loss in regenerator cyclone efficiencies. From a catalyst perspective, if you are trying to put in a new catalyst, that catalyst will have different attrition properties. Evidence of mechanical damage in the unit that results in more catalyst fines will have the same effect. Increasing fresh catalyst additions will produce the same result, because fresh catalyst contains more fines. Also, as you start processing resid or heavy metal gasoil and your sodium and vanadium levels go up, the catalyst will start to get stickier and will create eutectic mixtures at certain temperatures, resulting in more catalyst sticking on the blades. From an operational standpoint, if you increase flue gas rates when you increase air rates, the amount of catalyst loss through the cyclones will increase and result in more deposition. 

At some sites, we found additional steam being injected upstream of the expander, a practice that is done for different reasons. One reason could be to maintain temperature and pressure on the inlet of the expander in order to keep up the efficiencies, but that is not a recommended solution because it actually makes the situation worse on the turbo expander. So, those were the causes. 

What is mitigation? It is important to understand the cause of the depositions. You want to make sure you analyze the deposits. If you do not have that luxury, then track the 20-microns or less size range, the catalyst’s physical properties, and monitor the e-cat metals. From a mechanical perspective, you want to do a routine monitoring of the bearing temperatures and vibration, and then check the process temperatures and pressures to identify if you are having a problem. One option to help reduce the turbine blade deposition is to run close to the design temperature and pressure. 

Consequences: These are obvious consequences. You lose power generation efficiency, but more serious are excessive depositions. If you have deposition on the blade and have uneven breaking off the deposits, the expander can become unbalanced, which goes back to the previous point about monitoring vibrations for early detection. How do people remove the deposits? First, you need to monitor the deposits via a viewport using a strobe light. This light will allow you to quickly detect the buildup of deposits. This monitoring is more important because if the deposit just builds up, it will be easier to remove the deposits. You can even do that with an online riser walnut shell cleaning. Some people have resorted to thermal cycling by bypassing the expander. Again, that is also a thermal shock to the unit, so we do not recommend it. Finally, if the deposits have been there for a long time, you do not have any choice but to shut down the expander. 

FEDERSPIEL (W.R. Grace & Co.) 

Like what was previously mentioned, some of the hard deposits that may form on the expander blades might consist of fine catalyst particles but may be enriched with other contaminants like sodium, potassium, calcium or chlorides, vanadium, iron, and other trace elements. The theory is that they might form a eutectic that drops out on areas of high velocity and pressure drop. It is a little counterintuitive; but the expander, of course, is one such place. Further, if you build up the deposit to a sufficient thickness where it starts to cause friction on the machine, those deposits can then be enriched with the expander metallurgy. And if they get hot enough, those deposits could then sinter and be very hard to remove. 

So, one of the key takeaways is to send in the deposit for analysis. We can do chemical analysis and look at what material is actually there. There are more advanced techniques as well, like microprobe or line scanning, which can tell us where the different elements are lining up in that deposit or how they are formed over time. Even SEM (scanning electron microscopy) and X-ray diffraction can look at not just the shape of the deposit but can also identify crystal structure. 

PUI-NANG LIN [Flint Hill Resources (FHR)] 

Another area we found very important for the expander fouling is the quality of your expander cooling and impingement steam. That is another source of sodium that can accelerate the blade deposits. 

ALEXIS SHACKLEFORD (BASF Corporation) 

Another element you should look for is sulfur. Sulfur is often enriched in these deposits. Occasionally, you may also see evidence of refractory in these deposits. Please see BASF’s response in the Answer Book that shows you what these deposits look like compared to e-cat and compared to fines samples.  

MELIKE YERSIZ (Chevron U.S.A., Inc.) 

How often do you recommend inspecting the blades with the strobe lights? 

BHARGAVA (KBC Advanced Technologies, Inc.) 

If you have a viewport in a strobe light, then you should be routinely monitoring the blades once every couple of weeks, depending on the severity of the situation. You can then increase or reduce the frequency, depending on how your deposition goes. 

FEDERSPIEL (W.R. Grace & Co.) 

When I worked at Hovensa, we would do it weekly. We often found out that staying ahead of the problem was a lot easier than trying to address it after it became an issue. So, if you have the facilities there that look to the viewport or take the pictures, it is probably easiest to set it up on a regular basis – like on Saturday – just to have the Inspection guys go out, take the pictures, and monitor it. 

PHILLIP NICCUM (KP Engineering, LP) 

I want to make a reference to a paper written by David Linden with Ingersoll Rand back in the 1980s. The topic of the paper was the composition of these deposits. It is a seminal work on this subject, and I recommend it. In the paper is a reference to tables of these eutectic mixtures, and I have some of them. It is pages and pages of eutectic mixtures with many elements from FCC with which you are very familiar. It is quite a useful reference.30 

BOB LUDOLPH [Shell Global Solutions (US) Inc.] 

Calcium and iron also play into those eutectics as well and can have a dramatic effect. As far as the sodium goes, make sure your desalter is being checked for its effectiveness, because a dramatic shift in its performance could really result in a much larger change in the expander operation. 

SANJAY BHARGAVA (KBC Advanced Technologies) 

Deposit formation is usually linked to catalyst depositing on the turbine blades. The deposits are mostly a function of catalyst loading of the inlet flue gas. An increase in catalyst loading could be due to several factors, including performance of the third-stage separator, loss in cyclone efficiency, change in catalyst attrition properties, excess fines in fresh catalyst, an increase in fresh catalyst additions, and/or an increase in flue gas rates due to higher air rates.  

In addition, KBC has seen locations where steam is introduced between the regenerator outlet and tunable diode laser spectroscopy third-stage separator. The addition of steam is used to control the temperature within the guidelines of the expander. The use of steam and the added “humidity” increase the deposition potential of the catalyst on the blades. Further, high levels of sodium and vanadium on equilibrium catalyst can also form a sticky, eutectic mixture which would tend to stick to the blades more easily and lead to accelerated deposition rates. 

Rigorous monitoring of catalyst balance and fines generation – specifically, sub-20-micron particles – helps us understand the deposition rate on the blades. Minimizing deviations from design pressure and temperature also helps reduce the rate of deposition. Monitoring catalyst physical properties and Na/V (sodium/vanadium) to ensure good cyclone efficiencies is as important. Early or regular action to correct the deposition problem by routine monitoring of bearing temperatures and vibration, along with process temperature and pressure, will provide the required information for corrective course of action. 

Deposits can be monitored via a view port with a strobe light to allow weekly photographing of the blades. This information can then be used to establish the frequency of regular online cleaning of the blades. Cleaning can be done with rice, walnut shells, or a less preferred method of thermal cycling of the turbine by partial bypassing of the flue gas to cool down the blades. If the deposits are allowed to build up over extended time periods, online cleaning is not recommended as chunks of deposits are likely to be removed non-uniformly, which can unbalance the blades and result in excessive vibration. 

MICHAEL FEDERSPIEL (W.R. Grace & Co.) 

The paper, “Catalyst Deposition in FCCU Power Recovery Systems” by David H. Linden31 at Ingersoll-Rand, refers to four types of deposits that occur in flue gas lines, equipment, and power recovery turbines. The first type (A) are powdery catalyst deposits that cling to surfaces in the flue gas train. The second type (B) occurs when those powdery catalyst deposits get wet and then harden after drying out. Upon analysis, these deposits appear similar in chemical makeup to equilibrium catalyst or third-stage separator fines. 

The third type (C) of deposit is made up of very small catalyst particles, along with elevated levels of alkali metals (sodium, potassium, and calcium), chlorides, vanadium, and iron, as well as other trace contaminants from unique or challenging feedstocks. It is theorized that these contaminants form a low melting point eutectic. These deposits are very hard and are often found in areas of high gas velocity and pressure drop, and the expander is just such a place. If these deposits form along the expander blade tips to sufficient thickness to cause rubbing, the friction will increase the temperature enough to sinter the deposit into a new type (D) of deposit which has increased metals content from the expander metallurgy.  

By reducing catalyst traffic to the flue gas and preventing condensation, these types of deposits can be minimized. Keeping the regenerator cyclones and third-stage separator mechanically healthy and operating within design specifications will help. Ensuring catalyst coolers are leak-free will prevent boiler feed water chemicals from further contaminating catalyst and eliminate an attrition source. Running the expander at design conditions can keep the flow path through the expander fully developed, thus mitigating eddy currents and dead spaces. Suppressing the contaminant metals (particularly those that accumulate on the surface of the catalyst particle and then abrade off) or using a catalyst with a low attrition tendency can help to reduce blade depositions. There are also several options for expander coatings which can reduce or prevent the accumulation of deposits. 

Monitoring deposits during the run can help a refiner be prepared for work that needs to be done during the next available outage. Taking pictures through view ports and vibration monitoring are two common methods used to quantify and track deposit formation.  

Once formed, deposits can cause loss of expander efficiency and threaten the mechanical integrity of the machine, as well as force a shutdown due to high vibrations. While these deposits can be removed through the injection of walnut shells or rice, a regular program aimed at preventing the formation of deposits is generally more effective at achieving longer run lengths than attempting to fix a vibration issue after it develops. Varying the size of the media can help reach the different places these deposits form. Other methods include thermal shocking or thermal cycling of the expander, which takes advantage of the different thermal expansion coefficients of the deposits and the turbine metallurgy to release accumulated deposits from the surface of the blades.  

ALEX MANNION (BASF Corporation) 

Below are deposit examples from a cyclone and expander blade. Typically, these deposits are rich in elements such as Fe (iron), Ca, Na, Mg and S. These elements can act as a “glue,” binding catalyst fines together. 

Factors that could lead to deposit formation include high catalyst attrition, high catalyst losses from cyclones, poor water quality being used from steam injection, water condensation and precipitation at cold spots, and high metals content in the FCC feed. 

Several measures can be taken to avoid deposits from laying down on the blades. To minimize fines generation, an attrition-resistant catalyst can be used, and fresh catalyst additions should be minimized. Also, all steam ROs (restriction orifices) should be in place, and excess velocities should be avoided. High-quality water should always be used while minimizing water injection, when possible. “Cold spots” in the overhead line or expander should be avoided. Crude desalting additives can be used to minimize metals (e.g., Ca) in the feed. Finally, effective soot blowing of CO boiler tubes can mitigate deposit formation as well. 

Deposits can lead to expander vibrations and blade erosion, potentially leading to catastrophic equipment failure. If deposits have begun to form, regular walnut shell cleaning can be conducted and thermal spalling if required.