At the most recent National Petroleum Refiners Association question and answer session on refining and petrochemical technology, refiners and a panel of experts exchanged experiences on hydrotreater operations.
Topics addressed included reactor pressurization, scale basket removal, and the use of antifoulants in effluent exchangers.
For details on the format of this meeting, held Oct. 11-13 in Washington, D.C., see the first installment of this series (OGJ, April 24, p. 61).
What is the industry practice used to speed up the pressurization of 2.25 Cr/1 Mo reactors during start-tip? Is there any relationship between reactor skin temperature and pressure used?
TREESE:
I wish I could say I had news here, but there is no way around the heat-up and pressurization steps.
In recent years, the chemistry of the 2.25 Cr/1 Mo steels has been improved. This has improved their resistance to chrome-molybdenum temperature embrittlement. However, heating up to the cold pressurization temperature remains a required and necessary step in the unit start-up.
To speed up pressurization, it is necessary to speed up heating at the cold pressuring temperature. This can be done by maximizing the pressure, as limited by the cold pressurization level, and maximizing the mass flow to the reactor.
Heater outlet temperatures should be adjusted so the heat-up is less than 75 F./hr, or the differential between the bed temperatures and the skin tempera-lures on the outer wall is less than 150 F. Both restrictions are to avoid thermal stresses within the reactor wall. Reactor skin couples aid in monitoring metal temperatures so the refiner is able to make a safer decision on when to pressure the reactors.
Unocal practices in the heat-up of 2.25 Cr steels vary among our locations, depending on the vintage of the reactors.
The following guidelines are generally observed:
We keep the reactor below 20% of hydrotest pressure until the wall temperature is over 250 F., and this is sometimes taken as 20-30% of full pressure, which is relatively conservative. If you work it out, it is about 30%. Above 250-275 F., the reactor is brought up to full pressure.
The exact pressure/temperature increase schedule for a set of reactors is dependent on the reactor vintage and the chemistry. On some of the newest reactors, you only need to go to about 200 F. before pressuring up. On old reactors with known defects identified by detailed fracture mechanics analysis, we have actually gone all the way up to 350 F. before pressuring up. The best guidelines you will find are from your reactor manufacturer.
On very-high-pressure hydrotreating units, heat-up time is a problem. This is because a very large mass of catalyst and reactor is present, and, generally, there is not a huge volume of gas to heat it up, at least not compared to a hydrocracker.
Spiking the recycle gas with some methane helps you get a little bit more molecular weight, a little bit more mass, through the reactors. That can help, but you must keep in mind that you are limited in how fast you can heat it up by some of the guidelines mentioned above.
LAUX:
We also follow our manufacturer's recommendations on the minimum pressurization temperature for these reactors, even for 1993 vintage reactors, although this requires heating up to 275 F. before the pressure is raised to the 20% limit, as Mr. Treese mentioned. That is about 500 psig for our reactors.
We do monitor skin thermocouples to give us an indication of where we are. The actual heat-up time to get to the 275 F. is about 24 hr, but we do not see any reason to shortcut this step of the procedure.
PORTER:
We limit the operating pressure to one-quarter of the design pressure when the external skin temperature of the reactor at any location is below a specified minimum pressurizing temperature. Each reactor is assigned a minimum pressurizing temperature based upon its particular steel, and takes into account "temper embrittlement" of the 2.25 Cr/1 MO material from long-term operating service.
SAYLES:
I agree with the comments that were made previously. The one caution I have is to make sure everyone agrees on the minimum temperature measurement point and the minimum temperature value.
D'AURIA:
Our guidelines are similar to those already stated: The reactor pressure should be limited to 20% of hydrostatic test pressure or 25% of operating pressure. This is for 2.25 chromo/1 molybdenum reactors. For newer 3 chrome/1 molybdenum reactors, this minimum temperature can be lowered to 200 F.
DEATON:
We do not take any short-cuts for pressurization during start-up for fear of inducing brittle fracture. We follow very strict guidelines for our Isomax hydrocracker, which operates at 1,600 psi.
Our guidelines are not to exceed 25% of operating pressure, or 400 psi, until reactor skin temperature is above 350 F. We watch the reactor bed temperatures and know from experience that a 400 F. bed temperature corresponds to a 350 F. wall temperature.
During the warm-up period below 400 F., we raise bed temperatures at a maximum rate of 50 F./hr. If we are starting up in the dead of winter, when temperatures are frequently near 0 F. in Chicago, initially we raise the bed temperatures at a very conservative 25 F. /hr.
CHARLES MCCOY
(MCCOY CONSULTANTS INC.):
One serious point to add to what Mr. Treese and others have said: Skin temperatures must be measured, not just on the shell of the reactor, but on nozzles and man-ways too. You have to get the whole reactor hot. You do not want any part of it to fail from brittle fracture.
On a more humorous side, two of our clients have come up with interesting ways to speed this up. One refinery in Australia has an idle boiler plant, and they have steam-traced their reactor. They just fire up the idle boiler and heat it up with steam while the rest of the plant is coming out of the shop. The Taiwanese, not to be outdone, put a giant electric blanket on their reactor and have a giant electric plug into which to plug it.
ANDERS NIELSEN
(HALDOR TOPSOE A/S):
When pressuring a vessel in 2.25 Cr/1 Mo, you have to consider the toughness of the material vs. temperature, as in some of the examples given by the panelists. However, you have to consider two additional points:
- One, during heatup, the delta T across the wall may be very significant.
- Two, if the vessel has been operating in hydrogen at high temperature, there will be hydrogen left in the material. Such a vessel should never be exposed to a high pressure at a low temperature. The hydrogen moves the toughness vs. temperature curve so that you need a higher temperature in order to apply a certain pressure.
Has anyone removed scale baskets from a hydrotreating reactor and compared operations before and after? If so, were there any noticeable differences? Why?
BOYCOTT:
We have operated both with and without trash baskets in our naphtha hydrotreater. This unit experiences significant pressure drop problems due to the processing of coker naphtha.
The typical arrangement for our reactor is 3-ft trash baskets with a graduated bed of inert spheres. The trash baskets extend through the graduated bed and slightly into the bed of catalyst.
On one occasion, in order to try to extend runs, and due to the fouling problems, we installed a guard bed of highsurface-area active support media in this reactor. We did not use trash baskets when we installed this bed.
The results were almost identical to the runs utilizing inert spheres and trash baskets-actually close enough to be indistinguishable. Since that time, we have gone back to the inert spheres and trash baskets.
I would point out that one useful test, and something that we have considered, would be a run with the high surface area, active support media with the trash baskets installed.
CUNEO:
We have had experience both with and without the trash baskets. In most cases, the baskets have been removed because there was little or no accumulated scale. We have observed no loss of performance in the cases where they were removed.
D'AURIA:
The advantage of trash baskets is that they can extend the time on stream before pressure drop becomes a problem. There are some drawbacks, the most significant being poor liquid distribution. UOP does not recommend trash baskets, but instead we prefer to install graded material at the top of the reactor. This system provides trash removal capability with better liquid distribution.
ARMBRESTER:
Our general practice is to use scale baskets in the light hydrocarbon services where the coking tendency is small. But in heavier services such as fluid catalytic cracking (FCC) feed hydrotreating, we use a distributor plate, which is primarily designed for improved mixing and distribution of the feed and hydrogen, but also serves to protect the bed from scale and debris.
We have had a couple of cases at our St. Paul Park refinery where we removed the scale baskets for one run. We did see considerably faster buildup of pressure drop in those reactors and, as a result, we have reinstituted the use of the scale baskets.
We have one other case in which we removed the scale baskets in a solvent hydrotreating reactor and replaced them with a distributor plate, and in that case we have not seen an increase in pressure drop.
BARLOW:
We have observed the use of baskets for many years, and the comment that we would make is that, if you have a pressure drop problem due primarily to migratory foulant material, we would not recommend removing the baskets. We have seen that done in one case, and the subsequent run was disastrous due to high delta P. The result was shortened run length.
If it is not migratory material, then other alternatives could be attempted.
LAUX:
We have removed the scale baskets from our No. 1 naphtha hydrotreater reactor and have not seen any increase in the pressure-drop buildup over the course of the run. Based on that experience, we did not install any on our new No. 2 naphtha hydrotreater reactor, and we also deleted the scale baskets that were designed for the first reactor in our hydrocracker. We have not seen any increased pressure drop due to scale or trash in either of these reactors either.
Our reasons for removing the baskets are that it does allow us to load more catalyst, and it also makes for much easier catalyst dump without the baskets inside the reactor.
TREESE:
Our experience has been much the same. The scale baskets have been of benefit where significant scale accumulation is expected, just because they increase the area of influx into the bed.
We have had experience where, when the reactor was opened up, the scale baskets were found to be completely filled with fines and scale material and, in that case, they probably helped the unit stay on-line.
If there is really little scale coming into the reactor from migratory materials, as Mr. Barlow mentioned, the scale baskets probably are not effective and you would be better off with a graded bed-type design. A graded bed is good enough for pressure-drop protection.
Recently, we removed the scale baskets in the hydrotreating reactor at one of our Unicrackers and replaced them with just a graded catalyst bed, and we have seen no problems and no unusual pressure drop accumulation. This is consistent with the fact there was no scale ever collected in the scale baskets.
CHARLES MCCOY
(MCCOY CONSULTANTS INC.):
I have preached against scale baskets for a long time because of the reasons that the panel has said. They may or may not extend run length, but they definitely extend the turnaround time. It takes a lot longer to get rid of the pressure drop and to clean up the reactor if you have to get those baskets out of there. I think Mr. Treese said it right a moment ago: Good bed grading will do the same job and still not interfere with the turnaround.
BOYCOTT:
I would like to make one comment on the removal of trash baskets during turnaround. The method that we have found extremely successful is to chain the baskets together and to the top head. We simply pull the baskets out with the chain. This does not extend the turnaround time.
What is the industry experience with the use of antifoulants for hydrocracking or hydrotreating reactor effluent exchangers?
ARMBRESTER:
Ashland has about 14 years of experience with the use of antifoulants in our distillate desulfurizer unit at Catlettsburg. In 1980, we conducted a 90-day, side-by-side test of two parallel heat exchange trains in this unit, with one train treated with an antifoulant while the other train was untreated.
This unit was designed for blocked-out operation on either kerosine or diesel feed. The diesel stream was a mixture of virgin diesel and light cycle oil from the FCC unit.
At the time of the testing, we had been experiencing rapid fouling of the reactor feed/effluent exchangers, resulting in excessive pressure drop through the exchangers and increased firing duty on the reactor charge Theaters. Run lengths between exchanger cleanings were typically 4-8 months, resulting in excessive downtime and high maintenance costs.
This side-by-side test demonstrated that run length could be nearly doubled with the use of antifoulant, with corresponding increases in heat-transfer efficiency. Consequently, we have used an antifoulant continuously since that time.
In 1993, a new diesel desulfurizer was commissioned at Catlettsburg to meet the low-sulfur diesel requirements, and this unit hydrotreats the mixed diesel/cycle oil stream exclusively, with the original desulfurizer now used to treat only kerosine.
We have used a Petrolite antifoulant at a 20 ppm treat rate in the new diesel desulfurization unit since the first day of operation, and have now operated for more than 13 months with less than a 10% decline in heat transfer. During the same time frame, the kerosine treater has been operated at an antifoulant rate of 9.5 ppm, and the heat transfer has declined by less than 5%.
BARLOW:
We are involved right now in treating over 50 hydrotreaters. We have been involved historically in monitoring and treating over 100 hydrotreaters. I break this question down into two parts: feed-side fouling, and effluent-side fouling of the feed /effluent exchangers.
On the feed-side fouling, we focus much of our efforts away from the traditional use of dispersant chemistries. Dispersant antifoulant chemistries are applied to try to prevent agglomeration of foulant material. They are effective on iron sulfide-type foulant materials, but not as effective, or not as efficient, on polymeric materials.
We concentrate on feedstock characterization and feedstock handling, as well as applying the right polymerization inhibitor, whether it is a free radical fouling problem or a condensation fouling problem. You also need to go upstream in your analysis if you are using tankage. Take a look at the tank handling of your feeds and what kind of oxygen intrusion you incur.
Mixing of feed is very important, too. You may have a straight-run feed that does not have an oxygen-initiated polymerization problem, but when you take that through tankage and combine it with a cracked feed such as a coker material, you can create a mess. So we have incorporated mechanical and operational changes for feedstock handling and inhibitor-type antifoulants to provide solutions to fouling problems.
On the effluent side, our primary experience focuses on ammonium chloride deposition. Water wash is historically used and is quite effective at controlling ammonium chloride deposition. If you do have underdeposit corrosion, you may need to apply a filmer. At fairly low treatment levels, an economical treatment program can be used to control corrosion.
If for some reason you have a water handling constraint, you can use detergent/dispersants to essentially move the ammonium chloride downstream to where you can better tolerate the deposition or remove it from the system.
BOYCOTT:
We have experienced severe fouling of the feed side of our naphtha hydrotreater feed/effluent exchangers. This problem is caused by the polymerization of gum precursors in the coker naphtha charge to the unit.
The coker has been expanded twice. Prior to the first expansion, the concentration of coker naphtha in the hydrotreater was 10%. At that time, run lengths were not limited by feed/effluent exchanger fouling.
After the first coker expansion, the concentration of coker naphtha in the hydrotreater charge increased to 15%. This caused a dramatic increase in feed/effluent exchanger fouling. Runs were limited to 2-3 months as a result of the fouling.
Two chemical addition systems were installed to address this problem: an antioxidant in the coker naphtha rundown stream and a dispersant in the naphtha hydrotreater charge stream. Runs were increased to 6-9 months.
After the second coker expansion, the concentration of coker naphtha in the naphtha hydrotreater charge rose to 25%. At that point, the chemical additions proved adequate and other measures were taken.
DEATON:
In our naphtha hydrotreaters, fouling and corrosion on the effluent side of the exchangers is due to ammonium chloride salt deposits. To protect the effluent coolers, we continuously water wash at a rate of about 1 gpm/1,000 bbl, which is sufficient to keep the stream entering the coolers below the dew point.
On the feed side of the exchangers, we inject antifoulant at a 12 ppm dosage rate. This antifoulant contains dispersants, corrosion inhibitors, and antioxidants. We regularly import naphtha at about 10% of total naphtha charge during the peak season and these feedstocks are often thought to cause fouling because of oxygen-induced polymerization.
We usually get a 2-year run on these exchangers before they are pulled and cleaned.
LAUX:
Like other people, our No. 1 naphtha hydrotreater is fed a stream from an unblanketed storage tank. The feed/effluent exchangers had a severe fouling problem on the feed side. They actually would plug off.
We monitored pressure drop instead of heat transfer to determine run length. Some runs were as short as 6 months. We then started using an antifoulant and, since then, we have not had much of a fouling problem. Run lengths are up over 2 years.
We have not seen any fouling on our McKee hydrocracker feed /effluent exchangers on the feed side in over 5 years of operation. However, this unit is fed directly from the crude units and there is no intermediate tankage.
KENNETH D. PETERS (UOP):
The need for antifoulant injection can be reduced significantly, and most often eliminated, by removing the oxygen from the naphtha by process means or preventing contact of the naphtha with the atmosphere. Particularly in the design of some of our overseas units that import significant quantities of naphtha by barge or tanker, we have had to add a reboiled oxygen stripper column, rather than a gas stripper, to eliminate the potential oxygen fouling problem in hydrotreating combined-feed exchangers.
Other than that, gas blanketing of the intermediate storage tanks between the crude unit and the naphtha hydrotreating unit seems to be mandatory to prevent fouling of the combined feed exchangers.
HEROS DERGREGORIAN
(GIANT REFINING CO.):
In case of gas blanketing of the feed tank, has anybody been successful to prevent feed-side fouling? Also, is the naphtha hydrotreater effluent exchanger, on the effluent side, using inhibitors? Have they experienced any problem in the reformers, as far as causing problems with catalyst poisoning?
BOYCOTT:
Yes, with regard to gas blanketing of the charge tank. This was one of a number of things that we did in addressing the feed-side fouling on our feed/effluent exchangers. We were successful in reducing the fouling, and I am sure that had some effect on it.
JAMES KELLY
(CITGO REFINING & CHEMICALS CORP.):
Several people have mentioned fouling on the effluent side. We, at Corpus Christi, had a problem in our catalyst feed hydrotreater and we developed pressure drop on the effluent side, which we think was ammonium chloride.
We injected water and the pressure drop went away, but we never could find the source of the chlorides. We could not see the chlorides in the feed, and the hydrogen was clean. Has anybody detected chlorides in the feed-organic or inorganic?
ARTHUR J. SUCHANEK
(CRITERION CATALYST CO. L.P.):
At the risk of having bricks or something thrown at me by the additives people, there is something that is in between the feed and the effluent. It is called the reactor, and it is full of catalyst. I would suggest that you talk to your additive suppliers to make sure what it is you are buying as your additive, because sometimes they can cause you troubles.
MOHAMMAD AL-SHAHRANI
(SAUDI ARAMCO):
Is there an oxygen scavenger that can be used to remove oxygen from hydrocarbon?
BARLOW:
Yes, there is. Hydrazine compounds or hydro-quinone-type compounds have been used. One drawback is that they do not mix well, so we have developed some oil-soluble materials that work. But we find oxygen scavenging is rather inefficient because you have to feed on a stoichiometric amount, which is not cost-effective when dealing with a hydrotreater fouling problem.
It is preferable to feed an inhibitor to deal with the polymer precursors. That has been our experience.
We have seen imported feed as a major problem, especially when barges have been used. There are a couple of things we have done successfully to address this problem.
First we monitor filterable solids on that feed before it is offloaded into the charge tanks. This gives you a very good indication as to whether you have a polymer problem.
Second, when you do identify that there are polymer precursors and you can get back to the source, putting in inhibitors at the source-not dispersants of course-has been very cost effective.
Copyright 1995 Oil & Gas Journal. All Rights Reserved.
