Refincor Ammonium Bisulfide References


M. C. Flickinger reported on a failure in an effluent system. They do use continuous wash water injection, but to lower ammonium bisulfidefrom a typical 5% to 7% concentration to a more desirable 2% level would require an increase from 284 L/min (75 gal/min) to about 1,136 L/min (300 gal/min). Their water is condensate with 100 ppb oxygen. He asked for advice on water wash rates and acceptable oxygen levels.


R. L. Piehl responded with an opinion their oxygen level was too high at 100 ppb. Use of stripped sour water, instead of condensate, should improve this situation, if oxygen leaks are avoided. When separator water ammonium bisulfidelevels are less than 2 wt% to 2-1/2 wt%, problems are usually nil, but there is no sudden abrupt change because other factors are also important.


J. D. McCoy stated that surplus injection water is needed to avoid full evaporation. McCoy uses 25% excess water to assure full saturation. Typical wash rates for McCoy would be 190 to 380 L/min (50 to 100 gal/min) for a 25,000 to 35,000 BPD unit, whereas W. H. Sharp reported over 760 L/min (200 gal/min) for a 20,000 BPD unit at 149øC (300øF) and 10 MPa (1,500 psig).


E. F. Ehmke agreed too much water shouldn't erode if gas velocities are less than 6.1 m/s (20 ft/s). Proper spray nozzles are important for good mixing.


G. F. Rak said that they have also seen more salt plugging in their effluent exchangers.


J. E. Cantwell commented that the problems might be caused by ammonium hydrosulfide due to lower charge rates, resulting in lower temperatures in the effluent train. At higher temperatures, the ammonium hydrosulfide would pass through the exchangers into the separator. D. R. Clarida agreed that this was a possibility on some of units in question.


Dannie Clarida, CONOCO, reported a problem that occurred in their hydrodesulfurizer effluent train, which had a low rate of continuous water wash at 1/2 gpm/1,000 bbl. In the process, at those operating conditions, the actual liquid water was 0.05 to 0.1 gpm/1,000 bbl, making conditions worse by concentrating corrodents and severe corrosion occurred in the carbon steel pipe downstream from the injection point as well as 3 heat exchangers.


Dannie Clarida, CONOCO, asked whether anyone was having problems with reactor effluent water wash. The amount of liquid water at temperature can lead to corrosion if there is only a small quantity present after injection.


D. R. Clarida (Conoco) described a recent corrosion problem in the bottom of a diesel fractionator downstream from the separator. Corrosion was in the form of localized, deep pitting that did not appear to have been caused by deposits. The material was carbon steel and service temperature was 585øF. What deposits there were consisted of iron sulfide, iron sulfate, ammonia, and other nitrogen compounds.


R. J. Horvath (Shell) asked what maximum velocity is being used for alloys in wet ammonium bisulfide service. He said that they were using 20 fps for CS but there would be an incentive to go to higher velocities with alloys in order to reduce costs.


Dick Horvath (Shell) described a recent occurrence, not totally understood or resolved yet. They've had a localized corrosion problem in piping and related blistering and HIC cracking in cat cracker feed hydrotreater vessels at around 1,200 psig. The unit takes a fairly high nitrogen, moderate sulfur feed. The effluent line is Incoloy. Effluent KP, as described in the Bob Piehl paper from 1976, is 3 to 4, which is very corrosive. They elected to minimize wash water and use alloy. There was some very localized piping corrosion in the overhead line from the high- pressure low-temperature (HPLT) separator, and also in the vapor line from the low-pressure low-temperature (LPLT) separator. Both of these carbon steel lines had vapor flow velocities of about 40 feet/second with insignificant corrosion for over twenty years.

With a recent catalyst change in the reactors, they got more de-nitrification of the feed. Ammonia levels went up significantly and KPs in these vapor lines tripled. They ended up with very localized corrosion washout areas, classic ammonium bisulfide corrosion. In hundreds of feet of pipe there were a few washout spots the size of a clenched fist. In the case of the LPLT separator vapor line, they could explain that they went from a noncorrosive to a corrosive range, using Piehl's KP. In the case of the HPLT separator vapor line, even though the ammonia levels had tripled and the KPs had tripled from 0.02 to 0.06, they still wouldn't expect a problem. In a discussion with Bob Piehl, he commented that his KP was derived from an empirical fit of data obtained years ago from effluent air coolers.

The HPLT area may not be the same conditions; it doesn't have a lot of water; it is a vapor that's basically saturated with water and the only water in the line is small amounts of condensate that accumulates down the line, causing the localized washouts. They suspect that they went from compositions below 10% ammonium bisulfidein that condensate to the 15% to 20%. In addition to the piping, they looked at the separators. In the HPLT separator they had done a thorough internal inspection and WFMP check in 1988; the vessel was in excellent shape.

This time there were massive blisters in the bottom head, about 4" in diameter, ruptured wide enough to stick a credit card into the rupture. There were a number of cracks that went down at least 5/8"; most of the HIC damage was into the plate about 3/4". The head was in the worst condition. All of the material was A212B. WFMP looked like a Christmas tree on both sides of welds and on the head-to-shell weld, with cracking aligned at the toe of the weld both on the shell can side and on the head side. HIC was so extensive they condemned the vessel. The LPLT separator also had some cracking, but they were able to repair that vessel. They feel they went over a threshold, probably somewhere within that 10% to 15% ammonium bisulfidelevel. The inspector commented that they used to have a nice, tight, hard, dense, sulfide scale. This time it was loose and "fluffy." This scale seems t Monday, August 15, 2005 12:25 PM ry little corrosion in the vessels; there was one washout area about 1/4" deep at an interface level, probably from a whipping action from the gas inlet.


Joerg Gutzeit (Amoco) asked whether they could have been seeing under-deposit corrosion, or whether cyanides were involved.


Dick Horvath (Shell) responded that the spots were obvious washouts; the scale around them was intact and the washout areas were devoid of scale. There was almost no loss at all surrounding the fist-sized washouts. Just downstream were big, voluminous deposits, in which one could see flow effects. There was no evidence of cyanide, as indicated by Prussian blue coloration or sampling and testing.


Joerg Gutzeit (Amoco) had some related comments. They have several amine unit stripper overheads with a lot of ammonium bisulfide and KPs as high as 60. Ammonia levels are around 8%, and H2S is about 5% in the water. 304 SS seems to corrode out and they've gone to 316 SS, which seems to be holding up.


Don Monday, August 15, 2005 12:25 PM ound-color: #ffff90">ammonium bisulfide, not in terms of KP. They try to keep below 10% in effluent air coolers. Chevron uses a lot of carbon steel there. They control velocities, and in the amine stripper overhead reflux line they try to control the velocities also to 20 feet/second.


Joerg Gutzeit (Amoco) said they also use % by weight, which is easier to interpret than KP, but they use 2% as a value that they believe comes closer to Bob Piehl's KP guidelines.


Don Truax (Chevron) responded that they believe, as a rule of thumb, that with less than 2% there probably is no corrosion problem; if you're higher than that, maybe to 10%, controlling the velocity helps out; if you're a lot higher than that, you'll get corrosion even if there's no velocity.


Ara Bagdasarian (UNOCAL) said that they use 8% as their guide in effluent air coolers, but it depends on the symmetry of the air cooler, piping, etc. All configurations are not necessarily good with 8%.


John Coombs (Arco) said that their KP criterion is 0.07, which comes from the Piehl paper. Above this value they'll alloy. It's a significant point that Dick Horvath's corrosion occurred at a KP of 0.06. He added that Piehl's original KP was calculated from table values, whereas the bisulfide concentration is measured in the field. His concern was that there may be a difference between calculated and measured values, possibly because of the problems in getting good measurements.


Dick Horvath (Shell) has not used KP a lot; they usually use ammonium bisulfidelevel. Looking at KPs as a postscript didn't leave them very comfortable. They measured ammonium bisulfideand historically had not exceeded 10%, typically around 8% to 11%, but during this run it was 14% to 16%, which was a significant jump.


Cathy Shargay (ARCO) reported on an HDS unit where they wanted to reduce the water injected upstream of the effluent air cooler in the overhead of the hot flash drum system. They have Incoloy 825 tubes in the air cooler and 825 piping downstream of the water injection point. The concern is with the carbon steel piping at the outlet of the air cooler. A mass balance determined that the ammonium bisulfidelevel was 9.5%. The original design was for 8%. Inspection had not shown any problems. Two corrosion probes were installed in 1992 in the piping. Wash water rates were increased in expectation of an increase in denitrification due to a catalyst change that would increase the level of ammonium bisulfide. The increased denitrification did not materialize, and they had 5.8% ammonium bisulfide. Therefore, they reduced the water rate, which increased the ammonium bisulfide to about 8.7%. The corrosion probes showed a rate of 6 to 11 mpy, which they consider unacceptably high due to the potential severity of this type of corrosion. They are now increasing the water rate and decreasing the ammonium bisulfide content to about 7.2%. They will continue to monitor and will optimize the wash water rate using the corrosion probes. The air cooler outlet piping goes to a cold flash drum. They have not detected hydrogen activity in the drum, but they do not have hydrogen probes. She emphasized that the only corrosion they have seen so far has been on the probes, not in any equipment.


Dick Horvath (Shell) reported a similar situation in a hydrotreater unit. They saw increased hydrogen activity in the effluent train separator drum downstream of the air coolers with about 7 to 7.5% ammonium bisulfide. After a run, they did not see considerable corrosion or blistering but found HIC damage. The system is similar to the one Cathy Shargay described. The effluent flows to a hot flash drum that operates above the ammonium bisulfide range. The vapors leave the drum and are water washed before passing through the air cooler to the carbon steel cold flash drum. It operates at about 120 to 130øF and 1150 to 1200 psi pressure. They did a fitness-for-purpose analysis and believed they did not need to make repairs.


Ray Konet (Amoco) - They experienced erratic readings during UT measurement of cat feed hydrotreater unit effluent air cooler outlet piping to the trim coolers. The piping was 12-in. diameter carbon steel and the ammonium bisulfideKp was 0.8. During a B-scan of the suspect area, they found a total loss of back reflection which was caused by blisters in elbows with no general corrosion. Piping was seamless, and the elbows had been hot bent. The microstructure showed some banding.


Dannie Clarida (Conoco) described recent failure in a gas oil hydrodesulfurizer that treats feed for an FCC unit. There was a leak and a fire at an elbow where the piping comes off the shell side of the trim cooler before the high-pressure cold separator (corrosion pretty well localized to the elbow). The line temperature was approximately 49øC (120øF) and velocity calculated to be 12 m/s (39 ft/s). The material was carbon steel and the ammonium bisulfidecontent in the sour water off the cold separator was 10 wt.%. The average corrosion rate was estimated at 1.2 mm/y (47 mpy) over the period from 1990 to 1996 with a rate of 1.4 mm/y (56 mpy) from 1990 to 1992, 0.15 mm/y (6 mpy) from 1992 to 1994, and an accelerated rate of 1.8 mm/y (69 mpy) over the 20-month period from 1994 to 1996 (average corrosion rate based on 0.36 cm [0.14 in.] of metal remaining at point of hole-through). Since start-up the feed rate has increased from a design of 2,000,000 to 3,600,000 L/d (13,000 to 22,500 BPD) and the nitrogen content of the feed has increased. Together, these changes increased the ammonium bisulfideconcentration, but the water wash rate was not changed from the original design because the pump was too small for an increased rate and the water wash rate was not evaluated as a part of the expansion.


Jim Feather (Exxon) described similar problems in two different hydrotreaters (one lube, the other distillate). In both cases the location was between the final exchanger or condenser in the hot separator overhead system and the cold separator. In the distillate hydrotreater the velocity was approximately 11 to 12 m/s (35 to 40 ft/s) and the ammonia bisulfide level was about 10 wt.%. The carbon steel corrosion rate in a T-piece was found to be approximately 5 mm/y (200 mpy); it didn't leak but was found by inspection. This was repaired by a weld overlay of Type 316L SS because a similar area had already been overlaid in a similar manner and the overlay was found to be in excellent condition. In the lube hydrotreater the velocity was approximately 11 m/s (35 ft/s) and the ammonium bisulfide level was about 5.6 wt.%. The carbon steel corrosion rate was about 3.8 mm/y (150 mpy). Again, this was something that was caught by inspection because it was targeted as potentally vulnerable because of the known high velocities.


Dannie Clarida (Conoco) - described a sensitization problem that occurred in some seamless ASTM A 312 type 321H SS FCC feed HDS furnace tubes. The unit was started up in 1985 and revamped in 1989. the tubes were designed for 510̊C (950̊F), 9.96 MPa (1,430 psig) with an operating skin temperature of 455̊C (850̊F). The tubes were in the helical radiant section and had been in service for 87,000 hours, 7 years on hydrogen and 4 years on oil and hydrogen. On startup in 1996, 14 through-wall cracks were found. The cracking was determined to be chloride SCC initiated from the outside. The source of chlorides was refractory debris and steam in the fire box. Additionally, it was found that the tubes had a significant amount of sensitization, which was attributed to the use of the high-temperature solution anneal temperature of 1,121̊C (2,050̊F), which can make the material more susceptible to sensitization at the operating temperature of the tube skin. Some internal cracking due to polythionic acid SCC was also found, but these cracks were not through-wall. The coil was replaced with type 321H SS annealed at 1,065̊C (1,950̊F) followed by a lower-temperature stabilizing anneal at 885̊C (1,625̊F) for 4 hours.


Dannie Clarida (Conoco) - described a tube leak problem that occurred in a type 347H SS vertically tubed recycle gas furnace. In this case, the furnace had hydrotested with a soda ash solution and an attempt had been made to blow the tubes dry of residual solution. The furnace was then fired to dry the tubes. The leakage that occurred was due to chloride pitting corrosion initiated by residual chloride in the soda ash solution. The next time the furnace was tested, some holes were drilled in the "U" bends to remove the residual soda ash solution.



The safe operation of oil refineries in the United States is under constant more
When: January 30, 2017 - February 2, 2017 Where: Galveston Island more

View all articles

Stay Current

Sign up for our quarterly newsletter

covering updates on corrosion