Forum Replies Created
If your LPG is failing copper strip corrosion, there is sulfur present in the LPG. Copper strip corrosion detects the presence of H2S and elemental sulfur. Polysulfides can also cause the Copper strip corrosion test to fail. As the lead acetate paper was not black, then H2S and polysulfides are not the cause. Elemental sulfur can be formed at ambient conditions when oxygen, H2S and liquid water are in contact with one another.
If the problem is elemental sulfur, then caustic circulation rates are not the important factor. Find the source of oxygen and eliminate it. The H2S will then react with the caustic and the LPG should pass the Copper strip corrosion test.
Another sulfur compound that can cause LPG to fail the Copper strip corrosion test is COS. It will hydrolyze back to H2S and CO2 and the H2S will cause the Copper strip to fail. It can sometimes be difficult to detect this mechanism. To remove COS from the LPG, a stronger caustic is needed. Typically KOH in methanol is used to remove COS. Another method uses a primary amine, MEA, to remove the COS.
There are test procedures from UOP that will test for mercaptans and polysulfides.
I think the rational to change the packing is driven by the tower internal inspection criteria. Once every 10 years is what is recommended in API 510 Pressure Vessel inspection.
There is a discussion about vessel internals, “When vessels are equipped with removable internals, internals may need to be removed, to the extent necessary, to allow inspection of pressure boundary surfaces. The internals need not be removed completely as long as reasonable assurance exists that damage in regions rendered inaccessible by the internals is not occurring to an extent beyond that found in more accessible parts of the vessel.” that would allow the packing to stay in place longer as long as all the conditions are met.
I agree with the other comments concerning the fouling of the packing. Replace the packing and investigate why coke is being formed or entrained into the packing.
The primary control of sodium in the feed to the Coker is the performance of the desalter in the Crude unit. Caustic addition in the Crude unit due to poor desalter performance or high salt content in the incoming crude oil all contribute to the sodium content of the Coker feed. This leads to monitoring the desalter and caustic addition will provide some insight into the sodium content in the feed to the Coker.
As you know, sodium in the feed will catalyze coke formation. At sodium concentrations below 10 ppm, this effect is not significant. Even up to concentrations of 15 ppm the increased fouling rate is not very noticeable. Above that concentration, the fouling rate can be excessive.
I recall a presentation many years ago from a Brazilian refiner that was contaminating the crude oil with sodium by using desalter wash water from a sour water stripper that was processing spent caustic. The caustic in the desalter wash water was entrained with the crude oil and ended up in the feed to the Coker. Once that source was removed, the Coker heater fouling rate decreased.
Monitoring the desalter performance and caustic addition in the Crude unit are probably the most direct methods of preventing sodium in the feed to the Coker.
It is very common to flush the vacuum residue line from the vacuum unit to the Coker with a lighter hydrocarbon that will not set up. Diesel is commonly utilized for this service. Light cycle oil from the FCC is sometimes used to flush the lines in this service.
If the line is adequately heat traced, it would not need to be flushed. There is process risk in leaving vacuum residue in the line and hoping the heat tracing prevents the line from setting up. If it does set up, temporary heat tracing using electric resistance coils have been used to heat the line up enough to get the residue flowing again.
For an extended outage, I would prefer to flush the line with diesel.
There are several methods to recycle oily sludges within the Delayed Coker. Processing a water slurry of these sludges during the water quench phase of the coke drum cycle was pioneered by Mobil Oil and was called MOSC for Mobil Oil Sludge Coking (at least that is what I think the abbreviation means). Another method is to inject an oil slurry into the coke drum either into the overhead of the drum or into the feed line downstream of the heater. Another method is to bring the dewatered oily sludge into the blowdown system to dry it out completely and then use the circulating oil from the blowdown system as overhead vapor line quench so that the oily sludge gets processed during the coking phase.
Veolia did a presentation back in 2006 about these various options. The address is listed below: https://refiningcommunity.com/presentation/process-solutions-oily-waste-processing-and-coker-injection-technologies/
Mitch Moloney of ExxonMobil did a presentation on sludge coking for one of the conferences. I think it was a CokingCoker conference back in 2004 or 2005. He listed the various methods of sludge coking and some of the processing rates. At that time ExxonMobil was licensing the sludge coking technology.
I am not aware of any other technology or significant improvements in the existing technology to recycle oily sludge into a Coker.
The formation of elemental sulfur requires H2S, oxygen and liquid water. For elemental sulfur to form in the spent air stream there would have to be H2S that was not contacted by NaOH in the prewash caustic or in the extractor. Maybe the caustic is becoming too far spent before being recharged and that is when the H2S is breaking through.
A thorough review of the LPG caustic treater operation may reduce the problems you are experiencing. If not, then consider installing a larger knock out pot on the spent air line to prevent the liquid carry over you mentioned from getting into the incinerator.
Part of the reason I prefer not to put the spent air stream into the flare header is oxygen, H2S and liquid water will form elemental sulfur at ambient temperatures. Flare headers nearly always have H2S, so the likelihood of forming solid elemental sulfur is quite high. Is the sulfur in the form of elemental sulfur? That would indicate H2S breakthrough from the caustic treating system, which should not happen unless the caustic gets completely spent. This seems unlikely as the mercaptan removal would suffer before the H2S broke through.
The spent air stream should not be as corrosive as you indicate. What is the operating temperature of the caustic treating system? What is the oxygen content of the spent air? Excessive oxygen can aggravate corrosion.
My preference is to still route the spent air stream to a fired heater or incinerator rather than to the flare.
Coker LPG caustic treatment is normally for removal of mercaptans and COS through a UOP Merox unit or a Merichem Thiolex unit. The spent air is typically routed to a fired heater to be combusted. In your case you use the incinerator of the SRU, which is fine.
I have seen this stream routed to a flare as the spent air stream was not near any fired heaters. My preference is to not route this stream to a flare, but find a heater or incinerator close to the caustic treating unit and route the spent air to that destination. I do not like to put any air stream into the flare header, which is my main objection to this routing. Additionally, whatever is plugging the spent air line will also deposit in the flare header and restrict the flow of hydrocarbons out the flare.
Do you have any calcium or magnesium (hardness) in the water used to dilute the caustic to proper densities? Hardness can cause emulsions in the caustic and calcium salts can be very difficult to remove. Entrained caustic with calcium in it could be the source of the deposits. Have you analyzed them?
The duration of the shutdown is important to how the unit should be laid up. Removing the hydrocarbon from the main tower and blanketing it with nitrogen is a good practice. Similarly, doing the same thing in the gas plant should allow the unit to return to service with the minimum number of issues.
I would normally have steamed the heater coils out into the main tower and then blinded the fuel and pilot gas and any waste gas streams that are routed to the heaters. You will likely spall some coke off of the heater tubes with these actions and that coke will end up in the main tower. Removing coke from the heaters is typically a good thing. As long as you have good steam flow during the heater steam out, the heater should come back on-line with limited issues.
The only time I have left pilots on while the unit was down was when the duration of the outage was expected to be short, only a few days. If it were to go beyond that time, I would steam out the heater and blind off the fuel gas.
New Delayed Cokers have interlocks to prevent the inadvertent valve misalignment that would allow hydrocarbons to be released to the atmosphere. Opening a vent valve on a coking drum is only one of several different potential valve misalignments that could result in a release of hydrocarbons. Interlocks require actuators on the valves that can prevent the valve from opening without the proper permissives. Typically these actuators are electric motor driven and the permissives prevent the operation of the motor. The handwheels have to be controlled as well as the manual operation of the actuator.
Typically, the vent line is routed to a safe location which provides a layer of protection for this hazard. A Layers of Protection Analysis (LOPA) can be used to determine if your safeguards are adequate to reduce the probability of the hazard causing injury. A key piece of this analysis is the initiating cause frequency. The CCPS (Center for Chemical Process Safety) has a large library of books dedicated to process safety including LOPA. In the book on LOPA there is a reference to the frequency of operator errors on a routine procedure for personnel that were well trained, unstressed and not fatigued to be at 10-2 (one in one hundred) opportunities. A 2-drum Coker on a 16-hr fill cycle has 548 venting, draining and deheading opportunities per year. The CCPS value would then indicate that there are 5.5 incidents per year where the procedure was not properly followed. This seems high to me and I think not all incidents would result in an injury. If 1 in 10 of these incidents resulted in an injury, there would be 0.55 incidents per year as an initiating event frequency for a 2-drum Coker, which seems reasonable. A 4-drum Coker would have twice the incidents and a 6 drum Coker would have 3 times the incidents.
Using the initiating event frequency above and using a targeted event frequency of 10-4 or 10-5 injuries per year (This is the risk tolerance your company has for large process safety events) it is not possible to achieve this frequency without an instrumented system. A SIL 1 system would be my minimum recommendation and depending on the risk tolerance the SIS might need to be a SIL 2 system.
FCC slurry has been successfully processed in Delayed Cokers. I have seen catalyst concentrations in the slurry that range from 0.16% by wt (1600 ppm) to 5% (50,000 ppm). The very high solids loading is more likely to cause erosion than the lower concentrations, but the erosion rate from the slurry catalyst content is low.
Normally, FCC slurry is processed at rates of 10% of the total fresh feed to the Coker or less. That dilutes the solids concentration. Erosion is dependent on a number of factors that include solids content, particle size, velocity, flow regime and a number of others. At typical catalyst concentrations the erosion rates are expected to be low. Many Cokers practice on-line spalling and the metal loss during that practice can be very high if velocities are not managed. Erosion from that practice can be much higher than any calculated erosion from catalyst fines.
Processing slurry oil in a Coker requires higher heater outlet temperatures to convert the material all the way to coke as those molecules have already been cracked at high temperature with catalyst, so the conditions in the Coker need to be more severe to fully convert them.
The catalyst fines will be contained in the coke produced in the Coker and will increase the ash content in the petroleum coke.
The typical strategy to prevent corrosion of the fractionator is to have the tower top temperature higher than the water dew point and the ammonium chloride salt point. This type of corrosion is usually caused by wet salts. Ammonium chloride salts that are dry are not corrosive; however, they can adsorb moisture from the vapors and become wet. These wet salts can be very corrosive. Additionally, these salts can cause the pressure drop through that section of the tower to increase such that the tower will flood at lower than design rates. These salts can also plug the tower and prevent liquids from traveling down the trays. Monitoring the tower pressure drop is a way to know if these salts are forming. If they have formed there will be associated corrosion.
The concentration of chlorides and ammonia in the overhead accumulator water can be used to calculate the salt point at the top of the tower. Chloride concentrations of as low as 5 ppm can correlate to a salt temperature as high as 125 C. There are a number of assumptions in that comment. Still, if the Coker fractionator tower top temperature is less than 125 C and the chlorides in the overhead water are over 5 ppm, there is a very good likelihood that ammonium chloride salts are depositing inside the tower.
I usually recommend the tower top temperature be a minimum of 15 C higher than the calculated water dew point. I then calculate the ammonium chloride salt point. If that temperature is higher than the water dew point, I would recommend increasing the tower top temperature to prevent ammonium chloride salt deposition while attempting to reduce the chloride content in the Coker feed.
Improving desalter performance to reduce the chlorides in the feed to the Coker is the first line of defense. Raising the Coker fractionator tower top temperature is the second option. There are ammonium chloride salt dispersants that have been supplied by the various chemical vendors, but those dispersants do not remove the salts from the tower.
I am not aware of any other technique to measure the tower wall thickness while the unit is in service.
I think you are measuring the necessary components to have notification of a change in the corrosion in the overhead system. If you have a cooling water condenser in the overhead circuit you should measure hardness in terms of calcium and magnesium ions in the sour water. A tube leak can put cooling water into the overhead accumulator and that hardness will foul the sour water stripper.
Corrosion in the overhead system requires liquid water, so calculating the water dew point and keeping the tower top temperature at least 25 F higher than that calculated temperature is important to keep water from forming inside the tower and corroding the tower trays or the vessel itself. Sometimes that 25 F margin is not enough, typically due to ammonium chloride salts. Using the information from the analysis of the sour water and the operating conditions of the tower the ammonium chloride salt point can be calculated. As the Delayed Coker generates ammonia as one of the products of the thermal decomposition of the feed, keeping the chlorides in the feed to a minimum by optimizing the desalter in the crude unit is the primary control strategy for minimizing chlorides in the overhead of the Coker fractionator.
Preventing water from condensing until the overhead condenser is the typical corrosion control strategy for Coker overhead systems. At the overhead condenser, wash water is added to force a liquid water phase. Enough water is added to dilute the acidic materials so they do not cause any metal loss. An adequate water wash is also used to prevent damage from cyanides in the overhead system.
It is common to have inspectors measure the overhead piping wall thickness on some regular schedule using UT while the unit is in service. This gives you information on the day of the inspection but does not provide any ongoing information. I have been involved in gluing an UT transducer to the piping and routing that signal into the control room. There is real time wall thickness information from that one point. Having multiple transducers provides a more general metal loss picture.
I am aware of a technique called pulsed eddy current testing that can provide the average wall thickness of the piping between the pulse generator and the detector. I do not have any experience with that technique.
Coke drums will last as long as you work on them. The amount of work that they require increases with time. If you are trying to prevent a through wall crack while the drum is in service, the amount of inspection and repair increases dramatically as the drum age.
For coke drums making shot coke or bonded shot coke, the number of cycles to the first through wall crack has ranged from a low of about 2000 cycles to a high of 5000 cycles. Normally, the first through wall crack occurs when the drums have accumulated approximately 70% of the maximum number of cycles for the material. Drums in shot coke service will normally be replaced by about 8000 cycles. Replacing the damaged material or performing weld overlay will change the number of cycles the drum has left before it is economic to replace.
Cokers producing sponge coke normally have lower peak stresses during the cycle so drums on those units normally experience between 5000 and 7000 cycles before the first through wall crack with a maximum of approximately 10,000 cycles before the drum is replaced.
In the late stage of coke drum life, there can be a large number of cracks that have to be repaired before the drum or drum section is replace. The number of cracks that occur is exponential with time, so even a slight delay in doing the repairs or replacement can result in a lot of through wall cracks. Mostly, these cracks are identified during the water quench phase, still some are not identified until the drum is in the coking phase and there is a release of hydrocarbons that has caught fire in some of those instances.
There are a several drums in service that have used the CB&I vertical plate technology. The thinking was that the circ welds in a normal plate arrangement were contributing to the cracking of the drums as the circ welds were slightly different metallurgically than the base metal. These drums have been in service less than 20 years and depending on the cycle time have roughly 5000 cycles (some more and some less), so it is likely too soon to tell if this design provides a difference in overall drum life.
Some of the drums with this design have had bulges and cracks, similar to drums with the conventional design. The stresses imposed on the coke drum during the water quench phase may be high enough that the orientation of the plate and the number of circ welds does not influence the fatigue damage. The drum wall thickness is a factor in the fatigue damage. The type of coke produced, the length of the fill cycle and how well the drum was warmed up prior to switching feed back into the drum all impact the fatigue damage on the drum.