Advances In Delayed Coking

ADVANCES IN DELAYED COKING By Norman Lieberman – Chemical Engineer Dealing With Shot Coke – Shot coke is produced when the ratio of asphaltines in coker feed divided by total coke production exceeds a specific value. This value depends on factors favoring shot coke production such as: • • •

Agitation in coke drum. Salt content of coker feed. Exposure of hot coker feed to oxygen.

The main difficulty in handling a drum of shot coke is the sudden and violent dumping of the entire drum contents as the bottom head is removed. Even with modern, remotely operated bottom unheading devices, the abrupt emptying of an entire coke drum is best avoided. Angle of Repose – One simple way of retaining the shot coke in the coke drum is to take advantage of the angle of repose of the shot coke. The ideal (actually used in commercial delayed cokers) described by one attendee is shown in Figure I. The 1/2" x 1/2" ring is welded at the top edge of the 72" I.D. bottom head opening. This simple idea is also the basis for the familiar Roman Arch. One attendee showed a video of what many consider the state-of-the-art in bottom unheading. In addition to the now common automated features, this method incorporates additional automated features: • • •

Unbolting and sliding back the bottom head. De-coupling the Grey-lock fitting on the feed line. Raising the telescoping chute.

The new devise has a steel cylinder which rises from the bottom deck, which encloses the above components from unheading through coke drum cutting. The presenter of this portion of the seminar felt sure that it had already prevented several accidents and injuries in the past year or two. Why Some Coke Drums Dump? – Most drums of shot coke mainly stay in the drum – at least until the pilot hole is cut. Some do not. Even with the protective cylinder described above (see Figure II), dumping an entire drum of shot coke can do mechanical damage or create coke handling problems. One primary reason discussed during the seminar for shot coke drum dumping is the use of the top quench water. If difficulty is encountered in forcing quench water in the bottom of the coke drum, some operators unwisely use top quench water. The top quench water may continue to percolate down through the coke after the bottom head is removed. As the use of top quench water is certain to create hot spots in the coke bed, the descending water may flash with explosive force to steam. The resulting surge in steam pressure will blow shot coke, boiling water and steam from the bottom head. To a lesser extent, poorly drained coke drum, which has been incompletely quenched, will act the same way. Sometimes the decoking crew will add top water while removing the top head to suppress steam evolution. The use of top quench water is potentially dangerous during the interval after the bottom head is removed and the telescoping chute is raised and secured to the coke drum bottom head. One accident described at the seminar apparently occurred due to such inappropriate use of top quench water.

Dealing With a Partly Coked Drum– A rather serious accident resulting in a number of fatalities occurred in the last year or two when unheading a partly coked drum. Loss of feed to a coke drum before the drum is up to a reasonable coking temperature is a common problem. For virgin vacuum resid to turn into solid coke requires a coke drum vapor line outlet of at least 775°F and 45 minutes residence time (790°F and one hour would be a much safer goal). The best way to finish coking a drum is to put the drum on gas oil recycle for a few hours while maintaining vapor outlet at 800°F. However, what would one do if the feed line is plugged. This can and did happen in the accident noted above. The 920°F coker feed solidified in the inlet line coincident with the power failure which was followed by loss of steam. Adding top water may help – but would the top water penetrate into the semi-coked liquid in the drum? Partly coked virgin resid has the appearance and consistency of black, brittle glass at temperatures below 300°F. In contact with water if may then be impermeable. Adding top water may not quickly cook the drum’s contents. The top water may create an additional safety hazard when the bottom head is removed due to undrainable boiling water that is trapped in the drum. Auto-ignition– The heavy, semi-coked liquid in the drum will have an auto-ignition temperature somewhat above 300°F. This means viscous liquid drained from the coke drum will burst into flame upon exposure to air without any external source of ignition. The coke drum should be cooled below its auto-ignition temperature. However, cooling it too far will cause the drum’s contents to become so viscous that drainage of water will be impossible. The attendees at our seminar did not have a complete solution to this puzzle. But certainly it is worth thinking about before the event. In the event the drum was unheaded (after charging for one or two hours), after waiting 1-1/2 days. The contents of the drum blew out (likely due to water in the drum flashing to steam) with some force and auto-ignited. The author calculated that at least ten days were needed before there was a possibility that the drum could have been opened if only ambient cooling was used. More rigorous calculations indicated that a period of 100 days or more was needed for ambient cooling to bring the contents of the drum below its auto-ignition temperature. Side Entry Feed Nozzles– Some new coke drums are equipped with several smaller quench nozzles located above the bottom skirt attachment. The suggestion was made that such nozzles could be used to add steam and water to a coke drum with a plugged bottom feed nozzle. Unfortunately, those attendees who have such nozzles report that the connections are either plugged or have suffered from mechanical failures due to thermally induced stresses. Automated Switch Valve– All new cokers are built with a motor operated switch valve. The days of the WilsonSnyder manually operated, three-way switch valve appear to be numbered. Discussions with some operators indicated the following problems with the MOV switch valve: • • •

•

About 2,100 lbs./hr. of purge steam are used to retard coke formation in the valve. This flow is a significant incremental load for coker fractionators. Switching drums quickly promotes foam-overs. But switching slowly promotes coking of the MOV valve. Switching speed could be more easily controlled with the Wilson-Snyder valve. The MOV valve at some point during the switch creates a larger delta P then the Wilson-Snyder valve. This is of little consequence at the coker heater inlet. However, safety relief valves located at the heater outlet may be forced open due to increased back pressure during drum switching. A partly coked transfer line or a high coke drum pressure may magnify the effect of the high switch valve delta P. One operator with 15 years of hands-on experience with the Wilson-Snyder switch valve and five years of experience with the new MOV automated switch valve questioned the benefits of the new valve. Other than reducing physical effort, this operating supervisor did not see any advantage to employing the MOV valve.

Indeed he felt that a more controlled drum switch (i.e., better control over the rate of fall in the coke drum pressure) was obtained with the Wilson-Snyder manually operated valve. Vapor Valve Failure– Perhaps the most valuable portion of inter-company gathering occurs at the cocktail hour. Here is one recent, but not well publicized incident: • • • • • •

One of the two vapor valves from the coke drum to the fractionator was inoperable and had remained in an open position for many switches. Both vapor valves were electrically operated. The MOV to the working vapor valve was not de-energized. The valve had been closed and its associated coke drum had both heads removed. The MOV was accidentally switched on. The valve opened and permitted 800°F vapor to flow into the top of the open drum. The hot vapors auto-ignited. Air, drawn up through the bottom head due to the resulting draft, increased the intensity of the fire at the top head. In the course of replacing the bottom head to control the fire, an operator was injured when flames escaped from the bottom of the coke drum.

Coke drums are exempt from normal blinding requirements because of the use of double block valves plus "cushion steam" as shown in Figure III. If one of these block valves cannot be closed, it is the same as opening an unblinded vessel. Cycle Time– The attendees were polled as to the cycle time employed on their units. Responses indicated: • • • •

Standard 24 hours – 15% 18-22 hours – 35% 14-16 hours – 40% 11-12 hours – 10%

The best performance for minimizing coke drum cycle time is on a two drum coker with 27’ diameter coke drum running (but not consistently) on a 10-1/2 hour cycle. Shot coke is produced 100% of the time. Some of the prerequisites for short coke drum cycles are: • • • • • • •

A clean, dry source of air to power the drill stem turning air motor and the drill stem hoist air operated motor. Wet, contaminated air will damage the brass bearings in these motors and interrupt the cycle. Preventing plugging of the blow-down quench tower air cooled condensers. Plugging of these exchangers creates back-pressure on the coke drums and extends quench time. A large (10" to 12") atmospheric steam vent to be used at the end of the quench cycle to rapidly de-pressure the quenched drum. A remotely operated bottom unheading device that permits safe unheading even when the coke drum is not properly drained after quenching. A totalizer on the quench water flow to determine if the correct amount of quench water has actually been added. This is proof-positive that the drum is cool enough to unhead and cut. Sufficient capacity of the wet gas compressor motor driver to permit steaming out the filled coke drum to the fractionator at a high steam rate. A secure place for the decoking crew to shelter from drum blowouts when drilling the pilot hole in a marginally quenched drum of shot coke.

Heavy Coker Gas Oil Quality– Based on field data the following was reported during the seminar regarding entrained residual components in the HCGO product: • • • •

The conradson carbon, metals and asphaltines observed in HCGO is due mainly to entrainment from the coke drums and not volatile components. Assume the con carbon of a HCGO is 2.0 wt. % with no wash oil sprays and no de-entrainment grid. Adding open sprays with no grid so that 4 wt. % on fresh feed of recycle is generated, will reduce the con carbon to 1.0 wt. %. Adding 36" of a heavy duty (but not flow-through) grid plus a spray wash to induce 4 wt. % on fresh feed of recycle, will reduce the con carbon to 0.5 wt. %.

One attendee noted an interesting method to curtail the observed con carbon in HCGO. On a two-drum coker the HCGO was only sampled shortly after the warming of the empty drum commenced. For a period of 30 minutes the reduced vapor flow to the fractionator improved de-entrainment and hence the HCGO color. Caution – the color and quality of heavy coker gas oil will vary by a very significant degree during different portions of the delayed coking cycle. Enhanced Liquid Yields– Reducing coke drum pressure by 6 psi will increase liquid yields in a delayed coker by roughly one liquid volume percent based on fresh feed. One cost effective method to reduce the coke drum pressure noted at the seminar was improved rates of condensation in the fractionator overhead condenser. In many cokers, 1/3 to 1/2 of the pressure difference between the coke drums and the suction of the wet gas compressor results from fractionator condenser pressure drop. Periodic slug washing the condenser tubes to remove accumulated, but water soluble, ammonia salts and external detergent washing of the air fin–fan tubes will enhance condensation efficiency and lower the condenser pressure drop. For one, two-drum coker the observed delta P through the condenser was reduced from nine to six psi. The calculated incremental liquid yield increase was 180 BSD of HCGO product at the expense of shot coke and dry fuel gas. At current crude prices this is worth roughly one million dollars a year in increased product value. The source of the water for the slug washing may be the steam condensate in the water draw-off boot from the fractionator reflux drum. This will avoid creating incremental sour water. Also, the required pump already exists, so that only minor piping changes are needed for the slug washing operation.

Carry over - emergency handling

1. Check that the antifoam system is working, and increase its rate to four times the normal as a maximum. Cut out sludge injection if being used. 2. Raise combination tower pressure 3 5 psig to pressure up coke drum(s) and slow down the carryover. 3. Reduce the furnace charge 25 percent every five minutes until foaming subsides or until the minimum furnace charge rate is reached. (Increase furnace velocity steam when furnace charge rates approach 50 percent of capacity.) 4. Start or increase diluent into the tower bottom. 5. Switch out of the full drum as soon as the new drum is hot enough. (Make a smooth switch to avoid a combination tower upset, which could aggravate the situation.) 6. If a drum switch cannot be accomplished, trip the furnace and charge pump; or cut the furnace coil outlet temperature rapidly to 750 775°F, bypass the full drum, and put the unit on circulation. A temporary outage is preferable to the alternative of extended downtime for decoking.

Good Practices 1. Eliminate the feedline flanges or couplers. 2. Industry has moved to automating (Hydraulic cylinders for lifting and lowering chute - this gets rid of chain falls) 3. How do you deal with a drum that has started to fill, but did not coke: • • • •

Maximum steam into drum, if you have steam. While steam system is sagging, it's time to put in steam water. Go back into the drum with feed, and you will get back in. Rock feedline, alternating steam/water, steam/water to open.

Coke Drum Foam-Overs Causes & Cures By Norm Lieberman There are two types of coke drum foam-overs - bad and very bad: • •

Drums that carry-over during filling - Bad Drums that carry-over after being filled -Very Bad

Results of a drum carry-over are quite variable, depending on how much coke has been carried into the fractionator, and now the fractionator is designed to deal with carry-overs. On most newer cokers, a circulating fractionator bottoms pump, and an external filter permit modest amounts of coke (a few tons) to be extracted from the fractionator in a day or two. A typical 12' diameter fractionator can tolerate a single carry-over of 20 or 30 tons of coke. Carry-over amounts greater than these amounts will: 1. Cause cavitation of the heater charge pump 2. Increase heater tube skin temperatures. 3. Knock the fractionator off-line. Carry-Over After Switching I've had a wonderful opportunity to study this problem in great detail last month. My client has four coke drum density level indicators, drum top pressure indicators and the combined drum outlet temperature located on the switch deck. This allowed me to observe the response of the coke drums, as the operators manipulated the drain, steam, switch and vapor valves. This coker has a tendency to carry-over on almost every switch. My observations: Made after watching half a dozen switches indicated. • •

• • • • •

•

Foam-overs occur from the full drum, due to a loss in drum pressure. Pressures must be measured at the top of the coke drum, not 100 feet away at the pressure gauge located upstream of the vapor valves. For a few minutes after the switch, these pressures may be moving in opposite directions. A drum pressure drop of one psi in two minutes, is sufficient to initiate a foam-over. Once started, foam-overs are harder to suppress, then if never allowed to start in the first place. Rapid coke drum switching, does not necessarily promote an increase in foam front height, if the operators closely controls the full coke drum pressure. A 10% - 15% reduction in coke drum density, at the lower drum level indicators proceeds, by a few minutes, a drum carry-over. However, this cannot be used as a warning for action. It's already too late. The drain valve from the warming drum must be closed before switching drums. However, closing the drain valve 3 or 5 minutes before switching is looking for trouble. The warm-up condensate can fill the feed line to the empty coke drum with lighter hydrocarbons. Then, when the 920°F resid hits the 300°F to 500°F condensate in the feed line, the resulting surge in vapor pressures-up both coke drum. If the pressure in the full coke drum then is permitted to drop back to it's pre-switch pressure, a foam-over from the full drum is likely. Naturally, an incompletely drained drum may cause a massive pressure swing. At the now defunct Western Slope Refinery, I observed a 15 psi pressure swing due to a poorly drained drum. The resulting coke drum carry-over knocked the heater charge pumps off-line. Maintaining an ascending pressure at the top of the full coke seemed to be the best way to suppress foamovers. For a period of 15 minutes, starting with the time the drain valve is closed, the full coke drum pressure ought to be increased by "x" psig. But what does "x" depend on? A typical valve for "x" is one psi per five minutes. However "x', the rate of the increase in the pressure drop profile needed to suppress foamovers, is a function of: 1. Coke drum outage. A small outage being 18' to 22' 2. Low coke drum top temperature (i.e. less than 815°F) promotes unstable foam fronts, which requires a more positive ascending pressure profile to suppress. 3. Lighter coker feeds, slop in feed, low coke drum pressure, high recycle rates, excess steam in the heater passes, and other factors that increase drum vapor velocities, require a larger "x" valve. 4. Many of my clients use a tremendous amount of steam in the coke drum structure, most of which goes into the coke drums, which increases velocity.

Many operators will double the amount of anti-foam injection a few minutes before switch time, then reduce it back to normal, when the drum pressures have stabilized (10 or 15 minutes later). I believe this to be a good practice. Total silicon contamination of coker naphtha will only increase by about 3%. On the other hand, this will not compensate for a sudden drop in coke drum pressure of 2 or 3 psi. Also, once the coke bed in the full drum starts to really fluff-up, a coke drum carry-over can't be positively stopped with anti-foam injection. To achieve an ascending pressure profile, the operators have a number of choices. My favorite one is to just pinchback on the vapor valve from the full drum. This requires a local pressure indication from the top of the coke drum, transmitted down to the vapor valve. A more conventional method is to hold back pressure with the combined vapor line warm-up valve, which is often an HIC control valve. This method however, has the disadvantage of increasing the pressure of both the full and empty drums. That is, five times as much coke drum volume has to be pressured-up, as compared to my favorite method. Increasing the fractionator pressure, by slowing down the wet gas compressor, and/or reducing the HCGO pump around rate also works. But this involves increasing the pressure of the fractionator and both coke drums. This is even less responsive than the previous suggestions. Regardless of the method chosen to obtain an ascending drum pressure after the switch, the hotter the coke drum vapor outlet (upstream of the overhead vapor line quench), the more stable the coke bed will be. Also, as liquid yields increase (about one volume percent for every increase of 8°F in drum top temperature), the higher temperature off-sets the loss in liquid yield due to the 40 minutes period, when the coke drum pressure is higher than optimum. (Drum pressure should not be decreased until the steamed drum is lined-up to the blow down system). Steaming out the coke drum When the "little" steam (3,000 - 4,000 lbs./hr) is introduced into the coke drum, the foam front, at first, is actually suppressed. That's because the steam will, for 10 or 15 minutes, cause an increase in the pressure in the drum being steamed. However this effect only lasts a short time. Then, the pressure in the drums will start to fall, unless the operator intervenes, to maintain an ascending pressure profile. If the empty coke drum being switched into is cold (300°F condensate outlet), then the tendency to lose pressure, in the full drum being switched out of, is greatly enhanced. That's because the combined vapor flow from both drums will be low, until the empty drum heats to full coking temperatures (790°F). The combined effect of steaming a full drum, when the pressure in the drum is falling, can get pretty ugly. But, simply shutting off the "little" steam, once the coke bod begins to expand will - well, start polishing up your resume, because that will accelerate the rate of pressure decline, and pull the drum over. Coke Drum Carry-Over While Filling This occurs less frequently, then a foam-over after the switch. Often, the anti-foam is not added properly; diluent is added, but no silicon; a wrong anti-foam flow is displayed; the anti-foam injection point is plugged. Or "your cup runneth over." The gravity of the coker feed (°API) drops, and the operators neglects to cut coker feed. The bearings on the drill steam hoist motor of "A" drum are gone, and the "B" drum cycle needs to be extended just a little bit too long. Properly calibrated K-ray level detectors give adequate warning to the alert operator. Some refiners have vibration probes (like those used to detect pump shaft vibration) on the overhead vapor line. At least for shot coke, this will

identify when a coke carry-over begins. Reducing the heater outlet temperature, by 5°F to 10°F will temporarily stop the carry-over, by reducing the vapor generation in the coke drum. Summary Maintaining a gradually increasing pressure in the drum to be quenched, of several psi, for about 15 minutes after the switch, will off-set to some extent, marginally low drum temperatures; shortened coke drum cycles; low drum outages; higher feeds; low coke drum pressure and rapid coke drum switches.

Signs of foam 1. Early, unusual or sudden drum level detector activity by any one of the drum level detectors. 2. Unusual blips, sudden or even gradual increases in coke drum pressure (sign of liquid/foam carrying over and briefly restricting normal vapor flow). 3. Increase in wet gas production (causes higher drum velocities and foaming). 4. Sharp increase(s) in combination tower level. 5. Decrease in fresh feed when recycle coking (reductions caused by material entering the tower from the coke drum overhead line). 6. Decrease in combination tower chimney temperature (temperature will drop as less drum vapors and more heavy liquid enters the tower in its place).