REFINERS EXCHANGE EXPERIENCES ON FCC PROBLEMS, COKING OPERATIONS

At the most recent National Petroleum Refiners Association question and answer session, refiners and technical specialists exchanged experiences on, among others things, mechanical and operating problems of heavy oil conversion units such as fluid catalytic crackers (FCCs) and cokers.

Subjects receiving much attention included:

• Common operating problems of FCC electrostatic precipitators

• Methods of cleaning FCC turbine blades

• Factors affecting coke quality

• Advanced process control of coking units.

The 1993 NPRA Q&A session on refining and petrochemical technology was held Oct. 20-22 in Dallas. For details on the proceedings see OGJ, Mar. 28, p. 41.

What are the main operating problems with electrostatic precipitators (ESPs) in FCC flue gas systems? What are the main maintenance problems? What preventive maintenance (PM) is required?

Osborn: We have experienced the following operating/maintenance problems over the years with our precipitators:

• Inlet slide gates stuck due to catalyst fines collecting behind the seals.

• Wires breaking, causing loss of efficiency and/or shorting of field.

• Wearing of keyhole support and/or wearing of the button at the top of the wire due to wires spinning while in operation.

(Some have worn to the point of failure, causing the wire to fall through with loss of efficiency and/or shorting off. Repairs included welding new keyhole bars over the originals, tack-welding the buttons to the keyhole, and using heavier wire weights).

• Bridging of catalyst in the bottom of the precipitator.

• Mechanical failure of conveyor augers.

• Leaks in the steam coils on the hoppers. (We discontinued the use of these heating coils and have looked into replacing them with electric heaters, but have not installed them vet due to cost considerations.)

• Cracked insulators.

Preventive maintenance includes routine inspections during turnarounds, sometimes using outside experts. After 12 years of use since it was built, we replaced all the wires at our last turnaround and plan to change them every 6 years, which is every other turnaround.

Brierley: Our main process operating problems are variable solids loading and variable gas loads, particularly SO3 and ammonia. The main maintenance problems are caused by leakage through the hopper valves.

Air ingress through the valves leads to condensation, corrosion, and hopper pluggage. The moisture also contributes to cracking of the porcelain support insulators.

Hopper valves are no", replaced "on the run" on a planned basis, and maintenance on the external electrical components is conducted on the run on individual transformer rectifier systems.

Murphy: Nalco FuelTech has experience with ESPs on FCCUs in one refinery in California. Here, the FCC CO boiler is treated with NOxOUT. The NOxOUT system has been run in a manner as to generate more ammonia slip than would normally be done.

This additional ammonia slip served to augment the existing downstream ammonia injection. It is believed that the ammonia released by the NOxOUT enhanced the precipitator performance because of the longer contact time of the ammonia and the catalyst fines.

Paules: Some of the main problems associated with electrostatic precipitators are smaller-than-normal fines, excessive flue gas rates, lower-than-design operating temperatures, drain-leg plugging, and bad rapper control.

Precipitators must be thoroughly purged to remove SO2 before cooling to the dewpoint or severe damage to the grids can occur. Some refiners have installed digitally controlled rapper systems to more effectively clean the grids during normal operation.

At one of our refineries, over the course of time, the ESP performance will deteriorate. The ESP is then bypassed, the grids are cleaned with a water jet spray and then returned to service. This procedure apparently removes very small fines which tend to build up on the grids and affect performance.

Pedersen: At the Mongstad refinery we have been operating a 40,000 b/sd UOP resid catalytic cracking (RCC) unit for approximately 4 years, with a set of electrostatic precipitators downstream of the CO boilers, followed by a seawater scrubber. We have not experienced any serious problems with those precipitators and maintenance required has been very little.

We do not have any preventive maintenance program in place yet. However, as we gradually have increased the throughput above the original design level, the particulate content in the flue gas leaving the precipitators has increased. Over the last 2 years, we have experienced fines collecting in the flue gas system between the scrubbed and the stack, at times causing sour particle emission.

Roy: Most of the problems experienced in our ESP operation are a result of catalyst buildup in the hoppers, which, when overfull, causes the electrical grids to go to ground. We have had to do some shell crack repairs in the vicinity of our precipitator vibrators.

We have also completely corroded the ESP blowers due to catalyst buildup. The buildup also causes vibration problems.

Hoppers need to be kept insulated with the steam tracing in good working order in order to keep moisture from condensing inside the hopper. Normal monitoring of the blower and motor operations, including periodic checks of the vibrators, constitute part of the PM program.

J. Williams: One of the prime operating problems of ESPs is their tendency to drift from the optimum settings of the rapper system with changing unit conditions.

Optimum conditions should be reestablished by the instrument shop in a routine preventive maintenance program. In addition to checking the actual function of each rapper, the optimum setting for rapper frequency, intensity, and duration should be verified for changing conditions.

Kenneth A. Peccatiello (Grace Davison): We have seen where high temperatures on the ESP itself, being either from a CO boiler problem or some other effect, cause the plates to warp. Once the plates warp, it becomes very easy for the wires to short out against the plates.

The other problem with a high temperature is that people sometimes try to control the ESP inlet temperature with the use of flue-gas duct spray boiler feed water. If there is not enough of a knockout area, the spray water will also end up shorting out the wires.

Joseph W. Wilson (Caltex Services): Mr. Paules, you mentioned that you bypass the electrostatic precipitator. Do you do this on-line? What type of hardware do you have to do this; i.e., the bypassing equipment? Can you actually get in there? Do you bypass it effectively where you can open it up and do the wash?

Paules: Yes, that is right. Our ESP has an isolation valve on its flue gas inlet and outlet and a bypass valve in a flue gas line around it. The block valves are 66-in. butterfly valves and the bypass is a 44-in. butterfly valve.

When taking the ESP out of service we first open the by ass valve then block the inlet and outlet isolation valves. Blinds are then installed on the ESP side of the isolation valves before any work on the ESP begins.

The ESP will cool to a temperature that will allow entry in less than 1 day by simply removing the man-ways and allowing it to radiate heat to the atmosphere. It should be mentioned that we do go out of compliance on the stack opacity while we perform this procedure.

Osborn: We also bypass; we have two banks of precipitators and are able to bypass one side using the main slide gate to isolate.

TURBINE BLADES

What is the panel's experience with thermal shocking of flue gas turbine blades to remove catalyst fines and recover turbine efficiency? How is the quench (shock) done? How fast ( F./hr)? For how long? Do vibrations (wheel/coupling) increase during shock cycle? Will this procedure remove catalyst adhering to stator blades?

O'Brien: Our procedure calls for starting with an initial expander inlet temperature of about 1,320 F., and cooling at a rate not to exceed 100 F./hr, to the range of 1,120-1,130 F. Typically this cooldown can be done using the flue gas quench system.

Vibrations typically increase during this procedure as the fouling deposits come off the blades, and then decrease when the blades are clean. We have had to reduce charge and decrease the expander inlet temperature to near 1,000 F. to clean the expander blades well. It is important to cool slowly to avoid thermal stress and increased cycle fatigue.

Critical monitoring includes rotor-casing, vibrations, bearing temperatures, and flue-gas pressure and temperature. Operators should be familiar with procedures in the event an emergency arises.

Visual inspection of the rotor prior to and after cleaning should be made to determine the effectiveness of the thermal shock. We have had no success with mini-quenches that others have used or walnut hulls.

Paules: We used to use thermal shocking of our flue gas turbine to remove catalyst fines, but have discontinued this procedure. The procedure was to inject quench water in the flue gas outlet and drop the expander inlet in 50 F. increments over 20 min and hold at each increment for an additional 15 min.

This was down done to an expander inlet temperature of 900 F., where this temperature was held for 1 hr. The procedure was reversed to bring the unit back to normal operation.

Vibrations occurred during the procedure and many times afterward due to uneven spalling of the catalyst fines and the resulting imbalance. This is why the procedure is no longer used.

Rajguru: I have input from a Gulf Coast refiner. It says, "we use two procedures to remove catalyst from the shroud: walnut blasting and thermal shock."

Thermal shock procedure is designed to cool inlet temperature to the expander to 600 F. by reducing regenerator bed temperatures. To achieve this temperature, the feed is removed from the unit.

The drop in temperature is about 300 F./hr. Temperature is maintained at 600 F. for 1-3 hr, or until catalyst is spalled from the shroud. Inspection is performed using a visual inspection of the shroud to ensure the procedure was effective prior to putting feed back into the riser.

Roy: At our Pasadena plant we have tried thermal shocking of expander blades to remove catalyst deposits. The latest attempt involved lowering the temperature set point on the flue gas quench control by 100 F./hr, trying for a net lowering of 400 F.

We only achieved about 200 F. lower temperature, but recovered 5-8% of our lost horsepower. The whole process may take 6-8 hr, but usually depends on prevailing conditions or operation at the FCC unit.

Some increase in vibration may occur during the spalling due to additional imbalance. This should be closely monitored with a vibration reading and video camera, or photographs taken from the view ports so as to avoid any blade-tip rub. Spalling is expected to remove catalyst from stator blades, but is not in view from our inspection windows.

COKE QUALITY

What feedstock analyses are used to predict delayed coke quality such as bulk density, volatile combustible material (VCM), and fines production? What processing changes can be made to improve coke quality?

Roy: We have observed VCM to be the only parameter in the question that is affected by feedstock property, namely API gravity. The higher the feedstock API gravity, the higher the VCM for a constant cokedrum vapor line temperature.

Decrease of 1 wt % VCM requires an increase of 7-9 F. coke-drum vapor line temperature. Decreasing cycle time by 6 hr increases coke VCM by 1 wt %.

Bulk density of the coke is a function of the amount of steam in the heater. The higher the steam, the greater the bulk density. Fines production is a function of the amount of water used for drilling. If the drill nozzles are worn out causing less back-pressure on the pump and hence more water to flow for a fixed drilling cycle time, more fines will be produced.

For other processing changes that can be made to improve coke quality, please look up Norm Lieberman's article in Oil and Gas Journal, Mar. 10, 1986, p. 53.

Rules of thumb for delayed cokers:

1. Each 8 psi reduction in coke-drum pressure reduces coke yield on feed by 1.0 wt %. (the term "feed" refers to fresh resid).

2. Each 8 psi reduction in coke-drum pressure increases liquid yields by 1.3 vol % on feed.

3. Each 10 F. increase in coke-drum vapor line temperature increases gas and distillate by 1.1 vol % on feed.

4. Each 10 F. increase in coke-drum vapor line temperature decreases coke yield by 0.8 wt % on feed.

5. Each decrease of 1.0 wt % VCM requires an increase of 7-9 F. coke-drum vapor line temperature.

6. Reducing recycle by 10% on feed reduces coke yields by 1.2 wt % on feed.

7. Reducing the virgin gas oil content of coker feed by 10% reduces coke yield by 1.5 wt % on feed.

8. Reducing coke yield by 1.0 wt % on feed raises liquid yields by 1.5 vol % on feed.

9. Decreasing cycle time by 6 hr increases coke VCM by 1.0 wt %.

One item that is not mentioned in the article is iron in coke. Reducing the crude atmospheric-tower overhead corrosion rate and reducing iron coming into the crude by proper desalter operation can significantly decrease iron in the product coke.

Pedersen: We produce anode-grade coke with relatively low sulfur and low reactivity specifications. We are concerned about the density, viscosity, Conradson carbon residue, sulfur, sodium, calcium, and vanadium in the feedstocks.

Feedstock density, viscosity and Conradson carbon are relevant to predicting product yields. Sulfur, sodium, calcium, and vanadium are of concern to keep sulfur and reactivities within coke specifications. Increasing the aromaticity in the feed will directionally decrease the bulk density of the coke.

Soft coke and fines production are usually associated with high-VCM coke. We like to keep our coke VCM between 7 and 9 wt %.

Results from pilot plant tests show that coke VCM depends upon the following coking conditions:

• Drum vapor outlet temperature

• Drum operating pressure and recycle ratio.

Decreasing the drum outlet temperature or the recycle ratio increases the core VCM. Lowering the drum operating pressure has a small effect toward production of harder coke.

Drum cycle time is also known to have a significant effect on coke VCM. It is generally accepted that reducing the cycle time by 4 hr. will raise the coke VCM by approximately 1%.

Bonelli: The principal analysis we use for coke quality prediction is the Conradson carbon residue test on vacuum residual charge to the cokers. We have worked with one of our coke customers to produce some proprietary testing methods.

There are other indicators of coke quality rather than just Conradson carbon. I encourage you to go search for them, you will find them.

I have one addition to the previous comments regarding iron and coke.

We made a concerted effort in both of our cokers to reduce the iron content of our coke. We discovered that operating the desalter in a mode to remove the tramp-iron solids that had come in with our crude oil was effective in improving coke quality. However, the solids overwhelmed our API separators and wastewater treatment plant. So we have not been successful at reducing our iron.

I also agree with Mr. Roy's comments on the literature in coking providing some insight in how to improve your coke quality. One other thing-we have found fines production to be a maintenance problem at our facility. It is related to the output water pressure, the decoking water pump, and the condition of the nozzles in the cutting tool. If those are rigorously maintained, it is not generally a problem for US.

Osborn: Certainly, poor condition of the cutting nozzles will sometimes increase fines, as will poor cutting technique. You need to be sure that the stem is continually moving downward and hitting in an area at least 18 in. away from the previous spot.

Additionally, Great Lakes Carbon Corp. performs a test called the coefficient of thermal expansion. They can laboratory-pot coke, crudes, or coker charge, and do that determination and give a good idea of what quality the coke will be for many different crudes.

Jerry Lacatena (Foster Wheeler Corp.): Bulk density, volatile combustible material, and fines production-depend more on coke-drum operating conditions than feedstock quality. For a constant set of coke-drum operating conditions, the boiling range, asphaltene content, and aromaticity of the feed can influence these coke properties. However changes in coke-drum operating conditions can negate some of these effects.

In general, increasing the coking temperature will reduce the VCM, increase bulk density and coke hardness, and reduce fines production.

PROCESS CONTROL

What has been the recent experience with advanced control systems on cokers to minimize operational upsets during drum switching and warm-up phases of the coking cycle?

Laabs: We have had success with this type of advanced control on the delayed coking unit at an affiliated refinery. The coker is a two-drum unit, which is likely to have more severe fractionator and recovery-section upsets than a four-drum unit.

They use fractionator top-reflux flow rate as a measure of tower slump during the switch and warm-up procedures. The advanced control system installed on this unit has been able to control the top reflux rate within 20% of the desired optimum, immediately following the drum switch, when the upset is at its most severe. The overhead product quality control is maintained during this time.

We also have another site that is implementing multivariable control for this service. They are confident that the installation will stabilize operation and improve unit profitability. The project payback is estimated at 1 year.

Bonelli: We also commissioned an advanced control on our delayed coker several years ago to do a hands-off switch. Once the operator initiated coke-drum switch, he could sit and watch the coker switch itself.

One of the problems we had was that coker operation is not the most stable in the refinery and it was frequently difficult for the advanced control to determine what was operating and what was a coke-drum switch, so that was a tuning problem for us.

We also used a number of different variables to have the control initiate the switch. What we found was the control was extremely maintenance-intensive, in terms of requiring a control engineer's time. We no longer use the control because we are not able to economically justify the manpower to keep it up to date.

Sloan: I will just add that the payback on these systems has been so good that a lot of operators are installing them on their cokers almost as a matter of course without going through full economic justification.

Frank J. Kleinschrodt (Setpoint Inc.): I just want to give a little bit of information about how the coke-drum switch controls actually work. What you do is you model the dynamic behavior of the unit in various modes, responding to the cycle key steps, such as the addition of steam, warm-up, cool-down-all the issues.

Then, you must identify some method of detecting the transition; either by some switch, etc., that is occurring in the cycle, or some variable moving. With the models that were identified earlier, a zero-gain impulse is input to the feed-forward models to simulate the disturbance, which is then minimized with the controller.

These can be implemented either in a traditional feed-forward sense (which Setpoint has been doing for a number of years) or in a multivariable controller, as recently done on two coking units.