DESIGN IMPROVEMENTS INCREASE RUN LENGTH
John R. Wood
Petroleum Refining Consultants Christiansted,
St. Croix U.S. Virgin Islands
Clement K. Marino
Hawaiian Independent Refining Inc.
Ewa Beach, Ha.
The Hawaiian Independent Refinery (HIRI) visbreaker recently completed 843 stream days of operation before its first heater decoke and major turnaround and inspection.
This is believed to be a record and shows that, with proper design criteria, longer run lengths and lower maintenance and operating costs can be achieved than previously thought possible.
From this commercial experience it can be seen that it is feasible to design coil-type visbreakers to achieve very long run lengths, and because the coke formation that does occur can be quickly and easily removed, better on-stream time and lower maintenance costs can be achieved.
The capital and other costs associated with soaker drums and the difficulty and expense of removing and disposing of the coke are well known, as are the problems of fuel oil degradation with soaker drums without complex internals.
This demonstrated improvement in run length and on-stream time, together with other improvements made in the past 10 years, gives modern coil design a definite advantage in most visbreaking applications. Many of the concepts employed are applicable to existing units and other processes where undesirable coke formation is normally a problem.
BACKGROUND
In 1985, HIRI decided to consider the possibility of building a visbreaker at its refinery on Oahu. A feasibility and cost study was carried out by Jacobs Engineering Group's (JEG) Houston office with technical advice and support from Petroleum Refining Consultants (PRC). This study provided yield, capital, and operating expense estimates for both coil and soaker visbreaker designs.
After considering all factors, HIRI selected the coil design. The process package was developed in 1986, again as a cooperative activity of JEG and PRC. The engineering, procurement, and construction activities were directed by the HIRI project staff, with engineering and technical support from JEG and PRC. The project was well-managed, and the unit started up without incident, on schedule, and under budget in September 1987.
DESIGN REQUIREMENTS
The design charge was 13,000 b/d of Alaska North Slope (ANS) vacuum pitch with an alternate of Ardjuna (Indonesian) pitch. Because the refinery at times ran certain low sulfur crude which would not be visbroken, the unit had to be capable of running at 8,000 b/sd and have a hydraulic capacity to handle up to 15,000 b/sd.
Intermediate storage was provided for the visbreaker to stay on-line during some of the time that low sulfur crude was being run.
Run length between decokes was specified as 9 months minimum. Because the refinery crude/vacuum heaters were running at capacity, heat exchange was to be provided from the visbreaker to the crude exchanger train. The unit was also designed so that, in the future, a heavy visbreaker gas oil stream could be produced with only minor modifications.
For initial operations, the refinery downstream facilities did not have the capacity to effectively process the visbreaker heavy gas oil. However, the major equipment was sized and designed for heavy gas oil recovery from the visbreaker main fractionator without the need for a downstream vacuum unit.
A heavy fuel oil yield reduction in excess of 20%, without producing heavy gas oil, also had to be met. At the same time, product quality had to satisfy the quite stringent stability and compatibility specifications common in the Pacific fuel market.
DESIGN PRACTICES
To meet the minimum run length requirements, particularly with the wide variation in charge rate, better design criteria and techniques were required than those that had been applied to coil visbreakers in the past. These older units typically experienced short run lengths and high maintenance costs because of excessive coke formation in the fractionator, heater, or transfer line.
Coking can be virtually eliminated in the fractionator by the use of internals that do not provide a location on which coke can form, and by controlling temperatures below critical levels. Proper metallurgy and quenching should be used in certain areas, and in some circumstances, simple flush arrangements may be appropriate.
Since coking in the fractionator is not a problem in a properly designed unit, there is no need for "coke catchers" on the suction of the fractionator bottoms pumps. The HIRI unit, like other properly designed units, does not have them.
It is not possible to completely eliminate coke formation in the heater, or the consequent spalling of some of this coke. It is therefore necessary to provide a simple system to collect any such spalled material to prevent it from reaching the pumps. Because the amount of potential coke spalling from the heater is very small, it is easy to design a system from which coke removal is required every 2-3 years.
HEAT
Excessive heater coking, resulting in run lengths of 9 months or less, can be due to a number of factors. Of greatest significance, perhaps, has been the historical practice of designing visbreaker furnaces with excessive radiant heat absorption rates.
This attempt to minimize the capital cost of the furnace is usually false economy. With today's furnaces using internal return bends, blanket insulation, etc., the incremental cost of adding more radiant tube area is very low.
There are numerous other areas of visbreaker design where truly effective savings can be made to offset the slightly higher heater cost. In all situations the feed temperature to the furnace should be 640 F. or higher, using preheat exchangers if necessary. Additionally, some amount of plug flow soaking can be used, without the use of a soaking drum, to reduce the required heater outlet temperature.
In determining how much soaking should be provided, one needs to consider both the required temperature and heat content of the inlet to the fractionator so that the required products can be produced (possibly including a heavy gas oil stream), and to achieve the overall heat integration of the visbreaker with the rest of the refinery, for energy conservation. For the HIRI design, a fairly conservative heat absorption rate and an intermediate amount of downstream soaking was considered appropriate.
VELOCITIES
Another common design and operating weakness has been the lack of attention to velocities in all circumstances. It is important not only to check the velocity and flow regimes for the design case, but also for reduced throughputs, severities, startups, and shutdowns.
For example, if severity is reduced 25% by reducing heater outlet temperature without any other changes, firing rates will change only slightly.
However, linear velocity, particularly near the outlet, will change significantly.
Injection of water or steam and adjustments to coil pressure can easily be made to satisfy the appropriate criteria under all operating conditions. Additionally, a system must be provided to properly shut down the heater and purge the coil during loss of feed conditions (e.g., power failure).
The same principles regarding velocities should also be applied to the transfer line design.
Finally, certain feed contaminants (e.g., sodium) can have a significant impact on the rate of coke formation. These can generally be reduced to a satisfactory level with little or no cost.
HEAT TRANSFERS
For reasons previously noted, provisions were made to transfer heat from the visbreaker to the crude unit exchanger train. During design, it was determined that there might be some benefit to increasing this transfer of heat, and also to extending visbreaker run lengths beyond the original 9 months.
The increase in heat transfer to the crude unit would result in a decrease in the approach temperature to the visbreaker and a consequent increase in heat absorption rates in the visbreaker furnace. Therefore, it was decided to use Alonized tubes in the radiant section to compensate for these higher rates, which at the same time extended the run length. This change was easily incorporated in the heater, which had already been specified, bid, and the vendor selected.
PROCESS DESCRIPTION
A process schematic of the unit is shown in Fig. 1. Following are comments on some of the significant aspects of the design. The vacuum pitch is run down to a standard (but insulated) cone roof tank at about 300 F.
The transfer pump to the unit is relatively low head, minimizing the pressure ratings and cost of the feed heat exchangers. Even with the low head transfer pump, the charge booster pumps downstream of the exchanger train only needed to be two stage.
To maintain proper velocities throughout the heater and transfer line for the wide charge rate variations anticipated, a back-pressure control valve was provided to vary the coil outlet pressure, and water injection facilities provided at the inlet of the furnace. Overhead condenser water was used because the main refinery steam supply was at an insufficient pressure, and use of condensate would increase boiler feed water consumption and sour water production.
The fractionator and flash zone conditions were designed so that it was possible to use less than 3,000 lb/hr of low-pressure stripping steam to achieve a satisfactory flash point on the visbreaker tar product. Hence, less than 3,000 lb/hr of sour water is generated. No chemical injection was specified or has been used on the unit at any location.
The heater, as designed, is a standard horizontal bottomfired unit, with ceramic fiber insulation and internal return bends. The fractionator was designed and constructed for heavy gas oil recovery, but was not piped for initial operations.
The offgas compressor was a post-design addition, included when it was found necessary to raise the originally specified battery limit pressure requirements for the offgas. It was cheaper, considering the status of the project, to simply add a small booster compressor rather than redesign the unit for a slightly higher operating pressure.
Features not shown on the schematic include a simple gas supply system to purge the furnace during emergency shutdown (e.g., power failures), a cutterstock flush system for flushing the unit, and equipment to facilitate turnaround and maintenance work.
OPERATING EXPERIENCE
The unit charge rate has varied as expected from slightly below 8,000 b/sd to over 14,000 b/sd, with an average rate of about 10,500 b/sd.
As noted previously, because of firing limitations on the crude/vacuum furnaces, it was anticipated that more heat would be transferred from the visbreaker than contemplated in the original design. In fact, this became a practice shortly after start-up, Consequently, the approach temperature to the visbreaker furnace has been significantly lower than design. In turn, the heater outlet temperature had to be increased to maintain cracking severity.
This temperature has averaged about 915 F. The combination of these two factors has resulted in firing and heat absorption rates in the visbreaker furnace of 20% above the original design for any given charge rate.
Severity has been slightly higher than design, and the reduction in fuel oil yield, as a percentage of what would have been produced without the visbreaker, has been about 25%. The stability of the visbreaker bottoms product, as measured by sediment (Shell Hot Filtration procedure or IP375), has been well below the 0.15 wt % specification limit common in the Pacific fuel oil market.
During periods of low sulfur crude runs, despite the existence of visbreaker feed storage, there were 17 visbreaker shutdowns due to lack of feed or rundown storage. Power outages resulted in 11 additional shutdowns.
There were also three shutdowns for other causes: One was due to a motor bearing failure on a charge pump while the spare pump was under repair; another was due to a heater hanger failure, and one just prior to the decoke and turnaround was precipitated by an unexpected tube failure.
Despite these numerous shutdowns and start-ups, the wide variation in charge rates and consistent over-design firing of the furnace due to increased heat transfer to the crude unit, the unit has operated for 843 days over a period of about 2 years and 9 months without decoking or major turnaround. It has consistently met or exceeded yield and product quality design criteria. Aside from the start-ups and shutdowns mentioned above, operations have been largely uneventful.
MAINTENANCE AND INSPECTION
There have been a few, mostly minor, mechanical problems. The conventional angle-type, back-pressure control valve had the usual tendency to coke up around the valve stem, despite purge provisions. By exercising the valve on a regular basis as per normal operating practice, this was eliminated.
Mechanical seal problems on the two-stage charge booster pumps were fairly accurate during initial operations. All the pumps in the unit were equipped with mechanical seals. Those on the fractionator bottoms pumps (which operate at higher temperatures than the charge pumps) were of identical design to those in the charge pump, and have operated without any problems. In fact, one of these bottoms pumps has not required maintenance since start-up.
Minor mechanical modifications to the charge pump seal system have largely resolved the problem there.
Relatively few problems have occurred with the heater. During periods when the unit has been shut down due to lack of feed, the heater has been inspected for evidence of any localized coke buildup or other problems. Pressure drop, skin temperatures, and other operating observations had not indicated any such problems.
X-rays showed no coke buildup and ultrasonic thickness (UT) wall measurements indicated original wall thickness throughout the furnace. Three inspections had been made, the most recent being in June 1990. The tubes in all these inspections showed no signs of overheating or other degradation, and were remarkably straight.
The most significant problem with the furnace has been hanger failures. When the heater was purchased, the originally quoted number of supports was increased by 25%, but due to a communications error, the details of the hanger design were not properly analyzed. This was an oversight because "standard" hanger designs are not an uncommon problem, particularly in cracking furnaces. The original hangers are being replaced, as necessary by supports fabricated at the refinery.
Shortly after the June 1990 heater inspection, a surprising tube failure occurred in the center of the furnace, many tubes away from the outlet. On either side of the failure, there was no apparent coke buildup or tube degradation, but an adjacent tube did show evidence of overheating in the sam area.
The equivalent tubes on the opposite side of the furnace, and, for that matter, the remainder of the tubes, showed no evidence of any problems. it must be concluded that the failure was largely due to a very localized "flame impingement," which is sometimes difficult to visually observe in gas-fired furnaces.
The shielded skin couples provided in the original design have always functioned well; but because none were located in the area of overheating, they did not indicate a problem, In reviewing the performance of the heater, it is necessary to consider not only the wide range of charge rates and the number of planned and emergency shutdowns and subsequent start-ups, but also the fact that for any given charge rate, the firing rate has consistently been more than 20% over the original design.
The large number of shutdowns and start-ups, mostly due to external causes, certainly aggravated the hanger problem.
With regard to coking, experience with other furnaces under similar conditions would indicate that the originally specified 9-month run length would not likely have been exceeded had we not chosen Alonizing to resolve the higher-than-design firing rate.
Since decoking was not necessary until after more than 2 years of on-stream time, it seems reasonable to conclude that Alonizing had the effect of at least doubling the run length.
Skin couple monitoring during operation indicated a very low increase in tube wall temperatures due to some coke laydown, but the amount of coke was so small that it did not noticeably affect coil pressure drop.
it was surmised, and subsequently confirmed, that the small amounts of coke being deposited tended to spall off, either during operations or, more likely, during start-ups and shutdowns.
This reduction in rate of coke laydown and lower adhesion to the tube walls is consistent with the many technical articles concerning Alonizing and the mechanisms of coke formation and deposition, and with commercial experience in similar applications.
The fractionator has been designed anticipating that such spalling would occur. The spalled coke was found where it was expected to collect, and was easily removed. A single and localized coke deposit was found in the splash zone area of the tower, but not of significant size or location as to interfere with operations in any way.
Inspection of the remainder of the unit did not disclose any significant corrosion or other problems of note.
The improvements in the visbreaker have resulted largely from better design concepts and practices, particularly in respect to better heater design criteria for all anticipated operating conditions. Although, as a consequence, the heater is slightly more expensive, the additional costs are more than offset by savings in other areas. Among these we have noted the elimination of the coke catchers for the bottoms pumps, and a flush system, which not only facilitates operations and maintenance, but basically eliminates the need for steam tracing.
Also the need for highhead, multistage pumps and high pressure rating exchangers is avoided, and in many other areas the design has been simplified to avoid unnecessary piping, connections, relief valves, etc. The anticipated advantages of Alonizing have also been demonstrated.
Many of these improvements can be applied to existing visbreakers and other units where coke formation is a problem. In considering what might be beneficial, it is advisable to look at all the options, because every refinery situation is, in some sense, unique and, in some cases quite significant differences may exist. For units with unsatisfactory run lengths or maintenance and other problems, significant attractive improvements can likely be achieved with a concurrent reduction in energy consumption, although in some cases it may not be economically appropriate to extend run length as long as 2-3 years.
REFERENCES
- Wood, John R., "Coil design visbreaker for HGO recovery has advantages," OGJ Apr. 22, 1985, pp. 80-84,
- Albright, Lyle F.. and McGill, W.A., "Benefits of Alonized tubing demonstrated in ethylene plant data," OGJ Aug. 31, 1987, pp. 46-50.
Copyright 1991 Oil & Gas Journal. All Rights Reserved.