TECHNOLOGY Texas plant retrofit improves the throughput, C 2 recovery

June 3, 1996
Joe T. Lynch Ortloff Engineers Ltd. Midland, Tex. Richard N. Pitman GPM Gas Corp. Goldsmith, Tex. GPM Gas Co.'s Goldsmith (Ector Co., Tex.) plant was recently converted from a nominal 90% ethane recovery and 86 MMscfd design capacity two-stage expansion process to a 95% ethane recovery and 135 MMscfd capacity. The project used the Gas Subcooled Process (GSP) design of Ortloff Engineers Ltd., Midland, Tex. The conversion required modification of existing expanders and chillers and addition
Joe T. Lynch
Ortloff Engineers Ltd.
Midland, Tex.


Richard N. Pitman
GPM Gas Corp.
Goldsmith, Tex.

GPM Gas Co.'s Goldsmith (Ector Co., Tex.) plant was recently converted from a nominal 90% ethane recovery and 86 MMscfd design capacity two-stage expansion process to a 95% ethane recovery and 135 MMscfd capacity.

The project used the Gas Subcooled Process (GSP) design of Ortloff Engineers Ltd., Midland, Tex. The conversion required modification of existing expanders and chillers and addition of a plate-fin exchanger, an absorber column, and a set of pumps.

Time from project approval through start-up was 5 months. NGL production was interrupted for 10 days while the plant was down for tie-ins and checkout.

Plant throughput was compression-limited to operation at 130 MMscfd through late 1995. Compression to allow throughput of greater than 130 MMscfd was operational in late 1995.

The demethanizer column and six of nine heat exchangers were reused in the Ortloff process retrofit. The demethanizer internals were changed out in 1995 in anticipation of higher throughput with the new compression. The two expanders were modified for parallel expander and booster compressor operation. Expander replacement was unnecessary.

Similar retrofits of other GPM plants using GSP are currently under study.

Goldsmith plant

Fig. 1 [18769 bytes] shows a block diagram of the Goldsmith plant. The plant includes inlet compression from 5 psig and 50-psig gathering systems, DEA treating to remove CO2 and H2S, cold-bed-adsorption (CBA) sulfur recovery, dehydration, treated-gas compression to 860 psig, cryogenic NGL recovery, residue recompression to 620 psig, and supporting utility systems.

The original refrigerated single-stage expansion cryogenic plant was installed in 1976 with an inlet rate of 78 MMscfd. It was changed to a two-stage expansion design for processing 86 MMscfd of gas in 1982.

The two-stage plant allowed operation at rates up to 110 MMscfd, but with a relatively poor ethane recovery of 70%. At 110 MMscfd, the throughput exceeded the capacity of the expanders; the J-T valves around both expanders were therefore always partially open. Excessive pressure drop existed across one plate-fin exchanger.

GPM wanted to increase the facility throughput and to improve substantially product recovery at minimum cost. Described here are the design review process and changes made to the two-stage cryogenic plant to increase the design throughput. Changes were planned for other systems in the plant to support the higher cryogenic plant throughput, but only the changes to the cryogenic plant and the resulting operation are discussed here.

Redesign checklist

In many cases, the ultimate capacity of the cryogenic unit will set the design rate for modification of all the other plant systems. Following are some details which must be checked during any cryogenic plant redesign study:

  1. Equipment design pressures

  2. Piping and equipment metallurgy and original cold and hot design temperatures

  3. Demethanizer sizing and internals

  4. Expander frame size and lube-oil system limitations

  5. Expander-feed separator separation capacity and liquid surge time

  6. Demethanizer side heater and thermosiphon reboiler hydraulics

  7. Heat-exchanger pressure drops, heat-exchange surface area limitations, and potential for tube vibration in shell and tube exchangers at higher throughput

  8. Pressure drops and/or velocities in the existing piping

  9. Pressure-relief system capacity, control-valve sizing, orifice-plate sizing, and instrument ranges

  10. Compression capacity and horsepower limits

  11. Existing equipment performance problems or known maintenance items

  12. Availability of surplus equipment

  13. Economic basis for recovery vs. throughput vs. capital cost decisions.

Assuming that the goal is to increase both the throughput and the ethane recovery of the existing facility, the study will generally include the following steps:

  • The original process design is simulated with the original inlet-gas composition to establish a check on the original design basis and provide baseline design data if information on the existing equipment is not readily available.

    Any process equipment changes made since the plant was built are then incorporated into the simulation and the design changes confirmed. Good agreement with all available data sheets should be obtained or any deviations explained and noted.

  • Model the current plant operation.

    The feed composition is updated, current pressures and temperatures are input, and the resulting simulation is checked against current field test data to identify any equipment items which do not match the expected performance.

    The simulation is then adjusted until the best agreement between the field data and the model is obtained.

  • The resulting exchanger performance and pressure drops can be used as a starting point for the redesign.

    Any significant differences between the actual equipment performance and the original design must be analyzed and resolved before the study can proceed. For example, current heat-exchanger performance may not meet the original design specifications.

    A decision must be made to use either the current heat-transfer capability for the redesign or to plan on restoring the original exchanger performance during a shutdown.

    For a given process redesign, a trial flow rate and recovery must first be simulated with appropriately factored exchanger heat-transfer (UA) values and pressure drops. The compression requirements are checked to be sure they are reasonable.

    The existing tower sizing is checked and the flow rate and/or recovery changed to accommodate the available tower capacity, if necessary.

    The performance of the expanders and expander feed separators is then checked. If there are no limitations identified at the trial flow rate, the trial flow rate and/or recovery is increased incrementally until the compression, expander, or tower size becomes limiting.

  • At this point, a design rate and recovery are established and each equipment item is rigorously checked. The expanders and plate-fin exchangers are best re-rated by the suppliers.

    Each shell and tube exchanger must be checked for thermal performance, pressure drop, and tube vibration. The demethanizer bottoms pumps and pipeline pumps must be checked for the higher product rates.

All the major process lines must be checked for pressure drop and velocity. The tower side reboilers and reboiler hydraulics are checked. Vessels are checked for adequate separation and liquid surge capacity.

The design-case simulation is then updated with the heat transfer, pressure drop, and efficiency data from the detailed equipment checks, and a list of items which must be either replaced, modified, or reused is compiled.

The external plant systems are checked for the higher rates, and a cost estimate is generated so that the overall project economics can be evaluated.

Additional iterations may be needed to determine the optimum overall plan for the facility as decisions are made regarding compression and throughput trade-offs and the limitations of upstream and downstream plant systems are determined.

Debottlenecking

This general sequence of events was conducted for GPM's Goldsmith facility in late 1993. GPM desired to maximize the throughput of the existing facility and to improve the ethane recovery.

The original DEA gas treating system was capable of treating 300 MMscfd. The limit on sulfur dioxide emissions required that the sulfur plant be converted to Amoco's CBA process for higher sulfur recovery before the plant throughput could be increased.

(The conversion to Amoco's CBA process was also designed by Ortloff as a separate project.)

GPM's specific guidelines for debottlenecking the Goldsmith cryogenic plant included the following:

  1. Rework (but not replace) the existing expand- ers.

  2. Keep the demethanizer tower; change internals as necessary.

  3. Add no refrigeration compression.

  4. Determine the maximum achievable throughput allowing for additional inlet and residue compression.

  5. Be ready for tie-ins during a scheduled plant shutdown in May 1994.

  6. Have the modified unit up and running by June 1, 1994.

  7. Minimize the loss of NGL production by minimizing time required for tie-ins and restart of the modified unit.

  8. Specify any new instruments for use with future distributed control system (DCS) equipment.

Tie-in piping for conversion to the GSP process was routed above the cramped original process skids while the Goldsmith plant remained in operation.

Limitations

The process flow diagram for the two-stage cryogenic unit prior to retrofit with the GSP design is shown in Fig. 2 [32149 bytes]. The design included two levels of refrigeration and two expansion stages.

The demethanizer (C-1) had a trayed upper section and packed middle and lower sections. The side heater (E-8) and reboiler (E-7) were combined into one plate-fin exchanger.

The cold gas exchanger (E-5) and the cold-cold gas exchanger (E-6) were also plate-fin exchangers. All remaining exchangers were shell-and-tube type. The propane refrigeration system included three 880-hp reciprocating compressors.

The two-stage design and equipment were first checked at a 110 MMscfd inlet rate (28% over the original design) and the following changes were determined to be necessary to achieve reasonably good ethane recovery:

  1. Replace both ex panders. In the two-stage process, each expander sees the full cryogenic plant inlet rate (less separator liquids), and each would need to be replaced to permit the J-T valves to remain closed at the 110 MMscfd inlet rate.

  2. Replace both expander-feed separators with larger vessels to avoid liquid carryover into the expanders at the higher rate.

  3. Replace the plate-fin cold-cold gas exchanger (E-6) to eliminate a 40-psi, residue-gas side pressure drop.

  4. Replace two piping runs in the residue-gas piping with larger piping to minimize pressure drop at the higher rate.

  5. Replace the intermediate gas exchanger (E-3) with a new shell-and-tube exchanger to avoid tube vibration and possible fatigue failure from the much higher-than-design shell side residue-gas flow rate.

  6. Add refrigerant vapor nozzles to both chillers to reduce the vapor velocity. The kettle diameters on both chillers were found to be marginal even for the original process conditions. The additional nozzles would reduce the carryover problem.

  7. Add a booster-compressor discharge cooler between the boosters and the residue-gas recompressors.

    Plant operating experience confirmed the bottlenecks determined from calculations. The plant had been operating at 110 MMscfd, but with a higher-than-desired tower pressure resulting in poor ethane recovery. Recovery could be improved if the pressure-drop problems were eliminated and compression were added to support a lower tower pressure.

If all these changes were made to the two-stage plant, the estimated ethane recovery would increase from 70% up to 87% at the 110 MMscfd rate. These changes, however, did not meet the constraints defined by GPM for the project.

Increasing the throughput for the two-stage design to greater than 110 MMscfd was impractical because of pressure drop and tower-capacity limits. Because the debottlenecking changes involved replacing existing equipment, the changes would all have to be made during an extended facility shutdown.

GSP process

Ortloff has previously described1 the retrofit of its patented GSP technology2 3 to industry-standard, single-stage plants. The process has been used extensively in new cryogenic plants built recently and has been widely used in new plants built since 1990.

Simplified process flow diagrams for both the typical industry-standard, single-stage (ISS) design and the GSP design are shown in Figs. 3 [24759 bytes] and 4 [26161 bytes], respectively.

The differences between the ISS plant and the GSP process include the following:

  • Several fractionation stages are used above the expander feed in the GSP design. The design's column top feed or reflux is obtained by condensing and flashing a portion of the high-pressure feed gas.

    In the ISS design, any ethane which is not condensed in the expander outlet is lost to the residue stream, as the upper section of the demethanizer is simply a vapor/liquid separator with the vapors becoming part of the residue stream.

  • The reflux stream for the GSP design is controlled to maintain a constant flow ratio in relation to the inlet-gas flow rate. Cold residue gas from the demethanizer overhead is used to condense the reflux stream.

    The refluxed fractionation stages above the expander feed increase ethane recovery by condensing ethane vapors from the expander outlet that would have escaped to the residue stream in the ISS design.

  • In the GSP design, the flow through the expander is no longer the total plant inlet vapor (less separator liquids) because a portion of the inlet gas is split off ahead of the expander to provide the reflux stream. The flow through the expander is 60-80% of the total plant inlet flow rate for the GSP design.

  • The cold separator temperature is warmer for the GSP design than in the ISS design since some of the cooling available from the residue gas in the ISS design is used to condense the reflux stream in the GSP design.

  • The expander horsepower will typically be higher for the GSP design than for the ISS design because of the warmer cold-separator temperature. This is true even though the mass flow through the expander is less.

    Therefore, when compared to the ISS design, either the external residue horsepower required for a given throughput and recovery will be less for the GSP design, or the throughput capacity for a given external horsepower and recovery will be higher for the GSP design.

  • The introduction of heavier components at the top feed contributes to ethane retention, and the resulting warmer temperature profile increases the tolerance of CO2 in the column.

A retrofit of a single-stage plant with the GSP design can be accomplished (Fig. 5 [30497 bytes]). In smaller plants, the additional packed-bed absorber section can sometimes be added to the top of the existing demethanizer, precluding the need for the cold pumps.

Larger plants are usually retrofit with a separate absorber column and a pair of pumps. (The pumps can be deleted at a cost of around 4% in ethane recovery.)

The absorber functions as an extension of the existing demethanizer and the bulk of the fractionation traffic is moved from the existing demethanizer to the new absorber section. The existing demethanizer becomes significantly unloaded in the GSP retrofit design.

Goldsmith retrofit

The flow diagram for GSP retrofit of the Goldsmith plant is shown in Fig. 6 [34813 bytes].

In the design effort, the plant configuration was first converted back to a single-stage design, then the GSP retrofit previously described was applied.

For the Goldsmith plant, a separate absorber column and pumps were used for the fractionation stages above the expander feed. The expander outlet piping was routed to the bottom of the new absorber, as was the demethanizer overhead. The absorber liquids were pumped to the top of the existing demethanizer through the old expander outlet line.

The cold gas separator (V-2), cold-cold gas exchanger (E-6), cold-gas exchanger (E-5), and the feed flash separator (C-2) were removed from service. The cold-cold gas exchanger (E-6) was not needed for the single-stage GSP design and was too small to be reused in the subcooler service.

The cold-gas exchanger (E-5) added pressure drop to both gas paths without resulting in any significant cooling of the inlet gas because of changes in the plant temperature profile. This exchanger was also too small for the subcooler service.

Because the expander feed separators serve the same function as the cold gas separator (V-2) in the two-stage design, this separator was deleted in the GSP retrofit design.

The two expanders were converted for parallel operation for both the expander and booster compressor services, splitting the inlet flow equally between the two modified machines. Flow to the expanders was split just ahead of the two expander-feed separators.

Paralleling the two expander-feed separators and expanders allowed reuse of the existing expanders, piping, and separators. It should be noted, however, that paralleling the equipment was possible because all the intermediate pressure equipment and piping added during the conversion to the two-stage process in 1982 had a design pressure equal to the existing high-pressure equipment.

For the Goldsmith retrofit, the inlet gas split to the subcooler was taken from the warm plant inlet stream ahead of the inlet exchangers rather than at the expander-feed separators. This ar-rangement reduced the flow rate through the existing exchangers by the amount of the subcooler flow, 27% of the inlet for Goldsmith's gas composition.

An additional warm gas/ gas exchanger (E-10) was required to cool the inlet gas down to the desired subcooler inlet temperature. This exchanger was physically combined with the subcooler (E-11) in a single new plate-fin exchanger.

A sidestream was withdrawn from the residue gas side between the two exchanger sections and routed through the old residue gas flow path consisting of E-3 and E-1 to provide some inlet-gas cooling in addition to that required for the reflux stream. Because the residue-gas flow rate through these exchangers was only a fraction of the original design flow rate, tube vibrations were no longer predicted, and the E-3 exchanger was reused without modification in the GSP retrofit design.

When this design was first applied to the Goldsmith process at a 110 MMscfd rate (that is, the highest practical throughput rate for the two-stage design), significant additional capacity remain- ed.

After several more iterations, a design rate of 135 MMscfd at 95% ethane recovery was selected as the optimum design point for the Goldsmith retrofit. This design point inlet flow rate was an increase of 57% over the nominal two-stage, 86-MMscfd design capacity at comparable inlet and tower pressures, and 23% higher than could be achieved by debottlenecking the two-stage design alone.

The increase in throughput was possible while significantly increasing the ethane recovery over what was being achieved with the unmodified plant operating at a 110 MMscfd inlet rate.

Rigorous checks of the existing equipment at the 135-MMscfd inlet rate resulted in identifying the following required changes in addition to the new GSP equipment:

  • Replace the expander and booster compressor wheels, shafts, and variable nozzle assemblies in both expanders. For parallel operation, modify the two machines to be as similar aerodynamically as possible even though the machines originated from two different manufacturers.

    Modify the lube-oil system of one of the expanders for increased shaft speed by installing a larger lube oil cooler and changing the 1,800 rpm lube-oil pump motors to 3,600 rpm motors.

  • Install two new vapor outlet nozzles on the shell of the intermediate level chiller (E-2) to reduce the vapor velocity in the kettle.

  • Reuse the cold gas separator vessel (V-2) as a low-level chiller vapor separator to avoid welding additional nozzles to the low-temperature carbon steel shell of the low-level chiller (E-4).

  • Replace the existing plate-fin reboiler exchanger (E-7) with a larger shell-and-tube reboiler exchanger (E-12). Reuse the side heater (E-8) section of the existing exchanger as-is.

  • Replace the trays in the upper section of the existing demethanizer (C-1) with packing and increase the packing size in the middle and lower packed sections.

  • Replace the demethanizer bottoms pumps (P-1) to accommodate the substantial Increase in NGL product flow.
Two gas turbine-driven inlet/treated/residue gas compressors (left) with inlet separators and discharge coolers (foreground) were added in 1995 to replace aging reciprocating compressors.

Debottleneck, retrofit

The GSP retrofit had significant advantages over debottlenecking the two-stage plant: The GSP retrofit had much higher plant throughput capability of 135 MMscfd vs. 110 MMscfd and higher ethane recovery at all flow rates.

Additionally, the cost of the retrofit was not significantly higher than debottlenecking the two-stage design. Capital cost of the new absorber and plate-fin exchanger for the GSP retrofit were balanced against the cost of two new expanders, two feed separators, and the replacement cold-cold gas exchanger required for debottlenecking the two-stage design.

GPM chose to proceed with the GSP retrofit using the nominal 135 MMscfd design plant inlet flow rate. That rate was then used as the minimum required flow in specifying the future compression requirements.

The final cryogenic-plant retrofit conceptual design, scope, cost, and schedule were defined by Ortloff in late December 1993. The project was approved by GPM on Dec. 30, 1993, with the "must complete" date set at June 1, 1994.

All tie-in work was scheduled for completion during a total facility maintenance turnaround scheduled for mid-May 1994.

Three 50% capacity stainless steel Sundyne pumps were purchased for the absorber bottoms-pump service rather than two longer delivery 100% capacity pumps. Reconditioned, multistage centrifugal pumps were provided from GPM surplus for the demethanizer-bottoms service to meet delivery requirements.

The Goldsmith plant's cryogenic process unit was originally installed with plenty of open area surrounding the unit. The cryogenic unit is separated from the compressor buildings with no adjacent processing units. The available plot area permitted the new equipment to be safely installed adjacent the existing operating cryogenic unit.

During detailed design, 82 piping and instrumentation tie-ins were identified, including many maintenance items which were in addition to the retrofit scope of work.

The maintenance items, conversion of the expanders and booster compressors to parallel operation and removal of the existing second-stage equipment from service, increased the number of tie-ins over what would be expected for a more typical retrofit.

Construction, tie in

Ref-Chem Corp., Odessa, Tex., moved onto the site on Mar. 1, 1994.

It was possible to complete construction around the absorber, subcooler, absorber bottoms pumps, reboiler, and switchgear before the facility shutdown. One expander was shut down for 1 day and disassembled for measurements and inspection before the facility shutdown.

Installation of new equipment progressed according to plans, and preparations were completed for tie-ins and facility shutdown which occurred on schedule on May 15, 1994.

During the shutdown, all tie-ins were completed, the product pumps were changed out and re-piped, and both expanders were removed and reworked. The tube bundle was removed from the intermediate level chiller (E-2) so that the additional nozzles could be installed in the shell.

All tie-in welds were 100% radiographed instead of hydrotested to prevent water from entering the system. All the new piping up to the tie-in points had been hydrotested and dried out before the shutdown.

The demethanizer was not entered during the 1994 shutdown because the original internals were acceptable for the conditions that were expected before the new compressors were commissioned in late 1995.

Commissioning; results

Work in the other Goldsmith units was completed on schedule, and the modified cryogenic unit was unblinded, purged, and pressurized for dryout flow on May 25. J-T operation and NGL delivery were resumed on May 26, 11 days after shutdown. Both modified expanders were commissioned on May 28.

Problems encountered during restart of the plant included trash in the absorber bottoms pumps screens, two gasket leaks, and miscellaneous instrument tuning and electrical interlock corrections.

Two process shutdowns were required to repair gasket leaks. The trash in the absorber bottoms pump screens appeared to be pieces of the "super-sac" material similar to that in which the absorber packing material was delivered to the site.

The pump-screen plugging problems persisted, and the packing was changed out in 1995 to solve the problem. There were no significant problems with the cold absorber bottoms pumps themselves or their seals. Both expanders were also operated without design related problems.

The molecular-sieve dehydrators must perform well to avoid freeze-ups and pressure-drop problems in the plate-fin exchangers and columns at the cryogenic temperatures.

There were some problems initially with achieving consistently low water dew points at the dehydrator outlet. The dehydration system performance has since been improved by increasing the regeneration cycle heating time and by replacing one leaking switching valve.

Initial operation was limited to the same 110 MMscfd inlet-gas rate available before the shutdown. The facility's inlet rate was increased to 125 MMscfd 7 days after startup, and this rate has been fairly constant since mid-1994.

With the reciprocating compression, the cryogenic plant's inlet pressure was 50 psig lower, and the tower pressures were about 35 psig higher than the retrofit design.

Ethane recovery has been 91% with the existing compression at the 125 MMscfd average inlet rate. The measured ethane recovery is consistent with the calculated recovery for the pressures available with the reciprocating compression.

There were intermittent problems from treated-gas-compressor lube oil. The reciprocating treated-gas compressors boost inlet-gas pressure to 800 psig after dehydration at 500 psig.

Any lube oil in the inlet-gas stream that is not removed by the existing discharge lube-oil coalescers collects on the new plate-fin subcooler exchanger and reduces the heat transfer, thus raising the reflux temperature, and reducing the recovery. (The lube-oil problem was eliminated when the new centrifugal compressors with dry-gas seals were commissioned in late 1995.)

The expander/compressors work well in their new parallel arrangement. Differences in the piping, nozzles, and nozzle actuators between the two machines prompted a unique control system in the retrofit design to keep the flow rates evenly split between the two machines.

The process pressure controller output is sent to both expander nozzle positioners, but the signal to one unit is biased 10% with a speed controller to maintain a speed match with the other machine.

The changes to the chiller vapor nozzles have greatly improved the operation of the refrigeration system by reducing the fouling of the economizer exchanger and eliminating liquid carryover back to the compressor suction scrubbers.

New equipment installed and tested with the Goldsmith plant in operation included the new demethanizer reboiler, subcooler exchanger, absorber column, and absorber bottoms pumps.

1995 changes

Following are changes made during a scheduled July 1995 shutdown in preparation for operation at the higher plant inlet rates possible with the new compressors:

  • Piping and valved tie-ins for the new compressors were installed.

  • Demethanizer trays and packing were replaced with all new packing suitable for the higher rates.

  • Absorber packing was replaced with new, clean packing. The new packing was not contaminated with any "super-sac" material and the pump screen plugging problems were eliminated.

GPM completed the major modifications to all Goldsmith systems in 1995 with installation of two large gas-turbine-driven inlet/treated/residue gas compressors. With the new compressors in service, the plant inlet rate can be increased to 135-150 MMscfd, as needed. Ethane recovery at the 150 MMscfd rate is expected to decrease to 92% from the design recovery of 95% at 135 MMscfd.

References

1. Wilkinson, J.D., and Hudson, H.M., "Improving Gas Processing Profits with Retrofit Designs for Better Ethane Rejection/Recovery," Permian Basin Regional Meeting of the Gas Processors Association, May 13, 1993, Midland, Tex.

2. U.S. Patent No. 4,157,904.

3. U.S. Patent No. 4,278,457.

The Authors

Joe T. Lynch has been a senior consulting engineer with Ortloff Engineers Ltd., Midland, Tex., since 1992. From 1985 to 1992, he was a plant engineer and then plant manager of Amerada Hess' Seminole gas plant. From 1973 to 1985, he was a mechanical engineer and later engineering group leader in Ortloff.

Lynch holds a BS (1971) in engineering science and an MS (1973) in engineering, both from the University of Texas at Austin. Lynch is a professional engineer in Texas.

Richard N. Pitman is Goldsmith area asset superintendent for GPM Gas Services Co., a division of Phillips Petroleum Co., in Goldsmith, Tex. Since joining Phillips Petroleum in 1981, he has been in plant process engineering in Borger, Tex., in process engineering and the engineering and construction group in Stavanger, and in processing engineering and his current position in Odessa, Tex.

Before joining Phillips, Pitman held various engineering positions with such companies as Shell Oil, Bingham Pump Co., and Solar Gas Turbines. Pitman holds a BS (1968) in mechanical engineering from the University of Washington.

Copyright 1996 Oil & Gas Journal. All Rights Reserved.