Research targets lower gas-processing operating costs

Howard S. Meyer, Dennis Leppin
Gas Research Institute
Chicago

Increasing natural-gas demand and declining gas quality at the wellhead require the gas-processing industry to look to new technologies to stay competitive.

The Gas Research Institute (GRI), Chicago, is managing a research, development, design, and deployment program that could save industry $230 million/year in operating and capital costs from NGL extraction and recovery, dehydration, acid-gas removal/sulfur recovery, and nitrogen rejection.

Three technologies are addressed here:

• Multivariable control (MVC) technology for predictive process control and optimization is installed or in design at 14 facilities, treating a combined total of more than 30 billion normal cu m/year (bcmy; 1.1 trillion standard cu ft/year, tcfy). Simple paybacks are typically less than 6 months.
• A new acid-gas-removal process based on N-formyl morpholine (NFM) is being field tested that offers 40-50% savings in operating costs and 15-30% savings in capital costs relative to a commercially available physical solvent.
• The GRI-MemCalc computer program for membrane separations and the GRI-Scavenger CalcBase computer program for scavenging technologies are screening tools that engineers can use to determine the best practice for treating their gas.

Research

The gas-processing program at GRI is directed at new technologies to reduce the ultimate cost of bringing wellhead gas to the U.S. interstate pipeline system. The program includes fundamental, exploratory, and applied research in various processing unit operations, including dehydration, acid-gas removal and sulfur recovery, nitrogen rejection, and NGL extraction. Overall objectives are two-fold:

1. To develop new technologies and improve existing ones which will lower the cost of producing natural gas for sale to the pipeline system and ultimately to the consumer
2. To develop software tools and data bases to provide information for processing decision making. The focus is on impact to the industry during 1996-2000 with some longer-term basic and exploratory research to set the foundation for future technology needs.

Technologies from the gas-processing program are currently already making a significant impact on the industry:

• R-BTEX process is being used at 11 installations for the removal and recovery of benzene, toluene, ethylbenzene, and xylene (BTEX) from dehydration vent streams. These facilities have a combined capacity of approximately 20 bcmy (700 bcfy).¹
• GRI-GLYCalc computer program for estimating emissions of BTEX from glycol dehydrators has been distributed to more than 800 individuals and been recommended by several state agencies for permitting use.²
• Results from field testing of liquid redox and scavenging processes for removal of H2S from high-pressure natural gas has been used by the industry as the basis for process selection.^{3 4}
• Lithium-bromide absorption chillers are providing a consistent hydrocarbon dew point with no environmental impact at a cost less than the conventional propane chiller. Field evaluations have shown that the dew point can be controlled from -17° to 27° C. (1- 80° F.).⁵

Gas-composition data base

Natural-gas composition varies greatly across the U.S. and from well to well.

Before introduction of raw natural gas to the U.S. pipeline system, the gas generally requires treatment to ensure that the more than 50 million residential, industrial, and commercial customers receive a high-quality fuel. Reducing these treatment costs allows gas to remain competitive at the burner tip.

Improved gas-processing technology can also improve the marketability of subquality natural-gas resources that are now uneconomical to produce because of the processing costs.

All natural gas is sold and transported under contracts with pipelines that specify the maximum acceptable impurities in the gas. There are variations between contracts on the actual quantities specified.

The range of typical specification in the standard contracts of U.S. pipelines is shown in Table 1 [13,011 bytes].⁶

To date, available information on the quantity of produced natural gas and lower-quality fields has been limited and has existed in several separate data systems and reports.

To provide an improved data base that could be used to establish a clear rationale for a gas-processing R&D program, Energy & Environmental Analysis, under contract to GRI, has integrated existing information and augmented it with data obtained through a national survey of producers and pipeline companies to assemble a single data base that characterizes gas reserves by composition.⁷

Table 2 [16,659 bytes] summarizes the chemical composition of raw, Lower 48, nonassociated reserves, production, and resources. The total raw gas reserve is estimated at 144.2 tcf. The amount considered subquality is 52.9 tcf or 37% of the raw resources. About 31% of the total Lower 48 annual production of 15.4 tcf is considered subquality in composition.

Conditioning, processing plants

In 1993, Purvin & Gertz, Houston, assembled for GRI a data base on U.S. gas conditioning and processing. ⁸

Purvin & Gertz compiled a data base on 1,427 gas processing and conditioning plants in the Lower 48 U.S., including 218 which have shut down. There were NGL extraction facilities at 857 plants including many which had two separate types of NGL-extraction on-site; these may be counted as 2 plants.

H₂S and/or CO₂ removal had been accomplished at 617 plants. Sulfur recovery through a Claus sulfur plant downstream of H₂S removal was identified at 84 plants; another 41 plants used a direct-conversion process for removing of H₂S.

Nitrogen rejection and helium-recovery capabilities were identified at 12 and 8 plants, respectively. In an independent study, Wright Killen & Co., Houston, estimated that there are approximately 42,000 dehydrators across the U.S.⁹

Almost 50% of the operating plants are located in Texas and Louisiana. There are 220 operating plants built before 1970 which represent almost 40% of the combined inlet capacity of the 1,209 operating plants.

More than 50% of the currently operating plants have a design inlet capacity of 7,000 cmd (approximately 25 MMscfd) or less, and 6% have more than 250 MMscfd.

Purvin & Gertz estimated that existing gas conditioning and processing facilities have a replacement investment of more than $7 billion.¹⁰ Based on forecasts of natural-gas production, the consulting company estimated that an additional $1.7 billion of capital investment will be needed over the next 10 years.

The new facilities are expected to be somewhat smaller on average than existing ones, primarily because of a reduction in the average size of new fields. The consultant also estimated that used equipment could account for 5-25% of the projected number of units. The cost to refurbish and relocate may be similar to the cost of a new facility, but the time to deploy can be much faster.

Estimated gross revenues from sales of NGL extracted as by-products by U.S. natural-gas processors totaled $9 billion in 1991.¹¹ Gross revenue from sales of gas at the producer-pipeline interface was roughly $38 billion.

By-products helium (about $160 million) and sulfur (about $154 million) each contributed far less to the economics of natural-gas production. The value of CO₂ and N2 recovered as by-products of natural-gas processing were not determined because they are recovered largely for internal use where no transfer price is visible.

The economic outlook for extraction and recovery of NGL is positive: Growing NGL demand will push up NGL prices.

The helium market is uncertain at this time in light of the 1996 Helium Privatization Act that will end the U.S. Government acting as a producer and marketer of helium.

Production of sulfur is expected to increase because of increased restrictions on sulfur emissions, resulting in a decrease in sulfur prices through the decade.

Multivariable control

Natural-gas plants are excellent applications for continuous, on-line, inexpensive, easy to use and understand optimization technology.

Such inlet conditions as flow rate, temperature, pressure, and gas composition change continuously. Market conditions for expenses, such as the cost of utilities, and for revenue, such as the price for ethane, can also vary greatly.

Gas-processing plants, however, are largely under-penetrated with such optimization controls for several reasons, including:

• The high degree of process nonlinearity that results in a difficult multivariable control problem
• A lack of industry sophistication in advanced control and optimization
• Varying operating philosophies which can differ significantly from plant to plant
• Software costs that are not justifiable for most medium and small facilities.

Several multivariable control approaches have been developed, including dynamic linear models and rigorous steady state models.

A unique, hybrid approach, known as MVC and developed by Continental Controls, Houston, combines reduced, rigorous, fundamental models with plant dynamics and economics to produce a set of nonlinear multivariable controller/optimizer solutions designed to handle an operator's particular requirements.¹²

MVC uses a set of reduced equations derived from well-known steady state process simulators, such as Hysim, Pro II, Chemshare, and Tsweet. A "core" model is achieved by closely matching simulation results to actual plant operating data.

The reduced equations that predict rigorous simulation results are combined with plant dynamics and operating constraints derived from plant tests to form a module. Each module is application specific to the particular process type, for example, cryogenic, lean oil, fractionator, compressor, amine and sulfur recovery, and then customized as necessary for the idiosyncrasies of the particular operation.

The modules are then integrated with the multivariable controller, economic optimizer, and a set of generic tools to yield the system. This approach produces a solution that allows multiple variables to be manipulated simultaneously to achieve several operating objectives economically within operating constraints.

Fig. 1 [67,583 bytes] shows the MVC controller.

Predictor models are used to predict the effects of any disturbances in a dynamic manner by polling multiple, key variables so that appropriate manipulated variables are moved to maintain the controlled variables at their economic optimal value.

The controller includes a means to correct any errors between actually measured values and predicted values, as well as the means for compensating for offset and drift of key sensors including analyzers and indicators. Therefore, the approach is self-adapting to changes in processing efficiency and instrument drift.

This technology has several important features that distinguish it from other multivariable controller/optimizers:

1. Reduced nonlinear models more closely represent the actual process over the wide, relevant operating range of the plant. It has been demonstrated that better models improve the performance of the multivariable controller. The models effectively provide adaptive gain that eliminates controller "hunting" and "zigzag."
2. Feedback from the plant at each control cycle provides the "sanity" check to allow automatic on-line [bias] adjustment of the model predictor. Therefore, predictive equations need not be perfect.
3. Standard features include compensation for measured variable drift and offset, including inferred properties and on-line analyzer comparison to laboratory results.
4. Separate and distinct allowances are made for process dynamics as well as the sensor or analyzer delays.
5. A constrained economic optimum, which allows solutions anywhere at or inside constraints, is calculated each control cycle (typically, 30 sec to 1 min).
6. The theory behind MVC is relatively easy for operators, plant engineers, and process engineers to understand.
7. MVC's low cost, simplicity, and minimal on-site support brings optimization opportunities to all size gas-processing plants.
8. MVC commissions faster than comparative technologies because the model gains are not based on step test data.
9. MVC can integrate easily with the existing control system, from a pneumatic system to a distributed control system. MVC usually resides on a dedicated personal computer but has been installed on systems such as Moore APACS and Honeywell SCAN 3000.

Results; features

The initial MVC installation considerations for a Ryan-Holmes CO ₂-recovery plant are documented elsewhere. ¹³

Other units installed or in design include two lean-oil plants, six cryogenic plants, four sulfur plants, six amine treaters, two compressor services including a compression-allocation system, one nitrogen-rejection plant, and five fractionation columns. The total installed capacity of these plants is greater than 1.1 tcfy.

These projects have been implemented within 4-6 months with a simple payout of less than 6 months to a year by increasing NGL revenues and decreasing plant-operating costs.

Recent results from an application of the MVC technology in a 100-MMscfd plant are seen in Fig. 2 [41,069 bytes] and Fig. 3 [40,337 bytes]. In both figures, a comparison is made between similar 5-day periods before and after applying the MVC technology.¹⁴

In Fig. 2, the average daily profit and barrels per day recovery are shown each day during the different 5-day periods. The daily profit is defined simply as plant share (after shrinkage) of gas plus liquids revenue minus fuel costs associated with propane refrigeration and recompression costs minus any product specification quality penalties as defined by the liquids pipeline contract.

On average, the MVC technology improved daily profit by $2,665 based on the then-current gas and liquids pricing yet maintained bottoms liquids C₁/C₂ specification at 2.0% (liquid volume) basis.

Daily liquids production was improved an average of 63.5 cu m (400 b/d). On an annual basis with a 95% plant service factor, the MVC technology benefit was about $925,000.

The combined economic optimization coupled with a nonlinear multivariable controller technology, on average, increased ethane and propane recovery (Fig. 3) about 4.6% and 0.6%, respectively, while maintaining the bottom liquids C₁/C₂ specification.

In addition, the MVC technology improved plant operating stability by reducing control variable cycling, which had been induced by conventional feedback control response to widely changing inlet feed disturbances.

Fig. 4 [43,653 bytes] shows the improvements that resulted with the upgrading of the process control system at a 100-MMscfd cryogenic plant. This facility moved from a manual controlled system to computer controlled, installation of programmable logic controllers (PLC), and then the operation of MVC in a supervisory and then direct-control mode.

Implementation of MVC increased the NGL product by about 20% while increasing the C₂ recovery by about 25%. Again, the C₁/C₂ deviation was significantly decreased, moving the product closer to the economical end of the specification range. MVC achieved payback in about 4 months at this plant.

The MVC technology offers the gas processor a robust economic optimization tool to adapt aggressively to gas-processing market demands. With the rapid changes occurring in the marketing of gas processing services, it is likely that reduced costs or the maximization of profit will make gas processors better able to meet challenges.

Acid-gas removal

With support from GRI, the U.S. Department of Energy (DOE), and the Institute of Gas Technology (IGT) Sustaining Membership Program, IGT is developing technology that will reduce gas-processing costs for current production and allow subquality gas to be economically produced.

The experimental program, in progress since 1990, has focused primarily on the evaluation of N-formyl morpholine (NFM) as a physical solvent for the cost-effective upgrading of subquality natural gas to pipeline-quality gas.¹⁵

Selection of NFM for this program was based on previous work conducted by IGT on the selective removal of hydrogen sulfide and carbon dioxide from coal gasifier effluents. That work showed that the use of NFM resulted in a significant cost advantage over 107 other solvents for that application.

Since 1968, NFM has been successfully used by both Huntsman Specialty Chemical Corp. and Krupp Koppers GmbH in the recovery of high-purity aromatics.

NFM, a derivative of morpholine, exhibits very high selectivity with good solvency, is environmentally compatible, is nontoxic and biologically degradable, and is widely known in refinery technology. This work will extend its application to gas sweetening.

The VLE apparatus developed at IGT is capable of handling all sulfur species present in natural gas and operating at pressures 0.17-21 MPa (25-3,000 psia) and temperatures 60-300° F.

More than 1,000 data points were collected for mixtures of NFM and water, MEA, DGA, MDEA, NMM, and NEM over ranges of conditions (60-300° F. and 25-1,150 psia)and gas components (single component:

CO₂ , H₂S, COS, C₃H ₈ , to C₆s; and mixtures: CH₄ , CO₂ , H₂S,

C₂H ₆ , C₃ H ₈ , to C₆s, BTEX).

The ASPEN Plus process-simulation software package was used to regress VLE and mass-transfer data to obtain the appropriate equation-of-state coefficients. Peng-Robinson, Redlich-Kwong-Soave, Redlich-Kwong-ASPEN, and Redlich-Kwong-UNIFAC thermodynamic models were investigated.

As a reality check, a solvent manufacturer conducted simulations of the same cases and conditions its proprietary models for a commercial physical solvent. Table 3 [10,019 bytes] shows a comparison between NFM and the commercial solvent where the requirements of the commercial physical solvent are referenced as 100%.

NFM has high capacity for H₂S and CO₂ absorption, low capacity for hydrocarbons, and needs little or no refrigeration. According to the models, there would be substantial savings in plant operating costs (40-50%), as well as in capital costs (15-30%).

Structured packing

Structured packing is well established, having been used successfully in distillation columns, specifically in low-pressure and vacuum distillation, because of its high efficiency, low pressure drop, low "height equivalent to a theoretical plate" (HETP) and high capacity.

This technology has also shown excellent performance in natural-gas dehydration at pressures of 1,000 psig and greater. ARCO Oil & Gas Co. showed that, for the same process conditions in a triethylene glycol (TEG) dehydration column, the structured packing was able to reduce the column size by 260%, vessel weight by 200%, internals by 200%, and cost by 30%.

Surprisingly, this technology has not been popular in natural-gas sweetening, there being little documented experience reported. The process conditions for natural-gas sweetening are closer to natural-gas dehydration than high-pressure distillation, and both natural-gas sweetening and natural-gas dehydration are absorption-type systems.

Potentially, benefits such as increased capacity, increased efficiency, lower pressure drop, and lower residence time (which may translate into increased selectivity) may be realized by this technology in natural-gas sweetening.

In conjunction with a program to evaluate structured-packing contractors for gas-sweetening applications, IGT has initiated a series of tests to evaluate NFM under field conditions. Although NFM VLE and computer modeling results were very encouraging, it was necessary to test NFM on a large scale.^16-18

IGT has conducted tests to determine NFM's performance in processing subquality natural-gas streams with 14% and 43% total acid-gas concentrations. The acid-gas loading of the NFM solvent increased from 3 cu ft/gal from 14% CO₂ inlet-gas concentration tests to 11.5 cu ft/gal at the higher partial pressure of CO₂ in the feed stream.

The field test unit (Fig. 5 [44,171 bytes]) is currently installed beside a Shell Western E&P Inc. processing plant near Zapata, Tex. The Shell plant processes gas from 18 wells nearby, and it has an operating capacity of 70 MMscfd.

The feed gas for IGT's field test unit is drawn from Shell's Fandango plant, and the processed gas from IGT's test unit is returned to the suction side of Shell's recycle compressor. Flash gas and acid gas from IGT's skid-mounted unit are combined and returned to Shell's vent system.

Preliminary testing from the 1-MMscfd pilot plant absorption unit operating at Shell's Fandango facility indicates the use of NFM as a treating solvent is extremely promising. Forty-three tests have been completed.

The data, when used in a sophisticated process simulation, predict a 40% savings for treating high-CO₂ streams, whether or not H₂S is present.

The data indicate that it might be possible to eliminate completely the heat input to the reboiler even with appreciable H₂S concentrations in the gas. This would result in significant savings and eliminate a high maintenance, high-cost utility boiler currently used for regeneration steam.

Further, these results were achieved without refrigeration. The current physical process at Shell Western's facility uses refrigeration to 25-30° F. levels to achieve comparable results. As expected with the short (20 ft column), the pipeline-gas specifications for CO₂ and H₂S were not met in the pilot plant.

Tests with structured packing of two types were conducted recently. The structured packing provided greater absorption, higher capacity, and lower pressure drop, all of which have the potential to improve significantly the expected savings of using the process.

By using a slipstream of recycled flash gas from the existing process as feed to the pilot unit, results were obtained at up to 43% CO₂ concentrations in the feed.

Further work in 1998 will include tests of a new solvent combination which will allow the use of refrigeration. This has benefits where it is desired to increase the capacity of a plant beyond the capabilities of current physical solvent formulations.

IGT and Krupp uhde have concluded an agreement that enables Krupp Wilputte to offer this technology to potential users worldwide. GRI, IGT, and the U.S. DOE are sponsors of the work.

Computer programs

GRI is exploring many means of technology transfer to communicate the results of its research to industry. The audience is generally the process or operating engineer who must make a selection between several alternative processes to identify the most cost-effective for a specific need.

Software products, as shown in this section, greatly augment the written reports and allow the engineer to focus quickly his studies on viable options.

These programs not only estimate performance characteristics, but also provide "expert-type" advisory information and educational information regarding the use of the technologies as a unit operation.

This educational information is in the form of hypertext, pictures, diagrams, and reference sources. Context-sensitive definitions are provided throughout the program, and extensive error trapping is provided both for procedural and conceptual errors.

The principal uses of these programs are for screening of new or unfamiliar technologies and for education of inexperienced users. The screening helps the engineer determine the best practices for any gas-separations problem and as a second opinion or check on the performance of the technologies.

Once screening has been done, electronic forms are provided to be completed and faxed to the major manufacturers to obtain detailed design information. These programs expand the public-domain knowledge of the industry and are expected to decrease the cost of removing undesirable components from natural gas.

MemCalc has been developed for simulating the performance of membranes for CO₂ removal from natural gas and other hydrocarbon streams.¹⁹ Scavenger CalcBase has been developed for simulating and comparing, on a consistent cost basis, the performance of scavengers for small amounts of H₂S from natural gas.²⁰

Both programs are PC-based for Microsoft Windows operating system.

Over the past several years, the use of membranes for natural-gas treating (specifically CO₂ removal) has grown for economic reasons, especially for small-scale applications.

To exploit the use of membranes fully, the engineer must have access to appropriate software tools necessary to integrate membranes for an optimized system. While software models exist for membrane-unit operations, the performance and costs generally are available only from commercial membrane vendors, typically on a case-by-case basis.

Therefore, the engineer may be unable readily to explore the options necessary to determine the optimum process configuration for a specific gas stream. Furthermore, without a screening tool, the engineer must rely solely on membrane manufacturers to narrow the choice of operating parameters.

Gas industry use of H₂S scavengers for removing low concentrations of H₂S (typically up to about 200 ppmv H₂S depending on the gas flow rate) where conventional amine treating is not economical has been growing significantly.

For years, the iron sponge and other nonregenerable processes were widely used by the industry to treat these gases. Heightened concerns about the safety and environmental impact associated with spent-materials disposal, as well as an increased use of scavengers, have prompted introduction and use of many new scavenger technologies.

As environmental regulations have become stricter, a keen interest in identifying better and more environmentally acceptable H₂S scavenging technologies has emerged, thereby generating a need for information on the application and performance of these technologies.

GRI has conducted extensive research in both of these areas. Laboratory and field evaluations of membranes and scavengers have been conducted and reported elsewhere.^{21 22} These studies were augmented by engineering evaluations to integrate these technologies into a total system on a consistent basis.

Expert information; "How to"

Both programs also include "expert-like" process advisory information.

In cases when the user should be aware of other considerations or makes a poor decision, "expertlike" advisory information is made available. For instance, a saturated-feed condition and high C₃+ content in the feed will elicit a warning to check the inlet dew point.

Another example of "expert-like" advisory information would be if the MemCalc user were to enter a permeate pressure of 150 psia. In this event, a warning message would recommend lowering the permeate pressure to achieve better membrane system performance.

This high permeate pressure may be desirable, however, to achieve an overall system performance optimum.

A context sensitive "HowTo" hypertext document is used to explain input fields, menu items, engineering terminology, and error messages, and to indicate limitations and intended uses of the software. Any unfamiliar terms used in the hypertext documents can be clicked on for additional information.

Besides the vendor-supplied "Help" files, there are several other such files accessible from each of the programs. A main "Help" file, which includes information about the program and its operation and a general introduction to the technology, is a starting point for access to all of the other help files.

Hypertext versions of GRI evaluation studies are available on-line through the help files. The hypertext documents are accessible from within the program's title screen, pull-down menus, and keyboard function key (F1).

Alternatively, the information is accessible from Windows' Program Manager as a standalone application.

What's coming

Current efforts in gas processing expected to reach fruition within the next 3-4 years include advanced membrane materials development for CO ₂ removal from natural gas, PC-based expert systems for process selection, and the CrystaSulf process for direct, high-pressure removal and conversion to sulfur of hydrogen sulfide in natural gas, ²³ a "best practices" assessment of direct injection of scavenger chemicals in high-pressure natural gas, ²⁴ evaluation of direct oxidation of amine offgas using a new catalyst. ²⁵

Plans for testing hydrocarbon dewpointing membranes for offshore applications are being advanced. Evaluations under way include a technology assessment of BTEX/VOC emissions from dehydrators control technology and assessment of a new approach to improving ethane recovery in refrigeration plants.

These results are expected to be available in early 1998.

References

1. Fisher, K.S., Rueter, C., Lyon, M., and Gamez, J., "Glycol dehydrator emission control improved," OGJ, Feb. 27, 1995, p. 40.
2. Rueter, C.O., and Gamez, J., "GRI's R&D Program on Natural Gas Dehydration," presented to the 74th Annual Convention of the Gas Processors Association, San Antonio, Mar. 13-15, 1995.
3. Dalrymple, D.A., and Wessels, J., presentation to the 1995 GRI Sulfur Recovery Conference, Austin, Tex., Sept. 24-27, 1995.
4. Leppin, D., and Dalrymple, D., "Overview of Liquid Redox Technology for Recovering Sulfur from Natural Gas," presented to the 1996 Laurance Reid Gas Conditioning Conference, Norman, Okla., Mar. 3-6, 1996.
5. Lane, M.J., and Huey, M.A., "Lithium bromide absorption chiller passes gas conditioning field test," OGJ, July 31, 1995, p. 70.
6. Meyer, H.S., Leppin, D., and Gamez, J., "Improvements in Natural Gas Processing," presented to Natural Gas Quality & Energy Measurement: Practices & Applications, Orlando. Feb. 20-22, 1995.
7. Hugman, R.H., Springer, P.S., and Vidas, E.H., "Chemical Composition of Discovered and Undiscovered Natural Gas in the United States-1993 Update," GRI-93/0456, December 1993.
8. Tannehill, C.C., and Galvin, C., "Business Characteristics of the Natural Gas Conditioning Industry," GRI-93/0342, May 1993.
9. Graham, J.F., et al., "Natural Gas Dehydration: Status and Trends," GRI-94/0099, January 1994.
10. Tannehill, C.C., and Gibbs, J.E., "Gas Processing Industry-Lower 48 States," GRI-91/0232, July 1991.
11. Tannehill, C.C., Echterhoff, L., and Trimble, K., "Assessing the Value of NGLs in Natural Gas," presented to 74th Annual Convention of the Gas Processors Association, San Antonio, Mar. 13-15, 1995.
12. Colwell, L.W., Poe, W.A., Papadopoulos, M.N., and Gamez, J.P., "What's New in Multivariable Predictive Control," presented to 74th Annual Convention of the Gas Processors Association, San Antonio, Mar. 13-15, 1995.
13. Berkowitz, P.N., Chou, K., Clay, R. M., Gamez, J.P., and Papadopoulos, M.N., "Multivariable control system installed at ARCO West Texas gas plant," OGJ, Nov. 16, 1992, p. 53.
14. Berkowitz, P.N., Colwell, L.W., and Gamez, J.P., "Gas Plant Economic Optimization is More than Meeting Product Specification," presented to 75th Annual Convention of the Gas Processors Association, Denver, Mar. 11-13, 1996.
15. Semrau, J.T., Palla, N., and Lee, A.L., "Subquality Natural Gas Sweetening and Dehydration Potential of the Physical Solvent N-Formyl-Morpholine," presented to the 1995 Laurance Reid Gas Conditioning Conference, Norman, Okla., Feb. 26-Mar. 1, 1995.
16. Palla, N., et al., "An Acid-Gas Removal System for Upgrading Subquality Natural Gas," presented to XII Jornadas de Gas, Puerto La Cruz, Venezuela, May 8-11, 1996.
17. Gross, M., et al., "Commercial Development of New Gas Processing Technology," presented to Natural Gas in the Americas IV-Lighting the Flame, July 14-16, 1997, Barbados.
18. Gross, M., and Menzel, J., "Acid Gas Removal for Upgrading Natural and Synthesis Gas," Hydrocarbon Engineering, November 1997.
19. Reading, G.J., and Meyer, H.S., "MemCalc: Computer Program for Membrane Separation Simulation," presented to the 1996 Membrane Technology/Separations Planning Conference, Newton, Mass., Oct. 28-30, 1996.
20. Guillory, G., and Leppin, D., "GRI Scavenger CalcBase Software, Comprehensive Scavenging Resource," presented to the 1996 Laurance Reid Gas Conditioning Conference, Norman, Okla., Mar. 3-6, 1996.
21. Feldkirchner, H.L., and Lee, A.L., "Field Evaluation Supports Applicability of Membrane Processing," GasTIPS, Vol. 2, No.

    2 (Spring 1996), p. 39.

22. Fisher, K., "Hydrogen Sulfide Scavenging Research Accelerates the Use of Improved Technologies," GasTIPS, Vol. 2, No. 2 (Winter 1995/1996), p. 43.
23. Dalrymple, D.A., and DeBerry, D., "CrystaFulf: A non-aqueous solution for H₂S problems," presented to GRI Eighth Sulfur Recovery Conference, Austin, Oct. 12-15, 1997.
24. Fisher, Kevin, "Initial Results from GRI's 30 MMscf/day Direct Injection H₂S Scavenging Test Facility," presented to GRI Eighth Sulfur Recovery Conference, Austin, Oct. 12-15, 1997.
25. Srinivas, G., and Karpuk, M., "Selective Oxidation of Hydrogen Sulfide to Sulfur," presented to GRI Eighth Sulfur Recovery Conference, Austin, Oct. 12-15, 1997.

The Authors