CA2174341C - Membrane and non-membrane sour gas treatment process - Google Patents

Abstract

Improved processes for treating gas streams containing hydrogen sulfide, car bon dioxide, water vapor, and methane, particularly natural gas streams. The processes rely on the availability of two membrane types, one of which has a hydrogen sulfide/methane selec tivity of at least about 40 when measured with multicomponent gas mixtures at high pressure. Based on the different permeation properties o f the two membrane types, optimized separation processes can be designed. The memb rane separation is combined with non-membrane treatment of the residue and/or permeate streams.

Description

x k ,. P~

MEMBRANE AND NON-MEMBRANE SOUR GAS TREATMENT PROCESS
FIELD OF THE INVENTION
The invention relates to processes for removing acid gases from gas streams.
More particularly, the invention relates to a membrane process, and to combinations of membrane and non-membrane processes, for removing hydrogen sulfide and carbon dioxide from gas streams, such as natural gas.
BACKGROUND OF THE INVENTION
Natural gas provides more than one-fifth of all the primary energy used in the United States.
Much raw gas is "subquality", that is, it exceeds the pipeline specifications in nitrogen, carbon dioxide and/or hydrogen sulfide content.
The best treatment for natural gas right now is no treatment. Raw gas that is lrnown to be high in nitrogen content, high in nitrogen plus carbon dioxide content, or high in hydrogen sulfide content is usually leh in the ground, because it cannot be extracted and treated economically with present processing technology.
There are several aspects to the problem of treating natural gas to bring it to pipeline specifications. The first is the removal of impurities, primarily water, hydrogen sulfide and carbon dioxide; the second is loss of methane during processing. Processes that remove hydrogen sulfide and carbon dioxide may also remove a portion of the methane. Losses of less than about 3% are nonmally acceptable; losses of 3-10% may be acceptable if offset by other advantages;
losses above 10% are normally unacceptable. A third aspect is the fate of the impurities once removed. Carbon dioxide can be discharged or reinjected, but hydrogen sulfide, which is toxic even in low concentrations, must be treated.
If the waste stream containing hydrogen sulfide can be concentrated sufficiently, it may be passed to a Claus plant for conversion to sulfur. Waste streams containing low concentrations must be disposed of in some other way, such as a redox process of the LO CAT or Stretford type, for example, or, less desirably, flaring.
Choice of appropriate treatment is, therefore, not straightforward, and depends on the feed gas composition, the size and location of the plant and other variables.
When.natural gas is treated, most plants handling large volumes of sour gas containing greater than about 200 ppm hydrogen sulfide use amine-based technology for acid gas removal. Amines commonly used include MEA, DEA, DIPA, DGA and MDEA. The plants can remove both carbon dioxide and hydrogen sulfide. When the amine solution is spent, the acid gases are flashed off and the 75136=9 solution is regenerated The mxhanicai equipment in an amine plant makes it susceptible to failure. The plant includes heaters, aerial coolers, pumps, etc. and requires frequent quality checks and maintenance, making operations! reliability probably the weakest feature of the technology.
Amine plants do not sorb mdhane to any signiG~t tent, so methane Loss is not an issue in this case. However, the hydnng~-sulGde~ontaining gas stre~tt when the sorbent is rcgentratod must still be treated, subjoct to the same constraints as above.
As an alternative to amine sorption, or as a polishing step following any process, specialized scavrnging or sulfur t~ecovery processes, such as Sulfa-Scrub, Sulfa-Check, Chernsweet, Supertron 600, solid iron sponge or solid zinc oxide may be used for low~volume streams containing less than about 100 ppm hydrogen sulfide. Many scavengers present substantial disposal problems, however. In an increasing number of states, the spent scavenger constitutes toxic waste.
A considerable body of literature exists regarding membrane-based treatment of natural gas, mostly using cellulose acetate (CA) membranes to remove carbon dioxide.
Although cellulose aoetatc membrane platys are designed to remove carbon dioxide, cellulose acetate merabrarxs also have selectivity for hydrogas sulfide over methane, sa thty tend to cocxtract small amounts of hydrogen sulfide. Unless the raw gas stream cattains very high concattrations of carbon diocide, however, it is not possible to reduce a stream containing even modest amounts of hydrogen sulfide to pipeline specification (usually 4 ppm hydrogen sulfide) without vastly ovaprocessing as far as the carbon dioxide specification is concerned. If such ovcrprocessing is performed, large amounts of methane are lost in the membrane permeate stream, and this is normally unacceptable.
Only a few of the many literature refenmcex rztating to membrane-based carbon dioxide treatment spxifxallydiscuss removal ofhydrogat sulfide in conjunction with the carbon dioxide. A paper by W.J.
Schell et al. ("Separation of COI Gem Mixtures by Membrane Permeation", presented at the Gas Conditioning Conference, Uttiversity of Oklahoma, March 1983) says that "If the H=S level is low enough, the membrane system can also be used to melt pipeline specification for this component without arty further lrcatmcnt required." The papa shows a case what a cellulose acetate membrane system can be used to reach pipeline spxification for carbon dioxide and hydrogen sulfide in two stages, starting with a Cecd content of 15% carbon dioxide and 250 ppm hydrogen sulfid and c, points out that, for high concsntrationt of hydrogen sulfide, "a much larger number of elements are required to reduce the H=S
levels to pipeline specification (I/4 grain) than for CO= (3%)." The costs of membrane trcatrnent are estimated to be more than 100% highs than conventional amine treatment in this case.
A report by N.N. Li et al. to the Depa:trrttttt of Energy ("Membrane Separation Processes in the " ~ r~

Petrochemical Industry", Phase II Final Report, September 1987) examined the effect of impurities, including hydrogen sulfide, on the ability of cellulose acetate membranes to remove carbon dioxide from natural gas. The reporters found that the membrane performance was not affected significantly by hydrogen sulfide alone. However, dramatic loss of membrane permeability was observed if both hydrogen sulfide and water vapor were pit in the food. The authors concluded that "successful use of these CA
based membranes must avoid processing gas which simultaneously has high H20 and H2S concentrations".
Another problem associated with cellulose acetate membranes is water, which is always present in raw natural gas streams to some extent, as vapor, entrained liquid, or both. The gas separation properties of cellulose acetate membranes are destroyed by contact with liquid water, so it is normally necessary to provide pretreatment to knock out any liquid water and to reduce the relative humidity low enough that there is no risk of condensation .of water within the membrane modules on the permeate side.
For example, the above-cited paper by W.J. Schell et al. ("Separation of COZ
from Mixtures by Membrane Permeation", presented at the Gas Conditioning Conference, University of Oklahoma, March 1983) points out that '°Even though membrane systems simultaneously dehydrate while removing C02, care must be taken to avoid contacting the membrane with liquid water. Feed gas streams saturated with water are normally preheated to at least I 0 ° above the water dew point at the feed inlet pressure and the pressure tubes and inlet piping are insulated to prevent condensation."
The above-cited report by N. N. Li et al. ("Membrane Separation Processes in the Petrochemical Industry," Phase II Final Report, September 1987) presents data showing the effect of water vapor on membrane flux for cellulose acetate membranes, and concludes that "for relative humidifies of 30% and higher, the flux decline is large, rapid, and irreversible". E.W. Funk et al.
("Effect of Impurities on Cellulose Acetate Membrane Performance", Recent Advances in Separation Techniques - III, AIChE
Symposium Series, 250, Vol 82, 1986) advocate that "Moisture levels up to 20%
RH appear tolerable but higher levels can cause irreversible membrane compaction".
U.S. Patent 4,130,403 to T.E. Cooley et al. (Removal of HZS and/or COz from a Light Hydrocarbon Stream by Use of Gas Permeable Membrane, 1978, Col. 12, Dines 36-39) states that "It has been discovered that in order to function effectively, the feed gas to the cellulose ester membrane should be substantially water free". A second paper by W.J. Schell et al.
(°'Spiral-Wound Permeators for Purification and Recovery", Chemical Engineering Progress, October 1982, pages 33-37) confums that "Liquid water is detrimental to the perfonmance of the membrane, however, so that the feed gas is delivered to the membrane system at less than 90% relative humidity."
In other words, although cellulose acetate membranes will permeate water preferentially over f , - ~ 17 4 3 41 pCT~594/12100 methane, and hence have the capability to dehydrate the gas stream, care must be taken to keep the amounts of water vapor being processed low, and, according to some teachings, as low as 20-30% relative humidity.
In light of these limitations, considerable effort has been expended over the last few years in the search for membrane materials that would be better able to handle streams containing carbon dioxide plus secondary contaminants, notably hydrogen sulfide and water.
For dense polymer membranes, the combined effect of the sorption and diffusion phenomena determines the selectivity of the membrane. The balance between mobility, or diffusion, selectivity and sorption selectivity is different for glassy and rubbery polymers. In glassy polymers, the mobility term is usually dominant, permeability falls with increasing permeant size and small molecules permeate preferentially. In rubbery polymers, the sorption term is usually dominant, permeability increases with increasing permeant size and large molecules permeate preferentially. Since both carbon dioxide (3.3 A) and hydrogen sulfide (3.6 A) have smaller kinetic diameters than methane (3.8 A), and since both carbon dioxide and hydrogen sulfide are more condensable than methane, both glassy and rubbery membranes are selective for the acid gas components over methane. To date, however, most membrane development work in this area has focused on glassy materials, of which cellulose acetate is the most successful example.
In citing selectivity, it is important to be clear as to how the permeation data being used have been measured. It is common to measure the fluxes of different gases separately, then to calculate selectivity as the ratio of the pure gas permeabilities. This gives the "ideal"
selectivity for that pair of gases. Pure gas measurements are more commonly reported than mixed gas experiments, because pure gas experiments are much easier to perform. Measuring the permeation data using gas mixtures, then calculating the selectivity as the ratio of the gas flu.~ces, gives the actual selectivity that can be achieved under real conditions. In gas mixtures that o~tain condensable components, it is frequently, although not always, the case that the mixed gas selectivity is lower, and at times considerably lower, than the ideal selectivity. The condensable component, which is readily sorbed into the polymer matrix, swells or, in the case of a glassy polymer, plasticizes the membrane, thereby reducing its discriminating capabilities.
A technique for predicting mixed gas performance under real conditions from pure gas measurements with airy reliability has not yet been developed. In the case of gas mixtures such as carbon dioxidc/methane with other components, the expectation is that the carbon dioxide at least will have a swelling or plasticizing effect, thereby changing the membrane permeation characteristics. This expectation is borne out by cellulose acetate membranes. For example, according to a paper by M.D.

~~ ~ ~ ~ r c 21 l 4 3 41 pCT~S94/12100 WO 95/11739 ' Donahue et al. ("Permeation behavior of carbon dioxide-methane mixtures in cellulose acetate membranes", Journal of Membrane Science, 42, 197-214, 1989) when measured with pure gases, the carbon dioxide permeability of asymmetric cellulose acetate is 9.8 x 10-5 em3/cm2-s-kPa and the methane permeability is 2.0 x 10~ cm3/cm2~s~kPa, giving an ideal selectivity of about 50. Yet the acfual selectivity ~ 5 obtained with mixed gases is typically in the range 10-20, a factor of 3-5 times lower than the ideal selectivity. For example, the report to DOE by Norman Li et al., discussed above, gives carbon dioxidc/mefhane selxtivities in the range 9-15 for one set of field trials (at 6000-6240 kPa (870-905 psi) feed pressure) and 12 for another set (at 1483 kPa (200 psig) feed pressure) with a highly acid feed gas.
The W.J. Schell et al. Chemical Engineering Progress paper, discussed above, gives carbon dioxide/methane selectivities of 21 (at 1828-3207 kPa (250-450 psig) feed pressure) and 23 (at SG20 kPa (800 prig) feed pressure). Thus, even in mixed gas measurements, a wide spread of selectivities is obtained, the spread depending partly on operating conditions. In particular, the plasticizing or swelling effect of the carbon dioxide~on the membrane tends to show pressure dependence, although it is sometimes hard to distinguish this from other effects, such as the contribution of secondary condensable components.
The search for improved membranes for removing acid components from gas streams, although it has focused primarily on glassy membranes, encompasses several types of membranes and membrane materials. A paper by A. Deschamps et al. ("Development of Gaseous Permeation Membranes adapted to the Purification of Hydrocarbons", LLF - LLR - Commission A3, Paris, 1989) describes work with aromatic polyimides having an intrinsic material selectivity of 80 for carbon dioxide over methane and 200,000 for water vapor over methane. The paper defines the target selectivities that the researchers were aiming for as 50 for carbon dioxide/methane and 200 for water vapor/methane.
The paper, which is principally directed to dehydration, does not give carbon dioxide/methane selectivities, except to say that they were °'gencxally low", even though the experiments were carried out with pure gas samples. In other words, despite the high intrinsic selectivity of 80, the lower target value of 50 could not be reached.
British Patent number 1,478,083, to Klass and Landahl, presents a large body of permeation data obtained with methane%arbon dioxide/hydrogen sulfide mixed gas streams and polyamide (nylon 6 and rrylon 6/6), polyvirryl alcohol (PVA), polyacrylonitrile (PAN) and gelatin membranes. Some unexpectedly high selactivities are shown. For the rrylon membranes, carbon dioxide/methane selectivities of up to 30, and hydrogen sulfide/methane selectivities up to 60, are reported. The best carbon dioxide/methane selectivity is 160, for PAN at a temperature of 30 °C and a feed pressure of 448 kPa (65 psia); the best hydrogen sulfide/methane selectivity is 200, for gelatin at the same conditions. In both cases, however, the permeability is extremely low: for carbon dioxide through PAN, less than 5 x 10'' Barer and for ~ ,. t~s ,-e., ~
WO 95!11739 ~ ~ ~ PCT/US94/12100 hydrogen sulfide through gelatin, less than 3 x 10'' Barrer. These low permeabilities would make the transmembrane fluxes miserable for any .practical purposes. It is also unknown whether the gelatin membrane, which was plasticized with glycerin, would be stable much above the modest pressures under which it was tested.
U.S. Patent 4,561,864, also to Klass and Landahl, incorporates in its text some of the data reported in the British patent discussed above. The '864 patent also includes a table of calculations for cellulose acetate membranes, showing the relationship between "Figure of Merit", a quantity used to express the purity and methane recovery in the residue stream, as a function of "Flow Rate Factor", a quantity that appears to be somewhat akin to stage-cut. In performing the calculations, separation factors (where the separation factor is the ~ of the carbon dioxide/methane selectivity and the hydrogen sulfideJmethane selectivity) of 20 to 120 are assumed. The figures used in the calculations appear to range from the low end of the combined carbon dioxide and hydrogen sulfide selectivides from mixed gas data to the high end of the combined selectivities calculated from pure gas data.
A paper by D.L. Ellig et al. ("Concentration of Methane from Mixtures with Carbon Dioxide by permeation through Polymeric Films", Journal of Membrane Science, 6, 259-263, 1980) summarizes permeation tests carried out with 12 different commercially available films and membranes, using a mixed gas feed containing 60% carbon dioxide, 40% methane, but no hydrogen sulfide or water vapor. The tests were carried out at 2,068 kPa (about 300 psi) feed pressure. The results show selectivities of about 9-27 for cellulose acetate, up to 40 for polyeihersulfone and 20-30 for polysulfone. One of the membranes tested was nylon, which, in contradiction to the results reported by Klass and Landahl, showed essentially no selectivity at all for carbon dioxide over methane.
The already much-discussed DOE Final Report by N.N. Li et al. contains a section in which separation of polar gases from non-polar gases by means of a mixed-matrix, facilitated transport membrane is discussed. The membrane consists of a silicone rubber matrix carrying polyethylene glycol, which is used to facilitate transport of polar gases, such as hydrogen sulfide, over non-polar gases, such as mclhane. In tests on natural gas streams, the membranes exhibited hydrogen sulfide/methane selectivity of 25-30 and carbon dioxide/methane selectivity of 7-8, which latter number was considered too low for practical carbon dioxide separation. The membrane was also shown to be physically unstable at feed pressure above about 1276 kPa (170 psig), which, even if the carbon dioxidelmethane selectivity were ~iequate, would render it unsuitable for handling raw natural gas streams.
U.S. Patents 4,608,060, to S.
Kulprathipanja, and 4,606,740, to S. Kulprathipanja and S.S. Kulkarni, of Li's group at UOP, present additional data using the same type of glycol-laden membranes as discussed in the DOE report. In this ~ WO 95/11739 PCT/US94/12100 case, however, pure gas tests were performed and ideal hydrogen sulfideJmethane selectivities as high as 115-185 are quoted. It is interesting to note that these are 4-8 times higher than the later measured mixed gas numbers quoted in the DOE report. The same effect obtains for carbon dioxide, where the pure gas selectivifies are in the range 21-32 and the mixed gas data give selectivities of 7-8.
Similar in concept is U.S. Patent 4,737,166, to S.L. Matson et al., which discloses an immobilized liquid manbrane typically containing n-methylpyrrolidone or another polar solvent in cellulose acetate or any other compatible polymer. The membranes and processes discussed in this patent are directed to selective hydrogen sulfide removal, in other words leaving both the methane and the carbon dioxide behind in the residue stream. As in the UOP patents, very high hydrogen sulfide/methane selectivities, in the range 90-350, are quoted. Only pure gas data are given, however, and the feed pressure is 793 kPa ( 100 psig). There is no discussion as to how the membranes might behave when exposed to multicomponent gas streams and/or high feed pressures. Based on the UOP
teachings, the mixed gas, high-pressure results might be expected to be not so good.
U.S. Patent 4,781,733, to W.C. Babcock et al., describes results obtained with an interfacial composite membrane made by a polycoc~nsation reaction between a diacid-chloride- terminated silicone rubber and a diamine. In pure gas experiments at 793 kPa (100 psig), the membrane exhibited hydrogen sulfide/methane selectivities up to 47 and carbon dioxide/methane selectivities up to 50. No mixed gas or high-pressure data are given.
U.S. Patent 4,493,716, to RH. Swick, reports permeation results obtained with a composite manbrane consisting of a polysulfide polymer on a Gorete~c (polytetra-fluoroethylene) support. Only pure &'~~ I~'-pressure test cell permeability data are given. Based on the reported permeabilities, which only give an upper limit for the methane permeability, the membrane appears to have a hydrogen sulfide/mcthane selectivity of at least 19-42 and a carbon dioxide/methane selectivity of at least 2-6. Some results show that the carbon dioxide permeability increased after exposure to hydrogen sulfide, which might suggest an overall decrease in selectivity if the membrane has become generally more permeable, although no methane data that could confirm or refute this are cited.
U.S. Patent 4,963,165, to I. Blume and I. Pinnau reports pure gas, low-pressure data for a composite membrane consisting of a polyamide-polyether block copolymer on a polyamide support.
Hydrogen sulfide/methane selectivities in the range 140-190, and carbon dioxide/methane selectivities in the range 18-20, are quoted. Mixod gas data for a stream containing oxygen, nitrogen, carbon dioxide and sulfur dioxide are also quoted and discussed in the text, but it is not clear how these data would compare with those for methane- or hydrogen-sulfide-containing mixed gas streams.

~, ~: ~~ a- ~ ~~.; 21. 7 4 3 41 Despite the many and varied research and development efforts that this body of literature represents, cellulose acetate membranes, with their attendant advantages and disadvantages, remain the only membrane type whose properties in handling acid gas streams under real gas-field operating conditions are reasonably well understood, and the only membrane type in commercial use for removing acid gas components from methane.
U.S. Patent 4,589,896, to M. Chen et al., exemplifies the type of process that must be adopted to remove carbon dioxide and hydrogen sulfide from methane and other hydrocarbons when working within the performance limitations of cellulose acetate membranes. The process is directed at natural gas streams with a high acid gas content, or at streams from enhanced oil recovery (EOR) operations, and consists of a multistage membrane separation, followed by fractionation of the acid gas components and multistage flashing to recover the hydrogen sulfide. The acid-gas-depleted residue stream is also subjected ~ ~~ ~~nent to recover hydrocarbons. The raw gas to be treated typically contains as much as 80%
or more carbon dioxide, with hydrogen sulfide at the relatively low, few thousands of ppm level. Despite the fact that the ratio of the carbon dioxide content to the hydrogen sulfide content is high (about 400:1), the raw gas stream must be passed through a minimum of four membrane stages, arranged in a three-step, two-stage configuration, to achieve good hydrogen sulfide removal. The goal is not to bring the raw gas stream to natural gas pipeline specification, but rather to recover relatively pure carbon dioxide, free from hydrogen sulfide, for further use in FOR The target concentration of carbon dioxide in the treated hydrocarbon stream is less than 10%, which would, of course, not meet natural gas pipeline standards.
The methane left in the residue stream after higher hydrocarbon removal is simply used to strip carbon dioxide from hydrogen-sulfite-rich solvent in a later part of the separation process; no methane passes to a natural gas pipeline. Despite the multistep/multistage membrane arrangement, in a representative example, about 7% carbon dioxide is felt in the hydrocarbon residue stream alter processing, and about 12% hydrocarbon loss into the permeate takes place.
It is n to combine treatment by membranes with treatment by non-membrane processes.
As a few sample references, the DOE Final Report by N.N. Li et al., Figure 1, shows such a membrane system upstream of an absorption unit and a Claus plant. U.S. Patent 4,737,166, to S.L. Matson et al., shows an immobilized liquid membrane unit combined with sulfur recovery from the permeate stream and methanation of the residue stream. The W.J. Schell et al. paper presented at the Gas Conditioning Conference, Figure 6, shows conventional treatment, such as amine absorption, of the membrane residue stream. A paper by D.J. Stookey et al. ("Natural Gas Processing with PRISM~
Separators", Environmental Pro~~ress, August 1984, Vol 3, No. 3, pages 212-214) shows various figures in which membrane separation is combined with non-membrane treatment processes. A paper by W.H. Mazur et al. ("Membranes for Natural Gas Sweetening and COz Enrichment", Chemical Engineering Progress, October 1982, pages 38-43) shows several membrane/non-membrane treatment schemes.
In summary, it may be seen that there remains a need for improved membranes and improved processes for handling streams containing methane, acid gas components and water vapor. Such improved membrane processes could, in turn, be combined with non-membrane treatment techniques to provide improved "hybrid" processes.
SUMMARY OF THE INVENTION
In one aspect, the invention provides a process for treating a gas stream comprising hydrogen sulfide and methane, said process comprising: (a) carrying out a membrane separation process, comprising: (i) passing said gas stream across the feed side of a membrane having a feed side and a permeate side; (ii) withdrawing from said feed side a residue stream depleted in hydrogen sulfide compared with said gas stream; (iii) withdrawing from said permeate side a permeate stream enriched in hydrogen sulfide compared with said gas stream; said membrane separation process being characterized in that said membrane, when in use in said process, exhibits a selectivity for hydrogen sulfide over methane of at least 35, measured with a mixed gas stream containing at least hydrogen sulfide and methane, and at a feed pressure of at least 3,552 kPa (500 psig); and (b) passing said permeate or residue stream to a non-membrane process for additional treatment.
The invention provides improved membranes and improved membrane processes for treating gas streams 9a containing hydrogen sulfide, carbon dioxide, water vapor and methane, particularly natural gas streams. The processes rely on the availability of two membrane types: one, cellulose acetate, or a material with similar properties, characterized by a mixed gas carbon dioxide/methane selectivity of about 20 and a mixed gas hydrogen sulfide/methane selectivity of about 25; the other an improved membrane with a much higher mixed gas hydrogen sulfide/methane selectivity of at least about 30, 35 or 40 l0 and a mixed gas carbon dioxide/methane selectivity of at least about 12. These selectivities must be achievable with gas streams containing at least methane, carbon dioxide and hydrogen sulfide and at feed pressures of at least 3550 kPa (500 psig), more preferably 5621 kPa (800 psig), most preferably 7000 kPa (1,000 psig).
An important aspect of the invention is the availability of membranes with much higher hydrogen sulfide/methane selectivities than cellulose acetate. This provides the flexibility to choose between the membrane with the higher carbon dioxide/methane selectivity, in treating streams containing little hydrogen sulfide relative to carbon dioxide; the membrane with the higher hydrogen sulfide/methane selectivity, in treating streams containing substantial amounts of hydrogen sulfide relative to carbon dioxide; and a mixed membrane configuration in treating streams in the intermediate category.
The availability of the two membrane types enables treatment processes balanced in terms of the two membranes, so as to optimize any process attribute accordingly, to be designed. Based on the different permeation properties of the two membrane types, we have discovered that it is possible, through computer modeling, to define gas composition zones in which a particular treatment process is 9b favored. For example, if it is the primary goal to minimize methane loss in the membrane permeate, it may be better to carry out the treatment using only the more hydrogen-sulfide-selective membrane, only the more carbon-dioxide-selective membrane or a mixture of both, depending on the particular feed gas composition. Similar determinations may be made if the amount of membrane area used is to be minimized, the costs and a~agy of rxomprtssion are to be ktpt below a target value, the hydrogen sulfide concentration in the pc~ncate is to be nra,omiaed, the ovcall operating costs arc to be reduced, or any other mtmbratte process attribute is to be the kry design factor.
If a combination of the two membrane types is to be usod, the preferred configuration is to pass the gas stream fast through modules containing the one membrane type, then to pass the residue stream from the fast bank of modules through a second bank containing membranes of the other type. If the raw gas stream contains signif cant amounts of water, for example, it is prcferabk to use the more hydrogen-sulZde-sdcctive membrane first. These mtsabrarres arc not usually damagod by water, and can handle gas streams having very high relative humidifies, up to saturation. Furthermore, the membranes an: very 10 permeable to water vapor, and so can be used to dehydrate the gas stream before it passes to the secmtd bank of modules.
Any membranes that can achieve the nocessary carbmt dioxid~tltane and hydrogat sulGdclmethar~e selxtivities under mixod gas, high-pressure conditions, plus provide commercially useful transmcmbrsne fluxes, can be usod. The most proCemed material for the more carbon-dioxide-sdective membrane is cellulose acetate or its variants. The most preferred material for the more hydrogen-sulfide~
selective membrane is a polyamide-polyether bkxk copolymers having the general formula HOC-PA-C-O-PE---C ~--H
~i a O O
where PA is a polyamide se~nent, PE is a polycther segment and n is a positive integer. Such polymtrs n~
2o are available commercially as Pebax~ from Atochem lnc., Glen Rock, New Jersey or as Vestamid~ from ~.
Nuodex inc., Piscataway, New Jersry.
In their most basic embodiments, the processes of the imrention make use of a one-stage mcmbr5ne design, if a single membrane type is indicated, and a two-step membrane design, in which the residue from the first step becomes the Toed for the second step, if a combination of membrane types is indicated. It is possible, however, to optimize the process in light of the various aspects of gas treatment discussad above, namely removal of impurities, less of mdharre, and ultimate fate of the impiaities. To simultaneously moat pipeline spocificatiarLS, minimize methane kiss and produce a waste stream containing a high hydrogen sulfide cocrcartratian, it may be desirable, for example, to use a two-stago (or more axnplicatod) membrane caa~figuration, in which the permeate from the first stage bocatret the fend for the 3o scud. This will both increase the concentration of hydrogen sulfide in the socond stage pate and reduce the methane loss.
An important aspect of the imatbion is that the manbrarte process is oi>;eu combined with ante or E .- ."~' A
~WO 95111739 21 T 4 3 41 p~~s94/12100 more non-membrane processes, to provide a treatment scheme that delivers pipeline quality methane, on the one hand, and that concentrates and disposes of the acid-gas-laden waste stream, in an environmentally acceptable manner, on the other.
The processes of the invention exhibit a number of advantages compared with previously available acid gas treatment technology. First, provision of a membrane with much higher selectivity for hydrogen sulfide over methane makes it possible, for the first time, to apply membrane treatment efTiciently to gas streams characterized by relatively high concentrations of hydrogen sulfide. Secondly, the processes are much better at handling gas streams of high relative humidity. Thirdly, it is sometimes possible to bring a natural gas stream into pipeline specifications for all three of carbon dioxide, hydrogen sulfide and water vapor with a single membrane treatment. Fourthly, overprocessing of the gas stream by removing the carbon dioxide to a much Beater extent than is actually necessary, simply to bring the hydrogen sulfide content down, can be avoided. Fifthly, much greater flexibility to adjust membrane operating and perforn~ance parameters is provided by the availability of two types of membranes. Sixthly, the process can be optimized for any chosen process attribute by calculating the appropriate membrane mix to use.
BRIEF DESCRIPTION OF THE DRAWINGS
Figure 1 is a diagram showing zones in which particular membranes should be used to separate hydrogen sulfide and carbon dioxide from methane.
Figure 2 is a basic schematic drawing of a one-stage membrane separation process.
Figure 3 is a gaph showing the effect of water vapor on carbon dioxide flux through cellulose acetate membranes.
Figure 4 is a gaph showing the effects of hydrogen sulfide and water vapor on the performance of cellulose acetate membranes.
Figure 5 is a basic schematic drawing of a typical two-stage membrane separation process.
Figure 6 is a basic schematic drawing of a two-step membrane separation process.
Figure 7 is a basic schematic drawing of a two-step/two-stage membrane separation process.
Figure 8 is a basic schematic drawing of a two-stage membrane separation process with an auxiliary membrane unit forming a second-stage loop.
Figure 9 is a diagram showing zones in which particular membranes should be used to separate hydrogen . sulfide and carbon dioxide from methane, based on different hydrogen sulfide/methane selectivities.
Figure 10 is a diagram showing zones in which particular membranes should be used to separate hydrogen WO .95!11739 ° f a '~, '; ° '.; ~ ~ ~ ~ PCT/US94/12100 sulfide and carbon dioxide from methane, based on different carbon dioxide/methane selectivities.
Figure 11 is a diagram showing zones in which particular membranes should be used to separate hydrogen sulfide and carbon dioxide from methane, for different feed gas pressures.
Figure 12 is a basic schematic drawing of a membrane process combined with non-membrane processes to treat both the residue and permeate streams from the membrane unit.
DETAILED DESCRIPTION OF THE INVENTION
The term intrinsic selectivity, as used herein, means the selectivity of the polymer material itself, calculated as the ratio of the permeabilities of two gases or vapors through a thick film of the material, as measured with pure gas or vapor samples.
The term ideal selectivity, as used herein, means the selectivity of a membrane, calculated as the ratio of the permeabilities of two gases or vapors through the membrane, as measured with pure gas or vapor samples.
The terms mixed gas selectivity and actual selectivity, as used herein, mean the selectivity of a membrane, calculated as the ratio of the permeabilities of two gases or vapors through the membrane, as measured W th a gas mixture containing at least the two gases or vapors in question.
The invention has several aspects. In one aspect, the invention concerns processes for treating gas mixtures containing carbon dioxide in certain concentrations, hydrogen sulfide in certain concentrations and methane, to remove the carbon dioxide and hydrogen sulfide.
In another aspect, the invention concerns optimizing such membrane separation processes in terms of a particular process attribute. This optimizing may be done to minimize the methane loss from the membrane process, to maximize the hydrogen sulfide cor~ntration in the permeate stream, or to provide the best fit between the membrane process and a non-membrane process or processes acting together as a "hybrid" process, for example. In yet another aspect, the invention concerns membranes that maintain high hydrogen sulficklmethane selectivities when challenged with mixed gas streams under high pressures. In yet another aspect, the invention concerns combinations of membrane and non-membrane treatment processes.
The processes of the invention rely on the availability of two membrane types:
one, cellulose acetate, or a material with similar properties, characterized by a mixed gas carbon dioxide/methane selectivity of about 20 and a mixed gas hydrogen sulfide/methane selectivity of about 25; the other a membrane with a much higl~r mixed gas hydrogen sulfide/methane selectivity of at least about 30, 35 or and a mixed gas carbon dioxide/methane selectivity of at least about 12. These selectivities must be achievable with gas streams containing at least methane, carbon dioxide and hydrogen sulfide and at feed °

~ ;": :~ ?~. ~". ~, ; ' pressures of at least 3550 kPa (500 psig), more preferably 5620 kPa (800 psig), most preferably 7000 kPa (1,000 psig).
The invention provides three forms of basic membrane treatment process:
1. Using only the more hydrogen-sulfide-selective membrane 2. Using only the more carbon-dioxide-selective membrane 3. Using a combination of both types of membrane.
Based o~n the different per~ation properties of the two membrane types, we have discovered that it is possible, through computer modeling, to define gas composition zones most amenable to each one of these three types of basic processes. In performing the computer calculation, a specific process attribute is used as a basis for calculating the boundaries of the gas composition zones. It will be apparent to those of ordinary skill in the art that any one of many process attributes could serve as the basis for the calculation. Representative, non-limiting, examples include methane loss, membrane area, stage cut, energy consumption, annual operating costs, permeate composition, residue composition, best match with other processes in the treatment train, volume%omposition of recycle streams, and so on.
Loss of methane is usually one of the most important factors in natural gas processing. On the one hand, pipeline grade methane is the desired product, and substantial losses of product have a substantial adverse effect on the process economics. On the other hand, large quantities of methane in the acid gas stream make further handling and recovery of any useful products from this stream much more diihcult. As a general rule, a successful natural gas treatment process should keep methane losses during processing to no more than about 10%, and preferably no more than about 5%.
For simplicity, therefore, most of the discussion and examples have been directed to processes designed to minimize methane losses, although it should be appreciated that the scope of the invention is intended to encompass any process design calculations done with the same goal, namely, defining zones applicable to the various processing options made possible by the two membrane types.
We believe the concept of these zones, how to calculate them and how to use them, is new, and will be useful in treating any gas stream that comprises methane, carbon dioxide and hydrogen sulfide.
such streams arise from natural gas wells, from carbon dioxide miscible flooding for enhanced oil recovery (EOR) and from landfills, for example. We believe that it will be particularly useful in the sweetening of natural gas containing acid gas components.
Referring now to Figure 1, this shows a typical zone diagram, with feed gas carbon dioxide concentration on one axis and hydrogen sulfide concentration on the other. The diagram was prepared by running a series of membrane separation computer simulat5ons for hypothetical three-component (methane, carbon dioxide, hydrogen sulfide) gas streams of particular slow rates and compositions. In all cases, the target was to bring the stream to a pipeline specification of 4 ppm hydrogen sulfide and 2% carbon dioxide. The membrane properties were assumed to be as follows:
$ Mnre C'p~~elective membrane: Carbon dioxide/methane selectivity: 20 Hydrogen sulfide/methane selectivity: 25 Methane flux: 7.5 x 10'~ cm'(STP)/cm2~s~cmHg ore I-I~"S-selective membrane: Carbon dioxide/methane selectivity: 13 Hydrogen sulfidelmethane selectivity: 50 Methane flux: 7.5 x 10~ cm'(STP)/cm2~s~cmHg In each case, the methane loss into the permeate stream that would occur if a one-stage membrane separation process were to be carried out was calculated, and was used to define zones of least methane loss. As can be seen, Figure 1 is divided into four zones. In zone A, no treatment is required, because the gas already contains less than 2% carbon dioxide and less than 4 ppm hydrogen sulfide. In zone B, methane loss is minimized if the more hydrogen-sulfide-selective membrane alone is used. In zone C, methane loss is minimized if the more carbon-dioxide-selective membrane alone is used. In zone D, m~hane loss is minimized by using a combination of the two membrane types. The zones are calculated based on the membrane selectivity and their exact position will change if the membrane selectivity changes. Figures 9 and 10 show the change in the B/D boundary for hydrogen sulfide/methane sclectivities of 30, 40 and 50 and for carbon dioxide/methane selectivities of 10, 13 and 15. As can be seen, the Zone B/D boundary moves to the right as the ability of the membrane to separate carbon dioxide improves. Likewise, the boundary moves to the right as the selectivity for hydrogen sulfide over methane decreases. Although the area where the more hydrogen-sulfide-selective membranes should be used is larger at lower hydrogen sulfide/methane selectivity, the methane losses encountered in using the membrane will be greater. Figure 11 shows the change in the B/D boundary for different feed pressures.
As can be seen, the zone boundary is relatively insensitive to changes in the fced pressure.
The zone diagram may be used directly to determine the best type of membrane to use for a specific separation by reading off the zone into which the feed composition fits.
Another way to use the diagram is to define concentration bands that can serve as guidelines in selecting a membrane process. Again refer ing to Figures 1, 9 and 10, we have discovered that, as a guide, three carbon dioxide concentration bands may be defined, thus:

t-. !': :_. ' 21 l 4 3 41 p~~/~594/12100 1. (a) If the feed gas to the membrane system contains less than about 3%
carbon dioxide to less than about 10% carbon dioxide and more than about 10 ppm hydrogen sulfide to more than about 300 ppm hydrogen sulfide, with the lower end of the carbon dioxide range corresponding to the lower end of the hydrogen sulfide range (<i% carbon dioxide; > 10 ppm hydrogen sulfide) and the upper end of the carbon dioxide range ~ponding to the upper end of the hydrogen sulfide range (<10% carbon dioxide;
>300 ppm hydrogen sulfide), then the most favorable process, in terms of minimizing methane loss, is carried out using the more hydrogen-sulfide-selective membrane only.
(b) If the feed gas contains less than about 10% carbon dioxide to less than about 20% carbon dioxide and more than about 300 ppm hydrogen sulfide to more than about G00 ppm hydrogen sulfide, 10 with the lower end of the carbon dioxide range corresponding to the lower end of the hydrogen sulfide range (<10% carbon dioxide; >300 ppm hydrogen sulfide) and the upper end of the carbon dioxide range corresponding to the upper erg of the hydrogen sulfide range (QO% carbon dioxide; >600 ppm hydrogen sulfide), then the most favorable process, in terms of minimizing methane loss, is carried out using the more hydrogen-sulfide-selective membrane only.
15 (c) If the feed gas contains less than about 20% carbon dioxide to less than about 40% carbon dioxide and more than about 600 ppm hydrogen sulfide to more than about 1%
hydrogen sulfide, with the lower end of the carbon dioxide range c~ponding to the lower end of the hydrogen sulFde range (QO%
carbon dioxide; >G00 ppm hydrogen sulfide) and the upper end of the carbon dioxide range corresponding to the upper end of the hydrogen sulfide range (<40% carbon dioxide; >1%
hydrogen sulfide), then the most favorable process, in terms of minimizing methane loss, is carried out using the more hydrogen-sulfide-selective membrane only.
Also, three hydrogen sulfide concentration bands may be defined, thus:
2. (a) If the feed gas contains less than about 5 ppm hydrogen sulfide to less than about 50 ppm hydrogen sulfide and more than about 3% carbon dioxide to more than about 15%
carbon dioxide, with the lower end of the carbon dioxide range corresponding to the lower end of the hydrogen sulfide range (<5 ppm hydrogen sulfide; >3% carbon dioxide) and the upper end of the carbon dioxide range corresponding to the upper end of the hydrogen sulfide range (<50 ppm hydrogen sulfide; > 15% carbon dioxide), then the most favorable process, in terms of minimizing methane loss, is carried out using the more carbon-dioxide-selective membrane only.
(b) If the feed gas contains less than about 50 ppm hydrogen sulfide to less than about 250 ppm hY~g~ ode and more than about 15% carbon dioxide to more than about 50% carbon dioxide, with the lower end of the carbon dioxide range corresponding to the lower end of the hydrogen sulfide range _.... 2 ~ X4341 1'CTIUS94/12100 (<50 ppm hydrogen sulfide; >15% carbon dioxide) and the upper end of the carbon dioxide range corresponding to the upper end of the hydrogen sulfide range (<250 ppm hydrogen sulfide; >50% carbon dioxide), then the most favorable process, in terms of minimizing methane loss, is carried out using the more carbon-dioxide-selective membrane only.
(c) If the feed gas contains less than about 250 ppm hydrogen sulfide to less than about 500 ppm hydrogen sulfide and more than about 50% carbon dioxide to more than about 85%
carbon dioxide, with the lower end of the carbon dioxide range corresponding to the lower end of the hydrogen sulfide range (<250 ppm hydrogen sulfide; >50% carbon dioxide) and the upper end of the carbon dioxide range con~sponding to the upper end of the hydrogen sulfide range (<500 ppm hydrogen sulfide; >85% carbon dioxide), then the most favorable process, in terms of minimizing methane loss, is carried out using the more carbon-dioxide-selective membrane only.
Also:
3. For feed gas compositions outside the ranges specified in points 1 and 2 above, the most favorable process, in terms of minimizing methane loss, is carried out using a combination of the more hydrogen-sulfide-selective and the more carbon-dioxide-selective membranes.
Another way to express the teachings of the invention is simply to define single limits for the carbon dioxide and hydrogen sulfide concentrations that are best treated by different types of membrane.
This approach gives a less accurate result in any individual circumstance than the zone or band approaches, but gives a broad guide that is useful irrespective of the particular process attribute that is of most concern. Specifically:
1. If the carbon dioxide content of the stream is less than about 40% and the hydrogen sulfide content is more than about 6,000 ppm (1%), the more hydrogen-sulfide-selective membrane should be used.
2. If the carbon dioxide content of the stream is less than about 20% and the hydrogen sulfide content is more than about 500 ppm, the more hydrogen-sulfide-selective membrane should be used.
3. If the carbon dioxide content of the stream is less than about 10% and the hydrogen sulfide content is more than about 10 ppm, the more hydrogen-sulfide-selective membrane should be used.
4. If the hydrogen sulfide content of the stream is less than about 25 ppm and the carbon dioxide content is more than about 10%, the more carbon-dioxide-selective membrane only should be used.

5. If the hydrogen sulfide c~tent of the stream is less than about 100 ppm and the carbon dioxide content is more than about 15%, the more carbon-dioxide-selective membrane only should be used.

6. If the carbon dioxide content of the stream is in the range about 5-20%
carbon dioxide and the hydrogen sulfide content is in the range 10-1,000 ppm, a combination membrane system may be used.

iv ~ ~ 7 4 3 4 l pCT/US94/12100 7. If the carbon dioxide content of the stream is in the range about 10-25%
carbon dioxide and the hydrogen sulfide content is in the range 50-5,000 ppm, a combination membrane system may be used.

8. If the carbon dioxide content of the stream is greater than about 25%
carbon dioxide and the hydrogen sulfide content is greater than about 200 ppm, a combination membrane system may be used.

9. If the carbon dioxide content of the stream is greater than about 40%
carbon dioxide and the hydrogen sulfide content is greater than about 600 ppm, a combination membrane system may be used.
If a combination of the two membrane types is to be used, the simplest configuration is to pass the gas stream first through modules containing the one membrane type, then to pass the residue stream from the first bank of modules through a second bank containing membranes of the other type. The order in which the membrane types are encountered by the gas stream can be chosen according to the specifics of the application. If the raw gas stream contains significant amounts of water and hydrogen sulfide, for example, it is preferable to use the mode hydrogen-sulfide-selective membrane fu~st, since cellulose acetate membranes have been shown to lose both selecfivity and permeability substantially if exposed to combinations of water vapor and hydrogen sulfide. They also do not withstand relative humidifies above about 30% very well. The polyamide-polyether block copolymer membranes that are preferred as the more . hydrogen-sulfide-selective membrane, on the other hand, are not usually damaged by water or hydrogen sulfide, and can handle gas streams having high relative humidifies, such as above 30% RH, above 90%
RH and even saturation. Furthermore, the membranes are very permeable to water vapor, and so can be us~i to dehydrate the gas stream before it passes to the second bank of modules. If humidity and hydrogen sulfide content are not issues, and no other factors that affect only one of the membrane types are at work, then the total methane loss into the permeate streams and the total membrane area required to perform the separation should be essentially independent of the order in which the membranes are positioned.
Arty marrbranes that can achieve the necessary carbon dioxidc/methane selectivity and hydrogen sulfideJmethane selectivity, plus commercially useful transmembrane fluxes, can be used. Preferably the membranes should be characterized by transmembrane methane fluxes of at least I x 10~ cm'(STP)/cm2~s~cmHg, most preferably by transmembrane methane fluxes of at least 1 x 10-5 cm'(STP)/cm2~s~cmHg.
For the more carbon-dioxide-selective membrane, the preferred membranes are the cellulose acetate membranes that are ali~eady in use. Other candidates include different cellulose derivatives, such as ethylcellulose, methylcellulose, nitrocellulose and particularly other cellulose esters. Otherwise, membranes might be made &nrtr polysulfone, polyethersulfone, polyamides, polyimides, polyetherimides, polyacrylonitrile, polyvinylalcohol, other glassy materials or any other appropriate material. Usually, glasry materials have enough mechanical strength to be fornned as integral asymmetric membranes, the production of which is well known in the art. The invention is not intended to be limited to arty particular membrane material or membrane type, however, and encompasses arty membrane, of arty material, that is capable of meerirrg the target pem>cation properties, including, for acample, homogeneous membranes, composite membranes, and membranes incorporating sorbents, carriers or plasticizcrs.
For the more hydrogen-sulfide-seleetivt membrane, the most preferred membranes have hydrophilic, polar elastomeric selective layers. T'he mobility selectivity of such materials, although it favors hydrogen sulfide and carbon dioxide over methane, is modest compared to glasry materials.
Bocat>,sc the membrane is hydrophilic and polar, however, the sorption selectivity strongly favors hydrogen sulfide, carbon dio~tide and water vapor over non-polar hydrophobic gases such as hydrogsn, , propane, bumne, etc. Although the silxtivity of such materials is affected by swelling in the presence of conc~sable components, we have discovered that hydrogen sulGdelmethane selectivities of at least 30 or 35, sometimes at last 40 and sometimes 50, 60 or about can be maintained, even with gas mixtwes containing high acid gas concentrations, even at high relative humidity, and even at high fend pressures up to 3530 kPa (500 psig), 5621 kPa (800 psig), 7000 kPa (1,000 psig) or above. These are unusual and very useful properties. Those properties rarda the membranes unusually suitable for treating natural gas, which offer cattair>s multiple companarts, has high humidity and is at high Preferred merrrbrarte materials are those that exhibit water sorption greater than 5'/0. more preferably greater than 10'/., when exposed to liquid water at room temperatwc. Particularly preferred arc segrsKnted or block copolymers that form two-domain structures, one domain being a Bolt, rubbery, hydrophilic region, the other being harder and glassy or more glassy. Without wishing to be bound by any particular theory of gas transport, we believe that the soft, rubbery domains provide a preferential pathway for the hydrogen sulfide and carbon dioxide components; the harder domains provide mahanical strength and prevent excessive swilling, and hence loss of selectivity, of the soft domains. Polyetha blocks are prefernd for forming the colt flexible domains; most preferably these blacks incorporate polyethylene glycol, polyte;cramethylene glycol or polypropylene glycol, to irtcreasc the sorption of polar molecules by the membrane material.
One specific acample of the most preferred membrane materials that could be used for the more hydrogen-sulfide selective membrane is polyamide-polyctlrer block copolymers having the general formula HO-~C-PA-C-O-PE-C ~--H
~~ ~~ n O O
where PA is a polyamide segment, PE is a poiyelher segment and n is a positive integer. Such polymers are available coerrmereially as Pebsc~ from Atocixm Inc., Gkn Rock, New Jersey or as Vestamid~ from 75136=9 Nuod~c Inc., Piscataway, New Jcrsty. The po!yamide block gives strength and is believed to prevent the membrane swelling excessively in the pr~esettet of water vapor and/or carbon dioxide.
thher specific examples include polyether- and polyester-based polyurethanes.
Iteprescntativc polymer formulations and recipes are given, !err example, in U.S. Patent 5,096,592, in which the oopolymas ate made by first pctparing a pr~epolytnet by combining simple diols and aliphatic or aromatic dicarboxylie acids with an exerss of diacid to prepare diacid-terminated blocks, then chain-extending these with appropriately selected polypropylene or polyethylase glycol segments.
Usually, rubbery materials do not have enough mechanical strength to be formed as integral asymmetric membranes, but are instead incorporated into composite membt~anes, in which the rubbery selective layer is supported on a microporous substrate, otlcn made from a glassy polymer. The prcparatiort of composite membranes is also well latown in the art. It is commonly thought that rubbery composite manbrarxs do not withstand high-pressure operation well, and to date, such membranes have not boon ge~ally used in natural gas treatment, whore fend gas pressura are often as high as 3550 kPa (300 psig) or 7000kPa (1,000 psig). We have Cound, however, that composite membranes, with thin enough rubbery scioctivc layers to provide a transmembrane mcthans flux of at least 1 x 10~ em'(STP)km~~s~cmHg, can be used satisfactorily at high food pressures and not only maintain their integrity but continue to exhibit high selectivity for hydrogtn sulfide over methane.
In their most basic embodiments, the processes of the invention make use of a one-stage membrane design if a single membrane type is indicated, and a two-step membrane design, in which the residue from the first step becomes tbc feed for the second step, if a combination of tnanbratte types is indicated, It will be apparent to those of ordinary skill in the art that more sophisticated embodiments are possible. For example, a two-stage (or more oompiicatod) membrane configuration, in which the pamtate from the fast stage bocames the feed for the second, may be used to further enrich the acid gas content oC
the patruatc stream and to reduce methane losses. It is envisaged that a two-stage msmbrane configuration, using like or unlike membrane types in the two stages will ofiut be used_ In such arrangements, the residue stream from the second stage may be rocirculated for further treatment in the fast stage, or may be passed to the gas pipeline, for example.
M important aspoct of the invrntiat is that the manbrane process is often combined with one or more non-mernbranc processes, to provide a "hybrid" treatment scl>eme that dciivers pipeline quality methane, on the one hand, and that concentrates and disposes of the acid-gas-laden waste stream, in as environmentally acceptable manna, on the other. Given the diversity of flow rates, coenposilions and locations of natural gas wells, it is envisioned that the membrane separation process will often form part ~513s-9 of a hybrid trtatma~t schane that includes multiple treatment steps in an integrated tratmrnt train. Such schenxs might gma~ally be described as having four components, any subset of which may be needed in a specific situation, namely:
1. Primary bulk separation of acid gas components from.the feed gas stream.
5 2. Additional treatment of the swoeteaod product gas stream to melt pipeline specifications.
3. Additional trcatmait of the waste gas stream to canoartratc the by&ogcn sulfide and to reduce methane losses.
4. Disposallcanversioa of the hydrogen sulfide.
In the processes of the invention, the first component of the scheme, primary bulk separation, is 10 accornplishod by membrane separation. Depending on the feed composition, configuration of the membrane system and operating conditioru, additional non-membrane treatment of the sweetened product gas strrram may or may not be necessary. If a non-membrane treatment process is used, it may be of arty appropriate type, such as absorption, adsorption, chanical reactiocs a the like. Absorption processes using alkanolamines are widely used in the gas industry at present. The reactivity and relatively low cost, IS particularly of MEA (monoethanolamine) and DEA (diethanolamine), has made them the workhorse sorbrnts of the industry. The absorption process involves passing the acid-gas-laden stream into an aqueous solution oC the amine of choice, which reacts with the hydrogen sulfide and carbai dioxide in the stream. The amine solution is regenerated for further use by heating.
Alternatively, other sacbent solutions, such as hot potassium carbonate, may be used, particularly 20 if the gas stream contains a large amount of acid gas. Potassium carbonate solutions may be regenerated by steam stripping. Pmmotcrs or activators, for example DEA (8e~eld process), arsenic trioxide, selenous acid and tcllurous acid (Giammarco-Vetrocoke process), can be added to the basic potassium TIA
carbonate solution. In applications for the removal of hydrogen sulGdc, tripotassium phosphate (Shell Development Canpany) may be used.
As yet another aliernativc, physical sorbcnts may be used. Representative absorption processes that make use of physical sorbents include the Selexol process f,Norion Comparry), which uses dimethyl ether of polyethylene glycol, the Rectisol and Purisol pivoccsscs (,Lurgi Gesdlschait f-ur Warmetoclrnik), TY TV
Estasolven proctss (Friedrich Uhde GmbI~ and the Sulfinol process (Shell International Research).
A few representative examples of the other types of non-membrane process inchrde specialized ,M ,w >r 3o scavenging or sulfur recovery processes, such as Sulfa-Scrub, Sulfa.Chock, ChemsWreet, Supatron 600, solid iron sponge or solid rinc oxide.
The manbrane process itself can usually provide the additional treatment netded as the third of the four components in the procsss scheme listed above, namely to achitvc sufficient conetntratia~n of hydrogen sulfide, and to reduce methane loss. This is typically accomplished by adding a second stage to the membrane unit, so that the permeate from the first bulk separation stage is subjected to a second treatment.
The fourth compmtertt of the proaxs train is disposal or conversion of the hydrogen sulfide concentrated stream. Preferably, the hydrogen sulfide is concentrated rnough for treatment by a sulfur-fixing process. The most preferred p<ooas is the Claus process, which cossverts hydrogen sulfide to high-quality, saleable sulfur. The basic steps in the process involve burnirsg the acid gas with stoichiometric amounts of air so that about 113 of the hydrogen sulfide is oxidized to sulfur dioxide. The incinerated stream a passed through a waste heat boiia and over a bauxite catalyst at about 370-400°C(700-750°I~.
Unc~r these conditions, the sulfur dioxide and hydrogen sulfide react to fornt elemental sulfur, which is condensed at about 160°C (320°F). The process can be repeated in second and third stages to increase the sulfur yield With a two-stage plant, sulfur removal cfTicicncics of 95%
are typical. The tail-gas frmtt the plant can be treated to meet environmental standards before discharge. For efficient operation of the I 5 Claus plant, the hydrngm sulfide content of the incoming stream should be above about 8 vol%, and most preferably should be higher, such as above 10 vol°/. or above 20 vol°/..
Alternatively, camersion of th: hydrogen sulfide to elemental sulfur can be carried out using s redox process. Such processes arc usually based on bringing the hydrosen sulfide into contact with a liquid suspension of oxidants such as polythionate, iron-cyanide complexes, iron oxide, lhioarsenates or organic catalysts. After several reaction steps, elemental sulfur precipitates. The solvent can then be tooxidizod and rwsed Various corrrmemial rodac prooasas are available, including Manchester, Stretford n A.D.A., Takaha.~~ Thylox, Giammarco-Vetrocoke and Shell Sulfolane. Typically, rodox processes are mane applicable to the recovery of small tautages of sulfur than the Claus process. The sulfur quality is poser than that fran a Claus plant and furiha refuting is needed to make it saleable. Such processes can, houreva, be run with relatively low inlet hydrogen sulfide concrntrations, such as about 2 vol% or more preferably above 4 vol'/°. As a less preferred alternative, the waste stream containing hydrogen sulfide may be flared or used as fuel gas.
A gata~al scharmtic of a process in which both the membrane residue stream and the membrane permeate svrcam are subjected to further non-membrane treatments is shown in Figure I2. Referring to this figure, feed stream 51 enters the membrane unit 50 for treatment. The membrane unit rosy contain one or more barks of membrane modules, of the same a ditlareat types, arranged in any desired oonGgurution, such as ano-stage, multistage, multistep and variations thereof.
As just a few n~-limiting r ~ ;.
E ... ,~. : ~~ ~ ~ 7 4 3 41 examples, the arrangements shown in Figures 5, 6, 7 or 8 could be used.
Partially sweetened residue stream 52, depleted in acid gas content compared with feed stream 51, passes to an non-membrane treatment process 53 far additional acid gas removal. This process may be any appropriate process known in the art, such as absorption and the other processes discussed above.
Treated stream 54 exits the process and passes to a pipeline or other destination. Stream 55, which is enriched in acid gas content compared with stream 5 l, passes to non-c~nbrane treatment process 56. This process may also be any appropriate process known in the art, but is preferably a process such as a redox process or a Claus process that produces sulfur as a useable product. In this case, stream 57 indicates the sulfur product stream.
Otherwise, the process may treat, fix or contain the acid gases in some other fashion, so that stream 57 is simply the discharge stream from that process.
In the zone calculations, the target pipeline specification for the treated gas was assumed to be no more than about 2 vol% carbon dioxide and 4 ppm hydrogen sulfide, which is typical pipeline specification. However, depending on the destination of the gas and s~cific standards to which the gas is subject, it is believed that a carbon dioxide content below about 3 vol%
and a hydrogen sulfade content below about 20 ppm will be acceptable in many situations.
The processes of the invention exhibit a number of advantages compared with previously available acid gas treatment technology. First, provision of a membrane with much higher selectivity for hydrogen sulfide over methane makes it possible, for the fast time, to apply membrane treatment efficiently to gas streams characterized by relatively high concentrations of hydrogen sulfide compared to carbon dioxide. This expands the range of utility of membrane separation substantially. Since membrane systems are light, simple and low-maintenance compared with amine plants, the enhanced ability to use membranes as a treatment option facilitates the exploitation of gas fields off shore or in remote locations.
Secondly, the processes are much better at handling gas streams of high relative humidity, so that less pretreatment of the raw gas stream is necessary. Thirdly, it is sometimes possible to bring a natural gas stream into pipeline specifications for all three of carbon dioxide, hydrogen sulfide and water vapor with a single membrane treatment. Fourthly, overprocessing of the gas stream by removing the carbon dioxide to a mph greater extent than is actually nary, simply to bring the hydrogen sulfide content down, can be avoided. Fifthly, much greater flexibility to adjust membrane operating and performance parameters is provided by the availability of two types of membranes. Sixthly, the process can be optimized for any chosen process attribute by calculating the appropriate membrane mix to use.
The invention is now further illustrated by the following examples, which are intended to be illustrative of the invention, but are not intended to limit the scope or underlying principles of the invention 75136=9 in any way.
EXAMPLES
The examples arc in five seas.

Examples 1-10 are comparative examples that illustrate the performance of various glasry and rubbery polymers exposed to acid gases undo a variety of conditions.
example 1. Pure gas measurements. Po1 n~m~ membranes of two erades (a) A thret-layer composite manbrane was prrpared, using a microporous polyvinylidene fluoride (PVDF) support layer. The support was first coated with a thin, high-Ilex, scaling Iayu, then with a TN
selective layer of polyimide (Matrimid Grade 5218, Ciba-Geigy, Hawthornc, N~.
Membrane stamps were mounted in a test cell and the permeation properties of the membrane were tested with pure carbon dioxide and with pure n>ethane at a Coed pressure of 448 l:Pa (50 psig). The results are listed in Table 1.
(b) A three-layer composite membrane was prepared, using a microporous polyvirrylidene fluwide IS (PVDF) support layer. The support was first coated with a thin, high-flux, sealing layer, then with a selective layer of polyimidc (custom-made 6FDA-IPDA). Membrane stamps were mounted in a test cell and the permeation properties of the membrane were tested with pure carbon dioxide and with pure methane at a feed presswe of 448 kPa (50~ prig). The results are listed in Table 1.
Example 2 Mi~ccd~a,~measuremcnts Polyimide~,membranes of twos (a) Throe-layer composite membranes as in Example I (a) were tested with a gas mixture consisting of 800 ppm hydrogai sulfide, 4 vol°/r carbon dioxide, the balsna mdhane.
The food pressure was 2,793 kPa (390 prig). The results are listed in Table 1.
(b). Thrx-layer compasitc membranes as in Example 1 (b) were tested with a gas mixture consisting of 800 ppm hydrogen sulfide, 4 vol'/. carbon dioxide, the balance methane. Two feed pressures, 2,807 kPa (392 prig) and 4,890 kPa (694 prig), were used. The rrsults arc listed in Table 1.
Example 3 Pwe ear mcaswements PTMSP membrane A composite membrane was prepared by coating a polytrimelhyl-silyflxOpync (PTMSP) layer onto a polyvinylidcnc fluoride (PVD~ support membrane. Membrane stamps were momttod in a test cell and the permeation properties of the membrane was tested with port carbon dioxide and with pure methane at a feed pressure of 448 kPa (50_ prig). The results are listed in Table 1.

, r. ~ ~ >., ° ' WO 95!11739 2 ~ 7 4 3 41 PCT/US94/12100 F~xamgle 4 Mixed eas measurements PTMSP membrane Composite membranes as in Example 3 were tested with a gas mixture consisting of 800 ppm hydrogen sulfide, 4 vol% carbon dioxide, the balance methane. The feed pressure was 2,793 kPa (390 psig). The results are listed in Table 1.
Example 5 Pure eas -measurements Silicone rubber membrane A composite membrane was prepared by coating a silicone rubber layer onto a microporous support membrane. Membrane stamps were mounted in a test cell and the permeation properties of the membrane were tested with pure carbon dioxide and with pure methane at a feed pressure of 448 kPa (50 psig). The results are listed in Table 1.
Example 6 Mixed has measurements Silicone rubber membrane Composite membranes as in Example 5 were tested with a gas mixture consisting of 650 ppm hydrogen sulfide, 4 vol% carbon dioxide, the balance methane. The feed pressure was 759 kPa (95 psig).
The results are listed in Table 1.
Example 7 Pure gas measurements Pol~utadiene membrane A c~nposite membrane was prepared by coating a polybutadiene (Scientific Polymer Products, Ontario, NIA layer onto a PVDF support membrane. Membrane stamps were mounted in a test cell and the permeation properties of the membrane were tested with pure carbon dioxide and with pure methane at a feed pressure of 448 l:Pa (50 psig). The results are listed in Table 1.
E~ple 8 Mixed g_as measurements. Polybutadiene membrane Composite membranes as in Example 7 were tested with a gas mixture consisting of 800 ppm hydrogen sulfide, 4 vol% carbon dioxide, the balance methane. The feed pressure was 2,821 kPa (394 psig). The results are listed in Table 1.

~513s=9 a is The higl~st selectivity for hydrogen sulfide ova nwas only 10.5, which was achievod wi th a poiyimide membrane at about2,862 kPa ( 400 prig) fad pressum, _ _ ." . ..a cr va This comparative e.~campic is from the report by N.N. Li et al. to the ("Membnme S ~p~nt of Enemy epin the P~ ~° ~ n Final Report, Septaaber 1987).
Li et al. acamined the e~~ of water vapor in a fend gas stream of carbon dioxide on trans~b~
flw4 F~~e 3~ ~~ ~ ~ art summarizes their data. For relative humidity of 10% ~ )~ ~
is appreciabk effort ors the carbon dioxide ilu.~c For relative humidifies in the range 18 23'/e, the !>vac 30'/e compared to the dry gas ilu~c, but recovered when the feed was switci~
b~ to as.
~Yg ~~ h~~~es of 3D'/o and higher, the flux declirK was found to be large, rapid end irrcv~.ble.
x~plc 0 Bc~vi~r ~rr.n..m_ . __. . _ . _ V~oar This example is also taken fi,~m the Li et al. report. Figure 4 ~ the data.
Hydrogen Permeation Proputies of Various Glassy and Rubbery Polymer Membranes 75136=9 sulfide has a negligible effect on membrane performance if the feed gas is dry. if both hydrogen sulfide and water vapor arc present, horveva, the trarLSmmembrane ihnc is substantially reduced. Li et al. conclude that the processing of streams containing both high concentrations of hydrogen sulfide and water vapor must be avoided with cellulose acetate membranes.

Examples 11 and 12 show the performance of polyamide-polycther membranes exposed to pure gases. These ~camples are Cran curlier work at Membrane Technology and Research, as already reported in U.S. Patent 4,963,165, since we were not able to make measurements with pure hydrogen sulf~dc.
Exam,~le 11. P~,vamide-~yether membranes. Pure as data A multilaya composite membrane was prepared by coating a polysuifone support membrane first TM
with a thin high-flax, scaling layer, then with a 1 wt% solution of Pebax grade 401 I in i-butarrol. The membrane was t~ with ptuc gases at a temperature of 20°C and a feed pressure of 448 kPa (50 psig).
The results are shown in Table 2.
Examnle 12. Polvamide-polvether membranes. Pure eas data A sxond membrane was prepared using the same materials and technique as in Example I 1. The results oCpure gas tuts with this manbrane are also shown in Table 2. That is good agreement between the sets of results from Examples I 1 and 12.

TM
Permeation Properties of Pcbax 4011 Composite Membranes Tested with Pure Gases r :4 rr.~ ~,C~ ~9u:... o.~ '~4 .~~tE ~j 4 H .tq ~y. ~5.: ~ :.._ ' ~~ '~ ~~ i ~ % Y
L Sf o-c ~
~ 1 ~ . ''' t9 '~
'~
~

a n..? ... ?6 :.,R;:.:::... , .,..Z-_.5;,.,. 4 ,.3:..a.2 .:.. .::t' ,?.g, .
~:;.r,:,k~:5t;..... d: ~..:.. e ;:Y:ai '...v.v,:,:.,&z.:.:r~~... ::::??.... :a"~~
:%~ sF . .
.t ;; ~..:. ::. ~~
::??a. ~~~~~;a .: .>:
' ~
~L.a i. ~,K:h,~~
~~~C :(~

~ . . . I-r 2'v. . ? d ~ ~
- ?1a ''~ ~F~ '~ 4~
~~' r9~? ' x' ~
' S' ' ~

t ~,y o~ l?W'ezcoal; ~~ W .VSS
S2 a, w i }s?
::..o:?.. ..r ::.%Y :~',i ~ '. C : ~ :c r .r' ~~. tir~~.~?~
Wa _ .. ..s: '2 ~.?
' ..::af~:. ~. r.
~ ..yfo..: . ~ ~.
_?li'i.: ' _ .y,::;,'/'i(~ ...
':~%.F. . Z : ;
~~..SFarF,.. , p .
~ .. ~ ' ~ ,:...,, ~
~kyxiY~~ ~
~~
~
:
:
.
' ...
..

: . ..
ZS ~:. i . .~% a:.:~ _.r ~:: .f.:; . y~
uv.\ , :: ::::h, v:J%V.ww:..:: .;~ : r::
.W ~'.::.3..~ t: F ~ ::.j.
i:v;; ...~'. : ...:
:;: ..'. ?. z .. ..: . . . ':
:..::r:::.:~... .......:
.. a\:~:.'.?. .:% : ::. ~'' .:
t%::r.;::;:_-;. ... ..<w?
:. r. ':St :..,,~ ......... : ;
:::::%::::::, .. ..
~ ':S: ' ..... ............ a~ .'.:?i . . ..i ... . . R N:,?~....... . i. ~:
:;'y~.:a...L.c::: .fb' .,r.... >7: ~ :::;,;... :~:v .....~a:.......,:??v:'% .:...,.::...:.::.,..::. :u: . . : . .
:...:..:.;. . ,_ ::
...;;..".,r.:..........:~:::.:.............,......5:..........:i.:y..F?
.,.,.5~?)f...:.~~~~...f.'~<.vF.;.
....... ,'.~:
.......... r.;~.v ':.
....
w:.
.:
,........
..2...
.
~::.
...:...
.:....:
......,.Fs..........:7~..
.,.5>K.>.<:....

11 448 (50 1,650 219 11.9 139 18 prig) I2 448 (50 I,750 185 9.19 190 20 psig) Exam_yles 13-1$ show the performance of polyamide-polyether membranes exposed to gas mixtures under a variety of conditions.
Ex ie A composite membrane was prepared by coating a layer of a polyamide-polyethcc copolyrrxr TM
(Pebax grade 4011) onto n polyvirtylidene fluoride (PVDF) support membrane using the same g~naal techniques as in Example 11. The membrane was tested with a two-component gas mixture c~taining < <. r, r,- ~ ; 1,7 4 3 41 ~WO 95/11739 PCT/US94/12100 4 vol% carbon dioxide, 96 vol% methane at three different feed pressures:
2,807 kPa (392 psig), 4,165 kPa (589 psig) and 6,724 kPa (960 psig). In all cases the permeate side of the membrane was at, or close to, atmospheric pressure and the membrane was at room temperature (23 ° C).
The penmeation results are listed in Table 3.
Example 14 The same type of membrane as in Example 13 was prepared and tested with a two-component gas mixture consisting of 970 ppm hydrogen sulfide, 99.9 vol% methane at three different feed pressures:2,779 kPa (388 psig), 4,159 kPa (588 psig) and 6,793 kPa (970 psig).
In all cases the permeate side of the membrane was at, or close to, atmospheric pressure and the membrane was at room temperature (23 °C). The permeation results are listed in Table 3.
Example 15 The same type of membrane as in Example 13 was prepared and tested with a three_component gas mixture consisting of 870 ppm hydrogen sulfide, 4.12 vol% carbon dioxide and 95.79 vol% methane at three different fend pressures: 2,765 kPa (386 psig), 4,165 kPa (589 psig) and 6,821 kPa (974 psig). In all cases the permeate side of the membrane was at, or close to, atmospheric pressure and the membrane was at room temperature (23 °C). The penmeation results are listed in Table 3.
Example 16 The same type of membrane as in Example 13 was prepared and tested with a three-component gas mixture consisting of 0.986 vol% hydrogen sulfide, 4.12 vol% carbon dioxide and 94.90 vol%
methane at three different feed pressures: 2,786 kPa (389 psig), 4,145 kPa (586 psig) and 6,800 kPa (971 psig). In all cases the permeate side of the membrane was at, or close to, atmospheric pressure and the membrane was at room temperature (23 °C). The permeation results are listed in Table 3.
Example 17 The same type of membrane as in Example 13 was prepared and tested with a three_component gas mixture consisting of 1.83 vol% hydrogen sulfide, 10.8 vol% carbon dioxide and 87.34 vol% methane at a feed pressure of 6,759 kPa (965 psig). The permeate side of the membrane was at, or close to, atmospheric pressure and the membrane was at room temperature (23 °C).
The permeation results are listed in Table 3.
xam le 18 The same type of membrane as in Example 13 was prepared and tested with a three-component gas mixture consisting of 950 ppm hydrogen sulfide, 8.14 vol% carbon dioxide and 75136=9 9L77 vol% methane at three diffcrcrtt fled pressures: 2,800 kPa (391 psig), 4,138 kPa (585 psi~ and 6,793 kPa (970 psig). In all cases the permeate side of the membrane was at, or close to, atmospheric prGSSiac and the membrane was at room temperature (23 ° C). The permeation results are listed in Table 3.

Permeation Properties of a Pebax~ 401 I Composite Membrane with Various Food Gas Compositions at Thret Feed Pressures x ' ...~:':.'>' .: ?~,~.,.:~b. ,.°'~ . . :ri'a ., ~ ~.: .#~'w..~ s' , y. ~_ ~"lennx'~~'i#', ~':.~1&~f~...c..: >r05 r~ <: ~ TA o-:: ~,>': .:. . ..: i». :: :: :. . '~.u.~. . ':'x:.. %:x ..5:~ ~ 6..:.'. ,t.,<.
: f:;.p~y.,~~y.;:~' .. ':~:srR::: ' ~ :~ ,~q , .
c;:in.. ..~';'.;.::...:.,.., t ..eo-., :..~ :a! ..k1 : _ n. , ~ _.
::~-'~ .. .. it : ".fi' ~. ~F :.
~:2:._ 'qty .!;~....;::..'.\.,,t: : .. 't. :.:
},a .,.>:~ ~~a.~ S
::..:.~lCr::..:., :.. : . . . .:'a~ ::.; .:...:::.: C~~ ) ~~:,;, ::.... .: ..
~ .: , .: . . . ;.
.'~Y~' 'A. 'x:n..'Tti:<'.fkk:
S -:i>."~
.;~ ,Si ~.::Mn:/' iy: , v~
,.W
".'~~ .' :_X .,S n ~i::~;
'j: "'.,..' .:.:2k~ '':t .: ~ ;:
5....
ta~:Y... 'r'' Y .':, 1. s'N ' '.r~'n . .5:. : n .. .:
:, x...... °', ~,Y ~' ~'~' : :: ..a~ ,:.~.w.4: .
..... ..
1 ':..<::~1...........
.::.''': . ;;:. :..
,.. :...:: ~ .:.::
~i .:~.~.. X
n%a~... .i .(?'c.. _ ~ 2. '' .
v. :.'.\ _.
:: vv-s~
. . r~:,:.::y . '~'.. .. .::>. -~ ~~ :~'.r''c.: ::
.,. $:,:..,.. ..: . :~. ,.',a~ . .. :. a.~..... '"':.w.'...':... :'y: :.::.
a...... ;.: s ve': ~..
:,. ~.:..~,~:-:;h<.... S~.~. ~C .. >. .. . .CN .. :..~W
..:.~.::. ~? v o S .
s -~.au~o .~ ~Y a.<: sax ~< .. ,'. ~<~ ~~~~'' 1$ 3,~ '~4" 't"~'xi~ . st ~h<
f~r~ :.:~
S, .... x .,<.~~~.. ~. r. :.;~."~ ~ . < w . ~ .,.,.: :~.e.sr~,l~'~ ., ...; ,~.
..
'x.~:~~xs rxx ~::;x 2,807 (392 prig) - 31 1.9 - 17 13 4,165 (589 psig) - 30 1.9 - 16 6,724 (960 psig) - 29 2.0 - 15 2,779 (388 psig) 91 - 1.8 5t 4,159 (588 psig) 74 - 1.8 41 6,973 (970 psig) 73 - 1.8 41 2,765 (386 psig) 140 31 1.9 70 16 4,165 (589 psig) 115 30 2.0 Sb 15 6,821 (974 psi ) I 10 29 2.2 52 14 2,786 (389 psig) 113 32 2.0 55 16 4,145 (586 psig) 103 31 2.0 51 15 6,800 (971 psig) 97 29 2.0 48 14 17 6,739 (965 prig) 121 34 2.4 50 14 2,800 (391 prig) 93 26 1.6 58 16 18 _ _ 4,138 (585 psig) 108 32 2.0 52 15 ~- , 6,793 (970 psig) ! 93 ~ 28 i 1.9 48 14~
The following observations can be made from the data of Examples 13-18:
1 _ The prcsa~ce of carbon dioxide in the food gas appears to increue the fluxes of both hydrogen sulfide and nx;thanc through the membrane. Far example, a comparison oC the rr-salts of Exampk 14, in which the food micame did not contain any carbon dioxide, with those of Examples IS-18, shows that the hydrogen sultide !)axes are about 25% Iowa and the methane fluxes are about 15% lower in Example 14.
The increased flux may be due to swelling of the manbrane by dissolved carbon dioxide.
2, In general, the pressure-normalizod fluxes of hydrogen sulfide and carbon dioxide decrease with inaeasing feed pressure, whereas those of methane increase. The decrease in the hydrogen sulfide and carbon dioxide fluxes may be due to compciitivc sorption, which results in a lower solubility coe~cient (the ratio of concenaation in the polymer to partial pressure) for each component. At the sane time, the polymer swells, rauiling in a higher diffusivity for all components, including methane. The net result is an increase in the methane ilaK and a decrease in the fluxes of the acid gases (hydrogen sulfide and carbon dioxide).
3. The hydrogen sulfiddmcthane scicctivity for three-component mixtures varies from a low of 48 to a high oC70, although all of the measura>ients were made at fairly high feed prcssurcs. The carbon dioxidclmethane selectivity, also at high pressure, is about 14-16.
Example 19. Gas streams containing water vapor The trpaiments oCExamplc 15 were rapeatod using Toed gas streams saturated with water vapor i 5 by bubbling the fend gas through a water reservoir. The acpaiments were carried out at food prasures of 2,772 kPa (387 psig), 4,159 kPa (588 psig) and 6,793 kPa (970 psig). The pctmcate side of the membrane was at, or close to, atmospheric pressure and the membrane was at room icmperature (23 °C).
The permeation results are listed in Table 4.

Permeation Properties of Pebax 4011 Composite Membranes Tested with Water-Sattuatod Gas Mixtures ~\ ~\Ci :G-' ~~'o fov y,,~. -aC' '.'--:. ~<1 Y.\': f".6 ~. . . < ~..' 1 ~ ~ ' .Ct ~.~~. T:;~~rf ? .'?'~ ~ ' ~.. ~
r ~
'S

, ~ahxrd Flu~c , .' .. . : xv0s . : ~rlembtvattaw.c ~ .~,.: . . lecti: .
. ': . .yg:
n , .
:; .
Fccd ..:.
.
': s s.
.
.
;ssf :::
.
,.
, , :.
' . :. r. . W
..:<4. . . a~\ ..
\\'a.,~ > ..T > :
.fi.:;~H r ice. ~;;>:c: ..~~ K
~ ..:." ..G . .~ ' : ..
x.'; . ' ' :k~'.~: Fy .. r4 ,: ~:. ~r 1. ' ~e rnrtt $ ~~~~~
; r Yfc ~.' : :~~ : ' - M
:. \,:.. ~ B:~
'. ,.:7!..>.~...f:\ ~f :: . ~
..~..:~W
a.:~:..a~~?:
x' \ "~,:.i ~ ~~~

Fw . : , ;: : ~ : :
s S?:,.1''..,.,;s~e~,.-~.~i,..x:;,'s:"~....
,.s;N..:.:~..~~~'~:~ ~ .; ,yC
:::::::;J:::::.:>:~: :a . .. .:.,.~ :. :;~'~,....,.~.,::
.. ..~ ..,..,~: . ?:.., :: v!v,,~: ;;.; :F9..H 2.
:~ , i... ... k...~~:i..~; i ~.:o,.::~:o::::::;:. . .~f#.'.
:: i? . ~ .
;.,.. :n~... . y:::i. y. .:..sR ~.'~>4.~
.: :>:..:.., >v'~~' ~i,.
~......;...:..:...., . ~::i .. :~33~.?~..,..
............::$: .:.. . ~.. :..
n. .C. .,.: :~ :. .. ::.~.:. . ./!;'.
;. ........ ... : ':~? , p.. ..7.. ~ . ' :'<:, . . :~.: .. .. . ' .... ' .. (:
...f.. ..;;.;.Y :%:::::::~ . .:,i. .
::'o:':'c::.c::..::.... f 2 . :.:.:::.. : :
.. .. :' .. .. . :. . . .. ..
......,.::.'?:f n ' : .:: .. ...
~: . . te . :..s s . :.~', ,. ' r .., ~
Y. ~ ~
..r..... ~ ?
r.
:..y .::::
.
.., ~
~5 :~fs:;:;:?. ' , .:::i'v:, . . ' :~(3"~:...
iy4,i.'S:, . ,. .. :
\ ... . ~ . ! :
v. : , .. , n1 ' .:...::.. i,'~,in'.'Y . .... ~: : ~ .
:..;,.... iYF .:...:'?..! '..
....\.. .. ~. ._ f ~~ . .~
. v . . . .
. :::::r:a i::.:
~~. x::> :..::> ..:. ' ' .. : F
. ..::R..ia.. . iS
' :.,s.._:; :Y "' ~
' Y ~ :
.. M..v'.. ' ; ' " _. ;'~:::\' ' w: ~-S~'' ~' .; .~ t~s ~~
~ . ~:YS,~~i ~
.~. ; .~c>.
x~, ....
;.... ..::.:, ", .:.rt>-~.,?.
;...
.3W....:
>
;-s..

:i::
:

.... . . .
. :: : .
:.. y .. . , .
k . :.
.,a r: ,r :..,>;':,..
.:::;;......R....a :~.::::.:. . :
5, r i .:
.::: ~..:n: ;: ':~f';;i:a:-X::
:':";,:..: ..... ;;n ,-:.: . '' :df~ ?
.. .. .:..is?'...... . .,.a;~ '.
: x::: . . . .. .. . ".xi~
... .....~.... . ......r::', ~~ : :>.~
. ........::... :v"%.. .i0..u..y ~.
. . ...v .:x: . 25..:3i.
f.

2,772 (387 77.0 18.9 1.03 ?4.9 18.4 psig) 4,159 (588 73.5 20.1 1.20 61.4 16.9 psig) b,793 (970 68.6 18.1 1.17 58.8 15.5 psig) Comparing these results with those of Table 3, it can be seen that the fluxes are considerably lower (about 40-45°/. lower) than those obtained in the absence of water vapor. Neither the hydrogen sulfrdeJrnetharte nor the carbon dioxidrhn«hane scbctivitics, howevsr, change significantly. Furthermore, whrn the membranes were retested with a dry gas strew the fluxes returned to the original values.

w-.v-<<vw~~ 274341 Examples 20-24 show calculations of the performance of representative membrane processes using the more hydrogen-sulfide-selective membrane only. Any of these could be combined with additional non-membrane treatment of the residue or permeate streams.
5 Example 20 A very simple one-stage membrane process was designed to handle a gas stream containing 100 ppm hydrogen sulfide, 0.1 vol% water vapor, 4 vol% carbon dioxide and the remainder methane, at a feed pressure of 6,897 kPa (1,000 psia). A basic schematic of the process is shown in Figure 2, where numeral 1 indicates the bank of membrane modules, and the feed, residue and permeate streams are 10 indicated by numerals 2, 3 and 4 respectively. The process was assumed to use one bank of more hydrogen-sulfide-selective membranes having the following characteristics:
Hydrogen sulfide/methane selectivity: 80 Water vapor/methane selectivity: 1,000 Carbon dioxide/methane selectivity: 12 Methane flux: 1 x 10'~ cm'(STP)/cm2~s~cmHg The compositions and slow rates of the permeate and residue streams were calculated and are given in Table 5.

STREAM FEED RESIDUE PERMEATE

Flow rate (Nm'/min) 28.3 (1,000 25.6 (903 2.7 ( 97 scfm) scfm) scfm) CH4 conc. (vol%) 95.9 98.1 75.6 COZ conc. (vol%) 4.0 1.9 23.2 HZS conc. (ppm) 100 ~ 4 995 Water vapor conc. (vol%)0.1 2 ppm 1.0 The membrane area used to perform such a separation was calculated to be about 70 m2. The stage cut was just under 10% and the methane loss into the permeate was 7.5%.
The process produces a residue stream that simultaneously meets pipeline specification for carbon dioxide, hydrogen sulfide and water vapor. The low grade permeate gas could be sent to the foul gas line.
xample 21 ~
The simple design of Example 20 is only possible for certain cases where the raw stream to be . . treated contains an appropriate balance of hydrogen sulfide and carbon dioxide. In many cases, a more WD 95/11739 ; ~~ ~ ~-' s 21 ~ 4 3 ~ ~ PG"T/US94/12100 complicated, optimized design is needed to improve the methane recovery and meet pipeline specifications without overprocessing.
A process was designed to handle a 28.3 Nm'/min (1,000 scfm) gas stream containing 1,000 ppm hydrogen sulfide, 0.1 vol% water vapor and the remainder methane, so as to keep methane loss in the permeate stream below 2%. The process uses a two-stage membrane separation system in which the permeate from the first bank of membrane modules becomes the feed for the second bank. A basic schematic of the process is shown in Figure 5, where numeral 10 indicates the first stage bank of membrane modules and numeral 18 indicates the second stage bank of membrane modules. The incoming gas stream 9 is at 6,897 kPa (1,000 psia) and is mixed with the residue stream 20 from the second stage to form the feed gas stream 21 to the first membrane stage. The permeate stream 12 from the first stage is mcompressed to 6,897 kPa (1,000 psia) in compressor 13. The compressed stream 14 passes to chiller 15, where water vapor is condensed and water is removed as liquid stream 16.
The non-condensed stream 17 enters the second membrane stage 18, where further separation of hydrogen sulfide takes place. The residue stream from this stage is recirculated within the process. Both membrane stages were assumed to use more hydrogen-sulfide-selective membranes having the following characteristics:
Hydrogen sulfide/methane selectivity: 50 Water vapor/methane selectivity: 1,000 Methane flux: 7.5 x 10~ cm'(STP)/cmz-s-cmHg The compositions and flow rates of the first and second stage permeate and residue streams were calculated and are given in Table 6.
~reur~c STREAM FEED RESIDUE PERMEATE

FIRST STAGE

Flow rate (Nm'/min)34.0 (1,200 27.9 (985 6.1 (215 scfm) CH4 cone. (vol%) scfm) scfm) 98.99 99.82 99.99 Water vapor conc. 0.08 0.0 0.45 (vol%) H2S conc. (vol%) 0.10 4 ppm 0.55 SECOND STAGE

Flow rate (Nm'/min)6.1 (215 scfm)5.7 (202 scfm)0.4 (13 scfm) CH,, conc. (vol%) 99.42 99.89 92.12 Water vapor conc. 330 21 5,015 (ppm) HZS conc. (vol%) 0.55 0.1 7.4 ~' r e' a '' ' ~ 2 ~ l 4 3 41 pCT~S94/12100 WO 95111739 . . .

The membrane area used to perform such a separation was calculated to be about 280 m2 total, 265 m2 in the first stage and 1 S m2 in the second stage. The residue stream 11 from the first stage meets pipeline specifications. The permeate stream 19 from the second stage contains a high enough concentration of hydrogen sulfide to be Passed to a Claus plant foc sulfur recovery unit, or to a liquid redox process, such as LO-CAT, Sulfemac, Hypeaion or Strotfard The overall methane loss into the second stage permeate is very low, at just about 1%.
~ple 22 A process was designed to handle a 28.3 Nm'lmin (1,000 scfm) gas stream containing 1,000 ppm hydrogen sulfide, 4 vol% carbon dioxide and the remainder methane.
The process uses a two stage membrane separation system in which the permeate from the first bank of membrane modules becomes the feed for the second bank. The process schematic is as shown in Figure 5, except that no condenser 15 is used. Numeral 10 indicates the first stage bank of membrane modules and numeral 18 indicates the second stage bank of membrane modules. The incoming gas stream 9 is at 6,897 kPa ( 1,000 psia) and is mixed with the residue stream 20 from the second stage to form the feed gas stream 21 to the first membrane stage. In this case, the permeate stream 12 from the first stage as recompressed to 6,897 kPa (1,000 psia) in compressor I3, then passed without any condensation taking place as compressed stream 17 to the second membrane stage 18, where further separation of hydrogen sulfide takes place. The residue stream from this stage is rccirculated within the process. Both membrane stages were assumed to use more hydrogen-sulfide-selective membranes having the following characteristics:
Hydrogen sulfide/methane selectivity: 50 Carbon dioxide/methane selectivity: 13 Methane flux: 7.5 x 10'~ cm'(STP)/cmZ~s~cmHg The c~npositions of the first and second stage permeate and residue streams were calculated and are given in Table 7.

75136=9 STREAM FEED RESIDUE PERMEATE

FIRST STAGE

Flow rate (Nm'Imin)34.6 (1,220 27.3 (964 7.3 (256 scam) sefm) scam) CH, coot. (vol%) 93.0 98.86 71.8 CO= coot. (vo1/.)G.9 1.14 28.3 H,S cone. (ppm) 1,000 4 4,733 SECOND STAGE

Flow talc (Nm'/min)7.3 (25G scfm)G.2 (220 1.1 (3G
scfm) scfm) CH, coot. (vol%) 71.8 80.0 19.6 CO, coot. (vol%) 28.3 19.9 77.7 H,S cone. (voP/o)0.47 0.1 2.7 The membrane urea used to perform such a separation was calculated to be about 244 m' total, 232 m~ in the first stage and 12 ms in the second stage, The residue stream 11 from the first stage moots pipeline specifications. The permeate stream 19 from the second stage contains a high cr~tgh oancattration of hydrogen sutlde to be passod to a Claus plant far sulfur rocovery unit, or to a liquid re~oc process, such as LO-CAT, Sulfcroc, Hyperion or Stn:tford. The overall methane loss into the second stage permeate is very low, at about 0.7°/..
Example 23 The calculations of Example 22 were repeated with a 28.3 I3m'/cnin (1,000 scfm) gas stream containing 10,000 ppm hydrogen sulfide, 4 vol% carbon dioxide and the remainder methaae. The results arc given in Table 8.

~~ 7~3~-1 WO 95!11739 ' ' PCTIUS94/12100 +~ ~ _.

STREAM FEED RESIDUE PERMEATE

FIRST STAGE ' Flow rate (Nm'/min)37.7 (1,330 26.9 (950 10.8 (380 scfm) scfm) scfm) CHI conc. (vol%) 91.0 99.4 70.0 COZ conc. (vol%) 8.0 0.6 26.5 HZS conc. (ppm) 10,000 4 3.5 vol%

SECOND STAGE

Flow rate (Nm'/min)10.8 (380 sefm)9.4 (330 scfm)1.4 (50 scfm) CH, conc. (vol%) 70.0 78.6 16.2 COZ cone. (vol%) 26.5 20.4 64.6 HzS conc. (vol%) 3.5 1.0 19.2 The membrane area used to perform such a separation was calculated to be about 353 m2 total, 339 m2 in the first stage and 14 m2 in the second stage. The residue stream 11 from the first stage meets pipeline specifications. The permeate stream 19 from the second stage contains a very high hydrogen sulfide concentration. The methane loss is less than 1%.
Examples 21-23 illustrate the benefits of two-stage processes in both reducing methane loss and raising the hydrogen sulfide tration of the waste stream. In Examples 21-23, the feed composition, both raw and after mixing with recycle stream 20, is in zone B.
Example 24 A process was designed to handle a 28.3 Nm'/min (1,000 sefm) gas stream containing 1,000 ppm hydrogen sulfide, the remainder methane. The process uses a membrane separation system as shown in Figure 8. Numerals 38, 44 and 47 indicate the three banks of membrane modules: all contain the more hydrogen-sulfide-selective membrane. The incoming gas stream 36 is at 6,897 kPa (1,000 psia) and is mixed with the residue stream 49 from modules) 47 to form the feed gas stream 37 to the first membrane stage. The permeate stream, 40, from the first stage is recompressed in compressor 42.
Compressor 42 drives two membrane units, the second stage unit, 44, and an auxiliary module or set of modules, 47, that are connected on the permeate side either directly or indirectly to the inlet side of the compressor, so as to form a loop. Thus, permeate stream 48 may be merged with permeate stream 40 to fpm c~nbined stream 41. The recompressed; combined stream, 43, passes as feed to membrane unit 44, and the residue stream, 46, from membrane unit 44 passes as feed to membrane unit 47. Permeate is a ~, ~ ~ :, ~ t'. r withdrawn from the loop as stream 45 and the treated residue exits as stream 39. This system configuration is particularly useful in situations where the hydrogen sulfide content of the raw stream is relatively low, yet flaring is not an option and the stream has to be brought up to a viable concentration for sulfur recovery. A series of calculations was carried out by keeping the area of membrane unit 38 constant, but varying the relative areas of membrane units 44 and 47. The characteristics of the membrane were assumed to be as in Example 22. The results of the calculations are given in Table 9.

Membrane Permeate conc Area (m2) .
Unit Unit 44 Unit 47 Total (vol%) 242 0 18 260 2.65 242 10 11 263 4.26 242 15 8 265 5.77 242 20 6 268 8.92 242 35 2 279 19.7 242 50 0.4 292.4 55.0 The residue stream 39 from the fu~t stage meets pipeline specifications. A
high concentration of hydrogen sulfide in the waste permeate stream can be achieved with an appropriate choice of membrane areas.
This type of design could also be used in situations where combinations of the two membrane types are indicated.

Examples 25-28 deal with streams in which the feed composition is in zone D, so that a combination of membrane types is indicated. Again, any of these membrane processes could be combined with a non-membrane treatment of the residue and/or permeate streams.
Example 25 A process was designed to handle a 28.3 Nm'/min (1,000 scfm) gas stream containing 60 ppm hydrogen sulfide, 15 vol% carbon dioxide and the remainder methane, a composition that falls in Zone D
of Figure 1, but close to the boundary between zones C and D. The process uses a membrane separation system as shown in Figure 7. NurneraLs 23, 26 and 32 indicate the three banks of membrane modules; 23 contains the more hydrogen-sulfide-selective membrane; 26 and 32 contain the more carbon-dioxide-selective membrane. The incoming gas stream 22 is at 6,897 kPa (1,000 psia) and is mixed with the residue stream 34 from the second stage to form the feed gas stream 35 to the first membrane stage. The -a g ;~ 36 -.
residue stream, 24, from the first bank of modules passes as feed to the second bank of the first stage, 26.
In this case, the permeate streams 25 and 28 from the two steps of the first stage are combined as stream 29 to be re~npressed in compressor 30, then passed as compressed stream 31 to the second membrane stage 32. It will be apparent to those of ordinary skill in the art that two separate compressors could be used and the stream combined after compression. Also, in cases where the stream to be treated contains .
water vapor, the system could include a condenser as in Figure 5 to condense permeating water vapor.
The composition of stream 31 was in Zone C, so that the more carbon-dioxide-selective membrane was chosen for the second stage. The characteristics of the two types of membrane were assumed to be as follows:
More hydrogen-sulfide-selective membrane:
Hydrogen sultide/methane selectivity: 50 Carbon dioxide/methane selectivity: 13 Methane flux: 7.5 x 10'~ cm'(STP)/cmz~s~cmHg More carbon-dioxide-selective membrane:
Hydrogen sulfide/methane selectivity: 25 Carbon dioxide/methane selectivity: 20 Methane flux: 7.5 x 10'~ cm'(STP)/cmz~s~cmHg The compositions of the various streams were calculated and are given in Table 10.

Stream CH, conc. (vol%)HZS conc. (ppm)COz cone. (vol%) #

22 85.0 60 15.0 35 84.0 60 16.0 24 90.2 10 9.8 27 98.0 1 2.0 25 36.1 456 63.9 28 52.3 51 47.7 31 45.5 223 54.5 33 7.5 407 92.5 34 78.6 60 21.4 ,~,rf~.; ~'.
~WO 95/11739 21 T 4 3 41 p~~S94/12100 The membrane areas required were as follows: 66 m2 for membrane 23, 120 m2 for membrane 26 and 22 m2 for membrane 32. The residue stream 27 from the first stage meets pipeline specifications. The permeate stream 33 from the second stage contains about 400 ppm hydrogen sulfide and the overall methane loss is about 1%.
xam le 26 A process was designed to handle a 28.3 Nm'/min (1,000 scfm) gas stream containing 200 ppm hydrogen sulfide, 15 vol% carbon dioxide and the remainder methane, a composition that falls in Zone D
of Figure 1. The process uses a membrane separation system as shown in Figure 7. Numerals 23, 26 and 32 indicate the three banks of membrane modules; 23 and 32 contain the more hydrogen-sulfide-selective membrane; 26 contains the more carbon-dioxide-selective membrane. The incoming gas stream 22 is at 6,897 kPa ( 1,000 psia) and is mixed with the residue stream 34 from the second stage to form the feed gas stream 35 to the first membrane stage. The residue stream, 24, from the first bank of modules passes as feed to the second bank of the first stage, 26. As in Example 25, the permeate streams 25 and 28 could be combined before or after recompression, and a condenser to remove water vapor could be included.
1 S The characteristics of the two types of membrane were assumed to be as follows:
More hydrogen-sulfide-selective membrane:
Hydrogen sulf ide/methane selectivity: 50 Carbon dioxide/methane selectivity: 13 Methane flux: 7.5 x 10'~ cm'(STP)/cm2-s~cmHg More carbon-dioxide-selective membrane:
Hydrogen sulfide/methane selectivity: 25 Carbon dioxide/methane selectivity: 20 Methane flux: 7.5 x 10'~ cm'(STP)lcm2~s~cmHg The compositions of the various streams were calculated and are given in Table I 1.

WO 95111739 ' '', ' ~ 'Y ~ ;~' PCT/US94/12100 Stream # CH, conc. (vol%) H2S cane. (ppm)C02 conc. (vol%) 22 85.0 200 15.0 35 64.0 200 36.0 24 68.0 130 32.0 27 98.0 4 2.0 25 13.2 1,443 86.7 28 26.7 294 73.3 31 25.2 427 74.8 33 3.0 1,447 97.0 34 29.9 200 70.1 The membrane areas required were as follows: 21 m2 for membrane 23, 248 m2 for membrane 26 and 17 m2 for membrane 32. The reside stream 27 from the first stage meets pipeline specifications. The permeate stream 33 from the second stage contains about 1,500 ppm hydrogen sulfide and the overall methane loss is about 0.4%. The feed stream to the second stage bank of modules, 32, contains 427 ppm hydrogen sulfide and 75 vol% carbon dioxide, a composition that falls in the more carbon-dioxide-selective membrane zone of the zone diagram. However, since it is not required to meet pipeline specification for the residue stream from the second stage, an optimized design provides better hydrogen sulfide recovery if the more hydrogen-sulfide-selective membrane is used.
Example 27 A process was designed to handle a 28.3 Nm'/min (1,000 sefm) gas stream containing 1,000 ppm hydrogen sulfide,15 vol% carbon dioxide and the remainder methane, a composition that falls in Zone 1) of Figure 1, but close to the boundary of Zone B. The process uses a membrane separation system as shown in Figure 7. Numerals 23, 26 and 32 indicate the three banks of membrane modules; 23 and 32 contain the more hydrogen-sulfide-selective membrane; 26 contains the more carbon-dioxide-selective membrane. The incoming gas stream 22 is at 6,897 kPa (1,000 psia) and is mixed with the residue stream 34 from the second stage to form the feed gas stream 35 to the first membrane stage. The residue stream, 24, from the first bank of modules passes as feed to the second bank of the first stage, 26. As in Examples 25 and 26, the permeate streams 25 and 28 could lx combined before or after recompression, and a condenser to remove water vapor could be included. The characteristics of the two types of membrane "';~ '~ ~~. ~ ~ 2174341 were assumed to be as follows:
More hydrogen-sulfide-selective membrane:
Hydrogen sulfide/methane selectivity: 50 Carbon dioxide/methane selectivity: 13 Methane flux: 7.5 x 10'~ cm'(STP)/cm2-s-cmHg More carbon-dioxide-selective membrane:
Hydrogen sulfidelmethane selectivity: 25 Carbon dioxide/methane selectivity: 20 Methane flax: 7.5 x 10~ cm'(STP)/cmz~s-cmHg The compositions of the various streams were calculated and are given in Table I2.

Stream CHI cone. (vol%)HZS cone. COZ cone. (vol%) # (ppm) 22 84.9 1,000 15.0 35 63.9 1,000 36.0 24 79.7 70 20.3 27 98.0 4 2.0 17.0 3,770 82.7 28 37.4 221 62.6 25 31 26.6 2,084 73.2 33 3.1 7,390 96.2 34 31.5 1,000 68.4 The membrane areas required were as follows: 119 mz for membrane 23, 188 m2 for membrane 2G and 17 m2 for membrane 32. The residue stream 27 from the first stage meets pipeline specifications.
The permeate stream 33 from the second stage contains about 0.7 vol% hydrogen sulfide and the overall methane loss is about 0.4%.
As with Example 25, an optimized design uses the more hydrogen-sulfide-selective membrane for the second stage.

.. ' ~' 2 ~ 7 4 3 41 pCT/US94/121()0 ~,Xample 28 A process was designed to handle a 28.3 Nm'/min ( 1,000 scfm) gas stream containing 100 ppm hydrogen sulfide, 4 vol% carbon dioxide and the remainder methane, a composition that falls in Zone B
of Figure 1, but so close to the bouc~dary of Zone D that the composition is just within Zone D after mixing 5 with the recycle stream from the second membrane stage. The process uses a membrane separation system as shown in Figure 7. Numerals 23, 26 and 32 indicate the three banks of membrane modules; 23 and 32 contain the more hydrogen-sulfide-selective membrane; 26 contains the more carbon-dioxide-selective membrane. The incoming gas stream 22 is at 6,897 kPa (1,000 Asia) and is mixed with the residue stream 34 from the second stage to form the feed gas stream 35 to the first membrane stage. The residue stream, 10 24, from the fast bank of modules passes as feed to the second bank of the first stage, 26. As in Examples 25-27, the permeate streams 25 and 28 could be combined before or after recompression, and a condenser to remove water vapor could be included. The characteristics of the two types of membrane were assumed to be as follows:
More hydrogen-sulfide-selective membrane:
15 Hydrogen sulGde/methane selectivity: ~0 Carbon dioxide/methane selectivity: 13 Methane flux: 7.5 x 10'~ cm'(STP)/cm2~s~cmHg 20 More carbon-dioxide-selective membrane:
Hydrogen sulfide/methane selectivity: 25 Carbon dioxide/methane selectivity: 20 25 Methane flux: 7.5 x 10'~ cm'(STP)/em2~s~cmHg The compositions of the various streams were calculated and are given in Table 13.

:- . ~. 2I 74341 ~V0 95/11739 ~ w '- ' ' PCT/US94/12100 Stream CH4 cone. (vol%)HZS cone (ppm)COZ cone. (vol%) #

22 96.0 100 4.0 35 94.0 100 6.0 24 98.0 4 2.0 27 98.1 4 1.9 25 70.3 741 29.7 28 75.7 57 24.3 31 70.3 737 29.7 33 20.0 3,680 79.6 34 I 81.2 I 99 I 18.8 The membrane areas required were as follows: 131 mz for membrane 23, 1 m2 for membrane 26 and 9 m2 for membrane 32. The residue stream 27 from the first stage meets pipeline specifications. The permeate stream 33 from the second stage contains about 0.4 vol% hydrogen sulfide and the overall methane loss is about 0.5%.
xam le 29 Example 29 also deals with streams in which the feed composition is in zone D, so that a combination of membrane types is indicated, but in this case a simple, one-stage, two step, membrane process is used. The gas stream was assumed to contain 100 ppm hydrogen sulfide, 10 vol% carbon dioxide,1,200 ppm water vapor and the remainder methane, at a feed pressure of 6,897 kPa (1,000 psia).
The process uses a combination process design as in Figure 6, where numeral 23 indicates a more hydrogen-sulfide-selective bank of membrane modules and numeral 26 indicates a more carbon-dioxide-selective bank of membrane modules. The incoming gas stream 22 is at 6,897 kPa (1,000 psia). The residue stream 24 from the first bank of modules forms the feed to the second bank.
The more hydrogen-sulfide-selective membrane was assumed to have the following characteristics:
Hydrogen sulfide/methane selectivity: 50 Water vapor/methane selectivity: 200 Carbon dioxide/methane selectivity: 13 Methane flux: 7.5 x 10'~ cm'(STP)/cmZ~s~cmHg The more carbon-dioxide-selective membrane was assumed to have the following characteristics:

n a ~'; ;. '.'"
WO 95111739 ~ 2 ~ ~ ~ 3 41 PCT/US94/12100 Carbon dioxide/methane selectivity: 20 Hydrogen sulfide/methane selectivity: 25 Water vapor/methane selectivity: 200 Methane flux: 7.5 x 10'~ cm'(STP)/cm2~s~cmHg The compositions and flow rates of the permeate and residue streams from each bank of modules .
were calculated and are given in Table 14.

TABLE

STREAM ~ FEED ~ RESIDUE ~ PERMEATE

FIRST MODULE BANK
(more hydrogen-sulfide-selective membrane) Flow rate (Nm'/min)28.3 (1,000 25.5 (900 sefm)2.8 (101 scfm) scfm) CH4 conc. (vol%) 89.9 94.38 50.35 COZ conc. (vol%) 10.0 . 5.62 49.0 HZO conc. (ppm) 1,200 4 1.18 vol%

HZS conc. (ppm) 100 14 866 SECOND MODULE
BANK (more carbon-dioxide-selective membrane) Flow rate (Nm'/min)25.5 (900 scfm)23.0 (811 scfm)2.5 (88.7 scfm) CHI conc. (vol%) 94.38 97.98 61.47 C02 conc. (vol%) 5.62 2.01 38.53 H20 conc. (ppm) 4 0 40 HZS conc. (ppm) 14 4 105 The total membrane area used is about 135 mz. Residue stream 27 from the second stage meets pipeline specification. If the permeate streams 25 and 28 from the two banks of membrane modules are pooled, the pemxate compositi~ is 507 ppm hydrogen sulfide, 44 vol% carbon dioxide, 0.62 vol% water vapor and 55.5 vol% methane. The methane loss in the pooled permeates is about 12%. This loss could be reduced if the process were optimized.

The examples in Set 5 show s~cific representative combinations of membrane and non-membrane treatment.
Example 30 Membrane plus scav~nrain~ process A process was designod to handle a gas stream contammg 1,000 ppm hydrogen sulfide, 0.1 vol%
water vapor, 4 vol% carbon dioxide and the remainder methane, at a feed pressure of 6,897 kPa WO 95/11739 t t~ ~ '' x~. 1 ~ 217 4 3 41 pCT~S94/12100 (1,000 psia). The process inclu~s a one-stage membrane separation step, followed by a scavenging step to bring the hydrogen sulfide concentration down further to 4 ppm. The scavenging step could be carried out using an iron sponge, for example. The process was assumed to use one bank of more hydrogen-sulfide-selective membranes having the following characteristics:
Hydrogen sulfidelmethane selectivity: 80 Water vapor/methane selectivity: 1,000 Carbon dioxide/methane selectivity: 12 Methane flux: 1 x 10~ cm'(STP~cmZ~s~cmHg The compositions and flow rates of the permeate and residue streams were calculated and are given in Table 15.

STREAM FEED RESIDUE PERMEATE

Flow rate (Nm'/min)2.8 ( 100 scfm)2.5 (90.3 scfm)0.3 (9.7 scfm) CHI conc. (vol%) 95.8 98 74.9 COZ conc. (vol%) 4.0 1.9 23.1 H2S conc. (ppm) 1,000 40 990 Water vapor conc. 0.1 2 ppm 1.0 (vol%) The membrane area used was calculated to be about 70 mz. The stage cut was just under 10% and the methane loss into the permeate was 7.6%. The process produces a residue stream that meets pipeline specification for carbon dioxide and water vapor, but needs further polishing to remove hydrogen sulfide.
Example 31 Process includin amine plant for hyd--~ro yen sulfide removal A process was designed to handle a gas stream containing 0.5 vol % hydrogen sulFde, 20 vol%
droxide and the remainder' rr>ethar~, at a foetl pressure of 6,897 kPa ( 1,000 psia). The process uses a one-stage membrane separation step to carry out a first separation of carbon dioxide and hydrogen sulfide, followed by an amine plant to bring the stream to pipeline specification. The process was assumed to use one bank of more hydrogen-sulfide-selective membranes having the following characteristics:
Hydrogen sulfide/methane selectivity: 50 Carbon dioxide/methane selectivity: 13 Methane flux: 7.5 x 10'~ cm'(STP)/cm2~s~cmHg The compositions and flow rates of the permeate and residue streams were calculated and are given in Table 16.

. . . ~ ... . ;-.
WO 95/11739 2 ~ 7 4 3 41 PCT/US94112100 STREAM FEED RESIDUE PERMEATE

Flow rate (Nm3/min)28.3 (1,000 23.8 (840 scfm)4.5 (160 scfm) scfm) CH, cone. (vol%) 79.5 88.8 30 COZ cone. (vol%) 20 11.1 67 HZS cone. (vol%) 0.5 0.05 2.9 The membrane area used was calculated to be about 70 m2. The stage cut was just under 16% and the methane loss into the permeate was 6%. The process produces a residue stream from which 90% of the hydrogen sulfide and about 50% of the carbon dioxide has been removed.
This residue stream passes to the amine plant for additional treatment to bring it within specification for carbon dioxide and hydrogen sulfide.
Example 32 A process was designed to handle a 28.3 Nm'/min (1,000 scfm) gas stream containing 2 vol%
hydrogen sulfide, 5 vol% carbon dioxide,1,200 ppm water vapor and the remainder methane. The process uses a two-stage membrane separation system in which the permeate from the first bank of membrane modules becomes the feed for the second bank. The basic schematic of the process is as shown in Figure 5, except that in this case, no condenser 15 is used. Both merzbrane stages were assumed to use more hydrogen-sulfide-selective membranes having the following characteristics:
Hydrogen sulfide/methane selectivity: 50 Water vapor/methane selectivity: 500 Carbon dioxide/methane selectivity: 15 Methane flux: 7.5 x 10'~ cm'(STP)/cm2~s~cmHg The compositions and flow rates of the product residue (first stage) and product permeate (second stage) streams were calculated and are given, together with the raw (unmixed) feed figures, in Table 17.

STREAM FEED RESIDUE PERMEATE

Flow rate (Nm'lmin)28.3 (1,000 26.4 (932 scfm)1.9 (68 sefm) scfm) HZS cone. (vol%) 2.0 50 ppm 29.3 COZ cone. (vol%) 5.0 1.0 59.3 Water vapor cone. 1,200 0.0 1.76 vol%
(ppm) CHI cone. (vol%) 92.9 99.0 9.6 ~~~341 W. _c, ~.' r .~
WO 95111739 - ~ ~ -° PCT/US94/12100 The membrane area used to perform such a separation was calculated to be about 250 mz total, 240 mz in the first stage and 10 m2 in the second stage.
The membrane system was considered to be part of a process as shown in Figure 12. Residue stream 52, which, after the membrane treatment meets pipeline specification for carbon dioxide and water 5 vapor, but is still over spec. for hydrogen sulfide at SO ppm, is passed to an iron sponge, 53. The iron sponge removes the remainder of the hydrogen sulfide down to below 4 ppm.
Stream 54 emerging from the iron sponge unit meets pipeline specification for all gases. Permeate stream 55 contains 29.3 vol%
hydrogen sulfide and has a slow rate of 1.9 Nm'/min (68 scfm). This stream is passed to a Claus plant, 56, for conversion to sulfur. Based on the stream content and flow rate, the typical yield, 57 is 10 1,100 kg/day of elemental sulfur. The overall methane loss from the total process is only 0.7%.
Bxample 33 A process was designed to handle a 28.3 Nm'/min (1,000 scfm) gas stream containing 1,000 ppm hydrogen sulFde, 1,200 ppm water vapor, 10 vol% carbon dioxide and the remainder methane. The process uses a membrane separation system as shown in Figure 8. Numerals 38, 44 and 47 indicate the 15 three banks of membrane modules: all contain the more hydrogen-sulfide-selective membrane. The incoming gas stream 36 is at 6,897 kPa (1,000 psia) and is mixed with the residue stream 49 from modules) 47 to form the feed gas stream 37 to the first membrane stage. The permeate stream, 40, from the first stage is racompressed in compressor 42. Compressor 42 drives two membrane units, the second stage unit, 44, and an auxiliary module or set of modules, 47, that are connected on the permeate side 20 either directly or indirectly to the inlet side of the axnpressor, so as to form a loop. Thus, permeate stream 48 may be merged with penmeate stream 40 to form combined stream 41. The recompressed, combined stream, 43, passes as feat to membrane unit 44, acid tl~ residue stream, 46, from membrane unit 44 passes as fend to membrane unit 47. Permeate is withdrawn from the loop as stream 45 and the treated residue exits as stream 39.
25 The characteristics of the membrane were assumed to be as follows:
Hydrogen sulfide/methane selectivity: SO
Carbon dioxide/methane selectivity: 15 Water vapor/methane selectivity: 500 30 Methane flux: 7.5 x 10'~ cm3(STP)/cm2~s~cmHg The compositions and flow rates of the product residue (first stage) and product permeate (second stage) streams were calculated and are given, together with the raw (unmixed) feed figures, in Table 18.

T k. 1 ~'. ~:.
a s '.
WO 95111739 21 l 4 3 41 PCTlUS94/12100 TART.R 18 STREAM FEED RESIDUE PERMEATE

Flow rate (Nm'/min)28.3 (1,000 27.8 (983.5 0.5 (16.5 sefm) scfm) scfm) HZS conc. (ppm) 1,000 400 3.7 vol%

COZ cone. (vol%) 10.0 8.7 86.9 Water vapor cone. 1,200 50 6.6 vol%
(ppm) CHI conc. (vol%) 89.8 91.3 2.8 The membrane area used to perform such a separation was calculated to be about 36 m2 total, 31 m2 in the first stage, 2 m2 in the second stage and 3 mz in the auxiliary stage.
The membrane system was considered to be part of a process as shown in Figure 12. Residue stream 52, which, after the membrane treatment is still substantially over spec. for both hydrogen sulfide and carbon dioxide is sent to an amine absorption unit, 53. The absorber removes the remainder of the hydrogen sulfide to 4 ppm and the carbon dioxide to 2 vol%, and produces a small stream containing 0.5 vol% hydrogen sulfide, the remainder carbon dioxide, which can be flared.
Permeate stream 55 contains 3.7 vol% hydrogen sulfide and has a flow rate of 0.5 Nm'/min (17 scfm). This stream has a flow rate and concentration that is on the low side for a Claus plant. The stream is passed to a redox unit, 56, for conversion to sulfur. Based on the stream content and flow rate, the typical yield, 57 is 35 kg/day of sulfur. The total methane loss is essentially zero, at about 0.05%.
Example 34 A calculation as in Example 33, using the membrane design of Figure 8 and the overall design of Figure 12 was performed. The process was again designed to handle a 28.3 Nm'/min (1,000 scfm) gas stream containing 1,000 ppm hydrogen sulfide, 1,200 ppm water vapor, 10 vol%
carbon dioxide and the remainder methane. In this case, however, more membrane area was used and the system was run at a slightly higher stage cut to extract more hydrogen sulfide and carbon dioxide into the permeate stream.
The characteristics of the membrane were assumed to be as follows:
Hydrogen sulfideJmethane selectivity: 50 Carbon dioxide/methane selectivity: 15 Water vapor/methane selectivity: 500 Methane flux: 7.5 x 10'~ cm'(STP)/cm2~s~cmHg The c~npositions and flow rates of the product residue (first stage) and product permeate (second stage) streams were calculated and are given, together with the raw (unmixed) feed figures, in Table 19.

2 ~ 14341 STREAM FEED RESIDUE PERMEATE

Flow rate (Nm'/min)28.3 (1,000 27.5 (970 scfm)0.8 ( 30 scfm) scfm) HZS cone. (ppm) 1,000 100 3.0 vol%

COZ cone. (vol%) 10.0 7.5 90.7 Water vapor cone. 1,200 0 4.0 vol%
(ppm) CH, cone. (vol%) 89.8 92.5 2.3 The membrane area used to perform such a separation was calculated to be about 93 m2 total, 80 m2 in the first stage, 3 mz in the second stage and 10 m2 in the auxiliary stage.
The membrane system was considered to be part of a process as shown in Figure 12. Residue stream 52, which, after the membrane treatment is still substantially over spec. for both hydrogen sulfide and carbon dioxide is sent to an amine absorption unit, 53. The absorber removes the remainder of the hydrogen sulfide to 4 ppm and the carbon dioxide to 2 vol%, and produces a small stream containing 0.1 vol% hydrogen sulfede, the remainder carbon dioxide, which can be flared.
Permeate stream 55 contains 3 vol% hydrogen sulfide and has a flow rate of 0.8 Nm'/min (30 scfm).
This stream has a flow rate and concentrntion that is on the low side for a Claus plant. The stream is passed to a redox unit, 56, for c~version to sulfur. Based on the stream content and flow rate, the typical yield, 57 is 52 kg/day of sulfur. Comparing this example with Example 33, it may be seen that the sulfur yield is much higher in this case. The methane loss is still very small, at 0.08%.
Example 35 A calculation as in Examples 33 and 34, using the membrane design of Figure 8 and the overall design of Figure 12 was performed. This time, the process was again designed to handle a 28.3 Nm'/min (1,000 scfm) gas stream containing 5 vol% hydrogen sulfide, 1,200 ppm water vapor, 10 vol% carl~n dioxide and the remainder methane. The characteristics of the membrane were assumed to be as follows:
Hydrogen sulfide/methane selectivity: 50 Carbon dioxide/methane selectivity: 15 Water vapor/methane selectivity: 500 Methane flux: 7.5 x 10~ cm'(STP)/cm2~s-cmHg The compositions and flow rates of the product msidue (first stage) and product permeate (second stage) streams were calculated and are given, together with the raw (unmixed) feed figures, in Table 20.

'. ' A , ~ ~~ ~ 174341 STREAM FEED RESIDUE PERMEATE

Flow rate (Nm'/min)28.3 ( 1,000 26.3 (927 sefin)2.0 (73 scfm) scfm) HZS cone. (vol%) 5.0 0.5 62.4 COZ cone. (vol%) 10.0 8.0 35.2 Water vapor cone.1,200 0 1.6 vol%
(ppm) CH, cone. (vol%) 84.9 91.5 0.8 The membrane area used to perform such a separation was calculated to be about 91 m2 total, 76 m2 in the first stage, 4 m2 in the second stage and 11 m2 in the auxiliary stage.
The membrane system was considered to be part of a process as shown in Figure 12. Residue stream 52, which, after the membrane treatment is still substantially over spec. for both hydrogen sulfide and carbon dioxide is sent to an amine absorption unit, 53. The absorber removes the remainder of the hydrogen sulfide to 4 ppm and the carbon dioxide to 2 vol%, and produces a small stream containing 5.8 vol% hydrogen sulfide, the remainder carbon dioxide, which can be flared or sent to further treatment or conversion. Permeate stream 55 contains 62.4 vol% hydrogen sulfide and has a flow rate of 2.0 Nm'/min (73 scfm). This stream is sent to a Claus plant, 56, for conversion to sulfur. Based on the stream content and flow rate, the typical yield, 57 is 2364 kg/day (2.6 ton/day) of sulfur.
Example 36 A process was designed to handle a 28.3 Nm'/min (1,000 scfm) gas stream containing 5,004 ppm hydrogen sulfide, 5 vol% carbon dioxide and the remainder methane. The process uses a two-stage membrane separation system as in Example 32. Both membrane stages were assumed to use more hydrogen-sulfide-selective membranes having the following characteristics:
Hydrogen sulfidelmethane selectivity: 50 Water vapor/methane selectivity: 500 Carbon dioxidelmethane selectivity: 15 Methane flux: 7.5 x 10'~ cm3(STP)/cm2~s~cmHg The compositions and flow rates of the product residue (first stage) and product permeate (second stage) streams were calculated and are given, together with the raw (unmixed) feed figures, in Table 21.

~_ ~! ~.
WO 95111739 , PCT/US94/12100 STREAM FEED RESIDUE PERMEATE

Flow rate (Nm'/min)28.3 (1,000 27.0 (954 1.3 (46 scfm) scfm) scfm) HZS conc. (ppm) 5,000 50 10.6 vol%

COZ canc. (vol%) 5.0 . 1.5 76.7 CHI conc. (vol%) 94.5 98.5 12.7 The membrane area used to perform such a separation was calculated to be about 192 m2 total, 182 m2 in the fn~st stage and 10 mz in the second stage.
The membrane system was considered to be part of a process as shown in Figure 12. Residue stream 52, which, aittr the membrane treatment meets pipeline specification for carbon dioxide, but is still over spec. for hydrogen sulfide at 50 ppm, is passed to an iron sponge, 53.
The iron sponge removes the remainder of the hydrogen sulfide down to below 4 ppm. Stream 54 emerging from the iron sponge unit meets pipeline specification for all gases. Permeate stream 55 contains 10.6 vol% hydrogen sulfide and has a flow rate of 1.3 Nm'/min (46 scfm). This stream is passed to a redox plant, 56, for conversion to sulfur. Based on the stream content and flow rate, the typical yield, 57 is 288 kg/day of elemental sulfur.
The overall methane loss from the total process is only 0.6%.

Claims (25)

We claim:

1. A process for treating a gas stream comprising hydrogen sulfide and methane, said process comprising:
(a) carrying out a membrane separation process, comprising:
(i) passing said gas stream across the feed side of a membrane having a feed side and a permeate side;
(ii) withdrawing from said feed side a residue stream depleted in hydrogen sulfide compared with said gas stream;
(iii) withdrawing from said permeate side a permeate stream enriched in hydrogen sulfide compared with said gas stream;
said membrane separation process being characterized in that said membrane, when in use in said process, exhibits a selectivity for hydrogen sulfide over methane of at least 35, measured with a mixed gas stream containing at least hydrogen sulfide and methane, and at a feed pressure of at least 3,552 kPa (500 psig);
and (b) passing said permeate stream to a non-membrane process for additional treatment.

2. The process of claim 1, wherein said permeate stream is sufficiently enriched in hydrogen sulfide for treatment in a sulfur-fixing process.

3. The process of claim 1, wherein said permeate stream contains at least about 2 vol% hydrogen sulfide.

4. The process of claim 1, wherein said permeate stream contains at least about 4 vol% hydrogen sulfide.

5. The process of claim 1, wherein said permeate stream contains at least about 8 vol% hydrogen sulfide.

6. The process of claim 1, wherein said non-membrane process comprises an oxidation process.

7. The process of claim 1, wherein said non-membrane process comprises a Claus process.

8. The process of claim 1, wherein said gas stream contains carbon dioxide, hydrogen sulfide and water vapor, all in concentrations above pipeline specification, and wherein said residue stream meets pipeline specifications for carbon dioxide, hydrogen sulfide and water vapor.

9. The process of claim 1, wherein said residue stream is subjected to an additional membrane separation step.

10. The process of claim 1, wherein said residue stream is subjected to a non-membrane treatment process.

11. The process of claim 1, wherein said permeate stream is subjected to an additional membrane separation step prior to said non-membrane treatment process.

12. A process for treating a gas stream comprising hydrogen sulfide and methane, said process comprising:
(a) carrying out a membrane separation process, comprising:
(i) passing said gas stream across the feed side of a membrane having a feed side and a permeate side;
(ii) withdrawing from said feed side a residue stream depleted in hydrogen sulfide compared with said gas stream;
(iii) withdrawing from said permeate side a permeate stream enriched in hydrogen sulfide compared with said gas stream;
said membrane separation process being characterized in that said membrane, when in use in said process, exhibits a selectivity for hydrogen sulfide over methane of at least 35, measured with a mixed gas stream containing at least hydrogen sulfide and methane, and at a feed pressure of at least 3,552 kPa (500 psig);
and (b) passing said residue stream to a non-membrane process for additional treatment.

13. The process of claim 12, wherein said non-membrane process comprises an absorption process.

14. The process of claim 12, wherein said non-membrane process comprises an amine-based absorption process.

15. The process of claim 12, wherein said residue stream is subjected to an additional membrane separation step prior to said non-membrane treatment process.

16. The process of claim 12, wherein said permeate stream is subjected to an additional membrane separation step.

17. The process of claim 12, wherein said permeate stream is subjected to a non-membrane treatment process.

18. The process of claim 1 or claim 12, wherein said permeate stream has a methane content such that methane loss from said gas stream is no more than about 5%.

19. The process of claim 18, wherein said methane loss is no more than about 2%.

20. The process of claim 1 or claim 12, wherein said selectivity for hydrogen sulfide over methane is at least 50.

21. The process of claim 1 or claim 12, wherein said feed pressure at which said selectivity can be obtained is at least 7,000 kPa (1,000 psig).

22. The process of claim 1 or claim 12, wherein said membrane comprises a composite membrane having a selective layer comprising a polymer that is rubbery under operating conditions of the process.

23. The process of claim 1 or claim 12, wherein said membrane comprises a block copolymer containing a polyether block.

24. The process of claim 1 or claim 12, wherein said membrane comprises a polyamide-polyether block copolymer having the general formula wherein PA is a polyamide group, PE is a polyether group and n is a positive integer.

25. The process of claim 1 or claim 12, wherein said gas stream comprises natural gas.