H2 PSA WITH MODIFICATION OF THE FEED GAS FLOW

Abstract

A process for the production of a gas stream with a hydrogen concentration equal to or greater than 99.9% utilizing a pressure swing adsorption unit with a main gas stream having at least 70 mol % of hydrogen, wherein a secondary stream, representing less than 20% of the molar flow rate of the main gas stream and having a hydrogen content of less than 25 mol %, is introduced into the main gas stream upstream of the PSA is presented.

Claims

1.-14. (canceled)

15. A process for the production of a gas stream with a hydrogen concentration equal to or greater than 99.9% utilizing a pressure swing adsorption unit with a main gas stream comprising at least 70 mol % of hydrogen, wherein a secondary stream, representing less than 20% of the molar flow rate of the main gas stream and having a hydrogen content of less than 25 mol %, is introduced into the main gas stream upstream of the PSA.

16. The process as claimed in claim 15, wherein the secondary stream represents less than 10% of the molar flow rate of the main gas stream and has a hydrogen content of less than 15 mol %.

17. The process as claimed in claim 15, wherein the secondary stream has a hydrogen content of less than 1 mol %.

18. The process as claimed in claim 15, wherein the pressure swing adsorption unit utilizes at least one adsorber, comprising an adsorbent or a group of adsorbents and the constituents of the secondary stream are not more adsorbable with regard to the adsorbent or the group of adsorbents used in the PSA than the constituents of the main gas stream.

19. The process as claimed in claim 15, wherein the main gas stream has a hydrogen content of greater than 85 mol.

20. The process as claimed in claim 15, wherein the secondary stream consists, to more than 50 mol %, of methane.

21. The process as claimed in claim 15, wherein the secondary stream consists, to more than 50 mol %, of CO.sub.2.

22. The process as claimed in claim 15, wherein the secondary stream consists, to more than 50 mol %, of a mixture of CO.sub.2 and methane.

23. The process as claimed in claim 15, wherein the main gas stream is a stream resulting from a cryogenic hydrogen/carbon monoxide separation unit or from a refinery gas partial condensation unit.

24. The process as claimed in claim 23, wherein the hydrogen/carbon monoxide cryogenic separation unit comprises an operation of washing with methane.

25. The process as claimed in claim 24, wherein the secondary stream is essentially methane withdrawn from the methane wash.

26. The process as claimed in claim 15, wherein the main gas stream is a synthesis gas.

27. The process as claimed in claim 26, wherein the secondary stream comprises more than 50 mol % of methane and is natural gas.

28. The process as claimed in claim 15, wherein the secondary stream flow rate is regulated as a function of the flow rate of the main gas stream and/or as a function of the content of a constituent in the main gas stream or in the main gas stream and secondary stream mixture.

Description

EXAMPLES

Example 1

[0058] In the first example, a PSA unit which produces hydrogen of very high purity from a waste gas resulting from a hydrogen/carbon monoxide cryogenic separation unit, the mean composition of which is 98.3% H.sub.2, 0.15% N.sub.2, 0.5% CO, 1% CH.sub.4 and 0.05% CO.sub.2, at 22 bar and 40 C., is considered. The high pressure of the cycle is 22 bars abs. The low pressure of the cycle is 1.6 bar. The specifications of the top product are a minimum of 99.9% of H.sub.2 but with a maximum of 10 ppm of CO, the latter constituent being found to be a poison for the downstream process using hydrogen. The adsorber is formed to approximately 20% of activated carbon and to 80% of molecular sieve.

[0059] Access can be had to a CH.sub.4 source, available at a pressure at least equal to the pressure of the initial feedstock gas of the PSA, a small fraction of which source will be mixed with the feedstock gas at the inlet of the PSA in order to slightly weighten the feedstock gas of the H.sub.2 PSA in methane.

[0060] In this example, the flow rate of waste gas rich in hydrogen resulting from the cryogenic separation unit is fixed, as is the very-high-purity hydrogen production flow rate to be provided. The standard yield of the PSA associated with this production flow rate is 86.5% in order to meet demand. The aim is thus to minimize the capital cost of the PSA while keeping a hydrogen yield at least equal to 86.5%. The cycle chosen in order to obtain this yield is a cycle having two equilibratings.

[0061] In a first step, the equilibrating stages are not modified, the cycle after addition of the methane remaining identical to the base cycle determined on the main feed alone. The differences in performance obtained as a function of the amount of CH.sub.4 mixed with the feedstock gas are presented in table 1 below. Case 1 corresponds to the reference case (where more CH.sub.4 is not added to the feedstock gas), for which the yield is 86.5%. Case 2 corresponds to the case where a further 2% as molar flow rate of CH.sub.4 are added to the initial feedstock gas. Finally, case 3 corresponds to the case where a further 4% as molar flow rate of CH.sub.4 are added to the initial feedstock gas. Vads, which is the increase in volume of adsorbent of the PSA necessary in order to obtain the purity required for the hydrogen, is revealed in this instance in the table. In practice, and in particular for case 2, there would be no reason to change the design of the PSA, the regulation with regard to the purity taking charge of reducing by a fraction of a second the duration of the adsorption stage, which would not have a secondary effect on the separation. In that way, a gain in yield related to the addition of the methane to the initial feedstock gas is demonstrated. It should also be noted that the change in the performance qualities (/Vads) is not linear as a function of the amount of methane injected. This comes from the fact that, for cases 1 and 2, it is the specification of 10 ppm maximum of CO which is determining with regard to obtaining the purity required for the hydrogen. In case 3, the nitrogen, which is adsorbed with difficulty, begins itself also to be forced out of the adsorbent and the methane itself to leave. It is the constraint related to the specification of 99.9% H.sub.2 minimum which becomes determining and begins to require an increase in the volume of the beds of adsorbants, even if this increase is slight and has a virtually negligible effect on the overall capital cost of the PSA unit.

TABLE-US-00001 TABLE 1 Case Perf. 1 2 3 ref. +0.8 pt +1 pt Vads ref. +0.5% +2.5%

[0062] The gain with regard to the hydrogen production, of the order of 200 Sm.sup.3/h for a PSA producing 20 000 Sm.sup.3/h of hydrogen, is not negligible but it is not in this instance what is being looked for.

[0063] Consequently, in a second step, the number of equilibratings will be reduced for each of cases 2 and 3, so as to regain a yield of 86.5% while this time gaining in productivity. The new differences in performance obtained with respect to case 1 are presented in table 2. It is noticed that, for a given yield of 86.5% (=0), that is to say in fact for a fixed hydrogen production, the fact of adding methane to the initial feedstock gas makes it possible to gain up to 22% less of volume of adsorbent to be installed (cf. change in Vads).

TABLE-US-00002 TABLE 2 Case 1 2 3 ref. +0 pt +0 pt Vads ref. 22% 21%

[0064] Such a gain in productivity may appear surprising as related to a variation of approximately 1 point in the yield (see table 1) but this is due in large part to the fact that the yield desired is particularly high for a gas of this type and that each additional point is difficult to obtain. By reasoning with regard to the losses of hydrogen in the waste product, it is seen that, on passing from case 3 to case 3, an H.sub.2 loss of 8% higher (loss of 12.5% changing to 13.5%) is allowed, which is no longer negligible seen from this angle.

[0065] Such an effect can be demonstrated experimentally, in particular on a pilot unit, by successively treating gases of different composition. This can be a validation stage before the application on an industrial unit. However, it is seen that the amount injected has to be precise and corresponds to a high-level optimization. The only truly industrial means of obtaining such results is to use software for the simulation of adsorption processes which are suited to PSAs. Such software now exists commercially and/or has been developed internally by companies working in the field of the separation of gases. With such tools, it is now possible to change step-by-step the amount of the secondary feed and to search automatically for an optimum with regard to criteria fixed beforehand by the user.

[0066] It would thus have been possible to test the effect of an injection of CO.sub.2 in place of or in addition to methane, it being known that the PSA is already planned to halt that present in the main feed. Nevertheless, on the site envisaged, there is no available source of CO.sub.2 under sufficient pressure and any compression unit, even for a small flow rate, adds an additional cost and an additional complexity which is not desirable. Conversely, there is/are generally one or more sources of methane or of gas very rich in methane (for example with a content of greater than 90 mol %) available at high pressure. Among these sources, mention may be made of natural gas, which is generally one of the starting materials used in the upstream units. This natural gas can undergo various pretreatments intended to remove possible impurities from it, such as sulfur-comprising products, traces of mercury, certain unsaturated hydrocarbons, cyclic compounds, and the like. The fraction constituting the secondary feed will be withdrawn at the most appropriate location, ordinarily after purification.

[0067] In the precise case of example 1, the H.sub.2/CO cryogenic separation unit comprises an operation of washing with methane. The mixture feeding the cold box, namely essentially hydrogen and carbon monoxide also containing of the order of a percent or a few percent of nitrogen and methane, is cooled and then injected at the foot of a column fed at its top with a flow of subcooled liquid methane. On descending in the column, packed with plates or packing, the methane liquefies the CO, a large part of the nitrogen and the methane of the feed. The top of the column is hydrogen containing the residual nitrogen and carbon monoxide and also the amount of methane in equilibrium with the liquid phase which is, at this level, virtually pure methane. This content is of the order of a percent and varies little from one separation unit to another, the column top temperature being approximately180 C. in order for the washing to be effective, while remaining slightly above the solidification point of methane. The washing methane circulates in the unitusing a pump which compresses it from the low to the high pressurewith an outlet via the hydrogen produced (approximately 1% of this flow rate) and an inlet via the feed gas. The inlet being very generally higher than the outlet, the excess methane is normally purged at low pressure. Provision is made in this instance to use the methane of the washing circuit after compression to the washing pressure as makeup constituting the secondary feed. This stream is then at the valid pressure to be injected directly into the main feed without requiring additional means. According to the organization of the line of heat exchanges between the heating fluids and the refrigerating fluids, the makeup can be injected at cryogenic temperature, for example in the form of liquid droplets, into the top fraction of the washing column or else after re-evaporation at ambient temperature. It may be possible to imagine operating the top of the washing column at a higher temperature, in order to directly have 2% or 3%, for example, of methane in the hydrogen stream, but this would be done, except in specific circumstances, to the detriment of the recovery of CO, whereas this is the main production of the cryogenic separation unit, or would complicate the upper part of the washing column.

[0068] From these alternative forms, it is seen at the injection of the secondary stream into the main feed gas may not be done at the actual inlet of the H.sub.2 PSA but, for example, before a heat exchanger or a separator which are located on the main feed circuit. It is the fact of deliberately injecting a gas corresponding to the claimed characteristics in order to modify the composition of the feed of the PSA which matters and not the exact position of the injection, which can in particular be carried out further upstream if the process lends itself thereto.

Example 2

[0069] The second example is that of a PSA unit which produces hydrogen of very high purity from a synthesis gas, itself already rich in hydrogen, but from which it is desired to extract certain impurities in a very high amount, such as N.sub.2, CO, CH.sub.4 and CO.sub.2. In this case, it may also be advantageous, according to the composition of the synthesis gas, to increase the amount of CH.sub.4 in the feedstock gas. This is because: [0070] CH.sub.4 is coadsorbed well with CO.sub.2 on the activated carbon, and is coadsorbed well with CO and N.sub.2 on the molecular sieve. It will thus take the place of the hydrogen on the activated carbon and the molecular sieve, while only weakly replacing the CO.sub.2, CO and N.sub.2 which it is desired to halt on these beds. It should be noted that this phenomenon, which means that there is after all little in the way of interactions between adsorbates (CO.sub.2, CH.sub.4 or CO, N.sub.2, CH.sub.4) is limited to composition ranges or more exactly partial pressure ranges of the different constituents. A very large methane content, corresponding to several bars of partial pressure, would have the consequence, in this case, of seriously interfering with the adsorption of CO, for example. It is seen that the optimization will be a question of proportioning the compositions. [0071] The CH.sub.4 specification in the top product is often relatively flexible. The methane is rarely the impurity which determines the design of the PSA. In many cases, it is thus possible to increase its content in the feed gas without fear of a sudden fall in the productivity, while taking into account the comment of the preceding paragraph. The main constraint is generally the content of CO, which is a poison for many catalysts, in particular hydrogenation catalysts. For commercial hydrogen, specifications requiring a minimum of 99.9% of H.sub.2 and a maximum of 10 ppm of CO are typically found.

[0072] Likewise, it may be advantageous to increase the CO.sub.2 content in the feedstock gas. This is because, by increasing the CO.sub.2 content, the amount of H.sub.2 coadsorbed on the activated carbon is decreased, which makes it possible to reduce the H.sub.2 losses during the regeneration.

[0073] Generally, depending on the new composition of the feedstock gas after addition of certain compounds, it may be necessary to adjust the distribution of the adsorbents, indeed even to add a layer of a new adsorbent. Typically, if light hydrocarbons, ranging from ethylene to pentane, are added to the feedstock gas of a PSA, it is then necessary to install a layer of silica gel dedicated to halting them, upstream of the layer of activated carbon. Likewise, if the content of CO.sub.2 is substantially increased in the feedstock gas of the PSA, it is then necessary to increase the proportions of activated carbon necessary to halt it.

[0074] More specifically, in this instance a PSA unit is considered which produces hydrogen from a synthesis gas resulting from steam reforming, the composition of which is 73.5% H.sub.2, 0.5% N.sub.2, 3% CO, 6.5% CH.sub.4, 16% CO.sub.2, at 25 bar and 40 C. The high pressure of the cycle is 25 bars. The low pressure of the cycle is 1.6 bar. The specifications of the top product are at least 99.9% of H.sub.2 with a maximum of 100 ppm of N.sub.2 and 10 ppm of CO. The adsorber is formed to 60% of activated carbon and to 40% of molecular sieve. The bed of activated carbon is designed in order to halt the CO.sub.2, whereas the bed of molecular sieve is designed in order to halt the CO, the N.sub.2 and the CH.sub.4 not halted on the activated carbon, at the required specifications.

[0075] In this example, the aim is to enhance the hydrogen yield as much as possible in value, so as to produce as much hydrogen as possible at the required purity for a given flow rate of synthesis gas at the inlet. A cycle having four equilibrating stages is thus chosen.

[0076] As above, it is assumed that access is had to a CH.sub.4 source, for example natural gas, available at a pressure at least equal to the pressure of the initial feedstock gas of the PSA, which is very generally the case as it is the starting material for synthesis gas; a fraction of this secondary stream is injected into the main feedstock gas of the PSA in order to slightly change the composition thereof by enriching it in CH.sub.4.

[0077] The differences in performance in terms of hydrogen extraction yield () and of additional volume of adsorbant to be installed (Vads) which are obtained as a function of the amount of CH.sub.4 mixed with the feedstock gas are presented in table 3 below. Case 1 corresponds to the reference case, where more CH.sub.4 is not added to the feedstock gas. Case 2 corresponds to the case where a further 2% as molar flow rate of CH.sub.4 are added to the initial feedstock gas. Finally, case 3 corresponds to the case where a further 4% as molar flow rate of CH.sub.4 are added to the initial feedstock gas.

TABLE-US-00003 TABLE 3 Case Perf. 1 2 3 ref. +0.5 pt +1 pt Vads ref. +2.5% +5%

[0078] It is noticed that the addition of methane to the feedstock gas makes possible significant gains in yield, while the increase in the volume of adsorbants necessary to treat the new feedstock gas remains relatively low. In case 3, 1 point of yield is thus gained, for an installed volume of adsorbants only 5% higher. It should be noted that this increase in volume takes account both of the change in composition of the feedstock gas and of its increase in flow rate.

[0079] For each of the three cases simulated, the design of the bed of molecular sieve was restricted by the specification of 100 ppm of N.sub.2. There was no change in the determining impurity between case 1, case 2 and case 3. This probably explains in part the virtually linear change in the performance qualities as a function of the amount of methane added, in contrast to the case of example 1.

[0080] The injection of methane was favored with respect to that of CO.sub.2, not only because this first fraction is available under pressure but because such a natural gas fraction would have been injected in any case into the low pressure waste product of the H.sub.2 PSA in order to increase the calorific value thereof, this waste gas being used as fuel in the process for the manufacture of the synthesis gas.

[0081] Used as rinse gas, that is to say after the adsorption stage, such a low flow rate of methane (or natural gas) would have only a negligible effect on the performance levels of the PSA, its effect being limited to the inlet zone of the PSA, whereas, as a mixture, it makes it possible to displace admittedly less hydrogen but in practice over the entire volume and not over a very thin layer of adsorbent.

[0082] The two preceding examples are based on a synthesis process starting from natural gas. This natural gas is generally subjected to various pretreatments before passing into the synthesis reactor proper. Throughout these treatments, it remains under pressure and can thus be withdrawn at the most appropriate point in order to act as secondary stream in the H.sub.2 PSA.

[0083] Other gases or mixtures of gases, if they are available, might be used, provided that the simulations show the advantage of such an addition. In practice, mixtures of gases containing constituents which are too difficult to desorb will be avoided. A first approach consists in not using a mixture containing several percent of a constituent which would be more adsorbable than the constituents already present in the main feed. The water which may be present in the main feedstock gas is not taken into account in this rule, as it is a constituent apart which is very adsorbable on many materials and which it is desired to rapidly halt on a first layer of adsorbent which is often dedicated to it.

[0084] As a rule of thumb, the aim may be to increase the content of a constituent which, due to its partial pressure in the main feed and the choice of the adsorbents, lies in the Henry region, that is to say that its adsorption capacity on the adsorbent selected is then virtually proportional to its content in the gas. As long as the conditions are within this region, there is no need in theory for more adsorbent in order to halt the additional amount of impurity: if the amount of impurity is increased, for example by 15%, the adsorption capacity of this impurity will also increase by approximately 15%.

[0085] Nevertheless, it has been observed, on an industrial unit for the production of hydrogen by PSA starting from an H.sub.2, CO, CH.sub.4 and CO.sub.2 mixture in which the hydrogen content was approximately 80 mol % and the CO.sub.2 content was a little more than 10%, that to further add a substantial amount of CO.sub.2 made it possible, here also, to produce more hydrogen, after simple adjusting of the PSA. A simulation proceeds in the same direction but the phenomena coming into play are more complex. The additional CO.sub.2 drives off the hydrogen from the activated carbon but also a portion of the methane, which experiences an increase in its content in the second half of the adsorbent, itself also acting as displacer of the hydrogen, as in the preceding examples.

[0086] On this subject, it may be supposed that the effect described would be less substantial in the case of a PSA having multiple layers of adsorbents, that is to say deploying more than .sub.4 or 5 successive layers of different adsorbants, each well adapted to a specific constituent or even to a given partial pressure range and particularly selective with regard to this constituent. This is another way of optimizing a PSA for a well fixed feed. The disadvantage is that the multiplication of the layers complicates the filling, the various interfaces having to be really horizontal in order not to create imbalances in the adsorber. Such a highly optimized arrangement with regard to a composition can be counterproductive as the composition of the feedstock gas can vary as a function of the operating conditions of the upstream units. A constituent extending too far into the PSA, that is to say into a layer of adsorbent not provided for it, can be adsorbed too strongly and be difficult to regenerate. In the context of the invention, provision can easily be made to regulate flow rate with regard to the secondary stream so as to retain, over time, one and the same content in the overall feed, for example 10 mol % of methane, the content in the main feed varying from 5 to 8 mol %. In the catalytic reactor case, such variations are frequent and a beginning of run composition and an end of run composition varying by several percent are often given. A controlled injection of a secondary stream can make it possible to bring the two compositions closer or to choose a more favorable composition which it is then possible to obtain or at the very least to approach at the beginning and end of life of the catalyst.

[0087] It is thus seen that the principle of the invention, by its aspect of regulation of an overall composition, can be applied to adsorbers comprising a plurality of adsorbing layers which are optimized as a function of the change in the composition of the gas inside the adsorbent.

[0088] The invention is limited to the cases of H.sub.2 PSAs in which a gas essentially containing impurities is injected with the aim of improving the performance qualities, which goes against what would appear to be indicated by simple common sense, which generally is in favor of the feed which is the richest possible in hydrogen. Beyond these applications, the teaching of this development is that, for a given PSA cycle and a typical feed (that is to say, resulting from a known process which is used time and time again, such as synthesis gas reactors, partial cryogenic condensations of refinery gas or of H.sub.2/CO mixture, and the like), there exist compositions more or less favorable to the separation envisaged. The addition of a small amount of a second gas as described in the context of the invention is the solution selected in this instance but there may exist other means of slightly modifying the composition of a gas in order to render it eventually more optimum with respect to the PSA. The tendency will then be to leave more impurities in the gas than in current practice. In a cryogenic process in which a light gas (helium, hydrogen, carbon monoxide) is obtained by partial condensation of the other heavier constituents (that is to say, more easily condensable constituents), the temperature of the gas/liquid separation can thus be reheated by a few degrees. In that way, the gas obtained will contain more condensable constituents, for example from 6 to 8 mol % of methane in hydrogen instead of targeting from 3 to 4%. Likewise, it is possible to adjust the degree of conversion of a catalytic reactor by acting on the reaction temperature or the top composition of a column by acting on the reflux of said column. A person skilled in the art will henceforth have to check whether a simple modification of the upstream process, which may moreover result in a saving in capital cost or in energy, does not result in a saving with regard to the PSA located downstream.

[0089] Generally, this invention can apply to any type of H.sub.2 PSA, in particular to processes employing N adsorbers or N groups of adsorbers, N being between 2 and 24, M of which are simultaneously in the adsorption phase, with M between 1 and N-1, and comprising P equilibratings, P being between 0 and 5. Group of adsorbers is understood to mean adsorbers operating completely in parallel. It is possible, for example, to use .sub.4 adsorbers in parallel, rather than to employ an adsorber with twice the diameter. There is no theoretical limit to the number of adsorbers of a PSA. The biggest units approach 20 adsorbers in service. Beyond this, it is probable that a good solution will be to install two units of 50% size.

[0090] It will be understood that many additional changes in the details, materials, steps and arrangement of parts, which have been herein described in order to explain the nature of the invention, may be made by those skilled in the art within the principle and scope of the invention as expressed in the appended claims. Thus, the present invention is not intended to be limited to the specific embodiments in the examples given above.