by GP TOWLER · Cited by 25 — Chemical absorption of gas-phase compounds into a liquid sorbent is an important industrial process. Examples of this process include recovery of hydrogen

126 KB – 10 Pages

PAGE – 1 ============
Improved Absorber-Stripper Technology for Gas Sweetening to Ultra-Low H2S Concentrations G.P. TOWLER, H.K. SHETHNA, UMIST, Manchester, United Kingdom B. COLE, B. HAJDIK, Bryan Research & Engineering, Inc., Bryan, Texas INTRODUCTION Chemical absorption of gas-phase compounds into a liquid sorbent is an important industrial process. Examples of this process include recovery of hydrogen sulfide and carbon dioxide from acid gases such as natural gas, refinery gas and coke-oven gas using aqueous alkanolami ne solutions, and absorption of sulfur dioxide using alkali metal sulfite-bisulfite solutions. 1,2 The solvents for such processes are chosen such that the absorption step can be reversed by changing the conditions of temperat ure and pressure. Absorption takes place at high pressure and low temperature, giving high loadings of absorbed component in the solvent. The solvent is then sent to a regenerator, which operates under co nditions of high temperature and low pressure, causing desorption of the absorbed components and regenerating the solvent. Heat is added in the regenerator to provide for the heat of reaction, the sensible heat change of the solvent, and to generate a vapor flow for stripping by partially reboiling the solvent. ABSTRACT The removal of trace components from a gas by absorption using a chemical solvent is of importance to the gas processing industry . There is a growing interest in reaching lower outlet concentrations fo r reasons of health and safe ty; however, this requires very high energy use for solvent regener ation. Instead, solid-adsorption-based processes are often used as a secondary treatment step. We have developed new processes for liquid absorption that exploit better understanding of the thermodynamics of chemisorption processe s in mixed solvent systems. The new processes use any conventional solven t and incorporate recycles between the absorber and stripper, by means of which the thermodynamic and process conditions for stripping are optimized to reduce the pr ocess heat requirement at high separation efficiency. Using these new processes it is possible to reach s ub-ppm concentrations of acid gas with considerable savings in ene rgy costs and without requiring use of solid sorbents. The new technology is based on conventional vapor- liquid contacting equipment and is suitable for retrofit to existing plant. Proceedings of the Sevent y-Sixth GPA Annual Conv ention. Tulsa, OK: Gas Processors Association, 1997: 93-100. Bryan Research & Engineering, Inc. Visit our Engineering Resources page for more articles. Bryan Research and Engineering, Inc. – Technical PapersPage 1of 10 Copyright 2006 – All Rights Reserved Bryan Research and Engineerin g, Inc.

PAGE – 2 ============
Alkanolamine absorption processes are efficient for bulk acid gas removal to concentrations up to about 10 ppm, but for lower outlet concentrations the energy required for st ripping increases rapidly, Fi gure 1. Because of this, processes based on adsorption onto a solid sorbent are us ually used to recover the last few ppm of acid gas. 3 These processes can be reversible, for example, pressu re-swing-adsorption, or irre versible. The reversible processes have the advantage that they can process a much larger volume of gas before the bed must be replaced and they can tolerate higher inlet concentrations of acid gas than the irrevers ible processes. They face the disadvantage that they generate a low-pressure off- gas that requires further treatment before it can be emitted. The irreversible processes are usually cheaper to install, allow grea ter flexibility, and can achieve very low outlet concentrations, but they must be fed with a gas containing a low fraction of acid ga s, otherwise the bed is saturated too quickly, and they produce solid waste that is not always suitable for regeneration. 4 There is considerable interest in treatm ent of natural gas to achieve lower H 2S specifications for pipeline distribution. Studies in the UK have sh own that there is a link between pipeline H 2S concentrations and the failure of gas metering and supply equipment. 5 This poses a potential health and sa fety risk, and it has therefore been suggested that the pipeline specific ation for natural gas in the UK should be lowered from 3ppm to 1ppm. 6 This has created a need for the de velopment of technology t hat will allow the lower H 2S specifications to be reached by making simple modifications to existing equipment, with out the installation of an additional plant, and with no new negative impacts on the environment. LNG production also requires CO 2 removal to very low levels. Figure 1. Typical plot of the variation of outlet H2S consum ption with energy use for the conventional single-loop design and the conventional split-loop design. Bryan Research and Engineering, Inc. – Technical PapersPage 2of 10 Copyright 2006 – All Rights Reserved Bryan Research and Engineerin g, Inc.

PAGE – 3 ============
Several process arrangements have been developed to reduce energy use. 7,8 As early as 1934, Shoeld proposed a split-loop absorption cycle, in whic h the bulk of the solvent is remo ved from an intermediate stage of the stripping column and recycled to an in termediate stage of the absorber, Figure 2. 9 In this arrangement only a small portion of the solvent is stripped to the lowest concentration, and a high vapor to liquid ratio for stripping is achieved in the bottom trays of the absorber. This resu lts in lower energy use at low outlet concentrations; however, the reductions in energy use are not enough to make split-loop absorption cheaper than alternative processes and split loop cycles have conseq uently not found much application in industry. 10,11 This poor performance is largely due to thermodynamic inefficiencies in stripping, which will be addressed in this paper. By recognizing that thermodynamic inefficiency results from va riations in the solvent com position as it circulates within the split loop, we are able to propose new designs that permit economic absorp tion to sub-ppm levels. These designs will be illustrated usin g the example of absorption of H2S and CO 2 from natural gas. SIMULATION METHODS This work required study of different proc ess configurations for absorption of H 2S under industrial conditions. Since such studies would be difficult and expensive to carry out on an industrial plant or laboratory scale model, the processes were instead modeled using commercial simu lation software provided by Hyprotech Ltd. (HYSIM) and Bryan Research and Engineering (TSWEET). A discuss ion of the assumptions used in these models and the approximations introduced into the anal ysis by use of modeling in place of experimentation is given by Shethna. 12Both of these simulation packages have been industrially validated, although TSWEET is more widely used in the gas processing industry. Care was ta ken to evaluate the perf ormance of both simulators in predicting the absorption equilibrium at low concentrations of H 2S and CO 2 by comparison with experimental data. Both programs use thermodynamic models based on that developed by Kent and Eisenberg, though each has been fitted using proprietary data. 13 It was found that both programs tend to slightly overestimate the H 2S partial pressure at concentrations below 1ppm; therefore, both programs will underestima te the energy that is required for stripping at very low outlet conc entrations. Since we are primarily intere sted in comparison between different absorption flowsheets, this error is not of concern, as it will affect all of our results equally; however, it should be taken into account if a comparison between absorption and adsorption processes is made. Figure 2. Conventional split-stream absorber-stripper arrangement. 9Bryan Research and Engineering, Inc. – Technical PapersPage 3of 10 Copyright 2006 – All Rights Reserved Bryan Research and Engineerin g, Inc.

PAGE – 4 ============
CONCENTRATION BE HAVIOR OF MIXED SOLVENT SYSTEMS Phase equilibrium of mixed solvents The design of chemis orption processes requires a clear understanding of the equilibri um between the solvent and the dissolved gas. In general, the so lvent consists of an ac tive component, such as an alkanolamine, together with diluents, physical sorption promoters and corrosio n inhibitors. Because of t he presence of these additional components the solubility of t he dissolved gas (solute) is usually given in moles of solute per mole of active sorbent (known as solvent loading). At constant solute partial pr essure, the solubility of the dissolved gas varies with the liquid concentration of the active component. For example, Figure 3 shows the partial pressure of H 2S over methyldiethanolamine (MDEA) solutions of different concentration. It is cl ear from this figure that the more concentrated MDEA solution exerts a higher pa rtial pressure at the same solvent loading. In designing an absorption process, we wish to ac hieve a specified outlet c oncentration of the absorbed component in the absorber column. To achieve this, it is necessary that the st ripped solvent leaving the regenerator must contain a concentration of solute less than that whic h would be in equilibrium with the gas leaving the absorber at the c onditions at the top of the absorber colu mn. The design problem therefore specifies a required solvent loading for the regenerated solution. Si nce a solution that contains a higher concentration of active component exerts a higher partial pressure of solute , it is easier to strip such a solution to achieve the required solvent loading. Thus a high concentration of active so rbent improves the efficien cy of stripping. Since regeneration is the most energy-intens ive stage in the process, the usual choice is to operate with the highest possible concentration of active sorbent, subjec t to the constraints imposed by corrosion. Variation of concentration in absorber-stripper processes Significant variation in solvent concent ration occurs during stripping. Th e vapor for stripping is generated by reboiling the solvent, and becomes enriched in the more volatile components of the solvent. For example, in aqueous alkanolamine systems the vapor is almost entir ely water, because of t he lower volatility of the alkanolamines. Some of this vapor conde nses inside the column to provide t he heat of desorption and sensible heat of the liquid. Since the bulk of desorption occurs on the top few trays of the stripper column, this condensation can be considered to occur almost entire ly on the trays immediately below the liquid feed. The remainder of the vapor is usually recovered in an overhead partial condenser and returned to the process to control the solvent concentration. Typi cally this is achieved by refluxing t he condensate to the stripper column. This reduces the condenser duty, but also reduces the solvent concentration in t he stripper, and consequently gives a less than optimal concentration of solvent for regeneration. Alternat ively, the condensate can be recycled to the stripper column reboiler. This maintains a highe r concentration of alkanolamine in the stripper, which Figure 3. Partial pressure of H2S vs. mol ratio of H2S to alkanolamine for different alkanolamine concentrations at 40 oC as obtained from acid-gas equilibrium calculation using flash stages. Bryan Research and Engineering, Inc. – Technical PapersPage 4of 10 Copyright 2006 – All Rights Reserved Bryan Research and Engineerin g, Inc.

PAGE – 5 ============
improves regeneration performance in si ngle-loop absorber-stripper processes. In split-loop processes, the variation of solvent composition is further comp licated by the presence of the side- draw from the stripper column. Referring to Figure 2, the solvent in streams S and B must be in mass balance with the solvent in feed F. Since the bulk of vapor condens ation occurs in region 1 of the regenerator, the side stream S will be enriched in the more volatile components of the solvent (typically water) and consequently the bottom stream B will be depleted. In t he case of alkanolamine abs orption, this means that the bottom stream will have the maximum concentration permitted, while the feed and side streams will be more dilute. This reduces stripping efficiency and causes an increa sed overall solvent flow, as the carryi ng capacity of the bulk solvent is also reduced. This increased solvent flow has been observed experimentally, alt hough they did not comment on the cause. 11 IMPROVED PROCESS FOR SPLIT-LOOP ABSORPTION The conventional split-loop absorption proc ess can be considerably improved by altering the design to give better control of the solvent concentration, in order to ac hieve the optimum conditions for regeneration. These improvements become particularly beneficial as we seek to achieve lower outlet concentrations of the absorbed component. Two modifications are necessary. The first is to control the composition of the side stream leaving the stripper column by placing a reboiler on this stream to boil off enough water to maintain the same concentration in the side stream and bottom stream. Th e vapor generated by this reboiler can be returned to the upper section of the stripping column. This eliminates the change in concen tration due to condensation in the upper section of the column. Increasing the side stream conc entration (and hence the bulk solvent co ncentration in the stripper feed) increases the solution loading at the bottom of the abso rber, and hence the overall solvent circulation is reduced. In processes using MDEA in water as so lvent, this modification typically r educes the liquid circulation by 20 %. Further improvements can be made by designing the st ripper to create optimum concentration conditions for regeneration. As stated above, the condensate must be returned to the system to maintain the solvent concentration. If the condensate is refluxed to the stripp er column then the concentration of alkanolamine in the liquid in the stripper column will be reduced, reducing the partial pressure of H 2S and hence reducing stripping efficiency. Instead, if the condensate is returned to a lower point in the column then the liquid on the trays above that point will be at th e optimum condition for stripping. The best option is therefore to return the condensate to the lowest point possible in the stripper column, i.e., to the column reboiler. In sp lit-loop processes, however, direct return of the condensate to th e reboiler is not desirable, as the cond ensate will generally be saturated with H2S and returning it directly to the reboiler causes undes irable back-mixing, reducing th e outlet concentration of solute achieved. Instead, the dissolved gas concentrat ion in the condensate can be reduced by returning the condensate to the column a few stages above the solvent feed, and then removing the condensate from the column instead of allowing it to flow down into the stripper. This part ially strips the conde nsate and allows some direct heat transfer between the condensate and the stripper off-gas, reducing the duty of the condenser. This gains all the advantages of refluxing the condensate to th e column without the disadvantage of producing a sub- optimum concentration on the stripping stages. If very low outl et concentrations of acid gas are required, the condensate can be further stripped using a small amount of steam injection in a side stripper column, prior to recycle to the stripper reboiler. If live steam is used, th e process water balance is maintained by taking a water effluent from the condensate or from the glycol dryi ng unit that typically fo llows acid-gas removal. Bryan Research and Engineering, Inc. – Technical PapersPage 5of 10 Copyright 2006 – All Rights Reserved Bryan Research and Engineerin g, Inc.

PAGE – 6 ============
A modified split-loop process embodying these modifications is illustrated in Figure 4. The condensate from partial condenser EX4 is sent to the top sect ion S3 of the stripper column, where it undergoes partial stripping, and is then further stripped to a very low c oncentration of dissolved gas in side-stri pper S4, before being returned to the stripper bottom reboiler EX6. The intermediate reboiler EX5 is used to maintain the same concentration of solvent in the bottom stream and side stream. Only a small portion of the total solvent (typically less than 20%) is stripped to the ultra-low concentration, thus allowing the process to achieve low outlet concentrations with low energy use. The performance of this process will be discussed below. NEW SPLIT-LOOP PROCESS USING VAPOR SUBSTITUTION Shethna and Towler showed that as the ou tlet gas concentration of an absorber-stripper process is reduced to very low values the incremental heat requirement is almo st entirely used to generate vapor flow in the stripper column.14 This results from the non-linear behavior of the vapor-liquid equilibrium in chemisorption processes. The reboiler heat duty can be reduced by substituting a different vapor for solv ent boil-up. An attr active possibility is to use a small recycle of the lean gas from the top of the absorber column, since this gas is almost completely free of H 2S. If the sections of th e stripping column are separated then this recycle gas can be returned to the process feed or put to some other use, d epending on the process and site conditions. Figure 4. A new flow scheme for chemisorption with ultr a-high recovery, thermodynamically efficient regeneration system. Bryan Research and Engineering, Inc. – Technical PapersPage 6of 10 Copyright 2006 – All Rights Reserved Bryan Research and Engineerin g, Inc.

PAGE – 8 ============
a Claus process tail-gas unit or integrated wit h other downstream or upstream processes. There are a number of ways in which this process can be advantageously combined with fixed-bed irreversible absorption technology. For example, the recycle gas can be passed over a bed of solid sorbent to remove H 2S prior to combustion. This substantially reduces the flowrate of gas and the acid gas concentration that must be treated in the fixed bed, thus ensurin g a long bed life. Sorbent beds can also be used to guarantee achievement of ultra-low outlet concentrations duri ng start-up and plant upsets. A furthe r advantage of incorporating fixed-bed absorption processes is that they can remove COS, allowing total removal of sulfur if required. An additional advantage of this design is that the use of a substitute vapor instead of reboiled solvent lowers the partial pressure of solvent vapor in the column and allows the secondary stripper column S2 to operate at a lower temperature than the primary stripper S1. This reduces t he corrosivity of the solvent and might allow the use of cheaper materials of construction such as carbon st eel in place of the conventional stainless steel. 1 It would be relatively straightforward to consider retrof it of an existing plant to this new design. The bulk absorption stages of the split-loop de sign behave similarly to a conventional single-loop absorber and the flows in the polishing section are small. It is therefore possible to use the existing plant for t he bulk separation stages. It may be necessary to re-tray the top stages of the absorber to allo w for the reduction in liquid flow that will occur. The additional equipment needed is limited to a sma ll secondary stripping column and any other equipment associated with the gas recycle. The compressor shown in Figure 5 may not be necessary, depending on the use of the recycle gas. CASE STUDIES Methods for the design of modified split-loop abso rption processes have been described by Shethna and Towler. 14 Optimized computer simulations of the split-loop processes can be generated in a few hours if a good initialization is made using the shortcut design procedures described by Shethna. 12 The performance of the different flowsheets presented above will be illustrated through two case st udies. The first of these is non- selective removal of H 2S from 4 mol% to 0.25 ppm in the presence of 2 mol% CO 2 in CH4, and the second is selective removal of H 2S from 2 mol% to 4 ppm in the presence of 15 mol% CO 2 in CH4. In both cases the feed was taken as 5000 kmol/hr at 35 oC and 40 bar, and the solvent used was 50 wt% MDEA in water. Although MDEA is primarily used for selective absorption, absorption of CO 2 is also possible if monomethyl- monoethanolamine or piperazine is us ed as an absorption activator. 15 Annualized costs for the process equipment were calculated based on correlations given by Douglas. 16 Details of the assumptions made in costing the processes are given by Shethna. 12 In both cases the new designs are compared with conventional flowsheets, although it should be no ted that for the non-selective case it is not possible to achieve an outlet concentration of 0.25 ppm H 2S using single-loop absorption. The results for this process are therefore given for an outlet concentration of 10 ppm, and it is assumed that an additional po lishing process would be necessary. The costs of such polishing processes cannot be assess ed accurately using data available in the literature. Table I. Cost comparison for non-selective removal (inlet gas: 4% H2S, 2% CO2) Process Single-loop absorption Conventional split-stream (Figure 2) New scheme I (Figure 4) New Scheme II (Figure 5) H2S Specification (ppm) 100.25 0.25 0.25 Energy use (kg steam/m3 solvent) 117.7 273.8 82.0 89.5 Solvent flow (m3/hr) 162.1 193.22 167.9 167.0 Power use (kw) 234.4 272.2 233.8 404.2 Annual energy cost (M$) 1.25 2.71 0.91 1.05 Annualized capital cost (M$) 0.97 1.52 0.89 1.12 Capital breakdown (M$) Columns 0.43 0.74 0.33 0.48 Exchangers 0.54 .78 0.56 0.41 Bryan Research and Engineering, Inc. – Technical PapersPage 8of 10 Copyright 2006 – All Rights Reserved Bryan Research and Engineerin g, Inc.

PAGE – 9 ============
Costs for the non-selective absorption ca se are given in Table I, from which it can be seen that the conventional split-stream absorption process is subs tantially more expensive than single -loop absorption to 10 ppm, confirming that conventional split-loop absorpti on is uneconomic compared to single-lo op absorption followed by a polishing process based on adsorption. Both new processes are si gnificantly cheaper than th e conventional split-stream process, mainly due to the reduction in solvent flow and e nergy use. In this particular case, the modified split-loop design of Figure 4 is the cheapest, al though in other cases the gas-recycle design of Figure 5 is better. In the selective removal case, Table II, the three designs achiev e roughly the same selectivity of H 2S removal over CO 2, but the energy use of the new split -loop processes is considerably le ss than that of the conventional single-loop process, as a result of t he improvement in the thermodynamic effici ency of stripping. This reduction in energy costs leads to savings of almo st 30 % in total annualized cost. For the selective removal case the benefits of using the modified split-loop proc esses are even greater than in th e non-selective absorption case. CONCLUSIONS Two new designs for absorption processes have been introduced, based on improved understanding of the solvent behavior in stripping processes. By creating the optimum co nditions for stripping it is possible to achieve significant reduction in process energy use, which allows ultra-low outlet concentrations to be achieved using absorption alone. The first new design is a modification of the conventional split-loop proc ess that allows better control of the solvent composition. The second is an ent irely new process, in which solvent polishing is achieved by recycle of a small portion of the treated gas. The case studies presented showed that these processes have significant cost advantages relative to the conventional technology. ACKNOWLEDGEMENT This research was funded by the members of the UMIST Process Integration Research Consortium. The authors are grateful to Hyprotech Ltd. and Bryan Research and Engineering for providing academic licenses of their software. REFERENCES 1. Kohl, A. and Riesenfeld, F., Gas Purification , Gulf Publishing Co., Houston, 1985. 2. Bailey, E.E. and Heinz, R.W., “SO 2 Recovery Plants – Materials of Construction”, Chem. Eng. Prog ., 71(3): 64- 68, 1975. 3. Collins, C., Durr, C.A., de la Vega, F.F. and Hill, D. K., “Liquefaction Plant Design in the 1990s”, Hydr. Proc., 74(4): 67-76, 1995. 4. Carnell, P.J.H., Joslin, K.W. and Woodham, P.R., “Fixed-bed Processes Provi de Flexibility for COS, H 2S Removal”, Oil & Gas J ., 93(23): 52-55, 1995. 5. Wilson, G., Gas Works Assoc., District Councils Rev., 10, July 1995. 6. Jones, A., “Safety Aspects of Hydrogen Sulphide Concentration in Natural Gas”, HSE Consultative Document, HMO Publishers, London, 1996. Compressor 0000.23 Total annualized cost (M$) 2.22 4.23 1.8 2.16 Bryan Research and Engineering, Inc. – Technical PapersPage 9of 10 Copyright 2006 – All Rights Reserved Bryan Research and Engineerin g, Inc.

PAGE – 10 ============
7. Sigmund, P.W., Butwell, K.F. and Wussler, A.J., “HS Process Removes H 2S Selectively”, Hydr. Proc., 60(5): 118-124, 1981. 8. Benson, H.E. and Parish, R.W., “Hipure Process Removes CO 2 and H 2S”, Hydr. Proc., 53(4): 81-82, 1974. 9. Shoeld, M., “Purification and Separation of Ga seous Mixtures”, U.S. Pa tent 1,971,798, 1934. 10. Wang, M. and Wei, S., “Energy Conservation of Ab sorption-Stripping Processes with Split-flow Cycles”, J. Chin. Inst. Chem. Eng ., 15(1): 111-120, 1984. 11. Wang, M., Chang, R., Cheu, T., “Ana lysis of Absorption-stripping Processe s with Split-flow C ycles for Energy Saving”, J. Chin. Inst. Chem. Eng ., 16(1): 1-9, 1985. 12. Shethna, H.K., Thermodynamic Analysis of Chemisorpt ion Systems for Acid-gas Removal , PhD. Thesis, UMIST, Manchester, Un ited Kingdom, 1996. 13. Kent, R.L., and Eisenberg, B., “Better Data for Amine Treating”, Hydr. Proc., 55(2): 87, 1976. 14. Shethna and Towler, “Gas Sweetening to Ultra-low Concentrations Using Alkanol amine Absorption”, Paper No. 46f, AIChE Spring Meeting, New Orleans, 1996. 15. Meissner, R.E. and Wagner, U., “Low Energy Process Recovers CO 2”, Oil & Gas J ., 81(5): 55-58, 1983. 16. Douglas, J.M., Conceptual Design of Chemical Processes , McGraw-Hill, Ne w York, 1988. copyright 2001 Bryan Research & Engineering, Inc. Bryan Research and Engineering, Inc. – Technical PapersPage 10of 10 Copyright 2006 – All Rights Reserved Bryan Research and Engineerin g, Inc.

126 KB – 10 Pages