Simulation of Water-Gas Shift Membrane Reactor for Integrated ...

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911-926 Paper received: 15.03.2011DOI: 10.5545/sv-jme.2011.100 Paper accepted: 28.07.2011Simulation of Water-Gas Shift Membrane Reactor forIntegrated Gasification Combined Cycle Plant with CO 2CaptureLotrič, A. ‒ Sekavčnik, M. ‒ Kunze, C. ‒ Spliethoff, H.Andrej Lotrič 1,* ‒ Mihael Sekavčnik 1 ‒ Christian Kunze 2 ‒ Hartmut Spliethoff 21University of Ljubljana, Faculty of Mechanical Engineering, Slovenia2Technische Universität München, Institute for Energy Systems, GermanyThe effectiveness of energy conversion and carbon dioxide sequestration in Integrated GasificationCombined Cycle (IGCC) is highly dependent on the syngas composition and its further processing. Watergas shift membrane reactor (WGSMR) enables a promising way of syngas-to-hydrogen conversion withfavourable carbon dioxide sequestration capabilities. This paper deals with a numerical approach tothe modelling of a water gas shift reaction (WGSR) in a membrane reactor which promotes a reactionprocess by selectively removing hydrogen from the reaction zone through the membrane, making thereaction equilibrium shifting to the product side. Modelling of the WGSR kinetics was based on Bradfordmechanism which was used to develop a code within Mathematica programming language to simulatethe chemical reactions. The results were implemented as initial and boundary conditions for the tubularWGSMR model designed with Aspen Plus software to analyze the broader system behaviour. On the basisof selected boundary conditions the designed base case model predicts that 89.1% CO conversion can beachieved. Calculations show that more than 70% of carbon monoxide conversion into hydrogen appearsalong the first 40% of reactor length scale. For isothermal conditions more than two thirds of the heatreleased by WGSR should be extracted from the first 20% of the reactor length. Sensitivity analysis of theWGSMR was also performed by changing the membrane’s permeance and surface area.©2011 Journal of Mechanical Engineering. All rights reserved.Keywords: IGCC, water-gas shift reaction, membrane reactor0 INTRODUCTIONGasification is an exothermal chemicalprocess at high temperatures and pressuresbetween a hydro-carbonaceous material and anoxidizer (air, oxygen and/or steam). In general,the feedstock used for gasification consists ofhydrocarbons and can be in different forms likecoal, oil, heavy refinery residuals, coke, biomassor even municipal waste. As the oxygen supply islimited (generally 20 to 70% of the oxygen neededfor complete combustion) only partial combustionof the feedstock occurs. The released heat from thechemical reaction drives the secondary reactionthat further converts organic material to syngas.Syngas is mostly a mixture of CO, H 2 ,CO 2 and some CH 4 , H 2 O and N 2 . The ratiobetween CO, H 2 and CO 2 can be quite differentas it depends on the type of a gasifier (movingbed,fluid-bed or entrained-flow gasifier) andthe gasifying feedstock. To achieve the desiredcomposition of syngas the correct operatingtemperature – hence the correct amount ofoxidizer for partial combustion – and pressure ofgasification must be selected.The operating temperature can rangebetween 800 and 1800 °C and the pressurebetween 10 and 100 bar.Gasification is one of the main processesin integrated gasification combined cycle (IGCC)energy systems. In general, all existing IGCCpower plants follow the chain of events presentedin Fig. 1 with an exception of carbon dioxidesequestration (CCS). This chain can be brokendown into ten key processes [1]:1. Air separation unit (ASU) separates air intooxygen to supply the gasifier and nitrogen,which can be used as a carrier, sweep ordilution gas.2. Coal particles are transferred – usingpneumatic conveying or in a form of watercoalslurry – into a gasifier.3. By-products captured in the gasifier (ash andslag) can have commercial value, dependingon local market conditions.*Corr. Author’s Address: University of Ljubljana, Faculty of Mechanical Engineering,Laboratory for Heat and Power, Aškerčeva 6, 1000 Ljubljana, Slovenia, andrej.lotric@gmail.com911

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911-9264. The syngas also has small amounts of otherimpurities (e.g. H 2 S, COS, NH 3 and alsochlorides, fluorides, mercury etc.) which areremoved during the gas clean-up. Before theclean-up process syngas is cooled, usuallyin a syngas cooler, where part (or all) ofthe steam used in the gasification process isproduced.5. Around 99% of H 2 S is separated from thesyngas and converted to elemental sulphur orpossibly sulphuric acid.6. Nowadays, the syngas is passed through a twostage WGS reactor and reacted over a catalystwith added steam to convert the majority ofthe CO into CO 2 and additional H 2 . The H 2is separated from the gas mixture usually byusing the pressure swing adsorption (PSA)process.7. The future potential of IGCC systems is theCO 2 removal, leaving H 2 -rich syngas behind.8. Syngas can also be used in variety ofprocesses; like production of synthetic fuelsor chemicals and pure H 2 can be used in fuelcells or combined cycle to produce electricity.9. The main cause for IGCC plants being moreefficient than conventional power plants isthe combined use of a gas and a steam turbineto produce electricity.10. Much of the water used in this processis recycled in the plant, while some isevaporated in a cooling tower.1 WATER-GAS SHIFT REACTION PROCESS1.1 Membrane ReactorIn general, the membrane reactor (MR) isa device that combines a membrane separationprocess with a chemical reactor in one unit. TheMR is capable of promoting a reaction process byselectively removing at least one of the productsfrom the reaction zone through the membrane,making the equilibrium reaction shifting to theproduct side.Fig. 1. Schematic presentation of the syngas production, hydrogen separation, CCS and hydrogenutilisation in the IGCC912 Lotrič, A. ‒ Sekavčnik, M. ‒ Kunze, C. ‒ Spliethoff, H.

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911 -926Fig. 2. WGSR process in the MRThe membrane used is highly selective tothe product of interest therefore, the product can bedirectly recovered during the reaction, eliminatingthe need for additional product purification steps.The rate of the product flux throughthe membrane is proportional to the product’spartial pressure difference between the process(feed) side and the permeate side; therefore, theextracted product is recovered at a lower pressurethan the process stream pressure. To increase thepressure difference, the inert diluting (or carrier)gas is sometimes used. The diluting gas is carryingaway the permeated product therefore, reducingthe concentration on the permeate side and thus,increasing the driving force for permeation. Thepressure drop through the membrane is materialdependent, and it obeys different laws for differentmembrane materials.With MR the yield can be increased (evenbeyond the equilibrium value for equilibriumreactions) and/or the selectivity can be improvedby suppressing other undesired side reactions orthe secondary reaction of products. Due to theintegration of reaction and separation, chemicalprocesses become simpler leading to a muchlower processing cost.According to Eq. (1) changing the pressureof the WGSR should not have any considerableeffect on changing the equilibrium concentrationsbecause the equation is equimolar. However, theexperimental results, obtained from [2], showthat very high pressures slightly favour the COconversion which can be seen in the Table 1.Table 1. Effect of pressure on equilibriumCO concentrations (inlet dry gas: 13.2% CO,10.3% CO 2 , 35.3% H 2 , 41.2% N 2 , steam-to-drygas = 0.5)Temperature[°C]p =3.04 bar[% CO]p =30.39 bar[% CO]p =303.9 bar[%CO]200 0.12 0.12 0.07300 0.68 0.65 0.48400 1.98 1.94 1.61500 3.93 3.88 3.46600 6.15 6.10 5.68700 8.38 8.34 7.951.2 Water-Gas Shift ReactionThe equilibrium of reversible reactionscan be shifted toward more product formationby changing reaction conditions such as pressureand temperature or concentrations of reactants orproducts.The main focus of this paper is theWater‐Gas Shift Reaction (WGSR), which is alsoan equilibrium reaction and it is represented withthe following equation:CO + H 2 O ↔ CO 2 + H 2 ‒ 41 MJ/kmol . (1)Fig. 3. Reactant conversion for the WGSR whereH 2 and CO 2 are not present at the beginning [3]On the other hand, it is well known thatincreasing the temperature has a decreasingeffect on equilibrium CO conversion. Therefore,H 2 yield can be considerably enhanced at highSimulation of Water-Gas Shift Membrane Reactor for Integrated Gasification Combined Cycle Plantwith CO 2 Capture913

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911-926reaction temperatures, at which the equilibriumCO conversion would otherwise be low, byextracting either CO 2 or H 2 from the reactionmixture.1.3 Water-Gas Shift Kinetics1.3.1 General Definition of Rate LawChemical kinetics investigates howdifferent experimental conditions can influencethe chemical reaction rate. The main factors thatcan speed up the reaction rate include increasein the concentrations of the reactants, increasein temperature or pressure at which the reactionoccurs and whether or not any catalysts are presentin the reaction.Usually, what appears to be a single stepconversion is in fact a multistep reaction and thus,the reaction mechanism is a step by step sequenceof elementary reactions. Each step has its ownrate law and molecularity (the number of speciestaking part in that step); therefore the slowest stepis the one that determines the overall reactionrate – the rate limiting step.Only for simple (elementary) reactionsa partial order of reaction is the same as thestoichiometric number of the reactant concerned.For stepwise reactions there is no generalconnection between stoichiometric numbers andpartial orders. Such reactions may have morecomplex rate laws and the orders of reaction are inprinciple always assigned to the elementary steps.By conducting experiments involvingreactants A and B, one would find that the rate ofthe reaction r is related to the concentrations [A]and [B] in a rate law as:r = k [A] a [B] b , (2)where k is the rate constant and the powers a andb are called the partial orders of the reaction. Theoverall order of the reaction is found by adding upthe partial orders. The rate constant is not a trueconstant because it is dependent on the reaction’stemperature T and the activation energy E a and isdefined via Arrhenius equation as:Ea−k = k ⋅eRT ⋅0 , (3)where k 0 is the pre-exponential factor.A pair of forward and backward reactionsmay define an equilibrium process where A and Breact into C and D and vice versa (a, b, c and d arethe stoichiometric coefficients):aA+ bB ⇔ cC+ d D(4)Assuming that above reactions areelementary, the reaction rate can be expressed as:r = k 1·[A] a·[B] b ‒ k 2·[C] c·[D] d , (5)where k 1 is the rate coefficient for the forwardreaction which consumes A and B; k 2 is the ratecoefficient for the backward reaction, whichconsumes C and D.In a reversible reaction, like WGSR,chemical equilibrium is reached when the rates ofthe forward and backward reactions are equal andthe concentrations of the reactants and products nolonger change (r = 0 in balance).1.3.2 Bradford Mechanisma) Backward WGSRBradford [4] assumed that the reactionmechanism would follow the simple gas-phasechain-reaction mechanism given below (Mindicates any gas-phase collision partner). Thechain is initiated by the gas-phase dissociationof hydrogen (Eq. (6)). Propagation steps arerepresented by reactions in Eqs. (7) and (8). Thetermination step corresponds to the gas-phase reassociationof H 2 in Eq. (9), consuming the chaincarriers.H2k1+ M ⎯⎯ →2H+M, (6)2H+ CO ←⎯⎯⎯⎯ →CO+OH, (7)2k−2k2 k2k3OH + H ←⎯⎯⎯⎯ →H O+H, (8)−3k−12 MH+ ⎯⎯ →H 2 + M. (9)With the assumption of low conversionand a stationary state for the concentrations ofthe intermediates (H and OH concentrations donot change significantly with respect to time), thefollowing rate equation r b is obtained,rdCOdtkk[ ] b = = ⎛ ⎝ ⎜ 1⎞−112 /12 /⎟ ⋅k2⋅[ H2] ⋅[ CO2]. (10)⎠914 Lotrič, A. ‒ Sekavčnik, M. ‒ Kunze, C. ‒ Spliethoff, H.

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911 -926Therefore, the rate constant for thebackward WGSR k b may be expressed as:12 /kkb = ⎛ ⎝ ⎜1⎞⎟ 2k−1⎠where the temperature dependence of the rateconstant is described with the Arrhenius Eq. (3).Consequently, the expression for the rate ofreaction in terms of k b becomes:rb⋅ k , (11)d [ CO ] 12 /= = kb⋅[ H2] ⋅[ CO2],. (12)dtb) Forward WGSRThe gas-phase, chain-reaction mechanismproposed by Bradford [4] can also be used todescribe the forward WGSR. Reaction in Eq.(13) provides the chain initiation by the reactionof H 2 O with any gas-phase molecule (designatedby M). Reactions in Eqs. (14) and (15) are thepropagation steps, while the reaction in Eq. (16) isthe termination step.4HO+ M ⎯⎯ →H+ OH + M, (13)2k5CO + OH ←⎯⎯⎯⎯ →H+CO2 , (14)k − 5k2 2k − 6k6HO+ H←⎯⎯⎯⎯ →OH+H , (15)k−H+ OH+ M ⎯⎯ 4→H O + M. (16)The stationary-state approximation for theconcentration of the chain-carriers (H and OH)under the conditions of low conversions leads tothe following expression for the forward WGSRrate of reaction r f :2rdCOdt2f = [ ] =⎛ k4⎞= ⎜ ⋅k5⋅k6⎝ k−4⎠12 /12 /⎟ ⋅[ CO] ⋅[ HO]2.(17)The rate constant for the forward WGSR k fis defined as:12 /⎛ k4⎞k f = ⎜ ⋅k5⋅k6⎟, (18)⎝ k−4⎠and the rate can be expressed as:d CO2 12 /rfk f COdt[ ] = = ⋅[ ] ⋅[ HO]2 MODEL OF WATER-GAS SHIFTMEMBRANE REACTOR2. (19)At the beginning it was assumed thatthe use of different unit models integrated insimulation software Aspen Plus will be sufficientto model the MR. However, it turned out thatthe model could not correctly predict the processtaking place inside of the reactor. This is why thedecision was made to include the WGSR kinetics,which gave a better overview of what occursinside the reactor.The modelling of the WGSMR wasconducted by using both the Mathematicaprogramming language and Aspen Plus. Thecalculations made in Mathematica were based onWGSR kinetics and used to predict the reactionFig. 4. Boundary conditions for the modelled system and link between Mathematica and Aspen PlusenvironmentSimulation of Water-Gas Shift Membrane Reactor for Integrated Gasification Combined Cycle Plantwith CO 2 Capture915

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911-926rate and the permeation through the membrane.The calculated data were then imported in AspenPlus as input (or initial) values for simulations.Aspen Plus was used to predict thebehaviour of a WGSMR by using basic engineeringrelationships, such as mass and energy balances,and phase and chemical equilibrium. According tothe calculated data from Mathematica, predictedoperating conditions, and different design models,the behaviour of MR was simulated.around 1.1 can be reached with quench waterat 260 °C.2.1 Initial and Boundary ConditionsIn IGCC system the MR will be situatedafter the gasifier and before the gas turbinecombustor, replacing the WGS reactor and PSAsystem (see Fig. 1).The boundaries for the modelled systemare marked with the dashed line in Fig. 4. Inthis figure n represents the molar flow, x thecomposition, T the temperature and p the pressureof gas stream. Heat duty is represented with Qand CO conversion with X CO . Index GAS stands forsyngas, R for retentate side and P for permeate sideof the membrane.Based on the position in the chain ofprocesses that take part in the IGCC, the initialand boundary conditions for “base case” studieswere defined:a) Based on gasifier:• The modelling is based on a dry-feedentrained-flow gasifier with water quench.Therefore, the maximum outlet pressure fromthe gasifier is presumed to be around 50 bar.• The outlet temperature from the gasifier is setto 1500 °C.• Composition of syngas has a major role indefining the steam-to-CO ratio and especiallyin defining the H 2 partial pressure.b) Based on quench:• The quench temperature is set to 800 °C. Inorder to reach this temperature the syngas hasto be quenched with water and not steam.• With pressure of 50 bar, equal to syngas,quench water can only reach temperaturesup to 264 °C, otherwise it would begin tovaporise.• Steam-to-CO ratio is also dependent on thetemperature of quench water. Maximum ratioFig. 5. Schematic presentation of the reactormodelc) Based on MR:• For modelling the tubular MR with gaugemeasurements L = 40 m and a = 10 m wasselected (see Fig. 5). The outer tubularmembrane diameter was set to the value 2r =7.5 cm.• The operational temperature range of the MRis presumed to be between 700 and 900 °C.• The pressure on the retentate side is definedby the outlet from the gasifier. In this study itis presumed to be 50 bar.• Low pressure on the permeate side increasesthe driving force through the membrane.However, from the point of energy lossneeded for compression the minimumpressure should not be less than 1 bar.• In order to improve the driving force the N 2sweep gas is introduced. In view of syngasthe flow of N 2 is in counter-current directionwhere H 2 -to-N 2 molar ratio at the exit is 1:1.• Since the WGSR is an exothermal reaction,the reactor is water cooled. In the presentstudy the MR will operate at isothermalconditions.• The desired CO conversion is set to be around95%.• Based on the temperature range, pressureconditions, hydrogen purity and permeability,the selected membrane material isPalladium based with permeance:k′ = 3·10 -4 mol m -2 s -1 Pa -0.5 .916 Lotrič, A. ‒ Sekavčnik, M. ‒ Kunze, C. ‒ Spliethoff, H.

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911 -926Fig. 6. Presentation of the numerical model for calculations of WGSR kinetics• Due to the high temperatures no catalyst isused for the WGSR.• In this model it is presumed that the membranematerial can withstand the acid environmenttherefore no pre-cleaning of syngas is needed.• The model of the reactor is based on a plugflow reactor. This means that there is nochange in concentration in radial direction.• In the kinetic model it is presumed that thereare no pressure losses on the retentate or onthe permeate side.2.2 Numerical ModelThe numerical model in Mathematica wascreated in such a way that each reactor tube wasdivided in N O = 10,000 infinitesimal sections withlength Dx. The calculation started with valuesobtained from initial and boundary conditions,which gave the values for the first section. Ingeneral, the calculations were conducted in sucha way that the results from section marked withindex i were used to calculate the values indexedas i+1 until reaching the last N Oth section.a) Retentate side:Combining Eqs. (5), (12) and (19) thereaction rate for section i+1 was calculated as:12 / 12 /i f i b i ir+ 1 k CO HO 2 k H2CO 2= ⋅[ ] ⋅[ ]− ⋅[ ] ⋅[ ]. (20)The flux in section i+1 was calculated withthe following Eq.:Ji+ 1 = k ′⋅ ( pRH2, i − pPH2, i ), (21)where permeance was held at constant valuek′ = 3·10 -4 mol m -2 s -1 Pa -0.5 .. This value wasselected based on the experimental study madeby Ciocco et al. [5] where the permeance of purePd membrane k′ = 3·10 -4 mol m -2 s -1 Pa -0.5 . wasachieved.Based on Fig. 6 the Law of Conservationof Mass must apply for each section therefore, themolar flow of each specie was calculated with thefollowing set of Eqs.:n = n + n − n(22)RH 2, i+ 1 RH2, i reaction, i+ 1 H2, i+1,n = n + n(23)CO2, i+ 1 CO2, i reaction, i+1,nCO, i+ 1 = nCO, i −nreaction , i+1 , (24)n = n − n. (25)H2O, i+ 1 H2Oi , reaction, i+1,In the above set of Eqs. n reaction,i+1 andn H2,i+1 were defined as:nreaction, i+ 1 = ri+1⋅NTUBES⋅ ∆ V,(26)n = J ⋅N ⋅∆ A (27)H2, i+ 1 i+1 TUBES .The molar flow of the retentate gas streamwas a sum of all calculated molar flows and theinert species in this study (Ar, H 2 S, N 2 , COS, HCl,CH 4 and others):nRi , + 1 = nRH2, i+ 1+ nCO2,i+1 +(28)+ n + n + nCO, i+ 1 H2O, i+1 inert .The molar fraction of each species wascalculated using the equation:Simulation of Water-Gas Shift Membrane Reactor for Integrated Gasification Combined Cycle Plantwith CO 2 Capture917

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911-926Table 2. Activation energies and pre-exponential factors for forward and backward WGSR obtained fromliterature; the rate constants for WGSR were calculated at operating temperature 1073 KSourceE af[J/mol]k 0f[(l/mol) 0.5 s -1 ]E ab[J/mol]k 0b[(l/mol) 0.5 s -1 ]k f[(cm 3 /mol) 0.5 s -1 ]k b[(cm 3 /mol) 0.5 s -1 ]Culbertson [8] 334.72 7.7·10 45·T-10 (i) / / 6.10 /Graven & Long [9] 281.58 5.0·10 12 238.49 9.5·10 10 3.10 7.37Bustamante [10] 253.55 1.52·10 10 218.40 6.65·10 8 0.22 0.49(i)The pre-exponential factor is also temperature dependant therefore the correct unit is K 10 (l/mol) 0.5 s -1 .xji , + 1n=nji , + 1Ri , + 1,(29)where j represents individual specie (H 2 , CO 2 , COor H 2 O).Similarly, the concentration of each specieswas calculated:nji , + 1 pRpRc ji , + 1 = ⋅ = x ji , + 1 ⋅ , (30)nRi , + 1 RT ⋅ RT ⋅where the pressure on the retentate side was heldat constant value p R = 50 bar. The partial H 2pressure on the retentate side was calculated as:p = x ⋅ p(31)RH2, i+ 1 RH2, i+1 R .In order to calculate heat duty the heat ofreaction DH R needs to be calculated. This wasdone by using the van’t Hoff Relation based onsource [6]:2 ∂∆HR=−R⋅T⋅ ln Keq( T, p ), (32)∂Twhere the logarithm of equilibrium constantK eq was obtained with the equation reported inBustamante [10] which was based on the studyfrom Singh and Saraf [7]:⎛ 9998.22 ⎞⎜ − 10 . 213 +T⎟⎜⎟ln K. Teq = 1.⋅ −3⎜ + 2 7456⋅10⋅ − ⎟.(33)1 987 ⎜−6 ⎜ −0.453⋅10⋅T2 ⎟− ⎟⎜⎝ −0. 201⋅lnT⎟⎠And the CO conversion X CO was deducedas follows:nCO, in − nCO,i+1XCO,i+1 =. (34)nCO,inb) Permeate side:Similarly as on the retentate side the Lawof Conservation of Mass must also apply for eachsection on the permeate side. The molar flow ofH 2 and N 2 mixture was calculated as:nPi , + 1 = nPi , −nH 2, i+1.(35)Because the molar flow of the N 2 sweepinggas is constant throughout the reactor, the molarflow of H 2 on the permeated side can be writtenas:n = n − n(36)PH2, i+ 1 Pi , + 1 N2.Samilarly as on the retentate side, themolar fraction of each species was calculatedusing the Eq.:nji , + 1x ji , + 1 = ,(37)nRi , + 1where j represents individual specie (H 2 or N 2 ).The partial H 2 pressure on the permeateside was calculated as:pPH2, i+ 1 = xPH2, i+1⋅ pP. (38)where the pressure on the permeate side was heldat the constant value p P = 1 bar.2.3 Model Design in Aspen PlusThe simulation was performed using theAspen Plus reactor model unit where chemicalreactions of WGS and side reactions (H 2 S andCOS formation) were modelled. The inputdata were temperature, pressure, molar flow,composition of syngas and formulas of chemicalreactions that take place inside the reactor. Theshortcoming of the model was that the programhad no information about the kinetics of the WGSRand hence no knowledge of the extent of the COconversion and the H 2 permeation through themembrane. These two parameters were calculatedbased on the kinetics of the WGSR and importedfrom Mathematica (see Fig. 4). That gave us moreaccurate predictions of the molar flow of H 2 in thepermeate stream and the achieved CO conversion918 Lotrič, A. ‒ Sekavčnik, M. ‒ Kunze, C. ‒ Spliethoff, H.

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911 -926in the WGSMR. Based on these two parametersand the initial and boundary conditions (seesubchapter 2.1) the simulation of the WGSMRwas performed.The model of the MR was designed toachieve the isothermal temperature distributionthroughout the entire reactor. This is done insuch a way that indirect water cooling is appliedin a co-current direction to syngas flow. To attainefficient heat removal, the water evaporatesalmost throughout the entire reactor and getsslightly superheated at the end of the reactor. It ispresumed that the WGSR will be most intensive atthe beginning of the reactor thus, most of the heatwill be released in the first part of the reactor. Thesweeping N 2 gas enters the reactor in a countercurrentway. The reason for that is to keep the H 2partial pressure difference reasonably high in allparts of the reactor in order to achieve better fluxthrough the reactor’s membrane.3 RESULTS AND DISCUSSION3.1 Calculations in MathematicaIn the subchapters bellow the WGSRkinetics are discussed and calculations for the basecase model are presented. A sensitivity analysis ofthe model is made by varying different parameterslike the membrane’s permeance, length (indirectlythe surface area) of the reactor and the molarfraction of H 2 in the permeate stream.3.1.1 WGSR KineticsThere is some discrepancy between theuncatalysed WGSR kinetic data available fromliterature. For this purpose, three different studieswere compared all using the Bradford mechanism[4] to describe the WGSR. Table 2 presentsactivation energies and pre-exponential factorsobtained from these studies.The rate coefficients for forward andbackward reaction were calculated according toEq. (3). As shown in Table 2, the rate coefficientscalculated with values obtained from Bustamante[10] are one order smaller than those obtainedwith values from Graven and Long [9]. Eventhough Bustamante’s research [10] was conductedat elevated pressures (16.21 bar), and Graven andLong’s [9] only at atmospheric pressure, the resultsfrom Graven and Long’s study [9] are in betteragreement with Culbertson [8] who conducted hisstudy at even higher pressures and temperatures(196.5 to 496.4 bar and 1200 to 2100 K).The calculations showed that the reactionrate according to Bustamante [10] is one ordersmaller than the one predicted from Graven andLong [9] (see Fig. 7). This has also been confirmedin the study from Culbertson [8], therefore thedecision was made that the coefficients deducedfrom Graven and Long [9] will be used in theWGSR kinetic study.3.1.2 Base Case StudiesOne of the purposes of this study was todetermine if the permeation through the membraneis sufficient or if there is a build-up of H 2 on theretentate side. First, the comparison was made byusing the molar flows of both processes calculatedby Eqs. (26) and (27).Only in the first two sections the reactionrate is slightly faster than the permeation (on Fig.a) Bustamante’s model b) Graven & Long’s modelFig. 7. Rate of reaction based on different modelsSimulation of Water-Gas Shift Membrane Reactor for Integrated Gasification Combined Cycle Plantwith CO 2 Capture919

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911-9268 the difference is negative) after which the rateof permeation becomes faster than the productionof H 2 . As seen from Fig. 8, the maximumdifference is reached around the 400 th sectionand the permeation rate becomes almost equal tothe reaction rate around the 4000 th section. Therate of reaction rises slightly at the end becauseat that point the concentration of H 2 is closeto zero and the effect of the backward WGSRalmost disappears. The increase in the reactionrate also causes the rate of permeation to increase,because more H 2 is produced, which immediatelypermeates through the membrane because theprocess is not permeation limited.in the retentate stream at the beginning and H 2converted from CO until that point permeatethrough the membrane. From that point on, onlythe H 2 that is converted from CO is permeatingthrough the membrane.Rapid permeation of H 2 is responsible formolar fractions of H 2 O and CO to experiencethe first inflection point around the 400 th section.The fractions reach the second inflection pointaround 2000 th section, where majority of H 2 onthe retentate side already permeated through themembrane. From that point on both fractionsgradually reduce towards the end of reactor wherethe consumption of H 2 O and CO again slightlyincreases on behalf of increase in the reaction rate.The molar fraction of CO 2 is increasing fastin the first 2000 sections because of H 2 permeationand because of CO conversion. This can be seenfrom a graph where after 2000 sections the rate ofCO 2 formation starts to decrease since virtuallyall of H 2 that was present in the stream permeatesthrough the membrane. At the end, the rate ofCO 2 formation slightly increases on behalf of theincrease in the reaction rate.Fig. 8. Difference between the rate of H 2production and the rate of H 2 permeation in eachsection of the reactorAs it can be seen from Fig. 9, the molarfraction of H 2 actually rises at the beginning eventhough the rate of permeation is greater as the rateof formation. By constantly extracting H 2 fromthe retentate stream, the joint molar flow alsodecreases. Since at the beginning the differencebetween the permeation and the production ratesof H 2 is not big enough, the H 2 molar flow inretentate stream is decreasing but not as fast as themolar flows of H 2 O and CO. As a consequence,the molar fraction of H 2 actually rises slightlyuntil the 64 th section where, on behalf of a sharpincrease in the permeation rate, H 2 starts to rapidlypermeate through the membrane.After reaching the peak at the 64 th sectionthe molar fraction of H 2 starts to decrease fastuntil the 1000 th section. Around the 2000 th sectionthe permeation already slows down considerablyand around the 4000 th section most of H 2 presentFig. 9. Change in molar fractions of species andthe CO conversion throughout the reactorBased on the input values, the calculatedmolar composition of the retentate stream ispresented in Table 3. Below, these values are alsocompared to the simulation results from AspenPlus.In order to keep the MR operating atisothermal conditions, approximately two thirdsof all heat, produced from the WGSR, would needto be removed from the reactor in the first 2000sections (see Fig. 10). The cooling tubes should bedistributed in such a way that they would insure920 Lotrič, A. ‒ Sekavčnik, M. ‒ Kunze, C. ‒ Spliethoff, H.

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911 -926the reactor to operate at isothermal conditions andto prevent the formation of any hotspots in thereactor.Table 3. Input and output values from MathematicaprogramInput values (SYNGAS)species H 2 O CO H 2 CO 2 inertx i 0.410 0.375 0.154 0.052 0.009Output values (GAS-2)species CO 2 H 2 O CO inert H 2x i 0.754 0.148 0.080 0.018 3.0∙10 -4600 sections. Beyond that point permeation isfaster but it remains fairly low. At the beginningpermeation slightly rises because of the H 2 buildupwhich increases the H 2 partial pressure andenhances the flux through the membrane. TheWGSR proceeds fast in the first 1000 sections andthen it remains low through the entire reactor. Theblue, dashed line shows the difference betweenboth processes.It should also be noted that by introducingthe cooling tubes into the reactor the COconversion may suffer minor penalties because thereaction volume for the WGSR will be reduced.The shape of the curve resembles the curveof the CO conversion, because with more COconverted more heat is released. The calculatedcumulative heat duty released from the WGSR isQ = 63.4 MW.Fig. 11. Difference between the rate of H 2production and the rate of H 2 permeation withone order smaller permeanceFig. 10. Cumulative distribution of heat releasedin the WGSMRThe graph in Fig. 12 shows the build-upof H 2 molar fraction in the retentate stream. Afterthe reaction rate becomes steady, the H 2 fractionstarts to gradually decrease on behalf of slowpermeation. As the permeation is very limited,all of H 2 does not succeed in permeating throughthe membrane, therefore in the end there is stillaround 4% of H 2 left in the retentate stream.3.1.3 Influence of Membrane’s Permeance on COConversionThe influence of membrane’s permeanceon the CO conversion was studied using the MRdesigned for base case studies. The only parameterthat was changed was the membrane’s permeance.In this analysis the permeance was set to the valuek′ = 3·10 -5 mol m -2 s -1 Pa -0.5 . The calculationsshow that in the MR with such material permeancethe processes would become permeation limited.The graph in Fig. 11 shows that the reactionrate is faster than the permeation rate in the firstFig. 12. Change in molar fractions of speciesand the CO conversion with one order smallerpermeanceSimulation of Water-Gas Shift Membrane Reactor for Integrated Gasification Combined Cycle Plantwith CO 2 Capture921

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911-926The reactor achieves only 50% COconversion because the extraction of H 2 is tooslow and it does not shift the equilibrium as far tothe product side as in the base case scenario.3.1.4 Influence of Reactor’s Surface Area andH 2 Molar Fraction in Permeated Stream on COConversionThe influence of the reactor’s surface areaon CO conversion was studied by varying thelength of the reactor. Additionally, the influenceof different molar compositions in the permeatestream was studied. By changing both parametersTable 4 was obtained.The table shows that reducing the molarfraction of H 2 in the permeate stream wouldfavour the CO conversion. This is due to the factthat the partial pressure of H 2 is lower and thus,enhances the flux through the membrane whichindirectly affects the CO conversion.Increasing the reactor’s surface, byextending the length of the reactor, also favoursthe CO conversion. By enlarging the membranesurface area, more H 2 can permeate through,which shifts the WGSR more to the product side,and helps improve the CO conversion.To further determine the influence ofsurface area, the reactor with 70 m in length andH 2 molar fraction x H2 = 0.45 was simulated. Thereactor achieved only 92.3% CO conversion. Thisshows that extending a reactor further would helpachieve better conversions, but from economicpoint of view investment costs would be enormousto build such a big reactor. As discussed in earlierchapters, it would be better to use the catalystbecause the surface area is already large enoughand in the latter part of the MR the limitingprocess is the WGSR rate.3.2 Simulation Results in Aspen PlusThe model simulated on the basis of thecalculations obtained from the Mathematica basecase model is presented in the previous subchapter.Table 4. CO conversion at different H 2 molar fractions and lengths of reactorL = 30 m L = 40 m L = 50 mX COn N2[mol/s]n P,H2[mol/s]X COn N2[mol/s]n P,H2[mol/s]X COn N2[mol/s]n P,H2[mol/s]x H2 0.5 87.0 2664.53 2664.49 89.1 2707.63 2707.57 90.2 2730.35 2730.30.45 88.0 3281.43 2684.73 90.1 3335.05 2728.57 91.2 3363.35 2751.820.55 86.0 2162.35 2642.9 88.0 2196.87 2685.16 89.1 2215.03 2707.32Table 5. Molar compositions of gas streams(see Fig. 13)Rawgas (on the exit from the gasifier)species CO H 2 H 2 O CO 2 Ar H 2 S N 2 COS HClx i 0.560 0.230 0.120 0.077 4.8∙10 -3 4.5∙10 -3 4.0∙10 -3 3.9∙10 -4 6.7∙10 -5Syngas (after quench)species H 2 O CO H 2 CO 2 Ar H 2 S N 2 COS HClx i 0.410 0.375 0.154 0.052 3.2∙10 -3 3.0∙10 -3 2.7∙10 -3 2.6∙10 -4 4.5∙10 -5Gas-2 (retentate stream on the exit of the MR)species CO 2 H 2 O CO Ar H 2 S N 2 H 2 COS HClx i 0.769 0.140 0.072 6.3∙10 -3 5.9∙10 -3 5.3∙10 -3 5.3∙10 -4 5.2∙10 -4 8.8∙10 -5Gas-3 (after catalytic burner)species CO 2 H 2 O SO 2 Ar N 2 O 2 CO H 2 H 2 Sx i 0.835 0.146 6.4∙10 -3 6.2∙10 -3 5.2∙10 -3 2.0∙10 -3 0 0 0GT-gas (permeate stream on the exit of the MR)species H 2 N 2 CO H 2 O CO 2 Ar H 2 S COS HClx i 0.5 0.5 0 0 0 0 0 0 0922 Lotrič, A. ‒ Sekavčnik, M. ‒ Kunze, C. ‒ Spliethoff, H.

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911 -926This chapter deals with the influence of absolutepressure on the permeate side as regards to thepower demand for compression of H 2 /N 2 mixture.3.2.1 Base Case StudiesThe input variables for the simulation inAspen Plus were the same as the values for modelin Mathematica. The only additional data thatthe model in Aspen needed were the values forCO conversion and the molar flow of permeatedH 2 . On the basis of the simulation, the molarcompositions of gas streams were obtained (seeTable 5). The other calculated data are presentedin the flowsheet diagram presented in Fig. 13.The molar fraction of CH 4 was left outof this table because the gasifier model in AspenPlus predicts that only a small fraction of methanex CH4 =3.8·10 -5 is formed.The values predicted from the simulation inAspen Plus and the calculations from Mathematicaare in good agreement. In general, the model inAspen predicts that more CO should be convertedfor a given CO conversion, because the fractionsof CO 2 and H 2 are slightly higher and the fractionsof CO and H 2 O are slightly smaller as predicted inMathematica. The calculated heat duty is in verygood agreement between both models.At the exit of the catalytic burner, thetemperature of CO 2 rich stream reaches 1200 °C;therefore, there is a great amount of heat availablefor other processes. A considerable amount of SO 2present in the stream can be cleaned of sulphurusing the conventional Flue gas desulphurisation(FGD) technologies. At this point it should bementioned that the MR can also be used for directthermal decomposition of H 2 S. Studies on thistopic have already been conducted by severalauthors [12] to [14] and this could be one of thefuture features integrated in the WGSMR.3.2.2 Influence of Absolute Pressure on PowerDemand for CompressionThe influence of the absolute pressureat the permeate side on the power demand forcompression of H 2 and N 2 mixture was studied atthree different pressures and at constant H 2 molarfraction x H2 = 0.5.The change in the absolute pressure on thepermeate side effects the H 2 permeation throughthe membrane because the partial pressure is alsochanged. As a consequence different H 2 molarflows on the exit of the MR are obtained. Toadequately compare the power demand at differentoperating conditions, the specific work w comp wasintroduced and defined as:Fig. 13. Flowsheet diagram used for WGSMR simulationsSimulation of Water-Gas Shift Membrane Reactor for Integrated Gasification Combined Cycle Plantwith CO 2 Capture923

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911-926wcompPcomp= , (39)n where P comp is the joint power demand of all stagesand n P the molar flow of the permeate stream. Thecalculated data are presented in Table 6.Table 6. Comparison of specific work forcompression of H 2 to 26 barp [bar] 1 2 3P comp [MW] 68.497 51.127 45.475n P [mol/s] 5415.2 5222.2 5075.6w comp [kJ/mol] 12.65 9.79 8.96The calculated data show that specificpower demand for the base case (1 bar) isrelatively high compared to other two cases. Incases of 2 and 3 bar the power demand is notthat different; therefore, operating at 2 bar wouldbe an inviting option. In the future, it wouldbe reasonable to study the trade-off betweenthe higher H 2 production and the lower powerconsumption for compression.P4 CONCLUSIONSThe WGSR kinetics was described withBradford mechanism [4] together with thekinetic data obtained from Graven and Long’s [9]experimental results. The calculations performedin Mathematica were used to calculate the COconversion and the H 2 flux through the membrane.The calculated data were then imported in AspenPlus as input (or initial) values for simulations.The model in Aspen Plus was designed as ablack box model, simulating the processes basedon the initial and boundary conditions and thecalculations obtained from Mathematica. Theresults, derived from both models, are in goodagreement with each other and support thecorrectness of the designed model.Based on the operational conditions ofthe designed reactor the membrane materialbased on Pd was selected. The calculations showthat the membrane material with permeancek′ = 3·10 -4 mol m -2 s -1 Pa -0.5 . would be sufficientfor the CO conversion because the process wouldbe reaction and not permeation limited. The COconversion could be enhanced by increasing thereaction rate with the use of high temperaturecatalyst or with higher process temperatures.Other studies emphasized that Pd also has somecatalytic properties for the WGSR but this wasnot included in the designed model. However, inthe future it would be reasonable to determine thecatalytic properties of Pd to accurately predict theWGSR rate, as well. It was established that oneorder smaller membrane permeance would reducethe CO conversion to 50% because the processwould be permeation limited.Based on the selected boundary conditionsthe designed model predicts that 89.1% COconversion can be achieved. With length increaseof 75% and H 2 molar fraction of x H2 = 0.45 in thepermeate stream 92.3% conversion was achieved.But from economic point of view the gain in COconversion would not justify the investment costfor such a big reactor.Changing the absolute pressure on thepermeate side shows that the trade-off betweenthe CO conversion and the power demand forcompression of H 2 /N 2 gas stream needs to beapproached by further research. The specificpower demand shows that it would be reasonableto set the absolute pressure to 2 bar.The retentate stream is fed to the catalyticburner, where the unconverted CO, unpermeatedH 2 and H 2 S are burned to produce additional heat.Afterwards, the stream contains the SO 2 which canbe removed with conventional FGD technologies.Another possibility for the MR in the future maybe the thermal decomposition of H 2 S. However,in order to incorporate the process into the MR, itneeds to be studied further.In general, the study shows that theWGSMR is feasible, but there are still manyissues that need to be addressed. First of all, themembrane material must retain its structuralstability and permeability during the operatingconditions that are stated. If the material cannotwithstand the acid environment, pre-cleaning ofsyngas would be needed. Proper cooling of thereactor also needs to be designed to avoid hotspotsand to keep the reactor operating at isothermalconditions. By introducing cooling tubes intothe reactor, the CO conversion may suffer minorpenalties because less reaction volume will beavailable for the WGSR to take place.924 Lotrič, A. ‒ Sekavčnik, M. ‒ Kunze, C. ‒ Spliethoff, H.

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911 -9265 ACKNOWLEDGEMENTThe authors would like to thank thereviewers for their constructive criticism, whichhelped to improve the quality and understandingof this article.6 NOMENCLATUREGeneral acronymsASU Air Separation UnitCCS Carbon Capture and StorageFGD Flue Gas DesulphurisationIGCC Integrated Gasification Combined CycleMR Membrane ReactorPSA Pressure Swing AdsorptionWGS Water-Gas ShiftWGSR Water-Gas Shift ReactionWGSMR Water-Gas Shift Membrane ReactorPhysical quantities2r Outer diameter of tubular membrane, [m][A] Concentration, [mol/m 3 ]a Partial order of reactiona Width and height of MR, [m]c Molar concentration, [mol/m 3 ]∆A Membrane’s surface area of infinitesimalsection, [m 2 ]∆H R Heat of reaction, [J/mol]∆V Volume surrounding infinitesimal sectionof the membrane, [m 3 ]∆x Length of infinitesimal section, mE a Energy of activation, [J/mol]E abEnergy of activation for backwardWGSR, [J/mol]E afEnergy of activation for forward WGSR,[J/mol]i index representing the i th section innumerical calculationsj index representing individual gas specieJ Flux through membrane, [mol/(m 2 s)]k Rate constant, [(cm 3 /mol) 0.5 s -1 ]k b Rate constant for backward WGSR,[(cm 3 /mol) 0.5 s -1 ]k f Rate constant for forward WGSR,[(cm 3 /mol) 0.5 s -1 ]k 0 Pre-exponential factor, [(l/mol) 0.5 s -1 ]k 0b Pre-exponential factor for backwardWGSR, [(l/mol) 0.5 s -1 ]k 0f Pre-exponential factor for forwardWGSR, [(l/mol) 0.5 s -1 ]k’ Membrane’s permeance,[mol m -2 s -1 Pa -0.5 .]K eq Equilibrium constantL Length of reactor, [m]n Molar flow, [mol/s]n R Molar flow of retentate gas stream,[mol/s]n P Molar flow of permeate gas stream,[mol/s]n CO Molar flow of CO, [mol/s]n CO2 Molar flow of CO 2 , [mol/s]n H2 Molar flow of H 2 permeating throughmembrane, [mol/s]n H2O Molar flow of H 2 O, [mol/s]n inert Molar flow of inert gases, [mol/s]n N2 Molar flow of N 2 (carrier gas) onpermeate side of the membrane, [mol/s]n reaction Molar flow of species produced orconsumed during the WGSR, [mol/s]n PH2 Molar flow of H 2 on permeate side of themembrane, [mol/s]n RH2 Molar flow of H 2 on the retentate side ofthe membrane, [mol/s]N TUBES Number of tubes in MRp Pressure, [bar]p P Pressure on permeate side of MR, [bar]p PH2 Partial pressure of H 2 on permeate sideof the membrane, [bar]p R Pressure on retentate side of MR, [bar]p RH2 Partial pressure of H 2 on retentate side ofthe membrane, [bar]P comp Power demand for H 2 compression, [W]Q Heat duty, [W]r Rate of reaction, [mol/(cm 3 s)]r b Rate of reaction for backward WGSR,[mol/(cm 3 s)]r f Rate of reaction for forward WGSR,[mol/(cm 3 s)]R Universal gas constant, [J/(mol K)]T Temperature, [K]w comp Specific work for H 2 compression,[J/mol]x Molar composition or molar fractionx RH2 Molar fraction of H 2 on retentate side ofthe membranex PH2 Molar fraction of H 2 on permeate side ofthe membraneCO conversionX COSimulation of Water-Gas Shift Membrane Reactor for Integrated Gasification Combined Cycle Plantwith CO 2 Capture925

Strojniški vestnik - Journal of Mechanical Engineering 57(2011)12, 911-9267 REFERENCES[1] Donaldson, A.M., Mukherjee, K.K. (2006).Gas turbine “refueling” via IGCC POWERMagazine, from: http://www.powermag.com/gas/Gas-turbine-&quotrefueling%22-via-IGCC_540_p3.html, accesed on 2009-06-14.[2] Ladebeck, J.R., Wagner, J.P. (2003). Catalystdevelopment for water-gas shift. Handbook ofFuel Cells - Fundamentals, Technology andApplications, vol. 3, part 2, p. 190-201.[3] Killmeyer, R., Rothenberger, K., Howard,B., Ciocco, M., Morreale, B., Enick, R.,Bustamante, F. (2003). Water-gas shiftmembrane reactor studies. US DOE, NETL,Berkeley.[4] Bradford, B.W. (1933). The water-gasreaction in low-pressure explosions. Journalof the Chemical Society, p. 1557-1563,DOI 10.1039/jr9330001557.[5] Ciocco, M.V., Iyoha, O., Enick, R.M.,Killmeyer, R.P. (2007). High-temperaturewater-gas shift membrane study. NETL,South Park.[6] Lebon, G., Jou, D., Casas-Vazquez, J.(2008). Understanding Non-equilibriumThermodynamics: Foundations, Applications,Frontiers. Springer-Verlag, Berlin, DOI:10.1007/978-3-540-74252-4.[7] Singh, C.P., Saraf, D.N. (1977). Simulationof high-temperature water-gas shift reactors.Industrial and Engineering Process Designand Development, vol. 16, no. 3, p. 313-319,DOI:10.1021/i260063a012.[8] Culbertson, B., Sivaramakrishnan, R.,Brezinsky, K. (2008). Elevated pressurethermal experiments and modeling studieson the water-gas shift reaction. Journal ofPropulsion and Power, vol. 24, no. 5, p.1085-1092, DOI:10.2514/1.31897.[9] Graven, W.M., Long, J.F. (1954). Kinetics andmechanisms of the two opposing reactions ofequilibrium CO + H 2 O = CO 2 + H 2 . Journalof the American Chemical Society, vol. 76, p.2602-2607, DOI:10.1021/ja01639a002.[10] Bustamante-Londono, F. (2004). The hightemperature,high-pressure homogeneouswater-gas shift reaction in a membranereactor. Ph.D. Thesis, University ofPittsburgh, Pittsburgh.[11] Enick, R., Bustamante, F., Iyoha, O., Howard,B., Killmeyer, R., Morreale, B., Ciocco, M.(2004). Conducting the homogeneous watergasshift reaction in a palladium/copper alloymembrane reactor at high temperature andpressure. NETL, Pittsburgh.[12] Edlund, D. (1996). A membrane reactor forH 2 S decomposition. Bend Research, Inc.,Morgantown.[13] Ma, Y. H., Moser, W.R., Pien, S., Shelekhin,A.B. (1994). Development of hollowfiber catalytic membrane reactors forhigh temperature gas cleanup. WorcesterPolytechnic Institute, Worcester.[14] US 7,163,670 B2 - Patent, Agarwal, P.K.,Ackerman, J. (2007). Membrane forHydrogen Recovery from Streams ContainingHydrogen Sulfide, University of Wyoming,Wyoming.926 Lotrič, A. ‒ Sekavčnik, M. ‒ Kunze, C. ‒ Spliethoff, H.