FINAL TECHNICAL REPORT January 1, 2013, through December 31, 2013

Transcription

1 FINAL TECHNICAL REPORT January 1, 2013, through December 31, 2013 Project Title: CFD MODELING OF A REGENERATIVE MGO-BASED PROCESS FOR CO 2 CAPTURE IN IGCC (PHASE 3) ICCI Project Number: Princial Investigator: Other Investigators: Project Manager: 12/US-3 Javad Abbasian, Illinois Institute of Technology Hamid Arastooour, Illinois Institute of Technology Debalina Dasguta, ICCI ABSTRACT The overall objective of this joint ICCI/USDOE rogram is to develo a Comutational Fluid Dynamic (CFD) model and to erform CFD simulations using Poulation Balance Equations (PBE) to describe the heterogeneous gas-solid absortion/regeneration and water-gas-shift (WGS) reactions for a regenerative magnesium oxide-based (MgO-based) rocess to simultaneously cature CO 2 and enhance hydrogen roduction in coal gasification rocesses. The work erformed in this ICCI-sonsored roject is art of the joint USDOE/ICCI 3-yr roject sonsored by the USDOE/NETL. The objective of the ICCI-sonsored ortion of the work for the third year was geared towards erforming CFD simulations for the regenerative carbon dioxide cature rocess. During the third year of the roject, a CFD model was develoed to simulate the absorber reactor and to redict the extent of CO 2 cature and enhancement of H 2 roduction in a regenerative MgO-based rocess for Integrated Gasification Combined Cycle (IGCC). A 3D CFD model was develoed and used to erform cold flow simulations on the riser art of the existing NETL Carbon Cature Unit (C2U) circulating fluidized bed unit. The results indicate that the CFD model can accurately cature the chugging behavior that was exerimentally observed in the NETL-C2U unit. The ressure dro in the lower mixing zone of the riser redicted by the CFD model was also shown to be in good agreement with the exerimental data. The results of the CFD simulation of the CO 2 cature in the absorber with a (50/50) CO 2 /N 2 mixture indicate that the extent of CO 2 cature at the NETL baseline condition is less than 15%. However, by increasing the solid circulation rate, the extent of CO 2 cature can be increased to about 40%. Furthermore, the results also indicate that, by decreasing the gas inlet velocity by 35% (and therefore increasing gas residence time), the extent of CO 2 cature can aroach u to 60%. The results of the CFD simulations with a simulated syngas containing 20% CO 2, 20% H 2 O, 30% CO, and 30% H 2 indicate that u to 60% CO 2 can be removed in the absorber, while the hydrogen content of the reactor exit gas can be enhanced through the Sorbent Enhanced Water-Gas-Shift (SEWGS) reaction to 65%. 1

2 EXECUTIVE SUMMARY Advanced ower generation technologies such as Integrated Gasification-Combined Cycle (IGCC) are among the leading contenders for ower generation in the 21st century, because such rocesses offer significantly higher efficiencies and suerior environmental erformance, comared to coal combustion rocesses. Develoment of IGCC rocesses is esecially advantageous to high sulfur Illinois coal, in comarison to western coal, because of its high Btu content. Global warming, which has been associated with the increasing concentration of greenhouse gases, mainly carbon dioxide (CO 2 ), is regarded as one of the key environmental issues in the 21 st century. Near-term alications of re-combustion CO 2 cature from IGCC rocesses will likely involve hysical or chemical absortion rocesses. However, the commercially available rocesses (e.g., SELEXOL) oerate at low temeratures, imarting a severe energy enalty on the system and, consequently, their use can significantly increase costs of electricity roduction. Therefore, develoment of high temerature regenerative rocesses based on solid sorbents offer an attractive alternative otion for carbon cature, at cometitive costs. The gas searation research team at the Illinois Institute of Technology (IIT) has develoed a regenerative high temerature CO 2 cature rocess that is caable of removing over 98% of CO 2 from a simulated coal gas mixture at the IGCC conditions using highly reactive and mechanically strong MgO-based sorbents. The sorbent also exhibited some catalytic activity for the water-gas-shift (WGS) reaction at 300C, increasing hydrogen concentration from 37% (inlet) to about 70% in the reactor exit. Initial results of modeling of the sorbent/catalyst erformance in a acked bed indicate that hydrogen concentration in the simulated coal gas mixture can achieve over 95% cature of CO 2. These encouraging results indicate that both CO 2 cature and enhancement of hydrogen roduction can be carried out in a single unit. Although the results obtained in the revious rojects are romising, there are several key issues which need to be investigated to rovide sufficient information for scale-u of the rocess. Comutational Fluid Dynamics (CFD) rovides an attractive otion to accomlish this goal in a systematic and economically feasible way. However, in order to use CFD to erform simulations of the regenerative CO 2 cature rocess, a model accounting for the variation of the article orosity and its effects on the hydrodynamics is required. In ursuance of this goal, the overall objective of this joint ICCI/USDOE rogram was to develo a Comutational Fluid Dynamic (CFD) model and to erform CFD simulations using Poulation Balance Equations (PBE) describing the heterogeneous gas-solid absortion/regeneration and water-gas-shift (WGS) reactions for a regenerative magnesium oxide-based (MgO-based) rocess to simultaneously cature CO 2 and enhance hydrogen roduction in coal gasification rocesses. 2

3 Work erformed in this ICCI-sonsored roject is art of the joint USDOE/ICCI 3-yr roject sonsored by the USDOE/NETL University Coal Research (UCR) rogram. The ICCI deliverable for this third year of funding is develoment of a CFD model to redict the extent of CO 2 cature and enhancement of H 2 roduction in the regenerative rocess. During the third year of the roject, a CFD model was develoed to simulate CO 2 cature and enhancement of H 2 roduction in the regenerative MgO-based rocess in IGCC rocess. A oulation balance equation (PBE) aroach was used to describe the evolution of the article orosity distribution and was linked with the multihase flow dynamics governing equations. The variable diffusivity reaction rate model (that was develoed in the second year of the ICCI roject) was used to describe the orosity variation of a single sorbent article as a function of sorbent conversion and was incororated into the CFD model to redict the extent of the overall absortion/regeneration reactions. The PBE was solved with an efficient comutational technique known as Finite size domain Comlete set of trial functions Method Of Moments (FCMOM), recently develoed by the IIT multihase flow research team. Cold flow simulations were erformed on the riser section of the existing cold flow Carbon Cature Unit (C2U) at the National Energy Technology Laboratory (NETL) circulating fluidized bed using the 3D CFD model to comare the ressure dro across the absorber redicted by the model with the exerimental data reorted by the NETL and to redict the chugging effect observed in the NETL exeriments. Results indicate that the CFD model can accurately cature the chugging behavior that was exerimentally observed in the NETL-C2U unit. The ressure dro in the lower mixing zone of the riser redicted by the CFD model using a detailed Energy Minimization Multi-Scale (EMMS) aroach was shown to be in good agreement with the exerimental data and caable of accurately redicting both the average and fluctuating ressure dro. Additionally, it was shown that, although a traditional drag model which assumes a homogenous distribution of solid articles in the gas medium, does not redict any major fluctuations but can redict average values with accetable accuracy. These results suggest that given the significantly shorter comutational time required with the traditional drag model, all simulations can be erformed using the simler traditional model, while the more detailed EMMS model should be used when more accurate information on instantaneous behavior of flow is required. The results of the CFD simulation of the CO 2 cature in the absorber with a (50/50) CO 2 /N 2 mixture indicate that the extent of CO 2 cature at the NETL baseline condition is less than 15%. However, by increasing the solid circulation rate, the extent of CO 2 cature can be increased to about 40%. The results also indicate that, by decreasing the gas inlet velocity by 35% (and therefore increasing gas residence time), the extent of CO 2 cature can aroach u to 60%. The results of the CFD simulations with a simulated syngas containing 20% CO 2, 20% H 2 O, 30% CO, and 30% H 2 indicate that u to 60% CO 2 can be removed in the absorber, while the hydrogen content of the reactor exit gas can be enhanced through the Sorbent Enhanced Water-Gas-Shift (SEWGS) reaction to 65%. 3

4 Secific objectives of this roject were to: OBJECTIVES Develo a CFD model to determine the extent of CO 2 cature in the gas/solid CO 2 absortion and sorbent regeneration reactions Exerimentally determine key gas-solid reaction arameters for the absortion/regeneration and WGS reactions Perform CFD simulations of the regenerative carbon dioxide cature rocess Develo a reliminary base case design for scale u To achieve the roject objectives, a 3-year rogram comrising of extensive exerimental, analytical, and CFD modeling work was undertaken consisting the following four tasks: Task 1. Develoment of a CFD/PBE model accounting for the article (sorbent) orosity distribution and of a numerical technique to solve the CFD/PBE model Task 2. Determination of the key arameters for the absortion/regeneration and WGS reactions Task 3. CFD simulations of the regenerative carbon dioxide cature rocess Task 4. Develoment of reliminary base case design for scale u INTRODUCTION AND BACKGROUND Coal-fired ower lants currently account for about 50% of the electricity used in the United States 1. With diminishing etroleum sulies, ublic concern regarding the overall safety of nuclear ower, unredictability of natural gas rices, and unavailability of alternative large-scale sources of energy, coal continues to lay a leading role in the total energy icture. Advanced ower generation technologies such as Integrated Gasification-Combined Cycle (IGCC) are among the leading contenders for ower generation conversion in the 21 st century, because such rocesses offer significantly higher efficiencies and suerior environmental erformance, comared to coal combustion rocesses. It is envisioned that these advanced systems can cometitively roduce low-cost electricity at efficiencies higher than 60% with coal while achieving near-zero discharge energy lants, if the environmental concerns associated with these rocesses, including the climate change, can be effectively eliminated at cometitive costs. 2 Develoment of IGCC rocesses is esecially advantageous to high sulfur Illinois coal, in comarison to western coal, because of its high Btu content. With any coal (high or low sulfur), a gas cleaning treatment is always needed due to rocess requirements in the downstream units. The cost of sulfur cature is determined rimarily by the gas throughut of the system rather than the sulfur content of the coal. Another advantage associated with the use of high sulfur Illinois coal is the value of the increased sulfur 4

5 roducts generated in the cleaning rocess. Global warming, which has been associated with the increasing concentration of greenhouse gases, mainly carbon dioxide (CO 2 ), is regarded as one of the key environmental issues in the 21 st century. Continued uncontrolled atmosheric emission of CO 2 is believed to significantly contribute to undesirable climatic changes. Given that a large fraction of the total CO 2 emissions is from large stationary sources such as fossil fuel-based ower lants, it is logical to assume that the initial efforts toward reducing CO 2 emissions should focus on develoment of imroved and novel technologies for cature of CO 2 from conventional or advanced ower generation rocesses. Near-term alications of CO 2 cature from re-combustion systems will likely involve hysical or chemical absortion rocesses. However, these commercially available rocesses (e.g., SELEXOL) oerate at low temeratures, imarting a severe energy enalty on the system 3 and, consequently, their use can significantly increase the costs of electricity roduction. Therefore, develoment of high temerature regenerative rocesses based on solid sorbents can offer an attractive alternative otion for carbon cature, at cometitive costs. The gas searation research team at the Illinois Institute of Technology (IIT) has develoed a regenerative high temerature CO 2 cature rocess that is caable of removing over 98% of CO 2 from a simulated coal gas mixture at IGCC conditions using a highly reactive and mechanically strong MgO-based sorbents (see Figure 1a) Furthermore, the catalytic activity has been exhibited by the sorbent for the water-gasshift (WGS) reaction at 300C (shown in Figure 1b), increasing hydrogen concentration from 37% (inlet) to about 70% in the reactor exit. 4,5 While CO 2 absortion reaction is accomlished at IGCC rocess conditions, the regeneration reaction is carried out by using a CO 2 free regeneration gas (e.g., steam) to roduce a concentrated stream of CO 2, to be utilized in another industrial alication or sequestered. The cyclic chemical reactions for CO 2 cature involving magnesium oxide are: MgO + CO 2 MgCO 3 (CO 2 Absortion Reaction) (A) MgCO 3 MgO + CO 2 (Regeneration Reaction) (B) CO + H 2 O CO 2 + H 2 (Water-Gas-Shift Reaction) (C) 5

6 (Mole of CO 2 Out) / (Mole of CO 2 In), %mol System Pressure= 20 atm CO 2 inlet= 27 %mol (dry base) H 2 inlet= 37 %mol (dry base) CO inlet=36 %mol (dry base) 425 C 400 C 300 C 350 C Time, min Molar Percent (dry base) H 2 CO Bed Temerature= 300 C System Pressure= 20 atm Inlet Total Flow Rate= 200 cm 3 /min CO 2 inlet= 27 %mol (dry base) H 2 inlet= 37 %mol (dry base) CO inlet=36 %mol (dry base) 10 CO Time, min Figure 1: a (to), Extent of CO 2 Cature at Different Oerating Temeratures; b (bottom), CO 2 Cature and H 2 Production at 300 C The regenerability and long term durability of the sorbent has been demonstrated over 25 consecutive absortion/regeneration cycles, indicating that the sorbent is suitable for long-term alications. It was also shown that only the outer layer (40-50 μm thick) of the sorbent (article diameter μm) reacted with CO 2. 6,7 Therefore, it is logical to assume that by reducing the article size to about μm, the CO 2 absortion caacity of the sorbent can be significantly increased. A two-zone exanding grain model was develoed to describe the gas-solid reaction to redict the sorbent erformance in the cyclic rocess and the extent of CO 2 cature and hydrogen roduction if the sorbent is mixed with an aroriate WGS catalyst. 5,7 The results of theoretical modeling of the sorbent/catalyst erformance in a acked bed indicated that hydrogen concentration in the simulated WGS mixture can exceed 95%. These encouraging results indicate that both 6

7 CO 2 cature enhancement of hydrogen roduction can be achieved through a Sorbent Enhanced Water-Gas-Shift (SEWGS) reaction involving MgO-based sorbents. Although encouraging results were obtained in the revious related rojects, there are several key issues that remain to be investigated to rovide sufficient information to roerly scale-u the rocess. Comutational Fluid Dynamics (CFD) rovides an attractive otion to accomlish this goal in a systematic and economically feasible way. However, in order to use CFD to erform simulations of the regenerative CO 2 cature rocess, a model accounting for the variation of article orosity and its effects is required. In this roject, a oulation balance equation (PBE) aroach was used to describe the evolution of the article orosity distribution and was linked with the multihase flow dynamics governing equations. The variable diffusivity reaction rate model (that was develoed in the 2 nd year of the ICCI roject) was used to describe the orosity variation of a single sorbent article as a function of sorbent conversion and was incororated into the CFD model to redict the extent of the overall absortion/regeneration reactions. The PBE was solved with an efficient comutational technique known as Finite size domain Comlete set of trial functions Method Of Moments (FCMOM), recently develoed by the IIT multihase flow research team. THEORETICAL APPROACH In this roject, a regenerative CO 2 cature rocess was considered. The rocess consists of two reactors, the first being the absorber and the second being the regenerator. Tyically, in a coal gasification ower lant, the coal gas is first cleaned from its contaminants (articulates, sulfur, etc.) and then is sent to a water-gas-shift (WGS) reactor, and then to the CO 2 cature unit. In the regenerative rocess considered in this roject, both the WGS reaction and the CO 2 cature by sorbent absortion take lace in the absorber. The H 2 -rich syngas exiting the rocess can be converted to electrical or thermal ower, while the sorbents enter the regenerator reactor, where CO 2 is released in the resence of a regeneration gas (steam). Therefore, both in the absorber and in the regenerator u to three hases can be distinguished, i.e. the sorbents, the WGS catalyst (if needed), and a gas hase (syngas in the absorber, and CO 2 lus the regeneration gas in the regenerator). In the CFD/PBE model develoed in this roject, the gas and catalyst hases were described by standard multihase governing equations 8. However, as far as the sorbent hase is concerned, an original aroach was used based on a oulation balance equation (PBE) governing the evolution of article density distribution. The equations governing the gas hase fluid dynamics are the mass, momentum, energy and secies balances 8 : g t g v N g R g g g gn n1 (1) 7

8 v g g g t v v g g g g τ I g R v v g g g gi g g gi gi i gi g ic, s ic, s T t g g g Cg vg Tg qg H g X g g gn t v X D X R g g g gn gn gn gn In equations 1 through 4, the subscrits g, c, s, and m denote gas, catalyst, sorbent, and generic solid hase, resectively. The subscrit n denotes a comonent of a hase, is the microscoic (material) density, is the volumetric void fraction, v is the convective velocity, R gn is the reaction (roduction) rates of the N g secies in the gas hase, and and are the ressure and the stress tensor, resectively. The constant, g, is the gravitation acceleration and I gi reresents the force (er unit volume) exerted by the gas hase on the solid hase i (interface momentum transfer). The last summation on the right-hand side of equation 2 reresents the effect of the mass transfer rate R gi from the gas hase to the solid hase i. The temerature is denoted by T, the secific heat by C, the heat flux by q, and the heat of reaction by H. X gn is the gas hase mass fraction of the n th secies and D gn is the diffusion coefficients [kg/(ms)]. The sorbent articles are characterized by a distribution of the article orosity. The article orosity can be related to the article density s, using the variable diffusivity model. The article orosity distribution (PPD) is denoted by f(, t, x) and is defined such that f, d, and dx give the number of articles with orosity in the range +d at the location x+dx at the time t. The oulation balance equation governing the PPD evolution is: f G f v t s f D f t (5) n eq. (5), G=G(, t, x) is the article orosity growth rate due to the absortion reaction, which can be comuted from the absortion reaction rate. The article orosity growth rate can be negative, i.e. the article orosity decreases. D t =D t (, t, x) is the article turbulent diffusivity and is assumed to be isotroic. The sorbent hase volume fraction can be comuted from the PPD: 3 P s Ds f d 6 Therefore, the mass balance is not required for the sorbent hase (it is already included in the PBE). The mass balance of the catalyst hase is exresses as 8 : (2) (3) (4) (6) 8

9 c c c c c t v 0 The momentum, energy and granular temerature balances of the two solid hases (catalyst and sorbent) are 8 : v t m m m v v m m m m τ I I g R v v m g m gm im m m gm gm m gm g ic, s T C v T H t q m m m m m m m m 3 m J m 2 t v q τ : v m m m m m m m m m m In eq. (10), m is the hase m granular temerature (one-third the mean square of article velocity fluctuations), q m the granular energy heat flux, m is the dissiation of granular energy due to gas-solid interaction, J m is the granular energy collisional dissiation. Referring to the sorbent hase, it should be noticed how the PBE (5) is linked to the equations (8), (9) and (10). The PBE (5) rovides a distribution of article orosities, i.e. article (microscoic) densities s. On the other side, equations (8), (9) and (10) are derived on the assumtion of one uniform value of s. Therefore, the value of s in equations (8), (9) and (10) is to be interreted as an average value obtained using the PPD comuted through the PBE (5). Additionally, the sorbent hase convective velocity in the PBE (5) is obtained from equation (8) and is an average velocity (indeendent of the article densities). Finite Size Domain Comlete Set of Trial Functions Method of Moments (FCMOM In the numerical solution of the CFD/PBE model defined above, in the revious section (eqs. 1-10), secial care must aid to the PBE (5) (in fact, standard numerical are available in literature and in CFD codes to solve the other governing equations). The PBE (5) was solved in this roject using the FCMOM (Finite Size Domain Comlete Set of Trial Fnctions Method Of Moments), recently develoed by the IIT multihase flow research team. The FCMOM is a moments based numerical technique to solve monovariate and bivariate PBE 9,10. The FCMOM is comutationally very efficient and rovides accurate reconstructions of the article internal variable distribution. The FCMOM was validated both for homogeneous and for in-homogenous (satially not uniform) systems To aly FCMOM to PBE (7), the articles are characterized by the article orosity, whose values range between min (t,x) and max (t,x) (minimum and maximum values). 9 (7) (8) (9) (10)

10 Using the following coordinate transformation: t, x t, x t, x t, x min max max min 2 2 The dimensionless internal variable ranges in the domain [-1, 1]. The article orosity distribution (PPD) can be defined in dimensionless terms as f (, t, x)= f (, t, x)/f sc, where f sc is an aroriate scale factor. The dimensionless moments are defined as: i 1 ' i f d 1. (12) The dimensionless PPD f and is exressed as a series exansion truncated after the first M terms: M 1,, x n, xn ' f t c t n0 Where, n ( ) are the ortho-normal functions associated to the Legendre olynomials 12 and the coefficients c j can be exressed in terms of the moments. Therefore the PPD f is comletely determined by the moments. M is the number of moments considered. 1 (11) (13) The moments evolution equations are derived from PBE (5): i t vs i D i MB MBconv MBdiff 1 MBdiff 2 t 2i 1 i ' G f 1 max min 1 d where: 1 min max 1 MB i i 1 i 1 i t t 1 1 MBconv i i 1 v s max min i i s 1 v (14) MB i 1 D i 1 D 1 t t diff 1 i1 max min i 10

11 t 1 t 1 MBdiff 2 2 D i max min i D i i t 2 ' ' D f i f 2 max min 1 ii1 i2 1 1 t ' ' 2 D f i1 f 2 max min 1 i1 i i1 1 1 t ' ' D 2 f i2 f 2 1 i 2 i 1 i t i1 D 2 2 t ii 1 D 2 2 i 2 max min max min In eq. (14), f ' f, ' 1 1 denote the values of the 1 and 1, resectively; max is equal to f ' derivative with resect to at min. Eq. (14) was derived in ' the assumtion that f is equal to zero at 1 and 1 ; the general form of eq. (14) has also been derived and is available but is omitted here for sake of clarity. The structure of equation (14) is as follows. On the left-hand side, the first term is the moments accumulation rate, the second term reresents convection, the third term reresents article diffusion. The remaining terms on the left-hand side MB, MBconv, MBdiff 1, MBdiff 2 are due to the coordinate transformation (12): MB accounts for the temoral variation of the boundaries min and max and is resent also in homogeneous rocesses 9. MBconv accounts for the satial variation of the boundaries in resence of MB diff MB diff 1 2 convection. and account for the satial variation of the boundaries in resence of article diffusion. On the right-hand side of equation (14), there is a term due to the article orosity growth rate G. Moments evolution equations must be couled with moving boundary conditions min roviding the governing equations for t, x and max t, x. Such boundary conditions were derived both for homogeneous and for in-homogeneous systems by Strumendo and Arastooour 9,11. As an examle, in a homogeneous (satially uniform) system in which the article orosity changes are only due to the article orosity growth (absortion/regeneration kinetics), the boundary conditions are obtained alying the growth rate G directly to min and max. For classical PBE in which the articles are characterized by the article size (instead of the article orosity which defines the article state in this roject), the FCMOM was validated for the cases of constant, linear, diffusion-controlled growth, with and without

12 article formation, article dissolution (negative growth), article aggregation, simultaneous article aggregation and growth, article diffusion and article convection by Strumendo and Arastooour 9,11. In Figure 2a, the results of the simulations for the case of diffusion-controlled growth are lotted. The article size growth rate (in terms of article radius) is inversely roortional to the article radius, through a constant K=0.78 micron 2 /sec. The numerical solution of the article size distribution, obtained with 10 moments, was comared with the analytical solution at the final time (Figure 2a); the numerical solution closely reresents and has converged to the exact size distribution. Figure 2a shows the simulation results when the FCMOM is alied to article dissolution (negative growth). In this alication, articles decrease their size, because mass is transferred from the articulate hase to the fluid hase, until they dissolve. The roduct between the constant dissolution rate K and the final time t fin is K.t fin =0.5 micron. Again, the comarison between the solution by the FCMOM (with 10 moments) and the analytical solution shows excellent agreement. a b Figure 2: Particle size distribution (PSD) vs. article radius: in Figure 2a (left), comarison between the numerical solution by the FCMOM and the exact solution for the diffusion-controlled growth rate. The final time is 20 sec. In Figure 2b (right), the numerical solution by the FCMOM vs. exact solution for article dissolution. In both Figures 2a & 2b, M=10. Simulations of the CO 2 cature regenerative rocess The regenerative CO 2 cature rocess consists of two reactors, i.e. the absorber and the regenerator. The syngas mixture (CO, CO 2, H 2 O, H 2, CO 2, diluents), the sorbent articles and the WGS catalyst (if needed) are fed to the absorber, where the CO 2 and H 2 are roduced by the WGS reaction and CO 2 is removed from the syngas by absortion. At the exit of the absorber, the H 2 -rich syngas is sent to ower roduction; while the carbonated sorbents are sent to the regenerator reactor and can be artially recycled to the absorber (the catalyst can be either recycled to the absorber or sent to the regenerator with the sorbent). In the regenerator, the CO 2 is released from the sorbent articles to a regeneration gas (steam); at the regenerator exit, the steam/co 2 stream is searated from the regenerated sorbent articles, which are fed again in the absorber. 12

13 The objective of this roject was to assess the erformance of the sorbent for caturing CO 2 in a regenerative rocess and to rovide a base case design for a circulation fluidized bed reactor. To achieve this objective, we used the NETL Carbon Cature Unit (C2U) design as our starting oint. The data rovided directly by NETL and some details has been reorted by Clark et al 13. The system is a bench scale CO 2 cature unit using a solid sorbent to cature CO 2 from a flue gas stream. The exeriment data on hydrodynamics and oeration of a continuously looing CO 2 cature system was used as a baseline for validation of our CFD model. As the first ste, we used the cold, non-reacting conditions to study the gas-article flow behavior of the system. A schematic icture of the exerimental unit is shown in Figure 3. The unit has an overall height of 3.35 m, a width of 1.16 m, and a deth of 0.71 m. The system is made of clear olycarbonate material which allows visual observations and image recordings of article flow. The regenerator and loo seals are cm ID clear olycarbonate. Piing is 5.08 cm ID clear olycarbonate excet for the 2.54 cm iing from the riser to the cyclone. Two-ring heating coils are located in the regenerator and loo seal 1 and reduce cross-flow area from cm 2 to an effective flow area of cm 2. The CO 2 -containing gas enters the bottom of the absorber and mixes with fresh sorbent. In the cold exeriments with inert sorbent, air is used as the fluidization media. The sorbent articles mix with the feed gas absorbing CO 2 into the article through chemical reaction. The carbonated articles flow u the riser, turn and flow into the cyclone. In a reacting C2U exeriment, CO 2 -lean gas is searated from articles in the cyclone and exit the system, and the carbonated articles ass through a loo-seal and into the regenerator where CO 2 is released from the sorbent-articles by heating the sent sorbent in a regeneration gas stream (e.g., steam). The CO 2 gas exits the C2U system and the regenerated sorbent articles continue through the loo to the next loo-seal. The fresh sorbent articles ass through the loo-seal to the absorber and the rocess continues. To maintain gas-article flow in a CO 2 cature loo, gases are injected throughout the system to kee articles fluidized. The main gas inlet is the absorber inlet. Other gases are injected into the system to kee articles fluidized and flowing. In an active CO 2 transort loo, fluidizing gases, injected downstream of the cyclone will be the regeneration gases such as steam. In the non-reacting cold-flow exeriment the solid articles were assumed to be sherical with mean diameter of 185 µm and density of 2480 kg/m 3. The system is oerated at room temerature and atmosheric ressure and only air is used as the fluidization media. Table 1 rovides the flow rates of gases and injection oints throughout the system. Cold flow simulations were erformed on the riser section of the NETL C2U circulating fluidized bed unit using our (Abbasi and Arastooour 14 ) 3D Eulerian-Eulerian CFD model. Our initial focus was directed toward the absorber and riser section of the circulating fluidized bed as shown in Figure 3. The objective of this stage was; 1) to calculate the ressure dro across the absorber and, 2) caturing the chugging effect as observed during the exeriments. This section has an overall height of 3.35 m. For our 13

14 simulations, the initial solid height in the absorber was set to 30 cm and at minimum fluidization state. Fluidizing gas was injected from the bottom of the absorber with a mass flow rate of kg/s while the solid mass flow rate was set to kg/s and injected from the side. Outlet boundary condition was set to atmosheric ressure. Cyclone Riser Regenerator Loo seal 1 Adsorber Figure 3. NETL C2U exerimental setu Table 1: Inlet flow rates (Positions shown in Error! Reference source not found.). Flow Location Absorber Loo Seal 1 Regenerator Loo Seal 2 Nominal Design Flow Rate (kg/s) 5.0e e e e-4 14

15 Following the cold flow simulations of the riser section of the NETL C2U circulating fluidized bed unit using our 3D Eulerian-Eulerian CFD model, the focus of our research was directed toward adding carbonation and water gas shift (WGS) reactions to the system to investigate the erformance of the MgO based sorbent in caturing CO 2 and enhancing H 2 roduction in the reactor. RESULTS AND DISCUSSION According to Clark et al. 13, chugging occurs when a large mass of articles lifts from the fluidized bed and moves into the cone leading into the riser. The cone-constriction revents articles from flowing smoothly into the riser and articles lug the riser ie. As shown in Figure 4, chugging was observed during the exeriment and the same behavior was catured by our CFD model as shown in Figure 5. A comarison of the redicted ressure dro in the lower mixing zone of the riser with the exerimental data is resented in Figure 6, indicating that the model rediction is in good agreement with the exerimental data. The average calculated ressure dro from the simulation is 4.29 kpa averaged over 12 sec comare to 3.87 kpa average ressure dro measured in the exeriment over 150 sec. One of the factors contributing to the slight differences between the redicted and the measured ressure dro can be due to the difference in averaging time, while overestimation of drag force between gas and articulate hase (calculated by ordinary drag models) may also be another contributing factor. Ordinary models assume a homogenous disersion of article hase in the fluid, and as a consequence, the calculated drag force between hases is higher than the actual drag force, in which articles make clusters in the fluid hase. In order to eliminate model over-rediction, an Energy Minimization Multi-Scale aroach (EMMS) 15 was incororated in the model. The EMMS aroach is based on the assumtion of minimizing energy transfer between the clusters and dilute hase inside a gas-solid environment, while the traditional drag models (e.g. Syamlal and O brien 16 ) assume a homogenous distribution of solid articles in the gas medium, and therefore neglect all heterogeneities inside the flow. Figure 7 shows the comarison between EMMS and Syamlal and O brien 16 model for a bubbling fluidized bed. The results clearly indicate that EMMS rovides an accurate rediction of both the instantaneous fluctuations and the average and ressure dro. While Symlal and O brien 16 model does not redict any major fluctuation, it can redict the average value with accetable accuracy. Given the significantly shorter comutational time required with the Syamlal and O brien 16 model, all simulations should be erformed with Syamlal and O brien 16 model, and EMMS should be used when a more accurate data on instantaneous behavior of flow is required. 15

16 Figure 4. Chugging effect reorted during NETL exeriments Figure 5. Chugging effect catured in simulations 16

17 Figure 6. Exerimental ressure dro data comare to the simulation results across the lower art of the adsorber Figure 7. Comarison between EMMS and Syamlal drag models 17

18 After validating the cold flow hydrodynamics model against the exerimental data at low ressure and temerature, the model was used as a base-case for simulations of reactive flow at elevated ressure and temeratures. Because of the effect of temerature and ressure on the viscosity and density of the fluidizing gas, the hydrodynamics of the system is different at elevated ressures and temerature, a reliminary study was erformed to determine the required inlet gas velocity at the target ressure and temerature (i.e., T=425 C and P=20 atm). The goal was to find an oerating condition that rovides the same solid inventory in the riser. Table 2 shows the baseline oerating condition for the reactive case. Table 2. Baseline Oerating Condition Pressure, atm 20 Temerature, C 425 Inlet gas velocity, m/s 0.15 Inlet solid rate, kg/s Inlet CO 2 mole fraction, v/v% 50 The carbonation reaction between CO 2 and the MgO-based solid sorbent was added to the model using the variable diffusivity shrinking core model (Abbasi et al. 17 ). To investigate the effect of oerating conditions on the CO 2 cature erformance, a set of simulations were erformed at different solid circulating rates and gas residence time. Table 3 shows the inlet gas velocity and solid circulating rates for different cases: Case Table 3. Test Cases Oerating Condition Inlet gas velocity (m/s) 5xSolid xSolid % less gas % less gas Solid circulating rate (g/s) Figure 8 shows the effect of solid circulation rate on the CO 2 cature. At the NETL baseline condition, the extent of CO 2 cature is less than 15%. By increasing the solid circulation rate by factors of 5 and 10, the extent of CO 2 cature increased to 40%, indicating that the solid hold u in the riser has reached a level that no longer controls the extent of the CO 2 cature. This can be attributed to the denser solid flow in the lower art of the riser and also denser solid flow in the transort zone. Figure 9 shows the effect of gas inlet velocity on the extent of CO 2 cature, decreasing gas inlet velocity by 35% (and therefore increasing gas residence time) u to 60 % CO 2 cature can be achieved. Following the imlementation of the carbonation reaction in the model, the water gas shift reaction is added to the reactive system to investigate the erformance of the MgO-based solid sorbent in the sorbent enhanced water gas shift (SEWGS) reaction. Hassanzadeh 7 evaluated the same MgO-based dry

19 sorbent in a SEWGS reaction in acked-bed reactor at 350 C and showed that the sorbent is caable of caturing over 90% of the CO 2 and increase the hydrogen mole fraction from 37 mol % (dry basis) in the inlet to 55 mol % in the outlet. In this study the same assumtions were made and erformance of the sorbent was evaluated for a SEWGS rocess. In this CFD/PBE simulation model, the inlet gas was a simulated syngas with a comosition of 20% CO 2, 20% H 2 O, 30% CO, and 30% H 2. Initially, the reactor contained only steam (H 2 O). The oerating condition corresonds to the 5 x Solid case excet the temerature was at 350 C. Equilibrium limit Figure 8. CO 2 outlet to inlet mole fraction ratio vs. time, at different solid circulating rates Equilibrium limit Figure 9. Effect of gas inlet velocity on CO 2 cature Figure 10 shows the gas comosition (wet basis) in the reactor outlet corresonding to about 60% CO 2 cature with and exit gas containing 65% H 2. Since the carbonation reaction is limited by equilibrium, in this case due to the lower concentration of CO 2, the extent of CO 2 catured by the sorbent is lower than that in the cases with 50% inlet CO 2. 19

20 Equilibrium limit Figure 10. SEWGS gas outlet concentration at 350 C CONCLUSIONS AND RECOMMENDATIONS A Comutational Fluid Dynamic (CFD) model was develoed to simulate CO 2 cature and enhancement of H 2 roduction in the regenerative MgO-based rocess for CO 2 cature in Integrated Gasification Combined Cycle (IGCC). A oulation balance equation (PBE) aroach was used to describe the evolution of the article orosity distribution and was linked with the multihase flow dynamics governing equations. The variable diffusivity reaction rate model (that was develoed in the revious hase) was used to describe the orosity variation of a single sorbent article as a function of sorbent conversion and was incororated into the CFD model to redict the extent of the overall absortion/regeneration reactions. The PBE was solved with an efficient comutational technique known as FCMOM recently develoed by the IIT multihase flow research team. Cold flow simulations were erformed on the riser art of the existing C2U cold flow unit at the NETL circulating fluidized bed unit using a 3D CFD model. The objective of this work was; 1) to calculate the ressure dro across the absorber and, 2) caturing the chugging effect observed in the NETL exeriments. The results clearly indicate that EMMS rovides an accurate rediction of both the instantaneous fluctuations and the average and ressure dro. While Symlal and O brien 16 model does not redict any major fluctuation, it can redict the average value with accetable accuracy. Given the significantly shorter comutational time required with the Syamlal and O brien 16 model, all simulations should be erformed with Syamlal and O brien 16 model, and EMMS should be used when a more accurate data on instantaneous behavior of flow is required. The results of the CFD simulation of the CO 2 cature in the absorber with a (50/50) CO 2 /N 2 mixture indicate that, at the NETL baseline condition, the extent of CO 2 cature is less than 15%. By increasing the solid circulation rate by factor of 10, the extent of CO 2 cature can increase u to 40%, while by decreasing the gas inlet velocity by 35% (and therefore increasing gas residence 20

21 time), the extent of CO 2 cature can aroach u to 60%. The results of the CFD simulations with a simulated syngas containing 20% CO 2, 20% H 2 O, 30% CO, and 30% H 2 indicate that u to 60% CO 2 can be catured in the absorber, while enhancing the hydrogen content of the reactor exit gas to 65% through the Sorbent Enhanced Water-Gas-Shift (SEWGS) reaction. 21

22 REFERENCES 1. US Deartment of Energy, Carbon Sequestration Technology Roadma and Program Plan, US Deartment of Energy, Financial Assistance Funding Oortunity Announcement for Suort of Advanced Coal Research at U.S. Colleges and Universities, FOA Number: DE-FOA , Tontiwachwuthikul P., Research and Develoment Activities on High Efficiency Searation Process Technologies for Carbon Dioxide Cature from Industrial Sources at University of Regina, Canada, Energy Conversion and Management, v 37, n 6-7, (1996). 4. Hassanzadeh A., Abbasian J., A Regenerative Process for Pre-Combustion CO 2 Cature and Hydrogen Production in IGCC, Proceedings of the 26th Annual International Pittsburgh Coal Conference, Hassanzadeh and J. Abbasian, Regenerable MgO-Based Sorbents for High- Temerature CO2 Cature from Syngas. 1. Sorbent Develoment, Evaluation, and Reaction Modeling. Acceted for ublication in FUEL, 89, , Abbasian J., Hassanzadeh A., A Regenerative Process for CO 2 Cature and Hydrogen Production in Integrated Gasification Combined Cycle (IGCC) Processes, Paer resented at the Sring 2009 AIChE National Meeting, Tama, FL, Hassanzadeh A., A Regenerative Process for Pre-Combustion CO 2 Cature and Hydrogen Production in IGCC, PhD dissertation, Illinois Institute of Technology, Benyahia S., Syamlal M., O Brien T.J., Summary of MFIX equations , from URL htt:// July Strumendo M., Arastooour H., Solution of PBE by MOM in Finite Size Domains, Chemical Engineering Science, 63, (2008). 10. Strumendo M., Arastooour H., Solution of Bivariate Poulation Balance Equations Using the FCMOM, Industrial and Engineering Chemistry Research, 48(1), (2009a). 11. Strumendo M., Arastooour H., Solution of Poulation Balance Equations by the FCMOM for In-homogeneous Systems, Industrial and Engineering Chemistry Research, under revision (2009b). 12. Courant R., Hilbert D., Methods of Mathematical Physics, Interscience Publishers, New York,

23 13. Clark, S., Snider, D.M., and Senik, J., CO 2 Adsortion Loo Exeriment with Eulerian Lagrangian Simulation. Powder Technology (2013). 14. Abbasi, E. and Arastooour, H. CFD Simulation of CO 2 Sortion in a Circulating Fluidized Bed Using the Deactivation Kinetic Model, Proceeding of the 10th International Conference on Circulating Fluidized Beds and Fluidization Technology, CFB 10, Oregon, USA, (2011). 15. Nikolooulos, A., Nikolooulos, N., Charitos, A. Grammelis, P. Kakaras, E. Bidwe, A.R Varela, G., High-resolution 3-D Full-loo Simulation of a CFB Carbonator Cold Model, Chemical Engineering Science, Volume 90, , (2013). 16. Syamlal, M. O'Brien T. J., "Fluid Dynamic Simulation of O 3 Decomosition in a Bubbling Fluidized Bed", AIChE Journal Volume 49, Issue 11, ages , (2003). 17. Abbasi, E., Hassanzadeh, A., Abbasian, J., Regenerable MgO-based sorbent for High Temerature CO 2 Cature from Syngas: 2. Two-zone Variable Diffusivity Shrinking Core Model with Exanding Product Layer, Fuel, 105, , (2013). 23

24 DISCLAIMER STATEMENT This reort was reared by Javad Abbasian, Illinois Institute of Technology, with suort, in art, by grants made ossible by the Illinois Deartment of Commerce and Economic Oortunity through the Office of Coal Develoment and the Illinois Clean Coal Institute. Neither Javad Abbasian & Illinois Institute of Technology, nor any of its subcontractors, nor the Illinois Deartment of Commerce and Economic Oortunity, Office of Coal Develoment, the Illinois Clean Coal Institute, nor any erson acting on behalf of either: (A) (B) Makes any warranty of reresentation, exress or imlied, with resect to the accuracy, comleteness, or usefulness of the information contained in this reort, or that the use of any information, aaratus, method, or rocess disclosed in this reort may not infringe rivately-owned rights; or Assumes any liabilities with resect to the use of, or for damages resulting from the use of, any information, aaratus, method or rocess disclosed in this reort. Reference herein to any secific commercial roduct, rocess, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imly its endorsement, recommendation, or favoring; nor do the views and oinions of authors exressed herein necessarily state or reflect those of the Illinois Deartment of Commerce and Economic Oortunity, Office of Coal Develoment, or the Illinois Clean Coal Institute. Notice to Journalists and Publishers: If you borrow information from any art of this reort, you must include a statement about the state of Illinois' suort of the roject. 24