Availale online at www.sciencedirect.com Energy Procedia (29) (28) 655 66 Energy Procedia www.elsevier.com/locate/procedia www.elsevier.com/locate/xxx GHGT-9 Pre-comustion CO 2 capture for IGCC plants y an adsorption process Johanna Schell, Nathalie Casas, Marco Mazzotti* ETH Zurich, Institute of Process Engineering, Sonneggstrasse 3, CH-892 Zurich, Switzerland Elsevier use only: Received date here; revised date here; accepted date here Astract A promising approach to provide in the near future electricity from fossil fuels, for the worldwide increasing energy requirements with near-zero CO 2 emissions is the IGCC (Integrated Gasification Comined Cycle) technology with pre-comustion CO 2 capture. One important aspect within this technology is the development of advanced processes for the capture of CO 2, i.e. its separation from hydrogen, with improved efficiency to decrease the energy consumption in this step. In our laoratory, the CO 2 /H 2 separation y a pressure swing adsorption process is investigated, including oth the experimental characterization of suitale commercial and new adsorents as well as the development, design and optimization of a proper process concept including the required purification steps y a process simulation. c 29 Elsevier td. Open access under CC BY-NC-ND license. IGCC; CO 2 Capture; PSA; Adsorption Isotherms; Process Modeling. Introduction Caron dioxide capture and storage (CCS) is a set of technologies for the capture of CO 2, its transport to a storage location, and its isolation from the atmosphere. CCS allows avoiding the increase of atmospheric CO 2 concentration and in this way mitigating climate change, while at the same time enaling a continued use of fossil fuels. A promising approach for near-zero CO 2 emission power plants to e realized within the near future is the IGCC technology (Integrated Gasification Comined Cycle), where the fuel, e.g. coal, is gasified and converted mainly to H 2 and CO 2 in a shift reactor. Before entering the turine, CO 2 is removed from the gas. This is a so-called pre-comustion capture and represents one of the key challenges within this process. One possile method for CO 2 removal is adsorption, which can e applied in a pressure swing adsorption process. To assess the potential of this option three different aspects have to e considered, namely the characterization of suitale adsorents for CO 2 /H 2 * Corresponding author. Tel.: +4 44 6322456; fax: +4 44 6324. E-mail address: marco.mazzotti@ipe.mavt.ethz.ch Both authors contriuted equally towards the completion of this work. doi:.6/.egypro.29..86
2 656 Johanna Schell, J. Schell Nathalie et al. Casas / Energy / Energy Procedia Procedia (29) 655 66 (28) separation, the design of an appropriate pressure swing adsorption process and the development of an overall process concept, including also the preliminary purification steps. Different existing adsorents are capale of separating CO 2 and H 2, however to make the technology profitale, improvements in the adsorption ehavior are needed. Therefore, activated caron, a well-known adsorent, is chosen to form a reference case that can e used for comparison of newly developed, tailored adsorents. One important characteristic of an adsorent is its equilirium adsorption isotherm. The adsorption isotherms of CO 2 and H 2 on activated caron have een measured at 45 C up to 9 ar y a gravimetric method and the results are presented. For the design of a pressure swing adsorption process various possiilities exist to comine the four asic steps, namely pressurization, adsorption, lowdown and purge. To find the est configuration and the optimal conditions for the process, conceptual models ased on equilirium theory as well as rigorous process simulation are required. A rigorous PSA model is presented and three different known asic PSA configurations are descried in this work. 2. Experimental section 2.. Materials Activated caron (AP3-6) was otained from Chemviron Caron (Neu-Isenurg, Germany). According to the manufacturer the diameter of the pellets was.3 mm. Prior to the adsorption measurements the sample was dried under vacuum at 5 C for one day. The gases used in this study were otained from Pangas AG (uzern, Switzerland), namely CO 2 and H 2 at a purity of 99.9 and He at a purity of 99.999%. The critical properties of the adsorates are as follows: T c (He) = 5.26 K, P c (He) = 2.26 5 Pa, c (He) = 69.3 kg/m 3, T c(co2) = 34.K, P c (CO 2 ) = 73.7 5 Pa, c (CO 2 ) = 467.6 kg/m 3, T c (H 2 ) = 33.9 K, P c (H 2 ) = 3. 5 Pa, c (H 2 ) = 3 kg/m 3. 2.2. Adsorption measurements The adsorption experiments were performed in a Ruotherm magnetic suspension alance (Ruotherm, Bochum, Germany), which allows measurements at pressures and temperatures up to 45 ar and 25 C, respectively. The sample weight is measured with an accuracy of. mg. Furthermore, the fluid ulk density can e otained in situ y measuring the uoyancy of a titanium sinker, whose volume has een calirated independently. The experimental setup and the measurement procedure are descried in more detail in [,2], nevertheless they are shortly summarized in the following. After having placed aout 2.5 g of adsorent in the alance, the system is evacuated at a temperature of 5 C and the weight M, which consists of the mass of the metal parts and the mass of the activated caron sample, met ac i.e. m m, is measured: M m m = met ac () Then the system is filled with helium, and the volume of the adsorent and the metal parts V where V ac met V M, M M, T ac + V met is otained: (2) T is the mass measured at the given experimental conditions, i.e. at density and temperature T. After evacuating it again, the system is then filled with the fluid to e adsored, namely CO 2 or H 2. As the volume of the adsored phase cannot directly e measured, the adsorption is commonly represented y the excess adsorption,t which is defined according to the first part of equation (3). From the measurement at a given pressure and temperature the excess adsorption can e calculated according to the second part of equation (3).
, T m V ads ads Johanna J. Schell, Nathalie et al. / Energy Casas Procedia / Energy Procedia (29) 655 66 (28) 657 3 M, T M V V met ac ads ads where m and V are the mass and the volume of the adsored phase, respectively. The experimental results reported in the next section are given in terms of the molar excess adsorption n ex per ac unit mass of coal m, calculated with the molar mass of the adsored gas M m. (3) n ex M m (, T ) ac m (4) 2.3. Experimental results and discussion The isotherm measurements were performed at 45 C and at pressures up to 5 ar. In Figure, the molar excess adsorption of CO2 on AP3-6 is shown against the ulk density. The experimental data exhiit the usual ehavior of excess adsorption isotherms: at low ulk densities the excess amount adsored increases with increasing ulk density to reach a maximum value, whereas at high ulk densities the molar excess adsorption decreases almost linearly with increasing ulk density. Additionally in Figure the molar excess of H 2 on AP3-6 is represented as a function of the density. For the H 2 excess isotherm no maximum is oserved; this is due to the fact that the working temperature is much higher than the fluid critical temperature. Compared to the CO 2, a higher pressure is therefore needed to reach the fluid critical density. Figure : Molar excess isotherms of CO (lue triangles) and H (red circles) on activated caron AP3-6 at 45 C. 2 2 At the same process conditions, i.e. at the same pressure and temperature, CO 2 adsors significantly more than H 2, suggesting that the two gases can e separated through the PSA process. However, this conclusion needs to e supported y competitive adsorption measurements of mixtures of these two components. As it has een anticipated aove, the activated caron corresponds to the reference case scenario and there these results show the lower limit of adsorption performance for all newly developed materials.
4 658 Johanna Schell, J. Schell Nathalie et al. Casas / Energy / Energy Procedia Procedia (29) 655 66 (28) 3. Mathematical model The pressure swing adsorption process is descried y a one-dimensional solid surface resistant model, in which temperature as well as concentration of the solid phase is descried using lumped equations. The following simplifying assumptions are made: ideal gas ehavior, negligile radial temperature and concentration gradients as well as temperature independent diffusivities and physical properties. Furthermore, the pressure is assumed to e constant during the adsorption and the purge step. The overall mass alance for every component i in the column is given as follows: c uc i i qi c D i i t * z t * z (5) z where c i and q i are the fluid and the adsored phase concentration of species i, respectively; u is the superficial gas velocity; * and are the overall and the ed void fraction, respectively; is a dimensionless relation of the overall void fraction, * / * ; is the axial dispersion coefficient of species i; t and z are the time and space Di coordinate. By expressing the mass transfer rate with a linear driving force (DF) model, the material alance in the adsored phase takes the following form: qi k ( * ) miap qi qi (6) t where k mi is the linear lumped mass transfer coefficient for species i; a p is the specific surface of the adsorent particles and q i * is the adsored phase concentration in equilirium with the gas phase, which can e calculated using the extended angmuir isotherm: q* q kp i i i si n kp where Pi is the partial pressure and qsi and kiare the saturation capacity and the angmuir equilirium constant. To account for temperature changes during the PSA process, heat alances can e written for the fluid and the adsored phase: T ha s s p q ( T Ts ) ( H ) t C C t s n s (7) (8) T T ( ut) q 2h K T t t z C t r C C z z s n ( H ) ( T Tw ) * g i* g * g (9) where T, T and T are the fluid, the wall and the solid phase temperature, respectively; is the dimensionless w s relation etween Cs C g, which are the heat capacity of the solid and the fluid, respectively; H is the heat of adsorption of species ; h and h are the heat transfer coefficient for the fluid phase and wall and the fluid and solid s phase, respectively; K is the thermal axial dispersion coefficient. The change in pressure during the pressurization and the lowdown step is descried y the Ergun equation: 2 p 5 ( ).75( ) u uu () z d d 3 2 3 p p where is the dynamic viscosity; the particle diameter and is the fluid phase density. d p
Johanna J. Schell, Nathalie et al. / Energy Casas Procedia / Energy Procedia (29) 655 66 (28) 659 5 3.. Simulation Results The finite differences method has een applied to discretize in space the partial differential equations. The ordinary differential equations are then solved numerically in Fortran y using a commercial ODEs solver. In Figure 2 the results of a simulation of the adsorption step are shown. The column was initially filled with H 2. The feed gas consists of CO 2, H 2, CO and CH 4, with a feed gas composition of.7,.755,.4 and.35, respectively. However for the sake of clear illustration H 2 is not shown in Figure 2, since its composition can e readily otained y alancing the other components to one. All other input parameters are chosen according to [3]. CO 2 CO CH 4 Figure 2: Simulated exit profiles CO 2, CO and CH 4. By evaluating pressure, temperature and composition profiles, the model descried can e used to define the final process design. In particular, this is done y optimizing the sequence of the four PSA steps, namely pressurization, adsorption, lowdown and purge and the process conditions. In the next section three different PSA configurations are presented. 4. PSA Process configurations Three different asic PSA process configurations are descried in literature [4]. The process concept is chosen depending on the purity requirements of the components, which have to e separated. Most industrial units are of the so-called stripping PSA type, where a highly pure light product (less adsoring component) is otained (see Figure 3a). For the sake of simplicity the pressurization and the lowdown steps are not shown in the figure. To achieve a high purity of the heavy product (strongly adsoring component) another process configuration, called rectifying PSA is applied (Figure 3). Comining the two process configurations in the so called dual-reflux PSA, allows oth the heavy and the light products to e produced at a high purity, as shown in Figure 3c.
6 66 Johanna Schell, J. Schell Nathalie et al. Casas / Energy / Energy Procedia Procedia (29) 655 66 (28) a) Stripping PSA ) Rectifying PSA c) Dual-reflux PSA x xh H x F( (/) P H /( P ) Heavy product x H Rectifying reflux R R Heavy product x H x F Adsorption P H Purge P Purge P H Adsorption P x F Feed P H Rectifying section Strippin gsection P x F Stripping Feed ight product x x x x x F( (/) P / P H ) F H ight product x reflux R S Figure 3: Three different PSA process configurations[4] a) Stripping PSA; ) Rectifying PSA and c) Dual reflux PSA. 5. Conclusion The development of a H 2 /CO 2 separation process for an IGCC plant with pre-comustion CO 2 capture needs a fundamental material evaluation as well as a rigorous process development including modeling. To estalish a reference case, the excess adsorption isotherms of CO 2 and H 2 at 45 C on activated caron AP3-6 were measured up to 5 ar. The higher adsorption of CO 2 on the activated caron at the same process conditions is required for an effective separation process. The potential of novel adsorents will e evaluated with respect to this reference case. A model used to descrie the PSA process is presented, consisting of oth material and heat alances. Evaluating temperature, pressure and concentration profiles leads to a final PSA process configuration, which meets the oundary conditions requested y the IGCC process, e.g. the constraints given y the gas turine. Acknowledgement The research leading to these results has received funding from the European Union s Seventh Framework Program (FP7/27-2) under grant agreement n 297 (the DECARBit proect). References. Ottiger, S., et al., Environ. Prog. 25 (26) 355. 2. Pini, R., et al., Adsorption 4 (28) 33. 3. Park, J.-H., et al., Chem. Eng. Sci. 53 (998) 395. 4. Diagne, D., et al., J. Chem. Eng. Jpn. 27 (994) 85.