Simulation study. November 2015

Transcription

1 Simulation of the post-combustion CO2 capture process by absorption-regeneration using amine based solvents: application to Brevik cement plant flue gases Simulation study Dr Lionel Dubois Research Coordinator CO2 Capture and Conversion Scientific Coordinator of the ECRA Academic Chair University of Mons Faculty of Engineering (FPMs) Chemical and Biochemical Engineering Department November 2015 Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 1

2 Summary In the context of the Norcem Brevik project in Norway where different post-combustion CO2 capture technologies are tested on a real flue gas in the Brevik Cement plant, and especially the absorption-regeneration process (Aker Clean Carbon), the purpose of the present study is to develop a simulation model (using Aspen Hysys TM ) allowing to simulate the application of the absorption-regeneration CO2 capture process to the Brevik cement plant flue gas, especially to highlight the interest of using alternative process configurations in order to reduce the energy consumption of the process. The simulations are carried out considering a pilot unit used during a previous European project, namely the CASTOR/CESAR one (see Fig. 1). In a first step, these simulations are performed considering the benchmark monoethanolamine (MEA) 30 wt.% as solvent but in a second step, other solvents will be considered (as piperazine (PZ) or methyldiethanolamine (MDEA) activated by PZ (@MDEA)). Figure 1: Comparison between MTU from Aker and CASTOR/CESAR pilot The interests of the present work for HeidelbergCement is triple: 1/ All the parameters related to the Mobile Testing Unit (MTU) of Aker Clean Carbon are not available for carrying realistic simulations (such as the solvent, some operating parameters, the details of the process configuration, etc.) whereas all the parameters in relation with the considered pilot are available in literature; 2/ The CASTOR/CESAR pilot has a CO2 capture capacity 10 times higher than MTU of Aker, which will give to HeidelbergCement some data in relation with a bigger unit; Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 2

3 3/ The objective of the present work is complementary to Aker one: Aker applied a precise process (MTU configuration and optimized solvent) and in the present case the simulation will highlight the interest of different alternative configurations and with different solvents. In addition to the interests for HeidelbergCement, this study will give the opportunity to submit a scientific publication in a Journal (such as the International Journal of Greenhouse Gas Control - Impact Factor of and 5-Year Impact Factor of 4.727) and to present an interesting paper during a future international conference on CO2 capture (certainly at the GHGT-13 Conference in Lausanne, November 2016). Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 3

4 Paper to be submitted for publication in (journal to be confirmed): Comparison of various configurations of the absorption-regeneration process for the post-combustion CO2 capture applied to cement flue gases Lionel Dubois, Julien Gervasi, Guillaume Pierrot and Diane Thomas* Chemical and Biochemical Engineering Department, Faculty of Engineering, University of Mons, 16 Place du Parc, 7000 Mons, Belgium. Corresponding author. Tel.: ; fax: address: (D. Thomas). Abstract ( words) Carbon Capture Storage and Utilization (CCSU) has gained widespread attention as an option for reducing greenhouse gas emissions from power plants but specific developments are still needed for the application to cement plants (flue gas with higher CO 2 content in comparison with power plants ones). More precisely, the post-combustion CO 2 capture process by absorption-regeneration using amine based solvents is the more mature technology for the application in the cement industry but it is still needed to reduce its costs. The present study is focusing on the Aspen Hysys TM simulation (with conventional monoethanolamine - MEA 30 wt.% as solvent) of different CO 2 capture process configurations (namely Rich Solvent Recycle (RSR), Solvent Split Flow (SSF), Lean/Rich Vapor Compression (L/RVC)) applied to the flue gas coming from the Norcem Brevik cement plant. For each configuration a parametric study was carried out in order to identify the operating conditions ((L/G) vol. ratio, split fraction, flash pressure drop, etc.) minimizing the solvent regeneration energy. It was shown that these configurations allow energy savings in the range 4 to 14%, LVC and RVC leading to the higher ones. A decrease of 50% of the condenser cooling energy was also noted. As perspectives, such study will be carried out with other solvents such as piperazine (PZ) (alone or as activator). Keywords: Post-combustion CO 2 capture; Absorption-regeneration process configurations; Aspen Hysys TM simulation; Rich solvent split flow; Lean vapor compression; Cement flue gases; Highlights (to be submitted separately, max 85 characters including spaces): The post-combustion CO 2 capture absorption-regeneration process was simulated. Different process configurations were simulated using Aspen Hysys TM. The rich solvent split flow and the lean vapor compression were considered. The specificity of the work is the treatment of cement flue gases. Significant process energy savings were highlighted. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 4

5 1. Introduction Based on the report (IEA, 2013) published by the International Energy Agency (IEA), Carbon Capture Storage and Utilization (CCSU) process chain is absolutely needed for an efficient reduction of the greenhouse gases from industrial plants and especially of carbon dioxide. Regarding more specifically the carbon capture technologies, several studies have already been carried out for their application to power plants and real installations are now under operation, such as for example at Boundary Dam (Saskatchewan, Canada) where the post-combustion absorption-regeneration process using amines based solvents is used as carbon capture technology. A lot of large pilot and demonstration plants are presented in (Idem et al., 2015). Nevertheless, another big CO 2 emitting sector which contributes to around 5% of the global CO 2 emissions and to 30% of the industrial emissions, namely the cement industry, still needs a lot of research studies and technical developments before implementing CCSU. In this context, the European Cement Research Academy (ECRA) and the University of Mons (UMONS) established in 2013 a new Academic Chair with the purpose of studying the applicability of Carbon Capture and its Reuse for its specific application in the cement industry where the CO 2 content of the gas to treat (namely between 17% and 35% depending on the plant) is higher than for power plant flue gases (typically between 5% and 15%). This Academic Chair is also supported by HeidelbergCement. Indeed, Norcem AS (HeidelbergCement Group) have joint forces with ECRA to establish a small-scale test centre for studying and comparing various post-combustion CO 2 capture technologies (namely amine scrubbing, membrane technology, solid adsorbent technology and carbonate looping process), and determining their suitability for implementation in modern cement kiln systems (Bjerge and Brevik, 2014). The project does not encompass CO 2 transport and storage. The small-scale test centre has been established at Norcem s cement plant in Brevik (Norway). This project (launched in May 2013) has received funding from Gassnova through the CLIMIT program and it is scheduled to conclude in spring 2017 (Test Step 1). Regarding more specifically the post-combustion CO 2 capture technology applying amines based absorption-regeneration process, the key point for allowing a large deployment of this technology is clearly its cost. In order to improve this process and to reduce its cost, three ways are possible. Firstly, the development of new solvents (new solutions or new blends) with high absorption capacity while keeping good absorption kinetics allows to reduce significantly the solvent regeneration energy. An example is given in the works of (Singh et al., 2013) showing that new solvents have the potential to reduce the reboiler energy requirements by 20-50% in comparison with a 31 wt% MEA solution as reference solvent. Secondly, implementing more efficient equipment such as new gas-liquid contactors or new packings leads to an improvement of the absorption performances, which allows the solvents to reach a CO 2 loading which is close to its maximum absorption capacity while reducing the OPEX and CAPEX of the process. The works of (Zhao et al., 2011) show that the use of new packing elements involves a 20-30% higher mass transfer coefficients when compared to conventional Mellapak 700Y and 13 mm Pall Ring, allowing a lower packed column height. Finally, the third way to improve the global performances of the absorption-regeneration technology is implementing new process configurations in order to take advantage of a better energy integration and to reduce its energy consumption. (Le Moullec et al., 2014) give more details regarding the technical description and the interest of other process configurations). The present paper focuses on this third possibility and especially on the energy savings linked to the implementation of advanced process configurations for a flue gas issued from a cement plant. The flue gas coming from Norcem Brevik Cement plant was considered and the CO 2 capture pilot simulated in Aspen Hysys TM software was based on the installation used during the CASTOR/CESAR European Projects (Knudsen et al., 2009), all the design and operating parameters being available in order to validate the simulation. The configurations considered are described in section 2.2 and the conventional monoethanolamine (MEA) 30 wt.% was selected as solvent. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 5

6 2. Simulation of different process configurations 2.1. Aspen Hysys TM modeling parameters The modeling was developed in Aspen Hysys TM v.8.6 software using the acid gas package and the conventional efficiency mode. The solvent selected for our simulations is monoethanolamine - MEA (HO(CH 2) 2NH 2) 30 wt.% and the reactions included in the acid gas package are: HO(CH 2) 2H + NH 2 +H 2O HO(CH 2) 2NH 2 + H 3O + (1) 2 H 2O H 3O + + OH - H 2O - + HCO 3 H 3O CO 3 CO 2 + OH - - HCO 3 - HCO 3 CO 2 + OH - (2) (3) (4) (5) HO(CH 2) 2NH 2 + H 2O + CO 2 HO(CH 2) 2NHCOO - + H 3O + (6) HO(CH 2) 2NHCOO - + H 3O + HO(CH 2) 2NH 2 + H 2O + CO 2 (7) The acid gas package developed by Aspen allows to simulate the removal of acid gases as CO 2 and H 2S. It includes the physicochemical properties of these acid gases, water, amines alone (such as MEA considered in the present case) and also several mixtures (such as amdea), a rate-based calculation model and a makeup unit operation to compensate for losses in water and amine in the system. The thermodynamic models used are e-nrtl for the liquids and the Peng-Robinson equation of state for the gaseous phase. The CASTOR/CESAR pilot unit was selected as case study because all the design and operating parameters are available in literature, which is not the case with most of the other installations. The pilot is sized to handle a flow of 5000 Nm³/h (at the inlet of the treatment line). This flow of 5000 Nm³/h at the inlet of the line results in a flow of 4000 m³/h at the inlet of the absorption column (after removal of a large portion of water, cooling and compression). Among the parameters imposed, it must be pointed out: the CO 2 recovered purity, fixed at 98 mol% (classical value); and the absorption ratio, equal to 90%mol (90% of the molar flow rate of CO 2 entering the absorption column is recovered at the outlet of the regeneration column) The dimensions of the absorber and the stripper, and the operating conditions in each column are also set (see Table 1). Table 1 Dimensions and operating conditions of the columns Absorber Stripper D (m) H (m) 17 (17 x 1 m) 10 (10 x 1 m) Packing used Random packing IMTP 50 Random packing IMTP 50 T up,liquid ( C) P bottom (kpa) The global illustration of the simulations principles is given on Fig. 1. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 6

7 G to treat = 4000 m³/h y CO2 = 20% 90% CO 2 abatement Q CO2 1.5 t CO2/h Brevik Cement plant CASTOR/CESAR pilot CO 2 recovery flow Fig.1. Illustration of the case study (base case) considered for the ASPEN Hysys TM simulations Note that water wash section present in the real pilot for keeping the water balance was not simulated, the makeup unit ensuring the water and MEA balance. With the internal heat exchanger the temperature of the liquid at the inlet of the stripper is fixed at 110 C as base case (corresponding to a pinch of around 10 C at the hot side of the exchanger). In addition, the linear pressure drop per unit height in the absorber and the stripper are fixed to 0.5 kpa/m. The flowsheet developed in Aspen Hysys TM is illustrated on Fig. 2 for the conventional process configuration. Fig. 2. Aspen Hysys TM fow sheet for the conventional process configuration The flue gas leaves the Brevik cement plant at 165 C and 100 kpa. It is compressed to 120 kpa and cooled down to 40 C before entering the flow sheet on Fig. 2. The conditioned gas ( gas to treat, 4000 m³/h, see composition in Table 2) is sent in the absorber where the CO 2 is captured by an amine-based solvent, aqueous MEA 30 wt %. in this study ( lean solution in abs in the flow sheet). The amount of CO 2 absorbed into the column is calculated using the rate-based model. At the outlet of the column, the Rich solution before preheat is pumped and preheated to 110 C thanks to the internal heat exchanger. Then, into the regeneration column, the gas is stripped thanks to the heating power and the CO 2 is recovered at the top of the regeneration column ( produced CO 2 in the flow sheet). The regeneration occurs at 200 kpa and at around 120 C (corresponding to the boiling point of aqueous 30 wt.% MEA at such pressure level). The condenser cooling energy is automatically adjusted in order to satisfy the CO 2 purity specification. The regenerated solvent ( lean solution before cooling ) which still contains some CO 2 is pumped through the heat exchanger in order to be cooled down. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 7

8 Table 2 Composition of the gas to treat (G = 3997 m³/h, 40 C, 120 kpa) after conditioning Component mol frac. N CO H 2O O CO SO NO NO To compensate possible losses in amine and water at the outlet of the absorber and the regenerator, a makeup unit is added in order to automatically adjust the total flow of liquid while reaching the desired concentration of amine. After this make up unit, the lean solution is cooled down to 40 C before entering in the absorption column and beginning a new absorption-regeneration cycle Types of process configuration improvements considered As indicated previously, many process configurations and the technical details associated are given in (Le Moullec et al., 2014). Three categories of process improvements are envisaged, namely: 1. Absorption enhancement: the purpose of this process modifications is to increase the CO 2 loading at the absorber bottom or to reduce excessive driving force in the absorber section. In the present case, the Rich Solvent Recycle (RSR) configuration (see Fig. 3) will be considered. Fig. 3. Aspen Hysys TM fow sheet for the Rich Solvent Recycle (RSR) process configuration The principle of this configuration is to recycle into the absorber a part of the rich solution coming from the bottom of this column. The rich solution going back to the column can also be cooled down in order to promote the CO 2 absorption. In addition to the liquid flow rate (liquid to gas ratio, L/G) that must be optimized for all the configurations, the other parameters that must be specifically adjusted for RSR configuration in order to minimize the energy consumption of the system (especially the solvent regeneration energy) are: the fraction of the rich solution recirculated, the temperature of the solution before re-injection into the column and the level of reinjection into this column. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 8

9 2. Exergetic or heat integration: the general idea of these process modifications is to perform heat integration between the different process streams in order to reduce the heat losses and the solvent regeneration energy. The process modification corresponding to this category and considered in the present case if the Solvent Split Flow (SSF) configuration (see Fig. 4), also called Rich Solvent Splitting. With this configuration, a part of the rich solution coming from the absorber is directly sent at the top of the regeneration column without being preheated by the internal heat exchanger. This arrangement leads to a modification of the temperature profile into the stripper (it is more smoothed than with conventional configuration) and the heat recovered from hot lean solvent is maximized. Furthermore, thanks to the cold solution injected at the top of the stripper, the condenser cooling energy is reduced. Fig. 4. Aspen Hysys TM fow sheet for the Solvent Split Flow (SSF) process configuration The parameters that must specifically optimized for SSF configuration are: the fraction of the cold rich solution by-passing the internal heat exchanger and the injection levels of the solutions (cold rich solution and preheated rich solution) into the stripper. 3. Heat pumps: the idea of this process modification category is to use a heat pump effect in order to increase the heat quality provided to the system. As described by (Le Moullec et al., 2014), this effect enables the valorization of heat available at a too low quality level or, more generally, when increasing the quality level is profitable. Two configurations corresponding to this category are considered in the present study, namely the Lean Vapor Compression (LVC) (see Fig. 5) and the Rich Vapor Compression (RVC) (see Fig. 6). Fig. 5. Aspen Hysys TM fow sheet for the Lean Vapor Compression (LVC) process configuration Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 9

10 Fig. 6. Aspen Hysys TM fow sheet for the Rich Vapor Compression (RVC) process configuration The principle of the LVC configuration is as follows: the lean solvent at the bottom of the stripper is flashed in order to produce a gaseous stream (mainly composed of water and carbon dioxide) which is compressed and fed back to the stripper. Such process modification reduces the reboiler steam demand and cools down the lean solvent going to the internal heat exchanger. Furthermore, the lean solvent exits the flash tank at lower temperature and heats the rich solvent in the heat exchanger also to a lower temperature, making the top of the stripper a bit colder which reduces the cooling requirement in the condenser. Based on operational experiences such as in the CASTOR/CESAR project (Knudsen et al., 2009), this modification is generally accompanied with an expansion or a modification of the internal heat exchanger in order to reduce the hot pinch of this exchanger to 5 C. Indeed, even if injecting the rich solvent into the stripper at a lower temperature is beneficial in terms of condenser cooling energy, it is counterbalanced by an unfavorable larger temperature difference between the inlet liquid temperature and the boiling temperature. Moreover, as the vapor coming from the compressor installed after the flash unit is very hot (which could lead to a hot stop inducing degradation problems into the bottom of the stripper), a supplementary heat exchanger can be installed in order to cool down this vapor (to 120 C) while giving a preheating complement to the rich solution before entering the regeneration column. Regarding the Rich Vapor Compression (RVC) configuration, the principle is quite similar as the LVC one even if it is the hot rich solvent which is flashed instead of the lean solution in order to produce a gaseous stream sent into the bottom of the stripper. As for LVC configuration, an expansion of the internal heat exchanger or a supplementary heat exchanger can be installed in order to cool down the vapor stream to 120 C and to give a supplementary preheating to the rich solution before its regeneration. In addition to the liquid flow rate, the important operating parameter in relation with LVC and RVC configurations is the flash pressure. This pressure depends on the stripper pressure but it has to be noted that a too low pressure (under atmospheric one) is not advised for economic reasons. As a conclusion of this section, it must be noted that four different configurations are considered in the present work, namely RSR, SSF, LVC and RVC, corresponding to the three categories of process modifications. The focus is put to these configurations because it does not imply too much modifications of the conventional process and some of them (such as LVC) have already shown interesting results for the application to power plants which have to be confirmed for the application to cement plants. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 10

11 3. Results and discussion Before analyzing the results, it must be noted that the simulation method was previously validated with the use of CASTOR/CESAR projects results (application to power plant) and also with the use of another study (other pilot design but same modeling method) concerning the application to a cement plant (see (Gervasi et al., 2014) for more details). Concerning the methodology used in this work, the objective was to minimize the solvent regeneration energy (E regen [GJ/t CO2]) defined as: E regen = Ф boiler G CO2,produced (8) where Ф boiler [GJ/h] is the heat duty provided at the bottom of the stripper and G CO2,produced [t CO2/h] the amount of CO 2 generated at the top of this column (outlet of the condenser). Note that no compression of the produced CO 2 is considered in the present case because in accordance with ECRA (European Cement Research Academy) the focus is put on CO 2 valorization options for which the level of CO 2 compression can be different depending on the CO 2 conversion process considered. As some configurations imply the optimization of several operating parameters, systematic parametric studies were carried out in order to identify the realistic conditions leading to the minimum of E regen Base case configuration Optimization of the liquid to gas flow rate ratio for the base case For the absorption-regeneration CO 2 capture process, it is conventional to optimize the (L/G) vol. ratio in order to minimize the regeneration energy. The simulated results for different (L/G) vol. ratio (which means different liquid flow rates due to the fact the gaseous flow rate was kept constant) are presented on Fig.7. Fig. 7. Regeneration energy as a function of the (L/G)vol. ratio First of all, it must be observed that the parabolic tendency of this graph is quite conventional for such process. The minimum regeneration energy was identified for a liquid flow rate of 22 m³/h ((L/G) vol. = ) leading to E regen = 3.36 GJ/t CO2. This value is in the range of conventional values measured for power plants (between 3 and 4 GJ/t CO2, see for example (Knudsen et al., 2009)). Nevertheless, it must be pointed out that 3.36 GJ/t CO2 is close to the minimum values conventionally measured for MEA 30 wt.% which could be justified by the fact that a higher CO 2 content in the gas to treat (20 vol.% for the cement plant considered in the present case in comparison with the range 5-15 vol.% for power plants) is favorable to the absorption process. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 11

12 Global results analysis for the base case The detailed results corresponding to the base case optimal operating conditions are presented in Fig. 8., Table 3 and Table 4. Fig. 8. Temperature profiles into the absorber and the stripper Regarding the temperature profiles into the absorber and stripper presented on Fig. 8., first of all it must be specified that the temperature mentioned for stage 17 of the absorber corresponds to the temperature of the inlet liquid (40 C) and that the temperature for stage 0 of the stripper corresponds to the boiler temperature (121.8 C). The profiles are quite conventional for such operation units even if the maximum temperature reached into the absorber (around 85 C) is a little bit higher than other values (75-80 C) generally measured. This could be linked to the higher CO 2 content of the gas to treat (20 mol.% for the cement plant considered) in comparison with power plants (5-15 mol.%) which induces a higher heat of reaction. Table 3 Simulation results for the base case ((L/G)vol. = ) Parameter E regen E condenser E pumps α CO2,rich α CO2,lean Value 3.36 GJ/t CO GJ/t CO GJ/t CO mol CO 2/mol MEA mol CO 2/mol MEA Concerning the other parameters indicated in Table 3, in addition to the regeneration energy (namely 3.36 GJ/t CO2) commented in previous section, it can be seen that even if it is not an issue for the absorption-regeneration process, the condenser cooling energy (E condenser) is significant (-1.90 GJ/t CO2) and reducing it thanks to the use of alternative configurations would be also benefic in practice (reduction of the water flow rate circulating into the condenser). Regarding the consumption of the liquid pumps (E pumps) equal to GJ/t CO2, it corresponds to only 0.5 % of the regeneration energy and is thus not significant for the evaluation of the overall energy consumption of the process. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 12

13 Moreover, the CO 2 loadings value for the rich (α CO2,rich) and the lean (α CO2,lean) solutions are equal respectively to and mol CO 2/mol MEA, which is conventional for MEA 30 wt.% even if it must be noted that the value slightly higher than 0.5 is possible thanks to the CO 2 content of the gas to treat ( 20 mol.%). Table 4 presents the gaseous compositions of the gas to treat, the gas treated and the produced CO 2. Table 4 Gaseous compositions for the base case ((L/G)vol. = ) in mol. fraction Component Gas to treat Gas treated Produced CO 2 N CO H 2O O CO SO NO NO MEA As no specific reactions were added concerning the other gaseous species (SO 2, NO, NO 2, etc.), the decrease of their concentrations into the absorber can only be associated to the modification of their solubility into the liquid phase. Regarding the MEA, only very small quantities are present into the treated gas and the produced CO 2. These results confirm the absorption ratio of 90% fixed as simulation parameter, and also the fact that the produced CO 2 contains 98 mol.% of CO 2, the rest being mainly composed of water. Finally, it has to be highlighted that the presented results were obtained for a rich solution injected at stage 9 into the stripper. Indeed, as presented on Fig. 9, injecting at level 9 or level 10 (top of the column) gives similar results in terms of regeneration energy while the condenser cooling energy is a little bit lower when the rich solution is not injected too close to the condenser, namely at stage 9. Fig. 9. Eregen and Econdenser as a function of the injection of the rich solution into the stripper Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 13

14 3.2. Rich Solvent Recycle (RSR) configuration As indicated in the previous section, in addition to the liquid flow rate, the other parameters that must be optimized for the RSR configuration are the fraction of the rich solution recirculated, the temperature of the solution before re-injection into the column and the level of re-injection into this column. In a first step, for a fixed re-injected amount of 15% (without any cooling), different re-injection levels were considered between 17 (top of the column) and 1 (bottom of the column) and for each reinjection level, the (L/G) vol. ratio was varied for a liquid flow rate in the range 20 to 30 m³/h (the gas flow rate being kept constant at 4000 m³/h). The results are presented in Fig. 10. Fig. 10. Eregen as a function of the re-injection level of the rich solution into the absorber for different (L/G)vol. values (15% re-injection and no cooling) RSR configuration It could be highlighted that the optimum liquid flow rate is situated in the range between 23 and 25 m³/h (around for the (L/G) vol.) and that the regeneration energy is increasing if the rich solution is recirculated at a level higher than stage 6. Considering a liquid flow rate of 24 m³/h, 15% of solution reinjection into the stripper at stage 4 (intermediate level between stage 1 and 6 in the optimal zone), Fig. 11 illustrates the effect of cooling the re-injected solution on the CO 2 loading of the rich solution and on the regeneration energy. Fig. 11. αco2,rich and Eregen as a function of the re-injected solution temperature into the absorber (15% re-injection at stage 4) RSR configuration As expected by an absorption enhancement process modification as RSR, the recirculation into the absorber of a cooled solution allows to increase the CO 2 loading of the rich solution (from 0.50 to 0.51 mol CO 2/mol MEA in the presented case when the solution is cooled from 50 C to 5 C) which leads to a decrease of the regeneration energy from 3.3 to 3.12 GJ/t CO2. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 14

15 Nevertheless, cooling the solution to a too low level would be unfavorable in terms of cooling energy. In practice, considering a cooling of the solution from 60 C (temperature of the solution at the outlet of the absorber) to 40 C (same temperature of the solution at the top of the absorber) seems more feasible and will be considered as reference for the results comparison. Finally, the parameter that must be optimized is the fraction of the rich solution which is cooled and reinjected into the column (fixed at 15% for the previous results). Fig. 12 illustrates the regeneration energy calculated by the simulation as a function of the fraction of the rich solution which is recirculated (reinjected at stage 4 and cooled down to 40 C). Fig. 12. Eregen as a function of the percentage of re-injected solution into the absorber (re-injection at stage 4 and cooled down to 30 C) RSR configuration It can be seen that the minimum of the regeneration energy is obtained for a fraction of the re-injected solution corresponding to 35% and leading to E regen equal to 3.07 GJ/t CO2 which means an energy saving of 8.54% in comparison with the base case. The other results corresponding to these operating parameters are given in Table 5. Table 5 Simulation results for the RSR configuration ((L/G)vol. = , re-injection of 35% of rich solution at stage 4 and 40 C) Parameter E regen E condenser E pumps α CO2,rich α CO2,lean Value 3.07 GJ/t CO GJ/t CO GJ/t CO mol CO 2/mol MEA mol CO 2/mol MEA Concerning the other simulation results, it can be seen from Table 5 that the RSR configuration leads to a slightly higher values of the CO 2 loadings of the rich and lean solutions in comparison with the base case configuration which certainly induces a lower regeneration energy. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 15

16 3.3. Solvent Split Flow (SSF) configuration Three operating parameters must be optimized with the Solvent Split Flow (SSF) configuration, namely the (L/G) vol. ratio (as for all the configurations), the fraction of the rich solution ( split fraction ) which is injected without being preheated ( cold solution ) and the re-injection level of the preheated solution into the regenerator (injected at stage 8 in a first step). Regarding the cold solution, it is being conventionally injected at the top of the stripper (stage 10 in a first step) in order to reduce the condenser cooling energy. In practice, for a defined heat exchanger, splitting the rich solution will change the liquid flow rate inside this exchanger leading to a modification of the temperature of the rich solution at its outlet (T rich,preheat), equal to 110 C in the base case. As used for example in (Sanchez Fernandez et al., 2012) for another process configuration, the internal heat exchanger can be simulated using the product UA [kj/ C h] of the heat transfer coefficient (U [kj/h m² C]) and area (A [m²]) corresponding to the base case. Based on the same principle, using the UA parameter coming from the base case configuration (without SSF), simulations were performed for different liquid flow rate at the inlet of the internal heat exchanger in order to establish the relation between this flow rate and the temperature at the outlet of the internal heat exchanger (T rich,preheat). It was shown that for the maximum split fraction considered (0.4), T rich,preheat reaches 113 C, which corresponds to an increase of 3 C in comparison with the base case. Using this method, Fig. 13 presents the simulation results (E regen) obtained with the SSF configuration for different split fractions of the rich solution and for different liquid flow rates. Fig. 13. Eregen as a function of the split fraction for different liquid flow rates (in the stripper: cold solution injected at stage 10, hot solution injected at stage 8) SSF configuration It can be seen that the optimum liquid flow rate (minimizing E regen) is situated in the range m³/h and that the optimum split fraction value is between 0.15 and Considering the optimum split fraction for each liquid flow rate, Fig. 14 illustrates the relation between E regen and the (L/G) vol. ratio. This graph highlights that the optimum (L/G)vol. ratio minimizing E regen is (which corresponds to a liquid flow rate of 23 m³/h and which was obtained for a split fraction fixed at 0.25). Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 16

17 Fig. 14. Eregen as a function of the (L/G)vol. ratio considering the optimum split fraction for each liquid flow rate value (in the stripper: cold solution injected at stage 10, hot solution injected at stage 8) SSF configuration Considering the optimum liquid flow rate of 23 m³/h, Fig. 15 illustrates more precisely the influence of the split fraction on the regeneration energy. It is shown that the split fraction minimizing the regeneration energy is 0.26 even if it must be noted that in the range 0.22 to 0.30 E regen does not vary a lot. Fig. 15. Eregen as a function of the the optimum split fraction for L = 23 m³/h (in the stripper: cold solution injected at stage 10, hot solution injected at stage 8) SSF configuration Finally, for the optimum liquid flow rate (23 m³/h) and split fraction (0.26), a last sensitivity study presented on Fig. 16 shows the influence of the injection stage of the hot and cold rich solutions into the stripper. As expected, the better results are obtained when the cold solution is injected at the top of the stripper (stage 10). Regarding the hot solution, it seems that the better situation is to keep 2 or 3 stages between the cold and the rich solutions, namely injecting the hot solution at stage 7 or 8, E regen being minimized for an injection at stage 7. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 17

18 Fig. 16. Eregen as a function of the injection stage of the hot solution into the stripper for different injection stages of the cold solution with L = 23 m³/h and the split fraction of the rich solution equal to 0.26 SSF configuration The simulation results corresponding to the optimum case are summarized in Table 6. Table 6 Simulation results for the SSF configuration ((L/G)vol. = , split fraction of 26% of rich solution, hot solution injection stage = 7, cold fraction injection stage = 10) Parameter E regen E condenser E pumps α CO2,rich α CO2,lean Value 3.22 GJ/t CO GJ/t CO GJ/t CO mol CO 2/mol MEA mol CO 2/mol MEA The regeneration energy was minimum for a liquid flow rate of 23 m³/h ((L/G) vol. = ) and a split fraction of 26%, leading to a regeneration energy of 3.22 GJ/t CO2, which means 4.2% energy savings in comparison with the base case. Concerning the other parameters, the CO 2 loading of the rich and lean solutions, and the energy for pumping are quite similar to the base case. Regarding the condenser cooling energy, it decreases by around 45% in comparison with the base case. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 18

19 3.4. Lean Vapor Compression (LVC) configuration In addition to the (L/G) vol. ratio (as for all the other configurations, especially the liquid flow rate, the gaseous flow rate being fixed), the important operating parameter regarding the LVC configuration is the flash pressure (flash pressure drop ( p) from 0 to 100 kpa). It must be also reminded that this process modification is generally accompanied with an expansion (or a modification) of the internal heat exchanger in order to reduce the hot pinch of this exchanger to 5 C and that another heat exchanger cools down the LVC flash vapor to 120 C in order to avoid any hot spot leading to degradation. Fig. 17 illustrates the simulation results in terms of E regen as a function of the LVC flash pressure and for different liquid flow rates. It can be noted from this graph that despite the decrease of the temperature of the rich solution at the outlet of the heat exchanger (see Fig. 18), a high p value is preferred for reducing the solvent regeneration energy. p equal to 100 kpa will be thus considered as the optimum condition for LVC. Fig. 17. Eregen as a function of the LVC flash pressure drop ( p) for different liquid flow rate LVC configuration Fig. 18. Temperature of the rich solution at the outlet of the internal heat exchanger as a function of the LVC flash pressure drop ( p) for a liquid flow rate fixed at 21 m³/h LVC configuration As evocated previously, Fig. 18 highlights that a higher LVC flash pressure drop leads to a decrease of the temperature of the rich solution at the outlet of the internal heat exchanger. Note that after the internal heat exchanger, the other heat exchanger gives a complement preheating to the rich solution (from 97.3 C to 99.3 C at 100 kpa for L = 21 m³/h) thanks to its heat exchange with the hot vapor coming from the LVC unit. Regarding the optimum value for the (L/G) vol. ratio, Fig. 19 shows that the regeneration energy is minimized at (L/G)vol. equal to , which corresponds to a liquid flow rate of 21 m³/h. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 19

20 Fig. 19. Regeneration energy as function of the (L/G)vol. ratio for a p of 100 kpa LVC configuration It can be seen that the minimum of the regeneration energy is 2.91 GJ/t CO2 which means an energy saving of 13.4% in comparison with the base case. The other parameters corresponding to the optimized case are given in Table 7. Table 7 Simulation results for the LVC configuration ((L/G)vol. = and p = 100 kpa) Parameter E regen E condenser E pumps E LVC-compressor α CO2,rich α CO2,lean Value 2.91 GJ/t CO GJ/t CO GJ/t CO GJ/t CO mol CO 2/mol MEA mol CO 2/mol MEA Concerning the other simulation results, it can be seen from Table 7 that the LVC configuration leads to a clearly lower condenser cooling energy in comparison with the base case (around the half) which is linked to the lower temperature at the inlet of the stripper (<100 C). Regarding the energy consumption of the pumps and LVC-compressor, even if not negligible it is clearly lower than E regen. The CO 2 loadings of the rich and lean solutions have the same order of magnitude as for the base case even if it must be noted that α CO2,lean is a little bit lower than for the base configuration due to the fact that the optimum value of the liquid flow rate is also lower for a same absorption ratio (90%). Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 20

21 3.5. Rich Vapor Compression (RVC) configuration As for the LVC configuration, in addition to the (L/G) vol. ratio, the important operating parameter regarding the RVC configuration is the flash pressure (flash pressure drop ( p) from 0 to 100 kpa). A supplementary heat exchanger also cools down the RVC flash vapor to 120 C in order to avoid any hot spot in the stripper leading to degradation. Fig. 20 illustrates the simulation results in terms of E regen as a function of the RVC flash pressure and for different liquid flow rates. Fig. 20. Eregen as a function of the RVC flash pressure drop ( p) for different liquid flow rate RVC configuration It can be noted from this graph that as for LVC configuration, a high p value is preferred for reducing the solvent regeneration energy. p equals to 100 kpa will be thus considered as the optimum condition for RVC. Moreover, it can be seen that E regen is sharply decreasing when the liquid flow rate is increased from 20 to 26 m³/h and then slightly decreasing in the range from 26 to 30 m³/h. Regarding the optimum value for the (L/G) vol. ratio (and thus for the liquid flow rate), Fig. 21 confirms that the regeneration energy reached a minimum once the (L/G) vol. ratio equals , which corresponds to a liquid flow rate of 29 m³/h. Fig. 21. Regeneration energy as function of the (L/G)vol. ratio for a p of 100 kpa RVC configuration As a very slight increase of E regen is observed for higher liquid flow rates, 29 m³/h ((L/G) vol. of ) was considered as the optimum flow rate. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 21

22 It can be seen that the minimum of the regeneration energy is 2.95 GJ/t CO2 which means an energy saving of 12.1% in comparison with the base case. The other results corresponding to the optimized case are given in Table 8. Table 8 Simulation results for the RVC configuration ((L/G)vol. = and p = 100 kpa) Parameter E regen E condenser E pumps E RVC-compressor α CO2,rich α CO2,lean Value 2.95 GJ/t CO GJ/t CO GJ/t CO GJ/t CO mol CO 2/mol MEA mol CO 2/mol MEA Concerning the other simulation results for the RVC configuration, it can be seen from Table 8 that the RVC configuration leads to lower condenser cooling energy in comparison with the base case (around 20% decrease) but not as low as for the LVC configuration. Regarding the energy consumption of the pumps and RVC-compressor, it is the same order of magnitude as for the LVC even if the compressor energy consumption is higher than for the LVC. Finally, it must be noted that the CO 2 loadings of the rich and lean solutions are respectively lower and higher in comparison with the other configurations (lower α CO2 which corresponds to α CO2,rich α CO2,lean) due to the fact that the optimum value of the liquid flow rate is clearly higher for a same absorption ratio (90%) Global comparison of the energy savings thanks to different process configurations A summary of the simulations results corresponding to the optimal operating conditions for each configurations is provided in Table 9 and in Fig. 22. Table 9 Summary of the simulation results of the different configurations Base case RSR SSF LVC RVC Operating conditions (L/G) vol,opt (m³/m³) Split fraction rich sol. (%) Re-injection sol. temp. ( C) Re-injection abs. stage (N ) Hot sol. stripper stage (N ) Cold sol. stripper stage (N ) Flash pressure drop (kpa) α CO2,rich (mol/mol) α CO2,lean (mol/mol) Energy consumptions E pump (GJ/t CO2) E condenser (GJ/t CO2) E LVC/RVC,compressor (GJ/t CO2) E regen (GJ/t CO2) E regen savings /Base case (%) Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 22

23 Fig. 22. Optimum regeneration energy for each process configurations (left) and energy savings linked to the use of alternative configurations in comparison with the base case (right) Among the different process configurations simulated with MEA 30 wt.% as solvent, it was shown that the heat pump modification (namely LVC and RVC process modifications) gives the best regeneration energy savings (around 14%) leading to E regen lower than 3 GJ/t CO2. Regarding the two other categories, namely absorption enhancement (RSR) and the exergetic integration (SSF) smaller energy savings were measured (between 4 and 8 %) which can be justified by the fact that even with the conventional process configuration, the rich CO 2 loading of the MEA 30% is close to its equilibrium value, thus the potential for increasing this loading is very limited (which is normally the purpose of RSR configuration for example). Concerning the SSF one, the major advantage is linked to the decrease of the condenser cooling energy (which is also clearly decreased thanks to the LVC configuration). Finally, even if not negligible, the pumping and LVC/RVC compressor energy consumptions have an order of magnitude clearly lower than the regeneration energy. 4. Conclusions and perspectives In the context of reducing the CO 2 emissions from the cement industry, and in order to reduce the CO 2 capture costs specifically for this industry, the present study focused on the Aspen Hysys TM simulation of different configurations of the absorption-regeneration CO 2 capture process using amine based solvents (MEA 30 wt.% in the present case) and applied to the flue gas coming from a cement plant: the Norcem Brevik Cement plant in Norway was taken as reference. The design of the CO 2 capture plant considered for the simulation was based on the CASTOR/CESAR European Projects pilot. Four process modifications were investigated, namely RSR, SSF, LVC and RVC, in order to be representative of the three categories of process modifications: absorption enhancement, exergetic or heat integration and heat pump effect. For each configuration, a systematic parametric study on operating parameters ((L/G) vol. ratio, split ratios, flash pressures, etc.) was carried out in order to identify the conditions minimizing the regeneration energy (taken as comparison factor for each configuration). Based on the simulations carried out, it was highlighted that the heat pump modifications LVC and RVC lead to the higher energy savings (around 12 and 13% respectively) while also reducing significantly the condenser cooling energy. The energy savings linked to RSR and SSF modifications were lower (namely 8.5 and 4%). As perspectives, other solvents will have to be considered for future simulations, such as piperazine (PZ) alone or used as absorption activator, because the energy savings linked to the use of alternative configurations strongly depend on the solvent properties. Other configurations will be also investigated such as the combination of two configurations (for example RSR and RVC). Finally, in addition to the interest in terms of OPEX, the consequence in terms of CAPEX will have to be estimated in order to be able to evaluate more precisely the global economic interest of using alternative process configurations. Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 23

24 Acknowledgments The authors would like to acknowledge the European Cement Research Academy (ECRA) and HeidelbergCement Company for their technical and financial support accorded to the ECRA Academic Chair at the University of Mons. References International Energy Agency (IEA), Technology Roadmap Carbon Capture and Storage 2013 Edition, Idem R., Supap T., Shi H., Gelowitz D., Ball M., Campbell C., Tontiwachwuthikul P., Practical experience in postcombustion CO2 capture using reactive solvents in large pilot and demonstration plants, Int. J. Greenhouse Gas Control, Bjerge L.-M. and Brevik P., CO2 Capture in the Cement Industry, Norcem CO2 Capture Project (Norway), Energy Procedia 63, , Singh P., Van Swaaij W.P.M., Brilman D.W.F., Energy Efficient Solvents for CO2 Absorption from Flue Gas: Vapor Liquid Equilibrium and Pilot Plant Study, Energy Procedia, 37, , Zhao X., Smith K.H., Simioni M.A., Tao W., Kentish S.E., Fei W., Stevens G.W., Comparison of several packings for CO2 chemical absorption in a packed column, Int. J. Greenhouse Gas Control, 5, , Le Moullec Y., Neveux T., Al Azki A., Chikukwa A., Hoff K.A., Process modifications for solvent-based post-combustion CO2 capture, Int. J. Greenhouse Gas Control, 31, , Knudsen J.N., Jensen J.N., Vilhelmsen P-J., Biede O., Experience with CO2 capture from coal flue gas in pilot-scale: Testing of different amine solvents, Energy Procedia, 1, , Gervasi J., Dubois, L. and Thomas D., Simulation of the Post-combustion CO2 Capture with Aspen Hysys TM Software: Study of Different Configurations of an Absorption-regeneration Process for the Application to Cement Flue Gases, Energy Procedia, 63, , Sanchez Fernandez E., Bergsma E.J., de Miguel Mercader F., Goetheer E.L.V, Vlugt T.J.H, Optimisation of lean vapour compression (LVC) as an option for post-combustion CO2 capture: Net present value maximization, Int. J. Greenhouse Gas Control, 11S, S114 S121, Dr L. Dubois ECRA Academic Chair Faculty of Engineering UMONS Belgium 24