CET Volume 86 DOI: 10.3303/CET2186180 Paper Received: 8 September 2020; Revised: 26 January 2021; Accepted: 27 April 2021 Please cite this article as: Kratky L., Kolacny J., Sulc R., 2021, Experimental Study on CO2 Membrane Separation from Flue Gas , Chemical Engineering Transactions, 86, 1075-1080 DOI:10.3303/CET2186180 CHEMICAL ENGINEERING TRANSACTIONS VOL. 86, 2021 A publication of The Italian Association of Chemical Engineering Online at www.cetjournal.it Guest Editors: Sauro Pierucci, Jiří Jaromír Klemeš Copyright © 2021, AIDIC Servizi S.r.l. ISBN 978-88-95608-84-6; ISSN 2283-9216 Experimental Study on CO2 Membrane Separation from Flue Gas Lukas Kratky*, Josef Kolacny, Radek Sulc Czech Technical University in Prague, Faculty of Mechanical Engineering, Department of Process Engineering, Technicka 4, 160 00, Prague 6, Czech Republic Lukas.Kratky@fs.cvut.cz The paper scoped to evaluate the process characteristics of polyetherimide-polyimide hollow fibre membrane module to separate CO2 of the model flue gas. The model gas was composed of 15.2 vol % CO2 + 4.5 vol % O2 + 80.3 vol % N2, as referenced to the typical composition of flue gas emitted by an LPG-fired power plant. The effect of process variables on process characteristics was researched for various differential pressures between permeate 0.2 - 0.5 MPa (a) and retentate 0.5 - 1.0 MPa (a), feed flowrates 100 g h-1 and 200 g h-1, all under the gas temperature of 30 °C. It was found out that the value of the separation factor (CO2/N2) was 6.9. The decrease of the separation factor (CO2/N2) by 4.1 compared to the ideal selectivity (CO2/N2). The separation factor (CO2/O2) was shallow; its maximum value of 1.8 was reached. 1. Introduction CO2 separation from flue gas by membrane process is nowadays viewed as a young and promising method with a very high commercial potential in Carbon Capture and Utilization technologies. As highlighted by Song et al. (2019), it is given by its simplicity in technical set-up, compactness, modularity, its simplicity in operation, control and maintenance, in its energy efficiency and scale-up flexibility. Gas membrane separation also evinces safety and environmental friendless because there is no requirement to use hazardous chemicals, e.g. as demanded by conventional CO2 separation by absorption (Boulmal et al., 2017). Generally known, gas membrane separation is a pressure-driven process that uses different diffusivities of individual gas components in a membrane cartridge to separate them. One or multi-stage configurations of laboratory or pilot gas membrane separation systems under TRL 3-6 are typically used to remove a given gas component from a feed. The configuration itself is primarily affected by flue gas composition, membrane characteristics (selectivity, permeability), and demanded permeate purity. Nevertheless, flue gas membrane separation has several limitations: low-temperature resistance up to 100°C, sensitivity to acid gases (SOX, NOX and H2S), varying long-term performance, and low selectivity and resilience (Song et al., 2019). Moreover, Xu et al. (2018) discussed that it is impossible to reach CO2 purity higher than 95 % with low energy demand and capture cost. Gas membrane separation is still under intensive development in laboratory or pilot plant scales aiming to find suitable membrane material, ambient operating conditions and low energy demand under favourable capture cost. Scientific activities historically dominantly scoped to test the efficiency of CO2 removal from binary flue gas mixtures. Liu et al. (2020) modelled flue gas as binary mixture CO2 + N2 and used CHA-type membranes to remove CO2 from the model gas at ambient temperatures. Torstensen et al. (2019) studied CO2 removal from model flue gas as a mixture of 10 vol % CO2 and 90 vol % N2 in feed using PVA/nanocellulose nanocomposite membrane. Lee et al. (2015) tested a commercial hollow fibre module's separation efficiency to remove greenhouse gas impurities from industrial flue gas composed of 86.4 vol % N2 + 13.4 vol % O2 + 951 ppm SF6 + 72.2 ppm CF4 + 1198 ppm of CO2. Choi et al. (2013) modelled LNG-fired flue gas as 10.8 vol % CO2 + 89.2 vol % N2 to experimentally develop multi-stage membrane process using PES hollow fibre membrane module. Ayadani et al. (2021) reported the potential of CO2 separation from binary mixtures of CH4, N2 and H2 using an SZZ-13 zeolite membrane. Nevertheless, there is also an effort to test the efficiency of CO2 removal multicomponent gas mixtures. E.g. Liang et al. (2017) tested separation characteristics of 1075 PDMS/PAN composite membranes for the synthetic ternary gas mixture composed of 15 vol % CO2 + 5 vol % O2 + 80 vol % N2. Bruntetti et al. (2020) investigated the use of poly-decyl methylsiloxane composite membrane to separate CO2 from the model flue gas with composition 15 vol % CO2 + 5 vol % O2 + 80 vol % N2. This information follows that gas membrane separation processes nowadays primarily identify process characteristics of gas membrane separation for a given component from binary gas mixtures. There is little information about gas membrane separation from multicomponent gas mixtures like industrial flue gases. Thus, the paper deals to analyze and model separation fundamental process characteristics of membrane CO2 removal from the model flue gas for a brand new developed membrane module in dependence on process parameters of the gas membrane separation process. 2. Methods 2.1 Experimental set-up The laboratory membrane unit RALEX GSU-LAB-200 was used to carry out all the experiments. Its experimental set-up is plotted in Figure 1. The tertiary gas mixture, the modelling flue gas produced by the LPG-fired power plant, was tested. The feed composition was set-up as 15.2 vol % CO2 + 4.5 vol % O2 + 80.3 vol % N2. Two flow rates of the gas mixture being 100 g h-1 and 200 g h-1 were used in experiments under one mean process gas temperature of 30 °C. The mixture was fed into the membrane module with polyetherimide- polyimide hollow fibres (Membrain Ltd, CZ). The internal diameter of fibre was 0.188 mm, the wall thickness of fibre was 0.06 mm, and the active mass transfer surface 0.8 m2. The effect of process variables on process characteristics was researched for various differential pressures between retentate and permeate set according to working pressure ranges for permeate 0.2-0.5 MPa(a) and retentate 0.5-1.0 MPa(a). The mass gas flow rate was measured in feed and retentate streams. The volumetric concentrations of individual gas components in feed, permeate, and retentate streams were continuously analyzed by gas analyzers Emerson at steady-state conditions of the gas membrane separation process. Figure 1: Technological scheme of the membrane unit. 2.2 Data analysis The data analysis and evaluation supposed the ideal gas behaviour. The data evaluation was based on the application of the law of mass conservation, general definitions of the amount of substance, molar mass for individual components and gas mixtures. Based on this data, CO2 stage cut θCO2 (-), permeate flux JCO2 (mol m-2 s-1), CO2 permeability PCO2 (barrer), CO2 permeance PCO2/l (GPU) and separation factor SF (-) were calculated as: 𝜃𝐶𝑂2 = 𝑚𝐶𝑂2 𝑃 𝑚𝐶𝑂2 𝐹 (1) 𝐽𝐶𝑂2 𝑃 = 𝑛𝐶𝑂2 𝑃 𝐴 (2) 𝑃𝐶𝑂2 𝑙 = 𝐽𝐶𝑂2 𝑃 𝑝𝐶𝑂2 𝑃 ∙ 3.35 ∙ 10−10 (3) 1076 𝑆𝐹 (𝐶𝑂2 𝑖⁄ ) = 𝑐𝐶𝑂2 𝑛𝑃 𝑐𝑖 𝑛𝑃 𝑐𝐶𝑂2 𝑛𝐹 𝑐𝑖 𝑛𝐹 (4) where A (m2) represents the active mass transfer surface of the membrane module, l (mm) is the wall thickness of membrane fibre, 𝑝𝐶𝑂2 𝑃 (Pa) defines partial pressure of CO2 in permeate stream, and i (-) means a gas component, CO2 excluded. 3. Results and Discussion The experimental analysis of process characteristics for the tested module was done both for single gases and the model flue gas. Figure 2: CO2 and N2 permeances in dependence on the partial pressure difference between feed and permeate (the lines represent only the trend). The use of single gases, i.e. nitrogen and carbon dioxide, was applied to characterize the permeability of the membrane cartridge for the individual gases in dependence on partial pressure difference of given component between feed and permeate stream. The results are plotted in Figure 2. The CO2 flowrate from feed to permeate stream was firmly linearly proportional to the given component's partial pressure difference between feed and permeate stream. Similar trends were reported by Choi et al. (2013). It was determined that the maximum CO2 permeance of 32.09 GPU was an order of magnitude higher than the maximum N2 permeance being 1.15 GPU. Based on these values, the ideal selectivity of the given module was calculated as: 𝛼𝐶𝑂2 𝑁2⁄ = 𝑃𝐶𝑂2 𝑙 𝑃𝑁2 𝑙 = 32.09 1.15 = 28 (5) The ideal selectivity 𝛼𝐶𝑂2 𝑁2⁄ is equal to 43 GPU for polyimide and to 20 GPU for polyetherimide (Šípek, 2014). From this follows that the ideal selectivity 𝛼𝐶𝑂2 𝑁2⁄ of the tested module being 28 GPU is realistic regarding the given selectivities of its material composition and its process behaviour are closer to the polyetherimide 1077 matrix. Finally, the comparison of its parameters with other membrane modules is plotted in Figure 3, being at the upper bound as defined by Robeson (2008). Figure 3: The implementation of the experimental results to the Robeson diagram (Robeson, 2008). Figure 4: CO2 concentration in permeate in dependence on stage cut θCO2 and feed flowrate mF (the lines represent only the trend) at constant permeate pressure of 2 bar(a). The dependence of CO2 concentration on stage cut is expressed in Figure 4. CO2 concentration in permeate is increasing with the increase in stage cut and feed flowrate. Nevertheless, the maximum concentration of 38 % vol. of CO2 was reached for the stage cut is higher than 0.4 at a feed flow rate of 200 g h-1. This effect was caused by mass transfer limitations given by process parameters and available mass transfer surface. E.g. Liu et al. (2020) regarded the same trends due to the influence of concentration polarization. Figure 5 depicts the dependence of individual separation factors on the permeability of CO2. It is evident that the separation factor decreases with an increase in CO2 permeability and feed flowrate. The maximum separation factor SF(CO2/N2) of 6.9 was reached at CO2 permeability of 478 barrer. Nevertheless, the separation factor SF(CO2/O2) evinces shallow values. Its maximum value of 1.8 was reached at CO2 permeability of 478 barrer. The presented experimental data are following the findings served by Brunetti et al. (2020). The team also reported moderate selectivities CO2/N2 = 8.7 at permeability close to 1000 barrer for CO2 removal from model flue gas using poly- 1078 decyl methylsiloxane composite membrane. Regarding the presented experimental data, the authors finally stated that the tested module is efficient to separate CO2 from N2, but there is very low selectivity for CO2 to O2 mixture. Figure 5: Separation factor SF (CO2/O2) and SF (CO2/N2) in dependence of CO2 permeability and feed flowrate mF(the lines represent only the trend). The summarization of the most important experimental results is overviewed in Table 1. The separation of the gas mixture achieves significantly worse parameters than in testing, which would take place only on pure gases. Experimentally identified CO2 permeance for model flue gas is more than two times lower than the permeance of single pure gas. The same behaviour is also evident for selectivity. The achieved separation factor is approximately four times smaller than the achieved ideal selectivity. It is evident that there is a mutual interaction among individual gas components resulting in lower permeances and separation factors in the gas mixture than single gases. E.g. Liang et al. (2017) reported the mutual effect of individual gas components on separation factor and presented separation factor CO2/N2 = 11.1. Choi et al. (2013) found that the gas component from the gas mixture was lower than the pure component's ideal selectivity. Thus, it is a practical example of the interaction of components as they pass through the membrane, as pointed out by Robeson (2008), Choi et al. (2013) or He and Wang (2018). 1079 Table 1: The overview of characteristics for the given membrane module at process temperature 30 °C. flue gas single gas mF = 100 g h-1 mF = 200 g h-1 membrane module pure polyetherimid* pure polyimide* permeance CO2 (GPU) 13.1 13.8 32.1 52.5 735.0 permeance N2 (GPU) 2.1 2.8 1.2 not defined not defined permeance O2 (GPU) 5.3 5.9 not analyzed not defined not defined ideal selectivity CO2/N2 (-) not defined not defined 28.0 20.0 43.0 separation factor CO2/N2 (-) 6.6 6.9 not defined not defined not defined separation factor CO2/O2 (-) 1.5 1.8 not defined not defined not defined * referenced by Šípek (2014) 4. Conclusions The experimental identification of process characteristics was carried out for the given membrane module and model tertiary gas mixture that corresponded with its composition to the LPG-fired power plant. It was found that the separation factor may not reach the ideal selectivity. When separating the modelled flue gases, the value of the separation factor CO2/N2 was almost 7. The decrease of the separation factor CO2/N2 compared to the ideal selectivity (CO2/N2 is approximately four times. The separation factor CO2/O2 is deficient; its value is between 1 and 2. The CO2 permeation of the module decreased two to three times in the case of the separation of the gas mixture compared to the separation of pure carbon dioxide. Acknowledgements This work was supported by Ministry of Education, Youth and Sports of the Czech Republic under OP RDE grant number CZ.02.1.01/0.0/0.0/16_019/0000753 "Research centre for low-carbon energy technologies". References Ayadani A., Brunetti A., Maghsoudi H., Barbieri G., 2021, CO2 separation from binary mixtures of CH4, N2 and H2 by using SSZ-13 zeolite membrane, Separation and Purification Technology, 256, 117796-117804. Boulmal N., Rivera-Tinoco R., Bouallou C., 2017, Lipophilic amine aqueous solutions for CO2 capture: comparison of absorption rates for hexylamine (hxa), dimethylcyclohexylamine (dmca) and hxa-dmca mixtures, Chemical Engineering Transactions, 61, 1201-1207. Brunetti A., Zito P.F., Borisov I., Grushevenko E., Volkov V., Volkov A., Barbieri G., 2020, CO2 separation from humidified ternary gas mixtures using a polydeclymethylsiloxane composite membrane, Fuel Processing Technology, 210, 106550-106559. Choi S.H., Kim J.H., Lee Y., 2013, Pilot-scale multi-stage membrane process for the separation of CO2 from LNG-fired flue gas, Separation and Purification Technology, 110, 170-180. He Z., Wang K., 2018, The ideal selectivity vs true selectivity for permeation of gas mixture in nanoporous membranes, IOP Conference Series: Materials Science and Engineering, 323, 1757-8981. Lee S., Choi J.W., Lee S.H., 2015, separation of greenhouse gases (SF6, CF4 and CO2) in an industrial flue gas using pilot scale membrane, Separation and Purification Technology, 148, 15-24. Liang C.Z., Yong W.F., Chung T.S., 2017, High-performance composite hollow fibre membrane for flue gas and air separations, Journal of Membrane Science, 541, 367-377. Liu B., Tang C., Li X., Wang B., Zhou R., 2020, High-performance SAPO-34 membranes for CO2 separations from simulated flue gas, Microporous and Mesorporous Materials, 292, 109712-109722. Robeson L.M., 2008, The upper bound revisited, Journal of Membrane Science, 320, 390-400. Šípek M. (Ed), 2014, Membránové dělení plynů a par. University of Chemistry and Technology in Prague, CZ. Song Ch., Liu Q., Deng S., Li H., Kitamura Y., 2019, Cryogenic based CO2 capture technologies: State of the art developments and current challenges, Renewable and Sustainable Energy Reviews, 101, 265-278. Torstensen J.Ф., Helberg R.M.L., Deng L., Gregersen Ф.W., 2019, PVA/nanocellulose nanocomposite membranes for CO2 separation from flue gas, International Journal of GHG Control, 81, 93-102. Xu J., Wu H., Wang Z., Qiao Z., Zhao S., Wang J., 2018, Recent advances on the membrane processes for CO2 separation, Chinese Journal of Chemical Engineering, 26, 2280-2291. 1080