Computer-aided Molecular Design of Phase-Change CO 2 Capture Solvents considering Thermodynamics, Reactivity and Sustainability

Transcription

1 Computer-aided Molecular Design of Phase-Change CO 2 Capture Solvents considering Thermodynamics, Reactivity and Sustainability A.I. Papadopoulos, P. Seferlis Centre for Research and Technology Hellas (CERTH), GR F. Perdomo-Hurtado, C. S. Adjiman, A. Galindo, G. Jackson Imperial College London, UK G. Shavalieva, S. Papadokonstantakis Chalmers University of Technology, SE V. Papaioannou, T. Lafitte, Costas Pantelides Process Systems Enterprise Ltd, UK 8 th CCUS KOREA, Jeju Island, January 25 th, 2018 H2020r.

2 Phase-Change Solvents Phase-change solvents form two distinct liquid phases: Directly after reaction with CO 2. After reaction with CO 2 and subsequent heating at approximately 80 o C 90 o C. The 2 liquid phases may be easily separated by a decanter. A CO 2 rich phase goes to regeneration and a CO 2 lean phase goes back to the absorber.

3 Update on current status DMX TM process: The solvent is 1,3-dipropyl-methyl-xanthine. [1] Lipophilic amines [2] DEEA-MAPA process: Solvent mixture of 5M diethylethanolamine (DEEA) and 2M 3-(Methylamino)propylamine. [3] 3H Self-concentrating process: An amine + non-aqueous solvent. [4] Advantages: Separation of the phases allows reduction in the liquid flow rate in the stripper, and reduces the energy requirement for regeneration of the solvent. Regeneration energy at approx. 2.3GJ/ton CO 2 captured compared to GJ/ton CO 2 captured of MEA. [1] Raynal L, Bouillon PA, Gomez A, Broutin P, 2011, Chem Eng J, 171(3), [2] J. Zhang, Y. Qiao and D.W. Agar, Chem. Eng. Res. Des., 2012, 90, [3] Pinto DDD, Zaidy SAH, Hartono A, Svendsen HF, 2014, Int J Green Gas Cont 28, [4] Hu, L., 2010, Methods and systems for deacidizing gaseous mixtures. US Patent

4 Motivation Very few such solvents reported to date, even fewer tested at pilot plant scale. Solvent selection is made on heuristic grounds. Very few selection criteria are considered (e.g. sustainability is an afterthought). Systematic solvent design methods have the potential to make a significant impact.

5 Modeling detail How can we design solvents? Molecularatomic scale Molecularprocess interface Computational chemistry methods Equations of state Processscale Group contribution property prediction models Range of molecular options In design we need to accurately determine an optimum molecular structure out of millions of options Computational chemistry methods (e.g. DFT, MD etc.): Robust determination of molecular chemistry (e.g. electronic structure etc.) Time consuming simulations-limited range of molecular options Equations of state: Bridging the gap between molecular characteristics and macroscopic, processrelated properties May enable predictions of VLLE mixture properties Data not enough yet for a wide range of molecular structures

6 Group Contribution (GC) Methods Molecular groups e.g. -CH 2 -, OH, -CH 3, -CF 3 -COOH etc. Evaluation of chemical feasibility Molecule e.g. CH 3 -CH 2 -CF 3 Property calculation (GC) They use the contribution of functional group to calculate molecular property: Result from experimental data fitted to a model. Contribution of a functional group is transferable to different molecules. Calculations are fast for both pure component and mixture properties. Sufficient to identify few molecular structures of desired properties, which can then be investigated through computational chemistry models, EoS or experiments. Wide range of molecules may be investigated

7 Computer-Aided Molecular Design (CAMD) Molecular simulator CAMD Optimization algorithm Molecular groups e.g. -CH 2 -, OH, -CH 3, -NH 2 -COOH etc. Evaluation of chemical feasibility Molecule e.g. CH 3 - NH 2 GC Property calculation Evaluation of performance measure Generation of new molecular structure GC methods with optimization algorithms. Employ molecular properties as selection criteria. Suitable for efficient: Screening of large databases including pre-existing molecular structures. Synthesis of novel molecular structures. Previous applications in the design of solvents for separations and reactions, refrigerants, polymers.

8 The approach Decision levels Evaluation approach Here we will discuss only the Molecular and Interface levels.

9 Optimization-based CAMD/Molecular level Molecular simulator Groups i.e. CH 2 -, OH, -CH 3 -CH 2 NH 2 etc. Chemical constraints Molecular structure i.e. CH 3 -NH 2 Property calculation The molecular structure is a decision variable in optimization. In case of pre-existing molecules the property calculation features may be used independently. What are the properties that may lead to optimum solvents suitable for CO 2 capture? Optimization algorithm New molecular structure i.e. CH 3 -CH 2 -OH Decision making Evaluation of performance criteria Papadopoulos, A.I.,Linke, P., AIChE J., 52(3), (2005)

10 Performance criteria/thermo & Reactivity Solvent property Impact on absorption/desorption process Solubility parameter (δ t -RED) Vapor pressure (P vp ) Liquid heat capacity (C p ) Density (ρ-v m ) Basicity (pk a ) Viscosity (μ) Boiling temperature (T bp ) Melting temperature (T m ) Ability to dissolve CO 2 / All design & operating parameters Solvent losses Sensible heat requirements / Desorption reboiler duty Equipment size-capacity Reaction rate / Size of absorption column Mass transfer coefficients-packing material characteristics Solvent evaporation losses Solvent solidification Solubility parameter is a solvent-co 2 mixture property All properties may be calculated through GC methods, except for pk a

11 Evaluation of phase-change potential The evaluation of liquid-liquid phase change in the presence of CO 2 and H 2 O necessitates the use of a model that accounts for non-ideal interactions. A preliminary approach is to use the solubility parameter (δ). We are interested in solvents which remain in one liquid phase at T Abs. s HO 2 T Abs The desired solvent should dissolve as much CO 2 as possible at T Abs s CO2 TAbs The desired solvent should form two liquid phases at T Des s HO 2 TDes CO 2 should dissolve as little as possible at the desired solvent at T Des s CO 2 Des

12 Sustainability metrics Life Cycle Assessment (LCA): Environmental impacts of solvent production Includes extraction, manufacture and disposal Sustainability Metrics Life Cycle Assessment Hazard Assessment Environmental Health and Safety (EHS) Hazard Assessment: Health Safety Harm potential of using the solvent within a CO 2 capture unit. Mainly dependent on solvent physical properties. Assessment using the EHS method [1,2] Environment [1] Koller G., Fischer U., Hungerbühler K., Ind. Eng. Chem. Res., 39, (2000) [2] Sugiyama H., Fischer U., Hungerbühler K., Hirao M., AIChE J., 54, (2008)

13 Performance criteria-lca Indicators for Life Cycle Assessment: Cumulative Energy Demand (CED): A single-score method assessing the amount of primary energy equivalents required for the finished product, counting direct, indirect and feedstock energies. The CED considers resource extraction and use. Global Warming Potential (GWP): A single-score method focusing on the effects of greenhouse emissions on the climate and temperature. Ecoindicator-99 (EI 99) : A damage-oriented multi-score assessment method providing results in the endpoint impact categories Human Health, Ecosystem Quality and Resources. (May be aggregated into single score) Life Cycle Assessment Production Data Ecoinvent Database & Plant Data Estimations FineChem Tool

14 Performance criteria-ehs Indicators for Environmental Health and Safety Hazard Assessment: Safety: Flash point Fire & Explosion Boiling point Mobility Auto-ignition temperature Decomposition Health: Toxic exposure limits Acute & Chronic Toxicity (Risk phrases/ EU classification) PH/dermal exposure limits Irritation Environment: Lethal aquatic concentrations Halflife in the environment Bioconcentration factor Chronic effects Water mediated effects Degradation Accumulation Air mediated effects

15 Interface Level Decision levels Evaluation approach

16 Statistical Associating Fluid Theory (SAFT) Aqueous mixtures of amines and CO 2 have complex behaviour. Classical approaches (UNIQUAC, NRTL) require extensive experimental data and are only applicable to liquid phase. SAFT is derived from molecular theory and statistical mechanics. SAFT requires a reduced number of temperature-independent parameters than can be estimated from VLE data. SAFT is a versatile approach: Polar fluids (e.g. CO 2, refrigerants) Strong hydrogen bonding (e.g. water) Mixed electrolytes (e.g. charged surfactants) Polymers Gas hydrates Mac Dowell et al., Ind. Eng. Chem. Res., 49, 1883 (2010). Rodriguez et al., Mol. Phys., 110, 1325 (2012).

17 Physical modelling of reactions Reactions are mediated through association sites. Chemical and physical equilibrium is approached by considering the reaction products as associated physical complexes of the reacting molecules. No need to postulate new compounds or treat as electrolytes. Mac Dowell et al., Ind. Eng. Chem. Res., 49, 1883 (2010). Rodriguez et al., Mol. Phys., 110, 1325 (2012). Jou et al., Can. J. Chem. Eng., 73, 140, (1995). Böttinger et al., Fluid Phase Equilib., 263, 131, (2008).

18 SAFT-γ: Modelling framework SAFT-VR molecular parameters Main Difference SAFT - group parameters m No of segments m i segment diameter i dispersive energy i dispersive range i association sites hb iiab, r c iiab shape factor segment diameter dispersive energy dispersive range association sites SAFT-VR: each molecule is a separate entity, the segments are non-overlapping and of the same type. SAFT-γ: Groups are used to build molecules, segments are of varying size and type. In we are developing more predictive models that allow the performance of new solvent molecules to be predicted by using a group contribution framework. S k hb kkab, r c kkab A. Lymperiadis, C. S. Adjiman, G. Jackson, and A. Galindo, Fluid Phase Equilibria, 274, 85 (2008). V. Papaioannou, C. S. Adjiman, G. Jackson, and A. Galindo, Fluid Phase Equilibria, 306, 82 (2011).

19 Solvent design and selection problems Formulation of solvent design problem max min,,, s.t. P c RED T T vp m bp p T T Abs Des s HO 2 TAbs s CO2 TAbs s HO 2 TDes s CO2 TDes Obtain Pareto front f 1 f 2

20 Solvent design and selection problems Generate all isomers of the Pareto solvents Calculate pk a Calculate EHS, CED, GWP, EI99 Formulation of solvent selection problem Formulate as multi-criteria assessment max, pk a min P, c,, RED, EHS, GWP, CED, EI99 vp p Calculate augmented index min J ig i jp r a x i, j i, j, pk Des a, RED, CED, GWP, EI99 T, Pr P vp, cp,, RED Generate Pareto fronts of J vs. each Pr J i Pareto front Pr

21 Implementation details We used 23 different functional groups. From the solvent design stage we obtained 120 molecular structures, and investigated approx. 900 isomers. This set was refined to 12 final solvents in the solvent selection stage. We considered aliphatic, non-cyclic and cyclic groups. We considered known phase-change options as reference solvents such as DMCA, MCA, DPA, DsBA [1] to compare the performance of the designed solvents. We provide results without considering sustainability properties at this stage. [1] J. Zhang, Y. Qiao and D.W. Agar, Chem. Eng. Res. Des., 2012, 90,

22 Pareto fronts of molar volume vs index J

23 Pareto fronts of viscosity vs index J A previously used solvent appears in the Pareto front

24 Pareto fronts of vapor pressure vs index J A previously used solvent appears in the Pareto front

25 Pareto fronts of heat capacity vs index J

26 Pareto fronts of basicity vs index J

27 Pareto fronts of RED vs index J

28 Designed structures DMCA, DsBA, MCA, DPA, MAPA are experimentally documented and highly performing phase-change solvents. They were all designed by the CAMD approach. MCA, DsBA and MAPA are in the finally selected solvents. Several other structures are very similar to those ones. Designed structures Known phase-change solvents

29 VLLE prediction of Hexylamine T = 353 K p = 0.1 MPa Hexylamine Carbon Dioxide Water Hexylamine is a known phase-change CO 2 capture solvent. We illustrate VLLE predictions obtained using SAFT-VR. Work is underway to predict the VLLE of the designed solvents using the GC version of SAFT.

30 Conclusions Demonstrated a framework for optimization-based design of phase-change CO 2 capture solvents. Utilized numerous properties associated with thermodynamics, reactivity and sustainability of solvents. Showed how few molecules can be selected, which can then be validated using a rigorous EoS. Previous application of framework in non-phase change CO 2 capture solvents: A.I. Papadopoulos, S. Badr, A. Chremos, E. Forte, T. Zarogiannis, P. Seferlis, S. Papadokonstantakis, C.S. Adjiman, A. Galindo, G. Jackson, 2016, Molecular Systems Design and Engineering, 1, This project has received funding from the European Union s Horizon 2020 research and innovation programme under the grant agreement H2020-LCE /H2020-LCE-2016-RES-CCS-RIA