Fast and accurate classifier-based surrogate models ensuring miscibility in optimisation-based design of solvent mixtures Tanuj Karia, Benoît Chachuat, Claire S. Adjiman Imperial College London, United Kingdom Solvent mixtures have the potential to achieve better processes (1) and products (2) relativeto using pure solvents. Given the large number of possible mixtures, it is desirable to use a systematic optimisation-based framework for solvent mixture discovery, such as Computer-aided mixture/blend design (CAM b D) (3). A key consideration during the design of solvent mixtures is to ensure that the designed mixture exhibits a single, stable liquid phase at the conditions of interest. Checking for stability is a challenging problem, which is often tackled using simplifying assumptions (4), to make the optimisation model is tractable (5,6). This can be complemented by a rigorous stability check post-optimisation, (7,8). This, however, entails re-solving the optimisation model specifically when the designed mixture is found to be immiscible (9) and often results in lower mixture performance. In this work, we propose to embed an artificial neural network-based classifier-surrogate model in the CAM b D framework to capture the phase stability constraint. The performance of the classifier surrogate is tested on two CAM b D case studies (10,11) for solvent design. Using these case studies, we demonstrate the effectiveness of the surrogate classifier in enabling the in silico design of miscible mixtures without the need for re-optimisation. Finally, the use of the proposed surrogate classifier provides the probability of miscibility as a practical, interpretable metric. References 1. Chai S, Song Z, Zhou T, Zhang L, Qi Z. Computer-aided molecular design of solvents for chemical separation processes. Current Opinion in Chemical Engineering 2022;35:100732. 2. Enekvist M, Liang X, Zhang X, Dam-Johansen K, Kontogeorgis GM. Computer-aided design and solvent selection for organic paint and coating formulations. Progress in Organic Coatings 2022;162:106568. 3. Karunanithi AT, Achenie LE, Gani R. A new decomposition-based computer-aided molecular/mixture design methodology for the design of optimal solvents and solvent mixtures. Ind Eng Chem Res 2005;44(13):4785-4797. 4. Smith JM, Van Ness HC, Abbott MM. Introduction to Chemical Engineering Thermodynamics. McGraw-Hill; 2001. 5. Buxton A, Livingston AG, Pistikopoulos EN. Optimal design of solvent blends for environmental impact minimization. AIChE J 1999;45(4):817-843. 6. Dahmen M, Marquardt W. Model-based formulation of biofuel blends by simultaneous product and pathway design. Energy Fuels 2017;31(4):4096-4121. 7. Michelsen ML. The isothermal flash problem. Part I. Stability. Fluid Phase Equilib 1982;9(1):1-19. 8. Baker LE, Pierce AC, Luks KD. Gibbs energy analysis of phase equilibria. Society of Petroleum Engineers Journal 1982;22(05):731-742. 9. Conte E, Gani R, Ng KM. Design of formulated products: a systematic methodology. AIChE J 2011;57(9):2431-2449. 10. Jonuzaj S, Akula PT, Kleniati P, Adjiman CS. The formulation of optimal mixtures with generalized disjunctive programming: A solvent design case study. AIChE J 2016;62(5):1616-1633. 11. Watson OL, Jonuzaj S, McGinty J, Sefcik J, Galindo A, Jackson G, et al. Computer aided design of solvent blends for hybrid cooling and antisolvent crystallization of active pharmaceutical ingredients. Organic Process Research & Development 2023;25(5):1123-1142.
IP09
© The Author(s), 2023
Made with FlippingBook Learn more on our blog