Massive, automated solvent screening for minimum energy demand in hybrid extraction-distillation using COSMO-RS

An automated approach for large-scale solvent screening is presented based on a comprehensive process-level assessment. In this solvent screening approach, COSMO-RS is used to efficiently predict physical properties for large numbers of solvents without the need for experimental data. The predicted thermodynamical behavior is used in pinch-based separation models for a thermodynamically sound and robust calculation of the minimum energy demand. With this approach, the performance of a hybrid extraction-distillation process is evaluated fully automated for more than 4600 solvents. The massive solvent screening approach is successfully applied to purification of the bio-based platform chemical γ-valerolactone (GVL). Novel promising solvents are identified. A reduction of 63% is achieved in minimum energy demand using the best predicted solvent in comparison to the literature benchmark. Restricting the approach to known classes of solvents, we still find a reduction of 31%. The process-level assessment overcomes the limitations of heuristics based on physical properties only, and allows for efficient and robust solvent screening.