Optimization-Based Design of Solvent-Based Separation Processes

As a major emitter of greenhouse gases, the chemical industry is at the center of the transformation needed to fight climate change. Therefore, the transition to renewable feedstocks and the downstream processing of bio-based materials are essential but introduce complex (azeotropic) separation tasks. Proven methods for the separation of these mixtures are solvent-based separation processes such as extractive distillation (ED), heteroazeotropic distillation (HAD) and liquid-liquid extraction (LLX), referred to as hybrid processes due to their inherent need for solvent recovery. In this thesis, robust optimization-based design methods are developed to enable informed decisions in early design phases based on rigorous models. An innovative topology-based initialization strategy for the design of HAD processes is presented, which enables the use of robust process models for various mixture topologies. Furthermore, a hierarchical approach to solvent selection in LLX processes is presented that exploits successive model refinement based on ab-initio property predictions, thus eliminating early-stage dependence on experimental data. A case study for the purification of gamma-valerolactone demonstrates its capabilities with cost reductions of over 50%, while another case study evaluates the process performance of an innovative deep eutectic solvent for the purification of furfural. Finally, the comprehensive design of ED processes combining well-understood existing methods is addressed. As a novelty, solvent selection is performed simultaneously with the determination of an appropriate energy-integrated process.