COSMO‐CAMPED – Solvent Design for an Extraction Distillation Considering Molecular, Process, Equipment, and Economic Optimization

Abstract

For an extraction process to be economically feasible, selecting a suitable solvent is imperative. This work extends the computer-aided molecular and process design (CAMPD) framework COSMO-CAMPD for solvent design in an extractiondistillation process by replacing a pinch-based process model with a hybrid, rate-based extraction-distillation process model. The resulting CAMPD framework is able to evaluate solvent candidates by investment costs in addition to operating costs in a fully predictive manner. In a case study for purifying acetone, we show that our framework's guided exploration of a vast design space yields solvents with superior performance compared to a benchmark solvent from the literature.

Multi-objective optimization strategy for green solvent design via a deep generative model learned from pre-set molecule pairs

“…50 In these studies, the property prediction in a molecular scale is addressed using COSMO approaches while the process model can be a pinch-based model based on a minimum solvent flow rate and minimum energy demand 49 or a more rigorous rate-based model. 50 The use of such process models is relevant for an accurate process design but there exist simpler criteria for assessing extractive distillation feasibility, such as solvent capacity and selectivity, 51 which are further related to infinite dilution activity coefficients (IDAC), and univolatility curves. 52 In this study, we propose a molecular multi-objective and multi-scale optimization framework for the combined molecular and process design with the predicted process constraints (solvent selectivity and capacity based on IDAC) where the process-related properties are directly used to train the molecular structure optimization model, with the help of deep-learning techniques.…”

“…For example, some authors have proposed a framework for the integrated design of a solvent and extractive distillation process by solving a multi-objective optimization problem addressing constraints related to thermodynamic process feasibility, along with process operation, a process model, and molecular constraints, 49 or a more rigorous rate-based model. 50 In these studies, the property prediction in a molecular scale is addressed using COSMO approaches while the process model can be a pinch-based model based on a minimum solvent flow rate and minimum energy demand 49 or a more rigorous rate-based model. 50 The use of such process models is relevant for an accurate process design but there exist simpler criteria for assessing extractive distillation feasibility, such as solvent capacity and selectivity, 51 which are further related to infinite dilution activity coefficients (IDAC), and univolatility curves.…”