(65b) Process-Based Solvent Screening for Efficient Extractive Distillation

Extractive distillation is a commercially viable method for separating azeotropic and close-boiling mixtures, which is superior to traditional distillation. Due to the huge design space of solvents, the search for optimal solvent candidates for use in extractive distillation remains a challenging endeavor. Clearly, a systematic solvent screening approach is required. The traditional solvent screening methods rely on the calculation of simple performance metrics (e.g., selectivity and working capacity) and shortcut process models. However, those methods cannot represent the real process performance leading to suboptimal solvent selection. In this work, process-based solvent screening is investigated and exemplified via the separation of ethylbenzene/styrene close-boiling (boiling point: 409 vs 418 K) mixture. Annually, more than 500,000 tons of the mixture are produced through the catalytic dehydrogenation of ethylbenzene. Many researches have shown that the pyrolysis of plastic wastes, particularly polystyrene, can generate a significant amount of the mixture. So far, extensive researches have already been conducted on screening ionic liquids for separating the mixture via extractive distillation. Unfortunately, none has demonstrated superior overall performance compared to the benchmark solvent sulfolane that is currently used for separating the ethylbenzene/styrene mixture. The major challenge in the modelling and optimization of extractive distillation processes is the complex thermodynamics. Considering the simplicity and accurate interpolation, rational functions are used in this work as surrogates to substitute the complex thermodynamics. In doing so, ED processes can be optimized using off-the-shelf solvers within seconds. Given such an accelerated optimization of extractive distillation processes, an automated framework is developed to quickly screen around 4000 potential organic solvents and evaluate their actual process performance. As a result, a number of possible solvents are discovered that could compete against the benchmark sulfolane in terms of total annualized cost.