Hybrid separations combining distillation and organic solvent nanofiltration for separation of wide boiling mixtures

•Design method for hybrid separations combining nanofiltration and distillation.•Optimisation under uncertainty with multi-objective optimization algorithms.•Applicability is illustrated for the separation of a wide boiling mixture.•Best membrane was found in membrane screening experiments.•Model validation and parameterisation based on detailed experiments is performed. Membrane assisted hybrid separations offer tremendous potential for process intensification which aims at increasing resource efficiency as well as decreasing operating and capital costs. Design of such processes is challenging due to large number of degrees of freedom but also due to large experimental effort necessary for membrane screening and for characterising membranes in whole operating range. To address these issues, this paper elaborates a four-step design method for combination of organic solvent nanofiltration (OSN) and distillation in a hybrid separation of wide boiling mixtures. The design method is applied in a case study which is the separation of small amounts of heavy boiler from a mixture containing a mid- and a light-boiler. In the first step, different process options are generated based on heuristics and engineering judgement and screened for feasibility. In the second step, the options are evaluated based on quantitative metrics using rigorous models. In this step the unknown key parameters are identified, and their influences on the process performance are quantified in a detailed a priori process analysis. If hybrid separations with OSN show to be promising when compared to stand-alone distillation, experiments are conducted to (i) identify the best membrane for the operating window in which the hybrid process operates and (ii) to perform model validation and parameterisation in the third step. In the last (fourth) step, an optimisation is performed to identify the best (cost optimal) process using the experimental data gained in step three.