Data driven CFD based design optimisation of flow pattern in a gravitational mixer settler

Gravitational mixer settlers (GMXSs) are widely used for liquid-liquid extraction (LLE). Immiscible fluids are mixed to promote the transfer of compounds, then separated by gravity in a settling chamber. Micromechanical fluid interactions decisive to the separation process are complex, however studies have shown that by optimising settler flow pattern, separation performance can be significantly improved. In this paper, an optimisation framework for GMXSs designs is investigated which uses experimentally validated single phase Computational Fluid Dynamics (CFD) and residence time distribution (RTD) analyses to identify optimal combinations of design features which maximise desirable characteristics such as resident time and pressure drop. The design of the settler is formulated in terms of two design variables: flow rate and position of the inlet baffle. A Radial Basis Function (RBF)-based surrogate modelling approach using a Design of Experiment (DOE) technique and a permutation genetic algorithm was used to establish optimal process parameters. A Pareto front is built which enables designers to explore appropriate compromises between designs with small residence time and those with small pressure drop.