Poster 139: Model Identification of Pressure Drop in Membrane Channels with Multilayer Artificial Neural Networks
TimeThursday, 21 November 20198:30am - 5pm
DescriptionThis poster presents the work of identifying a data-driven model of pressure drop in spacer-filled reverse osmosis membrane channels and conducting CFD simulations. The established model correlates the pressure drop with a wide range of design objectives, which enables a quantitative description of the geometric structures and operation conditions for improvement. This way, it aims at optimizing the spacer geometry with minimal effort. Furthermore, a high-performance computing strategy is employed to tackle the resulted intractable computational task in the identification procedure and CFD simulations.