AI-assisted maldistribution minimization of membrane-based heat/mass exchangers for compact absorption cooling

Abstract

Flow maldistribution has been a major challenge for heat/mass exchangers, which is a particular concern in compact membrane-based absorbers used in absorption refrigeration systems driven by renewable/waste energy. Herein, we construct an artificial intelligence (AI) tool coupling a 3D CFD model, a discrete model, and an optimization algorithm for the development of highly efficient and compact plate-and-frame membrane-based absorbers (PFMAs). In the AI-assisted tool, CFD simulations demonstrate that the PFMA suffers from more severe flow maldistribution as the number of channels increases. The average absorption rate is decreased by 21.44% as the number of channels increases from 5 to 21. The heat and mass transfer performance of the 5-channel and 21-channel models is reduced by 3% and 22%, respectively. Meanwhile, a simple and universal discrete model is developed and validated to predict the flow distribution in PFMAs, with a maximum deviation of 10.18%. To minimize the flow maldistribution, an optimization structure with a uniform distributed flow field is determined by developing and coupling a rapid optimization algorithm. After optimization, a reduction of about 10 times in the flow maldistribution can be achieved, and the heat and mass transfer performance deterioration caused by the flow maldistribution can be minimized to about 1%.