Performance enhancement of a solar-driven DCMD system using an air-cooled condenser and oil: Experimental and machine learning investigations

Abstract

Solar-driven direct contact membrane distillation systems (DCMD) are disadvantaged by low freshwater productivity and low gain-output-ratio (GOR). Consequently, this study aims to achieve two primary objectives: i) improving the solar DCMD performance, and ii) harnessing machine learning models for precise and straightforward modeling of the solar DCMD system. To achieve these goals, a novel solar DCMD system powered with oil-filled heat pipe evacuated tube collectors (HP-ETCs) and equipped with an air-cooled condenser was used for the first time. The system was evaluated under eight different scenarios covering both its energy and economic performances. The performance prediction of three different machine learning models including ANN, SVR and RF was assessed for the proposed system. The results showed that integrating an air-cooled condenser and oil-filled HP-ETCs into the solar DCMD system significantly improved the performance and reduced freshwater cost, resulting in: a 35.39–37 % increase in freshwater productivity; a 30.64–31.57 % enhancement in GOR; a 35–38 % rise in daily efficiency; and a 20 % decrease in freshwater cost. The results demonstrate that ANN and SVR have excellent performance for modeling the solar-driven DCMD system, achieving MAPEtest values of approximately 1 % and 4 % for predicting permeate flux and GOR, respectively.