Machine learning-driven prediction and optimization of turbulence-enhanced multi-effect distillation system for efficient water treatment

Abstract

Multi-effect distillation (MED) is one of the most energy-efficient thermal desalination technologies for freshwater production using low-grade heat; however, its performance is constrained by limited heat-transfer efficiency and the lack of accurate predictive tools for process optimization. This study presents an integrated experimental and explainable machine-learning investigation of a six-effect vertical tube evaporator (VTE)-based MED system operating under smooth-tube and turbulence-enhanced conditions. The effects of the feedwater temperature (70–90°C), steam temperature, and brine temperature on the distillate output, heat transfer rate (HTR), overall heat transfer coefficient (OHTC), gain output ratio (GOR), recovery rate (RR), and distillate temperature were experimentally evaluated. A normalized multi-parameter optimization approach was employed to determine the optimum operating conditions, while an XGBoost ensemble learning model was developed for performance prediction and compared with the ANN, SVM, and Random Forest models. The turbulence-enhanced configuration increased the distillate output, HTR, OHTC, GOR, and recovery rate by approximately 25–30%, with the optimum performance achieved at a feedwater temperature of 90 °C, while maintaining a stable distillate temperature and acceptable hydraulic performance. The XGBoost model demonstrated the highest prediction accuracy (R2 = 0.972, RMSE = 0.87, and MAE = 0.61), outperforming the other machine-learning models. SHAP-based feature interpretation identified feedwater temperature as the dominant factor governing MED performance. The proposed hybrid experimental–machine-learning framework provides an accurate, scalable, and energy-efficient approach for intelligent optimization of MED systems, supporting sustainable freshwater production and the advancement of thermal desalination technologies. The results confirm that turbulent enhancement as a scalable and energy-efficient strategy for MED performance enhancement, supporting efficient water treatment aligned with the United Nation’s sustainable development goals (SDGs), such as SDG-6 and SDG-11.