Solar flat plate collectors with internally grooved absorber tubes for enhanced efficiency: Application of eXplainable artificial intelligence using cooperative game theory-based Shapley values

This study investigates the impact of grooved absorber tubes on the efficiency of solar flat plate collectors (SFPCs). Conducted in Vellore, India, it compares plain tubes (case 1) with grooved tubes (cases 2 and 3) using water and glycol-water mixtures. Results show that grooved tubes significantly improved efficiency, particularly with glycol-water mixtures. Efficiency increased by up to 12 % in transient flow conditions and by 8 %–20 % across various flow rates compared to plain tubes. A 5 % increase in solar radiation led to a 7 % rise in effectiveness. Heat removal factors ranged from 0.66 to 0.69 for water and 0.76–0.78 for glycol-water. Case 2 exhibited the highest efficiency under transient flow, while case 1 had the lowest under laminar flow conditions. These findings suggest that grooved tubes significantly enhance SFPC performance. Three machine learning approaches—AdaBoost, K-Nearest Neighbors, and Tweedie regression—were employed for model prediction, utilizing SHapley Additive exPlanations for interpretability. The AdaBoost model outperformed with an MAE of 0.0150, MSE of 0.0003, RMSE of 0.0175, and R2 of 0.8494. Key features identified by the model included groove pitch, ambient and glass plate temperatures, and solar radiation. These findings offer insights into enhancing SFPC efficiency through absorber unit adjustments.