Development of a XGBJSO Hybrid Machine Learning Model for Predicting Heat Transfer Coefficient of Water and Al2O3 Nanofluid in Flow Boiling

A flow boiling examination of water along with a nanofluid composed of water-Al2O3 flowing through an annulus has been carried out. The examination was carried out for the mass flux, pressure, and heat flux. The volumetric concentration of Al2O3-water nanofluid was constant at 0.10 %. Experimental results were validated with Chen correlation which showed good agreement. Results demonstrated that the heat transfer coefficient of both water-Al2O3 nanofluid and water enhanced with the enhancement in pressure, heat flux, and mass flux. Water-Al2O3 nanofluid showed better performance than water and the best heat transfer coefficient was observed for water-Al2O3 nanofluid at pressure 1.5 bar, heat flux 144 kWm-2, and mass flux 1,015 kgm-2s-1. A linear heat transfer coefficient correlation was generated for water-Al2O3 nanofluid. This research created the XGBJSO machine learning model (XGBoost with Jelly Fish Search Optimizer), which showed remarkable predicted accuracy for the heat transfer coefficient of Al2O3 nanofluid in water. A high R2 score of 0.99986 for the training set and 0.98453 for the testing set, showed a significant correlation between predicted and actual values. The XGBJSO framework offers a scalable and adaptable solution for thermal system optimization. This work advances the field of heat transfer prediction and provides a foundation for future research in thermal management.