Extra tree regressor and Tree-structured parzen estimator based machine learning model for predicting nanofluid’s Nusselt number

Abstract Boiling is a very effective method of heat transfer process, which is characterized by phase change, due to phase change high transfer rates of heat occur at much smaller temperature differences between the heated surface and the fluid. By introducing nanofluids, which is a dispersion of nanoparticles in basic fluids when incorporated in flow boiling systems, it can be convenient to enhance energy efficiency and ultimately reduce world energy consumption. The intent of this research is to evaluate the practicality of using machine learning (ML) models as a substitute for Computational Fluid Dynamics (CFD) in heat transfer simulation. This research uses Extra Tree Regressor (ETR) with Tree-structured Parzen Estimator algorithm (TPE) to estimate the Nusselt number in water-based nanofluids that consist of Al 2 O 3 and TiO 2 , nanoparticles. ETR is tree based machine learning algorithm and TPE is used to tune the hyper parameter of ETR. The ETR-TPE model is used to establish a correlation among nanoparticle parameters such as type of nanofluids, Reynolds number, size, volume percentage and Nusselt number. Various statistical measures and scatter plots are used to compare and estimate the performance of the proposed ETR-TPE model. The model has excellent predictive accuracy, as shown by a R 2 value of 0.980381 and 0.986313 for Al 2 O 3 and TiO 2 respectively. The Root Mean Square Error (RMSE) is computed and found to be 12.96 and 10.01 for Al 2 O 3 and TiO 2 respectively. The proposed ETR-TPE model demonstrates a strong correlation in accurately estimating the Nusselt Number based on experimental data.