Performance of the Artificial Neural Networks for Estimation of Pool Boiling Heat Transfer Coefficient of Alumina Water-Based Nanofluids

Due to the transformation of the latent heat during the boiling process, it appeared as an efficient process for heat transfer purposes. Boiling heat transfer coefficient is the most important parameters for calculation of the amount of energy that can be transferred by the boiling process. Recently, new class of fluids namely nanofluid is widely used to improve performance of the boiling process. In spite of the large amount of experimental researches on the pool boiling heat transfer coefficient, theoretical and modeling basics of this parameter was less taken into account.Therefore, in this study, pool boiling heat transfer coefficient of water-alumina nanofluid is tried to be predicted using artificial neural networks. Excess temperature, operating pressure, diameter and weight fraction of nanoparticles are used for modeling of the considered coefficient of the nanofluids. At first, four different types of the artificial neural networks, and various empirical correlations are employed for estimation of the pool boiling heat transfer coefficient of water- alumina nanofluid. Thereafter, the best intelligent model is selected through comparing the predictive accuracy of various intelligent/empirical approaches.

The results show that a two-layer feedforward neural network with fourteen hidden neurons is the best model for estimation the pool boiling heat transfer coefficient of water-alumina nanofluid. This model was able to predict the considered coefficient with overall MSE=3.98, AARD%=9.80, and R²=0.98.