Anfis modeling of boiling heat transfer over tube bundles

Abstract

The article describes the application of artificial intelligence technique artificial neuro-fuzzy inference system (ANFIS) to predict the flow boiling heat transfer coefficient for distilled water on individual row in plain tube bundles. The variation of row-wise heat transfer coefficients is discussed with respect to the operating conditions such as mass flux, heat flux, and pitch to distance. A semi-empirical correlation is also formulated to predict the flow boiling Nusselt number taking the Peclet number, Froude number, and pitch-to-diameter ratio as inputs. The experimental data are predicted with �15% accuracy by the semi-empirical correlation, whereas the ANFIS model is capable to predict within a maximum error of �10%. � Springer Nature Singapore Pte Ltd. 2019