Characterization of heat transfer and artificial neural networks prediction on overall performance index of a channel installed with arc-shaped baffle turbulators

Abstract

Influences of baffle pitch ratio (p/w) and attached angle of arc-shaped baffles (AB) on the overall performance index (OPI) of a channel installed with AB have been carefully studied. In addition, an artificial neural network (ANN) model for predicting the OPI of the channel was reported. The arc-shaped baffle (AB) showed a significant effect on the augmented heat transfer and friction loss penalty as compared to a smooth channel. As the attached arc shaped angle (θ) increased, both Nusselt number and friction factor intensified. The Nusselt number values at θ = 90° were higher than those at θ = 20°, 40°, 60°, and 80° by up to 5.8%, 3.9%, 2.3% and 2.5%, respectively. The Nusselt number increased when the p/w was raised from 4.0 to 8.0 while the opposite trend was observed when the p/w was raised from 8.0 to 12.0. The maximum OPI of 1.43 was achieved by using the baffles with θ = 90° and pitch ratio of 8.0 at Re = 4000. For the development of ANN models for predicting the OPI, it was found that the best predictive performance was (R2) of 0.99843407 for ANN model of 3-50-50-1 with Tanh-Tanh activation function at epoch of 1200.