MLP-driven prediction of heat and mass transfer performance of tangent hyperbolic fluid flow over a rough rotating disk with variable thickness

Abstract

The objective of the study aligns with the United Nations sustainable development goals by targeting enhanced energy efficiency and sustainable resource management. Keeping this in mind, the authors have worked on to examine heat and mass transfer performance of three-dimensional magnetohydrodynamic flow of tangent hyperbolic fluid over a rotating disk and leveraged a deep learning-based prediction framework to forecast the critical parameters, radial and tangential skin friction coefficients, local Nusselt number, and local Sherwood number. This work explores the second law of thermodynamics, which pertains to irreversibility. For the study of deep learning, multilayer perceptron's architecture is implemented to accurately predict the parameters. The dimensionless governing equations are solved numerically by applying the bvp4c solver. From the outcomes, the profiles of the Bejan number continuously decrease with increasing thickness coefficient of the disk. The multilayer perceptron's model achieved a $R^2-$value of 82.32 % on testing data and 99.73 % on training data.