Artificial Neural Network for Simulation of Friction drag and heat transfer enhancement in non-Newtonian nanofluid flow over a circular cylinder: Non-similar solution

Abstract

The primary purpose of the investigation is to highlight the significance of artificial intelligence and machine learning techniques in engineering and fluid mechanics problems. With the advancement in Artificial Intelligence (AI) and Machine Learning (ML) techniques, the computational efficiency and accuracy of numerical results have enhanced. The present study aims to examine MHD boundary layer flow of Casson Nanofluids (CNFs) over a cylinder with thermal radiation effects. Tiwari and Das' model is used to develop a governing Casson nanofluid problem that contains nonlinear PDEs. ANN and numerical models are utilized to examine the effect of numerous physical parameters on heat and fluid transfer properties. The Keller-Box technique, a second-order accurate implicit finite difference scheme, is used to solve the modified fundamental differential equations computationally. However, the predicted solution is examined with MLP-ANN. The efficiency of the proposed model is examined using MSE, the correlation index, and the optimal curve fitness function. An optimal performance of MLP-ANN is examined with the computation of MSE [2.56×10^(-8), 4.8×10^(-8), 4.42×10^(-8), 3.23×10^(-8), and 4.81×10^(-8)] is attained against the epoch [1000, 608, 555, 822, and 787] for scenarios 1-5 with case 1. Graphs and tables illustrate how the Nusselt number ( Gr^(-1/5) ) and skin friction (Gr^(1/5) C_f ) affect several physical variables related to the Casson nanofluid flow. One of the most important conclusions is that (Gr^(1/5)  C_f) decreases and (Gr^(-1/5)  increases when the Casson parameter increases.