Artificial neural networks analysis of magneto-boundary layer flow of partially ionized Prandtl tri-hybrid nanofluid with entropy generation over heated cylinder

Tri-hybrid nanofluid (THNF) using Prandtl fluid model (PNM) is considered under the influence of viscous dissipation, partially ionized, Hall current, thermal radiation, and nonlinear heat source across heated cylinder. The novel contribution of the present study is to investigate the influence of thermal radiation, entropy optimization, and nonlinear heat source across heated cylinder using artificial neural networks. The dimensionless differential equations are solved numerically by Keller Box method to generate dataset for different scenarios. Two different artificial neural network models have been established with different input to examine the local skin friction coefficient (LSFC) and local Nusselt number (LNN). In the present artificial neural network models (ANNMs), 70 %–15 %–15 % data is used for training-validation-testing, respectively. Similarly, 20 %–5 %–5 % dataset is utilized for training-validation-testing of LSFC and LNN. The effect of the emerging factors such as Hall currentδH, ionized slipδi, elastic factorδ, magnetic factorM, curvature parameterβ, and Prandtl first parameter ε are discussed through graphs. Comparative analysis for LSFC and LNN are documented for the ternary hybrid nanofluid (Al2O3+ CuO + TiO2 + PNF), hybrid nanofluid (Al2O3+ CuO + PNF), and nanofluid (Al2O3 + PNF) and found that profile for (Al2O3+ CuO + TiO2 + PNF) is more significant compared to others. For the sake of validation, the present work is compared with the published work.