Convective flow dynamics with suspended carbon nanotubes in the presence of magnetic dipole: Intelligent solution predicted Bayesian regularization networks

Abstract

This paper narrates an investigation for the convective flow dynamics involving single and multiwall carbon nanotubes as well as methanol hybrid nanofluids in the presence of magnetic dipole that has many applications in pharmacy, medication, electronics, energy, and biomedical engineering. Mathematical modeling has been performed for the conversion of physical model in to set of mathematical equations. The derived PDEs are reduced to system of ODEs by utilizing the suitable similarity variables. Adams numerical solver is utilized to solve the reduced differential systems for the formulation of reference datasets nanofluidic model for sundry scenarios by variation of ferromagnetic interaction parameter, Biot number, thermal radiation parameter and viscous dissipation parameter. Intelligent solution predicted strategy via supervised artificial neural networks (ANNs) with a Bayesian Regularization (BR), i.e., ANN-BR, is employed to get immediate numerically approximated results via simulations on training, testing and validation samples from reference datasets of complex geometry. Numerical outcomes of exhaustive simulations, in term of negligible error, i.e., MSE ≤ 10−12, unit value of regression metric, i.e., R ≈ 1 and histogram with majority of instance, i.e., ≥ 90%, closed to reference line, certainly endorsed/verified the correctness of ANN-BR for solving each variation in magnetite nanofluidic system model. Validation of the presented computational model, analysis of flow instabilities, effects of heat transfer properties and optimization of suspended nanoparticles are the conducted perquisites of the study.