Exploring the influence of nanolayer morphology on magnetized tri-hybrid nanofluid flow using artificial neural networks and Levenberg–Marquardt optimization

Abstract

This study investigates the impact of morphological nanolayers on the thermal conductivity of magnetized tri-hybrid nanofluid flow using artificial neural networks (ANNs) employing the Levenberg–Marquardt (LM-ANN) scheme. Nonlinear partial differential equations (PDEs) are transformed into nonlinear ordinary differential equations (ODEs) using similarity variables. A dataset covering various parameters such as nanolayer thickness, particle radius, magnetic parameter, expansion/contraction ratio, Schmidt number, volume fraction, and Prandtl number is generated using the Bvp5c numerical algorithm. LM-ANN is then used for testing, training, and validation to analyze approximate solutions for individual cases. Histogram analysis, mean square error (MSE), and regression investigations validate the LM-ANN. The proposed approach closely aligns with suggested and reference findings, showing an accuracy level within 10⁻¹⁰ range. Increasing nanolayer thickness leads to higher effective thermal conductivity. The Nusselt number increases notably with higher hybrid nanoparticle volume fraction. Laminar shape consistently exhibits superior heat transfer performance compared to other shapes. Increasing nanoparticle volume fraction enhances the Nusselt number and reduces skin friction coefficients (SFC). Industry recommendations include integrating nanolayer technologies for improved thermal efficiency.

Keywords:

• Artificial neural network
• magnetohydrodynamics
• morphological nanolayer
• tri-hybrid nanofluid

Disclosure statement

No potential conflict of interest was reported by the authors.