Machine Learning technique to simulate the impact of nonlinear thermal radiation and electric field on hydrothermal features of dielectric nano-suspension

Background

Non-traditional designs were found to provide more effective heat transfer because of the availability of modern manufacturing techniques. This study examines porous nanofluid flow behavior inside a corrugated enclosure using machine learning (ML) and artificial neural networks (ANN). The influence of electrohydrodynamics (EHD) and nonlinear thermal radiation are conducted within a dielectric TiO₂H₂O nano-liquid porous enclosure with three cooling channels. In the proposed ML model Rayleigh number, conduction-radiation parameter, Eckert number, Lorentz force number, charge diffusivity number, surface-temperature parameter, Darcy number, and domain’s length along the vertical axis are used as input, while their associated numeral estimated values of electric charge density, temperature, stream function, velocities, and Nusselt number are considered as output targets for the ML model.

Methods

This investigation applies one of the powerful numerical methods called the finite element method (FEM) to gain an errorless solution of governing equations (GEs) of buoyancy-driven dielectric nano-suspension in a porous domain with channels inside and influenced by non-linear thermal radiation and EHD. The use of ANN-based machine learning techniques to determine the thermal behavior is a novel approach in this study. The performance of the approach is measured using measures such as coefficient of regression and mean squared error.

Significant findings

This study shows that the rate of heat transmission increasesby 1.74 times with the rise of the Rayleigh number (Ra), while the Lorentz force number (SE) shows opposite trend. The Eckert number (Ec) reduces the Nuaveby 13.28 %. The stream velocity is effectively increased by 31.64 % by the thermal radiation while moving from Rd=1 to 2. Increasing the Darcy number (Da) enhances the flow circulation leading to a 9.95 % increase in the magnitude of stream function. Results demonstrate that the used ANN model produces minimal mean squared error (MSE) with high accuracy. The study findings contribute to the development of efficient heat transfer systems and advancements in the fields of heat exchangers, electronic and battery cooling systems, and thermal and chemical engineering.