Work

Neural network-based investigation of MHD radiative flow of ternary hybrid nanofluid in porous media

Year: 2025

Type: article

Abstract: Scientific research in many disciplines, such as aerospace engineering, industrial cooling, energy systems, biomedicine, geophysics for oil recovery, and geothermal energy extraction, can greatly benefit from the magnetohydrodynamic radiative flow of a ternary hybrid nanofluid (MHD-RFTHN) over a rotating disk in a porous media. Artificial recurrent neural network enhanced with Bayesian regularization (ARNN-BR) is also used in advanced technologies like neurological disorder detection, speech recognition, mechanism translation, and automatic text translation. This work proposed a supervised learning approach based on ARNN-BR method for obtaining numerical solutions of MHD-RFTHN by considering the rotating disk in a porous medium. It investigates. the thermophysical characteristics of ternary nanomaterials ( C u , M n Z n F e 2 O 4 , and A g ) dispersed in pure water as the base fluid. Compared to traditional numerical methods, ARNN-BR increases accuracy and stability as well as lowers computational cost. Intelligent computing facilitates enhanced prediction of fluid flow behavior in response to variations of numerous physical parameters while aptly capturing the influences of porosity, MHD, thermal radiation, and suction or injection for a more comprehensive analysis. The system of governing partial differential equations (PDEs) is transformed into a dimensionless ordinary differential equation (ODEs) using similarity transformations and then solved using the ARNN-BR technique. The proposed method is trained on numerical data generated by varying multiple physical parameters, including the magnetic parameter, nonlinear thermal radiation, porous parameter, suction or injection parameter, volume fraction, temperature ratio parameter, and Prandtl number. The ARNN-BR technique has been validated using performance, regression index evaluation, error histogram analysis, autocorrelation investigations, and convergence curve analysis. With a minimal mean square error ranging from 10 − 10 to 10 − 3 , the results demonstrate the method's high accuracy and effectiveness. The increase in porosity parameter decreased the radial velocity. One significant finding is that the temperature profile of the ternary hybrid nanofluid (THNF) can be improved by increased nanoparticle volume ratio. The temperature of the fluid also increases with an increasing temperature ratio and a thermal radiation parameter, while the fluid temperature decreases with an increasing Prandtl number. Predictions from the ARNN-BR process indicated that the method can efficiently estimate the fluid behavior most accurately in the MHD radiative flow situations.

Source: Journal of Radiation Research and Applied Sciences

DOI: 10.1016/j.jrras.2025.101448