Comparative Analysis of Artificial Neural Networks and Response Surface Methods for Mixed Convective Heat Transfer in a Rectangular Channel with Heated Diamond-Shaped Obstacles

Articles in Press

Document Type : Research Paper

Authors
Noor-A- Siddiki
Shahinur Islam
Farjana Habiba
Stamford University Bangladesh
10.22055/jacm.2026.49656.6137
Abstract
The mixed convective heat transfers and flow characteristics of a non-Newtonian power-law hybrid nanofluid, 〖Al〗_2 O_3 CNT/water) in an open rectangular channel containing two heated diamond-shaped blocks, with a bottom inlet, two top outlets, and isothermal cold side walls, are investigated numerically in the present study. The non-dimensional controlling transport equations are computed numerically using COMSOL Multiphysics while the left vertical wall is kept at an identical cold temperature (Tc). The analysis considers a wide range of controlling parameters: nanoparticle volume fraction (ϕ = 0.00,0.01,0.02,0.04), Richardson number (Ri = 1.0,2.0,4.0,6.0)), Reynolds number (Re = 50,100,200,400), Power law index (n = 0.6,0.8,1.0,1.2), and inlet/outlet width (Win = Wout = 0.1). The results exhibit that the governing parameters have a substantial effect on the thermal performance of the rectangular channel: higher values of the power-law index (n) reduced heat transfer, while higher values of the nanoparticle volume fraction (ϕ), Richardson number (Ri), and Reynolds number (Re) increase it. While comparative analyses using Response Surface Methodology (RSM) and Artificial Neural Networks (ANNs) confirm their efficacy in accurately predicting the average Nusselt number, Nu_av detailed flow and thermal characteristics are presented, providing effective tools for the design and optimization of hybrid nanofluid-based thermal systems.
Keywords
Nanofluids
Streamline
Isotherm
Power law index
Comsol Multiphysics
Subjects
Journal of Applied and Computational Mechanics

Articles in Press, Accepted Manuscript
Available Online from 30 September 2026