Scholarlink

<p>This study presents a comprehensive mathematical modeling, numerical simulation, and artificial neural network (ANN)-based prediction of steady natural convection within a square porous enclosure saturated with Al₂O₃–water nanofluid and containing internally heated obstacles of various geometries <b>(</b>square, triangular and thin plate<b>)</b>. Momentum transport within the porous medium is represented through the Darcy–Brinkman extension of the two‑dimensional, steady, incompressible Navier–Stokes equations. The thermophysical characteristics of the Al₂O₃ water nanofluid is determined using the Tiwari–Das formulation, which accounts for nanoparticle concentration. The governing relations are reformulated in a nondimensional vorticity–stream function framework and discretized via a finite difference method with second‑order spatial precision. In addition to the numerical simulation, an artificial neural network (ANN) model is developed to predict the local Nusselt number distribution based on the numerical dataset. The current work generalizes previous studies by including comparative assessments of square, triangular, and thin-plate obstacles in porous nanofluid enclosures and also utilizing a machine learning approach to optimize heat transfer via ANNs. A detailed series of parametric investigations is carried out to evaluate how the Rayleigh number (Ra), Darcy number (Da), nanoparticle volume fraction (ϕ), and the geometry of the internal obstacle influence the fluid flow and heat transfer behavior. The results reveal that increasing Ra or Da enhances buoyancy-driven circulation and heat transfer, whereas increasing doping with Al₂O₃ nanoparticles i.e. greater volume fraction, ϕ, suppresses convection due to viscosity-induced damping. Among the geometries considered, the square obstacle yields the strongest vortical motion and the highest average Nusselt number, followed by the flat-plate and triangular configurations. Quantitatively, a square internal obstacle produces 10–15% higher heat transfer than a thin plate obstacle and 20–25% higher than a triangular obstacle across all examined values of Ra, Da, and . The study establishes that geometric sharpness and porous permeability jointly determine the efficiency of convective transport in nanofluid-saturated porous media enclosures and provides a validated numerical framework for optimizing, for example, hybrid fuel cells and porous thermal systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Mathematical modelling, numerical simulation and artificial neural network (ANN)-based model for nanofluid convection in a porous enclosure with obstacles

  • 期刊论文

  • 发表日期：

    2026年6月30日

查看全文
  • K. Venkatadri,
  • O. Anwar Bég,
  • T. A. Bég

摘要

This study presents a comprehensive mathematical modeling, numerical simulation, and artificial neural network (ANN)-based prediction of steady natural convection within a square porous enclosure saturated with Al₂O₃–water nanofluid and containing internally heated obstacles of various geometries (square, triangular and thin plate). Momentum transport within the porous medium is represented through the Darcy–Brinkman extension of the two‑dimensional, steady, incompressible Navier–Stokes equations. The thermophysical characteristics of the Al₂O₃ water nanofluid is determined using the Tiwari–Das formulation, which accounts for nanoparticle concentration. The governing relations are reformulated in a nondimensional vorticity–stream function framework and discretized via a finite difference method with second‑order spatial precision. In addition to the numerical simulation, an artificial neural network (ANN) model is developed to predict the local Nusselt number distribution based on the numerical dataset. The current work generalizes previous studies by including comparative assessments of square, triangular, and thin-plate obstacles in porous nanofluid enclosures and also utilizing a machine learning approach to optimize heat transfer via ANNs. A detailed series of parametric investigations is carried out to evaluate how the Rayleigh number (Ra), Darcy number (Da), nanoparticle volume fraction (ϕ), and the geometry of the internal obstacle influence the fluid flow and heat transfer behavior. The results reveal that increasing Ra or Da enhances buoyancy-driven circulation and heat transfer, whereas increasing doping with Al₂O₃ nanoparticles i.e. greater volume fraction, ϕ, suppresses convection due to viscosity-induced damping. Among the geometries considered, the square obstacle yields the strongest vortical motion and the highest average Nusselt number, followed by the flat-plate and triangular configurations. Quantitatively, a square internal obstacle produces 10–15% higher heat transfer than a thin plate obstacle and 20–25% higher than a triangular obstacle across all examined values of Ra, Da, and . The study establishes that geometric sharpness and porous permeability jointly determine the efficiency of convective transport in nanofluid-saturated porous media enclosures and provides a validated numerical framework for optimizing, for example, hybrid fuel cells and porous thermal systems.