Scholarlink

<p>This study numerically investigates thermosolutal magnetoconvection and entropy generation in a sinusoidal plus-shaped enclosure containing a centrally embedded wavy cold cylinder filled with a water-based hybrid nanofluid and subjected to a uniform horizontal magnetic field. The finite element method (FEM) implemented in COMSOL Multiphysics is employed to solve the steady, laminar, incompressible governing equations under the Boussinesq approximation, and the numerical model is validated against benchmark solutions. The effects of key dimensionless parameters, including the Rayleigh number, Hartmann number, hybrid nanoparticle volume fraction, Lewis number, buoyancy ratio, and cylinder position, on flow structure, heat and mass transfer, and thermodynamic irreversibility are systematically examined. The analysis of results indicates that there is a notable improvement in convection with the increment of the Rayleigh number, which is manifested in a 78.8% boost of the average Nusselt number and a 213% increase in the Sherwood number. Conversely, using greater magnetic fields significantly increases entropy formation while reducing fluid velocity and heat transmission by 29.6%. Stronger gradients in the presence of hybrid nanoparticles have the dual impact of enhancing thermal performance and raising entropy generation. When Rayleigh and Hartmann numbers rise, entropy analysis shows that thermal irreversibility gives way to viscous irreversibility. Apart from the numerical simulations, an artificial neural network (ANN) model was developed to estimate the average Nusselt number using the Rayleigh number, Hartmann number, and nanoparticle concentration as inputs. As a result, there is almost perfect correlation and an extremely low mean squared error (3.82 × 10⁻<sup>1</sup>). The findings not only provide important physical insights but also make it possible to more precisely predict the design of energy-efficient thermal systems, such as electronics cooling devices, energy storage units, and chemical reactors where the effects of heat, solute, and magnetism are coupled.</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

FEM–ANN-based thermosolutal and entropy analysis of MHD hybrid nanofluid flow in complex wavy geometry

  • 期刊论文

  • 发表日期：

    2026年3月6日

查看全文
  • Noor Zeb Khan,
  • Asad Ullah,
  • Rawda A. I. Suliman,
  • Mahmoud M. Selim

摘要

This study numerically investigates thermosolutal magnetoconvection and entropy generation in a sinusoidal plus-shaped enclosure containing a centrally embedded wavy cold cylinder filled with a water-based hybrid nanofluid and subjected to a uniform horizontal magnetic field. The finite element method (FEM) implemented in COMSOL Multiphysics is employed to solve the steady, laminar, incompressible governing equations under the Boussinesq approximation, and the numerical model is validated against benchmark solutions. The effects of key dimensionless parameters, including the Rayleigh number, Hartmann number, hybrid nanoparticle volume fraction, Lewis number, buoyancy ratio, and cylinder position, on flow structure, heat and mass transfer, and thermodynamic irreversibility are systematically examined. The analysis of results indicates that there is a notable improvement in convection with the increment of the Rayleigh number, which is manifested in a 78.8% boost of the average Nusselt number and a 213% increase in the Sherwood number. Conversely, using greater magnetic fields significantly increases entropy formation while reducing fluid velocity and heat transmission by 29.6%. Stronger gradients in the presence of hybrid nanoparticles have the dual impact of enhancing thermal performance and raising entropy generation. When Rayleigh and Hartmann numbers rise, entropy analysis shows that thermal irreversibility gives way to viscous irreversibility. Apart from the numerical simulations, an artificial neural network (ANN) model was developed to estimate the average Nusselt number using the Rayleigh number, Hartmann number, and nanoparticle concentration as inputs. As a result, there is almost perfect correlation and an extremely low mean squared error (3.82 × 10⁻1). The findings not only provide important physical insights but also make it possible to more precisely predict the design of energy-efficient thermal systems, such as electronics cooling devices, energy storage units, and chemical reactors where the effects of heat, solute, and magnetism are coupled.