New Higher-Order Super-Compact Finite Difference Scheme to Study Three-Dimensional Natural Convection and Entropy Generation in Non-Newtonian Fluids

TL;DR

This paper presents a novel high-order, super-compact finite difference scheme for 3D natural convection and entropy generation in non-Newtonian fluids, achieving high accuracy with minimal stencil points, and explores the effects of fluid properties and Rayleigh number on flow and heat transfer.

Contribution

It introduces the first higher-order finite difference scheme for 3D natural convection in non-Newtonian fluids, combining efficiency and accuracy in complex flow simulations.

Findings

Shear-thinning fluids exhibit higher convection efficiency.

Increasing Rayleigh number raises average Nusselt number.

The scheme shows excellent agreement with benchmark results.

Abstract

This work introduces a new higher-order super-compact (HOSC) implicit finite difference scheme for analyzing three-dimensional (3D) natural convection and entropy generation in non-Newtonian fluids. The proposed scheme achieves fourth-order accuracy in space and second-order accuracy in time while utilizing only seven directly adjacent grid points of the compact stencils at the $(n+1)$ time level, making it highly efficient and super compact. To the best of our knowledge, this is the first higher-order accurate finite difference scheme proposed to study 3D natural convection and entropy generation in non-Newtonian fluids. A time-marching technique is applied, where pressure corrections are addressed using a modified artificial compressibility method. The scheme is applied to the power-law model of non-Newtonian fluids, investigating both shear-thinning and shear-thickening effects on natural convection and entropy generation within a 3D cubic cavity. In the numerical simulations, the Prandtl number remains constant at $Pr = 1.0$, while various Rayleigh numbers ($Ra = 10^2$, $10^3$, $10^4$, $10^5$) and power-law indices ($n = 0.75$, $1.0$, $1.25$) are considered. Results are presented in terms of isotherms, streamlines, Nusselt numbers, Bejan numbers, and local entropy generation rates. The validation of the proposed scheme demonstrates excellent agreement with existing benchmark results. The numerical study reveals that the $n$ and $Ra$ have significant impacts on flow dynamics, heat transfer, and entropy generation. As $Ra$ increases, the maximum value of average Nusselt number ($Nu_{\text{avg}}$) also increases, whereas an opposite trend is observed with $n$ values. Shear-thinning fluids demonstrate the highest convection efficiency compared to Newtonian and shear-thickening fluids at any specific $Ra$.