Background: This research investigates the magnetohydrodynamic flow of a chemically reactive Casson hybrid nanofluid within a Sodium Alginate base, flowing over a curved stretching surface in a porous environment. The analysis accounts for internal heat sources, magnetic field influence, reactive diffusion, and thermophoretic effects to improve thermal performance. Methodology: The model considers transport effects, including Brownian motion, thermophoresis, internal heating, viscosity, and Arrhenius-type reactions. Similarity transformations reduce the governing PDEs to ODEs, which are solved using MATLAB’s BVP4c. The sensitivity of thermal and flow parameters is further evaluated using Multiple Linear Regression (MLR). Core findings: Results indicate that elevating the Biot number can boost the Nusselt number by approximately 42%, emphasizing improved heat transfer at the surface. The heat generation parameter exerts the strongest effect on thermal output, with a sensitivity index peaking at 2.8673. Furthermore, the curvature parameter plays a significant role in modulating surface shear. The sensitivity analysis pinpoints parameter combinations that yield optimal performance, reinforcing the utility of machine learning in fluid system optimization. Validation: Comparisons to previous studies demonstrate excellent agreement, as deviations remain under 1.6% for skin friction and 2.3% for the Nusselt number when the curvature parameter equals zero. These results affirm the robustness of the applied transformations and numerical approach. Furthermore, the MLR model perfectly matches numerical outputs, reaching an <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11043_2025_9812_Article_IEq1.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="21" /> </InlineMediaObject> <EquationSource Format="MATHML"><math> <msup> <mi>R</mi> <mn>2</mn> </msup> </math></EquationSource> <EquationSource Format="TEX">$R^{2}$</EquationSource> </InlineEquation> score of 1.0, confirming predictive accuracy. Applications: The findings reference engineering applications, specifically solar thermal systems, HVAC equipment, and miniaturized heat exchangers. By combining numerical modeling with machine learning, this study offers a reliable approach for designing and controlling energy-efficient thermal systems under varying physical conditions.

Thermal and flow analysis of chemically reactive Casson hybrid nanofluids with machine learning validation

期刊论文
发表日期：

2025年8月20日

查看全文

P. Priyadharshini,
M. Sowndharya,
Ali J. Chamkha

摘要

Background: This research investigates the magnetohydrodynamic flow of a chemically reactive Casson hybrid nanofluid within a Sodium Alginate base, flowing over a curved stretching surface in a porous environment. The analysis accounts for internal heat sources, magnetic field influence, reactive diffusion, and thermophoretic effects to improve thermal performance. Methodology: The model considers transport effects, including Brownian motion, thermophoresis, internal heating, viscosity, and Arrhenius-type reactions. Similarity transformations reduce the governing PDEs to ODEs, which are solved using MATLAB’s BVP4c. The sensitivity of thermal and flow parameters is further evaluated using Multiple Linear Regression (MLR). Core findings: Results indicate that elevating the Biot number can boost the Nusselt number by approximately 42%, emphasizing improved heat transfer at the surface. The heat generation parameter exerts the strongest effect on thermal output, with a sensitivity index peaking at 2.8673. Furthermore, the curvature parameter plays a significant role in modulating surface shear. The sensitivity analysis pinpoints parameter combinations that yield optimal performance, reinforcing the utility of machine learning in fluid system optimization. Validation: Comparisons to previous studies demonstrate excellent agreement, as deviations remain under 1.6% for skin friction and 2.3% for the Nusselt number when the curvature parameter equals zero. These results affirm the robustness of the applied transformations and numerical approach. Furthermore, the MLR model perfectly matches numerical outputs, reaching an $R^{2}$ $R^{2}$ score of 1.0, confirming predictive accuracy. Applications: The findings reference engineering applications, specifically solar thermal systems, HVAC equipment, and miniaturized heat exchangers. By combining numerical modeling with machine learning, this study offers a reliable approach for designing and controlling energy-efficient thermal systems under varying physical conditions.