<p>Energy storage devices in thermal solar plants play a crucial role in controlling the energy and power demand. Their performance is significantly influenced by the thermal capacity of the materials used. Motivated by the growing need for enhanced thermal energy efficiency, a Williamson ternary hybrid nanofluid is used to examine the non-steady magnetohydrodynamic (MHD) flow through a porous stretching cylinder containing gyrotactic microorganisms. Physics-informed neural network (PINN) with GaussSwish hybrid activation function is utilized in this study. The network minimizes the residuals of the governing equations together with boundary constraints using automatic differentiation and the NADAM optimizer until it converges to the optimal loss. The effects of different flow parameters on temperature, momentum, concentration, and motile density are analyzed. Magnetic and electric field parameters show a drop in the momentum profile, whereas an inverse trend is noticed in the temperature profile. Weissenberg number, curvature, and heat sink parameters contribute to elevate the temperature. Schmidt number lowers the concentration profile; on the other hand, the curvature parameter exhibits an opposite relation. Peclet and bioconvection Lewis number cause the motile microorganism density to decline. Ternary hybrid nanofluid achieves up to <InlineEquation ID="IEq1"> <EquationSource Format="TEX">$24.3\%$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>24.3</mn> <mo>%</mo> </mrow> </math></EquationSource> </InlineEquation> greater heat transfer, <InlineEquation ID="IEq2"> <EquationSource Format="TEX">$29.7\%$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>29.7</mn> <mo>%</mo> </mrow> </math></EquationSource> </InlineEquation> mass transfer, and <InlineEquation ID="IEq3"> <EquationSource Format="TEX">$34.1\%$</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mn>34.1</mn> <mo>%</mo> </mrow> </math></EquationSource> </InlineEquation> higher motile microorganisms density than the hybrid nanofluid, confirming its potential for advanced thermal energy storage systems. The results further show the effectiveness of physics-informed neural networks in handling complex fluid problems.</p>

Scientific computing for thermal analysis in ternary hybrid nanofluid flow through cylinder with gyrotactic microorganisms: thermal storage applications

期刊论文
发表日期：

2026年1月20日

查看全文

Ahmad,
Husna Zafar,
Muhammad Jawad,
Usama Arif,
Nayyar Ijaz Dar,
Muhammad Noveel Sadiq

摘要

Energy storage devices in thermal solar plants play a crucial role in controlling the energy and power demand. Their performance is significantly influenced by the thermal capacity of the materials used. Motivated by the growing need for enhanced thermal energy efficiency, a Williamson ternary hybrid nanofluid is used to examine the non-steady magnetohydrodynamic (MHD) flow through a porous stretching cylinder containing gyrotactic microorganisms. Physics-informed neural network (PINN) with GaussSwish hybrid activation function is utilized in this study. The network minimizes the residuals of the governing equations together with boundary constraints using automatic differentiation and the NADAM optimizer until it converges to the optimal loss. The effects of different flow parameters on temperature, momentum, concentration, and motile density are analyzed. Magnetic and electric field parameters show a drop in the momentum profile, whereas an inverse trend is noticed in the temperature profile. Weissenberg number, curvature, and heat sink parameters contribute to elevate the temperature. Schmidt number lowers the concentration profile; on the other hand, the curvature parameter exhibits an opposite relation. Peclet and bioconvection Lewis number cause the motile microorganism density to decline. Ternary hybrid nanofluid achieves up to $24.3\%$ $24.3 %$ greater heat transfer, $29.7\%$ $29.7 %$ mass transfer, and $34.1\%$ $34.1 %$ higher motile microorganisms density than the hybrid nanofluid, confirming its potential for advanced thermal energy storage systems. The results further show the effectiveness of physics-informed neural networks in handling complex fluid problems.