Highlights • Thermal transport of a temperature-dependent Al2O3–EG/Water nanofluid is investigated. • A curved-corner enclosure with a cold inner cylinder is examined under HFBC and IBC thermal conditions. • Nuavg rises notably with higher fluid temperature ( T 0 ) and nanoparticle loading ( ϕ ). • Heat transfer performance (ECOP) is lower in IBC compared to HFBC in all conditions. • Sensitivity analysis shows Ra significantly impacts Nuavg, followed by T 0 , ϕ , and r . Reliable natural convection is essential in passive thermal systems, heavily influenced by enclosure geometry and fluid temperature. This study analyzes the convective heat transfer characteristics of a nanofluid with Al 2 O 3 nanoparticles in a 20:80 ethylene glycol-water mixture across varying temperatures around a cold body. By integrating experimental data on thermal conductivity and dynamic viscosity, the research aims to provide a more accurate depiction of nanofluid behavior in natural convection. Using the finite element method within the Galerkin weighted residual framework, the study solves nonlinear partial differential equations subject to two boundary conditions: isothermal (IBC) and constant heat flux (HFBC). Quantitative assessments involve calculating average Nusselt number ( Nu avg ), average entropy ( S avg ), and average Bejan number ( Be avg ), while varying parameters like nanoparticle volume fraction ( ϕ = 0 % ∼ 1.5 % ) , nanofluid temperature ( T 0 = 20 o C ∼ 60 o C ) , cylinder size ( r = 0.10 ∼ 0.20 ) , and Rayleigh number ( R a = 10 3 ∼ 10 6 ) . Results show that as Ra increases, both Nu avg and S avg rise. For example, increasing Ra from 10 3 to 10 6 at T 0 = 35 ∘ C , r = 0.15 , and ϕ = 1.5 % results in a 99.11% increase in Nu avg under HFBC and a 135.49% increase under IBC, with IBC consistently yielding higher values. Additionally, increasing T 0 enhances Nu avg ; specifically, varying T 0 from 20 ∘ C to 60 ∘ C at R a = 10 6 causes a 21.73% increase in HFBC and a 22.35% increase in IBC at r = 0.15 and ϕ = 1.5 % . Additionally, the study employs Response Surface Methodology (RSM) to analyze parameter interactions, yielding high predictive accuracy ( R 2 = 0.9815 for HFBC and R 2 = 0.9891 for IBC). Sensitivity analysis indicates that Ra is the most critical factor, followed by T 0 , ϕ , and r . The integration of numerical modeling and response surface methodology (RSM) optimization establishes a robust framework for predicting and enhancing thermal performance in curved corner geometries that utilize temperature-dependent nanofluids.
Dipu et al. (2026) studied this question.