Modeling of heat transfer correlations for supercritical n-decane pyrolysis reactive flow based on a decoupling and dimensionality-reduction strategy

In regenerative cooling of hypersonic vehicles, traditional heat transfer correlations fail under supercritical pyrolysis due to neglected chemical heat absorption and thermochemical coupling effects. Starting from the energy equation for fully developed turbulent flow, a semi-analytical model is developed to identify and quantify the contributions of convective transport, conductive transport, species-enthalpy diffusion, and chemical reaction terms to reaction-enhanced heat transfer. Dimensional analysis reveals three governing criteria: the chemical heat absorption fraction 𝜉 c h e m , reaction temperature sensitivity 𝛩 ∗ , and Reynolds number 𝑅 ⁢ 𝑒 𝐷 . An artificial neural network surrogate model is built, and sensitivity analysis reduces these to two principal parameters ( 𝜉 c h e m and 𝑅 ⁢ 𝑒 𝐷 ), enabling the fitting of explicit linear and quadratic polynomial correlations. Weighted least-squares fitting shows that the quadratic correlation, which captures nonlinear enhancement, achieves an average relative error of 1.56% in high-response regions (global 𝑅 2 = 0 . 8 3 3 8 ), significantly outperforming the linear form. Validation against simulations confirms good accuracy and generalizability under unseen operating conditions, providing an effective tool for regenerative cooling design. The proposed framework reveals that reaction-induced heat-transfer enhancement originates from both chemical heat absorption redistribution and reaction-driven modification of near-wall thermophysical properties.