Bayesian-driven thermal adaptation design of multi-channel regenerative cooling structures under non-uniform scramjet combustor heat loads
Year: 2026
Type: article
Abstract: Thermal adaptation of regenerative cooling structures is critical for scramjet combustors subjected to highly non-uniform heat loads. This study develops a Bayesian-driven framework for adapting a multi-channel regenerative cooling panel to a combustor-side heat load that is non-uniform in both the streamwise and lateral directions. The thermal boundary, represented by the adiabatic wall temperature and gas-side heat-transfer coefficient, is extracted from three-dimensional reacting-flow simulations. A fast model couples a one-dimensional plug-flow-reactor (PFR) coolant solver with two-dimensional solid heat conduction, and station-wise sequential Bayesian optimization redistributes the lateral channel positions by minimizing the area-weighted cross-sectional solid-temperature non-uniformity. The discrete solution is reconstructed as smooth three-dimensional channels and assessed by conjugate heat-transfer and thermo-mechanical simulations. Relative to the uniform baseline, the reconstructed layout reduces the coolant outlet-temperature standard deviation from 16.66 to 2.61 K and the maximum solid temperature from 972 to 943 K; its channel pressure drops remain within 6.27–6.49 kPa, and the maximum equivalent stress decreases modestly from 345 to 336 MPa. The results show that the framework improves thermal matching by redistributing the available cooling capacity toward local high-heat-load regions.
Source: Applied Thermal Engineering
Authors: Lang Luo, Yu Chen, Xiangyuan Li
Institution: Sichuan University
Language: English
Subfield: Computational Mechanics
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $3,660
Award: G255001997
OpenAlex ID: W7212305591