Experimental investigation and data-driven modeling of nanofluid pool boiling under rotational hypergravity

The coupling effects of rotational hypergravity (1–3.16g) and nanoparticle concentration on the nucleate pool boiling heat transfer of Al ₂ O ₃ -water nanofluids were experimentally investigated. Contrary to the monotonic deterioration reported in previous literature, this study reveals a distinct enhancement-deterioration non-monotonic trend in the heat transfer coefficient (HTC). The HTC peaks at approximately 1.41g due to buoyancy-assisted bubble detachment but deteriorates sharply at 3.16g. Mechanism analysis elucidates that this degradation is governed by hypergravity-induced sedimentation, where Stokes settling overrides Brownian diffusion, forming a compact thermal resistance layer. Furthermore, the nanoparticle concentration exhibits a complex non-linear impact: while intermediate concentrations cause surface clogging, an optimal concentration of 0.015 wt.% yields maximum enhancement by reconstructing the surface into a three-dimensional (3D) porous structure with strong capillary wicking effects. To address the failure of classical semi-empirical correlations in this coupled regime, machine learning (ML) algorithms were implemented. The XGBoost model demonstrated exceptional fidelity with coefficient of determination (R ² ) = 0.998, root mean square error (RMSE) = 1.2% and mean absolute error (MAE) = 0.8%, successfully capturing the complex non-linear boundaries of hypergravity and surface fouling. These findings provide robust theoretical insights and predictive tools for aerospace thermal management systems.