Cova-PINN: Cross-Domain Conservation Physics-Informed Neural Network for Fluid-Solid Conjugate Heat Transfer in Complex Geometries

2026-10-07

AI Summary

This study addresses the inaccuracies in energy transfer and outlet temperature prediction arising from independent sampling in multi-domain Physics-Informed Neural Networks (PINNs) for fluid–solid conjugate heat transfer. To overcome this limitation, we propose Cova-PINN, a framework that innovatively incorporates conservation support into multi-domain PINNs. By jointly optimizing local composite control volume balances with global paired-wall closure conditions, the method effectively aligns cross-domain thermal interaction pathways and extends naturally to complex triply periodic minimal surface (TPMS) geometries. Experimental results demonstrate that, compared to baseline models, Cova-PINN reduces outlet temperature errors and device-level closure errors by 37.7% and 60.2%, respectively. These improvements significantly enhance full-field accuracy and the reliability of thermal load predictions.

Abstract

Multi-domain physics-informed neural networks (PINNs) flexibly model medium-specific representations to solve fluid--solid conjugate heat transfer (CHT). However, standard multi-domain PINNs enforce governing equations and interface conditions on separately sampled domain supports, which can yield plausible temperature fields but inaccurate end-to-end energy transfer and outlet temperatures. We propose Cova-PINN, a multi-domain PINN framework that aligns conservation support with thermal interaction paths in complex geometries. Cova-PINN jointly optimizes cross-domain composite control-volume balances at the local scale and paired-wall closure at the global exchanger scale. We evaluate Cova-PINN on four triply periodic minimal surface (TPMS) heat exchangers and a geometrically distinct DualMS design against CHT-specific, optimization-oriented, and complex-geometry PINN baselines under a common protocol. Relative to the closest baseline, MUSA-PINN-CHT, Cova-PINN reduces average outlet-temperature and device-level closure errors across the four TPMS topologies by $37.7\%$ and $60.2\%$, respectively, while also improving full-field and heat-duty accuracy, with consistent gains on DualMS.

Problem

• Conjugate Heat Transfer
• Physics-Informed Neural Networks
• Complex Geometries
• Energy Conservation
• Multi-domain Modeling

Innovation

• Physics-Informed Neural Networks
• Conjugate Heat Transfer
• Cross-Domain Conservation
• Control-Volume Balance
• Complex Geometries