Probabilistic modeling of coupled heat transfer: A step towards optimization based on multiphysics Monte Carlo simulations

This article presents a modular framework allowing to construct probabilistic models of coupled heat transfer problems in complex systems. First, a substructuring approach has been applied to formalize the problem. This process allowed for the coupling of physical field submodels, in our case temperature and radiative intensity. Each physical model was established according to the conservation law inside of its domain (solid and fluid) and the continuity laws at interfaces. Then, these models have been rewritten from the deterministic point of view to a probabilistic one. This enables a recursive Monte Carlo algorithm to estimate the desired values. After a validation stage, against academic cases, this framework is applied to examples emulating heat transfer in buildings. This approach presents a major beneficial behavior for complex systems optimization: only the influential parts of the problem have an effect on the computational time. These regions are automatically identified in a self-adaptive way, even in intricate or extensive geometries. •Framework to construct probabilistic models of coupled heat transfer is proposed.•Coupled heat transfer problems are solved by a single recursive Monte Carlo algorithm.•Framework has been validated against analytical results.•Computational time depends on the only inuential parts of the problem.