Dual-Mesh Algorithm for Transient Coupled Heat Transfer in Multi-Body Enclosed Systems

Currently, numerical algorithms for simulating transient coupled heat transfer in multi-body closed systems — involving solid heat generation and conduction, gas-phase natural convection, and surface radiation — still suffer from limitations in stability, convergence, and computational efficiency. To address these issues, a dual-mesh spatio-temporal region-decomposition and multi-time-step boundary coupling algorithm (ATUTE) is proposed. The proposed method divides the computational domain according to different heat transfer mechanisms, in which the fluid region is further decomposed into a dense fluid mesh and a sparse fluid network. Based on the disparity between the flow evolution timescale and the thermal response rate of solid components, the continuous unsteady natural convection process is decomposed into a series of quasi-steady stages. When the wall temperature rise of a solid exceeds a prescribed threshold ΔT, the algorithm triggers a steady-state solution of the dense fluid mesh sub-process to update the convective heat transfer characteristics for the current stage. The steady natural convection field obtained from the dense fluid mesh provides the flow-driving characteristics, while the sparse fluid network is used to construct a thermal equilibrium model coupled with the solid domain. By dynamically adjusting the time step according to the spatio-temporal response of each region, the global solution frequency is effectively reduced. This establishes a high-efficiency modeling framework that combines regional decoupling, boundary coupling, and multi-time-step advancement, achieving a balance between local accuracy and overall computational efficiency. Numerical validations show that the proposed ATUTE method exhibits excellent accuracy and efficiency in typical natural convection and multi-body coupled heat transfer cases. Compared with the commercial software Fluent, the maximum relative error of the average device temperature is only 0.17%, while the total computation time is approximately 1/63 of Fluent’s. These results demonstrate that ATUTE can significantly enhance the computational efficiency, scalability, and applicability of transient thermal simulations in complex closed systems without sacrificing accuracy.