A multiscale thermo-fluid computational model for a two-phase cooling system

• We model a novel two-phase cooling system for power electronics.
• The two-phase coolant flows in a network of pipes.
• Dissipated heat is removed through air convection.
• Multiscale techniques are used for model complexity reduction.
• Approximation with monotone and flux-conservative finite elements.

In this paper, we describe a mathematical model and a numerical simulation method for the condenser component of a novel two-phase thermosiphon cooling system for power electronics applications. The condenser consists of a set of roll-bonded vertically mounted fins among which air flows by either natural or forced convection. In order to deepen the understanding of the mechanisms that determine the performance of the condenser and to facilitate the further optimization of its industrial design, a multiscale approach is developed to reduce as much as possible the complexity of the simulation code while maintaining reasonable predictive accuracy. To this end, heat diffusion in the fins and its convective transport in air are modeled as 2D processes while the flow of the two-phase coolant within the fins is modeled as a 1D network of pipes. For the numerical solution of the resulting equations, a Dual Mixed-Finite Volume scheme with Exponential Fitting stabilization is used for 2D heat diffusion and convection while a Primal Mixed Finite Element discretization method with upwind stabilization is used for the 1D coolant flow. The mathematical model and the numerical method are validated through extensive simulations of realistic device structures which prove to be in excellent agreement with available experimental data.