Thermal-Flow Network Modeling for Virtual Prototyping of Power Electronics

Abstract

Fast and accurate thermal analysis is critical in designing power electronics converters through virtual prototyping to achieve high power density, efficiency, etc. Virtual prototyping can save cost and time in comparison to hardware prototyping and experimental test and can lead to an overall optimized design. For removing the heat in power converters effectively, forced convection, such as air cooling or liquid cooling, is usually adopted. To estimate the temperature distribution and assist the layout optimization without building physical prototypes, fast and accurate thermal prediction is required, which is an essential area in virtual prototyping of power electronics. In this article, a thermal-flow network (TFN) modeling method is proposed to solve the flow and temperature distribution in a power converter. The flow distribution is calculated by flow network consisting of flow resistances, flow branches, and sources. The temperature distribution is then estimated by the thermal network, where the related boundary conditions (heat transfer coefficient and local fluid temperature) are assigned using empirical equations. The fundamental and basic circuit elements for the TFN modeling method have been implemented in software, which helps to build the TFN modularly. The comparison between the proposed method, computational fluid dynamics (CFD), and experiment is provided to show the effectiveness of the TFN modeling method. The proposed modeling and analysis process can also be adapted for other forced cooling methods.