Thermal analysis of an electronics enclosure: coupling flow network modeling (FNM) and computational fluid dynamics (CFD)

The objective of the present study is to perform thermal analysis for the design of a complex electronic cabinet used as a high speed Internet subscriber device. The novelty of the paper lies in the approach used for analysis of the flow distribution in the system of interest. It utilizes Flow Network Modeling (FNM) and Computational Fluid Dynamics (CFD) in complementary and interactive manner for quick and accurate thermal analysis of the entire system. The enclosure consists of thirteen PCB's and two axial fans. In the analysis of the flow distribution, the flow through the passages formed by adjacent PCBs along with the inlet region is analyzed using the CFD technique to generate flow impedance characteristics. A flow network model of the entire system is then constructed by interconnecting the various components of the system in a manner that represents the paths followed by the air as it moves through the system. The CFD-based impedance characteristics in combination with the loss characteristics available from handbooks are ascribed to individual components in the network model. The FNM-based analysis of the entire system accounts for the interaction of the fan curve and the flow impedances to predict the flow distribution of air throughout the system. These results are, in turn, used to provide boundary conditions in the CFD analysis for the prediction of detailed flow distribution in individual card passages in order to obtain a thermal map of the PCBs. The predicted flow rates through the individual card passages are within 10% of the experimentally measured values. The analysis approach couples the power of CFD with the speed and flexibility of the FNM technique to enable accurate prediction of the flow distribution throughout the system in the most efficient manner. Further, the modularity of the proposed approach allows quick and scientific examination of the design changes such as use of different filters, screens, or fans and easy identification of performance-limiting components. The complementary use of CFD and FNM reduces the time required for thermal analysis by an order of magnitude over the approach that uses only the CFD technique.