Abstract
The ability to identify heat sources and predict their temperatures across a variety of operating conditions is key in the design of a reliable electronic system such as a computer, a server, or any other system containing a printed circuit board assembly (PCA). The goal of this thesis is to explore a process on how one should approach thermal design to improve the probability to design a reliable electronic system. During the process, one should be able to define system requirements, understand operational environment conditions, choose an appropriate cooling method, and understand how the system will behave with these defined constraints. Using these constraints, one can perform an initial study with hand calculations to help determine a high-level feasibility of the design. Computational Fluid Dynamics (CFD) can then be used to create a more complex study that will provide detailed results. Finally, a thermal mock-up should be built to compare experimental results to CFD results or to check how well the empirical data correlates to the theoretical data. Improvements to the CFD model can increase the accuracy of the correlated data and increase the confidence of results for future optimization studies. A natural convection study of a heat sink was used to demonstrate this thermal design process. Multiple fin heights of a heat sink at different ambient temperatures were analyzed by hand calculations, CFD simulations were performed using ANSYS Icepak 14.0, and a mock-up was built using a copper slug with a Minco heater attached to simulate a processor on a PCA. Results from this experiment demonstrate how by using known correlated data to future investigations one can improve the accuracy of results. That is why it is always a benefit to assist CFD simulations with empirical data. Empirical data helps improve the accuracy of a CFD model and adds credibility to the theoretical results. This allows for improved optimization studies and helps reduce design cycle time.