Design & Analysis of Simulation Experiments (DASE) Approach to Circuit Card Assembly Thermal Analysis

Abstract

A multi-parameter sensitivity study was performed on a circuit card assembly using Design & Analysis of Simulation Experiments (DASE), a powerful tool to determine the statistical effects of input design parameters. Several input parameters were screened to determine their impact on thermal margin, and the results indicate that the dominant factors affecting thermal margin are the top cooling method (bare, heat plate and thermoelectric), the thermal interface material characteristics, power dissipations, and substrate through-plane thermal conductivity. The finite element model was simplified to a linear regression model using a modified Latin Hypercube Monte Carlo sampling method. This model allowed for many more variable combinations to be examined and to explore variable interaction. Although the particular results of this case study are important for this hardware design, even more important is the demonstration of the DASE process to model hardware and identify the statistical contribution of multiple design parameters to thermal performance. I. Introduction The DASE process is very similar to the Design of Experiments (DOE) process in many respects. Both processes are useful in statistically determining the relative importance of design factors or variables on resulting metrics. Both are used to determine not only the significance of individual effects, but also the significance of how design factors interact with one another to impact the resulting metrics. The primary difference between the two processes is that DASE is applied to analytical and numerical simulations, while DOE is applied to experiments. This is important because random variations do not necessarily exist in simulations as they do in experiments. If the exact same inputs are used during two experimental runs, slightly different results can be produced due to random variations. Simulations, however, are far more deterministic and identical results can be obtained in two different simulation runs provided the same inputs are used. The DASE process can be applied to any simulation model. In this particular case study, the DASE process is shown to be an effective analysis element of a robust design process. The purpose of this study is to illustrate how the DASE process can be applied to thermal analyses and sensitivity studies. The particular results of this study, such as which design parameters have the most significant effect, are not applicable to general circuit card thermal analyses.