Prognostics of Solder Joint Reliability Under Vibration Loading Using Physics of Failure Approach.

Physics-of-failure (PoF) is an approach that utilizes knowledge of a product's life cycle loading and failure mechanisms to perform reliability modeling, design, and assessment. Prognostics is the process of predicting the future reliability of a system by assessing the extent of deviation or degradation of a product from its expected normal operating states. When prognostics is combined with physics-of-failure models, it is possible to make continuously updated reliability predictions based on the monitoring of the actual environmental and operational conditions of each individual product. A literature review showed that the research on prognostics of solder joint reliability under vibration loading is very limited. However, personal portable electronic products are no longer used exclusively in a benign office environment. For example, any electronic component (throttles, brakes, or steering) in an automobile should be able to survive in a vibration environment. In this thesis, a methodology was developed for monitoring, recording, and analyzing the life-cycle vibration loads for remaining-life prognostics of solder joints. The responses of printed circuit boards (PCB) to vibration loading were monitored using strain gauges and accelerometers, and they were further transferred to solder strain and stress for damage assessment using a failure fatigue model. Damage estimates were accumulated using Miner's rule after every mission and then used to predict the life consumed and the remaining life. The results were verified by experimentally measuring component lives through real-time daisy-chain resistance measurements. This thesis also presents an uncertainty assessment method for remaining life prognostics of solder joints under vibration loading. Basic steps include uncertainty source categorization, sensitivity analysis, uncertainty propagation, and remaining life probability calculation. Five types of uncertainties were categorized, including measurement uncertainty, parameter uncertainty, model uncertainty, failure criteria uncertainty, and future usage uncertainty. Sensitivity analysis was then used to identify the dominant input variables that influence model output. After that, a Monte Carlo simulation was used for uncertainty propagation and to provide a distribution of accumulated damage. From the accumulated damage distributions, the remaining life was then able to be predicted with confidence intervals. The results showed that the experimentally measured failure time was within the bounds of the uncertainty analysis prediction.