The objective of this collaborative research project is to predict product reliability in the earliest design stages, such as concept design. The prediction is based on information gathered from a variety of sources such as previous components and products, expert opinions, early prototype testing, and simulations. The project uses a Bayesian framework that aggregates and processes uncertain information. Quantifying product reliability in early design stages helps reduce risk and avoid costly and unnecessary design changes. This project uses a graphical model of probabilities to represent system reliability and a combination of subjective and objective information, using reliability-related data that are scattered, in different formats, at different levels of details, from various sources. Pulling together all this data allows for more accurate reliability prediction, leading to more effective actions identified early to prevent potential failures or reduce their likelihood.
If successful, this project will improve design practices for all kinds of products, because reliability is a core element of product performance and directly determines customer satisfaction, product market share, and product safety. Specifically, this project will advance engineering design theory and methodology and expand the scope of reliability engineering. By quantitatively predicting product reliability early, the project provides engineers with a better way to achieve high reliability with reduced cost.