Collaborative Research: Quantitative Reliability Prediction in Early Design Stages

Project: Research project

Project Details


Collaborative Research: Quantitative Reliability Prediction in Early Design Stages Collaborative Research: Quantitative Reliability Prediction in Early Design Stages Objective: The objective of this project is to predict product reliability in early design stages using information from various sources. The information will be extracted from heterogeneous, multilevel sources such as previous products, expert opinions, component and early prototype testing, and initial simulations. The major approach is the development of a Bayesian framework that aggregates and processes the information and then quantitatively predicts product reliability in the early design stages, including design conceptualization and embodiment. The quantitative reliability prediction will then assist to make decisions in the two early designs stages. Intellectual Merit: Quantifying product reliability is crucial in early design stages because it helps reduce risk and avoid costly and unnecessary design changes. However, the current reliability methodologies, such the Failure Modes and Effects Analysis, are mostly qualitative. The challenge is that the reliability-related information in early design stages is scattered, in different formats, at different levels of details, and from various sources. This research will address the challenge with a quantitative and systematic way to predict product reliability in the early design stages. Specifically, the establishment of the novel probabilistic graphical models and Bayesian network will not only specify the complex structure of the prediction system but also integrate both subjective and objective information. The uniqueness of accommodating all the aforementioned heterogeneous data will allow for more accurate reliability prediction, thereby more effective actions identified early to prevent potential failures or reduce their likelihood. The originality of the project is also characterized by its potential ability of making more accurate estimation on product warranty and maintenance costs because reliability is a major driving factor in such costs. As a result, design concepts can be evaluated more rigorously in terms of reliability and cost, and more reliable design decisions can be made. This project will be conducted through collaborative efforts between two universities, involving expertise in both engineering design and reliability engineering. Broader Impact: Reliability is a core element in the overall performance of almost all products. It directly determines customer satisfaction, product market share, and product safety. By quantitatively predicting product reliability in early design stages, engineers will have a better way to achieve high reliability with reduced cost. As a result, this research will impact the design practices for all kinds of products. Specifically, the project will benefit both engineering design, including the methodology and theory of conceptual design, and reliability engineering. The results from this project will be publicly available through conference presentations, journal articles, and internet. In addition, the technology transfer will be accelerated by teaming up with a manufacturing company and a software company. Through a problem-based, data-centric teaching method, the research results from this project will be transferred to classroom learning. At least three courses, across sophomore to graduate levels, at Arizona State University and Missouri University of Science and Technology, will be enhanced by this project. Two doctoral students will be supported; female and minority students will be aggressively recruited. A miniseries of lecture videos on early reliability prediction in design will be produced and made available to general public through the project website. REU: Collaborative Research: Quantitative Reliability Prediction in Early Design Stages
Effective start/end date8/1/137/31/17


  • National Science Foundation (NSF): $245,191.00


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.