Graduation Year


Document Type




Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Industrial and Management Systems Engineering

Major Professor

Mingyang Li, Ph.D.

Committee Member

Tapas K. Das, Ph.D.

Committee Member

Devashish Das, Ph.D.

Committee Member

Lu Lu, Ph.D.

Committee Member

Qing Lu, Ph.D.


Acceptance Uncertainty, Multi-criteria Evaluation, Multiple Failure States, Optimal Design, Reliability Assurance


Reliability demonstration test (RDT) is one of important reliability assurance activities to demonstrate products' quality over time. Binomial RDT (BRDT) is one class of RDTs with appealing features, such as less failure monitoring efforts and fewer reliability modeling assumptions. Integrating with Bayesian method further allows prior knowledge incorporation for potential test sample size reduction. However, conventional designs often assume the binary failure states (i.e., success and failure) and consider a single objective of minimizing the testing cost with limited planning horizon. In this dissertation, a series of RDT designs are proposed and studied by advancing the conventional Bayesian BRDT designs from three aspects. First, the multi-state RDT designs are proposed to demonstrate product reliability either at multiple time periods or with multiple failure modes. They relax the simplified assumption of single time period or failure mode in conventional designs. Second, a BRDT design with extended planning horizon is proposed to consider both the testing acceptance uncertainty, and the anticipated cost impacts on subsequent reliability assurance activities, such as reliability growth and warranty services. Third, a multi-objective RDT design is formulated to simultaneously consider various RDT performance evaluation criteria, such as consumer's risk, producer's risk, acceptance probability and different cost components. The trade-offs among different conflicting objectives are leveraged by a Pareto Front approach. Case studies are provided to illustrate the proposed methodological frameworks and further demonstrate their superior performance and/or validity with comprehensive evaluation and analysis. The proposed work allows practitioners to develop more advanced Bayesian BRDTs to meet the increasingly complex reliability requirements from consumers and/or manufacturers.