Reliability Predictions Using Model-Based Criticality-Associated Similarity Analysis

In this paper, we will describe a methodology called Model-Based Criticality-Associated Similarity Analysis (CASA). The CASA methodology was first introduced to the Reliability Engineering community nearly 20 years ago, at the 48th Reliability and Maintainability Symposium (RAMS) in Seattle, WA [Ref. 1]. This methodology systematically develops a reliability prediction by applying the following empirical-based step-wise conjecture: The ratio of predicted to demonstrated reliability for a new product (i.e., a product that has never been placed in-service) is equal to the corresponding ratio for a similar in-service product that has both its predicted and demonstrated reliabilities adjusted to reflect all the failure and sneak modes, mechanisms, and root causes of the new product. The CASA methodology is practical and efficient when there is newly designed electronics equipment that is sufficiently similar to in-service electronics equipment that has a demonstrated reliability. The application of this methodology will result in a reliability prediction that is more precise than those obtained by using traditional reliability prediction methodologies. With that said, the prerequisite for successful application of the CASA methodology is availability of detailed design and operational/field data. This paper describes an example application of the CASA methodology, in the rapid development of a new and more technologically advanced product that is required to have higher operational reliability and lower cost per unit-function than the predecessor in-service product. Fault/Failure-based modeling can yield meaningful comparisons between the relative design reliability of a new product and the operational/field reliability of a similar inservice product. It can also be used to perform complex reliability assessment in less time. Quantifying design differences allows one to determine adjustment factors that can be applied to the field reliability of the in-service product to obtain a precise and repeatable prediction for the “expected” field reliability of the new product. Since no field or test data are available for a new product design, the characteristics data of a similar in-service product must be used to achieve a degree of confidence in the reliability prediction of the new product. This type of reliability prediction is of great value during the development of the new product's design reliability features because CASA makes use of knowledge about the impact of fault/failure modes, mechanisms, and root causes that occurred in the field, but which may not be considered by the designers prior to product manufacture and delivery.