Nondestructive Evaluation (NDE) Technology Initiatives (NTIP). Delivery Order 0039: Statistical Comparison of Competing Material Models
Abstract
The ability to decide objectively between competing material models has always been an important engineering task. For a variety of reasons, simple inspection of the general "fit" of various models to the appropriate data does not always result in an obvious choice for the superior model, and heretofore there has been no statistical measure that can be used to choose between models (except in the simple case of "nested" models). Bayes factors overcome these previous limitations and can provide a valid objective statistical measure for choosing between competing material models. This study explores the use of Bayes factors as an objective measure in choosing between competing material models. The focus is on the use of this method by engineers who may or may not be well versed in statistical methods. In addition, the statistical background necessary to gain an in-depth understanding of the Bayes method is included as appendices. Also discussed in some detail is the Random Fatigue Limit s-N model proposed by Pascal and Meeker, which holds considerable promise for describing HCF behavior, as well as the applicability of applying Bayes factors to this model.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 2003
- Accession Number
- ADA415275
Entities
People
- Charles Annis