Exploring the Statistical Advantages of Nondestructive Evaluation Over Destructive Testing

Abstract

In this paper, we propose a framework for exploring the statistical advantages of nondestructive evaluation (NDE) over destructive testing (DT). Two cases are considered: (1) 0–1 or pass/fail data, and (2) continuous measurement data. While NDE data are less expensive to collect, they are less precise than DT data. However, taking more NDE data provides more precision. The proposed framework provides a way to calculate equivalent sample sizes of NDE data for given sample sizes of DT data. We illustrate the proposed framework with a radiographic example.

• Bayesian
• Continuous and Discrete Measurements
• Metropolis–Hastings Algorithm

Additional information

Notes on contributors

C. Shane Reese

Dr. Reese is an Associate Professor in the Department of Statistics at Brigham Young University. His email address is [email protected].

Paul Deininger

Mr. Deininger is retired and previously was a Technical Staff Member in the Weapons System Engineering Division at Los Alamos National Laboratory.

Michael S. Hamada

Dr. Hamada is a Technical Staff Member in the Statistical Sciences Group at Los Alamos National Laboratory. His email address is [email protected].

Robert Krabill

Mr. Krabill is a Technical Staff Member in the Weapons System Engineering Division at Los Alamos National Laboratory. His email address is [email protected].