Abstract

A measurement system is a necessary component for many quality improvement initiatives. For measurement systems such as destructive testing, however, the part characteristic is measured as the part is destroyed so that it is no longer available for additional measurements. There are statistical methods to estimate the repeatability and reproducibility (R&R) components in destructive testing if it is assumed that a batch of parts is enough alike that each component may be considered identical. This assumption of a homogenous batch justifies using an R&R study. When a homogenous batch is too small to provide multiple parts for each test operator, it is appropriate to use the nested or hierarchical model applied in a test to measure the impact strength of steel. The results of an R&R approach to assess a destructive measurement system are not as straightforward as those used in a nondestructive test because of the inability to distinguish repeatability variation from the within-batch variation. If these conditions make the destructive measurement system unacceptable from a repeatability standpoint, the homogenous batch assumption should be questioned.