The Appropriate Verification and Validation Methodology:
An Issue of Shared Systematic Error Sources
Jeffrey Weathers, Robert Taylor, James Weathers, Jonathan Woolley
International Journal of Cast Metals
Status: In Submission, submitted May 2009
ABSTRACT
Traditionally, validation metrics have been derived using univariate statistical methods. However, a univariate statistical analysis may be inappropriate when the verification and validation (V&V) exercise includes multiple realizations. Such circumstances may call for the use of multivariate V&V methodology. In the present paper, experimental and simulation results corresponding to an A356 plate cast in a resin bonded sand mold are subjected to a multivariate V&V effort. Multivariate V&V methodology is explained in detail through the sand casting V&V effort, and the consequences of inappropriately using univariate V&V methodology are also addressed.
