An approach to the validation of computational solid mechanics models for strain analysis

Abstract

The need to provide strong evidence of the validity of predictions from computational solid mechanics models used in engineering design decisions is discussed. A new procedure is proposed, based on image decomposition, for reducing the dimensionality of strain field data from models and experiments and then comparing the resultant feature vectors via a simple linear correlation in which validation is deemed to be achieved when the coordinate pairs from the two feature vectors lie within a scatter band defined by the minimum measurement uncertainty. The procedure is illustrated by some simple examples that allow the advantages and drawbacks of the approach to be highlighted. It is anticipated that the procedure could become part of a corporate plan or regulatory process for verification and validation of computational solid mechanics models.