The Determination of Stresses Using a Combined Theoretical and Experimental Analysis Approach

Abstract

With the rapid increase in the size and power of digital computers over the past decade, the direct numerical analysis approach to engineering design is now adopted on a routine basis. In the field of linear elastic stress analysis in particular, the finite element method is now used almost indiscriminately, and the results produced are often accepted without question. The purpose of this paper is to demonstrate by example that belief in calculated results should be tempered with caution. The first example demonstrates how the accuracy of a finite element analysis can be limited by the simplifying modelling assumptions that are imposed by practical and commercial constraints, and illustrates the necessity for testing the validity of the assumptions made. It is shown that a combination of theoretical and experimental work may be necessary before a full understanding of the physical behaviour of the component under investigation can be achieved. The second example discussed in the paper shows that even if a given theoretical approach provides an apparently satisfactory solution, full confidence in the solution is not justified unless the theoretical approach has been validated, either against another independent theoretical technique or against experiment. Validation of one theoretical analysis against another, when both are based on the same fundamental approach (such as the finite element method) is shown to be insufficient.