On the edge of reality –an evaluation of parametric representations of creep strain and rupture data

Abstract

Advances in the understanding of the physical processes that underlie creep deformation, damage and failure have led to the development of powerful models with demonstrable predictive ability. Some are specific to classes of material, or to particular creep mechanisms; others aim to be more general in application. The insights which they provide are of considerable value, and certain of them have proved sufficiently robust to be incorporated in the more advanced practical engineering procedures.Nevertheless, almost all design and most assessment work makes use, not of these theoretical models, but of simpler parametric descriptions of creep rupture, and occasionally of creep strain. Originating at a time when theories were less advanced and when much practical computation was by hand methods, their practical advantages are clear. They have also become familiar and it is therefore not unreasonable that they continue in use for basic calculation. However familiarity – though necessary – is not sufficient for understanding, and comparison of established and more recently published creep and rupture parameters reveals some features that continue to give rise to concern.This paper evaluates a number of creep rupture and creep strain parameters, contained within published standards and procedures, in terms of their descriptive accuracy and practical use. It is shown that the differences between them can often be attributed to:the distribution of the source datathe influence of other mechanismsthe form of mathematical representationSince these differences can be sufficiently large to have led to disputes between manufacturers and operators of plant, it is important that the reasons be identified and understood in each case. Recently, however, parameters have been published that in form and calculated behaviour are at considerable variance with the previous consensus. Whilst these features are demonstrable, they are not so readily explicable.As it is not reasonable to leave practical engineers with methods that predict lives differing by several orders of magnitude, the paper concludes with some provocative suggestions as to what a parametric creep description should be and should do.