A Stochastic Model for Assessment of Creep damage

Abstract

Experimental investigations on the creep behaviour of engineering materials are frequently conducted for the purposes of design, development and service life prediction. However, tests on a single specimen yield a definite value for each material parameter. But when a number of specimens are tested, the parameter values randomly fluctuate from specimen to specimen. The scatter observed in creep deformation and failure data is of considerable technological importance because it greatly complicates the task of making accurate deformation and lifetime estimates for high temperature components. Several studies, employing theory of probability concepts have been developed to predict and characterize the variations in creep damage parameter. So far the deterministic approach has been used which is based on only two limiting states of the material i.e., ω= 0, no damage and ω = 1, completely damaged. This imposes severe limitations on the validity and practical applications of this model. For instance, it is not possible to trace the damage accumulation during service using the NDT techniques.