TWI Industrial Member Report Summary 381/1988
By M Matsumoto and T Jutla
Currently a number of procedures are available that allow defects in structures to be assesed.These procedures vary considerably in complexity and their use depends entirely on the structure under consideration and the extent of information available. A common problem with all fracture assessment procedures is that the data for the various parameters in the formulation are input as single deterministic values. No provisions are made for treating variability in the data and for evaluating significance of the variability on the predicted parameter whether it is crack size, stress or toughness.
In order to asses how the predicted values are affected when variability in certain parameters is introduced in the fracture assessment procedure a probabilistic analysis has to be performed. This report describes a series of sensitivity analyses on a collapse modified strip yield model for surface breaking cracks subjected to tensile loading. The analyses were carried out using the Monte Carlo method on four wide plates that failed at different nominal strength over yield strength ratios. The material fracture toughness, primary stress and residual stress were treated as random variables. Three methods of calculating the net section stress were also included in the investigation.