TWI Industrial Member Report Summary 191/1982
By A A Willoughby
Background
When carrying out defect assessments on elastic-plastic materials it is important to prevent the strain in the remaining ligament ahead of a defect from becoming excessive, otherwise the driving force becomes very high. Failure would then occur at a stress below that predicted by a simple fracture analysis. In practice this is prevented by ensuring that plastic collapse of the net section does not occur. In many practical situations, particularly for thin sections of ductile steels, plastic collapse becomes the criterion governing the allowable defect size.
The plastic collapse analyses incorporated in three defect assessment methods (PD6493, CEGB R6 and the Battelle method for pipes) are examined and compared. PD6493 is found to be more conservative as the defect length becomes small since all defects are treated as being infinitely long, and R6 is more conservative for larger defects under tension loading, owing to the assumption of pin jointing. The Battelle methods are generally the least conservative for pipelines.
Comparison of the various analyses with experimental data indicates that corrections can readily be made to PD6493 to allow for defects that are not infinitely long, that in many cases the assumption of pin jointing in R6 is unduly conservative (especially for embedded defects), and that the Battelle methods should be regarded as critical, rather than safe, analyses.
Improved methods of assessment for a number of specific defect geometries are proposed.