Validation of some FM-based fitness for purpose methods

The reliability of several FM-based fitness-for-purpose methods has been investigated on a number of objects for which accurate fracture data were available from experiments or from practice, viz. 23 wide plates, 30 mm thickness (surface and through thickness cracks, cracks at holes, with and without welds), 45 pipelines sections with cracks, pressure vessels and a T-joint. The methods applied mainly comprise ASME XI, PD 6493 and R6. This contribution reviews the results. Within the restricted scope of these investigations some specific and general conclusions can be drawn: - wide plate and pipe tests: No method excludes unsafe predictions; The least safe predictions stern from application of methods outside the range of failure behaviour for which they can be considered applicable (e.g.: pipe burst tests with Batelle-flow stress method in actually toughness dominated range; wide plate tests with LEFM ASME XI in plastic collapse dominated range). All unsafe predictions fall below Kr/Sr= 1 (except pipe burst with wrong model indicated above). The safety ratio ranges from 0.6 to over 3; most data fall between 0.8 and 2. Scatter varies considerably per method. Scatter increases towards both the lower and the upper end of the fracture behaviour range. The CEGB 2 criteria method obviously yields the best results. No safety ratio falls below 0.8, data are closest to unity (i.e. accurate) and show strikingly less scatter. - pressure vessels: redactions are safe (and rather accurate for the boiler drum), but can turn into unsafe depending on the applied method and the applied toughness data (HSST vessels). - T-branch: in fact this case represents a series of examinations, i.e. the actual structure representative for a complex geometry) and small and large models. The partially unsafe predictions imply a warning, especially in view of the safe to very safe predictions for the models. In general it can be concluded that no one of the widely applied methods considered here excludes unsafe predictions, even for laboratory-type cases including extensive analytical and experimental investigations CEGB 2-criteria method results show the highest overall accuracy and the least overall scatter. Methods should not be applied outside the range of failure behaviour for which they are considered applicable. This should be verified upon application. (author)