(Received 1 May 1989; accepted 8 November 1989)
| ||Format||Pages||Price|| |
|PDF (424K)||19||$25||  ADD TO CART|
It is desirable in design codes to characterize material strength by a low fractile of its distribution. This low fractile is known as a characteristic value. The material strength variable considered here is fracture toughness.
Ten realistic fracture toughness distributions were defined, and two characteristic values were estimated on the basis of simulated observations from these parent distributions. The estimates were then compared with the actual characteristic values to evaluate the quality of the estimators. Three estimators were evaluated: one estimator was based on the assumption that the parent distribution was log-normal; the second estimator assumed that it was a two-parameter Weibull distribution with a shape parameter equal to two; and the third was a power limit estimator, which assumed little about the parent distribution's parametric form.
The results suggest that in practical situations the log-normal method gives the best estimator, and that ten is a reasonable number of fracture toughness tests for the purposes of characteristic value estimation by the log-normal method.
The Welding Institute, Abington Hall, Abington, Cambridge,
Stock #: JTE12468J