Volume 17, Issue 5 (September 1989)
Regression Analysis of Single Specimen Fracture Toughness Data with Serially Correlated Errors
The statistical properties of errors associated with J-integral estimates and crack growth measurements obtained by single specimen fracture toughness test methods are examined. It is shown that the errors in the J estimates can be serially correlated. Slope and intercept estimates and their standard errors are derived for the fitted line on the basis of a regression model, suggested by Mandel, which makes allowance for the serial correlation of errors in J. A theoretical model is developed to compare the scatter apparent on a J/Δa plot with the true error inherent in the data.