NOISE SENSITIVITY OF FIT PROCEDURES FOR WEIBULL PARAMETER EXTRACTION

The noise sensitivity of three fit methods frequently applied to determine the parameters of a Weibull distribution has been analyzed. Using the estimator P = (i - 0.5)/N and a linear least-squares fit procedure with no weight function, the least biased results are obtained, although the amount of bias strongly depends upon the Weibul modulus m and the amount of noise. From the results it can be concluded that the larger the value of the Weibul modulus m, the smaller the amount of noise should be (< 1%, e.g.) to avoid systematic large underestimation of m. Therefore increasing experimental accuracy is required, to be able to discriminate the effects of material improvement (higher Weibull modulus).