Unbiased estimation of the Weibull scale parameter using linear least squares analysis

Unbiased estimation of the Weibull scale parameter using unweighted linear least squares (LLS) analysis was investigated in this work as a function of specimen number, N. Mean, median and mode values of the Weibull scale parameter were estimated using a Monte Carlo procedure and general probability estimator of F = (i - a)/(N + b) with a and b varying between zero and unity. Unlike the mean and mode cases, the median value of Weibull scale parameter was essentially independent of Weibull modulus, m, with the coefficient of variation decreasing with b and being a factor of five smaller compared to that of m. Optimum values of a as a function of N were obtained using the median of the Weibull scale parameter with a being found to vary between 0.4796 for N = 10 and 0.2698 for N = 200 with b = 0.