Combining Data for Improved Weibull Parameter Estimation

Abstract

Weibull parameters are often used by experimentalists to characterize the fracture behavior of structural materials. Frequently, several types of tests are performed, and theory indicates that the parameters for each type should be the same or have some specific relation to each other. In such cases, it is advantageous to combine the data to find the parameters best representing all the data. The present paper discusses techniques for carrying out this objective and analyzes two examples in detail to illustrate the principles involved. The first example compares uniaxial and equibiaxial fracture statistics and the second discusses fracture statistics for single fibers of several lengths.