COMPARISON OF EVALUATION PROCEDURES FOR THE SUBCRITICAL CRACK-GROWTH PARAMETER-N

The uncertainty in the determination of the exponent n in the equation of the subcritical crack velocity from four-point-bending experiments is investigated with respect to different evaluation methods. If the bending strength values are Weibull distributed, and the crack-extensi on parameter n is calculated by linear regression of the bending stren gth values of a number of experiments at different loading rates, an a nalytical solution can be given for the mean value and the standard de viation. It turns out that both the mean value and the standard deviat ion depend on the Weibull modulus m and the true value no. The analyti cal solution illustrates the essential features of this dependence on m and no. For other evaluation methods, e.g. the one proposed as the C EN standard, this dependence is investigated by a Monte-Carlo simulati on for different crack-extension parameters no and different Weibull m oduli m. The standard deviation, which is calculated is the theoretica l lowest limit for certain evaluation procedures. By this, the estimat ion of the margin of error is put on a firm ground Since the standard deviation increases with no, there is only a limited range in which n can be determined by four point-bending tests. A new evaluation method , which gives a better approximation than the method proposed as the C EN standard, is presented. The computational effort of this evaluation method is only slightly larger. It furthermore allows the number of e xperiments to be analytically calculated, which is necessary to obtain a certain accuracy.