Invariance problem in structural non-probabilistic reliability index

The non-probabilistic reliability index has been extensively used to evaluate the safety degree of structures with limited experiment data. By dealing the uncertain parameters of structures with the ellipsoidal model, this paper investigates the invariance problem in the non-probabilistic reliability index. A prerequisite of the existence of the invariance problem is first given. An investigation of whether the two non-probabilistic first order reliability methods, namely the mean-value and design-point methods, encounter the same problem is then presented. A comparison of the precision of these two methods is further conducted through three numerical examples, and based on which some significant phenomena are summarized.