Proof Loading for Target Reliability

In absence of reliable analytical procedures, the adequacy of structural resistance can be judged from proof-loading experiments. Neither a unified method nor a clear rationale exists now for the design of such experiments. Reliability approaches have been only partially successful, due to their oversimplification of the test, especially with regard to the information that is obtained from proof reading, which is reduced simply to failure or survival. Classical models of this type are reformulated and extended to include multivariate resistance and the possible presence of gross errors. More realistic models are also proposed, that utilize indirect information on ultimate resistance (e.g., loss of serviceability, cracking, softening). Gross errors and indirect information are of critical importance in determining the intensity of proof loading.