The research will develop new reliability models of load testing that use information on the performance of series-produced or similar structures, success or failure in different test and service modes, risk of test damage, load history of a structure, and other structural indicators such as stiffness and permanent deformation. In the approach planned, a first estimate of strength or reliability is combined statistically with new information from a load test or other source, such as the observed performance of a related structure. A revised estimate is obtained containing both new and old performance information, that can be used in subsequent evaluation of the structure. The quality of information, which varies from high, for actual proof tests of a structure in all important service modes, to low, for completely subjective evaluations of performance, will be reflected in model development. The expected value of information to be gained from a test or non- test alternative will be balanced against its cost (difficulty), and against the performance of other competing options. Current load testing procedures will be reviewed and compared to model characteristics, and examples and case studies will be used to compare model performance in a variety of load testing applications. The research will have an important impact on safety verification, risk management and maintainability of structures, screening of gross errors and deterioration, and many other decision problems in structural engineering.
