Description and objectives

Long-term testing of reinforced and prestressed concrete elements is crucial due to its significant implications for structural safety, durability, and service life estimation. Ideally, such tests should provide information on the behavior of structures over extended periods, enabling engineers to assess long-term structural reliability, identify potential degradation mechanisms, and develop effective maintenance and rehabilitation strategies.

Non-destructive testing (NDT) methods, such as pulse-echo measurements, electrical resistivity tomography, and impact-echo tests, are generally used to evaluate the condition of reinforced and prestressed concrete structures. However, recent research suggests that NDT methods are generally weakly predictive. Furthermore, the most reliable NDT techniques, such as digital radiography, tend to be the most expensive, making large-scale application challenging.

For this reason, tests with high predictive capability, such as endoscopy or direct testing, are preferable, although they are more invasive and provide localized results. By employing active learning techniques and Dynamic Bayesian Networks, properly calibrated on case study data, ADEPT addresses these needs by providing a space-time adaptive optimization framework with respect to the limit states under consideration. Consequently, ADEPT maximizes the informational content of the tests, controls inspection costs, reduces invasiveness on the structures, and improves structural safety.