Inverse identification of creep of concrete from in situ load–displacement monitoring

Conventional algorithms for damage identification are effective in detecting local damage such as cracks. On the other hand, they are not directly applicable in identifying overall change in structural behavior. Considering that many structures show overall degradation without local cracks, alternative methods of identifying change in structural behavior are of great need. In this paper, we present a method that characterizes the overall time-dependent behavior of concrete using in situ load–displacement measurements. The method is based on a neural network constitutive model, finite element method, and numerical integration. Only by using the global load–displacement measurements, the method can compute approximate local stress–strain behavior at any location, enabling long-term monitoring of overall behavior of a structure with a small number of sensors. Application of the method to the identification of creep of a concrete beam shows a good match between the model prediction and the experimental measurement.