A Gaussian damage function combined with sliced finite-element meshing for damage detection

Bridges are among the most important components of transportation systems. Timely damage detection of these structures not only ensures reliability but also prevents catastrophic failures. This paper addresses the damage assessment of bridges based on model updating techniques. Artificial damage was introduced to a beam that was a part of a real prestressed concrete bridge. The magnitude of the damage was increased stepwise, and static loading experiments were conducted in each step. A linear Finite-Element (FE) model with solid elements that were clustered into slices was utilised. A Gaussian bell-shaped curve was used as a damage function to describe the crack location using only three parameters. The experiments focused on sagging under dead load. Damage identification was performed in two steps using a coarse and a refined model. Initially, the FE model with a coarse mesh was updated to approximately localise the damage. Then, the FE model is refined in the vicinity of the approximately localised damage, and damage identification was accurately achieved. The results show that after the second step, the maximum error value of damage localisation is less than 0.5%. This approach could be later used to detect small damages that are not visible.