ABSTRACT
In this paper the feasibility of a data-driven method, namely the Interpolation Damage Detection Method (IDDM), is investigated to estimate the magnitude of damage in a structure permanently monitored by a network of accelerometers.
The adopted method had been previously successfully used for damage localization purposes. The investigation is carried out based on the numerically simulated responses obtained from the finite element model of a benchmark suspension bridge. The structural response to a seismic excitation is computed both in the undamaged and in several damaged configurations. Damage is simulated through a reduction of the elastic modulus of the material in selected structural elements.
For each damage scenario several damage severities are modeled and the correlation between the damage feature and the severity of damage is investigated.
The possibility to deploy a permanent network of sensors on the structure and to apply an efficient damage identification algorithm can greatly reduce the cost of structural maintenance while improving safety. In critical situations the possibility to have, almost in realtime, a complete description of the damage state of the structure, can be strategic for the management of both the emergency and repairing interventions.
Data-driven methods are attractive for adoption within an automated monitoring system but generally limited to detection and localization of damage. Model-based methods may allow a deeper knowledge about the health of the structure, allowing the damage severity estimation, but are hardly compatible with real time on line damage identification algorithms. The possibility to use a data driven method to estimate the magnitude of damage can thus greatly enhance the effectiveness of an on line damage identification system.
In this paper preliminary results of a numerical study are reported. They are focused on the possible extension to damage severity assessment of the Interpolation Damage Detection Method, previously applied for damage localization purposes. Structural responses were obtained using a refined finite element model of a suspension bridge by simulating several damage scenarios.
Results show that there is a proportional relationship between the damage severity and the damage parameter that allow to consider the IDDM a promising method, hopefully extendable to level 3 methods according to Rytter (1993), able to detected, localize and assess the severity of damage. This would be a remarkable performance, being the IDDM a non-model-based damage identification method.
At present, however, the relationship between the value of the damage index and the loss of stiffness is still unknown. Further research efforts will be devoted to this aspect in the future.
