A limiting factor for many current structural health monitoring methods is that in order to locate damage, modeling of the structural response is required. The structural model itself can introduce significant errors, in addition to sensor noise, both through limitations in the assumptions applied and manufacturing variations. This work presents a method of localizing damage that eliminates the requirement for an independent structural model. The method is based on the flexibility parameters of the structure in pre- and post- damage states. An technique has been developed that allows the required information to be constructed from only sensor and actuator data. From the sensor data, a set of damage location vectors are determined. These vectors are shown to localize damage via two methods: The first analysis reapplies each set damage location vector as applied forces to the structure. The second, more applicable to real-time health monitoring, locates lowest values of the damage location vectors themselves. Simulations on a plate are performed for two sensor meshes (eight and thirty-two locations). The results demonstrate excellent damage localization, and some indication of damage severity. Finally an experimental demonstration of the method utilizing eight sensors surface mounted to an aluminum plate is presented.
Fiber optic sensors have the capability to simultaneously measure multi-scale data (e.g. strain, strain gradients, and integrated strains) for the purpose of structural health monitoring of structures. Presented here is a technique that can detect and localize damage in a structure suitable for the fusion of such multi-scale data. The flexibility method has been previously proposed for damage localization using changes in natural frequencies and mode shapes. However, such changes are difficult to apply in a real world application due to the extreme accuracy required for the input excitation frequencies. In addition, computer modeling of the undamaged structure to compare with the damaged structure can lead to significant errors due to imperfections. The flexibility method for solving indeterminate structures using static conditions presented here avoids these difficulties. The method follows a procedure of applying a known load to determine the flexibility matrices of the structure in pre- and post- damaged states. The introduction of fiber optic strain and displacement measurements permits the calculation of these matrices from the static loading conditions. These matrices are subtracted and the resulting null space calculated. When reapplied to the original system this null space locates the damage. To demonstrate the validity of this process numerical models are created for a simple truss structure and a plate instrumented with several fiber optic sensors. The examples show that damage can be located in both cases.