Transient Load Recovery Using Strain Measurements and Model Reduction

Abstract

A transient load is defined as a loading condition where the magnitude of the load changes rapidly in a short period of time. Impact loads are a common example of transient loading. It is well known that impact loads can have disastrous effect on the structure compared loads applied over a longer period of time. An identification of impact loads is an important aspect in the design of structures. A direct identification of applied force on a structure through use of force transducers is not possible under all situations. In such cases, the structural response could be used to recover the imposed loading. Various structural responses such as displacement, stress or strain could be used to recover the imposed loads. However, this thesis focuses on a use of strain data to recover impact loads acting on a component. The strain data is extracted by placing strain gages at different locations on the component. A selection of the location for strain gages is tricky because the accuracy of the load recovery is sensitive to the position of the sensors. A D-optimal technique is used in this thesis to determine the optimum location of sensors so that most accurate results for load estimates are obtained. With sensors placed at the optimum locations, the strain data was extracted. The extracted strain data was used in conjunction with component’s modal data to approximate mode participation factors. The approximated mode participation factors and displacement mode shapes were next used to approximate displacements, velocities and accelerations. Finally, this information was used to estimate the loads acting on the component. A drawback of this approach is that it requires modal information of the entire structure be available. However, practical computational considerations limit the use of information of all modes. To overcome this difficulty, reduced order modeling based on Craig-Bampton model reduction is used. It is seen that Craig-Bampton reduction allows for an accurate estimation of imposed loads while utilizing only a small subset of available modal information.