Dynamic response reconstruction using passive components

Doctoral thesis, 2012

In many testing applications within the field of structural dynamics, such as noise and vibration harshness, rattle and squeak, or durability testing in the automotive and truck industries, large hydraulic or electromagnetic actuators are used to excite the tested structure. Such actuators are mounted in the laboratory in a fixed configuration referred to here as a test rig. A frequent method for calculating the input to such test rigs is to replicate a certain reference signal, for instance dynamic responses from driving tests, to ensure that the test case is physically motivated. Versions of the Time Waveform Replication (TWR) method are often used to this end, iteratively calculating an input sequence used to excite a system such that its response replicates a desired reference. The TWR algorithm essentially estimates the unknown input by pseudo-inversion of the frequency domain transfer function. In most situations, TWR will ensure a test output which is close to the reference signal; however, if there are eigenmodes of the test specimen in the desired frequency range of the test which are uncontrollable, it can be shown that the required input force and/or the error between test output and desired reference will be large. Such problems remain even when reference-contributing modes are but marginally controllable. In a test rig such as described above, this controllability shortcoming may be impossible to prevent in the usual manner, \emph{i.e.} through changing the input configuration. This can cause poor reference replication whereby the validity of the test may be put into question. For such specific cases, where the possibilities of changing the input configuration is slim and controllability is lacking, a different approach is required. This thesis presents such an approach. It introduces the concept of passive control to the problem of response reconstruction: By attaching a modifying component to the tested structure, designed specifically to improve response reconstruction, the controllability of the system can improve. Two methods for designing such passive components are described. The first uses a high-quality finite element model of the tested structure and parameterized passive component, performing synthetic TWR experiments to evaluate its rig control properties. The second uses instead an experimentally derived model of the tested structure, which is coupled to the analytical model of the passive component through an experimental/analytical substructuring technique.