Dynamics & Vibration Research Group

In the theory of structural vibration, the modelling of effects of mass and stiffness is well understood. This allows routine computation of undamped vibration of any linear structure, for example by Finite Element methods. Vibration damping is much less well understood, and the usual approaches to this problem are entirely ad hoc. Recent work has shown that, provided the damping is light, it is possible to incorporate the effects of rather general models of damping into the usual theoretical framework. This means, for example, that Finite Element predictions can be post-processed to allow for any desired damping model, and show the effect on natural frequencies, mode shapes and transfer functions.

Project Details

Using this approach, the current project aims to address a number of general questions:

• Is it possible to determine the "correct" model for the damping of a given structure by experimental measurements?
• Is it possible to determine the spatial distribution of damping by measurement?
• Can more than one damping model account equally well for the measured behaviour of a structure?
• What are the consequences of choosing the wrong damping model? Which aspects of vibration prediction are robust, and which are sensitive to choice of damping model?

Useful progress has been made on these questions, by developing algorithms to fit various possible damping models to measured responses, using the complex frequencies and mode shapes. Extensive simulation studies have been used to explore the strengths and weaknesses of the algorithms. Some conclusions are that, at least with simulated data, it is often possible to get a good idea of the spatial distribution of damping. One can also get clues as to the right damping model to choose, because a wrong choice produces recognisably non-physical predictions: a coefficient matrix which should be symmetric turns out to be non-symmetric after the fitting process.