Computing Transfer Functions from Mass Loaded Response of Structures 2004-01-0780

This paper outlines a method for computing the transfer functions of structures using their mass loaded responses. According to the method, scaled transfer functions are computed from the response of a structure and without any knowledge of the input forces. The paper outlines the analytical approach, develops the necessary equations for the computation of transfer functions between a mass loading point and other points on a linear dynamic system. A numerical example to show the validity, advantages and limitations of the method is also provided.

Currently, the method can be applied to the responses obtained from analytical simulations where it may be necessary to compute coupled response of a simulated dynamic system with other dynamic systems that are not (or cannot be) included in a simulation. It is not uncommon that many dynamic simulations exclude certain coupling effects between the main and the auxiliary systems. This could be due to the limitations of the simulation software or simply related to practical reasons originating from large model size and excessive computation time. Consequently, the predictive ability of a dynamic simulation is sometimes diminished to a degree that renders its results useless. The method presented in this paper can remedy this limitation by facilitating transfer function computation under the actual operating conditions implemented in a simulation. Of course, once driving point transfer functions at some point on a dynamic system are obtained, many other techniques can then be utilized to compute any coupling or external force effects. Hence, response of a large dynamic system can be obtained from independent (uncoupled) simulation of its smaller parts.

The significance of this method would be even greater if it could be applied to measured response of real dynamic systems. However, initial investigations of its application to measured response data indicated that experimental use does present certain difficulties and requires further development. Nevertheless, the paper establishes a theoretical basis and shows that transfer function computation is possible from the response of a structure alone.