Dynamic modeling of structural joints

Thesis Type: Postgraduate

Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Faculty of Engineering, Department of Mechanical Engineering, Turkey

Approval Date: 2012




Complex systems composed of many substructures include various structural joints connecting the substructures together. These mechanical connections play a significant role in predicting the dynamic characteristics of the assembled systems accurately. Therefore, equivalent dynamic models of joints that consist of stiffness and damping elements should be developed and the joint parameters should be determined for an accurate vibration analysis. Since it is difficult to estimate joint parameters accurately by using a pure analytical approach, it is a general practice to use experimental measurements to model joints connecting substructures. In this study an experimental identification method is suggested. In this approach the frequency response functions (FRFs) of substructures and the coupled structure are measured and FRF decoupling method is used to identify equivalent dynamic characteristics of bolted joints. Since rotational degrees of freedom (RDOF) in connection dynamics is very important, a structural joint is modeled with translational, rotational and cross-coupling stiffness and damping terms. FRF synthesis and finite-difference formulations are used for the estimation of unmeasured FRFs and RDOF related FRFs, respectively. The validity and application of the proposed method are demonstrated both numerically and experimentally. In simulation studies, simulated experimental values are used, and it is seen that the identification results are prone to high errors due to noise in measurement and the matrix inversions in the identification equations. In order to reduce the effect of noise, it is proposed to extract the joint properties by taking the average of the results obtained at several frequencies in the frequency regions sensitive to joint parameters. Yet, it is observed in practical applications that experimental errors combine with the measurement noise and the identification results still may not be so accurate. In order to solve this problem, an update algorithm is developed. In the approach proposed, the identified dynamic parameters are used as initial estimates and then optimum dynamic parameters representing the joint are obtained by using an optimization algorithm. The application of the proposed method is performed on a bolted assembly. It is shown with experimental studies that this method is very successful in identifying bolted joint parameters. The accuracy and applicability of the identification method suggested are illustrated by using a dynamically identified bolt in a new structure, and showing that the calculated FRFs in which identified joint parameters are used, match perfectly with the measured ones for the new structure. In this study, the effects of bolt size and quality of bolts, as well as the bolt torque on the joint properties are also studied by making a series of experiments and identifying the joint parameters for each case.