An implementation of structural change detection procedure based on experimental and numerical model correlation Damir Sedlar n , Zeljan Lozina, Damir Vucina University of Split, Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture, 21000 Split, Croatia article info Article history: Received 14 December 2010 Received in revised form 5 December 2011 Accepted 17 February 2012 Handling Editor: I. Trendafilova Available online 14 March 2012 abstract A correlation of the experimental and the numerical model is implemented in order to detect structural change. The implementation of the method involves two steps: in the first the respective location is detected, and in the second the structural change is quantified. The method can detect changes in mass, stiffness or both. The numerical simulations demonstrate that the method is accurate and reliable. The method is implemented analytically and experimentally in parallel with laboratory test-cases. Improvements of the procedure are suggested and implemented in order to study the applicability to real-world cases. This study shows that model update significantly contributes to the successful structural change detection. The influence of the model reduction and model update on the detection accuracy is studied. & 2012 Elsevier Ltd. All rights reserved. 1. Introduction Identification of structural changes using techniques that examine changes in measured structural vibration response is a very important topic of research. The underlying principle is that modal parameters are functions of the mass, stiffness and damping. Structural changes lead to changes of the natural frequencies, damping loss factor and mode shapes. These changes make it possible to detect structural changes by correlating vibration response of a structure before and after the changes have occurred (original and modified structure). However, the success of this methodology depends on many factors that are associated with analytical, numerical and experimental problems. Many non-destructive techniques have been proposed or being used, Doebling et al. [1] gave a detailed overview of the vibration-based change detection methods. Salawu [2] proposed damage detection methods based on changes in frequency. Kim et al. [3] have also proposed a method to locate and estimate the size of damage using the natural frequency and mode shape. Changes in structural systems due to damage can also be detected using mode shape curvature [4], strain energy [5] and changes in flexibility [6]. In this paper, an (improved) version (modification) of the method based on the residual force vector [7] is implemented. Many researchers have introduced the assumption that there is no change in mass, but only in the stiffness, which simplifies the problem [8–10]. Huynh et al. [8] define the damage location vector in order to locate the damage and determine its extent in space truss and plate structures. The damage location vector requires the spatial model in the form of mass matrix and stiffness matrix of the undamaged structure. Moreover, the experimentally obtained frequency response function of the damaged structure is required. The incompatibility of the degrees of freedom (dofs) between the theoretical model and the experimental model is solved using dynamic expansion. Yang and Liu [9] use the residual force Contents lists available at SciVerse ScienceDirect journal homepage: www.elsevier.com/locate/jsvi Journal of Sound and Vibration 0022-460X/$ - see front matter & 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.jsv.2012.02.019 n Corresponding author. Tel.: þ385 21 305 967; fax: þ385 21 463 877. E-mail addresses: damir.sedlar@fesb.hr (D. Sedlar), lozina@fesb.hr (Z. Lozina), vucina@fesb.hr (D. Vucina). Journal of Sound and Vibration 331 (2012) 3068–3082