A novel structural damping identification method is presented. The approach is robust with respect to measurement noise and makes use of highly effective reduced-order-models (ROMs). Several different methods are currently available for damping identification. Most of these techniques can be grouped into two types based on the nature of the system information that is needed for the damping identification. The first type involves measuring damped eigenvalues and mode shapes, and does not require measurements of the system excitation. The second type involves measuring the forces applied to the system and constructing (full) frequency response functions. In contrast to existing techniques, the proposed method avoids complications involved in measuring damped modal characteristics or applied forces, while identifying structural damping only from displacement or velocity measurements. The focus of this work is identification of damping in systems with high modal density (such as cyclically symmetric systems) exemplified by blisks and bladed disks. First, a novel, general methodology for identifying (uniform) structural damping is presented. This method uses undamped tuned system mode shapes and a minimum of two measurements. Next, a more general methodology is formulated, which incorporates stiffness mistuning and uses ROMs for enhanced robustness and fast calculations. Validation of the damping identification is done by comparing the performance of the viscous damping method by Lee et al. with the proposed method. In Lee's method, the complex frequency response function is used to determine a viscous damping matrix. This method is adjusted to identify structural damping where the damping matrix is now diagonal. For a low dimensional system and noiseless measurements, both Lee's method and the proposed approach correctly identify the structural damping. Introducing measurement noise causes inaccuracies in the identification results obtained using Lee's method, while the proposed method remains accurate. Next, two measurement filters are proposed to further increase the accuracy and robustness of the proposed damping identification by reducing the effect of measurement noise. The first filter applies to measurements which are approximately equal in amplitude and phase although they occur at different frequencies. The second filter removes measurements where the magnitude of the response is low. These filters are implemented for a complex validation structure: a one-piece bladed disk with stiffness mistuning. Simulated forced response measurements are generated by ANSYS and corrupted by noise. Next, measurements of the modal amplitudes and phases for the blisk are obtained through an elaborate and complex process of measurement point selection, mode selection, and data filtering similar to the one associated with mistuning identification. These filtered measurements are then shown to be accurate for use in the novel damping identification methodology.
展开▼
