Dynamic response estimation and fatigue prediction in a linear substructure of a complex mechanical assembly

Abstract

In this work, a computational framework is proposed for online estimation of fatigue damage, in a linear steel substructure of the entire body of a lignite grinder assembly at a PPC power plant. The proposed method is based on sparse vibration measurements and recursive Bayesian filtering for response estimation, in presence of unknown nonstationary excitations. First, a discrete FE model of steel base is developed; the resulting model is then updated for matching its dynamic characteristics with those measured from experiments on a support-free state of the structure. In doing so, measured FRFs are used for estimating the natural frequencies and the damping ratios of the substructure. Structural identification methods are used for estimating the parameters (material properties) of the FE model, based on minimizing the discrepancy between the experimental and analytical modal characteristics, in order to develop a high fidelity FE model. Fatigue is estimated using the Palmgren-Miner damage rule, S-N curves, and rainflow cycle counting of the variable amplitude time histories of the stress components. Incorporating a numerical model of the structure in the response estimation procedure, permits stress estimation at unmeasured spots. Therefore, fatigue estimates could be available at any point on the structure for drawing a complete fatigue map. These stress response characteristics are predicted from by means of a limited number of vibration sensors.