Output-Only Estimation of Wind Induced Stresses in Structures

Abstract

The use of output feedback as a means to monitor stresses and fatigue in metal structures has been the subject of recent work by researchers in the field of vibration based structural health monitoring. The methodologies that have been proposed in the literature employ the Kalman filter as an estimator and operate under the premise that excitations are realizations of Gaussian white noise. In practice two issues arise, excitations cannot always be approximated by realizations of Gaussian white noise and performance requirements are such that engineers require the use of large finite element models to accurately estimate structural response. This is especially true in the case of wind excited structures, such as tall wind turbines and complex buildings. In the present paper we propose a model based state estimator that can be implemented as a modified finite element model of the system subjected to corrective forces proportional to the measurements. This allows for direct implementation within the context of an arbitrarily large finite element model (FEM) and is of significant importance for implementation purposes given the difficulties typically encountered when implementing the standard Kalman filter in large FEM. Additionally, the proposed estimator is able to explicitly account for the non-white nature of wind loads by using knowledge of the power spectral density of the process. The proposed methodology is successfully applied to estimate the time history of bending moments throughout the height of a simulated tall vertical structure subject to stationary turbulent wind using only sparse measurements of dynamic response and a FEM.