Vehicle fatigue damage prediction based on estimated strains and road measurements

This paper presents a study of fatigue damage predictions in vibrating dynamic systems with unknown inputs. The fatigue damage prediction is based on strains estimated from sparsely measured vibration responses and a numerical model. Components mounted on a truck chassis are studied, and measured acceleration responses from several hundreds of kilometres of driving are analysed. A linear finite element model is used to describe the dynamics of the vehicle, a numerical model which contains significant model errors. To reduce the effect of model errors and measurement noise, the probabilistic fixed-lag Kalman smoothing algorithm is used to estimate the strains in critical regions in the components. Due to the vast length of the analysed measurements, Dirlik’s spectral fatigue method is used to compute the fatigue damage in a computationally efficient way. The main objective of this study is to evaluate the applicability of the proposed fatigue damage prediction methodology for online health monitoring of heavy vehicles. This study is unique in its complexity as it covers dynamic system in combination with the full-scale measurements analysed. The predicted damage is validated by comparison to fatigue failures reported from trucks operating in the same regions as where the chassis acceleration responses were measured. The proposed methodology results in predicted fatigue damages which correspond well to reported failures, and it is concluded that the methodology has potential for future online vehicle fatigue damage predictions.QC 20210503