Transient and Spectral Fatigue Analysis for Random Base Excitation
Examensarbete för masterexamen
Applied mechanics (MPAME), MSc
Madhava Acharya, Prithviraj
This thesis work gives an insight on how to estimate the fatigue damage using transient and spectral fatigue analysis for both uniaxial and multiaxial stress states in case of random vibration base excitation. The transient method involves a rainflow count algorithm that counts the number of cycles causing the fatigue damage, while the spectral method is based on probabilistic assumptions from which an expected value of fatigue damage can be estimated. The purpose was to compare the fatigue damage obtained from the transient and spectral approach, and evaluate the performance of the spectral method. In this study, the Dirlik’s empirical formula has been selected for the spectral method, partly because it has proven to be a good implementation in structural mechanics fatigue. In order to account multiaxial fatigue in the calculations, the authors decided to use Dang Van equivalent stress for the transient analysis. In addition to that, the cycles were counted applying Wang and Brown’s method, which can be seen as a more general extension of the original rainflow count algorithm. In the spectral analysis, the well known equivalent von Mises stress (EVMS) has been selected in order to account for multiaxial spectral analysis. The two methods were studied in association with an air dryer component that s attached to a chassis frame of a truck. The air dryer is subjected to random vibration via the mounting interface. The vibration was simulated by acceleration base excitation. Both uniaxial and multiaxial base excitation were investigated and the fatigue life was estimated in three selected hotspots on the surface of the air dryer component. The hotspots were chosen based on modal analysis. The Dirlik’s empirical formula was showing promising estimation of the fatigue life similar to the rainflow count. In most cases, the difference n fatigue life between the two methods was less than 30 % for both uniaxial and multiaxial stress. However, Dirlik’s formula was mostly showing more conservative results compared to the rainflow count. The cause of this could either be errors in the calculations or too short input signals. In some cases the difference between the methods were more significant, showing 200 % difference in fatigue life. The authors believe that this is most likely caused by mid stress effects in the Dang Van equivalent stress.
Spectral fatigue analysis , transient fatigue analysis , power spectral density (PSD) , transfer function , base excitation , Dirlik's empirical formula , equivalent von Mises stress (EVMS) , rainflow count , Wang and Brown's method , Dang Van equivalent stress , Palmgren-Miner rule