Uncertainty quantification and sensitivity analysis for buckling characterisation using the vibration correlation technique

Kurzfassung

The Vibration Correlation Technique (VCT) is becoming an established non-destructive method to predict buckling loads for imperfection imperfection-sensitive structures. While it has been successfully used to validate numerical models and to predict experimental buckling loads, recommendations on how to set- up the experimental test for VCT usage are relatively scarce. In this paper, the sensitivity towards the number of load steps and the maximum axial load level of the measurements is studied. An uncertainty quantification of the measured frequency affecting the VCT buckling load prediction was also performed. The numerical study is based on a series of FE models validated by experiment, representing a family of nominally identical cylinders. Furthermore, the experimental data available in the literature was also used to validate the trends observed in the numerical study. When no frequency deviations were introduced in the numerical data, a strong correlation between the VCT predictions and the maximum load used for measurements was found, while the number of load steps used was only relevant in reducing the errors induced by the frequency deviations. Introducing measurement deviations in the frequencies deterred the VCT predictions correlation with the maximum load level, while using more load steps reduced this influence. The sensitivity study using experimental data confirmed most of the trends observed in the numerical analysis, the most significant difference being a poor sensitivity of the prediction as a function of the maximum load ratio.