Spectral and statistical analysis of flow-induced vibrations

Abstract

A general approach that utilizes both spectral and extremal statistical methods are utilized to investigate the time series of flow-induced response behavior of a flexible horizontal cylinder subject to both random waves and constant current conditions. The cylinder model was 29 m long and had a slenderness ratio of approximately 760. The random waves were generated using a JONSWAP wave amplitude spectrum. In addition, for some tests, the cylinder was towed at two different speeds to simulate the combined loading of random waves and constant current conditions. The data were initially analyzed using standard spectral analyses to interpret the cylinder's flow-induced response behavior and relate the findings to traditional deterministic parameters. Further analyses were performed using a generalized extreme value (GEV) distribution procedure that involved dividing the time series into blocks and fitting the block maxima of the extreme values in the measured response time series data. The Anderson-Darling (AD) test criterion and quantile plots were then used to assess whether the GEV distribution provides a satisfactory fit to the data capturing the statistical characteristics in the flexible cylinder's flow-induced response behavior, which was stochastic in nature. For the data set analyzed, the extremal GEV methodology presented was observed to provide excellent results for the random wave cases and moderately good-to-good results for the combined random wave and constant current cases.