Modal Analysis of an Offshore Platform using Two Different ARMA Approaches

In the present investigation, multi-channel response measurements on an offshore platform subjected to wave loads is analysed using Auto Regressive Moving Average (ARMA) models. Two different estimation schemes are used and the results are compared. In the first approach, a scalar ARMA model is used to estimate the poles (eigenfrequencies and damping ratios), and then the mode shapes are found by fitting an analytical form to the empirical estimates of the covariance functions using the estimated poles. In the second approach, a full vector model is used, and the pole and mode shapes are estimated in one step by solving an eigenvalue problem. Both of the models assume 14 modes, some of them are considered non-physical. Results for the 8 most significant modes in the estimations are compared.