On the Estimation of Directional Wave Spectrum Based on Stationary Vessels 1^st Order Motions: A New Set of Experimental Results

The practicability of estimating directional wave spectra based on a vessel 1^st order response has been recently addressed by several researchers. The interest is justified since on-board estimations would only require only a simple set of accelerometers and rate-gyros connected to an ordinary PC. The on-board wave inference based on 1st order motions is therefore an uncomplicated and inexpensive choice for wave estimation if compared to wave buoys and radar systems. The latest works in the field indicate that it is indeed possible to obtain accurate estimations and a Bayesian inference model seems to be the preferable method adopted for performing this task. Nevertheless, most of the previous analysis has been based exclusively on numerical simulations. At Polytechnic School, an extensive research program supported by Petrobras has been conducted since 2000, aiming to evaluate the possibility of estimating wave spectrum on-board offshore systems, like FPSO platforms. In this context, a series of small-scale tests has been performed at the LabOceano wave basin, comprising long and short crested seas. A possible candidate for on-board wave estimation has been recently studied: a crane barge (BGL) used for launching ducts offshore Brazil. The 1:48 model has been subjected to bow and quartering seas with different wave heights and periods and also different levels of directional spreading. A Bayesian inference method was adopted for evaluating the wave spectra based on the time-series of motions and the results were directly compared to the wave spectra measured in the basin by means of an array of wave probes. Very good estimations of the statistical parameters (significant wave height, peak period and mean wave direction) were obtained and, in most cases, even the directional spreading could be properly predicted. Inversion of the mean direction (180° shift), mentioned by some authors as a possible drawback of the Bayesian inference method, was not observed in any case. Sensitivity analysis on errors in the input parameters, such as the vessel inertial characteristics, has also been performed and attested that the method is robust enough to cope well with practical uncertainties. Overall results once again indicate a good performance of the inference method, providing an important additional validation supported by a large set of model tests.