DYNAMIC SELECTION OF SHIP RESPONSES FOR ESTIMATION OF ON-SITE DIRECTIONAL WAVE SPECTRA

Knowledge of the wave environment in which a ship is operating is crucial for most on-board decision support systems. Previous research has shown that the directional wave spectrum can be estimated by the use of measured global ship responses and a set of transfer functions determined for the specific ship. The approach can either be based on parametric or Bayesian (non-parametric) modelling, and in both cases a set of three ship responses usually provides the best estimation. The optimal response combination of three responses at any time depends on the environmental conditions and the operation of the ship. Since measurements of more than three responses are usually available, a quick, dynamic selection procedure of the three signals best suited for the wave spectrum estimation procedure is essential. In the present paper the concept of a selection method based on a simple pre-estimate of the wave spectrum is suggested. The selection method needs to be robust for what reason a parameterised uni-directional, two-parameter wave spectrum is treated. The parameters included are the zero up-crossing period, the significant wave height and the main wave direction relative to the ship's heading. The procedure basically seeks to minimise the difference between a set of spectral moments derived from a measured response spectrum and the corresponding response spectrum calculated from the parameterised wave spectrum and the transfer function for any given response. Subsequently, the three responses with the best overall agreement are selected for the actual estimation of the directional wave spectrum. The transfer functions for the ship responses can be determined using different computational methods such as strip-theory, 3D panel codes, closed form expressions or model tests. The uncertainty associated with transfer functions depends on the computational method used, relative heading, frequency and type of response. This uncertainty is conceptually taken into account in the selection procedure using the First Order Reliability Method (FORM). However, specific assessment of the uncertainties has not been carried out, but is subject to later studies.