Identification of response amplitude operators for ships based on full scale measurements

Abstract

Heerema Marine Contractors (HMC) operates several crane vessels, barges and tug boats for the transportation, the installation and the removal of offshore facilities. To achieve safe and successful projects, the accurate prediction of vessel motions is of great importance. Vessel motions are calculated on the basis of wave energy spectra and motion Response Amplitude Operators (RAOs). The wave energy spectra are received from meteorological services or from wave rider buoys and the RAOs are typically calculated using diffraction software, based on potential theory. In order to validate the numerical models, HMC regularly monitors the motions of its vessels. In a number of cases, HMC has noticed that the calculated vessel motions do not fully match with the full scale measurements. This inaccuracy can occur due to several reasons such as forward speed of the vessel, awkward draught, viscous damping forces or the variations of mass distribution during the projects. The main objective of this graduation study is to develop a mathematical method which determines the motion RAOs using the full scale motion measurements and the available wave data. Basically, it is examined whether it is possible to calibrate the following model properties of the vessels: The potential added mass, the potential damping and the potential wave forces. The radii of gyration, the viscous damping matrix and the coordinates of the CoG. The procedure for the identification of the RAOs is applied on a specific vessel with a certain draft: the semi-submersible crane vessel Thialf at 22m draft in deep water. The first challenge of this graduation study is to find a method to express each element of the hydrodynamic database with a limited number of parameters for the entire frequency range. This is accomplished by applying the vector fitting method. According to this method, the hydrodynamic elements are fitted to approximation functions with a certain number of coefficients. In order to calibrate the motion RAOs, a sensitivity analysis is first be performed. The sensitivity analysis leads to the minimization of the number of parameters that is to be examined. The identification process is repeated several times until the best possible modifications are found. In order to choose the correct modification, the normalized root mean square error between the calculated and the measured vessel motions is determined. Apart from the normalised root mean square error, the selection of the possible modifications should be based on logical criteria as well. For this graduation study, the identification method is tested by several cases. These test cases are based on simulated data. The results of this study shows that we can identify several parameters that lead to more accurate response spectra. However, in most cases the solution is not unique and we have to choose between the calibration of two parameters or more. For instance, the same results can be achieved by either changing the radius of gyration or a diagonal element of the hydrodynamic added mass and damping. This can be solved by examining the causes of inaccuracies. If the draft of the vessel is such that the awkward draft does not occur, then the hydrodynamic data base should not be modified. Small changes of the viscous damping could not be identified for the SSCV ‘Thialf’. The viscous damping is much lower than the hydrodynamic damping and as a consequence, it doesn’t have noticeable impact on the vessel responses. However, this is not valid for rolling ships such as the deep-water construction vessel Aegir. Finally, it should be mentioned that the errors of the data processing influence the accuracy of the identification procedure.