Operational optimization of vessels is valuable for the planning and execution of maritime operations. Accurate and efficient models to predict vessel motions are needed to make reliable operational decisions. The wave-induced vessel response can be modelled in terms of a Response Amplitude Operator (RAO) and a wave spectrum. Uncertainties related to the parameters that govern the RAO can significantly influence the reliability of the vessel motion prediction. To decrease these uncertainties, the maritime sector has realized the potential of using vessel motion measurements. As a result, it is envisioned that a vessel response model might include an identification module that searches for model parameters using measurements of responses to make a reliable prediction.

This study presents an identification procedure to handle the inherent uncertainties of vessel model parameters, aiming to improve vessel motion prediction. The identification procedure identifies the vessel's RAO by the measured response spectrum and nowcast wave spectrum, with the goal of finding the heave and roll natural frequencies. The natural frequencies provide information on the vessel’s parameters. This is used to identify the parameters related to the mass distribution and damping of the vessel. These were found by minimizing a cost function, that quantified the difference between the measured and predicted response spectrum, using an optimization method. Identifiability analyses of the parameters were performed on two case studies.

For the first case study, a synthetic data set is created with the vessel response model to simulate the vessel motions. Tests were conducted with five different wave spectra and several vessel headings, constituting diversified scenarios. The RAO was identified by the measured response, the wave spectrum, and a sinusoidal function to describe the directional dependency of the RAO. Using the synthetic data set, the identification algorithm successfully identified the parameters with good agreement to their actual values. The second case study involved the examination of parameter identification on real onboard vessel motion measurements. In most of the cases, the RAO could be identified from the measurements and the natural heave and roll frequency was found. The identified parameters resulted from the identification procedure and improved the vessel motion prediction compared to the initial prediction, but still, deviations remained. The identified parameters are verified against a different measured data set. The results show that the identified response spectra approach the measured responses, indicating that the identified parameters are reusable.

In summary, it was found that the parameters have a great influence on the output of the vessel response model. Therefore, it is essential to have a thorough understanding of the correct operational parameters for accurate motion prediction. The established identification procedure shows to be a good addition to existing vessel motion models to identify input parameters at relatively low computational cost.