Deterministic prediction of the ship motions induced by sea states – PREDEMO-NAV

The following project falls within the general scope of enhancing the safety for operations at sea and increasing the operability of naval floating structures, for defense as well as civilian purpose. This can be achieved through the deterministic prediction of the motion of a ship operating in waves to a forecast horizon of about 1 minute. The forecast horizon for up to date technics does not exceed ten seconds for class 5 sea states though, through autoregressive methods based on the history of the ship motion alone. For more severe sea states, the natural variability of the waves is quite important but not accounted in those models. Indeed, increasing significantly the forecast horizon requires knowledge of the direct environment around the ship, which includes waves. The main technologies for remote sensors able to measure surfaces of oceans from a ship are related to band X RADAR. Nonetheless, one of the main limitation of this technology remains the poorly identified relation between the intensity of the backscattered signal and the actual surface elevation.
This project targets different systems and methods aiming to constitute the links in a deterministic forecast chain for the motion of a vessel at sea. For a forecast horizon of about one minute, three main bottlenecks need to be properly addressed:
• Measuring sea surface from grazing angles from a moving platform
• Reconstructing, assimilating and propagating measurements in real time down-wave to the platform
• Estimating the response in time of the vessel
The proper performances of the tools in terms of computational time are compulsory in order to prevent a degradation of the forecast horizon.
On a first hand, this project includes the evaluation of a novel remote sensing technique of the ocean surface, based on a 3D Flash-LIDAR camera emitting a matrix of laser rays to reconstruct ocean wave profiles. Coupled to an inverse algorithm, those data in space and time will provide a reconstruction of the surface over an observation area located several hundreds of meters ahead of the camera.
Including nonlinear hydrodynamic processes is a key step, not only for the ocean surface reconstruction, but most of all for its forecast in time. On a second hand, the reconstructed data will then enable the down-wave accurate propagation in a non-linear wave model. This work includes the in depth analysis of the accuracy for the Lagrangian method developed for the specific purpose of accelerated computation of the non-linear propagation problem in real time on a dedicated architecture (GPGPU). This evaluation will be achieved in regard to a reference hydrodynamic method (High Order Spectral Method) and measurements database.
On a third hand, this project will address the determination of a ship motion response accurately and in real time. The problem will be studied for two geometries for the frigate class (generic and proprietary) and one generic geometry for a different class. Several specific tools from BEM(linearized, ALE, weak-scatterer) and CFD methods (LS-SWENSE) will be complementarily used for the proper evaluation in real time of the ship motion, including its appendages. Their accuracy will be evaluated by comparison with experimental database of ship motion in waves.
This project will finaly provide a global evaluation of the prediction chain generated from the previously described links.