Hull Generation for Fast Concept Exploration

Abstract

Damen Shipyards Gorinchem has found deficiencies in the process of developing concept designs of Offshore Patrol Vessels (OPVs). There is a need to increase the efficiency of this process, meaning that (1) the speed has to be increased, (2) the accuracy of designs has to be increased and (3) the risk of deviations of the project plan has to be reduced. Analysis of the current concept design phase shows that the designer should get more insight in the design challenge, and that a more systematic and uniform design approach is required amongst different involved departments. Additionally, the lead time of developing concept designs is mostly delayed by two major bottlenecks: development of the hull shape and obtaining a resistance prediction. In this thesis a conceptual design tool is developed to resolve these deficiencies, by providing a platform for fast hull concept exploration. Key of this approach is to pre-compute time-intensive design aspects such that the most critical activities are done before starting the design process. It uses a set of parent hulls for which the resistance is obtained by RANS CFD simulations. Based on the parents, a semi-parametric modelling technique is used to create new hull geometry. A kriging surrogate model is used to provide a RANS CFD resistance prediction to that new hull geometry. A brute-force approach is then adopted to fill a database with feasible hull shapes, which can be used by the conceptual design tool for hull shape concept exploration. To test this concept a proof-of-concept database of 245.005 hulls of double-ended ferries is generated in a few days on a laptop with good memory capabilities. The geometry results and resistance predictions in the database have been assessed with a sensitivity study, cross-validation and comparison with the parent hulls. It is shown that resulting geometry is smooth and well-faired. The resistance results are sufficiently accurate for use in the concept design phase, but applicability is limited for low prismatic coefficients. For application to OPVs further research into application of this approach to hull geometries with varying local details (bulb shape, tunnels etc.) is necessary. The approach seems promising to improve the concept design phase, as it (1) reduces time pressure in the design process significantly, (2) provides insight in how the resistance is affected by main dimensions of the vessel, and (3) reduces the number of departments which is actively involved in the design process. Putting the design tool in practice should reveal if this is indeed the case.