Aim: Optimisation of the hull for a new operational speed profile and on the design draught. Changes in the hull were only permitted within certain boundaries at stern and bow, according to constraints given by the shipyard.
This optimisation was carried out in 2013, for a well-known European cruise liner shipyard under the banner “ShipOpt”, a successful joint venture between Rodrigo Azcueta of Cape Horn Engineering and Jason Ker of Ker Yacht Design, merging the outstanding CFD capabilities and expertise of Cape Horn Engineering with the innovative optimisation techniques developed by Jason for the design of his successful sailing yachts. For more details on ShipOpt read at the end of this case study.
To increase simulation accuracy, a virtual propeller (actuator disk model) was used to simulate the thrust and torque of the propeller, meaning that all the results included the ‘thrust deduction’ component of the hull efficiency, which includes the suction of the propeller onto the hull.
The overall change in weighted power requirement was a 6.1% reduction. The breakdown for the new operational speed profile was:
- 9kts ( 10% weighting ) = Drag Reduction 7.9%;
- 17kts (55% weighting) = Drag Reduction 9.4%;
- 21kts (35% Weighting) = Drag Reduction 3.2%.
Background on ShipOpt Technology
Our optimisation system was developed for extracting tiny performance improvements from America’s Cup yachts, the Formula One of the Seas, in order to obtain a performance advantage in an environment where every second counts. This technology was perfectly transferable to ship hull optimisation.
We used proprietary optimisation tools and methods based on highly accurate RANS CFD simulations combined with surrogate model optimisation techniques. Hundreds of shapes were created using a parametric CAD model and the drag of those shapes was computed with CFD. We then trained multivariate response surfaces to recognize the correlation between the parametric shape combinations and drag in the considered conditions. Once the response surfaces are trained they are used to compute the drag of over a million virtual shapes, from which a short-list of candidates is decided upon and tested again with CFD, which are then compared against the baseline hull supplied by the client. We regularly use two of the best RANS CFD codes available on the market, and in that way are able to cross-check that the computations are well set up before starting the optimisation process, and then also cross-check its results.