Blind Validation Case Study – CFD Simulations vs Towing Tank and Sea Trial Results Motor Yacht Fleming 58
The Australian company Norman R. Wright & Sons Pty Ltd commissioned Cape Horn Engineering to conduct a blind CFD validation study using towing tank data for the motor yacht Fleming 58. The aim of the validation was to increase confidence in the current CFD simulations and to demonstrate that they are a good alternative or complement to tank testing for future projects. The 58 feet motor yacht from Fleming Yachts in Taiwan was tank tested at the Australian Maritime College towing tank in Tasmania.
Cape Horn Engineering conducted the CFD simulations without having access to the towing tank results and sea trial data. Only a detailed description of the towing tank setup and tested conditions was known before the CFD study was completed.
The CFD programme included the following tasks and aims:
• Running the CFD at model scale matching the tank setup as closely as possible to validate the CFD simulations
• Running CFD at full scale with the bare hull as tested in the tank to verify the towing tank extrapolation procedures
• Running at full scale with the fully appended yacht and compare with sea trials to verify the designers and towing tank extrapolation procedures and Overall Propulsive Coefficients
The simulations were carried out following Cape Horn Engineering’s best practices for motor yacht simulations, with 2 Degrees-Of-Freedom (sinkage and trim), free surface deformation, viscous and turbulent flow using the best in class software package Star-CCM+ in its latest version.
The simulations at model scale were conducted for 2 conditions as described in the AMC report, one without and one with a bulbous bow. The yacht displacement at full scale is 41 tonnes. 7 and 5 speeds were run for the 2 conditions, respectively. Great care was taken to reproduce the tank conditions as closely as possible, i.e. same water properties, same position of centre of gravity and pull force in the simulations as in the tank.
Figure 1 shows the external force applied during the simulation to model the tank pull at model scale (top image) and the propeller thrust at full scale (bottom image). In the model scale simulations, it is a horizontal force at the towing tank pull position and in the full-scale simulation it is a force aligned with the propeller shaft at the propeller position. Both types of propulsion methods have a different effect on the final pitch of the yacht. This effect can be taken into account in the CFD, however, such corrections are usually not used for towing tank full-scale extrapolation.
Figure 1: External force modelling the tank pull and the full-scale propulsion
The volume meshes have resolutions of around 3.3 million cells for the bare hull simulations both at model scale and full scale (one side of the model). The fully appended cases have around 4 million cells, see Figure 2. The mesh resolution on the appendages has been kept deliberately low to speed up the analysis. Resolutions of 3 to 4 million cells are not necessarily high. However, the purpose of this exercise was to validate CFD simulations performed on meshes that are appropriate for providing a standard service to the industry, rather than using a setup with the highest possible resolution which is only affordable as an academic exercise. Services provided to the industry by Cape Horn Engineering use such mesh resolutions as a standard and follow best practices gathered during years of experimentation, so that the services are attractive for projects in the recreational industry with typically moderate budgets.
Figure 3 shows the drag differences as computed with CFD and measured in the towing tank for the 2 investigated condi- tions. There are some uncertainties related to the towing tank setup. Firstly, the vertical position of the centre of gravity VCG was not known. It is not common to determine the VCG of the model in the towing tank, due to the complexity of that measurement. However, VCG has a (minor) effect on the longitudinal stability of the vessel and thus the final pitch of the model at speed. For all simulations VCG was assumed to be at the water plane. Secondly, the boat speeds reported in the tank report had 2 significant decimal places, and a conversion to knots gives slightly different values to the ones used in the simulations.
Despite these uncertainties, the correlation is acceptable. There are some differences at intermediate speeds but for most of the other speeds the differences are less than 2 %. The measured and computed running trims are also in very good agreement.
Figure 3: Simulations and towing tank delta drag comparison in % (model scale)
Figure 4 shows the differences in effective power to the full scale simulations with the bare hull. The orange line corresponds to the fully appended simulations and it shows more drag as expected compared to the bare hull, around 5 to 7 %. The yellow line compares the effective power reported by the towing tank following the 1957 ITTC Performance Prediction Method. The grey line adds a correction of 3 to 5 % for appendage drag following the client’s own procedures based on experience and handmade calculations.
The effective power from the tank extrapolation is in general 5 to 10 % higher than for the simulations. This is expected since the ITTC 57 ship-model correlation line follows a conservative approach widely used by towing tanks that includes a 10% form factor compared to other correlation lines. Also, compared to the full scale simulations, there is no correction undertaken in the extrapolation procedure to account for the difference in propulsion force alignment.
Figure 4: Delta power comparison (in % to bare hull simulation)
Figure 5 shows the Overall Propulsive Coefficients (OPC = Effective Power/Engine Power) as determined using simulation results and engine power measured during sea trials. OPC values from the CFD simulations and from the client’s own workings are in very good agreement, in particular for the top 2 speeds.
Figure 5: Overall Propulsive Coefficient curves from simulations
It has to be noted that no corrections for windage and hull roughness have been added to the simulation results, thus the OPC values might be overestimated to a certain extent. Only windage of the hull topsides in zero wind condition and with a flat deck was considered in the simulations, but not windage of the superstructure. The hull surface was considered as a hydraulic smooth surface which is standard practice for RANSE CFD simulations with turbulent models. No attempt has been made to increase roughness numerically in the simulations.
A more extensive version of this report is available under request.