Wind propulsion systems (WPS) are major investments and the decision to install them requires careful consideration of many complex questions. One of the recurring and challenging issues for ship owners is the choice of a suitable WPS for a specific ship and a specific operational pattern. Today most WPS providers offer on-demand case studies, but obviously the underlying performance prediction methodologies differ from provider to provider. This makes comparing different technologies from competing suppliers next to impossible. In this paper we present a numerical platform to compare different WPS of different makes, sizes, and costs in a fair way. The fundamental idea is to use aerodynamic WPS datasets that are independently verified by SSPA through wind tunnel test, sea trials or extensive CFD. This is combined with a hydrodynamic dataset from SSPAs database of tank tests. The same performance prediction method, identical routes and weather statistics are then used to determine Key Performance Indicators and financial metrics of the competing wind propulsion technologies. The purpose is to provide guidance for shipowners at the early concept stage of a vessel and help them select a system that suits their particular requirements.

Accurate performance prediction is necessary when designing/optimising wind propulsion systems (WPS). An independent, trustworthy prediction of the energy-saving potential is also needed to support the ship owner’s decision to invest in new technology. By using weather statistics along with a mathematical model of ship performance, route simulations can estimate the time and power required for transit of a route. Such simulations are commonly used today to optimise the design and operation of conventional ships. The introduction of WPS poses additional challenges for route simulations. WPS performance must be predicted at all points along the route, with wind of differing velocity and direction. The apparent wind will vary vertically (twist), due to the interaction between the ship velocity and the atmospheric boundary layer. Also, many proposed concepts use multiple WPS, introducing additional complexity, such as independent spin ratios/ sheeting angles. 3D CFD simulations capture the complex physics, including vortex formation and interaction effects, providing accurate performance prediction and an understanding of the flow. However, 3D CFD is costly, and it would not be possible to simulate all conditions at a reasonable cost. We present simplified approaches to modelling of WPS, using a limited number of CFD simulations, either in 2D or 3D, which are then extrapolated such that 3D effects are represented, and all conditions covered. The methodology is demonstrated on rotor sails and wing sails.

Full scale measurement, model testing and a hybrid CFD method were used to characterize the Underwater Radiated Noise (URN) of a ship at design speed. The CFD method consists of a multiphase Delayed Detached Eddy Simulation and the Ffowcs-Williams Hawkings acoustic analogy. The paper discusses the correlation of the noise spectra with the observed cavitation behavior and compares the measured full scale data with those predicted by the model testing and the CFD method. The comparison shows that the sheet cavity and Tip Vortex Cavitation (TVC) predicted by the model testing are in reasonably good agreement with the full scale observations. The pressure pulses are somewhat higher than the full-scale data. Overall, the predicted URN has a good correlation with the noise spectra obtained from the sea trial. The CFD method shows the potential to resolve turbulence eddy structures in the wake. It captures the dynamic development of sheet cavitation and the collapse and rebound of TVC as observed in the model test and the sea trial, but under-predicts the extent of TVC. The pressure pulses and tonal noise are in close agreement with the respective measured data for the first five orders of blade passing frequency. The method underestimates the broadband noise level in the frequency range 50-112 Hz where the TVC is expected to have an important contribution. The maximum under-prediction in this range is about 28 dB at 72 Hz. At frequencies above 200 Hz, the broadband noise becomes more and more under-predicted with increasing frequency. 

The paper presents SSPA’s work in the EU project AQUO to predict underwater radiated noise (URN) generated by a coastal tanker with a cavitating propeller. A CFD method, consisting of a multi-phase Delayed Detached Eddy Simulation (DDES) and a Ffowcs Williams-Hawkings (FWH) acoustic analogy, is applied to predict the cavitation, pressure pulses and radiated noise for the ship at model and full scale. In comparison with the data obtained from the model test and full scale measurement, it is found that the predicted sheet cavity correlates quite well with the observed ones in the experiment and sea trial. Some success is made in predicting the collapse and rebound of tip vortex cavitation (TVC) at model scale, yet the extension of TVC is under-predicted. The predicted pressure pulses agree reasonably well with the measured ones at the first three harmonics, deviation becomes larger at higher harmonics. The tonal noise has fairly good agreement with the measured signal at both scales up to 5th harmonics. The simulation however under-predicts part of broadband noise that is caused by the TVC, mainly due to an under-resolution of the flow in the tip region and the propeller wake. The agreement with the data for the model scale case is slightly better than that for the full scale case.

The paper presents SSPA's work in the EU project AQUO to predict underwater radiated noise (URN) generated by a coastal tanker with a cavitating propeller. A CFD method, consisting of a multi-phase Delayed Detached Eddy Simulation (DDES) and a Ffowcs Williams-Hawkings (FWH) acoustic analogy, is applied to predict the cavitation, pressure pulses and radiated noise for the ship at model and full scale. In comparison with the data obtained from the model test and full scale measurement, it is found that the predicted sheet cavity correlates quite well with the observed ones in the experiment and sea trial. Some success is made in predicting the collapse and rebound of tip vortex cavitation (TVC) at model scale, yet the extension of TVC is under-predicted. The predicted pressure pulses agree reasonably well with the measured ones at the first three harmonics, deviation becomes larger at higher harmonics. The tonal noise has fairly good agreement with the measured signal at both scales up to 5th harmonics. The simulation however under-predicts part of broadband noise that is caused by the TVC, mainly due to an under-resolution of the flow in the tip region and the propeller wake. The agreement with the data for the model scale case is slightly better than that for the full scale case.

The paper presents a study of using a modified k-ω model to predict the unsteady cavitating flows around 2D and 3D hydrofoils in the framework of multi-phase mixture flow RANS approach. The cavitation is modeled by Schnerr-Sauer’s cavitation model. A 2D NACA0015 foil at cavitation number σ=1.0 (unsteady with cloud shedding) is studied first, followed by the Delft twisted 3D foil. It is found that the present RANS method is able to predict the essential features like re-entrant jets, the periodic shearing and shedding of cloud cavities. Two distinct shedding dynamics are noted for the 2D foil: (a) Shedding of medium to large scale structures (at low frequency); (b) Shedding of a series of secondary vortex cavities (at high frequency). For the 3D twisted foil, the collaborated effect of re-entrant jets from the curved closure line to break up a primary cavity, as well as the formation, roll-up and transport of cavitation vortices that are observed in the experiment are truly reproduced in the simulation. The method is found to have a tendency to under-predict the lift coefficients.

Abstract: The paper presents a validation study of a RANS prediction of open water characteristics of a highly skewed propeller in full range advance-ratio. Computational results are compared with open water data, and extensively validated at one loading with paint test, cavitation inception test and PIV measurements. Except at two extreme off-design loads, the difference between prediction and measurement is less than 3% for KT and less than 5% for KQ at a wide range of advance ratios. While this can be considered as good agreement with experiment for KT and KQ, an attempt is made to critically analyze the predicted flow detail and some possible deficiency of the existing RANS method is discussed. Part of the work was carried out within the EU-funded research project "LEADING EDGE2 2 Leading Edge: Prediction of Leading Edge and Tip Flow for the Design of Quiet and Efficient Screw Propellers (Contract No: G3RD-CT-2002-00818). 

The paper presents a validation study of a RANS prediction of open water characteristics of a highly skewed propeller in full range advance-ratio. Computational results are compared with open water data, and extensively validated at one loading with paint test, cavitation inception test and PIV measurements. Except at two extreme off-design loads, the difference between prediction and measurement is less than 3% for KT and less than 5% for KQ at a wide range of advance ratios. While this can be considered as good agreement with experiment for KT and KQ, an attempt is made to critically analyze the predicted flow detail and some possible deficiency of the existing RANS method is discussed.