ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN

ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN EJ de Ridder [1] G Gaillarde [2] F van Walree [3] ABSTRACT Since the beginning of high level sailing events, like the America s Cup and Volvo Ocean Race, designers have used a varied and increasing set of techniques on the experimental and later on the numerical field in design sailing yachts. This paper will give the latest developments for the different fields at MARIN. In the field of CFD calculations, the program RAPID has been extended with a sailing yacht module, which solves the exact, fully non-linear potential flow problem by an iterative procedure. This paper highlights the comparison of CFD results with some experimental results, for a small range of different sailing yacht designs. A small introduction will be given about the new underwater, three-component particle image velocimetry system (3c-PIV or stereo- PIV). With aid of such a system it is possible to measure and visualise the flow around appendages such as keel, winglets bulb and rudder. Even if investigated in many occasions in the past, mainly through calculations, performances in dynamic conditions are not yet part of the standard design spiral. Seakeeping and manoeuvring aspects require more variables than steady conditions and require more complex numerical and experimental tools, which probably explain partly their nearly total absence into the design process. At a time where optimisation has reduced drastically the scatter into hull performance, seakeeping and manoeuvring aspects can be seen as new fields open for potential increase in performance and create the difference with other designs. [1] Project Manager Ships-Propulsion [2] Project Manager Ships-Seakeeping [3] Project Manager Ships-Seakeeping Maritime Research Institute Netherlands (www.marin.nl) The present paper discusses the current state at MARIN for each of those fields, their applicability and limitations. The demand for high accuracy tools is increasing, and techniques to further improve the sailing yacht design are becoming available. List of symbols Fn = Froude Number Lwl = Waterline length Vs = Forward boat speed Rrh = Residuary Resistance Hull Vrh = Frictional resistance hull Rn = Reynolds number INTRODUCTION The Maritime Research Institute Netherlands, MARIN, is a research institute that offers hydrodynamic services for the maritime industry. The aim of the company is to be on the leading edge of the technology field. As such MARIN is active in different fields of the maritime industry. MARIN is involved in day-to-day model testing of floating structure, merchant vessel, naval ships or motor yachts, and nowadays more often in testing sailing yachts models. We entered this field in 198 when most people thought that the development of the America s Cup yacht of those days, the International 12-Meter, was fully developed and no design breakthroughs could be made. Our involvement resulted in the development of the winged keel of the Australia II. After this we carried out experiments for a number of syndicates for the 1987 AC. Since then the Cup went through a transition period in which MARIN was not involved. In this period MARIN worked for a number of oneoff sailing yachts for Dutch yards. We developed a close relation with the Technical University of Delft in this field of expertise. For the 23 America s Cup MARIN carried out experiments for the Mascalzone Latino syndicate. Nowadays MARIN is involved in research projects for the 25 Volvo Ocean Race and the Americas Cup of 27. ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 1

This paper describes the various aspects of different performance prediction tools for sailing yachts at MARIN. Below we first describe our in-house potential and viscous codes. We will show some result of the potential CFD Code RAPID which is developed by our CFD department and which is continuously under further development to improve the efficiency and accuracy. Because the use of viscous-flow predictions is now increasing, we will also discuss the usefulness of our code called PARNASSOS, to attack possible design problems. Both programs have a very good reputation in merchant shipping, and RAPID has recently been extended with the possibility to take in to account the drift and heel angle of the yacht during computations. The second section will give a small introduction about PIV measurements and the possible usefulness of this system. The last section will give the different possibilities on the numerical and experimental field for sailing yachts in waves. New computational seakeeping methods such as our in-house programs PANSHIP or PRETTI as well as a new experimental facility dedicated for free sailing tests in all types of wave directions, in service since 6 years, may open new field of investigations and optimisation for designers and syndicates. POTENTIAL AND VISCOUS FLOW METHODS The rapid development of computational Fluid Dynamics (CFD), i.e. the prediction of flow phenomena by numerical solution of a mathematical model of the flow, has resulted in a strongly increasing role of computations in hydrodynamic sailing yacht design. Not only for the design of racing yachts but also for cruising yachts. With the aid of this kind of codes mainly the wave resistance and side force production of a design can be calculated relatively quickly and easily, compared to model testing. Another significant advantage is the possibility to predict the effect of very small and local design changes, which are normally not well predicted by VPP programs. Moreover, the output from CFD codes, in the form of detailed flow visualisations, provides a much more comprehensive insight in the flow characteristics than a model test. Subsequent analysis of the results by an experienced designer provides valuable indications on hull form modifications which are likely to improve the flow and reduce the resistance. This section will give a short explanation of the potential and viscous flow codes developed at MARIN and their applicability in sailing yacht design. Potential flow code: RAPID Free-surface potential flow is a mathematical model of a flow that neglects viscous effects (friction, diffusion, boundary layers, wakes, breaking waves), but takes into account the presence of a water surface that deforms due to wave making. This makes it an adequate flow model to predict the wave pattern and wave resistance, the side force on keel and rudder, the induced resistance; but not for the viscous resistance, scale effects, or viscous effects on (stern) wave patterns. Notwithstanding the limitations, this class of methods is the most widely and successfully used in hydrodynamic ship design. At MARIN, in 199-1994 a method has been developed to compute steady nonlinear freesurface potential flows, the code RAPID [1, 2]. It is used extensively in ship design since 1994, at MARIN and by several licencees worldwide, and has proved indispensable in that regard. RAPID is well known for its advanced numerics and high accuracy, and for various innovations introduced in its development, e.g. [3, 4]. At MARIN, dense discretisations are used as a standard, in order to achieve best numerical accuracy of the results. A range of validations has been made over the years, see e.g. [4]. One example is shown in Figure 1, which shows longitudinal cuts through the wave pattern at a distance of.48 ship length out of the centreplane, for two variations of a design of a RoRo vessel [5]. Clearly, there is excellent agreement between the computed and measured wave pattern, even at this relatively large distance, a feature that is not easy to achieve but which is essential to provide the right analysis of the wave making and its causes. Also the difference between the design variations is very well predicted, permitting a precise hull form optimisation based on computations. ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 2

Figure 1 The method A potential flow can be generated as the sum of many flow fields induced by so-called sources, dipoles or vortices, plus the uniform onset flow due to the ship s speed. For a normal ship hull, sources are located on the wetted hull surface, spread out over source panels. The sum of all induced flow fields must be such that the flow passes around the hull surface, not through it. From this requirement, the source strengths can be solved, and the entire flow field can be derived. However, the free water surface, the shape of which is not known beforehand, poses additional requirements: again the flow must pass along that surface, but also the pressure must be atmospheric there. These boundary conditions require the use of additional panel distributions for a part of the water surface surrounding the ship. Again the boundary conditions can be expressed in terms of the unknown source strengths. The result now is a large system of equations for the (few thousands) source panel strengths on the hull, and the several thousands of panels on the wave surface; which can be solved to find the source strengths and again the entire flow field and wave pattern. But as mentioned, the shape of the wave surface, at which the boundary conditions must be satisfied, is not known beforehand. As in RAPID the conditions are imposed without any approximation, in fully nonlinear form, a stepwise approach to the solution needs to be used. In several iterations the wave surface is successively improved. After each iteration, the attitude of the ship is adjusted so the weight distribution and the hydrostatic and hydrodynamic forces are balanced, to take into account the dynamic trim and sinkage; the wave surface is updated; and the source panel distributions on the hull and wave surface are adjusted so their intersection is precisely matched. When in the iteration process no further change occurs in the wave surface shape and trim and sinkage, the final solution has been obtained. Usually, 1 to 3 iterations are needed. On a modern PC with sufficient memory, a computation for a bare hull sailing yacht with 5 panels can be completed in just a few minutes, to 2 minutes for an AC yacht with keel, rudder and winglets under heel and drift as shown in figure 21. Extensions for sailing yachts In 1997, extensions of the code were made for sailing yachts [6]. First requirement was to model a-symmetric ships (sailing yachts at heel) and ships with a leeway angle. Also, lifting surfaces had to be added. Flow around a lifting surface poses additional requirements and cannot be modelled by source panels alone. In RAPID, the solution is based on source panel distributions on the outside of the lifting foils, plus additional internal vortices that extend aft to infinity as trailing vortices. The strengths of these follows from the requirement that the flow must leave the trailing edge of the foils smoothly. This requirement is added to the same set of equations and solved simultaneously. Thus we obtain the lift on the foils and the precise spanwise and chordwise lift distributions, in interaction with the flow around the hull and the wave making. Figure 24 shows a panel distribution as used for an AC yacht sailing under heel and drift. Recent Developments Recently the code has been further extended and improved for more extreme geometries and conditions. In addition to the more substantial additions, a variety of small issues was solved to deal with the complicated geometries, derive all desired hydrodynamic quantities etc. The following main adjustments have been incorporated in the code: When calculations are performed for sailing yachts with large overhangs the underwater hull shape changes considerably in the course of the iteration process, due to the formation of the waves. This not only applies to the stern overhang, but also the bow overhang present on today s ACC yacht ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 3

hulls. Adjustments were made in the automatic determination of the intersection of hull and wave surface panel distributions, and the hull repanelling in the course of the iteration, to cope with this problem. For strongly flared hull and large overhangs, free-surface panels close to the hull may become much narrower at a wave crest, much wider at troughs. The quick changes and large variations in panel size may destabilize the iteration process and reduce the numerical accuracy. To avoid this problem and to improve the calculations, the program now repanels the wave surface when the geometry of the panels changes too much. As mentioned, the determination of trim and sinkage for normal vessels just balances the weight and pressure distribution on the hull. This was extended to make it possible to take in to account the sail forces and moments, as is common in model tests experiments. The post processing has been extended resulting in better visualizations, which is indispensable for inspection of the flow field, analyzing phenomena as lift carry-over and the interaction of side force and wave making, and so on. Some of the figures illustrate the possibilities of such tools. Lifting components cause a so-called induced resistance, connected with the generation of trailing vortices. This resistance component is hidden in the resistance found by integration of pressure forces over the hull, which in addition includes the wave resistance. For analysis and optimizations purposes, it is very interesting to be able to distinguish these two different forces, since they respond to different design parameters. In the code this is made possible by a separate determination of the induced resistance far aft of the ship, by a so called Trefftz-plane analysis. How to do this in the context of a nonlinear free-surface panel method is, however, far from obvious. A brief study has been conducted on this subject and on the behaviour of the spanwise loading near a free surface, and a practical approach has been found for an approximate determination of the induced drag. Validation and comparison In this section, some comparisons will be presented of calculations and experimental data for a sailing yacht. The model test data of the Sysser hulls were kindly made available by the Ship Hydromechanics laboratory of Delft University of Technology and were produced during the master thesis of E. Lataire. Comparison of calculated and measured wave profile along the hull By comparing the calculated wave profile along the hull with the measured profile in the tank, a more accurate validation can be made of the code than by only comparing the forces; although there is some uncertainty of the experimental wave pattern due to scale effects, spray and other effects. In these calculations the mesh on the hull consisted of about 2 panels, and the free-surface mesh consisted of a total of 25 panels per symmetric half. The first comparison of the wave system is made for Sysser hull 27, which is a large displacement slender hull; the main dimensions are shown in table 1. Because of the available experimental data, the comparison is made for a bare hull yacht with no heel and leeway. The results for Fn=.3 and Fn =.45 are presented in the figures 12 and 13 for the full scale yacht. Two different calculated wave profiles are shown: The first wave profile is the one at the centre of the free-surface panels along the hull, which actually are located at half a panel width off the waterline; The second profile is obtained from the hull pressure distribution, and is directly on the hull. In these cases the two profiles agree closely. In figures 15 and 16 the wave pattern is shown for Fn =.45 and Fn =.6 for Sysser 26 which is a low-displacement design, the main dimensions are shown in table 1 LWL BWL TC Volc Sc [m] [m] [m] [m^3] [m^2] Sysser 26 1 2.54.194 1.975 17.163 Sysser 27 1 2.224.94 7.948 21.61 Table 1 From the comparison it follows that in general there is good agreement between ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 4

the measured and calculated wave profiles. The largest difference occurs at the bow, and can be mainly explained by the thin sheet of water climbing along the hull surface which was present during the tank test; and the spray which results in an over prediction of the bow wave during the visual inspection of photographs of the tank tests to determine the experimental profile. Although RAPID is a potential-flow code, the wave profile is predicted very accurately even up to the stern, where viscous effect could play a role. For slender hulls like these sailing yachts, such good wave pattern predictions can be obtained. This is very important to predict the dynamic waterline length for sailing yachts with large overhangs at the stern. Figure 14 and 17 give an overview of the total wave pattern of both hulls at Fn=.45. The large difference of the wave height can be explained by the difference in displacement. The overview also shows the difference in strength of the transverse wave system, which can be partly explained by the differences in canoe body depth. Comparison of calculated and measured forces, trim and sinkage In this section the calculated resistance forces, trim and sinkage are compared with the measured values in the tank. For a single calculation RAPID will calculate the following forces: Wave resistance Induced resistance Sum of pressure drag, vertical and horizontal lift of all components Pressure Drag, vertical and horizontal lift for every lifting appendage (if present) separately. The wave resistance is calculated by some different methods. The most straightforward is a summation of all longitudinal pressure forces over the hull up to the wavy waterline. This is made numerically more accurate by correcting for the integral of the hydrostatic pressure forces up to the static waterline. Nevertheless, rather dense panel distributions on the hull are needed to get accurate results. Besides, a wave pattern resistance Rw3 is obtained by an analysis of the computed wave pattern at a distance aft of the ship. A special technique is applied based on eight transverse cuts of the wave pattern in order to capture all the radiated wave energy. The first is more complete and closer to the measured values, the latter is computed in a different way so, if any error is present during the computation, it should be easy to detect it by comparing the results. The pressure integration results R w1 are the values used in this comparison. As the calculation only considers the inviscid flow, the comparison requires a correction for the viscous effects. To this end we added an estimated viscous resistance to the calculated values. This estimation was based on the calculated dynamic wetted surface and the same form factor was used as measured in the tank. Figure 19 shows the resistance curve for a Volvo 7 yacht in upright position without leeway. It is a low wetted surface and narrow design which has been tested at MARIN at a scale of 1:3, for a total of 2 different test points. For the extrapolation of the test results, the dynamic wetted surface and different form factors measured in the tank were used. It is clear that the resistance as predicted with RAPID is in close agreement with the measured resistance in the tank, both for the low and high Froude range. Because nowadays CFD calculations are used as tools to optimize a design, it is very important that the resistance is predicted accurately for a range of design variations. This resistance can then be used as input in a VPP in which the final design choice can be made. To validate RAPID for design variations we have made the following comparison; Figure 2 shows the total upright calculated and measured resistance of Sysser 26 relatively to Sysser 27. As already mentioned in the previous section Sysser 26 is low displacement design and Sysser 27 is a high displacement design. ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 5

%-diff [-]. -1. -2. -3. -4. -5. -6. -7. -8. -9. Upright resistance for 26 relatively to 27 Measured & Caculated M easured in tank Calculated RAPID.35.4.45.5.55.6.65 Vs [Fn] Figure 2 From the comparison it follows that the code predicts the differences in resistance well, only at the lower Froude numbers there is a small difference. In Figures 2, 22 and 23, a comparison is made of the measured and calculated trim for Sysser 26, 27 and the Volvo 7. The sinkage and trim of the models is generally well predicted, only at higher Froude numbers the calculated values are sometimes further away from the measured values. Drag [Kn] 3.5 3 2.5 2 1.5 1.5 Influence of the drift angle on the resistance forces heel= Vs=1-7 -6-5 -4-3 -2-1 Induced drag theory Wave+Induced Resistance RAPID Wave Resistance RAPID Induced resistance RAPID Drift angle[deg] Figure 3 During the post-processing fase, also the lift forces, yaw moment and the induced resistance of the lifting surface are calculated. The side force is evaluated by integration of the pressure forces over the wetted area below the actual (wavy) waterline. The induced resistance is included in the total pressure resistance Rw 1, but in addition it is calculated from the trailing vortex system Because it is not possible to compare the induced resistance with tank values, we have made a comparison with a calculated induced resistance according to by the Faulkner thin airfoil theory. Figure 3 shows the wave resistance and induced resistance calculated by RAPID and by the theory, for an AC type of yacht sailing under 1 knots of speed and no heel angle. Viscous flow code The code developed and used at MARIN to calculate the viscous flow around the hull of a ship is called PARNASSOS [7]. This solves the steady RANS equations for the flow around the hull. It provides detailed information on the velocity and pressure field around the hull, keel and rudder and the viscous resistance. PARNASSOS distinguishes itself from other RANS codes by its high numerical accuracy and very large computational efficiency. Validations have shown a generally good prediction of the flow and separation phenomena. Figure 25 gives an example of the flow around a skeg for a hull under 1 degrees of leeway. The flow separation presented as the black vortices at windward side of the skeg is clearly visible. At the moment, RANS codes are the state of the art computer programs used in sailing yacht design. Most of the time these programs are used in the field of Americas cup and Volvo ocean race yacht design. Due to increasing computer capacity, the use of those programs will be more common and very valuable in other design areas as well. One of the areas that could benefit are the large cruising yachts, for which the trend is visible that the yachts are increasing in length and not in draft, due to the restricted depth in the harbours. And the easiest way to decrease the draft of a large sailing yacht is by decreasing the draft of the keel, which can result that only 1 / 3 or only ¼ of the total draft is caused by the keel. Due to the low aspect ratio of this kind of keels and the large viscous effects over the keel and the aftbody of the yacht, the effect of the lifting surface is not modeled accurately in potential flow programs like RAPID, and viscous-flow programs will give more reliable results. ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 6

Because this code solves the RANS equations, it is the most appropriate model of the flow around the keel under the hull, as it includes all viscous effects and will predict the vortex system from the trailing edge and tip of the keel without any prior assumptions. By carrying out such a computation for the flow around the hull at a given heel angle and leeway angle, the result will show the vortex systems generated by the keel and its junction with the hull; may indicate possible improvements and will predict the lift, drag and balance of the yacht. The visualisation of the flow around the keel derived from the calculations (streamline pattern, pressure distribution, vorticity and velocity field etc.) will give a good insight in the nature of the flow over the keel and will indicate possible problems like flow separation or imperfect alignment of edges. PARTICLE IMAGE VELOCIMETRY A new non-intrusive technique that has recently entered the maritime research market is Particle Image Velocimetry (PIV) ([8], [9]). The experimental tool PIV yields unsteady and averaged 3D flow field. These data can be used to gain insight in the dynamics of unsteady flows and in the interaction of unsteady flows with immersed bodies. PIV is also a powerful tool to validate CFD-tools, by comparing both experimental en numerical flow fields. Further, it is possible to measure quickly the mean ship wake. The flow measurement with PIV is based on the measurement of the displacements x of particles in a target plane between two successive light pulses with time delay. The flow is seeded with particles and the target plane is illuminated with a light sheet. An overview of the measuring method is presented in Figure 26. The particle positions are recorded by two special digital cameras. One PIV-image consists of two image frames belonging to the two successive light pulses. Special image processing software analyses the movements of the group of particles in subsections of the PIV-image using correlation techniques. The output is an instantaneous velocity field in the measuring plane. The third velocity component perpendicular to the measuring plane can be derived by using two cameras in a stereoscopic arrangement. Therefore, this is called stereoscopic-piv or stereo-piv. More background information about this method can be found in relevant handbooks, for example [1]. A new underwater three component Particle Image Velocimetry system (3C-PIV or stereo-piv) is available now at MARIN for detailed flow measurements in towing tank and offshore basin. This system has been successfully used for the EU-project LEADING EDGE in 24. The challenge was to measure the flow near a rotating propeller in open-water condition. Figure 27 presents non-dimensional velocities in axial direction as measured by the PIV-system. These data are compared with computations. Another recent successful PIV campaign was focussed on measuring the three-dimensional flow around a manoeuvring ship in shallow water. The analysing of the huge amount of data is still being in progress. A typical result presented in Figure 28 shows clearly a vortex under the ship at lee side. SEAKEEPING General aspects All developments of IACC designs and sailing yachts in general are nowadays extensively relying on hydrodynamic tools, numerical or experimental. With hydrodynamic is meant the performance in calm water with constant speed and steady flow around the hull. However, larger dynamic effects are taking place in real life, as the flow around the hull is unsteady as well as the ship motions. Waves and ship motions are influencing the performance of the vessel. The contribution of the dynamic unsteady effects (summarised as seakeeping aspects) on the overall performance is depending on the environment in which the yacht is sailing but also on the motion characteristics of the yacht itself. Such influence of seakeeping aspects is quite clear when sailing in significant sea conditions, with sometimes rather dramatic evidence such as hull or rig damage or excessive ingress of water. However, even in light wind and wave conditions, seakeeping aspects will affect the overall performance mainly in terms of added ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 7

resistance in waves and lift and drag degradation (or increase) of appendages in unsteady conditions. Due to the complexity of such studies, adding a large amount of variables into the (already complex) design spiral, they are never undertaken completely nowadays and the real influence of seakeeping aspects on the overall performance is not clearly identified. However, in view of the development reached in calm water performance and the fact that most designs have now reached very similar maximum calm water speed, seakeeping aspects represent an underinvestigated field of development that may yield to a high potential gain in terms of yacht performance and differences with other designs. This section intends to present new calculations and experimental techniques and their potential power though a recent example. A full investigation having not yet been conducted, the following section place the first stones of what could potentially be an interesting new field of development. races the climate definition is obviously much wider but design can be optimised for segments of the race, when prevailing climate are steady and prominent for the issue of the race. Concerning the Valencia area, the wind and wave data were obtained from the Spanish harbour authorities (www.puertos.es) and were analysed in-house. Depending on the type of measurement systems that were installed of the coast of Valencia, different types of data are available. Significant wave height and periods were available from 1985 to 25. Wave direction is only measured by means of a Triaxys buoy since 25. Wind velocity and direction was available from 1997 to 22. Coastal buoy from Valencia harbour authorities (Tryaxis type) Latitude: Longitude: 39º.9' N º 12.3' W Water depth: 48 m As a general remark, related to any vessel or yacht, the seakeeping performance will depend on the following aspects: prevailing climate (wave + wind) on the area at the time of the races; motion characteristics of the design; motion response (behaviour) in given wave/wind conditions; control of active appendages (steering and sails tuning). The three above first points are illustrated by means of an example study performed for an America s cup yacht. As previously stated, the principle of a seakeeping performance study is highlighted hereafter but does not represent the results of a full investigation. Climate for the 32 nd America s Cup in Valencia On of the most important aspect in the evaluation of the performance of a concept at sea is to obtain accurate climate conditions in which it will operate. This applies to the full maritime industry, from oil platforms to ferries operating between two harbours. It is obviously true for the Louis Vuitton s and America s cup as the location where races will take place is fixed and the time of the year is fixed as well. For Ocean Figure 4 Occurrences of wind directions are shown in figure 5, for the month of May and only concerning afternoon conditions. The polar plot shows the fraction of time (occurrence) for each wind direction. It shows that wind conditions are relatively steady in direction. The same applied for wind speed. Histogram - May (Afternoon) ο N 33 ο 3 ο.5.4 3 ο.3 6 ο.2.1 W 27 ο 9 ο E 24 ο 21 ο 18 ο S 15 ο 12 ο Figure 5 ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 8

Concerning wave conditions, the following figure shows scatter diagram for May and June period, only afternoon data as well (from 12. to 18.). Figures are also shown in figures 29 through 31 at the end of the paper. H s [m ] 4 3 2 1 Scatter diagram - May (Afternoon) 3 6 9 12 T [s] p Figure 6 The above scatter diagram was obtained by taking wave conditions measured during the month of May, from 12. to 18., over a period of 2 years. It provides a very representative overview of all type of conditions that will be encountered during the Louis Vuitton s Cup. The density of the measurements (not directly shown herewith) also provides the percentage of occurrence of each combination of significant wave heights and wave periods. Concerning IACC yachts, the most unfavourable wave peak period lays around 4 to 5 seconds for pitching motions. Such wave periods, combined with significant wave heights around.5 to 1. meters are relatively frequent for that period of the year. Steeper and shorter waves also occurs, with periods around and below 3 seconds, which are typically growing sea states induced by afternoon thermal wind. From a seakeeping point of view, significant pitching will only occur in wave period larger than about 3.5 seconds. Numerical approach and correlation with sea trials encounters should be gathered. The wave conditions for the simulations were taken from the on-line buoy measurements and checked with onboard observations. The program Panship, in-house developed at MARIN, was used to reproduce sailing conditions that were measured during the sea trials. The results of one windward condition are presented hereafter. Panship contains a time domain panel method for seakeeping of ships with lifting surfaces. Use is made of the transient Green function to describe free surface effects. This implies that free surface conditions are linearised about the mean free surface. However, wave exciting forces are evaluated on the instantaneous wetted surface. More information is provided by Van Walree in Ref. [11]. The hull surface (below and above the mean waterline) and appendages were represented by about 32 quadrilateral panels with a constant source and/or doublet strength. The required mean heel angle was obtained by applying a constant external heel moment to the vessel. During the simulations all modes of motion were considered, except for the forward speed which was forced to be constant. Course keeping was provided by an autopilot actuating the rudder. No attempt was made to derive autopilot coefficients that are representative for a human helmsman. The duration of each simulation corresponds to about 3 sec, which means that about 18 waves are encountered. This is sufficient for an accurate assessment of the motions of the vessel. Figures 7 and 8 show the panelling arrangement and the unsteady vorticity on the wake sheets (history of encountered waves) during a short sequence of the run. These figures are also shown at the end of the paper in figures 32 and 33. Sea trials were performed in the month of May in very typical wave conditions for that period of the year. For correlation purpose, one windward condition was chosen that had duration of about 3 minutes. This is obviously not a typical racing condition, but for reliable statistical comparison, enough ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 9

Z Panneling arrangement Y X Unsteady vorticity on keel, winglets and wake sheets Figure 9 Figure 7 Figure 8 shows the results of measured and calculated pitch angles. Standard deviation of pitch and distribution of positive peak values illustrate the good agreement between sea trials and model tests. Due to ship motions (mainly heave and pitch) and with a lower contribution wave orbital velocities, false angle of attacks of the flow on the appendages are taking place. Figure 1 shows for example part of the time history of the lift and drag coefficient on part of the appendages..6.6 Upwind Panship Measurement conditions standard deviation Pitch [deg].984.889 Heave [m].117.13 1 Cd-wing PS.4.2 Cd Cl.4.2 Cl-wing PS -.2 Probablity of exceedance [%] 1 1.1 1 2 3 4 Pitch angle [deg] Panship simulation. Sea trials measurement Figure 8 More interesting for our purpose are the dynamic effects on the appendages, inducing dynamic fluctuations of lift and drag. Such unsteady conditions that are occurring in real life deviate from the quasisteady approach which is used in standard design approach. Unsteady effects could potentially bring further development in the performance of appendages in particular. Figure 9 shows the history of vorticity on keel and winglets for several wave encounters. -.2 8 9 1 11 -.4 12 Time [sec] Figure 1 Making use of these dynamic fluctuations of the angle of attack on part of the appendages could potentially bring interesting re-thinking of their design and location. Current development shows from a numerical approach even potential reduction of the overall resistance, reduction which increases with increasing wave height. Optimum appendage settings could be sea state dependant, when all contributions are investigated in details (calm water part and dynamic part in waves). The trade-off between optimum appendage configuration for steady calm water and unsteady conditions in wave should be investigated in details in the future. This would lead to a choice of appendage configuration depending on the forecast weather conditions prior to each act for example. ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 1

First results of calculations indicate that an overall decrease of mean resistance (calm water +added resistance in waves) could be in the order of 1 to 4%. Experimental approach Simulations could provide good results thanks to sea trials results, in particular for the input of the steady heel angle, leeway and ship speed. In order to explore more design modifications and check their impact on performance, model tests could provide a good approach. Model tests were performed in oblique seas in the aftermath of the Australia II victory, as shown by figure 11 and 34, but since then no use of seakeeping basin was made. resistance, trim and sinkage is in good agreement with the measurements for different designs. In the second section we gave a short explanation of P.I.V measurements during tank tests. Sailing yacht performance in waves is an under investigated part of the design spiral, and the last section shows some potential openings on the numerical and experimental field at MARIN to further improve a sailing yacht design. ACKNOWLEDGEMENTS The authors are indebted to J.A. Keuning, E. Lataire and K.J. Vermeulen of the Ship Hydromechanics Laboratory of the Delft University of Technology for providing a part of the experimental results in this paper Further more we would like to thank the ABN AMRO Volvo Ocean Race team, for the use of model test data. Special thanks go to H.C. Raven and J. Tukker for there contribution and assistance. REFERENCES Figure 11 The Seakeeping and Manoeuvring Basin at MARIN, in service since 2, represents an ideal platform to test free sailing models in waves from arbitrary direction with respect to the course of the vessel (figure 35). Dynamic towing points to simulate sail forces are currently under development in order to apply the sail propulsion (figure 36). This represents a further step to provide a new tool dedicated to sail yacht optimisation, especially for America s Cup yachts. FINAL OBSERVATIONS AND CONCLUSIONS In this paper we gave a presentation of the different possibility at MARIN to be used during the designing phase of a sailing yacht. First we presented the modification and new features added to our non-linear CFD code RAPID for sailing yachts. The results of the improved code were compared with model test measurements. From this comparison it followed that the wave profile along the hull, 1. Raven, H.C., A practical nonlinear method for calculating ship wavemaking and wave resistance, 19th Symp. Naval Hydrodynamics, Seould, Korea, 1992. 2. Raven H.C, 1996, A solution method for the nonlinear ship wave resistance problem, PhD thesis, Delft University of Technology. 3. Raven, H.C., Prins, H.J., "Wave pattern analysis applied to nonlinear ship wave calculations" Workshop Water Waves and Floating Bodies, 1998, Alphen a/d Rijn, Netherlands. 4. Raven, H.C., "Inviscid calculations of ship wave making capabilities, limitations and prospects" 22e Symposium on Naval Hydrodynamics, Washington DC, U.S.A., August, 1998 5. F. Valdenazzi, S. Harries, C.E. Janson, M. Leer-Andersen, J.J. Maisonneuve, J.Marzi, H.C.Raven: The FANTASTIC RoRo: CFD optimisation of the forebody and its experimental verification, NAV 23 Symposium, Palermo, Italy. 6. M.M.D. Levadou, H.J. Prins, H.C. Raven (1998). Application of advanced computational Fluid Dynamics in Yacht Design. ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 11

7. Hoekstra M., 1999, Numerical simulation of ship stern flows with a space-marching Navier-Stokes method, PhD thesis, Delft University of Technology. 8. Calcagno, G.; Felice, F. Di; Felli, M.; Pereira, F., Propeller Wake Analysis Behind a Ship by Stereo PIV, 24th Symposium on Naval Hydrodynamics, Fukuoka, 22. 9. Tukker, J., Blok, J.J., Kuiper, G. & Huijsmans, R.H.M. Wake Flow Measurements in Towing Tanks with PIV. Proceedings of the Ninth International Symposium on Flow Visualization, 2. 1. Raffel, M., C. Willert and J. Kompenhans; Particle Image Velocimetry, a practical guide; Springer, 1998. 11. Walree F. (22), Development, validation and application of a time domain seakeeping method for high speed craft with a ride control system, 24 th symposium on Naval Hydrodynamics, Fukuoda, Japan, 8-13 July 22. ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 12

1.8.6.4.2 Sysser 27 wave elevation along the hull for Fn.3-2 2 4 6 8 1 12 -.2 L [m] -.4 -.6 -.8-1 Wave elevation [m] measured wave model RAPID free surface RAPID hull pressure Figure 12: Comparison of the measured and calculated wave profile along the hull for Fn=.3 1.8.6.4.2 -.6 -.8-1 Sysser 27 wave elevation along the hull for Fn.45-2 2 4 6 8 1 12 -.2 L [m] -.4 Wave elevation [m] measured wave model RAPID free surface RAPID hull pressure Figure 13: Comparison of the measured and calculated wave profile along the hull for Fn=.45 Figure 14: Overview of the total wave pattern for Fn=.45 for the Sysser 26 hull ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 13

1 Wave elevation [m].8.6.4.2 Sysser 26 wave elevation along the hull for Fn.45-2 2 4 6 8 1 12 L [m] -.2 -.4 measured wave model RAPID free surface RAPID hull pressure Figure 15: Comparison of the measured and calculated wave profile along the hull for Fn=.45 1 Wave elevation [m].8.6.4.2 Sysser 26 wave elevation along the hull for Fn.6-2 2 4 6 8 1 12 L [m] -.2 -.4 measured wave model RAPID free surface RAPID hull pressure Figure 16: Comparison of the measured and calculated wave profile along the hull for Fn=.6 Figure 17: Overview of the total wave pattern for Fn=.45 for the Sysser 26 hull ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 14

Figure 18: Overview of the total wave pattern for Fn=.45 for a Volvo 7 hull 12. Total resistance Volvo 7 Measured & Caculated 1. Rrh+Vrh [Kn] 8. 6. 4. 2. RAPID tank data..25.3.35.4.45.5.55.6.65.7 Vs [Fn] Figure 19: Comparison of the measured and calculated total upright resistance for a Volvo 7 hull.6.5 Sinkage Volvo 7 Measured & Caculated.2. -.2 Trim Volvo 7 Measured & Caculated Vs [Fn].4 -.4.3 -.6 -.8.2-1. Rapid Rapid.1 tank data -1.2 tank data Vs [Fn]. -1.4.25.3.35.4.45.5.55.6.25.3.35.4.45.5.55.6 Figure 2: Comparison of the measured and calculated trim and sinkage for a Volvo 7 hull ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 15

Figure 21: A sailing yacht under heel and drift angle calculated by the CFD code RAPID 1.. Trim Sysser 26 and 27 Measured & Caculated 1 m Lwl Vs [Fn] RAPID Sysser 26 Trim [DEG] -1. -2. -3. Tank data Sysser 26 RAPID Sysser 27-4. Tank data Sysser 27-5..25.3.35.4.45.5.55.6 Figure 22: Comparison of the measured and calculated trim for Sysser 26 and 27.16 Sinkage Sysser 26 and 27 Measured & Caculated 1 m Lwl Sink [m].14.12.1.8.6.4.2 RAPID Sysser 26 Tank data Sysser 26 RAPID Sysser 27 Tank data Sysser 27..25.3.35.4.45.5.55 Vs [Fn].6 Figure 23: Comparison of the measured and calculated sinkage for Sysser 26 and 27 ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 16

Figure 24: panel distribution on the hull and the free surface, as used in the CFD code RAPID Figure 25: Flow around a skeg for a hull under 1 degrees of leeway. The flow separation is presented as the black vortices at the windward side of the skeg ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 17

mm 11 1 9 8 7 6 5 4 3 2 1-1 -2-3 -4-5 -6-7 -8 Streamlines: Streamlines 5 levels -2-19 -18-17 -16-15 -14-13 -12-11 -1-9 -8-7 -6-5 -4-3 -2-1 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 mm 17 mm 11 1 9 8 7 6 5 4 3 2 1-1 -2-3 -4-5 -6-7 -8 Streamlines: Streamlines 5 levels -2-19 -18-17 -16-15 -14-13 -12-11 -1-9 -8-7 -6-5 -4-3 -2-1 1 2 3 4 5 6 7 8 9 1 11 12 13 14 15 16 mm 17 2 º Simposio Internacional de diseño y producción de yates de motor y vela. Figure 26: An overview of the P.I.V. measurement method Figure 27: Visualisation of averaged flow fields measured at 8 planes with 3C-PIV. Figure 28: Ship in shallow water at 1 and 15 degrees of drift angle ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 18

Histogram - May (Afternoon) ο N 33 ο 3 ο.5.4 3 ο.3 6 ο.2.1 W 27 ο 9 ο E Histogram - Jun (Afternoon) ο N 33 ο 3 ο.5.4 3 ο.3 6 ο.2.1 W 27 ο 9 ο E 24 ο 12 ο 24 ο 12 ο 21 ο 18 ο S 15 ο 21 ο 18 ο S 15 ο Figure 29 4 Scatter diagram - May (Afternoon) 3 Wave periods with maximum pitch response for IACC H s [m] 2 1 3 6 9 12 T [s] p Figure 3 4 Scatter diagram - Jun (Afternoon) 3 H s [m] 2 1 3 6 9 12 T [s] p Figure 31 ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 19

Z Panneling arrangement Y X Figure 32 Panel arrangement for an AC type of yacht Z Unsteady vorticity on keel, winglets and wake sheets Y X Figure 33: Vorticity calculated at each encountered wave (mainly due to pitch and heave) ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 2

Figures 34: Model testing in waves in the seakeeping and manoeuvring basin at MARIN ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 21

Figure 35: Overview of the seakeeping and manoeuvring basin at MARIN Figure 36: Measurement method for testing sailing yachts in oblique waves ADVANCED AND FUTURE HYDRODYNAMIC OPTIMISATION TOOLS IN SAIL YACHT DESIGN. 22