Seahorse International Sailing
October 1996
The use of Computational Fluid Dynamics (CFD) in the design of modern racing yachts has received a great deal of publicity recently. The actual impact of this technology on the design process is less well defined and discussed. While many articles centre around breathtaking computer images of yachts sailing in 'digital oceans' the technical merit of CFD is often left undiscussed. It is important to view all new technical development tools with some scepticism until their worth has been shown. We believe that CFD may someday be an extremely reliable, cost effective and versatile tool for use in yacht design but the present state of the art precludes it's widespread use on every day design projects.
Firstly, let's explain the term Computational Fluid Dynamics. Put simply, CFD is the study of aero and/or hydrodynamics using analytic methods which necessitate the use of computers. But fluid flows are inherently difficult and complex things to understand.
In laboratory conditions the flow of a viscous liquid over an inclined plate can be easily demonstrated and recorded. But the accurate prediction of this flow over the plate and the ensuing disturbance downstream using any analysis method, including CFD, is almost impossible. Very good estimations can be obtained for the flow over the plate, but he flow downstream of the plate provides far too much complexity to be modelled. So it should be apparent that the analysis of an entire yacht's hull and underwater appendages operating in an unsteady water-air interface will be challenging to say the least.
Over many years scientists and engineers have developed mathematical laws which describe the way fluids behave. These equations are complex and can only be solved using numerical methods. To find how any flow behaves, no matter how complex, it is a matter of solving these equation s for every point in that flow. In reality, there are an infinite number of points in any flow, so we cannot solve the equations for every point. This is the fundamental problem of modern CFD - we must find some method of approximating the mathematical laws governing fluid flows allowing relatively easy solution yet still yielding realistic results.
Programs such as SPLASH and VS-AERO use what is termed 'panel' analysis to approximate fluid flows. All surfaces including solids, water-air interfaces and wakes are defined as panels. The analysis program then determines the flow characteristics such as pressure and velocity at each panel. The accuracy of these packages is often the source of considerable argument. Another CFD analysis method often used is called 'finite element analysis'. In the method the flow field is meshed and effectively broken into a very large number of small elements. These elements may be two or three dimensional and the fluid flow characteristic for each element are calculated by the analysis program.
It seems likely that if CFD were to find wide spread acceptance as a reliable design tool it would have been established firstly in the Aerospace industries. In a business where the research and development budgets for new products run into the billions, not millions, much effort has been expended on CFD. It is wind tunnel testing, not CFD, which provides the bulk of the analysis information concerning the aerodynamics of a new military or commercial aircraft. For example, Boeing, in their development of the new 777, undertook thousands of hours of tunnel testing with limited emphasis on computational analysis.
The design of the racing yacht can prove to be an incredibly complex task if efforts are made to quantify every element of the aero/hydro dynamics. The two fundamental reasons for this are the air-water interface and the unsteady state of the sea and, to a lesser degree, air. In the America's Cup, the pinnacle of racing yacht design and development, CFD has been used extensively to understand performance variations from minor and major hull and appendage modifications. With every boat being designed for the same operational conditions it could be thought that they would all be very similar in underwater design. The bulb unveiling showed this to be far from the truth with obvious, large variations in bulb shape. These variations shows that computational methods have a long way to go before converging at the highest level.
It is a fundamental of yacht design that yachts must perform in a variety of conditions. Even America's Cup boats must perform in varying sea states and wind strengths. Weather and sea studies will inevitably result in generalisations being made about the conditions in which the boat will sail. Yachts that race in the open ocean have an even broader operational regime. this requires most racing yachts to be versatile craft, able to be competitive when sailed in conditions they are not specifically designed for. While ocean racing yachts are designed with a bias (Whitbread 60's are primarily designed for reaching), they must be able to sail competitively at every point of sail.
The extensive use of CFD in a yacht design program is limited to a large degree by finance. 80 hours of computational CRAY time is seen as an acceptable amount of analysis for the development of a single optimum hull shape. This is not an excessively long time, but at around $3000/hour this becomes prohibitively expensive for all but high-end projects. On top of this cost, validation of the CFD models must be undertaken. This will involve the use of tank testing hull shapes and wind tunnel testing of appendage shapes. In a budget-constrained ocean racing yacht project it is unlikely that $500,000 will be available for this type of analysis.
With this in mind we see the selective application of CFD as useful in design process. In the context of offshore racing yachts the use of highly efficient keel and rudder planforms and sections is fundamental to optimum performance. In smaller projects the extent of CFD use may only run to section design. On larger projects we may use panel methods to analyse keel-hull, keel-bulb joins an bulb shapes as well as planform and section optimisation. However it is very important that throughout the entire design process the fundamentals of aerodynamics are not ignored. The huge body of information produced by fluid dynamists over the last 100 years gives a large amount of technical information readily applicable to yacht design.
Because of the technologically based nature of CFD it must be seen as a developing tool. Presently, we are not satisfied with the ability of commercial software to perform 'out of the box'. With the present requirement of complex validation using tunnel and tank testing, it seems that the widespread use of CFD is limited to the budgets of the America's Cup contestants. It is also clear that future developments in both CFD software and hardware will bring computational analysis into the area velocity prediction programs hold today. It is conceivable that in 5 years time both CFD and VPP's will share space on every designer's desktop PC.
Paul Bury, B Eng (Aero)
Design Engineer
Jutson Yacht Design