America’s Cup Sail Design and Optimization

Sailing yacht flows offer a classic example of problems that demand RANS-based design methods.  Boundary layers separate around the forestay and mast, but reattach further back on the sail.  Viscous and vortex wakes that shed from upstream sails impinge on the downstream ones.  Onset flow is both viscous and rotational since yachts are immersed in an Earth’s boundary layer, and the topsides and deck pose major disruptions near the sail foot.  The embedded movie demonstrates some of these complications by showing typical Americas Cup class yacht sail surface pressures, velocity and vorticity cutting planes, surface streamlines, and off-body streamlines.  Potential flow based design methods cannot capture these demanding flow details, and Reynolds numbers are typically too high for wind tunnel testing. 

As if physics don’t present enough of a challenge, the number and range of design variables is huge.  Sail chord, camber, entry angle, and position of maximum camber can all vary along the span.  They can also change independently between foresail and mainsail.  Angle of attack can change both locally (as trimmers affect the relative angles between sails, add or subtract twist, or flatten camber), and globally (as the helmsman steers to adjust apparent wind angle).  Finally, the resulting designs must be able to work across a wide range of environmental conditions such as boat heel, wind speed, and wind angle. 

Additional issues arise in the particular case of America’s Cup Class design because the difference between competing boats is so small.  Less than 1% boat speed often separates winners from losers, so design changes smaller than this must be accurately resolved.  This poses a huge demand on any design tool, let alone those that omit the (much larger) physical effects mentioned above.  Even with state-of-the art RANS methods, care must be taken in making design decisions.  Isolated forces cannot be used to guide the process since the relationships between boat speed, drag force, drive force, side force, roll moment, and righting moment is so complex. 

AFT circumvents each of these problems using design tools developed over many years.  RANS is used to simulate the complex aerodynamic and hydrodynamic flows, and to provide the required forces.  Efficiency and robustness are maintained using automated grid generation and run management techniques; and in-house supercomputers are employed to complete large numbers of simulations in quick succession.  Performance prediction and design decision-making are handled using AFT’s in-house Velocity Prediction Program (VPP) and optimization capability.  This is the only VPP currently known to work solely with high-resolution RANS data. 

More detail of AFT’s RANS-based design capabilities are described in “Reynolds-Averaged Navier-Stokes in an Integrated Design Environment,” and VPP optimization capabilities in “Performance Prediction without Empiricism:  A RANS-Based VPP and Design Optimization Capability.”