Software rules the seas in the America's Cup race

When two seconds over a three-hour period is all that makes or breaks the winner of the most heralded boat race in the world, design details are vital. America's Cup engineers can spend years perfecting the most minor items -- the placement of an airfoil or the slant of the bulb.

Traditionally, engineers analyzed these details by testing one-quarter-size scale models in 300-yd-long, 40-ft-wide towing tanks and wind tunnels. Weeks, even months, passed before they had results.

Today, some America's Cup engineers are using high tech computer workstations from Silicon Graphics (SGI) and computational fluid dynamics (CFD) software programs to crank out findings in days, allowing for multiple iterations of various design considerations.

"We are always up against the clock," says Dave Egan, design coordinator for Team Prada, the Italian syndicate. "The race day doesn't move. To be able to examine 10 or 20 designs in a week instead of year because of improved computer hardware and software tools is tremendous. By the next America's Cup, we may be doing these calculations in an hour or less."

Analyzing the forces impacting a yacht racing through the water is not easy.

"The yacht functions as an interface vehicle between air and water," says Tom Schnackenberg, design coordinator for the New Zealand syndicate. The hull contacts both the water and air. The deck, crew, sails, and mast are above the water. The keel, bulb, rudder, and fins below.

The dual relationship means engineers must calculate both air and water flows for the same components. But these flows are neither static, stable, nor predictable. The water line on a yacht constantly changes with weather, waves, and direction of sail. Float and drag forces shift from moment to moment.

CFD software allows Team New Zealand and Team Prada engineers to simulate the flow of water or air past a hull, underwater appendages, or sails, and calculate the pressure and drag forces.

Before, engineers had to assume the hull was on a flat surface, says Egan. But software advances make it possible for them to use a "moving" mesh or grid of elements that changes as the pressures on the boat change from the activity of the sea and winds.

Computers also enhance collaboration. When testing in a towing tank, people often forget their ideas between the actual test and the wait for results. Using computers, all the members of the team can brainstorm on the spot.

Then there is communicating the design information. The insight of the engineers must be shared with the sailors, as they are a major part of the yacht design as well. For example, the person trimming the sail on the day of the race has the final word on its shape, says Schnackenberg. "When he or she is pulling on the line, that action is the last part of the design process," he says. So sailors spend time in classrooms with engineers and CFD simulations learning how the air and water pressures affect various areas of the boat.

The average person isn't going to understand a lot of mathematical equations, says Egan, but he or she will understand the graphical representation of wind and water flow.

Team New Zealand uses CFD software from Fluent (Lebanon, NH), combined with a mix of Silicon Graphics^(R) workstations and servers which include Origin^TM 200, SGI Octanes^(R), and two SGI O2^(R).

Team Prada uses a homespun CFD package comprised of CAD software Pro/ENGINEER from Parametric Technology Corp. (Waltham, MA), a pre- processing program Hexa from ICEM-CFD (Arden Hills, MN), and optimization software EnSight from CEI International (Morrisville, NC) on SGI Origin 200 and 2000 servers.

"We wanted to use off-the-shelf components as much as possible, but add a bit of our code as well," says Egan. "Commercial software didn't achieve the accuracy that we needed." To work quickly and accurately, the programs must be integrated. "When ICEM joined with Pro/ENGINEER we had our answer. Once the programs started talking together, the system was much faster and easier."

Team New Zealand has implemented a velocity-prediction program that makes connections between the weight of the boat, the righting moment, lift over the sail, lift over the keel, and motion of both air and water over the yacht. This solves for equilibrium of the vessel.

After practice runs and races, the design teams feed navigational data from on-board equipment into workstations to analyze and optimize the performance. Both teams will have their computer systems on hand at the race, calculating each day's race and updating what they can on the spot.

Old and new work together. In spite of their reliance on CFD, both teams are quick to point out that software will not replace the more traditional testing methods. "CFD is not a replacement for the towing tank," Egan says, but it does compliment the experiment. Before, CFD supported the results of the tank. Today, the tank supports results of CFD, he says.

The Italian team has so much faith in its computer analyses that when a particular model ran well in the tank but not in the computer, the team opted to discard that design completely. "Only history will tell whether this was a wise decision or not," Egan says.

"A year ago, I would have been happy going to the America's Cup with two workstations," he continues. "Now we have eight and I want more. It has been a huge learning curve, but the results have been phenomenal."

"Four years ago CFD was a black box that no one saw or used but the specialist and it was a battle to turn results around quickly," says SGI's Roger Rintala. For these reasons, engineers used CFD sparingly. Today, simulation is pervasive, he says. More than half of the syndicates are using it to help design faster boats in less time.

"If one designs a keel without CFD, he has a 95% chance of arriving at the correct answer," says Egan. "But it is that last 5% that is hard to obtain and could make all the difference between winning and losing."

"Computers haven't helped us design any faster or even more efficiently, but they have helped us design more knowledgeably," Schnackenberg says. "I personally have more confidence in the design when I sign off on it because we tried more alternatives."

As servers and workstations get faster, programs become better and more sophisticated. "There are huge opportunities we haven't explored yet," he adds.

The Louis Vuitton Cup began in Auckland, New Zealand, October 18, 1999. The format includes three round robin series, a semi-final series and a final. The winner of the Louis Vuitton Cup will meet Team New Zealand in the 30th America's Cup Match beginning February 19, 2000.

Engineering considerations

Among the many design considerations that America's Cup engineers must balance against each other:

Length, thickness, and slump of the bulb. In the America's Cup Class, yachts carry 20 tons of lead or ballast in the bulb. The bulb hangs from the keel, and together they form the appendage package, projecting below the boat. This keeps the yacht from capsizing as well as supplying most of the hydrodynamic lateral force that enables the boat to sail upwind. Analyses show that an oval shape resembling the Goodyear Blimp produces the least amount of drag. But, the more squash to the bulb, the lower its center of gravity. While this increases drag, it also increases stability, especially important to counteract the heeling moment from the sails.

Position of the wings. Each bulb may have 1- to 1.5- meter-long "wings" or winglets attached. These extend out to the sides of the bulb. They may be twisted, cambered, and angled, just as wings on an airplane and their positioning plays a part in their effect on the water flow. Although a minor part of the overall design, every detail adds or subtracts from that two-second advantage needed to win the race.

The shape, length, and thickness of the fin and tab combination. Together these form the airfoil, whose shape counteracts the side force from the sails. Engineers ask themselves, "how long should the fin chord be and how thick?" Unequal shaping of the foil could cause vibration, which creates drag.

The shape of the sail and the size of the mast. Engineers estimate that one-third of the total drag off the hull is from wind friction caused by invisible vortices that stream off the sails, much like the air streams off the wing-tips of an airplane. Because air flows from high to low pressure, air spills over the edges of the sail and creates a spiral of slower moving air that slows the boat. "We spend a lot of time making these vortices smaller by changing the shape of the sail or by trimming," says Tom Schnackenberg, design engineer for New Zealand.

How Team New Zealand used CFD

In yacht design and marine engineering in general, engineers are primarily concerned with unbounded external flows, and much of the work pioneered by the aerospace industry is applicable.

For many years now the mainstay of maritime CFD has been the zero-viscosity or inviscid boundary element techniques, or so called "panel" codes. These are computationally economical, reasonably robust and give fairly sensible answers. The basic assumption that the viscosity-affected region is a relatively small volume of the total flow field is sound if one is looking for "macroscopic" physics, such as lift induced drag.

Team New Zealand used Fluent as an inviscid code. For the lifting flows, engineers added an artificial interior boundary (or wake sheet) to contain the shed vorticity from the lifting surfaces.

The first image shown is one of a number of candidate designs run in Fluent as an inviscid case. The effective span calculated from the results is in close agreement with the tank test data, and the calculated load on the winglets also agrees well with full-scale measurements on the yachts. The grid-generation time using fully unstructured grids makes this type of analysis extremely cost effective when compared to wind tunnel and towing-tank studies.

The second figure shows detail views of the fin. Extracting the numerical data for sections along each foil allows engineers to feed the load distribution information into the 2D foil section design program and optimize each section (shape and chord length). The results from the 2D program then feed into the next 3D model.

Other information derived from models like these include:

Keel wake trajectory visualized using pathlines emitted from the trailing edge of the fin. This data is useful for optimizing the rudder position.

Optimal paddlewheel location the paddlewheel is the boat speed sensor for the instruments. Boat speed is used in a large number of calculations for wind speed, tide, time to next mark etc., so positioning it away from local flow accelerations improves this information in the racing environment.

Another area that Team New Zealand used Fluent to study is the 2D foil sections at or beyond stall. Typically, keel areas are chosen to minimize drag in the upwind sailing condition. During the tight maneuvering of prestart circles and mark roundings, the lift coefficient required to stop the yacht from drifting sideways can force the appendages to stall. Candidate sections have been modeled in the stalled condition to find those sections with softer stall characteristics and higher lift in the stalled condition.