Sail first on the computer

ByJoseph Ogando, Materials Editor

On location in Hauraki Gulf, Auckland, New Zealand -The America's Cup may be decided on the water, but it begins in the glow of workstation monitors long before the final races. By the time the sailors get their feet wet, teams of design engineers have already spent years navigating the drop-down menus of CAD, computational-fluid-dynamics (CFD), and structural-analysis programs.

To see just how pervasive computer simulation has become in sailing's most celebrated contest, take a look at this year's competitors. The winner, Team New Zealand, designed its hull and keel in Unigraphics CAD software from Unigraphics Solutions, and performed CFD analyses with software from Fluent. America True, the San Francisco Yacht Club's entry, put SDRC I-DEAS to use. And Prada Challenge, which posted a 31-8 record in qualifying rounds and won the Louis Vuitton Cup to qualify for the cup finals, employed PTC's Pro/ENGINEER with the new Behavioral Modeling Extension (BMX), and also used custom and commercial CFD codes.

Despite Prada's ultimate loss in the cup finals, its engineers and designers took the use of PTC's software to new heights, says Wayne George, senior vice president at PTC. Prada's chief design engineer David Egan used Behavioral Modeling to incorporate design objectives, analysis work, functional behavior, and other expressions of design intent into the geometry model as features. In fact, Egan and his fellow designers had enough confidence in their computer-aided engineering methods that they eliminated the traditional physical testing of the "appendages"-the boat's keel, bulb, and wings. "They were all designed numerically," Egan says.

Through BMX, Prada's CFD and structural analyses existed as model features in Pro/ENGINEER. So did the interface to the mesh generator, ICEMCFD from ICEM. So did the two hulls Prada used in competition, allowing Egan to place either hull in whatever boat configuration he happened to be analyzing. And so did the bulk of Prada's "design intelligence"-a large collection of functional requirements, specifications, and design constraints imposed by America's Cup rules.

BMX immediately saved Prada time by treating the meshing of the sea surface that surrounds the boat as a model feature, which ensured that sea-surface mesh and the boat geometry always match one another: Change the model geometry and the mesh changes too, eliminating any manual mesh trimming. "Geometry changes became painless," Egan recalls. And with thousands of CFD runs to perform-each taking from three to 40 hours-keeping the sea-surface mesh and model geometry up to date is no trivial matter.

More than any other piece of Prada's engineering-software system, the BMX also made parametric design optimization a reality, according to Egan, who notes that it supports two related optimizations tasks. For one, BMX provides built-in "goal seeking" functionality that lets it look for only those design solutions that satisfy multiple design objectives. Egan points to bulb design, for example, where BMX enabled him to balance drag reduction against volume, mass, and surface-area constraints.

Also, BMX can co-exist with external automatic optimization methods that generate a series of small parametric design changes for CFD analysis. At first, Egan accomplished this task through Unix shell scripts that systematically vary key design attributes. Later on, he used iSight software from Engineous to optimize these parametric changes in what could be described as a virtual design of experiments.

Before all the automatic optimization, finding the best design required designers to make many manual design changes, send them through the CFD code, and evaluate the consequences of each change. "It could take days or weeks," Egan says. And with the automatic optimizing process? On a typical day, Prada's designers ran between 20 and 50 configurations through the main computer, an Origin 2000 from Silicon Graphics.

The flotation problem. Prada's key piece of custom software addressed what Egan calls "the flotation problem"-predicting how water flows around the boat. As the boat bobs up and down in the water, changes orientation while sailing, and interacts with waves, "different amounts of the boat touch the water," explains Egan. This constant change in the amount of wetted surface area makes it difficult to predict drag. "We haven't solved the problem yet," Egan says. To get close, Egan had to write his own CFD solver. Called Flow-Logic, it predicts drag forces by computing the hydrodynamic flows around the boat and a pressure distribution. One thing Egan's code doesn't do is handle skin friction. For this component of drag, the team used several commercial Navier-Stokes codes.

Though some commercial CFD programs contain flotation models, Egan found them lacking because they "assume no sea-surface change" and instead base their flotation calculations on a flat, "mean surface." Flow-Logic continually generates new sea-surface shapes and yacht positions. Because the sea and boat act on each other, Egan stresses that "the sea surface shape and yacht position have to be computed simultaneously to get accurate results."

Above deck too. Prada's use of Behavioral Modeling didn't stop with CFD analysis of the boat's appendages-or "the stuff under the water," as Egan puts it. BMX also helped optimize parts of the boat's structural design in conjunction with PTC's Pro/ MECHANICA, whose composites-analysis functionality includes the influence of temperature settings, resin percentage, and orientation. Like the CFD optimization process, the structural analyses existed as model features and drove the parametric geometry changes, according to Andrea Avaldi, Prada's structural designer and an aerospace engineer. BMX built-in goal-seeking functionality to fine-tune the attachment of the keel to the bulb and to the hull, among other parts of the boat.

What's more, in a process similar to that for the CFD optimization, Avaldi also used Behavioral Modeling with iSight to identify the optimum design and set-up for the 32-meter-long mast and the cables-or "stays"-that support it. For sails to work efficiently, the masts must bend to a specific shape under the tension of the stays, which can exceed 10 tons. So Avaldi's analysis not only had to consider the flexural and structural strength of the mast itself but also the stay position and tension. It wasn't an easy analysis. "Masts can behave very strangely," Avaldi says, explaining that they bend unpredictably because of the complex interplay between their bending moments, their axial stiffness, and axial loads that vary along their length. Fortunately for Prada, which did suffer a snapped mast early in the qualifying rounds, the software ultimately did a "good job predicting the mast's non-linear behavior," Avaldi says.

The limits of CFD. With all the computer-aided engineering (CAE) work done on Prada, the boat's hull stands out for its lack of reliance on CAE. In part, the hull underwent less CAE because it had already been designed by the time BMX became available. Yet, hull design also runs up against some of the limits of current CFD technology, restricting its use even further. As Egan points out, skin friction-or viscous drag-takes on a greater importance with hull design than with the appendages. And these are precisely the type of CFD predictions that have the least accuracy and take the longest, according to Egan. With CFD runs that go on for days, it took Egan one-and-a-half years to do five computational studies of the hull's viscous drag.

Still, Egan expects that increasing computer power and falling computation times will render hull design more CFD-friendly, making yacht models as complete as he wants them to be. "The more you model," he says, "the less you have to guess."

Distribute the loads

ByJohn Lewis, Northeast Technical Editor

What technical element (sails, hull, computers, rig, keel, etc.) has the greatest impact on winning the America's Cup? Everyone has a different opinion. But the story is that winglets on the ballast bulb at the bottom of the keel strut propelled Australia II to victory in 1983, by providing extra lift as the boat heeled. Every year since, America's Cup engineers have been tweaking keel design. This year, engineers on the Young America team pioneered a new approach for flanged keel-fin design: open-die forging.

Their design objectives for the keel fin included minimizing hydrodynamic drag, keeping the fin weight as low as possible in order to design more weight into the bulb, and keeping deflection of the fin (distortion of the bulb angle while heeling, reducing the righting moment) to an absolute minimum.

"For maximum boat stability, it's essential to minimize bulb deflection," says the team's Design/Technology Project Manager Duncan MacLane. "If there is any deflection in the fin because it is not stiff enough, or if there is any deflection at the point where the fin connects to the boat, stability suffers and results in a loss of speed." (During the second round of racing, Young America's USA 53 nearly sank when the deck failed and the boat folded in on itself while sailing upwind. A preliminary investigation determined the failure was not the result of engineering design.)

To meet its objectives, the team developed a fin design with:

thin and lightweight airfoil profile that minimizes drag

stiff construction that minimizes bulb deflection

exceptional strength and no internal imperfections that could cause cracks

integral flanges to allow bolt-on mounting to the bottom of the hull.

The integral flanges and bolt-on feature represent a distinct difference from most keel designs-which typically protrude up through the hull in a tongue-in-slot configuration-and was included in the specifications in order to deliver two important advantages.

First, the rigid bolt-on connection reduces bulb deflection compared with a typical, less sturdy, tongue-in-slot connection. Second, bolting the fin onto the bottom of the hull spreads the load out into the hull of the boat instead of at the connection point, allowing for a lighter yet stiffer hull foundation. "This concentrates more weight in the bulb for increased stability," explains MacLane. Because the bulb constitutes as much as 80% of the overall weight of the boat, all of these advantages add to the boat's speed.

The team chose open-die forging of the fins and flanges instead of casting primarily because, they reasoned, forging's directional grain flow ensures excellent structural integrity even at the flange points. This directional grain flow, resulting in maximum metallurgical soundness, part strength, and fatigue resistance, is not found in casting. And without porosity or other defects inherent to castings, open-die forging offers unmatched structural integrity throughout parts. These benefits, designers were convinced, would provide exceptional strength even in a fin that was extremely thin and light.

Engineers worked with Scot Forge (Spring Grove, IL). Metallurgists decided upon a modified 4340 steel for the keel fin, which would be formulated especially for this project. This specialty metal offered the exceptional strength, stiffness, and toughness specified by the design team-surpassing other IACC-approved materials, such as carbon-based composites, in each of those areas, according to Chief Metallurgist Tony Biell. "Carbon-based composites have been tried because their lower density makes them lighter, but they are more prone to bending and deflection," he says. "The modified 4340 steel we provided also offers excellent directional strength, very high yield strength, and excellent toughness under the entire range of potential water temperatures. And it permits the use of a high tempering temperature, minimizing residual stresses and distortion during machining. As a bonus, the material is more cost-effective than stainless steel, while offering the same or better performance properties."

Forging the fin. The process for forging the fin included open die forging, heat treating, machining, and testing-all at Scot Forge.

"Forging allowed us to integrate the best available steel with the best manufacturing process," says MacLane. "By giving us the ability to make the fin as light as we needed it to be, forging enabled us to push the design to its limit, instead of making the normally expected design trade-offs. As a result, we were able to realize performance gains that we had never seen with other keel manufacturing techniques."

Analysis of the design prior to production using computer simulations and testing actual P/3-size scale models in the towing tank and wind tunnel, indicated that the design not only performs effectively, but that it could potentially shave seconds off boat speed performance times. Computer simulations with South Bay Simulations'(Babylon, NY) Small Perturbation Linear Analysis of Surface Hydrodynamics (SPLASH) condensed the overall testing process from months to days.

Upon completion, Scot Forge tested the keel fin in a variety of ways including Charpy impact tests, mechanical testing, and ultrasonic and magnetic particle nondestructive testing to verify that the forging was defect free.

In actual waters, the keel fin helped win speed competitions in the Louis Vuitton Cup in New Zealand, and is now being considered by Farr Yacht Design as a major design component in numerous yachts that will compete in the upcoming Volvo Round the World Race.