Speedo Computes Next-Generation Athletic Swimwear Designs on SGI® Altix®

To create the designs and materials to give the Summer Games and competitive athletes a winning edge, Speedo, currently the world's largest selling swimwear brand, is designing the next-generation of their FASTSKIN FSII athletic wear and their next-generation goggles using the power of Silicon Graphics technology. Performance-intensive computational fluid dynamics (CFD) analysis based on data of real Summer Games swimmers-monitored in Speedo's Aqualab research and development facility-is run on an SGI Altix high-performance compute system to better understand how to optimize the flow of water around a swimmer. According to Speedo, the company is the first-and only-swimwear company to use CFD analysis to design swimwear and athletic wear.

Speedo swimwear, originally developed in Australia, has been a staple of the Summer Games since 1932, and athletes sponsored by the Speedo brand include Greg Louganis, Janet Evans, Michael Phelps, Amanda Beard, Dawn Fraser and Kosuke Kitajima. At the 2004 Summer Games in Athens, 47 competitors wearing Speedo athletic wear earned medals. FASTSKIN FSII ICE, an offshoot of FSII for swimmers, debuted at the 2006 Winter Games in Turin, Italy. More than 30 U.S. and international athletes wore the FSII ICE speed suit in the bobsled and skeleton competition. The American bobsled team of Shauna Rohbock and Valerie Fleming became the first-ever athletes to win a Winter Games medal in Speedo FSII ICE with their Silver Medal finish in the women's bobsled competition.

Rerunning CFD models of previous FASTSKIN FSII designs in order to create newer, even more hydrodynamic models will further reduce high skin-friction drag, Speedo engineering consultant Barry Bixler is now using the SGI Altix system to develop the next generation of FSII and the new goggles, and doing much more sophisticated analysis than ever before.

"I do the number crunching with the SGI Altix and then I get the result files, which I can just post-process on my PC home computers, and visualize the results in Fluent software," says Bixler, who had the idea, several years ago, of using his "day job" CFD experience as an aerospace engineer for swimwear design when his daughter took a serious interest in competitive swimming. "We're looking at the speed of the water on the surface, the velocity. Right at the surface it's zero, of course, but right below the surface, it varies. We look at the speed and the direction of the flow, and where it leaves the body, and where it reattaches itself. And of course where the skin friction drag is: where it's high, where it's low."

Most competitive swimmers, male and female, wear some type of full body suit that either covers wrists to ankles, or sleeveless, with torso and legs covered; some prefer torso and a leg length to just above the knees. Speedo makes all the varieties and combinations, yet Bixler (and many others) believe athletes can swim faster in the longer suits because water on human skin creates more friction, which slows a swimmer down.

Amazing amounts of research go into the carefully designed fabrics of competitive swimwear. FSII material, for example, more closely resembles a shark's skin than human skin, down to the identical roughness. It so resembles the shape and feel of a shark's denticles (small, pointed surface projections) that FSII mimics the denticle variations along a shark's body that manage the flow of water. Managing water flow means reducing passive drag, which affects a swimmer in the streamline position-generally after the initial dive into the water and following a turn. FSII is said to reduce passive drag in the water by as much as 4% compared with Speedo suits used in the Australian Summer Games several years ago.

"The days of wearing a skimpy little thong are gone," continues Bixler. "Athletes can simply swim faster with the longer suits, and that's where the CFD analysis comes in. I've got several CFD models of actual Summer Games swimmers and at Aqualab we've analyzed those swimmers in a streamlined position. We've also tested those swimmers in a flume, in a streamlined position."

Bixler explains that a flume is like a wind tunnel, except with water. Swimmers in the Speedo Aqualab slip into the water and hold on to an underwater handle so passive drag can be measured on their bodies while wearing different suits. The underwater handle is hooked up to a small cylinder, which goes ahead of the swimmer in the flume, up to a big strut, which has equipment to measure force. The force is what the drag on the swimmer is, minus the drag on the strut. This gives Bixler an experimental drag force and then, from his CFD model on the SGI Altix, he computes an analytical drag force, which is compared to the computer model, which is then tweaked in Fluent to make sure he's getting the right results and the proper modeling of fluid flow around the swimmers.

"CFD is kind of a black art, so to speak," says Bixler, with a laugh. "You can use whatever numbers; you can tweak it to get whatever answer you want. It's always good if CFD is used in conjunction with the actual test results. Once you get your model tuned, then you can start making variable changes to that model-put different suits on that model, put the orientation of the fabric in different ways."

All fabric has an orientation, or grain, and CFD modeling shows how the water flows with-or against-the grain. Water does not go straight around the body; in fact, when you get to the legs, it actually curves and swirls around the legs as they kick while swimming. CFD analysis on the SGI Altix led Speedo designers to orient the grain of the fabric-as well as the seams of the swimsuit itself-to follow the water flow at optimal hydrodynamic placement. Obviously, the grain of the fabric and swimsuit seams need to be parallel to the water flow. If the flow is curving around the leg, the fabric and the patterns are cut and sewed together to follow that flow curve. Different suits are designed for male and female athletes because CFD analysis has determined there is a more separated flow, where the water actually leaves the surface of the swimmer, for females than males, because of their extra curves.

"We analyzed different suits on the CFD model on the Altix, and we could see which way the water flow goes along the swimmer, and how fast the flow was at certain points," continues Bixler. "In some places it goes along the body at 2mm per second, but other places where there are bumps-like the butt or the breasts, the shoulders and the head-you have to speed up the flow to move around those bumps. What the CFD on the Altix showed was that I could get contour plots of the skin friction drag along the body and also the pressure drag, which is the force perpendicular to the surface where skin friction is parallel to the swimmer. We can see where the water leaves the body, where it reattaches, and we can see the areas of high skin friction drag. In our testing, in the flume, we noticed that some fabrics use lower drag at lower speed and other fabrics give lower drag at higher speeds, so we use two different fabrics in designing our swimsuits."

Bixler gets much faster results over PC clusters by using the SGI Altix, which runs eight Intel® Itanium® 2 processors in an SGI shared-memory architecture Linux® environment.

"I chose the SGI Altix for one reason: speed," concludes Bixler. "CFD is one of the most time-consuming, intensive number-crunching activities that you can use a computer for, and I can use all of the eight processors running in parallel on the Altix. When I was using a PC cluster, some jobs would take almost a week to run. I can now run similar jobs on the Altix overnight-start it off in the evening, wake up in the morning and it will be done. I get really good turnaround time and the fact that I can run a job overnight instead of having to wait a week, that's an incredible increase in my productivity."

The shared memory architecture of the SGI Altix system, which speeds up the design process by shortening data access time and computational runs, is the perfect tool for complex CFD analysis, whether designing next-generation swimsuits, cars, airplanes or rockets, where speed of analysis is paramount in time to market and maintaining a competitive business advantage.