Different Strokes

CFD analysis helps optimize a swimmer's performance

By Karen Moltenbrey

In competitive swimming, a race can be won by a single stroke-sometimes it's a stroke of luck, but more often it's a stroke of the swimmer's arm. No one is quite sure just how much of an effect the subtle change in positioning of a swimmer's hand and arm has on peak performance, since conventional theories haven't been proven through extensive analysis. New research using 3D graphics, however, may soon show that certain swimming techniques really do hold water.

Until recently, most swimming research has been strictly experimental, centering on physical drag testing in a wind tunnel and water flume, neither of which produced entirely accurate results, says Scott Riewald, biomechanics director of USA Swimming, the national governing body for competitive swimming in the US. The reason, he notes, is that these tests focus on a single component-the hand, arm, leg, or body-rather than the effect they all have on the overall movement. But a year ago, USA Swimming found a more effective solution from Barry Bixler, an aerospace engineer at Honeywell Engines and Systems in Phoenix. Bixler, whose daughter swims competitively on a regional level, began applying computational fluid dynamics (CFD)-a technique he uses at Honeywell for evaluating airflow to optimize the design of jet engines-to understand the fluid mechanics of swimming.

Working as a hobbyist, Bixler started "playing around with the technology" a few years ago to investigate the effects of fluid flow in swimming. In 1998, he began to work in earnest on his model. Using software from Fluent Inc. (Lebanon, NH), Bixler began running computer simulations of a swimmer's hand and arm, altering certain variables such as water turbulence and the velocity and position of the hand and arm during a swim stroke. Encouraged when his preliminary CFD results compared well with traditional physical experimental data, he contacted USA Swimming and forged a relationship with the organization to help fine-tune the strokes of its elite swimmers.

The results of a computational fluid dynamics (CFD) analysis superimposed on a swimmer's arm and hand shows the progression of the water across the skin surface during a freestyle stroke. (Image courtesy Fluent Inc.)

When Bixler first began experimenting with his analyses, he used crude techniques in his garage to create a physical model of his own hand and arm. After making many cross-section cuts through his model, he digitized points around the perimeter of each cross section. "Now I'd definitely use a 3D laser scanner or a similar method," Bixler says. "But back then, I was just doing this for fun and didn't think it would turn into anything. So I used whatever materials were readily available."

After those preliminary steps, however, Bixler relied strictly on professional tools and methodologies. The engineer read those points into MSC.Software's (Los Angeles) Patran program to create a 3D surface mesh of his arm and hand. He then used Fluent's TGrid program to create a volume mesh of tiny fluid cells (to represent water) around the surface of his digital arm model. Each of those cells contained node points at the corners, where complex equations of fluid flow were solved iteratively throughout the model using Fluent's CFD program.

Conducting such an analysis with high-end tools is usually too costly for a hobbyist, but Bixler was able to use the Fluent software running on a Unix workstation at Honeywell. As his work became more time-consuming and involved, Fluent donated a copy of its CFD software to Bixler for minimal cost, which he runs at home on a dual-processor Pentium III. Intel also has pledged to provide Bixler with the additional computer power needed to process the extensive data that will arise from his further analyses.

A CFD model surface mesh of the swimmer's hand and fingers is refined during the analyses.

Bixler's ultimate goal is to design the optimum stroke for swimmers-how they should move their hands and arms through the water to maximize the propulsive force that they're applying to the water to propel themselves. "There's been a lot of debate over the most effective way to generate propulsion and what your hand path should be," says Riewald. "But that's been difficult to measure experimentally with an athlete." However, through Bixler's early re search, he has established a firm analytical foundation upon which to build more complex analyses. To determine the best stroke, he is using a three-phase approach.

Last year, Bixler completed the first phase: analyzing the hand and arm at various velocities and orientations for a variety of water turbulence conditions. All analyses in this phase were steady state, where the velocity and orientation of the arm and hand did not change. He then plotted drag and lift coefficient curves as a function of attack angle and water turbulence, and found that the aerodynamic efficiency of the hand is significantly less than an airfoil of similar aspect ratio. "The idea is to develop the highest propulsive drag and lift possible on a swimmer's hand as it strokes the water, while minimizing the resistance drag on the swimmer's body," he explains.

While Bixler found the optimal stroke using steady-state solutions, that position cannot be verified for the dynamic state until the final phase of analysis is completed, possibly next year. In fact, he compares the results from this first phase of analysis to those from the traditional physical tests in that neither takes into account interference drag. "By analyzing each piece separately, you're chopping off flow at one end and changing the drag," he says. "That's not the way it happens in swimming. Swimming is a complex motion that depends on numerous factors and interrelationships."

In this CFD model, the gray area (in the middle) shows where the fluid leaves the skin's surface, indicating increased drag as the hand moves through the water.

For Bixler, this first phase was a springboard to more complicated analyses. "I wanted to start out simple and get some information that I was fairly confident with, based on wind-tunnel and tow-tank testing," he says. "I was able to compare my steady-state lift and drag coefficients with the experimental data and prove to myself that the analytical methods and techniques I chose for modeling were accurate. That gave me the confidence to move into phase two, where there's not a lot of experimental data with which to compare my results."

In phase two, Bixler increased the complexities of the problems, examining the effects of accelerating and decelerating the arm during its motion. To analyze this unsteady flow required transient analyses, for which he solved the problem at one point in time, then increased the velocity slightly over time and solved it again. Eventually, Bixler obtained a time-history plot of the drag and lift forces on the hand and arm versus time that showed these forces to be affected unequally by acceleration and deceleration.

In phase three, which will begin next year, Bixler will examine the effects of the arm and hand rotation through the water-which will be close to simulating an actual stroke. "We'll have acceleration, deceleration, and direction and rotation changes of the arm. Then we'll be in a position to try to optimize all those variables as they come together to calculate the largest propulsive force," he says. "Until we complete that final phase, I can't make recommendations for stroke improvement, since phase three will affect the findings in the previous phases." Therefore, the analysis results were not used to coach this year's US Olympic swim team, but Bixler predicts they will impact the performance of the 2004 Olympians.

Even with these early analysis results, Riewald is extremely optimistic. "I believe that we'll be able to make some statements and claims soon," he says. "I've seen the preliminary results of the studies, and they look encouraging. I believe they will be able to provide new insight into what occurs during competitive swimming." For instance, swimmers competing in the freestyle and butterfly events presently are instructed to move their hands in a reverse "S" path. According to Bixler, his preliminary test results indicate that the optimum stroke will most likely follow a straighter path through the water, emphasizing the drag component of propulsion rather than the lift component.

In conjunction with the last phase of testing, Bixler plans to conduct a separate full-body analysis-something that may be of particular interest to Speedo, maker of the full-body swimsuits that made such a splash at this year's international games. According to Bixler, Fluent's moving mesh capability that is currently under development by the company would further that effort by enabling him to apply a swimmer's butterfly or scissors kick to his 3D model. "Ultimately, I'd like to create a single model of a swimmer's body with moving arms and legs."

A vortex appears on the downstream side of this hand model as water passes over the skin, indicating high turbulence caused by increased propulsion.

Currently, Bixler could not estimate the time-savings that would result from a swimmer using the "optimal stroke." However, in competitive swimming, races are often won by hundredths of a second over short distances, so every incremental change that increases a swimmer's time will be seen as an advantage. "We're looking to make small changes to hone a person's skills to where the swimmer may get that extra tenth of a second," adds Riewald, "which could be the difference between winning the gold or fourth place."

Fluent, Fluent Inc. (www.fluent.com)

CG images courtesy Barry Bixler.