Streamlining the time trial apparel of cyclists: the Nike Swift Spin project

Transcription

1 Streamlining the time trial apparel of cyclists DOI: /jst.12 Technical R&D Article Streamlining the time trial apparel of cyclists: the Nike Swift Spin project Len Brownlie 1,, Chester Kyle 2, Jorge Carbo 3, Nate Demarest 3, Edward Harber 3, Richard MacDonald 3 and Matt Nordstrom 3 1 Aerosports Research, Canada 2 Mechanical Engineering Department, California State University, Long Beach, USA 3 Nike Inc., USA This paper documents the development of aerodynamic apparel for the Tour de France individual time trial (TT), the Olympic TT, and track cycling races. A wind tunnel and metric balance were used to measure the drag force (F d )and wind tunnel air velocity on cylinders, limb models, and live cyclists clad in samples or suits sewn with one or more of 200 stretch fabrics. A concurrent measurement of model dimensions and frontal areas provided the nondimensional drag coefficient (C d ) and Reynolds Numbers (Re) that characterized the ability of the various fabrics and suits to reduce frictional drag and induce a drag crisis (DC) or premature flow transition. DC defines a critical air velocity over the body segments at which the airflow transitions from laminar to turbulent, yielding a smaller wake behind the body segment and a corresponding decrease in F d. A number of fabrics triggered DC on cylinders and limb segments, reducing cylinder and limb C d by over 40 per cent. Several methods of lowering the F d of cycling apparel proved effective, including custom fitting, aligning seams with the airflow, and matching fabric textures to body segments. Repeated drag measurements of the same cycling suit provided a mean drag of approximately 3200 g with a standard error of 729 g. The final 2005 individual TT suit design, worn by a pedaling cyclist, had a measured drag at 53 kph, which was 125 g less than typical 2001 TT cycling suits worn by competitors (F d reduction per cent). A mathematical model predicts that a drag difference of this magnitude would provide a time saving of approximately 44 s in a 55-km Tour de France TT. In and 2007, the production version of the Swift Spin TT suit was worn by the winner of the Tour de France; by the women s hour record holder and by road; TT and track cyclists who set four world records, six Olympic records, and won seven medals in individual cycling races at the 2004 Athens Olympics. & 2009 John Wiley and Sons Asia Pte Ltd Keywords:. aerodynamic drag. TT suit. NIKE Swift Spin. flow transition. zoned fabrics 1. INTRODUCTION *Aerosports Research, 5761 Seaview Place, West Vancouver, B.C. V7W 1R7, Canada firstsilver@hotmail.com In an individual time trial (TT) cycling race on a level course, aerodynamic drag (F d ) can be over 90 per cent of the total Sports Technol. 2009, 2, No. 1 2, & 2009 John Wiley and Sons Asia Pte Ltd 53

2 Technical R&D Article Len W. Brownlie et al. retarding force that impedes the forward velocity of the cyclist [1,2]. In many world class cycling races, the margin of victory is exceedingly small, so that subtle reductions in aerodynamic drag may be the difference between victory and defeat. Since the human body provides at least two-thirds of the air resistance of the bicycle and cyclist unit [3], racing cyclists have attempted to decrease the air drag of the body by lowering the riding position to increase aerodynamic efficiency [3]; however, the next most important method would be to streamline the cyclist s TT apparel. Kyle [4] was among the first to propose that aerodynamic apparel was important in cycling. Kyle and Caiozzo [5], Brownlie et al. [6], and Brownlie [7] noted the influence of apparel on running, cycling, and skiing aerodynamics. The implication of these publications was that specially-designed aerodynamic clothing would improve the performance of any high-speed sport, including cycling. This paper describes a multifaceted research program designed to develop more aerodynamic TT racing apparel that would permit higher speeds, with no obvious change in the appearance or functionality of the clothing. Previous aerodynamic measurements of athletic clothing have shown that the aerodynamic drag of certain fabrics can be quite different, even though there is little visible difference in the surface texture or appearance of the fabrics [1,7 11]. 1.1 Pressure Drag and Skin Friction As a cyclist moves forward, the air provides a retarding drag force (F d ), measured in Newtons or grams of force (g): F d ¼ 0:5 C d A p r V 2 ð1þ where C d is a non-dimensional drag coefficient, A p is the athlete s frontal area (m 2 ), r is the air density (kg m 3 ) and V is the relative velocity (defined as the vector sum of model and wind velocities, m s 1 ). Aerodynamic apparel may reduce total body F d by decreasing pressure or frictional drag on the cyclist. Pressure drag is caused by differences in air pressure between the leading and trailing surfaces of objects, such as cylinders or body segments, in cross-flow. As a smooth human limb moves through the air at low speed, the airflow on the leading surface is laminar and will separate from the limb near the widest point, leaving a large low-pressure wake (Figure 1). At higher limb and air velocities, airflow on the leading edge of the limb becomes turbulent, and these turbulent air packets have increased momentum and will follow the limb circumference past the limb s widest point, leaving a much smaller low-pressure cavity in the wake. At this critical flow velocity (termed the drag crisis [DC] or flow transition), the F d will quickly decline, with the magnitude of this reduction dependent upon the air velocity and surface texture characteristics of the limb. If a cylinder or limb segment has a smooth texture, the DC can be generated prematurely at lower velocities by covering the limb with fabrics that display surface textures of a specific roughness. The body segments that are adaptable to this strategy require a vertical or yawed orientation to the incoming airflow. In TT cycling, only the arms and legs can go through the DC, Figure 1. China clay flow visualization under UV light shows the line of flow separation on the upper arm of a cycling mannequin in the time trial position. since the torso is nearly horizontal and is parallel to the flow, where skin friction predominates. Skin friction is the second type of drag that may be reduced with aerodynamic apparel. Air molecules near the body surface of a cyclist will adhere to the surface and impede the flow along the surface, producing a shear force. Smooth surfaces generally create a lower shear force than rough surfaces since they trap less air. At high limb speeds, beyond where a smooth surface will spontaneously goes through the DC, rough surfaces will have a higher skin friction, as the rough surface causes a turbulent boundary layer, and this turbulent layer provides additional viscous stress. Furthermore, before the DC occurs, rougher surfaces also have a higher frictional drag than smooth surfaces. Thus, there is a limited range of velocities within which rougher surfaces provide an aerodynamic advantage. In areas of totally separated flow, such as the lower back, the surface texture does not affect drag and any appropriate fabric may be used in the apparel. These regions may be identified by flow visualization with oil smoke flow, china clay, or tufts of thread. In the TT position of cycling, separated flow occurs behind the legs and arms and on the lower back. Using this principle, the final TT suit was constructed such that most fabric seam lines ran parallel to the airflow and mesh cooling areas were restricted to the wake area behind the shoulders. Based on the characteristics of airflow around a cylinder and flow visualization on cyclists, we hypothesized that 54 & 2009 John Wiley and Sons Asia Pte Ltd Sports Technol. 2009, 2, No. 1 2, 53 60

3 Streamlining the time trial apparel of cyclists apparel sewn with composite fabrics covering the various body segments and torso could reduce both pressure drag and skin friction on a cyclist. 2. METHODS 2.1 Wind Tunnel Tests Aerodynamic measurements of the F d of cylinder models and other body segment models were made in the Parkinson wind tunnel of the University of British Columbia (UBC-P; Vancouver, Canada) and the wind tunnel at the Irvine Aerodynamics Laboratory, University of California (UCI; Irvine, CA, USA). Full-scale tests were conducted in three low-speed wind tunnels: (i) the Kirsten wind tunnel (Department of Aeronautical and Astronomical Engineering, University of Washington, Seattle, WA, USA); (ii) the boundary layer wind tunnel (UBC, Canada); and (iii) the low-speed wind tunnel (San Diego Air and Space Technology (San Diego, CA, USA). The aerodynamic characteristics of these tunnels are presented in Table 1. To measure drag forces, metric balances were used to collect and average either 1000 or 2000 drag samples for a given dynamic pressure over 20 s (static tests) or 60 s (pedaling tests). The longer sampling window was necessary in the pedaling tests to dampen out variations in inertial forces and drag during pedaling. The resolutions of the balances were between 1.2 and 5.9 g. In all wind tunnels, air velocity was calculated from the dynamic pressure, q, where: q ¼ 0:5 r V 2 and V was varied by a solenoid control on tunnel fan rotation speed. All cylinder and limb segment F d measurements were recorded at between four and nine V between 4.8 and 35.4 m s 1. For the cylinder tests, F d and V measurements were converted to non-dimensional Reynolds numbers (Re) and C d values to determine the DC or skin friction characteristics of each fabric. The Re characterizes the flow regime over ð2þ geometrically similar objects. In this paper, Re is defined as: Re ¼ l V v 1 where l is a representative width or length of a body segment or cylinder (m), V is the segment s maximum velocity, and v is the kinematic viscosity (m 2 s 1 ) of the air at a particular temperature and pressure. For the live cyclist tests, F d measurements were recorded at four evenly-spaced V between 13.3 and 16.1 m s 1. A simple linear regression equation was fitted to the F d and V 2 data from each test run. In general, the correlation between V 2 and F d was linear, and R 2, the regression coefficient, was essentially 1.00 for all model conditions that did not undergo whole body flow transition, indicating that C d was constant. This relationship was utilized to compare the F d of each model at a reference velocity. 2.2 Models Fitted sleeves of approximately 200 stretch fabrics were placed over plastic (8.9-cm diameter), aluminum (10-cm diameter), and galvanized steel (20.3-cm diameter) cylinders, and full-scale fiberglass leg models. Domed caps were placed on the top of the cylinder and leg models to smooth airflow over the model top. All models were vertically positioned in the wind tunnel with the 10-cm diameter aluminum cylinder also mounted horizontally on two aerodynamic struts in order to measure the frictional drag characteristic of various fabrics. In the full-scale tests, a modern TT or racing bicycle was fixed to the metric balance through a motorized mount. One national class triathlete (JL, height 180 cm; weight: 73 kg) served as the main test subject; however, a total of four national or international class cyclists and triathletes served as test subjects over the course of the development program. The A p of the main subject, as measured by planimetry, was 0.31 m 2, and the A p of the TT bicycle was 0.11 m 2. Once in the tunnel, the cyclist was positioned in a streamlined racing ð3þ Table 1. Aerodynamic characteristics of wind tunnels used in this study. Wind tunnel name Location Test section dimensions: height width (m) Test section cross-sectional area (m 2 ) y Maximum wind speed (m 1 ) Parkinson Boundary layer Irvine Aerodynamics Laboratory, University of California Kirsten: Low-speed wind tunnel: San Diego Air and Space Technology Center Department of Mechanical Engineering, University of British Columbia, Vancouver B.C., Canada Department of Mechanical Engineering, University of British Columbia, Vancouver B.C., Canada University of California, Irvine, CA, USA Aerodynamics Laboratory, University of Washington, Seattle, WA, USA San Diego, CA, USA y Excludes corner fillets. Sports Technol. 2009, 2, No. 1 2, & 2009 John Wiley and Sons Asia Pte Ltd 55

4 Technical R&D Article Len W. Brownlie et al. Figure 2. Position of the cyclist on the time trial bicycle in the wind tunnel. position on the TT bike with elbows resting on standard aerodynamic bar extensions (Figure 2). The cyclist s body position was fixed by reference to a side-view video image that was projected on a computer monitor. A silhouette of this image was drawn on the monitor with an erasable marker and used by the cyclist to orientate his body position before each test run. The bicycle crank was locked in position with the right crank top vertical during the first series of tests. A second series of dynamic tests had the cyclist pedaling at a cadence of RPM, as measured by a cadence sensor attached to a power measurement ergometer at the back of the bike mount. The front wheel remained stationary for all tests during , but was rotated at approximately 13 m s 1 during the tests in A third series of tests involved positioning the cyclist in an upright position, with the hands on the brake hoods in a simulated peloton or touring riding position. In all of the tests, the cyclist wore stretch shoe covers and an aerodynamic helmet. The object of the research program was to develop several methods of reducing the air resistance of cycle clothing. 3. RESULTS 3.1 Flow Transition of Fabric-Covered Cylinders Hoerner [12] glued sand grains of various sizes to the surface of cylinders to demonstrate that varying degrees of surface roughness could trigger an early boundary layer transition at Re from approximately to Hoerner found that the point of transition was determined by the ratio of the sand grain width to the cylinder diameter, with larger grains causing earlier transition. Given a large and small cylinder covered in sand grains of identical size, a small cylinder will transition at a lower Re than a large cylinder, because the roughness-to-diameter ratio of the small cylinder is greater. The UBC-P and UCI tests confirmed Hoerner s observation that certain textured stretch fabrics were found to trigger an early DC around a cylinder, with an often dramatic reduction in C d of over 40 per cent. This DC is repeatable in different tunnels and with different cylinder diameters. The UBC-P and UCI tests also confirmed that different fabrics cause a reduction in C d at different Re with rougher textured fabrics causing a DC at lower Re and smoother fabrics causing a DC at higher Re (Figure 3a). For example, in Figure 3a, fabric ]1 is an impermeable polyurethane-coated stretch fabric with a very smooth and shiny surface texture. This fabric did not cause the cylinder to undergo a DC in the range of velocities tested; however, this fabric did display a very low skin friction (Figure 3b), which would make it suitable for use in blocky areas (such as over the shoes) where a DC is difficult to generate, but where an aerodynamic advantage may be achieved through reduced skin friction and the development of a more streamlined shape. Fabric ]4 is a smooth, uncoated stretch knit fabric with a very subtle knit pattern. This fabric caused the cylinder to undergo a DC with a minimum C d of 0.54 at a relatively high Re of (or a velocity of approximately 31.5 m s 1 ). The combination of DC and skin friction characteristics (Figure 3b) of fabric ]4 led to its selection as a torso fabric in the 2005 cycling suit. Fabric ]40 is a multilayer stretch knit fabric with a pronounced surface knit pattern and an internal membrane that limits air permeability through the fabric. This fabric is commonly used in winter sports apparel, such as competitive ski suits. Fabric ]40 caused the cylinder to undergo an early DC with a minimum C d of 0.72 at a Re of (or a velocity of approximately 17 m s 1 ). Unfortunately, at higher Re, the surface texture characteristics, and perhaps the limited air permeability, cause the post-transition C d to plateau at approximately Fabric ]7 is a stretch knit fabric with a very pronounced surface texture that resembles a series of raised dimples and depressed hollows. At higher wind speeds, the raised dimples were observed to flatten, reducing the surface texture of the fabric. Fabric ]7 caused the cylinder to undergo an early DC with a minimum C d of 0.62 at a Re of The post-transition C d of the cylinder covered in fabric ]7 gradually increased from 0.72 to 0.76; however, this fabric provided one of the best patterns of flow transition and was used on the arms and lower thigh regions of the final cycling suit. As the cylinder diameter increased, we observed that the minimum C d shifted to a higher Re (i.e. to the right; Figure 4). In Figure 4, the subcritical values of the C d are different because the 10-cm diameter cylinder with a domed cap did not fully span the height of the test section, while the 20-cm diameter cylinder fully spanned the height of the test section. In Figure 4, the 10- cm diameter cylinder was covered in stretch fabric ]7 and went through DC at a Re of , while the 20-cm diameter cylinder covered in the same fabric went through DC at a Re of , which agrees reasonably well with Hoerner [12]. In addition, we found that covering the top and bottom portions of a cylinder in different fabrics leads to a pattern of flow transition, which is a composite of the pattern of each individual fabric. We also found that covering a fabric-covered cylinder with another stretch fabric results in a pattern of flow transition characteristic of the top fabric: when fabric ]40 is applied over fabric ]4, the C d versus Re curve more closely resembles the curve for fabric ]40 than fabric ]4 (Figure 3a). This result suggested that those areas of the final competition suit that required thicker fabrics for muscle heat retention could be 56 & 2009 John Wiley and Sons Asia Pte Ltd Sports Technol. 2009, 2, No. 1 2, 53 60

5 Streamlining the time trial apparel of cyclists (a) Cd Re x 10 5 (b) Re x over Fabric #7 on 10 cm cylinder Fabric #7 on 20 cm cylinder Figure 4. Drag coefficient (C d ) versus Reynolds number (Re) for 10- and 20-cm diameter cylinders covered in the same stretch fabric. As the cylinder diameter increases, the minimum C d is shifted to a higher Re (i.e. to the right) Re x diameter aluminum cylinder when it was orientated horizontally on two aerodynamic struts and placed streamwise to the airflow (Figure 3b). In general, the smoothest fabrics (e.g. fabrics ]1 and 4) provided the least frictional drag, with fabrics that have rougher surface textures (e.g. fabrics ]7 or ]40) providing higher skin friction. Similar results were also recorded for various styles of seaming, with low-profile flat, bonded, or taped seams providing less drag than exposed surged seams. Surface wrinkles also create excessive frictional drag; however, the measurement of the effects of wrinkling is best done on finished garments, as there is an interaction between the fit of the suit in the position of motion and the degree of wrinkling of the garment Figure 3. (a) Drag coefficient (C d ) versus Reynolds number (Re) for a vertically-orientated 10-cm diameter cylinder covered in various stretch fabrics. Where pressure drag is the predominant form of drag, different fabrics cause a reduction in C d at different Re, with the reduction in C d related to the surface texture of the fabric. (b) C d versus Re for a horizontally-orientated 10-cm diameter cylinder covered in various stretch fabrics. Where skin friction is the predominant form of drag, fabrics with a smoother surface texture create less frictional drag. covered with a more aerodynamic fabric to improve the overall aerodynamic characteristic of the suit. Although we measured the physical surface texture characteristics of many fabrics with a non-contacting surface profile meter, we were unable to accurately correlate surface profile measurements to the fabric s aerodynamic performance and continue to determine this quality experimentally in the wind tunnel. 3.2 Skin Friction of Fabric-Covered Cylinders We measured the difference in skin friction between various stretch fabrics by measuring the F d of the domed 10-cm 3.3 Drag of the Lower Leg We observed that the same stretch fabric that initiated a DC on the 8.9-cm diameter cylinder would also trigger a DC on the full-scale lower leg model (Figure 5). At a V of 18 m s 1, a lower leg covered in this fabric would create 110 g (17.9 per cent) less drag than the bare leg. The bare leg model and the leg clad in smooth fabrics did not experience a DC. Premature flow transition of limb segments at the velocities encountered in athletic events has been hypothesized, but has been difficult to measure, possibly because the tapered diameter of the limbs will cause a partial DC at a variety of air velocities [7,10]. 3.4 Drag of Cycling Suits In a TT, the air flow velocity over the arms and legs in cross-flow is essentially equivalent. Given the results of the cylinder tests, rougher textures should be required to induce transition on narrower diameter limbs, such as the upper arms, than on the wider diameter limbs, such as the thighs or calves. The horizontally-orientated torso is more streamlined, and the Sports Technol. 2009, 2, No. 1 2, & 2009 John Wiley and Sons Asia Pte Ltd 57

6 Technical R&D Article flow stays attached until it finally separates at the rear of the cyclist. In areas of attached flow, smooth fabrics that minimize skin friction should be used. The concept of placing fabrics with optimal aerodynamic properties over specific regions of the body, termed zoned fabrics, was shown to be effective in full-scale tests of TT apparel. While traditional cycling suits are generally composed of one or two fabrics, the current research demonstrated that zoned fabric suits containing up to five fabrics provided less air drag than these single or dual fabric suits. Table 2 shows the drag of a statically-positioned or pedaling cyclist clad in one of eight suits at a velocity of 53 kph. The 2001 TT suit and 2002 track racing suits are representative of current designs and contain one or two fabrics. The suits were either new production-zoned fabric suits or experimental prototypes. Test conditions for all tests were identical (Figure 2). Repeat drag measurements of the suits in Table 2 provided a mean drag of approximately 3200 g, with a standard error of 729 g or less. The static tests, with the cranks positioned vertically, had a higher drag than the pedaling tests. Since drag on the thighs, lower legs, and feet varies around the crank Drag Coefficient Bare leg Lower leg covered in #23 fabric Reynolds Number x 10 5 Figure 5. Drag coefficient versus Reynolds number for an 8.9-cm diameter cylinder and a full-scale lower leg model. cycle, the pedaling tests provide an average of the drag when pedaling at a cadence of RPM (Figure 6). Since leg motion will trigger flow transition, the difference in drag between suits was higher in the static tests than in the pedaling tests. Suits that did not generate flow transition when static often improved if the leg motion during pedaling forced transition. Although aerodynamic drag reductions of up to 50 per cent are possible with fabric-covered cylinders, the drag reductions possible with cyclists are of lesser magnitude ( per cent). The human body is a combination of complex, tapered oval cylinders, where the flow field of each cylinder affects the flow field of the other cylinders and where transition is initiated first on the largest diameter cylinder, then proceeding towards the smallest diameter segment. For example, in the TT racing position, a portion of the undisturbed air will become turbulent as it encounters the handlebar extensions, hands, and upper arms and then sweeps around the shoulders and head and continues over the back. This airflow, once detached from the upper arms, is not amenable to further surface roughness-induced flow transition. On the lower body, the rapid change from vertical to near horizontal orientation of the thighs, calves, and feet during a pedal stroke combined with the wake generated by each limb segment and the initial turbulence created by the spinning front wheel reduce the effectiveness of surface roughness-induced DC on any part of the leg. Brownlie et al. [11] found that a combination of cylinders does not undergo a sudden transition, but rather, the drag coefficient gradually declines as the air velocity increases and portions of various cylinders undergo transition or suffer Drag (g) Angle (degees) Drag (g) Len W. Brownlie et al. Figure 6. The crank angle on drag of a cyclist. A vertical right crank is zero degrees. (Kyle et al. [9]). Table 2. Aerodynamic drag of a cyclist and time trial (TT) bicycle at 53 Kph Suit Aero drag (g) Drag difference (g) y Drag increase (%) Time difference in 55 Km s TT suit, static TT suit, pedaling track suit, pedaling TT suit, static TT suit, static (Ref) +0 (Ref) 0 (Ref) 2003 TT suit, pedaling (Ref) +0 (Ref) 0 (Ref) 2004 TT suit, pedaling TT suit, pedaling track suit, pedaling prototype tracksuit and leggings, pedaling y All differences are statistically significant (Po0.05) & 2009 John Wiley and Sons Asia Pte Ltd Sports Technol. 2009, 2, No. 1 2, 53 60

7 Streamlining the time trial apparel of cyclists post-transition increases in C d. This pattern was also observed in the current investigation, where during several hundred individual wind tunnel runs conducted at velocities between 48 and 58 kph, the variation in the whole body drag coefficient at four speeds was normally less than 3 per cent. In a typical wind tunnel session, the least squares linear regression of F d versus q for each run had an average variability of R (n 5 51). 3.5 Effect of Apparel Size and Wrinkling The effect of suit wrinkling and fit were examined by testing a variety of suit patterns and sizes on a pedaling cyclist. Intuitively, tight-fitting apparel with few wrinkles should provide less frictional drag than a loose or improperly-sized garment. For example, we measured the F d of the subject at a velocity of 40 kph (11.17 m s 1 ) when he was wearing one of four sizes of streamlined cycling jersey or a standard cycling jersey (Table 3). The streamlined cycling jerseys were essentially the top portion of the TT suit, with a contoured fit and aerodynamic-zoned fabric composition. Drag measurements were taken when the cyclist was in an upright position, with the hands on the brake hoods in a simulated peloton or touring position. The small-size jersey was stretched very tightly over the cyclist, with little wrinkling; however, this size would not be comfortable to wear over the course of an extended Grand Tour stage. As the jersey size increased and became less form fitting, extensive wrinkling and billowing of the jersey fabric was observed, with increased drag related to increased jersey size. All of the streamlined jerseys were found to provide less F d than the a medium-size standard cycling jersey, illustrating that seam placement and garment pattern also contribute to the aerodynamic characteristics of cycling apparel. 3.6 Time Savings Associated with Aerodynamic Cycling Apparel Utilizing a mathematical model developed by Bassett et al. [3], time savings were calculated for a 55-km TT at an average velocity of 53 kph (Table 2), or a 190-km solo Grand Tour stage at an average velocity of 40 kph (Table 3). Comparing the 2005 TT suit worn by the winner of the Tour de France with the older 2001 TT suit, a time advantage of 44 s (pedaling) is predicted. When the Tour is very close, 44 s could be enough to win the entire race. In the long, flat stages of the Grand Tour races, cyclists typically ride together in a peloton, where the effects of aerodynamic drag on individual riders are reduced by mass drafting. However, cyclists who participate in long, solo breakaways would benefit from more aerodynamic jerseys. As an extreme example, a solo cyclist clad in a medium-size streamlined cycling jersey would generate 192 g less drag and would complete a 190-km flat stage 6 minutes and 38 s ahead of an identical athlete who was clad in a loose-fitting jersey. As shown in Table 2, leggings (tights) reduce the drag of the cyclist. In extended races, tights may impair thermoregulation; however, in shorter races, tights may provide a distinct aerodynamic advantage. Tights have not been traditionally worn in cycle racing, so customs and rules may govern their future utilization more than their aerodynamic potential. 3.7 Streamlined Apparel Usage in Cycle Racing The Swift Spin suit has been worn by a number of successful athletes. At the 2004 Athens Olympics, these suits were worn by cyclists in both the road TT and the track cycling events. As most track cyclists wore these suits for the first time in Athens, there was the opportunity to compare pre-olympic personal best performances with the cyclists performances in Athens. We focused our attention on four events (women s 500-m TT, Women s 3000-m individual pursuit, men s 1000-m TT, and the men s 4000-m individual pursuit), where the athletes performed at near maximum velocity for the entire race distance. We compared the Olympic performance of all athletes in these four events to their performances in the October 2003 Stuttgart and May 2004 Melbourne World Championships. In Athens, the average improvement from the World Championship time of cyclists wearing Swift Spin suits was 2.73 per cent (n 5 16), compared to 0.96 per cent (n 5 49) for cyclists wearing competing brands of apparel, a difference of 1.77 per cent. Cyclists who wore these suits in Athens set four world records; six Olympic records, and won seven medals. Perhaps the most noteworthy performance improvement was that of New Zealander Sarah Ulmer, who bettered the 2000 Sydney gold medal performance by 4.14 per cent and bettered her own 2004 World Championship world record in the 3000-m Individual Pursuit (IP) by 2.88 per cent when wearing the suit. The Swift Spin suit was also worn by the winner of the and 2007 Tour de France in TT events and by the current women s hour record holder. Table 3. Aerodynamic drag of a cyclist and bicycle at 40 Kph with the cyclist in an upright position with the hands on the brake hoods. Jersey size Aero Drag Grams Drag difference (g) Drag increase (%) Time difference in 190(min:s) Small s Medium :48 Large :22 Extra large :04 Medium 2002 TdF leader jersey :26 TdF, xxx. Sports Technol. 2009, 2, No. 1 2, & 2009 John Wiley and Sons Asia Pte Ltd 59

8 Technical R&D Article Len W. Brownlie et al. The cyclist s helmet, shoes, and gloves may also affect aerodynamic performance and most modern time trialists utilize streamlined helmets and stretch-coated fabric shoe covers. However, optimization of these items may provide only a fraction of the aerodynamic benefit of streamlined cycling suits [1]. The significance of zoned aerodynamic apparel may be proportionately larger in sports performed at higher velocities. Zoned apparel should be developed in these sports, because the resulting improvements in performance require neither changes in the athletic training regime or technique and are in a sense free speed. Acknowledgements Figure 7. Prototype suit with minimal seaming in cross-flow. This research was funded by Nike Inc. REFERENCES 4. CONCLUSION Through extensive wind tunnel measurements, we determined that the aerodynamic drag of TT cyclists can be significantly reduced through the development of streamlined TT apparel. The most effective apparel design innovations proved to be: 1. The use of lightweight stretch fabrics to allow adequate thermoregulation during extended periods of exercise. 2. The development of a garment pattern that produces the fewest wrinkles when the suit is worn in the TT position. 3. Custom fitting of each suit to the cyclist to minimize wrinkles or areas of loose fabric. 4. In regions of attached flow, the use of flat seams aligned parallel to the flow direction. 5. Placement of the majority of seams towards the rear or in regions of separated flow, where they have no impact on drag (Figure 7). 6. Placement of necessary cooling panels in shielded regions out of the air stream. 7. In zones of attached airflow, such as the torso, smoother fabric textures were utilized to minimize skin friction. 8. Appropriate fabric textures were carefully placed in crossflow zones to induce boundary layer transition from laminar to turbulent, and thereby reduce the low-pressure wake behind the segment. 9. The upper arms, thighs, and calves encounter cross-flow where appropriate surface textures of fabric were found to provide less air drag than bare skin. Although longer sleeves and tights might provide enhanced aerodynamic benefits, garment length appears to be determined by customs, regulations, cooling requirements, and factors other than aerodynamics. 1. Kyle CR. Improving the racing bicycle. Mechanical Engineering 1984; 106(11): Kyle CR, Weaver MD. Aerodynamics of human-powered vehicles. Proceedings of the I Mech E Part A Journal of Power and Energy 2004; 218: Bassett DR, Kyle CR, Passfield L, Broker J, Burke ER. Comparing cycling world hour records, : modeling with empirical data. Medicine and Science in Sports and Exercise 1999; 31(11): Kyle CR. Fast fashion: the aerodynamics of bicycle clothing. Bicycling 1985; 26(5): Kyle CR, Caiozzo VJ. The effect of athletic clothing aerodynamics upon running speed. Medicine and Science in Sports and Exercise 1986; 18(5): Brownlie L, Mekjavic I, Gartshore I, Mutch B, Banister E. The influence of apparel on aerodynamic drag in running. Annals of Physiological Anthropology 1987; 6: Brownlie L. Aerodynamic characteristics of sports apparel (Dissertation). Simon Fraser University: Burnaby B.C., Kyle CR. Athletic clothing. Scientific American 1986; 254: Kyle CR, Brownlie LW, Harber E, MacDonald R, Shorten MR. The Nike Swift Spin cycling project: reducing the aerodynamic drag of bicycle racing clothing by using zoned fabrics. In: Hubbard M, Mehta RD, Pallis JM, eds. The Engineering of Sport 5. International Sports Engineering Association, Sheffield, 2004; Brownlie LW, Kyle CR, Harber E, MacDonald R, Shorten MR. Reducing the aerodynamic drag of sports apparel: Development of the NIKE Swift sprint running and SwiftSkin speed skating suits. In: Hubbard M, Mehta RD, Pallis JM, eds. The Engineering of Sport 5. International Sports Engineering Association, Sheffield, 2004; Brownlie LW, Gartshore I, Chapman A, Bannister EW. The aerodynamics of cycling apparel. Cycling Science 1991; 3(3 4): Hoerner SF. Fluid dynamic drag. Hoerner Fluid Dynamics: Albuquerque, Received 30 November 2007 Revised 8 August 2008 Accepted 8 August & 2009 John Wiley and Sons Asia Pte Ltd Sports Technol. 2009, 2, No. 1 2, 53 60