2010 Human Kinetics, Inc. Marinho is with the Department of Sport Sciences, University of Beira Interior, Covilh, Portugal, and the Centre of Research in Sports, Health, and Human Development, Vila Real, Portugal. Barbosa is with the Centre of Research in Sports, Health, and Human Development, Vila Real, Portugal, and the Polytechnic Institute of Bragana, Bragana, Portugal. Reis is with the Centre of Research in Sports, Health, and Human Development, Vila Real, Portugal, and the University of Trs-os-Montes and Alto Douro, Vila Real, Portugal. Kjendlie is with the Norwegian School of Sport Sciences, Oslo, Norway. Alves is with the Fac- ulty of Human Kinetics, Technical University of Lisbon, Lisbon, Portugal. Vilas-Boas is with the Faculty of Sport, University of Porto, Porto, Portugal. Machado is with the Faculty of Sport, University of Porto, Porto, Portugal. Silva is with the Centre of Research in Sports, Health, and Human Development, Vila Real, Portugal, and the University of Trs-os-Montes and Alto Douro, Vila Real, Portugal. Rouboa is with the University of Trs-os-Montes and Alto Douro, Vila Real, Portugal, and the Department of Mechanical Engineering and Applied Mechan- ics, University of Pennsylvania, Philadelphia, PA. TECHNICAL NOTES Swimming Propulsion Forces Are Enhanced by a Small Finger Spread Daniel A. Marinho, Tiago M. Barbosa, Victor M. Reis, Per L. Kjendlie, Francisco B. Alves, Joo P. Vilas-Boas, Leandro Machado, Antnio J. Silva, and Abel I. Rouboa The main aim of this study was to investigate the effect of fnger spread on the propulsive force production in swimming using computational fuid dynamics. Computer tomography scans of an Olympic swimmer hand were conducted. This procedure involved three models of the hand with differing fnger spreads: fngers closed together (no spread), fngers with a small (0.32 cm) spread, and fngers with large (0.64 cm) spread. Steady-state computational fuid dynamics analyses were performed using the Fluent code. The measured forces on the hand models were decomposed into drag and lift coeffcients. For hand models, angles of attack of 0, 15, 30, 45, 60, 75, and 90, with a sweep back angle of 0, were used for the calculations. The results showed that the model with a small spread between fngers presented higher values of drag coeffcient than did the models with fngers closed and fngers with a large spread. One can note that the drag coeffcient presented the highest values for an attack angle of 90 in the three hand models. The lift coeffcient resembled a sinusoidal curve across the attack angle. The values for the lift coeffcient presented few differences among the three models, for a given attack angle. These results suggested that fngers slightly spread could allow the hand to create more propulsive force during swimming. Keywords: hand shape, numerical simulations, computational fuid dynamics, forces, competitive swimming The study of human swimming propulsion is one of the most complex areas of interest in sport biomechanics (Payton et al., 2002). Over the past decades, research in swimming biomechanics has evolved from the observa- tion of a subjects kinematics to a basic fow dynamics approach, following the line of the scientists working on this subject in experimental biology (Dickinson, 2000; Arellano et al., 2006). Computational fluid dynamics (CFD) is one of the recent methodologies used to achieve this goal. This methodology allows us to analyze the water fow around the human body, to understand the magnitude of drag forces resisting forward motion (Silva et al., 2008; Marinho et al., 2009), and to compute the propulsive forces produced by the propelling segments (Bixler & Riewald, 2002; Lecrivain et al., 2008). Computational fuid dynamics could help coaches, in the short term, with technique prescription. Moreover, this methodology could provide answers to some practi- cal issues that remain controversial. The fngers relative position during the underwater path of the stroke cycle is one of these cases. A large intersubject variety of relative fnger positioning can be observed during training and competition. Some swimmers (i) maintain the fngers closed together (not spread apart), (ii) others have a small distance between fngers, and (iii) still others have a large distance between fngers. Indeed, the propulsive repercussions of those three possibilities remain unclear for swimming coaches and scientists. There is a lack of research on this issue, and some ideas are passed among members of the swimming community with little empiri- cal (experimental or numerical data) support. Experi- mental data are controversial: for example, Schleihauf (1979) showed that the fngers closed together and the 88 Marinho et al. thumb partially abducted allow higher propulsion and Berger (1996) concluded that fnger spreading does not infuence propulsion. But a more recent paper suggests that fngers closed together induces less propulsion than fngers spread (Sidelnik & Young, 2006). To our knowl- edge, there is no research published using a numerical approach on the effect of fnger spreading and with anthropometrical data of elite swimmers hands. Therefore, the main aim of this study was to inves- tigate the effect of fnger spread on propulsive force production in swimming using CFD. Methods Three-Dimensional Model Scanning. To obtain the geometry of the hand, eight cross-sectional scans of the right hand of an elite swimmer (Figure 1) were conducted using a Toshiba Aquilion 4 computer tomography scanner. Computer tomography scans were obtained with confguration of V2.04 ER001. A 2-mm-slice thickness with a space of 1 mm was used. The subject was an Olympics-level swimmer who participated in the 2004 Olympic Games, in Athens. The subject was lying prone, with his right arm extended ahead and fully pronated. This procedure was conducted with different fnger spreads: fngers closed together, fngers with a small spread (an intrafnger distance of 0.32 cm, from fngertip to fngertip), and fngers with a large spread (0.64 cm, from fngertip to fngertip) (Schleihauf, 1979). This protocol has been approved by the appropriate ethical committee of the institution in which it was performed and the subject gave informed consent to participate in this work. Data Manipulation. The transformation of values from the computer tomography scans into nodal coordinates in an appropriate coordinate system warrants the use of image-processing techniques. The image-processing program used in this study was the Anatomics Pro (Anatomics, Saint Kilda, VIC, Australia). This program allowed us to obtain the boundaries of the human segments, creating a three-dimensional reconstruction of the hand. At frst, before processing and converting procedures, the data were prepared by observing the computer tomography data and erasing the irrelevant parts of the anatomical model. This step was also conducted using the software FreeForm (SensAble Technologies, Woburn, MA, USA). Finally, the data were converted into an IGES format (*.igs), which could be read by Gambit/Fluent software (Fluent Inc, Lebanon, NH, USA) to defne the fnite elements approach through the three- dimensional surfaces (Figure 2). CFD Study The Fluent code solves fow problems by replacing the Navier-Stokes equations with discretized algebraic expressions that can be solved by iterative computerized Figure 1 Anthropometric characteristics of the swimmers hand. Hand length (1): 20.20 cm, index breadth (2): 1.50 cm, index length (3): 8.10 cm, palm length (4): 9.50 cm, and hand breadth (5): 8.90 cm. Propulsion Forces and Finger Spread 89 calculations. Fluent uses the fnite volume approach, where the equations are integrated over each control volume. The dynamic fuid forces produced by the hand, lift (L) and drag (D), were measured in this study. These forces are functions of the fuid velocity and they were measured by the application of the Equations 1 and 2, respectively: D = C D 1/2 A v 2 (1) L = C L 1/2 A v 2 (2) In Equations 1 and 2, v is the fuid velocity, C D and C L are the drag and lift coeffcients, respectively, is the fuid density, and A is the projection area of the model for the angles of attack used in this study. Preprocessing. The whole domain was meshed with a hybrid mesh composed of prisms and pyramids. Signifcant efforts were conducted to ensure that the model would provide accurate results by decreasing the grid node separation in areas of high velocity and pressure gradients. Solving Steady Flow. For the calculations, hand model angles of attack of 0, 15, 30, 45, 60, 75, and 90, with a sweep back angle of 0 (thumb as the leading edge) were used (Schleihauf, 1979). Steady-state CFD analyses were performed using the Fluent code, and the drag and lift coeffcients were calculated for a fow velocity of 2.0 ms 1 (Lauder et al., 2001; Rouboa et al., 2006). We used the segregated solver with the standard K-epsilon turbulence model because this turbulence model was shown to be accurate with measured values in previous research (Moreira et al., 2006). All numerical computational schemes were second order, which provides a more accurate solution than frst- order schemes. We used a turbulence intensity of 1.0% and a turbulence scale of 0.10 m. The water temperature was 28 C with a density of 998.2 kgm 3 and a viscosity of 0.001 kg(ms) 1 . Incompressible fow was assumed. The measured forces on the hand models were decom- posed into drag (C D ) and lift (C L ) coeffcients, using Equations 1 and 2. Results Figures 3 and 4 show the values of C D and C L , respec- tively, obtained for the hand model with different fnger spreads. One can note that the C D presented the highest values for an attack angle of 90 in the three hand models (0.90 < C D < 1.10). In the three models, the C D increased with the attack angle. Moreover, it was possible to observe that for attack angles greater than 30, the model with the small distance between fngers presented higher values of C D when compared with the models with fngers closed and with large fnger spread. This last model presented the lowest values of C D . For attack angles of 0, 15, and 30, the values of C D were very similar in the three models of the swimmers hand. The C L resembled a sinusoidal curve across the attack angle. Maximum values for any hand model occurred near 3045 (C L 0.60). Furthermore, the C L seemed to be independent of the fnger spreading, thus presenting little differences among the three models. However, it was possible to note slightly lower values for the position with a larger distance between fngers, especially for attack angles ranging from 15 to 60. Figure 2 Computational fuid dynamics model geometry with the hand inside the domain (the model with fngers closed). 90 Figure 3 Values of C D obtained for the different attack angles and for the different fnger spreads. Sweepback angle = 0 and fow velocity = 2.0 m/s. Figure 4 Values of C L obtained for the different attack angles and for the different fnger spreads. Sweepback angle = 0 and fow velocity = 2.0 m/s. Propulsion Forces and Finger Spread 91 Discussion The main aim of this study was to analyze the effect of fnger spread in the swimming propulsive force produc- tion, through CFD. Results suggested that fngers slightly spread could allow the hand to create more propulsive force during swimming. In this study, we tried to clarify one technical concern of the swimming community: which should be the best fnger position to improve force production by the hand during swimming? Therefore, three models with differing fnger spread were chosen for the analysis, addressed to characterize different swimming strategies. In addition, the option to analyze one position with fngers closed, one with a small distance between fngers, and another with a large distance between fngers was based on the pioneering study of Schleihauf (1979). Despite some theoretical assumptions and expert opinions (e.g., Coun- silman, 1968; Colwin, 1992; Maglischo, 2003), there are few experimental studies to clarify this issue (Schleihauf, 1979; Takagi et al., 2001; Berger, 1996; Sidelnik & Young, 2006). Rather than an experimental analysis, the current study applied the numerical techniques of CFD to compute the forces produced by the model of the swim- mers hand. Bixler et al. (2007) has already demonstrated the validity of CFD analysis as a tool to examine the water fow around a swimmers body. Nevertheless, it is very important that the digital model corresponds to a truthful representation of the human segment to ensure accurate numerical results (Candalai & Reddy, 1992; Lecrivain et al., 2008). Indeed, the computer tomography scans allowed the creation of a true digital model of the swimmers hand (Aritan et al., 1997). Moreover, precise images of complex 3-D shapes, such as a human hand, obtained by imagiography are becoming widely used in reverse engineering (Lecrivain et al., 2008). The main fnding of the present research was that the model with the small distance between fngers presented higher values of C D than the models with fngers closed together and with fngers spread widely. Furthermore, the C L seemed to be independent of the fnger spread, presenting few differences among the three models. These results suggest that the use of a position with a small dis- tance between fngers seems to be gainful for swimmers. The hand position with the small distance between fngers seemed to increase the projection area of the hand, thus increasing force production. The distance between fngers seemed not enough to allow the water to fow freely. Indeed, a turbulent fow between the fngers may be formed, creating some kind of barrier. Nevertheless, regarding the C L , the values for the position with the small fnger spread and for the position with fngers closed were very similar. For attack angles lower than 90, the fow above the dorsal surface of the hand, fowing at high velocities, could prevent the fow between fngers. In this condition, assuming that the higher velocity difference between the two surfaces of the swimmers hand will occur at the attack angle corresponding to the higher C L (in this case, between 30 and 45), it will thus be expected that the so-called barrier will be stronger at those C L values. As can be seen from Figures 3 and 4, at = 45 a relative increase of the C D value is perceptible. This curve tendency corresponds to the maximal C L value obtained for the slight spreading condition, and for all studied conditions, indicating the higher fow velocity difference between both faces of the hand. Concerning this, Ungerechts & Klauck (2006) did suggest having fngers slightly spread to induce fow around the hand at the beginning of the arm cycle. However, this gain did not occur when we analyzed the greater distance between fngers. In both C D and C L
coeffcients, for the position with large fnger spread, the values were lower when compared with the positions of fngers closed and slightly spread. For the C D and for attack angles higher than 30, the position with more distance between fngers presented lower values. This position presented also lower values in C L . It seems that there is a critical distance between fngers beyond which the force production became compromised. Schleihauf (1979) has already reported an identi- cal situation. The C D for the fngers closed and slightly spread positions presented higher values than the large spread position. In contrast, the values of C L increased in indirect proportion to fnger spread for attack angles ranging between 0 and 60. Berger (1996) reported that spreading the fngers did not infuence propulsive force. Moreover, lift force at attack angles between 60 and 80 was higher when spreading the fngers (Berger, 1996). In a recent experimental study, Sidelnik & Young (2006) determined that a hand with 10 of separation between fngers created more stroke force than a fngers-together confguration, across all attack angles tested. Furthermore, C D presented the highest values for an attack angle of 90 in the three hand models (0.90 < C D < 1.10), whereas C L resembled a sinusoidal curve across the attack angle (C L 0.60). These results are quite similar to the ones already described with experimental methodologies (e.g., Schleihauf, 1979; Berger et al., 1995; Takagi et al., 2001). In summary, this study showed that CFD methodol- ogy can be an important tool for coaches and swimmers to improve performance. However, the present results were obtained using steady fow simulations. Further studies should include the unsteady effects of motion, such as accelerations, decelerations, and rotations (Sand- ers, 1999). It would be interesting to observe whether the results would be the same as suggested by Ungerechts & Klauck (2006). These authors proposed the use of fngers slightly spread to induce fow around the hand at the beginning of the arm cycle and to create unsteady fow to allow a marked increase of propelling momentum. Although the results of the present numerical research showed that fngers slightly spread created more force, this is a comparison of only three hand positions. In the future, there are many hand shape parameters that could be included by varying for instance wrist angle, thumb abduction, and hand confguration (fat vs. cupped palm and fexed vs. extended interphalangeal joints). 92 Marinho et al. Acknowledgments This work was supported by the Portuguese Government by a grant of the Science and Technology Foundation (SFRH/ BD/25241/2005; PTDC/DES/098532/2008). We would like to acknowledge the staff of the Radiology Department of Hospital de So Joo, in Porto, Portugal, and personally to the depart- ment director, Isabel Ramos. We would also like to express our gratitude to the Med Mat Innovation Company, in Maia, Portugal, and especially to Jos Domingos Santos and Bruno S for their contributions. References Arellano, R., Nicoli-Terrs, J.M., & Redondo, J.M. (2006). Fundamental hydrodynamics of swimming propulsion. Portuguese Journal of Sport Sciences, 6(Suppl. 2), 1520. Aritan, S., Dabnichki, P., & Bartlett, R. (1997). Program for generation of three-dimensional fnite element mesh form magnetic resonance imaging scans of human limbs. Medi- cal Engineering & Physics, 19(8), 681689. Berger, M. (1996). Force generation and effciency in front crawl swimming. PhD Thesis. Amsterdam: Faculty of Human Movement Sciences, Vrije Universiteit. Berger, M.A., de Groot, G., & Hollander, A.P. (1995). Hydrody- namic drag and lift forces on human/arm models. Journal of Biomechanics, 28, 125133. Bixler, B., Pease, D., & Fairhurst, F. (2007). The accuracy of computational fuid dynamics analysis of the passive drag of a male swimmer. Sports Biomechanics, 6, 8198. Bixler, B., & Riewald, S. (2002). Analysis of swimmers hand and arm in steady fow conditions using computational fuid dynamics. Journal of Biomechanics, 35, 713717. Candalai, R.S., & Reddy, N.P. (1992). Stress distribution in a physical buttock model: effect of simulated bone geometry. Journal of Biomechanics, 15, 493504. Colwin, C. (1992). Swimming into the 21th century. Champaign, Illinois: Leisure Press. Counsilman, J. (1968). The science of swimming. Englewood Cliffs: Prenctice-Hall Inc. Dickinson, M.H. (2000). How animals move: an integrative view. Science, 288, 100106. Lauder, M.A., Dabnichki, P., & Bartlett, R. (2001). Improved accuracy and reliability of sweepback angle, pitch angle and hand velocity calculations in swimming. Journal of Biomechanics, 34, 3139. Lecrivain, G., Slaouti, A., Payton, C., & Kennedy, I. (2008). Using reverse engineering and computational fluid dynamics to investigate a lower arm amputee swimmers performance. Journal of Biomechanics, 41, 28552859. Maglischo, E.W. (2003). Swimming fastest. The essential reference on training, technique and program design. Champaign, Illinois: Human Kinetics. Marinho, D.A., Reis, V.M., Alves, F.B., Vilas-Boas, J.P., Machado, L., Silva, A.J., et al. (2009).Hydrodynamic drag during gliding in swimming. Journal of Applied Biomechanics, 25, 253257. Moreira, A., Rouboa, A., Silva, A., Sousa, L., Marinho, D., Alves, F., et al. (2006). Computational analysis of the turbulent fow around a cylinder. Portuguese Journal of Sport Sciences, 6(Suppl. 1), 105. Payton, C., Baltzopoulos, V., & Bartlett, R. (2002). Contribu- tions of rotations of the trunk and upper extremity to hand velocity during front crawl swimming. Journal of Applied Biomechanics, 18(3), 243256. Rouboa, A., Silva, A., Leal, L., Rocha, J., & Alves, F. (2006). The effect of swimmers hand/forearm acceleration on propulsive forces generation using computational fuid dynamics. Journal of Biomechanics, 39, 12391248. Sanders, R.H. (1999). Hydrodynamic characteristics of a swim- mers hand. Journal of Applied Biomechanics, 15, 326. Schleihauf, R.E. (1979). A hydrodynamic analysis of swimming propulsion. In J. Terauds & E.W. Bedingfeld (Eds.), Swim- ming III (pp. 70109). Baltimore: University Park Press. Sidelnik, N.O., & Young, B.W. (2006). Optimising the free- style swimming stroke: the effect of fnger spread. Sports Engineering, 9, 129135. Silva, A.J., Rouboa, A., Moreira, A., Reis, V., Alves, F., Vilas- Boas, J.P., et al. (2008). Analysis of drafting effects in swimming using computational fuid dynamics. Journal of Sports Science and Medicine, 7(1), 6066. Takagi, H., Shimizu, Y., Kurashima, A., & Sanders, R. (2001). Effect of thumb abduction and adduction on hydrodynamic characteristics of a model of the human hand. In J. Black- well, & R. Sanders (Eds.), Proceedings of Swim Sessions of the XIX International Symposium on Biomechanics in Sports (pp. 122126). San Francisco: University of San Francisco. Ungerechts, B., & Klauck, J. (2006). Consequences of unsteady flow effects for functional attribution of swimming strokes. Portuguese Journal of Sport Sciences, 6(Suppl. 2), 109111.