The Engineering of Sport 7 Vol. 1
Springer Paris Berlin Heidelberg New York Hong Kong Londres Milan Tokyo
Margaret Estivalet Pierre Brisson
The Engineering of Sport 7 Vol. 1
Margaret Estivalet ESTIA Technopole Izarbel 64210 Bidart France
Pierre Brisson UTC 66, avenue de Landshut 60200 Compiègne France
ISBN-13: 978-287-09410-1 Springer Paris Berlin Heidelberg New York
© Springer-Verlag France, Paris, 2008 Printed in France Springer-Verlag France is member of groupe Springer Science + Business Media Apart from any fair dealing for the purposes of the research or private study, or criticism or review, as permitted under the Copyright, Designs and Patents Act 1998, this publication may only be reproduced, stored or transmitted, in any forrn or by any means, with the prior permission in writing of the publishers, or in the case of reprographic reproduction in accordance with the terms of licenses issued by the copyright. Enquiry concerning reproduction outside those terms should be sent to the publishers. The use of registered names, trademarks, etc., in this publication does not imply, even in the absence of a specific statement that such names are exempt from the relevant laws and regulations and therefore free for general use.
Cover design: Jean-François Montmarché
ISEA 2008, just before the summer olympic games! What a fantastic opportunity to present a compilation of more than 160 articles talking about sports engineering, analysing the coefficients of friction between the balls and the rim and back-board for leather and synthetic basket balls, extracting the aerodynamic force data during real ski jumping flights, optimizing new prosthesis of the lower human leg, analysing the golf ball spin rate after impact, analysing the most common injury in sport climbing using eight fresh dozen cadaver fingers, describing the heat transfer in footwear using finite elements, measuring the aerodynamic performance of cycling time trial helmets, etc, … What a challenge too to be honnest1 A huge diversity of articles, top level contributions to sports engineering. Today the world is convinced sport is not only fun but economically a sector, a multi sector, which is not only growing if you only take into acccount the total turn over but is becoming one of the fast growing business. Sport is not any more reserved for top sporters who want to maintain a certain level in some disciplines, it became a new philosophy of life, a new trend, a way to cope with aging population, with the reality of the society today. Our every day life is concerned with sport or sport derived products or services, it is in our shoes, our suits, our car, our bike, at home, when we eat, when we drink, when we sleep, relax, when we look at TV for international events, when we listen, watch the news, for fun.The sports engineering community as it was noted two years ago keeps growing. We have to admit it was a very difficult task to review all the contributions and to come down to 150 articles; It was very difficult too to allocate reviewers to contributions because a lot of articles were proposing not only scientific contributions but also engineering solutions and methodologies. Some groups of articles could have been selected as a basis for a workshop in itself! In front of such a diversity of contributions we have decided not to group the articles by families, by themas, by keywords, by branches, by sports, by subjects, by numbers of contributions but we decided to regroup it in two different volumes without any introduction which we thought would not bring anything to the readers, just proposing the articles in a natural order creating of course some surprises, but it was a choice! Of course there is a table listing the articles with their authors and co-authors and the programme will indicate evey time the article number. Complex to read? Difficult to apprehend? We thought it would give the best way to understand the complexity of sports engineering today; An article about football in a ball section of proceedings, in the shoes section, in the field surface section, in the injury section, in the training part, in the video group, in the sliding effect paragraph, in the referee point of view chapter, in the leather section may be, why not the aerodynamics or the finite elements analysis, may be in the professional sports section or the leisure, the TV business, the star system, … So many possibilities, we just did it in the way we were convinced would be the most open!
VI The Engineering of Sport 7 - Vol. 1 What we wanted to do is to provide the readers with the best sports engineering contributions in 2008, before the biggest sports event on earth, the olympic games, in front of 5 billions telespectators who will enjoy the show and for many of them start again sporting, or just start a new sport, realising what they can do, discover a new passion, using in any case the brain storming of the world of engineering contributors to improve our every day life. This is the magic of sport1 Margaret ESTIVALET & Pierre BRISSON
Contents
Effects of Body Weight on Ski Jumping Performances under the New FIS Rules (P3) ..............................................................................................
1
Luca Oggiano, Lars Sætran
Calculated Golf Ball Performance Based on Measured Visco-hyperelastic Material Properties (P5) ......................................................
11
Khairul Ismail, Bill Stronge
Interaction of Flexor Tendons and Pulleys in Sport Climbing (P6) .................................................................................................................
19
Andreas Schweizer, Beat Moor, Hans-Peter Bircher
Friction Between Players’ Hands and Sports Equipment (P7) .....
27
S.E. Tomlinson, R. Lewis, M.J. Carré
Development of a Comfort Model for Cricket Leg Guards (P9)
35
James Webster, Jonathan Roberts, Roy Jones
Enabling Technologies for Robust Performance Monitoring (P10) .................................................................................................................................
45
Laura Justham, Sian Slawson, Andrew West, Paul Conway, Michael Caine, Robert Harrison
An Objective Performance and Quality Comparison of Drivers from Different Market Sectors (P11) ............................................
55
Jeff Brunski, John Rae
Defining Strategies for Novel Snowboard Design (P12) .......................
65
Aleksandar Subic, Patrick Clifton, Jordi Beneyto-Ferre
Business Process Modelling and its Use Within an Elite Training Environment (P15) ...........................................................................
73
Laura Justham, Sian Slawson, Andrew West, Paul Conway, Michael Caine, Robert Harrison
Accelerometer Profile Recognition of Swimming Strokes (P17) .
81
S.E. Slawson, L.M. Justham, A.A. West, P.P. Conway, M.P. Caine, R. Harrison
Evaluation of Start Techniques in Sports Swimming by Dynamics Simulation (P18).................................................................................................................................... 89 Thomas Härtel, Axel Schleichardt
VIII The Engineering of Sport 7 - Vol. 1 A simulation of outrigger canoe paddling Performance (P19) .............................................................................................................................
97
Nicholas Caplan
The Dynamic Compaction of Cricket Soils for Pitch Preparation (P20) ................................................................................................................................. 107 Peter Shipton, Iain James
Experimental Validation of a Finite-element Model of a Tennis Racket String-bed (P21)...................................................................................... 115 Tom Allen, Simon Goodwill, Steve Haake
Experimental Validation of a Tennis Ball Finite-element Model for Different Temperatures (P22) ..................................................................................... 125 Tom Allen, Simon Goodwill, Steve Haake
Nonlinear Dynamics of a Simplified Skateboard Model (P24) ..... 135 Alexander S. Kuleshov
Cricket Batting Stroke Timing of a Batsman When Facing a Bowler and a Bowling Machine (P26) ..................................................................... 143 Alex Cork, Laura Justham, Andrew West
Estimation of a Runner’s Speed Based on Chest-belt Integrated Inertial Sensors (P27) ...................................................................................................................... 151 Rolf Vetter, Emanuel Onillon, Mattia Bertschi
Design and Construction of a Custom-made Lightweight Carbon Fibre Wheelchair (P28) .......................................................................................... 161 Marc Siebert
Design and Implementation of a Rugby-specific Garment Evaluation Trial (P30)..................................................................................................................... 169 Bryan C. Roberts, Gareth Williams, Mike P. Caine
Open Rotator Cuff Surgery in Swiss Elite Rock Climbers (P31) . 177 Hans-Peter Bircher, Christoph Thür, Andreas Schweizer
A Quantitative Analysis Of Beach Casting (P33) ........................................... 183 Benjamin Charles, Darryl P Almond, Aki I T Salo, Presented by Alan N Bramley
An Assessment of Sensing Technologies to Monitor The Collision of a Baseball and Bat (P34)............................................................... 191 Lawrence Fallon, James Sherwood, Michael Donaruma
Correlation Between the Linear Impulse and Golf Ball Spin Rate (P35) ........................................................................................................................................ 199 Woo-Jin Roh, Chong-Won Lee
The Engineering of Sport 7 - Vol. 1
IX
Dynamics-based Force Sensor Using Accelerometers-application of Hammer Throw Training Aid- (P37) .................................................................... 207 Ken Ohta, Koji Umegaki, Koji Murofushi, Ayako Komine, Chikara Miyaji
Influence of Pedal Foot Position on Muscular Activity during Ergometer Cycling (P39) .............................................................................................................. 215 Stefan Litzenberger, Sandrina Illes, Martin Hren, Martin Reichel, Anton Sabo
Accurate Trajectory and Orientation of a Motorcycle derived from low-cost Satellite and Inertial Measurement Systems (P42)............. 223 Adrian Waegli, Alain Schorderet, Christophe Prongué, Jan Skaloud
Wireless Impact Measurement for Martial Arts (P43) ............................ 231 J.I. Cowie, J.A. Flint, A.R. Harland
A Comparative Study of Ball Launch Measurement Systems; Soccer Case Study (P44) ................................................................................................................ 239 Jouni Ronkainen, Chris Holmes, Andy Harland, Roy Jones
Testing Protocol for Quantitative Comparison of Top of the Range Soccer Boots (P45) ........................................................................................................... 247 Jouni Ronkainen, Dan Toon, Joe Santry, Tom Waller
Development of a Measurement-Prosthesis for a Ski Boot Test Bench (P48)..................................................................................................................................... 255 M. Reichel, A. Haumer, H. Schretter, A. Sabo
Development of Multi-platform Instrumented Force Pedals for Track Cycling (P49) ................................................................................................................. 263 Jean-Marc Drouet, Yvan Champoux, Sylvain Dorel
In-Situ Measurement of Clipless Cycling Pedal Floating Angles (P51) ................................................................................................................................................ 273 Yvan Champoux, Daniel Paré, Jean-Marc Drouet, Denis Rancourt
Correlation Between Treadmill Acceleration, Plantar Pressure, and Ground Reaction Force During Running (P52) .................................. 281 Alex, J. Y. Lee, Jia-Hao Chou, Ying-Fang Liu, Wei-Hsiu Lin, Tzyy-Yuang Shiang
Development of Immediate Feedback Software for Optimising Glide Performance and Time of Initiating Post-Glide Actions (P56) .............................................................................................................. 291 Roozbeh Naemi, Serdar Aritan, Simon Goodwill, Steve Haake, Ross Sanders
Rod Response Analysis to Fish Bite Based on Multi-link Model Solved by Lower Triangularization of Sparse Symmetric Coefficient-matrix (57) .................................................................................................................. 301 Shigeyuki Yamabe, Hiromitsu Kumamoto, Shingo Nishioka
X
The Engineering of Sport 7 - Vol. 1
Design and Manufacture of Customised Orthotics for Sporting Applications (P62) ............................................................ 309 Paul Crabtree, Vimal Dhokia, Martin Ansell, Stephen Newman
Analysis of Snowboard Stiffness and Camber Properties for Different Riding Styles (P65) .................................................................................................. 319 Aleksandar Subic, Patrick Clifton, Jordi Beneyto-Ferre, Arnaud LeFlohic, Yoshiki Sato, Victor Pichon
The Fluctuating Flight Trajectory of a Non-Spinning Punted Ball in Rugby (P67) .......................................................................................................................................... 329 Kazuya Seo, Osamu Kobayashi, Masahide Murakami
Aerodynamics of Bicycle Helmets (P68) ................................................................... 337 Firoz Alam, Aleksandar Subic, Aliakbar Akbarzadeh
Aerodynamics of Cricket Ball-an Effect of Seams (P70) ....................... 345 Firoz Alam, Roger La Brooy, Aleksandar Subic, Simon Watkins
Numerical Modelling of the Flow Around Rowing Oar Blades (P71) .................................................................................................................................... 353 Anna Coppel, Trevor Gardner, Nicholas Caplan, David Hargreaves
The Acute Response to a Garment-based Elastic Thoracic Load, Applied During Exercise on Inspiratory Muscle Strength and Pulmonary Function (P72) ....................................................................................................... 363 Ashley R. Gray, Dr Tom M. Waller, Prof Mike P. Caine
Aerodynamic Performance of Cycling Time Trial Helmets (P76)............................................................................................................................. 371 Kim B. Blair, Ph.D., Stephanie Sidelko
Physical Motion Analysis of Nordic Walking (P77) .................................... 379 Takayuki Koizumi, Nobutaka Tsujiuchi, Masaki Takeda, Yusuke Murodate
Driving Performance Variability Among Elite Golfers (P79).......... 387 Ian C. Kenny, Eric S. Wallace, Steve R. Otto
Power Measurement in Cycling using inductive Coupling of Energy and Data (P80) ............................................................................................................ 397 Reinhardt Tielert†, Norbert Wehn, Thomas Jaitner, Roland Volk
Online-Monitoring of Multiple Track Cyclists During Training and Competition (P81) ....................................................................................... 405 Thomas Kuhn, Thomas Jaitner, Reinhard Gotzhein
A Model Predictive Controller for Sensor-based Training Optimization of a Cyclist Group (P82) ...................................................................... 413 Ankang Le, Lothar Litz, Thomas Jaitner
The Engineering of Sport 7 - Vol. 1
XI
A Dynamic Heart Rate Prediction Model for Training Optimization in Cycling (P83) ............................................................................................. 425 Ankang Le, Thomas Jaitner, Frank Tobias, Lothar Litz
Stability Training and Measurement System for Sportsperson (P84) ................................................................................................................... 435 S.N. Omkar, D.K. Ganesh
SRM Torque Analysis of Standing Starts in Track Cycling (P85) 443 Paul Barratt
Aerodynamic Study of Ski Jumping Flight Based on High-Speed Video Image (P86) ............................................................................................ 449 Masahide Murakami, Nobuyuki Hirai, Kazuya Seo, Yuji Ohgi
The Role of Materials and Construction on Hockey Ball Performance (P88)............................................................................................................................... 457 Dan Ranga, James Cornish, Martin Strangwood
Shape Optimization of Golf Clubface using Finite Element Impact Models (P90)......................................................................................................................... 465 Willem Petersen, John McPhee
An Examination of Cricket Bat Performance (P92) ................................... 475 Lloyd Smith, Harsimranjeet Singh
Forces Applied on Rowing Ergometer Concept2®: a Kinetic Approach for Development (P94) ......................................................... 483 Nicolas Découfour, Franck Barbier, Philippe Pudlo, Philippe Gorce
JUMPICUS – Computer Simulation in Ski Jumping (P95)................ 491 Heike Hermsdorf, Falk Hildebrand, Norman Hofmann, Sören Müller
Kinematic Response to Variations in Natural Turf During Running (P96)........................................................................................................................................... 499 Stiles, V. H., Dixon, S.D., Guisasola, I.N., James, I.T
Finite Element Simulation of Ice Pick Torquing (P97) ........................... 509 Rae S. Gordon, Kathryn L. Franklin
A Sociological Analysis of a Controversy in French Sport Science Field: How to Manage Teams Specialising in Technological Innovation (P99).................................................................................................................................... 519 Philippe Terral, Cécile Collinet
Biomechanical Ingredients Measurement: A New Vision-Based Approach (P102) .................................................................................................................................... 529 Mohammad Reza Mohammadi, Hadi Sadoghi Yazdi
XII The Engineering of Sport 7 - Vol. 1 How Optimal Baseball Swings Change for Three Levels of Play (P103) ............................................................................................................................................ 539 Ann Chase, Mont Hubbard, Chris Ray
Graduated Compression Stockings and Delayed Onset Muscle Soreness (P105) ....................................................................................................................................... 547 Stéphane Perrey, Aurélien Bringard, Sébastien Racinais, Kostia Puchaux, Nicolas Belluye
A Study of Knuckling Effect of Soccer Ball (P106) ....................................... 555 Takeshi Asai, Kazuya Seo, Yousuke Sakurai, Shinichiro Ito, Sekiya Koike, Masahide Murakami
Ball and Racket Movements Recorded at the 2006 Wimbledon Qualifying Tournament (P109) ........................................................................................... 563 Simon B Choppin, Simon Goodwill, Steve Haake, Stuart Miller
Ball Spin Generation at the 2007 Wimbledon Qualifying Tournament (P110) ............................................................................................................................ 571 John Kelley, Simon Goodwill, Jamie Capel-Davies, Steve Haake
Analysis and Optimization of the Sliding Properties of Luge Steel Blades on Ice (P111) ........................................................................................................... 579 Mathieu Fauve, Hansueli Rhyner
Brake Induced Vibration in Mountain Bikes (P112) ................................. 587 Robin C. Redfield
Aerodynamic Optimization and Energy Saving of Cycling Postures for International Elite Level Cyclists (P114) ............................. 597 Luca Oggiano, Stig Leirdal, Lars Sætran, Gertjan Ettema
A Comparison of Test Methodologies to Enable the Improved Understanding of Soccer Boot Traction (P115) ............................................... 605 J.D. Clarke, M.J. Carré, R.F. Kirk
How to Build an Optimized Movement Analysis Laboratory for High Performance Athletes of Various Sport Disciplines (P116) ................................................................................................................................ 613 Lars Janshen
Analysis of the Wobble of a Spinning Disc at Launch (P117) ......... 623 William Rae, Mont Hubbard
A Study of the Influence of the Environmental Condition and the Garment in Skin Temperature in Sport Activity (P119) ............ 631 Natividad Martínez, David Rosa, Javier Gámez, Juan Carlos González, Carlos Chirivella, José María Gutiérrez, Jaime Prat, José Javier Sánchez
The Engineering of Sport 7 - Vol. 1
XIII
Compression Sleeves Significantly Counteracts Muscular Fatigue During Strenuous Arm Exercise (P124) ............................................. 641 Thibaud Thedon, Nicolas Belluye, Stéphane Perrey
Development of a New System for Measuring Tennis Court Pace (P126) ................................................................................................................................ 649 Simon Goodwill, Steve Haake, James Spurr, Jamie Capel-Davies
A Feedback System for Coordination Training in Double Rowing (P127) .......................................................................................................................................... 659 Arnold Baca, Philipp Kornfeind
Modelling the Oblique Impact of Golf Balls (P128) ................................... 669 James Cornish, Steve Otto, Martin Strangwood
Modelling and Stability Analysis of a Recumbent Bicycle with Oscillating Leg Masses (P131) ............................................................................................... 677 Brendan Connors, Mont Hubbard
Computerised Games for Balance Training: A Pilot Study on Collegiate Females (P135) ................................................................................................. 687 Jonathan S. Wheat, Ben Heller, Stephanie Lovick
Effects of Turbo-jav Release Conditions on Distance of Javelic Throw (P136) ............................................................................................................................................. 697 M. Maeda
Differences Between Leather and Sybthetic NBA Basketballs (P137) ................................................................................................................................ 705 Hiroki Okubo, Mont Hubbard
Subject Index .............................................................................................................................................. 713
Effects of Body Weight on Ski Jumping Performances under the New FIS Rules (P3) Luca Oggiano1, Lars Sætran2
Topics: Aerodynamics. Abstract: Based on the results of several different experiments, it has been concluded that the weight of a ski jumper is crucial in performing a long ski jump. In response to this conclusion, many of the best ski jumpers in the world began dieting to reduce their weight, resulting in many underweight athletes and some incidents of anorexia. In order to deal with this problem the International Ski Federation (FIS) introduced a new rule where the ski length is determined by both the jumper’s height and weight. An athlete with a Body Mass Index (BMI) of less than 20 must reduce the length of his or her skis. To evaluate the effect of the new rules a numerical and experimental investigation on the effects of the BMI on ski jumper's performances has been done. A numerical model has been built in order to evaluate the effects of BMI on the final speed in the in-run path. The numerical results obtained from the model match experimental data present in the literature. Experiments in the wind tunnel have been made in order to evaluate the aerodynamic forces acting on the ski jumper and on the skis during the flight path according to the new FIS rules. Experiments have been carried out on a doll mounted on a 6 components balance and different positions and ski length have been tested. The data acquired have been introduced into a numerical model and the final jump length has been then estimated. In conclusions it has been found out that the current FIS rules do reduce the problem addressed but experiments shows that it is still more advantageous to lose weight and consequently cut the skis, compared to gaining weight in order to keep the full ski length. Keywords: Aerodynamic, Ski jumping, Drag, Lift.
1. Norwegian University of Science and Technology, Faculty of Engineering Science and Technology, N-7491 Trondheim, Norway - E-mail:
[email protected] 2. Norwegian University of Science and Technology, Faculty of Engineering Science and Technology, N-7491 Trondheim, Norway - E-mail:
[email protected] 2
The Engineering of Sport 7 - Vol. 1
1- Introduction Ski Jumping is a sport discipline which involves different engineering fields. In this paper we will focus on the effect that BMI (and then the body weight) has on ski jumper’s performances and especially on the jump length. The increase the BMI has 2 main effects on ski jumpers performances. It has a positive effect during the in-run (higher weight gives a higher speed at take-off) and it has a negative effect during the flight path (the higher the weight is, the shorter the jump is. The 2 effects are not balanced. The negative effect during the flight is much stronger than the positive one during the in run. The final effect of BMI increasing is a shorter jump length. [SM1]. Because of this advantage of being light in terms of jump length, athletes began to lose weight, and many cases of underweight and some of anorexia athletica [S] came up. This alarming trend forced FIS to create new rules in order to reduce the problem with underweight ski jumpers. Under the old rules, an athlete's ski length was determined by the athlete’s height only but, in 2004, the rules changed, and today the ski length is determined by both height and BMI. Under the new rules, an athlete with a BMI of less than 20 must reduce the length of his skis according to a table made by FIS. Any athlete who has a BMI below 17.5 is not allowed to participate in the competitions. The rules change required underweight ski jumpers to use shorter skis than the jumpers’ height had formerly allowed. The intention was to reduce the positive lift forces acting on a ski jumper and his equipment during the flight, hence reducing the positive effect of being light. When the new rules became operative, there was a general belief within the ski jumping community that it would be beneficial for the athletes to gain weight by building up their thigh muscles. The weight gained would cause an increase in BMI and the athletes could then keep their original ski length and, theoretically, increase the power generated at the jump. The trend that the athletes followed it has not been the same that FIS expected. The average BMI among all the athletes present in the Olympic Games in Turin 2006 has been 19.41, 0.5 less than the average BMI measured in 2000 [SM1]. In order to determine whether the changes to the rules are justified, experiments have been conducted in a wind tunnel with 1:1 model of skis and ski jumper and a numerical model for In-run and Flying-path has been made and used to compute the final jump length. Different positions and angles for both jumper and skis have been analyzed and tested. Experimental data from previously studies and wind tunnel experiments [SM2] have been used in order to determine the position of skis and body angle. An adaptive numerical model has been built trying to describe as close as possible the real flight path.
The Engineering of Sport 7 - Vol. 1
3
1.1 BMI The BMI is a relation between a person’s weight and height. It’s defined as: (1)
Table 1 - BMI compared to weight status.
From results regarding the decrease of ski jumpers BMI acquired from 1973 to 2000 and shown by Vaverka [V] and Müller [SM1] it has been noticed that the average BMI among ski jumpers dropped from a value above 23 in the past to a value under 20 in 2000 (see Fig. 1). The last data about 106 ski jumpers who participated to the Olympic Games in Turin 2006 give an average BMI of 19.41 (ca. 0.5 less than what was measured in 2000). This shows that the trend of losing weight for obtaining better performances has not been stopped with the introduction of the new FIS rules.
Figure 1 - BMI trend in ski jumping competitions during the last 40 years.
1.2 Forces acting on a ski jumper Several forces act on a ski jumper during the three phases of the jump (in-run, take off, flight and landing). The combination of these forces will decide whether the ski jump is successful or not.
4
The Engineering of Sport 7 - Vol. 1
Roughly dividing the jump into 2 parts, the in-run and the flight, the main forces acting during these parts are drag, gravity, ski-snow friction during the in-run and drag, lift and gravity during the flight.
2- Experimental Setup For the experiments, the wind tunnel of NTNU (Norwegian University of Science and Technology) in Trondheim has been used. The test section of the wind tunnel is 12.5 meters long, 1.8 m high, and 2,7 m wide. The wind tunnel is equipped with a 220KW fan that can produce a variation of speed between 0.5 - 30 m/s.The balance (Carl Schenck AG) used is a six component balance capable to measure the three forces and the three momentums around the three axes. Variations of forces and moments are measured using strain gauges glued to the balance body. The voltage outputs are measured by a LABVIEW based PC program.
3- Numerical simulation of the in-run In order to evaluate the effect of BMI on the take off speed a numerical model has been built. The mathematical model here presented includes friction forces between ski and snow, mass forces due to gravity and aerodynamic drag forces acting on the ski jumper. The simulation has been divided in three parts, following the different shapes of the in run path. The path (see Fig. 2) is divided in three parts one straight part with a constant heeling-angle 1 (no curvature), one curved part with a constant radius of curvature r and then another straight part with constant heeling angle 2.
Figure 2 - In-run path. The in-run path is divided into 3 different parts: 2 straight parts and a curved one between the 2 straight ones.
The results obtained applying the model show that the BMI does not have a huge influence on the speed at take off. The difference between the take-off speed calculated for an athlete with a BMI around 30 (weight 85kg and height 170cm) and the speed calculated for an athlete with a BMI around 15 (weight 45kg and height 170cm) is about
The Engineering of Sport 7 - Vol. 1
5
1m/s (see Fig. 3). Considering that most of the ski-jumpers have a BMI between 17 and 21, this difference is reduced and it has been calculated to be about 0.2 m/s.
Figure 3 - Effect of increase of BMI in the speed at take off calculated for the Granåsen jumping hill in Trondheim (Norway). The curve which reaches the lower speed has been calculated for a speed skater with BMI 16 while the curve which shows the higher speed has been simulated for a ski jumper with BMI 25. L1=50m, L2=6.8m, r=110m, 1=34.5˚, 2=11˚, TOTlength=101.9m.
4- Experimental investigation on the aerodynamic forces 4.1 Skis and doll position in the wind tunnel
Figure 3 - Different angles between ski jumper and wind direction. is the angle between wind direction and skis, is the angle between wind direction and ski jumper’s body and is the angle between wind direction and horizon line.
6
The Engineering of Sport 7 - Vol. 1
Skis and the doll have been tested separately since they can be considered as two separate systems. The ski’s wake does not in fact interact with the ski jumper’s body. This means that the 2 aerodynamic parameters (Lift and Drag) can be measured separately. They were both connected to a shielded support connected to the 6 component balance. The doll used for the test is 170cm tall and his position has been varied from 10degrees to 60degrees. The suit used is the same suit used by the Norwegian skijumpers during the Olympic Winter Games in 2006. The correct angles were adjusted using the joints on the support. Trying to model a ski jump in the most realistic way possible, the angles were adjusted in relation to the flying path, wind direction and the tilt of a skier’s ankle. In the experiment, 3 different velocity levels have been used:13 m/s, 20 m/s and 27 m/s, respectively. The ski-length is about 268cm and it has been tested at 6 angles relatives to 6 different flight positions. In order to evaluate the effect of the new FIS rules, the skis got cut in the back end for 5cm at a time and the same test has been done until a ski length of 248cm has been reached.
4.2 The flight path In order to evaluate the effects of the new FIS rules on the aerodynamic forces, data regarding the positions assumed by a ski jumper during his flying path were needed. These data have been acquired by Schmölzer, & Müller [SM2]. The simplified flight path used for the model here presented has been divided in five different parts, assuming aero dynamical forces to be constant in each part. Forces during the flight path have been decomposed in vertical and horizontal forces [R]. - Take off 1 (t2.5 ton, would be advantageous in increasing b to a greater depth. The horizontal component of movement is important as this is thought to cause shearing of roots in the soil, leading to the adverse phenomenon of 'root breaks' where a sheared layer between 25 and 50 mm allows dense horizontal root growth, creating a 'spring' in the cricket pitch profile. This technique will be used to explore this hypothesis further, and future studies will include vertical reinforcement from live grass roots,
The Engineering of Sport 7 - Vol. 1
113
which is expected to reduce this horizontal component. Future work will also investigate moisture content and roller diameter variables.
4- Conclusions The instrumented rolling rig provides a tool for the development of steel smoothwheeled rollers for cricket pitch preparation in a controlled environment. The resultant data, which in this case show that pressure at 80 mm is negligible, even with relatively heavy rollers is significant in understanding the role of rolling in pitch consolidation. To increase density throughout the whole profile requires effective drying by the plant – it cannot be achieved by rolling alone. The horizontal component of deformation is considered significant and should be minimised in the rolling of cricket pitches. Further investigation will combine this data with field data in the form of a model to deliver optimised parameters for cricket pitch rolling and preparation.
5- References [AG1] Adams, W. A. & Gibbs, R. J. Natural Turf for Sport and Amenity. CAB International, UK, 1994. [BH1] Baker, S.W., Hammond, L.K.F., Owen, A.G. and Adams, W.A. Soil physical properties of first class cricket pitches in England and Wales. 1. Classification system for soil characteristics. In Journal of Turfgrass and Sports Surface Science, 79: 2-11, 2003. [JC1] James, D.M., Carre, M.J. and Haake, S.J. The playing performance of county cricket pitches. In Sports Engineering, 7: 1-14, 2004. [JD1] James I; Dixon S; Blackburn K; Pettican N. The Measurement of Applied Pressure at Depth with Two Natural Soil Surfaces at Different Densities. In: The Engineering of Sport 6, Volume 2: Developments for Disciplines. E F Moritz & S Haake (Eds.), 2006, Springer, NY, USA. ISBN-10: 0387-34678-3, p 173-178, 2006 [SJ1] Shipton, P.M.R., James, I.T. and Vickers, A. The mechanical behaviour of cricket soils during preparation by rolling. In: The Engineering of Sport 6, Volume 1: Developments for Sports. E F Moritz & S Haake (Eds.), Springer, NY, USA: 229-234, 2006.
Experimental Validation of a Finite-element Model of a Tennis Racket String-bed (P21) Tom Allen, Simon Goodwill, Steve Haake1
Topics: Tennis & other Rackets Sports, Modelling. Abstract: An explicit finite-element (FE) model of a tennis racket string-bed was produced in Ansys/LS-DYNA 10.0. This model was used to simulate a range of impacts between a tennis ball and string-bed, which were validated against experimental data. The laboratory validation was undertaken by firing balls, with backspin in the range from 0 to 600 rad·s-1, from a pitching machine onto a head-clamped tennis racket. Inbound velocities and angles in the range from 20 to 30 m·s-1 and approximately 20 to 60° respectively were tested. Results were obtained for rebound spin, angle and velocity, with good agreement between the model and experiment. Keywords: Tennis, Finite Element Analysis, String-bed, Spin.
1- Introduction The tennis ball-rigid surface impact has been modelled successfully using the finiteelement (FE) technique (GK1, AG1). A number of authors have attempted to include a string-bed in an FE model of a tennis racket with varying success (WM1, WM2, KN1). Kanda et al. (2002) produced an FE model of a tennis ball impacting perpendicular to a freely suspended strung tennis racket. The ball was modelled as a pressurised rubber core, with linear material properties. In addition the felt cover was not incorporated into the model even though it has been found to influence rebound spin for impacts on a rigid surface (GK1); there was also no reference of the ball being independently validated. A number of impact locations were simulated and the coefficient of restitution (COR) was found to decrease with increasing string tension and to be highest between the geometric centre of the string-bed and the throat, in agreement with other authors (BC1, GH1). However, as with previous publications (WM1, WM2), the strings were assumed to be fixed at their intercepts, effectively ignoring the effect of string to string friction. This is clearly not a realistic representation of reality and subsequent errors would become apparent if simulating an oblique impact. Sports Engineering Research Group, Sheffield Hallam University, UK - E-mail:
[email protected]; s.r.groodwill,
[email protected] 116 The Engineering of Sport 7 - Vol. 1 Cross (2000) produced a mathematical model of a ball impacting obliquely on a string-bed at 10 m·s-1, to analyse the effect of the sliding friction between the ball and the strings. Cross experimentally obtained sliding and rolling friction coefficients for 5 different strings, which were 0.27-0.42 and 0.05, respectively. However, he calculated sliding friction by placing a 10 kg mass on a ball and dragging it across the string-bed at a constant velocity, a method not representative of a typical high speed collision. The critical value of sliding friction between the ball and string-bed was found to be 0.3; below this the ball’s rebound angle and range drops significantly, which would result in a detrimental effect on the player’s performance. As well as not validating the model against experimental data, Cross made a number of assumptions and simplifications. His results were based on the assumption that the ball impacts at the centre of the stringbed, which is not representative of a shot during play (CG1). He didn’t consider the effects of friction between the strings and their movements. However, currently there is no published data which relates to the method of obtaining string to string fiction. Goodwill and Haake (2004) experimentally analysed the impact of an oblique spinning ball on a head-clamped racket. They tested inbound velocities of 23 and 31 m·s-1 at an angle of 39° to the normal, with backspin in the range from 0 to 420 rad·s-1. It was concluded that a rigid body mathematical model under-predicted the rebound spin of the ball. Furthermore, the experimentally measured spin was found to be higher than that associated with rolling. A deformable ball mathematical model was produced and it was concluded that the ball starts to over-spin at the mid point of the impact, resulting in the friction force reversing direction, which hence produces an increase in the horizontal velocity. This reversal of the friction force acting on the ball was in agreement with the findings of numerous authors for oblique impacts on a rigid surface (C2, HC1, HC2). However, Goodwill and Haake’s (2004) model did not have the capacity to calculate the rebound spin of the ball. Further testing including higher spin rates and a range of angles is required to gain insight into the ball’s characteristics when impacting with a string-bed. In this investigation an FE model of a tennis racket string-bed will be validated against experimental data for nominal impact angles of 20°, 40° and 60°, relative to the racket normal. The ball’s rebound velocity, angle and spin will be compared against inbound spin, for a set angle and velocity. For a set material and gauge (diameter), string-bed stiffness will be determined by string tension, which has a typical range of 220-310 N (BC1). The agreement of the FE model with the experimental data for rebound spin will provide an initial indication as to whether the ball-string friction is in the correct range.
2- FE model An explicit FE model of a tennis racket string-bed, consisting of 16 main and 19 cross strings, was created in Ansys/LS-Dyna 10.0 (Fig. 1). The overall dimensions of the stringbed were 0.331 0.253 m, with the strings having a diameter of 1.32 10-3 m, MAT_ELASTIC was selected as the linear material model for the strings. The method for obtaining the dynamic stiffness of a tennis string, using a hammer strike, was deve-
The Engineering of Sport 7 - Vol. 1
117
loped by Cross et al. (2000); in this FE model, the Young’s modulus of 7.2 GN·m-2 was obtained using KL/A, where K is the dynamic stiffness of the string and L and A are the length and cross sectional area of the string, prior to the hammer strike. The value of K used (30822 N·m-1) was taken from Lindsey (2006), which was a value consistent with the particular nylon strings used in this model. The density of 1100 kg·m-3 was provided by the manufacturer and the Poisson’s ratio was assumed to be 0.3. SOLID164 3-D 8 node bricks, with single point integration or constant stress, were used to mesh the main and cross strings, which consisted of 19,624 and 17,712 elements, respectively. This particular value of mesh density was selected as it produced elements of a similar size to those in the ball and hence prevented contact instabilities. Solid elements were selected, above 1-dimensional elements, as they were a better representation of reality; they have the ability to accurately represent the shape and structure of the string-bed, in particular the 3D deformation characteristics of individual strings.
Figure 1 - String-bed model.
Contact between the ball and string-bed was defined as CONTACT_SURFACE_TO_SURFACE with a friction coefficient of 0.4 (C1). The same technique was used for string to string contact, except the friction coefficient was set to 0.1. A rigid cylinder 1 10-3 m in length and 1.32 10-3 m in diameter was attached to both ends of every string. This FE model consists of two phases: the dynamic relaxation phase, in which the static pre-loads are applied to structures (string tension and internal ball pressure), which occurs prior to the transient stage of the model, in which the ball is set in motion. The end of one phase and the beginning of the next is triggered when the displacements of the structures, following their loads, start to “settle” to within their convergence tolerances, which is a value that denotes how closely the displacements must reach the value they are converging to. A load of 150 N was applied to each of the rigid cylinders, in the required direction during the dynamic relaxation phase, to produce a total tension on every string of 300 N. These rigid volumes were then fixed in position during the transient phase of the simulation, effectively resulting in a headclamped racket. The convergence tolerance for dynamic relaxation was 0.01. Applying constraints to rigid cylinders as opposed to directly to the ends of the nylon strings, prevented element distortion, hence resulting in a more stable model. A pressurised rubber core with a felt cover was used to simulate the ball; the material models were MAT_OGDEN_RUBBER and MAT_LOW_DENSITY_FOAM, respectively. Details of the ball model, including the material properties and its validation, can
118 The Engineering of Sport 7 - Vol. 1 be found in Allen et al. (2007). The ball’s initial velocity and spin were assigned using INITIAL_VELOCITY_GENERATION. Due to the ball having three different inbound angles, the string-bed was translated in the horizontal direction, prior to the simulation, in order to ensure the ball contacted at the correct string-bed location. This was achieved using a piece of software, specially developed in MS Visual Basic 2005, which modifies the DYNA text file by selecting the nodes corresponding to the string-bed and updating their locations. Translating the ball, rather than the string-bed, to allow for the correct contact positions, would make calculating spin difficult, thus this method enabled the ball to be kept in the same location, with its initial position at the intercept of the x, y and z axes.
3- Method Tennis balls were projected against a head-clamped racket (Fig. 2) using a modified pitching machine device. An aluminium tube was fitted to the pitching machine to provide greater consistency for the balls inbound angle. A carbon fibre tennis racket with a head size of 0.063 m2 was used for all tests. Two groups of four rackets were used in the investigation, strung at 200 N and 289 N, respectively. To ensure consistent and accurate results the string-bed deflection of the rackets was measured directly before and after testing using a Babolat RDC, the mean of these 2 values are quoted in this paper as opposed to the stringing tension. Ball inbound angles and velocities in the range from 20 to 60° and 20 to 30 m·s-1 were analysed. The inbound angle was adjusted by tilting the racket as opposed to adjusting the cannon and flight path of the ball. Around 20 impacts were undertaken for every racket at each angle; the inbound backspin was varied from 0 to 600 rad·s-1.
Figure 2 - Experimental setup.
The flight of the balls was recorded as a series of bitmap files, using a Phantom V4.1 high-speed video camera, positioned 8 m from the racket in the direction of its longitudinal axis and recording at 1000 fps. Inbound and rebound velocities, angles and spins were measured from the recorded images using Richimas (bespoke image analysis software). The mean inbound angles were found to have deviated slightly from those
The Engineering of Sport 7 - Vol. 1
119
predicted from the experimental set-up; the calculated values are shown in table 1. For simplicity the inbound conditions will be referred to by their nominal values. The mean horizontal distances from the ball’s impact location to the geometric centre of the stringbed, were calculated for each set of impacts. This was achieved by estimating each ball’s impact position relative to the string-bed centre, from its initial location upon exiting the pitching machine and the calculated mean angle corresponding to the experimental set-up (Table 1). FE simulations were undertaken with inbound velocities, angles and impact locations identical to those in the laboratory experiment. For each angle and velocity pair, simulations were undertaken with backspin ranging from 0 to 600 rad·s-1, at 200 rad·s-1 increments. Table 1 - Inbound angles, velocities and impact locations relative to the centre of the string-bed.
4- Results Figure 3a-f shows that the model results for rebound velocity are in good agreement with the experimental data. Although the model marginally under-estimated the rebound velocity of the ball at 20° and 20 m·s-1, for a backspin of 200 rad·s-1 (Fig. 3a). The model also appears to slightly over-calculate rebound velocity at 60° and 30 m·s-1 (Fig. 3f). Overall the results from both the model and experiment show that rebound velocity decreases as the inbound angle, relative to the racket normal, increases. Rebound velocity can also be seen to decrease with increasing inbound backspin; with the decrease becoming more pronounced as the inbound angle increases. Close inspection of figures 3e and 3f show that the rate of decrease in rebound velocity drops significantly for inbound backspins greater than around 400 rad·s-1. This non-linear relationship of the data appears to be evident in both the model and experiment.
120 The Engineering of Sport 7 - Vol. 1
Figure 3 - Rebound velocity against inbound spin at a) 20°, 20 m/s, b) 20°, 25 m/s, c) 40°, 20 m/s, d) 40°, 30 m/s, e) 60°, 20 m/s, f) 60°, 30 m/s
Figure 4 shows that the model results for rebound angle are in good agreement with the experimental data. However, the rebound angles for the 20° simulations are slightly higher than the experimental values (Fig. 4a and b). The general trend is that the rebound angle of the ball increases with the inbound angle. The results also show that rebound angle decreases as inbound backspin increases. However, figure 5e and f show that the reduction in rebound angle with increasing inbound backspin appears to become less pronounced in the experimental data, for backspins greater than approximately 350 rad·s-1. This non-linearity, which is in agreement with the results for rebound velocity, can be observed in the FE model with a nominal inbound angle of 60° at 20 m·s-1 (fig 3e) but not at 30 m·s-1 (fig 3f).
Figure 4 - Rebound angle against inbound spin at a) 20°, 20 m/s, b) 20°, 25 m/s, c) 40°, 20 m/s, d) 40°, 30 m/s, e) 60 °, 20 m/s, f) 60°, 30 m/s.
The Engineering of Sport 7 - Vol. 1
121
Figure 5 shows that all the model results for rebound spin are in good agreement with the experimental data. The results show that rebound spin decreases with increasing backspin, whilst it increases with the inbound angle. Again a non-linearity can be observed, for both model and experiment, in the rebound characteristics of the ball for inbound backspins above approximately 350 rad·s-1 at a nominal inbound angle of 60° (Fig. 5e and f).
Figure 5 - Rebound spin against inbound spin at a) 20°, 20 m/s, b) 20°, 25 m/s, c) 40°, 20 m/s, d) 40°, 30 m/s, e) 60 °, 20 m/s, f) 60°, 30 m/s.
Figure 6a shows that the horizontal force acting on the ball, comprising of friction and a string-bed horizontal reaction force, switches direction just after the midpoint (2.85 ms) of the impact. The initial horizontal force is negative which means that the force is acting in the opposite direction to the ball motion. At a time of approximately 2.85 ms the horizontal force becomes positive, implying that it is in the same direction as the ball motion. This causes an increase and decrease in the horizontal and angular velocities of the ball, respectively (Fig.6b and c). The horizontal force switches direction again at around 4.2 ms, resulting in a very slight decrease in the horizontal velocity and an increase in spin. As the horizontal force acting on the ball switches direction during impact then there will be an instance at which the coefficient of friction between the ball and strings will equal zero. This illustrates that any model that assumes a simple, linear relationship between a friction and vertical reaction force is invalid.
122 The Engineering of Sport 7 - Vol. 1
Figure 6 - Inbound velocity 30 m·s-1, angle 40° and backspin 200 m·s-1 a) Vertical and horizontal force, b) Horizontal velocity and c) Spin.
5- Discusssion The FE model has been found to be in good agreement with the experimental data for rebound velocity, angle and spin, albeit a few marginal discrepancies. It was found that the horizontal force acting on the ball, switches direction at approximately the midpoint of the impact when the inbound velocity, angle and backspin were 30 m·s-1, 40° and 200 rad·s-1, respectively. This implies that the ball spin rate has exceeded that associated with rolling and is over-spinning, as found by Goodwill and Haake (2004). The fact that the friction force becomes negative again indicates a tennis ball impacting obliquely to a string-bed converges towards a rolling state. Further research would be required to quantify this hypothesis. This investigation has shown that the current FE model is capable of replicating a single impact location on a string-bed for different inbound velocities, angles and spins. This model has also allowed realistically-strung strings to move independently, providing a more realistic representation of string to string friction. However, further research should be undertaken in order to determine a more precise value for both ball, and string, to string friction. In reality, a ball will impact at a variety of locations on a stringbed during play (CG1). Therefore, to ensure that the model accurately represents reality it must be validated for different impact positions on the string-bed. These impact positions should be extremes in the longitudinal and lateral directions, to allow the largest possible area of the string-bed to be validated. This would also allow the string-bed to be encompassed into a frame to create an accurate FE model of a racket, which could be used to simulate all the typical impacts encountered during a game of tennis. This model could be used by manufacturers to evaluate the performance of different frame and string constructions, to complement their existing experimental testing.
6- Conclusion An FE model of a tennis racket string-bed has been produced and successfully validated against experimental data for different impacts. These impacts had inbound backspins in the range from 0 to 600 rad·s-1, with nominal velocities and angles from 20 to 30 m·s-1
The Engineering of Sport 7 - Vol. 1
123
and 20 to 60°, respectively. The experimental validation was undertaken using a pitching machine to project balls onto a head-clamped tennis racket. It was found that a ball will enter into an over-spinning stage during the impact, except when there is a combination of a high inbound angle (60°) and backspin (>400 rad·s-1). Extending the validation to include different impact positions on the string-bed would increase its applicability. This would also allow for a realistic FE model of a tennis ball-racket impact to be produced.
7- Acknowledgements The authors would like to thank the International Tennis Federation (ITF), for the use of their testing facilities. They would also like to thank Prince for sponsorship of the project.
8- References [AG1] Allen, T., Goodwill, S. & Haake, S. Experimental validation of a tennis ball finite-element model. Tennis Science and Technology 3, 1, 21-30, 2007. [BC1] Brody, H., Cross, R. & Lindsey, C. The Physics and Technology of Tennis. 1st edn. Racquet Tech Publishing, 2002. [CG1] Choppin, S., Goodwill, S., Haake, S. & Miller, S. 3D player testing results from the Wimbledon Qualifing Tournament. Tennis Science and Technology 3, 1, 341-348, 2007. [C1] Cross, R. Effects of friction between the ball and strings in tennis. Sports Engineering, 3, 8597, 2000. [C2] Cross, R. Grip-slip behavior of a bouncing ball. American Journal of Physics, 70, 1093-1102, 2002. [CL1] Cross, R., Lindsey, C. & Andruczyk, D. Laboratory testing of tennis strings. Sports Engineering, 3, 219-230, 2000. [L1] Lindsey, C. String Selection Map 2006. Racket Tech, 1, 22-28, 2006. [GH1] Goodwill, S. and Haake, S. Modelling of an impact between a tennis ball and racket. Tennis Science and Technology 2, 1, 79-86, 2003. [GH2] Goodwill, S.R. & Haake, S.J. Ball spin generation for oblique impacts with a tennis racket. Experimental Mechanics, 44, 195-206, 2004. [GK1] Goodwill, S.R., Kirk, R. & Haake, S.J. Experimental and finite element analysis of a tennis ball impact on a rigid surface. Sports engineering (Sheffield, England), 8, 145-158, 2005. [HC1] Haake, S., Carre, M. & Goodwill, S. Modelling of oblique impacts of tennis ball impacts on tennis surfaces. Tennis Science and Technology 2, 1, 133-137, 2003. [HC2] Haake, S.J., Carre, M.J., Kirk, R. & Goodwill, S.R. Oblique impact of thick walled pressurized spheres as used in tennis. Proceedings of the Institution of Mechanical Engineers C, Journal of Mechanical Engineering Science, 219, 1179-1189, 2005. [KN1] Kanda, Y., Nagao, H. & Naruo, T. Estimation of tennis racket power using three-dimensional finite element analysis. 218, 2002. [WM1] Widing, M. & Moeinzadeh, M. Finite Element Modeling of a Tennis Racket with Variable String Patterns and Tensions. International Journal of Sport Biomechanics, 6, 78-91, 1990. [WM2] Widing, M. & Moeinzadeh, M. Nonlinear finite element analysis of a frame stiffened with tension members. Computers and structures, 33, 233-240, 1989.
Experimental Validation of a Tennis Ball Finite-element Model for Different Temperatures (P22) Tom Allen, Simon Goodwill, Steve Haake1
Topics: Tennis & other Rackets Sports, Modelling. Abstract: An explicit finite-element (FE) model of a pressurised tennis ball was produced in Ansys/LS-DYNA 10.0 and validated at room temperature. This model was successfully updated to simulate temperatures of 283.15 and 313.15 K (10 and 40 ºC), by adjusting the internal pressure and material properties of the ball’s rubber core. The validation experiment was undertaken using an impact rig in a climate chamber, for perpendicular impacts on a rigid surface with inbound velocities in the range from 15 to 30 m·s-1. The impact rig consisted of an air-cannon, for firing the balls, a set of light gates for calculating coefficient of restitution (COR), and a force plate for measuring contact time. The model was found to be in good agreement with the experimental data across the entire range of temperatures tested. Keywords: Tennis, Finite Element Analysis, Ball, Temperature.
1- Introduction A number of authors have successfully modeled ball impacts using finite-element (FE) technique (CS1, HS1, GK1, AG1, Goodwill et al. (2005) created and validated an FE model of a pressurized tennis ball and pressurized rubber core, for impacts with a rigid surface at a range of velocities. Allen et al. (2007) modified this model to simulate a tennis ball from a different manufacturer, whilst extending the validation process to include pressurized and punctured, cores and balls. This was to independently validate the three separate parts of the model; the felt cover, rubber core and internal pressure. However, both of these studies only resulted in the production of FE models, with the capacity to simulate impact characteristics of a tennis ball at room temperature. Rose and Coe (2000) experimentally tested the effect of temperature on the static stiffness and COR of tennis balls for perpendicular impacts up to 45 m·s-1. Within the range of 273.15 to 313.15 K, they found static stiffness to effectively remain constant, whilst COR increased with temperature. Downing (2007a) experimentally measured the Sports Engineering Research Group, Sheffield Hallam University, UK - Email:
[email protected]; s.r.groodwill,
[email protected] 126 The Engineering of Sport 7 - Vol. 1 effect of temperature, in the range of 283.15 to 313.15 K using a climate chamber, on COR and contact times for perpendicular impacts. Downing used an impact rig, consisting of an air-cannon to fire the balls, a set of light gates to measure velocity and a force plate to record contact times, testing inbound speeds in the range from 15 to 30 m•s-1. COR was found to increase with temperature, which agreed with Rose and Coe (2000). Contact times were also found to increase with temperature, indicating a reduction in the ball structural stiffness (BC1, CL1, GK1), which was concluded to be caused by a change in the material properties of the rubber. However, neither of these experimental investigations provided an insight into how the individual changes in internal pressure and rubber properties with temperature, could effect the rebound characteristics of a tennis ball. As rubber increases in temperature, its stiffness decreases. However, it is currently not known to what extent the material properties of a tennis ball core could change in the range from a typical, to an extreme, playing temperature. In addition to the change in rubber properties, the internal pressure of a tennis ball increases with temperature. It is possible to accurately calculate the change in internal pressure with temperature, using Gay-Lussac’s law by assuming the volume to remain constant. Therefore, it is feasible to produce an FE model of a tennis ball at different temperatures, by updating the internal pressure and then adjusting the material properties, until the rebound characteristics are in good agreement with the experimental data. Downing (2007b) examined the effect of temperature on surface pace rating (SPR) for an acrylic and synthetic carpet surface. SPR was found to decrease with temperature, indicating an increase in the coefficient of friction (COF) between the ball and court. Although, due to only two surfaces being tested this investigation does not provide an overall picture of how the SPR of tennis courts changes with temperature. A reliable FE model, which simulates an oblique impact between a tennis ball and different surfaces, could be used to determine how SPR changes with temperature. The first stage of producing such a model would be the creation and validation of a tennis ball for different temperatures. This paper aims to create and validate explicit FE models of pressurized tennis balls, for temperatures of 283.15 and 313.15 K. An initial investigation will be undertaken on the effect of only modifying the internal pressure of the model, to simulate the two temperatures. Following this, a separate study will be conducted on independently adjusting the static rubber material properties. Finally, the internal pressure and material properties, both static and dynamic, of the tennis ball in the FE model will be updated to simulate the different temperatures. The FE models will all be validated against the experimental data from Downing (2007a).
2- FE model The initial FE model of a tennis ball, used in this investigation, was identical to the one produced by Allen et al. (2007). The temperature, at which this original FE model’s material testing or validation was undertaken, was not recorded. For the purpose of this investigation both of these temperatures were considered to have been 295.15 K (22 ºC); assuming this to be a realistic estimation of room temperature. In the current FE model,
The Engineering of Sport 7 - Vol. 1
127
the internal pressure is simulated as an airbag. The relationship between the pressure (P) and volume (V) during impact was assumed to be adiabatic and defined by PV1.4 equal to a constant. This adiabatic assumption, for which there is no heat transfer between the air enclosed by the ball and its surrounding, was based on the high rate at which the volume changes during impact, combined with the insulating properties of the rubber and felt. During the experiment, the balls were acclimatised to each temperature before testing (refer to D1 for details); hence there was sufficient time for heat transfer between the enclosed volume of air and the surroundings. Therefore, in the modified models the initial pressure was calculated at each temperature (T) by PV/T, which is equal to a constant. For simplicity the volume was assumed to remain constant. As with the original model the relationship between the pressure and volume during impact was assumed to be adiabatic. Figure 1a shows the relationship between the internal pressure and relative volume of the ball, for temperatures of 283.15, 295.15 and 313.15 K. The relative volume of the ball is defined as the actual volume divided by the original volume. The ball was simulated in the model using a pressurised rubber core with a felt cover; the material models were MAT_OGDEN_RUBBER and MAT_LOW_DENSITY_FOAM, respectively. Details of the ball model, including the material properties and its validation, can be found in Allen et al. (2007). The effect of altering the static stiffness of the rubber core was achieved by adjusting the uniaxial data in the MAT_OGDEN_RUBBER material model (Fig. 1b). To produce models, which simulated the full effects of adjusting temperature, both the internal pressure (Fig. 1a) and static and dynamic material properties of the rubber were modified. The static modulus of the rubber was altered by ±10 % for the models at 283.15 and 313.15 K, respectively (Fig. 1b). The dynamic modulus of the rubber in the model was increased by 75 % to simulate a temperature of 283.15 K and it was decreased by 20 % for 313.15 K. The apparatus required to obtain the material properties of the rubber at different temperatures was not available to the authors; therefore, the stated changes in the material properties of the rubber were used as they were in good agreement with the experimental data, in terms of contact times and COR, obtained by Downing (2007a), refer to figures 6 and 7 for more details. The properties of the ball, in both the model and the experiment, were measured at each temperature in terms of contact time and COR. The material properties of the felt were not modified with temperature, as any changes were assumed to have an insignificant effect on the rebound characteristics of the ball.
Figure 1 - a) Tennis ball internal pressure against relative volume, for temperatures in the range from 283.15 to 313.15 K, b) Static rubber core material properties.
128 The Engineering of Sport 7 - Vol. 1
3- Method The FE models were validated by comparing COR and contact times, for perpendicular impacts with a rigid surface, with the experimental results published by Downing (2007a). Details of this experimental procedure can be found in Downing (2007a), whilst the original validation of the FE model can be found in Allen et al. (2007). The first stage of this investigation was to analyse the effect of only adjusting the internal pressure of the model for temperatures of 283.15 and 313.15 K. Following this, an analysis was undertaken to identify the effect of increasing and decreasing the static stiffness of the rubber by 10 %, whilst keeping the original internal pressure. Finally, the internal pressure and static and dynamic material properties of the rubber were all updated to simulate the two temperatures under investigation.
4- Results The results from Downing (2007a) at 298.15 K were initially compared to the FE model and experimental data published by Allen et al. (2007). This was necessary due to different balls being used in each experiment, in addition to discrepancies in the test temperatures. Figure 2 shows that COR and contact times were found to be in good agreement between the experimental data from Downing (2007), Allen et al. (2007) and the original FE model. However, contact times were marginally higher at inbound velocities below 20 m·s-1 for the validation data and the FE model in comparison to Downing (2007a).
Figure 2 - a) COR and b) contact time, comparison between the experimental data from Allen et al. (2007) (295.15 K) and Downing (2007) (298.15 K) and the original FE model.
Figure 3 shows that when only the internal pressure was adjusted, the model overpredicted COR when the temperature was 283.15 K. At 313.15 K the model is in good agreement with the experimental data, although it marginally under-predicted COR for inbound velocities below 20m·s-1. The FE model and experimental data both show increasing COR with temperature.
The Engineering of Sport 7 - Vol. 1
129
Figure 3 - COR for adjusted internal pressure at temperatures of a) 283.15 K and b) 313.15 K.
Figure 4b shows that that when only the internal pressure was adjusted, the FE model was in relatively good agreement with the experimental data at both temperatures, for contact time. However, the experimental data shows increasing contact time with temperature, whilst the FE model has the opposite trend.
Figure 4 - Contact time for adjusted internal pressure at temperatures of a) 283.15 K and b) 313.15 K.
Figure 5a shows that increasing the static stiffness of the rubber in the FE model by 10 % results in a marginal increase in the COR (dynamic stiffness kept constant). The opposite was the case for a 10 % reduction in static stiffness. Figure 5b shows that reducing the static material stiffness by 10 % results in a significant increase in contact time. Again, the opposite was the case for a 10 % increase in stiffness. The range in both COR and contact time between both FE models (20 % change in rubber material stiffness) is approximately equal to the range of scatter in the experimental data, for a set inbound velocity.
130 The Engineering of Sport 7 - Vol. 1
Figure 5 - Effect of the static rubber material stiffness on a) COR and b) contact time.
Figure 6 shows that the FE model, with the updated internal pressure and static and dynamic rubber material properties, is in good agreement with the experiment for COR at 283.15 K (static and dynamic rubber modulus 10 and 75 % higher) and 313.15 K (static and dynamic rubber modulus 10 and 20 % lower). The model and experiment both show increasing COR with temperature.
Figure 6 - Complete model - COR a) 283.15 K and b) 313.15 K.
Figure 7 shows that the complete model is also in good agreement with the experimental data for contact time at both temperatures. Both the model and experiment show increasing contact times with temperature.
Figure 7 - Complete model - Contact time a) 283.15 K and b) 313.15 K.
The Engineering of Sport 7 - Vol. 1
131
Figure 8 shows that the maximum impact force is marginally higher at 283.15 K in comparison to 313.15 K for all the inbound velocities under investigation. It can also be observed that the maximum force occurs earlier into the impact for the model at 283.15 K for inbound velocities of 15 and 20 m·s-1. At inbound velocities of 25 and 30 m·s-1 the maximum impact force at 283.15 K lags that of the model at 313.15 K.
Figure 8 - Complete model force plot a) 15 m·s-1, b) 20 m·s-1, c) 25 m·s-1, b) 30 m·s-1
5- Discussion Increasing the internal pressure of a tennis ball, in isolation, raises its structural stiffness, thus reducing its contact time. Hence, when only the internal pressure of the tennis ball was adjusted, contact times were found to decrease with temperature. This was in contradiction to the experimental data, where contact times increased with temperature. When the effect of changing the material properties with temperature is added into the model, contact time and COR are in good agreement with the experimental data. Therefore, in agreement with Downing (2007a), it is concluded that the change in a ball’s material properties with temperature, have a greater influence on its rebounds characteristics than the alteration in internal pressure. In this paper it was found that a greater change in the dynamic material properties of the rubber was required to simulate a temperature of 283.15 K (75 % increase), in comparison to 313.15 K (20 % decrease). This was also found by Downing (2007a), who demonstrated that for normal impacts there is a greater difference in both COR and contact times between 25 and 10 ºC, in comparison to 25 and 40 ºC.
132 The Engineering of Sport 7 - Vol. 1 Independently adjusting the static material properties of the rubber resulted in an increase in COR and a reduction in contact time, with increasing stiffness. A tennis ball’s static structural stiffness is predicted to be affected by temperature in two ways; i) the material and hence structural stiffness of the ball is reduced with increasing temperature and ii) the lower static stiffness of the rubber at higher temperatures results in the ball expanding more from the applied internal pressure, which in turn increases its volume lowering its initial internal pressure. The diameter of the ball in the FE model was found to increase from 0.066004 to 0.066204 m when the simulated temperature was increased from 283.15 to 313.15 K. This equated to a 0.62 % increase in the frontal area or drag force acting on the ball during flight. Further research is required to determine whether temperature has a significant effect on tennis ball’s cross sectional area and hence flight characteristics. The results show that when just the internal pressure of the ball was updated to simulate a temperature of 283.15 K, the model over-predicted COR. It was also found that increasing the static stiffness of the rubber, to simulate a drop in temperature, resulted in an increase in the COR, making the model over-predict the COR even more. Therefore, if only the internal pressure and static stiffness of the rubber were adjusted the model would over-predict COR at 283.15 K, whilst the opposite would be the case at 313.15 K. An increase in damping results in a decrease in COR (DH1); hence including the change in the dynamic material properties of the rubber resulted in strong agreement with the experimental data. This investigation has provided an indication as to how the static and dynamic material properties of a tennis ball rubber core change with temperature. However, the intention of this study was only to provide a gauge of the extent to which the material properties of a tennis ball change with temperature. In this investigation the static material properties were assumed to adjust with temperature, whilst following the trend of the original data. In reality, it is very unlikely that this would be case, especially in the range from 298.15 to 283.15 K which experienced the largest change in rebound characteristics. In addition, more precise material testing would be required to determine how the static properties of rubber cores change with temperature. Indeed, a more accurate methodology needs to be developed in order to experimentally obtain both the static and dynamic material properties of the ball at different temperatures. The change in felt material properties with temperature may also have an influence on the rebound characteristics of a tennis ball. If the felt were to become less stiff with increasing temperature, then the ball would stretch more from the internal pressure. This would, in turn, increase its initial volume, reducing its internal pressure and hence structural stiffness. However, it is predicted that the stiffness of the felt would change by a very marginal amount within the temperature range used in this investigation. In-depth material testing would be required to quantify how the felt properties change with temperature.
The Engineering of Sport 7 - Vol. 1
133
6- Conclusion An FE model of a tennis ball, validated at room temperature, has been updated to include simulations at temperatures of 283.15 and 313.15 K for inbound velocities in the range from 15 to 30 m·s-1. This was achieved by modifying the internal pressure in accordance with the laws of thermodynamics, whilst simultaneously estimating the change in the rubber core material properties. The model was found to be in good agreement with the experimental data for the entire range of velocities under investigation, at both temperatures. Overall, the change in rubber properties with temperature was found to have a more significant effect on the rebound characteristics of a tennis ball than the change in internal pressure (Downing 2007a). In-depth material testing would be required to determine precisely how the rubber core and felt cover properties change with temperature.
7- Acknowledgements The authors would like to thank the International Tennis Federation (ITF), in particular Matt Downing, for providing the experimental data. They would also like to thank Prince for sponsorship of the project.
8- References [AG1] Allen, T., Goodwill, S. & Haake, S. Experimental validation of a tennis ball finite-element model. Tennis Science and Technology 3, 1, 21-30, 2007. [BC1] Brody, H., Cross, R. & Lindsey, C. The Physics and Technology of Tennis. 1st edn. Racquet Tech Publishing, 2002. [CS1] Calder, C. & Sandmeyer, B. Modelling The Bat-ball Impact Using Finite Element Analysis. 158-159, 1997. [CL1] Cross, R. & Lindsey, C. Technical tennis. 1st edn. Racquet Tech Publishing, 2005. [DH1] Dignall, R.J. & Haake, S.J.. Analytical Modelling of the Impact of Tennis Balls on Court Surfaces. In Tennis Science and Technology. (edited by S.J.Haake and A.O. Coe), First edn. pp. 155162. Blackwell science Ltd., 2000. [D1] Downing, M. The effects of climate changes on the properties of tennis balls. Tennis Science and Technology 3, 1, 49-55, 2007a. [D2] Downing, M. The effect of temperature on the court pace rating of tennis surfaces. Tennis Science and Technology 3, 1, 81-86, 2007b. [GK1] Goodwill, S.R., Kirk, R. & Haake, S.J. Experimental and finite element analysis of a tennis ball impact on a rigid surface. Sports engineering (Sheffield, England), 8, 145-158, 2005. [HS1] Hubbard, M. & Stronge, W.,J. Bounce of hollow balls on flat surfaces. Sports Engineering, 4, 49-61, 2001. [RC1] Rose, P. & Coe, A. The variation of static and dynamic tennis ball properties with temperature. Tennis Science and Technology, 1, 169-174, 2000.
Nonlinear Dynamics of a Simplified Skateboard Model (P24) Alexander S. Kuleshov1
Topics: Skate & other Urban Sports; Extreme Sports; Modelling. Abstract: In this paper the further investigation and development for the simplified mathematical model of a skateboard with a rider are obtained. This model was first proposed by Mont Hubbard (Hubbard 1979, Hubbard 1980). It is supposed that there is no rider’s control of the skateboard motion. To derive equations of motion of the skateboard the Gibbs-Appell method is used. The problem of integrability of the obtained equations is studied and their stability analysis is fulfilled. The effect of varying vehicle parameters on dynamics and stability of its motion is examined. Keywords: Skateboard; Nonholonomic Constraints; Integrability; Stability of Motion.
1- Introduction Skateboarding is one of the most popular extreme sports of today. In 2003 skateboarding had over 11 million participants in the U.S. alone (Frederick et al. 2006), putting the sport on a par with tennis and volleyball in terms of participant levels. However, despite of the growing number of participants skateboarding is poorly represented in the scientific literature. At the late 70th – early 80th of the last century Mont Hubbard (Hubbard 1979, Hubbard 1980) proposed the two mathematical models describing motions of a skateboard in the presence of a rider. To derive equations of motion he used the principal theorems of dynamics. In this paper the further development of the simplest skateboard model proposed by Hubbard is given.
Figure 1- The Skateboard Side View. 1. Department of Mechanics and Mathematics, Moscow State University, Main building of MSU, Leninskie Gory, Moscow, 119991, Russia - E-mail:
[email protected] 136 The Engineering of Sport 7 - Vol. 1 The skateboard typically consists of a board, two trucks and four wheels (Figure 1). The modern boards are usually from 78 to 83cm long, 17 to 21cm wide and 1 to 2cm thick. The most essential elements of a skateboard are the trucks, connecting the axles to the board. Angular motion of both the front and rear axles is constrained to be about their respective nonhorizontal pivot axes, thus causing a steering angle of the wheels whenever the axles are not parallel to the plane of the board. The vehicle is steered by making use of this kinematic relationship between steering angles and tilt of the board. In addition, there is a torsional spring, which exerts the restoring torque between the wheelset and the board proportional to the tilt of the board with respect to the wheelset. We denote the stiffness of this spring by k1.
2- Formulation of the Problem. Equations of Motion. The simplest skateboard model assumes that the rider, modeled as a rigid body, remains perpendicular with respect to the board. Therefore, when the board tilts through γ, the rider tilts through the same angle relative to the vertical (Figure 2). Let us introduce an inertial coordinate system OXYZ in the ground plane. Let FR = a is the distance between two axle centers F and R of a skateboard. The position of the line FR with respect to the OXYZ - system is defined by X and Y coordinates of its centre and by the angle θ between this line and the OX - axis (Figure 3). The tilt of the board is accompanied by rotation of the front wheels clockwise through δf and rotation of the rear wheels anticlockwise through δr. The wheels of the skateboard are assumed to roll without lateral sliding. This condition is modeled by the constraints, which are nonholonomic as can be proved:
Figure 2 - The Skateboard Rear and Top View.
(1)
The Engineering of Sport 7 - Vol. 1
137
. . Equations (1) can be solved with respect to X and Y: (2)
Thus the velocities of the points F and R will be directed horizontally and perpendicularly to the axles of the wheels and there is a point P on the line FR which has zero lateral velocity and hence only forward speed u. It may be shown, that (3) Using the results obtained in (Hubbard 1979, Osterling 2004) it is easy to find the following relations between the tilt of the board and the steering angles f and r (4) where f and r are the fixed angles which the front and rear axes make with the horizontal. . . According to (4) equations (2) for x and y can be rewritten as follows: (5)
Figure 3 - The Basic Coordinate Systems.
138 The Engineering of Sport 7 - Vol. 1 Expressions (3) become (6)
Suppose that the board of the skateboard is located a distance h above the line FR. The length of the board is also equal to a. The board center of mass is located at its center. The rider’s center of mass is not be located above the board center of mass, but it is located over the central line of the board a distance d from the front truck. Let l be the height of the rider’s center of mass above the point P. Other parameters for the problem are: mb is the mass of the board; mr is the mass of the rider; Ibx, Iby, Ibz are the principal central moments of inertia of the board; Irx, Iry, Irz are the principal central moments of inertia of the rider. The following parameters will be used below:
It can be proved (Osterling 2004, Kuleshov 2007) that the variables u and satisfy the following differential equations (7)
Here A, …, E, K are the functions of the parameters, namely
Thus, equations (5)-(7) form the close system of equations of the skateboard motion. Moreover equations (7) admit the energy integral
The Engineering of Sport 7 - Vol. 1
139
Equations (7) have the particular solution (8) which corresponds to uniform straight-line motion of a skateboard. Consider the problem of stability of this particular solution. Setting u = u0 + and keeping for its notation we write the equations of the perturbed motion (9)
. Here and are functions depending on , and whose development as the series in powers of said variables starts with terms of at least the second order. Moreover, . these functions identically vanish with respect to when = 0 and = 0:
The characteristic equation corresponding to the linearized system (9) has the form (10) When conditions (11) are fulfilled equation (10) has one zero-root and two roots in the left-half plane. . Since the functions and identically vanish for = 0, = 0, then under conditions (11) the critical case of one zero-root takes place and solution (8) is stable with respect . . to , and u (asymptotically stable with respect to and ). Since a condition is valid for all values of parameters then stability conditions (11) can be finally written in the form (12)
(13) If at least one of the conditions (12)-(13) is not fulfilled then equations (9) has the root in the right-half plane and solution (8) will be unstable.
140 The Engineering of Sport 7 - Vol. 1 One can make now simple conclusions about stability of a straight-line motion of the skateboard using conditions (12)-(13). Note, first of all, that the expression on the left-hand side of inequality (12) contains u0 as a multiplier. This means that the stability of motion depends on its direction. If one direction of motion is stable the opposite direction is necessary unstable. Such behavior is peculiar to many nonholonomic systems. First of all this effect exists the classical problem of motion of a rattleback (aka wobblestone or celtic stone (Bondi 1986, Garcia and Hubbard 1988)). In this problem the stability of permanent rotations of a rattleback also depends on the direction of rotation. Suppose that u0 > 0, f = r = and condition (13) is valid. In this case the stability of motion (8) depends on the location of the rider. If the rider stands closer to the front truck d < a/2, the motion will be stable and when the rider stands closer to the rear truck d > a/2 it will be unstable. When u0 = 0, the skateboard is in the equilibrium position on the plane. In this case the characteristic equation (10) has one zero-root and two pure imaginary roots under the condition (14) It can be proved (Hubbard 1979, Osterling 2004, Kuleshov 2007) that condition (14) is necessary and sufficient condition for stability of the skateboard equilibrium position.
3- Integrable case The characteristic equation (10) can have one zero-root and two pure imaginary roots also in the case when B = 0. In particular, this condition is valid when the skateboard is symmetric f = r and the rider stands in the center of the board d = a/2. One can see that in this case equations (5)-(7) can be completely solved in terms of quadratures. Indeed, in the case B = 0 equations (7) will have a form (15) with the energy integral (16) By introducing a new independent variable in the first of equations (15) this equation can be transformed to the form (17) Equation (17) is differential equation with separate variables. Its solution gives the variable u as a function of : u = u(). Substituting this function into the energy integral
The Engineering of Sport 7 - Vol. 1
141
. (16) and solving it with respect to one obtains the following differential equation for : (18) Equation (18) is also differential equation with separate variables. The solution of this equation gives the angle as a function of time: = (t). By substituting this function in the expression for u = u() this expression can be rewritten as follows: u = u((t)) = u(t). Having the functions = (t) and u = u(t) it is easy to find all the remaining variables as functions of time. Indeed the angle satisfies the following differential equation
and therefore (19) The coordinates X and Y of the board center of mass satisfy the following differential equations
Since the functions u = u(t), = (t), = (t) are already known, it is easy to obtain by integration (20)
Thus, in the case B = 0 all unknown functions in the problem can be expressed as functions of time by formulas (17)-(20).
4- Conclusions In this paper the problem of motion of a skateboard with a rider was examined. This problem has many common features with other problems of nonholonomic dynamics. In particular it was shown that the stability of motion of the skateboard depends on the direction of motion. The similar effects have been found earlier in the classical problem of a rattleback dynamics. It was found also the integrable case in the problem. Note that the integrable problems are very rare in nonholonomic mechanics and therefore these results seem to be interesting and helpful.
142 The Engineering of Sport 7 - Vol. 1
5- Acknowledgements This research was supported financially by the Russian Foundation for Basic Research (grant 08-01-00363).
6- References [B1] Bondi H. The rigid body dynamics of unidirectional spin. In Proceedings of the Royal Society of London. Series A, 405: 265-274, 1986. [FD1] Frederick E.C., Determan J.J., Whittlesey S.N. & Hamill J. Biomechanics of skateboarding: Kinetics of the Ollie. In Journal of Applied Biomechanics, 22: 33-40, 2006 [GH1] Garcia A. and Hubbard M. Spin reversal of the rattleback: theory and experiment. In Proceedings of the Royal Society of London. Series A, 418: 165-197, 1988. [H1] Hubbard M. Lateral Dynamics and Stability of the Skateboard. In ASME Journal of Applied Mechanics, 46: 931-936, 1979. [H2] Hubbard M. Human Control of the Skateboard . In Journal of Biomechanics, 13 : 745-754, 1980. [K1] Kuleshov A.S. Mathematical Model of a Skateboard with One Degree of Freedom. In Doklady Physics, 52(5): 283-286, 2007. [O1] Osterling A.E. MAS 3030. On the skateboard, kinematics and dynamics. School of Mathematical Sciences. University of Exeter, United Kingdom. 2004. http://www.longboard.nu/files/theSkateboard.pdf
Cricket Batting Stroke Timing of a Batsman When Facing a Bowler and a Bowling Machine (P26) Alex Cork1, Laura Justham1, Andrew West1
Topics: Lawn Sports (Hockey, Cricket). Abstract: Cricket batsmen must evaluate each delivery and select a shot to play from temporal and spatial information gained from the movements and anatomical position of the bowler (pre-release cues), initial ball flight and their own previous experience. Current batting training methods often make use of bowling machines; however these machines offer the batsman no pre-release visual cues pertaining to ball type. Previous research has suggested that highly skilled batsmen are superior to less skilled players in the pick-up of pre ball flight information and that the bowling arm provides referential information regarding ball type. A pilot study has been conducted to establish the different approaches adopted by a batsman when faced with a human bowler and a bowling machine. The movements and reaction times of an amateur (English premier club level) batsman are compared for both scenarios. Front and side on high speed video footage was recorded of the batsman when facing a random selection of deliveries from a human bowler and a bowling machine. Front on high speed footage of the bowler/bowling machine was simultaneously recorded allowing the responses of the batsman to be analysed when facing a human bowler (with visual cues) and a bowling machine (without visual cues). Keywords: Cricket Batting, Bowling Machine, Movement.
1- Introduction An express paced delivery (90mph and higher) as classed by Abernethy (1981) takes, approximately, 450msecs to travel from the bowler’s hand to striking the bat. Previous experimentation has shown that combined choice reaction time and movement time for a choice of just four possible strokes would equal approximately 700msecs, so, effective batsmen must use some form of anticipation when facing even moderately fast bowling (Gibson and Adams, 1989). The ability to accurately predict the length, direction and pace of the delivery from the movements of the bowler prior to delivery is therefore a 1. Sports Technology Research Group, Sports Technology Institute, Loughborough Science and Enterprise Park, 1 Oakwood Drive, Loughborough, Leicestershire, LE11 3QF, UK - Email: A.E.J.Cork, L.Justham,
[email protected] 144 The Engineering of Sport 7 - Vol. 1 skill that will greatly enhance a batsman’s ability. This skill becomes increasingly important when facing express paced bowlers where the time constraints imposed by ball velocity will typically exceed the time available to process information or high class spin bowlers who are able to disguise the changes in their action that lead to subtly different deliveries. The movements made by batsmen in cricket are the result of feedback gained from the movements of the bowler and the ball in the air during the delivery. Information about an individual delivery arising from the movements and anatomical positions of the bowler will be added to prior knowledge the batsman has about the pace and style of the bowler, the pitch conditions and their own form, culminating in a decision being made as to which stroke they elect to play. Using existing bowling machines coaches are unable to present the batsman with a realistic match environment in with to train.“Coaches operating bowling machines typically provide some cue as to the time at which they place the ball in the machine” (Gibson and Adams, 1989) usually by raising their hand prior to dropping a ball into the machine. This clearly does not provide batsmen any pre-release information pertaining to ball type, neither does it offer batsmen the opportunity to look for such information. Some bowling machines operate with warning lights on the front of them; this is usually in the form of a “traffic-light style” countdown to release. However in a match or when facing a bowler in the nets a batsman must recognis the moment of release during the delivery stride, there are no warning lights or raised hands to help. Further to this the direction and length of the delivery can be ascertained from viewing the angle of the head of the bowling machine as recognised by Gibson and Adams (1989). This is clearly not a match realistic pre release cue and could cause deficiencies in technique to develop in batsmen. Abernethy has deduced that “for skilled players, information to help predict the length of a bowled ball is available prior to release” (Abernethy et al., 2005). Thus if a batsman is using the angle of the bowling machine head to determine the line and length of delivery, when presented with a match situation he is poorly prepared to face deliveries as he has little exposure to the pre release cues emanating from a bowler’s action. In cricket expert batsmen have shown a persistent capability to use early sources of information to aid shot selection which other skill groups are not attuned to (Muller et al, 2006). There are three key abilities encompassed within the anticipatory skill demonstrated by expert batsmen; (i) visual search: selecting the areas the eyes will focus upon during the delivery stride and release, (ii) selective attention: picking out the key events within the bowling action that relate to ball type and finally, (iii) discrimination ability: being able to recognise the movements of the bowler and interpret them into the resultant ball type. It is therefore reasonable to assume that each of these abilities is aided by experience. Perhaps the area that is most enhanced by experience is discrimination ability where players are provided with a wider store of potentially relevant memories and a rapid and automatic access to these (Schneider and Fisk, 1983). In conclusion; does training with bowling machines add to this store of relevant memories and aid batsman in match scenarios when attempting to evaluate an approaching delivery from the movements of the opposing bowler?
The Engineering of Sport 7 - Vol. 1
145
The pilot study described in this paper focuses upon the movement patterns of a premier league club cricketer when facing a human bowler and a typical bowling machine. The aim of this study was to see whether there were differences in the timing of movements of the batsman when faced with the two different scenarios. While there is no doubting the benefit of using bowling machines to teach and practice the technique of particular shots, they do not provide batsmen with relevant information about approaching deliveries, neither do they add to the memory store of previously faced deliveries available to batsmen. From the results of this study it is possible to make an initial judgement on the validity of using bowling machines to train for match scenarios. While there have been previous studies conducted in this area, notably Gibson and Adams (1989), it was felt that with highly accurate, modern equipment available such as a number of high speed video cameras offering a high image resolution, the HawkEye ball tracking system and a realistic training environment for testing to be conducted within, additional knowledge could be developed within this area of research.
2- Method Player testing was conducted at the ECB National Cricket Centre, Loughborough. An amateur batsman, who plays for a premier league club side was filmed batting against a non-familiar human bowler of the same standard and a Bola bowling machine. The movements of both the batsman and bowler/bowling machine operator were recorded from the initiation of the bowler’s run up/machine operator’s signal to the completion of the batsman’s shot. The batsman was filmed from both a front on and side on perspective and the bowler/bowling machine was filmed from a front on perspective. The cameras were synchronised and recording was initiated using an SV TTL trigger when the bowler broke a laser beam positioned at the beginning of the run up. For the deliveries bowled by the bowling machine, recording was initiated when the bowling machine operator broke a laser beam when signalling an imminent delivery to the batsman by raising his arm. A specifically designed test bed was assembled to house two high speed video cameras, one was focussed upon the batsman and the second was focussed upon the bowler, both from a front on perspective. The test bed was positioned in the centre of the designated pitch at a distance of 6 metres from the stumps at the bowler’s end. Both cameras were Photron SA1 High Speed Video Cameras. The cameras were set up to sample at 500 frames per second which enabled data to be recorded at resolution of 1024 x 1024 pixels. A third high speed camera was positioned orthogonally to the pitch, in line with the batsman’s crease on a standard Manfrotto tripod at a height of 1.2 metres, focussed upon the batsman. This camera was a Photron Ultima APX high speed video camera which also sampled at 500 frames per second at a resolution of 1024 x 1024 pixels. Additional data was recorded for each delivery using the HawkEye system installed within the indoor facility. HawkEye is a computer based ball tracking system that uses three orthogonal cameras recording at 140 frames per second to capture the trajectory of the ball before and after it bounces. Mathematical algorithms are then used to predict the complete ball trajectory. These data were used to give an overview of each delivery faced by the batsman and enabled the ball flight characteristics of the deliveries bowled to be quantified.
146 The Engineering of Sport 7 - Vol. 1
3- Procedure The batsman was allowed a short period to warm up and faced two deliveries from the human bowler prior to the commencement of recording to test the synchronisation of the equipment and to acclimatise the batsman to the style and pace of the bowler. The batsman then faced twelve recorded deliveries. He was asked to bat normally and without any irregular movements that may have been intended to affect the bowler. The bowler was asked to bowl consistently, at his normal pace, line and length. After a prolonged break testing resumed as before with a bowling machine replacing the human bowler. The batsman faced two deliveries, as with the human bowler, to check the synchronisation of equipment and to allow acclimatisation to the pace and style of the bowling machine deliveries. The average speed of the human bowler’s deliveries were determined from the HawkEye data and the bowling machine was set to replicate this. The bowling machine operator warned the batsman of an imminent delivery by raising his hand with the ball prior to inserting it into the machine; this was considered to be the most commonly used method adopted by coaches. Between deliveries, the angle of the head of the bowling machine was marginally altered in an attempt to reproduce some of the variability in length seen in human bowling, without any short balls or Yorkers. The line of the deliveries was kept consistent; on the stumps.
4- Results Each of the video cameras sampled at 500 frames per second, therefore each frame equated to 0.002 seconds. It was subsequently possible to establish the timings of events during each ball faced. Mean times were calculated for each event during the batting stroke, and the mean times for foot movements were also recorded. Figure 1 is a series of still images taken from the front and side on high speed footage of the batsman during the study at six key points identified during the batting stroke.
Figure 1 - Still images taken from the high speed video of the batsman from both front and side on at six key positions during the batting stroke (from left to right); Bat back lift, Bat held level, Movement upwards, Bat reaches top of backswing, Movement down, Bat and ball contact.
The Engineering of Sport 7 - Vol. 1
147
The mean times for each of the events during the batting stroke with respect to ball release for all of the deliveries faced from a human bowler and a bowling machine are displayed in table 1. The events are chronologically ordered. The point of release was considered to be the first frame in which the bowler no longer had contact with the ball or the moment the ball emerged from the “mouth” of the bowling machine, examples of these points can be seen in figure 2. On the occasion that there was no bat and ball contact made then the frame in which the ball was level with the batsman’s bottom hand was assumed to be the point of contact. The sixth column of table 1 gives the calculated t-values for each of the events. This value indicates whether there are significant differences when the batsman faces the bowler and bowling machine. It should be noted that against the bowling machine there was only a secondary front foot movement seen on nine of the twelve deliveries faced. When facing the bowler, the batsman had a secondary movement of the front foot on all twelve of the deliveries faced. Thus the mean and t values for the secondary front foot movement have been calculated for nine deliveries, with the corresponding delivery numbers being selected from each of the data sets for analysis. Table 1 - The mean times, in seconds, of specified events during the batting stroke with respect to ball release by a bowler or bowling machine. * Denotes a significant difference between the two sets of data. Critical t: (22df, two tailed, of 0.05) = ±2.074, (16df, two tailed, of 0.05) = ± 2.120.
From the data in table 1 it is possible to say that there are significant differences in the timings of the batsman’s bat back lift and front foot movement up and down when facing a bowler and a bowling machine. Against a bowling machine the batsman’s bat back lift was significantly earlier than against a bowler. His front foot movement, both up and down was significantly later against a bowling machine than against a bowler. Once the front foot has been planted, there is little difference in the timings of the batsman’s movements up until bat and ball contact. The discrepancy between the values for secondary movement down of the front foot can be explained by the fact that on two occasions, when facing deliveries from the human bowler, the batsman played shots with only his back leg on the ground and only planted his front foot once he had played the shot. The timings of events during the operation of the bowling machine and the delivery stride of the human bowler are displayed in table 2. These events are highlighted in figure 2. From the figures it can be said that there is more variation in the times of events for
148 The Engineering of Sport 7 - Vol. 1 the operation of the bowling machine than the delivery stride of the human bowler. The initiation of arm movement upwards by the bowling machine operator commenced, on average, 1.131 seconds before the ball was released. In contrast to this, the human bowler began the bound of the delivery stride 0.697 seconds prior to releasing the ball.
Table 2 - The mean times, in seconds, of specified events during the operation of the bowling machine and run up and delivery stride of the human bowler during testing with respect to ball release.
Figure 2 is a series of still images of both the human bowler and the bowling machine taken at key points during the delivery stride and operation of the machine respectively. The images have been arranged sequentially with respect to release. It should be noted that there are only three images for the bowling machine; this is because there was no discernable difference between the signal and release stages from the batsman’s perspective.
Figure 2 - The visual information available to the batsman from a human bowler (top) and a bowling machine (bottom) (from left to right); Top: The Bound, Rear foot impact, Front foot impact, Release. Bottom: Initiation of arm movement, Signal to batsman, Release.
The Engineering of Sport 7 - Vol. 1
149
The data displayed in table 3 are the mean values for the physical features of the deliveries bowled by both the human bowler and the bowling machine. The data is concerned with the release and pitching position of the deliveries and their release speed.
Table 3 - Mean values of features of the deliveries bowled by the human bowler and bowling machine during testing. Features are measured in metres unless stated, negative values indicate distances towards the off stump.
5- Discussion The results gained from this pilot study offer an insight into the different batting approaches employed by a batsman when facing a bowler and a bowling machine. Against a bowling machine the batsman takes longer to reach the top of his backswing and although he reaches the top at a similar time with respect to ball release, he initiated the movement considerably earlier. The timings of the machine operator’s signal and the bound of the bowler’s delivery stride are very similar. It could therefore be assumed that the batsman is taking his cue to initiate bat pick up from the earlier occurring initial arm movement of the machine operator rather than waiting for his signal. The data in table 3 confirm that there were differences between the deliveries that the batsman faced from the human bowler and the bowling machine. The human bowler released the ball from wider in the crease; he also pitched the ball wider than the bowling machine and considerably shorter. Both the bowler and the bowling machine were consistent in the line they bowled, however there was considerable variation in the length that the human bowler bowled. There was also greater variation in the position from which the ball was released. Despite this the batsman was still more consistent in the early part of the batting stroke, through to the top of the backswing, when facing the human bowler. The batsman commented after the test that he felt that he was waiting at the top of his backswing in anticipation of the ball being released by the bowling machine. This is reflected in the earlier pick up of his bat and the prolonged time it took him to reach the top of his backswing in contrast to facing a human bowler. A very clear indication of the visual differences between each scenario from a batsman’s perspective can be seen in figure 2. One of the clearest indicators of pre-release information being used is the consistently earlier movement of the batsman’s front foot when facing the human bowler. The technique observed by most batsmen is to move your front foot forward, towards the pitch of the ball if it is a full delivery, or to move your feet backwards if the delivery is short. For the batsman to be moving his feet before the ball has been released, he must
150 The Engineering of Sport 7 - Vol. 1 be interpreting information received from the movements of the bowler and predicting the length of the imminent delivery.
6- Conclusion It is clear that there is a need to conduct this testing with a larger player collection of varying standards, before it will be possible to form a clear judgement on the differences observed in players batting against a human bowler and a bowling machine. It is however possible to conclude from this initial pilot study and previous research conducted, that there appears to be a different approach adopted by batsmen when facing bowling machines to that seen against human bowlers. Further work should include the testing of players of different abilities; this may allude to the importance of visual cues to higher skilled players with the hypothesis that more experienced players are more reliant on earlier visual cues to interpret an imminent delivery. They will gain information later and therefore have to react more quickly to the bowling machine where the only information is available post release. Further to this it would be of benefit to monitor the batsmen’s eye movements from the start of the bowler’s run up to the point of bat and ball contact either through the use of an eye tracking system or a screen based test.
7- Acknowledgements The authors would like to thank the ECB for the use of the National Cricket Centre. They would also like to thank the participants in this study for their time.
8- References [A1] Abernethy, B. (1981) Mechanisms of Skill in Cricket Batting. Australian Journal of Sports Medicine 13: 3-10. [AM1] Abernethy, B. Muller, S. Farrow, D. Guy, W. Barras, N. (2005) Dealing With Natural Constraints: The Timing of Information Pick Up by Cricket Batsmen of Different Skill Levels. In Proceedings of the ISSP 11th World Congress of Sports Psychology (CD). Sydney: International Society of Sports Psychology. [GA1] Gibson, A.P. and Adams, R.D. (1989) Batting Stroke timing with a Bowler and a Bowling Machine: A Case Study. The Australian Journal of Science and Medicine in Sport 21(2): 3-6. [MA1] Muller, S. Abernethy, B. Farrow, D. (2006) How Do World Class Cricket Batsmen Anticipate a Bowler’s Intention? The Quarterly Journal of Experimental Psychology 2006, 59(12): 2162-2186. [SF1] Schneider, W. Fisk, A.D. (1982) Degree of consistent Training: Improvements in Search Performance and Automatic Process Development. Perspectives in Psychophysics, 31(2): 160-168.
Estimation of a Runner’s Speed Based on Chest-belt Integrated Inertial Sensors (P27) Rolf Vetter, Emanuel Onillon, Mattia Bertschi1
Topics: Athletics, biomechanics, fitness. Abstract: In long distance running, real time monitoring and performance optimization has been recently rendered possible through the commercialization of a large variety of running computers. Generally, such computers simultaneously monitor several parameters such as heart rate, running speed, stride frequency and stride length. More precisely, stride and speed information is obtained using foot located inertial sensors. Unfortunately, due to its leg extremity location, the foot inertial sensor presents some disadvantages: it may be perceived as cumbersome; it requires supplementary telecommunication facilities as well as local power supply; it may increase a runner’s energy expenditure even though it weights only a few tens of grams. To overcome the above-mentioned drawbacks we have developed a method for the estimation of a runners speed, stride frequency and stride length based on inertial sensors which may directly be integrated in classical chest-belt heart-rate-monitors. The method processes various kinetic features from chest located accelerometers. The instantaneous speed is then estimated through a linear mapping of the most reliable kinetic features. The determination of the most reliable kinetic features as well as the mapping coefficient is achieved through a minimal variance approach on a calibration run. A validation has been performed on 9 subjects on various running surfaces and over durations of 1 year in order to highlight the consistency of the proposed approach. We have found that different runners necessitate different optimal weighting factors of the kinetic features. This may be related to a runner’s stride efficiency, that is, more or less energy wasting and bouncing trajectories for similar forward speed. However, for a given runner we obtained consistent results namely a speed estimation accuracy of about 7% in a range of 60% to 110% of his maximal aerobic speed. Keywords: running speed; stride length; stride frequency; inertial sensor; chest-belt. 1. Swiss Center of Electronics and Microtechnology, Neuchâtel, Switzerland - E-mail: rve, eon,
[email protected] 152 The Engineering of Sport 7 - Vol. 1
1- Introduction Recent progresses in micro-electronics, biomedical signal processing, biomechanics, and telecommunications have allowed the development of various portable devices for monitoring vital parameters for outdoor activities. Whereas during the past the monitoring abilities of such devices focussed mainly on cardiovascular parameters such as heart rate, there is actually a tendency to extend their features to kinetic properties such as acceleration, velocity, movement frequency, and position of the subject. The motivations of such a tendency include subject specific personal interest, training and performance considerations of serious athletes, rehabilitation of disabled, injury prevention, and the design and analysis of footwear. In running or jogging activities, there is also a demand for monitoring heart rate, speed and step frequency in view of performance analysis and optimization [ML1]. Various portable devices for running speed estimation have been recently developed. They can be mainly categorized in the following classes with respect to the underlying technologies: inertial sensors, GPS, Doppler-microwave, and pedometers. Devices based on Doppler-microwave techniques and pedometers are not currently able to provide sufficient accuracy for jogging or running activities. GPS techniques provide high accuracies in open field use but they are not able to provide information about step frequency which is an important parameter in running performance optimization. A very attractive solution is based on inertial sensors. Polar has brought recently such a solution on the market [FR1]. Experimental evidence has shown that the estimated speed is highly accurate if a preliminary calibration is performed. At the calibration speed one obtains an accuracy of about 2%. This accuracy degrades slightly as one runs much slower or faster than the calibration speed. An inconvenient of such a device resides in the fact that it requires an accelerometer to be fixed on the running shoe [RF1]. This is somewhat cumbersome and the supplementary weight may decrease a runner’s performance. Moreover, such a solution requires a supplementary telecommunication link between the foot-located sensor and the processing device and thus increased power consumption. To avoid the need of a supplementary accelerometer device on the lower leg extremity we have developed a method based purely on an accelerometer integrated in an existing chest belt [V2].
Figure 1 - Illustration of variations in trunk inclination over a whole running cycle of two strides. The dashed lines illustrate at each phase of the cycle the inclination of the accelerometer plane located in a classical commercial chest belt. On the right side we show the two accelerometer signals a1 and a2 with respect to the dashed plane.
The Engineering of Sport 7 - Vol. 1
153
2- Methods 2.1 Developed System and data acquisition In order to develop and validate an algorithm for the estimation of a runner’s speed we have developed a portable system consisting of a 3d accelerometer integrated in a commercial chest belt and an associated wire-linked Holter. The Holter records the accelerometer data at a sampling rate of 256 Hz after anti-aliasing filtering and 12 bit analogto-numeric conversion. The recorded data were then uploaded on a PC where data processing was achieved offline.
2.2 Background The focus of the proposed approach is to exploit acceleration measured by an inertial sensor located in a chest belt to estimate a runner’s speed. Trunk located acceleration measures in experimental studies have been used to assess the transmission and attenuation of heel-strike since the early nineties [V1]. However, they were not exploited for speed estimation. In contrast, studies using force platforms have focused on various aspects of a runner’s kinetic behavior [C1,C2,M1]. Notably, it has been shown that the integral of the vertical ground reaction force over the whole stance period is related to a runner’s speed [M1]. More recently Hunter et al have highlighted the relative importance of propulsive, vertical and breaking impulse in sprinting [H1]. The different impulses have been processed using ground reaction forces obtained through force platforms. They concluded that speed estimation could be performed using the propulsive and breaking impulse with a relative importance of 57% and 7% respectively. The main differences between ground reaction forces and chest located accelerometer measurements reside in the stability of their measurement referential. Ground reaction forces are measured by force platforms which have time invariant measurement referential. In contrast, chest located accelerometer change their inclination during a whole running cycle as illustrated in Figure 2. Changes may depend on a runner’s stride efficiency, his technique, and his core stability [C1]. Therefore, chest-located accelerometers grasp a runner’s vertical or forward acceleration up to an instantaneous rotation. If the trunk inclination angle was known the instantaneous rotation could be compensated through matrix processing. However, the rotation angle is not known and we measure on the two acceleration signals a1 and a2 mixtures of a runner’s vertical and frontal acceleration. In long distance running, stride wise vertical acceleration variations are much larger than variations in frontal acceleration. Thus, rotational artifacts on accelerometer signal a1 may be neglected while this is not the case accelerometer signal a2.
2.2 Developed method The global concept of the algorithm is represented in Figure 2 and operates as follows. At each stride 3 features which may convey information about a runner’s speed are extracted from the acceleration signals. Firstly, we used a feature whose correlation with speed has been assessed during previous studies on force platforms. The feature F1 is obtained
154 The Engineering of Sport 7 - Vol. 1 through the average value over the stance phase of the accelerometer signal a1 such as to mimic the stance wise average value of the vertical ground reaction forces [M1]. Our tentative to process features from the acceleration signal a2 to mimic breaking and propulsive impulses as they have been used on force platforms [H1] was unsuccessful. As highlighted in the previous section, this may be due to rotational artifacts on the signal a2 mediated to a runner’s varying trunk inclination throughout a stride.
Figure 2 - Bloc scheme of proposed algorithmic concept for the estimation of a runner’s speed.
Secondly, we processed a feature whose relation to speed became apparent in the present study. Figure 1 illustrates this experimental evidence on the average stance stride pattern of accelerometer a1 at various running speeds. One can observe that the slope of the signal at toe-off, namely at t = 0 sec, depends on a runner’s speed. Thus, the feature F2 was processed as the minimum value of the instantaneous derivate of the accelerometer signal a1 around toe-off.
Figure 3 - Typical vertical acceleration pattern for the stance phase recorded by the proposed portable device. The pattern we show have been obtained through averaging subsequent instantaneous stride patterns over 800m run on a track at the indicated speed.
The Engineering of Sport 7 - Vol. 1
155
A third feature was processed exploiting knowledge of advanced signal processing together with kinetic considerations. Indeed, a runner’s instantaneous speed reaches its maximum during the flight phase. The distance he covers during the flight time depends mainly on his total speed and thus on his impulse at toe-off. Stride wise impulses are related to stride wise speed variability. Thus, in order to grasp the total impulse in the vectorial sense we proceeded to a stride wise principal component analysis (PCA) of variability of speed, namely the integral of accelerometer signals a1 and a2 after highpass filtering at 0.5 Hz. The instantaneous PCA achieves a rotation of the observation space [H2] and may thus compensate rotational artifacts due to varying trunk inclination of a runner. Furthermore, PCA’s rotation is performed in such a way that the first principal component yields in our case the norm of the maximum speed variation in the vectorial sense, that is, a variable related to the norm of the vectorial impulse over the stance period. Further features have been processed and tested but not retained due to their inability to represent accurately a runner’s speed. The developed algorithm processes in a subsequent step for each feature and at each stride an estimation of a runner’s speed through a linear affine mapping. It is important that the map is linear and affine such as to minimize the unknown parameters. Indeed, in a linear affine function only one parameter is unknown. Therefore, it can be estimated during one calibration run. The calibration run should typically be performed on a distance of 800m to 2000m. The average calibration speed is processed as the ratio of calibration distance to calibration run duration. For each feature we obtain the associated calibration coefficient as follows. (1)
Three speed estimates are finally obtained by multiplying the three features by their respective calibration coefficient. During the calibration run we selected the most reliable feature for a given runner through a minimum variance approach. Indeed, the variance of the stride wise estimates of the three speeds provides us information about their reliability. The feature providing speed estimates of minimum variance over the calibration run is automatically the best according to statistical considerations [H1]. Finally, to further improve the statistical reliability of the estimated speed a median filter over 40 strides was applied.
3- Validation The validation of the proposed algorithm has been achieved in two phases: in a first phase we have assessed whether features and running speed could be related through linear relationship during 800m running intervals on a track; in a second phase we focussed on the assessment of instantaneous speed estimation in off-track running. The features processed by our method were as explained in the previous section: F1 the average of the accelerometer signal a1 over the stance segment; F2 the slope of the acce-
156 The Engineering of Sport 7 - Vol. 1 lerometer signal a1 at toe-off; F3 the maximum Eigenvalue of a principal component analysis of the two dimensional stance wise speed variability obtained through integration of accelerometer signal a1 and a2.
3.1 Assessment on 800m track intervals This validation was performed on a 400m track on 17 subjects ranging form occasional joggers to elite runners. Subjects were requested to run five 800m intervals at regular speeds raging from 60% to 110% of their maximal aerobic speed. The covered speed range of the total database extended from 7 to 24km/h. The duration of each interval was recorded to evaluate the average speed. The average value of feature F1 was then processed for each interval and the accuracy of its linear relationship with the average interval speed assessed. We found that a linear relationship could provide average speed estimates with mean absolute relative error over the whole database of 3%. Subject specific results of the mean absolute relative error ranged from 1% to 5%. The mean correlation coefficient over the whole database was 0.96 with a subject specific minima and maxima of 0.94 and 0.99. This validation allows us to ascertain that the proposed method allows an estimation of an average speed using the feature F1 and a linear mapping. However, it does not provide any information about the ability of the proposed method to track variations in speed as they may be observed in long distance runners. Moreover, it is important to evaluate the invariance of the mapping function and the performance on various running surfaces.
3.2 Assessment on off-track running In order to gain further insights on the performance and to highlight the consistency of the proposed approach a second validation has been performed on nine subjects, various running surfaces and over a period of 1 year. Runners achieved during the test period 3 to 15 evaluation sessions including a calibration run of 800m and, subsequently a validation run of 20 to 60 minutes. As a reference for the calibration and the evaluation run we used a previously calibrated RS800 of Polar. Results for the runners are shown in Table 1. A brief analysis of the mean absolute relative error (MARE) between the estimations of the proposed algorithm and the RS800 points out that MARE are considerably higher than in the previous validation on the track. Notably, the MARE for feature F2 ranges from 13% to 24% whereas we obtained 1% to 5% on the track. The reason for this degradation may be related to the nature of this feature. Indeed, F2 is a derivative feature which implies that it is very sensible to noise. In the validation on the track an average value over 800m has been computed and noise related inaccuracies may thus be diminished through averaging. In contrast, in the present validation short term estimation has been achieved and therefore estimator noise may appear clearly.
The Engineering of Sport 7 - Vol. 1
157
Table 1 - Detailed speed estimation results for the nine subjects and various recordings over a duration of one year. We show mean ± standard deviation of the mean absolute relative error (MARE) and the mapping coefficients of the three features.
A further analysis highlights that best results in terms of MARE are obtained for feature F3 in 6 out of 9 subjects. For the subjects No 3, 6 and 7 the feature F1 performs better. However, even if for each subject the best feature was selected we end up with MARE with respect to the RS800 of Polar ranging from 5% to 8% (mean=7.1%, standard deviation=1.2%). The causes for this high values may be numerous. Firstly, the experimental protocol may have been too ambitious. The evaluation runs have been executed on a large variety of surfaces such as tar, track, gravel, grass and hilly forest trails. It is obvious that the different surfaces and profiles may alter a runner’s stride and thus the kinetic features of a runner’s chest [C1]. Secondly, it is questionable if the RS800 of Polar yields highly accurate estimations for such a large variety of running surfaces.
Figure 3 - An example of speed estimation during a progressive run.
Finally, an analysis of the mapping coefficient of the most reliable results, namely, c1 or c3 shows that they are subject-dependent. Values for coefficient c1 are distributed in uniform manner from 62 to 99. Coefficient c3 cover also a range from 53 to 85. However, one can observe that they are mainly grouped in two clusters: the first around 82 is put in evidence in the table through bold fonts; the second around 53 is highlighted through
158 The Engineering of Sport 7 - Vol. 1 italic fonts. The fact that different runners necessitate different mapping coefficients may be related to a runner’s stride efficiency [CO1,V2]. On one hand, stride cycles are accomplished by runners with more or less energy wasting, bouncing trajectories for similar forward speeds [C1]. The amplitudes of the bouncing trajectories are related to vertical speed changes or impulses [H1]. On the other hand, the instantaneous forward speed of a runner changes slightly during each stride cycle. It slows down at touch down during the early stance phase due to the breaking impulse and increases again during the late stance phase towards toe-off due to the propulsive impulse [H1]. Elite runners with high running efficiencies are able to come along with lower vertical, breaking and propulsive impulses at a given frontal speed than novice runners [CO1]. Features F1 and F3 are a function of stride wise speed variations in the plane a1-a2 (see Figure 1) and the mapping coefficient are therefore related to the relationship between mean forward speed and stride wise speed variations. Thus, runners with lower speed variations for a given forward speed and which may be considered as more efficient in accordance with the above considerations, have larger mapping coefficients c1 and c3. In Figure 4 we show a typical example of a speed estimation using the proposed method. The device has been calibrated in a run over 800m on the track. After a recuperation of 10 minutes the runner started off for two progressive segments of about six minutes. One can observe that the speed estimated by the proposed approach follows with small errors the speed estimated by the Polar RS800 as long as the running speed is not too slow. Notably, during the walking segments in the beginning and at time t=600 seconds, the estimation errors become considerable. A reason for this erroneous behavior of the algorithm resides in the fact that walking and running are completely different form the biomechanical point of view [N1]. In walking, potential and kinetic energy reach their maximum in phase opposition while in running both of them are in phase.
4- Conclusions We have developed a new method for the estimation of a runner’s speed based on a chest located accelerometer. The method has been validated in two different phases. Feasibility and linear mapping characteristics have been assessed on a track where average speed could be estimated with mean absolute relative errors of 1% to 5% in 17 subjects and for a large speed range. The ability of the proposed method to provide instantaneous speed estimation for a large variety of outdoor running surface was assessed on 9 subjects and over duration of 1 year. Higher mean absolute errors of about 5% to 9% have been obtained when compared with the commercial system RS800 of Polar. However, taking into account of the large variety of running surfaces and the lack of an absolutely accurate reference system for such conditions, the results are very promising. A major limitation of the proposed solution is the mandatory user specific calibration. The advantages of the proposed solution, when compared with classical foot located sensing, are manifold: The accelerometer can be integrated directly in the existing chest belt used for the pulse measurement; The power consumption is reduced since no telecommunication facilities are required for a supplementary link with the foot located sensor; Compactness and effectiveness for the runner are improved since no foot located sensor is required any more.
The Engineering of Sport 7 - Vol. 1
159
5- References [C1] P.R. Cavanagh. Biomechanics of distance running. Human kinetics books, 1990. [C2] G.A. Cavangna. Force platforms as ergometers. Journal of applied physiology, vol. 39 pages 174-179, 1985. [CO1] J. Cornuz, E. Onillon, R. Vetter. Procédé et dispositif de mesure de l’efficacité d’un geste sportif, EP1586353, 2005. [FR1] K.R. Fyfe, J.K. Rooney, K.W. Fyfe. Motion analysis system, US 2002/0040601, 2002. [H1] J.P. Hunter, R.N. Marshall, P.J. McNair. Relationships between ground reaction force impulse and kinematics of sprint-running. Journal of applied biomechanics pages 31-43, 2005. [H2] S. Haykin. Neural networks. Prentice Hall, 1994. [M2] D.I. Miller. Ground reaction forces in distance running. Chapter 8, Biomechanics of distance running, Human Kinetics Books, pages 203–224., 1990. [M3] C.F. Munro. Vertical ground reaction forces in running. Journal of biomechanics, vol. 20, pages 147-155, 1987. [ML1] M. Marquardt, C. Loeffelholz, B. Gustafsson. Die Laufbibel. Spomedis, Hamburg, Germany, 2006. [N1] T.F. Novacheck. The biomechanics of running. Gait and posture, vol. 7, pages 7795., 1998. [V1] G.A. Valinat. Transmission and attenuation of heel-strike accelerations. Chapter 9, Biomechanics of distance running, Human kinetics books, pages 225-247, 1990. [V2] Rolf Vetter. Procédé et dispositif de détermination de la vitesse d’un coureur. WO2007017471, 2007. [RF1] J.K. Rooney, K.W. Fyfe, K.R. Fyfe, W. Bortz. Shoe clip. US 2003/0000053, 2003.
Design and Construction of a Custom-made Lightweight Carbon Fibre Wheelchair (P28) Marc Siebert1
Abstract: Mobility can be equated with quality of life. For handicapped (paraplegic) people, a lightweight wheelchair means an enormous facility. Particularly with regard to independent mobility the weight of the wheelchair is very important, because the user has to lift the wheelchair very often especially when getting into his vehicle [S1]. The aim was to create an extreme light and perfect fitted wheelchair to increase the mobility of handicapped people. Therefore in a first step, dynamic and static tests have been done to learn more about the loads of a wheelchair. In a second step, an adjustable measurement wheelchair has been developed. With the measurement wheelchair it is possible to adjust the seating position separately from the ride characteristics. It is not a static process and not only the measurement of body lengths, the handicapped person can ride the wheelchair, feel the seating position and the driving behaviour and make sure for example, if the breakover point is perfect. After the fitting process in a third step the computer aided construction starts by using the data of the measurement chair to create a 3D-model of the wheelchair. The last step is the manufacturing of the wheelchair and therefore a special flexible manufacturing system has been developed. To get a high performance and extreme lightweight wheelchair, carbon fibre reinforced plastic (cfrp) has been used. A couple of different proceedings for the parts, such as filament winding, hand wrapping and compression methods have been used. For the wheelchair frame the tube-to-tube-process and special adhesive joining techniques have been used. As a result of the application of high tech materials and joining methods it was possible to realize an unique lightweight 5.9 kg carbon wheelchair for the daily use. Keywords: design, measuring wheelchair, manufacturing system, carbon fibre wheelchair.
1. Universität Kassel Mönchebergstr. 3, 34125 Kassel, Germany - E-mail:
[email protected] 162 The Engineering of Sport 7 - Vol. 1
1 - Mechanical loads The aim of this project was to develop an extreme lightweight and perfect fitted wheelchair. For the optimum dimensioning it is necessary to know as much as possible about the mechanical loads. To learn more about the load types, several wheelchair users were accompanied during their daily routine. As a first result it can be said, that the maximum (impact) load was caused by the jump off the kerb [VCRB1] (Figure 1). The height of a standard kerb according to DIN EN 1340 and DIN EN 483 is about 120 mm. To simulate the jump off the kerb a vertically adjustable stage (100-440 mm) was built. To collect the data (vertical force) a Kistler load cell was used (Figure 2). Two wheelchair users were recruited, a male and a female. Tyres, pressure of the tyres and height of the stage were changed. Before the dynamic test the static loads and the load distribution were measured (Table 1). The biggest part of the static loads was found on the rear wheels (77-93%). To determine the dynamic loads in connection with the stage height, the subjects were instructed to do five jumps from 100 mm, 200 mm and 300 mm stage height. The male subject, the winner of three gold medals at the paralypic winter games in Torino 2006, the professional monoski rider Martin Braxenthaler was able to jump off 440 mm, so that it was possible to get extreme data.
Figure 1 - Jump off the kerb.
Figure 2 - Kistler load cell.
The Engineering of Sport 7 - Vol. 1
163
Table 1 - Wheelchair characteristics and static loads.
The vertical force is about 2300-2500 N caused by the jump off the kerb. With a 200 mm stage height the vertical force is higher than 3800 N. The maximum force was collected at 440 mm stage height (5000 N). Figure 3 shows the results of the measurements.
Figure 3 - Vertical force in connection to stage height.
Further dynamic loads appear as a result of collisions between barriers and the front wheels (Figure 4). Depending on the moving direction the collisions between barriers and the front wheels involves planar stresses or torsion in the structure. Simulations of the collisions between barriers and the front wheels were realized by the fixation of different rails (20 mm and 30 mm). To determine the dynamic loads in connection with the rail height, the subjects were instructed to do five high speed collisions (straight and rotating) with 20 mm and 30 mm rail height. The maximum horizontal force (straight collision) is about 2500 N.
164 The Engineering of Sport 7 - Vol. 1
Figure 4 - Collision test.
2- Measurement and adaptation process Normally the fitting process and the selection of a wheelchair result from the measurement of body data [HLMSZW1]. The advisor measures the femoral and lower leg length and some more body data and selects the wheelchair model and size. Rarely it is possible for the customer/patient to try the selected wheelchair to check the seating position and the driving behaviour. In many cases it is impossible to change the seating position without changing the driving behaviour. The perfect wheelchair offers the optimum seating position, an adequate driving behaviour and minimum weight. To achieve the aim of the best possible seating position and driving behaviour a multi adjustable measurement/fitting wheelchair has been developed. The measuring/fitting chair was realized as a lightweight construction. To be as close as possible to the geometry and handling behaviour of the customised wheelchair, the same moulded cfrp-parts has been used. With the new measuring/fitting wheelchair (Figure 5) it is possible, to adapt and optimize the seating position separately from the driving behaviour. An enormous advantage is that the customer can feel the seating position and check it with his therapist concerning to medical requirements. The customer can also verify the driving behaviour, especially the break over characteristics.
The Engineering of Sport 7 - Vol. 1
165
Figure 5 - Measuring/fitting wheelchair.
3- Data collection and design The next step after the fitting process is the collection of the adjustment parameters. For a definite acquisition of the seating position and the chassis, several parameters like seat width, seat height, incline of the seat, height of the backrest, incline of the backrest, position of the axle (horizontal), distance between front and rear wheels, wheel size and the camber of the rear wheels are necessary. The adjustments represent the input for special parametrised three-dimensional computer model. With this computer model it is possible to get a true to scale design of the customised wheelchair. The designer is able to do an interference check (free rotatability of the front wheels without touching the foot plate or tubes). Furthermore the three-dimensional is the basis for computational analysis, such as finite element analysis.
4- Manufacturing As just noted, the determination of the optimum seating position and handling behaviour on the basis of measured body data is imprecise. The manufacturing of custom made wheelchairs is state of the art, but imprecise too. The tolerance of commercially available custom made wheelchairs is about ±1-1.5 cm. To reduce the tolerance and to facilitate an exact transfer of the measured data, a flexible manufacturing device has been developed (Figure 6). The manufacturing device consists of aluminium T-slot profiles and special fixation elements for the tubular wheelchair structure. The link-up of the three-dimensional computer model and the manufacturing device allows the exact adjustment of the fixation elements and with it the manufacturing of custom made wheelchairs with a tolerance about ±1 mm.
166 The Engineering of Sport 7 - Vol. 1
Figure 6 - Manufacturing device with three-dimensional model of the wheelchair (true to scale).
5- Ultralight wheelchair The explanations about measuring/fitting, design and the manufacturing system are independent of the used material of the custom made wheelchair. The aim of this project was to create a perfect fitted and extreme light wheelchair. We decided not to give too detailed information about dimensioning, Finite Elemente Analysys and other very technical data. The following informations should give a review about materials, manufacturing processes and joining technology. Lightweight materials like carbon fibre reinforced plastic (cfrp), aluminium and titan were used. For the frame (tubular structure) cfrp was used. To manufacture the different components, a wide range of processes were used. The formed components were manufactured via blow-moulded process, tubular components were manufactured via filament winding or wrapping processes. Planar structures like the fenders were manufactured vacuum assisted. The joining of the components was very complex. For the wheelchair frame the tube-to-tube-process and special adhesive joining techniques have been used.
The Engineering of Sport 7 - Vol. 1
167
6 - Conclusions The system, consisting of measuring/fitting chair, three-dimensional design tool and manufacturing device enables for the first time the determination of the optimum seating position, adequate handling, computer aided design and exact manufacturing. A special feature is that the system is independent form material. Exclusive application of steel, aluminium or carbon fibre reinforced plastic is as possible as multi material design to manufacture custom made wheelchairs. As a result of the application of high tech materials and joining methods it was possible to realize a unique lightweight 5.9 kg carbon fibre wheelchair for the daily use (Figure 7). With such a lightweight wheelchair it is possible to increase the mobility and reduce the physical strain of handicapped people.
Figure 7 - Carbon fibre wheelchair.
7- References [HLMSZW1] Harfich, K.-H., Lex, W., Mertsch, S., Sharma, J., Zurheide, M., Weege, R.-D. Leitfaden zur Rollstuhlversorgung, Manuelle Greifreifenrollstühle, 2. Auflage, 2001 [S1] Sonderhüsken, H.: Rollstuhltauglichkeit von Pkw’s. In the journal „Der Paraplegiker“, Heft 3/2001 [VCRB1] VanSickle, D.P., Cooper, R.A., Robertson, R.N., Boninger, M.L. Determination of wheelchair dynamic load data for use with finite element analysis. In journal of Rehabilitation Engineering, Volume: 4: 161-170, 1996
Design and Implementation of a Rugbyspecific Garment Evaluation Trial (P30) Bryan C. Roberts1, Gareth Williams1, Mike P. Caine1
Topics: Apparel and rugby. Abstract: A structured rugby-specific wearer trial to extract meaningful on-field garment performance data does not exist. To address this, a novel rugby-specific garment evaluation protocol was developed and trialled using a rugby-union international prototype shirt. Three wearer trial elements were investigated namely, rugby-specific wear-service conditions, players’ perceptions of the shirt and, the determination of garment performance on the field of play. Methods: A field test mimicking the demands of the game was devised using published international match-play time-motion analysis. Fifteen non-professional club players (age 26.1 yrs ± 5.2 SD, height 1.83 m ± 0.05 SD, pre-trial body mass 95.8 kg ± 10.7 SD) participated and completed the aforementioned field test. Heart rate was recorded throughout and protocol intensity determined. Garment performance was assessed using controlled visual assessment techniques and dimensional stability measurements. Structured questionnaires were administered during and post-trial to determine player perceptions. Results: Forwards’ mean and peak heart rates (153.3 bpm ± 24.1SD and 181.9 bpm ± 10.0 SD) were lower than match-play target values. Backs’ mean and peak heart rates (147.2 bpm ± 26.8 SD and 185.0 bpm ± 10.7 SD) were similar to target values. Dimensional changes and defects were identified and quantified successfully. Players conformed well to the protocol and responded favourably to the questionnaire. Conclusion: The protocol described represents the first rugby-specific garment evaluation protocol to be documented. It is hoped that this will be adopted and refined thereby adding structure to the sports-specific garment development process. Keywords: time-motion analysis, garment durability, wearer trial, player perception, heart rate.
1. Sports Technology Research Group, Loughborough University Sports Technology Institute, 1 Oakfield Avenue, Loughborough University, Loughborough, Leicestershire, UK, LE11 3QF - E-mail:
[email protected] 170 The Engineering of Sport 7 - Vol. 1
1- Introduction Innovation in the rugby union (rugby) team-wear industry is cyclic in nature; predominately driven by major tournaments. Competition at the highest level is seen as a visible platform to promote brand identity through innovative product design. Prior to 1999 Rugby World Cup, rugby team-wear design was predominately fashion led. Focus has since shifted to the improvement of athletic performance via enhanced garment functionality therefore prototype evaluation is critical to assess functionality whilst also ensuring garments can withstand the rigours of the game. Rugby-specific wearer trials are often used to assess the performance of, and wearers’ physiological response to, apparel during training sessions and simulated match-play. However, personal experience and interviews with manufacturers suggest that the current wearer trial process is ad-hoc and has limitations. Test time with elite players is often limited, end-use conditions may not replicate the intensity of match-play and, feedback may be biased. Post-trial interviews are typically informal and unstructured to create a rapport with the athlete rather than maximise opportunity for comprehensive user insight capture. Finally, when tested over a number of trials, the process is timeconsuming and expensive. Wearer trials enable the validation of expected wear performance of a textile in its intended end-use conditions (ASTM D3181 1995) and are defined as any controlled tests carried out on an item of textile apparel in which the item is worn (BSI 7754 1994). Two international standards summarise generalised wearer trial procedures (BS 7754 1994 and ASTM D3181 1995). To date, there are no published sports-wear performance evaluation protocols. The development of a rugby-specific wearer trial was dependent on the development of a rugby-specific exercise protocol to elicit an applicable physiological response to exercise, a thorough knowledge of player perception techniques and structured garment assessment protocols to predict on-field performance. This was to ensure that the shirts and players were subjected to demands similar to that of the game. It is evident that a structured, controlled and repeatable rugby-specific wearer trial is necessary to extract meaningful garment performance and player perception data over a short period of time. The current paper details a pilot study detailing the design and implementation of a rugby-specific wearer trial, however the principles outlined may be applied to other sports.
2- Methods Fifteen international match-play prototype rugby shirts, worn by fifteen non-professional competitive club players (age 26.1 yrs ± 5.2 SD, height 1.83 m ± 0.05 SD, pre-trial body mass 95.8 kg ± 10.7 SD) from each rugby union position, were evaluated using a rugby-specific wearer trial protocol. Eight additional players acted as opposition to the tasks where necessary and all tasks were performed on an outdoor turf rugby pitch. All players provided written informed consent in accordance with generic clearance from the University’s ethical advisory committee. Ambient temperature and humidity were 15.8°C ± 2.1 SD and 55% ± 10 SD respectively.
The Engineering of Sport 7 - Vol. 1
171
Figure 1 - Wear-service conditions.
The prototype shirts were constructed from 78% polyester 22% cotton COOLMAX® weave blend (main body panels), 65% nylon, 21% polyester, 14% LYCRA® blend (arms and side panels) and, 100% cotton twill (collar). Grip panels were positioned on the chest, shoulders and upper back. Time-motion analysis data, used to derive the exercise portion of the rugby-specific wearer trial (Figure 1), were extracted from televised game analysis of the Rugby World Cup 2003 (IRB 2003). Wear-service conditions focussed primarily on non-running intensive exertion (rucking/mauling, lineouts and scrummaging) and game-specific activities (kicking) rather than movement patterns (standing still, walking, jogging, cruising, sprinting and utility) as equipment to classify individual movement patterns throughout the trial were not available. The exercise section of the trial was completed in 60 minutes.
172 The Engineering of Sport 7 - Vol. 1 Protocol intensity was estimated through determination of sweat loss and heart rate which were compared to target values. Nude body mass was measured pre- and posttrial using digital weighing scales (2204-Duo Model, Tanita UK Ltd; accuracy ± 0.2 kg) to derive an approximation of sweat rate however, players were allowed to consume water freely as in match-play. Heart rate (HR) was monitored every 5 seconds using wireless heart rate monitors (Vantage NV, Polar Electro, Finland) to determine mean and peak heart rates. Robergs & Landwehr (2002) conducted a review of age-predicted HRmax and concluded that, to date, the most suitable for the general population was proposed by Inbar et al., (1994) which was used in this study (Equation 1). Mean percentages of time in each HR zone (95% HRmax) were also determined. Scrum-half data were not included in the backs category due to the distinctive nature of the position and for comparison with previous time-motion analysis studies. The movement patterns of a scrum-half are distinctly different from the forward and backs category (Duthie et al., 2005). HRmax = 205.8 - (0.685*age) (1)
Figure 2 - Measurements taken to assess dimensional stability of rugby shirts and garment defect location.
Garments were measured pre-trial, post-trial and post-laundering to assess dimensional stability (accuracy ± 0.5 mm) using a steel rule. Measurement points, as specified by the manufacturer, were 1) half chest width (measured 40 mm vertically downward from the arm-hole), 2) half hem, 3) sleeve length, 4) half sleeve opening, 5) underarm length, 6) neck width, 7) front neck drop, and 8) centre back length, as shown in Figure 2. Visual garment assessment techniques were devised from previous rugby shirt durability testing, literature reviews and interviews with the manufacturer. Scales and failure criteria were assigned to each defect category. The reader is directed to Saville (1999) for further classification and examples of dense, distinct, moderate and slight defects. Defects were described by their defect category, size, severity, shirt location, specific location (anatomical or shirt feature) and failure criteria (Table 1). Failure in a particular defect category is fulfilled when classification matches the number denoted by ‘failure = x’ in Table 1.
The Engineering of Sport 7 - Vol. 1
173
Questionnaires, devised from interviews with players, manufacturer consultation and previous wearer trial information, were distributed during the position-specific drills contained predominately quantitative questions. The aim was to gauge opinion, in a short period of time, regarding the effects of the shirt on performance during a particular activity. Post-trial questions were concerned with the shirt performance throughout the complete trial. Four discrete areas of interest were identified to aid analysis. Firstly the player’s history was gauged to understand personal experience, and perhaps, aptitude for the test. Secondly, general garment functionality was assessed by determining player preference for fit, style, strength, thermal comfort, thermal sensation and, weight. Then rugby-game specificity was assessed such as the need for grip and its positioning, difficulty to tackle, ease of binding, affects during scrummaging and, range-ofmovement whilst handling the ball. Finally, the players were given an opportunity to give additional comments on any aspect of the shirt design.
Table 1 - Visual assessment defect classification post-trial.
3- Results 3.1 Physiological profiles Mean body mass losses were 1.23 kg ± 0.70 SD, 1.70 kg ± 0.37 SD, 0.69 kg ± 0.59 SD for the whole squad, forwards and backs respectively with a mean fluid deficit of 1.3 % initial body mass. Mean heart rate, mean peak heart rates and, percentage of time spent in each HR zone are shown in Table 2.
174 The Engineering of Sport 7 - Vol. 1
Table 2 - Estimated maximum heart rate and mean (SD) participant heart rate data.
3.2 Garment performance Few major defects were observed post-trial; there were three instances of slight seam puckering, two instances of slight aesthetic deterioration, one instance of moderate aesthetic deterioration and two instances of slight seam slippage across all shirts. However, all shirts exhibited staining post-laundering, particularly soil retention. The garments initially increased in size post-trial; the main body material (78% polyester 22% cotton COOLMAX® weave blend) exhibited the most dimensional instability where half hem differences varied 20 mm ± 2.2 SD post-trial. Differences were outside the limits as specified by manufacturers (± 1 mm) therefore the garments failed this aspect of the test.
3.3 Players’ perception of the rugby shirt design Of the 15 respondents, eight felt there were no fit-related issues. Of the other seven, four indicated a preference for a looser fit around the abdomen. The majority of players thought the shirt was stylish and both the fabric and seam construction appeared strong. Ten players felt warm during the trial but only one player felt uncomfortably warm. Three players described particular areas of discomfort; the side, chest and armhole of the shirt respectively. Players stated that the shirt did not restrict passing, tackling, running, scrummaging, or mauling. Six players believed the shirt was too light whereas the rest felt it was “just right” in both wet (after sweating) and dry (initially donned) conditions respectively. Nine players considered the shirt difficult to tackle but none thought it was easy. Finally, the majority of players found the grip, a hexagonal raised silicon transfer measuring 0.5 mm in height on the front and back of the shirt, beneficial when catching and ball carrying (11 players) but felt no benefit during tackling (9 players), binding (9 players), driving (10 players) or lifting (12 players).
4- Discussion A rugby-specific garment evaluation trial was devised to replicate the demands of rugby union using time-motion analysis of game-specific activities and non-running intensive exertion. The aim of the study was to apply the knowledge gained from a number of wearer trials conducted by rugby manufacturers to create a time-saving, structured and controlled rugby-specific wearer trial. This was achieved in this pilot study by assessing the functionality of fifteen international prototype shirts using structured garment evaluation and player perception techniques. The first aim of the investigation was to determine protocol (or wear-service condition) physical intensity. Although overall intensity was lower than a real-life match situation, the exercises were deemed structured and easy to replicate in further trials;
The Engineering of Sport 7 - Vol. 1
175
particularly since the trial was administered by team coaches through a single verbal and written communication prior to exercise. Lower than target heart rates may also be attributed to a lack of competitiveness during the trial, lack of opposition and, periods of rest to respond to questionnaires. In this study, mean heart rates recorded during the study were 153.3 ± 24.1 SD and 149.0 ± 26.7 SD compared to 175 ± 5 SD and 150 ± 5 for forwards and backs respectively (Doutreloux et al. 2002). Players also exhibited a mean fluid deficit of 1.3 % initial body mass whilst still consuming fluid freely throughout the 60 minute trial. Burke (1997) suggests the mean total sweat losses, for the whole squad, during a game of rugby union (~80 minutes) lie between 1.7 to 2.3 L per match corresponding to a mean deficit of 1.5 % initial body mass. Future experiments should incorporate maximal exercise tests to accurately calculate HRmax, thus enabling the percentage of time spent in each heart rate zone to be derived more accurately. It is hoped that the exercise protocol be further refined to elicit a more accurate forwards physiological response and that repeatability be assessed through testing a number of rugby clubs. Two methods of physical garment assessment were employed namely, dimensional stability and visual defect assessment. The pass-fail criteria for both were determined from a series of match-play trials held prior to this study. Defect category and classification were also determined from interviews with manufacturers and analysis of prototype garment standards. In this study, few defects were observed post-trial, however, all shirts exhibited a moderate degree of staining, particularly soil retention. Seam slippage, aesthetic deterioration, and seam puckering were minimal. Only one shirt failed completely due to logo deterioration on the back of the collar. The main body material exhibited signs of garment instability however, according to players’ opinion, performance was not affected. In future, garment features or defects should be given a weighting of importance and with further research, garment defects should be compared to visual grades to more accurately classify each defect. Player perceptions were gauged during and post-trial through position-specific qualitative and quantitative questionnaires. Questions were determined from previous international player interviews, match-play analysis and interviews with rugby-shirt designers. Personal or group interviews were not appropriate for this process due to the limited amount of time on the field of play therefore short quantitative questionnaires were optimal. In this study, players’ opinions were favourable, mostly approving of the shirt design features and style. Player perceptions, in this test, were used to increase understanding and aid product development of rugby shirt design.
5- Conclusions Current garment evaluation trials are often unstructured, time-consuming and potentially biased. A review of the rugby-specific wearer trial development process has been detailed in the current paper. A rugby union specific field test was developed to replicate the non-running intensive exertion and game-specific activities of the game. Wearservice conditions and players’ physiological response to exercise were monitored to assess game intensity, which is invaluable for the reliable assessment of sports garment
176 The Engineering of Sport 7 - Vol. 1 performance. Structured garment assessment techniques were successfully implemented post-trial. All trial prototype shirts passed the wearer trial with few negative player perceptions or functional defects, however, the test procedure was able to identify and distinguish even very minor defects. It is hoped that testers will apply the principles outlined to develop additional sports-specific wearer and equipment trials.
6- References [ASTM1] ASTM D3181-95 Standard guide for conducting wear tests on textiles, ASTM. [BS1] BS 7754:1994 Code of practice for garment evaluation by wearer trials. BSI. [B1] Burke, L.M. Fluid balance during team sports. Journal of Sports Sciences, 15, 287-295, 1997. [DW1] Docherty, D., Wenger, H. A. & Neary, P. Time motion analysis related to the physiological demands of rugby. Journal of Human Movement Studies, 14, 269-277, 1988. [DT1] Doutreloux, J.P., Tepe, P., Demont, M., Passelergue, P. & Artigot, A. Exigences énergétiques estimées selon les postes de jeu en rugby. Science and Sports, 17(4), 189-197(9), 2002. [DP1] Duthie, G., Pyne, D. & Hooper, S. Time motion analysis of 2001 and 2002 super 12 rugby. Journal of Sports Sciences, 23(5), 523 – 530, 2005. [IO1] Inbar, O. Oten, A., Scheinowitz, M., Rotstein, A., Dlin, R. and Casaburi, R. Normal cardiopulmonary responses during incremental exercise in 20-70-yr-old men. Medicine and Science of Sport and Exercise, 26(5), 538-546, 1994. [IRB1] IRB. RWC 2003: Statistical review and match analysis. Dublin: International Rugby Board (IRB), 2003. [M1] McLean, D.A. Field testing in rugby union football. In: Macleod et al., [Eds.] Intermittent high intensity exercise: preparation, stresses, and damage limitation. London: E & F N Spon, 1993. [RL1] Robergs, R.A. & Landwehr, R. The surprising history of the “HRmax=220-age” equation. Journal of Exercise Physiology, 5(2), 1-10, 2002. [S1] Saville, B.P. Physical testing of textiles. Woodhead Publishing Ltd, Cambridge, 1999.
Open Rotator Cuff Surgery in Swiss Elite Rock Climbers (P31) Hans-Peter Bircher1, Christoph Thür2, Andreas Schweizer3
Topics: Climbing; Extreme sport. Abstract: Rock and indoor climbing has become a very popular sport. For moving in a vertical or even overhanging wall an intact shoulder function is mandatory beside a good grasp function. Hand and particular finger injuries in climbing are well described in the literature. Almost no report is avaible about shoulder injuries in climbers. In our retrospective case study 21 shoulders in 20 climbers (18 male, 2 female, age 28 to 65) with open rotator cuff surgery between 1998 and 2006 were analysed. The ability of climbing before and after the operation is a main outcome control. The level of difficulty in climbing is given by the french difficulty scale ranging from three to nine whereas the top grade nine is performed by only few climbers worldwide. In 10 climbers a single shouldertrauma while in 7 a repetetive traumas and in 3 chronic overuse prevented a situation compatible with rock climbing. There are 7 complete rotator cuff lesions and 14 partial tear of the supraspinatus tendon. In all climbers an open rotator cuff surgery was done including 17 enlargement of the subacromial space and 15 tenodesis of the long head of biceps. 6 months after the operation in average the climbers started again. Average time to return in a similar level of difficulty was 16 months in all but one climber. From the point of biomechanics, pathologies of climbing and swimming shoulder are discussed. Keywords: sportclimbing, injury, rotator cuff, open surgery, returning in sportclimbing.
1- Introduction Rock and indoor climbing has become a very popular sport. Sportclimbing is moving with the use of both arms and legs in steep or even overhanging walls or rock. This activity goes back on a rebelish group of young mountaineer in the early seventies in the USA revolting the old fashioned climbing style using technical aid to overcome difficulties. They created the motivation to climb „by fair means“ which excluded technical aid. The climbing gear is used for potection only. The difficulties in climbing increased since, 1. Orthopaedic Clinic, Hospital of Zug, 6300 Zug, Switzerland - E-mail:
[email protected] 2. Shouldersurgery and Traumatology, 8001 Zürich, Switzerland - E-mail:
[email protected] 3. Universityclinic Balgrist, 8008 Zürich, Switzerland - E-mail:
[email protected] 178 The Engineering of Sport 7 - Vol. 1 which is equivalent to a decrease of grips and steps. The stress on the upper extremity in this sport is high. To climb difficult routes, a power and endurance training particular for the upper extremities is mandatory. Several tries in a projected route are needed before climbing it sucessfull. Finger injuries, particular strain or rupture of flexor tendon pulley, are frequent and well described [B1][B2][MG1][LO1] [RM1][S1][SJ1]. It is been estimated that up to 20% of elite climbers suffer shoulder problems [JA1][R1]. Very few information was found about general shoulder problems among climbers in literature. No publication was found about a specific shoulder problem in climbing. The presented paper analyses rotator cuff pathologies in elite climbers and recreational climbers.
2- Material and Methods In a retrospective case study 21 operated shoulders with rotator cuff pathologies in 20 climbers were controlled in two orthopaedic clinics (Clinic of Shouldersurgery and Traumatology, Zürich and Orthopaedic Department, Hospital of Chur, Switzerland). 13 were professional (10 mountain guides, 2 competitive climbers, 1 gym instructor). 7 were recreational climbers on a high level. The avarage age was 44 years (28- 65). In this collective the comparison in climbing prae- and postoperativly is used as a quality control for the operative treated rotator cuff problem. The difficulties in climbing were rated in the world wide accepted french scale, ranging from three to nine. Three is a simple climbing, nnie is an extreme difficult climbing. Each routerating is set by the climber who climbed the wall or rock the first time. Grade 9 is reached by a few climbers world wide only. Each degree is subdivided in a, b and c, rising in difficulty. 10 climbers had a causual shoulder trauma with a tear/distorsion. 7 climbers had repeated shouldertraumas including contusion. 3 climbers showed an overuse of the shoulder before the onset of the dysfunction. Conservative treatment was performed in 16 patients (7 physiotherapy, 2 infiltration with steroids, 2 alternative treatment and 5 with regularly analgetic medication). None of the collective was able to climb even near his top level because of shoulder dysfunction. An arthro-MRI was obtained of all operated shoulders. The intervall between onset of shoulder pain and /or trauma and operation was 16 months (3-24). All patients were operated in general anaesthesia with a supplementary scalenus block. 7 complete and 14 partial tears were diagnosed (Fig. 2). The complete tears were in 6 cases limited on the supraspinatus tendon (1x2 cm to 3x4 cm) In one case the tendon rupture was 5x4 cm including the upper part of the infraspinatus- and subscapularis tendon. The 14 partial tears were subdivided in 2 PASTA, 4 bursal -, 4 articular side and 4 distinct inflammation in the rotator interval and supraspinatustendon. (Table 1)
The Engineering of Sport 7 - Vol. 1
179
Table 1 - Rotator cuff, biceps and SLAP pathologies.
tendon
complete 7
partial 14
supraspinatus
1x2, 2x3,2x3, 2x3, 1x2,3x4
PASTA articular bursal inflammation
2 4 4 4
rupture part inflammation Typ I Typ II
2 3 1 2 2
supraspinatus, infraspinatus subscapularis long head of biceps
5x4
rupture part. tear inflammation
3 4 2
SLAP
In 10 patients a diagnostic arthroscopy was performed. All operations on the rotator cuff were done in an open way. Reinsertion was done with braided non- absorbable sutures (Ethibond No. 2). Intertendinous sutures were fixed with absorbable monofil material (PDS No. 1). In 9 cases a transosseus supraspinatus reinsertion with a doubling of the tendon was done. In 5 cases a supraspinatus reinsertion and in 2 cases a debridement were performed. An acromioplasty was made in 17 shoulders. (13 acromion lift up osteotomies [TJ1] and 4 modified plasties according to Neer). 15 tenodesis of the long head of biceps were done (11 key houle and 4 suture fixation). Additionally two SLAP II lesions were refixated arthroscopicly with absorbable suture ancors (Panalok loaded with PDS 1). Two SLAP I lesions were debrided arthroscopicly. (Table 2) Table 2 - operativ procedure in rotator cuff.
tendon
open operation
arthroscopic operation
supraspinatus
reinsertion and doubling of supraspinatus reinsertion deltoid flap
9 5 1
diagnostic tenotomy biceps debridement
10 8 2
tenodesis - key hole - suture
tenotomy biceps
8
11 4 refixation debridement
2 2
supraspinatus infraspinatus subscapularis long head of biceps
SLAP
Postoperativly 16 patients were treated with an shoulder abduction splint for 6 weeks. In 5 cases the shoulder was immobilized for 6 weeks in a shoulderwest. Physical therapy started the first day after operation obtaining range of motion passivly: abduction/adduction 90-40-0 and internal/externalrotation 20-0-20. After the forth postoperative week
180 The Engineering of Sport 7 - Vol. 1 isometric abduction in 90° with flexed elbow was allowed. A clinical and radiological controll was done 6 week after operation. Active physiotherapy is started to build up a full range of motion. The patients were seen 12, 26 and 52 weeks ambulatory after operation. A telephonic questioning was done in average 49 months postoperativly (12-108). A systematic questionnaire about pain, range of motion, restart of climbing and top level was used.
3- Results In a healty shoulder situation praeoperativly the top level of 5 climbers was grade 8 (1x8c, 3x8b, 1x8a). 4 climber were familiar with grade 7 (1x7c, 1x7b, 2x7a). 9 climbers made grade 6 (7x6c, 2x6b) and 2 climbers grade 5 (2x5c). (Table 3) We saw no complication in the operated collective. The ability of working was given after 4 months postop (1-6). Restart in climbing was in average 8 months after operation (6-12). The top level after the operation was reached 16 months after the operation by the collective (8-24). The top level after the operation of grade 8 climbers was similar to the praeoperativ level in 3. Two grade 8 climber reached 8a instead of 8b praeoperativly. In grade 7 two climbers did the same top level as before the operation. One climber stopped his carrier out of other reason than the operated shoulder. One climber is performing grade 6c instead of 7a praeoperativly. In grade 6 four climbers had small loss, but remained on the 6. degree. In grade 5 there was no difference. (Table 3) Table 3 - Climber and his top level prae- and postoperativly (french scale).
climber A.R. G.H. M.A. A.L. A.L. M.W. R.B. S.A. L.S. N.E. D.M. T.U. L.C. D.Z. C.A. A.R. D.M. A.R. Z.B. F.J. H.D.
prae-op
post-op
8c 8b 8b 8b 8b 8a 7c 7b 7a 7a 6c 6c 6c 6c 6c 6c 6c 6b 6b 5c 5c
8c 8b 8a 8a 8a 8a 7c 7a 6c 6c 6c 6c 6c 6b 6b 6a 6b 6a 5c 5c
The Engineering of Sport 7 - Vol. 1
181
4- Discussion The epidemiology of shoulder injuries in climbing is given by 15-20% of elite and recreatinal cimber in the literature. Finger is the most commen injuiered part of the upper exremity in climbing. Finger injuries are supposed to be three to four times more frequent than shoulder. But shoulder dysfunction is a real handicap. The aethiology of shoulder dysfunction is either cuff tear by overload. Further repetitive overstrain may result in cuff tear and /or subacromial bursitis with a following outlet impingement. In our collective almost every climber mentioned several contusion or distorsion of shoulder while climbing in the past. The most commen reason of these accidents is a fall into the securing rope. This occurs usually during exploring a difficult climbing part. In our experience rotator cuff pathologies and subacromial impingement problems represent the main part of shoulder problems. Beside these we observed in climbers instabilities, omarthritis, AC- dislocation, AC- arthritis and distal tendon rupture of biceps. The cuff and biceps pathologies in the presented collective show a large spectrum. There seems not to be one typical pathology. It is difficult to compare the shoulder in climbers with other sport. Due to a distribution on both arms for moving and a similar load subdivision between arms and legs [PT1] the shoulder of swimming may be comparable with the climbing shoulder. Competitive swimmer perform nearly 500’000 stroke revolutions per arm and season [RJ1]. The swimmer’s shoulder has an element of general laxity due to repetitiv arm motion. It is thought that a loss of static components in stabilisation require a greater contribution from the rotator cuff [RB1][ZM1]. A certain laxity in climbing shoulder is found as well. This kind of laxity in the climber shoulder may be explained by the repetitive grasping toward the top. But the number of tractions and the dynamic load in the pull through or climb through part of movement is difficult to compare. On the side of the impingement, Swimmer’s shoulder show moments of subacromial but also moments of nonoutlet type of impingement [YH1][YH2]. Due to the pathology of cuff lesions in the climbing shoulder a certain nonoutlet form of impingement must be discussed. Open cuff surgery shows an excellent outcome in our collective. Postoperativly 19 of twenty climbers took up climbing in a similar level including the elite. The postoperativ rehabilitation is a key point in teh treatment as well. The climbers showed a progressiv strength in their operated shoulder up to 2 years. The top level in climbing reach the climbers in average 8 months after the restart.
5- References [B1] Bollen S., Soft tissue injury in extreme rock climbers. Br J Sports Med, 22(4): 145-7, 1988 [B2] Bollen S., Upper limb injuries in elite rock climbers. J R Coll Surg Edinb. 35(6 Suppl): S1820, 1990 [JA1] Jones G, Asghar A, Llewellyn DJ. The epidemiology of rock climbing injuries. Br J Sports Med, 2007
182 The Engineering of Sport 7 - Vol. 1 [LO1] Largiader U., Oelz O., [Analysis of overstrain injuries in rock climbing]. Schweiz Z Sportmed, 41(3): 107-14, 1993 [MG1] Moutet F., Guinard D., Gerard P., Mugnier C., [Subcutaneous rupture of long finger flexor pulleys in rock climbers. 12 case reports]. Ann Chir Main Memb Super, 12(3): 182-8, 1993 [PT1] Pink M., Tibone J.,The painful shoulder in the swimming athlete. Ortop Clin Noth Am, 31(2): 247-261, 2000. [R1] Rooks MD. Rock climbing injuries. Sports Med, 23(4): 261-70, 1997 [RB1] Rupp S., Berninger K.,Hopf T., Shoulder problems in high level swimmers – impingement, anterior instability, muscular imbalance? In Int J Sports Med, 16(8):557-562, 1995. [RJ1] Richardson A., Jobe F., Collins H., The Shoulder in competitive swimming, Am J Sports Med, 8(3):159-163, 1980. [RM1] Rohrbough J., Mudge M., Schilling R. Overuse injuries in the elite rock climber. Med Sci Sports Exerc, 32(8): 1369-72, 2000 [S1] Schweizer A., Lumbrical tears in rock climbers. J Hand Surg [Br], 28(2): 187-9, 2003 [SJ1] Schoffl VR, Jungert J. Closed flexor pulley injuries in nonclimbing activities. J Hand Surg [Am], 31(5): 806-10, 2006 [TJ1] Thür C., Jülke M., Bircher H., [Lifting osteotomy of the acromion as a new principle in treatment of impingement syndrome, especially in correlation with reconstruction of large rotator cuff lesions] Unfallchir 101(3):176-83, 1998 [YH1] Yanai T., Hay J., Miller G., Shoulder impingement in front crawl swimming: I. A method to indentify impingement. Med Sci Sports Exerc, 32(1):21-29, 2000. [YH2] Yanai T., Hay J., Shoulder impingement in front crawl swimming: II. Analysis of striking technique. Med Sci Sports Exerc, 32(1):30-40, 2000. [ZM1] Zemek M., Magee D., Comparison of glenohumeral joint laxity in elite and recreational swimmers. In Clin J Sport Med, 6(1): 40-47, 1996 Jones G, Asghar A, Llewellyn DJ. The epidemiology of rock climbing injuries. Br J Sports Med, 2007 Largiader U, Oelz O. [Analysis of overstrain injuries in rock climbing]. Schweiz Z Sportmed, 41(3): 107-14, 1993 Rohrbough JT, Mudge MK, Schilling RC. Overuse injuries in the elite rock climber. Med Sci Sports Exerc, 32(8): 1369-72, 2000
A Quantitative Analysis Of Beach Casting (P33) Benjamin Charles1, Darryl P Almond1, Aki I T Salo2 Presented by Alan N Bramley1
Topics: Sea fishing, video analysis. Abstract: Video analysis techniques have been used to analyse quantitatively the overhead casting method employed by sea anglers fishing from beaches. The techniques have been used to estimate launch velocities and launch angles achieved by this casting method. Cast distances achieved have been compared with results of projectile calculations, using observed launch angles and velocities. The measured cast distances have been found to be between two and three times shorter than predicted by basic projectile calculations. Further experiments indicate the drag of the running line to be the main cause of the reduction in cast distance. Keywords: video analysis, projectile dynamics, sea fishing, casting.
1- Introduction Beach casting is a very popular form of fishing which involves using a long rod to cast hooked bait or lures from a beach, pier, breakwater or off rocks into the sea. A lead weight is attached to the end of the line to ‘pull’ the hooked bait or lures out from the shoreline and across the water during a cast. The main objective of the cast is to project the hooked bait or lures as far as possible from the shoreline into the sea. Despite the enormous number of people participating in this and other forms of angling, there has been no quantitative scientific analysis of the casting method. There have been some publications on fly casting (Robson 1990, Spolek, 1986) and various forms of beach casting and the equipment have been described (Holden, 1982). We report here the results of a detailed study of beach casting in which the casting process has been subjected to video analysis and casting distances achieved are compared with the predictions of conventional projectile calculations. The method of casting studied here is the “overhead” or “ground” cast which is one of the most commonly used methods. At the start the lead weight at the end of the line 1. Department of Mechanical Engineering, University of Bath, Claverton Down, Bath, BA2 7AY, UK. E-mail:
[email protected];
[email protected] 2. Sports and Exercise Science, University of Bath, Claverton Down, Bath, BA2 7AY, UK - E-mail:
[email protected] 184 The Engineering of Sport 7 - Vol. 1 is suspended from the rod tip by between one and two metres of line (the “leader length”) and laid on the beach behind the angler. The angler then rotates the rod rapidly in an arc set in a vertical plane. The weight is towed around, following the rod tip, until the line is released when the angle between the rod and the sea is approximately 45. The mechanism is very similar to that employed by the medieval siege engine, the trebuchet, to project missiles over great distances. The momentum of the weight at the point of release determines the distance it travels across the sea. This distance is also affected by the aerodynamic drag of the weight and the bait or lures and by the drag of the line as it is drawn off the fishing reel and through the line guide rings along the length of the rod. There is a number of other casting methods, notably the pendulum cast, that involve rotating the weight in a near horizontal plane to build up momentum prior to release. These methods have not been investigated here because of the complexity that arises in video analysis of a 3D motion of this type. The advantage of the overhead or ground cast is that the motion occurs, to a good approximation, in the vertical plane alone, allowing it to be recorded by a single video camera.
2- Equipment The rod used was a 12 foot Shakespeare Targa Beachcaster. 12 feet (3.66 metres) is the most commonly used length for beach cast fishing. The rod is made of carbon fibre which is currently the standard material for fishing rod manufacture. Since the rod is manufactured from carbon fibre, it has a dark almost black appearance. White tape was therefore applied to the rod at intervals along its length to ensure it stood out in the video recordings. The reel used was a Targa Beachcaster Fixed Spool Reel, as supplied with the rod. The line, supplied with the reel, was a 15 pound (6.8 kilogram) test strength nylon “running line” that has a diameter of 0.5 mm. In the tests reported here, only a weight was cast. Three different weights were used, 2, 3 and 4 ounce (57, 85 and 113 grams). These were standard torpedo shaped lead weights. One of the weights was attached to a ~ 4 metre length of 50 pound (24 kg) test strength nylon line, known as the “shock leader”. This was in turn tied to the running line. The purpose of the shock leader is to withstand the high centrifugal forces exerted by the weight on the line during casting that could break the lighter running line. During casting, the angler holds the end of the shock leader line with one finger against the rod handle so that casting forces are exerted on the shock leader line and not on the running line. At the release point of the cast, this finger is raised, releasing the shock leader which is drawn through the line guide rings by the motion of the weight, which in turn draws the running line off the reel and through the same rings and away from the rod. The casts were recorded using a Sony DCR-TRV 900E Mini-DV Camcorder with an effective video resolution of 420 k pixels. This camera records at 25 frames per second, thus yielding 50 fields per second for the analysis.
The Engineering of Sport 7 - Vol. 1
185
3- Casting experiments A large number of casts were video recorded and the corresponding cast distances were measured. Different combinations of weight and shock leader lengths were tried. Casts of one particular combination of shock leader length and weight (4 foot (1.22m) and 3 ounce (85 grams) were repeated 14 times to study the variability in cast distance. The video analysis was completed using Vicon Peak Motus software [Vicon Motion Systems, Inc., Centennial, CO]. The digitizing process of the video sequences involves progressing through each field in the video (there are two fields per frame, and thus 25 frames per second (yielding 50 fields per second) and marking all the relevant locations the user wishes to analyse further. For this analysis, nine points were selected along the rod, and one for the weight. The points used were where the white tape had been added to the rod, mentioned above. The locations on the rod were not equal distances apart as fewer are required towards the base of the rod which has a thick section that bends little during casting.
4- Results
Figure 1 - A sequence of digitised stills from the video recording of a cast.
Figure 1 is a selection of images from the video recording of a cast. The magnitude of the velocities of the weight and the rod tip during the cast, given by the Peak Motus software, are shown in figure 2. Figure 2 shows the weight and rod tip to have similar velocities for the majority of the cast as the weight is towed up from the ground.
186 The Engineering of Sport 7 - Vol. 1
Figure 2 - Resultant velocities of the weight and the tip of the rod plotted against time. Data is obtained from the digitised images, figure 6, using Peak Modus software.
The period of acceleration, figure.2, corresponds with the bending of the rod evident in figure 1. This bending is a reaction to the force being used to accelerate the weight. Measurements of the stiffness of the rod confirm the bends in figure 1 to match the acceleration of the weight indicated in figure 2. It can be seen from figure 1 that towards the end of the cast, when the rod is approximately vertical, the weight begins to orbit the rod tip. It is during this phase of the cast that the line is released and the precise time of release determines the angle at which the weight moves away from the rod. The centrifugal forces in the shock leader line are at their maximum during this phase of the cast. The data at times after the line release were used to estimate the magnitude of the weight velocity and the angle at which it was moving at line release. An error analysis indicated an uncertainty of ± 1 ms-1 in estimations of weight launch velocity and ± 2o in launch angle. The results shown in figure 3 were obtained using three different weights, 2, 3 and 4 ounce (57, 85 and 113 grams) and shock leader lengths of 0, 2, 4, 6 and 8 ft (0, 0.61, 1.22, 1.83 and 2.44 m)
The Engineering of Sport 7 - Vol. 1
187
Figure 3 - Comparison of predictions of cast distance, calculated with and without weight drag, with actual experimentally measured distance.
5- Analysis of cast distances The first approximation is to assume the weight to be a freely moving projectile and to estimate cast distances from observed launch angles and launch velocities. However, it is clear from figure 1 that the weight is released from a point several metres above ground and this factor must be added to calculations at the outset. It is well known that the distance, s, travelled by a freely moving projectile launched from a point on the ground is given by: (1) In which u is the launch velocity, is the launch angle and g is acceleration due to gravity. It can be seen from this equation that the optimum launch angle is 45° as this leads to sin2 taking its maximum value of 1. It can also be seen that projectile distance is fairly insensitive to launch angle because of the slow variation of sin2 for values of around 45°. If the projectile is launched at a distance h above the ground, the projectile distance becomes: (2)
188 The Engineering of Sport 7 - Vol. 1 Where ~ 45° and h 40’000€) and is therefore unsuitable for many sports. Recently, smaller and lighter equipment consisting of low-cost MEMS triple axis accelerometer and gyroscopes together with inexpensive L1 GPS receivers were introduced by (Waegli and Skaloud 2007). As it has been shown, the combination of these sensors helps to overcome the lack of continuity in the reception of the GPS signals in obstructed environment and to determine accurately the orientation of the MEMS-IMU sensor. In this article, the design of the GPS/MEMS-IMU system is described. Then, its performance is illustrated based on an experiment where the system was installed on a motorcycle. In the second part of the paper, applications of motorcycling are highlighted where the use of such a system can bring considerable benefits. Firstly, an example of trajectory analysis is given. Secondly, the computation of the lateral slipping of a motorcycle based on the GPS/MEMS-IMU trajectory is explained. Lastly, it will be shown how GPS/INS derived parameters can be exploited in order to characterize the performance of tires.
2- GPS/MEMS-IMU Integration Performance MEMS-IMUs are subject to large random and systematic errors (biases, scale factors, misalignment, and noise) which need to be suppressed in order to provide accurate orientations. For instance, a typical bias of 0.5m/s2 of a MEMS accelerometer would result in positioning errors of 50m after 10s. Representative MEMS gyroscope biases are up to 1°/s and would lead to orientation errors of 10° after only 10s. A solution for calibrating these errors consists in the integration of MEMS-IMU with satellite positioning. However, the conventional GPS/INS integration strategies (Titterton and Weston 1997) need to be slightly adapted in order to take into consideration the error characteristics of the MEMS-IMU sensors (Waegli, Skaloud et al. 2007). A motorbike was equipped with MEMS-IMU (xsens MT-i) which is rigidly fixed to the GPS antenna. A low-cost mono-frequency receiver (u-blox AEK4) was used together with a dual-frequency GPS receiver from Javad for reference. The results presented below integrate differential L1 GPS solutions of low-cost receivers at 1Hz with the triple-axis accelerometer and gyroscope measurements of the MEMS-IMU which are at 100Hz.
Figure 1 - GPS/MEMS-IMU system setup.
The Engineering of Sport 7 - Vol. 1
225
The performance of the low-cost L1 GPS/MEMS-IMU system was investigated by (Waegli and Skaloud 2007) in Alpine skiing by comparing its performance to a higher order GPS/INS system. The mean accuracies are summarized in table 1. The system parameters converge rapidly which permits bridging of GPS data gaps of up to 10s with inertial navigation without loss of accuracy. Using dual-frequency receivers would increase the position accuracy to decimeter-level while velocity and orientation quality would remain at a similar level. Table 1 - System setup and mean accuracy of the presented differential GPS/MEMS-IMU system.
3- Applications to Motorcycling Accurate position, velocity, acceleration and orientation data is crucial in many domains in motorcycling. In the sequel, applications where the GPS/MEMS-IMU system provides innovative approaches are described.
3.1 Trajectory Analysis The precise trajectory allows visualizing and comparing any parameters related to the performance (Waegli and Skaloud 2007). This may be quantities derived from GPS/MEMS-IMU trajectory (e.g. velocity, acceleration, tire slips) or parameters related to the motorcycle (e.g. throttle, suspensions) or to the athlete (e.g. heart-rate). Figure 6 gives an example where the lateral slipping of the back wheel of the motorbike is visualized on two turns of the track.
3.2 Solution to the Reference Frame Problem When studying the motorcycle performance, various parameters are determined in different reference frames, e.g. in the reference frame fixed to the motorcycle (abbreviated with b for body) and in the reference frame fixed to the track (l for local-level frame, figure 2). Measurements (e.g. force or torque) can be converted between the two frames thanks to the rotation matrix Rlb which expresses the orientation of the motorcycle with respect to the track and which is estimated in the GPS/MEMS-IMU integration. xl = Rlb • xb where x stands for observations in either frame.
(1)
226 The Engineering of Sport 7 - Vol. 1
Figure 2 - Illustration of the reference frames.
3.3 Computation of the Lateral Slipping of Tires The lateral slipping of tires can be observed directly with the use of GPS/INS derived results. First, the direction of the trajectory can be derived from the velocity components vNorth and vEast (Figure 2) : v tan() = North vEast
(2)
Furthermore, the GPS/INS integration yields the heading hd of the motorcycle. Combining the two variable leads to the slip of tires :
= hd –
(3)
Figure 3 and figure 4 illustrate the slip angle with respect to the throttle, traction/braking torque at the rear wheel, the roll angle and the lateral acceleration respectively. The confidence level (1 1°) highlights the accuracy of the slips. The presented experiment was conducted in winter and the dynamics were correspondingly small. Nevertheless, statistically significant drifts were observed during the turns. The lateral acceleration is consistent with the roll angle. It can be noted that the motorcyclist was inclined even on a great portion of the straight lines (Figure 5). As seen from figure 4, this inclination was compensated by a lateral acceleration.
The Engineering of Sport 7 - Vol. 1
227
Figure 3 - Throttle with slip angle and accuracy indicator (1). The black vertical lines indicate the beginning of the turns.
Figure 4 - Torque, lateral accelerations and roll during the same period as figure 3. The black vertical lines indicate the beginning of the turns.
Figure 5 - Lateral slipping angles visualized on the GPS/MEMS-IMU derived trajectory.
228 The Engineering of Sport 7 - Vol. 1
3.4 Evaluation of the Characteristics and Performance of Tires The “magic formula tire model” defined by (Pacejka and Bakker 1992) provides a mathematical expression from which forces and moments acting longitudinally or laterally from the road on the tire can be related to its slip performance: y(x) = D • sin(C • arctan(B • x – E • (B • x – arctan(B • x)))) Y(X) = y(x) + y x = X + x
(4)
where – X represents the lateral slipping angle or the longitudinal slip . – Y stands for the lateral force Fy, the aligning torque Mz or longitudinal force Fx. – B, C, D, E, x and y are constant coefficients.
Figure 6 - Typical tire characteristics indicating the meaning of some coefficient of equation 4.
The constant coefficients are usually determined by laboratory experiments (on drum). The GPS/MEMS-IMU system permits the calibration of these deterministic parameters in situ. First, force and torque measurements need to be referenced with respect to the road which becomes possible due to the orientation determined by GPS/INS. The lateral slip angle is directly computed from GPS/INS and needs only to be corrected for the steering of the front wheel, the tire radius variation due to speed, load and roll angle as wells as the suspension pitch which can be measured directly by means of linear potentiometers. The longitudinal slip requires the knowledge of the longitudinal velocity (vsx, from GPS/INS) and of the longitudinal velocity of the tire at the contact patch (vx, from digital speed sensors, which must be compensated for the tire radius variation ): v k = – vsx x
(5)
Hence, combining all these measurements permits to characterize tires in the field. This calibration reflects the actual characteristics of the surface (temperature and road roughness) and therefore refines laboratory findings. The peak value of is often situated
The Engineering of Sport 7 - Vol. 1
229
at approximately 10° of slipping and slipping angles up to 30° can be modeled. In this range, the accuracy of the slipping angles accounts therefore for 3-10% of the error, whereas the torque and force measurements can be reached with a level of 5%.
3.5 Tire related vibrations Attitude determination also provides essential data to the motorcycle vibration analysis. Indeed, tire-related vibrations are today very important in motorcycle racing. The socalled “chattering” vibrations may appear at the curves’ entry, apex or exit and enforce the rider to reduce its speed. To solve this problem, the phenomenon is studied by means of numerical and experimental approaches (Duvernier, Fraysse et al. 2002). One approach consists in using hybrid eigenvector bases to perform a modal synthesis (Schorderet and Gmür 2004) (Schorderet 1997). The vibratory behavior of a given tire is depending on two external parameters: ground load (vertical) and roll angle. To build the modal basis, a laboratory experimental modal analysis provides the eigenfrequencies and eigenmodes for discrete values of these parameters. Then, vibration measurements are realized on the track and dedicated software is applied to identify the frequencies where chattering exists (rider data). The hybrid model is then used to reduce or cancel the chattering vibrations. The efficiency of this predictive model is depending on the experimental data quality. The use of the low-cost L1 GPS/MEMS-IMU system combined with a force measurement unit (body reference frame) allows determining accurate roll angles and evaluating the vertical force in the local reference frame. These two parameters are required for the definition of the precise dynamic conditions under which the vibrations appear.
4- Conclusion A low-cost GPS/MEMS-IMU system was presented which is suitable in terms of cost and ergonomy for deployment in motorcycling. The system continuously tracks the 3Dtrajectory of the motorcycle which allows monitoring and comparing a large number of parameters related to the performance. Any performance parameter can be represented with respect to the trajectory and compared to subsequent turns or laps. Its accuracy (50cm in position, < 0.2m/s for velocity and 1-2° for orientation) opens new possibilities in analyzing many factors related to motorcycle performance. The system provides orientations which enable the computation of the lateral slipping angles in relation to the track and measurements of the motorcycle (e.g. force, torque measurements). Combining all this data enables the calibration of tire models in situ.
5- Acknowledgement This research is financed by TracEdge, based at Grenoble, France. The experimental tests have been performed using torque sensors provided by NRCtech, based at EPFL Science Park, Lausanne, Switzerland.
230 The Engineering of Sport 7 - Vol. 1
6- References Duvernier, M., P. Fraysse, et al. (2002). Tyre Modelling for NVH Engineering in ADAMS. 1st MSC.ADAMS European User Conference, London. How, J., N. Pohlman, et al. (2002). GPS Estimation Algorithms for Precise Velocity, Slip and Racetrack Position Measurements. SAE Motorsports Engineering Conference & Exhibition. Pacejka, H. B. and E. Bakker (1992). “The magic formula tyre model.” Vehicle System Dynamics International Journal of Vehicle Mechanics and Mobility 21: 1-18. Schorderet, A. (1997). Synthèse modale et problème inverse en dynamique des structures. Lausanne, EPFL. PhD: 1698. Schorderet, A. and T. Gmür (2004). “Structural dynamics optimization based on a hybrid inverse synthesis method using a quadratic approximation.” ASME Transactions : Journal of Vibration and Acoustics 126(2): 253-259. Skaloud, J. and P. Limpach (2003). Synergy of CP-DGPS, Accelerometry and Magnetic Sensors for Precise Trajectography in Ski Racing. ION GPS/GNSS 2003, Portland. Titterton, D. H. and J. L. Weston (1997). Strapdown inertial navigation technology, Peter Peregrinus Ltd. Waegli, A. and J. Skaloud (2007). Assessment of GPS/MEMS-IMU Integration Performance in Ski Racing. ENC, Geneva, Switzerland. Waegli, A. and J. Skaloud (2007). “Turning Point – Trajectory Analysis for Skiers.” InsideGNSS(Spring 2007). Waegli, A., J. Skaloud, et al. (2007). Assessment of the Integration Strategy between GPS and Body-Worn MEMS Sensors with Application to Sports. ION GNSS, Fort Worth, Texas. Zhang, K., R. Grenfell, et al. (2003). Towards a Low-Cost, High Output Rate, Real-Time GPS Rowing Coaching and Training System. 16th International Technical Meeting of the Satellite Division of The Institute of Navigation, Portland.
Wireless Impact Measurement for Martial Arts (P43) J.I. Cowie1, J.A. Flint1, A.R. Harland1
Topics: Apparel; Measurement Systems Abstract: The integration of electronics into performance sports equipment has increased in recent years and with it the facility to provide more accurate judging, improved coaching, participant health monitoring and to enhance the entertainment of spectators. The focus in this paper is to respond to recent interest in various martial arts in quantifying and categorising impacts which occur during competition. An impact measurement technique has been developed which incorporates a non-invasive sensor system into a body protector worn in Taekwondo. The demonstrator system is integrated into the garment and utilises Bluetooth technology to transmit sensor readings back to a computer for analysis. The sensors considered for impact measurement include thin film piezoresistive force and pressure sensors, and also much newer technologies such as MEMS (Micro-Electro-Mechanical Systems) accelerometers. Impact analysis has been carried out in both the frequency and time domain in order to determine the suitability of the different methods and the bandwidth requirements. The key metrics of force, pressure, velocity and impact duration have been assessed. It has been shown by field experiments and by employing standardised test methods that impacts can be best characterised by making use of various types of sensor in combination. Bluetooth allows for 100 m range; however it was found that the limited bandwidth afforded makes real time data capture problematic unless the sampling rate, bit resolution or number of channels is kept relatively small. In conclusion, a highly compact measurement system has been realised which is capable of quantifying martial arts impacts. The Bluetooth standard is able to support the data transfer requirements, however development of a ‘smart’ on-garment sensor system would be the next logical step to manage a multi-sensor measurement system. This way the gathered data can be compressed or coded to reduce the transmission bandwidth. Keywords: Impact; Measurement; Piezoresistive; MEMS Accelerometers; Wireless; Bluetooth.
1. Loughborough University, Loughborough, England - Email: J.I.Cowie, J.A.Flint,
[email protected] 232 The Engineering of Sport 7 - Vol. 1
1- Introduction In Taekwondo points are scored for a punch or kick to the thorax or head, these are recorded by four judges situated around the match arena, decisions are made purely on visual inspection and the officials judgement. The World Taekwondo Federation has identified the requirement for an electronic scoring system in the game to assist the match officials. As well as assisting officials, the integration of an electronic measurement system could be tailored to a number of other functions; to monitor the health of the players to determine if too many impacts have been received in the same location, and for the purposes of coaching to allow coaches to monitor technique, for example where players are dropping their guard or where they are not being the most effective during attacks. In order to achieve this instrumentation would be required in the protection equipment that is worn in matches and coaching sessions. This instrumented protective clothing would need to measure one or more of the key metrics of the impacts; force in the order of 2kN (Pierce et al. 2004), pressures of several hundred kPa, the velocity of the impact up to 10ms-1 (Gulledge, Dapena 2008) and displacements of tens of millimetres. This paper describes the measurement system for martial arts which incorporates piezoresistive sensors, MEMS accelerometers and a Bluetooth radio system. An overall system diagram is shown in Figure 1. The individual elements of the system will be discussed in the following sections. The problem can be broken down into a number of areas. A sensor technology is required to detect the impacts of kicks and punches. A microprocessor system needs to be developed to perform some signal processing on the sensor outputs and encode them for transmission. Radio frequency transceivers and antennas are needed to transmit the data from the protective equipment to a computer. Software also needs to be developed to take the data received, store them and present them for judges to make decisions.
Figure 1 - Block diagram of the system.
The Engineering of Sport 7 - Vol. 1
233
2- Sensor Technologies 2.1 Piezo Transducers Piezoresistive and piezoelectric force and pressure sensors have been used for some time (Millet et al. 1998, Bachus et al. 2006, Schmidt 2007) in sports applications (impacts, skiing and golf) to gather data generated by human interactions. Piezoresistive materials function by having a resistance which varies with the force which is applied to it. Piezoelectric materials generate a voltage across them when a force is applied to them, this voltage can be very large (much greater than the 10V maximum input to most analogue to digital converters). An instrumented punch bag with a piezoelectric strip to measure force has been used to demonstrate piezoelectric sensors. Figure 2 shows a voltage trace from a punch impact to the bag. This clearly demonstrates a known problem with the piezoelectric materials as the higher values have been clipped due to the capabilities of the data acquisition system.
Figure 2 - Voltage response from piezoelectric force sensor after a gentle punch.
2.2 MEMS Accelerometers One of the key metrics to be determined is velocity of the impact this can not be measured directly, and can not be determined from force pressure data. The use of accelerometers to measure soft tissue displacements has been shown (Boyer, Nigg, 2006). The use of Micro-Electromechanical Systems (MEMS) accelerometers has been studied to measure accelerations, velocities and displacements. These devices are becoming more accessible due to their mass marketing in commercial items such as games consoles and mobile phones. MEMS accelerometers operate using a mass spring system, as the device
234 The Engineering of Sport 7 - Vol. 1 is accelerated the mass’s inertia deforms the spring, this deformation is then measured. A punch bag has been instrumented with an accelerometer and various drop tests and punches have been carried out. An ADXL320 accelerometer from Analogue Devices has been used this is mounted with support circuitry on flexible PCB (Printed Circuit Board) substrate, Figure 3. A National Instruments USB data acquisition card is used to capture and digitise the data from the accelerometer. Figure 4 shows the resulting trace from an accelerometer during a punch impact.
Figure 3 - Accelerometer on flexible PCB attached to punch bag.
Figure 4 - Acceleration response of a gentle punch.
The Engineering of Sport 7 - Vol. 1
235 (1) (2)
From the acceleration an integration (Eq. 1) can be carried out to give a velocity profile and a double integration (Eq. 2) can be carried out to determine the displacement. Numerical integration methods were used to calculate velocity and displacement as can be seen in Figure 5. Errors have emerged in the velocity and displacement responses due to the constants of integration, the displacement has a falling response following c1t+c2. This problem is solved in post processing by the inclusion of a high pass filter with cut off frequency around one hertz.
Figure 5 - Velocity and displacement data from accelerometer.
3- Bluetooth Transmission The transmission from the garment is achieved using the Bluetooth transmission protocol operating in the ISM (Industrial Scientific and Medical) 2.4GHz license free band. The Bluetooth class 1 band offers a range of 100m utilising a low power requirement it also gives a bandwidth of 1Mb/s which can be enhanced to 3 Mb/s. For the purposes of this project a commercially available Bluetooth transceiver is used, a Roving Networks BlueSentry XP is used. The BlueSentry contains a RN-41 class 1 Bluetooth module for transmission, this is controlled by a Microchip PIC16F73 microcontroller. The BlueSentry also includes a Texas Instruments ADS8344 analogue to digital converter allowing the connection of sensors directly to the device. The analogue to digital converter allows for 8 channels to be used with a 16 bit resolution, this gives 4 sensors
236 The Engineering of Sport 7 - Vol. 1 differentially connected or 8 single ended sensors. Access to the device is given through the use of the Bluetooth Serial Port Profile, SPP. The SPP simulates an RS232 serial port giving the ability to send the device simple ASCII instructions to set up the analogue to digital converter and transmitter. Software has been written in visual basic to allow a Bluetooth enabled computer to interface with the BlueSentry, this software gathers data from the digital channel displays it graphically on screen and logs it to a comma separated values file with a time stamp for later offline analysis. The 8 channels available is a limitation of the Bluetooth protocol, the bandwidth offered will not allow for any more data to be transmitted continuously.
4- Integration In order to bring these systems together a method for integrating them with the personal protective equipment has been devised. This includes a study investigating where movement occurs in the human to determine the best location to place rigid electronics and to calculate the strains required to be withstood by the interconnections and transmission lines within the garment. The use of flexible printed circuit board substrates has been demonstrated with the accelerometers, further use of this material is required. Work has also been carried out on the use of conductive yarn to provide flexible interconnections in fabrics. A silver coated nylon thread is being used, this yarn can be sewn using a commercially available domestic sewing machine as well as by hand. It has a resistance of 3k per meter, this relatively high resistance can be lowered by stitching the same path multiple times, effectively increasing the number of strands in the interconnecting wires. The use of a very fine stitch pattern also helps in reducing the resistance, with these to methods a resistance of 200 per meter has been demonstrated. The integration of stitched wires into a fabric allows for the creation of a fully flexible circuit.
5- Conclusion A couple of methods of measuring impacts has been presented and a method for transmitting the data from the protective clothing to the judge has been outlined. Due to the limited bandwidth available from the Bluetooth protocol only a small number of sensors or an unacceptably low sampling frequency on the analogue signals is possible. To over come this problem a system of microprocessors with a certain level of intelligence will need to be developed to allow a large number of highly coupled sensors to operate simultaneously, whilst only relevant data is transmitted, this will reduce the amount of data which is needed to be transmitted while increasing the amount of data captured on the garment. With the use of multiple sensor types impacts in martial arts can be fully categorized and identified using the important metrics of force, pressure, acceleration, velocity and displacement. A highly capable wireless measurement system has been developed for use in Taekwondo or similar contact sports to assist in judging, coaching and health monitoring. It makes use of wireless technologies, micro-electromechanical systems, materials and microprocessor technologies. It was found that one limitation is the amount of information versus bandwidth and the need for a strategy to manage this information. There is also the possibility to use other sensor technologies not mentioned
The Engineering of Sport 7 - Vol. 1
237
here and the need to develop suitable antennas for use in a garment. Further work needs to be carried out to make the system robust enough to be used in match situations so that it can withstand all the movement and impacts that are associated with the sport.
6- References [BD1] Bachus, K. N., DeMarco, A. L., Judd, K. T., Horwitz, D. S. and Brodke, D. S. Measuring contact area, force, and pressure for bioengineering applications: Using Fuji Film and TekScan systems. Medical Engineering & Physics 28: 483–488, 2006 [BN1] Boyer, K. A. and Nigg, B. M. Soft tissue vibrations within one soft tissue compartment. Journal of Biomechanics 39: 645–651, 2006 [GD1] Gulledge, J. K. and Dapena, J. A comparison of the reverse and power punches in oriental martial arts. Journal of Sports Sciences 26(2): 189 – 196, 2008 [MH1] Millet, G. Y., Hoffman, M. D., Candau, R. B. and Clifford, P. S. Poling forces during roller skiing: effects of grade. Medicine and Science in Sports and Exercise 30 (11): 1637-1644, 1998 [PR1] Pierce, J., Reinbold, K., Lyngard B., Goldman, R., and Pastore1, C. Direct measurement of punch force during two professional boxing matches. Journal of Sport & Exercise Psychology 26(Suppl.); 150, 2004 [S1] Schmidt, E. R. Measurement of grip force and evaluation of its role in a golf shot. April 2007
7- Acknowledgments The authors would like to thank the co-investigators on this project Roy Jones and Richard Hague and also the IMCRC (Loughborough University Innovative Manufacturing and Construction Research Centre) who with EPSRC (UK Engineering and Physical Sciences Research Council) have funded it.
A Comparative Study of Ball Launch Measurement Systems; Soccer Case Study (P44) Jouni Ronkainen, Chris Holmes, Andy Harland, Roy Jones1
Topics: Measurement systems, Soccer. Abstract: The sports ball market is extremely competitive and in the US alone valued in excess of $1200 million [SG1]. In order to research and develop sport balls it is vital to quantitatively measure the launch and flight characteristics of the ball. Original equipment manufacturers (OEM’s) are currently using a wide range of systems to measure these parameters, allowing direct comparison between products. The purpose of this investigation is to compare some of the most methods currently available to measure soccer ball launch characteristics. The three measurement systems used were an optical system, a radar system and a high speed video (HSV). A detailed operational description of each method is provided which highlights system strengths and weaknesses. All systems were tested by assessing the launch characteristics of 30 kicks representing a maximal velocity strike, along with 30 impacts representing a curve kick. The kicks were carried out using a purpose built mechanical kicking simulator developed at Loughborough University (LU) in order to obviate inconsistencies achieved with player testing. The main findings from the work showed statistically that the optical system gave a higher soccer ball velocity, whereas the radar system gave a lower launch angle to the other systems. The measured spin rates for kicks highlighted the limitations of current measurement systems, due to discrepancies between all measured spin rate values. In summary this is the first comprehensive study to compare current soccer ball launch measurement systems using a highly repeatable kicking simulator. The study highlighted the uncertainties involved and particular attention was given to the fidelity of the spin measurement. Keywords: High speed video, launch characteristics, soccer, spin, measurement systems.
1- Introduction The sports ball market is very competitive; valued in excess of $1200 million in the US alone, comprising of £100 million in soccer ball sales [SG1]. In order to research and develop sport balls it is vital to quantitively measure the launch and flight characteristics 1. Sports Technology Institute, Loughborough Science & Enterprise Park, Loughborough, Oakwood Drive, Leicestershire, UK E-mail: J.A.Ronkainen, C.E.Holmes, A.R.Harland,
[email protected] 240 The Engineering of Sport 7 - Vol. 1 of the ball. OEM’s are currently using a wide range of systems to measure these parameters, allowing direct comparison between products. It is fair to state that almost all soccer ball manufacturers have their own testing protocols in place to develop their latest range of soccer balls. Initially OEM’s are concerned with the ball achieving the FIFA ‘approved’ or ‘inspected’ insignia, however once they have achieved this status, they must design and develop their balls to outperform the competition. This study compares some of the most advanced methods currently available to measure soccer ball launch characteristics, these devices are essential in benchmarking balls so that future modifications can be objectively assessed against predecessors or competitor products.
2- Methodology The four main pieces of equipment used for this study are outlined in detail, highlighting the strengths and weaknesses of each device. The testing protocol is clearly defined.
2.1- Equipment 2.1.1- High Speed Video Camera The initial flight of the soccer ball, after impact with the mechanical kicking simulator, was captured using a Photron APX HSV camera. The camera was operating at a resolution of 1024 512 pixels, recording at 1000 fps with a shutter speed of 1/3000 seconds in order to improve the clarity of the image. A voltage output from the mechanical kicking simulator was used to trigger the camera at a known time interval, ensuring the capture of the entire ball flight. Figure 1 depicts a composite image of the soccer ball (20 ms apart) in flight during a straight kick (a) and curve kick (b). The strength of the system is that the HSV is a well established technique for launch measurement. Weaknesses include the manual nature of the digitisation process and using one camera only allows 2D analysis.
Figure 1 - Composite high speed images of soccer ball in flight, (a) straight kick, (b) curve kick.
2.1.2- Optical system This system, as shown in Figure 2, was purpose built for soccer ball launch characteristic measurement. The optical system which has previously been described in detail [NJ1 & NJ2], is an image based acquisition system. The device captures two images of the soccer ball; (1) an image of a static ball at approximately 1.25 metre stand off distance and (2)
The Engineering of Sport 7 - Vol. 1
241
an image of the ball, triggered by the impact sound, 6 ms after impact. The system calculates launch angle and velocity of the ball using elementary image processing techniques. The ball is marked with 50 dots; 25 blue and 25 yellow, the dots are arranged so that the relationship between each one is unique. It is inconsequential at what orientation the ball is viewed, avoiding the need to pre-align the ball. The spin rate is calculated by the system locating 7 dots on the ball, in each acquired image. In conjunction with the ball launch characteristics measurement, the system also captures the player’s kicking technique using a video acquisition device. The strengths of the system are that it can calculate the ball velocity, launch angle, spin rate and spin axis of the ball. The main weakness of the system is, in order to calculate the spin rate and spin axis, the fiducials must be present on the ball.
Figure 2 - The optical system.
2.1.3 Radar system The technology has descended from missile tracking applications, utilising Doppler radar technology, operating at 10.5 GHz bandwidth. The system is able to locate the ball in 3D space using the monopulse principle [R1], using one micro wave emitter and three receivers. A continuous measure is achieved, so that the position of the ball is located throughout the flight of the ball. The system is able to calculate the ball velocity, vertical and horizontal launch angle and spin rate. The lift and drag coefficients of the ball are plotted for the ball throughout flight. The system has been primarily designed for golf measurement, and this is the first time the radar based system is reported for soccer ball flight measurement. The prodigious strength of the system is that it is able to measure the ball throughout flight, the system is effortless to setup and once acquisition is initiated data is recorded indefinitely. A weakness in the system is due to the low spin rates achieved in soccer, the manufacturer claims that spin rates under 300 RPM are not demodulated accurately by the system. The system outputs total spin rate but spin axis is not defined.
242 The Engineering of Sport 7 - Vol. 1
2.1.4 Mechanical kicking simulator Athletes have been extensively used by manufacturers during the development and testing of sporting equipment. However athletes will tire during testing and therefore provide inconsistent performance. Mechanical simulators have been developed to obviate the issue arising from athlete fatigue and have been used since 1926. “Smart structures have been used increasingly over the last four decades to recreate consistent athlete/sports motions which offer increased repeatability” [H1]. Mechanical simulators are now used by manufacturers, governing bodies and academic institutes to provide accurate and repeatable simulations. In order to produce soccer ball impact conditions accurately and reliably, the simulator, as shown in Figure 3, consists of a simple leg rotation device, designed around a rigid A-frame, manufactured from 50 50mm steel box section. The adjustable ball teeing system is capable of manipulation about three axes, allowing the creation of different flight characteristics. The drive system uses a Lenze 6.9 kW asynchronous geared servo motor to rotate the kicking leg which is made from two aerospace grade aluminium plates. The leg impacter is capable of accelerating to a maximum velocity of 37 ms-1 in 270°. The machine’s software allows the length and rate of acceleration and deceleration to be specified, allowing different impact conditions to be achieved. The maximum ball velocity achieved during preliminary testing was 45ms-1, with the repeatability of the leg speed calculated to 0.032 ms-1 (2 SD).
Figure 3 - CAD images depicting the mechanical kicking simulator.
2.2- Protocol The purpose of this investigation was to compare three current measurement systems. All systems were tested simultaneously, using the mechanical kicking simulator impacting the ball at 16, 20 and 23 ms-1 leg velocity. A straight kick and curve kick were recreated by the mechanical kicking simulator by varying the position of the ball at impact, using the adjustable teeing mechanism, as shown in Figure 4. Five elite dot marked soccer balls, inflated to a pressure of 1 bar, were impacted a total of 10 times, for each leg velocity and kick type, with the initial launch characteristics measured using each system. Figure 5. shows the test and instrumentation arrangement. The radar and optical systems use a series of inbuilt algorithms within the analysis software to produce launch data, whilst the recording from the HSV was later digitised in order to allow comparison.
The Engineering of Sport 7 - Vol. 1
243
Figure 4 - Impact position for (a) straight kick, (b) curve kick.
Figure 5 - Plan view of testing protocol.
2.3 Analysis The HSV recording was manually digitised using Image Pro Plus, which determined the x and y pixel co-ordinate values for the centre of the ball at two positions a known time interval apart, by encompassing the circumference of the ball with a circle. The centre co-ordinate positions were used to calculate the ball velocity and elevation angle in the plane of the camera. The soccer balls were marked with a circumferential line which allowed the spin rate to be determined by analysing the number of frames taken to complete half a revolution. The uncertainty of the high speed video measurement Q Total can be defined using Equation 1. The uncertainty of the device Q Device was assumed to be very small by comparison to other uncertainties that exists within the measurement procedure. The
244 The Engineering of Sport 7 - Vol. 1 uncertainty of the setup Q Setup was attributed to the ball flight not being perpendicular to the camera placement. Q Setup was difficult to measure and was controlled by accurate alignment of the camera position, and mechanical kicking simulator. Uncertainty of the analysis Q Analysis can be described as a measure of repeatability. The repeatability of the analysis was calculated by analysing the same image a number of times and defined for velocity as ±1.01 ms-1 (95% confidence) and for launch angle ±0.14° (95% confidence). (1) Equation 1 - Uncertainty of the measurement procedure.
3- Results and Discussion The results were achieved successfully and the weather conditions were, dry and cool (10 to 13°C) throughout the testing protocol. Figure 6 - 7 display the mean ± 1 S.D for the velocity and elevation angle, measurements during straight kicks (a) and the curve kicks (b). Figure 8 displays each successful spin measurement taken during the entire testing protocol, for the straight kicks (a) and the curve kicks (b).
Figure 6 - Launch velocity measurements (average ± SD), (a) straight kick, (b) curve kick.
All the results in Figure 6 consistent indicated by low standard deviations, although a slight increase was observed for the curve kicks. The ball velocities are considerably lower for the curve kick than the straight kick, due to leg energy being transferred into rotational and translational energy. As the leg velocity of the robot increases, a consequent increase was observed in the ball velocity. A drastic reduction in measured velocity, was observed with the HSV measurement during the curve kick. This was accounted due to the HSV measuring 2D velocity, observed in Figure 1 (b), as the ball travels out of plane in the camera field of view. The optical system measured considerably higher velocities than the other two systems, which was due to the system acquiring two images; one static image and one dynamic image, therefore the impact point in time must be known precisely, otherwise errors are induced in velocity calculations.
The Engineering of Sport 7 - Vol. 1
245
Figure 7 - Launch angle measurement (average ± SD), (a) straight kick, (b) curve kick.
The HSV and optical systems recorded statistically similar results for the straight kick launch angle at the 95% confidence level, as shown in Figure 7. This was calculated using a one way ANOVA test and this was the only comparison to provide statistically similar results between the measurement systems. Similar results were achieved because both systems measured the launch angle in an identical way. It was noticeable that the curve kick results show a larger launch angle for the optical and HSV measurements, this was due to the alteration of the ball teeing position, in order to replicate a realistic curve kick. Due to the position of the radar system, it was not possible to measure the ball immediately after impact, which accounts for the low measured launch angle, as the ball was approaching its apex.
Figure 8 - Spin rate measurements, (a) straight kick, (b) curve kick.
Figure 8 highlights the fidelity of the soccer ball spin rate measurement and shows the results exhibit considerable variation. The radar system appears to be the worst performer regarding spin rate measurement, highlighted during the 20 ms-1 straight kick, where the spin rate measurement ranged from ~170 to 650 RPM. The HSV results show the most consistency, however spin rate was only measured around one axis, which was assumed to give a good representation for the straight kick. For the curve kick compound spin was placed on the ball and the HSV method will underestimate the spin rate. For this reason the optical system was expected to give the most accurate spin rate measurements for the curve kick. It must be noted that the systems all measured spin differently; the HSV measured spin around one axis only, the radar method measured total spin rate and the optical system measured total spin rate and spin axis.
246 The Engineering of Sport 7 - Vol. 1
4- Conclusions Uniform results were achieved for launch velocity and launch angle for all measurement systems. For comparative studies all the measurement systems give consistent results, therefore all systems could be used for ball dynamic measurements. However differences were observed between all systems, therefore users should be cautious when using the methods for benchmarking. If a new system is acquired for development purposes, it is recommended that a calibration test is carried out so that results from both systems can be compared. The test outlined here is recommended for this purpose. The only results between systems to show significant results were between the HSV and optical systems for straight kick launch angle measurement. The most difficult variable to measure was the spin rate, which was proven by the large variation in results. This work highlighted the need to develop more advanced spin rate measurement techniques with traceability. The optical system was proven to be a useful tool for soccer ball launch characteristics measurement, however this work recommends that both images captured for analysis are acquired post launch, as it would remove rotational and translational accelerations associated with the impact. This is the first comprehensive study to compare current soccer ball launch measurement systems using a highly repeatable kicking simulator. The results have shown the difficulty in measuring ball launch characteristics in a consistent and accurate manner, therefore an industry standard is required in order to calibrate new and existing measurement systems.
5- References [H1] Harper, T., 2007. Robotic simulation of golf swings. PhD thesis, Loughborough University, UK. [NJ1] Neilson, P., Jones, R., Kerr, D., and Sumter, C., 2004. An automated system for the measurement of soccer ball flight characteristics. The Engineering of Sport, 5, (2), pp. 180-6. [NJ2] Neilson, P., Jones, R., Kerr, D., and Sumter, C., 2004. An image recognition system for the measurement of soccer ball spin characteristics. Institute of physics publishing, Measurement Science and Technology, 15, pp. 2239-47. [R1] Rhodes, D., R., 1959. Introduction to monopoles. McGraw-Hill. [SG1] Sporting Goods Manufacturing Association, (SGMA)., 2007. Manufacturers sales by category report, US wholesale value of annual manufacturers’ shipments ($million).
Testing Protocol for Quantitative Comparison of Top of the Range Soccer Boots (P45) Jouni Ronkainen1, Dan Toon1, Joe Santry2, Tom Waller1
Topics: Soccer. Abstract: Soccer manufacturers are spending increased amounts of time and money developing their soccer boots. Claims such as more grip, more control and more speed have been used in advertising campaigns and no doubt will be used again in the future. The proof behind such claims is often not realistic to match play situations therefore this investigation set out to compare three top of the range branded boots and one pair of prototype boots using University level players to strike an instep swerving free kick. Subjective player feedback was given regards to the comfort and feel of each boot and how it felt to strike the ball. Objective results were obtained for each strike using a soccer launch monitor, used to measure the launch characteristics for each kick, allowing a direct measure of ball velocity, launch angle and spin rate. Player testing was used in order to achieve feedback on the boots, this presented an inherent problem of inconsistent strikes. Therefore ten repeats of each kick were carried out and no miss kicks were recorded. Since soccer is predominantly played outdoors, the weather conditions can drastically influence the players kicking ability. In wet conditions an aquaplaning effect is observed when the boot contacts the ball reducing the amount of grip between boot and ball. Therefore it was vital to test the boots in wet and dry conditions. The results obtained showed clear trends, in the wet conditions less spin and less velocity were imparted on the ball. No significant differences were observed between the launch characteristics measured between the different boots, suggesting that boots performed similarly, however distinguishing factors such as comfort, feel, aesthetics and branding were deemed very important by the players. Keywords: Soccer, soccer boots, launch characteristics, comparative study.
1. Sports Technology Institute, Loughborough Science & Enterprise Park, Loughborough, Oakwood Drive, Leicestershire, UK E-mail: J.A.Ronkainen, D.Toon,
[email protected] 2. Progressive Sports Loughborough Science & Enterprise Park, Loughborough, Oakwood Drive, Leicestershire, UK E-mail:
[email protected] 248 The Engineering of Sport 7 - Vol. 1
1- Introduction The governing body for soccer, the Fédération Internationale de Football Association (FIFA) has 208 national affiliated associations, consisting of more member states than the United Nations. FIFA contains confederations on 6 continents which act as umbrella organisations, thus the global presence of the game is undoubted. A key element of the game is the footwear that the player’s play in. In the US alone, in 2006, the wholesale value of soccer footwear was $296 million [SG1]. Soccer manufacturers are spending increased amount of time and money developing their soccer boots, in order to maintain competitive advantage. Claims such as more grip, more control and more power have been used in advertising campaigns and no doubt will be used again in the future. The proof behinf such claims is often not realistic to match play situations, therefore this investigation sets out to compare three top of the range branded soccer boots as well as a pair of novel prototype boots. It is not questioned that the sports industry is marketing orientated, therefore boot contracts with players such as Ronaldinho are lucrative for the players and the companies alike, however it is the performance aspect of boots that are investigated here.
2- Methodology Four boot types (A-D) from four different brands were selected for inclusion in the investigation. Three of the four boots were commercially available and considered market leading. Boot selection was based on retail price point and advice from players and coaches. An additional prototype boot was also included in the current investigation. The testing was split into two different categories: (1) a mechanical friction testing protocol was carried out in order to deduce if the grip of the material influences factors such as more control and more grip. (2) a player testing protocol was devised in order to achieve objective measurements of players striking the ball. The kick was designed to replicate a free kick scenario, considered a key set piece in soccer. The different testing protocols are outlined as follows.
2.1 Mechanical friction testing protocol Static and dynamic frictional resistance of the four soccer boot uppers (uppers A, B, C and D) were determined to assess the force necessary to initiate sample movement, and the force required to sustain sample movement respectively. A weighted sled with the attached boot sample was pulled a distance of 100 mm (BSISO: 15113-2005) across a surface constructed from a net form FIFA approved football (2007 Umbro England X ball). The ball was mounted upon the bed of a steel friction test rig. The force required to pull the sled and its displacement were recorded at 40 ms intervals using a Lloyd LRX constant rate extension machine, the resolution of measurement was stated at ± 0.02 Newtons. The friction testing setup is shown in Figure 1.
The Engineering of Sport 7 - Vol. 1
249
Figure 1 - Friction testing setup.
The boot samples were attached to the sled base using Cyanoacrylate adhesive, the front of the sample was rolled over the edge of the sled to reduce the amplified mechanical friction that would be expected from an edge interaction. A normal load placed upon the sled held the boot sample against the ball base during the test. Two displacement speeds were used during the testing, 50 mm·min-1 (BS-ISO: 15113-2005; speed 1) and 1000 mm·min-1 (speed 2). BS-ISO: 15113-2005 stipulates that a velocity of 1000 mm·min-1 (the maximum of the machine) may cause frictional heating consequently altering frictional properties. It must be noted that speeds of this magnitude and much higher are expected during matchplay. The friction tests were repeated in wet conditions; the surface of the ball was lightly sprayed with water from an atomised spray. Each individual test combination was repeated 5 times.
2.2 Player testing protocol Three players were selected to participate in the current investigation (age 23.7 ± 3.2 years, shoe size UK 9). The players selected had extensive competitive experience at University level and above. All testing was carried out on a long pile (65mm) third generation in-fill (sand and rubber) artificial pitch. Prior to testing a health screening questionnaire was completed and informed written consent was obtained in accordance with Loughborough University’s ethical advisory regulations. The test procedure was outlined to the players at the start of each test. The players were asked to perform a swerve kick from the instep of the foot. The kick was chosen since out of the 147 goals scored in the 2006 World Cup in Germany, 33% of the goals were scored from set pieces [AY1]. Set pieces are predominantly carried out using the instep kick. A typical instep kick consists of controlled components of spin and power,
250 The Engineering of Sport 7 - Vol. 1 but each athlete was encouraged to perform the kick in their own style to facilitate individual kicking reproducibility. The ball was positioned at a fixed point before each kick and targets were arranged in the field at fixed heights to promote consistent kicks with curved flight. The main marker was positioned so that when the ball travelled past it, the ball launch angle was circa 15°, in order to simulate a free kick going over the wall. The players wore their own boot on the supporting foot as a control boot, and the test boot on the kicking foot; this allowed for any differences in the outsoles of the boots to be eliminated. Soccer is predominantly played outdoors, the weather conditions can drastically influence the players kicking ability. In wet conditions an aqua planning effect is observed when the boot contacts the ball reducing the amount of grip between boot and ball. Therefore the test procedure was repeated for both wet and dry conditions. The wet conditions were simulated by lightly coating the kicking surface of the boot with a fixed amount of water using an atomising spray and dipping the ball in water prior to each kick. Boots were presented to the player in randomised order and each player performed ten successful kicks in each boot. The dynamics of ball flight were determined for every kick using a bespoke ball strike launch monitor system (Quinspin, Sports Dynamics Ltd.), positioned 1.25 metres from the soccer ball, perpendicular to the initial direction of travel. Spin rate and ball velocity were analysed for each kick and the data presented is a mean + 1 standard deviation (SD) of ten kicks. The player testing setup is shown in Figure 2. A one-way ANOVA was conducted to determine if any of the boot conditions generated significantly different ball velocity or spin. Accepted levels of significance were set at P0.05 and all statistical analysis was conducted using SPSS v.13.
Figure 2 - Plan view of player testing setup.
The Engineering of Sport 7 - Vol. 1
251
3- Results 3.1 Friction testing protocol Figure 3 illustrates the static and dynamic frictional force (Mean +1 SD) of the boot samples across the flat football surface. Both speeds and conditions are displayed. In Figure 3 (a) upper D shows a significant (P0.05) increase in frictional resistance for all ‘speed 1’ and ‘wet’ conditions against all other samples. However due to its large standard deviation, during tests under the higher speed, it does not show a significant increase over upper B in the condition ‘dry speed 2’. Upper B exhibits significantly (P0.05) higher frictional resistances than upper C and A in all conditions apart from ‘dry speed 1’ where it is only significantly (P0.05) higher than upper A. Upper C demonstrates significant (P0.05) increases in friction over upper A in all conditions excluding the condition ‘dry speed 2’ where no significant difference was observed. The dynamic frictional force between samples showed very similar trends and hierarchies to the static frictional results. The graphed data illustrates lower mean forces and decreased standard deviations throughout the data, as shown in Figure 3 (b). Uppers A, B, and C showed similar percentages of decrease in force and therefore remained in the same hierarchal order as in the static test. However, upper D showed a larger percentage decrease over the other three uppers. This is illustrated with upper B showing significantly (P0.05) higher dynamic frictional resistances in both ‘dry’ conditions over upper D, however no significant difference is seen between the ‘wet’ conditions with these two samples. Upper D still shows significantly (P0.05) higher dynamic frictional forces than uppers C and A.
Figure 3 - Mean frictional forces, (a) Static condition, (b) Dynamic condition.
3.2 Player testing protocol Figure 4 illustrates the mean velocity for all players and all kicks recorded for each of the boots in wet and dry conditions. No significant differences were exhibited between the boots although there is a consistently lower velocity achieved in the wet condition, the mean wet condition velocity was 1.46 ms-1 less than in the dry for all boots. On average, approximately a 6%
252 The Engineering of Sport 7 - Vol. 1 decrease in measured ball velocity was achieved in the wet. Boot B showed the greatest difference between wet and dry ball velocity.
Figure 4 - Mean velocity measurements (Mean + 1 SD).
Figure 5 illustrates the mean spin generation for all players and all kicks recorded for each of the boots in wet and dry conditions. No significant differences were exhibited between the boots although there is a consistently lower spin rate achieved in the wet condition, the mean wet condition spin rate was 82.3 RPM less than in the dry for all boots. Wet spin rates for boot B and boot C were noticeably lower than the equivalent dry spin rates.
Figure 5 - Mean spin rate measurements (Mean + 1 SD).
The Engineering of Sport 7 - Vol. 1
253
4- Discussion The testing protocol was carried out successfully and reports one of the first investigations to measure soccer ball launch characteristics using an automated measurement system. The measured velocities for the instep kick are comparable with previous results, which range from 18 to 31 ms-1 [AA1, N1, P1, RM1], however this is the first study to show the effects on velocity due to wet versus dry conditions of the boot and ball. The prototype boot tested against market leading boots showed that the acquired results were comparable. The large variability in results does not allow for statistically significant results. Player testing is accounted for the large variation in results. In order to obviate this, a mechanical kicking simulator could have been used, currently however they do not mimic foot kinematics satisfactorily and obviously would not allow for player feedback on the boots. It is thought that using professional players more consistent kicks could have been achieved, which could have led to statistical significance between boots. Anecdotal evidence proves that the comfort, feel, aesthetics and branding were deemed important by the players even if the boots performed similarly. Feel of the ball on the foot was important and it was stated that the prototype boot compromised this, even though the strike ‘felt good’. Players stated that when striking the ball with boot B and D they felt as if they ‘imparted more spin’ on the ball, the ‘material feels really rough, grippy’, which correlates with the friction testing findings, however strikes with boot C seemed to give the highest spin rate values. It was assumed that the large friction values would correlate to higher spin rates, since more grip would facilitate longer contact time between foot and ball on impact, allowing more spin to be imparted upon the ball. Ideally a larger number of boots would have been tested, and a bottom of the range boot could have been selected, it is thought that this would have performed statistically worse than the other boots. This would confirm the capability of the testing protocol to measure differences using player testing for velocity and spin measurement. Boot D appeared to be the most consistent performer in both wet and dry conditions regarding velocity and spin rate. This was the only material to elicit an increase in friction conditions when wet.
5- Conclusions The quantitative testing protocol allows the investigator to substantiate claims such as more power, more control and more swerve, since absolute values can be measured for successive strikes of the ball. The test shows interesting findings, clear trends are observed between wet and dry conditions when striking the ball. Noticeably lower launch velocities and spin rates are measured in the wet condition. This is the first study to report such behaviour. Since football is predominantly played outdoors, the testing suggests that footwear selection should perhaps be based on weather condition. Based on this, playing in the dry condition, from a performance point of view, boot B would be recommended, however if playing in wet conditions boot B seemed to be the worst performer. This is another
254 The Engineering of Sport 7 - Vol. 1 consideration for boot manufacturers in the design and development process for new boots. The testing showed that the prototype boot B, that was tested against market leading soccer boots was very comparable in relation to measured performance characteristics. It is therefore concluded that the boot is ready to be released for the mass market.
6- Acknowledgements The authors wish to thank the external collaborator for funding this work, the enthusiasm shown by the soccer players and Mr R. Waters and Mr A. Gray for assistance with the player testing protocol.
7- References [AA1] Asai, T., Akatsuka, T., and Haake, S., 1998. The physics of football. Physics World, 11, (6), pp. 25-7. [AY1] Acar, M., F., Yapicioglu, B., Arikan, N., Yalcin, S., Ates, N., and Ergun, M., 2007. Analysis of goals scored in 2006 World Cup, VIth World Congress on Science and Football, book abstracts, January 16-20th, Turkey. [N1] Neilson, P., J., 2003. The dynamic testing of soccer balls. PhD Thesis. Chapter 4 – Determination of soccer ball performance parameters, pp. 59-74. [P1] Plagenhoef, S., 1971. Patterns of human motion, a cinematographic analysis, Prentice Hall. [RM1] Roberts, E., M., and Metcalfe, A., 1968. Mechanical analysis of kicking. Biomechanics I, Baltimore University Press, pp. 513-9. [SG1] Sporting Goods Manufacturers Association (SGMA)., 2007. Manufacturers sales by category report. s
Development of a Measurement-Prosthesis for a Ski Boot Test Bench (P48) M. Reichel1, A. Haumer1, H. Schretter2, A. Sabo1
Topics: Ski & other Winter Sports; Biomechanics; Materials; Measurement Systems; Shoes; Testing, Prototyping, Benchmarking; Abstract: In cooperation between HTM Sport- und Freizeitgeräte AG (Tyrolia) and the University of Applied Sciences Technikum Wien a ski boot test bench for stiffness measurement was developed. The former implemented prosthesis of the lower leg was only equipped with a hinged ankle joint. To optimize the measurement system a new prosthesis of the lower human leg was designed. Due to significant differences between ski boots for men and for women, prosthesis for male and female has to be distinguished. The differences are the varying sizes of feet for men and women and the different shapes of the lower legs. The foot is a “Greissinger plus Fuß” made by Otto Bock Healthcare Products GmbH with mobility in the forefoot area. The ankle joint of the new prosthesis is based on the human ankle joint, so the axes conform to the axes of the upper and lower ankle joint with anatomically range of motion. The mobility of the joint also conforms to the anatomic range of motion and is limited by elastomers. Furthermore the ankle joint is linked to a tube by an adapter of Otto Bock, which is used to transmit force. A lower leg is positioned around the tube which guarantees a natural fit inside the ski boot. The developed prosthesis was compared to the former prosthesis, at the ski boot test bench. The analysis addresses both the reproducibility of the two prostheses and the interpretation of the differences in measurement results for ski boot stiffness. The results of the study demonstrate that measurements of the new prosthesis are more reproducible and that lateral movement of the ski boot, previously inhibited by the single hinged ankle joint, is now possible with the new prosthesis. Keywords: Ski Boot, Flex, Stiffness, Prosthesis, Ankle Joint, Test Bench.
1. University of Applied Science, Technikum Wien, Sports-Equipment Technology, Vienna, Austria E-mail:
[email protected] 2. HTM Sport- und Freizeitgeräte AG, Schwechat, Austria - E-mail:
[email protected] 256 The Engineering of Sport 7 - Vol. 1
1- Introduction For ski control the ski boot is the most important element beside ski and binding. Forces are needed to turn initiated by the rider are transmitted to the ski via boot and binding. Beside force transmission the ski boot should fulfil several other main criteria. For an average customer the wearing comfort is very important. The more comfortable a boot the softer all materials should be, like cushioning elements and the shell as well as the shaft. This means that a softer boot will deform more than a stiffer one. Manufacturers declare and often indicate this by a flex-index; the higher the index the stiffer the boot. These characteristics of ski boots can be measured in the lab, where boundary conditions can be held more or less constant. For this purpose a ski boot test bench has been developed at HTM Sport- und Freizeitgeräte AG, Schwechat, Austria (Tyrolia) in cooperation with the University of Applied Sciences Technikum Wien [A2]. This test bench is able to measure the applied force, which results in deformation in longitudinal (heel toe) and transversal direction, and the forces as well as the torques in all spatial directions. Furthermore boundary conditions can be changed like temperature, deformation speed, canting adjustments and boot angle in relation to direction of applied force. In the strict sense, the force is applied to a prosthesis of the lower leg, which is positioned into the boot. In principle this prosthesis consists of a foot and a shank connected by a hinge. Tyrolia uses a prosthesis where the foot and the shank are connected by a simple uniaxial hinge joint, which enables only longitudinal forward a backward movements of the shank. This means that no anatomical human movement is possible, which will result in measure errors especially in cases of nonlongitudinal force applications. A further goal is to consider differences between male and female, because female boots have got a softer shank top, mostly a fleece lining, a female last as well as heel wedges due to anatomical conditions. These features should give a higher wearing comfort, especially for the lower female calf onset to maintain good blood circulation. From this point of view, an improved model of a prosthesis should be developed to consider most of above mentioned factors and finally to compare to the old model.
2- Fundamentals The goal is to develop a new prosthesis, consisting again of jointed foot and shank, but the joint should be based on an anatomical human ankle joint. For female a size of “Mondopoint 25” and for male “Mondopoint 27” was chosen, but the male and female prosthesis should also differ in shape of the shank. The ankle joint should be a “standard” ankle joint and used for both male and female prosthesis.
2.1 Ski-boot test bench The ski boot test bench mainly consists of three elements (Figure 1): a base frame, a boot-fixing plate (“binding”) mounted on a 3D force plate (SHUNK) and a lever arm to apply the force to the prosthesis by a servo drive. The applied force is measured by a load cell and the resulting displacement is measured by a path measurement sensor. Figure 1 shows also the calibrated coordination system for torque and force measure-
The Engineering of Sport 7 - Vol. 1
257
ment. The boot is fixed in the x-y-plane, where an angle of +90° to -90° in steps of 5° can be adjusted to simulate more realistic skiing conditions. Force is always applied in xdirection.
Figure 1 - Ski-boot test bench with measurement coordination system (left) and classic “hysteresis” curve of ski boot flex (right)
The arrangement allows measurement of reaction forces and torques between ski and boot as well as applied force and the flex angle of the prosthesis or the boot respectively. For this purpose well known “hysteresis” curves (Figure 1b) can be achieved [S1]. A personal computer is connected to the test bench via an A/D-converter controlled by a “Labview” application to collect all measurement data and store them to a data base.
3- Methods Nowadays several measurement prostheses are available, like e.g. prosthesis with an uniaxial hinge joint (Tyrolia), human anatomic based ankle joint without elastomers [E1] (Sports-Equipment and Material Group at the Technical University of Munich) and the gait analysis ankle joint with elastomers but not for use in ski boots (Technology Development Group of “Fraunhofer Institut”). To develop a human anatomic bases ankle joint with implemented elastomers to simulate joint forces and also useable in ski boots, an anatomical model with axis and positions of upper and lower ankle joint is used (Figure 2, [I1] and [I2]). Considering the movements of upper and lower ankle joint all rotations are monoaxal, but can individually vary in a wide range [A1, D1, G1, F1]. The mean values are an inclination of 79° and deviation of 84° in the upper as well as an inclination of 23° and a deviation of 41° in the lower ankle joint. The distance of the axes is 5mm [I1, I2]. Based on these values an accordingly ankle joint can be constructed with the help of the software “ProEngineer Wildfire 2”. The joint consists of 3 parts, which are a base plate, a middle plate and a top plate. The parts are connected by fit screws which axes correspond to the upper and the lower axes of the ankle joint (Figure 3). The inserted elastomers in the upper ankle joint are of
258 The Engineering of Sport 7 - Vol. 1 diameter 20mm (70 Shore A) and in the lower of diameter 16mm (90 Shore A for outer and 70 Shore A for inner side).
Figure 2 - Superior (a) and lateral (b) view of the axis of the upper and lower ankle joint.
Figure 3 - 3D-view of the ankle-joint (left) and with inserted elastomers (right).
Figure 4 - Assembled measurement prosthesis (left) and its components (right); (1) “Greissinger plus Fuß”, (2) ankle joint with elastomers, (3) adapter to join tube and ankle joint, (4) tube with drilling to adjust shank, (5) fork joint to apply force, (6) adjust pin, (7) milled shank.
The Engineering of Sport 7 - Vol. 1
259
To compare the new developed anatomically based prosthesis (ABP) to the old one uniaxial hinged (UHP), several tests have to be performed: • Reproducibility (7 measurement cycles with Head Edge 9) • Comparison with boot fixation in x-y-plane at 0° • Comparison with boot fixation in x-y-plane at 20° The comparison tests are performed with 3 different boots corresponding to different flex indices: • Boot1 (B1): Head Ezon 8.5 (soft, index = 60) • Boot2 (B2): Head Edge 9 (middle, index = 70) • Boot3 (B3): Head Raptor Supershape (stiff, index = 100)
4- Results For analysing the results the pathway of the servo motor – moving the prosthesis – at 4 defined measurement points is determined; at +100Nm, +200Nm, -100Nm and 200Nm (Figure 5).
Figure 5: Hysteresis curve of ski boot flex measurement; defined points for analysing steps at 100Nm and 200Nm in forward as well as -100Nm and -200Nm in backward direction.
4.1 Reproducibility For the measurement series to determine reproducibility, each prosthesis is used within the boot of middle stiffness (Head Edge 9) in 7 separate measurement cycles. Old (UHP) and new (ABP) prosthesis are measured alternatively with a break of 10min between each cycle to enable complete restoring of boot deformation. Table 1 shows the determined values and deviations at the defined points.
260 The Engineering of Sport 7 - Vol. 1 Table 1 - Values of reproducibility measurement with hinged (UHP) and anatomical (ABP) prosthesis; values in bracket indicate the measurement torque My, M/SD…mean values and standard deviation of 7 cycles.
4.2 Comparison at 0° The comparison measurement between the 2 prostheses at 0° should give the possibility to analyse mainly the behaviour of the ski boot in forward and backward direction (Table 2). Table 2 - Values of comparison measurement at boot fixation of 0° in x-y-plane with hinged (UHP) and anatomical (ABP) prosthesis; values in bracket indicate the measurement torque My, maximum torque Mx and Mz are measured in (for)ward and (back)ward direction.
4.3 Comparison at 20° The comparison measurement between the 2 prostheses at 20° should give the possibility to analyse mainly the behaviour of the ski boot in partially transversal direction (Table 3), like it can occur in onset period of a turn. Table 3 - Values of comparison measurement at boot fixation of 20° in x-y-plane with hinged (UHP) and anatomical (ABP) prosthesis; values in bracket indicate the measurement torque My, maximum torque Mx and Mz are measured in (for)ward and (back)ward direction.
The Engineering of Sport 7 - Vol. 1
261
5- Discussion This work should point out, if there is a need for a new prosthesis based on a human anatomical ankle joint due to measurement errors of a prosthesis with just a uniaxial hinged ankle joint. The new prosthesis is able to fulfil anatomical correct movements which mean that for the ski boot also a freely transversal movement (in y-direction) is possible. This is for lab conditions more close to practical slope behaviour than it is possible with a uniaxial hinged ankle joint. The results of the reproducibility measurements show (Table 1), that the deviations in values of deformation x are lower with the anatomically based prosthesis and therefore the reproducibility is higher. Loading the prosthesis in forward direction at a boot fixation of 0° (Table 2) B3 is the stiffest, B2 is middle and B1 is the softest boot for both types of prosthesis. In backward direction is B2 the stiffest, B1 the middle and B3 the softest boot. In spite of the elastomers the range of motion is higher for ABP than for UHP, probably caused by blocking of UHP. If the torque Mx is high, the shaft of the prosthesis would move utwards but it is blocked by the boot or the joint of the prosthesis. Table 2 shows, that Mx is lower for ABP than for UHP which means, that UHP blocks the naturally movement of the shank and therefore the movement of the shaft of the boot. Also the values of torque Mz is a sign of blocking due to UHP. Measurements with all 3 boots have higher values with UHP than with ABP because ABP enables twisting in z-axis. Table 4 - Values of differences between boot fixation of 20° and 0° in x-y-plane with hinged (UHP) and anatomical (ABP) prosthesis; values in bracket indicate the measurement torque My, M/SD…mean values and standard deviation at each torque value.
Loading the prosthesis in forward direction at a boot fixation of 20° (Table 3) B3 is the stiffest, B2 is middle and B1 is the softest boot for both types of prosthesis. In backward direction is B2 the stiffest, B3 the middle and B1 the softest boot. In spite of the elastomers the range of motion is higher for ABP than for UHP, probably caused by blocking of UHP. At the boot fixation of 20° the difference between UHP and ABP should clearly be seen. Therefore mean differences between fixation of 20° and 0° in displacement of prosthesis at the defined measuring points (200/100/-100/-200Nm) have been worked out (Table 4). The values show, that both types of prosthesis have clear mean differences in forward direction, but the deviation is much higher in UHP which lowers reliability dramatically. In backward direction even the differences are not clear for UHP and also the deviation is very high, which approves the data from forward direction. Mx is similar in UHP and ABP either in forward or in backward direction and Mz is higher in UHP than in ABP, which underlines the block of twisting in UHP (Table 3).
262 The Engineering of Sport 7 - Vol. 1
6- Conclusion Summarising the results it can be stated, that the newly developed human anatomically based prosthesis (ABP) is able to perform measurements with higher reproducibility applied to the ski boot test bench. To analyse ski boots it is necessary to move the measuring prosthesis in all special directions concerning the geometry of ski boots. With the uniaxial hinged prosthesis (UHP) these kind of movements are not or hardly possible because this kind of joint blocks torques in x- and zdirection (Mx, Mz). With ABP shank and shaft movements in all relevant directions can be performed caused by its anatomically based ankle joint.
7- References [A1] Alt, W. W., 2001. Biomechanische Aspekte der Gelenkstabilisierung, 1. Auflage. Geislingen: C. Maurer [A2] Amara, T., 2007: Inbetriebnahme und Evaluierung einer Skischuhprüfmaschine und Entwicklung von Mess- und Auswertungsmethoden. Diplomarbeit. FH Technikum Wien [D1] Dettwyler, M. T., 2005: Biomechanische Untersuchungen und Modellierungen am menschlichen oberen Sprunggelenk im Hinblick auf Athroplastiken. Dissertation. Eidgenössische technische Hochschule Zürich [E1] Ebert, C., 2007: Die Interdisziplinarität in der Sportgeräteentwicklung am Beispiel des alpinen Skischuhs: Suche und Bewertung funktionsrelevanter Konstruktionsparameter. Dissertation. TU München [F1] Faller, A., 1999. Der Körper des Menschen: Einführung in Bau und Funktion, 13. Auflage. Stuttgart: Thieme Verlag [G1] Galik, K., 2002: The effect of design variations on stresses in total ankle arthroplasty. Dissertation. University of Pittsburgh [I1] Inman, V. T., 1976. The joints of the ankle. Baltimore: Williams & Wilkins [I2] Isman, R. E., 1969. Anthropometric studies of the human foot and ankle. Bulletin of Prosthetic Research [S1] Senner, V., 2001: Biomechanische Methoden am Beispiel der Sportgeräteentwicklung. Dissertation. TU München
Development of Multi-platform Instrumented Force Pedals for Track Cycling (P49) Jean-Marc Drouet1, Yvan Champoux1, Sylvain Dorel2
Topics: Bicycle, Measurement Systems. Abstract: The aim of this research was to develop instrumented force pedals that meet the specific requirements of track cycling. Both pedals are instrumented with eight strain gauges and provide the ability to measure normal and tangential pedalling forces. Rotary encoders are used to determine the angular position of the pedals relative to the crank arm and the angular position of the crank arm relative to the bicycle frame. One original feature is that the instrumented pedals can be fitted with interchangeable pedal platforms: the clipless LOOK CX7 and the Shimano 600 (PD-6400) with a toe-clip and strap for sprint and kilometre time trial events. It is therefore possible to measure pedal loads for all track cycling disciplines. The pedals have a very high mechanical resistance in order to withstand the pedal loads produced by track sprinters which are the highest encountered among all cycling disciplines. Their mechanical design allows the pedals to be installed on any crank arm model without requiring any crank modification. With the data acquisition system attached to a modified Camelbak pack carried by the cyclist, the pedals permit on-track pedal load measurements. Post-processing software was developed to calculate derived parameters, which include the effective power, the effectiveness index and two components of the total force: the effective force component that is the component normal to the crank arm and the force component in line with the crank arm. The derived parameter calculations and analysis can be done on site for each leg and allow specific qualities to be evaluated (peak power, peak force, etc.). Typical results for these parameters are presented in this paper. Keywords: instrumented force pedals, track cycling.
1- Introduction The measurement of pedal loads is essential in acquiring a better understanding of the pedalling process as well as providing load data for bicycle design. In designing the proposed instrumented force pedals for track cycling use, specific requirements must be 1. Mechanical Engineering Department, VélUS Group, Université de Sherbrooke, Sherbrooke, Canada E-mail: Jean-Marc.Drouet,
[email protected] 2. Laboratoire de Biomécanique et Physiologie, INSEP, Paris, France - E-mail:
[email protected] 264 The Engineering of Sport 7 - Vol. 1 addressed. One of these requirements is that the instrumented force pedals have a very high mechanical resistance to withstand the pedal loads produced by track sprinters. These loads are the highest encountered among all cycling disciplines. Even with a high load capacity, the pedals must accurately measure the pedal loads throughout the loading range. In order to be useful for all track cycling disciplines, another requirement is that the instrument force pedals be fitted with two different pedal platforms: clipless pedals and, toe-clip and strap pedals. The latter are used in sprint and kilometre time trial events. Also, considering the many different crank arm lengths used in track cycling, it would more practical and cost effective to have an instrumented force pedal design that allows for normal installation on the crank arm rather than to have a design that requires a dedicated crank arm (Alvarez and Vinyolas 1996, Rowe et al. 1998, Reiser et al. 2003). Because on-track situations such as full-power starts from a dead stop, drafting during a team event and riding on a banked track are difficult or impossible to replicate in a laboratory, on-track measurements are required to obtain realistic pedal load data in these situations. Many different pedal dynamometers have been described in the literature. Some of these dynamometers are restricted to laboratory use (Bolourchi and Hull 1985, Boyd et al. 1996, Davis and Hull 1981, Newmiller et al. 1988, Wheeler et al. 1992) while others permit load measurement outside of the laboratory. In the latter category, Alvarez and Vinyolas (1996) proposed a pedal dynamometer for road use based on a Time clipless pedal platform; Rowe et al. (1998) developed a pedal dynamometer for off-road bicycling; and Reiser et al. (2003) proposed instrumented pedals based on Shimano PD-6500 clipless pedals. These three pedal dynamometers require a dedicated crank arm with integrated bearings. Also, the maximum pedal load capacity reported by Rowe et al. (1998) and Reiser et al. (2003) is not high enough for track cycling use. Since none of these previously described pedal dynamometers satisfy the specific aforementioned requirements for track cycling use, the aim of this research was to develop instrumented force pedals that meet these requirements.
2- Methods The instrumented force pedals can be fitted with two interchangeable pedal platforms: the clipless LOOK CX7 platform (LOOK CYCLE International, Nevers, France) and the Shimano 600 (model PD-6400, Shimano Inc., Osaka, Japan) toe-clip and strap platform (Fig. 1a and 1b). In order to ensure that the instrumented force pedals are functionally equivalent to the original pedals, some parts of the original pedals have been used for the construction of the platform-specific bodies. These platforms are bolted to a pedal base assembly (Fig. 2), which consists of three principal elements: the ball bearing assembly, the instrumented spindle and the rotary encoder. The bearing allows the rotation of the instrumented spindle relative to the crank arm. A thin delrin rod is inserted into the hollow of the instrumented spindle and connects the encoder axle to the bearing assembly, thus allowing the encoder to measure the angular position of the instrumented spindle relative to the crank arm.
The Engineering of Sport 7 - Vol. 1
265
Figure 1 - Photographs of the instrumented force pedal with the LOOK CX7 and the Shimano 600 platforms.
Figure 2 - Photograph of the instrumented force pedal base assembly (exploded view).
The instrumented force pedals measure only the mutually orthogonal force components Fx and Fz (Fig. 3). These forces were measured using a total of eight strain gauges on each pedal (Fig. 2). The strain gauges were arranged in two full Wheatstone bridges, one in the x-y plane (Fx) and the other in the y-z plane (Fz). Theoretically, the position of the strain gauges and their interconnection give bridge signals that are independent of the location of forces Fx and Fz as well as insensitive to the unmeasured loads (Fy, Mx, My and Mz). The instrumented force pedals maximum load is 2500 N. In order to withstand the pedal load, a double row angular contact ball bearing with a dynamic load rating of 10600 N was used. High strength materials were also used for the construction
266 The Engineering of Sport 7 - Vol. 1 of the pedals. Heat-treated 17-4 PH stainless steel (yield strength (Sy) = 1250 MPa) was used for the spindle and the bearing assembly. For the two platforms, 7075-T651 aluminium (Sy = 500 MPa) was used. The mass of one instrumented force pedal is 422 g with the LOOK CX7 platform and 512 g with the Shimano 600 platform (original pedals typical mass: 200 g for LOOK and 282 g for Shimano).
Figure 3 - Pedals local coordinates system with three force components (Fx, Fy, Fz), three moment components (Mx, My, Mz), the pedal angle relative to the crank ( ) and the crank angle relative to the bicycle frame () (right pedal shown).
A supplemental rotary encoder located on the bicycle frame was used (Fig. 4) to measure the angular position of the crank arm relative to the bicycle frame (). Two pulleys (transmission ratio: 1:1) and a synchronous belt were used to link the left crank to the encoder.
Figure 4 - Crank angle () measurement system.
The Engineering of Sport 7 - Vol. 1
267
Measurement data from the instrumented pedals and from the encoder located on the bicycle frame are collected by a data acquisition system (model pro7, ISAAC Instruments inc., Chambly, Canada) attached to a modified Camelbak pack carried by the cyclist (Fig. 5). The pedals as well as the encoder located on the bicycle frame are wired to the data acquisition system. The electrical cables are attached to the cyclist’s legs with elastic bands and Velcro fasteners. The mass of the data acquisition system (including the Camelbak pack) is 1.2 kg. Post-processing software was developed using Matlab to calculate derived parameters, which include the effective power, the effectiveness index and two components of the total force: the effective force component that is the component normal to the crank arm and the force component in line with the crank arm. The derived parameter calculations and analysis can be done on site for each leg and allow specific qualities to be evaluated (peak power, peak force, etc.).
Figure 5 - Track cyclist and bicycle equipped with the instrumented force pedals (Shimano 600 toe-clip and strap platform) and data acquisition system attached to a modified Camelbak pack.
Calibration was performed by applying force and moment loads to measure the direct sensitivity of the in-plane loads (Fx and Fz) and both the calibratable and noncalibratable cross-sensitivities (Rowe et al. 1998). The calibratable sensitivity matrix (V/N, normalized for gain and input voltage ) and the non-calibratable sensitivity matrix (V/N or V/Nm, normalized for gain and input voltage) are given by equations (1) and (2) respectively. Moment My was not considered because it is produced by the pedal bearing friction and can be ignored. (1) (2) Using the extreme loading amplitudes indicated in Table 1, the maximum total root mean square error of the instrumented pedal was found to be 1.6% FS (Full Scale) for Fx and 1.5% FS for Fz. The hysteresis was also determined from the calibration data and it was
268 The Engineering of Sport 7 - Vol. 1 found that hysteresis introduced a maximum error of 0.9% FS. The instrumented pedal repeatability and reproducibility have been assessed by successive calibration. They were found to be less than 1% FS. The natural frequency along the x and z axis was determined using the pedal stiffness and assuming half the weight of a 75 kg cyclist clipped onto the pedal. The natural frequencies for both directions were about 125 Hz. The resolution of the rotary encoders mounted on the instrumented force pedals was 0.4º. A zeroing adjustment for both components of force (Fx and Fz) and pedal angle ( ) was carried out before each measurement session. All the signals were acquired at a sampling rate of 1 kHz (USB data acquisition, ISAAC Instruments inc., Chambly, Canada) and stored on a computer.
Table 1 - Pedal loads for the evaluation of the direct sensitivities and the cross-sensitivities.
3- Results Different experimental sessions in field conditions (i.e. on a track) were carried out, demonstrating the ability of these instrumented force pedals to provide a good quantity of relevant information. The present data describe an example of some measured and calculated derived variables obtained for an elite cyclist (i.e. world class sprinter) during a specific 125 m all-out effort. The athlete was asked to perform maximal effort and encouraged to produce the greatest possible acceleration. A starting block was used for the start and the cyclist naturally adopted a standing position throughout the exercise. Following the test session, raw data stored in the acquisition system were uploaded on a computer for subsequent analysis. The typical time course of the raw data (Fx, Fz, pedal angle of the left pedal and crank angle) measured by the device are presented on Fig. 6. Fz reached very high negative values during each downstroke phase and especially at the beginning of the sprint (>2200 N) while non-negligible positive values (almost 400 N) were observed during the upstroke phases. Oscillations between positive and negative values were also observed for Fx but the magnitude remained much lower (between -500 N and +600 N).
Figure 6 - Measured and calculated derived variables for a world class sprinter during a specific 125 m all-out effort.
The Engineering of Sport 7 - Vol. 1
269
Based on Fx, Fz, the pedal angle and the crank angle (Fig. 6), the total force was calculated by trigonometry and resolved into two components: one orthogonal to the crank (effective force) and another along the crank (ineffective force). A time derivative of the crank angle was used to obtain crank angular velocity and hence the power output (i.e. the product of effective torque and crank angular velocity). Evolution of total and effective forces on the left pedal, crank angular velocity and total power output (i.e. on both pedals) is depicted in Fig. 6. The effective force remained positive throughout the test (except during the last 2-3 seconds): although the effective force produced during the downstroke was very high (peak value > 2200 N), the value during the upstroke was also significant (>500 N), which highlights the importance of all zones of the crank cycle in carrying out the maximal acceleration of the system.
4- Discussion One of the design requirements for the instrumented pedals was that they be able to withstand the very high loads applied by track sprinters. As can be imagined, the severe consequences for the cyclist in the event of mechanical failure of the pedals during high force and high-speed runs are such that this design criterion is an important safety issue. Finite elements analysis (FEA) was used to evaluate stress levels in the pedal. The selected ball bearing was also experimentally tested to verify its mechanical resistance. Functional equivalence of the force pedals with the original LOOK CX7 and Shimano 600 pedals was also a design requirement. With the LOOK CX7 platform, the instrumented force pedals allow for normal engagement and disengagement of the LOOK Delta cleat and for full use of the cleat’s floating angle. For the Shimano 600 platform, the original toe-clips are used. The double straps go underneath the pedals and are held securely in place using tie-wrap fasteners. To avoid contact between the right pedal and the track when riding on the steep banking of a track, another consideration was that the pedals be compact. The use of high strength materials permitted us to reduce the pedals size. FEA was also used to ensure that no mechanical interference occurs between the pedals components under maximum load conditions. A consequence of reducing the pedal size was the reduction of pedal mass, which is of importance considering the high accelerations encountered in track cycling. During the design process, FEA was used for the optimization of the area moment of inertia of the instrumented area of the spindle in order to increase sensitivity. It was also used to determine the corner radius at each end of the instrumented area of the spindle. The stress concentration in the vicinity of these radii was taken into account to determine the location of the strain gauges and to numerically ensure low cross-sensitivities. The accuracy of the in-plane forces was established through calibration by evaluating the direct sensitivity and also by measuring the influence of the other load components. Extreme loadings were considered and it was established that the influence of non-measured loads is small. The direct cross-sensitivity between measured forces Fx and Fz is also small and does not contribute significantly to measurement error. For the measured forces, the linearity is very good and the hysteresis is small. The first natural
270 The Engineering of Sport 7 - Vol. 1 frequency of 125 Hz is high enough to assume a dynamic flat response of the instrumented pedals within the operational measured frequency band of interest of 0-30 Hz. The aerodynamic drag of the data acquisition system located on the back of the cyclist was a concern in high-speed runs (over 60 km h-1). Slower times were observed for 200-m sprints during which the cyclist is in a low crouch and the acquisition system is exposed to airflow. The electrical cables connecting the pedals to the acquisition system were attached to the external side of the cyclist’s legs. It was reported that when routed this way, the cables do not interfere with leg movement even at very high pedalling cadence (over 160 rev min-1). The data presented in this paper constitute a typical sample from among the different possibilities allowed by this new device. The device allows researchers and coaches to analyse the data using practical concerns while providing relevant and useful information that reflects not only the muscular capacity but also the technical abilities of the cyclist. Among the technical aspects that can be measured are the maximal effective force and power output, the index of asymmetry and the index of effectiveness.
5- Conclusion In this paper, instrumented force pedals that meet the specific requirements of track cycling were presented. They are functionally equivalent to the original LOOK CX7 and Shimano 600 pedals and provide accurate measurements of the mutually orthogonal force components Fx and Fz with very low cross-sensitivities. Considering that these instrumented force pedals are a research tool as well as a tool for use in the field, multiple perspectives are offered for the future in terms of scientific approaches to training.
6- References [AV1] Alvarez G. and Vinyolas J. A New Bicycle Pedal Design for On-Road Measurement of Cycling Forces. In Journal of Applied Biomechanics, 12(1):130-142, 1996. [BH1] Bolourchi F. and Hull M.L. Measurement of Rider Induced Loads During Simulated Bicycling. In International Journal of Sport Biomechanics, 1(4):308-329, 1985. [BH2] Boyd T., Hull M.L. and Wooten D. An improved accuracy six-load component pedal dynamometer for cycling. In Journal of Biomechanics, 29(8):1105-1110, 1996. [DH1] Davis R.R. and Hull M.L. Measurement of pedal loading in bicycling: II. Analysis and results. In Journal of Biomechanics, 14(12):857-872, 1981. [HD1] Hull M.L. and Davis R.R. Measurement of pedal loading in bicycling: I. Instrumentation. In Journal of Biomechanics, 14(12):843-856, 1981. [NH1] Newmiller J., Hull M.L. and Zajac F.E. A mechanically decoupled two force component bicyle pedal dynamometer. In Journal of Biomechanics, 21(5):375-386, 1988. [RH1] Rowe T., Hull M.L. and Wang E.L. A Pedal Dynamometer for Off-Road Bicycling. In Journal of Biomechanical Engineering, 120(1):160-164, 1998. [RP1] Reiser II R.F., Peterson M.L. and Broker J.P. Instrumented bicycle pedals for dynamic measurement of propulsive cycling loads. In Sport Engineering, 6(1):41-48, 2003.
The Engineering of Sport 7 - Vol. 1
271
[WG1] Wheeler J.B., Gregor R.J. and Broker J.P. A Dual Piezoelectric Bicycle Pedal With Multiple Shoe/Pedal Interface Compatibility. In International Journal of Sport Biomechanics, 8(3):251258, 1992.
In-Situ Measurement of Clipless Cycling Pedal Floating Angles (P51) Yvan Champoux1, Daniel Paré, Jean-Marc Drouet, Denis Rancourt
Topics: Bicycle, Measurement Systems. Abstract: In cycling sports, clipless pedals are used by athletes and dedicated cyclists to attach the shoe to the pedal because this allows efficient energy transfer to the bike. Most clipless pedals now offer a degree of freedom (float) to the shoe around an axis normal to the pedal surface. This feature was originally introduced in an attempt to reduce knee injuries due to overuse. Most studies reporting on the influence of the clipless pedal floating angle on knee injuries have been carried out under laboratory conditions and little is known about the use of the float in real road conditions. This paper is an evaluation of the design and accuracy of a new apparatus that can measure the in-situ clipless cycling pedal floating angle. A high-sensitivity sensor that can measure magnetic field orientation is embedded in a commercial pedal. Small magnets temporarily clipped onto the cleat create the required magnetic field around the sensor. This measurement technology eliminates the need for a physical connection between the sensor and the shoe, thus allowing the pedal to be used normally. Static calibration and a subsequent accuracy check revealed that the angle measurement uncertainty was found to be within a range of ±0.25º with an hysteresis of less than 1% Full Scale. Typical in-situ sample floating angle measurements are included to demonstrate the ability of the instrument to provide useful information. Keywords: Measurement, clipless pedals, floating angle, cycling.
1- Introduction In cycling sports, 41% of overuse injuries/complaints occur at the knee. Several studies have investigated potential knee injury mechanisms in cycling (Ericson et al. 1984) (Ruby, Hull and Hawkins, 1992) ( Ruby et al. 1992) (Bailey et al. 2003). A clipless pedal was first commercialized by LOOK Cycle in 1984. This type of pedal attaches the shoe to the pedal and allows efficient energy transfer to the bike. Modifications of the clipless pedal have included the introduction of an additional degree of freedom to the shoe (float) around an axis normal to the pedal surface in such a way that the foot’s internal 1. Mechanical Engineering Department, VélUS Group, Université de Sherbrooke, Sherbrooke (Quebec) Canada J1K 2R1 E-mail: Yvan.Champoux,Daniel.Pare,Jean-Marc.Drouet,
[email protected] 274 The Engineering of Sport 7 - Vol. 1 and external rotation is allowed within a limited and set range of motion. The foot movement allowed by the float reduces pedal loads and varus/valgus knee moments (Ruby and Hull, 1993) (Boyd et al. 1997). To the authors’ knowledge, the studies reporting on the influence of the clipless pedal floating angle on knee injuries were carried out under laboratory conditions. Consequently, there is no information available on how cyclists actually use the floating angle in different road situations. The goal of this study was to develop and test an apparatus specially designed to provide accurate and reliable monitoring of the in-situ time variation of the floating angle of commercially available clipless pedals. The apparatus needed to be lightweight, small, easy to use and to not perturb the natural motion of the foot while allowing the cyclist to easily clip in and out. The floating allowed by the two commercial pedal models selected for use in this study (LOOK KéO and LOOK PP 336) was limited to an angular rotation range of 9° along an axis normal to the pedal surface, as shown in Fig. 1. The floating axis is located approximately 25 mm in front of the pedal axis. The 0º mark of the floating angle corresponded to the most counter-clockwise position of the floating range as seen from above. The positive floating angle corresponded to a clockwise rotation of either the left or right shoe. Most commercial clipless pedals are generally based on the same functioning principle and use similar components. A cleat is solidly attached to the underside of the shoe. The spring loaded pedal mechanism holds the cleat solidly in place and allows the cyclist to clip in and clip out of the pedal. Most clipless pedals offer a rotational floating along the Z axis but some of them also allocate a few centimetres of freedom along the Y axis.
Figure 1 - Clipless pedal floating representation. View from above the right pedal. Floating is identified by a rotation along the floating axis in –Z direction. Floating angle increases with a clockwise rotation of the shoe.
The Engineering of Sport 7 - Vol. 1
275
Within the floating 0º-9º range, the resistance moment of the pedal along the Z axis is small and is generated by dry friction between the cleat and the pedal. To disengage the foot from the pedal, the cyclist must apply a strong moment along the Z axis to rotate the shoe beyond the resistance-free 0°-9° range up to angles of approximately –16° and +25°. Consequently, at angles of less than 0° or greater than 9° the cyclist experiences external moment loads around the Z axis.
2- Measurement system The commercial clipless pedals were equipped with a Honeywell angular displacement sensor HMC 1501. This sensor is composed of a Wheatstone bridge and a high-resolution low-power magnetic resistance transducer that measures magnetic field angle direction with a resolution of 0.07º.
Figure 2 - Diagram of the apparatus
The bandwidth response is between 0-5 MHz. The transducer is very small and occupies a volume of only 5 mm x 4 mm x 1.2 mm. Unlike incremental encoding devices, the sensor detects absolute position and requires no indexing for proper positional output. The available full-scale output range is 120 mV for a bridge excitation of 5 V. The small transducer was held in place under the pedal with a plastic support glued to the pedal body as shown in Fig. 2. The additional weight of the device for one pedal is of the order of 80 g. A ceramic horseshoe magnet was anchored underneath the cleat with a small aluminium holder. Magnetic north and south poles created a strong magnetic field around the sensor. The rotation of the shoe/cleat modified the magnetic field which is detected by the sensor. The sensor was aligned along the axis of rotation of the shoe/cleat. There was a gap of 4 mm between the magnets and the top surface of the sensor. An ISAAC Instruments Dual action Wheatstone bridge conditioner model MODWBD-101
276 The Engineering of Sport 7 - Vol. 1 (3 mV/V Full Scale) was used to supply the proper signal conditioning to the sensor. An inductive proximity probe with sufficient sensitivity to detect the passage of the crank without requiring target or reflective tape was installed on the frame. This generated a one-pulse-per-revolution synchronisation signal which was used to segment the signal and to calculate cycle averages. A Model v7 Pro ISAAC Instruments Data Acquisition System was used to store the signals on three channels. With a sampling rate of 1 kHz and a memory capacity of 128 Mb, signals can be recorded for over 5 hours. The recorder and its battery pack were mounted on a modified Camelback® hydration pack. Small electric wires connecting the pedals to the recording system were routed along the cyclist’s lower limb. The recorded data was transferred to a computer with a USB connexion for post processing.
3- Calibration The Honeywell HMC 1501 sensor electrical voltage output V is V = VsS sin(2ø)
(1)
where Vs is the supply voltage, S is a constant determined by the material and ø is the angle between the orientation of the sensor and the magnetic field. For such a sinusoidal function at angles near 0°and for a small angle range (15º), a linear behaviour of the sensor can be assumed. For calibration, an accurate protractor was directly attached to a cleat in place of a shoe. Floating angles and sensor outputs were simultaneously recorded and used to determine the instrument’s direct sensitivity. A very good linearity was obtained (R2=0.999) and a nominal sensitivity of 0.45 mV/V/° was measured. Because small angle variations are measured, any unwanted relative displacement (besides the rotation related to the floating angle) between the magnet and the sensor provoked by pedal loadings would perturb the measurements. Static Loads (forces Fx, Fy, Fz; moments Mx) were individually applied to a cleat mounted on the instrument to verify the sensitivity of the measured floating angle to loads applied to the pedal. Moment My was not applied because it is produced by the pedal bearing friction and can thus be neglected. Moment Mz was also not considered because it is directly responsible for the floating angle variations. Using the load ranges shown in Table 1, maximum angle errors respectively for each load were measured. The total root mean square error related to pedal loads was found to be smaller than 0.25°. The hysteresis was also determined from the calibration data; it introduced a maximum error of less than 1% Full Scale. The calibration procedure was repeated at several occasions over a long period of time (more than 2 years) and the maximum variation of the sensitivity was less than 1% guarantying a good reproducibility in the measurements.
The Engineering of Sport 7 - Vol. 1
277
Table 1 - Pedal loads for the evaluation of the instrument’s cross-sensitivity.
4- Sample data for the in-situ floating angle To give examples of typical results and to demonstrate the ability of the floating angle measurement instrument, in-situ loading angles were measured in various conditions as shown in Figure 3. Each single floating cycle is described by its amplitude and angular position. As shown in Fig. 3c, the amplitude of the selected cycle (2.1°) corresponds to the peak-to-peak amplitude over the cycle. The angular position (3.5°) of the cycle corresponds to the central angular value of the cycle. A modified version of a polar plot provides an interesting global representation of a complete run, as shown in Fig. 4. Each cross denotes the amplitude (radial distance from the origin of the plot) and position (angular position) of a single cycle. The white dot in the centre of the grey square indicates the average angular position and amplitude for all the cycles under consideration. The two thin curved lines on both the left and right of the pedal polar plot indicates the zone limits within which the cyclist does not exceed the pedal floating range.
Figure 3 - Measured signals as a function of time. a) Synchronisation signals generated by the trigger signal b) Left pedal floating angle c) Right pedal floating angle.
278 The Engineering of Sport 7 - Vol. 1 The results in Fig 4a) show that for the cyclist being tested during flat road sitting, the right pedal floats near the centre of the available floating range. However, the results for the left pedal indicate that the cyclist consistently exceeds the limit of the available floating range and pushes the spring loaded pedal mechanism, increasing foot loads and creating additional stress to the knees (Ruby and Hull, 1993; Gregersen and Hull, 2003). Figures 4 b) and 4c) show the respective results for climbing in the sitting and standing position. The left foot shows large amplitudes that systematically exceed the range limit of the pedals (0-9°) for climbing.
5- Discussion The goal of this work was to design an instrumented clipless pedal that would provide an accurate measurement of the floating angle under real operating conditions. Using commercial pedals to develop the instrument ensures normal use of the pedal floating in order to obtain realistic measurements. The use of a magnetic sensor also guaranteed that the pedal could be used safely because the cyclist is able to dismount easily if needed. Upon calibration, the direct sensitivity is very linear and consistent and the hysteresis is small. A commercial clipless pedal is not an infinitely rigid structure and a slight deformation of the pedal body was observed when loads were applied. This was identified as the most important contributor to measurement error. Taking this fact into account, the influence of the pedal loading was measured and included in the measurement error. Tolerance variations of the shape of the cleats do not allow an exact repositioning of the magnet on each cleat. It was found that the instrument’s sensitivity is not significantly influenced by the variation in magnet position on the cleat. However, it may create a small angle measurement bias and zeroing is therefore required at each test to eliminate this bias. The measurement principle is based on the evaluation of the direction of a magnetic field. The magnet placed very close to the sensor succeeds in generating a sufficiently strong magnetic field to yield measurements that are not influenced by the surroundings. For example, it was verified that the use of a steel bike or the earth’s magnetic field did not influence the measurement. No significant drift was noted due to barometric pressure or temperature change. No special effort was taken to make the instrument waterproof and consequently, it is not recommended for use in wet conditions. The apparatus requires small wires to be attached along the cyclist’s legs. All of the cyclists tested stated they were not bothered by the wires.
The Engineering of Sport 7 - Vol. 1
279
Figure 4 - Polar plot indicating the amplitude and position of the cleat for each cycle and for the left and right pedals; MA = Mean Amplitude; MP = Mean Position. a) Flat road; Sitting; Number of cycles = 64; Left pedal: MA = 2.3º; MP = 1.4º; Right pedal: MA = 1.0º; MP = 5.2º b) Climbing and sitting; Number of cycles = 48; Left pedal: MA = 2.1º; MP = 5.5º; Right pedal: MA = 1.0º; MP = 6.1º c) Climbing and standing; Number of cycles = 20; Left pedal: MA = 3.4º; MP = 1.9º; Right pedal: MA = 0.9º; MP = 6.2º
280 The Engineering of Sport 7 - Vol. 1
6- Conclusion Although measurements of very small angle variations in the field is a difficult task, the authors were able to demonstrate that the instrumented commercial pedal suitable for in-situ measurement of the floating angle used in these tests provides reliable and accurate results. No other instrument enabling the in-situ quantification of the floating angle of a commercial pedal has yet been reported. This quantification is useful to investigate the influence of power and cadence on the in-situ use (hill climbing, sitting position) of the floating angle. It is also useful for the development of cleat fitting techniques to optimize the use of clipless pedal floating.
7- Acknowledgments The authors wish to thank ISAAC Instrument and LOOK Cycle International for their collaboration. This study received financial support from the National Sciences and Engineering Research Council of Canada (NSERC).
8- References [BM1] Bailey, M.P., Maillardet F.J. and Messenger, N. Kinematics of cycling in relation to anterior knee pain and patellar tendinitis. Journal of Sports Sciences, 21:649-657, 2003. [BN1] Boyd, T.F., Neptune, R.R. and Hull, M.L. Pedal and knee loads using a multi-degree-offreedom pedal platform in cycling. Journal of Biomechanics, 30:505-511, 1997. [EN1] Ericson, M.O., Nisell, R. and Ekholm, J. Varus and valgus loads on the knee joint during ergometer cycling. Scandinavian Journal of Sports Sciences, 6:39-45, 1984. [GH1] Gregersen, C.S. and Hull, M.L. Non-driving intersegmental knee moments in cycling computed using a model that includes three-dimensional kinematics of the shank/foot and the effect of simplifying assumptions. Journal of Biomechanics, 36:803-813, 2003. [RH1] Ruby, P. and Hull, M.L. Response of intersegmental knee loads to foot/pedal platform degrees of freedom in cycling. Journal of Biomechanics, 26:1327-1340, 1993. [RH2] Ruby, P., Hull, M.L. and Hawkins, D. Three-dimensional knee joint loading during seated cycling. Journal of Biomechanics, 25:41-53, 1992. [RH3] Ruby, P., Hull, M.L., Kirby, K.A. and Jenkins, D.W. The effect of lower-limb anatomy on knee loads during seated cycling. Journal of Biomechanics, 25:1195-1207, 1992. [WH1] Wilber, C.A., Holland, G.J., Madison, R.E. and Loy, S.F. An epidemiological analysis of overuse injuries among recreational cyclists. International Journal of Sports Medicine, 16:201206, 1995.
Correlation Between Treadmill Acceleration, Plantar Pressure, and Ground Reaction Force During Running (P52) Alex, J. Y. Lee1, Jia-Hao Chou1, Ying-Fang Liu2, Wei-Hsiu Lin3, Tzyy-Yuang Shiang4
Topics: Biomechanics, Innovation & Design Abstract: The purpose of this study was to investigate the correlation between peak treadmill acceleration (PTA), peak plantar pressure (PPP) and peak ground reaction force (PGRF) during running. Eight active college students (mean age: 21 ± 0.8 yrs; height: 169.9 ± 7.4 cm; weight: 62.5 ± 9.8 kg) wore standardized shoes and ran on a speed calibrated treadmill (95Te, Life Fitness, USA) at seven speeds (1.3, 1.8, 2.2, 2.7, 3.1, 3.6, and 4.0 m/s). PTA was measured by a dual axis accelerometer (MAX2312G/M, MEMSIC, Inc, USA) which plugged into the middle of the treadmill running board, with an MP100 data acquisition system (BIOPACK Systems, Inc, USA). In-shoe PPP and PGRF were measured by a wireless foot pressure measuring system (F Scan Mobile, Tekscan, USA). The PTA, PPP, and PGRF data were recorded for 10 seconds at a sampling rate of 500 Hz for each running speed. PPP and PTA data at different speeds were compared across the range of speeds by a repeated measures one-sample t-test. A standard linear least squares correlation was used to calculate the coefficients of determination (r2) between PTA, PPP, and PGRF. The running speed had a different effect on PTA, PPP, and PGRF at the seven speeds. The PTA, PPP, and PGRF during the fastest speeds (4.0 m/s) increased approximately 425% (2.2 g vs. 0.53 g), 216% (225.7 psi vs. 104.4 psi), and 228% (311.2 kgs vs. 136.6 kgs), respectively, when compared to the slowest speed (1.3 m/s). The coefficients of determination between PTA and PPP, PTA and PGRF, PPP and PGRF were 0.75, 0.78, and 0.91, respectively (p