Copyright is owned by the Author of the thesis. Permission is given for a copy to be downloaded by an individual for the purpose of research and private study only. The thesis may not be reproduced elsewhere without the permission of the Author. Quantification and description of braking during mountain biking using a novel brake power meter A thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Sport & Exercise Science at Massey University, Palmerston North, New Zealand. Matthew Curtis Miller Bachelor of Science in Exercise Science Master of Science in Exercise Science 2017 Student Declaration i Student Declaration I hereby declare that this thesis is my own work and does not, to the best of my knowledge, contain material from any other source unless due acknowledgement is made. This thesis was completed under the guidelines set by Massey University’s College of Health, for the degree of Doctor of Philosophy and has not been submitted for a degree or diploma at any other academic institution. Candidate: _____________________________________ Date: _________________________________________ Foreword ii Foreword When I came to New Zealand to do my PhD, I wasn’t very sure of what I wanted to study. I knew I wanted to study mountain biking, but it’s such a diverse sport with many genres, and I really had no clear direction of where my studies would go. However, one thing was for sure: I was in the right place! My supervisory team had what I felt was a good mixture of specialities within Sport & Exercise Science, and the further I went through my research, the more I understood the perfect mixture of talent surrounding me. Strangely, while this same team had previously paved the way to new ideas in mountain biking research, I was given full liberty to shape my own ideas and make my own mistakes. The brake power meter idea was born during an actual mountain bike competition. I found myself racing against my supervisor, Steve, who was much more fit than myself. As we continued the race and I could hear Steve’s squeaky brakes, I knew the only reason I was able to keep up with him was for not braking myself. Rather than being told it was a silly idea to measure braking for my PhD, I was taught how to apply for funding, given advice on what kind of variables we should measure, and had conversations on how we might run experiments. It was these kinds of events that taught me the depth of expertise and highly innovative scientists I’m surrounded by. Foreword iii I’ve been tested more than I ever expected throughout this process, but have gained knowledge and experience beyond that of sports experiments. Thank you for believing in me. Acknowledgements iv Acknowledgements Thank you to my family and friends for motivating me and assuring me that this process will be worth it in the end. GUNAX LBH GB ZL FGVAXL. Thank you to Giant Bicycles NZ and Crank It Cycles for their generous equipment support in each experimental study and for my own cycling adventures. Thank you to Massey Ventures Limited, Massey University and SENSITIVUS Gauge for supporting the commercial advancement of the Brake Power Meter. This is for the haters. Publications & Presentations v Publications & Presentations Publications Miller, M. C., Fink, P. W., Macdermid, P. W., & Stannard, S. R. (2017). Validation of a normalized brake work algorithm designed to output a single metric to predict non- propulsive mountain bike performance. (IN REVIEW). Miller, M. C., Fink, P. W., Macdermid, P. W., & Stannard, S. R. (2017). Calculation of brake power during skidding in road and off-road cycling conditions. (IN REVIEW). Miller, M. C., Macdermid, P. W., Fink, P. W., & Stannard, S. R. (2017). Magnitude differences in braking variables and their effects on performance when comparing experienced and inexperienced mountain bikers navigating and isolated off-road turn. (IN REVIEW). Miller, M. C., Fink, P. W., Macdermid, P. W., & Stannard, S. R. (2017). Quantification of brake data acquired with a brake power meter during simulated cross-country mountain bike racing. Sports Biomechanics (IN PRESS). Miller, M. C., Fink, P. W., Macdermid, P. W., Perry, B. G., & Stannard, S. R. (2017). Validity of a device designed to measure braking power in bicycle disc brakes. Sports Biomechanics, 1-11. Publications & Presentations vi Macdermid, P. W., Miller, M. C., Fink, P. W., & Stannard, S. R. (2017). The effectiveness of front fork systems at damping accelerations during isolated aspects specific to cross-country mountain biking. Sports Biomechanics, 1-13. Miller, M.C., Macdermid, P. W., Fink, P. W., and Stannard, S. R., (2017). Performance and physiological effects of different descending strategies for cross-country mountain biking. European Journal of Sport Science, 17(3): p. 279-285. Macdermid, P. W., Fink, P. W., Miller, M. C., and Stannard, S., (2017). The impact of uphill cycling and bicycle suspension on downhill performance during cross-country mountain biking. Journal of Sports Sciences, 1-9. Miller, M. C., Macdermid, P., Fink, P., and Stannard, S. Agreement between Powertap, Quarq and Stages power meters for cross-country mountain biking (2016). Sports Technology, p. 1-7. Stannard, S., Macdermid, P., Miller, M. C., and Fink, P., The power of cycling (2015). Movement, Health & Exercise, 4(2). Miller M. C., Macdermid P. W. Ergonomic Strategies Related to Health and Efficiency in Mountain Biking (2015). Journal of Ergonomics 5:e139. Publications & Presentations vii Miller, M. C., & Macdermid, P. (2015). Predictive validity of critical power, the onset of blood lactate and anaerobic capacity for cross-country mountain bike race performance. Sport and Exercise Medicine Open Journal, 1(4), 105-110. Macdermid, P. W , Miller, M. C., Macdermid, F. M., & Fink, P. W. (2015). Tyre Volume and Pressure Effects on Impact Attenuation during Mountain Bike Riding. Shock and Vibration, 10. Miller, M. C., Moir, G. L., & Stannard, S. R. (2014). Validity of using functional threshold power and intermittent power to predict cross-country mountain bike race outcome. Journal of Science and Cycling, 3(1), 16. Publications & Presentations viii Conference Presentations Matthew C. Miller, Philip W. Fink, Paul W. Macdermid, Steven R. Stannard (2017). The utilization of a bicycle brake power meter for cross-country mountain biking. Australian Strength and Conditioning Association, Gold Coast, Australia, 10-12 Nov Matthew C. Miller, Paul W. Macdermid, Philip W. Fink, Steven R. Stannard (2016). Performance and physiological effect of different pacing strategies for mountain bike racing. European College of Sports Science, Vienna, Austria, July 9 Matthew C. Miller, Chad A. Witmer, Gavin L. Moir, Shala E. Davis (2014). The Predictive Validity of Critical Power and Functional Threshold Power for Mountain Bike Race Performance. Tsukuba Summer Institute, Tsukuba University, Japan, July 23 Matthew C. Miller, Chad A. Witmer, Gavin L. Moir, Shala E. Davis (2014). The Predictive Validity of Critical Power and Functional Threshold Power for Mountain Bike Race Performance. ACSM Annual Meeting, Orlando, FL, USA May 27-31 Emily J. Sauers, Matthew C. Miller, Benjamin Sina, Bryce J. Muth, Brandon W. Snyder, Shala E. Davis (2014). Effects of Full-fat and Fat-free Chocolate Milk On Recovery Following Endurance Running. ACSM Annual Meeting, Orlando, FL, USA May 27-31 Abstract ix Abstract Olympic format cross country mountain biking is both physically and technically demanding. The demands of this cycling genre are in contrast to road cycling because of the demanding off-road terrain. With its many obstacles and different surfaces, riders must make their way up and over steep hills a number of times throughout a lap. It’s very easy to be able to measure the performance of the riders on ascending sections of the track thanks to on-the-bike personal power meter that measure the propulsive work rates in the pedals. However, there is currently no commercially available method to assess the way the rider handles the bike on descending sections. This thesis first highlighted the differences in physiological demand of descending on off-road versus on-road (Chapter 4). An interesting finding in Chapter 4 also showed that riders might be able to save energy by adopting a coasting strategy down hills. This caused the researchers to question the bicycle handling attributes that might allow this, which led to the development and validation of a device designed to measure how the rider uses the brakes while riding/racing (Chapter 5). From there, we completed an investigation akin to the early mountain biking descriptive studies (Chapter 6), but instead of focusing on data related to respiratory and metabolic load, the brake power meter was employed. The finding that braking patterns were related to mountain biking performance was not surprising, but being the first team to quantify this was very exciting. Since most of the braking was occurring on the descents in that study, we examined the differences in braking between training groups on an isolated turn (Chapter 7). The finding that inexperienced riders use their brakes differently—and that this results in reduced performance—left no doubt to the importance of braking. From there, we revisited the method used to calculate rear brake power, since current methods led to inaccurate measurement during skidding Abstract x (Chapter 8). This thesis culminated with the exploration of an algorithm that could quickly and easily describe mountain bike descending performance with one single metric (Chapter 9); the hope is that the normalized brake work algorithm should increase the utility of the brake power meter for training purposes and post-competition performance analysis. Overall, this thesis highlights the need, importance and utility of a bicycle brake power meter to assess mountain bike performance. List of Abbreviations xi List of Abbreviations ANOVA – analysis of variance AT – aerobic threshold CP – critical power DH – downhill (descending) terrain EK – kinetic energy F - force FTP - Functional threshold power HR – heart rate I - inertia IP – intermittent power J - joule LT – lactate threshold FLAT – flat terrain m – meters (omega) – angular velocity OBLA – onset of blood lactate r - radius RCP – respiratory compensation point rad - radians RMS – root mean square s - seconds SD – standard deviation t – time List of Abbreviations xii - torque TRIMPS – training impulse UP – uphill (ascending) terrain v - velocity VO2 – volume of oxygen uptake W – watt W1 – anaerobic work capacity XCO-MTB – Olympic format cross-country mountain bike racing Table of Contents xiii Table of Contents Student Declaration .................................................................................................... i Foreword .................................................................................................................. ii Acknowledgements .................................................................................................. iv Publications & Presentations ..................................................................................... v Abstract ................................................................................................................... ix List of Abbreviations ............................................................................................... xi Table of Contents ................................................................................................... xiii List of Figures ....................................................................................................... xvii List of Tables ......................................................................................................... xxi Chapter 1 ...................................................................................................................... 1 INTRODUCTION .................................................................................................... 1 Chapter 2 ...................................................................................................................... 4 REVIEW OF LITERATURE .................................................................................... 4 2.1 Demands of Olympic format cross-country mountain bike racing ......... 4 2.2 Indices of performance ....................................................................... 10 2.3 Performance analysis and technological tools ..................................... 15 2.4 Descending performance .................................................................... 23 Chapter 3 .................................................................................................................... 26 AIMS & STRUCTURE .......................................................................................... 26 3.1 Background ........................................................................................ 26 3.2 Aims .................................................................................................. 27 3.3 Thesis structure .................................................................................. 28 Chapter 4 .................................................................................................................... 34 Table of Contents xiv PERFORMANCE AND PHYSIOLOGICAL EFFECTS OF DIFFERENT DESCENDING STRATEGIES FOR CROSS-COUNTRY MOUNTAIN BIKING . 34 4.1 Abstract ............................................................................................. 34 4.2 Introduction ....................................................................................... 36 4.3 Methods ............................................................................................. 39 4.4 Results ............................................................................................... 43 4.5 Discussion .......................................................................................... 50 4.6 Conclusion ......................................................................................... 53 Chapter 5 .................................................................................................................... 55 VALIDITY OF A DEVICE DESIGNED TO MEASURE BRAKING POWER IN BICYCLE DISC BRAKES ..................................................................................... 55 5.1 Abstract ............................................................................................. 55 5.2 Introduction ....................................................................................... 57 5.3 Methods ............................................................................................. 60 5.4 Results ............................................................................................... 69 5.5 Discussion and Implications ............................................................... 73 5.6 Conclusion ......................................................................................... 79 Chapter 6 .................................................................................................................... 81 QUANTIFICATION OF BRAKE DATA ACQUIRED WITH A BRAKE POWER METER DURING SIMULATED CROSS-COUNTRY MOUNTAIN BIKING ...... 81 6.1 Abstract ............................................................................................. 81 6.2 Introduction ....................................................................................... 83 6.3 Methods ............................................................................................. 86 6.4 Results ............................................................................................... 91 6.5 Discussion and Implications ............................................................... 97 6.6 Conclusion ....................................................................................... 105 Table of Contents xv Chapter 7 .................................................................................................................. 107 MAGNITUDE DIFFERENCES IN BRAKING VARIABLES AND THEIR EFFECTS ON PERFORMANCE WHEN COMPARING EXPERIENCED AND INEXPERIENCED MOUNTAIN BIKERS NAVIGATING AN ISOLATED OFF- ROAD TURN ....................................................................................................... 107 7.1 Abstract ........................................................................................... 107 7.2 Introduction ..................................................................................... 109 7.3 Methods ........................................................................................... 112 7.4 Results ............................................................................................. 115 7.5 Discussion ........................................................................................ 122 7.6 Conclusion ....................................................................................... 127 Chapter 8 .................................................................................................................. 128 CALCULATION OF REAR BRAKE POWER DURING SKIDDING IN ROAD AND OFF-ROAD CYCLING CONDITIONS ................................................................ 128 8.1 Abstract ........................................................................................... 128 8.2 Introduction ..................................................................................... 129 8.3 Methods ........................................................................................... 133 8.4 Results ............................................................................................. 135 8.5 Discussion and Implications ............................................................. 138 8.6 Limitations and Conclusion .............................................................. 139 Chapter 9 .................................................................................................................. 140 VALIDATION OF A NORMALIZED BRAKE WORK ALGORITHM DESIGNED TO OUTPUT A SINGLE METRIC TO PREDICT NON-PROPULSIVE MOUNTAIN BIKE PERFORMANCE ....................................................................................... 140 9.1 Abstract ........................................................................................... 140 9.2 Introduction ..................................................................................... 142 9.3 Methods ........................................................................................... 146 Table of Contents xvi 9.1 Results ............................................................................................. 151 9.2 Discussion and Implications ............................................................. 155 9.3 Conclusion ....................................................................................... 162 Chapter 10 ................................................................................................................ 163 DISCUSSION ....................................................................................................... 163 10.1 General discussion ..................................................................... 163 10.2 Practical applications ................................................................. 167 10.3 Thesis limitations ....................................................................... 174 10.4 Conclusions ............................................................................... 177 Appendices ............................................................................................................... 178 Appendix I ............................................................................................................ 179 References ................................................................................................................ 201 List of Figures xvii List of Figures Figure 2.1 Elevation profile of the XCO-MTB World Championships in Canberra, Australia in 2009 was originally published by Abbiss et al. (2013). .............................. 5 Figure 2.2 Elevation profile, propulsive power output and physiological variables (heart rate and VO2) from one individual during a simulated XCO-MTB race. This figure appeared in Macdermid & Stannard, 2012. ................................................................... 8 Figure 2.3 Example of placement of accelerometers for use in measuring vibrations during cycling. This photo was first published by Macdermid et al. (2014). .................. 9 Figure 2.4 Example of a functional (albeit dirty) propulsive power meter located within the chainring spider of a mountain bike crank set. ....................................................... 17 Figure 2.5 Some of the cycling metrics used by cycling practitioners presented from TrainingPeaks software. It is not expected that all this data is useful to the untrained eye. ................................................................................................................................... 19 Figure 2.6 Power (W; pink), heart rate (BPM; red) and elevation (Feet; grey) collected from one competitive cyclist during actual racing using TrainingPeaks software. ....... 20 Figure 3.1. Thesis structure schematic. ........................................................................ 33 Figure 4.1 Descent profiles with elevation (m) for A. DHC and DHP (1.01 km, -16.3% gradient), and B. DHR (1.03 km, -15.7%). .................................................................. 40 Figure 4.2 Mean ± SD for values of A. Heart Rate (bpm) and B. VO2 (ml·min-1·kg-1) and C. Time (s). ................................................................................................................. 45 Figure 4.3 Mean ± SD of A. heart rate (bpm) and B. VO2 (ml·min-1·kg-1) continuously sampled and reduced to 15 s averages. ........................................................................ 47 Figure 4.4 Comparison between DHC, DHP and DHR for total accelerations. ............... 49 Figure 5.1 Non-drive side view of bike with A: data logger; B: front brake power meter; and C: rear brake power meter. ................................................................................... 62 List of Figures xviii Figure 5.2 Front brake power meter mounted on bicycle fork. A: machined aluminium block mounted to standard bicycle mounting posts; B: stainless steel rod; C: stainless steel fixing pin for load cell; D: load cell housed within aluminium block; E: sliding aluminium block with F: brake caliper mounted; G: magnetometer; H: brake rotor. ...................... 63 Figure 5.3. Correlation between total energy removed from the bicycle-rider system and the change in kinetic energy for A: 348 braking trials completed on a flat, tar-sealed road; and B: 60 braking trials completed on a flat dirt path. ................................................. 72 Figure 5.4. Braking example using both front and rear brakes highlighting measurements of brake torque (A-B), angular velocity of the brake rotor (C-D) and resultant brake power (E-F). Initial velocity was 3.7 m/s slowing a combined mass of 79.2 kg (rider plus cycling gear and bicycle) to 2.6 m/s. Sample duration was 0.9 s with a total brake work of 231 J. Mean brake power was 166 and 106 W for the front and rear brakes, respectively. ..... 76 Figure 5.5. Braking example using front brake only. Mean power was 105 and 6 W for the front and rear brakes, respectively. Sample duration was 2.9 s with a total brake work of 322.0 J. Initial velocity was 3.63 m/slowing a 79.2 kg mass to 1.7 m/s with subsequent ∆EK estimated at 316.8 J. ............................................................................................ 77 Figure 6.1. Lap profile for XCO-MTB time trial. The track was 1,240 m long and had 45 m elevation gain per lap. ............................................................................................. 87 Figure 6.2. A. Brake Work (J), B. Brake Time (s), and C. Brake Power (W) for front (grey) and rear (black) brakes in laps 1-3. ................................................................... 94 Figure 6.3. A. lap time (s) and B. average propulsive power (W) for laps 1-3. ............. 96 Figure 6.4. Example of braking and propulsive data for one participant on lap 1 (A) and lap 3 (B). The participant was a 22-year-old male weighing 68.5 kg with gear, had a VO2max of 73.1 ml/kg/min and was competitive in international races. For lap 1, braking variables were 20342 J, 28.3 s and 719 W for brake work, brake time and brake power, respectively. Average propulsive power was 328 W with resultant lap duration of 263.3 s. For lap 3, braking variables were 18461 J, 27.0 s and 683 W for brake work, brake time and brake power, respectively. Average propulsive power was 306 W with resultant lap duration of 263.8 s. ................................................................................................... 103 Figure 7.1. Schematic representation of the off-road track and turn highlighting profiles for terrain sections (A-D= distance, gradient; A= 14.50 m, -16%; B= 35.80 m, -16%; C= List of Figures xix 13.65 m, 0%; D= 13.65 m, 0%), which were measured with a trundle wheel. Black dots represent boundary for each section. ......................................................................... 119 Figure 7.2. Two-way analyses for relative brake work (J/kg; A), brake time (s; B) and relative brake power (W/kg; C) for Experienced and Inexperienced groups. .............. 120 Figure 7.3. Example trace of relative brake power of one experienced (black) and one inexperienced (grey) riders throughout respective (A-D) sections of the track. The respective mass of each rider (plus cycling gear) was not overly dissimilar (63.6 and 64.8 kg, respectively). Relative brake work, brake time, and relative brake power were 33.36 J/kg, 2.46 s, and 13.57 W/kg for the experienced and 37.16 J/kg, 4.53 s, and 8.21 W/kg for the inexperienced rider, respectively. Individual braking strategies elicited performance time of 14.08 and 15.87 s for the experienced and inexperienced rider, respectively. .............................................................................................................. 121 Figure 8.1. Correlation between the change in kinetic energy and adjusted energy removed from the bicycle-rider system. .................................................................... 136 Figure 9.1. Elevation profile of the descending track used for testing in this study. The total distance was 1.01 km with a total elevation loss of 165 m (average gradient of - 16.3%). This track was chosen based in its previous use which indicated that performance time was not dependent on propulsive work. ............................................................. 147 Figure 9.2. The relationship between Performance Time (s) on the mountain bike descent and A) relative brake work (J/kg); B) brake time (s); C) relative brake power (W/kg); and D) normalized brake work......................................................................................... 153 Figure 9.3. Graphical representation of normalized brake power (1/s) across a 150 m portion of the descent from one trial each by a high-performing and low-performing mountain biker. The normalized brake work was 0.13 and 0.80 for the high-performer and low-performer, respectively. The time to complete this section was 14.72 and 18.22 s for the high- and low-performer, respectively, which equated to 10.91 and 8.23 m/s, respectively. .............................................................................................................. 159 Figure 9.4. Frequency distribution of normalized brake power (1/s) comparing a high- and low-performing mountain biker. Normalized brake power was separated in to 10 bins in 0.05 1/s steps, with values below 0.05 1/s removed from analysis and values above 0.5 1/s combined in the same bin. ................................................................................... 161 List of Figures xx Figure 10.1 The Candidate with an early commercially viable prototype brake power meter. Photo c. David Wiltshire, Massey University. ................................................ 170 Figure 10.2 An example of helmet video recording of the rider's view of the trail and the braking trace completed throughout the trial. ............................................................ 171 List of Tables xxi List of Tables Table 2.1. These regression models were derived from Miller et al. (2014) and indicated the strength of cycling field tests for XCO-MTB race prediction. ................................ 12 Table 5.1. Mean ± SD and range for mean power and derivatives during braking. ....... 70 Table 5.2. Mean ± SD and range for calculated variables of energy loss during braking. ................................................................................................................................... 71 Table 6.1. Descriptive performance data (mean ± SD) per lap of a simulated XCO-MTB time trial. .................................................................................................................... 92 Table 6.2. Multiple regression model incorporating brake time, brake work and relative propulsive power to explain lap time. .......................................................................... 95 Table 7.1. Mean ± SD of performance and braking variables between Experienced and Inexperienced mountain bikers. ................................................................................. 117 Table 7.2. Mean ±SD starting (Sections B-D) and maximum velocity (Sections A-D) for Experienced and Inexperienced mountain bikers. ...................................................... 118 Table 8.1. Mean ± SD and range for calculated variables of energy loss during braking. ................................................................................................................................. 137 Table 9.1. Mean ± SD for performance and braking variables. .................................. 152 Chapter 1: Introduction 1 1. Chapter 1 INTRODUCTION Indices to predict sport performance have largely relied on results of field and laboratory tests due to their ease and reliability of measurement (Miller et al. 2014). Data garnered through these tests has led to fine-tuned physical training programs for athletes and has arguably contributed to the trajectory of world records throughout sporting history (Clarke & Skiba, 2013). While beneficial overall, the laboratory focus has overlooked the dynamic and technical nature of some sports, specifically in Olympic format cross- country mountain bike racing (XCO-MTB). XCO-MTB presents unique challenges compared to laboratory or road cycling, mostly relating to the undulating terrain and variable surface conditions on mountain bike tracks. While 40% of race distance might be uphill (Impellizzeri et al. 2007), races no longer feature one major ascent and descent; rather, the terrain is generally made up of multiple short, steep ascents separated by short, steep descents in succession throughout a lap. Clearly this is at odds with traditional steady-state laboratory correlates traditionally used in testing protocols. The sort of track profile in XCO-MTB requires intense intermittent bouts of propulsive power production on ascending segments, with recovery bouts occurring on the descents (Macdermid & Stannard, 2012). However, physiological demands are complicated by the uneven and variable riding surfaces, which can exacerbate oxygen demand due to the need to recruit additional musculature for vibration Chapter 1: Introduction 2 damping (Macdermid et al. 2014; Macdermid et al. 2015). This in turn reduces the ability to recover during non-pedalling bouts (Macdermid et al. 2017). These external variables lead to relatively steady and elevated heart rate profiles, despite high oscillations in power output (Stapelfeldt et al. 2004). Attempts to reduce the magnitude of the dissociation between work output and physiological response has garnered much research attention lately as scientists seek to utilise advances in technology for the athletes’ benefit (Hurst et al. 2017; Macdermid et al. 2015; Macdermid et al. 2017). Early reviews of the XCO-MTB have indicated that there is a need to quantify the way a rider negotiates terrain (Impellizzeri et al. 2007) due to the notion that this has some effect on performance. Broadly termed ‘technical ability’, it has been noted that some riders can increase their speed on descending sections of the track to make up for lost time on ascents (Mastroianni et al. 2000). While this raises particular health and safety concerns (Aleman et al. 2010), small differences in technical ability can be crucial for performance in XCO- MTB, though this area remains unexplored. The programme of research described in this thesis aims to ascertain the need and utility for technological developments to better understand dynamic XCO-MTB performance. These aims evolved after particularly interesting findings in the first experimental chapter (Chapter 4), which identified the need for a device to quantify mountain bike descending ability. Successive investigations indicated the validity of a purpose-built, novel brake power meter (Chapter 5), highlighted the enhanced understanding of XCO-MTB by collecting variables recorded with the brake power meter during simulated racing Chapter 1: Introduction 3 (Chapter 6), explored controlled and isolated aspects to better understanding of braking variables (Chapter 7), explored a method to correct shortcomings of the device (Chapter 8) and finally, reduced braking performance to a single metric designed to predict descending performance (Chapter 9). The experiments presented in this thesis pioneer the way toward the development of novel devices for cycling performance analysis, ultimately culminating by explaining new, complicated variables with easy-to-understand metrics. Findings are presented as stand- alone research manuscripts which are published or in-review, then compiled within this thesis as a complete program of research towards a Doctor of Philosophy qualification. Chapter 2: Review of Literature 4 2. Chapter 2 REVIEW OF LITERATURE 2.1 Demands of Olympic format cross-country mountain bike racing Competitive cross-country mountain biking is recognized by the International Olympic Committee (IOC) and was first held during the Atlanta games in 1996. Current XCO- MTB racing is governed by the Union Cycliste Internationale (UCI), which rules that competitions should last approximately 90 minutes (Stapelfeldt et al. 2004). To make up the race duration, relatively short (4-9 km) circuits are completed a number of times on a mixture of terrain that is mostly off-road (Lee et al. 2004). The terrain can feature a number of obstacles including roots, rocks, jumps and drops which can be natural or man- made and on flat, ascending or downhill terrain (Figure 2.1). Riding surface can be weather-dependent, meaning that a hard-packed, dry and dusty course could become muddy and slippery after a rain; however, by and large competitions go on as scheduled during inclement weather, with only minor changes made to the number of laps so as to stay within completion time restraints. Tracks may also feature grass, gravel or tar-sealed road sections, and it is expected that top athletes be able to navigate any type of terrain in expert-fashion. Chapter 2: Review of Literature 5 Figure 2.1 Elevation profile of the XCO-MTB World Championships in Canberra, Australia in 2009 was originally published by Abbiss et al. (2013). Races are mass-start, but quickly funnel into sections of track that are only one-rider- wide, which makes passing difficult (Macdermid & Morton, 2010). The 80% rule used in Elite racing governs that any rider who falls to below 80% of the pace of the fastest rider will be pulled from the race. This was made to reduce lapped traffic for the front competitors, but complicates race strategy for racers lining up in the back of very large (100+) rider fields (Macdermid & Morton, 2010). As such, race dynamics demand an all- out effort from the start to gain positional advantage before the first narrow section of track, after which intensity quickly reduces to a more sustainable effort (Viana et al. 2013). Despite this all-out start, elite riders are able to maintain lap time across the duration of the lap regardless of variation of speeds recorded in different parts of the track (Martin et al. 2012). Thus far this lap time maintenance stands as a testament to their peak physical capability. As expected, ascending portions are the slowest parts of the track, with technical ascents being even slower (Martin et al. 2013). Some 40% of the race Chapter 2: Review of Literature 6 distance could be ascending (Lee et al. 2002), which is course dependent. Analysis of the 2009 XCO-MTB World Championships revealed 168 m of elevation gain in every 6.47 km lap (Abbiss et al. 2013). These observations place a premium on ascending ability, which demands a high power:weight ratio (W/kg) (Gregory et al. 2007). The earliest descriptive study by Impellizzeri et al, (2002) observed that 82% of lap time was spent above the lactate threshold (LT), which highlighted the intense nature of the sport. In this study, riders maintained 90% of maximum heart rate (HR) for the duration of the race, which corresponded with 84% of maximum oxygen uptake ( O2max) measured in laboratory testing. Generally, HR recordings appeared to remain steady in the vicinity of the LT throughout races, which was supported by the relatively high levels of blood lactate measured during competitions (Impellizzeri et al. 2007). The LT had already been well-accepted as a strong indicator of endurance and road cycling ability (Sjodin et al. 1981; Coyle et al. 1991), and race-specific data tended to support its importance for XCO-MTB performance. While this may have been the case, the literature fell short of describing the demands of XCO-MTB in comparison with road cycling until the advent of on-the-bike personal power measuring devices (power meters); to that point XCO-MTB was considered to be equivalent to a shorter, intense road race (Fernández- García et al. 2000). Some years later, these early studies were revisited with the addition of power meters, and later, portable breath-by-breath O2 analysers, which increased our understanding of the uniqueness of XCO-MTB. In 2004, Stapelfeldt et al. published the first longitudinal analysis of power profiles from XCO-MTB competitors, indicating that average power output was relatively high (3.6 Chapter 2: Review of Literature 7 W/kg for males) throughout a season’s racing. The authors noted that 22% of power produced during XCO-MTB was supramaximal, which alerted sport scientists to the sport’s variable intensity. Power profiles from XCO-MTB races were contrasted with cyclocross races (Hansen et al. 1999). The authors noted that XCO-MTB race duration likely contributed to overall lower power output when compared with cyclocross racing (where XCO-MTB is slightly longer in duration), placing the intensity somewhere between cyclocross and road cycling with respect to demands. Nevertheless, it was confirmed that periods of high power output (250-500 W) were seen on ascents, which lasted several seconds to several minutes, and that power output was lowest on descents. Macdermid & Stannard (2012) were the first to include respiratory gas analysis in their comprehensive study on the demands of XCO-MTB. HR and power readings reverberated previous work, but the authors additionally highlighted that actual measured VO2 equated to be 77% of maximum values recorded in an exhaustive laboratory test. Throughout these later studies, it became clear that there was a phase-delay between actual propulsive power created through pedalling and the physiological strain placed on the rider, especially on descending sections (Figure 2.2). Chapter 2: Review of Literature 8 Figure 2.2 Elevation profile, propulsive power output and physiological variables (heart rate and VO2) from one individual during a simulated XCO-MTB race. This figure appeared in Macdermid & Stannard, 2012. The group of Hurst et al. (2006) were the first to explore the demands of downhill riding, and noted that the additional work of the upper body contributed to an elevated physiological strain in contrast with power output. Later studies investigated specific attributes of XCO-MTB riding surfaces, and indicated that the body adopts an autonomic strategy to reduce vibration exposure to the head and central nervous system (Macdermid et al. 2014). The bike and axial body are exposed to high frequencies and magnitudes of vibrations during simulated XCO-MTB racing (Figure 2.3), however musculotendonous damping and eccentric contractions reduce exposures before reaching the head and lumbar spine; this damping comes at an increased physiological cost when compared with road cycling (Macdermid et al. 2015). It has been postulated that vibration exposure— and thus damping requirements—increase linearly with velocity, which contends that the increases in both speed and damping requirement within descending track segments may Chapter 2: Review of Literature 9 explain increased physiological strain. Exacerbated oxygen demand has shown significant reduction in cycling economy on downhill sections when compared with ascents, and may reduce the rate of recovery between high-intensity ascents. Researchers have indicated the need to investigate strategies to reduce oxygen demand on descents to help improve overall performance (Macdermid et al. 2016), however attempts to do so have not been realised. Figure 2.3 Example of placement of accelerometers for use in measuring vibrations during cycling. This photo was first published by Macdermid et al. (2014). Chapter 2: Review of Literature 10 2.2 Indices of performance To date, much of the literature related to XCO-MTB has focused on finding the strongest indicator of actual performance measured from laboratory or field tests. The goal of this type of research is to identify attributes or indices separating the best performer from the least performers, and to henceforth utilise training interventions to improve these attributes/indices. As highlighted above, descriptive characteristics of XCO-MTB stalled for some years while scientists searched for additional variables to measure, and these predictive tests appear to be following a similar trajectory. The traditional exhaustive cycle ergometry test has proven to be popular, with different studies investigating various metabolic thresholds and correlating them with actual or simulated XCO-MTB performance. Amongst these it became clear early on that power output at the ventilatory thresholds on the lower end of the intensity spectrum were not suitable determinants of performance (r=-0.37) (Impellizzeri et al. 2005). VO2max proved to be slightly better (r=- 0.46), which is to be expected due to the very high oxygen uptake values previously recorded in XCO-MTB athletes during maximal testing (Impellizzeri et al. 2002; Impellizzeri et al. 2005). High-intensity testing continued to show promise with later investigations supporting the strong relationship with maximum power output (Costa et al. 2008). However, results from these tests overlooked the aerobic nature of the sport and lacked practical resolution because after all, they were completed in the laboratory and not in the field. By further analysing respiratory gas and metabolite concentrations, high-end aerobic metabolic thresholds in the vicinity of the LT and anaerobic threshold (AT) have Chapter 2: Review of Literature 11 repeatedly been shown to relate to strongly to performance and thus continue to be some of the better laboratory indices. Relative power output (W/kg) at the respiratory compensation point (RCP) was one of the first strong significant indicators of performance (r=-0.61) (Impellizzeri et al. 2005). That original data was supported by a small study reporting a significant relationship with performance time and the second LT (Costa et al. 2008). The findings of Gregory et al. (2007) reinforced this, highlighting that relative power output (W/kg) at the individual anaerobic threshold were significantly associated with a simulated XCO-MTB time trial (r=-0.78). In this same study, the researchers went further and looked purely at ascending performance, which was similarly strong (r=-0.75). Furthermore, it was uncovered that these relative power output values were stronger indices than absolute measures due to the effects of gravity on ascending. Prins et al. (2007) noted that relative power output at the onset of blood lactate (OBLA) was significantly related to an XCO-MTB-specific time trial (r=-0.74). This was followed up some years later (Miller et al. 2014), with the authors indicating that the functional threshold power (FTP) field test used to estimate power output at OBLA (Gavin et al. 2012; Allen & Coggan, 2012) was significantly related to race performance (r2=0.74), which was a successful bid to determine widely available and convenient testing methods (Miller et al. 2014). In an exploratory study, Miller et al. (2015) later suggested that relative critical power (CP) may strongly indicate XCO-MTB performance since it demarcates the boundary between aerobic and anaerobic energy production, however the concept has yet to gain momentum in the sport. The majority of these predictive studies have focused mainly on well-known steady-state aerobic or peak anaerobic values, which arguably overlooks the numerous intense periods Chapter 2: Review of Literature 12 within competition. While specific stationary XCO-MTB time trial performance is strongly related to actual performance (r=0.79) (Prins et al. 2007), this type of test is more difficult to measure than laboratory variables, and could be dependent on the course profile. Attributable to the intermittent nature of XCO-MTB, mean power throughout a 5 x 30 s Wingate was shown to indicate performances quite well (Inoue et al. 2012). Following these measurements, researchers developed a novel intermittent field testing protocol (Miller et al. 2014), noting that it predicted performance better than the FTP field test and could be completed in 20 minutes on a stationary trainer, which is appealing in its own right. Table 2.1. These regression models were derived from Miller et al. (2014) and indicated the strength of cycling field tests for XCO-MTB race prediction. Variable Model r r2 MSE Error Relative FTP Race time = 6317.224 + (-655.688 Relative FTP)* 0.858 0.736 92,190 303.6 Relative IP Race Time = 6662.768 + (-598.752 Relative IP)* 0.886 0.786 74,773 273.5 Note: Relative FTP = Functional Threshold Power; Relative IP = Intermittent Power; MSE = mean square error, calculated as the residual sum of squares divided by the degrees of freedom; Error = estimation error, calculated as the square root of MSE. * Both models p < 0.001 Chapter 2: Review of Literature 13 Promising new methodology incorporating multidimensional variables has further enhanced knowledge of XCO-MTB performance indices (Novak et al. 2017). Incorporating video footage from XCO-MTB trails, participants were timed on their ability to choose the fastest section of trail. Results from this test were henceforth utilised in various regression models, and highlighted that a combination of fitness and sport- specific decision-making variables could predict performance better than unidimensional models. However, these models were admittedly weakest when predicting technical and/or downhill sections. To date, little research has focused on predicting ability on downhill sections within XCO-MTB, with research relying on the few descriptive studies on descending performance. This type of research is important because there is an equal amount of elevation gain and loss due to the lap-style format of racing. Recreational participation in downhill mountain biking is physically demanding and has been likened to other moderate intensity exercise, with the magnitude of increase in oxygen uptake and heart rate likened to off-road motorsports (Burr et al. 2012). However, competitive outcomes are not reliant on propulsive power output (Hurst et al. 2006). In both types of riding, there are similarly notable reductions in handgrip strength due to fatigue, and additional energy that is potentially being directed towards concentration. The variation in descending performance has been suggested to be attributable to difference in technical ability (Mastroianni et al. 2000), which was supported by qualitative analysis (Chidley et al. 2014), though reports otherwise remain unquantified. It seems sensible to begins to describe or quantify descending performance—especially given the relative weakness of physical fitness indices—however, the technology to do so is not currently available. It is Chapter 2: Review of Literature 14 possible that these types of analyses may allow better identification and resolution of factors that performance. Chapter 2: Review of Literature 15 2.3 Performance analysis and technological tools The efficacy of training interventions has reliably been monitored by collecting variables such as oxygen uptake, the rate of substrate utilisation, and HR and comparing them to the rates of work done in the laboratory on cycle ergometers. To monitor training sessions, athletes and practitioners adopted metrics such as the training impulse score (TRIMPS) (Banister et al. 1985)—which accounted for exercise time above and below the threshold heart rate—to gain a better understanding of the stress from each session. This model has been adapted over time to weight high-intensity training and differences in sex characteristics (Morton et al. 1997). While these models are not perfect, the ease of collection and calculation—along with the easy-to-interpret single metric—have proven this metric to be valuable. As technology advanced, the advent of on-the-bike personal power meters (Figure 2.4) has revolutionised cycling, allowing training and race data to be collected and monitored across days, months and multiple seasons. To calculate propulsive power, pedalling torque and the angular velocity of the pedals are multiplied, given: (Eq. 2.1) Where, τ is torque, and is angular velocity. The unit for power is the watt (W). Typically, average power output is of interest, and it would be up to the coach or practitioner to decide the time interval of interest. Chapter 2: Review of Literature 16 The total mechanical work done is calculated as the product of average power and time, given: (Eq. 2.2) Where, t is the duration (s). The unit for work is the joule (J). These measurements are reflective of the rate of energy expended for pedalling through a race or section, which is valuable to the coach for understanding training/competition stresses. These measurements help the coach to guide training interventions aimed at improving performance. To this point, measuring propulsive power output has become extremely well-accepted in cycling, mainly due to its ease of collection and strong association with physical fitness (Jobson et al. 2009). This has led to a number of practitioner publications that put readings of power output into an easy-to-understand context for greater utility of data (Allen & Coggan, 2012). An effect of this has been the increasing availability of companies marketing these types of tools; to date there are over a dozen manufacturers marketing propulsive power meters. Chapter 2: Review of Literature 17 Figure 2.4 Example of a functional (albeit dirty) propulsive power meter located within the chainring spider of a mountain bike crank set. One area that remains debated is the accuracy and validity of devices marketed by different power meter manufacturers (Abbiss et al, 2009; Bertucci et al. 2005; Millet et al. 2003; Novak et al. 2015; Sparks et al. 2015). The SRM power meter is often considered the ‘gold standard’ of commercially available power meters (Hurst et al. 2015), which may be attributable to its longevity within the market. This device places strain gauges and accelerometers within the crank’s chainring spider to calculate pedalling torque and angular velocity, respectively, and it has long been used to compare the data collected from devices made by competing manufacturers. Nevertheless, despite the various placements or methods utilised to calculate both torque and angular velocity, algorithm development has allowed devices to attain similar readings (Gardner et al. 2004) regardless of the configuration the electronics are placed. This was of particular Chapter 2: Review of Literature 18 interest at the onset of the present research program; therefore, we undertook the investigation of three makes of power meters during both steady and XCO-MTB riding (Appendix I). This investigation concluded that power meters are equal during steady riding, but small differences arise during field use due to different technologies utilized to measure cadence. It has been surmised that the low error in these devices can reflect the small changes that affect performance in road cycling, and thus the propulsive power data collected remains highly regarded (Paton & Hopkins, 2001). The development of easy-to-understand metrics from propulsive power meters continued on from heart rate models. This has included the complicated Impulse-Response model (Busso et al. 1994; Morton, 1997), which calculates training stress curves and allows an easier understanding or prediction of longitudinal training stresses. Other scientific models, such as the critical power and W’bal model (Clarke & Skiba, 2013) have proven useful to calculate physiological reserves within training and racing situations to allow optimal pacing strategy. At the same time, coach practitioners have taught various additional models to predict peak performances and guide training (Allen & Coggan, 2012), however despite their wide acceptance these have yet to be fully accepted in the scientific community. Nevertheless, these single metrics help increase the utility and efficacy of power meters as training tools. Chapter 2: Review of Literature 19 Figure 2.5 Some of the cycling metrics used by cycling practitioners presented from TrainingPeaks software. It is not expected that all this data is useful to the untrained eye. At the same time, it is common for other types of simple data analyses (Stannard et al. 2015), even though not all have yet to be dissected in the literature. An example of some the metrics coaches may view can be seen in Figure 2.5. More common/well-accepted practices for power data analysis include: the visual inspection of data; normalization of power output (Allen & Coggan, 2012; Jobson et al. 2009); frequency histograms based on intensity zones; comparison of HR and power output over a time course or after repeating sections of a lap; detection of bilateral pedalling asymmetries (Carpes et al. 2010); and analysis of pacing strategy within races and training. The aforementioned analyses have allowed coaches and athletes the ability to fine-tune training and racing programs for peak physical ability C ha pt er 2 : R ev ie w o f L ite ra tu re 20 Fi gu re 2 .6 P ow er ( W ; p in k) , h ea rt ra te ( B PM ; r ed ) an d el ev at io n (F ee t; gr ey ) c ol le ct ed f ro m o ne c om pe tit iv e cy cl ist d ur in g ac tu al r ac in g us in g Tr ai ni ng Pe ak s so ftw ar e. Chapter 2: Review of Literature 21 These types of analyses have similarly become well accepted to fine-tune XCO-MTB performance (Macdermid & Stannard, 2012). An example of the analyses that can be completed can be seen above in Figure 2.6, which is a snapshot of real racing data from a competitive mountain biker. While the recordings from global positioning systems (GPS) lack finite resolution (Wing et al. 2005), a ‘good enough’ approach has been taken when analysing those types of data. As shown in the figure, coaches and athletes can view the elevation profile and corresponding power output; based on known physiological thresholds, HR and power data enhance understanding. Henceforth, the data can be used to determine strengths and weakness of individual cyclists, or to determine pacing strategy. Figure 2.6 highlights the high power output at the start of the race, and a reduction is seen thereafter despite a relatively steady elevated heart rate. One indication this individual analysis may give is that the starting intensity was too high, which is clear due to the fact that this power output was never achieved in the remaining hour of racing. Perhaps by knowing the rider’s physiological thresholds—such as power output at the lactate threshold—strengths or weaknesses may become apparent and can direct training interventions. For example, it very well may be that in this case the athlete was able to maintain power output at 100% of the lactate threshold on the track’s longest ascent, which may be a personal best after 30 minutes spent racing. However, it could be the case that the short ascents in the later part of the race were far below what the athlete has Chapter 2: Review of Literature 22 shown possible in training. In these cases, adjustments could be made to training to focus on improving weaknesses. One aspect of performance that is not apparent with current analysis tools is that of descending. It very well could be that this rider was only able to negotiate the descents on this track by getting off the bike and walking, then pedalling for short sections. This type of riding would be a serious detriment to performance because time would certainly be lost, however this is impossible to see based strictly on propulsive power meter data. Unfortunately, this type of analysis is difficult to quantify and thus left to guesswork. Chapter 2: Review of Literature 23 2.4 Descending performance As XCO-MTB race tracks roll through terrain, laps start and end in the same location. Accordingly, there are equal amount of elevation gain and loss, and thus similar amount of ascending and descending in a lap. However, ascending portions of the track are markedly slower as the riders compete against gravity, and because of this, over 40% of race time is spent ascending. Analysis amongst different groups competing on track has revealed that the best riders spend the least amount of time on technical ascents (Martin et al. 2012). GPS analysis has indicated that the majority of competition time is spent ascending on the track on various segments featuring a variety of steepness and technical demand (Abbiss et al. 2013). It is because of this that climbing ability has been indicated as the most important attribute for XCO-MTB race performance. While this is likely due to differences in physical fitness, it is in reference to these observations that recommendations have been made for riders to focus on technical uphill ability (Abbiss et al. 2013). It comes as no surprise then that most literature has focused on aspects that could explain or improve ascending performance. However, most literature has overlooked the importance of descending ability, demand and performance, with only a few refereed papers covering the topic. It became apparent early on that descending on a mountain bike in itself was not dependent on propulsive work (Hurst & Atkins, 2006), but was rather related to aspects of technical ability (Mastroianni et al. 2000). The group of Chidley et al. (2014) took a qualitative approach to gain an understanding of descending ability, however this required visual inspection of actual riding by an experienced coach. Some of the earlier studies helped explain that the Chapter 2: Review of Literature 24 technical nature of the track can affect the recruitment of upper body musculature, supposed mainly for steering and postural work throughout descents (Hurst et al. 2013). Later studies indicated that vibrations translated from the riding surface through to the body of the rider were particularly high on descents, and may be attributed to the high speeds (Macdermid et al. 2014). These vibrations are dampened by the body through an autonomic strategy adopted to protect the central nervous system (Miller & Macdermid, 2015); largely done by tiny eccentric contractions, this vibration damping comes at a measurable metabolic cost. These reports helped fill the gap in understanding of why heart rate and oxygen uptake remained elevated in XCO-MTB, and further explains the dissociation between propulsive work and physiological strain in XCO-MTB (Macdermid et al. 2015). As highlighted previously, researchers have understood well that descending performance contributed to overall performance—whether from the variations in time for competitors to complete these sections or due to the elevated physiological variables. Qualitative measures have not been able to answer all of the concerns, though they did reinforce the notion that propulsive power was not the answer to understanding these sections. A recent investigation by Macdermid et al. (2016), which highlighted very low cycling economy on descending section of XCO-MTB racing, suggested the following: “…competitive cross-country cyclists should focus on fitness aspects that will allow to recover quicker and develop technical aspects of bike handling…enabling greater efficiency during downhill riding…” Appropriately, it very well could be the case that riders can recover quicker by increasing their bike handling ability, or at the very least, improving bike handling on descents could Chapter 2: Review of Literature 25 improve overall performance. To date however, bike handling remains multivariate, immeasurable and difficult to understand; herein lies the issue. Chapter 3: Aims & Structure 26 3. Chapter 3 AIMS & STRUCTURE 3.1 Background Olympic format cross-country mountain bike racing features an equal amount of elevation gain and loss, though measures of cycling economy are significantly reduced on descents. The ability to navigate these sections more quickly without the addition of propulsive work could reduce overall oxygen uptake, however the quantification of non- propulsive cycling performance is currently inhibited by lack of technology available. With current tools, scientists are able to accurately assess mechanical propulsive work, but there is nothing commercially available to assess the economy by which a cyclist can navigate the bicycle. Chapter 3: Aims & Structure 27 3.2 Aims Given the background of work done in research surrounding XCO-MTB, there are clear pathways to new discoveries. Thus, the aims of this thesis were to: 1) compare the physiological demand of descending on-road versus off-road, and to determine the effects on performance of the elimination of propulsive work during off-road descending; 2) design and validate a device to measure brake power in bicycle disc brakes; 3) quantify data recorded on a brake power meter during simulated cross-country mountain bike racing to understand the effects of braking on performance; 4) compare braking work done between experienced and inexperienced mountain bikers; 5) determine a method that can account for brake work done when the rear wheel is skidding; 6) validate an algorithm that reduces brake data to a single-number metric that can explain variation in descending performance. Chapter 3: Aims & Structure 28 3.3 Thesis structure To address the aims of this thesis, six experimental investigations were devised and conducted. These are written and presented as manuscripts suitable for stand-alone publication in refereed academic journals. Individually, these manuscripts can stand alone, and are either published (in which case the publication is shown in the Appendices) or in review. These are written and presented as such because it makes the dissemination of the information much more practical and reduces the need for much more extraneous formatting and repetition. It is important to note that the direction of these individual manuscripts evolved throughout this thesis where new questions arose. Moreover, the hypotheses in each manuscript were derived from one preceding it, making this stand- alone research sequence follow a communicable story suitable for a doctoral thesis. Often times, these manuscripts were edited to suit specific Journal style (e.g. ‘Discussion & Implications’ for Sports Biomechanics versus ‘Discussion’ for European Journal of Sports Science). Below are individual, general abstracts for each experimental study. Study 1 (Chapter 4): Eight nationally competitive cross-country mountain bike racers completed a ramp-style laboratory test to exhaustion on a cycle ergometer. Through respiratory gas analysis, the power output at the respiratory compensation point was determined. Participants then undertook three experimental trial on a closed mountain bike track which incorporated an ascent completed at power output associated with the respiratory compensation point leading in to a descent that was completed at race pace. All participants rode the same bike that was outfitted with six accelerometers, and also Chapter 3: Aims & Structure 29 wore a portable metabolic gas analyser and heart rate monitor. In a random order, participants firstly descended an off-road track twice; in one instance they were allowed to pedal while they were instructed to coast in the second instance. During the last trial, they descended a forest road that ran parallel to the mountain bike descent, and were instructed to maintain the average velocity of the preceding two trials. Vibration exposure was significantly higher on the mountain bike descent when compared with the road descent, which led to significantly elevated heart rate and oxygen uptake. While vibration exposure was not different between the mountain bike descending pacing strategies, the coasting trial had significantly reduced oxygen uptake, and importantly, did not negatively affect performance. Thus, it was determined that a purposeful reduction in propulsive work during descents could positively benefit recovery during mountain bike competitions without impacting performance. It was henceforth determined that future investigations should seek to understand the efficiency of navigating the bicycle, which could have to do with the way the brakes are used. Study 2 (Chapter 5): A bicycle brake power meter was constructed to measure the work done through the use of the brakes during cycling. One experienced mountain biker completed 408 braking trials on a flat, tar-sealed road (n=348) and flat dirt path (n=60) over five days on the bicycle outfitted with this device. Energy lost to drag and rolling resistance was added to the work done through braking in each braking event; this total energy removed from the bicycle-rider system was compared with the change in kinetic energy of the bicycle-rider system. Analysis highlighted that the change in kinetic energy was not different from the total energy removed from the bicycle-rider system. Similarly, there was a strong linear relationship with these two values. It was therefore determined Chapter 3: Aims & Structure 30 that this bicycle brake power meter was a valid means of assessing the energy removed during braking while cycling, and it was recommended that future research should quantify braking during simulated racing. Study 3 (Chapter 6): Ten nationally competitive mountain bike racers undertook a three- lap simulated cross-country mountain bike time trial using the same mountain bike outfitted with the brake power meter and a propulsive power meter. Despite propulsive work significantly reducing following the first lap, performance time in each lap did not change. Brake power meter data intimated that the maintenance of lap time was due to reductions in brake work and the time spent braking following the first lap. It was surmised that braking efficiency increases after repeating a track at race pace, which counterbalances performance decrements due to fatigue. Most of the braking on the track used in this study occurred on descending and turning sections, and the determination was made to investigate these sections more closely. Study 4 (Chapter 7): A total of 18 volunteers participated in a trial that sought to determine difference in braking and performance between experienced (n=8) and inexperienced (n=10) mountain bikers. A 76.5 m track was taped to 1.5 m wide, which went down a hill on a forest road before leading in to a 180º turn on grass. Participants completed three trials each on the track while utilising a mountain bike outfitted with the brake power meter but no chain to prevent propulsive work. The mean of all variables were calculated from the second and third trials of each participant, and the track was virtually segmented by terrain type post-hoc. Brake power had a significant negative Chapter 3: Aims & Structure 31 association with performance time, which indicated that faster riders were completing brake work over a shorter period of time. Sectional analysis highlighted that experienced riders wait until longer to brake, which increased the time spent moving quickly and ultimately reduced performance time. Moreover, it was discovered that inexperienced riders are heavily reliant on the rear brake, which has a negative impact on overall braking. The findings surmised that inexperienced riders may be able to perform better on descents and turns by learning to use both brakes more evenly and waiting longer to brake. Study 5 (Chapter 8): It had been noted that skidding occurred during mountain biking, which equated to 0.3% of lap time. The methods utilised to calculate power relied on the angular velocity of each wheel; henceforth any time the wheel was skidding there was no work being recorded which was not reflective of actual brake work being done. This study tested the possibility of utilizing the angular velocity of the front wheel to calculate the rear brake power during braking events with a mixture of skidding and non-skidding. One experienced mountain biker completed 28 braking trial with a mixture of skids and non- skids on a mixture of a flat, tar-sealed road and a flat dirt path. Correcting for rear brake power by utilising front wheel angular velocity proved to be a valid means of calculate the change in kinetic energy during skidding and not skidding. It was concluded that rear brake power calculated with the present correction would reflect actual energy lost during braking. Chapter 3: Aims & Structure 32 Study 6 (Chapter 9): The multivariate performance models utilizing brake metric may be difficult to assess for some uses, which could be a barrier to brake power meter utilization. Nine mountain bikers descended mountain bike track three times using a bike outfitted with a brake power meter to determine the efficacy of brake power meter data to describe descending performance with a single metric. Traditional brakes measures were calculated to determine their association with performance; additionally, a normalized brake work algorithm was utilised which calculated brake power as a proportion of kinetic energy. The normalized brake work was significantly associated with performance time (r2=0.929) and had a stronger association than other measures investigated. This study underscored the validity of the normalized brake work algorithm to describe descending performance based on a single braking metric, which should enhance the value of brake power meter data for training purposes. C ha pt er 3 : A im s & S tr uc tu re 33 Fi gu re 3 .1 . T he sis st ru ct ur e sc he m at ic . St ud y 1 O bj ec tiv e: G ai n an un de rs ta nd in g of th e ph ys io lo gi ca l d em an d of X CO -M TB de sc en di ng Fi nd in g: C oa st in g do w n hi lls c an sa ve e ne rg y an d be ne fit pe rf or m an ce St ud y 2 O bj ec tiv e: D es ig n an d va lid at e a de vi ce to m ea su re b ra ki ng Fi nd in g: T he b ra ke po w er m et er is v al id fo r u se d ur in g cy cli ng St ud y 3 O bj ec tiv e: Q ua nt ify br ak in g du rin g XC O - M TB ra cin g Fi nd in g: B ra ki ng a ffe ct s pe rfo rm an ce , a nd m os tly o cc ur s o n de sc en di ng se ct io ns St ud y 4 Ob je ct iv e: Q ua nt ify br ak in g di ffe re nc es be tw ee n ex pe rie nc ed an d in ex pe rie nc ed co ho rt s Fi nd in g: In ex pe rie nc ed rid er s b ra ke lo ng er a nd do m or e br ak e w or k, w hi ch is d et rim en ta l t o pe rfo rm an ce St ud y 5 Ob je ct iv e: V al id at e a m et ho d to ca lcu la te th e en er gy lo ss es du rin g sk id di ng Fi nd in g: B ra ke w or k an d po w er c an b e ca lcu la te d in th e re ar w he el u sin g th e an gu la r v el oc ity o f t he fro nt w he el St ud y 6 O bj ec tiv e: V al id at e an al go rit hm to d es cr ib e de sc en di ng pe rfo rm an ce w ith o ne nu m be r Fi nd in g: N or m al ize d br ak e w or k is st ro ng ly as so cia te d pe rfo rm an ce w hi le de sc en di ng Chapter 4: Mountain bike descending strategies 34 4. Chapter 4 PERFORMANCE AND PHYSIOLOGICAL EFFECTS OF DIFFERENT DESCENDING STRATEGIES FOR CROSS-COUNTRY MOUNTAIN BIKING 4.1 Abstract This study investigated the performance-related feasibility and physiological benefits of purposefully eliminating propulsive work while descending in mountain biking and compared values to those measured during road descending. Participants cycled uphill on a road at race pace before descending over three conditions (off-road pedalling; off-road coasting; road coasting). Relatively low power output during off-road pedalling was associated with a greater oxygen uptake (p<0.001) when compared with off-road coasting despite no difference in vibration exposure (p>0.05). Importantly, pedalling did not invoke a performance benefit (p>0.05) on the descent used in this study. Significantly greater heart rate and oxygen uptake (both p<0.0001) were observed between road and off-road descending, likely caused by the increase in terrain-induced vibrations (p<0.0001) experienced between the bicycle and rider. Results indicate that reducing propulsive work during descending can improve recovery without being disadvantageous to performance. Similarly, the vibrations experienced during road descending are Chapter 4: Mountain bike descending strategies 35 relatively low, and further reduce oxygen cost. In an effort to increase efficiency, it is recommended that mountain bike athletes focus on skills to increase descending speed without the addition of pedalling, and that equipment be used to decrease vibrations nearer to those seen on the road. Chapter 4: Mountain bike descending strategies 36 4.2 Introduction Olympic format cross-country mountain bike racing (XCO-MTB) takes place over varying terrains, with high-intensity high-power ascending sections separated by relatively lower intensity descents. Despite a pattern of highly intermittent mechanical propulsive work (Macdermid & Stannard, 2012), measured physiological variables remain elevated (Impellizzeri et al. 2002; Stapelfeldt et al. 2004), even on the descents, as athletes contend with technical terrain and obstacles (Gregory et al. 2007). During sections where propulsive work requirement is low, such as on the descents, recovery is inhibited due to non-propulsive work from the musculo-tendonous damping of vibrations (Macdermid et al. 2014) as well as steering and postural work of the upper body (Hurst et al. 2012). The lack of change in physiological variables on the downhill sections are in contrast to road cycling where riders are seemingly afforded better recovery in competition (Lucia et al.1999; Padilla et al. 2000) which is likely due to reduced vibration exposure because of the smoother riding surface (Macdermid et al. 2015). Technological systems designed to reduce soft impacts and vibrations (Macdermid, et al. 2014; Macdermid et al. 2017) can potentially reduce non-propulsive work during XCO-MTB (Macdermid et al. 2015), but physiological demands remain higher than those on the road. As vibration exposure may be exacerbated with increases in velocity (Macdermid et al. 2014), descending parts of the course may indicate the greatest magnitude of difference between road and off-road riding, however no attempt has been made to quantify differences in off-road and road descending. Chapter 4: Mountain bike descending strategies 37 Other attempts to maximize energy efficiency in cycling sports have focused on pacing strategies. Effective road cycling time trial pacing strategies utilize increased power on uphill portions and reduced power on descents (Atkinson & Brunskill, 2000; Atkinson et al. 2007) to provide the best performances. These strategies appear to work because competitors stand to gain the least time on faster parts of the course (e.g. descents) and gain the most time on the slowest parts of the course (e.g. ascents). In practice, this strategy enhances recovery while descending to help the athlete exercise at a higher intensity on subsequent ascents. Such a tactic could be recommended to attenuate reduced efficiency and benefit overall performance in XCO-MTB as uphill climbing ability remains most important to race outcome (Abbiss et al. 2013; Martin et al. 2012). This is corroborated by previous work showing that downhill mountain biking has decreased dependence on propulsive work (Hurst & Atkins, 2006) and increased emphasis on rider skill (Chidley et al. 2014; Mastroianni et al. 2000). While these findings tend to suggest that propulsive work could be purposefully reduced during XCO-MTB descending to help the next ascent without contributing to an overall performance decrement, scientific investigations have yet to explore this area specifically. Therefore, this study was conducted to investigate the performance and physiological demand of either pedalling or not pedalling while navigating a XCO-MTB descent under the simulated intensity of a XCO-MTB race. A secondary aim was to compare the off- road strategies to reduce propulsive and non-propulsive work done with a road descent of similar duration and speed. It was hypothesized that off-road coasting would not alter performance but that the reduced propulsive work would lead to a decreased oxygen Chapter 4: Mountain bike descending strategies 38 uptake compared to pedalling. It was also supposed that a reduction in non-propulsive work during the road descent would further reduce oxygen uptake. Chapter 4: Mountain bike descending strategies 39 4.3 Methods Participants, Procedures and Equipment Eight nationally competitive XCO-MTB athletes (mean ± SD: age= 27 ± 7 years, height= 174 ± 8 cm, mass= 70.3 ± 11.6 kg, power at respiratory compensation point (RCP)= 300.1 ± 48.3 W, RCP (percent of maximal oxygen uptake)= 74.9 ± 3.7 %, VO2max= 68.1 ± 9.8 ml·min-1·kg-1) volunteered for this study which was approved by the institutional Human Ethics Committee. Testing involved one session in the laboratory and one session on a purpose built track (Kohitere Forest, Levin, NZ) at least 24 hours later. The laboratory session involved a ramp-style test to exhaustion (Lode Excalibur Sport, NL) commencing at 100 W and increasing by 25 W per minute until the participant could no longer maintain the required power output. Throughout this test heart rate (Polar Electro, Kempele, Finland), expired air (K4b2, Cosmed, Italy) and power output (W) were measured. Expired air data averaged every 15 s enabled calculation of VO2peak and RCP to determine exercise intensity for the hill climb component specific to that reported in race conditions (Lucía et al. 2000) for the field trial. Chapter 4: Mountain bike descending strategies 40 Figure 4.1 Descent profiles with elevation (m) for A. DHC and DHP (1.01 km, -16.3% gradient), and B. DHR (1.03 km, -15.7%). Off-road descents with coasting (DHC) or pedalling (DHP) and road descent (DHR). Participants used the same bike (Anthem 27.5 2, Giant Bicycles, NZ) set at personal preferences for the field trials. All participants were familiar with the track used (1.0 km, -16.3% gradient; Figure 4.1). Following an individualized race-specific warm-up for 15 minutes, participants completed a coasting familiarization trial. Each subsequent test trial began with an ascent at RCP to ensure descents were completed under racing stress. Participants completed in a randomized order the same descent at race pace while either pedalling (DHP) or coasting (DHC). The participants then completed a parallel descent on Chapter 4: Mountain bike descending strategies 41 a forest road of comparable characteristics (DHR; 1.0 km, -15.7% gradient; Figure 1) at the mean velocity of DHP and DHC. Data Recording Power output (SRAM Quarq S2275, USA) (Appendix I) and heart rate (Garmin Edge 510, USA) were continuously sampled and logged every second throughout the field trial, while VO2 (ml·min-1·kg-1) was measured breath by breath on the Cosmed K4b2 and reduced to 15 s averages. Data recorded from the power meter and GPS device were transmitted to a conventional personal computer and processed with the Garmin Training Centre software (version 3.6.5). Wireless, tri-axial, gyroscopic accelerometers (Emerald, APDM, OR, USA) sampled at 128 Hz were used in a synchronised data logging mode to measure accelerations in accordance with methods reported elsewhere in similar studies (Macdermid et al., 2015; Macdermid et al., 2014, Miller & Macdermid 2015). The accelerometers were placed on the lower left arm (frontal distal position); left lower leg (frontal, distal position); seat post (within 10cm of saddle-rider contact area); lumbar region of lower back; and medial forehead (Macdermid et al. 2014). The accelerometers were synchronised with the Garmin Edge 510 and the Cosmed K4b2 enabling specific terrain sections to be marked across all units. Data were analysed using MATLAB R2014b for total (XYZ) accelerations (Macdermid et al. 2014). Chapter 4: Mountain bike descending strategies 42 Statistical Analyses Descriptive data (mean ± SD) for time (s), power (W), HR, and relative oxygen uptake (VO2; ml·min-1·kg-1) separately for the ascent and descent of each trial and were analysed using one-way repeated-measures analysis of variance (ANOVA). Two-way repeated- measures ANOVA was used to compare total RMS data across each accelerometer for each trial (condition*location) as well as to compare changes in heart rate and oxygen uptake throughout each trial (condition*time). Correlations (Pearson’s r) were determined comparing all performance variables. Where significant difference was found the main effect was analysed using Bonferroni post-hoc testing. All statistical analyses were performed using GraphPad Prism 6, significance set at p<0.05. Chapter 4: Mountain bike descending strategies 43 4.4 Results Uphill variables including total acceleration amplitude (F(2, 84)=1.512, p=0.2265), heart rate (F(2, 14)=0.9831, p=0.3986), VO2 (F(2, 14)=0.4187, p=0.6659), power output (F(2, 14)=1.113, p=0.3559) and time (F(2, 14)=0.5568, p=0.5852) were not significantly different between trials, indicating the uphill portion of this study was well controlled and that the participants entered each descending trial in a similar physiological state. Maximum power during DHP was 526.4 ±151.0 W, while average power was 18.8 ±11.6 W from 19.9 ±9.8 s of pedalling. The power recordings during DHP were significantly greater (F(1, 7)=20.82, p=0.0026) than DHC and DHR, during which participants did not pedal and thus recorded values of 0 W. However, analysis of velocity showed no statistically significant difference (DHC= 7.3 ±1.4, DHP= 7.5 ±1.5, DHR= 7.3 ±1.4 m∙s-1; F(2,14)=0.9014, p=0.4283), and was accompanied by no significant difference in time (Figure 2C) to complete each trial (DHC=142.5 ±29.2, DHP=139.9 ±29.2, DHR=145.1 ±29.3; F(1.135, 7.948)=2.615, p=0.1438). Results recorded for VO2 (Figure 4.2B) were different between all trials (F(1.152, 8.062)=131.7, p<0.0001), equalling 31.7 ±3.9, 35.2 ±4.8, and 20.1 ±2.6 ml·min-1·kg-1, corresponding with 47.4 ±6.7, 52.5 ±6.7, and 30.1 ±5.2 % of maximal oxygen uptake for DHC, DHP and DHR, respectively. This was accompanied by significantly different heart rate recordings (DHC= 166.8 ± 7.7, DHP= 171.8 ± 6.6, DHR= 132.0 ± 10.4; F(1.519, 10.63)= 200.3, p<0.0001; Figure 4.2A) between trials, however Chapter 4: Mountain bike descending strategies 44 post-hoc testing revealed there was no significant difference between DHP and DHC (p>0.05). C ha pt er 4 : M ou nt ai n bi ke d es ce nd in g st ra te gi es 45 Fi gu re 4 .2 M ea n ± SD fo r v al ue s o f A . H ea rt R at e (b pm ) a nd B . V O 2 ( m l·m in -1 ·k g-1 ) a nd C . T im e (s ). Po st -h oc a na ly si s $ $$ $ (p <0 .0 00 1) w he n co m pa re d w ith D H P; ** ** (p <0 .0 00 1) w he n co m pa re d w ith D H R , # ## # (p <0 .0 00 1) w he n co m pa re d w ith D H C . O ff -r oa d de sc en ts w ith c oa st in g (D H C ) o r p ed al lin g (D H P) a nd ro ad d es ce nt (D H R ). Chapter 3: Aims & Structure 46 Maximum (p=0.501) and average (p=0.489) power output were not related to time for DHP. Time for DHP was significantly correlated (r=0.957; p<0.001) with time for DHC. Values for oxygen uptake were significantly related between DHP and DHC (r=0.971; p<0.0001) and between DHC and DHR (r=0.708; p=0.049) but not between DHP and DHR (p=0.070). VO2max was related to VO2 in DHR (r=-0.674; p=0.047), but not between any other variables. Analysis of physiological variables averaged every 15 s (Figure 4.3) indicated a significant main effect for condition (F(2, 184)=186.4, p<0.0001) and time (F(9, 184)=25.81, p<0.0001) with a significant interaction between condition*time (F(18, 184)=5.748, p<0.0001) for VO2. Analysis of heart rate also revealed significant main effects for condition (F(2, 183)=272.7, p<0.0001) and time (F(9, 183)=31.29, p<0.0001) and a significant interaction between condition*time (F(18, 183)=10.09, p<0.0001). Post hoc testing of O2 and heart rate (Figure 4.3) identified differences (p<0.05) at the 45 s time point when comparing DHR and DHP, and from the 60 s time point for both DHC and DHP compared with DHR, indicating an increased ability to recover during DHR. Chapter 4: Mountain bike descending strategies 47 Figure 4.3 Mean ± SD of A. heart rate (bpm) and B. VO2 (ml·min-1·kg-1) continuously sampled and reduced to 15 s averages. Post-hoc analysis * (p<0.05), **** (p<0.0001) significant difference between DHC and DHR; ### (p<0.001), #### (p<0.0001) significant difference between DHP and DHR. Off-road descents with coasting (DHC) or pedalling (DHP) and road descent (DHR). Chapter 4: Mountain bike descending strategies 48 Two-way ANOVA of total accelerations (Figure 4.4) revealed a significant main effect of location of accelerometer (F(5, 42)=39.44, p<0.0001) and downhill condition (F(2, 84)=324.3, p<0.0001) as well as a significant interaction between condition*location (F(10, 84) =22.86, p<0.0001). Post-hoc testing identified accelerations to be significantly lower (p<0.0001) during DHR at the wrist, ankle, handlebar, and seatpost when compared with DHP and DHC, but not significantly different at the lumbar spine or head (p>0.05), indicating increased vibration damping demanded during DHC and DHP. Chapter 4: Mountain bike descending strategies 49 Figure 4.4 Comparison between DHC, DHP and DHR for total accelerations. ++++ (p<0.0001) main effect of location; @@@@ (p<0.0001) main effect of condition; oooo (p<0.0001) interaction between condition*location. Post-hoc analysis **** (p<0.0001) when compared with DHR Off-road descents with coasting (DHC) or pedalling (DHP) and road descent (DHR). Chapter 4: Mountain bike descending strategies 50 4.5 Discussion The primary aim of this study was to investigate the performance and physiological demand of either pedalling or coasting while navigating a XCO-MTB descent at the simulated intensity of a XCO-MTB race. A secondary aim was to compare these demands to an equivalent road descent. The main findings were: a) downhill performance was not changed through the absence of propulsive work; b) oxygen uptake significantly increased with the addition of propulsive work to off-road descending; c) mechanical and soft tissue damping associated with road descents were less than off-road, resulting in significantly decreased oxygen uptake and heart rate. Uphill variables such as power output, heart rate and VO2 firstly indicated the demand of the ascent, and also highlighted that riders were stressed similarly during each downhill trial. The key finding is that coasting on the XCO-MTB descent (Figure 4.1) used in this study did not result in a significantly reduced velocity (DHC= 7.3 ± 1.4, DHP= 7.5 ± 1.5 m∙s-1) or increased time (Figure 4.2), indicating that this strategy did not negatively impact performance. While oxygen uptake and heart rate were not different at 15 second time points (Figure4. 3) during off-road descents, overall VO2 was significantly increased when pedalling versus coasting (Figure 4.2). Maximal power during DHP (526.4 ± 151.0 W) likely exacerbated oxygen uptake, however overall VO2 increases occurred despite relatively low average power output (18.8 ±11.6 W) and minimal time spent pedalling (19.9 ±9.8 s). Importantly, increased VO2 was not a result of non-propulsive work from Chapter 4: Mountain bike descending strategies 51 vibration damping (Figure 4.4), which may be explained by the similar velocity between trials. While better recovery during intermittent exercise may be important (Tomlin & Wenger, 2001), this study did not investigate how these two strategies affected a subsequent high-intensity ascent which warrants further investigation. As indicated by the non-significant difference in time (Figure 4.2) when pedalling versus coasting and the non-significant relationship between power output and time when pedalling, these findings agree that downhill performance has less dependence on power output (Hurst & Atkins, 2006) and supports reports suggesting that skill is most important to descending performance (Chidley et al. 2014; Mastroianni et al. 2000). While skill cannot currently be quantified, its importance is corroborated by the significant relationship between times in DHP and DHC, indicating that the fastest riders had the fastest times whether pedalling or not. It is important to point out that despite elimination of mechanical propulsive work (i.e. DHC), participants in this study displayed a wide range of descending performance (142.5 ±29.2 s). While the importance of ascending prowess for XCO-MTB performance cannot be overlooked, it can be surmised that riders who descend the slowest are disadvantaged before the next ascent regardless of any time they are able to make up by ascending more quickly. Considering the results of this study and the demands of XCO-MTB, athletes can benefit from improvement in specific skills that would allow them to decrease descending time without the addition of pedalling. More research is required to determine these necessary skills. Chapter 4: Mountain bike descending strategies 52 It was hypothesized that non-propulsive work would be reduced when descending on the road and accompany a decrease in physiological variables. As expected, vibrations were significantly reduced at the wrist, handlebar, ankle and seatpost during DHR (Figure 4.4). This reduced vibration exposure was associated with significantly lesser (Figure 4.2) oxygen uptake and heart rate in DHR compared with DHP and DHC. After 45 s (Figure 4.3), VO2 appears to plateau in DHP and DHC while values continue to drop throughout the duration of DHR. Riders who are have a higher level of physical fitness tended to recover to greater extent as seen by the negative correlation between VO2max measured in the laboratory and VO2 measured during DHR. We surmise that the reduced vibration exposure eventuated to a decreased musculo-tendonous work requirement (Hurst et al. 2012) which would normally absorb the accelerations in order to protect the head and lumbar spine (Macdermid et al. 2015; Macdermid et al. 2015; Macdermid et al. 2014; Miller & Macdermid, 2015a). However, as muscle activity was not directly measured, this explanation remains only speculative. Nevertheless, the data do indicate that reducing vibrations nearer to those seen on the road could further reduce metabolic demand riding off-road and therefore it is recommended that investigations continue to explore equipment (Macdermid et al. 2015) designs which can decrease vibration exposure. Overall, the data support our hypothesis that physiological demand is greater during off- road descending than on-road, and this may help explain differences in what constitutes optimum physiological rider profiles between XCO-MTB and road competitors. Chapter 4: Mountain bike descending strategies 53 4.6 Conclusion This is the first study to compare different strategies for approaching descents during XCO-MTB and also to compare off-road and road descending. On the off-road descent used in this study, pedalling was not advantageous and resulted in an exacerbated oxygen uptake. Therefore, it is recommended that athletes focus on skill increases that will allow them to complete descents faster without adding pedalling, which is an area that needs more research. Moreover, physiological responses were significantly greater during off- road descents when compared with the road and can be partly attributed to the terrain demanding greater vibration damping by the body. Future research should continue to investigate equipment that can minimize vibration exposure when riding off-road nearer to values seen on the road to ultimately reduce oxygen uptake. Chapter 4: Mountain bike descending strategies 54 Chapter 5: Validation of a brake power meter 55 5. Chapter 5 VALIDITY OF A DEVICE DESIGNED TO MEASURE BRAKING POWER IN BICYCLE DISC BRAKES 5.1 Abstract Real-world cycling performance depends not only on exercise capacities, but also on efficiently traversing the bicycle through the terrain. The aim of this study was to determine if it was possible to quantify the braking done by a cyclist in the field. One cyclist performed 408 braking trials (348 on a flat road; 60 on a flat dirt path) over five days on a bicycle fitted with brake torque and angular velocity sensors to measure brake power. Based on Newtonian physics, the sum of brake work, aerodynamic drag, and rolling resistance was compared with the change in kinetic energy in each braking event. Strong linear relationships between the total energy removed from the bicycle-rider system through braking and the change in kinetic energy were observed on the tar-sealed road (r2=0.989; p<0.0001) and the dirt path (r2=0.952; p<0.0001). The t-tests revealed no difference between the total energy removed and the change in kinetic energy on the road (p=0.715) or dirt (p=0.128). This study highlights that brake torque and angular velocity sensors are valid for calculating brake power on the disc brakes of a bicycle in field Chapter 5: Validation of a brake power meter 56 conditions. Such a device may be useful for investigating cyclists’ ability to trav