Russell 2012

Embed Size (px)

Citation preview

  • RESEARCH ARTICLE

    Computer models offer new insights into the mechanics of rockclimbing

    SHAWN D. RUSSELL1, CHRISTOPHER A. ZIRKER1, & SILVIA S. BLEMKER1,2,3,4

    1Department of Mechanical & Aerospace Engineering, University of Virginia, Charlottesville, VA, USA,2Department of Biomedical Engineering, University of Virginia, Charlottesville, VA, USA, 3Department of Radiology,

    University of Virginia, Charlottesville, VA, USA and 4Department of Orthopaedic Surgery, University of Virginia,

    Charlottesville, VA, USA

    (Received 7 April 2012; accepted 3 October 2012)

    AbstractThree computer models of varying complexity were developed in order to investigate the kinematics, kinetics, muscleoperating ranges, and energetics of rock climbing. First, inverse dynamic models were used to investigate the joint angles andtorques used in climbing and to quantify the total mechanical work required for typical rock climbing. Climbing experiencewas found to have a significant effect on the kinematics used in climbing; however, there were no significant differences inmechanical work. Second, a musculoskeletal model of the whole body was developed, this model combined with thekinematic data was used to analyze the operating ranges of the upper and lower limb muscles during climbing. In general, theexperienced climbers employed kinematic motions that corresponded to muscle fibers used for climbing operating muchcloser to their optimum length than the kinematics of inexperienced climbers. Third, a forward dynamic model wasdeveloped to predict the metabolic goal of climbing. The results of this model suggest that an experienced climbing styleminimizes the fatigue of muscles while an inexperienced climbing style minimizes the total joint torques generated.

    Keywords: climbing, biomechanical modeling, work, efficiency, optimum climbing strategies, inverse dynamics

    Introduction

    Rock climbing poses unique demands on the human

    musculoskeletal system. In standard walking, the

    lower body is primarily responsible for locomotion

    and support; however, in climbing, both the upper

    body and lower body provide both locomotion and

    support. Climbing entails significant motion in the

    vertical plane and is not primarily limited to the

    horizontal motion typically associated with walking

    (McIntyre, 1983). Climbing results in motion

    mechanics that are drastically different from those

    the body typically performs in everyday activities.

    These unique mechanics and energetics are currently

    poorly understood. An advanced understanding of

    climbing mechanics will elucidate new methods for

    training which will help decrease the loads during

    climbing, reducing the chance of injury; increase the

    efficiency of climbing, reducing the energy needed to

    climb; and offer a deeper understanding of human

    motor learning strategies and physical adaptation

    development for atypical motions.

    To date, research regarding the kinematics and

    dynamics of rock climbing has focused on analysis of

    the strategies the climbers use to maintain stability

    while holding a static posture during climbing.

    Marino and Kelly (1988) reported that as climbing

    slopes increased from 608 to 1208 (908 being vertical),

    the percentage of body weight supported by the

    upper body increased from approximately 20% to

    40%. Others have investigated the dynamic changes

    in the support forces, applied through the hands and

    feet, when transitioning from a static quadrupedal

    state (four-limb support) to a static tripedal state

    (three-limb support) by removing a support foot

    (Quaine, Martin, & Blanchi, 1997a; Quaine &

    ISSN 1934-6182 print/ISSN 1934-6190 online q 2012 Taylor & Francis

    http://dx.doi.org/10.1080/19346182.2012.749831

    Correspondence: S.D. Russell, Department of Mechanical & Aerospace Engineering, University of Virginia, 122 Engineers Way, P.O. Box 400746,

    Charlottesville, VA 22904-4746, USA. E-mail: [email protected]

    Sports Technology, AugustNovember 2012; 5(34): 120131

    Dow

    nloa

    ded

    by [R

    MIT

    Univ

    ersity

    ] at 2

    3:13 2

    3 Dec

    embe

    r 201

    2

  • Martin, 1999). Although these studies attempted to

    limit the center of mass (CoM) motion resulting in

    static trials, they found that to maintain balance,

    climbers began the transfer of reaction forces away

    from the limb to be removed prior to initiating the

    transfer from a quadruped to a triped state. Others

    have found similar balance strategies employed when

    varying wall slope (Noe, Quaine, & Martin, 2001)

    and for various optimal and suboptimal climbing

    postures (Quaine, Martin, & Blanchi, 1997b; Testa,

    Martin, & Debu, 1999). However, each of these

    methods limited the climber to static positions, and

    none have calculated the joint loads associated with

    the climbing moves.

    Previous work regarding the efficiency of rock

    climbing has been focused on comparing the

    anthropometry and physiology of climbers and non-

    climbers. This work has demonstrated that climbers

    are typically small in stature with elevated strength to

    body mass ratios (Grant, Hynes, Whittaker, &

    Aitchison, 1996; Watts, 2004). Aerobic power

    analysis has shown that climbers tend to have lower

    oxygen consumption compared with the mean

    endurance athlete (Booth, Marino, Hill, & Gwinn,

    1999; Mermier, Janot, Parker, & Swan, 2000),

    suggesting that their aerobic fitness level is consistent

    with one required for quick recovery from high

    intensity effort. Mermier, Robergs, McMinn, and

    Heyward (1997) found that oxygen consumption of

    elite rock climbers on moderate terrain is equivalent to

    running at 2.6 ms21. However, climbing is a

    stochastic activity with spurts of action interspersed

    with periods of resting/static support. It is, therefore,

    difficult to measure a steady-state VO2 consumption

    rate, and the reports of mean VO2 range widely from

    18.6 to 43.8 ml(kgmin)21, for similar climbing

    conditions (Billat, 1995; Mermier et al., 1997;

    Watts, Daggett, Gallagher, & Wilkins, 2000).

    Although these measurements offer insight into

    individual climbers and their climbing strategies, it

    is difficult to apply those strategies to the general

    population due to the effects of individual condition-

    ing, rest to climb ratio, and averaging.

    Computer models have the potential to offer

    insights into complex human motions. Previously,

    simple forward dynamics models with limited degrees

    of freedom have been used to demonstrate multiple

    complex properties of walking, including stability and

    control of joint angles (Garcia, Chatterjee, Ruina, &

    Coleman, 1998; Goswami, Espiau, & Thuilot, 1996;

    Morgan, Mochon, & Julian, 1982), and also the

    efficiency of motion due to the distribution of joint

    torque (Kuo, 2002). In addition, more complex

    models are often used in both forward and inverse

    dynamic simulations of human movement. These

    inverse models have been used to quantify joint

    torques and forces, muscle mechanics and motor

    control of human walking (Arnold, Ward, Lieber, &

    Delp, 2010), running (Edwards, Taylor, Rudolphi,

    Gillette, & Derrick, 2010), and jumping (Anderson &

    Pandy, 1999). However, these models have primarily

    focused on motions supported entirely by the lower

    extremity and were not developed to include the

    upper extremity as a part of the support and control of

    locomotion.

    In this paper, we describe our recent work using

    three computer models of climbing with varying

    complexity to answer a range of questions regarding

    the mechanics and energetics of climbing. The first

    inverse dynamics model was developed to quantify

    the kinematic, kinetic, and energetic differences

    between experienced and inexperienced climbing

    motions. The second musculoskeletal model was

    used to evaluate how differences in these kinematic

    strategies affect muscle force-generating capacity.

    Finally, the third forward dynamic model was

    developed to investigate the energetic goals of

    differing strategies for rock climbing. Each of these

    models was created based on a set of human climbing

    motion capture experiments that made use of a

    custom climbing wall instrumented with six force

    plates.

    Methods

    Human climbing experiments

    Twelve healthy participants participated in this study,

    including seven inexperienced climbers and five

    experienced climbers, where experienced was defined

    as comfortable climbing 5.10 on the YosemiteDecimal System (5.10 YDS, VII-UIAA, or 20Australia). The inexperienced climber group con-

    sisted of five males and two females averaging

    26.7 ^ 5.0 years of age, 177.1 ^ 5.7 cm in height,

    and 75.3 ^ 9.1 kg in mass. The experienced climber

    group included two males and three females averaging

    29.8 ^ 8.6 years of age, 169.8 ^ 9.4 cm in height,

    and 62.0 ^ 9.4 kg in mass. All tests were conducted in

    the Motion Analysis and Motor Performance Lab-

    oratory at the University of Virginia. Participant

    consent was approved by the University of Virginias

    Human Investigation Committee and was obtained

    for all participants.

    Participants were given up to 20 min of free

    climbing time prior to data collection to acclimate

    themselves to the climbing wall. They were then

    instructed to climb the wall (approximately 35 cm

    steps) using their self-selected climbing strategy.

    Participants were instructed to ascend then descend

    the climbing wall three times per trial, pausing briefly

    in a four-point stance when changing directions.

    Kinetic data were collected using a climbing wall

    instrumented with six custom force plates (Bertec,

    Mechanics of rock climbing 121

    Dow

    nloa

    ded

    by [R

    MIT

    Univ

    ersity

    ] at 2

    3:13 2

    3 Dec

    embe

    r 201

    2

  • Figure 1. Experimental set-up used in data collection, including the instrumented climbing wall with typical grip placement.

    S. D. Russell et al.122

    Dow

    nloa

    ded

    by [R

    MIT

    Univ

    ersity

    ] at 2

    3:13 2

    3 Dec

    embe

    r 201

    2

  • Columbus, OH, USA), each with seven possible

    grip-mounting locations (Figure 1). 3D kinematic

    data were collected using a six-camera Vicon Motion

    Analysis System (Oxford Metrics, Oxford, UK) at

    120 Hz, and a modified full-body Plug in Gait marker

    set, 35 markers. At least three ascents were

    performed and measurements were averaged for

    each trial, with the participants moving from one

    four-point posture to another four-point posture.

    Analyzed data began at the initiation of movement

    from the lower grip set and ended when the climber

    reached a neutral posture on the upper grip set.

    Inverse dynamic model

    To quantify the kinematics and joint kinetics of

    climbing, a 3D, 17 segment, 16 joint, participant-

    specific model (Figure 2) was created for each

    participant in MSC.Adams, using the LifeMod plug-

    in (Biomechanics Research Group, San Clemente,

    CA, USA), from individual anthropometric data

    (age, weight, height, and gender). The 17 model

    segments included the following: head, neck, upper

    torso, central torso, lower torso, upper arms (2),

    lower arms (2), hands (2), upper legs (2), lower legs

    (2), and feet (2) (Figure 2). The segments physical

    properties were defined using the Generator of Body

    Data (GeBOD) database (Cheng, Obergefell, &

    Rizer, 1994). The 16 joints were each specified as a

    ball joint with three degrees of freedom; however, the

    elbow and wrist joints are reduced to two axis and the

    knee joints are reduced to only one axis.

    Development of this model in the MSC.Adams

    environment facilitated the application of external

    forces (wall contacts) from two sources: (1) direct

    application of experimentally measured forces (Ber-

    tec force plates) and (2) the predictive modeling of

    wall contact forces. Wall contact, from grasping and

    releasing climbing grips, was also modeled in the

    MSC.Adams environment. For simulations

    described in this study, model inputs were the

    measured marker positions exported from VICON to

    the LifeMod model and measured contact for data

    and position, and outputs based on inverse kinematic

    and dynamic models were joint angles, forces, and

    torques. From these parameters, we can calculate

    energetic parameters as described below. An advan-

    tage of this model over others (Willems, Cavagna,

    & Heglund, 1995) is that it facilitates the quantifi-

    cation of energetics down to the level of each joint

    degree of freedom.

    Total work, Wtot, is the sum of the external work

    Wext, work done to move the system CoM, and the

    internal work Wint, work done to move the body

    segments about the CoM:

    W tot W int W ext X

    i

    jtiDuij; 1

    W ext X

    i

    jFjDSj j; 2

    where ti is the torque at joint i, ui is the angle of joint i,Fj is the composite force applied to the wall in the jth

    cardinal direction, and Sj is the composite body CoM

    displacement in the jth cardinal direction. All

    mechanical work data presented in this study have

    been normalized by climber mass and the vertical

    distance traveled, J(kgm)21.

    Musculoskeletal model

    The second musculoskeletal model was similar to the

    first in degrees of freedom and was developed in the

    Opensim environment. This full-body model

    (Figure 3) combines previously developed models

    of the upper extremity (Holzbaur, Murray, & Delp,

    Mono

    Colo

    Figure 2. Three-dimensional inverse dynamics model utilized for

    inverse dynamic simulations and work calculations.

    Mechanics of rock climbing 123

    Dow

    nloa

    ded

    by [R

    MIT

    Univ

    ersity

    ] at 2

    3:13 2

    3 Dec

    embe

    r 201

    2

  • 2005) and lower extremity (Arnold et al., 2010). The

    combined model includes 112 independent muscles

    (54 in the lower extremity and 58 in the upper

    extremity). Muscles that cross only the wrist and

    finger joints were excluded from the upper extremity

    model to reduce computation time and simplify data

    analysis. Measured motion data were used as input

    for the inverse kinematic simulations, which resulted

    in muscle parameters for each movement pattern as

    the output.

    Of particular interest are the fiber lengths of each

    muscle as the participants perform their climbing

    task. Muscle fiber length is directly related to skeletal

    kinematics, and the maximum force generated by a

    muscle is a function of fiber length (Zahalak &

    Motabarzadeh, 1997). This analysis allows one to

    determine how chosen climbing kinematics may

    affect the amount of available muscle force being

    used in climbing and the amount of strength in

    reserve. This has obvious implications for injury

    prevention, training, and also may have implications

    on the efficiency of force generation.

    Forward dynamic model

    A third, 2D sagittal plane, model with five degrees of

    freedom was created using Adams/View (MSC.Soft-

    ware Corporation, Santa Ana, CA, USA) to mimic the

    Figure 3. A full-body musculoskeletal model was used to

    determine muscle behavior during climbing.

    Figure 4. Forward dynamic climbing model created in

    Adams/View. Actuators (red arrows) and joints at contact points

    (blue arrows) are shown (Colour online).

    S. D. Russell et al.124

    Dow

    nloa

    ded

    by [R

    MIT

    Univ

    ersity

    ] at 2

    3:13 2

    3 Dec

    embe

    r 201

    2

  • sagittal plane motion which occurs during the

    standing/pulling phase of climbing (Figure 4). This

    model capable of both inverse and forward dynamic

    simulations was developed to predict the joint torques

    required for optimum efficiency of a desired climbing

    trajectory, given a specified cost function. The model

    consists of seven rigid bodies linked together by

    revolute joints. The seven bodies represent the foot,

    shank, thigh, torso/head, humerus, forearm, and hand

    while the joints between segments represent the ankle,

    knee, hip, shoulder, elbow, and wrist. The torso

    segment includes a rigidly attached circular body to

    represent the head and to maintain accurate

    distribution of mass between the segments. Each

    segment of the model is scaled based on the

    participant-specific anthropometry while the masses

    and (CoM) locations of each of the segments were

    chosen based on normalized values in the literature

    (Winter, 1990).

    For this model, two simplified (sagittal plane only)

    kinematic and CoM positioning strategies for the

    ascension phase (standing/pulling up) of climbing

    were analyzed (Figure 5).

    (1) Experienced style: Elbows extended, CoM further

    away from the wall.

    (2) Inexperienced style: Elbow bent, CoM close to the

    wall.

    Matlab-MSC.Adams co-simulation routines (Zirker,

    2011) were used with this model to calculate the joint

    torques required to perform these movements. This

    was done using two solution methods: first using

    measured wall reaction forces and inverse dynamics

    to calculate the actual joint moments used by the

    climber, and second using optimization routines for

    forward dynamic predictive simulations. The second

    solution method employed Matlab-based SQP

    optimization algorithms to calculate the optimum

    joint torques required to reproduce the reference

    kinematics.

    For optimization, three cost functions that

    characterized the efficiency of the climbing model

    in unique ways were implemented. The first cost

    function was the sum of total mechanical work done

    by each joint, as described in the inverse dynamics

    model. The second cost function was the sum of the

    square of the joint torques (Tsqr). The third cost

    function was the sum of the square of the normalized

    joint torque (Tmax), where the torque developed by

    each muscle group is normalized by the maximum

    isometric joint torque. These values have been shown

    to be representative of the efficiency of individual

    muscle groups to do work. The maximum isometric

    torques were found in the literature for the

    lower extremities (Arnold et al., 2010), shoulder

    (Holzbaur et al., 2005), elbow, and wrist (Holzbaur,

    Delp, Gold, & Murray, 2007). The trunk and wrist

    muscles were not included in the Tmax calculations.

    This was due to the limited data regarding

    maximum isometric torques related to our trunk

    joints, and the small values of wrist torque and

    associated noise.

    For comparison purposes, all results were normal-

    ized by the inverse dynamic solutions of an average

    climb. Total work is not dependent on the number of

    simulated time steps. However, Tsqr and Tmax values

    were dependent on the climbing duration, so these

    values were also normalized by the number of

    discretely simulated steps for each simulation.

    Statistics

    Repeated measures ANOVA were performed to

    determine any differences between the climbing

    styles. To quantify differences between performed

    Figure 5. Schematic detailing the kinematics of each climbing style over the course of a single climbing stride.

    Mechanics of rock climbing 125

    Dow

    nloa

    ded

    by [R

    MIT

    Univ

    ersity

    ] at 2

    3:13 2

    3 Dec

    embe

    r 201

    2

  • tasks, Students paired t-tests (two-tailed) were used.

    Data were considered significant for p , 0.05.

    Results

    Motion capture measurements of climbing demon-

    strated that the joint kinematic trajectories varied

    greatly both between climbers and between trials for

    individual climbers. However, the general climbing

    patterns used were similar to those reported for

    ladder climbing (Armstrong, Young, Woolley, Ash-

    ton-Miller, & Kim, 2009; Hammer & Schmalz,

    1992; McIntyre & Bates, 1982; McIntyre, 1983).

    These climbing patterns included diagonal and

    lateral gait, coordinated movement of the contral-

    ateral arm and leg, and the collateral arm and leg,

    respectively. Each climbing style had a temporal

    pattern of two beats, concurrent motion of limbs, or

    four beats, with slight delay between coordinated

    limb motions. In addition, we found some climbers

    employed a strategy not reported in ladder climbing.

    These strategies, leading limb climbing, employed

    four distinct movements (four beat motion) where

    the climbing gait was initiated with either, the

    reaching up with both hands followed by stepping

    up with both feet, or stepping up with both feet then

    reaching up with both hands.

    Generally the experienced group of climbers used

    a different kinematic strategy than the inexperienced

    climbers. Experienced climbers maintained a posture

    with a more extended elbow and flexed knee

    compared to inexperienced climbers (Figure 6).

    These differences in climbing kinematics can be seen

    in the differing joint force trajectories of a typical

    climb from each group (Figure 7). The result of these

    kinematics is that, contrary to the common assump-

    tion that experienced climbers tend to keep their

    bodies close to the climbing surface to reduce loads,

    the experienced climbers climbed with their CoM

    farther from the wall compared to the inexperienced

    climbers (Figure 8).

    Climbing requires substantially more mechanical

    work than walking: climbing in our study typically

    required over 10 times more work than walking;

    total mechanical work for climbing was 18.0^

    2.2 J(kgm)21 compared to the total mechanical

    work for healthy walking which has been previously

    reported to be 1.02.0 J(kgm)21 (Mian, Thom,

    Ardigo, Narici, & Minetti, 2006; Russell, Bennett,

    Sheth, & Abel, 2011; Willems et al., 1995). Despite

    the different kinematic strategies employed by the

    experienced and inexperienced climbers, there were

    no significant differences in the work done between

    the groups. Of the work done in climbing, Wextrepresented 62.2 ^ 6.3% of the total work, which is

    similar to normal walking where the mean contri-

    bution of Wext to Wtot is 55% (Willems et al., 1995).

    Further analysis showed the distribution of Wtot done

    by sections of the body was upper body

    36.7 J(kgm)21, trunk 12.4 J(kgm)21, and lower

    Figure 6. Comparison of a representative climbing stride, experienced (dark blue) and inexperienced (orange) climbers. As shown above,

    each climber stepped up with the left foot followed by the right, then stood/pulled up followed by reaching up with the left then the right hand.

    Joint angles (left) and torques (right) are shown for the ankle and knee joints (Colour online).

    S. D. Russell et al.126

    Dow

    nloa

    ded

    by [R

    MIT

    Univ

    ersity

    ] at 2

    3:13 2

    3 Dec

    embe

    r 201

    2

  • body 50.9 J(kgm)21, and experience had no

    significant effect.

    Analysis of the musculoskeletal model demon-

    strated that the kinematic movement patterns resulted

    in differences in operating ranges of muscles during

    climbing, between the experienced and inexperienced

    climbers. The force generating capacity of a muscle

    fiber is a function of the fiber length (background of

    Figure 9) where optimum fiber length corresponds to

    the length of maximum force generation (Zajac,

    1989). Normalized fiber lengths were found to vary

    between groups (Figure 9). Experienced climbers

    used climbing strategies that kept the operating fiber

    lengths of the biceps brachii closer to their optimal

    fiber lengths than the inexperienced climbers. Con-

    versely, the inexperienced climbers used strategies in

    which the triceps brachii operated closer to its

    optimum fiber length.

    Using the third model to explore which parameters

    climber might be trying to minimize, the efficiency

    Figure 7. Kinematic differences between the experienced (dark blue) and inexperienced (orange) were more pronounced in the upper

    extremities. Differences between the elbow and knee, minimum, mean, and maximum, joint angles are shown. 08 represents joint neutral

    (normal posture in extension) (Colour online). *p , 0.05, **p , 0.01, and ***p , 0.001.

    Figure 8. Distance of climbers CoM from the wall in normal

    climbing, experienced (dark blue) and inexperienced (orange)

    climbers. Minimum, mean, and maximum distances reported over

    onestride (Colouronline). *p , 0.05,**p , 0.01, and***p , 0.001.

    Mechanics of rock climbing 127

    Dow

    nloa

    ded

    by [R

    MIT

    Univ

    ersity

    ] at 2

    3:13 2

    3 Dec

    embe

    r 201

    2

  • measures, work, Tsqr, or Tmax. We found the relative

    efficiency of each climbing style (Figure 5) varied

    depending on the cost function. Calculations from

    actual joint torques used in each of the two desired

    climbing trajectories showed that experienced climb-

    ing strategy (straight arms) was the most efficient

    for all efficiency quantities (Figure 10). More

    interestingly, optimization of the joint torques

    resulted in kinetics with increased efficiency

    (decreased values of efficiency quantities) for all

    simulations conducted except for the Tmax value for

    the experienced climbing strategy which remained

    nearly the same as the inverse dynamic calculation.

    This may indicate that Tmax is a quantity that

    experienced climbers are minimizing through

    training.

    Figure 9. Minimum,mean, and maximum fiber lengths of experienced (dark blue) and inexperienced (orange) climbers are identified by asterisks

    or circles depending on significance of the differences between the two groups (Colour online). *p , 0.05, **p , 0.01, and ***p , 0.001.

    Figure 10. Total costs for inverse dynamic (measured) and optimized (simulated) joint torques for the experienced (dark blue) and

    inexperienced (orange) climbing styles. Note that the inverse dynamic Tmax for the experienced climbing style does not reduce with

    optimization (Colour online).

    S. D. Russell et al.128

    Dow

    nloa

    ded

    by [R

    MIT

    Univ

    ersity

    ] at 2

    3:13 2

    3 Dec

    embe

    r 201

    2

  • Discussion

    The three models presented here provide a new

    paradigm for the analysis of the kinematics, kinetics,

    efficiency, and control of dynamic rock climbing.

    Here, these models have been used to quantify differ-

    ences between experienced and inexperienced clim-

    bers. The first model showed that the experienced

    and inexperienced climbers used different joint

    kinematics at the elbow and knee when climbing.

    This resulted in differing CoM trajectories, but no

    difference in the total mechanical work done by each

    group. The second model depicts how the elbow

    kinematics employed by the experienced and

    inexperienced groups facilitate greater maximum

    force generation in the biceps brachii (elbow flexion)

    and triceps brachii (elbow extension), respectively.

    Simulations from the third model show that climbers

    may use different physiologic goals when climbing,

    inexperienced climbers minimize the magnitude of

    force they develop in climbing, Tsqr, whereas

    experienced climbers minimize magnitude of force

    generated relative to their maximum force generating

    capacity, Tmax. Each model has been developed and

    implemented to further our knowledge about a

    unique aspect of climbing.

    Kinematic differences between groups lead to

    experienced climbers maintaining their CoM farther

    from the climbing surface than the inexperienced

    climbers (Figure 8). Zampagni, Brigodoi, Schena,

    Tosi, and Ivanenko (2011) reported similar differences

    between the CoM position of experienced and

    inexperienced climbers. This was unexpected as the

    generally accepted method for efficient climbing is to

    minimize the distance of the CoM to the wall, thus

    reducing the increased load due to the increased CoM

    induced moment. This may be due to the size of the

    grips used in these studies, both used large easy to hold

    grips allowing for the distribution of force across the

    entire hand; expert climbing routs often incorporate

    small grips where few and occasionally only one finger

    is used to support the climbers load. In these cases, the

    added load caused by an increased CoM couple may be

    the difference in a successful climb or a fall. In cases of

    climbing with reduced grip size, studies may find that

    experienced climbers begin to move their CoM closer

    to the climbing surface. In addition, the kinematics

    used in climbing result in an energy-intensive method

    of locomotion. This is evident when the energy cost of

    climbing are compared to the energy required for

    typical bipedal walking on level ground (Mian et al.,

    2006; Russell et al., 2011; Willems et al., 1995).

    However, it is interesting that the total mechanical

    work done for experienced and inexperienced climbers

    was not different despite the varying kinematic

    approaches used by the two groups, and the qualitative

    differences observation in their energy levels at the end

    of data collection.

    In climbing, the elbow contributes to the vertical

    motion in the upward direction by developing a

    flexion moment. When overlaying the operating

    ranges of the normalized fiber lengths of the biceps

    brachii (primary elbow flexor), developed using the

    second model, on the normalized forcelength curve

    of muscle, it is apparent that the different kinematics

    used shifting the operating length of the muscle fibers

    closer to optimal fiber length for the experienced

    climbing group (Figure 9). The forcelength curve

    represents the maximum available force generation of

    the muscle at full activation. Muscles operating at or

    near the peak are able to generate much more force

    per activated fiber than one operating below the peak

    (Murray, Buchanan, & Delp, 2000), which would

    theoretically result in a more metabolically efficient

    development of joint torque. Thus, although doing

    the same amount of work as the inexperienced

    climber, the experienced climber may employ

    climbing kinematics that put the muscles used in

    climbing closer to their optimum fiber lengths,

    allowing them to do that work with more efficiency

    and with more force available in reserve.

    It is generally accepted that when humans move we

    do so with an objective of reaching a destination or a

    goal while minimizing some quantity. Although many

    parameters are likely included in this minimized

    quantity, i.e., distance, metabolism, pain, boredom,

    and weather, we often simplify it to only include

    some measure of energetics, such as work, force

    generation Tsqr, or fatigue Tmax. Our simulations

    using the third model indicate that the experienced

    climbers employed a climbing strategy that mini-

    mized the energetic quantity of Tmax. Tmax represents

    the percentage of the total available muscle force used

    in climbing and has been related to muscle fatigue

    (Ackermann & van den Bogert, 2010). This would

    indicate that experienced climbers have learned that

    fatigue is the limiting factor when climbing. However,

    the inexperienced climbing strategy was most efficient

    when the Tsqr value was used as the cost function,

    indicating that the inexperienced climbers are not

    worried about fatigue as much as they are about the

    metabolic cost of generating muscle force (Umberger

    & Rubenson, 2011).

    When the results are combined, the three models

    presented here are capable of offering a broad picture

    of how successful climbing is achieved, and the role

    of experience in developing climbing style. However,

    the models in their current forms have limitations.

    The first two models do not incorporate muscle co-

    contraction, the simultaneous contraction of agonist

    muscle pairs about a joint often employed to increase

    the stability of a joint often when learning new

    movements. The inclusion of co-contraction would

    Mechanics of rock climbing 129

    Dow

    nloa

    ded

    by [R

    MIT

    Univ

    ersity

    ] at 2

    3:13 2

    3 Dec

    embe

    r 201

    2

  • likely result in higher work numbers for all climbers,

    the current model reports the net work done at a joint

    while the actual work would include both the positive

    and negative work done by antagonist muscle. We

    hypothesize that inexperienced climbers would have

    a higher level of co-contraction, thus increasing the

    differences in work between groups. In the second

    model co-contraction may also result in changes in

    muscle length, co-contracting muscles would

    increase the net force on a muscle that may result

    in increased stretch of the muscle tendons and

    shortening of the muscle belly. Finally, the kin-

    ematics of the third model were fixed while the joint

    torques were optimized based on a given cost

    function. Relaxing the joint kinematics constraints

    may allow the model to be useful for the prediction of

    better, reduced work, peak loads, reaction forces,

    etc., climbing modalities. The accuracy of Tmaxresults was limited by the use of published average

    muscle volumes. Previous work has shown that

    experienced climbers tend to have a smaller and

    more compact physique compared to inexperienced

    climbers that may manifest in the difference between

    muscle volumes in the two groups.

    This work provides a basis for comparison for any

    future climbing studies that analyze joint kinematics

    or make quantitative comparisons between experi-

    enced and inexperienced climbers. Future studies

    should investigate whether climbing kinematics in

    general have an effect on work done. In addition, the

    models and techniques developed here should be

    used to elucidate the differences between how and

    where (joint specific) the experienced and inexperi-

    enced climbers generate the work done in climbing.

    This would allow us to better tailor training

    regiments to increase the efficiency of the climber,

    and more importantly it would allow us to under-

    stand the increased loads due to various climbing

    strategies, increasing our understanding of where

    and when injuries may occur. This may help to

    understand the overrepresentation of injuries to the

    upper body reported in sport climbers (Peters,

    2001). We are currently incorporating MR imaging

    and climber-specific muscle volumes into the models

    to increase the fidelity of the effects of musculature

    on climbing movements. This forward dynamic

    model demonstrates that simple models can offer

    unique insights into human movement. This paper

    described the use of three simple cost functions;

    however, the model is developed to allow the

    addition of other more complex cost functions for

    optimization. In addition, the kinematics of this

    model were constrained as an input; however, these

    constraints can be removed to allow the model to

    optimize both joint torques and kinematics. Such

    simulations could offer even more information on

    how climbers choose their climbing strategies and

    how those choices change with experience.

    Conclusion

    Rock climbing is increasing in popularity both as a

    competitive sport and as an outdoor adventure

    activity. With this increase in popularity, little has

    been done to investigate the biomechanics of

    climbing. The energetics and mechanics of climbing

    are complicated. This paper demonstrates how

    models can provide new insights into the complex-

    ities of climbing mechanics and energetics, which

    would not be achievable through experiments/obser-

    vation alone. These new insights can help us to better

    understand why we choose specific climbing kin-

    ematics. In addition, they may lead to new more

    efficient climbing and training strategies that mini-

    mize injury risk while maximizing climbing ability.

    Acknowledgements

    The authors would like to thank the staff at the

    Motion Analysis and Motor Performance Lab,

    KCRC, at the University of Virginia. This work was

    funded by the DARPA-DOD Z-Man Program.

    References

    Ackermann, M., & van den Bogert, A. J. (2010). Optimality

    principles for model-based prediction of human gait. Journal of

    Biomechanics, 43, 10551060.

    Anderson, F. C., & Pandy, M. G. (1999). A dynamic optimization

    solution for vertical jumping in three dimensions. Computer

    Methods in Biomechanics and Biomedical Engineering, 2,

    201231.

    Armstrong, T. J., Young, J., Woolley, C., Ashton-Miller, J., & Kim,

    H. (2009). Biomechanical aspects of fixed ladder climbing:

    Style, ladder tilt and carrying. Proceedings of the human factors

    and ergonomics society annual meeting, 53, 935939.

    Arnold, E., Ward, S., Lieber, R., & Delp, S. (2010). A model of the

    lower limb for analysis of human movement. Annals of

    Biomedical Engineering, 38, 269279.

    Billat, V. (1995). Energy specificity of rock climbing and aerobic

    capacity in competitive sport rock climbers. Journal of Sports

    Medicine and Physical Fitness, 35, 20.

    Booth, J., Marino, F., Hill, C., & Gwinn, T. (1999). Energy cost of

    sport rock climbing in elite performers. British Journal of Sports

    Medicine, 33, 1418.

    Cheng, H., Obergefell, L., & Rizer, A. (1994).Generator of body data

    (GEBOD) manual. Dayton, OH: Systems Research Labs, Inc.

    Edwards, W. B., Taylor, D., Rudolphi, T. J., Gillette, J. C., &

    Derrick, T. R. (2010). Effects of running speed on a

    probabilistic stress fracture model. Clinical Biomechanics, 25,

    372377.

    Garcia, M., Chatterjee, A., Ruina, A., & Coleman, M. (1998). The

    simplest walking model: Stability, complexity, and scaling.

    Journal of Biomechanical Engineering, 120, 281288.

    Goswami, A., Espiau, B., & Thuilot, B. (1996). Compass-like

    bipedal robot part I: Stability and bifurcation of passive gaits.,

    Technical Report 2996, INRIA.

    Grant, S., Hynes, V., Whittaker, A., & Aitchison, T. (1996).

    Anthropometric, strength, endurance and flexibility character-

    S. D. Russell et al.130

    Dow

    nloa

    ded

    by [R

    MIT

    Univ

    ersity

    ] at 2

    3:13 2

    3 Dec

    embe

    r 201

    2

  • istics of elite and recreational climbers. Journal of Sports Sciences,

    14, 301309.

    Hammer, W., & Schmalz, U. (1992). Human behavior when

    climbing ladders with varying inclinations. Safety Science, 15,

    2138.

    Holzbaur, K. R. S., Delp, S. L., Gold, G. E., & Murray, W. M.

    (2007). Moment-generating capacity of upper limb muscles in

    healthy adults. Journal of Biomechanics, 40, 24422449.

    Holzbaur, K. R. S., Murray, W. M., & Delp, S. L. (2005). A model

    of the upper extremity for simulating musculoskeletal surgery

    and analyzing neuromuscular control. Annals of Biomedical

    Engineering, 33, 829840.

    Kuo, A. D. (2002). Energetics of actively powered locomotion

    using the simplest walking model. Journal of Biomechanical

    Engineering-Transactions of the Asme, 124, 113120.

    Marino, G., & Kelly, P. (1988). Modifications of force distribution

    in novice rock climbing technique. Proceedings of the 6th

    international symposium on biomechanics in sports

    (pp. 347353).

    McIntyre, D. R. (1983). Gait patterns during free choice ladder

    ascents. Human Movement Science, 2, 187195.

    McIntyre, D. R., & Bates, B. T. (1982). Effects of rung spacing on

    the mechanics of ladder ascent. Journal of Human Movement

    Studies, 8, 5572.

    Mermier, C. M., Janot, J. M., Parker, D. L., & Swan, J. G. (2000).

    Physiological and anthropometric determinants of sport climb-

    ing performance.British Journal of SportsMedicine, 34, 359365.

    Mermier, C. M., Robergs, R. A., McMinn, S. M., & Heyward,

    V. H. (1997). Energy expenditure and physiological responses

    during indoor rock climbing. British Journal of Sports Medicine,

    31, 224228.

    Mian, O. S., Thom, J. M., Ardigo, L. P., Narici, M. V., & Minetti,

    A. E. (2006). Metabolic cost, mechanical work, and efficiency

    during walking in young and older men. Acta Physiologica, 186,

    127139.

    Morgan, D. L., Mochon, S., & Julian, F. J. (1982). A quantitative

    model of intersarcomere dynamics during fixed-end contrac-

    tions of single frog muscle fibers. Biophysical Journal, 39,

    189196.

    Murray, W. M., Buchanan, T. S., & Delp, S. L. (2000). The

    isometric functional capacity of muscles that cross the elbow.

    Journal of Biomechanics, 33, 943952.

    Noe, F., Quaine, F., & Martin, L. (2001). Influence of

    steep gradient supporting walls in rock climbing: biomechanical

    analysis. Gait & Posture, 13, 8694.

    Peters, P. (2001). Orthopedic problems in sport climbing.

    Wilderness & Environmental Medicine, 12, 100110.

    Quaine, F., & Martin, L. (1999). A biomechanical study of

    equilibrium in sport rock climbing. Gait & Posture, 10,

    233239.

    Quaine, F., Martin, L., & Blanchi, J. P. (1997a). Effect of a leg

    movement on the organization of the forces at the holds in a

    climbing position 3-D kinetic analysis. Human Movement

    Science, 16, 337346.

    Quaine, F., Martin, L., & Blanchi, J. P. (1997b). The effect of body

    position and number of supports on wall reaction forces in rock

    climbing. Journal of Applied Biomechanics, 13, 1423.

    Russell, S., Bennett, B., Sheth, P., & Abel, M. (2011). The gait of

    children with and without cerebral palsy: Work, energy, and

    angular momentum. Journal of Applied Biomechanics, 27,

    99107.

    Testa, M., Martin, L., & Debu, B. (1999). Effects of the type of

    holds and movement amplitude on postural control associated

    with a climbing task. Gait & Posture, 9, 5764.

    Umberger, B. R., & Rubenson, J. (2011). Understanding muscle

    energetics in locomotion: New modeling and experimental

    approaches. Exercise and Sport Sciences Reviews, 39, 5967.

    Watts, P. (2004). Physiology of difficult rock climbing. European

    Journal of Applied Physiology, 91, 361372.

    Watts, P. B., Daggett, M., Gallagher, P., & Wilkins, B. (2000). 185,

    190 Metabolic response during sport rock climbing and the

    effects of active versus passive recovery. International Journal of

    Sports Medicine, 21.

    Willems, P. A., Cavagna, G. A., & Heglund, N. C. (1995).

    External, internal and total work in human locomotion. Journal

    of Experimental Biology, 198, 379393.

    Winter, D. A. (1990). Biomechanics and motor control of human

    movement. NY: Wiley-Interscience.

    Zahalak, G. I., & Motabarzadeh, I. (1997). A re-examination of

    calcium activation in the Huxley cross-bridge model. Journal of

    Biomechanical Engineering, 119, 2029.

    Zajac, F. E. (1989). Muscle and tendon: properties, models,

    scaling, and application to biomechanics and motor control.

    Critical Reviews in Biomedical Engineering, 17, 359411.

    Zampagni, M. L., Brigadoi, S., Schena, F., Tosi, P., & Ivanenko,

    Y. P. (2011). Idiosyncratic control of the center of mass in

    expert climbers. Scandinavian Journal of Medicine & Science in

    Sports, 21, 688699.

    Zirker, C. (2011). Development of biomechanical models to

    identify efficient locomotive strategies in rock climbing. Masters

    of Engineering, University of Virginia, VA.

    Mechanics of rock climbing 131

    Dow

    nloa

    ded

    by [R

    MIT

    Univ

    ersity

    ] at 2

    3:13 2

    3 Dec

    embe

    r 201

    2