MUSCULOSCELETAL EFFICIENCY IN CYCLING AND … · EFFICIENCY IN CYCLING AND HANDCYCLINGAND HANDCYCLING Dr. Harald Böhm Orthopaedic Hospital for Children ... Fahrradphysik und Biomechanik.

MUSCULOSCELETAL EFFICIENCY IN CYCLING

AND HANDCYCLINGAND HANDCYCLING

Dr. Harald BöhmOrthopaedic Hospital for ChildrenAschau im Chiemgau, Germary

AgendaAgenda

• Index of efficiency: Definition and relevance in cyclingy g

• Musculoskeletal simulation of cycling

• Handcycling: the force generation abilities of the arms 

• Simulation of handcycling

M l l h d i 3D• Muscle‐length and momentarms in 3D

2

Only tangential force contributes to propulsion

F di lFradial

Ft ti l3

Ftangential

Pedal forces in cycling

360°/0°

270°

90°

270

180°

4

Böhm et al.  2008. Effects of short‐term training using SmartCranks on cycle work distribution and power output during cycling, European Journal of Applied Physiology, 103 (2), 225‐232

Index of effectivenes (IE)Index of effectivenes (IE)

%100

360

0tan∫

°

°

αdFIE

gential

%10036022

tan

0 •+

=

∫°

αdFFIE

radialgential0°

5

Davis & Hull 1981, Measurement of pedal loading in bicycling: II. Analysis and results. J Biomech 14:857–872

Four Sectors of a cycle (right) y ( g )

315° 45°push forward

push downpull up

135°225°pull backward

6

p

Sectors IESectors IE

45°

%100

135

45tan

•=∫°

°

αdFIE

gential

unten%100135

45

22tan

•+

=

∫°

°

αdFFIE

radialgential

unten unten

135°

7

Coyle et al. 1991, Med Sci Sports Exerc 23:93–107

IE at different power levelsIE at different power levels100

60

80

]

20

40

vene

ss [%

]

-20

0

x of

effe

ctiv

80

-60

-40

inde

30 40 50 60 70 80 90 100 110-100

-80

power [% maximum individual power]

8Böhm et al.  2008,  European Journal of Applied Physiology, 103 (2), 225‐232

p [ p ]

Options to increase IEOptions to increase IETraining

1. Technique specific Protocol 

Training

2. Biofeedback

3 SmartCranks3. SmartCranks

Hardware modifications

1 Elliptical chainrings

Hardware modifications

1. Elliptical chainrings

2. Variation of crank alignment

9

Relevance of IE in CyclingStudies relating energy consumption and power 

t t t b tt IE ( h d h )

Relevance of IE in Cycling

output to better IE (no hardware changes) 

Recreational cyclists: y1. Lafortune & Cavanagh 1983

1. Ericson & Nissel 19882 Coyle et al 19912. Coyle et al. 1991

Professionals:1 Lafortune & Cavanagh 19831. Lafortune & Cavanagh 1983 2. Hillebrecht et al. 19983. Cavanagh & Sanderson 19864 K t & F lt 19914. Kautz & Feltner 1991

Where is the problem?Where is the problem?

BiomechanicsMechanics Biology

+ =

I d fIndex ofEffectiveness

11

Hypothesis

For efficient c cling the m sc loskeletal

Hypothesis

For efficient cycling the musculoskeletal system requires radial forces

SummarySummary

• Index of effectiveness (IE)  is the relation of tangential forces to total forces applied to the g pppedal

• IE can be subdivided into four sectors defined by the action of the cyclist 

• IE does not show any relation to efficiency or power production in professional or recreational cyclists

13

Human modelHuman model

4 rigid segments

8 Hill‐type muscles [1]

metabolic energymetabolic energy consumption [2]

[1] Böhm et al. J Appl.Biomech 2006:  3‐13.

[2] Bhargava et al. J Biomech. 2004: 81‐8]

Muscle metabolic energy [1]++++=+= WBSMAWHE &&&&&&&&

Muscle metabolic energy [1]

=E& energy rate

=

WH&

& heat rate

ork rate )(f=

=

A

W&

work rate

activation heat rate ))/exp(,,(),(

tmactfvFf

stim

CECE

−==

τ

=

=

SM&

& maintainance heat rate

shortening heat rate ),),((

)/,,(

max

,

vFlFf

llmactf

CECECEiso

optCECE

=

=

=B&g

basal metabolic heat rate )(),),(( max,

mff CECECEiso

=

Bicycle modelBicycle model

seating position [2]

air resistance

inertia of bike, rider and ,wheels

free wheel clutchfree wheel clutch 

[2] GRESSMANN, M.: Fahrradphysik und Biomechanik. Technik, Formeln, Gesetze, (8.Auflage). Bielefeld: Delius Klasing, (2003).

Optimization of muscle activation

Genetic algorithm

Optimisation goal:Optimisation goal:a) reduction of metabolic energy consumptionb) increase of index of efficiencyb) increase of index of efficiency

ion

activati

Crank angle [°]

Validation Study

N]

y

• Ergometer 280 W, 90 rpm • 20 male amateur cyclists 

. force

[N

(28±7 y , 180±5cm, 76±8kg)

tang

orce

[N]

radial fo

Crank angle [°]

Böhm et al.  2008. Effects of short‐term training using SmartCranks on cycle work distribution and power output during cycling, European Journal of Applied Physiology, 103 (2), 225‐232

Validation of muscle activation

Neptune et al.: The effect of pedaling rate on coordination in cycling, Journal ofBiomechanics, 30 (10), 1997.

Simulation results

35 km/h

280 W 

90 rpm90 rpm

Simulation resultsSimulation results

E ti l IE optimalEnergy optimal IE optimal

Comparison of energy consumption and IE

d f ff [ ]Energy [J] Index of effectiveness [%]

optimal energy

optimal IE optimal energy

optimal IEgy

consumption gy

consumption

why it is so efficient to push downward ?

1. Anti‐gravity muscles are 

h t th th i momentarm hip

momentarm

much stronger than their 

antagonistsmomentarm 

knee

ee ip

2. long moment arms in top 

pedalntarm kn

ntarm hi

and bottom positionshigh joint torques required pedal 

force

mom

en

mom

enhigh muscle forces required

pedal force

Conclusioni l d lli h i l i

ConclusionIE optimal pedalling technique results in 2.5 times higher  energy consumptiong gy p

I d f Eff ti i t d t tIndex of Effectiveness is not adequate to quantify the quality of pedalling techniqueq y q y p g q

SummarySummary• Muscles energy consumption• Muscles energy consumption 

might be estimated using the Hill muscle model 

• Optimization for energy p gyconsumption results in realistic muscle activations

• Optimizing for IE results in considerably high energy comsumption

25

F

Question

α F

Fr

β FG

Ft

β

α

=Ft

β

Fr

F

Arms are shorter then legs, current crank system simply adopted from cycling might not be

FP

simply adopted from cycling might not be suitable for the hand arm system

push, pull (max. Isometric)

3        2        1       3      2       1Positions

Device to measure isometric joint torques

Maximum isometric joint torques at the elbow and shoulder

Muscles torque potential

Böhm H., Krämer C., 2007, Computational Imaging and Vision (36)

Torque potential forTorque potential for rowing and cycling

circularlinear

0°

45°

90°

135°

Conclusions from the measurementsCo c us o s o t e easu e e ts1. In rowing the ability to generate force is higher 

h i lithan in cycling. 2. In handcycling the maximal joint loads are l h i ilower than in rowing

C l iConclusion:Combine both advantages in an elliptic drive

Hypothesisypot es s

Th lli ti d i t bl th•The elliptic drive concept enables the muscles to work in a better range of  their force length relation, so that there is less activation required and less metabolic qenergy consumed than in a cyclic movementmovement

AgendaAgenda

• Index of efficiency: Definition and relevance in cyclingy g

• Musculoskeletal simulation of cycling

• Handcycling: the force generation abilities of the arms 

• Simulation of handcycling

M l l h d i 3D• Muscle‐length and momentarms in 3D

35

Human modelHuman model

4 rigid segments

6 Hill‐type muscles [1,2]

metabolic energymetabolic energy consumption [3]

[1] Böhm et al. J Appl.Biomech 2006:  3‐13.[ ] pp[2] Fagg, Technical Report Department of Computer Science,University of Massachusetts,Amherst[2] Bhargava et al. J Biomech. 2004: 81‐8

]

Muscle lengths of biarticular muscles usc e e gt s o b a t cu a usc es

Optimization procedure to obtain muscle activation

Starting position withg pinitial guess for activation functions of six muscles(Boehm et al JAB 2006)(Boehm et al. JAB 2006)

Calculate the cyclicCalculate the cyclicmotion and the cost function

Optimizer changes the muscle gActivation to get a global minimum of the cost function (Corana et al. ACM 13, 1987)(Corana et al. ACM 13, 1987)

Stop when minimum is reached

Example M.brachialis

Folie 14

Optimization• global Minimum• Multidimensional Problem (48 Variables)• Multidimensional Problem (48 Variables) 

Simulated Annealing (Corana et al ’87)

3

Simulated Annealing (Corana et al.  87)

1

2

-1

0

-3 -2 -1 0 1 2 3-3

-2

Folie 18

Simulated motion

cyclic drive (b/a=1) elliptical drive (b/a =0.5)

• m, 32 y, 170 cm, 72 kg

• drive resistance 20 Nm• drive resistance 20 Nm

• crank length 18 cm

All muscles’ metabolic energy consumption

elliptic drive

gy p

200

]elliptic drivecyclic drive

150

ener

gy [J

100

met

abol

ic

50

m

50 100 150 200 250 300 3500

angle [°]

Simulations

Subject Arm length

Muscles metabolic energy (Eelliptic – Ecyclic) / Ecyclic

m,170 cm,72 kg 58.0 cm - 3.5 %

m,180 cm,79 kg 59.8 cm - 3.8 %m,187 cm,88 kg 64.2 cm - 4.0 %

•The model suggest that the elliptic drive concept is more efficient

Krämer et al 2009 ErgonomicsKrämer  et al 2009, Ergonomics

Krämer et al 2009, European Journal of Applied Physiology

2D to 3D2D to 3D

M l l th d t i 3D• Muscle length and momentarms in 3D are not known for complex arm movements such as forearm pronation and supination  

The Visible Human DatasetThe Visible Human Datasethigh resolution images segmentation of boneshigh‐resolution images segmentation of bones

right arm

humerus

frontal view horizontal view

Triangulated surface meshesTriangulated surface meshes

Dijkstra algorithmDijkstra algorithmmuscle length = shortest path from origin to insertion

humerus - originhumerus - origin

m. triceps brachii caput laterale

ulna - insertion

Graphical user interfaceGraphical user interfacevirtual reality command window

Extensor: m triceps brachiiExtensor: m. triceps brachii

Polynomial fit to Fagg 2003Dijk t l ith

≈ 2 cm

Dijkstra algorithm

extended arm flexed arm

Flexor: m brachialisFlexor: m. brachialis

Polynomial fit to Fagg 2003

≈ 2 cm

Polynomial fit to Fagg 2003Dijkstra algorithm

extended arm flexed arm

ConclusionConclusion

All l l ti f i di id l l l th d i+ Allows calculation of individual muscle lengths during motion

+  Can be transferred to complex joint movements

+  Calculation is semi‐automatic

+  Works well for muscles close to the bony surface

Muscle deformation by minimizingdeformation energy

Krämer C.,  Böhm H., Senner V., 2008. Creating 3D muscle lengths and moment arms from the Visible Human Dataset, The Engineering of Sport 7, Springer, 2, 143‐148.

SSummary

• Elliptical drive was suggested to be more efficient in handcycling

• 3d simulation requires a lot more work 

54