Available online at www.sciencedirect.com

Procedia Engineering 00 (2009) 000–000

Procedia Engineering

www.elsevier.com/locate/procedia

8th

Conference of the International Sports Engineering Association (ISEA)

Systematic Design Customization of Sport Wheelchairs using the

Taguchi Method

Burton Ma*, Subic Aa, Mazur Ma, Leary Ma aSchool of Aerospace Mechanical and Manufacturing Engineering, RMIT University, Melbourne, Australia

Received 31 January 2010; revised 7 March 2010; accepted 21 March 2010

Abstract

This paper describes a systematic approach developed for customisation of wheelchairs for court-based sports, such as

wheelchair rugby. Novel predictive models and reference to prior experimental work has identified pertinent wheelchair

design parameters, including horizontal and vertical seat position and wheel camber. A purpose-built adjustable

wheelchair frame has been developed to experimentally assess the effects of these parameters on the performance of

individual athletes. The paper proposes a novel approach to determining optimal design parameters for a specific athlete

based on the Taguchi method. This approach is superior to traditional testing as it enables efficient characterisation of the

effect of design variables on wheelchair performance, including error checking to quantify the effect of noise variables

such as muscle fatigue.

© 2009 Published by Elsevier Ltd.

Keywords: Design Customization; Wheelchair Ergometer; Taguchi Method

1. Introduction

Interest and participation in wheelchair sports has grown over the years and there is an ongoing effort amongst different

organising bodies to popularise these events further at an international level. The sports have their roots in the 1940’s,

when wheelchair-based activities were first devised to provide rehabilitation exercise for returning servicemen using

regular “day chairs”. However, the equipment that is now used in wheelchair sports has evolved to become much more

specialised for different events. In fact, the level of specialisation now even extends to wheelchairs that are designed to

accommodate the requirements of different sub-classifications that exist within single team events, such as those which are

found in wheelchair rugby.

The performance of both athletes and the equipment that they use is becoming subject to more and more scrutiny.

Typically, wheelchairs that are used for sport are custom built for individuals by manufacturers who use anthropometric

data from the intended user to drive the design and provide a specific user-fit. However, current approaches to wheelchair

* Corresponding author. E-mail address: mike.burton@rmit.edu.au.

c© 2010 Published by Elsevier Ltd.

Procedia Engineering 2 (2010) 2659–2665

www.elsevier.com/locate/procedia

1877-7058 c© 2010 Published by Elsevier Ltd.doi:10.1016/j.proeng.2010.04.048

Open access under CC BY-NC-ND license.

Open access under CC BY-NC-ND license.

2 M. Burton et al. / Procedia Engineering 00 (2010) 000–000

customisation are mainly subjective, often relying on either the tacit knowledge of experienced users to provide a detailed

specification or a manufacture’s ability to interpret a user’s specific needs and translate them into a suitable design. This is

not exact and, since these wheelchairs represent a considerable capital investment, many stakeholders in these sports

believe that a more systematic approach is required.

A more comprehensive approach to design customisation of sport wheelchairs requires performance data to be related

to design parameters and for such relations to be customised for a particular user.

Previous studies show the use of an instrumented wheelchair ergometer to measure propulsion related performance

outputs [1, 2]. Separate research into the biomechanics of manual wheelchair propulsion highlights the importance of body

position relative to the wheelchairs rear wheel [3] and rear wheel camber angle [4]. This paper describes a novel

systematic approach to the design customisation of sport wheelchairs that optimise performance for specific athletes.

2. Methodology

2.1. Measuring Performance Outputs

Effective customization of any product to improve the performance of an individual user requires a reliable method

with which to measure performance outputs so as to accurately assess the effects that specific design changes have upon

them.

Wheelchairs that are used for sports differ from conventional “ day chairs” in that they are tailored for self-propulsion

and typically have cambered rear wheels for increased stability and improved turning capabilities. It is possible to divide

sport wheelchair design types into two main groups, racing wheelchairs and those that are used for court-based sports,

such as basketball, tennis and rugby. The focus of this research project is upon wheelchairs that are used for wheelchair

rugby, which is a fast paced, full contact ball sport that is played by two teams of up to twelve players on a level court. All

rugby wheelchairs must comply with regulations outlined by the International Wheelchair Rugby Federation [5]. Fig 1

shows a selection of wheelchairs that have been designed for different court-based sports, including rugby.

Basketball Rugby Tennis

Fig 1. Court-based Sports Wheelchairs (Images provided courtesy of Melrose Wheelchairs, New Zealand)

The functional requirements and design characteristics of a sports wheelchair differs from sport to sport and is also

affected by a user’s level of ability or their category ranking and the position that the play, if they are involved in a team

sport. However, research conducted at RMIT University suggests that key performance measures for wheelchair rugby

are: acceleration from standstill, and sustained sprinting over short distances. To quantifiably measure these and other

outputs, Devillard [1] and Faupin [2] describe the successful use and validation of a specific wheelchair ergometer that is

suitable for sports wheelchairs that have rear wheel camber. Hence it was decided that similar equipment could be used in

this project, as part of a systematic approach that can be used to assess the effect of design parameter changes on identified

performance outputs. The image in Fig 2 shows a picture of the VP Handisport-25 wheelchair ergometer that was procured

from the Techmachine subsidiary of the HEF Group in France for use in the RMIT Sports Technology Laboratory.

2660 M. Burton et al. / Procedia Engineering 2 (2010) 2659–2665

M. Burton et al. / Procedia Engineering 00 (2010) 000–000 3

Fig. 2. VP Handisport-25 Wheelchair Ergometer (unloaded and with a rugby wheelchair in place)

2.2. Propulsion Biomechanics and Relevant Wheelchair Design Parameters

Numerous studies have been conducted into the biomechanics of manual wheelchair propulsion and wheelchair design

parameters that affect propulsion. Vanlandewijck [6] provides an authoritative reference, in which the experimental work

of several other researchers is reviewed. Literature indicates that camber angle and seat position (vertical and horizontal)

are pertinent design variables.

2.2.1. Seat position

Kotajarvi [3] conducted experiments to assess the effect of seat position on propulsion biomechanics. For these

experiments, a wheelchair was modified to allow the position of the main wheel axle to be adjusted relative to the

wheelchair frame and seat. A total of nine fixed seat positions were tested, and performance was measured by recording

the associated push-rim forces. It was seen in these experiments that the lower tested seat positions had a positive effect on

push time, recovery time and push angle. The peak radial and axial push-rim forces were also seen to be significantly

higher in the lower seat positions. However, tangential force was largely unaffected by alterations to seat position.

Similar experiments were conducted by Masse [7], adjusting a wheelchairs seat relative to the axle in order to test six

different seating positions. It was observed that lowering the seat caused a decrease in pushing frequency for Low-Back

and Low -Middle seat positions. The opposite was found to be true when raising the seat, which generally resulted in

higher speeds and push frequency for High-Front, High-Middle and High-Back settings.

Richter [8] developed a simplified quasi-static model to simulate the effects of seat position on performance. This

model has been enhanced by the authors to enable the effect of vertical and horizontal position to be modified explicitly

and independently (rather than vertical position and seat angle as in [8]). This model has been applied to quantify the

effect of vertical and horizontal seat position on performance (Figure 3).

2.2.2. Camber angle

Camber angle of the rear wheels has also been shown to influence propulsion cycle efficiency. Faupin conducted a

study in which three different camber angles (9°, 12° and 15°) were tested, using a basketball wheelchair on a roller-type

ergometer, with test participants sprinting for durations of 8 seconds [4]. Higher camber angles were seen to increase both

residual torque and total power but decrease mean velocity. Propelling time (powered push phase duration) was seen to

increase with camber angle.

Veeger conducted an experiment in which eight non-impaired subjects participated in a test using a motor-driven

treadmill to study the effect of rear wheel camber on wheelchair ambulation [9]. The test consisted of four runs with rear

wheels set at four different degrees of camber (0°, 3°, 6°, and 9°) and at four different speed steps (2, 3, 4, and 5 km/h). No

significant effects upon oxygen consumption, heart rate or mechanical efficiency were observed in the tests. However, the

kinematic parameters of push time, push angle, and abduction showed differences between 3 and 6 degrees camber.

Seat position and wheel camber are pertinent to the magnitude and duration with which effective propulsion forces can

be applied to the rear wheels by a seated user. However, no examples could be found of experiments that combined these

different factors to establish the optimal parameter settings for a specific athlete.

M. Burton et al. / Procedia Engineering 2 (2010) 2659–2665 2661

4 M. Burton et al. / Procedia Engineering 00 (2010) 000–000

Fig. 3. Effect of seat position on shoulder and elbow torque based on a model developed by the authors.

2662 M. Burton et al. / Procedia Engineering 2 (2010) 2659–2665

M. Burton et al. / Procedia Engineering 00 (2010) 000–000 5

2.3. The Adjustable Wheelchair Frame

A mechanical sizing frame was developed that allows controlled incremental adjustment of pertinent design parameters,

namely the rear wheel camber angle, vertical seat position and horizontal seat position. Fig 4 shows the main variable parameters

that may be adjusted in this mechanism. The rear wheel axles have a range of adjustment that is comparable to the ergometer (i.e.

0° to 25°) so as to ensure that the fullest range of experimental and manufactured camber angles can be catered for. This range

exceeds that of previous studies [4, 9] and encompasses the camber angles typically seen in wheelchair rugby (14° to 20°). The

mechanism also allows a range of axel movement in horizontal and vertical planes, exceeding the simulated results (Fig 3) and

encompassing the parameter range that was observed in manufactured wheelchairs that are used for wheelchair rugby. In addition

to the experimental parameter adjustments, the angle of the seat and back rest can both be independently adjusted along with

height of the seat relative to the foot plate so as to suit the specific anthropometric requirements of each athlete. During

experiments, the adjustable frame is secured to the wheelchair ergometer using suitable straps. Participants accelerate from stand

still and maintain comfortable propulsion velocities for durations that vary from eight seconds, as seen in Faupin [4], up to twelve

minutes for a performing a standard “Cooper Test”, as described by Franklin [10].

Fig. 4. RMIT University wheelchair mechanism adjustment parameters

2.4. Experimental Design and Taguchi Method

The effect of these design parameters can be quantified at each unique combination of parameters, known

as full factorial analysis. This method is problematic for physical testing of athletes as:

• The number of experiments is very large, i.e. testing the three identified parameters at three levels (a

relatively sparse sample size for optimisation) would require 27 unique tests and five levels require

243 tests.

• Experimental testing of athletes is subject to systemic errors as human performance is not precisely

repeatable, and each experiment will effect subsequent trials due to muscle fatigue. To mitigate

systemic errors, it is imperative to allow: sufficient recovery time to avoid muscle fatigue; and, to

allow repetitions at each test level to allow an average response to be identified. These factors

significantly increase the required sample size.

The Taguchi method [11] provides an opportunity to dramatically reduce the number of required sample size. The Taguchi

method is implemented by an orthogonal experimental array (fractional factorial) which enables the effect of multiple parameters

to be assessed simultaneously. This approach significantly reduces the required sample size, thereby alleviating the systemic

errors typically associated with experimental testing of athletes. Furthermore, the Taguchi method allows noise variables, such as

the number of runs before extended rest, to be included to ensure that muscle fatigue does not compromise experimental results,

for example, Table 1.

M. Burton et al. / Procedia Engineering 2 (2010) 2659–2665 2663

6 M. Burton et al. / Procedia Engineering 00 (2010) 000–000

To aid the identification of optimal chair position, a response surface model (RSM) of the Taguchi method outcomes will be

developed in-situ during testing. This RSM will enable human-in-the-loop feedback optimisation. The initial test results

summarise the influence of the identified parameters on performance. Based on this summary, regions of high performance can

be investigated with greater resolution.

The predictive model of wheelchair propulsion developed by the authors has been applied to identify pertinent experimental

parameters (Section 2.2). The experimental outcomes will be compared with predicted performance to enable tuning of the

predictive model. It is also intended that the model outcomes will allow interpolation between experimental data points to further

reduce the time required to optimise seat position for a particular athlete.

Table 1. L9 (34

) orthogonal array.

3. Conclusion

The predictive model of wheelchair propulsion developed by the authors has been applied to identify the influence of vertical

and horizontal seat position on wheelchair performance. In combination with reported literature, these parameters, plus wheel

camber have been identified as pertinent to wheelchair performance. However, no examples could be found in the literature of

experiments that combined these different factors to establish what might be the optimum overall parameter settings for any

given user. An experimental apparatus (sizing-frame) has been developed to quantify the influence of these design parameters.

Taguchi method has been applied to identify the optimal settings for design customization of wheelchairs for specific sports.

References

[1] Devillard X, Calmels P, Sauvignet B, Belli A, Denis C, Simard C and Gautheron V, "Validation of a new ergometer adapted to all

types of manual wheelchair." European journal of applied physiology, Volume 85, Issue 5, 2001, pp 479-485

[2] Faupin, A, Gorce P and Thevenon A, "A wheelchair ergometer adaptable to the rear-wheel camber." International Journal of

Industrial Ergonomics, Volume 38, 2008, pp601-607

[3] Kotajarvi, B. R., Sabick M B, An K, Zhao K D, Kaufman K R and Basford J R, "The effect of seat position on wheelchair

propulsion biomechanics." Journal of Rehabilitation Research & Development, Volume 41, Issue 3b, 2004, pp403-414

[4] Faupin A, Campillo P, Weissland T, Gorce P and Thevenon A, "The effects of rear-wheel camber on the mechanical parameters

produced during the wheelchair sprinting of handibasketball athletes." Journal of Rehabilitation Research and Development, Volume 41,

Number 3b, 2004, pp 421-428

[5] International Wheelchair Rugby Federation, “International Wheelchair Rugby Federation Rules”, IWRF Technical Commission, 2008

[6] Vanlandewijck, Y., D. Theisen and Daly D, "Wheelchair Propulsion Biomechanics: Implications for Wheelchair Sports." Sports

Medicine, Volume 31, Issue 5, 2001, pp339-367

[7] Masse L C, Lamontagne M and O’Riain M D, "Biomechanical analysis of wheelchair propulsion for various seating positions."

Journal of Rehabilitation Research and Development, Volume 29, Issue3, 1992, pp12-28

[8] Richter, W. M. "The effect of seat position on manual wheelchair propulsion biomechanics: a quasi-static model-based approach "

Medical Engineering & Physics, Volume 23, Issue 10, 2001, pp707-712

Run Vertical

position

Horizontal

position

Wheel

camber

Runs between

extended rest

1 low rearward decreased 1

2 low neutral neutral 2

3 low forward increased 3

4 neutral rearward increased 3

5 neutral neutral decreased 1

6 neutral forward neutral 2

7 high rearward neutral 2

8 high neutral increased 3

9 high forward decreased 1

2664 M. Burton et al. / Procedia Engineering 2 (2010) 2659–2665

M. Burton et al. / Procedia Engineering 00 (2010) 000–000 7

[9] Veeger HEJ; van der Woude LHV and Rozendal RH, "The effect of rear wheel camber in manual wheelchair propulsion",

Journal of Rehabilitation Research & Development, Volume 26, Issue 2, 1989, pp37-46

[10] Franklin BA, Swantek KI, Grais SL, Johnstone KS, Gordon S, Timmis GC “Field test estimation of maximal oxygen

consumption in wheelchair users”, Archives of Physical Medicine and Rehabilitation, Volume 71, Issue 8, 1990, pp574-8

[11] Taguchi G, "System of Experimental Design", New York: UNIPUB/Kraus International Publications, ISBN 0-527-91621-8,

1987

M. Burton et al. / Procedia Engineering 2 (2010) 2659–2665 2665