Applied Ergonomics 41 (2010) 428–435

Contents lists available at ScienceDirect

Applied Ergonomics
journal homepage: www.elsevier.com/locate/apergo

Ergonomic evaluation and redesign of children bicycles based on

anthropometric data
Lambros Laios, John Giannatsis*
Ergonomics Research Group, Department of Industrial Management, University of Piraeus, 80 Karaoli & Dimitriou str., 18534 Piraeus, Greece

a r t i c l e i n f o a b s t r a c t

Article history: Proper bicycle fit is very important for cycling performance, efficiency, comfort and injury prevention.
Received 6 December 2007 This is especially true in the case of children cyclists that do not have the necessary cycling experience,
Accepted 15 September 2009 balance and the fully developed musculoskeletal system of the adults. Bicycle fit depends on both the
design and dimensions of the bicycle as well as on the anthropometric dimensions of the cyclist. In the
Keywords: present paper a case study concerning the ergonomic evaluation and redesign of a series of bicycles for
Bicycle fit
children and teenagers 7–14 years old is presented. The study has been commissioned by a major Greek
Bicycle design
bicycle manufacturer who wanted to gain competitive advantage by introducing new anthropometri-
Anthropometric design
cally-designed bicycles. Employing virtual modelling techniques and the method of Principal Component
Analysis, bicycle affordance for selected representative cases and various bicycle sizes has been exam-
ined. Based on the results of the study redesign recommendations that improved bicycle fit for specific
groups were proposed and a formal bicycle size selection method has been defined. The redesigned
bicycles are now in full production and distribution is underway in many commercial outlets in Greece.
! 2009 Elsevier Ltd. All rights reserved.

1. Introduction
Bicycles are the most common and widely used human powered
vehicle around the world. Cycling is nowadays considered not only
an efficient and healthy means of transportation but also a popular
recreational and sport activity. The expansion of cycling over the
last 50 years led to various bicycle designs, like sport/road bicycles,
mountain bicycles, BMX bicycles, standard ‘utility’ (commuter)
bicycles and more recently the hybrid bicycle (Wilson, 2004). All of
the above mentioned bicycle types are specialized variants of the
basic diamond frame bicycle, each one designed to meet the
specific needs posed by the intended use of the bicycle. These needs
and constraints are usually associated with cycling efficiency,
power, safety and comfort and their relative importance for the
intended bicycle application.
It is widely accepted that cycling efficiency and comfort depend
not only on bicycle design but also on bicycle fit (Silberman et al.,
2005; De Vey Mestdagh, 1998). Bicycle fit involves the selection of
the appropriate frame size and subsequently the adjustment of
various dimensions, like the seat height or fore-aft position
according to the anthropometric dimensions the cyclist, so that the
* Corresponding author. Tel.: þ30 210 41 42 390; fax: þ30 210 41 42 356.
E-mail addresses: llaios@unipi.gr (L. Laios), ggian@unipi.gr (J. Giannatsis).
best posture for the intended bicycle application can be obtained.
Bicycle size is defined mostly by the size of the frame and it is
usually referred by the corresponding wheel size or the seat tube
length, depending on the manufacturer.
Absolute bicycle fit is also very important for injury prevention
(Baker, 2000). Some of the most common health problems and
annoyances associated with cycling can be attributed to poor
bicycle fit and wrong cycling posture (Mellion, 1991; Schwellnus
and Derman, 2005). These problems are usually observed at three
areas of the human body, namely the neck, the lower back and the
lower limbs and most often the knee (De Vey Mestdagh, 1998).
Different handlebar positions, for example, affect spinal curvature
as well as the load imposed on the back/neck area and can, there-
fore, be associated with annoyances in these areas (Kolehmainen
et al., 1989). In the case of bicycles for children, proper bicycle fit is
of highest importance for both safety and injury prevention
(Nemours Foundation, 2006). Safety is extremely important for
relatively young children that do not have the cycling experience
and are therefore prone to accidents. Furthermore, children of all
ages do not have a fully developed musculoskeletal system which
implies that cycling discomfort can be more apparent and cycling
injuries may have more severe and possibly permanent conse-
quences. Besides safety and injury prevention, cycling comfort
should also be a major issue in the design of children bicycles not
only because they represent a significant market segment but also

0003-6870/$ – see front matter ! 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apergo.2009.09.006
 L. Laios, J. Giannatsis / Applied Ergonomics 41 (2010) 428–435
because of their future potential as customers of the bicycle
industry. It can be reasonably argued that the feeling of discomfort
during cycling prevents the frequent use of bicycles (Christiaans
and Bremner, 1998). Since cycling habit is usually acquired in the
younger ages, from childhood to puberty, it is obvious that the
corresponding bicycles should provide maximum accommodation
and comfort to encourage the use of bicycles and cycling in general.
Because of the wide variance in anthropometric dimensions
between children, even of the same age, bicycle manufacturers
offer a series of bicycles of various sizes. The design of these bicycles
involves the definition of frames of different size and dimensions as
well as the definition of the corresponding adjustment ranges for
the seat and the handlebar. In order to fully accommodate the
majority of possible users, the design of the bicycles should be
based on data that take into account the distribution of anthropo-
metric dimensions among the children. It is not uncommon among
bicycle manufacturers to define the frame dimensions of the chil-
dren’ bicycles, as well as the allowable adjustment range for the
various components, according to assumptions regarding the
dimensions of the children. According to the designers of the
examined bicycles, these assumptions were mostly based on prior
experience and data from other non-Greek bicycle manufacturers.
This practice, obviously, does not take into account the vast indi-
vidual differences in the anthropometric dimensions of the Greek
children.
In the present study the ergonomic evaluation of the design of
a series of child bicycles, produced by a major Greek bicycle
manufacturer, based on Greek children anthropometric data is
presented. This evaluation involved the control of the geometry and
dimension of the bicycles’ frames as well as the control of the
permissible adjustment range for the saddle and the handlebar,
with respect to the specific population. The objective of the study
was to investigate the affordance of the present bicycle design for
the investigated population and if necessary to redesign the bicy-
cles so that the best possible fit could be obtained, thereby
increasing cycling safety and comfort for the young cyclists.
2. Methodology
Following the basic guidelines proposed by the Human Factors
and Ergonomics Society (HFES) for the application of anthropo-
metric data on product design (HFES 300 Committee, 2004) the
study involved three basic steps:
1. Problem statement, which includes target population defini-
tion, task analysis and the identification of critical design
parameters and relevant anthropometric dimensions.
2. Collection and analysis of anthropometric data of the target
population and the selection of representative case studies.
3. Investigation of the selected case studies and analysis of the
corresponding results.
The constraints that should be considered in bicycle design can
be classified into four categories, namely clearance, reach, strength
and posture constraints. Clearance constraints ensure that the
bicycle design and dimensions allow the adjustment of the posture
and free movement of the limbs, especially for the members of the
target population with the largest anthropometric dimensions.
Reach constraints on the other hand define the minimum accept-
able limits for basic bicycle dimensions that are imposed by the
users with the smallest anthropometric dimensions. Strength
constraints are associated mainly with the use of specific compo-
nents that require a certain amount of force to be applied, like the
brakes or the pedal. The amount of force required depends on the
429
design characteristics of the components themselves but also on
their relative position in the bicycle design.
Posture constraints define the angles between the anatomy
parts of the cyclist and its relative position on the bicycle (De Vey
Mestdagh, 1998). The cycling posture is a very important parameter
in terms of power production, energy efficiency, cycling comfort
and injury prevention and hence has been the subject of many
studies. Most of the studies related with bicycle fit focus on
competitive cycling (racing or trail cycling) and the relationship
between posture and power maximization, energy efficiency and
prevention of injuries which can be caused through long periods of
cycling (Berry et al., 2000; Dal Monte et al., 1987; Hull and Gonzalez,
1988; Too, 1990; Yoshihuku and Herzog, 1990). The relationship
between posture and the cyclist’s feeling of comfort or safety, which
is probably the main issue in every-day cycling and the design of
commuter bicycles, has been less studied (Christiaans and Bremner,
1998). Comfort is a concept of rather subjective nature but it can
generally be defined as the absence of pain and any other similar
nuisance and is usually associated in the relevant literature with the
design and adjustment of the saddle (Bressel and Larson, 2003;
Groenendijk et al., 1992).
All aforementioned studies were conducted on or consider
adults who, as it can be reasonably argued, ride bicycles differently
from children or teenagers. In their study, Donkers et al. focused
specifically on bicycles for relatively young children and assessed
their design and dimensions in terms of fit to the Dutch children
population and the associated safety regulations (Donkers et al.,
1993). According to their findings designers should relate the
bicycle dimensions to stature and not the age of the child. Based on
the results and recommendations by Donkers et al., as well as on
the results of the most relevant of the adult studies (Baker, 2000;
Christiaans and Bremner, 1998; De Vey Mestdagh, 1998) and
international practice (Bikefitting.com, 2006; White, 2006) the
reference cycling posture has been defined (Fig. 1a and b). The main
guidelines that have been adopted for the construction of the
reference posture are:
" The child should be slightly leaning forward, approximately 15#
in respect to the vertical axis. This leaning of the upper body
increases the respiratory volume because it transfers part of
shoulders weight to the arms and reduces the load to the
lower-back area, without overloading the arms and the wrists.
" The knee angle (the angle between the thigh and the crus)
should not be more than around 150# when the pedal is in the
bottom dead centre position and no less than 65# when the
pedal is in the top position. This constraint prevents knee
problems due to excessive stretching and flexing of the knee.
" The arms should be slightly bent in the elbows, i.e. the angle
between upper and lower arm should be about 20# to reduce
the effect of possible vibrations to the shoulders. Furthermore
the handlebar should be wider than the shoulders width for
better steering.
The above guidelines define a relative up-right posture that is
assumed to favour comfort compared to the low bending posture of
race cycling that favours cycling performance in terms of speed and
endurance. Although, the riding posture of children can be different
from the posture of adults, for the purposes of the present study the
reference posture has been selected on the bases of safety, comfort
and control for commuter cycling in urban and high-traffic areas.
The extent at which the above posture can be obtained by
a specific cyclist depends on the bicycle design and of course the
cyclist’s anthropometric dimensions. In Fig. 1c the critical bicycle
dimensions that were examined in the present study are sche-
matically presented. These dimensions are:
430 L. Laios, J. Giannatsis / Applied Ergonomics 41 (2010) 428–435

a a

fa
c b c
h A
a

t B D

l
C E

Fig. 1. Basic anthropometric and bicycle dimensions examined in the present study.

" the saddle-handlebar horizontal distance A, which depends on quite small. It is, therefore, apparent that besides the frame
the length of the top tube of the frame, the saddle setback and dimensions, the ranges of the saddle and handlebar adjustment
the length of the handlebar stem (if present), mechanisms are also very crucial for the evaluation of bicycle fit to
" the saddle-pedal distance, when the pedal is at the top a specific user population.
(dimension B) and bottom (dimension C) dead centre of the The basic anthropometric dimensions that were considered (see
rotation circle (usually referred also as saddle height), which Fig. 1a), are:
depends on the saddle height adjustment and the crank length,
" the handlebar height adjustment dimension D, and " the thigh length t, the lower leg length l and the foot height f
" the vertical distance E from the saddle to the ground that is that relate to dimensions B and C of the bicycle,
defined by the saddle height adjustment and the frame and " the chest height c and abdomen height ab that relate to
wheel sizes. dimensions A and D of the bicycle,
" the upper arm length a, the forearm length fa and the hand
Since anthropometric dimensions vary to a great extent, bicycle length h that also relate to dimensions A and D of the bicycle,
dimensions should also be adjustable to the maximum extent " the inseam length or ‘inside’ leg length, which is defined as the
possible. In standard bicycle design, followed by most bicycle distance between the ground and the pubis symphysis and
manufacturers, two main adjustment mechanisms are incorpo- relates to bicycle dimension E.
rated; namely the saddle height adjustment and the handlebar
height adjustment. The adjustment ranges of these mechanisms in As noted above certain bicycle dimensions relate to specific
conjunction with the constant dimensions of the frame directly anthropometric dimensions of the cyclist and thereby affect the
define also the ranges of dimensions B, C, D and E for a specific obtained posture. The saddle-handlebar distance A and the
bicycle. Saddle and handlebar adjustment indirectly affect also the handlebar height adjustment D relate to the torso dimensions,
saddle-handlebar distance A which is mainly associated with the namely the chest height and the abdomen height, as well to the
length top tube of the frame. In most bicycles is also possible to upper limb dimensions, namely the arm length, the forearm length
adjust the saddle position horizontally through the fore-aft and the hand length, since the combination of these dimensions
adjustment mechanism, but the permissible adjustment range is implicitly defines the cyclist’s leaning angle required for reaching

75

70

65
Maximum
Minimum
Optimum saddle height (cm)

60 Average

55

50 26”
24”
45
20”

40

35
120-124

125-129

130-134

135-139

140-144

145-149

150-154

155-159

160-164

165-169

170-174

>175

He ig ht g ro up (cm)

Fig. 2. Evaluated ‘optimum’ saddle height and saddle height adjustment ranges for the 20, 24 and 26 in bicycles.
 L. Laios, J. Giannatsis / Applied Ergonomics 41 (2010) 428–435 431

Table 1
Presentation of the height groups and corresponding data for characteristic anthropometric dimensions.

Height Bicycle Sample (n) Upper arm length (cm) Forearm length (cm) Hand length (cm)
group size
5% 50% 95% s 5% 50% 95% s 5% 50% 95% s
(cm) (inches)
120–140 20 607 20.65 23.41 26.80 2.111 17.40 19.69 22.67 1.759 13.93 15.57 17.71 1.285
140–170 24 630 24.17 27.39 31.69 2.310 19.40 22.58 26.28 2.155 14.85 16.90 19.58 1.496
$160 26 165 28.50 31.08 34.27 1.808 24.97 26.92 29.73 1.592 18.59 20.24 22.33 1.172

Height Bicycle Sample (n) Thigh length (cm) Lower leg length (cm) Foot height (cm)
group size
5% 50% 95% s 5% 50% 95% s 5% 50% 95% s
(inches)
120–140 20 607 26.45 29.86 33.34 2.274 25.87 29.59 33.34 2.350 6.16 6.81 7.46 0.408
140–170 24 630 31.55 35.23 39.76 2.536 31.45 35.10 39.45 2.481 6.53 7.21 7.88 0.419
$160 26 165 35.68 39.20 43.22 2.198 35.45 38.80 42.78 2.136 6.99 7.68 8.47 0.424

Height Bicycle Sample (n) Inseam length (cm) Chest height (cm) Abdomen height (cm)
group size
5% 50% 95% s 5% 50% 95% s 5% 50% 95% s
(inches)
120–140 20 607 52.60 59.92 66.20 4.584 23.31 27.75 31.42 2.665 12.68 15.31 17.63 1.544
140–170 24 630 63.25 70.67 79.74 4.983 23.65 28.67 32.69 2.851 14.26 18.17 23.27 2.776
$160 26 165 86.37 78.10 86.33 4.355 24.95 28.86 33.40 2.741 18.62 22.13 26.27 2.346

the handlebars. Likewise, saddle-pedal distances B and C, as defined
by the saddle height adjustment, in conjunction with lower limb
dimensions (thigh length, lower leg length and foot height),
implicitly define the knee angle range during pedal rotation. Saddle
height combined with inseam leg length affect also the difficulty of
the task of getting on and off the bicycle.
The user group under investigation in the present study
comprised of children and young teenagers. The variability of
anthropometric dimensions in this group is very wide; hence it
cannot be accommodated by a single bicycle and a series of bicycles
of different frame dimensions and corresponding adjustment
ranges were examined. For the specific user group three bicycle
sizes were recommended by the manufacturer, namely the 20, 24
and 26 bicycle sizes. The aforementioned sizes of the bicycles
correspond to the diameter of their wheels in inches.
3. Anthropometric data analysis
The target population for the investigated bicycle series range
from 7 years old children to 14 years old teenagers. The required
anthropometric data were extracted from the database of Greek
Somatometric Institute, which contains the anthropometric
dimensions of 1247 individuals that were measured between 2003
and 2005. The first step of the study was the evaluation of bicycle
fitness for the existing bicycles. In order to do this, the target
population was divided into three groups, one for each bicycle size,
according to their age. Common practice followed by the manu-
facturer assumed that the 16 in size bicycle was appropriate for
children between 5 and 7 years old, the 20 in for children between
used methods for the definition of optimum saddle height is the
Guimard–LeMond method (De Vey Mestdagh, 1998). According to
this method the optimum saddle height is calculated by multi-
plying the cyclist’s inseam length to a constant, the Guimard–
LeMond factor, which equals 0.883. Employing this method, the
‘optimum’ saddle height for each individual of the sample was
evaluated and compared to the maximum and minimum allowable
saddle height for each bicycle, thus providing an indication of the
height range that better fits each bicycle size. In Fig. 2 the ‘optimum’
saddle height values (derived with the Guimard–LeMond method)
for specific height groups of the sample are presented. For every
height group the maximum, minimum and average ‘optimum’
saddle height were evaluated and compared to the saddle height
adjustment ranges of each bicycle size, represented by the corre-
sponding horizontal lines.
It can be observed the 20 and 24 in bicycles fit best children with
height 120–140 cm and 140–170 cm respectively, while for tall
(>160 cm) children and/or young teenagers the 26 in size seems to
offer the best fit. As can also be observed the saddle height
adjustment range for the bicycles overlap in a large extent, some-
thing which is desirable, in order to cover individual preferences
and cycling style. Based on these observations, three new children
subgroups were formed according to their height. The
4
90% Ellipse 3
2X

7 and 10, the 24 in for children between 10 and 12 and the 26 in for
X
X
children older than 12 years old. However, a primary analysis of 1
FACTOR 2

Characteristic
anthropometric data showed great variability of dimensions with Points
0
respect to the age of the children, age therefore was not a particu- -3
X
-2 -1 0 1
X
2 3 4
larly accurate index of the child’s anthropometric dimensions. -1
Height on the other hand, seemed to offer a better indication of X
X
a child’s ‘size’, although there is of course no absolute correlation -2
X
between height and the other anthropometric dimensions.
In previous studies saddle height, i.e. the distance from the top -3

of the saddle to the centre of the pedal at the bottom rotation

extreme (dimension C in Fig. 1c), has been identified as one of the -4
FACTOR 1
most important variables in terms of power, efficiency and knee
injury prevention (Christiaans and Bremner, 1998; De Vey Mest- Fig. 3. Scatter plot of the evaluated PCA factors for the individuals of the ‘120–140’
dagh, 1998; Silberman et al., 2005). One of the most commonly group and the corresponding 90% ellipse.
432 L. Laios, J. Giannatsis / Applied Ergonomics 41 (2010) 428–435

Table 2 (comprised of the thigh length, the lower leg length and the foot
Nominal and mannequin measured values of basic anthropometric dimensions for height). Each of the 1247 individuals was scored using their original
the two boys (all values in cm).
dimensions and the two corresponding PC eigenvectors. Based on
Anthropometric Older boy Younger boy the results of the analysis the three graphs (scatter plots) of the PC
dimension
Nominal Mannequin Nominal Mannequin values were drawn, and the 2-D ellipses capturing the 90% of the
Chest height (c) 28.5 28.8 25.5 25.2 corresponding population were fitted to the corresponding pop-
Abdomen height (ab) 15.0 14.8 14.5 14.3 ulation scatter, employing Matlab software. The ellipse was
Upper arm length (a) 23.2 23.8 21.5 22.2 employed in order to exclude from the analysis individuals with
Forearm length (fa) 20.0 20.3 19.2 19.5
extremely rare and perhaps falsely measured anthropometric
Hand length (h) 14.8 15.0 13.8 13.6
Inseam length 60.0 61.8 56.0 57.2
dimensions. In Fig. 3 the PCA graph and the corresponding 90%
Thigh length (t) 31.0 30.7 29.5 29.8 ellipse for the ‘120–140’ group is presented.
Lower leg length (l) 31.5 31.9 30.0 30.6 The cases that were examined comprised of the individuals
Foot height (f) 6.7 6.4 5.6 5.7 whose points coincide or are very near the eight characteristic
points of the ellipse presented in Fig. 3. These cases, also referred as
‘border’ cases, represent individuals with unusual combinations of
categorization of the sample according to height, the associated anthropometric dimensions (e.g. a child with relatively short torso
bicycle size, the sample size for each group (n) as well as some data but long limbs or the opposite). In order to check also the recom-
regarding important for the study anthropometric dimensions, mended (default) saddle and handlebar adjustment of the bicycle,
including sample measurements for the 5%, 50% and 95% of the a central case representing the average user was also examined. As
group population and the corresponding standard deviation (s), are the central case the individual represented by a point coinciding or
summarized in Table 1. the nearest to the axis intersection point was selected.

4. Selection of representative case studies 5. Investigation of representative cases

The next step of the study concerned the selection of case
studies representative of the variability and range of anthropo-
metric dimensions in each subgroup. In order to identify the indi-
viduals that represent interesting case studies, a graphical
representation of the anthropometric dimensions distribution was
required (HFES 300 Committee, 2004). The problem of bicycle fit, as
defined in Section 2, is multi-dimensional; in order, therefore, to
make a graphical presentation possible a reduction of the multi-
dimensional space was required. For this purpose the method of
Principal Component Analysis (PCA) has been employed. Through
the application of PCA it is possible to group highly-correlated
dimensions into factors (Principal Components) that can be
subsequently used for a graphical representation of the distribution
of the users’ anthropometric dimensions.
PCA has been performed for the three sample groups and two
Principal Components (PC) were identified, the first representing
torso dimensions (comprised of the chest and abdomen heights)
and upper limb dimensions (comprised of the upper arm, forearm
and hand lengths), while the second representing leg dimensions
For the investigation of the selected cases, the method of 3D
virtual modelling has been applied, due to the anonymity of the
anthropometric database; hence it was not possible to contact the
children representing interesting case subjects and directly involve
them in the study. Virtual modelling and analysis offers also the
advantage of speed, since a sufficient number of cases can be
studied in relatively short time and with relatively good measure-
ment accuracy, a fact of great value for the bicycle manufacturers.
Assuming that the virtual anthropometric models (mannequins)
represented with sufficient accuracy the actual anthropometric
dimensions of the individuals, virtual modelling could therefore
provide useful insights and information that can further support
and enhance the results of actual testing with humans.
For the present study NexGen Ergonomics – Mannequin ergo-
nomic analysis software has been used for the construction of the
custom mannequins with user-defined anthropometric dimen-
sions. The anthropometric characteristics of the mannequins are
defined through the definition of main body part dimensions, such
as the width, breadth and length of the thigh, the lower leg, the

Fig. 4. Photograph of the older boy (a) and schematic representation of the corresponding mannequin (b) on the 20 in bicycle.
 L. Laios, J. Giannatsis / Applied Ergonomics 41 (2010) 428–435 433

Table 3
Anthropometric dimensions of the nine cases/mannequins examined for the ‘120–140’ height subgroup (all values in cm).

Anthropometric dimension 1 2 3 4 5 6 7 8 9
Height 140.0 140.0 140.0 134.0 122.0 124.0 121.0 131.0 130.0
Chest height (c) 33.8 27.3 28.8 24.2 22.5 23.5 26.4 31.6 27.5
Abdomen height (ab) 17.8 19.1 16.1 13.5 12.6 13.1 14.7 16.6 15.3
Upper arm length (a) 26.2 26.8 27.0 25.0 21.7 21.7 19.5 21.9 23.5
Forearm length (fa) 20.8 22.2 21.5 20.0 17.3 17.3 15.5 17.1 19.7
Hand length (h) 16.3 16.5 15.6 14.3 13.8 13.8 12.4 13.4 14.6
Inseam length 60.2 66.8 66.3 64.9 59.7 56.3 51.2 52.4 59.4
Thigh length (t) 30.2 33.0 34.3 32.7 30.1 28.2 25.6 25.8 30.0
Lower leg length (l) 29.5 33.8 33.0 33.5 29.7 27.7 25.2 25.9 29.6
Foot height (f) 6.7 6.6 7.2 7.8 7.2 6.7 6.1 5.8 6.8

abdomen height, etc. In order to verify the representation accuracy
of the virtual models, two young boys, age 7 (height 126 cm) and 9
years old (height 139 cm), were measured and their respective
custom mannequins were constructed. The critical dimensions of
the two mannequins were subsequently measured, employing the
measuring tools of the software. The nominal and the measured
values for the basic anthropometric dimensions of the two boys
examined are summarized in Table 2. Direct comparison between
these two values reveals slight differences, which were considered
within acceptable for the purposes of the study limits. The two
mannequins were subsequently positioned in the optimum posture
on the 3D CAD model of the 20 in bicycle and the models and
a basic examination of bicycle fitness has been performed.
This investigation revealed that the specific bicycle accommo-
dated quite well the younger boy, who could obtain the reference
posture when the seat and handlebar height were near their default
(middle) adjustment height. The older boy on the other hand, could
also obtain near-optimum posture, but this was achieved by
defining the seat and handlebar height near their respective
maximum allowable height (about 5 cm less for saddle height),
leaving, therefore, limited bicycle fit allowance for the eventual
growth of the child (refer to Fig. 4a). These basic conclusions were
confirmed by direct observation of the two children in actual
cycling and a subsequent discussion, in which it became apparent
that although both children could use the bicycle, the older boy felt
more constrained by its size and would prefer a relatively bigger
bicycle.
The first step in the virtual analysis of the dimensions of the
three bicycles was the construction of the associated nine virtual
mannequins of the children, representing the cases identified
through PCA. As noted above, these cases were selected so that the
wide variability of anthropometric dimensions for each group can
be accounted for. The dimensional variability for the ‘120–140’
height subgroup can be observed in Table 3, where the basic
anthropometric dimensions for the eight border cases (manne-
quins 1–8) and the average (mannequin 9) are presented.
The children mannequins were next placed in the reference
posture on the corresponding bicycle model, which was designed
using 3D CAD software and imported in the NexGen Ergonomics –
Mannequin software. In Fig. 5 two of the children mannequins for
the 20 in bicycle, one of relatively small size (Fig. 5a) and another
one of relatively large (Fig. 5b), are presented.
Having defined the position and posture of a mannequins it was
relatively straightforward to examine visually to what extent the
reference or near reference posture can be obtained and measure
possible deviations in distance and angles. Upon completion of the
analysis, the investigation results as well as some general recom-
mendations were communicated to the company’s designers and
engineers to study and devise possible ways of implementing them
in the design of new bicycles. For example a relatively large increase
in the saddle to handlebar distance can be achieved through
increasing the length of the top tube(-s) of the frame, while
a relatively smaller increase may be also accommodated through
increasing the stem of the handlebar. Regarding saddle height most
changes involved the adjustment of maximum and minimum
saddle position obtained in each bicycle. Assuming that the same
crank length is employed, these adjustments were translated either
to small changes to seat tube length or to the seat post length.
Another general observation concerned the maximum handlebar
height possible, which in most bicycles was relatively low. To

Fig. 5. Children mannequins of relatively small (a) and large (b) size from the ‘120–140’ group on the 20 in bicycle.
434 L. Laios, J. Giannatsis / Applied Ergonomics 41 (2010) 428–435

expected to minimize problems caused by improper bicycle fit, for

the specific bicycle series and the investigated user population. It
should be noted that for the size selection, the anticipated child
growth should also be taken into account; hence, when a child
dimensions are near the upper adjustment limits of a bicycle size
and further child growth is anticipated, the selection of the larger
size would probably be the best choice.

6. Discussion and conclusions

The main objective of the study presented in this paper was to

investigate and possibly to improve the fitness of commercial
children bicycles for the Greek market. The investigation focused
on five critical bicycle dimensions; the saddle-handlebar horizontal
distance, the saddle height and its adjustment range, the handlebar
height and its adjustment range, the handlebar width and the
Fig. 6. Overlapping photos of the old and the redesigned 26 in bicycles. vertical distance from the saddle to the ground. Nine basic
anthropometric dimensions of children for the assessment of the
limbs length and the torso height were examined. Since the range
increase this adjustment setting either an increase in the steering
of these anthropometric dimensions in the target group of the
tube length or to the handlebar post was proposed. Regarding the
study, namely children age 7–14, is quite large, three bicycles with
handlebar width, the case study analysis showed that in nearly all
variable size were examined. Each bicycle size is targeted to
cases the handlebars were wider from the shoulder breadth
a specific children subgroup defined by their height. Bicycle
complying, therefore, with the constraints.
investigation involved the examination of nine cases-children
For the 20 and 24 in bicycles a reduction of 3# –5# of the seat tube
representative of the anthropometric dimensions of the corre-
angle has been proposed, in order to facilitate comfortable riding
sponding height subgroup. Since there were multiple relevant
and the perceived frequent changes of posture into more ‘out-of-
anthropometric dimensions the method of Principal Component
the-saddle’ riding that can be expected by relatively old children
Analysis was applied to reduce the problem to two dimensions and
(>10 years old) and young teenagers. This reduction of seat angle
hence make possible the graphical representation of data and the
led to an increase in saddle to handlebar distance, which in most
selection of cases.
cases was also desirable. For the 26 in bicycle recommended
For the assessment of fit of each bicycle size to the corre-
modifications included an increase of the average saddle and
sponding children subgroup the method of virtual modelling and
handlebar height as well as a small decrease in the saddle-
analysis was employed. Virtual modelling method was selected
handlebar distance. The implementation of these changes in
because it offers the advantage of cost-effective and fast exami-
combination with technical enhancements, such as the addition of
nation of multiple cases, assuming that the virtual mannequins
a shock absorber, resulted to the redesigned bicycle. The old and
accurately represent the actual human dimensions. In the present
redesigned 26 in bicycles are presented in overlapping mode in
study NexGen Ergonomics – Mannequin software was used. An
Fig. 6.
initial examination of the mannequin representation showed that
Besides the above described ‘fine-tuning’ of bicycle dimensions
mannequins reflected with sufficient, for the purposes of the study,
and adjustable ranges, company engineers noted that the study
accuracy the nominal dimensions of the children. Virtual analysis
aided them in defining a formal method for bicycle size selection
involved the placement of static mannequins in an ‘optimum’
for children. A common mistake when purchasing a new bicycle,
posture on 3D CAD bicycle models and the measurement of the
made usually by both the parents and salespersons, is to select its
associated deviations. The mannequins’ posture was defined
size based only on the age of the child. Size selection according to
according to recommendations for maximum safety and comfort of
the child’s age does not of course take into consideration the
previous studies on adult and children cyclists. Based on the
specific anthropometric dimensions as well as differences in frame
evaluation results adjustments in the dimensions of the frames
dimensions between bicycle manufacturers and may therefore be
and the saddle/handlebar height ranges were proposed. In addi-
misleading. To address this issue a general guide for selecting
tion to the proposed adjustments, a formal guide for the selection
bicycle size in the form of Table 4 has been constructed.
of bicycle size was also proposed. Purpose of the guide was to help
The application of this guide in conjunction with providing
the selection of bicycle size based on actual anthropometric
general fit and adjustment guidelines to prospective customers is
dimensions of the child, namely its height and inseam length. This
guide is aimed of course to the investigated population and the
specific bicycle series and should not be considered as a definite
Table 4 selection guide of bicycle size for other manufacturers and/or user
Size selection table according to child’s height and inseam length for the redesigned
groups.
bicycles.
Experience gained through the present study indicates that
Height (cm) Inseam length (cm) Bicycle size (in) virtual modelling and analysis, based on a representative range of
120–140 <52 Smaller size (16) selected anthropometric data, could be a valuable tool for the
52–74 20
bicycle design process. According to the designers involved, virtual
>74 24
modelling provides valuable insight in the initial design stages,
141–160 <62 20 where the main dimensions of the frame and the adjustment
62–80 24
ranges for saddle and handlebar are defined. However, due to
>80 26
representation accuracy issues and its relatively static nature,
>160 %74 24 virtual modelling results should be subsequently confirmed and
>74 26
supported by actual testing with human subjects representing the
 L. Laios, J. Giannatsis / Applied Ergonomics 41 (2010) 428–435
full range of the indented users. This can be a topic of a subsequent
research that poses some problems because the researchers must
identify a number of ‘‘boundary and internal cases’’ each defined
from various individual anthropometric dimensions. For the
purposes of the present first study this would prolong the duration
of the assignment, something not desirable by the bicycle manu-
facturer. Regarding the main problem of bicycle fit, it was observed
that most of the relevant literature was focused on adult pop-
ulations and mainly on sport cycling, where efficiency is the main
issue. Further investigation is, therefore, required for the verifica-
tion of the proposed ‘optimum’, in terms of safety and comfort,
posture for cycling in urban areas, as well as for the identification of
possible differences in cycling style and habits between children
and adults.
Acknowledgments
The authors would like to thank the Greek Somatometric
Institute for providing the data necessary for the study, and Field
S.A. bicycles manufacturing company for their support to the joint
bicycle design project.
References
Baker, A., 2000. Medical problems in road cycling. In: Gregor, R.J., Conconi, F. (Eds.),
Road Cycling. Blackwell Publ., London, pp. 68–118.
Berry, M.J., Koves, T.R., Benedetto, J.J., 2000. The influence of speed, grade and mass
during simulated off road bicycling. Appl. Ergon. 31 (5), 531–536.
Bikefitting.com, 2006. Bicycle fit theory. Available at: <http://www.bikefitting.com/>.
435
Bressel, E., Larson, B.J., 2003. Bicycle seat designs and their effect on pelvic angle,
trunk angle and comfort. Med. Sci. Sports. Exerc. 35 (2), 327–332.
Christiaans, H.H.C.M., Bremner, A., 1998. Comfort on bicycles and the validity of
a commercial bicycle fitting system. Appl. Ergon. 29 (3), 201–211.
Dal Monte, A., Leonardi, L.M., Menchinelli, C., Marini, C., 1987. A new bicycle design
based on biomechanics and advanced technology. J. Appl. Biomech. 3 (3),
287–292.
De Vey Mestdagh, K., 1998. Personal perspective: in search of an optimum cycling
posture. Appl. Ergon. 29 (5), 325–334.
Donkers, C.M., Toussaint, H.M., Molenbroek, J.F.M., Steenbekkers, L.P.A., 1993.
Recommendations for the assessment and design of young children’s bicycles
on the basis of anthropometric data. Appl. Ergon. 24 (2), 109–118.
Groenendijk, M.C., Christiaans, H.H.C.M., Van Hulten, C.M.J., 1992. Sitting comfort on
bicycles. In: Lovesey, E.J. (Ed.), Contemporary Ergonomics. Taylor and Francis,
London, pp. 551–557.
HFES 300 Committee, 2004. Guidelines for Using Anthropometric Data in Product
Design. Human Factors Ergonomics Society, Santa Monica, CA.
Hull, M.L., Gonzalez, H.K., 1988. Bivariate optimization of pedalling rate and crank
arm length in cycling. J. Biomech. 21 (10), 839–849.
Kolehmainen, I., Harms-Ringdahl, K., Lanshammart, H., 1989. Cervical spine posi-
tions and load moments during bicycling with different handlebar positions.
Clin. Biomech. 4 (2), 105–110.
Mellion, M.B., 1991. Common cycling injuries. Management and prevention. Sports.
Med. 11 (1), 52–70.
Nemours Foundation, 2006. Bike safety. Available at: <http://www.kidshealth.org/
parent/fitness/safety/bike_safety.html>.
Schwellnus, M.P., Derman, E.W., 2005. Common injuries in cycling: prevention,
diagnosis and management. S. A. Fam. Pract. 47 (7), 14–19.
Silberman, M.R., Webner, D., Collina, S., Shiple, B.J., 2005. Road bicycle fit. Clin. J.
Sport. Med. 15 (4), 271–276.
Too, D., 1990. Biomechanics of cycling and factors affecting performance. Sports.
Med. 10 (5), 286–302.
White, P.J., 2006. How to fit a bicycle. Available at: <http://www.peterwhitecycles.
com/fitting.htm>.
Wilson, D.G., 2004. Bicycling Science. MIT Press.
Yoshihuku, Y., Herzog, W., 1990. Optimal design parameters of the bicycle-rider
system for maximal muscle power output. J. Biomech. 23 (10), 1069–1079.