ARTICLE IN PRESS

Applied Ergonomics 39 (2008) 392398

www.elsevier.com/locate/apergo

Ergonomic effects of load carriage on the upper and lower back on

metabolic energy cost of walking
Daijiro Abea,, Satoshi Murakib, Akira Yasukouchib
a

Faculty of Human Sciences, University of East Asia, 2-1 Ichinomiya Gakuen-cho, Shimonoseki, Yamaguchi 751-8503, Japan
Research Laboratory of Ergonomics, Faculty of Design, Kyushu University, 4-9-1 Shiobaru, Minami-ku, Fukuoka 815-8540, Japan

Received 19 May 2006; accepted 19 July 2007

Abstract
We examined the effects of load carriage position on the energy cost of walking dened as the ratio of the 2-min steady-state oxygen
consumption to the speed and economical speed. Fourteen healthy men walked on a treadmill at various speeds without and with load on
the lower and upper back, which corresponded to 15% of their body mass. The energy cost of walking signicantly decreased during
walking with load than without load at slower speeds. A signicant decrease in the energy cost of walking was also observed while
carrying the load on the upper back than on the lower back at 6080 m/min. The economical speed signicantly decreased when carrying
the load on the upper and lower back, and it was signicantly correlated with body height. These ndings suggest that an optimal
carrying method is evident to reduce physical stress during walking with loads.
r 2007 Elsevier Ltd. All rights reserved.
Keywords: Locomotion; Optimal speed; Free-ride

1. Introduction
Energy expenditure during bilateral locomotion in
humans with and without load in a variety of occupations
has been broadly investigated (e.g. Knapik et al., 2004).
There have been two completely opposite opinions with
regard to the metabolic energy cost of walking during load
carriage. One of them is that the energy expenditure during
walking increases linearly with the weight of the carrying
load (Soule and Goldman, 1969; Gordon et al., 1983;
Francis and Hoobler, 1986). In contrast, other studies have
found that the energy expenditure did not necessarily
increase in proportion to the weight of the load (Maloiy
et al., 1986; Charteris et al., 1989a, b; Heglund et al., 1995;
Stuempe et al., 2004; Bastien et al., 2005a). For example,
it was interesting to note that the oxygen consumption did
not increase when the load was carried on the head if the
load was less than 20% of the subjects body mass
(Charteris et al., 1989a; Heglund et al., 1995). Those
authors termed such a phenomenon as free-ride. Bastien
Corresponding author. Tel.: +81 832 57 5166; fax: +81 832 56 1485.

E-mail address: daijiro@toua-u.ac.jp (D. Abe).

0003-6870/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.apergo.2007.07.001

et al. (2005a) recently found that a typical Nepalese porter

could carry an extraordinary load of up to 60% of the
body mass without a signicant increase in energy
expenditure. These previous studies suggested the importance of the position of the load and walking speed close to
the economical speed during load carriage (Bastien et al.,
2005b).
Abe et al. (2004) proposed that such a phenomenon
could be due to an interaction between the rotative torque
around the center of body mass treated as a positive effect
and a concomitant excessive burden on the lower leg
extremities treated as a negative effect (Fig. 1). The rotative
torque around the center of the body mass can be dened
as follows:
E AB  load weight,
where E is the rotative torque and AB is a radius of
rotation (Fig. 1). Thus, if the load position was located on
the upper or lower back, then it could be possible to adjust
the rotative torque without changing the load weight. If the
possible mechanism to explain free-ride is an interaction
between rotative torque and excessive burden on the lower
leg extremities, then the higher the position of the load, the

ARTICLE IN PRESS
D. Abe et al. / Applied Ergonomics 39 (2008) 392398

393

called the economical speed or optimal speed (Falola et al.,

2000; Bastien et al., 2005b). Factors affecting economical
speed seemed to be dependent on the leg length and gravity
(Grifn et al., 1999), however, as Bastien et al. (2005b)
discussed, little has been unveiled in the relationship
between load carriage and the economical speed. If the
rotative torque contributes to the forward propulsion
during walking, then the economical speed will also
increase during walking with load, in particular, on the
upper back. Thus, the second purpose of this study was to
examine the effects of load carriage on the economical
speed. It was hypothesized that the economical speed
would be greater during walking with load on the upper
back than carrying on the lower back and/or without load.
The economical speed may be estimated by the physical
characteristics using the Froude number system (Donelan
and Kram, 1997). The Froude number system is calculated
with leg length, and it is expected that leg length and body
height be correlated, suggesting that the body height is
expected to correlate to the economical speed. The third
purpose of this study was to examine how to estimate the
economical speed when the load was carried, on a
hypothesis that the body height will be the factor for
estimating the economical speed.
2. Methods
2.1. Subjects
Fig. 1. Schematic description of an interaction between the rotative
torque functioning around the center of body mass and the concomitant
excessive burden on the lower leg muscles during walking with load on the
back. A, center of body mass; B, center of mass of load; AB, radius of
rotation; C, rotation arc; D, excessive burden on the lower leg muscles; E,
rotative torque functioning around the center of body mass (Abe et al.,
2004).

lower the energy expenditure, because the rotative torque,

represented as E in Fig. 1, will contribute to the forward
propulsion (dened as horizontal acceleration of the body)
during walking. To test such a mechanism, the rst
purpose of this study was to examine the effects of load
carriage on the lower and upper back at various speeds
on the metabolic energy cost of walking. It was hypothesized that the metabolic energy cost of walking was lesser
when the load was carried on the upper back of the
subjects.
The most important factor of the energy-conserving
mechanism during walking has been explained by the
transfer efciency between kinetic energy (Ek) and gravitational potential energy (Ep) (Cavagna et al., 1963). Minetti
et al. (1995) reported that the transfer efciency between Ek
and Ep became maximal at around 7080 m/min during
walking, and it decreased gradually when walking at slower
and/or faster than that speed. Thus, it is expected that there
exists a specic speed that can minimize the metabolic
energy cost of walking. The walking speed corresponding
to the minimum energy cost per unit distance has been

Fourteen healthy young male (age 20.970.8 years old;

height 170.676.2 cm; body mass 59.673.6 kg) volunteered
for this study. These subjects were recruited based on their
similar body mass in order to equalize the physical stress
imposed by the load. Informed consent from each subject
and the approval from the ethical review committee were
obtained. One subject was excluded from a series of
measurements due to a health issue.
2.2. Measurements
Walking tests were performed on a motor-driven treadmill (Biomill BL-1000, S & ME, Tokyo). To become
accustomed to treadmill walking wearing a gas collection
mask, each subject performed at least two preliminary
trials on the same treadmill at several speeds. The speeds
selected were 30120 m/min with increments of 10 m/min,
meaning that previously employed walking speeds were
covered in the present study. The subjects performed 5-min
trials at each speed with and without load on the back. The
subjects wore underwear, shirts, socks, gym shorts, and
lightweight training shoes. The measurements were performed once a day on each subject.
The load to be carried consisted of a sand bag on a net
weight of 2.0 kg position-adjustable lightweight aluminum
frame (Carry-Bone EBB003, Evernew, Tokyo) with an
elastic belt (Figs. 2(a) and (b)). The total weight of the load
was set at 15% of each subjects body mass. The sand bag

ARTICLE IN PRESS
394

D. Abe et al. / Applied Ergonomics 39 (2008) 392398

of the steady-state V_ O2 to the walking speed in m/min:

C w V_ O2 =walking speed:
As shown by previous studies (Zamparo et al., 1992; Abe
et al., 2004; Bastien et al., 2005b), the relationship between
walking speed and Cw can be approximated by a U-shaped
quadratic equation, meaning that there exists a specic
walking speed corresponding to the minimum energy cost
per unit distance. The relationship between walking speed
and Cw can be mathematically described as the following
equation:
C w awalking speed2 bwalking speed c,
where ac are the constants determined by the least squares
with the actually observed Cw values at each walking speed.
Therefore, a differential equation of the original quadratic
equation of each experimental condition could be described
as follows:
C w 0 2awalking speed b.
The economical speed was then determined at the speed
when C w 0 equals zero. That is, the economical speed could
be observed as follows:
economical speed b=2a.
The estimated economical speed can be obtained using
the Froude number, which has been used to predict the
biomechanics of legged locomotion over a wide range of
body size, gravity and speed (Donelan and Kram, 1997).
For human walking, the inertial force commonly employed
is the centripetal force. The Froude number has been
considered to be a dimensionless ratio of inertial force to
the gravitational force, and is dened as follows:
Froude number body mass  walking speed2 =l=
body mass  g,
where l is the leg length, dened from the trocanter major to
the ground, and g the gravitational acceleration (9.81 m/s2).
Thus, this equation can be shortened as follows:
Fig. 2. Position-adjustable aluminum frame. Back view (a) and side view
(b).

was placed at the upper (U) and lower (L) end of the frame.
The control condition (C) was prepared with no load and
the order of three load conditions (U, L and C) was
randomized. Each subject chose a preferred step frequency
at each experimental condition.
Breath-by-breath oxygen consumption (V_ O2 in ml/kg/
min) was measured with a gas analyzer (AE-300S, Minato
Co. Ltd., Osaka), which was calibrated before the tests
with room air and reference gases of known concentrations. A single sample with an average 2-min V_ O2 value at
each walking speed was calculated to obtain the values of
energy cost of walking per unit distance (Cw in ml/body
mass+load/meter). The Cw was determined from the ratio

Froude number walking speed2 =9:81l.

The walking speed at which the highest exchange between
Ep and Ek occurred was when the Froude number was 0.25
(Minetti, 2001). The estimated economical speed was then
obtained as follows:
Estimated economical speed 1:566l 0:5 .
2.3. Statistical analysis
Cw values of each trial as function of speed were
interpolated by a quadratic equation as previously done by
Zamparo et al. (1992). Differences in the observed Cw
values at each walking speed were compared by one-way
analysis of variance. When the signicant F values were
present, Fishers least squares test was applied as the post
hoc test. The relationships between body height and
economical speed in each condition were evaluated by a

ARTICLE IN PRESS
D. Abe et al. / Applied Ergonomics 39 (2008) 392398

simple regression analysis. A comparison between the

economical speed obtained from the C condition and the
estimated economical speed was evaluated using a t-test.
The statistical signicance was established at the 0.05
probability level.
Table 1
Summary of statistically signicant results
Cw

Speed (m/min)

30

40

50

60

70

80

90

100

110

120

L Condition
U Condition

*
*

*
*

*
*

*#
*#

#
*#

#
*#

*Signicantly different from the control condition; # signicant difference

between the L and U conditions; not signicantly different from other
conditions.

395

3. Results
The statistical results are summarized in Table 1. Fig. 3
shows the relationship between the walking speed and Cw
for the three conditions. The regression line for the average
data did not seem to be an ideal quadratic shape, because
the Cw values obtained from four subjects always appeared
to be a collapsed U-shape in each condition. However, the
correlation coefcients of the quadratic relationship
between Cw and walking speed ranged from 0.93 to 0.99
in all subjects. At 6080 m/min, the Cw values obtained
from L and U conditions were signicantly lower than
those obtained from the C condition. This trend was
observed until 60 m/min in the L condition and 100 m/min
in the U condition (Fig. 3). The Cw values were signicantly lower in the U condition than in the L condition at
6080 m/min.
The economical speeds for U, L, and C conditions were
77.873.8, 78.172.8, and 81.473.4 m/min, respectively. A
signicant difference for the economical speed was
observed between C and other conditions. Compared to
the C condition, the percent decreases in the economical
speeds were 4.472.9% and 4.173.4% for the U and L
conditions, respectively. There was a signicant linear
relationship between the body height and economical
speed obtained during the U and C conditions (Figs. 4(a)
and (c)), but not during the L condition (Fig. 4(b)).
Signicant relationships were also observed between the
economical speed obtained during the C condition and
estimated economical speed (Fig. 5), although the average
estimated economical speed (88.472.9 m/min) was signicantly greater than the economical speed obtained during
the C condition.
4. Discussion

Fig. 3. Metabolic energy cost of walking (Cw) as a function of walking

speed. Statistically signicant results were summarized in Table 1.

We examined the effects of load position on the energy

cost of walking per unit distance as a function of speed. In
support of our rst hypothesis, the result of the present

Fig. 4. Relationships between body height and the individuals economical speed. Each panel represents the U, L, and C conditions.

ARTICLE IN PRESS
396

D. Abe et al. / Applied Ergonomics 39 (2008) 392398

Fig. 5. Relationship of the economical speeds obtained during the C

condition and Froude number system. Dotted line indicated identity line.

study was primarily characterized by the nding that

signicantly lower Cw values were observed in the U
condition than in the L condition at 6080 m/min (Fig. 3
and Table 1). These results are partly consistent with that
of Stuempe et al. (2004). The result of the present study
also conrmed that lower Cw values were observed in the
loaded conditions than in the unloaded condition (Abe
et al., 2004).
Previous studies have examined the metabolic responses
for different carrying methods and the weight of the load
carriage (Soule and Goldman, 1969; Kinoshita, 1985;
Francis and Hoobler, 1986; Hong and Cheung, 2003;
Stuempe et al., 2004; Abe et al., 2004). As reported by
Cavagna et al. (1963), the time courses of the gravitational
potential energy (Ep) and kinetic energy (Ek) were almost
out of phase during walking, and it was assumed that the
most signicant factor affecting the energy-conserving
mechanism during walking was explained by the pendulum-like energy transfer efciency between Ep and Ek. In
the present study, both the total mass and height of the
center of mass in the U and L conditions increased
compared to that of the C condition because the load was
carried on the back, indicating that the Ep must have also
increased in the U and L conditions than in the C
condition. It goes without saying that the increased Ep
could be larger in the U than in the L condition, because
the load was located relatively higher in the U condition
than in the L condition. The difference between these
conditions was expected to yield a difference in the
production of the rotative torque (Fig. 1). The increased
Ep could cause an increase in Ek if the transfer efciency
between Ep and Ek did not alter regardless of the load
carriage. Indeed, it was well known that the highest
transfer efciency between Ep and Ek was observed at
around 6080 m/min in adult men (Minetti et al., 1995;
Willems et al., 1995), assuming that the increased Ep could,
to a greater or lesser extent, have contributed to an increase
in Ek in our study. In other words, the increased rotative

torque functioning around the center of body mass would

contribute to an increase in the Ek in the U condition than
in the L condition, and it would save energy expenditure
during walking with load on the upper back at around
6080 m/min.
It was recently reported that the recoil of the elastic
energy acted as one of the energy-conserving mechanisms
(Fukunaga et al., 2001; Lichtwark et al., 2007; Orendurff
et al., 2005; Sasaki and Neptune, 2006). Our rst
hypothesis was that the vertical load as an excessive
burden could induce a negative effect for the forward
propulsion due to the requirement of more energy
expenditure by the lower leg muscles. However, if the
recoil of the elastic energy could be counted on as one
of the energy-conserving mechanisms during walking
regardless of the walking speed, then the vertical load
might also contribute as one of the energy-conserving
factors during walking by way of such physiological
mechanisms.
The economical speed without load obtained from the C
condition was 81.473.4 m/min, and was consistent with
the result of Bastien et al. (2005b). The results of the
present study were secondly characterized by the fact that
the economical speed was signicantly lower by around 4%
when the load was carried on the back. Thus, the second
hypothesis of this study was rejected. When obtaining the
economical speed from the quadratic relationship between
walking speed and Cw values (Fig. 3), the Cw values were
signicantly lower in the U and L conditions at slower
speeds than in the C condition, but not at faster speeds. It
resulted in a leftward shift of the U-shaped curved line for
the relationship between Cw values and walking speed. As
discussed before, the experimental setup of this study was
especially characterized by the fact that the difference of
the load position could induce an experimental manipulation of the rotative torque functioning around the center of
the body mass without a difference of the vertical load on
the legs (Fig. 1). There were no signicant differences in the
economical speed between U and L conditions, although
the economical speed obtained from the C condition was
signicantly higher than that obtained from other conditions. It was considered that the economical speed obtained
from each condition ranged around 80 m/min, meaning
that the change in the economical speed by the load
carriage could be practically negligible.
It was also interesting to note that the economical speed
was signicantly correlated with body height in the U and
C conditions (Figs. 4(a) and (c)), but not in the L condition
(Fig. 4(b)). Thus, the third hypothesis of this study was
partly supported. The present study suggests that the
economical speed was inuenced by the body height. This
was a logical outcome because the body height and leg
length were correlated (r 0.93, po0.05).
Fig. 5 further showed a signicant relationship between
the economical speed obtained from the C condition and
estimated economical speed. There has been no information available for a conventional estimation of the

ARTICLE IN PRESS
D. Abe et al. / Applied Ergonomics 39 (2008) 392398

economical speed. The Froude number system could be a

potential tool for estimating the economical speed. A
conventional estimation of the individual economical speed
will be useful for reducing the early onset of fatigue.
However, it was pointed out that the economical speed
obtained from estimation by the Froude number system
was considerably higher than that obtained from control
condition (Fig. 5). Note that the at 0.25 might not be
appropriate to estimate the economical speed at 1 times
gravity. If the average leg length (0.901 m) and economical
speed (1.357 m/s), which were observed in our study, were
used to calculate the Froude number, then the average
value of the Froude number became 0.208. This lower
Froude number was further supported by the results of
Bastien et al. (2005b). Based on the dynamic similarity
hypothesis, previous studies neglected the subjects muscle
strength of the lower leg extremities and/or the difference
of the body dimension in relation to the racial difference
when assessing the Froude number (Donelan and Kram,
1997; Grifn et al., 1999; Minetti, 2001). The Froude
number system derived from the dynamic similarity
hypothesis was established based on the simulated reduced
gravity environment. That is, it is assumed that the relative
importance of the muscle strength necessary to ght
against the gravity could be smaller in the simulated
reduced gravity environment than that on the earth. A
conventional estimation for the economical speed was also
possible in the U condition (Fig. 4(c)), but not in the L
condition (Fig. 4(b)). These results suggest that the
estimation of the economical speed in the simulated added
gravity environment needs some attention.
The carrying method with load on the back and in front
of the body is called doublepack system (Kinoshita,
1985). Here, it is important to note that the free-ride
phenomenon requires a particular condition that the load
weight was less than 20% of the body mass carried on the
head or on the back (Charteris et al., 1989a; Abe et al.,
2004, 2007). If the doublepack system is employed during
walking with load, the forward propulsion due to the
rotative torque will be cancelled. However, the body
posture and gait pattern for the doublepack system were
quite similar to those for the normal walking (Kinoshita,
1985), suggesting that the doublepack system, rather than
the normal backpack system, can be recommended for
carrying the load if the load weight exceeds more than 20%
of the body mass. That will be also available for reducing
early onset of fatigue. For another practical application
of the result of the present study for occupational safety
and/or military purposes, the heavier load should be
located on the upper back during walking, being associated
with slower walking speeds at around 6080 m/min. In
particular, this study showed possible information to
estimate the individuals economical speed based on the
body height and/or leg length. Information about the
alteration of the energy expenditure by the load position
could be available for reducing early onset of fatigue
during working.

397

5. Conclusion
Signicantly lower values of the energy cost of walking
per unit distance were observed when the load was carried
on the upper back than on the lower back at 6080 m/min.
A signicant decrease in the economical speed was
observed when the load was carried on the back regardless
of the location of the load. The observed economical speed
was signicantly correlated with the simulated economical
speed obtained by the Froude number system. However,
this resulted in overestimates of economical speed. The
present results indicated that the load should be carried on
the upper back at speeds of 6080 m/min to reduce
physiological stress.
Acknowledgements
This study was supported by a Grant-in-Aid for Young
Scientist (B) from the Japan Ministry of Education,
Culture, Sports, Science, and Technology (Grant no.
17770217 to DA) and in part by the Nakatomi Foundation
(to DA).
References
Abe, D., Yanagawa, K., Niihata, S., 2004. Effects of load carriage, load
position, and walking speed on energy cost of walking. Appl. Ergon.
35, 329335.
Abe, D., Muraki, S., Yasukouchi, A., 2007. Ergonomic effects of load
carriage on energy cost of gradient walking. Appl. Ergon., in press,
doi:10.1016/j.apergo.2007.06.001.
Bastien, G.J., Schepens, B., Willems, P.A., Heglund, N.C., 2005a.
Energetics of load carrying in Nepalese porters. Science 308, 1755.
Bastien, G.J., Willems, P.A., Schepens, B., Heglund, N.C., 2005b. Effect
of load and speed on the energetic cost of human walking. Eur.
J. Appl. Physiol. 94, 7683.
Cavagna, G.A., Saibene, F.P., Margaria, R., 1963. External work in
walking. J. Appl. Physiol. 18, 19.
Charteris, J., Scott, P.A., Nottrodt, J.W., 1989a. Metabolic and kinematic
responses of African women headload carriers under controlled
conditions of load and speed. Ergonomics 32, 15391550.
Charteris, J., Nottrodt, J.W., Scott, P.A., 1989b. The free-ride
hypothesis: a second look at the efciency of African women headload
carriers. S. Afr. J. Sci. 85, 6871.
Donelan, J.M., Kram, R., 1997. The effect of reduced gravity on the
kinematics of human walking: a test of the dynamic similarity
hypothesis for locomotion. J. Exp. Biol. 200, 31933201.
Falola, J.M., Delpech, N., Brisswalter, J., 2000. Optimization characteristics of walking with and without a load on the trunk of the body.
Percept. Mot. Skills 91, 261272.
Francis, K., Hoobler, T., 1986. Changes in oxygen consumption
associated with treadmill walking and running with light hand-carried
weights. Ergonomics 29, 9991004.
Fukunaga, T., Kubo, K., Kawakami, Y., Fukashiro, S., Kanehisa, H.,
Maganaris, C.N., 2001. In vivo behaviour of human muscle tendon
during walking. Proc. R. Soc. Lond. B 268, 229233.
Gordon, M.J., Goslin, B.R., Graham, T., Hoare, J., 1983. Comparison
between load carriage and grade walking on a treadmill. Ergonomics
26, 289298.
Grifn, T.M., Tolani, N.A., Kram, R., 1999. Walking in simulated
reduced gravity: mechanical energy uctuations and exchange. J. Appl.
Physiol. 86, 383390.

ARTICLE IN PRESS
398

D. Abe et al. / Applied Ergonomics 39 (2008) 392398

Heglund, N.C., Willems, P.A., Penta, M., Cavagna, G.A., 1995. Energysaving mechanics with head-supported loads. Nature 375, 5254.
Hong, Y., Cheung, C.K., 2003. Gait and posture responses to backpack
load during level walking in children. Gait Posture 17, 2833.
Kinoshita, H., 1985. Effects of different loads and carrying systems on
selected biomechanical parameters describing walking gait. Ergonomics 28, 13471362.
Knapik, J.J., Reynolds, K.L., Harman, E., 2004. Soldier load carriage:
historical, physiological, biomechanical, and medical aspects. Mil.
Med. 169, 4556.
Lichtwark, G.A., Bougoulias, K., Wilson, A.M., 2007. Muscle fascicle
and series elastic element length changes along the length of the
human gastrocnemius during walking and running. J. Biomech. 40,
157164.
Maloiy, G.M., Heglund, N.C., Prager, L.M., Cavagna, G.A., Taylor,
C.R., 1986. Energetic cost of carrying loads: have African women
discovered an economic way? Nature 319, 668669.
Minetti, A.E., 2001. Invariant aspects of human locomotion in different
gravitational environments. Acta Astronaut. 49, 191198.

Minetti, A.E., Capelli, C., Zamparo, P., di Prampero, P.E., Saibene, F.,
1995. Effects of stride frequency on mechanical power and energy
expenditure of walking. Med. Sci. Sports Exerc. 27, 11941202.
Orendurff, M.S., Segal, A.D., Aiona, M.D., Dorociak, R.D., 2005.
Triceps surae force, length and velocity during walking. Gait Posture
21, 157163.
Sasaki, K., Neptune, R.R., 2006. Muscle mechanical work and elastic
energy utilization during walking and running near the preferred gait
transition speed. Gait Posture 23, 383390.
Soule, R.G., Goldman, R.F., 1969. Energy cost of loads carried on the
head, hands, or feet. J. Appl. Physiol. 27, 687690.
Stuempe, K.J., Drury, D.G., Wilson, A.L., 2004. Effect of load position
on physiological and perceptual responses during load carriage with an
internal frame backpack. Ergonomics 47, 784789.
Willems, P.A., Cavagna, G.A., Heglund, N.C., 1995. External, internal
and total work in human locomotion. J. Exp. Biol. 198, 379393.
Zamparo, P., Perini, R., Orizio, C., Sacher, M., Ferretti, G., 1992. The
energy cost of walking or running on sand. Eur. J. Appl. Physiol. 65,
183187.