This article was downloaded by: [University of Kent]

On: 09 November 2014, At: 08:47

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number:
1072954 Registered office: Mortimer House, 37-41 Mortimer Street,
London W1T 3JH, UK

Ergonomics
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/terg20

Effects of box features

on spine loading during
warehouse order selecting
W. S. MARRAS , K. P. GRANATA , K. G. DAVIS ,
W. G. ALLREAD & M. J. JORGENSEN
Published online: 10 Nov 2010.

To cite this article: W. S. MARRAS , K. P. GRANATA , K. G. DAVIS , W. G.

ALLREAD & M. J. JORGENSEN (1999) Effects of box features on spine
loading during warehouse order selecting, Ergonomics, 42:7, 980-996, DOI:
10.1080/001401399185252

To link to this article: http://dx.doi.org/10.1080/001401399185252

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all
the information (the “Content”) contained in the publications on our
platform. However, Taylor & Francis, our agents, and our licensors
make no representations or warranties whatsoever as to the accuracy,
completeness, or suitability for any purpose of the Content. Any
opinions and views expressed in this publication are the opinions and
views of the authors, and are not the views of or endorsed by Taylor
& Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information.
Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities
whatsoever or howsoever caused arising directly or indirectly in
connection with, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study
purposes. Any substantial or systematic reproduction, redistribution,
 reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access
and use can be found at http://www.tandfonline.com/page/terms-and-
conditions
Downloaded by [University of Kent] at 08:47 09 November 2014
 E RGONO M ICS , 1999, VOL . 42, NO. 7, 980 ± 996

EŒects of box features on spine loading during warehouse order

selecting

W . S. M ARRAS *, K . P. G RANATA , K. G . D AVIS , W . G. A LLREA D and M . J. J ORGEN SEN

Biodynamics Laboratory, The Institute for Ergonomics, Ohio State University,
1971 Neil Avenue, 210 Baker Systems Columbus, OH 43210, USA

Keywords: Spine loading; Warehousing tasks; Handles; Box features.

Low back disorders in distribution centres or warehouses have been identi® ed as

Downloaded by [University of Kent] at 08:47 09 November 2014

an area of elevated risk in many industries. The task of an order selector requires
workers manually to lift boxes from storage bins to a mobile pallet. This study
explored the eŒect of box features and box location when lifting from a pallet in a
storage bin upon spine loading. Ten experienced warehouse workers were asked
to lift boxes from a pallet while the size, weight, handle features and location of
the box on a pallet were changed. An EM G-assisted model was employed to
assess spine compression, lateral shear and anterior-posterior shear during the
lifts. The position from which the worker lifted a box on a pallet had the most
profound eŒect on spine loading while the lower level of the pallet represented the
greatest loadings on the spine. Box weight did not appear to be a feasible means
of controlling spine loading unless its position on the pallet could also be
controlled. The inclusion of handles had an eŒect similar to reducing the box
weight by 4.5 kg, whereas box size did not eŒectively aŒect spine loading. The
mechanisms by which these factors aŒect spine loading are discussed.

1. Introduction
Low back disorders (LBD) continue to represent the most common and most costly
musculoskeletal disorder experienced in the workplace (Hales and Bernard 1996).
M any of these LBD s are associated with occupational factors (Spengler et al. 1986)
and signi® cantly increase workers’ compensations costs. For example, LBDs account
for approximately 16 ± 19% of all workers’ compensation claims, but 33 ± 41% of the
total cost of all work com pensation costs (Spengler et al. 1986, W ebster and Snook
1994).
M anual material handling (M M H) tasks have been associated with the majority
of lower back injuries (Snook et al. 1978 , Bigos et al. 1986). These tasks often require
the worker to be exposed to several known risk factors including lifting, bending,
twisting m otions, lateral bending motions, maintenance of static postures, carrying
heavy loads, and combinations of these (Bigos et al. 1986, Snook et al. 1978 , Kelsey
et al. 1984 , Keyserling et al. 1991, M arras et al. 1993). One of the most comm on
M M H tasks, especially in warehouses or distribution centres, consists of palletizing
and depalletizing, which can be de® ned as the transferring and stacking of material
(boxes and bags) onto pallets. These tasks continue to employ manual labour
because it remains a cost-eŒective method of material transfer (D rury et al. 1989).

*Author for correspondence.

0014± 0139 /99 $12.00 Ó 1999 Taylor & F rancis L td

 Box features on spine loadin g 981

The N ational Association of W holesale Grocers of America (NAW GA ) and the

International Foodservice Distribution Association (IFDA) report that 30% of the
injuries reported by food distribution warehouse workers were attributable to back
sprains / strains (W aters 1993). Hence, grocery item selectors have an incidence of
LBD at least as high as other M M H jobs. Thousands of workers are employed in the
food distribution industry, with most being male (W aters 1993). The job of the
grocery item selector requires lifting and lowering during the transferring of various
containers (boxes and bags) from pallets located in storage bins throughout the
warehouse to a pallet which is generally positioned on the front of a motorized fork
truck along the storage bins.
The boxes handled in food distribution centres vary in size and weight. Drury et
al. (1982) evaluated box characteristics and found the median box size to be 28.0 cm
long, 20.5 cm wide and 21.5 cm high. The eŒects of box size on the maximum
acceptable weight of lift (M A W L) have been evaluated by several researchers (Garg
Downloaded by [University of Kent] at 08:47 09 November 2014

and Saxena 1980 , Snook and Ciriello 1991). These studies have shown that as the
depth of the box increases, M AW L decreases. In theory, the larger the depth (in the
sagittal plane) of the box, the larger the moment arm about lumbosacral joint (L 5 /
S 1 ).
Additionally, the weights of the boxes vary drastically depending upon the
material contained inside, with about 6% of the boxes > 23 kg (50 lb) (M arras et al.
1996). Numerous studies have indicated that box weight is an im portant factor in
LBD risk (Herrin et al. 1986 , M arras et al. 1993 , 1995). However, a more signi® cant
consideration might be the moment im posed about the spine during the lift as
opposed to box weight alone. An increase in moment arm or weight would result in
an increase in external moment, which has been directly linked to increases in m uscle
activity, an d, ultimately, increases in spinal loading (compression and shear forces)
(Cha n an d Baker 1970 , Cha n and Park 1973, Andersson et al. 1976 , NIOSH
1981 , Schultz and Andersson 1981 , Seroussi and Pope 1987 , M arras and M irka 1990 ,
M arras and Sommerich 1991 b, W aters et al. 1993).
W hile the majority of the boxes transferred have no handles, those boxes that do
have handles are usually cutouts in the two ends of the box (Drury et al. 1982). In
1991 , NIOSH developed a lifting equation that incorporated new factors for handle
coupling and task asymm etry. A wide range of research investigating the eŒects of

Figure 1. Schematic view of the six regions of the pallet (A, front-top; B, back-top; C, front-
middle; D, back-middle; E, front-buttom; F, back-bottom).
 982 W . S. M arras et al.

handles during lifting has been conducted using psychophysical methodology (Garg
and Saxena 1980 , Smith and Jiang 1984). However, this method has not been
validated by correlating M AW L values to actual incidence rates (Leamon 1994).
Few studies have investigated the eŒects of handles on loading of the spine. Freivalds
et al. (1984) employed a biomechanical model to estim ate the eŒects of spine loading
due to handle use. They found loading increased with handles. However, the model
only evaluated the eŒects of loading due to one muscle in the back and the
investigation was performed under conditions that do not match those seen in
industry.
Another factor that might signi® cantly aŒect spine loading and risk of injury is
position of the box on the pallet. The position of the box on the pallet might aŒect
the external moment arm distance and corresponding trunk moments. Boxes at the
back of the pallet might require larger m om ent arm s about L 5 / S 1 , resulting in larger
trunk moments. This could especially be the case when the worker cannot step on the
Downloaded by [University of Kent] at 08:47 09 November 2014

pallet. Although the eŒects of external moment on spinal loading have been well
documented, the trade-oŒs associated with changes in moment arm resulting from
variations in box size, weight, handle presence and position on a pallet during
warehouse order selecting have not been investigated. Thus, it is hypothesized that
box features m ay aŒect the moment arm, trunk moment, and ultimately, spine
loading.
The assessment of spinal loading has evolved from very sim ple two dimensional
static models to complete three-dimensional dynam ic models with numerous trunk
muscle inputs. W e have developed an electromyographic (EM G)-assisted model that
is well suited to evaluate dynam ic lifting under realistic work conditions (M arras and
Reilly 1988, Reilly and M arras 1989 , M arras and Som merich 1991a, b, Granata and
M arras 1993, 1995, M arras and Granata 1995, 1997 , Davis et al. 1997). Therefore,
the objective of the current study was to employ biomechanical load assessment
measures to evaluate the eŒect of transferring boxes (depalletizing) of diŒerent size,
weight, and handle conditions from diŒerent positions on a pallet.

2. M ethods
2.1. Subjects
Ten male subjects who worked as item selectors at a local warehouse volunteered to
depalletize / palletize boxes with various features. The subjects’ ages ranged from 19
to 49 years (average 27.2 years), with a work experience range of 0.25 ± 23 years in a
warehouse setting. The average (SD) height of the selectors was 180.3 (7.1) cm and
the average weight was 89.1 (8.4) kg.

2.2. Experimental design

The experimental design used a four-way, within-subject design. The independent
variables were: (1) box size; (2) box-handle coupling; (3) box weight; and (4) pallet
region. Subjects served as a random eŒect. Two sizes of boxes were evaluated in this
study ± a `sm all’ box with dimensions of 20.3 ´ 40.6 ´ 30.5 cm (H ´ W ´ D) and a
`large’ box with dimensions of 28 ´ 49.5 ´ 30.5 cm, which corresponded to volumes
3
of 4023 and 6810 cm , respectively. H andle conditions consisted of (1) boxes with
cut-out handles and (2) boxes without handles. The cut-out handles were 8.9 cm
wide and 2.5 cm high, positioned at the centre of the longer sides of the boxes,
5.1 cm below the top of the box. The position and size of the han dles were sim ilar to
those commonly found on boxes used in warehouse environments. The weights of
 Box features on spine loadin g 983

the boxes in this study were 18.2, 22.7 and 27.3 kg respectively. These weights were
at the upper percentiles of typical box weights found in a common warehouse setting.
Therefore, these weights were chosen to evaluate the eŒects of heavier loads on the
low back and the subsequent risk of LBD.
Each of the pallets were divided into six regions corresponding to front-top,
back-top, front-middle, back-m iddle, front-bottom, and back-bottom areas. Figure
1 shows a schem atic view of these six regions on a standard pallet. The handles of the
boxes in each of the regions remained at a set level corresponding approxim ately to
front-top and back-top regions at 133.8 cm from the ¯ oor; front-middle an d back-
middle regions at 95.3 cm from the ¯ oor; and front-bottom and back-bottom regions
at 47.6 cm from the ¯ oor. The number of boxes in each region depended on the size
of the box. A pallet of large boxes had four boxes in front-top region, three boxes in
back-top region, and seven boxes in regions front-m iddle, back-m iddle, front-
bottom , and back-bottom, while the pallets of small boxes had eight boxes in each of
Downloaded by [University of Kent] at 08:47 09 November 2014

the six regions.

The boxes were stacked on a standard pallet generally found in a warehouse with
a width of 101.5 cm and a depth of 112 cm. The small box conditions used a double-
stacked pallet to allow the handles of the boxes in various regions to correspond to
the heights of the large boxes which were stacked on a single pallet. The height of the
fully loaded pallet was 138 cm. The small boxes contained boxes of salt while the
large boxes contained plastic bottles of water. In order for the boxes to have the
desired weight, speci® c am ounts of material were removed from the inside containers
(e.g. water was drained from the centre bottle). The pallets of small boxes contained
six rows of ten boxes, while pallets of the large boxes comprised ® ve rows of seven
boxes.
The dependent variables associated with spinal loading were evaluated using an
EM G -assisted biomechanical model (M arras and Reilly 1988 , Reilly and M arras
1989 , M arras and Somm erich 1991a, b, Granata and M arras 1993 , 1995 , M irka and
M arras 1993, M arras and G ranata 1995 , 1997 , Davis et al. 1997). The spinal loads
estimated in this study were the maximum values of compression force, anterior-
posterior (A / P) shear and lateral shear forces on the lower back at the lumbosacral
joint. The trunk moments included the m aximum values of sagittal bending, lateral
bending, and axial twisting moments.

2.3. Task
To sim ulate a `realistic’ warehousing depalletizing task, subjects transferred
boxes from one pallet to another. The depalletizing task started when the
participant grasped the box from the pallet in the storage bin and ended when
he crossed an imaginary line that coincided with the point at which the
participant was upright and facing the `palletizing’ pallet (pallet in the aisle
where the boxes will be stacked). Data were collected for only this interval of
time, although subjects completed the task. An overhead view of the
arrangement is shown in ® gure 2.
The lifting rate for all subjects was set at 166 boxes handled per h and was
determined from the minimum loading rate required at the local warehouse where
the subjects were employed. The actual lifting cycle was one box lifted every 10 s
(360 per h) which was signalled by a computer tone; however, the actual lifting rate
was adjusted to 166 lifts / h by including any down time (e.g. moving pallets, ® lling
out body part discomfort surveys by the subjects, lunch, and additional rest breaks).
 984 W . S. M arras et al.
Downloaded by [University of Kent] at 08:47 09 November 2014

Figure 2. Schematic view of the experimental layout.

2.4. Appa ratus

The lumbar motion monitor (LM M ) was used to collect the trunk m otion variables.
It is essentially an exoskeleton of the spine in the form of a triaxial electro-
goniometer that measured instantaneous three-dimensional position, velocity and
acceleration of the trunk. The light-weight design of the LM M allowed the data to be
collected with minimal obstruction to the subject’ s movements. For more
information on the design, accuracy, and application of the LM M , see M arras et
al. (1992).
Electromyographic (EM G) activity was monitored through the use of bipolar
electrodes spaced approximately 3 cm apart at the ten major trunk muscle sites. The
muscles were: right and left erector spinae; right and left latissimus dorsi; right and
left internal obliques; right and left external obliques; and right and left rectus
abdominus. For the standard locations of electrode placem ent for these muscles, see
M irka and M arras (1993).
A force plate (Bertec 4060 A; W orthington, OH, USA) and a set of electro-
goniometers measured the external loads an d moments placed on L 5 / S 1 during the
various calibration exertions. The purpose of the calibration exertions was to
determ ine the individual gain to be used in the `open-loop’ exertions. `Open-loop’
 Box features on spine loadin g 985

referred to exertions that use a predetermined gain (muscle gain represents the
maximum force per unit area of muscle) to calculate internal moments and forces,
rather than calculating a speci® c gain for each exertion. The electrogoniometers
measured the relative position of L 5 / S 1 with respect to the centre of the force plate,
along with the subject’ s pelvic angle. The forces were translated and moments were
rotated from the centre of the force plate to L 5 / S 1 through the use of the
electrogoniometers (Fathallah et al. 1997).
All signals from the aforementioned equipment were collected simultaneously
TM
through customized W indows -based software developed in the Biodynam ics
Laboratory. The signals were collected at 100 Hz and recorded on a 486 portable
computer via a 16-bit analogue-to-digital board.

2.5. Procedure
Upon arriving at the Biodynam ics Laboratory, subjects were given a brief
Downloaded by [University of Kent] at 08:47 09 November 2014

description of the study and what they would be required to do. Subjects read and
signed a consent form. Next, anthropom etric measurements were taken. The surface
electrodes then were applied using proper placem ent procedures to sample the
muscles of interest. Skin impedances were kept < 1 M l . The participant was placed
into a stable structure that allowed maximum exertions to be performed in six
directions, while a constant resistance was held against the participant. These
maxima were performed to allow all subsequent EM G data to be normalized. The
six exertions consisted of: sagittal extension with the trunk at a 20 8 forward ¯ exion
angle; sagittal ¯ exion at 0 8 ¯ exion; right lateral ¯ exion at 0 8 ¯ exion; left lateral
¯ exion at 0 8 ¯ exion; right twist at 0 8 ¯ exion; and left twist at 0 8 ¯ exion. After each
maximum exertion, 2 min of rest were given, in accordance with past research
(Caldw ell et al. 1974).
Before handling each pallet of boxes, the participant completed a set of
calibration lifts to determ ine muscle gain. During the calibration exertions, the
participant lifted a 22.7 kg box from a sagittally symmetric position at a slow,
smooth pace (controlled by the participant). The lift started at the subject’ s knee
height and ended in his upright position. The calibration lifts were run under `closed-
loop’ conditions; that is, internal moments were validated with m easured external
moments. The model performed well with physiologically reasonable gains
2 2
(average = 28.4 N / cm ), high R (average = 0.87) and low average absolute error
(average = 21.9 Nm) for the calibration exertions. Before and after each set of
calibrations, data were collected to determine the position of the LM M and the
relative position of L 5 /S 1 to the centre of the force plate m easured by the electro-
goniom eters for each participant standing erect.
Subjects then were instructed as to how to transfer boxes from one pallet to
another, that is, the order of the boxes to be lifted and not the style of lifting.
In general, subjects unloaded boxes from left to right and from front to back.
The participant was instructed not to begin depalletizing a new row until the
current row was completely ® nished. A computer-generated tone sounded to
indicate the participant was to begin lifting the next box. After one box was
transferred, the participant returned to a designated spot marked on the ¯ oor to
await the next tone. This procedure allowed the experimenters to control the
lifting rate of each participant. The subjects were only allowed to step on the
pallet for the bottom layer and were permitted to slide the boxes forward for
the upper layers.
 986 W . S. M arras et al.

The pallets’ positions corresponded to the positions commonly found in a

grocery warehouse setting. Each participant transferred all boxes on a pallet for all
the combinations of independent variables (e.g. 12 pallets total). The order in which
these 12 pallets were handled was randomized for each participant, and each
participant completed all 12 pallets on the same day.

2.6. Data analyses

The LM M voltages were converted into angles, velocities and accelerations
through a customized conversion software. The kinematic data were used in the
EM G-assisted spinal load model. The raw EM G signals were preampli® ed, high-
passed ® ltered at 30 Hz, low-passed ® ltered at 1000 Hz, recti® ed, and integrated
via a 20 ms sliding window hardware ® lter. The EM G data were normalized with
respect to the maximum output of the muscles (obtained at the beginning of the
testing) and muscle length ± strength and force ± velocity modulations by a
Downloaded by [University of Kent] at 08:47 09 November 2014

custom ized software program (Granata and M arras 19 93, Davis et al. 1998a).
The EM G, kinem atic, and `closed loop’ muscle gain data were imported into the
EM G-assisted model to calculate spinal forces and moments on the lumbosacral
joint.
D escriptive statistics were computed for all of the dependent variables. Analysis
of variance (ANOV A) statistical analyses then were performed on all the dependent
variables. For all signi® cant independent variables, post-hoc analyses, in the form of
Tukey multiple pairwise comparisons, were performed to determine the source of the
signi® cant eŒect(s).

3. Results
The ANOVA analysis indicated that com pression as well as both shear forces (lateral
and A / P) were signi® cantly in¯ uenced by box weight and the position of the box on
the pallet. The presence of handles was found to aŒect the level of A-P shear and
com pression force while box size only had an eŒect on the A-P shear force. In
addition, interactions of box size and handle with region resulted in statistically
signi® cant diŒerences for all three spine loading measures. A / P shear and
com pression both responded in a statistically diŒerent manner to the various handle
conditions and the weight by region interaction. Finally, A / P shear responded
diŒerently to the size conditions. Table 1 summ arizes the statistically signi® cant
diŒerences that were observed.
As expected, as the box weight increased, spine loading in all three directions
increased. Figure 3 shows that each loading measure increased by about 15% for
each 4.5 kg increase in box weight (slightly more for A / P shear). However, this ® gure
also indicates a large am ount of variation associated with each loading parameter. In
other words, because of the large distributions of forces, an 18.2 kg box often
produces loads on the spine that are very sim ilar to a 27.3 kg box. It appears that the
spine loading distributions have a great deal of overlap (as evidenced by the standard
deviation bars in ® gure 3) between the box weight conditions.
The position of the box on the pallet (its region) can explain much of the
overlap in spine loadings observed as a function of box weight. Figure 4 shows
how spine compression varies as a function of the box location. This ® gure
indicates that there is a signi® cant increase in spine compression in front-bottom
and back-bottom regions of the pallet. Speci® cally, the spine loading distributions
are signi® cantly shifted upward at these lower levels of the pallet. Similar trends
 Box features on spine loadin g 987

were observed for A / P shear and lateral shear, however, the magnitudes of the
loads were smaller with lateral shear displaying the lowest magnitude of loading
in all pallet positions.
The interaction of the box weight and region indicated that the weight of the box
was more of a problem at the lower regions of the pallet than at the upper or m iddle
layers of the pallet. Figure 4 shows that the greatest compression force occurred at
the lowest level of the pallet and this is where the diŒerences in weight resulted in the
largest diŒerences in spine loading. This ® gure indicates that even the lightest boxes

Table 1. Analysis of variance results for spinal loading.

Maximum
Maximum lateral anterior-posterior Maximum
shear force shear force compression force
Downloaded by [University of Kent] at 08:47 09 November 2014

Size (S) 0.1144 0.0317* 0.3116

Handle (H) 0.1032 0.0094* 0.0001*
Weight (W) 0.0001* 0.0001* 0.0001*
Region (R) 0.0009* 0.0001* 0.0001*
S´ H 0.3969 0.3989 0.3592
S´ W 0.2148 0.5429 0.5779
S´ R 0.0242* 0.0084* 0.0001*
H´ W 0.6630 0.8528 0.8025
H´ R 0.0120* 0.0001* 0.0001*
W´ R 0.4048 0.0001* 0.0001*
S´ H´ W 0.3797 0.5480 0.4552

*Signi® cant at p < 0.05.

Figure 3. Three-dimensional spine loading as a function of case weight.

 988 W . S. M arras et al.

averaged peak compressive loads that were > 3400 N in front-bottom and back-
bottom regions of the pallet. Similar trends were observed for the spinal A / P shear
forces as shown in ® gure 5.
Downloaded by [University of Kent] at 08:47 09 November 2014

Figure 4. Maximum compression force as a function of box weight and box location of
the pallet.

Figure 5. Maximum anterior-posterior shear force as a function of box weight and box
location on the pallet.
 Box features on spine loadin g 989

Handles decreased the compression and A / P shear forces on the spine by about
10 ± 15% . However, the A / P shear load relief was more prominent at the lower
regions of the pallet as shown in ® gure 6. The trend shown in this ® gure was also
representative of the compression and lateral shear loading pro® les.
Finally, handling the smaller box reduced the average maximum A / P shear force
by 50 N compared with the large box. The size by region interaction had a signi® cant
eŒect on all three spine loading measures. However, the trends am ong these
measures was not consistent. The smaller box resulted in slightly lower A / P shear
forces in the front-top, back-top, and front-bottom regions (® gure 7), whereas, the
larger box resulted in lowered lateral shear force in these same regions as well as in
the back-bottom region. Although the eŒect of box size on compression was
statistically signi® cant the practical eŒect was negligible.

4. Discussion
Downloaded by [University of Kent] at 08:47 09 November 2014

To get an accurate interpretation of the results, they must be viewed as a function of

the tolerance limits of the spine. The literature is rich with references to spine
compression tolerance. This is an area of much debate and the range of compression
tolerances is large (Cha n and Andersson 1991). M any analyses employ the limits
adopted by the NIOSH W ork Practices Guide for M anual Lifting (1981). This
document suggests that 3400 N is the point at which vertebral end plate
microfractures begin to occur in those < 40 years of age. It is believed that the
repair process of these microfractures facilitates the degeneration of the disc.
Furtherm ore, as spine compression increases to 6400 N about half of workers will
experience vertebral endplate microfractures. Shear tolerances are less prominent in
the literature. However, M cGill (1996) has recently suggested 1000 N as a tolerance

Figure 6. Maximum anterior-posterior shear force as a function of box handle coupling and
box location on the pallet.
 990 W . S. M arras et al.
Downloaded by [University of Kent] at 08:47 09 November 2014

Figure 7. Maximum anterior-posterior shear force as a function of box size and box location
on the pallet.

level for disc dam age due to shear loading. Thus, we will assum e these `benchmarks’
as values to consider when interpreting the signi® cance of these results.
This analysis has shown quantitatively that the greatest in¯ uence on spine
loading is the location from which the box is lifted from a pallet. Spine loading
measures have shown that the lower regions of the pallet (front-bottom and back-
bottom regions) contribute the most to spine loading. These regions also represent
the conditions where most of the spine com pression distribution and a large portion
of the A / P shear forces exceed the spine tolerance limits (N IOSH 1981 , M cGill
1996). The compressive forces exceeded the tolerance lim its of the spine m ore often
(a greater portion of the distribution) than did the A / P shear forces. In fact, even the
relatively light 18.2 kg box, without handles, resulted in > 70% of the lifts > 3400 N
spine compression tolerance limit. Therefore, a large bene® t can be derived by
elevating the pallet so that boxes can be transferred oΠan elevated or adjustable
platform such as a lift table. However, there are many situations where raising the
load in a bin might be impractical. For example, a distribution centre with thousands
of order-picking bins might ® nd this solution to be cost-prohibitive. The in¯ uence of
the position of the box relative to the spine has been described by many other studies
(Andersson et al. 1976, Cha n and Page 1984 , Serousi and Pope 1987, Davis et al.
1998a) . Similar to this study, these researchers found that the greater magnitudes of
trunk ¯ exion (due to lower box positions) resulted in higher muscle activity and
spinal loads.
This study also has shown that the manipulation of box features also m ay aŒect
the loading associated with a lift. Box size, weight and handle conditions all
displayed signi® cant in¯ uences on spine loading, especially when these factors were
considered as a function of box region. It is not surprising that an increase in spinal
loads was found as the box weight increased (Cha n and Baker 1970, Cha n and
 Box features on spine loadin g 991

Park 1973, M arras and Sommerich 1991 b). Of these features, size, although often
statistically signi® cant, appeared to have little in¯ uence in moving the spine loading
from above a tolerance benchmark to below the benchmark for either compression
or shear. A /P shear increased slightly for the large boxes in about half of the pallet
regions. This probably was due to an increase in the moment associated with a
slightly large box since the box centre of gravity would be slightly farther from the
body compared with a smaller box. In the sam e regions lateral shear decreased
slightly with the large box. This probably was due to reduced lateral bending
associated with large boxes, since subjects were not required to reach as far with a
larger box as com pared with a smaller box. For the sizes of boxes evaluated in this
study, the eŒects of A / P and lateral shear were oŒ-setting. Thus, manipulating box
size appeared to have little practical value.
Box handles, across all regions, tended to reduce the loading on the spine for a
given box weight. The 27.3 kg boxes with handles resulted in a mean maximum
Downloaded by [University of Kent] at 08:47 09 November 2014

compression force of 3275 N. This was slightly less than the 22.7 kg boxes without
handles (3399 N ). W hen handles were added to the 22.7 kg boxes, the mean
maximum compression force reduced to 2864 N, which was comparable with the
mean maxim um compression force from 18.2 kg boxes without handles (2925 N).
Therefore, the eŒect of including handle s on boxes was appro ximately equivalent to
reducing the box weight by 4.5 kg.
The reduction of the spinal loads through the use of handles was contrary to
Freivalds et al. (1984), who found that the peak com pression forces were higher for
the boxes with handles. Their assessment neglected multiple muscles by using a single
equivalent muscle, rigid link model, and, thus, failed to account for the higher
coactivity that occurred for the boxes without handles. On the other hand, similar
results of lower forces when lifting boxes with handles was found by Kromodihardjo
and M ital (1987) and Davis et al. (1998b).
Box weight and handle condition, together, did appear to have a great in¯ uence
on spine loading. To appreciate how spine compression loading changed as a
function of these features as well as the region factor, table 2 has been included to
indicate the percentage of observations that exceed the tw o NIOSH compression
force `benchmarks’ discussed earlier. It indicates how the inclusion of handles and
boxes of diŒerent weights interact with pallet region to de® ne loading, and therefore
risk of LBD . Similar analyses for the shear forces indicated that, for the most part,
spine compression was more problematic than the shear forces for the weight factor
alone. Table 2 indicates that the practical bene® t of handles is realized for the heavier
boxes located at the lower levels of the pallet. Speci® cally, regardless of box weight,
the using of handle s can reduce the percentage of observations that exceed the 6400 N
limit by appro ximately 50% . The eŒect of handles also interacts with the region
factor. Compression forces often exceed the tolerance limit for all box weights at the
lowest layer of the pallet. The decrease in compression force by the use of handles
most likely resulted from the decrease in the external m oments. This would be
expected since the handles were located at the top of the boxes and subjects were not
required to bend the torso to the sam e extent as when lifting was done without using
handles. This would reduce the external moment generated by the torso mass.
Further evaluation of the model results con® rmed that signi® cant diŒerences in
maxim um L 5 / S 1 compression force were associated with diŒerences in three of the
four maximum moment measurem ents (e.g. sagittal, twisting, and resultant
moment). The external moment associated with boxes containing handles was
 Downloaded by [University of Kent] at 08:47 09 November 2014
992

Table 2. Summary of the percentage of data within the benchmark zones for spine compression.

Box weight

18.2 kg 22.7 kg 27.3 kg

Region Benchmarks
on the pallet (n) Handles No handles Handles No handles Handles No handles

< 3400 100.0 100.0 100.0 99.2 99.2 100.0

Front-top 3400 to 6400 0.0 0.0 0.0 0.8 0.8 0.0
> 6400 0.0 0.0 0.0 0.0 0.0 0.0
< 3400 98.2 89.1 84.5 76.4 83.6 67.3
Back top 3400 to 6400 1.8 10.9 15.5 23.6 16.4 32.7
> 6400 0.0 0.0 0.0 0.0 0.0 0.0
< 3400 98.7 91.3 94.7 82.7 92.6 76.0
Front-middle 3400 to 6400 1.3 8.7 5.3 17.3 7.4 23.3
W . S. M arras et al.

> 6400 0.0 0.0 0.0 0.0 0.0 0.7

< 3400 88.7 82.0 80.7 75.3 76.7 64.7
Back-middle 3400 to 6400 11.3 18.0 19.3 24.7 23.3 34.6
> 6400 0.0 0.0 0.0 0.0 0.0 0.7
< 3400 45.3 30.0 29.3 14.0 16.0 3.3
Front-bottom 3400 to 6400 52.0 62.0 62.7 65.3 72.0 66.0
> 6400 2.7 8.0 8.0 20.7 12.0 30.7
< 3400 35.3 24.0 30.0 10.7 9.3 2.0
Back bottom 3400 to 6400 60.7 67.3 56.7 65.3 71.3 62.0
> 6400 4.0 8.7 13.3 24.0 19.3 36.0
 Box features on spine loadin g 993

signi® cantly less than the moment for boxes without handles. The most pronounced
reduction of maximum moment was for the sagittal moment, which reduced to
151.8 Nm from 176.4 Nm (a 13.9% decrease) by the inclusion of handles on the
boxes. Examination of the results also shows that the eŒect of including handles is to
raise the lift point of the box. This aŒects the length ± strength relationship of the
trunk muscles, moving the load position to a location where this relationship
becomes more optim al. Thus, including handles is an alogous to slightly changing the
position of the box on the pallet.
A large portion of the A / P shear distribution also exceeded the 1000 N tolerance
at the bottom regions when no handles were present on the box. Speci® cally, A / P
shear was found to be problematic for the 27.3 kg boxes in the front-bottom and
back-bottom regions. Handles were also found to reduce this average peak shear to
well below the 1000 N tolerance (® gure 6). Observations of the subjects indicated
that this reduction in shear associated with handles was due to the fact that subjects
Downloaded by [University of Kent] at 08:47 09 November 2014

could grab the han dle and slide the load toward them as opposed to reaching to the
far end of the box to lift, as was observed when lifting was done without using
handles.
These results based upon spine load ing also were compared to an analysis of the
same features evaluated with a LBD risk model (M arras et al. 1993). This analysis
resulted in very similar results (Allread et al. 1996). Thus, the spinal loading pro® les
were for the most part similar to trunk motion and workplace factor measures that
match trunk motions and workplace factors historically associated with jobs placing
workers at an increased risk of LBD .
Several potential lim itations of this study should also be acknowledged. First,
the work performed by the warehouse order selectors was performed in a
laboratory environment. Every eŒort was m ad e to simulate a distribution centre;
however, whenever data are collected via EM G, some realism is lost. Second, the
work was paced to simulate the speed observed in a distribution environm ent.
Even though the average pace matched that of a warehouse order selector’ s task,
in reality, a warehouse order selector works more in spurts (lifting several boxes
within a few minutes and then not lifting again until a new bin location is located).
Thus, these results may not re¯ ect the full impact of the dynam ic nature of the job.
Third, the model-predicted spine loads were indirectly validated through the
calibration results. H owever, the calibration results indicated that the model was
indeed robust (see the M ethods for model performance variables). Furthermore,
the model has been validated in three-dim ensions through several previous studies
(M arras and Reilly 1988 , Reilly and M arras 1989, M arras an d Sommerich 1991a,b,
Granata and M arras 1993, 1995 , M irka and M arras 1993 , M arras and Granata
1995 , 1997 , Davis et al. 1998a) . Finally, these results m ay not be generalizable to a
female population since it would be expected that the loading patterns would be
diŒerent. Future research should evaluate these population groups to enable a
wider application of the results.

5. Conclusions
(1) The presence of handles on the boxes had a profound eŒect of reducing mean
maxim um L 5 / S 1 compression forces and A / P shear forces. The reduction of
compression forces from the use of handles m ost likely resulted from the
decrease in external moments, which indicated that the load was able to be
held closer to the body. A relationship emerged between handles and box
 994 W . S. M arras et al.

weight, in that the com pression values for boxes lifted without handles were
approximately equivalent to lifting boxes with handles weighing 4.5 kg more.
(2) Box weight signi® cantly in¯ uenced spinal loading and, therefore, risk of
LBD . The mean maxim um L 5 / S 1 compression forces for the low er weight
boxes without handles positioned at the middle and top layers of the
pallet, were at or below the NIOSH acceptable 3400 N. However, these
were m ean maximum values, and the standard deviations around these
means suggests that there were observations (e.g. some subjects’ responses)
> 3400 N.
(3) The region of the pallet from which boxes were lifted also had a profound
impact on compression. The results implied that, as boxes were located lower
on a pallet, risk increased. Boxes that were lifted from the bottom layer of the
pallet, regardless of the weight or the handle condition, produced mean
maxim um L 5 / S 1 compression forces above the NIO SH acceptable compres-
Downloaded by [University of Kent] at 08:47 09 November 2014

sion force 3400 N. However, lifting the 18.2 kg box with han dles did
approach a low er risk compression force level (3855.4 N).
(4) The quantitative understanding of the conditions under which spine loadings
become excessive, as described in this study, can serve as a basis for job and
box feature redesign in distribution centres.

Acknowledgement
This work was supported, in part, from a grant from the Food M arketing Institute,
W ashington DC. W e thank M r Bryan Kirking for technical assistance.

References
A LLR EAD , W. G., M A RRAS , W. S., G RA NATA , K. P., D AVIS , K. G. and J ORG EN SEN , M. J., 1996,
The eŒects of box diŒerences and employee work experience on trunk kinematics and
low back injury risk during depalletizing operations, Proceedings of the Human Factors
and Ergonomics Society, (Santa Monica: Human Factors and Ergonomics Society),
651 ± 655.
A N DERSSO N , G. B., O RTEN GREN J. R. and N ACH EM SON , A. 1976, Quantitative studies of back
loads in lifting, Spine, 1, 178 ± 185.
B IGOS , S. J., B A TTIE , M. C., S PEN GLER , D. M., F ISHER , L. D., F OR DYCE , W. E., H ANSSON , T.
H., N ACH EM SON , A. L. and W ORTLEY , M. D. 1991, A prospective study of work
perceptions and psychosocial factors aŒecting the report of back injury, Spine, 16,
1 ± 6.
B IGOS , S. J., S PENG LER , D. M., M A RTIN , N. A., Z EH , J., F ISH ER , L., N A CHEMSO N , A., and W ANG ,
M. H., 1986, Back injuries in industry: a retrospective study II injury factors, Spine, 11,
246 ± 251.
C ALDW ELL , L. S., C H AFFIN , D. B., D UK ES -D OBOS , F. N., K RO EM ER , K. H. E., L AUBACH , L. L.,
S NOOK , S. H. and W ASSERM AN , D. E. 1974, A proposed standard procedure for static
muscle strength testing, American Industrial Hygiene Association Journal, 35, 201 ± 206.
C H AFFIN , D. B. and A N DERSSON , G. B. J. 1991, Occupational Biomechanics, 2nd ed ± (New
York: John Wiley).
C H AFFIN , D. B. and B A KER , W. H. 1970, A biomechanical model for analysis of symmetric
sagittal plane lifting, American Institute of Industrial Engineers Transactions, 2, 16 ± 27.
C H AFFIN , D. B. and P AG E , G. B., 1994, Postural eŒects on biomechanical and psychophysical
weight-lifting limits, Ergonomics, 37, 663 ± 676.
C H AFFIN , D. B. and P ARK , K. S. 1973, A longitudinal study of low-back pain as associated with
occupational weight lifting factors, American Industrial Hygiene Association Journal, 34,
513 ± 525.
D AVIS, K. G., K IR KING , B., G AU DES , L., J AY , W. and M ARRA S, W. S., 1997, The in¯ uence of load
moment and visual perception on psychophysically determined acceptable lifting
weight, Human Factors, 39, 312 ± 322.
 Box features on spine loadin g 995

D A VIS , K. G., M ARR AS, W. S. and W ATERS , T.R. 1998a, Evaluation of the nature and spinal
loading during lowering and lifting, Clinical Biomechanics, 13, 141 ± 152.
D A VIS , K. G., M A RRAS , W. S. and W ATERS , T. R. 1998b, Reduction of spinal loading through
the use of handles, Ergonomics 41, 1155 ± 1168.
D R URY , C. G., D EEB , J. M., H ARTM AN , B., W OOLEY , S., D RU RY , C. E. and G ALLAG HER , S. 1989,
symmetric and asymmetric manual materials handling part 1: physiology and
psychophysics, Ergonomics, 32, 467 ± 489.
D R URY , C. G., L AW , C. and P AWEN SK I, C. S. 1982, A survey of industrial box handling, Human
Factors, 24, 553 ± 565.
F ATH ALLAH , F. A, M A RRAS , W. S., P ARN IAN POU R , M. and G RAN ATA , K.P. 1997, A method for
measuring external spinal loads during unconstrained free-dynamic lifting, Journal of
Biomechanics, 30, 975 ± 978.
F REIVAL DS , A., C H AFFIN , D. B., G ARG , A. and L EE , K. S. 1984, A dynamic biomechanical
evaluation of lifting maximum acceptable loads, Journal of Biomechanics, 18, 571 ± 584.
G A RG , A. and S AXENA , U. 1980, Container characteristics and maximum acceptable weight lift,
Human Factors, 22, 487 ± 495.
Downloaded by [University of Kent] at 08:47 09 November 2014

G R ANA TA , K. P. and M A RRAS , W. S. 1993, An EM G-assisted model of loads on the lumbar

spine during asymmetric trunk extensions, Journal of Biomechanics, 26, 1429 ± 1438.
G R ANA TA , G. P. and M ARRAS , W. S. 1995, An EMG-assisted model of trunk loading during
free-dynamic lifting, Journal of Biomechanics, 28, 1309 ± 1317.
H A LES , T. R. and B ERNA RD , B.P. 1996, Epidemiology of work-related musculoskeletal
disorders, Journal of Orthopedic Clinics of North of America, 27, 679 ± 709.
H ERRIN , G. A., J ARAIED I, M. and A ND ERSON , C. K. 1986, Prediction of overexertion injuries
using biomechanical and psychophysical models, American Industrial Hygiene Associa-
tion Journal, 47, 322 ± 330.
K ELSEY , J. L., G ITH EN S, P. B., W HITE , A. A., H OLFOR D , T. R., W ALTER , S. D., O’ C ONN ER , T.,
O STFELD , A. M., W EIL , U., S OUTHW ICK , W. O. and C ALOGER O , J. A.. 1984, An
epidemiological study of lifting and twisting on the job and risk for acute prolapsed
lumbar intervertebral disc, Journal of Orthopedic Research, 2, 61 ± 66.
K EYSERLING , W. M., A RM STR ONG , T. J. and P U NN ETT , L. 1991, Ergonomics job analysis: a
structured approach for identifying risk factors associated with overexertion injuries
and disorders, Applied Occupational Environmental Hygiene, 6, 353 ± 363.
K R OM O DIHAR DJO , S. and M ITAL , A., 1987, Biomechanical analysis of manual lifting tasks,
Journal of Biomedical Engineering, 109, 132 ± 137.
L EAM ON , T. B. 1994, Research to reality: a critical review of the validity of various criteria for
the prevention of occupationally induced low back pain disability, Ergonomics, 37,
1959 ± 1974.
M ARRAS , W. S. and G RAN ATA , K. P. 1995, A biomechanical assessment and model of axial
twisting in the thoraco-lumbar spine, Spine, 20, 1440 ± 1451.
M ARRAS , W. S. and G RAN ATA , K. P. 1997, Spine loading during trunk lateral bending motion,
Journal of Biomechanics, 30, 697 ± 703.
M ARRAS , W. S. and M IRKA , G. A. 1990, Muscle activity during asymmetric trunk angular
accelerations, Journal of Orthopedic Research, 8, 824 ± 832.
M ARRAS , W. S. and R EILLY , C. H. 1988, Networks of internal trunk loading activities under
controlled trunk motion conditions, Spine, 13, 661 ± 667.
M ARRAS , W. S. and S OM M ERICH , C. M. 1991a, A three dimensional motion model of loads on
the lumbar spine: I. Model structure, Human Factors, 33, 129 ± 137.
M ARRAS , W. S. and S OMM ERICH , C. M. 1991b, A three dimensional motion model of loads on
the lumbar spine: II. Model validation, Human Factors, 33 139 ± 149.
M ARRAS , W. S., F ATHALLA H , F. A., M ILLER , R. J., D AVIS , S. W. and M IRK A , G. A. 1992,
Accuracy of a three-dimensional lumbar motion monitor for recording dynamic
trunk motion characteristics, International Journal of Industrial Ergonomics, 9, 75 ±
87.
M ARRAS , W. S., G RA NATA , K. P., D AVIS, K. G., A LLR EAD , W. G. and J OR GENSEN , M. J. 1996, The
eŒects of box weight, box size, and handle coupling on spinal loading and risk of low
back disorders in order selecting. (Washington: Food Marketing Institute), Final
Project, March 1996.
 996 W . S. M arras et al.

M AR RAS , W. S., L AVEN DER , S. A., L EURG ANS , S. E., F ATHALLA H , F. A., F ERGU SON , S. A.,
A LLREAD , W. G. and R AJULU , S. L. 1995, Biomechanical risk factors for occupationally
related low back disorders, Ergonomics, 38, 377 ± 410.
M AR RAS , W. S., L AVEND ER , S. A., L EUR GAN S , S. E., R AJULU , S. L., A LLREA D , W. G., F A TH ALLAH ,
F.A. and F ERGU SO N , S. A. 1993, The role of dynamic three-dimensional motion in
occupationally-related low back disorders, Spine, 18, 617 ± 628.
M C G ILL , S. M. 1996, Searching for the safe biomechanical envelope for maintaining healthy
tissue. In Pre-ISSLS Workshop: The Contribution of Biomechanics to the Prevention and
Treatment of Low Back Pain, (University of Vermont, 25 June).
M IRKA , G. A. and M ARR AS, W. S. 1993, A stochastic model of trunk muscle coactivation
during trunk bending, Spine,18, 1396 ± 1409.
NIOSH 1981, Work Practices Guide for Manual Lifting. NIOSH Technical Report No. 81 ±
122 (Cincinnati: US Department of Health and Human Services, National Institute for
Occupational Safety and Health).
R EILLY , C. H. and M ARRA S, W. S. 1989, SIMULIFT: a simulation model of human trunk
motion during lifting, Spine, 14, 5 ± 11.
Downloaded by [University of Kent] at 08:47 09 November 2014

S EROUSSI, R. E. and P OPE , M. H. 1987, The relationship between trunk muscle

electromyography and lifting moments in the sagittal and frontal planes, Journal of
Biomechanics, 20, 135 ± 146.
S CHU LTZ , A. B. and A N DERSSON , G. B. J. 1981, Analysis of loads on the lumbar spine, Spine, 6,
76 ± 82.
S M ITH , J. L. and J IANG , B. C. 1984, A manual materials handling study of bag lifting, American
Industrial Hygiene Association Journal, 45, 505 ± 508.
S NO OK , S. H. and C IRIELLO , V. M. 1991, The design of manual handling tasks: revised tables of
maximum acceptable weights and forces, Ergonomics, 34, 1197 ± 1213.
S NO OK , S. H., C AM PANELL I , V. and H A RT , J. W. 1978, A study of three preventive approaches to
low back injury, Journal of Orthopeadic Medicine, 20, 478 ± 481.
S PENGLE R , D. M., B IGOS , S. J., M AR TIN , N. A., Z EH , J., F ISHER , L. and N A CHEMSO N , A. 1986,
back injuries in industry: a retrospective study. I. overview and cost analysis, Spine, 11,
241 ± 245.
W ATERS , T. R. 1993, Workplace factors and trunk motion in grocery selector tasks. In
Proceedings of the Human Factors and Ergonomics Society 37th Annual Meeting, (Santa
Monica: Human Factors and Ergonomics Society), 654 ± 658.
W ATERS , T. R., P UTZ -A N DERSON , V., G AR G , A. and F INE , L. J., 1993, Revised NIOSH equation
for the design and evaluation of manual lifting tasks, Ergonomics, 36, 749 ± 776.
W EBSTER , B. S. and S N OOK , S. H. 1994, The cost of 1989 workers’ compensation low back pain
claims, Spine, 19, 1111 ± 1116.