Research Paper: PHdPostharvest Technology

The simulation of the impact damage to fruit during the

passage of a truck over a speed bump by means of
the discrete element method
Michael Van Zeebroeck
a,
*, Geert Lombaert
b,1
, Edward Dintwa
a
, Herman Ramon
a
,
Geert Degrande
b
, Engelbert Tijskens
a
a
Department of Biosystems, Katholieke Universiteit Leuven, Kasteelpark Arenberg 30, B-3001 Leuven, Belgium
b
Department of Civil Engineering, Katholieke Universiteit Leuven, Kasteelpark Arenberg 40, B-3001 Leuven, Belgium
a r t i c l e i n f o
Article history:
Received 11 December 2007
Received in revised form 10 March
2008
Accepted 2 June 2008
The discrete element method (DEM) was used to study fruit damage during transportation.
The DEM is a particle-based simulation technique which is well suited for the solution of
granular material related problems in food and agriculture. In this paper, the application
of DEM to food transport problems was demonstrated by simulation of bruising to apples
stored in bulk bins during the passage of a truck over a speed bump. The effects of truck
load, bulk bin position, suspension type and driving speed on damage were investigated.
The simulations showed that higher truck loads lead to less bruising and that apples in
bulk bins behind the rear axle suffered more damage than those in bulk bins in front of
the rear axle. Furthermore, a considerable reduction in the damage was predicted in sim-
ulations where the truck has a soft suspension. Independent of truck load, suspension type
and bulk bin position, the commercially signicant bruising (i.e. apples with bruise volume
of maximum bruise above 500 mm
3
) was predicted to be insignicant for driving speeds be-
low 20 kmh
1
. At higher driving speeds, the extent of commercially acceptable bruising de-
pended on various parameters. A reduction in the driving speed, an increase in the truck
load and a reduction in the suspension stiffness all helped to reduce the occurrence of fruit
damage.
2008 IAgrE. Published by Elsevier Ltd. All rights reserved.
1. Introduction
For most fruits bruising is the most common type of posthar-
vest mechanical injury. Fruit bruising does not only affect the
quality appreciated by the consumer, but bruises open the
pathways for pathogenic attack, even when they are small
in size. The present study deals with bruising caused by tran-
sient vibrations or shocks. This form of damage is more fre-
quently encountered than damage due to static compression
(Mohsenin, 1973).
In the literature, only a limited number of studies deal with
the effect of transient vibrations on fruit impact damage. Holt
* Corresponding author. Flemish Ministry of Agriculture and Fishery, Ellipsgebouw Koning Albert II, Laan 35, Bus 40, B-1030 Brussel,
Belgium.
E-mail address: michael.vanzeebroeck@gmail.com (M. Van Zeebroeck).
1
Postdoctoral Fellow of the Research Foundation Flanders (FWO).
Avai l abl e at www. sci encedi r ect . com
j our nal homepage: www. el sevi er . com/ l ocat e/ i ssn/ 15375110
1537-5110/$ see front matter 2008 IAgrE. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.biosystemseng.2008.06.003
b i o s ys t e ms e ng i ne e r i ng 1 0 1 ( 2 0 0 8 ) 5 8 6 8
andSchoorl (1985) andJones et al. (1991) developeda numerical
model for the simulation of impact damage of apples due to
transient shocks caused by road discontinuities such as
bumps and potholes. Their model accounted for the interac-
tion between the road, the vehicle and the load. The apples
were assumed to be stacked in vertical columns, so that there
are only two potential bruise sites on each apple. The follow-
ing three sub-problems were considered: (1) modelling and
characterisation of the road surface roughness, (2) prediction
of the vertical vehicle response for a given road unevenness,
(3) prediction of the contact interaction of the fruit (energy ab-
sorption in their study, but possibly also contact forces) and
correlation of fruit interactions with bruising. The equations
of motion of this system were solved numerically by means
of the RungeKutta or the AdamsBashforth method. In the
study by Jones et al. (1991), apple bruising was predicted as
a function of the truck load, the bulk bin position, the bump
height and the truck driving speed.
A more accurate representation of the dynamic behaviour
of the stacked apples can be obtained by means of the dis-
crete element method (DEM). The DEM is a numerical tech-
nique to model the kinematic and dynamic behaviour of
particles. In the case of packed fruit, each particle represents
a single piece of fruit. In the DEM all forces acting on the par-
ticles (gravity force and contact force) are considered and the
equations of motion of Newton and Euler are integrated to
obtain the velocity and position of the particles in the next
time step. The DEM was originally developed in the eld of
rock mechanics and has been used extensively for engineer-
ing materials, but applications for biological materials are
scarce, in particular for soft biological materials such as fruit.
When used for the simulation of the dynamic behaviour of
packed fruit, the rotation of the apples, the impact with all
surrounding objects and the friction between the apples
can be accounted for. Pioneering work with two-dimensional
DEM simulations of in-transit fruit damage was carried out
by Rong et al. (1993) for a total of 12 particles (i.e. apples).
The contact forces between the apples were simulated for
two types of road irregularities (curbs and ramps) and for
both soft and stiff vehicle suspensions.
In order to improve the control of mechanical fruit han-
dling with respect to bruising, a generic three-dimensional
DEM model was implemented in the DEMeter software by
Tijskens et al. (2003). Van Zeebroeck et al. (2006a,b) also applied
this model to simulate the dynamic behaviour of packed ap-
ples. Newmeasuring techniques that allowfor the determina-
tion of the impact parameters of biological materials in the
normal and tangential directions have also been developed.
Furthermore, the use of the DEM in the simulation of impact
damage of fruit has been validated and DEM simulations
have been applied to investigate impact damage of apples
due to continuous vibrations during transport (Tijskens
et al., 2003; Van Zeebroeck et al., 2003, 2004, 2006a,b; Dintwa
et al., 2004, 2005a,b).
In this paper, three-dimensional DEMsimulations are used
to investigate impact damage caused by the passage of a truck
on a speed bump for apples in completely lled commercial
bulk bins. Furthermore, the inuence of truck load, position
of the bulk bin, vehicle suspension type and vehicle speed
on fruit damage are investigated.
2. Materials and methods
2.1. Two-stage approach
The simulation of the apple impact damage during the pas-
sage of the truck over a speed bump was performed in two
stages. In the rst stage, a two-dimensional vehicle model of
a Volvo FL6 truck was used to predict the dynamic response
of the truck during passage over the speed bump. In the sim-
ulations, the static load of the bulk bins was taken into ac-
count, while the dynamic interaction between the vehicle
body and the bulk bins was disregarded. This is a reasonable
assumption when no loss of contact occurs between the ap-
ples, the bulk bin and the vehicle body and when the reso-
nance frequency of the bulk bin with the stacked apples is
relatively high compared with the resonance frequencies of
the vehicle. In the second stage, the dynamic response of
the vehicle body was used to excite the apples in the bulk
bin. The vertical response of the truck body depends on the
suspension stiffness, position, driving speed and total truck
load was investigated. A soft suspension, where the value of
the suspension stiffness was halved, and a stiff suspension,
where the stiffness was doubled, were also considered. The vi-
brations of the vehicle body were considered at three posi-
tions: (1) at the rear axle, (2) 2 m in front of the rear axle and
(3) 2 m behind the rear axle. The vehicle speed varied between
20 kmh
1
and 50 kmh
1
. Three different loading conditions
were considered for the truck: lightly loaded, half loaded,
and fully loaded.
The dynamic response of the vehicle body was subse-
quently used to determine the dynamic response of the apples
by means of the DEM simulations. The box (bulk bin) in the
simulations has a length of 1.15 m and a width of 0.96 m, cor-
responding to the size of commercial bulk bins used in Bel-
gium. The simulations were performed using DEMeter,
a library of C DEM routines developed at the Katholieke
Universiteit Leuven. Each simulation considered 1512 apples
in a maximum stack height of 0.57 m. It was assumed that
the diameter of the apples could be represented by a normal
distribution with an average value of 0.076 m and a standard
deviation of 0.0076 m, covering the entire range of commercial
apple diameter classes. The inuence of the aforementioned
parameters on the apple bruising was investigated by means
of a total of 90 DEM simulations. In each simulation, 7 s real-
time are simulated in a computation time of 40 min using
a computer equipped with a P4 processor operating at a speed
of 2.4 GHz and a dynamic memory of 512 MB RAM. In the sim-
ulation, the rst 2 s were used to obtain a natural stacking of
the apples in the bulk bin, which was simulated by dropping
the apples from a certain height into the box. This event
was not taken into account when determining bruising. The
remaining 5 s in the simulation covered the vertical excitation
of the box by passage over the speed bump.
2.2. Roadvehicle model
In this section, the dynamic response of a truck during the
passage on a speed bump is considered. The focus was on
the response at the rear of the vehicle body, where the cargo
b i os y s t e ms e ng i ne e r i ng 1 0 1 ( 2 0 0 8 ) 5 8 6 8 59
was present. The calculations were performed for the particu-
lar case of a two-axle Volvo FL6 truck. The Volvo FL6 truck has
a wheel base of 5.20 m and a maximum total mass of
14,000 kg. Similar types of trucks are frequently used for
road transport of fruit. The simulations were performed
with Matlab (Mathworks, Natick, MA, USA).
In the following, the case is considered where a sinusoidal-
shaped speed bump excites the vehicle. The longitudinal un-
evenness prole u
w/r
( y) of the speed bump represents the de-
viation of a travelled surface from a true planar surface at
each position y along the road:
u
w=r
y
H
2

1 cos

2py
L

for
L
2
y
L
2
(1)
In Belgium, the Royal Decree of 9 October 1998 prescribed
a height H0.12 m and a length L 4.80 m for this type of
speed bump (Fig. 1a) (NN, 1998). These parameters were cho-
sen such that the increase of the car body accelerations, and
hence the discomfort of the driver, was the strongest at
a speed of 30 kmh
1
. A forward Fourier transformation with
respect to the longitudinal coordinate y reveals the wavenum-
ber content ~ u
w=r
k
y
of the prole (Fig. 1b). The wavenumber
content showed zeros at k
yn
2pn/L (n >1) and was mainly sit-
uated in the range of wavenumbers k
y
below 2.6 rad m
1
or
wavelengths l
y
2p/k
y
longer than 2.4 m.
As the speed bump excited the vehicle simultaneously
along both wheel paths, a two-dimensional vehicle model suf-
ced. The main interest was in the calculation of the dynamic
response in a frequency range between 0 Hz and 20 Hz. In this
frequency range, the vehicle body and axles are assumed to be
rigid and models for the simulation of vehicle ride behaviour
as shown in Fig. 2 are used (Cebon, 1993). The vehicle body
and axles are represented by discrete masses, while the sus-
pension system and the tyres are modelled as a spring-
dashpot system.
As part of an experimental validation for the prediction
method for ground vibrations due to road trafc, a linear vehi-
cle model was developed for the Volvo FL6 truck (Lombaert
and Degrande, 2003). This is shown in Fig. 2. The parameters
of the inertial elements and the spring constants have been
determined fromweighings, information fromthe truck man-
ufacturer and an experimental modal analysis. The damping
constants have been estimated from a t of the predicted
and measured frequency content of the axle response during
a passage of the truck on a ply-wood unevenness at a speed
v 30 kmh
1
. The following values were used (Lombaert
et al., 2000; Lombaert and Degrande, 2003): the mass of the ve-
hicle body m
b
9000 kg, the rotational inertia of the body
I
b
35,000 kg m
2
. The position of the rear and front axles
with respect to the centre of gravity was l
1
1.49 m and
l
2
3.72 m, respectively. In the following, the position of the
rear axle position is referred to as the position y l
1
; the posi-
tion at 2 m in front of the rear axle is the position y l
1
2 m,
while the position at 2 m behind the rear axle is the
y l
1
2 m position.
The mass of the rear and front axles was m
a1
600 kg and
m
a2
400 kg, respectively. The spring constants of the rear
and front suspensions were k
p1
0.61 10
6
Nm
1
and
k
p2
0.32 10
6
Nm
1
. The corresponding damping constants
are c
p1
16,000 Ns m
1
and c
p2
10,050 Ns m
1
. The spring
constants of the rear and front tyres are k
t1
3.00 10
6
Nm
1
and k
t2
1.50 10
6
Nm
1
. The damping constants of the tyres
were assumed to be zero, so that c
t1
0 Ns m
1
and
c
t2
0 Ns m
1
. This mathematical model of the Volvo FL6
truck has been shown to successfully predict the ground vi-
brations during a passage of the truck over the ply-wood
bump for a range of vehicle speeds between 20 kmh
1
and
60 kmh
1
(Lombaert and Degrande, 2003).
2.3. Discrete element model
In the DEM simulations, the parameters for the normal and
tangential contact force models and the bruise prediction
Fig. 1 (a) The longitudinal road prole u
w/r
( y) of a sine-
shaped trafc bump as a function of the coordinate y along
the road and (b) ~ u
w=r
k
y
in the wavenumber domain.
Fig. 2 Two-dimensional 4DOF model for a vehicle with
two axles.
b i o s ys t e ms e ng i ne e r i ng 1 0 1 ( 2 0 0 8 ) 5 8 6 8 60
given by Van Zeebroeck et al. (2003) and Van Zeebroeck et al.
(2006a) were used. These parameters are briey summarised
here.
A pendulum device was used to develop a contact force
model that describes the apple-metal impact. The model con-
sisted of the parallel connection of a spring and a damper. For
the particular case of Jonagold apples, the following values
for the damping constants c (kg m
1/2
s
1
) and the spring con-
stants k (Nm
3/2
) have been determined experimentally by
Van Zeebroeck et al. (2003) (both R
2
0.38):
c
R
914 465v 8:7r (2)
c
G
914 465v 8:7r (3)
k
R
336712 434707v 38732r (4)
k
G
1 117131 434707v 11664r (5)
where the subscripts R and G refer to the red and green side of
the apple, respectively, while v denotes the impact velocity
(ms
1
) and r is the radius of curvature (mm). According to
the classical Hertz contact theory, the spring constant de-
pends on the effective radius of curvature. Kuwabara and
Kono (1987) derived the damping constant in a similar way.
Furthermore, the parameters also depended on the impact
velocity, as discussed in detail by Van Zeebroeck (2005).
Although the parameters were only valid for the impact of
apples on a hard wall, they could be modied to represent
the contact forces for contact between apples (Van Zeebroeck,
2005). The values above indicate that the damping and spring
constants were higher for the red side of the Jonagold apple
compared to the green side. However, in the DEMsimulations,
distinctions were not made between the sides and average
values were used to determine the contact force. In anticipa-
tion of the implementation of the viscoelastic extension of
the Mindlin and Deresiewicz model in the DEMeter soft-
ware (Dintwa et al., 2005a,b) a tangential contact force model
based on dry Coulomb friction was applied. A value of 0.27
was used for the dynamic friction coefcient in the tangential
contact force model since it had been determined experimen-
tally for appleapple contact (Van Zeebroeck et al., 2004). The
same value was also applied for the contact between the ap-
ples and the wall.
Based on the DEMresults for the contact forces, the level of
bruising was determined. The model of Chen and Sun (1981)
was used to predict bruise volume (BV) from the peak value
(PF) of the contact force. Aregression analysis based on results
from the pendulum device determined the following relation-
ship:
BV 15:81PF 608:90 (6)
with a coefcient of determination (R
2
) of 0.90. In the simula-
tions, the location of the impact on the apple surface was not
stored. This requires a local coordinate system for each parti-
cle, a feature that was currently, not then available, in the
DEMeter software. The BV was therefore interpreted as be-
ing the single impact BV. A validation of the model can be
found in Van Zeebroeck et al. (2006a). The calculation of the
BV was performed in Matlab, allowing for further analysis of
the occurred damage, such as the histogram of the number
of apples with the volume of the largest bruise in a certain
damage class and the number of apples with damage above
a certain threshold.
3. Results
3.1. Roadvehicle simulation
In the simulations, as the vehicle traversed the speed bump, it
was excited by an imposed displacement u
w/r
( y) at the contact
points between the vehicle and the road. This implied that the
vehicle was assumed to remain in perfect contact with the
road and that the exibility of the road was disregarded.
The time history of the excitation was determined by the ve-
hicle speed v and obtained from u
w/r
( y), replacing y by vt.
Fig. 3a shows the time history of the excitation at the front
axle for vehicle speeds of 10 kmh
1
, 30 kmh
1
and
50 kmh
1
. These three cases will be considered here. Apart
from a small time delay, the rear axle experienced the same
excitation.
As the vehicle model was linear, the equations of motion
were solved in the frequency domain. The frequency content
b u
w=r
u of the imposed displacement was calculated from the
representation ~ u
w=r
k
y
of the unevenness in the wavenumber
domain:
b u
w=r
u
1
v
~ u
w=r

u
v

(7)
Fig. 3 (a) Time history u
w/r
(t) and (b) frequency content
b u
w=r
u of the signal applied at the vehicle axles during the
passage on a speed hump at 10 kmh
L1
(dotted line),
30 kmh
L1
(solid line) and 50 kmh
L1
(dashed line).
b i os y s t e ms e ng i ne e r i ng 1 0 1 ( 2 0 0 8 ) 5 8 6 8 61
This expression shows how the vehicle speed v couples the
wavenumber content ~ u
w=r
k
y
of the unevenness prole to
the frequency content b u
w=r
u of the signal that is applied to
each axle of the vehicle. A disturbance with a wavelength l
y
corresponds to a harmonic excitation of the vehicle at a fre-
quency v/l
y
. Fig. 3b shows the frequency content of the excita-
tion for the same three speeds. As can be anticipated from Eq.
(7), the frequency content shifted to higher frequencies for in-
creasing vehicle speed. In the following, results are given for
the vehicle response and the reader is referred to Lombaert
et al. (2000) for more details about the equations of motion of
the vehicle.
Fig. 4a shows the displacement b uu of the vehicle body on
top of the rear axle. The displacement u(u) was obtained from
the displacement u
b
(u) at the centre of gravity of the truck
body and the rotation f
b
(u) about the centre of gravity as
u
b
(u) l
1
f
b
(u). At a speed of 10 kmh
1
, the vehicle reacted
nearly statically and the displacement of the vehicle body
(Fig. 4a) was similar to the imposed displacement (Fig. 3b).
At higher vehicle speeds, the dynamic excitation become
more important, and a resonance frequency near 1.4 Hz
appeared. This corresponds to a pitch and bounce mode of
the vehicle, where the rear part of the vehicle body rotates
about a rotational axis near the front axle. At low frequencies
between 0 Hz and 2 Hz, the response was larger for the speed
of 30 kmh
1
, as expected from the results in Fig. 3b. The
higher speed of 50 kmh
1
further shifted the response to
higher frequencies, but did not lead to a greater response.
Fig. 4b shows the time history u(t) of the displacement that
has been obtained as the inverse Fourier transformation of
b uu in Fig. 4a. At a time t 1.78 s, the front axle of the truck
was at the centre of the bump. However, it can be seen that
the excitation at the front axle is relatively unimportant to
the response at the rear part of the vehicle body. The response
was dominated by the passage of the rear axle at a time t
which was equal to 3.66 s, 2.41 s and 2.16 s for the vehicle
speeds of 10 kmh
1
, 30 kmh
1
and 50 kmh
1
, respectively.
At the lowest vehicle speed, the response showed quasi-static
lifting of the vehicle during the passage on the speed bump.
This was no longer observed in the response at the higher ve-
hicle speeds.
Fig. 5a and b shows similar results for the acceleration at
the rear of the truck body. As the frequency content b au of
the acceleration is obtained as the product of u
2
and b uu
(Fig. 4a), the higher frequency range is more important. The
maximum response in the time domain (Fig. 5b) was still
obtained for the vehicle speed of 30 kmh
1
.
In Fig. 6, the predicted inuence of the positionon the truck
body acceleration is shown for the passage of the partly
loaded truck on the speed bump at a speed of 30 kmh
1
. The
displacement of the vehicle body on top of the rear axle is
compared to the response at 2 m in front of the rear axle
( y l
1
2 m) and 2 m behind the rear axle ( y l
1
2 m). The
greatest frequency content is found at y l
1
2 m (Fig. 6a) or,
behind the rear axle. The peak response in the time domain
(Fig. 6b) is also found at this position. This is because the re-
sponse of the rear part of the truck was dominated by a pitch
and bounce mode that involves a rotation of the vehicle body
about an axis close to the front axle. The displacements and
Fig. 4 (a) Frequency content b uu and (b) time history u(t)
of the displacement at the rear of the truck during the
passage on a speed hump at 10 kmh
L1
(dotted line),
30 kmh
L1
(solid line) and 50 kmh
L1
(dashed line).
Fig. 5 Passage of the truck on a speed bump. (a) Frequency
content b au and (b) time history a(t) of the acceleration at
the rear of the truck during the passage on a speed hump
at 10 kmh
L1
(dotted line), 30 kmh
L1
(solid line) and
50 kmh
L1
(dashed line).
b i o s ys t e ms e ng i ne e r i ng 1 0 1 ( 2 0 0 8 ) 5 8 6 8 62
accelerations of the vehicle body therefore increased with the
distance from the front axle.
In Fig. 7, the predictions for the partly loaded truck are
compared to those for a lightly loaded truck (m
b
5200 kg,
I
b
19,200 kg m
2
) and a fully loaded truck (m
b
13,000 kg,
I
b
42,700 kg m
2
). The centre of gravity of the vehicle body
was recalculated by assuming that the centre of gravity of
the load coincided with the location of the rear axle. The
change of the vehicle body mass m
b
therefore mainly affects
the frequency at which the pitch and bounce mode of the
rear part occurs (Fig. 7a). The inuence of the mass m
b
can
be understood by considering the problem of a simple
sprung mass system, excited at its base by u
w/r
(u
r
). When
u
r
and x represent the resonance frequency and the damp-
ing ratio of the sprung mass system, respectively, a higher
mass m lowers both the resonance frequency u
r
and the
damping ratio x. The shift of the resonance frequency is
clearly observed in Fig. 7a. For the sprung mass system,
the acceleration at the resonance frequency u
r
equals
u
r
u
w/r
(u
r
)/2x and scales with the mass m as u
w=r
u
r
=

m
p
.
However, the decrease in acceleration at resonance with
an increasing mass m is partly compensated by the change
of the frequency content u
w/r
(u
r
) with the resonance fre-
quency u
r
(Fig. 3b). The change of the resonance frequency
will also affect the speed at which the maximum response
occurs, as can be understood from Fig. 3b. Fig. 7b shows
how the peak acceleration in the time domain decreases
with an increasing mass m
b
.
Fig. 8 compares the rear body acceleration for three values
of both the front and the rear suspension stiffness for the half-
loaded truck. Because the response of the vehicle body on top
of the rear axle is being considered, the results are mainly af-
fected by the change of k
p1
. For a simple sprung mass system,
higher suspension stiffness increases the resonance fre-
quency u
r
and lowers the damping ratio x. The shift of the
resonance frequency is clearly observed in Fig. 8a. The accel-
eration at the resonance frequency scales with the spring con-
stant k of the sprung mass system as k
1.5
. The strong increase
with k is only partly compensated by the generally lower
values u
w/r
(u
r
) (Fig. 3b) for increasing resonance frequencies
u
r
. The speed at which the maximum response is predicted
to occur in the frequency domain will shift to lower speeds
for the smaller and higher speeds for the larger suspension
stiffness. The peak acceleration in the time domain (Fig. 8b)
increases with the suspension stiffness k
p1
.
3.2. Discrete element simulations
Using DEM simulations, the dynamic response of the stacked
apples during the passage of the truck on the speed bump was
determined. Fig. 9 shows eight snapshots that visualise the
position of the apples at different times. These snapshots
have been determined by means of the DEMeter software.
At the time t t
1
, the initial position of the apples at a certain
height above the box is shown. The height of the domain
Fig. 6 Passage of the truck on a speed bump at 30 kmh
L1
.
(a) Frequency content b au and (b) time history a(t) of the
vehicle body acceleration at the 2 m in front of the rear axle
(dotted line), at the rear axle (solid line) and at 2 m behind
the rear axle (dashed line).
Fig. 7 Passage of the truck on a speed bump at 30 kmh
L1
.
(a) Frequency content b au and (b) time history a(t)of the
vehicle body acceleration for a lightly loaded (dotted line),
a partly loaded (solid line) and fully loaded truck (dashed
line).
b i os y s t e ms e ng i ne e r i ng 1 0 1 ( 2 0 0 8 ) 5 8 6 8 63
considered in the simulations is therefore larger than the ac-
tual height of a bulk bin. At the time t >t
1
, the apples were
dropped in the box (t t
2
), where they obtain their natural
stacking as shown for the time t t
3
. The bulk bin was excited
by the vertical response of the vehicle body at the times t >t
3
.
In the particular case of the stiff suspension, it can be ob-
served how all apples became airborne at a certain time
t t
4
. Some more images of the apples during the event are
shown for the times t t
5
, t t
6
and t t
7
, while t t
8
shows
the nal position of the apples when the vertical movement
of the vehicle body had diminished. These predictions show
that for the case of the stiff suspension, where some of the ap-
ples became airborne, the dynamic interaction between the
apples and the vehicle might affect the vehicle response.
Van Zeebroeck et al. (2006a,b) showedthat for eachsimulation,
a histogram of the bruising with a distribution of the apples
according to the BV of their maximum bruise can be given.
For the particular case of the lightly loaded truck, with a bulk
binat y l
1
2 manda driving speedof 35 kmh
1
, anexample
of histogram is shown in Fig. 10. Because 90 simulations were
carried out the information in the histograms was further re-
duced by considering only the number of apples with a maxi-
mumbruised volume above 500 mm
3
and 2600 mm
3
(Tables 1
3). This information is sufcient to evaluate bruising. Based on
the The United States Standards for Grades of Apples and
previous research, it can be stated that apples with a single
BV beyond 500 mm
3
can be considered as being injured,
which means that a defect is present that detracts from the
apples appearance, edibility, and shipping quality (NN, 2002;
Van Zeebroeck et al., 2007). Apples with a single BV beyond
2600 mm
3
can be considered as being seriously damaged, in-
dicating that a defect is present that seriously detracts from
the appearance, edibility, and shipping quality.
3.2.1. Effect of truck load (reference suspension) on apple
bruising
The truck load was predicted to have an important effect on
the amount of bruising for all driving speeds. Higher truck
loads tended to decrease the predicted apple bruising. For
lightly loaded trucks, only driving speeds below 30 kmh
1
Fig. 8 Passage of the truck on a speed bump at 30 kmh
L1
.
(a) Frequency content b au and (b) time history a(t) of the
vehicle body acceleration for suspension stiffness k
p1
/2
and k
p2
/2 (dotted line), k
p1
and k
p2
(solid line), and 2k
p1
and
2k
p2
(dashed line).
Fig. 9 Position of the stacked apples for the case of a bulk bin located at the rear axle in the case of a truck with a stiff
suspension crossing the speed bump at a speed of 50 kmh
L1
: t [t
1
initial position of the apples prior to the stacking (no
contact); t [t
2
the apples are dropped in the box; t [t
3
the apples have obtained their natural stacking; t >t
3
excitation of
the bulk bin by the passage on the speed bump; t [t
4
the apples become airborne; t [t
5
, t [t
6
and t [t
7
show some more
images of the apples during the event; t [t
8
shows the nal position of the apples.
b i o s ys t e ms e ng i ne e r i ng 1 0 1 ( 2 0 0 8 ) 5 8 6 8 64
gave rise to low bruising for bulk bins at y l
1
2 m and y l
1
.
In the case of y l
1
2 m, driving speeds below 25 kmh
1
are
necessary to achieve commercially acceptable bruising. It can
be concluded that for lightly loaded trucks and all bulk bin po-
sitions driving speeds below 25 kmh
1
are necessary to main-
tain commercially acceptable bruising.
For the half-loaded truck there was a clear peak in the pre-
dicted damage for all bulk bin positions around 2535 kmh
1
.
For bulk bins at y l
1
2 m and y l
1
, this is the only critical
driving speed. In this case, bruising below 25 kmh
1
and be-
yond 35 kmh
1
can be neglected (n.b. driving speeds beyond
50 kmh
1
were not simulated). For bulk bins at y l
1
2 m,
the 2535 kmh
1
zone was also critical, but beyond
35 kmh
1
the predicted bruising was still considerable. In
this case only driving speeds below 25 kmh
1
gave rise to
commercially acceptable bruising. Therefore, it can be con-
cluded that for half-loaded trucks and all bulk bin positions
driving speeds below 25 kmh
1
are necessary to maintain
commercially acceptable apple bruising.
For fully loaded trucks and bulk bins at y l
1
2 m and
y l
1
, the bruising was commercially acceptable for all driving
speeds (again remembering that driving speeds beyond
50 kmh
1
were not simulated). On the other hand, driving
speeds between 25 kmh
1
and 35 kmh
1
gave rise to consid-
erable predicted bruising for bulk bins at y l
1
2 m. In this
case, only driving speeds below 25 kmh
1
and beyond
35 kmh
1
gave rise to commercially acceptable bruising
(once again remembering that speeds beyond 50 kmh
1
were not simulated). It can be concluded that for fully loaded
trucks and all bulk bin positions, driving speeds below
25 kmh
1
and between 35 kmh
1
and 50 kmh
1
can lead to
commercially acceptable bruising.
The important inuence of the truck load on the amount of
apple bruising can be explained by the fact that the truck load
lowers the acceleration levels in the time domain (Fig. 7b). Fur-
thermore, the driving speed at which the peak damage occurs
also depends on the truck speed. The peak damage for lightly
loaded trucks occurs at a higher velocity (35 kmh
1
) than that
for fully loaded trucks (2530 kmh
1
). This can be explained
by the fact that higher truck load lowers the resonance fre-
quency u
r
(Fig. 7a) and therefore the critical driving speed at
which b u
w=r
u
r
is at a maximum. This was explained in detail
in Section 3.1.
The results in Tables 13 show that, for bulk bins at
y l
1
2 m, the number of apples predicted to have commer-
cial bruising in the case of the fully loaded truck is only 33% of
the number for lightly loaded trucks. For bulk bins at
y l
1
2 m and y l
1
, this effect is much more pronounced.
The number of apples with commercial bruising in the case
Fig. 10 Histogram of the number of apples with their
maximum single bruise volume in damage classes for
a bulk bin at y [l
1
D2 m and the passage of a lightly
loaded truck on the speed bump at a driving velocity of
35 kmh
L1
.
Table 1 The effect of driving speed, truck load and
suspension type on apple bruising for bulk bins at y [l
1
(rear axle)
Velocity
(kmh
1
)
Damage
>500 mm
3
Heavy damage
>2600 mm
3
Lightly loaded truck (reference suspension)
20 5 0
25 90 0
30 1456 619
35 1483 741
40 1469 626
50 1487 501
Half full truck (reference suspension)
20 27 0
25 732 46
30 1141 163
35 584 27
40 158 1
50 162 0
Full truck (reference suspension)
20 24 0
25 15 0
30 43 0
35 21 0
40 7 0
50 8 0
Stiff suspension (half full truck)
20 26 0
25 417 11
30 1500 1324
35 1500 1434
40 1500 1380
50 1481 634
Soft suspension (half full truck)
20 10 0
25 15 0
30 6 0
35 17 0
40 11 0
50 24 0
The second and third columns note the number of apples with at
least one bruise with a bruise volume higher than, respectively,
500 and 2600 mm
3
The bold lines indicate the velocity at which
the peak damage occurs.
b i os y s t e ms e ng i ne e r i ng 1 0 1 ( 2 0 0 8 ) 5 8 6 8 65
of the fully loaded truck is predicted to be less than 3% of that
for lightly loaded trucks.
3.2.2. Effect of bulk bin position on apple bruising
The position of the bulk bin position has a large inuence on
the amount of bruising predicted. Bruising increased from
y l
1
2 m over y l
1
to y l
1
2 m. This result could be
expected as during the passage of the rear axle on the speed
bump, the truck body exhibited a pitch and bounce motion
with a rotation centre near the front axle (Fig. 6b).
The results in Tables 13 show that the number of apples
predicted to have commercial bruising for the lightly loaded
truck where y l
1
2 m is half of the number in the case
where y l
1
2 m. For the half loaded and fully loaded trucks
the difference was even more pronounced with the number of
apples with commercial bruising at y l
1
2 m only 5% of the
number in the case where y l
1
2 m.
3.2.3. Effect of suspension type (half-loaded trucks) on
apple bruising
Suspension type is also an important inuence on the amount
of bruising. In the case of the soft suspension, the amount of
apple bruising predicted was negligible for all driving speeds
and bulk bin positions, whereas for the stiff suspension there
was considerable bruising. This result could be expected as in
Fig. 8b it was demonstrated that the acceleration levels in the
time domain increased with suspension stiffness. For the stiff
suspension commercial bruising was predicted to start at
30 kmh
1
for bulk bins at y l
1
2 m, y l
1
and 25 kmh
1
for bulk bins at y l
1
2 m.
Table 2 The effect of driving speed, truck load and
suspension type on apple bruising for bulk bins at
y [l
1
D2 m (in front of rear axle)
Velocity
(kmh
1
)
Damage
>500 mm
3
Heavy damage
>2600 mm
3
Lightly loaded truck (reference suspension)
20 19 0
25 16 0
30 692 23
35 1288 213
40 1178 105
50 381 6
Half full truck (reference suspension)
20 12 0
25 30 0
30 29 0
35 66 0
40 76 0
50 31 0
Full truck (reference suspension)
20 19 0
25 13 0
30 10 0
35 15 0
40 28 0
50 32 0
Stiff suspension (half full truck)
20 14 0
25 45 0
30 1305 214
35 1444 412
40 1485 530
50 1495 780
Soft suspension (half full truck)
20 7 0
25 5 0
30 8 0
35 9 0
40 9 0
50 14 0
The second and third columns note the number of apples with at
least one bruise with a bruise volume higher than, respectively,
500 and 2600 mm
3
. The bold lines indicate the velocity at which
the peak damage occurs.
Table 3 The effect of driving speed, truck load and
suspension type on apple bruising for bulk bins at
y [l
1
L2 m (behind rear axle)
Velocity
(kmh
1
)
Damage
>500 mm
3
Heavy damage
>2600 mm
3
Lightly loaded truck (reference suspension)
20 16 0
25 986 60
30 1499 1027
35 1494 829
40 1491 706
50 1480 525
Half full truck (reference suspension)
20 88 0
25 1492 1114
30 1494 1104
35 1310 211
40 468 14
50 1313 240
Full truck (reference suspension)
20 55 0
25 1234 193
30 746 30
35 35 0
40 35 0
50 179 0
Stiff suspension (half full truck)
20 48 0
25 1481 794
30 1500 1360
35 1500 1403
40 1500 1346
50 1499 904
Soft suspension (half full truck)
20 19 0
25 19 0
30 8 0
35 13 0
40 59 0
50 325 3
The second and third columns note the number of apples with at
least one bruise with a bruise volume higher than, respectively,
500 and 2600 mm
3
. The bold lines indicate the velocity at which
the peak damage occurs.
b i o s ys t e ms e ng i ne e r i ng 1 0 1 ( 2 0 0 8 ) 5 8 6 8 66
The results in Tables 13 show that the vehicle speed that
corresponds to the maximum predicted bruising increases
with the suspension stiffness. This is in accordance with the
larger resonance frequency of a stiffer suspension (Fig. 8a)
and the effect of the vehicle speed on the excitation.
Furthermore, these results show that for bulk bins at
y l
1
2 m, the number of apples predicted to have commer-
cial bruising in the case of the truck withthe soft suspension is
only 6%of the number for the stiff suspension. For bulk bins at
y l
1
2 m and y l
1
, this effect is even more pronounced.
The number of apples predicted to have commercial bruising
is only l% of the number in the case of the stiff suspension.
3.2.4. Effect of driving velocity on apple bruising
In all situations (different truck loads and suspension types)
driving speeds of 20 kmh
1
(and less) do not lead to commer-
cially signicant apple bruising. In almost all cases consid-
ered, the predicted peak damage occurred at a vehicle
speeds between 25 kmh
1
and 35 kmh
1
. However, for bulk
bins at y l
1
2 m and the case of the stiff suspension the
largest amount of damage was predicted for a vehicle speed
of 50 kmh
1
.
The vehicle speed at which the largest amount of damage
was predicted corresponded to the highest accelerations
levels of the vehicle body. For the particular case of the half-
loaded truck and a bulk bin at y l
1
, the largest amount of
predicted damage (Table 1) and the peak acceleration
(Fig. 5a) occurred at a speed of 30 kmh
1
.
4. Discussion
The previous remarks regarding the inuence of the truck
load, the driving speed, the bulk bin position and the suspen-
sion type on the apple bruising are in agreement with results
of simulations, experiments reported inliterature, and operat-
ing experience. However, relatively little is found in literature
on the effect of transient vibrations or shocks on fruit bruis-
ing. The study by Holt and Schoorl (1985) demonstrates the su-
periority of soft vehicle suspensions over hard suspensions in
reducing apple bruising due to shocks by potholes and bumps,
but to date no other experimental results on the inuence of
the truck load and the bulk bin position can be found in liter-
ature for transient vibrations or shocks. For continuous vibra-
tions, Hinsch et al. (1993) and Timm et al. (1996) demonstrated
the superiority of air-ride suspensions (soft suspension) in re-
ducing fruit bruising, as compared to steel spring suspensions
(hard suspension). Although Berardinelli et al. (2005), working
with pears, measured increasing continuous vibrations levels
from the front to middle position to rear position and they
gave no information on the level pear damage at the different
positions. As for transient vibrations, no data on the effect of
truck load on fruit bruising is found in literature for continu-
ous vibrations. Numerical simulations of fruit damage due
to transient or continuous vibrations are also rare. Schoorl
and Holt (1985) demonstrated, using a theoretical method
similar to that of Jones et al. (1991), the superiority of soft sus-
pensions over hard suspensions in reducing apple bruising
when shocks are provoked by potholes and bumps. The pio-
neering work of Rong et al. (1993) of DEM simulations also
demonstrated the clear effect of suspension type on apple
bruising. In the numerical study of Jones et al. (1991) the effect
of truck load, driving speed and bulk bin position on fruit
bruising was considered. Although a less elaborate model
was used to describe the dynamic behaviour of the stackedap-
ples, the results were in a very close agreement with those of
the present study. Jones et al. (1991) also found that increasing
truck load decreased the bruising. The interaction of load and
suspension, when load is treated as a single mass, was recog-
nised by Page (1973) who noted that as the sprung mass (or the
load) increased relative to the unsprung mass the acceleration
of the unsprung mass decreased. Jones et al. (1991) also found
that bruising increased as the stacked apples were moved
from 1.25 m in front of the rear axle, to above the rear axle,
and then to 1.25 m behind the rear axle. For lightly loaded
trucks (0.1 tonne) and for speed bumps with slopes at 45

,
a lengthof 0.24 mand the same height of 0.12 mas in the pres-
ent study, the peak damage is found at a similar speed of
35 kmh
1
. For 1 tonne loaded trucks, the peak damage occurs
at a lower speed of 30 kmh
1
, which agrees with the present
results. For more heavily loaded trucks (2 tonne and 4 tonne)
Jones et al. (1991) found the largest amount of damage for ve-
hicle speeds of 54 kmh
1
and 72 kmh
1
for the 2 tonne and 4
tonne truck load, respectively. In the present results, however,
the peak damage was predicted in all cases for speeds be-
tween25 kmh
1
and 35 kmh
1
, except for the soft suspension
simulation where the peak damage was at 50 kmh
1
(again
noting that higher speeds were not considered). The differ-
ence is probably due to the different shape of the speed
bump model used in the study of Jones et al. (1991).
In the present simulations with the sinusoidal-shaped
speed bump, the largest amount of damage generally occurs
at a speed close to 30 kmh
1
. This can be explained by the
fact that the speed bump had been designed such that the in-
crease of the vehicle body accelerations was greatest at this
speed (NN, 1998).
The shape of the speed bump is therefore an important pa-
rameter that affects the vehicle response and, therefore, the
amount of bruising. Extrapolations to other sinusoidal speed
bumps could be made, considering that the vehicle response
scales linearly with the height H of the speed bump and that
the predictions only depend on the ratio v/L of the vehicle
speed v and the length L of the bump.
Further experimental evidence can be found in Jones et al.
(1991), who observed during the unloading of produce vehicles
that a larger amount of fruit damage was found for lightly
loaded trucks and that the worst damage occurred over or be-
hind the rear axle and was exacerbated by a stiff suspension.
5. Conclusions
The DEM simulations clearly demonstrate how the suspen-
sion type, the truck load, the driving speed and the bulk bin
position affect apple bruising caused by the passage of a truck
on a speed bump. For fully loaded trucks no commercially sig-
nicant bruising is predicted for driving speeds between
20 kmh
1
and 50 kmh
1
for apple bulk bins positioned 2 m
in front the rear axle and on top of the rear axle. However,
driving at speeds between 25 kmh
1
and 35 kmh
1
must still
b i os y s t e ms e ng i ne e r i ng 1 0 1 ( 2 0 0 8 ) 5 8 6 8 67
be avoided because these speeds were predicted to cause con-
siderable bruising for apples in bulk bins at 2 m behind the
rear axle. For light and half-loaded trucks the bruising was
much more pronounced with up to 33 times more bruising
for lightly loaded trucks compared to fully loaded trucks.
With driving speeds below 25 kmh
1
, however, commercial
bruising was also avoided for these truck loads. The results
clearly demonstrate how soft suspension types considerably
reduce the amount of damage, even up to 99% when com-
pared to stiff suspensions for bulk bins behind the rear axle.
For lightly loaded trucks and half or fully loaded trucks, the
simulations predicted 50% and 5% less bruising, respectively,
for bulk bins in front of the rear axle compared to bulk bins be-
hind the rear axle.
Acknowledgements
The authors want to thank the Research fund of the Katho-
lieke Universiteit Leuven and the Scientic Research Fund
Flanders (FWO).
r e f e r e n c e s
Berardinelli A; Donati V; Giunchi A; Guarnieri A; Ragni L (2005).
Damage to pears caused by simulated transport. J Food Eng,
66, 219226.
Cebon D (1993). Interaction between heavy vehicles and roads.
Technical report, Cambridge University Engineering Department.
Chen P; Sun Z (1981). Impact parameters related to postharvest
bruising of apples. ASAE Paper No. 81-3041. St. Joseph,
Michigan.
Dintwa E; Van Zeebroeck M; Tijskens E; Ramon H (2004).
Torsional stiffness of viscoelastic spheres in contact. Eur Phys
J B, 39, 7785.
Dintwa E; Van Zeebroeck M; Tijskens E; Ramon H (2005a).
Determination of parameters of a tangential contact force
model for viscoelastic spheroids (fruits) using a rheometer
device. Biosyst Eng, 91, 321327.
Dintwa E; Van Zeebroeck M; Tijskens E; Ramon H (2005b). Torsion
of viscoelastic spheres. Granular Mat, 7, 169179.
Hinsch R T; Slaughter D C; Craig W L; Thompson J F (1993).
Vibration of fresh fruits and vegetables during refrigerated
truck transport. Trans ASAE, 36, 10391042.
Holt J E; Schoorl D A (1985). Theoretical and experimental analysis
of the effects of suspension and road prole on bruising in
multilayered apple packs. J Agric Eng Res, 31, 297308.
Jones C S; Holt J E; Schoorl D (1991). A model to predict damage to
horticultural produce during transport. J Agric Engng Res, 50,
259272.
Kuwabara G; Kono K (1987). Restitution coefcient in a collosion
between two spheres. Jpn J Appl Phys, 26, 12301233.
Lombaert G; Degrande G (2003). The experimental validation
of a numerical model for the prediction of the vibrations in
the free eld produced by road trafc. J Sound Vib, 262,
309331.
Lombaert G; Degrande G; Clouteau D (2000). Numerical modelling
of free eld trafc induced vibrations. Soil Dyn Earthq Eng, 19,
473488.
Mohsenin N N (1973). Physical Properties of Plant and Animal
Materials. Gordon and Breach Science Publishers, New York.
NN (1998). Koninklijk Besluit 9 oktober 1998. Koninklijk Besluit tot
bepaling van de vereisten voor de aanleg van verhoogde
inrichtingen op de openbare weg en van de technische
voorschriften waaraan die moeten voldoen. Belgisch
Staatsblad 28 oktober.
NN (2002). The United States Standards for Grades of Apples.
Department of Agriculture, USA.
Page J (1973). Dynamic behaviour of a single axle vehicle
suspension system: a theoretical study. T.R.R.L. (UK) Report
LR, 580, pp 113.
Rong G H; Negi S C; Jofriet J C (1993). DEM simulation of in-transit
fruit damage. ASAE Paper No. 93-6503.
Tijskens E; Ramon H; De Baerdemaeker J (2003). Discrete element
modeling for process simulation in agriculture. J Sound Vib,
266, 493514.
Timm E J; Brown G K; Armstrong P R (1996). Apple damage in bulk
bins during semi-trailer transport. Appl Eng Agric, 12, 369377.
Van Zeebroeck M (2005). The discrete element method (DEM) to
simulate fruit impact damage during transport and handling.
PhD no. 643, Faculty of Bio-engineering, Catholic University
Leuven.
Van Zeebroeck M; Dintwa E; Tijskens E; Deli V; Loodts J; De
Baerdemaeker J; Ramon H (2004). Determining tangential
contact force model parameters for viscoelastic materials
(apples) using a rheometer. Postharvest Biol Technol, 33,
111125.
Van Zeebroeck M; Tijskens E; Dintwa E; Loodts J; De
Baerdemaeker J; Ramon H (2006a). The Discrete Element
Method (DEM) to simulate fruit impact damage during
transport and handling: model building and validation of DEM
to predict bruise damage of apples. Postharvest Biol Technol,
41, 8591.
Van Zeebroeck M; Tijskens E; Dintwa E; Loodts J; De
Baerdemaeker J; Ramon H (2006b). The Discrete Element
Method (DEM) to simulate fruit impact damage during
transport and handling: case study of vibration damage during
apple bulk transport. Postharvest Biol Technol, 41, 92100.
Van Zeebroeck M; Tijskens E; Van Liedekerke P; Deli V; De
Baerdemaeker J; Ramon H (2003). Determination of the
dynamical behaviour of biological materials using a pendulum
device. J Sound Vibr, 266, 465480.
Van Zeebroeck M; Van linden V; Darius P; De Ketelaere B;
Ramon H; Tijskens E (2007). The effect of fruit factors on the
bruise susceptibility of apples. Postharvest Biol Technol, 46,
1019.
b i o s ys t e ms e ng i ne e r i ng 1 0 1 ( 2 0 0 8 ) 5 8 6 8 68