Sunteți pe pagina 1din 10

Traction versus slip in a wheel-driven belt conveyor

A.J.G. Nuttall
*
, G. Lodewijks
Delft University of Technology, Transport Technology and Logistics, Mekelweg 2, 2623 CD Delft, The Netherlands
Received 13 July 2005; received in revised form 15 December 2005; accepted 2 January 2006
Available online 2 March 2006
Abstract
This paper presents an extension of existing models, used for at belt conveyors, to describe the relationship between
traction and slip in a wheel-driven belt conveyor with a curved surface. The model includes the viscoelastic properties of
the rubber running surface in the form of Maxwell elements. After the application of a correction factor to account for the
interaction between adjacent elements, which is initially not modelled, experimental results show that the model generates a
satisfactory match and that belt speed has little eect on traction in the feasible speed range.
2006 Elsevier Ltd. All rights reserved.
Keywords: Rolling contact; Traction; Viscoelasticity; Maxwell model; Pouch belt conveyor; Curved belt surface
1. Introduction
Traditionally belt conveyors for transporting bulk material have a drive station at the head and/or tail of
the system where the belt is wrapped around a drive pulley, see Fig. 1. It is a well proven drive conguration
for belt conveyor systems with a single or dual drive stations. However, problems arise when more than two
drive stations are desired. Due to the fact that the drive pulley cannot be placed at any arbitrary location along
the carrying strand of the belt without interfering with the bulk material ow on the belt, it cannot take full
advantage of the benets a distributed drive system has to oer.
An alternative drive method, which oers greater layout exibility in a multiple drives system, is to imple-
ment drive wheels that directly press onto the belts surface to generate the desired traction force. In the
EnerkaBecker System (abbreviated EBS) for example motors with drive wheels mounted on their output
shafts form a drive pair that can be placed at virtually any location along the belt. Bekel [1] also proposed
vulcanising a drive strip to the bottom of a conventional at conveyor belt, so it could be driven by a pair
of drive wheels. The freedom to place the drive stations at any location along the belt gives the system designer
an opportunity to control the tension in the belt by balancing the installed drive power with the resistances
that occur in each section. This is the key to keeping the tension in the belt low, giving the opportunity to
0094-114X/$ - see front matter 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mechmachtheory.2006.01.005
*
Corresponding author. Tel.: +31 15 2782004; fax: +31 15 2781397.
E-mail address: A.J.G.Nuttall@wbmt.tudelft.nl (A.J.G. Nuttall).
Mechanism and Machine Theory 41 (2006) 13361345
www.elsevier.com/locate/mechmt
Mechanism
and
Machine Theory
use the same light belt construction regardless of the overall conveyor length. This results in reduced belt costs,
greater layout exibility and oers possibilities to standardise the system components.
For both the conventional drive pulley and drive wheel, like in the EBS, the generated traction force is
determined by the friction coecient and the contact force between the belt and the pulley or drive wheel sur-
face. However, with the drive wheel conguration the contact force is not primarily determined by the belt
tension, but by the normal force that is generated as a result of the weight of the belt and the bulk solid mate-
rial it carries and the force generated by a clamping roller. Due to this dierence with a drive pulley the Euler
equation [2], which is normally used to determine the maximum transferable eective traction in a conven-
tional belt conveyor, cannot be applied to a wheel-driven conveyor. Therefore, a new model needs to be for-
mulated that takes the material and geometric properties of the belt and drive into account.
This paper presents a model that describes the relationship between traction and slip in the rolling contact
patch of a wheel-driven belt conveyor like the EBS. The model includes the viscoelastic rubber properties of
the rubber belting material as an array of Maxwell elements and is compared to an elastic approach used by
Bekel [1]. Both models are also compared with experimental results. The tractionslip relationship is of interest
because the traction and slip combined with the applied normal contact force greatly inuence the wear rate of
the belts surface. To prevent the belt from wearing out before its guaranteed lifetime, a maximum allowable
wear rate has to be set, which can result in a derating of the maximum transferable traction.
2. Modelling contact forces based on viscoelastic properties
A number of researchers have used the Maxwell model to quantify the energy dissipation of a cylinder
rolling on a viscoelastic surface [35], which is comparable to a conveyor belt passing over an idler. As the
cover passes the idler the rubber surface compresses and relaxes in quick succession. Due to the viscoelastic
properties of the rubber cover material the relaxation will take some time. This causes an asymmetrical stress
distribution that results in a resistance force. To incorporate the viscoelastic properties and calculate the
indentation resistance, the Maxwell model has mainly been used in its three parameter form. One model in
particular, described by Lodewijks [6], combines the three parameter Maxwell model with a Winkler founda-
tion or mattress consisting of springs that do not interact with each other. Because shear forces between
adjacent spring elements are not considered calculations become less complex. Despite the simplication
results show that this representation of the belt cover behaviour generates satisfactory results. Therefore,
the combination of the Maxwell model and Winkler foundation will serve as starting point for the analysis
of the relationship between traction and slip of a wheel-driven conveyor belt.
In order to adopt a similar approach to describe the traction force exerted by a drive wheel in the EBS, the
model is expanded in two ways. Firstly, the number of Maxwell elements is increased to accommodate a match
wrap angle
drive pulley
troughed
conveyor belt
support roll
contact line
F
l
v
b
Fc clamping roller
drive wheel
F
l
pouch
conveyor belt
triangular profile
support rollers contact patch
v
b
bulk material
Fig. 1. Support and drive conguration of a conventional (left) and a pouch (right) belt conveyor.
A.J.G. Nuttall, G. Lodewijks / Mechanism and Machine Theory 41 (2006) 13361345 1337
between the model and the real rubber behaviour throughout the contact patch. Secondly, a brush model, also
used to describe the rubber tread behaviour of car tyres [7], is introduced to calculate the shear forces caused
by slip between the drive wheel and the belt.
The three parameter Maxwell model, consisting of a single Maxwell element in series with a spring, suces
for a conventional conveyor belt because the contact surface between the belt and idler can be described by a
line contact. With a constant contact length throughout the contact zone the model only has to match for a
single frequency of excitation, making a good approximation possible by tuning the time constant of the single
Maxwell element to this frequency. However, as a result of the curved running surface in the EBS, there exists
an elliptical contact zone. Due to the varying contact length in the elliptical patch, the model has to match for
a range of frequencies. Fig. 2 shows how the model represents the belt passing over an idler or drive wheel. A
rigid cylinder rolling with angular velocity x is pushed onto a curved viscoelastic surface moving with the belt
velocity v
b
, which results in the elliptical contact patch.
To match the model with the rubbers viscoelastic properties within the excitation range, additional
Maxwell elements are introduced. An array of Maxwell elements approximates the viscoelastic behaviour each
consisting of a spring with stiness E
i
and a dashpot with a damping coecient g
i
, as illustrated in Fig. 3.
Ideally this model would have an innitely large number of elements. However, due to practical and compu-
tational reasons the ideal situation is simplied by limiting the number of elements to m.
The total stress in this model is equal to the sum of the stress on the single spring and the stresses in each
Maxwell element, or
r r
0

m
i1
r
i
; 1
where r
0
is the stress in the single spring, which is directly related to the material strain e
r
0
E
0
e. 2
The stress in the remaining spring and dashpot elements is directly related to the local strains of the individual
element
r
i
E
i
e
E
i
; 3a
r
i
g
i
_ e
g
i
; 3b
where e
E
and e
g
represent the local strain of the spring and dashpot elements, respectively. The sum of the local
strains is equal to the total strain on the element. Taking the time derivative of the strains leads to
_ e _ e
E
i
_ e
g
i
. 4
The time derivatives of e
E
and e
g
can be found from the Eqs. (3a) and (3b)
z
y
v
b
R
2
R

1
curved running surface
rolling cylinder
contact patch
x
F
z
Fig. 2. Rigid cylinder rolling on a curved visoelastic surface.
1338 A.J.G. Nuttall, G. Lodewijks / Mechanism and Machine Theory 41 (2006) 13361345
_ e
E
i

_ r
i
E
i
; 5a
_ e
g
i

r
i
g
i
. 5b
Combining Eqs. (5a) and (5b) with Eq. (4) results the following relationship between the total strain and the
stress in each spring dashpot assembly
_ r
i
r
i
E
i
g
i
E
i
_ e. 6
Together with Eq. (1) and (2) the dierential equations (6) of all Maxwell elements form a set of equations that
when solved gives the normal stress in the contact plane.
The parameters of the Maxwell model have to be tuned to match its complex modulus of elasticity with
viscoelastic properties of the belt cover measured in oscillatory experiments where the material is subjected
to sinusoidal varying stresses and strains [8,9]. Fig. 4 shows the results of such experiments for the rubber used
in the EBS belt. The results of these experiments are typically expressed as the storage modulus E
0
, loss mod-
ulus E
00
and loss factor tand. Together they represent the complex modulus of elasticity and are related as
follows:

1
E
1
E
0

2
E
2

m
E
m
Fig. 3. Modelling the viscoelastic properties with Maxwell elements.
Fig. 4. Measured and approximated viscoelastic properties.
A.J.G. Nuttall, G. Lodewijks / Mechanism and Machine Theory 41 (2006) 13361345 1339
E

E
0
i E
00
; 7
tan d
E
00
E
0
. 8
To t the properties of the model onto the measured data the storage and loss modulus are expressed as a
function of the model parameters and the excitation frequency x. This is accomplished by eliminating the
stresses of the individual Maxwell elements from Eq. (1) with Eq. (6) and substituting the strain e with the
periodic function sin(xt), which results in
E
0
E
0

m
i1
x
2
g
2
i
E
i
x
2
g
2
i
E
2
i
; 9
E
00

m
i1
xg
i
E
2
i
x
2
g
2
i
E
2
i
. 10
The number of Maxwell elements m to be used in the model depends on the required accuracy of the complex
modulus of elasticity in a desired frequency range. With a possible operational belt speed of 1.610 m/s and an
approximated contact length of 0.02 m, the frequency of excitation ranges from 80 to 500 Hz. The accuracy
generally increases when more elements are added. However, with more elements the model also becomes more
complex, making computations more time consuming and the search for starting conditions that give a good
convergence of the optimisation routine during the matching procedure increasingly dicult. Furthermore, due
to the implemented least squares approach, the maximum number of elements is physically limited by the
amount of experimentally measured data. It is impossible to t a model with more parameters than data points.
Fig. 4 shows how the model ts onto the measured viscoelastic properties of the EBS when dierent num-
bers of Maxwell elements are used. The gure clearly illustrates the dierence between the simplest model with
one element (or three parameters) that gives an unsatisfactory approximation between 10 and 1000 rad/s and a
model with three elements (or seven parameters) with an improved accuracy. The seven parameter model was
nally chosen as a good match and used for further calculations.
3. Normal stress distribution
When a drive wheel applies a traction force to the conveyor belt within the traction limit, stick and slip-
zones exist in the contact plane. In the stick-zone only the rubber surface deforms due to the applied traction,
while in the slip-zone the rubber surface also slides over the wheels surface because the friction limit has been
reached. To determine the placement of the zones, friction is modelled according to the Coulombs dAmon-
tons law:
jsx; yj 6 lrx; y; 11
where l is the friction coecient.
To solve this equation, the pressure distribution r(x, y) in the contact plane is determined rst, by dening
the deformation of the viscoelastic surface in the direction of the z-axis (see Fig. 2). For this calculation an
assumption, also used by Johnson [3], is made that the shear stress does not inuence the normal stress dis-
tribution. If the contact zone is small compared to the curvatures of the rolling cylinder and rubber surface (so
x ( R
1
and y ( R
2
), and the cylinder is pressed into the surface with a distance z
0
, then the deformation of the
contact surface can be described as follows:
wx; y z
0

x
2
2R
1

y
2
2R
2
with z
0

c
2
2R
2
. 12
Under steady state conditions with a constant belt speed v
b

dx
dt
v
b
, using a Winkler foundation with thick-
ness h and the deformation equation (12) e
wx;y
h
, the dierential equation (6) for each Maxwell element can
be written as
or
i
ox
r
i
E
i
g
i
v
b
E
i
x
hR
1
. 13
1340 A.J.G. Nuttall, G. Lodewijks / Mechanism and Machine Theory 41 (2006) 13361345
This dierential equation can be solved by setting the pressure at leading edge a(y) of the contact plane equal
to zero or r (a, y) = 0 because at the rst point of contact no deformation has occurred yet. Solving the equa-
tion reveals the pressure distribution in the contact plane
rx; y
E
0
2R
1
h
a
2
x
2

m
i1
E
i
k
i
hR
1
x a a k
i
1 exp
x a
k
i
_ _ _ _ _ _
with k
i

g
i
v
b
E
i
. 14
The resulting normal force F
z
can now be calculated by integrating the stress distribution over the whole con-
tact region or
F
z

_
c
c
_
ay
by
rx; y dxdy. 15
The trailing edge of the contact plane positioned at b(y) is found by setting r(x, y) equal to zero.
4. Shear stress distribution
With the calculated pressure distribution and a measured friction coecient, most of the information is
available to determine the shear stress within the slip-zone, as determined by Eq. (11). The next essential step
to nd the shear stress distribution in the whole contact plane is the calculation of the shear stress in the stick-
zone.
In the stick-zone no sliding takes place between the contact surfaces. However, an apparent speed dierence
or creep does occur between the drive wheels outer diameter and the belt when a traction force is applied. This
apparent velocity is also know as the creep ratio d and is dened as follows:
d
v
b
xR
1
v
b
j j
; 16
where x is the angular velocity of the drive wheel.
The creep ratio is related to the shear angle by the following equation:
oc
ox

d
h
. 17
To establish a relationship between the creep ratio and shear stress distribution in the stick-zone, the Maxwell
model is combined with a brush model that describes shearing eects. The brush model depicted in Fig. 5 is a
simplied representation of the belt cover in the contact region. It consists of rigid elements that hinge and are
held in place by a torsion spring at their base. The behaviour of the torsion spring is also based on the Maxwell
model analogous to the spring element in Fig. 3.

z
x
R
1
v
b
F
z
h
M
d
hinge
brush element

Fig. 5. Brush model.


A.J.G. Nuttall, G. Lodewijks / Mechanism and Machine Theory 41 (2006) 13361345 1341
By replacing the modulus of elasticity E, stress r and strain e in Eq. (1), (2) and (6) with the shear modulus
G, shear stress s and shear angle c respectively equations are derived that describe the behaviour of the brush
elements. Under steady state conditions and using the deformation equation (17) the dierential equation
describing the shearing of each Maxwell element can be written as
os
i
ox
s
i
G
i
g
i
v
b
G
i
d
h
. 18
To derive the viscoelastic shear parameters, additional oscillatory experiments should be conducted where the
rubber test sample is subjected to shear stresses and strains. However, due to the fact that no results of shear
experiments were available, the shear parameters were derived from the normal stress experiments and con-
verted with the aid of the following equation:
G
E
21 m
. 19
If it is assumed that the stick-zone starts at the leading edge of the contact plane, a solution to dierential
equation (18) can be found, yielding the shear stress in the stick-zone
s
stick
x; y
d
h
G
0
a x

m
i1
dg
i
v
b
h
1 exp
G
i
x a
g
i
v
b
_ _ _ _ _ _
. 20
The contribution of both the stick and slip-zone can now be calculated by integrating the calculated shear
stress in each zone separately
F
traction

_
c
c
_
t
1
y
by
l rx; ydx
_
ay
t
1y
s
stick
x; ydx
_ _
dy; 21
where t
1
(y) represents the transition line separating the stick from the slip-zone. It represents the edge where
the shear stress reaches the friction boundary and it can be found by solving
s
stick
t
1
; y l rt
1
; y. 22
5. Correction factor
A correction factor f
s
is introduced to compensate for the fact that the Winkler foundation does not incor-
porate the shearing eect between adjacent spring elements and to match the stiness of the model with the
actual stiness of the layer. Under the condition that the speed dierence between the drive wheel and the belt
is small, the slip region at the trailing edge becomes vanishingly small. As there is virtually no slip in the con-
tact region, the occurring speed dierence or creep is predominantly determined by the layer stiness. The cor-
responding limit for the creep ratio, as derived by Johnson [3] using a half space approximation, is
d
aF
0
t
2RF
0
z
or F
0
t

2RF
0
z
a
; 23
where F
0
t
and F
0
z
are measured per unit length of the contact width.
The normal force F
0
t
can be expressed as a function of the distance to the leading edge a of the contact zone.
Bekel [1] derived the following equation, using the Hertz formulas:
a

8F
0
z
R1 m
2

pE
_
; 24
where E is the statically measured modulus of elasticity. With this equation the normal force F
0
z
is eliminated
from Eq. (23). To match the stiness of the brush model, the tangent at the start of the models traction curve
has to match the creep curve described by Eq. (23), which is calculated by
lim
d!0
F
0
t
f
s
_
a
b
s
stick
dy f
s
d
h
g
0
2
a b
2
v
b

n
i1
g
i
a b k
ig
1 exp
a b
k
ig
_ _ _ _ _ _
_ _
; 25
where k
ig

g
i
v
b
g
i
and f
s
is the correction factor.
1342 A.J.G. Nuttall, G. Lodewijks / Mechanism and Machine Theory 41 (2006) 13361345
Elimination of F
0
t
by combining Eqs. (23)(25) gives the following correction factor:
f
s

apEh
41 m
2
p
;
p
g
0
2
a b
2
v
b

n
i1
g
i
a b k
ig
1 exp
a b
k
ig
_ _ _ _ _ _
:
26
The stiness of the model is compensated by scaling the Maxwell parameter with the factor of Eq. (26).
6. Experimental validation
Experiments were conducted to measure the actual relationship between traction and slip at a drive station
in the EBS and validate the presented model. During the experiments two wheels were used, see Fig. 6. One
wheel made from steel represents the drive wheel and is driven by an electric drive motor. The other wheel,
representing the belt cover, has a rubber layer (h = 30 mm) vulcanised to it. It is also connected to an electric
motor that is used as an adjustable brake. Strain gauges on each motor shaft measure the produced torque. An
adjustable spring was also used to pull the brake wheel onto the drive wheel, making it possible to control the
contact force. The diameters (D
d
= D
b
= 500 mm) of both wheels were chosen such that their contact patch,
created when pressed against each other, is comparable with the patch between the drive wheel (D = 250 mm)
and the belt in the EBS.
At the start of each experiment the contact force and the drive wheel speed are set to a desired value. To
compensate for a decrease in brake wheel diameter due the indentation of the rubber layer, the speed of the
brake wheel is adjusted just below synchronous speed until the brake torque reduces to zero. From this point,
where the measured traction is zero, a traction slip curve is created by successively decreasing the brake wheel
speed and measuring the resulting increasing traction. Fig. 7 presents the results for dierent contact forces
and a constant speed. It also shows the curves that were calculated with the presented viscoelastic model
and the equations presented by Bekel that he used to describe the traction slip relationship for a wheel-driven
rubber strip [1]. Bekel used a similar half space approach as described by Johnson [3] for a line contact involv-
ing completely elastic material, which results in
d
al
r
R
1

1
F
t
lF
z
_ _
with
1
r
R

1
r
1

1
r
2
. 27
The results show that the presented Maxwell model gives a good match with the measured values for low con-
tact forces. As the contact force increases the model starts to underestimate the actual traction.
To assess the inuence of the viscoelastic properties on traction, dierent curves where calculated with vary-
ing speeds. Fig. 8 presents the results for a constant contact force with speeds ranging from the EBSs stan-
dard belt speed of 1.6 m/s to a potential high speed application with a belt speed of 10 m/s.
rubber layer
brake wheel
drive wheel
F
z
M
d
F
d
hinge
M
b
adjustable spring
Fig. 6. Experimental layout with drive and brake wheel.
A.J.G. Nuttall, G. Lodewijks / Mechanism and Machine Theory 41 (2006) 13361345 1343
The curves in Fig. 8 suggest that traction decreases with increasing speed, with the greatest reduction occur-
ring in the middle part of the slip range. However, this eect seems very small in the feasible speed range of a
belt conveyor. With the speed inuence in the same order of magnitude as the measurement error, it can be
concluded that in this case the viscoelastic part of the rubber properties has a small inuence on the relation-
ship between traction and slip.
7. Conclusion
This paper shows that it is possible to expand a three parameter Maxwell model, which is used to calculate
the rolling resistance of a cylinder rolling on a viscoelastic layer, and include the behaviour required to deter-
mine the relationship between traction and slip. Such a model has a number of simplications, making it
relatively simple and computationally friendly for a contact model that includes viscoelastic behaviour. After
the introduction of a correction factor to compensate for the stiness of the layer, results collected from exper-
iments with the cover material of the EBS conveyor belt show that the model generates satisfactory matches
with the measured values. Further analysis with the validated model shows that the speed dependency of the
traction slip relationship is small in the feasible speed range of a belt conveyor.
The knowledge gained from the relationship between traction and slip will be a valuable asset for the system
designer, when choosing the number of drive stations to install in a belt conveyor system like the EBS. As
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
0
500
1000
1500
slip
T
r
a
c
t
i
o
n

(
N
)
Maxwell model
Elastic half space
x Measurements
Fz = 500 N
Fz = 1000 N
Fz = 1500 N
Fig. 7. Comparison of experiments and model (v
b
= 1.6 m/s).
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
0
300
600
900
slip
T
r
a
c
t
i
o
n

(
N
)
1.6 m/s
5.0 m/s
10 m/s
Fig. 8. Speed inuence with F
z
= 1000 N.
1344 A.J.G. Nuttall, G. Lodewijks / Mechanism and Machine Theory 41 (2006) 13361345
both traction and slip generated by the drive wheels inuence the overall belt wear, the system designer will
have to determine the minimum number of drive stations that will give an acceptable wear rate or belt life.
However, little is known about the eect that traction and slip have on the wear rate in the EBS. Therefore,
further research will focus on this issue, using the presented model as a guide.
References
[1] S. Bekel, Horizontalkurvengangiger Gurtfo rderer mit dezentralen Reibradantrieben, Thesis, Hameln, 1992.
[2] A.C. Low, J.W. Kyle, Recommended Practice for Troughed Belt Conveyors, The Mechanical Engineers Association, London, 1986.
[3] K.L. Johnson, Contact Mechanics, Cambridge University Press, Cambridge, 1985.
[4] W.D May, E.L. Morris, D. Atack, Rolling friction of a hard cylinder over a viscoelastic material, Applied Physics 30 (1959) 17131724.
[5] S.C. Hunter, The rolling contact of a rigid cylinder with a viscoelastic half space, Applied Mechanics 28 (1961) 611617.
[6] G. Lodewijks, Dynamics of Belt Systems, Thesis, Delft University of Technology, Delft, 1995.
[7] H.B. Pacejka, The Role of Tyre Dynamic Properties, Smart Vehicles, Swets and Zeitlinger Publishers, Lisse, 1995, pp. 5568.
[8] G. Lodewijks, Determination of rolling resistance of belt conveyors using rubber data: fact or ction? BeltCon 12, 2324 July 2003,
Johannesburg, South Africa.
[9] A.N. Gent, Engineering with Rubber, Carl Hanser Verslag, 2001.
A.J.G. Nuttall, G. Lodewijks / Mechanism and Machine Theory 41 (2006) 13361345 1345

S-ar putea să vă placă și