Sei sulla pagina 1di 9

Non-Linear Model for Dynamic Analysis of Conveyors

David Beavers, David Morrison & David Rea


Sinclair Knight Merz
Abstract
This paper describes the dynamic analysis of a conveyor belt based on a finite element model. Each element in the
model contains a mass, stiffness and damping component. In particular the stiffness element contains a non-linear
characteristic to account for the stiffness behavior of the belt at low tensions. The dynamic model was produced
from the static analysis of the conveyor which included an analytical drag model based on the work of Jonkers [4],
Spaans [7], Wheeler [8] and SKF [2]. The analytical drag model has also been incorporated into the dynamic model
to show the change in drag due to different tensions throughout the belt. The drag model described is based on the
methods established in published papers. The result of this dynamic model is representation of the tensions in a
conveyor belt. In particular it enables low tensions to be modeled accurately.
Nomenclature
F(t) Forcing function
M Mass matrix
C Damping matrix
K Non-linear stiffness matrix
x Displacement
E Elastic modulus (kN/m)
q Belt mass (kg/m)
a Idler spacing (m)
T Belt tension (kN)
I Moment of inertia (m
4
)
V Velocity (m/s)

Introduction
In the vast majority of cases, the process of designing a conveyor belt involves a static approach that assumes the
belt to be moving at a constant velocity and the belt is an inextensible body. The purpose of this static analysis is to
ensure the correct components have been selected for the conveyor and to confirm the conveyor will operate
effectively under normal operating conditions. An important aspect of the static analysis is to determine the overall
drag in the conveyor and from that, the power requirements and tensions expected throughout the conveyor. For
short conveyors the standards (such as ISO5048) suggest a friction factor that can be used for the drag occurring
along the conveyor. This method is generally acceptable for a short conveyor as the magnitude of the error involved
is proportional to the length and thus relatively small. However, a small error in the friction factor is amplified for
any long conveyor due to the large number of idler sets and the interaction between the idlers and the belt. For this
reason it is important to use a more detailed approach when designing long conveyors. Overly conservative
assumptions used to determine the drag will lead to an over designed system with high capital and operating costs as
a result of over specified equipment being installed [1].
The static approach also doesnt account for the dynamic response of the conveyor belt and the resulting forces and
shockwaves that occur during transient events [5]. For that reason a dynamic analysis of a conveyor belt is often
used to investigate the transient response of the belt during different starting and stopping scenarios, such as
emergency stop and aborted start. Understanding the drag experienced by the conveyor under transient loading
events is therefore paramount in establishing a valid model of a conveyors behavior.
For a long belt conveyor the resistances and forces vary throughout the length of the conveyor. By applying a static
model with tensions averaged along the conveyor this variation in drag cannot adequately be taken into account.
Therefore the resistances are likely to be underestimated in areas of low belt tension and overestimated in areas of
the belt subject to higher tensions. For this reason a more accurate approach to model the conveyor is to use a non-
linear Finite Element Method (FEM) (see figure 1). This method takes into account the variations in both drag and
stiffness along the conveyor.
The important aspects to take into account when determining the dynamic response of a conveyor are the properties
of the belt and material, the resistive forces throughout the conveyor and the belt sag between idlers.
The general theory applied to the elastic response of the belt is to assume the stiffness of the belt is constant during
all loading conditions. However, discrepancies between field measurements and dynamic models show that this
theory is not valid when the belt is subject to low tensions. For this reason an algorithm to account for the non-linear
stiffness of the belt will be presented.

Figure 1: FEM Model

Drag Model
The drag experienced by a conveyor is split into four components; main resistances, secondary resistances, special
resistances and material lift. The main resistances occur at each idler station along the full length of the conveyor,
the secondary resistances occur at certain locations along the conveyor, e.g. loading zones and skirting boards, and
the special resistances occur on certain conveyor installations. For long conveyors the main resistances are a
significant proportion of the load on the conveyor. Therefore a more detailed approach to conveyor design is to
develop an analytical model of the main resistances instead of applying a general friction factor to the whole
conveyor. The resistance model presented is based on published research by Jonkers [4], Spaans [7] and Wheeler [8]
that includes indentation resistance, material flexure resistance, belt flexure resistance and rolling resistance.
The indentation model is based on the work of Jonkers [4] and looks at the deformation of the belt as it passes over
an idler roller and the resulting horizontal force developed from the asymmetrical stress distribution (see figure 2).


Figure 2: Stress distribution of belt on roller [6]
Material flexure is caused by the cyclic expansion and contraction of the material in both the transverse and
longitudinal direction. The transverse compression is due to the material conforming to the shape of the idler and the
longitudinal expansion and contraction is due to the sag of belt between idler sets. The compression of the material
creates a passive stress state and the relaxation creates an active stress state. The model adopted to describe this
resistance is based on the work by Spaans [7].
The flexure resistance of the belt is not the most significant contributor to the Main resistance; but is large enough to
warrant inclusion in the model. The resistance due to belt flexure is caused by the flexing of the belt as it passes over
an idler roll. As the belt approaches the idler roll it is bent up causing it to gain energy due to the bending moment.
As the belt leaves the idler roll gravity bends it back in the other direction, causing the belt to lose energy. If this belt
exhibited no flexural resistance then the energy lost would equal the energy gained. However, the properties of
rubber result in a loss during reverse bending known as a hysteresis energy loss. The hysteresis loss is equal to the
flexural resistance of the belt. The model for the flexural resistance is based on the work by Spaans [7].
The final main resistance in the conveyor is the rolling resistance; this is due to the drag developed in the idler
bearings, labyrinth seals and lip seals. The resistance in bearings is due to the sliding and rolling action between
bearing elements [2]. A labyrinth seal has resistance due to shear stresses developed between the grease and the fins
[8]. Finally, a lip seal has resistance due to frictional drag between the lip of the seal and the shaft [2]. Due to the
variation in geometry of idlers it is difficult to develop a general analytical model for the drag in each type of roller
available on the market. For this reason, where possible, the experimental results from idler suppliers are used to
determine the overall resistance. When this data is not available the set of equations from the SKF catalogue and the
research by Wheeler is used to predict a rolling resistance.
Dynamic Model
The standard approach for a dynamic model is to assume the belt has a linear stiffness. This concept is adequate
when the tensions in the belt are high and the belt is behaving like an elastic solid. It will be shown that as the
tension decreases the stiffness of the belt does not behave as an elastic solid and the response becomes non-linear.
This is due to the increase of the belt sag between the idlers / increased length of belt between the idlers.
For the dynamic analysis to be relevant the varying drag and belt stiffness due to belt tension, component selection,
and the operating conditions needs to be taken into account. These factors can be accurately modeled by determining
a number of system properties. This includes: mass of the belt, the burden, idler and pulleys; stiffness of the belt;
external forces such as drives, brakes and resistances in the conveyor; and the take-up. Typically the finite element
(see figure 3) used in this model to represent these properties is composed of a mass, stiffness and damping
component, with an external force applied to each element. The general equation of motion is given by equation 1.
Kx x C x M t F + + = ) ( (1)
The ends of the model are connected by the take-up, with the movement of the take-up calculated to be the
difference in displacement between the first and last elements.
In order to model the non-linearity of the belt stiffness, it was important to understand the physical events. Stiffness
by definition is the ratio of applied tension and elongation.

Figure 3: Representation of the element
Each element in the FEM model consists of numerous idler sets with a specified spacing and a tension applied at
each end. The displacement of these elements comprises two components; the elongation due to the strain of the belt
acting in the positive direction, and the reduction in length due to the belt sag acting in the negative direction (see
Figure 4). To calculate the actual deformation the resultant of these two displacements is required. To determine the
reduction in length an equation for the sagging length of belt between the idlers is required.

Figure 4: Sag of belt between idlers
If the length of belt between two idlers is considered, the deflection of the belt can be found by solving the following
differential equation for the belt under dynamic conditions.
E I
x

4
x
y

4
F
2
x
y

2
q
2
t
y

(2) [8]
As the belt hangs between the idlers it will take the shape of a catenary curve and this condition can be used to solve
for the amount of deflection (y(x)) in the belt. The solution of the equation above will take the form of equation 3,
where x is the distance from an idler:
(3) [8]

The actual length of the belt between the idlers can be found by applying the equation for an arc length to equation
4. This equation can then be solved by applying the limits as zero (k) and the distance between idlers (f).
(4)
There is no analytical solution to this integration; therefore a numerical solution was used. The method used was the
'n-point Gaussian Quadrature. This method uses an optimally chosen polynomial to approximate the integrand over
the required interval. The form that was used in this procedure is to use a uniform weighting over the interval and
evaluate the integral at a number of points over the interval determined by the roots of a Legendre Polynomial. For
the solution of this integral a 10
th
degree polynomial is used with the specified roots and weights. Using this
information the integral can be approximated by applying equation 5 and the standard equation for Gaussian
Quadrature

(3)
This equation can be applied to determine the actual length of the belt between the idlers and then the length of the
overall element can then be multiplied by the number of idlers in the element to obtain the actual length neglecting
the strain from the belt tension. The strain in the element due to the belt tension is then calculated by applying the
linear stress strain laws to the belt (6);

lin

E
:=

l
lin
l
0
:=
(6)
Equations 5 and 6 are now superimposed to calculate the displacement of the element due to sag and linear stress-
strain (7). The difference in these displacements is equal to the overall displacement of the element. From this the
strain, velocities, accelerations and tensions can be found for each element in the dynamic model.
(7)
As an example, figure 5 shows the Stress Vs Strain of an element subjected to tensions varying from 0.5 kN to
10 kN. Given that the stiffness of the material is the gradient of the stress strain curve the non-linear aspect can be
identified at the low tensions. Whereas at the high tensions the stiffness returns to an elastic solid. This model is
included in the dynamic analysis.
Stress V Strain
0.000
1.000
2.000
3.000
4.000
5.000
6.000
-3.000E-02 -2.000E-02 -1.000E-02 0.000E+00 1.000E-02 2.000E-02 3.000E-02
Strain
S
t
r
e
s
s
(
M
P
a
)
Linear
Non-Linear

Figure 5: Graph of Stress Vs Strain for linear and non-linear behaviour
The general equation of the dynamic model contains a stiffness matrix which accounts for the elastic modulus of the
conveyor belt. To incorporate the non-linear properties of the stiffness this matrix is recalculated each time step to
account for changes in tension and hence possible changes in stiffness. This is calculated by determining the stress
and strain of each element for each iteration. Assuming the change in tension is not excessive between iterations the
modulus can be approximated using equation 8. The subscript j denotes the iteration number.
E
nonlinear

j

j 1

j

j 1

:=
(8)
Case Study
The dynamic analysis presented in this paper was used on the Rio Tinto Coal Australia (RTCA) Clermont Coal
Mine Project to analyze the In Pit Crushing and Conveying (IPCC) system. This system comprises four relocate-able
conveyors that vary in length and lift over the lifetime of the mine. During the initial stages of design the worst
operating cases were determined and dynamic analyses were performed on these cases. The purpose of performing
the dynamic analysis was to determine the transient belt tensions in the conveyor, with particular attention paid to
the tail pulley and the take-up pulley. The following cases were analyzed, accelerating, braking, aborted start and
restart after braking. The dynamic analysis was also used to optimize the starting and braking routines to avoid too
low or high tensions.
As an example a brief review of the dynamic analysis that was performed on the Face Conveyor BC101 at stage 3B
is presented. The conveyor is transporting over-burden with a design capacity of 10,700 tph on an 1800mm wide
belt travelling at 5.75m/s. It has a length of 1820 m and a lift of 51m (see figure 6).

Figure 6: Conveyor Profile
The conveyor is driven by four 1350kW drives installed at the head end of the conveyor, each drive has a 2.7kNm
high speed brake installed. The take-up is a 132kW modulating winch which remains fixed during braking events.
The following plot (figure 7) shows the global velocity of the conveyor during the braking routine. The plot starts on
the return strand side of the take-up at station 1, the tail is station 14 and the primary drive is station 28.


Figure 7: Global velocity of conveyor during braking
During the braking routine a tension wave is developed when the drives are shut-off and the brakes are applied. This
wave is due to the effective tension across the drive pulleys dissipating its energy through the conveyor. For the
design of the conveyor it is important to determine the speed and magnitude of the wave to ensure the belt tension
stays within the design envelope for the selected carcass. It is also very important that the structural components are
designed for the transient tensions that occur at the take-up and the tail end of the conveyor. Plots A and B (figure 8)
show the tension and velocity of the belt at the tail pulley.

Figure 8: Plot A and B: Tension and velocity at tail pulley
The tension plot clearly shows the tension wave passing through the pulley after 3 seconds. The speed of this wave
can be predicted by determining the wave velocity (9) developed by Harrison.
b
b
o
q
E
V = (9)
An important characteristic of the plot is the peak magnitude of the tension wave passing through the tail. The
magnitude of this peak was predicted by the dynamic analysis to be more than twice the tension predicted by static
analysis. Therefore without the consideration of the transient tensions there could have been a risk that the tail pulley
frame would be under designed for these transient loads. Another feature that was identified during the braking
analysis was a period of low tension at the concave curve (figure 9) on the carry strand of the conveyor. This low
tension is caused by the mass of the material and the belt lacking the energy to travel up the hill and the effects of
the standing wave traveling through the belt. A method to counter this effect is to include a flywheel on the drives,
this reduces the effect of the standing waves through the belt as the high inertia drives effectively drive the conveyor
belt during deceleration.

Figure 9: Tension in carry strand of belt


Conclusions and Future Work
This paper presents an analytical static model of the main resistances in the conveyor that included indentation,
material flexure, belt flexure and rolling resistance. This model was then applied to the dynamic model of the
conveyor. Along with the drag model the dynamic model includes an algorithm to predict the effects of low tension
on the stiffness of the belt. This showed that at low tensions the belt no longer behaved as an elastic solid but
experiences non-linear stiffness due to the extra belt sag in the conveyor.

References
[1] Funke, H. and F.K. Konneker, Experimental Investigations and Theory for the Design of a Long Distance Belt
Conveyor System. Bulk Solids Handling, 1988. 8(5): p. 567.

[2] Gerber, C., SKF: General Catalogue. 1989, Germany: SKF. 56-63.
[3] Harrison, A., Stress Front Velocity in Elastomer Belts with Bonded Steel Cable Reinforcement. Bulk Solids
Handling, 1986. 6(1): p. 27-31..
[4] Jonkers, C.O., The Indentation rolling resistance of belt conveyors. Fordern und Heben, 1980. 30(4): p. 312-
317.
[5]. Nordell, L.K. and Z.P. Ciozda, Transient Belt Stresses During Starting and Stopping: Elastic Response
Simulated by Finite Element Methods. Bulk Solids Handling, 1984. 4(1): p. 93.
[6] Rudolphi, T.J. and A.V. Reicks, Viscoelastic Indentation and Resistance to Motion of Conveyor Belts using a
Generalized Maxwell Model of the Backing Material. Submitted to Rubber Chemistry and Technology, 2006.

[7] Spaans, C., The Calculation of the Main Resistance of Belt Conveyors. Bulk Solids Handling, 1991. 11(4): p.
809-825.
[8] Wheeler, C. Main Resistances. in An Intensive Short Course in Belt Conveying. 2005. University of Newcastle

Potrebbero piacerti anche