You are on page 1of 8

Wear 261 (2006) 435442

Development of wear model for cone crushers


M. Lindqvist , C.M. Evertsson
Department of Applied Mechanics, Chalmers University of Technology, SE 412 96 Goteborg, Sweden
Received 30 March 2005; received in revised form 2 November 2005; accepted 12 December 2005
Available online 19 January 2006

Abstract
Cone crushers are used in the aggregates and mining industries to crush rock material. A model to predict the worn geometry of cone crushers was
previously developed. In that model there was some disagreements between predicted and measured geometry and several effects were suggested to
explain the discrepancy in the model. In this study the effect of shear forces along the crushing surfaces was implemented in the model. Simulations
were compared to measurements on two different crushing chambers. The results show a significant improvement with respect to the discrepancy
between measured and simulated geometry. Measurements were made on a coarse crushing chamber where the operating parameters hydroset
pressure, power draw and capacity were tracked during the lifetime of the set of liners. The simulated operating parameters show some agreement
with measured data, but the crusher was not run under ideal conditions at all times.
2005 Elsevier B.V. All rights reserved.

Keywords: Comminution; Crushing; Modelling; Abrasive wear; Cone crusher

1. Introduction developed a flow model, a size reduction model and a pressure


response model.
Cone crushers are widely used in the mining and aggregates The geometry of the crushing chamber is crucial for the per-
industry to crush blasted rock material. The two main crushing formance. Due to wear the geometry of the liners will change,
parts are the mantle and the concave. The main shaft of the and hence the crusher performance will also change and some-
mantle is suspended on a spherical radial bearing at the top and times suffer. Therefore it is desirable to simulate the change of
in an eccentric at the bottom. A hydraulic cylinder supports the geometry and performance as the liners wear. A model for this
thrust bearing that carries the thrust force of the main shaft. The purpose was previously developed [3,4]. That model was based
hydraulic system can raise the main shaft in order to compensate on the results of Evertsson [1]. In the model for wear prediction
for the wear of the mantle and concave. The hydraulic pressure there was some discrepancy between the simulated geometry
in the cylinder that supports the thrust force from the main shaft and measured geometry in the upper part of the crushing cham-
is called the hydroset pressure. As the eccentric is turned the ber [3]. Several explanations of this discrepancy were suggested.
rock material will be squeezed and crushed between the liners It was first assumed that the work hardening behaviour of the
(see Figs. 14). liner material might depend on the applied pressure. In a study
Along its path through the crushing chamber, a rock parti- by the author [5] it was concluded that it was not a variation in
cle will be subjected to several crushing events. The shortest work hardening in the chamber that caused the discrepancy in
distance across the crushing chamber is called the closed side the wear model.
setting, CSS, and is an important variable for the performance of Among the other explanations for the discrepancy, that were
the crusher. The control system is calibrated regularly to main- proposed by Lindqvist and Evertsson [3], the prediction of
tain a constant CSS. Previous research [1,2] has made it possible pressure on the liners was assumed to be an important fac-
to model the behaviour of a given cone crusher. Evertsson [1] tor. To address this, an improved flow- and pressure model
was presented by Lindqvist and Evertsson [6]. That model
showed a significant improvement in prediction of the oper-
Corresponding author. Tel.: +46 31 772 13 76; fax: +46 31 772 3872.
ating parameters CSS, power draw and capacity, but only a
E-mail addresses: mats.lindqvist@me.chalmers.se (M. Lindqvist), slight improvement of wear prediction for a fine crushing
cme@mvs.chalmers.se (C.M. Evertsson). chamber.

0043-1648/$ see front matter 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.wear.2005.12.010
436 M. Lindqvist, C.M. Evertsson / Wear 261 (2006) 435442

Fig. 1. Cone crusher, schematic image.

Fig. 4. A new set of crusher liners, mantle and concave.

Other suggested explanations are non-linear dependency


between pressure and wear, shear stress at the interface between
rock and liner, dependency between particle size and wear rate.
Chenje and Radziszewski [7] showed that there was a non-linear
relationship between applied force and wear rate in a sliding
wear experiment. If Chenjes [7] results were also applicable
for the case of non-sliding wear in cone crushers, they would, at
least in part, explain the discrepancy. The technique used in the
present study, to measure the geometry of the liners, is similar to
the technique used by Rosario [8]. He has made measurements
of liner wear on gyratory crushers.
Among the possible explanations of the disagreement in the
model, shear forces in the contact between rock and liner is the
one that is addressed in this paper.
Fig. 2. Operating principle of cone crusher.

2. Method

2.1. Wear model

The wear model presented by Archard [9] suggests that wear


is proportional to sliding distance and applied pressure. In the
previous work carried out by the author [10] it was found that
wear occurs even if there is no macroscopic sliding motion
between rock material and liner. This is the case in a cone crusher
where there is no macroscopic sliding motion between liner and
rock. The mantle is free to roll against the bed of rock mate-
rial. On at least one point, the point of moment equilibrium for
the mantle, there is pure rolling between the mantle and bed
of material. At other points the relative sliding motion is very
small, since the concave is designed nearly as an ideal cone with
the generatrix of the mantle intersecting the pivot point of the
main shaft (see Fig. 5).
The wear model presented by Archard [9] suggests that the
wear rate is proportional to sliding velocity. If a worn crusher
liner is inspected, no ploughing grooves can be observed. The
wear mechanism is squeezing wear without macroscopic relative
motion between the bed of rock particles and the steel surface.
Fig. 3. An H8800 hydrocone crusher. This particular model is nearly 5 m high. On a small scale there is of course some relative motion since
M. Lindqvist, C.M. Evertsson / Wear 261 (2006) 435442 437

Fig. 6. The nip angle between the liners is larger for a coarse crushing chamber
(left) than for a fine chamber (right).

contact is here assumed to change the stress state around the con-
tact and increase the wear rate. As mentioned, it is not possible
to observe any ploughing grooves on a worn liner surface. This
indicates that there is no macroscopic sliding motion between
the rock particles and the steel surface and that friction is not
Fig. 5. Mantles are designed nearly as an ideal cone whose generatrix intersects fully developed.
the pivot point of the main shaft. If a particle is squeezed between oblique surfaces, the shear
force in the contact can be computed. Consider the particle
particles are rearranged as they are crushed, but the direction of squeezed between two oblique surfaces in Fig. 7. Since the par-
this motion is random. A wear model like Archards [9] that is ticle does not slip, the friction is not fully developed.
dependent of sliding velocity would in the case of cone crushers, The tangential frictional force Ft can be decomposed as the
yield no wear. Therefore, Lindqvist and Evertsson [3] adapted product of a frictional factor f times the normal force N. Since
the wear model used for cone crushers. friction is not fully developed f where is the coefficient
In the model for wear prediction, described by Lindqvist and of friction. With reference to Fig. 7, equilibrium require that
Evertsson [3] it is proposed that the amount of wear in a single

crushing action is proportional to the maximum average pres- N = F cos (2)
sure p that occurs during the crushing event (see Eq. (1)). In 2
this constitutive equation W is the wear resistance coefficient, a
material parameter unique for each combination of rock material fN = F sin (3)
2
and steel. Wear w is here expressed in mm, pressure in N/mm2 ,
and hence the unit for the wear resistance will have the unit
N/mm3 .
pmax
w = (1)
W
The average pressure expressed in Eq. (1), consists of a
large number of contact loads of different magnitude acting
on the steel surface. The wear that occurs is a function of the
mechanical properties of the steel, the number and magnitude
of the contact loads, and the shape and mechanical properties
of the rock particles. The wear resistance coefficient W is deter-
mined by the mechanical properties of the steel and rock, and is
verified in experiments or in full-scale measurements.
The wear resistance parameter W in Eq. (1) was found to be
94 kN/mm3 in a previous study [3]. The material was highly
abrasive quartzite in combination with austenitic manganese
steel. It was shown in that study that the wear model in com-
bination with the crusher model yielded an under-prediction of
wear in the upper part of the crushing chamber. The objective
here is to present a model that will address this discrepancy.
If a particle squeezed between oblique surfaces, as in Fig. 6,
the shear force increases as the nip angle increases. Among sev- Fig. 7. Shear forces are present when a particle is squeezed between oblique
eral mentioned and partly investigated reasons, a shear force in a surfaces.
438 M. Lindqvist, C.M. Evertsson / Wear 261 (2006) 435442

Fig. 9. Measurement rig.

2.2. Wear measurements

Fig. 8. Simulated pressure distribution on a mantle used with an H6800 EC A measurement rig that was previously developed for mea-
concave.
suring the worn geometry of cone crushers was used. The
method resembles the one used by Rosario (2004). The crusher
is stopped and a probe detects the location of the surfaces of
: fN cos = N sin (4)
2 2 the mantle and concave. The device is made of a frame that is
attached to the main shaft of the crusher (see Fig. 9). A step

motor moves a carrier by turning a threaded rod. Small stepping


motors send out probes. The number of pulses sent to the step
f = tan (5)
2 motor corresponds to a certain position relatively to the measur-
ing frame. When a probe contacts the liner the controller stops
If the computed factor f exceeds the coefficient of friction, the motor and the number of pulses is registered. The number
the particle will slide. of pulses is then converted into geometric coordinates.
In the crusher model, the pressure is computed according The measurements were carried out at the NCC quarry
to the pressure response model presented by Evertsson and located approximately 70 km:s east from Goteborg, Sweden.
Lindqvist [4]. The pressure response model relates compres- The crusher was a secondary SANDVIK H6800 crusher, with
sion ratio (i.e. the compressive engineering strain of the particle a coarse crushing chamber. The material fed to the crusher was
bed: deformation/original thickness), and variational coefficient 32250 mm granite that had previously been crushed in a pri-
of the particle size distribution to crushing pressure. A second- mary jaw crusher.
degree polynomial in two variables (compression and variational
coefficient of size distribution) was fitted to test results. The total 3. Results
pressure ptot is computed using the pressure response model (see
Fig. 8). 3.1. Measurements
So the shearstress pshear and normal pressure pnormal at the
surface is hence computed according to Eqs. (6) and (7). The coordinates from the measurements were transformed,
and the measured geometry was entered into a CAD-tool. Fig. 10
1
pnormal =  ptot (6)
1 + f2

f
pshear =  ptot (7)
1 + f2

where ptot is the total pressure computed from the pressure


response model. The proposed wear model hence looks as:

1
w = (pnormal + Kpshear ) (8)
W
Here K is a new model parameter that scales the effect of the
shear force when there is no slip. Sliding wear in a jaw crusher
has been found to be three to six times faster than squeezing-only Fig. 10. Measured geometry compared with a 3D-CAD model of mantle and
wear, at the same crushing load [10]. concave.
M. Lindqvist, C.M. Evertsson / Wear 261 (2006) 435442 439

Fig. 12. Simulated and measured amount of wear on the mantle of an H6800
EC liner set. The geometry was measured in the normal direction of the surface
after 385 h of operation.

was made with the flow model presented by Evertsson [1]. That
Fig. 11. Simulated geometry of a worn mantle profile at different times, using model is slightly different from the one used here. Fig. 12 shows
two different wear models. the wear on the mantle of a Sandvik, H6800 crusher, Fig. 13
shows the wear on a Sandvik H3000 MF concave. The mantle
shows the measured worn geometry, compared to a cross section and the concave have different local coordinate systems in the
of the nominal CAD-geometry. simulator, hence the difference in y-coordinate.

3.2. Simulation versus measurement of wear 3.3. Simulation versus measurement of operating
parameters
The worn liner profiles were computed using the crusher
model. Fig. 11 shows worn mantle profiles at different times, Hydroset pressure and power draw were read off the control
using the two different wear models. The left profile shows the panel of the crusher once every day. When the inlet bin of the
worn geometry obtained using the previous wear model that is crusher is entirely filled with rock material, the crusher is said
independent of shear forces. The right profile shows the worn to be choke fed, and this is the preferred way to operate a cone
geometry from the new shear-dependent wear model. There is crusher. Readings were taken during normal operation of the
an obvious difference between the two models in prediction of crusher, i.e. choke fed conditions. The feed was between 32 and
wear in the upper part of the chamber. The effect of non-sliding 250 mm and came from the primary crusher.
shear force is scaled so that simulations fit measured data. The
wear model parameter K in Eq. (8) was selected so that the wear
was correctly predicted at two points on the liner: where the
maximum wear occurs, near the bottom of the mantle, and on
one point located near the top of the liner, one-third of the cham-
ber height from the top. K = 50 gives the best agreement. A shear
wear factor of 50 may seem high, but the shear force factor f in
Eq. (5) is small, since the angle between the liners is small.
Fig. 12 shows the measured and simulated wear on an H6800
mantle. The wear is computed as the difference between nominal
new and worn geometry, measured in the normal direction of
the surface. As can be seen in Fig. 12, the new wear model
significantly improves the wear prediction in the upper part of the
crushing chamber compared to the old model. The flow model
used here was presented by Lindqvist [6].
Fig. 13 shows simulated and measured wear on the concave
of a worn SANDVIK H3000 MF chamber. The measurement
in Fig. 13 was made by Lindqvist and Evertsson [3]. Highly Fig. 13. Simulated and measured wear on a concave of a SANDVIK H3000 MF
abrasive quartzite was crushed. The simulation in reference [3] chamber. Measurements were made by Lindqvist and Evertsson [3].
440 M. Lindqvist, C.M. Evertsson / Wear 261 (2006) 435442

Fig. 14. Correlation between power draw and hydroset pressure.

Fig. 15. Power draw, simulation and measurement.

The wear model is indifferent to how time is scaled, and the


wear rate is exaggerated in the simulations, to save computa- there is an inaccuracy of less than 8 h as for when each read-
tion time. The wear was accelerated by a factor of 4700 times, ing was made. Simulated time has here been expressed as dates.
as compared to the wear rate found by Lindqvist and Everts- Simulated time corresponds to the time it takes for the model
son [3]. If the wear rate is accelerated too much, the simulated to produce the same amount of maximum wear on the mantle
worn geometry will deteriorate as compared to the measured as is measured. In other words, maximum simulated wear cor-
geometry. responds directly to simulated time. The maximum wear on the
Fig. 14 shows the correlation between power draw and mantle was 48 mm in the last measurement.
hydroset pressure. The model for flow and crushing pressure The measured capacity of the crusher was computed as the
require a validation of some model parameters [7]. In the sim- daily output divided by the number of hours the crusher was run
ulations made here, the model parameters were selected so that that day. The crusher is a secondary crusher. When designing a
power draw and hydroset pressure were predicted as accurately crushing plant the components are generally over-dimensioned
as possible with respect to average measured data. Power draw and the plant bottleneck is placed late in the process. This means
and hydroset pressure cannot be predicted accurately without that the rest of the plant, at times will be under-utilised, and the
taking losses into account. Losses in a cone crusher arise mainly crusher may not be choke fed. The measured capacity hence
in the electric motor, the belt drive, the roller bearings support- fluctuates. This is the reason why the capacity is generally over
ing the driveshaft. Frictional losses occur in the top bearing, the predicted.
eccentric bushings and the spherical thrust bearing who are all During some shifts, the plant was set to crush 090 material
boundary lubricated plain bearings. According to the machine in the secondary crusher, and during those shifts readings of
manufacturer, this particular crusher usually has an idle power power and hydroset pressure were not taken. Those shifts are
draw of 3035 kW. The mass of the main shaft corresponds to
a hydraulic pressure of 0.28 MPa. To adjust for losses, load
dependent and load independent losses were simply added to
the nominal data to make simulations match measured data. A
constant load independent loss of 35 kW was added to the power
draw and the load dependent loss was computed by dividing the
nominal power draw by the total efficiency. The efficiency used
here was 59%. If the losses are subdivided onto electric motor,
belt drive, driveshaft, bevel gear and eccentric bushing, the aver-
age efficiency of each of these power-transmitting components
will be about 90%. The two model parameters for an H3000 MC
chamber that were found by Lindqvist [6] were K1 = 0.312 and
K2 = 1.01. For this crusher, which is much larger, K1 = 0.3590
and K2 = 1.2387.
Readings of power draw and hydroset pressure were taken
during normal choke fed conditions once every day (see
Figs. 15 and 16). The time of these readings were only specified
by date. The number of hours per day each crusher was in oper-
ation was recorded, and was below 8 h every day. This means Fig. 16. Hydroset pressure, simulation and measurement.
M. Lindqvist, C.M. Evertsson / Wear 261 (2006) 435442 441

not representative for the most common use of the machine, cannot be explained by wear; consider for example readings dur-
which is set to crush 090 mm material less than 10% of the ing August when power draw and hydroset pressure is higher
time. than during the rest of the period. No corresponding trend in
capacity can be seen. The likely reason is that the properties of
4. Discussion the rock have changed during this time. The rock is blasted and
hauled to the crushing plant from different locations in the pit,
The aim of this study was to improve the crusher model by and rock properties generally differ between different locations.
adding a shear force dependent component in the wear model. As The crusher was not run choke fed at all times. This explains
mentioned, in a previous study, there was a discrepancy between the under-prediction of capacity (see Fig. 17). During less
simulation and measurement in the upper part of the chamber [4]. than 10% of the crushers operating time, it was set to crush
In the study by Lindqvist [6], an enhanced flow model was pre- 090 mm material, which means that capacity, power draw and
sented and the prediction of power draw, capacity and hydroset hydroset pressure all increase. Readings of hydroset pressure
pressure improved considerably. There was a slight improve- and power draw were not taken for those operating condi-
ment in prediction of wear for a fine crushing chamber. When tions. The wear may change the geometry in a different way
running simulations with the new flow model on a coarse cham- during that time. These periods of different operating con-
ber, it became apparent that the under-prediction in the upper part ditions, less than 10% of the time, were neglected in the
of the chamber was even worse (see Fig. 11). The disagreement simulation.
between the model and measurement was thus more pronounced
for the coarse chamber, where the nip angle between the liners is 5. Conclusions and future work
larger. The conclusion is that the inaccuracy in the previous flow
model was not what caused the discrepancy, even though there By adding a shear force dependent factor in the wear model
was an improvement in prediction of other operating parameters. for cone crushers, the agreement between simulated and mea-
Another proposed mechanism that could explain the discrep- sured worn geometry was significantly improved. The prediction
ancy is that wear is dependent on particle size and number of of operating parameters hydroset pressure and power draw was
contact points that occur. Furthermore, the rotation of the mantle satisfactory, but measured power draw and hydroset pressure
deviates from ideal rolling against the rock in the upper part of fluctuated. As was mentioned in previous sections, there are
the crushing chamber (see Fig. 5). None of these effects how- other possible explanations for the addressed model discrepancy,
ever, are likely to be more pronounced in a coarse chamber than for example, the particle size distribution or non-linear relation-
in a fine crushing chamber, since the mantle does not differ ships between wear rate and pressure. The crusher model used
much between the chamber types. The shape of the concave in this study can be described as a grey-box model. Introduc-
however, and the nip angle between the liners does indeed differ ing more model parameters to describe more phenomena might
between the chamber types (see Fig. 6). Therefore this issue was eventually make the problem of finding optimal model param-
addressed first. eters poorly conditioned. Even though the model presented
An important question is whether the model parameters here, has successfully solved the addressed problem, further
described by Lindqvist and Evertsson [6], remain valid as lin- work is necessary to fully understand the importance of other
ers are worn. The measured power draw and hydroset pressure variables.
fluctuates considerably (see Figs. 15 and 16). These fluctuations
Acknowledgements

The measurements were carried out at the NCC Ramnaslatt


Quarry in Southern Sweden. Thomas Gustafsson and the other
staff, are gratefully acknowledged. SANDVIK Rock Processing
AB is gratefully acknowledged for providing funding, informa-
tion and support.

References

[1] C.M. Evertsson, Cone Crusher Performance, Ph.D. Thesis, Dept. of


Machine and Vehicle Design, Chalmers University of Technology, Swe-
den, 2000.
[2] K. Gauldie, The output of gyratory crushers, Engineering (October)
(1953) 456458, October 16, 1953, 485486.
[3] M. Lindqvist, C.M. Evertsson, Prediction of worn geometry in cone
crushers, Miner. Eng. 16 (12) (2003) 13551361.
[4] C.M. Evertsson, M. Lindqvist, Power draw in cone crushers, in: Presen-
tation at the Minerals Engineering Conference, September 2527, 2002,
Perth, Australia, 2002.
[5] M. Lindqvist, P. Sotkovski, Work hardening in cone crusher liners,
Fig. 17. Simulated and measured capacity. in: Fourth International Conference on Computer Applications in the
442 M. Lindqvist, C.M. Evertsson / Wear 261 (2006) 435442

Minerals Industries, September 810, 2003, Calgary, Alta., Canada, [8] P.P. Rosario, R.A. Hall, D.M. Maijer, Liner wear and perfor-
2003. mance investigation of primary gyratory crushers, Miner. Eng. 17
[6] M. Lindqvist, C.M. Evertsson, Improved flow- and pressure model for (NovemberDecember (1112)) (2004) 12411254.
cone crushers, Miner. Eng. 17 (NovemberDecember (1112)) (2004) [9] J.L. Archard, Contact and rubbing of flat surfaces, J. Appl. Phys. 24 (8)
12171225. (1953) 981988.
[7] T. Chenje, P. Radziszewski, Determining the steel media abrasive [10] M. Lindqvist, C.M. Evertsson, Liner wear in jaw crushers, Miner. Eng.
wear as a function of applied force and friction, Miner. Eng. 17 16 (1) (2003) 112.
(NovemberDecember (1112)) (2004) 12551258.

You might also like