关于破碎机衬板磨损的实验

Liner wear in jaw crushers

M.Lindqvist *,C.M.Evertsson

Machine and Vehicle Systems,Chalmers University of Technology,Goteburg 41296,Sweden

Received 15June 2002;accepted 23August 2002

Abstract

Wear in rock crushers causes great costs in the mining and aggregates industry.Change of the geometry of the crusher liners is a major reason for these costs.Being able to predict the geometry of a worn crusher will help designing the crusher liners for improved performance.

A model for prediction of sliding wear was suggested by Archard in 1953.Tests have been conducted to determine the wear coe?cient in Archard ?s https://www.360docs.net/doc/829807765.html,ing a small jaw crusher,the wear of the crusher liners has been studied for di?erent settings of the crusher.The experiments have been carried out using quartzite,known for being very abrasive.Crushing forces have been measured,and the motion of the crusher has been tracked along with the wear on the crusher liners.The test results show that the wear mechanisms are di?erent for the ?xed and moving liner.If there were no relative sliding distance between rock and liner,Archard ?s model would yield no wear.This is not true for rock crushing applications where wear is observed even though there is no mac-roscopic sliding between the rock material and the liners.For this reason,Archard ?s model has been modi?ed to account for the wear induced by the local sliding of particles being crushed.The predicted worn geometry is similar to the real crusher.

A cone crusher is a machine commonly used in the mining and aggregates industry.In a cone crusher,the geometry of the crushing chamber is crucial for performance.The objective of this work,where wear was studied in a jaw crusher,is to implement a model to predict the geometry of a worn cone crusher.ó2002Elsevier Science Ltd.All rights reserved.

Keywords:Comminution;Crushing;Particle size;Modelling

1.Introduction

Jaw and cone crushers are commonly used in the mining and aggregates industry.Today,it is possible to predict the performance of a cone crusher,provided the geometry,crusher settings and the characteristics of the material fed into the crusher are known (Evertsson,2000).The geometry of the crusher will change because of wear.Being able to predict the worn geometry will help optimising the design of the new crusher for im-proved performance throughout its lifetime.

Current research and knowledge in the ?eld of wear is very extensive.However,much of this research is con-ducted from a material science perspective on a micro-scopic level.The objective is often to gain knowledge in material selection situations,heat and surface treatment and so forth.In this work,wear on a macroscopic level

has been studied,aiming to understand and prevent the damaging e?ect of the inevitable wear in cone crushers.1.1.Rock crushers

The operating principles of cone and jaw crushers are described in Fig.1.

In a cone crusher,the shaft of the inner mantle is suspended in two bearings,one eccentric at the bottom and one concentric at the top.When the eccentric at the bottom is turned,a vertical cross section of the inner mantle will su?er an oscillating motion.The rock par-ticles between the inner and concave will be squeezed,crushed,and fall down.When the material passes through the crusher,it will be subjected to several crushing actions.The properties and performance of the crusher are strongly dependent on the stroke and bed thickness.Because of wear,these geometric quantities will change during the life of the liners.In turn,when the geometry of the liners changes,the performance of the crusher will change.With very few exceptions,this will be a detrimental e?ect.

*

Corresponding author.

E-mail addresses:mats.lindqvist@me.chalmers.se (M.Lindqvist),magnus.evertsson@me.chalmers.se (C.M.Evertsson).

0892-6875/03/$-see front matter ó2002Elsevier Science Ltd.All rights reserved.PII:S 0892-6875(02)00179-

6

Minerals Engineering 16(2003)

1–12

In order to design a crushing chamber it is desirable to be able to predict the geometry of the worn chamber.The objective of this study is to develop a model for this purpose.With such a model,it will be possible to per-form simulations in order to design crusher chambers that are less sensitive to wear.1.2.Wear mechanisms

The research and literature on abrasive wear and wear mechanisms is very extensive.A lot of di?erent wear situations have been described,but generally four types of fracture are described as being present in abrasive wear:fatigue,shearing of junctions,microcut-ting and impact (Vingsbo,1979).There are also sec-ondary e?ects such as frictional heating and corrosion that a?ect the material/wear mechanism.Much of the research is done on a microscopic level.In this work,no further attention will be paid to such topics.

A model for predicting material removal due to wear was suggested by Archard (1953).In this model,it is assumed that wear is proportional to pressure and sliding distance.Hence,in order to use Archard ?s wear model,the local pressure and motion the crusher need to be known in.One di?culty in determining the wear resistance coe?cient in Archard ?s model is the fact that the abrasive particles are crushed during wear.Some results have been achieved by (Yao and Page,2001;Yao et al.(2000)),who have studied wear during crushing of silica sand.They have studied surface damage on a microscopic level after a single crushing event.To obtain the wear resistance coe?cients needed in Archard ?s model,it is necessary to make the measurements after a repeated number of crushing events.Their research in-dicates,however,that a testing device for determining

the relationship between wear,pressure and motion will need to resemble the process in a real crusher,where rock material is crushed,mixed and crushed again.After a large number of repeated crushing events,the worn geometry will be measured.In order to predict the worn geometry the components in Archard ?s wear model,the pressure and relative motion must also be known.1.3.Test equipment

A few schematic testing devices for determining wear coe?cients were suggested by Hutchings (1992).For crushing applications,methods such as the ‘‘pin on disc’’test have the drawback that the abrasive properties will change during crushing.This has also been proven by Yao et al.(2000),who have found that by appropriate control of pressure and shear force,a protective layer of material can be formed near the surface.This means that a testing device for determining wear coe?cients needs to resemble the conditions in a real crusher,where material is crushed,mixed and crushed again.It is also desirable to measure the crushing forces during crush-ing,to verify the pressure.

2.Experiments

A small jaw crusher originally designed for testing the properties of rock material was modi?ed (see Fig.2).Rock material is fed into the top of the crusher.The rotating eccentric shaft and the link give the right liner an oscillating motion that will crush the material.The left,stationary liner has three load cells.Two of them are horizontal,to measure normal forces,and one ver-tical to measure the frictional force.The force cells

have

Fig.1.Cone crusher (left)and jaw crusher in operation.Bed thickness and stroke are important properties for the crusher.

2M.Lind qvist,C.M.Evertsson /Minerals Engineering 16(2003)1–12

been designed to be insensitive to bending and torsion.They only register compressive/tensile forces.The liner material is manganese alloy steel,commonly used for crusher liners,with 1.2%C,12.5%Mn,0.6%Si and 1.5%Cr.Austenitic manganese steels are common in abrasive wear applications;they are well known for their excellent capacity for work hardening.Upon plastic deformation the austenite in this material transforms into martensite and becomes harder.There are various explanations for this strain-induced hardening;but the major mechanisms that drive the transformation are twinning and slipping of dislocations (El Bitar and El Banna,2000).

The small jaw crusher was used to study the wear as a function of force (i.e.pressure)and motion.The ex-periments were carried out using quartzite,since this material is known for being very abrasive.The size distribution of the feed material was 8–11mm.The closed side setting (the minimum distance across the crushing chamber)was set to 2,3and 5mm.

n com-pressive crushing,two modes of breakage were identi?ed by Evertsson (1998):inter-particle breakage and single particle breakage.The reason for selecting size distri-bution and crusher setting in this study was to ensure inter-particle breakage.Inter-particle breakage occurs when the size of the particles is smaller than the bed height i.e.particles will be crushed against each other;as opposed to single particle breakage when a single par-ticle is crushed between the two steel liners.

The forces were registered and the motion of the crusher was measured by recording a signal indicating

the eccentric angle of the main shaft.Wear of the liners was measured on a 13by 8grid with 10mm spacing on each liner.The average wear on each level in the downward direction was computed.

3.Modelling 3.1.Modelling ?ow

In order to predict the wear,the pressure and motion of the crusher need to be determined.Previous work has been carried out by Evertsson (1995,1998,1999)on ?ow and capacity modelling of cone crushers.A particle ?ow model has been implemented for the jaw crusher.In the particle ?ow model,three types of motion are de?ned:free fall,sliding and squeeze.Impact is modelled plas-tically,which means that the normal component of the velocity is annihilated and the tangential component is preserved upon impact (Evertsson,1999).

Fig.3shows the path of a particle through the crusher.This particle ?ow model overestimates the ca-pacity of the crusher.Due to dynamic ?lling e?ects,the actual capacity will be reduced in comparison with the particle model (Evertsson,1999).3.2.Modelling pressure

By crushing a bed of rock material in a cylindrical container (see Fig.4)it is possible to relate average pressure p to the compression ratio es =b T.Rock is

Fig.2.Small jaw crusher with three load cells.

M.Lind qvist,C.M.Evertsson /Minerals Engineering 16(2003)1–123

crushed in a cylindrical container.By measuring force along with compressiones=bTit becomes possible to?t a polynomial to the measured data.This function is de-pendent on grain size distribution.This work has been done elsewhere(Evertsson,2002).

The nominal compression ratioes=bT

nom is deter-

mined by the geometry of the crusher(horizontal stroke divided by horizontal chamber size).Due to dynamic

e?ects,the e?ective compression ratioes=bT

eff is less than

the nominal.In addition,?lling e?ects further reduces the e?ective compression.The compression ratio utilized

is denotedes=bT

u :es=bT

nom

>es=bT

eff

>es=bT

u

(Everts-

son,1999).By computing the e?ective compression from the particle?ow model and using the correlation be-tween pressure and compression obtained in the can test, it is possible to compute the pressure on the crusher liners(see Fig.5).The pressure is also dependent on size distribution.In these simulations,however,an average dependency betweenes=bTand pressure has been used, since the change in size distribution before and after the crusher is not signi?cant in terms of a?ecting the pres-sure.

3.3.Modelling forces

Using the simulated pressure distribution,it is pos-sible to compute the forces corresponding to the mea-sured forces on the test crusher.It is assumed here that the normal force on the moving liner is equal to the normal force on the stationary liner.In Fig.6,a prin-ciple image of the crusher is shown.

Equilibrium around point A in Fig.6yields the fol-lowing equations:

F2?

1

d

c

Z

py d y

àF3h

e1TF3?C

Z

l p d ye2TF1?F2àc

Z

p d ye3Twhere d is the distance between the horizontal force cells,c is the width of the plate and h is the distance between the liner surface and the vertical force cell.This calculation is performed for each time step.

3.4.Modelling motion

The geometry and motion of the crusher can easily be computed.During each time step,every point on the moving liner in squeeze will have a motion relative to the bed of rock material(see Fig.7).

It is possible to compute the sliding and squeezing distance from the velocities in the tangential and normal direction,as shown in Fig.

7.

Fig.4.The can test.

4M.Lind qvist,C.M.Evertsson/Minerals Engineering16(2003)1–12

V ?Z v t d t e4TS ?

Z

v n d t

e5T

3.5.Modelling wear

The liners will be worn during crushing,even if there is no sliding motion between the liner and bed of ma-terial.Archard ?s wear law will not yield any wear for this case.For this reason,Archard ?s model needs to be modi?ed to account for the local sliding that occurs when particles are crushed (see Eqs.(6)and (7)):w is the

wear,p pressure,V sliding distance.The coe?cients W 1and W 2are wear resistance coe?cients and will be de-termined by experiment.Archard ?s original wear model only included the ?rst term pv =W 1.The term p =W 2ac-count for the local sliding of particles during crushing.This component of the wear is thus considered as an event independent of time or velocity;only pressure causes this squeezing wear.During one stroke,the wear will be

D w ?1W 1Z t 0pv d t tp W 2e6T

For N strokes the total wear will be computed

as

w ?N D w e7T

This means wear resistance coe?cients W 1and W 2can be computed from the wear w observed in experiments.

4.Results

4.1.Force measurements

The signals from the three load cells shown in Fig.3were recorded when the crusher was in operation.The main eccentric shaft was run clockwise as viewed in Fig.3.During squeezing,the moving liner moves upward,which explains why the vertical force is negative (tensile)(see Fig.8).As expected,the lower force cell is subjected to the highest force,since the compression and,conse-quently,the pressure,is higher at the bottom of the crusher.

A signal indicating the eccentric angle was recorded together with the three forces.In this way,it is possible to identify a unique starting point for each crushing cycle.By adding the signals from a large number of cycles it is possible to compute the average force during one cycle (see Fig.9).

It can be seen in Fig.9that the maximum force does not occur in the two horizontal force cells simulta-neously.The reason for this is that the moving liner starts to close earlier at the top than it does at the bottom.The coe?cient of friction can be estimated by

dividing the vertical frictional force by the sum of the horizontal forces.It is not meaningful to compute the coe?cient of friction for times where crushing forces are close to zero.Coe?cient of friction l is about 0.3.A similar test (Yao et al.,2000)found that l is about 0.4–0.5when silica sand is being crushed.The result is shown in Fig.10.

The e?ective compression ratio es =b Teff was used to compute the pressure and the resulting forces are shown in Fig.11.

Apparently this procedure greatly overestimates the compression and pressure.The simulated force was found to be approximately 10times higher than the actual measured force.

According to Evertsson (1999),the utilized com-pression ratio,es =b Tu ,is smaller than the e?ective

ratio

6M.Lind qvist,C.M.Evertsson /Minerals Engineering 16(2003)1–12

due to the fact that at the beginning of the stroke,particles are not con?ned.A rearrangement of particles is necessary before any crushing can take place.This reduces compression signi?cantly compared to the nominal or e?ective compression ratio.A volumetric ?lling ratio g v <1was introduced to account for this ?lling e?ect.s b eff ?S eff

b eff

e8Ts b

u

?

S eff àe1àg v Tb eff

b eff g v

e9T

Before any crushing can take place,particles will be rearranged,and only a portion s utilized of the e?ective stroke s eff is utilized (see Fig.12and Eqs.(8)and (9)).Insu?cient ?lling has a dramatic impact on the utilized compression ratio.For example at the top of the crusher,s eff ?7mm,b eff ?55mm.Then es =b Teff ?0:12.Assuming that g v ?0:9yields es =b Tu ?0:03,in other words,a 10%reduction of ?lling reduces compression by 75%.

It is di?cult to accurately model or simulate the utilized compression since it is so sensitive to changes in ?lling ratio.It is necessary to measure the forces by some means in order to verify the pressure distribution.Reducing compression es =b Teff by a factor of 0.3gives a better correlation between measurements and https://www.360docs.net/doc/829807765.html,ing this value and Eqs.(8)and (9)yields a volumetric ?lling ratio of 0.9.This is a typical value for

cone

M.Lind qvist,C.M.Evertsson /Minerals Engineering 16(2003)1–127

crushers and this reduction in compression,gives a better correlation for all settings of the crusher(see Fig.

13).Losses due to wall friction are included in this factor.In a similar test,wall friction was about7%of applied pressure(Yao et al.,2000).

4.2.Wear results

During the test,the crusher was run under choke feed conditions.The wear of both liners,stationary and moving,was measured.The measurements were made on a13by8grid with10mm spacing with10min in-tervals and the average wear on the di?erent levels was computed.

It can be seen from Fig.14that in the upper section of the crusher,the wear is very small or material is even added.This is because the radius of the measuring probe is much bigger than the size of the surface asperities formed during the test;the probe detects the peaks of the craters.For this reason,the wear data from this region is not suitable for determining the wear resistance

coe?cients in Archard?s model.The last20mm of the liner the wear pro?le also deviates from what can be expected from the simulated pressure.Since the particles near the bottom are not properly con?ned,the pressure will be reduced here;rock material shatters during compression and falls out of the crusher.This is also understood by looking at the forces in Fig.13.The peak force on the lower force cell of the crusher is slightly overestimated and the force on the upper cell is under-estimated.Wear data from this region should also be ignored when calculating wear coe?cients.

The wear mechanisms are di?erent for the?xed and the moving liner.The graph in Fig.15shows the dif-ference.The wear rate is4–9times higher on the sta-tionary liner compared to the moving one.If the worn surfaces are studied,one can clearly see the

ploughing

grooves on the ?xed liner,but on the moving liner there are no such grooves (see Figs.16and 17).

This di?erence in relative motion between the bed and the two liners is governed by the angle of the liner,and the direction of motion.This is understood if the bed of rock material is regarded as a wedge (see Fig.18).

Equilibrium for the wedge of rock material will re-quire

!:N 1sin a tF 1cos a àF 2?0":N 2àN 1cos a tF 1sin a ?0

The ratio between normal forces N 1,N 2and frictional forces F 1,F 2is denoted f 1and f 2respectively.If the friction is fully developed at the left surface (f 1?l )then the coe?cient of friction at the right surface will be

f 2?

tan a tl 1àl tan a which is greater than the coe?cient of friction l ,and this is not possible.On the other hand,if the friction is

fully developed at the right surface (f 2?l )then the utilized friction at the left surface will be f 1?

l àtan a 1tl tan a

and this is less than the coe?cient of friction.This means that slip will begin at the right surface.

The length of the longest observable ploughing groove on the ?xed liner is about 5mm.The total sliding length during squeeze is about 14mm.This indicates that the pressure is high enough to cause ploughing grooves during roughly a third of the stroke.Because of the di?erence in wear mechanisms,two di?erent models are used depending on whether there is or is not any relative sliding motion between steel surface and bed.Eq.(10)is the model used when there is both sliding and squeezing wear,and Eq.(11)is used for pure squeezing wear.

D w ?1W 1Z t 0pv d t tp W 2

e10

T

Fig.16.Photo of the surface of the moving liner.Indentation marks are clearly observed whilst no ploughing grooves can be

seen.

Fig.17.Stationary liner with ploughing

grooves.

Fig.18.Wedge of rock material between two liners.

M.Lind qvist,C.M.Evertsson /Minerals Engineering 16(2003)1–129

D w ?

p W 2

e11T

In the model used here,sliding distance V and pres-sure are computed for each time increment.Wear re-sistance coe?cients W 1and W 2can be computed by applying the method of least squares.When these co-e?cients are obtained it is possible to compute the simulated worn geometry.The graphs in Figs.19and 20shows measured worn pro?le and corresponding simu-lated pro?le.

Note that the order of magnitude of the wear on the moving liner (pure squeezing wear)is much smaller than for the ?xed liner (sliding wear).The wear coe?cients obtained are (note that the units for sliding and squeezing wear are di?erent):

W 1?208kN =mm 2for sliding wear :W 2?274kN =mm 3for squeezing wear :

In Fig.21,maximum wear on each plate and for each test is plotted against time.The wear is proportional to time (i.e.sliding distance or number of strokes).This is in agreement with Archard ?s wear model.In Fig.20,it can be seen that for the moving liner on the upper sec-tion,it appears as if material has been added,which of course is not possible.The explanation for this is that the size of the surface asperities is of the same order of magnitude as the wear itself.After 30min of crushing at CSS 2mm,the maximum surface deviation R z is 92l m on the ?xed liner at the location of maximum wear.

5.Discussion

According to P ~o

dra,1997,the predicted life can de-viate by more than 90%when the linear wear equation is used.This is in agreement with Yao et al.(2000),who found a change in wear mechanism as pressure increased during crushing of silica sand.Hutchings (1992),made a distinction between wear caused by brittle fracture and plastic deformation.Based on simple theory,Hutchings suggests a linear wear model identical to Archard ?s for plastic deformation wear.For wear caused by brittle fracture,several non-linear models are suggested.In the non-linear methods quoted by Hutchings,wear is pro-portional to pressure raised to an exponent of between 1.125and 1.5.The pressure used in the wear prediction here has been calibrated by measuring the forces that occur during crushing.It has been possible to calculate the average pressure on the crusher liners from the measured forces and the geometry and motion of the crusher.

The feed material,quartzite of size distribution 8–11mm,was selected since it is known for being

very

10M.Lind qvist,C.M.Evertsson /Minerals Engineering 16(2003)1–12

abrasive and to ensure inter-particle breakage.The liner material,manganese steel,is well known for its excellent capability of work hardening.This e?ect is generally not observed when crushing quartzite,however.It may be necessary to take the e?ect of work hardening into ac-count when crushing other materials than quartzite.The in?uence of particle size distribution on wear rate may need to be investigated further.

Figs.19and20shows a similarity between simulated and measured geometry.The crushing chamber in the jaw crusher used in these experiments is120mm high. The boundary e?ects(poor con?nement of particle bed near top and bottom of crusher)will be less signi?cant in a cone crusher,since the crushing chamber is much higher compared to the bed thickness and stroke.An-other circumstance that may be bene?cial is that the motion of the cone crusher in a vertical cross section is synchronous.In a certain vertical plane,the closing and opening takes place simultaneously everywhere in a cone crusher.This is not the case for the jaw crusher,which opens at the top while closing take place at the bottom. Hence,discrepancies between modelled and measured wear at the top and bottom of the crusher,can be ex-pected to be smaller in a real crusher.The objective of this study was to predict the geometry of the worn crusher regardless of how long it takes.It is not so im-portant to be able to predict the wear rate.It is more important to predict how the geometry changes in the worn crusher,regardless of time.

6.Conclusions

The?ow model is not accurate enough to predict pressure in the jaw crusher without additional mea-surements.The relationship between wear and pressure has proven to be reasonably good(see Figs.5,19,20). The pressure distribution is similar to the wear distri-bution with the exception of the lower part of the crusher.Fig.21shows that wear is proportional to time. This is in agreement with Archard?s wear equation.This is promising since there are reasons to assume that wear prediction will be more accurate in a cone crusher.The next step will be to implement the wear model for the cone crusher(Figs.22and23).

Acknowledgements

The author would like to thank the following persons for support,advice and valuable discussions:Professor G€o ran Gerbert,Jan M€o ller and Mats Persson at Chal-mers University of Technology,Richard Bern,Torsten A hman,Anders Nilsson and Per Olsson-Artberger at

Sandvik Rock Processing,Claes L€o wgren at SSAB,Per Jonning at

NCC.

References

Archard,J.L.,1953.Contact and rubbing of?at surfaces.Journal of Applied Physics24(8),981–988.

El Bitar,T.A.,El Banna, E.M.,2000.Improvement of austenitic had?eld Mn-steel properties by thermo mechanical processing.

Canadian Metallurgical Quarterly39(3),361–368.

Evertsson,C.M.,1995.Prediction of size distributions from compress-ing crusher machines.In:Proceedings EXPLO95Conference, Brisbane Australia,pp.173–180.

Evertsson,C.M.,1998.Output prediction of cone crushers.Minerals Engineering11,215–232.

Evertsson,C.M.,1999.Modelling of?ow in cone crushers.Minerals Engineering12,1479–1499.

Evertsson, C.M.,2000.Cone Crusher Performance,Ph.D.thesis, Department of Machine and Vehicle Design,Chalmers University of Technology,Sweden.Evertsson, C.M.,Lindqvist,M.,2002.Power draw and pressure distribution in cone crushers.Accepted for presentation on Minerals Engineering-02Perth Australia.

Hutchings,I.M.,1992Tribology,friction and wear of engineering materials.Butterworth Heinemann,ISBN0-340-56184X.

P~o dra,P.,1997.FE Wear simulation of sliding contacts Ph.D.

dissertation,Royal Institute of Technology,Stockholm, Sweden.

Vingsbo,O.,1979.Wear and wear mechanisms.Reprint from Wear of materials1979,pp.620–635,ASME.

Yao,M.,Page,N.W.,2001.Friction measurement on Ni-Hard 4during high pressure crushing of silica.Wear249,117–126.

Yao,M.,Page,N.W.,Keys,S.,McMillan,W.,Cenna,A.,2000.A high pressure shear cell for friction and abrasion measurements.

Wear241,186–192.

12M.Lind qvist,C.M.Evertsson/Minerals Engineering16(2003)1–12