You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/43498056

A structured approach to modelling SAG mill liner wear - monitoring wear

Article · January 2011


Source: OAI

CITATIONS READS

2 589

2 authors:

Malcolm S. Powell Rajiv Chandramohan


The University of Queensland Ausenco
148 PUBLICATIONS   1,338 CITATIONS    7 PUBLICATIONS   30 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Malcolm S. Powell on 24 February 2017.

The user has requested enhancement of the downloaded file.


III-133 III-134

falls under maintenance engineering departments. The main objective


of these departments is to prolong the use of these liners. Prolonging
SAG the life became more important as the mineral prices and grades fell,
necessitating lower operational costs and reducing the downtime.
2006 Additionally, relining became a critical factor in maintaining high
throughputs of single line operations.
DEPARTMENT OF MINING ENGINEERING
UNIVERSITY OF BRITISH COLUMBIA The secondary use of liners is to transfer the rotary motion of the mill to
Vancouver, B. C., Canada the media. The rotary action and profile of the charge is directly
attributed to the lifter. Extensive studies of charge motion as influenced
by the lifters were conducted by McIvor (1983). He looked at the
A STRUCTURED APPROACH TO MODELLING SAG MILL LINER influence of lifter face angle and its effect on the projection of the
WEAR – MONITORING WEAR particle for given rotational velocity. Following this work, Powell (1991b)
extended McIvor’s model to include the effect of the number of lifters
and lifter height. The MillTraj software (2000) is based on the work of
Rajiv Chandramohan & Malcolm S Powell Powell (1991b) and is used extensively by liner and mill manufacturers.

Mineral Processing Research Unit, University of Cape Town, The progressive wear of mill liners was rarely considered as a variable
Rondebosch, 7700, South Africa. in design because the liner suppliers and manufacturers focused their
Corresponding author: rajiv@chemeng.uct.ac.za. attention on material composition. Because each milling condition is
different due to variation in ore type, the wear rates of liners varies
ABSTRACT considerably.

A Liner wear project has been established to monitor the progressive Factors that promote excessive wear, such as impacting on the liner,
wear of liners and to use this information to guide liner selection and incorrect ball loading and, dilute slurry, also affect the throughput of
design. Two measuring devices, a basic mechanical liner profile gauge, mills, Powell et al (2006). Thus monitoring the wear of mill liners
and a Laser operated electronic device which has improved compliments the optimisation of mill performance. The aim of this
measurement time and accuracy are presented. This data can also be project as part of ongoing work in liner design and selection is to
used calibrate and validate DEM wear models that will be used as a tool monitor the wear rate using simple measurement techniques.
to predict the wear of liners.
Current techniques for measuring the wear of mill liners
INTRODUCTION A range of techniques are used to monitor the wear of mill liners.
Semi Autogenous and Autogenous Grinding mills are an integral part of
the modern mineral processing plant. An understanding is steadily Visual inspection:
developing of mechanisms which drive the functionality of the mills. The most basic technique is to visually inspect the liner when the mill is
stopped for maintenance. This is based on experience in judging the
The mills should be designed according to the ore characteristics, but wear state of the liners, so is intrinsically qualitative and subject to error.
this is applied primarily in the sizing rather than in any manner that
‘Tunes’ a mill for specific ore. One of the main components in a mill that
is often overlooked is the liners / lifters. This is compounded if different people conduct the assessments and if
inexperienced staff are used. Although this method is not the best
The primary use of the liners in a mill is protection against the harsh technique, it is used by many mill operators throughout the mining
internal conditions. Because tumbling rock is highly abrasive, the liners industry to judge when a liner should be changed out.
take the brunt of the impact and abrasion during tumbling. It was
reported by Powell et al (2006) that the primary responsibility for liners
III-135 III-136

Profile Gauge: PERI Associates (www.processeng.com) have developed a laser based


gauge that has been through 2 major versions to the current one on
A quantitative technique is the physical measurement of the mill liners
offer, as shown in Figure 3. This utilises a mechanical drive to move a
using a profile gauge. It appears that the rubber liner manufacturers were
laser range meter between the vertical legs, and the vertical distance is
first to institute rigorous liner monitoring. Hammering a nail through the
logged electronically. It is considered that this is the best commercial
liner plate gives the plate thickness. Total lifter height or a profile of a few
device on offer.
points across the lifter bar was introduced by Skega in the 70’s.

Another form of profile measurement is a template of the new liner used


to measure the difference between that and the worn liner at a set of
predetermined intervals, as shown in Figure 1.

Figure 3 – PERI Laser profiler (www.processeng.com)

The University of Cape Town – Mineral Processing Research Unit has


developed a simple mechanical device which enables the user to
Figure 1 – Template profile measurement measure the profile of the liner, Powell (2006). The application of this
device is addressed in this paper. The profile measurement technique
is believed to be the best suited to accurately measure the wear and
Powell (1991a) used a custom built gauge to accurately monitor the liner monitor it progressively due to its ease of use, low cost and therefore
thickness at a number of points and the lifter bar height for liners with accessibility for regular use and operations.
removable lifters, shown in Figure 2. It was accurate to 1mm and
produced high quality wear data, but unfortunately not the lifter profiles.
Full liner profile:
Nordell of Conveyor Dynamics and Metso have developed a laser
profile scanner which can map the profiles of the interior of the mill. This
system provides a full image map of the mill interior including; charge
level and profiles of all visible liners. Scanning the interior takes
approximately 1 – 1.5 hours, depending on the required resolution.

Figure 2 – Liner thickness monitor (Powell (1991a))

Figure 4 – Nordell & Metso’s mill laser scanner


III-137 III-138

UCT LINER PROFILE GAUGES - DESIGN REQUIREMENTS AND


OBJECTIVES

The design of the liner gauge is based on the following design


objectives:

• Span between the liner joints: Profile of the liner should be taken
for the entire liner width.
• Time per profile: Because of production pressure, operators
don’t like to stop the mill, especially to measure the liner profiles.
Therefore, downtime should be less than 30 min, preferably Figure 5 – Measuring the thickness of the liner
20min.
• Reference thickness: The thickness of the liner plate must be Figure 6 illustrates the extremes cases for given similar profiles; CASE
measured at each edge. B is unsuitable and should be replaced immediately. When this critical
• Portability: The precision of the monitoring gauge is improved by measurement has been overlooked the operators are caught by
its repeatability. To facilitate the firm placement of the of the surprise when liners unexpectedly fail.
gauge, it should have a good central grip, be light enough to
comfortably hold in place with one hand, and allow the free hand
to align the profile pins or slide the laser head.
• Accuracy: the gauge should have accuracy better than 1mm in
height measurement.

Method of measuring the profiles

Measuring the mill liners should be conducted in a systematic procedure


in order to maintain consistency. To get a good set of profile data, the Figure 6 – Extreme cases for measuring the thickness
following vital measurements are required besides the liner profile.
Rubber liners: Nail is hammered through the rubber – measurement
Mill internal shell diameter: This information is usually gathered from the accurate, Figure 5A.
blue prints of the mill.
Steel liners: A hardened steel nail is hammered between liner joints,
Liner thickness: This is usually difficult to measure. Liner thickness is a Figure 5B. This is slow and difficult and readings can’t necessarily be
key piece of information for the monitoring the wear of liners and is the taken where desired due to peening closing up the gap and metal
basis of profile measurement. Figure 5 illustrates how the thickness of slivers wedging into the gap. Inaccuracies arise from the nail becoming
the liner is measured using a nail hammered through or between liner stuck or striking a steel obstruction in the groove. Listening to the
plates. The thickness of the liner is calculated as follows: where Tliner is resonance of the nail as it is struck can be a good indication of whether
the thickness of the liner and Trub is the thickness of the rubber backing, it has reached the shell. It is strongly advised that redundant
as illustrated in Figure 6: measurements are taken as a check. A solution to the problems with
steel liners is to install rubber wedges between the liner plates, which
Tliner = Lnail − H nail − Trub provides easy access to hammer the nail and measure the liner
thickness. Some liner manufacturers do supply wedges to prevent steel
The thickness of a new rubber backing is usually 6mm. To allow for packing between the liners, so as to reduce the risks from spread of
compression rubber backing thickness should be calibrated during the liners and to ease liner removal.
installation of new liners; otherwise a thickness of 4mm is used.
III-139 III-140

Marking off: Liner profile measurement should be conducted in a The y – component is the measurement of the length of profile rod, and
consistent manner where the same liner should be monitored over time. its source of error is the manual measurement of the protruding rod with
Marking the measured liner could either be done by counting the number a steel ruler, marked in 1mm segments as shown in Figure 8.
of rows to the measured liner from a reference point such as a mill hatch
bolt or the measured liner could be marked outside on the mill shell
(painting the bolt). When variations in thickness of 10mm are common
between liners, variation in data are unsuitable for wear prediction.

Lifter Face Angle: Lifter face angle is a critical feature which is constantly
changing. The function of the lifter is to propagate the motion of rotation
to the charge. As the lifter crashes into the rocks and balls at the toe of
the charge, and the then is subjected to material sliding over and off it, it
is subjected to high wear that results in dramatic changes to the lifter
profile over its life. As the lifter profile changes, the trajectory of the
cataracting charge concurrently changes. As part of monitoring the wear
of liners, measuring the change in the lifter angle is important. Figure 8 – Accuracy of measuring the mechanical profile gauge

Flexing of the rod is a problem when measuring the profile of the lifter.
LINER PROFILE GAUGES
This occurs when the user forces the rods onto the lifter face, thus
Mechanical Liner Profile Gauge providing false readings of the lifter height. Figure 9a illustrates the error
The design of the mechanical gauge enables the user to measure the caused by flexing of the rod and its effect on the profile length. The
profile of the liner by pushing on the fixed length rods as shown in Figure absolute errors ∆yf and ∆xf are a function of the liner face angle θ.
7. The two support legs on either side of the gauge are used as support Therefore as θ approaches 90°, ∆yf and ∆xf increases. It is noted that
of the gauge when mounting onto the liner plate and used as a reference they are both zero when θ is zero (flat surface of the liner). Flexing of
height when measuring the depth of each protruding profile rod. The the rod is also attributed to the diameter and length of the profile rod. In
gauge is custom made to fit the width of liner plate so that the user does order to maintain rigidity and reduce the total weight of the mechanical
not need to adjust the support leg placement in the gauge. gauge, the profile rods are manufactured from aluminium. A suitable
diameter of 6mm is selected for the profile rod and its length is custom
made accordingly to lifter height.

A larger diameter has a substantial effect on the measured profile of the


liner. As shown in Figure 9b, if the tips of the profiles were pointed, this
significantly reduces the error in the diameter offset ∆xd1. As shown in
the inserts in Figure 10, the profile rods have flat ends, and the graphs
were corrected for the diameter offset error.

Another factor which may hinder the accuracy of the Mechanical liner
gauge is the number of profile rods used for the measurement of the
liner profile. As shown in Figure 9c, decreasing the number rods results
in poor profiles of the liner. The mechanical gauge developed at the
Figure 7 – Mechanical Liner Profile Gauge University of Cape Town, uses 10mm rod spacing for a 600mm span.
Because the rod spacing has a significant effect for steep lifter profiles,
The accuracy for the mechanical gauge is strongly influenced by the the mechanical gauge is custom made to match liner profiles with steep
measurement of the length of the profile rods. As shown in Figure 8, the lifters. For such cases, a combination of 20mm gaps for horizontal
x – component has a fixed length and its source of error is accuracy in sections of the liner and 10mm gaps for steep profile sections are
aligning the holes for the profile rods. This is to within 0.1mm. configured.
III-141 III-142

Profile example and testing


Profile testing was conducted on a mock-up liner profile. Figure 10
illustrate profiles for 10mm and 20mm rod profiles.

As shown in the Figure 10, the comparison between actual and


measured profiles using 10mm rod spacing is good. For 20mm rod
spacing the results show inaccurate profiles around the corners and
edges for both trapezoid and the semi circular shapes. For a vertical
face there is an unavoidable error arises from the gauge not being able
to account for the 90 degree corner. The error is substantially reduced
for the closer rod spacing.

Electronic Liner Profile Gauge


Figure 9 – Sources of error from rod parameters
The design of the electronic gauge enables the user to measure the
profile of the liner by sliding the distance measuring laser along a fixed
20mm rod spacing Actual
250
arm. The position of the sliding laser is measured by another distance
measuring laser. The combined information is captured using an A/D
converter and downloaded into a computer for analysis. Figure 11
200
illustrates the electronic liner profile gauge with its 2 lasers. The design
of the gauge is such that it is fully adjustable to varying liner width. This
Y height (mm)

150 is done by sliding the support leg as shown in Figure 11, thus changing
the horizontal range of the laser B.
100

50

0
0 100 200 10mm300 400
rod spacing 500
Actual 600 700
X distance (mm)
250

200
Y height (mm)

150

100

50
Figure 11 – Electronic Liner Profile Gauge
0
0 100 200 300 400 500 600 700 The lasers used in the electrical gauge are two Wenglor Proximity
X distance (mm) Lasers. These lasers have an inherent dead zone where the laser
cannot measure distance within a certain range and the construction
Figure 10 – 10 & 20mm rod spacing profile and design must accommodate this. Laser B has a dead zone of
III-143 III-144

300mm, necessitating a protruding arm of 300mm to support it. Laser A x2 − x1


has a dead zone of 50mm, which is added to the length of the support For Laser B (x distance): Rx = ; xi = RxVi
legs. The support legs are 350mm long, maximum span is 700mm. V2 − V1
y − 50
For Laser a (Y distance): Ry = ; yi = R yVi
Capturing the data: V3
The output signals from the lasers are recorded as voltages that are Profile Results:
captured in an A/D converter wireless data logger. The device has a
maximum data capture rate of 20 kHz, and can sample up to 8 sensors The Electronic profiler was calibrated against the liner mock-up. The
simultaneously. The electrical gauge requires 2 analogue inputs with 10 sliding speed of Laser A is controlled by the user.
Volt sensor connections. The data logger has an in-built battery which
lasts up to 4 hours. The lasers require 24VDC supply which is supplied A slow slide is classified as time taken to slide across the span in
by 2X 12V batteries per laser. The battery pack as shown in the insert of greater than 5 seconds, and a fast slide is time taken to slide in less
Figure 11 encompasses 4 batteries and the data logger and is than 5 seconds. Figure 13 illustrates examples of sampling rates and
suspended by a shoulder strap. sliding speed. The resolution of the profile, in data points per unit length,
is proportional to sliding time. (X density ≡ sliding speed & sampling
The logger has limited in-built memory, so the captured data can be rate (s/s))
stored and retrieved later for analysis. Increasing the sampling rate
increases the resolution of the profile but limits the logging period. 50s/s slow Actual 10s/s slow Actual
250 250

Calibration 200 200

Y height (mm)

Y height (mm)
150 150
During each liner profile, the electronic profile gauge requires
recalibration for laser B as this distance is changed to suit the liner width. 100 100

The laser A does not require renewed calibration as this distance is 50 50


fixed. The voltage output from the laser is directly proportional to the
distance, so the proximity lasers are calibrated using the dead zone 0
0 100 200 300 400 500 600
0
0 100 200 300 400 500 600
50s/s fast Actual 10s/s fast Actual
distance as shown in Figure 12: X distance (mm) X distance (mm)
250 250

200 200

Y height (mm)

Y height (mm)
150 150

100 100

50 50

0 0

0 200 400 600 0 100 200 300 400 500 600


X distance (mm) X distance (mm)

Figure 13– Laser liner profiles

Figure 12 – Calibrating the electronic gauge The 50s/s and 10s/s for slow sliding rates are both in excellent
agreement with the actual profile. For the fast sliding rates, overall the
For Vi = Voltage at distance xi , Rx = laser B calibration and R y = laser A measured profiles show good correlation but show poor comparisons
around the corners for the trapezoids and the semi circular shapes.
calibration:
III-145 III-146

ERROR ANALYSIS Horizontal Error


As the X error is considerably lower than the leg positioning error, the
An error analysis was conducted to confirm the accuracy of the gauges
error in the X-plane Ex can be taken as Ex = ±1.1 mm.
with respect to the actual profile measured with a Vernier gauge to better
than 0.03mm. The equipment errors and the overall errors pertinent to Height Error
the liner wear data being collected are analysed. Location of the support This is determined from the comparison of measured and actual
legs with respect to the reference mark on the liner is conducted by profiles. The SD for all the flat surface points and average coefficient of
repeat tests by the user. Positioning of the support legs on the measured variance for the most error prone zone of the sloping faces are
liner is done by marking a strip on the surface and measuring the determined and given in Table 2.
distance to the profile rod number 1 as shown in Figure 14. A number of
repeat measurement were conducted, an SD of 1.1mm was established.
Table 2 – Error table

Vertical height
Variance, mm Area Angle

flat face (mm2) %error (deg) %Error


Actual 0.0 0.0 51135 0.0 70.0 0.0
Elec 50s/s slow 0.4 1.3 51709 1.1 70.1 0.2
50s/s fast 0.6 2.1 51261 0.2 69.9 0.2
10s/s slow 0.5 1.9 51527 0.8 70.0 0.1
10s/s fast 0.6 1.7 51896 1.5 69.7 0.5
Mech 10mm 0.0 0.8 54340 3.8 69.9 0.2
Figure 14 – Positioning of the support leg 20mm 0.0 1.2 54860 4.9 71.1 1.6

Intrinsic Errors Area Comparison


These are limits to accuracy imposed by the measurement techniques Calculating the area under the graph gives the cross-sectional area of
and are given in Table 1. The resolution of the lasers describes the the measured liner, which can be used to calculate the remaining mass
accuracy of the range and the linearity describes accuracy between the or mass loss. Table 2 shows that the error is significant but acceptable
measured distance and the output voltage. at 5% for the mechanical profiles. The electronic profile has a larger
error for the profiles which have fast a slide speed than for slow slide
Table 1 – Laser description speed. The slow slide electronic gauge provides excellent accuracy, of
below 0.6%.
Mechanical Gauge
Direction Lifter Angle
X 0.1mm in hole position To calculate the lifter face angle the data from the lifter face section is
Y Limited to within 1mm by steel ruler measurement selected, as shown in Figure 15. A linear fit is projected onto the data
Electronic Gauge set and the gradient calculated.
Direction Laser Working Range Resolution Linearity
X A 50 – 350mm 0.08mm 0.2% For a worn curved lifter, the zone on the lifter corresponding to the
Y B 300 – 1300mm 0.2mm 2% steepest angle is used. Generally a liner fit to this set of data is
adequate. Table 2 shows that the error in face angle is minimal for the
mechanical gauge with closer rods and negligible for the electronic
gauge for slow sliding.
III-147 III-148
View publication stats

The accuracy of the mechanical gauge is within 1mm in height, 5% in


area and 0.1° in face angle. The best design for the mechanical gauge
Angle 1
is with a rod spacing of 10mm. The error in height calculated for the
sloping faces for the mechanical gauge is good after a sloping
Angle 2 correction factor was added.

For the electronic gauge it was found that there was a significant error
in the height measurement on the sloping faces. This was found to be
attributed to the sliding direction of the laser A. Although the calculated
cross sectional area is good for the electronic gauge, the absolute Y
height on the slope shows poor overall error. This issue is under
investigation. The accuracy of the electronic gauge is less than 1.5 % in
Figure 15 – Calculating the lifter face angle area and better than 0.1° in face angle, which exceeds accuracy
requirements. The best suited design for the electronic gauge is
sampling at 10s/s for a 6-7sec slide.
Error in flexing (mechanical gauge)
The weak point in measurement for the steel liners is the plate
The error in flexing is a function of the lifter angle. Therefore the lifter thickness, and so a complementary device is under development to
angle has consequential errors on the lengths of the protruding rods. solve this issue.
With reference to Figure 9, the following is calculated:
For the mechanical gauge, experience shows that 6 to 8 profiles can be
Δy f taken in 30min. for the electronic gauge the profile reading takes less
tan θ = ⇒ Δy f = Δx f tan θ , ( Δy f is the rod length error) than 10s, so most of the time taken is in positioning the gauge and
Δx f
measuring the plate thickness. 15 to 20 readings can be taken in 20min,
But Δx f 0 (for large rod diameter, i.e. the rod becomes stiffer) so 4 to 5 profiles of 3 lines of liners can be taken in 20min, thus meeting
the objective of the design.
Therefore the flexing is proportional to the stiffness of the rods. For the
mechanical gauge, 6mm diameter rods are used. From an error test it REFERENCES
was determined that for these rods, the error is flexing is equal to
McIvor, R.E., (1983). Effects of speed and liner configuration on ball mill
approximately 3tan θ . To minimise this error, the user should push the performance. Mining Engineering, Jun.: 617-622.
rods carefully onto the steep angled lifters.
Powell, M.S., Smit, I., Radziszewski, P., Cleary, P., Rattray, B.,
Eriksson, K.G., Schaeffer, L., (2006). The Selection and the Design of
CONCLUSIONS Mill Liners. Advances in Comminution, Edited by Kawatra, S.K.,
Published by Society for Mining, Metallurgy and Exploration, Inc.
The measurement technique presented in this paper is a rigorous Powell, M.S., (2000). MillTraj – Liner Design Software. JKTech
methodology to measure the profile of liners. The tools used to measure application papers, Published by JKMRC, University of Queensland.
the profiles such as the mechanical and electrical gauges use a Powell, M.S. (1991a). The effect of liner design on the motion of the
systematic procedure to measure key features which are pertinent to outer grinding media in a rotary mill. International Journal Mineral
monitoring wear. The data is being built into a predictive software tool Processing 31: 163-193.
that can be used to help prolong the useable life of the liners by Powell, M.S. (1991b). The design of rotary-mill liners, and their backing
providing reliable wear rate and end of life predictions, and thus allow the materials. Journal of the South African Institute of mining and Metallurgy.
lifters to be pushed to a safe limit. 91(2): 63-75.

You might also like