Professional Documents
Culture Documents
Summary
1. Introduction
some of the earliest tests on wear in ball mills, Ellis (11 found no spalling
on 25 mm balls in a mill 30 cm in diameter and Norman and Loeb [ 21 like-
wise found no spalling on 75 mm balls in a mill 90 cm in diameter. In
marked-ball studies in commercial mills, Norman and Loeb [2] did observe
occasional spalling in a mill 2.75 m in diameter, the largest available at that
time. Although field data provide the best overall criteria for evaluating wear
performance, field tests lack control and are time consuming; furthermore,
the time at which spalling or fracture occurs cannot be clearly determined
under service conditions. Thus, the need for a controlled spalling test exists.
An apparatus for effectively producing spalling, fracture and surface degrada-
tion, without abrasion, under reproducible and repeatable conditions is
described in the present paper. Studies on the causes of spalling are under
way.
2. Experimental procedure
The concept of the present design evolved after many other designs had
been conceived and some of them constructed and tried. A ball-on-plate
machine, similar to the machines used by Dixon [3], Durman [4] and Farge
and Barclay [ 51 was constructed first, but it was realized that tremendously
long operating times would be required to produce the large numbers of im-
pacts necessary to produce spalling of the balls. The concept of the present
design is based on a device used in the demonstration of the physics of the
conservation of momentum. The device consists of a frame that supports
a line of six or seven balls in contact, each ball hanging from a pair of strings.
Letting a ball at one end swing into the second ball produces collisions down
the line of balls that make the ball on the other end swing away. We at-
tempted this with steel balls 75 mm in diameter suspended by wires 3 m in
length. Although the device was quite spectacular, any attempt to keep the
balls aligned during. operation was hopeless. Furthermore, the balls always
made contact at the same spot unlike a freely falling ball in a ball mill, which
receives impacts at random over the entire surface. Thus, the concept evolved
of confining the balls inside a tube and letting the initial ball fail freely.
A schematic ~angement of the equipment devised for producing large
numbers of impacts between balls is shown in Fig. 1. Typically, the system
operates with about 22 test balls, 18 are in the curved tube at any one time,
while four are in the return leg. From the top of the machine, a ball drops
a distance of 3.5 m to impact the line of balls confined in the curved tube.
The last ball in the tube immediately pops out with just enough energy to
send it down the ramp to the conveyer, where it is again lifted to the top.
The fall of 3.5 m is believed to represent impact conditions in a ball mill of
considerably greater diameter. The test apparatus is more severe than a real
mill because a real mill contains ore that cushions the blow.
363
8.1 cm ID x 8.9 cm 00
steel tube
Sound insulotion-
Frame
A pprox.
drop Conveyer
75 mm
40cm
dia test
apart
balls
intervals of 1.5 cm, and then bending and welding. Unwelded parts of the
kerfs and additional drilled holes allowed small pieces of ball debris to fall
out during testing. Because large pieces of fractured balls cause the balls to
jam in the curved tube, it was attached only with bolts at each end to facili-
tate its removal. The exit end of the curved tube has a guard that prevents
the balls from popping over the ramp.
From the exit end of the tube, a ramp carries the balls to the conveyer.
Figure 3 shows the end of the curved tube, the guard, the ramp and the
bottom of the conveyer with one of its plastic buckets. Midway along the
ramp are the optical pick-ups that detect each passing ball and actuate the
counter. The conveyer is a reinforced rubber belt 23 cm in width with a
bucket bolted every 40 cm; each bucket can hold two balls. The conveyer is
driven by an electric motor rated at 560 W at a speed of 38 m min-‘. The rela-
tively high speed of the conveyer is required to throw the ball out of the
bucket as it passes over the top pulley. The upper ramp is the width of two
balls and drops the balls onto a narrower single-file ramp located at a right
angle to the upper ramp. The single-file ramp guides the balls to the light-
weight steel tube where the balls drop. Safety-guards cover the upper ramps
and the conveyer belt to prevent the balls from dropping unexpectedly.
365
2.2. Procedure
To start up the system, the curved tube is first filled with approximate-
ly 18 balls. With the conveyer running, the remaining four balls are loaded
onto the bottom ramp. The conveyer carries the balls to the top and flings
them into the upper ramp, then the balls continue in the cycle. During
“steady state” operation the conveyer picks up one or two balls every one
to six buckets, depending on the total number of balls in the system and
their momentary grouping. The machine is run unattended, usually over-
night because of the loud noise.
A run is concluded when any of the balls suffers a failure such that
it jams the flow of balls through the tube or fails to roll down one of the
ramps because of a spall or fracture. A count of the impacts is then read
from the counter. To unjam the system, it may be necessary to unbolt the
curved tube and to pour the balls out; the machine can then be restarted for
continued testing.
The equipment allows occasional random reversals of pairs of balls.
Over a period of time, this allows each ball to impact all the other balls in
the system. This prevents the occurrence of a bias caused by the same pairs
of balls always impacting each other. The reversing is done by the arrange-
ment of single- and double-width ramps. Figure 3 shows the single-file ramp
that leads to the loading chute which is the width of two balls. Occasionally,
two balls will load into one conveyer bucket. After exiting onto the double-
width ramp at the top, the balls roll into the single-file ramp at right angles
to the double-width ramp. Here the balls will be in reverse order. Thus, the
order of impacting balls gradually changes as the test progresses.
366
3. Test specimens
balls were forged in closed dies and subsequently heat treated. All other balls
were cast in sand molds by a commercial foundry or by our laboratory
foundry, with three balls per casting. The metal was air melted in an induc-
tion furnace with charges of 15 - 30 kg per heat. Cast balls were heat treated
and tempered in a manner suitable for each composition. At temperatures
below 1000 “C the balls were heated in cast iron chips to prevent decarbu-
rization. When they were heated at higher temperatures, the balls were
wrapped in stainless steel foil to prevent oxidation and decarburization.
Some balls were tested in the as-cast condition.
rapII
Porting line, on
cast balls
roave from
8 1, used an
Fig. 4. Ball identification system showing the flat spots which were ground on the ball.
368
castirons
24 3C~kY.v alloy 627 3
41 SCr-Ni, Ni hard IV 622
42 15Cr-3Mo 540
19 ZOC-2Mo 680
43 ZOCr-ZMo~lCu, as cast 578
44 25Cr-HC-250 648
Cast stee/s
25 Low alloy, 12s 375
36 Low alloy, AR360 370
37 ZCr-0.5Mo 780
30 6Cr- 1Mo 643
Wrought steek
51 Commercial ball A 712 2
9 Commercial ball B 740 14
52 Commercial ball C 627 2
76 Commercial ball Cz 321 4
57 Commercial ball F 352 1
58 Commercial ball I 670 8
55 Univ. Calif. “Quatough” 411
23 ALRC Cr-MO 627
100 I50 200 250 300
IMPACTS, IO3
Fig. 6. Spalling of a ball 75 mm in diameter (specimen 41C; number of impacts, 192 000).
hardness ranging from 627 to 712 HB and on the manganese alloys, which
were quite soft initially but which work hardened to about 550 HB.
Flaking, the third type of failure observed, was the formation of very
thin flakes that resulted from the extreme surface fatigue and cold work of
the relatively soft low alloy steels. The hardness of balls that flaked was
below 400 HB. At least 50 000 - 100 000 impacts were required to produce
the degree of work hardening needed to develop flakes. The ball shown in
Fig. 8 has typical thin flakes attached to its surface. As the flakes developed,
371
they gradually came loose from the ball and eventually fell off. Flaking is
not observed in a real ball mill, undoubtedly because abrasive wear of the
surface by the ore dominates the wear process. Flakes were examined by
transmission electron microscopy. The thin regions of the flakes that could
be penetrated by electrons revealed extreme cold work and were estimated
to be less than 10m6 cm thick.
Breakage, the fourth type of failure, was a complete fracture of the ball
and generally occurred after relatively few impacts (a few thousand to
50 000). Many of these fractures passed near the center of the ball. Some-
times breakage took place spontaneously within several hours of the end of a
test run. Figure 9 shows a broken ball with its typical work-hardened rim
that extends about 5 mm below the surface. It is believed that the fracture
pattern reveals the rim region because of the high compressive stress there.
Static equilibrium requires that the outer compressive forces should be
balanced by inner tensile forces. It is believed that the ball breaks when the
tensile stress at some point in the core exceeds the fatigue strength. The
break is sudden, almost explosive, because of the stored energy in the rim
and core. The average hardness of the four alloys that failed by breakage was
667 HB.
Fig. 10. Energy distribution for 18 balls in the tube: 0, energy input; n, energy dissipated.
gy incident upon each ball. The energy dissipated at each ball is also shown
in Fig. 10 and ranges from 6.7 J initially to 0.7 J at the last ball. It is not
known what fraction of the energy is dissipated as friction between the ball
and tube, internal friction within the ball as heat, plastic deformation of the
ball as heat, and cold work of the ball as accumulative stored energy.
Measurements of the time of contact between commercial steel low
alloy balls gave approximately 0.1 ms, which agrees with theoretical calcula-
tions 163. The area of contact and the maximum force (stress) between balls
were not measured.
5. Conclusions
A test procedure was devised in which spalling, fracture and other sur-
face effects could be produced in relatively short times by multiple impacts
between balls. The machine can be run unattended. Severe spalling or break-
age of a ball stops the test, thereby allowing the number of impacts at failure
to be accurately determined. An accurate counting of the number of impacts
on each ball can be maintained while balls are added to or removed from
the test.
64 commercially forged and cast balls of 22 types were evaluated. In
general, the hardest balls (with hardness greater than 670 HB) failed by
breakage after relatively few thousands of impacts or suffered severe spalling
as the number of impacts increased beyond about 50 000. Balls that did not
break or spall eventually minispalled after 100 000 - 150 000 impacts. The
softer balls (with hardness less than 400 HB) neither broke, spalled nor
minispalled but after 50 000 - 100 000 impacts they developed an extremely
cold-worked surface that began to flake because of surface fatigue.
It is believed that the ability to produce large-scale spalling in the labo-
ratory under controlled conditions will lead to a better understanding of
the mechanism of spalling and how to reduce it.
References
1 0. W. Ellis, Proc. 34th Annu. Conf. of the Inst. Br. Foundrymen, June 1937.
2 T. E. Norman and C. M. Loeb, Jr., Trans. AZME, 176 (1948) 490 - 526.
3 R. H. T. Dixon, J. Iron Steel Inst., 197 (January 1961) 40 - 48.
4 R. W. Durman, Foundry Dade J., 134 (1973) 645 - 651.
5 J. C. Farge and G. A. Barclay, in R. Q. Barr (ed.), Proc. Symp. Materials for the Mining
Industry, Vail, CO, July 30 - 31, 1974, Climax Molybdenum Co., Greenwich, CT,
1975, pp. 189 - 199.
6 P. A. Engel, Impact Wear of Materials, Elsevier, New York, 1978.