Grinding Ball Rationing

Determining the optimum size assortment of grinding balls

for a makeup charge is a practical means of improving mill
operation.

by Walter L. Crow

M UCH has been published on highly technical

phases of grinding. Very little has been written
on how a mill man can improve his operations by
and some of the larger ones, have not the facilities
to blend mine run ore for mill feed, and with a great
variation in character of feed there may be a diffi-
determining the optimum size assortment of grind- cult problem in establishing even an optimum sin-
ing balls that should be added as a makeup charge. gle-size ball charge. Small plants may find that this
This ball rationing is not to be confused with the has taxed their personnel enough without going into
ball ration used as an initial charge when the mill ball rationing. This does not mean that the personnel
is started up. lacks technical ability, but that facilities, personnel,
Ball rationing is considered for one or more of the and sufficient time are not available for careful test-
following purposes: 1 ) to increase throughput of the ing. Moreover, the cost of such an effort may not be
mill, 2) to reduce the power required per ton of ore justified. Total savings in dollars and cents, through
ground to the desired size, 3) to reduce steel con- a relatively minor improvement in grinding prac-
sumption per ton of ore ground, 4) to improve prod- tice, will not be as great in a small operation as in a
uct size, and 5) to lower retention time of ore in larger one.
the mill. Ball Wear Pattern: When one-size balls are used
An increase in throughput of the ball mill offers a for addition, the seasoned charge in the mill ranges
big advantage. Since total milling cost is the sum of from balls of the original diameter to those small
fixed or indirect costs (capital investment, salaries, enough to purge from the mill. If a screen analysis
etc.) plus variable or direct costs divided by the tons of the charge is made using screens with openings of
of ore processed, and since indirect costs remain con- equal increments, such as 55 in., the weight of the
stant irrespective of increases in throughput, the balls on each screen will show a certain pattern, or
total mill cost (in cents per ton) will decrease. This distribution of the charge by weight, whether ball
advantage can be realized in plants that can raise wear varies in direct proportion to its surface area
the rate of mining and all steps in ore processing to as D' (attrition grinding), as maintained by Rit-
match the increase in throughput of the ball mill. tinger; or in direct proportion to its volume as D"
Indications That Rationed Charge Is Needed: (impact grinding), as maintained by Kick; or some-
When a one-size ball addition to a mill is being es- where between these two figures as suggested by
tablished, the following conditions may warrant Bond ( a combination of attrition and impact com-
ball rationing: minution).
1) There may be a certain amount of tramp over- These theories of ball wear do not hold true, in
size that can be reduced by replacing a portion of the writer's opinion, for apparently the rate of wear
the balls by a larger size. of different ball sizes in the charge is affected by size
2) There may be crowding of particles of reduced structure of the mill feed and by crystal size of the
size but not of finished size, showing a deficiency of minerals. Also the physical and chemical character-
small-size balls. istics of the ball may vary with the distance from its
When a new mill is started up, of course, a cer- center.
tain time is needed to get the processing running There is evidence that there is a difference in the
smoothly. During this period a further problem of wear pattern of grinding balls of different manufac-
ball rationing would be ill-timed. ture. Some appear to be self-rationing compared
Not all mills should attempt ball rationing, and with others.
these words of caution are offered. Most small mills, The advantage of large balls in the charge is that
.- they drop with greater impact and have a more
W. L. CROW is Production Engineer, Colorado Fuel & Iron Corp., effective nipping action on the larger particles. Small
Denver. balls make a greater number of contacts because

750-MINING ENGINEERING, JULY 1957

there are more of them, so that attrition grinding is pendent on the sample of ore submitted for labora-
more effective and there is a higher crushing inci- tory test and is limited by the shortcomings of labo-
dence due to nipping action on the smaller particles. ratory tests. Some people question that the factor F
Ball rationing is employed to change the size distri- is an ideal criterion.
bution of the ball charge to one that has a better When a new mining property is being developed
ratio of impact, nipping, and attrition. it is often not accessible, so that sampling is not
One-Size Ball Makeup for a N e w Mill: Before an thorough enough to indicate all the types of ore that
attempt is made to work out a rationed makeup eventually will be encountered. It is important that
charge, the best one-size ball makeup charge should the comminution problem of various ores be studied
be established. Ball size is determined mainly by the in terms of the subsequent metallurgical processes,
factors listed below: both physical and chemical. The degree to which the
Ball ores will be blended before entering the ball mill
1 ) Specific gravity (affected by voids in the ball) should be taken into consideration. If no blending or
2) Shape poor blending is anticipated the most difficult grind-
3 ) Homogeneity ing ore should be given the most weight in determi-
4) Relative cost of balls by diameter nation of ball size, although this size may be too
Mill large for the softer ore.
1) Inside diameter Through past experience some mill manufacturers
2) Speed (peripheral speed, rather than percent have prepared tables that specify ball sizes for each
critical speed) mill size. Tables are usually set up with different
Manner of Operation (Assuming One-Stage Grind- ball sizes recommended under ore classified as hard,
ing Only) medium, and soft. The question is, how hard is a
1) Open or closed circuit. (Percent circulating hard ore?
load of closed circuit) An important consideration is the ball size used
2) Mill pulp density (specific gravity of pulp con- by other mills with similar ores, especially when
stituents) similar ball mills are used.
Feed Material In an operating mill it is not difficult to determine
1 ) Size structure of mill feed whether the optimum one-size ball makeup charge
2) Desired particle reduction is being used. Variation in grindability of ore is the
3) Character of ore chief complicating factor. It is better to add too large
a ) Specific gravity of gangue and of mineral or rather than too small a ball, although fewer balls are
minerals used, giving fewer contacts and less attrition grind-
b ) Grindability characteristics ing. This is a precaution against encountering ores
1) Comminution to crystal sizes that are more difficult to grind, which may cause an
2) Comminution through crystal sizes increase of tramp oversize to be circulated through
3) Sliming characteristics the closed circuit, choking up the circuit and necessi-
Grindability tests have been made with labora- tating reduction of mill feed.
tory-size equipment by mill manufacturers, insti-
tutions, and mining companies. These tests indicate Too large a ball will reduce the larger feed parti-
grindability of the ore, but scale-size work has its cles with little tramp oversize, but the reduced par-
limitations. The ratio between mill diameter, mill ticles will crowd before they reach the required size
peripheral speed, ball diameter, and particle size for further processing, and excessive slimes may also
which is obtained in laboratory work is not the ratio be produced. In closed circuit, recirculation of the
that exists in the full-scale operation. A mathemati- crowded sizes will overload the classifier, requiring
cal solution of the problem may indicate the proper reduction in mill feed. Added to an open circuit mill,
ball size. The simplest mathematical aid is the Cog- grinding balls that are too small allow tramp over-
hill and DeVaney formula, D2= Kd, where D is the size to enter the next process, and balls that are too
ball diameter, d the particle size, and K a constant large do not produce the desired fineness of grind or
varying between 25 and 50 depending on the relative liberation size, while at the same time they may pro-
hardness of the ore. duce too many slimes.
This empirical formula cannot always be trusted, Prices of the different size balls are worth consider-
and it is evident that many factors listed on page 752 ing. The lowest-priced steel ball is commonly the 3-
under General Method of Rationing are ignored. in. diam ball. Larger sizes are slightly higher in cost.
Bond has gone further in developing the empirical For smaller sizes the price becomes rapidly higher.
formula : If 3 % and 3-in. diam balls give similar results the
3%-in. ball is preferable, as it gives insurance
FW, S against production of excess tramp oversize in case
the feed becomes more difficult to grind. However,
KC, the lower price of the 3-in. ball may be the deciding
factor. Balls of 3-in. diam are more commonly used
In this formula B is the diameter of the ball to be in beneficiation plants and are usually more avail-
charged, in inches; C is the percent critical speed of able for immediate delivery.
mill; D is the inside mill diameter, in feet; F is the It is an interesting contention that a grinding ball
size of screen opening, in microns, that 80 pct of the with a softer core is less subject to splitting from
new feed to the mill passes; K is a constant which impact and that when the ball is worn down to the
has a value of 200 for a closed circuit ball mill; and size considered of little value in comminution, it is
S is the specific gravity of the ore. Wi is called the then reduced in size at a faster rate by abrasion and
work index, which is determined in the Allis-Chal- thus ejected from the mill sooner, making way for
mers laboratory. This formula assumes the use of a the addition of balls of more useful size.
spherical steel or iron ball and takes into account the As in other experimental work in an operating
more important factors listed above. The factor circuit, it is good practice not to make too radical a
most difficult to determine is W,. It is of course de- change. If it is indicated that 4-in. diam grinding

JULY 1957, MINING ENGINEERING--751

 balls would be more satisfactol-y than the 3-in. diam charge in ball ration, it is important that the mill
balls in use, it would be better to test 3%-in. diam feed be kept at capacity at all times.
balls first and then check results. Or, if it is thought A study of the classifier sands using the graphic
that 2%-in. diam balls will improve results over the method presented by Coghill maybe made. The size
present use of 3-in. diam balls, it may be better to structure is plotted using the abscissa to represent
use only one quarter or one half the charge of each mesh size of the standard series screen and
2%-in. diam balls to check for improvement before using the ordinate to represent the respective per-
going to 100 pct 2%-in. balls as the makeup charge. cent weight of material on each screen. If the plot
General Method of Rationing: How can the ideal shows a bulge in finer sizes, a smaller ball should be
ratio of different sizes of balls for a mill be deter- considered in the ration. The size and number of balls
mined? How can the ratio be varied? The second must be determined by trial and error. All the sizes
problem is relatively easy. of balls in the mill do work in reducing all the sizes
Many operating personnel in the cement industry of ore particles, but each size of particle is most
believe that when raw materials and clinker a r e effectively reduced by a certain size of ball. This is
ground the seasoned charge contains too many balls simply a restatement of Davis's rule. As Davis sug-
worn to small sizes and irregular shapes, giving in- gests, best results in grinding are obtained when the
effective contact on impact. These take up space that size analysis of the classifier sands is uniform.
could be utilized by larger spherical balls that would W , in Bond's formula and the K in Coghill and
grind more effectively. To increase the ratio of larger DeVaney's are subject to the weaknesses of laboratory
balls in the charge the mills are periodically dumped, determinations and may be somewhat in error in
the charge is screened, and the larger balls are re- calculation of a one-size ball makeup charge. When
turned to the mill. Cull balls are replaced by new an optimum one-size ball makeup charge, B and D,
ones. has been proved by practice, the formulas can be
If it is determined that there is a shortage of solved for W i and K. By observing the bulges on a
small balls in the seasoned charge, the makeup Coghill type of plot of the classifier sand size struc-
charge may be partly replaced by one or more sizes ture, it is possible to estimate a new F and D. With
of the smaller balls. This is a rationed charge. the variables solved for, W, and K, and estimated, F
Steps in Working Out the Ball Ration: Complete and D, the formula can be solved for a new B and
records should be kept so that the throughput of the D, and these values can be used as an indication of
mill, power consumption, and ball wear are known. the ball sizes to be added to the ration.
It is also well to record liner wear, although this re- By screening a dumped ball charge as described
quires observation over a long period of time. It is under Ball Wear Pattern, it may be found that there
very important to keep records of feed, discharge, is a deficiency in certain ball size or sizes in the
classifier overflow, and classifier sands screen analy- wear pattern. This may be correlated with the
ses. Classifier sands size structure will give indica- bulges in the plotted curves of the screen analysis
tions as to whether smaller or larger balls in the of mill feed and classifier sands and with the crystal
charge will give better results. Davis's rules should sizes of the minerals being ground. This deficiency of
be used: a certain size of ball in the wear pattern may indi-
cate that this size or one slightly larger should be
1) Crowding will appear at the fine sizes of par-
ticles in the classifier sands, if the seasoned charge is added to the makeup charge for grinding improve-
graded too much towards the larger sizes of balls. ment.
Ideally, comparative tests of different ball rations
2) Crowding will appear at the coarse sizes in should be made in mills that are in parallel, with
the classifier sands, if the ball charge is graded too
much toward the smaller sizes. identical mill conditions, with feed as similar in
3) Best efficiency is obtained when there is a character as possible, and with independent control
minimum crowding at any size of the size structure of feed and water to each mill so that the mills may
of the classifier sands. be fed at capacity. If the same mill must be used for
comparative tests, any change in ore characteristics
Complete records should be kept to show the and liner condition must be correlated.
effect of any change in ball addition to the mill
throughput, power consumption per ton, and ball Conclusion
wear in pounds per ton. These changes will affect Attaining an ideal ball ration is not easy. How-
the size structure of the mill discharge and the clas- ever, by using common sense in applying established
sifier overflow and sands. Before the full effects of a principles of grinding, by keeping complete records
change in ball addition occur, the ball wear pattern and avoiding changes that are too radical, and by
should be given time to reach its new equilibrium. allowing enough time for changes to take effect,
This will require the time necessary for the old ball many mill operators can increase the efficiency of
charge to be entirely replaced by the new charge. their ball mills.
For example, if the ball load in the mill is to be
maintained at 100,000 lb and 1000 Ib of balls are Acknowledgments
added per day to maintain the ball level in the mill, The author wishes to acknowledge the assistance
it is reasonable to assume that more than three rendered by C. E. Golson and H. M. Holkestad of the
months will be needed to complete the modified ball Colorado Fuel & Iron Corp. and by D. J. Drinkwater
charge. of the Mine & Smelter Supply Co.
If the new makeup charge is one that would in-
References
. .
crease the rate of comminution of the ball mill, and
1 F. C . Bond: Crushing and Grinding Calculations. Allis-Chalm-
if the new ball charge reaches its equilibrium with- ers Mfg. C o . , S e p t . 6 1956.
2 L . M . Barker and E. G . L e w i s : Developments o f Ball Mill G r i n d -
out an increase in feed rate, the new charge will not ing Practices a t N e w Cornelia. AIME Trans., 1943, v o l . 153, pp.
show an increase in throughput, the steel consump- 333-344.
3 W H . Coghill: Evaluating Grinding Efficiency b y Graphical
tion will be too high owing to steel on steel wear, ~ e t h o d s Engineering
. and Mining Journal, 1928, v o l . 126, PP. 934-
and power consumption per ton of ore ground will 938.
4 E . W . Davis: Fine Crushing in Ball Mills. AIME T r a n s . , 1919,

be unnecessarily high. To prove the value of a v o l . 61, pp. 250-294.

752-MINING ENGINEERING, JULY 1957