Mill Grinding

Introduction to Process Engineering
Table of contents
3. MILL GRINDING THEORY
1. Introduction and Overview.............................................. 3
1.1 Introduction to Grinding......................................... 3
1.2 Grinding Mill Circuits.............................................. 4

1.2.1 Open - Circuit Grinding................................ 4
1.2.2 Closed - Circuit Grinding............................. 4
1.2.3 Principles of Closed Circuit Grinding.......... 8
1.3 Types of Mills.......................................................... 8

1.3.1 Ball Mills...................................................... 9
1.3.2 Roller Press Mills......................................... 9
1.3.3 Roller or Bowl Mills..................................... 9
2. Internal Ball Mill Parts..................................................... 12
2.1 Partitions................................................................. 12
2.1.1 Purpose of Partitions................................... 12
2.1.2 Double Wall Diaphragm Partitions.............. 12
2.1.3 Operation and Repair of the Diaphragm
Partition....................................................... 16
2.2 Liners...................................................................... 17
2.2.1 Purpose of Liners......................................... 17
2.2.2 Types of Liners............................................ 18
2.2.3 Liner Quality................................................ 24
3-1
3. Ball Mill Grinding and Related Topics............................. 25
3.1 Ball Milling Process (Basics and Fundamentals)... 25

3.1.1 Size Reduction............................................. 25
3.1.2 Bed Movement............................................. 26
3.1.3 Factors Influencing Size Reduction............ 27
3.1.4 Mill Critical Speed....................................... 28
3.1.5 Ball Size and Breakage Rates..................... 29
3.1.6 Ball Mill Liners............................................. 31
3.1.7 Mill Sweep Influence on Grinding............... 31
3.1.8 Mill Mass Transport and Bypass................. 33
3.2 Grinding Media........................................................ 39

3.2.1 Ball Types and Wear.................................... 39
3.2.2 Ball Coating................................................. 41
3.2.3 Ball Charge Design...................................... 41
3.3 Ball Mill and Circuit Evaluations and Tests........... 52

3.3.1 Mill Material Levels..................................... 52
3.3.2 Fineness Evaluation.................................... 54
3.3.3 Mill Retention Time..................................... 54
3.3.4 Circulating Load.......................................... 57
3.4 Ball Mill Control...................................................... 58

3.4.1 Basic Ball Mill Control Theory..................... 58
3.4.2 Mill Motor kW Control.................................. 58
3.4.3 Mill Sound Control....................................... 61
3.4.4 Discharge Bucket Elevator Motor kW......... 62
3.4.5 Rejects Flowrate......................................... 63
3.4.6 Rule Base Mill Control................................. 64
3-2
4. Air Separators.................................................................. 67
4.1 Introduction............................................................ 67
4.2 Types....................................................................... 67
4.2.1 Static Grit Separator................................... 68
4.2.2 Sturtevant Separators................................. 72
4.2.3 Raymond Separators................................... 77
4.2.4 High Efficiency Separators......................... 79
4.3 Separator Efficiency............................................... 87

4.3.1 Tromp Curve................................................ 87
4.3.2 Rosin-Rammler Number............................... 92
4.4 Mill Circulating Loads............................................. 93

4.4.1 Definition..................................................... 93
4.4.2 Circulating Load and Production Rates...... 94
4.4.3 L/D Ratio and Circulating Load................... 96
4.4.4 Circulating Load Calculations..................... 97
4.5 Qf/Qa Principle........................................................ 101

4.5.1 Bypass and Qf/Qa........................................ 102
3-3
3. MILL GRINDING THEORY
Foreword
The text of the following section on Mill Grinding was adapted

from a larger document entitled:
Lafarge Mill Grinding Reference, Edition 1

Volume 1: Practical Fundamentals of Ball Milling and Separation
Readers wishing to study the topic in greater depth and detail

are encouraged to obtain a copy of this and other appropriate
volumes which are available from CTS in hard copy or on diskette
for Macintosh.
Other volumes that will be available in the Lafarge Mill Grinding

Reference series (all of which will be available before the end of
1992) are:
Volume 2: Auxiliary Equipment in Mill Circuits

Volume 3: Process Methods and Theory
Volume 4: Roller Mills and Roller Presses
Volume 5: Wet Process Grinding
Volumes 1 & 2 were meant to be practical reference and idea

books for plant people; to help solve the dizzy array of problems
they encountered. The next three texts specialize and include
more theory where appropriate.
Reader should understand that the text presented here has been
edited, substantially, to suit the time allotted in this course
3-1
"Introduction to Process Engineering". Realistically we cannot

present and teach this segment which normally requires 9
working days (for Production Supervisors). Quite simply there is
too much material. Recognize that the topic "Mill Grinding and
Separation" is far more complex than most of us realize and in
any short course one can only cover the highlights which never
delves into the intricacies of this whole field. We can only
encourage you to ask questions and discover.
Sam Fujimoto Paul Ukrainetz

Process Engineer Process Engineer
Lafarge Canada Inc. Lafarge Canada Inc.
3-2
1. Introduction and Overview
1.1 Introduction to Grinding
In the cement industry we spend a lot of effort and money

into size reduction at various stages in the whole process.
At each of these steps we try to grind materials very finely
and within a specific size and surface area range. Why? A
given particle's size (or mesh) and surface area (or blaine)
significantly influences the speed and completeness of the
chemical reaction with that particle. For example a pound
of course particles will have a much smaller total surface
area and will react slower than a pound of very fine
particles which would have a much larger total surface
area. Moreover with the larger particles, there is the
chance that the reaction will not consume the whole
particle, leaving the centers untouched. Grinding allows
the cement manufacturer to influence and tailor the
process to achieve the desired result.
In raw mix grinding, the particle size has an important role

in the ability and ease to produce clinker, the clinker
quality and the efficiency with which we accomplish this in
the kiln. Most plants use a % passing 200 mesh (or 75
micron size) target as the index to determine, whether the
product has been properly ground. See also "The Impact of
Raw Grinding on Kiln Operation".
For coal mills, most plants use the 200 mesh target as well.
The particle size in fuels has a profound effect on the
flame's shape, temperature and stability, which ultimately
influences the clinker quality.
For cement finish mills, we use both a 325 mesh (45 micron)
and blaine (or surface area) targets. Different targets and
3-3
different emphasis are used depending on the type of

cement being produced and the desired performance. For
example, ASTM Type 1 cements characteristically has very
good early day strength gains but slows down at 28-days.
The 325 mesh in cements strongly influences late day
strengths and therefore most plants focuses on this target
closely. For ASTM Type 3 we typically raise the blaine or
surface area (the amount of super fines) in order to
dramatically raise the 1 and 3 day strength performance.
In short how we grind materials at the plants at each stage

plays a very major role in our success in producing
consistently high quality cements. Therefore we must all
understand grinding thoroughly if we are to stay in
business. We shall begin that process here.
1.2 Grinding Mill Circuits
A mill circuit is a combination and proper arrangement of

one or more grinding mills and the auxiliaries required to
convey, classify and to collect the ground product.
Let us first look at the Ball Mill grinding circuits. Once a

circuit is defined as either WET or DRY grinding, it need be
further defined as OPEN or CLOSED circuit.
1.2.1 Open-Circuit Grinding
Where the mill product is sent to storage silos without

sizing or returning the oversize to the mill for further
grinding. Figure 4.1.2.1 is a sketch of an open circuit mill.
3-4
Since size reduction must be accomplished in one pass,

open circuit mills tend to be very long.
1.2.2 Closed-Circuit Grinding
Where the mill product is sent to the separator and the

oversize returned to the mill for further grinding. The
oversize material can be called REJECTS, TAILINGS or
CIRCULATING LOAD. Figure 4.1.2.2 is a sketch of a closed
circuit mill. Figure 4.1.2.3 is a variation showing flash
drying.
3-5
Fig. 1.2.1 - Open Circuit Grinding
Sketch of an open circuit mill
1 2 1. Clinker silo
2. Gypsum silo
3. Mill feed
4. Grinding mill
5. Ground cement
3
5
4
3-6
Fig. 1.2.2 - Closed Circuit Grinding
Separator Feed (A)
Separator
Rejects (R)
Fines (F)
Fresh Feed (K)
Mill
3-7
Fig. 1.2.3 - Closed Circuit Grinding
With Drying of Feed
Separator Feed (A)
Separator
Fresh Feed (K)
Rejects (R)
Fines (F)
Mill
3-8
1.2.3 Principles of Closed Circuit Grinding
For most raw mix and cement grinding we used closed-

circuit grinding.
In a closed circuit system, the mill product is conveyed to a

classifier (separator) and the material coarser than the
required maximum size is returned to the mill feed. The
separator removes the fine (Finish Product), which is then
pumped to storage and the coarse particles (rejects) are
returned to the mill. Sufficient new feed material is added
as required to maintain the charge within the mill at the
operating capacity.
From the cost standpoint, it is less expensive to grind in

open circuit if the required product does not exceed
approximately 3300 Blaine. The mill system in open circuit
is simpler, but it is not as efficient as closed circuit
grinding. In general they tend to overgrind the product to
maintain a certain mesh target.
In the closed circuit, the mill has a greater capacity and the
finish product contains little or no oversize, depending upon
the adjustments made to the separator. Also for quality
control, closed circuit systems offer more options to adjust
the product particle size distributions for optimum
performance.
1.3 Types of Mills
The grinding mill is the main piece of equipment used in the

total finish grinding system. Grinding mills can be
classified into the following categories:
3-9
1.3.1 Ball Mills.
1.3.2 Roller Press Mills.
1.3.3 Roller or Bowl Mills.
3 - 10
1.3.1 Ball Mills
A ball mill is a cylindrical drum with varying ratio of

diameter to length, (also called tube mills). It has either
one, or two compartments, (sometimes three compt.). The
grinding media are usually steel or cast iron balls of
different diameters. Within Lafarge, the vast majority of the
grinding is accomplished with such mills and come in a
variety of arrangements reflecting the technology of the
time and what it is supposed to be grinding.
1.3.2 Roller Press Mills
A roller press as shown on the attached diagram is nothing

more than a pair or rolls placed in opposing position with a
small gap between the rolls. One of the rolls is stationary
(fixed roll) and on (moveable roll) is mounted on sliding
guide ways, with hydraulic cylinders applying force toward
the fixed roll. The moveable roll of the roller press is under
constant load from the hydraulic cylinders. It is an old idea,
re-invented for a new application. Thus far installations
have been applied to existing ball mills, to dramatically
improve grinding rates. However they are difficult to
balance and can be expensive to maintain.
1.3.3 Roller or Bowl Mills
Roller Mills consist of wheels (or rollers) mounted above a

rotating table. Fresh feed is dropped into the table is
ground between the wheels and table as the table turns.
Often these mills are air swept and usually come with there
own built in separator. Many of the installations in North
America are for grinding coal, with a few grinding raw
3 - 11
materials (eg. Balcones, Davenport and Demopolis).

Installations are usually compact and are ideal for relatively
soft materials, but normally have complex hot gas circuits
associated with them.
3 - 12
Roller Press Configuration
The basic configuration of the roller press consists of:

• 2 rollers
• Frame
• Roller support bearings
• Hydraulic cylinders
• Hydraulic accumulator
Use Word 6.0c or later to
view Macintosh picture.
3 - 13
3 - 14
2. Internal Ball Mill Parts
2.1 Partitions
2.1.1 Purpose of Partitions
Their primary purpose is to separate the different sizes of

the ball charge in order to roughly match the ball size to
particle size being ground. For example, many open circuit
mills were originally supplied with three compartments.
However most compartmented mills in Lafarge today have
just two. The material being ground in the first
compartment passes through the grate slots in the partition
to enter the second compartment, then leaves the mill
through the discharge grate.
A secondary purpose for a double intermediate partition is

to retain the insufficiently ground material in the first
compartment, by regulating the material level in it.
Properly designed the partition determines the flow rate
from one compartment to the next and thereby helps to
maintain a good filling ratio.
Of course it also roughly separates and retains the large

material particles in the first compartment.
3 - 15
2.1.2 Double Wall Diaphragm Partitions
These partitions are equipped with lifters which regulate

the flow rate of the material from the first to the second
compartment. The back plates are blind.
Feed End View of Partition As mill rotates, material is

lifted and dropped through
the opening in the center hub.
Material
Discharged
Airflow
Lifter or Scoop
Blind Plate
In the lower
half, material is
Air Sweep Center Screen pushed through
Center Hub or Cone Side View
the slots, filling
Slotted Grate Segment the empty lifter
chamber.
Standard Liftered Partitions
These are common and mostly suffer from the same

problem. The porosity of the first compartment charge is
high and with a fully-liftered partition (lifters right to the
shell) the material filling ratio (U) is pulled down well below
0.9 and ball wear is accelerated. Usually it is below the 0.6
point at which material breakage rates are reduced. (In
3 - 16
other words the voids are too empty - see also section on
Material Filling Ratio.) This is evidenced by no material
visible on the balls at the partition, sometimes one must dig
18" or more to 'hit' cement.
The French have attempted to overcome this through ball

charge modifications (more smaller balls) and reducing the
slot size. At Demopolis large holes have been cut in the
lifters to reduce their efficiency. Neither method has been
highly successful. Changing slot size and cutting lifters are
irreversible and plants have experienced very limited
success. Adding small balls to increase the filling ratio is
O.K. until the circulating load or material grindability
changes, at which time you must start over.
Other plants have installed partitions which are multi-

segmented. These allow you to replace slotted segments
with blind sections on the inlet side and thereby controlling
the total slot area. The Group does not have a lot of
experience with this type but it appears difficult to make
changes and it also chokes off mill sweep.
Adjustable Double Wall Diaphragm Partitions
This is a double wall partition with adjustable scoops

developed by SLEGTEN, and later offered by other
manufacturers.
The grinding efficiency of a mill depends directly on the

time of retention of the material within the ball charge.
This regulating partition thus permits adjustment of the
material level in the first compartment by using scoops
located between the slotted grates and the blind grates on
the second compartment side.
3 - 17
SLEGTEN DOUBLE - DIAPHRAGM PARTITION
Actual installations of the SLEGTEN regulating diaphragm

on finish mills in the Group are:
• Boussens
• Val d'Azergues
• Port-la-Nouvelle
• Contes
• Le Havre
• Alpena
• Balcones
• Bath
• Whitehall
• Wossingen
• L.F.I. at Dunkerque, Fos, and Norfolk
The Group strongly recommends SLEGTEN diaphragm

partitions over all others. In addition, the SLEGTEN
diaphragm partition is mechanically very well designed in
3 - 18
that it has a simple structure, well adapted to stresses

caused by the rotation of the mill shell, especially in large
diameter mills. However, recognize that the installation of
a regulating diaphragm requires a long commissioning
period, systematically making adjustments of the grinding
charge and partition to see an improvement.
2.1.3 Operation and Repair of the Diaphragm Partition
The diaphragm and grates are a rather weak (perhaps the

weakest) element inside the mill. If they are in poor shape,
the mill will function poorly.
• Slots in bad shape block easily.
• High pressure loss causes poor ventilation.
• Holes, excessive play between the plates, worn and

enlarged slots, all pass clinker particles (2, 5, 10
mm) which prevent the efficient action of the
second compartment ball charge.
It is therefore necessary to conduct a detailed inspection

every time the mill is entered.
The proper operation of the mill partition can be better

understood by the following two process elements:
• Fineness curves of samples taken along the ball

charge, (granulometry). If they show too many
large particles (2, 5, 10 mm) in the second
chamber, the partition needs attention.
• The level of the material in the first compartment.

A well adjusted partition will keep a level of
material equal to the level of the balls (even slightly
higher) in the first compartment.
3 - 19
If this is not the case, then you must play with:

• the position of the scoops, if installed
• the dimensions of the lifters, if possible
• the number and/or the dimensions of the slots
3 - 20
2.2 Liners
2.2.1 Purpose of Liners
The purpose of mill liners is not only to protect the mill

shell but they must also "grip" the grinding balls and
release them at the right height to obtain the cascading or
tumbling that maximizes the grinding rate. If the balls are
carried too high they will get thrown against the mill liners
on the opposite side where the balls or liners may break. In
addition, the grinding action could be limited to the toe of
the mill load. If the balls are not carried high enough then
the impact energies while tumbling are greatly reduced,
thus retarding the grinding action. In addition, recognize
that some sliding contact will occur which increases liner
wear.
Many of us lose sight of the fact that the balls should not be
sliding as they are being lifted up. Sliding between balls
and liners increases wear and can lead to premature
failures. Evidence of this can be seen in the form of
"racing". Racing are the grooves or rings worn into liners
and is a result of sliding contact between balls and liners.
Excessive racing will in turn promote more sliding contact
and thus accelerating the whole process. Recognize
though that as the liner rotates around and enters into the
toe of the ball charge some sliding contact will occur until
the liner has gripped the ball charge, to lift it. Some
manufacturers take advantage of this and market grooved
liners which are designed to do extra grinding in the
grooves themselves as the liners enter the ball charge.
Examples are Manoir, Armco-Delloye, and Owen Corp.
3 - 21
Liner design is critical to good grinding action and wear life.

Careful consideration must also be given to ball charge
design, material, material load and the grinding action
required in that compartment in order to design the
appropriate liner.
3 - 22
2.2.2 Types of Liners
A) First Compartment
The work done in this compartment is done by impact, thus

some call this the crushing compartment. Therefore the
liners must be lifting and are present in several forms,
block, wedge bar, Lorraine and Duo-lift. Their job is simple.
Lift the balls high to eject them from the charge so they fall
farther and hit harder. This is shown in the following
drawing showing the detachment point of the balls from the
lining for a 0.75 m Ø test mill, 75% critical speed, 25%
volume load, 30 mm balls, material filling ratio 1.0, feed <
3.15 mm.
3 - 23
Therefore saturating is supported and the fracture

mechanism enhanced. This means that breakage rates will
be shifted towards the coarser particle sizes, at the
expense of breakage rates on the finer ones.
This is clearly shown by tests conducted by Rogers et al.

("..Effect of Liner Design on Performance of a...Wet Ball
Mill") on a 0.91 m Ø mill. The lifters move the peak
breakage point to a coarser size, from 1.5 to 4 mm. The
breakage of finer material is much lower with lifters than
with corrugated lines. Therefore lifters don't belong in the
second compartment.
3 - 24
Thus this is the type of liner which is the first choice many
plants, (sometimes called shipload or wedge or stepped
liner).
3 - 25
Even so, there are other types still in use in North America.
For wedge bar, lorraine, block type liners the ratio of height
to spacing in between (usually matches bolting patterns) is
critical to achieving the correct trajectories. Surprisingly
many suppliers do not have sophisticated methods to
determine this. However this group of liner types seems to
be ideal in SAG mills used in mining or for mills with very
low % critical speed. Most single wave liners generally had
good lift characteristics, however on the down slope side,
media tended to slide which vastly accelerates wear. The
new “DuoLift” liner developed by SLEGTEN could provide an
interesting results.
3 - 26
The Duolift is especially designed to maintain the same lift

characteristics through 90% of it wear life by controlling
the wear pattern. In doing so Slegten claims that the
absorbed power and grinding efficiency will remain more
consistent, and wear life greatly extended. Presently this
liner is being tried in Demopolis and Whitehall.
B) Second Compartment
Corrugated/Wave Linings
These are the standard linings in North America that

original come with most mills. As with lifters they come in
all versions and perversions, wave, simple wave, double
wave, etc. Their purpose is to assist in the cascading
action of the balls. Therefore they assist in the attrition
mechanisms, chipping and abrasion, on smaller sized
particles. This means they do not belong in a first
compartment, but rather in the second. They are, however,
not the recommended second compartment liners.
Classifying Linings
This type of liner has several important advantages:
• ball size matches particle size along the mill;

• reverse and double-reverse classification is avoided.
They have also been shown to increase material transport

opposed to a non-classified charge. This is due to the
increased porosity (big balls) in the areas where the
particles are larger and less 'fluid'. This means it is
theoretically possible to reduce the final ball size in the mill
3 - 27
in comparison to an unclassified charge for better 'fit'

between balls and material without material transport
problems.
This fit is obviously the best when 'plug-flow' is present.

This means that material flows through the mill as though
in a pipeline and isn't mixed too much. The longer L/D in
the second compartment gives RTD's (residence time
distributions) which are closer to the plug flow limit than
the short L/D first compartment.
The advantages of a classified charge are greatest when it

allows a wide range of ball sizes to be used. The fact that
only 3 to 4 ball sizes are present in the more mixed first
compartment reduces to benefits of this lining in that area.
3 - 28
For each liner step for classifying liners in large mills over
4 metres (13 feet) in diameter, use two liners per step,
lengthwise of the mill.
Large media, 50 to 70 mm (2 to 2 3/4 in) diameter, can be

put in the same compartment as small media, as small as
15 mm (5/8 in) diameter in an open-circuit mill. In such a
compartment, the grinding charge can accept a very coarse
feed. Classifying liners permit a deteriorating operation in
the first compartment, or in the diaphragm partition.
Furthermore, they allow a better overall grinding
performance in a mill whose partition diaphragm location is
not ideal.
Except in special cases, which must be examined with

care, classifying linings are recommended for the second
compartment, and long single-compartment mills with lower
circulating loads, but not for first compartments.
2.2.3 Liner Quality
In North America, most liners today are white cast irons

with high chrome contents although there seems to be a
general movement to high carbon tool steel base. Different
alloying elements as well as carbon and chrome contents
are used, we suspect, to achieve desired properties and
performance, but within the confines of the technology
used at different foundries. This appears to vary even within
the same company. The observation is unconfirmed mainly
because none of the suppliers wish to divulge trade
secrets, however none of our experience disputes this.
In general, all liners attempt to produce grains of chromium

carbide, which are extremely abrasion resistant, held in a
martensitic steel (the hardest form of steel) matrix. Also
3 - 29
the best performing liners generally have very fine grain

microstructures (achieved through heat treatment and
proper quenching) which increases overall hardness and
abrasion resistance. By varying the carbide content,
alloying elements, and microstructures suppliers can alter
the liner performance with respect to abrasion and impact
resistance to suit the application.
However, foundry quality control is vital to good liner

performance. This seems to vary more widely in North
America than in Europe; to the point where some plants will
specify the foundry at which the liner is to be cast.
3 - 30
3. Ball Mill Grinding and related topics
3.1 Ball Milling Process (Basics and Fundamentals)
The basic objectives of ball milling is the size reduction of

materials.
3.1.1 Size Reduction
Size reduction in a ball mill is achieved primarily by the

motion of the balls in the mill. By rotation, the mill elevates
the ball charge and material and drops the load upon itself.
Three types of fracture of a particle can occur.
A) Fracture
This is a massive impact in which the particle disintegrates

into many very much smaller particles.
3 - 31
B) Chipping
This is a glancing blow which chips a corner off an irregular

shaped particle. The result is a large particle almost the
same size as the original and a very much smaller "chip".
This mechanism rounds irregular rock into roughly spherical
shapes, for example the pebbles formed in autogenous
grinding.
C) Abrasion
This is the wear of surfaces by rubbing. Again in

autogenous grinding this produces the smooth "beach"
pebbles. The products are once more the main particle,
with little change in size, and very minute particles.
These last two mechanisms are what are commonly

referred to as attrition and the first we observe as crushing.
In any mill these mechanisms overlap to produce particle
size distribution. However by studying the particle size
distributions produced (and seeing the build-up of grits" in
the second-compartment) Austin (Process Engineering of
Size Reduction - Ball Milling, L. G. Austin & R.R. Klimpel &
P.T. Luckie, published by Society of Mining Engineers, 1984)
has concluded that the last two mechanisms of attrition
account for the majority of size reduction which occurs in a
ball mill. Some fracture occurs in the first compartment.
3.1.2 Bed Movement
A) Cascading
3 - 32
At slow speeds the mass of balls are carried by friction up

the side of the mill but immediately slide down in the form
of a locked mass with a slip surface. As the mill begins to
turn faster the ball mass surface appears inclined as
tumbling action of the balls increases. The tumbling action
results in balls emerging, rolling down and re-entering the
surface. The bed is expanded allowing material to
penetrate the voids. The series of collisions which occurs
as the ball cascades down the surface (or streamline)
transfers stress to the material "nipped" between two balls.
B) Cataracting
As the mill speed is increased (or lifters installed) the balls

are lifted higher and more are ejected from the surface at
the top of the mill and a cataract is formed. These balls fall
freely and impact at the toe with great force, some of which
is recovered as they strike the "downhill" side of the mill.
The fraction of critical speed at which cataracting onsets is
a function of the filling ratio and lifter design.
At the 75% of critical speed, at which most of our mills run,

cascading is the primary movement as this is optimum for
the chipping and abrasion mechanisms which are the most
efficient for normal particle sizes. Lifters in the first
compartment assist in the cataracting to increase the
fracture mechanism necessary for large particles.
3.1.3 Factors Influencing Size Reduction
The efficiency of the size reduction in a ball mill is a

function of many factors. In no particular order they are:
3 - 33
A. The mill rpm or % of its critical speed.
B. The size of the grinding media.
C. The mill liners.
D. Air sweep
E. Mill Mass Transport

- Permeability and Retention Time
- Material Filling Ratio
- Discharge Grates
F. Ball Charge Design and Volume Loading (covered in

later sections)
G. Ball coating (covered in later sections).
H. Material characteristics (covered in later sections).
3.1.4 Mill Critical Speed
The critical speed of a ball mill is that speed of rotation at

which centrifugal forces overcome the force of gravity. At
that speed, the balls no longer fall or cascade but ride on
the liners around the full revolution of the mill. Obviously in
order to grind we must operate at a speed below this. As
operators we generally cannot change the mill's speed.
However when working with liner and grinding media
suppliers, they need to know the rotational speed relative
to critical, to properly design the liner and media.
To calculate a ball mill's critical speed use the following

equation:
Vc =
In Feet:
3 - 34
In Meters: Vc =
where Vc = mill critical speed in RPM

Du = useful internal diameter
For Example:
if the useful diameter of a mill was found to be 12 feet then:

U se W o rd 6 .0 c o r l a te r to
vi ew M a c i n to s h p i c tu re .
75% of this would give a mill speed of 16.6 RPM. Bear in

mind that feed material type, ball volume loading, ball
charge gradation and liner types can also affect the
grinding conditions as well. Note that liner selection and
volume loading will affect the grinding action at a given
speed.
3.1.5 Ball Size and Breakage Rates
The specific breakage rate is defined as the rate at which a

given particle size range disappears from the mill, by being
reduced to smaller sizes. This breakage rate is optimized at
a certain size particle depending on the ball size, material
characteristics, mill diameter, mill RPM and lining design.
As we as plant operators don't change mill geometry, speed
or material characteristics, we are left with ball size
selection and lining types.
Researchers, like J.P.Bombled, have performed testing on a

specific clinker to find the optimum ball size based on the
size of the clinker particles present to be ground at a
3 - 35
particular spot in the mill. His laboratory results are

summarized as follows:
What this data says is that for a given size of material to be

ground there is an optimum ball size for a given material
and mill characteristics. This ball size decreases with the
particle sizes to be ground.
The reasons for the optimum are as follows. If the balls are
too small then they don't have the inertia to properly nip the
material and break it. For example; Grits (or spitzers) in the
second compartment, the cascading small balls don't hit
with enough force to fracture the grains so they are worn
smooth at a slow rate and retained in the mill, causing
material transport problems.
If the balls are bigger than the optimum for the feed size
then the lower number of contact (nip) points reduces the
3 - 36
specific breakage rate. Note that if a particle size is

smaller than either the optimum for both the smaller or
larger ball the specific breakage rate of the smaller ball is
superior, this is why finer balls in the second compartment
work well. All balls must be bigger than the optimum, (for
example, according to Bombled, a 1/4" ball would optimum
for 45 µm particles as these balls wouldn't stay in the mill,
this size is obviously impractical). Generally, the
inefficiency of a small ball isn't as bad as a larger one on
that size particle.
3.1.6 Ball Mill Liners
In general balls will, to some extent, cascade within a

rotating mill; even with a perfectly smooth interior surface.
However, the cascade action appears to depend on the
percent of critical speed and the percent volume loading.
Moreover the cascading action also depends on the
grinding media sizes and the nature of the material being
ground as well.
The purpose of mill liners is not only to protect the mill

shell but they must also "grip" the grinding balls and
release them at the right height to obtain the cascading or
tumbling that maximizes the grinding rate. If the balls are
carried too high they will get thrown against the mill liners
on the opposite side where the balls or liners may break. In
addition, the grinding action could be limited to the toe of
the mill load. If the balls are not carried high enough then
the impact energies while tumbling are greatly reduced,
thus retarding the grinding action. In addition, recognize
that some sliding contact will occur which increases liner
wear.
3 - 37
Liner design is critical to good grinding action and wear life.

Careful consideration must be given to ball charge design,
material load and the grinding action required in that
compartment in order to design the appropriate liner. (See
also section on mill internal parts)
3.1.7 Mill Sweep Influence on Grinding
The mill air sweep (ventilation) affects the transport of

material through the mill. Basically, as the mill air sweep is
increased, the retention time of material in the mill is
shortened. Increasing the mill air sweep in the finish ball
mills will provide for the removal of fines from the mills as
they are created. This serves to unload the mill and
prevents the fines from being overground.
Moreover, the removal of fine particles from the mill as they

are created and enter the mill air sweep has a positive
influence on the size reduction function of the mill. It has
been shown that as a mill material load becomes finer, it
becomes more difficult to accomplish additional grinding
work. Therefore, removing fines promotes additional
grinding work. Also, fine particle removal inhibits ball
coating. Therefore, for several reasons, high ventilation
provides for a more efficient size reduction. In fact, in very
high sweep conditions such as those found in Demopolis
the better grinding allows for a finer ball charge, (according
to Slegten).
Other effects could be summarized as follows:
a) Higher mill sweeps will cool the mill requiring less

water spray (less chance of hydration), but depending
3 - 38
on spray configuration it may suck the spray onto the

discharge grates and plugging them.
b) Higher mill sweeps will reduce the quantity of

superfines and may lead to longer setting times and
lower 1 and 3 day strengths.
c) Increased sweep on mills will also vent moisture and

lowers humidity. This is an important consideration
on raw mills and for plants grinding masonry cements.
3 - 39
3.1.8 Mill Mass Transport and Bypass
If the mass transfer of material through the mill is inhibited

by plugging, loss of grinding aid, loss of mill ventilation, or a
ball charge which is too fine (too many small balls),
backspilling can result. This can limit the system
circulating load to a suboptimal level, which can cause
overgrinding in the mill and suboptimal production.
Mass transfer in a ball mill is a very complex process. The

following conditions contribute to the mass transport
function:
A) When the mill rotates, the ball motion makes the

material behave very much like a fluid.
B) When the material begins to flow, it becomes viscous;

and, thus, its movement is hampered by any fixed
surface.
C) The ball charge itself stands in the way of material

flow. It is the key factor in loss of head in the transfer
process.
D) The flow of the material through the mill is due to a

head of material created by the feed to the mill.
E) As the total mill feed increases (fresh feed plus

circulating load), the amount of material in the mill
increases and the ball load expands to a certain limit.
Beyond the limit, mill bypass occurs. Therefore, for a
given ball load, there is an optimum quantity of material
that the dynamic (in motion) ball load can contain.
3 - 40
A) Ball Charge Permeability and Retention Time
As mentioned previously, the mill ball charge is the major

factor in loss of material head or resistance to material flow
in the mill.
Big balls have a low specific surface area (Ft 2/Ft3) and,
therefore, offer little resistance to material flow; while
small balls have a higher specific surface and are
responsible for most of the resistance. For example, 5/8"
diameter balls have about six times more surface per Ft 3 of
balls than do 3 1/2" diameter balls.
Therefore, the second compartment of a ball mill offers

significantly more resistance to material flow than does the
first compartment. Thus, it also has a greater influence on
the mill retention time.
3 - 41
The use of a finer ball charge results in a higher mill

retention time and a higher mill material level for a given
mill throughput (fresh feed and clinker). As stated
previously, mill bypass begins to occur past a certain mill
material loading. Therefore, a mill with a fine ball charge
cannot handle a high mill throughput without operational
problems.
Tests carried out at the Lafarge Plant in St.Constant in

Canada, utilizing the same mill with two different ball
charges, have shown that for an equal mill throughput the
finer ball charge had the higher mill retention time.
Mill throughout could be easily increased with the coarser,

more permeable ball charge. But with the finer, less
permeable charge, mill bypass was encountered at a mill
throughput of 280 TPH.
These experiments proved that a permeable ball charge is

required in the second compartment of a finish grind mill to
accommodate high mill throughput and high circulating load
levels.
Influence of the ball charge on the material load
3 - 42
It is also worthwhile pointing out that mill retention is

highest at low throughput and decreases as throughput
increases. The shortest retention time is when bypass
occurs.
3 - 43
B) Material Filling Ratio
Related to permeability is also the material filling ratio,

(some will calculate the "steel to clinker ratio" which
expresses the same concept but produces different
numbers). Production is also affected by the material filling
in the ball charge. Slegten and others sell expensive
adjustable partitions for just this purpose. Ciment Lafarge
(France) has expended a lot of effort on first compartment
charge design and slot fiddling to get just the right level of
material in the charge.
The following graph plots the breakage rate against

material filling ratio for various ball volume loads.
Material filling ratio is = the volume of powder in the

mill
the volume of the voids in the charge
3 - 44
At low material filling ratios there is a reduced number of

ball-to-material contacts, due simply to an absence of
material in the nip points, therefore breakage rates are
reduced. Also the dynamics of breakage are modified as the
breakage occurring is preferential on the fine material,
resulting in overgrinding of the fines. At the same time, the
coarser particles may migrate into the emptier voids. This
explains the tendency observed by some plants of low 325
mesh coupled with high blaines when a mill is under fed
severely.
As the amount of material is increased the collision spaces

between the balls are filled and higher rates of breakage
are obtained. When all the effective spaces in which
collisions between tumbling balls are occurring are filled
with powder the rates of breakage reach a maximum.
Further increases in material does not give increased
3 - 45
breakage as the collision zones are already saturated. The

excess powder simply reservoirs inside the charge. A
plateau of almost constant breakage occurs, between 60%
and 110% of the calculated void volume. (U = 0.6-1.1).
Overfilling the charge expands the load and the collisions

become cushioned, reducing breakage rates. Of course,
bypass occurs at ratios above 1.1.
It is clear that an optimum is found between 60% and 110%

of the ball charge voids. The lower end of the optimum
should be avoided as the low material filling results in
increased ball-to-ball contact and the resulting higher ball
wear. Therefore mills are usually run at the higher end of
the range, between 90% and 110% of the voidage. This is
evidenced in a crash-stopped mill by material at, slightly
above (2-3"), or slightly below the ball level.
3 - 46
Steel to Clinker Ratio
Although the idea of material filling ratio is easy to

understand it is extremely impractical to measure. Instead
it is easier to calculate the steel to clinker ratio (sometimes
called C/M ratio, for ball charge / material weight). The
ratio is the total grinding media weight divided by the total
material load weight at any instant in time. Material load is
calculated by:
Material Load = Feed Rate X (1+ Circ. Load/100) X Mill Ret.

Time/60
(where feed rate is in tph)
3 - 47
Steel to clinker ratio of 8 roughly corresponds to material

filling ratios of 1.1 and a steel to clinker ratio of 12
corresponds to a filling ratio of 0.6. Bypass occurs when
the steel to clinker ratio is less than 7 and the mill is badly
underfed with steel to clinker ratios greater than 12.
3 - 48
3.2 Grinding Media
3.2.1 Ball Types and Wear
There are two major types of grinding media (balls) used in

North America. The least expensive type of balls are high
carbon steel forgings, usually called forged steel balls. The
other type found in most cement finish mills are the high
chrome white iron castings, normally referred to as cast or
high chrome balls.
The ball volume load can change because of the ball wear.
Wear is due to abrasion, which results from contact
between the ball and the mill feed, between the ball and the
liners, and between the ball and other balls. Abrasion
depends upon such factors as the amount of mill feed, mill
speed and diameter, the material characteristics, the ball
size, and chiefly the ball metallurgy.
In general, cast balls last longer than forged balls.
Typical figures are as follows:
High Wear Resistant Balls: .07 - .15 kg/MT of Cement

(Cast) .14 - .3 Lbs/Ton
Ordinary Balls: .2 - .4 kg/MT of Cement

(Forged) .4 - .8 Lbs/Ton
However recent studies and plant experiences suggest that

for very abrasive materials (more commonly experienced in
the mining industry) there will be a smaller to no difference
in wear rates between the the two types. Wear rates will
vary from plant to plant and given the traditionally higher
3 - 49
cost of chrome media, some experimenting is

recommended before switching from one type to another.
In general, it is not good practice to mix forged and cast

balls, or to mix balls of differing metallurgy. Harder balls
will cause softer balls to wear more rapidly than is usually
expected and cause the overall charge gradation to
deteriorate.
Ball destruction, that is, deformation, splitting, or spalling,

can occur as a result of a number of factors. Most ball
problems occur as a result of metallurgy and/or heat
treating problems. Various materials require balls of
various metallurgies and hardnesses. Different ball
metallurgies may be required for raw and finish grinding, for
instance. Also, different ball metallurgies may be required
in raw grinding, depending on the hardness of the materials
to be ground or if the process is wet or dry.
Ball splitting is normally the result of a ball hardness

problem or from problems with the forgings or castings,
such as voids inside the ball. Most hardness related
splitting is a result of balls being too hard, especially if the
smaller balls in a compartment are harder than the large
balls. Then, the smaller balls will tend to split.
Ball deformation is usually a result of too soft of a

metallurgical composition, or a heat treatment problem in
the entire charge or in isolated balls. Spalling is most often
a problem associated with the forging and casting of balls,
such as voids or remaining oxides in the metal.
Interestingly, recent studies show that minor ball
deformation (i.e. the ball is slightly out of round) does not
affect grinding efficiencies. In other words researchers
could not detect a production difference when using a ball
3 - 50
charge consisting of perfectly round balls versus a charge

with imperfections. However most plants do agree that
when balls become squared off, badly deformed or
concaved, it's time to change the charge.
In all cases, ball manufacturers should be able to help

troubleshoot ball problems and should stand ready to
correct problems resulting from the ball manufacturing
process or metallurgy specification.
3 - 51
3.2.2 Ball Coating
In dry process grinding, ball coating can occur in raw as

well as finish grinding. Ball coating can be a result of the
following conditions:
1. Inadequate Grinding Aid.
2. Poor Mill Ventilation.
3. Mill Material Overheating.
4. Too Much Moisture Input.
5. Mill Overloading.
In all cases, mills must be adequately ventilated. Mill

ventilation can serve two purposes to reduce ball coating in
mill temperature problem situations. It can take heat away
from the mill as it is created by the grinding, and it can take
water vapor away from introduction with the material or
from water sprays which provide additional cooling.
3.2.3 Ball Charge Design
The matching of grinding media size to particle sizes in the

correct location to obtain the most efficient grinding is
what ball charge design is all about. In the view of many it
seems to be more art that science, although this seems to
be changing rapidly. However the design can profoundly
affect production rates and quality significantly and should
not be taken lightly. It will require, in all cases, a lot of
follow-up, patience and detailed record keeping to find the
correct design.
3 - 52
Objectives and Design Considerations
In general, the ball charge design must produce the

maximum output of different types of optimum quality
cement from the mill department.
Of course different types of cement must be considered,

and the charge should be adjusted to the type most used.
To do this:
• Adjust the mill charge porosity, or permeability, to the

amount of circulating load chosen to be the best. In
other words, the type of separator has to be taken into
account.
• Adjust the charge to the material particle size. This is a

well-known theory, but difficult to achieve in practice.
• Adjust the level of charge, or volume loading, thought to

be most effective.
As well the emphasis on these considerations changes

slightly with the type of partition used.
Mills with Intermediate Double Partitions Equipped with

Lifters
For these, the most effective means of adjusting the level of

the material is the partition, which controls the material
flow rate through the mill. The permeability of the charge
in the first compartment is not the essential criteria for the
3 - 53
determination of the ball charge, if the partition is

adjustable.
Here, the requirements are to:
• Obtain a continuous change in material fineness.
• Use the available power most efficiently.
• Take into account the different action of the

charge with different products.
• Make the maximum use of the charge with a

minimum of wear.
3 - 54
Ball Volume Loading
A basic question asked is "what is the optimum ball

loading?" There are two answers to this. One is related to
maximum production from a mill and the other to minimize
kWh/t.
a. Maximum Production
The maximum production of a mill will occur at up to 40%

volume loading, depending on the given mill's drive power,
trunnion design and partition design. This is shown on the
following graph from Austin's Process Engineering...Ball
Milling.
This graph shows that maximum production occurred at

40% volume load. However, production fell only slightly on
either side of the optimum. At 35% and 45% volume load
3 - 55
production fell only 1.5%, therefore since we know that at

35% the mill would draw significantly less power this would
improve the kWh/ ton.
3 - 56
b. Minimum Grinding Energy (kWh/t)
In general, the specific grinding energy is the worst

(highest) at around 40% volume load. On the other hand,
the specific grinding energy is lowest in the 15-20% volume
load range.
In a review of 11 mills (Energy Input for Cement Grinding,

von H.-G. Ellerbrock and B. Schiller, Dusseldorf, ZKG - Nr.
2/1988) the authors showed the following results. They
used kWh/t of mill throughput as a bench mark given that
they were grinding to a mill exit of 1800 cm 2/g. All mills
were running at 55% to 75% of critical speed.
3 - 57
Once more a volume load of 25% or less was optimal for

minimum specific grinding energy. The authors go so far a
to recommend a 26% volume load as optimal.
However, there are other considerations one must take into

account. Firstly, liner wear is accelerated at low volume
loads. Secondly, the capital costs are increased because
for a given mill power a larger (longer) mill will be used to
produce the same output. To build mills running at 20-25%
volume loads would mean very large mills for the power
they consume and the cement they produce. In a given mill
system the production will drop if the volume load is
reduced.
c. Conclusions
Therefore if power is cheap, as it has traditionally been in

the U.S., then a small, low cost mill with high volume loads
is traditionally used to obtain maximum production, but at
poor specific grinding energies. In Europe, where power is
expensive, the mills are larger, higher cost units running
with reduced volume loads and better specific power
consumptions.
The conclusion is that a mill should be run with the

minimum charge (over 26%) that will allow production
targets to be met, if power costs are a factor. If maximum
production is the target, load the mill to the maximum
allowable (below 40%) by power available and trunnion and
partition designs.
d. Rules of Thumb for Volume Loading
3 - 58
In Europe experience indicates that the best volume

loading for cement mills is:
• 1st compartment : 30 to 35%
• 2nd compartment: 28 to 32%.
Other recommendations:
Slegten: 34%
Fuller: 40% (only if the mill allows)
A mill compartment can be overfilled if the primary need is

mill production rate. The upper limits then are the
maximum absorbed power allowed by the drive, and the
maximum level of the grinding charge with respect to the
trunnions and to the central partition vent opening. The
lower limit is determined by the amount of wear on the balls
and liners for a given level of grinding charge.
Porosity or Permeability
Europeans have used simple calculation as an index to
relate a given ball charge's porosity. It's called the average
ball weight. One simply calculates the total weight of balls
in a given compartment and divide by the total number of
balls. This is a useful index in designing ball charges
because one can clearly see the changes in porosity at the
same time one changes the average crushing force; to
ensure that the ball charge is not too porous or too
restrictive.
WARNING: Average ball weights can be calculated using

suppliers data or by sampling. However values can vary. If
you are calculating and comparing average ball weights be
sure that you are using the same data otherwise you will
get misleading results.
3 - 59
All average ball weights given in the handbook are

calculated using the following table:
Diameter # of Balls /mt Diameter # of Balls /mt

mm inches
90 334 3.5 350
80 474 3.0 570
70 709 2.5 990
60 1125 2.0 1920
50 1946
In North America, average ball weights for the first

compartment varies considerably from 1.3 kg/ball to 1.7
kg/ball, (with a few exceptions) and averages at about 1.5
kg/ball. In general, plants with soft clinkers use a smaller
average ball weight and plants with harder clinkers use a
larger one. Ciment Lafarge says that mills with adjustable
partitions can use a coarser first compt. ball charge and
their average ball weights vary between 1.85 and 2.3
kg/ball. Moreover recognize that by reducing the average
ball weight tightens the ball charge increases the mills
tendency to overfill and backspill. For example one plant
had a 1st compt. avg. ball weight of 1.31 kg/ball and
increased it progressively to 1.58 kg/ball to reduce
backspillage.
First Compartment Ball Charge Design
The General Objectives
The sizing of the grinding charge in the first compartment is

determined by three constraints:
3 - 60
• The highest crushing force needed which corresponds

to the number of large balls 80 to 90 mm (3 1/4 to 3 1/2
in) required.
• A charge porosity such that the level of material is

equal to that of the grinding charge.
• The sizing is an “equilibrium charge”, that is a charge

where the porosity is little changed during the life of
the balls, providing that the largest size balls are
added as needed (90 mm, 3 1/2 in).
Equilibrium Ball Charge Design
Developed for the mining industry, this charge design works

very well in mills, grinding extremely abrasive materials. It
is possible to design the charge such that the charge will
maintain the exact same size distribution, permeability and
surface area as the balls wear down from maximum to
minimum (almost nothing) and provided one tops up the ball
charge regularly with just the largest size. Such a design
works very well where wear rates are extremely high and if
the mill has only one chamber. It minimizes the need to
dump, screen and sort the ball charge. However in the
cement industry most mills are partitioned due to the very
different grinding actions required and thus equilibrium
charge formulas (if used) are usually applied to just the first
compartment. Consequently such first compartment
designs for cement mills are never true equilibrium charges.
It might be better if we were to call them "truncated"
equilibrium ball charge designs. An example truncated
equilibrium charge would look like:
3 - 61
3.5" Diameter 28.7% by weight

3.0" 43.7%
2.5" 27.6%
average ball weight = 1.6 kg/ball
However equilibrium charges are difficult to calculate.

Instead Ciment Lafarge uses a simple method (promoted by
Slegten, Polysius will give a similar result): Determine the
amount 3.5" needed to crush, then keep the total number
(not the % weight) of balls in each of the remaining sizes
the same. This approximates an equilibrium charge. This
method is used to determine the starting ball charge design
but then it is adjusted (over the years) to optimize
performance.
3 - 62
Discussion: Which is Better?

If we held the average ball weight the same we can then
compare the traditional design vs the method Lafarge
prefers:
Comparison of Starting 1st Compt. Designs:
3.5" 3.0" 2.5"

Traditional 45% 30% 25%
Lafarge 39.5% 38.4% 22.1%
average ball weight = 1.74 kg/ball
From the comparison there are a couple of observations we

can make:
a) The traditional design is slightly coarser (1.74 vs 1.5

Lafarge North American average, kg/ball). Given the
large percentage of 3.5" balls it was designed
probably to ensure good crushing and no backspillage.
Unless the plant has very hard feed materials the
traditional design will likely "over" succeed and will
wear balls faster (from too much steel to steel contact
and higher impact energy) and show very little
grinding for it.
b) The traditional design also has much more 2.5"

diameter balls than the Lafarge design. Therefore one
can expect the traditional design to choke itself off
faster. This may be one of the reasons why the
traditional design is coarser to begin with.
3 - 63
c) The Lafarge design will likely give a longer useful

charge life and a more effective and uniform grinding
performance over that life.
3 - 64
Here are some starting designs using the method preferred

by Lafarge.
Ist Compt Starting Ball Charge Designs

(%'s are by weight)
Avg. Ball Wt. 3.5" 3.0" 2.5"
1.75 kg/ball 40.71% 37.63% 21.66%
1.70 36.21% 40.48% 23.31%
1.65 31.44% 43.51% 25.05%
1.60 26.36% 46.73% 26.91%
1.55 20.97% 50.16% 28.88%
1.50 15.21% 53.81% 30.98%
1.45 9.05% 57.72% 33.23%
Caution: the designs listed in the table are not optimum

designs. Trial and error modifications will be required to
determine what best suits the needs of the individual mill.
Second Compartment Ball Charge Design
Ball Charge Size Distribution in the Second Compartment
In general, there are two design models used by Lafarge;

Polysius and Slegten. Equations and rules to follow will
generate a design curve; ball size versus mill length.
3 - 65
Many references are made to these models when designing

a ball charge. The actual equations are tedious and
laborious to calculate and therefore are not discussed here.
In general, the sizing of the charge in the second
compartment is determined by:
• Crushing work left over from the first compartment.

• Sufficiently fine to be effective.
• Sufficiently dense to retain the material.
• Allows the separator to function correctly.
• Discourages agglomeration (coating).
EXAMPLE: Bath FB Mill 2nd Compt. (with O'SEPA)
2.5" diameter 7% by weight

2.0" 7%
1.5" 7%
1.25" 8%
1.0" 21%
0.75" 49%
3 - 66
(modified Slegten Design, also shown on the comparison

graph)
3.3 Ball Mill and Circuit Evaluations and Tests
3.3.1 Mill Material Levels
The level of material found in each compartment of a ball

mill following a crash stop, can provide information on how
the mill is being operated. In general, if a significant depth
of materiel is discovered above the balls in any
compartment, then mill overloading and mill bypass, or
other mill problems can be suspected. In most mills, only a
few inches of material should be found above the ball
charge.
In the first compartment, there may be a significant amount

of material at the feed end of the mill resulting from a
material flush from the separator and its feed system, upon
crash stopping the mill system. In addition, in most cases,
the material load level in the first compartment will
disappear below the level of the ball charge some distance
from the partition grate wall. This is a result of the drain
action of the partition, together with a low resistance to
material flow offered by the first compartment charge.
If a high material level is found above the ball charge in the

first compartment of a mill, then either the first
compartment charge is too closed (too fine), or the partition
is not functioning properly. The efficiency of the partition
can be affected by the following factors:
(1) Coarse material grate slot plugging

(2) Grinding media grate slot plugging
(3) Grate slot peening
3 - 67
(4) Partition internal plugging

(5) Improper adjustment if the partition is adjustable
If the partition grate slots are plugged with material, either

the first compartment charge is inadequate (volume load or
gradation) or the mill has been overloaded with material or
moisture.
If the partition grate slots are plugged with spitzers (broken

or worn balls) or other metal, this indicates a media or liner
problem, especially in the second compartment. The
second compartment charge may be badly worn or the
small balls may be breaking.
If the partition grate slots are peened closed, this indicates

a metallurgy problem with the grates or the balls or both.
Either the grates are too soft or the balls are too hard
relative to the grates.
It is unlikely that a mill partition will become plugged

internally unless there exists a partition water spray with a
problem. An adjustable partition may require adjusting if
the material feed to the mill changes.
If none of these possible problems exist, the the mill is

either overloaded, or material is backing up from the
second compartment. In this case, a high level of material
above the ball charge will be found in the second
compartment.
If a high material level is found in the second compartment

of a ball mill, then either the second compartment charge is
too fine, the grates are almost totally closed, or the grates
are plugged as the result of a mill discharge water spray
problem.
3 - 68
Nearly all of the aforementioned problems will be precluded

by mill backspilling or puffing.
The other material loading problem, of course, is when a

mill or a compartment is underloaded or seemingly void of
material. This can result from a mill or compartment
charge being too coarse for the system surrounding it. In
other words, the mill system conveying equipment and/or
separator will not accommodate a high enough circulation
load to properly load the mill.
Most often, however, the first compartment of a mill is the

one which will lack material load. In this case, energy is
wasted and suboptimal performance is obtained form the
mill. This is usually the result of a ball charge which is too
coarse to provide resistance to material flow, an improperly
adjusted partition diaphragm, or over ventilation.
3 - 69
3.3.2 Fineness Evaluation (Granulometry)
The fineness of material along the length of a ball mill can

reveal good information on ball charge performance and
size reduction efficiency. The best way to make these
determinations is by axial sampling and particle size testing
on the resulting samples.
Axial Sampling Technique
To obtain representative axial samples from a two

compartment ball mill, the following procedures and
sampling frequencies should be applied.
(1) Five evenly spaces samples should be taken across the

width of the mill at each designated sampling point
along the mill length. Divide the useful width of the mill
at the ball charge level by six. One sample should be
taken at the mill center and two equally spaced
samples on each side of center. The two outside
samples should be one sixth of useful width from the
shell liners.
(2) In the first compartment, take samples every half meter

along the mill axis beginning a half meter from the head
liners.
(3) In the second compartment, take samples every meter

along the mill axis beginning one meter from the
partition discharge wall liners.
(4) In short single compartment mills, samples should be

taken every half meter beginning a half meter from the
head liners.
3 - 70
(5) Samples should be taken from the ball charge and not
from the material on top. This requires some digging
and small sampling scoops in the first compartment. In
the second compartment, a shovel in conjunction with
a a coarse mesh screen and a bucket may be utilized.
(6) To obtain representative individual samples for the

composite, the individual samples should be
approximately the same size before compositing. The
five individual samples taken at intervals along the mill
axis should be mixed thoroughly before splitting for
sieve and/or sedigraph analyses.
The results of the screen and sedigraph analyses should be

plotted for analyses.
Mill Granulometry - Bath FMB T1

80
Outlet
Partition
60
Percent Retained (%)
40 #8 Mesh
#50
#100
20 #200
#325
0
0 5 10 15
Distance (m)
5% Maximum #8 allowable @ partition
This technique is extremely useful in evaluating ball

charges before and after a design change. However, funny
results can occur and are usually traced back to sampling
errors.
3 - 71
3.3.3 Mill Retention Time
This test is performed to help in assessment of material

transport and is necessary information to calculate steel to
clinker ratio, since we must know how much material is in
the mill at any given time. In general, it gives us an idea of
the mill's overall permeability.
Progress of the material through the mill is monitored by

the use of a tracer dye, usually Fluorescein. This is used
because it can be detected in very low concentrations.
Usually about 2g of Fluorescein per tonne of mill production
(per hour) is used.
The dye is mixed with say 5 lb. of mill feed material. Which
should be the appropriate proportions of fresh feed and
separator rejects. The mixture is slurried with methyl
alcohol to ensure good dispersion of the dye and the
alcohol is subsequently evaporated by heat.
The dried slurry is dumped into the mill feed chute,

following which the mill discharge is sampled every minute
for 15 to 30 minutes (depending on mill characteristics).
Each sample is analyzed for fluorescein concentration,
using extraction and measurement with a fluorometer.
(send samples to CTS) A graph is plotted of concentration
versus time and the maximum (peak) obtained is the
retention time.
3 - 72
Fluorescence
Open circuit
of sample
Closed circuit
Time (min.)
3 - 73
3.3.4 Circulating Load
B
Feed
E
C ts
j ec F
Re
Finished
D
Product
Fresh A
Feed
First ensure that the mill circuit is stabilized at normal

operation conditions. This means points such as elevator
KW, mill sound etc, are achieved. Three people with
synchronized watches are required for the sampling. Agree
on a time to start sampling, allowing sufficient time for
each person to reach his sampling location.
At the given time, begin sampling at one minute intervals

for at least 5 minutes. Separator Feed can be sampled at
location a) or b) as shown in the diagram. Rejects can be
taken either in a gravity discharge pipe (from separator) on
in an airslide if one is present (point cord). Finished
product can also be collected in the separator discharge
pipe or an airslide (locations E or F).
Each set of samples must be thoroughly blended before

performing sieve analysis or particle size distribution
testing (sedigraph or laser).
3 - 74
Be sure record all necessary mill parameters during the

sampling period.
3 - 75
3.4 Ball Mill Control
3.4.1 Basic Ball Mill Control Theory
From the plant Production department's point of view, we

are interested in keeping the mill running as smooth as
possible without overloading the mill. Recognize that an
overloaded system generally makes a mess of the plant and
increases the likelihood of something breaking.
Essentially, plants strive to maintain the total throughput

rate (fresh feed plus separator rejects) at an "ideal
constant". Of course the ideal throughput will change as
conditions alter, such as circulating load, feed
characteristics, etc.- to suit quality targets and
requirements.
For many years, technology allowed us only to approximate

this indirectly using mill motor kW, mill sound or elevator
kW. Today, with the advent of newer weighfeeder
technologies we can now directly measure total throughput
with much better accuracy and consistency. This has led
to better mill control but not all plants can justify installing
an impact flowmeter. Thus some plants using existing
instrumentation have resorted to rule based control
programs (fuzzy logic) to respond to system changes
caused by feed material changes for example. Many plants
also have developed their own schemes by blending two or
more control methods together, and have operated this way
successfully. In all cases though mill control requires a lot
of study and persistence to perfect.
3.4.2 Mill Motor kW Control
3 - 76
In general, the mill motor kW will vary with mill feed, but it
is non-linear. It varies in the following way:
1) With the mill at rest and completely run out of feed

material, imagine the total weight of the ball charge
(W) to be concentrated at one point. This point is
called the center of gravity and is located a certain
distance from the mill center. As the mill turns, the
center of gravity becomes slightly offset from the
mill's vertical centerline by a distance "m", sometimes
called the moment arm. W X m = the torque required
(excluding friction) to turn the mill.
To illustrate, let's say that we've installed a badly

designed mill liner which causes the charge to climb
higher, then the moment arm "m" gets longer and
torque and therefore mill power increases. On the
other hand let's say that the liners are badly worn and
the charge slips down. In this case "m" becomes
shorter and torque and hence mill power decreases.
3 - 77
Motor kW Control Theory
Mill Rotation
Center of Gravity
Torque = W X m
m
2) Going back to the original case, and let's begin to add

feed. Under these circumstances, the voids in
between the balls that started out empty now begin to
fill. "W" increases since we are adding more mass to
the ball charge without increasing its volume. "m"
doesn't change therefore torque and mill motor kW
increases.
3) However, at a certain point as we continue to add

feed, the voids become completely full and the ball
charge starts to expand. At a critical point the balls
are pushed far enough apart that they lose contact or
"grip" with one another. Consequently the whole ball
charge has a tendency to slide down. This shortens
the moment arm "m" which reduces torque and mill
power despite the fact that we are continuing to add
3 - 78
feed. (Actually one must also remember that as one

adds feed, mill retention time decreases which
technically will hold "W" approx. constant.)
KW
Ball Charge Filling Ball Charge Expanding
A B TPH
Two production rates for the same motor kW value
From the graph we can see that there is one major problem
or flaw with mill motor kW to control mill feed. For a given
mill motor kW setpoint there are potentially two feed rates
which can cause a single loop controller to hunt or
oscillate. In addition, the kW value will change over a
relatively narrow range with a corresponding large change
in feed. This coupled with the difficulty in calibrating kW
meters for large HP motors accurately makes this control
scheme very difficult to tune. This type of control is never
used by itself but occasionally it is used with another type
of control loop.
3 - 79
3.4.3 Mill Sound Control
The basic principle of mill sound control is simple. A

directional microphone is used to pick-up sound generated
by the grinding media tumbling inside a mill chamber. When
the mill is empty of feed, metal to metal contact is at its
highest and therefore the microphone will record the
loudest decibels. As the mill fills, the cushioning effect
deadens the noise levels. In theory a mill that's completely
plugged such that grinding media cannot tumble will
produce no noise at all. Some plants report this value as
decibels. Others reverse the signal and express it as %
level or % full.
In general, mill sound is useful in determining whether a

given compartment is plugged or plugging or to show that
the compartment is emptying. With each compartment
equipped with microphones, mill sound is very useful in
monitoring and troubleshooting a cycling mill. Some plants
successfully use the 1st compartment sound to control mill
throughput.
However mill sound is very imprecise and is not considered

to be very repeatable (for example, 65% level which
corresponded to a backspilling condition one shift may not
repeat itself the next shift). Microphones can pick up noise
from other mills which can impair it's reliability.
Furthermore the microphones are easily damaged as well
dust can affect its performance. All of these things
generally makes mill sound microphones difficult to
calibrate. Despite these problems, a well isolated and
maintained microphone can be made to work well.
3 - 80
3.4.4 Discharge Bucket Elevator Motor kW
On most mill circuits, mill product is fed into a bucket

elevator which transports the product to the separator feed.
As the mass flow rate increases so does the motor load on
the bucket elevator, since it now has to move more
material. Many plants use this fact to control their mill
since, in most cases, elevator kW's fairly reflects the mill's
total throughput. In fact total throughput can be estimated
using this formula:
M = (Ka-Ke) x 3600 x E / 9.81 x H
where:
M is material flow in mtph
Ka is actual power measured in kW
Ke is power measured with elevator empty in kW
E is elevator efficiency
H is the inter axis height in meters
(Note that this give an approximate answer since the

above values will change with mechanical wear and
the amount of recirculation and/or boot digging that
occurs. Efficiency should be rechecked after each
major overhaul.)
For the most part elevator control works well. However

there are two major difficulties with this type of control:
a) Elevators are volumetric devices and are therefore

very susceptible to changes in bulk density or
flowability which leads to the boot overfilling. Many
mill bucket elevators operate with the boot full or
overfilled. (In such cases the elevator was probably
3 - 81
designed for mill product whose bulk density was

estimated at 90 to 100 lbs/ft3 but whose real bulk
density can be as low as 30 to 40 lbs/ft 3.)
Consequently any fluctuations in grinding aid, internal
water spray or airslide aeration can dramatically
change the product bulk density causing the boot to
overfill. The resulting digging and possible
recirculation that occurs alters the elevator kW's but
with no real change in mass flow. For example in a
few plants whose bucket elevators are fed with
airslides, reducing the under canvas pressure from 20
inwc to 10 inwc reduced motor kW by 8 to 10 % with
no change in flow.
b) Bucket elevators are located after the mill and

therefore controls will always experience a lag time.
Elevator controls should work well in short mills or
those with very short retention times, but will have
less success with long mills or long retention times.
Moreover since these are feedback loops often they
cannot detect 1st compartment plugging problems.
For example, suppose a mill is beginning to plug
somewhat in the mill's 1st compartment due to a bin
segregation problem. In such cases the second
compartment will start to empty which is detected by
the elevator kW control. However it assumes that the
mill is not getting enough feed therefore starts adding
more feed to maintain a setpoint. If the operator fails
to notice this then the mill will either backspill and/or
begin to cycle.
Most plants over come this difficulties by using more than

one type of control method together.
3 - 82
3.4.5 Rejects Flowrate
Ideally to know the true mill throughput one should be

measuring all streams entering the mill. By summing the
fresh feed rate and the rejects flowrate from the dynamic
separator one will know the throughput with no guesswork.
A flowmeter on the separator rejects will allow you to
determine immediately the circuit's circulating load, the
effects the separator adjustments has on the mill circuit
and evaluate the whole circuit retention time(s). However
it should be used in conjunction with other instruments
previously mentioned since total flow will tell you very little
about internal mill problems.
This is perhaps the most expensive control device to install

since in most cases it requires mechanical adaptations to
the system.
3.4.6 Rule Base Mill Control
As we have seen in the previous sections, each control type

has its own set of difficulties - to some degree depending
on the mill. Already many plants utilize more than one type
of control together, depending on the limitations of the
electronic equipment used. The most elaborate
combination would be to install a Rule Base control
scheme, which is a table of decisions a computer can take
for each possible combinations of conditions that could
occur. Such a table properly set-up will mimic the
responses a human operator should make when confronted
with the same set of conditions. Fuzzy logic type systems
operate on a similar principle. The following is an example
from Demopolis:
3 - 83
Elevator Mill #1 Compt #2 Compt Feed

KW KW Level Level Change
1 low ok ok low add
2 low ok ok ok add
3 low ok low ok add
4 low ok low low add
5 low high ok ok add
6 low high ok low add
7 low high low low add
8 low high low ok add
9 ok high ok ok add
10 ok high low ok add
11 ok high ok low add
12 ok high low low add
13 ok ok low low add
14 ok ok ok low add
15 ok ok low ok add
16 high ok ok high sub
17 high ok ok low sub
18 high ok ok ok sub
19 high ok low ok sub
20 high ok low high sub
21 high ok low low sub
22 high ok high high sub
23 high ok high low sub
24 high ok high ok sub
25 high low ok ok sub
26 high low ok low sub
27 high low ok high sub
28 high low low ok sub
29 high low low low sub
30 high low low high sub
31 high low high high sub
32 high low high ok sub
33 high low high low sub
34 ok low ok high sub
35 ok low ok ok sub
36 ok low ok low sub
37 ok low low ok sub
38 ok low low high sub
39 ok low low low sub
40 ok low high high sub
41 ok low high low sub
42 ok low high ok sub
43 ok ok high ok sub
44 ok ok high low sub
45 ok ok high high sub
3 - 84
Notes:
a) For each parameter there are 3 conditions (high, low,

ok). Then the number of possible combinations will
equal 3x, where x is the number parameters used. For
example for 3 parameters there are 33 = 37
combinations. For 4 parameters there are 3 4 = 81
possible combinations. For 5 parameters there will be
35 = 243 combinations, and so on. In the above
example there should be 81 combinations where only
45 are shown. Recognize that for the remaining 36
combinations that no action was deemed necessary.
b) Each combination must be studied in detail to

determine whether feed should be added or
subtracted (or no action's to be taken) and by how
much. Clearly certain combinations will call for a
stronger action than others. As well one must also
define what is high, what is low and what is OK.
c) Rule base controls work on a regular intervals. In

other words the rules are evaluated once each
interval and picks one action to take (including "do
nothing"). In between evaluations the program does
not make any moves. Intervals can be adjustable to
determine the appropriate frequency (ie. every 30
minutes or 25 minutes, etc.).
d) For each parameter there should be an adjustable

deadband. For example on paper strip chart a pen will
be erratic within a certain band but the average
continues to trend up or down. The width of this band
varies depending on mechanical wear, instrument
fatigue, etc. The deadband takes this into account,
thus only when the signal is outside of the deadband
is the parameter considered to be high or low.
e) Fuzzy logic or AI (artificial intelligence) systems work

on a similar principles.
3 - 85
f) WARNING: a considerable amount of study and tuning

is required to get the program just right. In other
words you must teach it how to operate the mill.
3 - 86
4. Air Separators
4.1 Introduction
Separators are utilized in a closed circuit grinding system

for the purpose of separating the fine and coarse particles
of the feed. The fine particles are passed through the
separator to become finished product while the coarse
particles are returned to the mill as rejects for further
grinding.
4.2 Types
First Generation Separators
These are separators with internal cyclones, e.g. Sturtevant

and Raymond.
Second Generation Separators
These are separators with external cyclones, e.g. Wedag-

Zub and O&K, (mainly in Europe). As a general rule their
efficiency in terms of bypass, fractional separation, and
grinding is superior to the first generation separators.
Some first generation separators can reach the same
performance levels with modifications.
High Efficiency Separators of the Third Generation
3 - 87
These are characterized, other than by external ventilation,

by the presence of a squirrel-cage, which allows an
improved fractional separation over other separator types.
Static Grit Separators
These permit the separation of large particles from the

material being transported in the mill air sweep.
4.2.1 Static Grit Separator
Principle of operation
During the process of grinding in the ball mill, an external

fan drafts fine and coarse particles from inside the mill
through the mill discharge. In process terms, the material
in swept from the mill.
The method of particle separation in static separator is a

very basic process. The dust laden air swept from the ball
mill flows upward into the bottom of the separator cone. As
the air (gas) rises, it flows between the inner and outer
cone until it reaches the inlet vanes. There it flows through
the vanes, imparting a circular motion, into the upper
section of the inner cone. Due to the much larger areas of
the inner cone, the velocity decays and the coarse particles
(grits) fall by gravity to the bottom of the inner cone out the
reject spout back to the grind circuit. In addition the
circular motion of the air (gas) improves the separation of
coarse and fines by centrifuge. The air (gas) with the fine
particles continues up through the thimble out of the
separator to elsewhere in the process.
3 - 88
Static Grit separators are commonly employed on semi air

swept finish mills and on air swept coal mills.
Components
This separator does not compare in design to the

conventional separator. Its design compares to a cyclone
with the exception that internal vanes can be adjusted.
a) Outer Cone
Can be described as the outer shell of the separator. The
bottom of the cone (inlet) receives unclassified material
from the ball mill sweep.
b) Inner Cone
Installed inside the outer cone. The bottom is piped outside
the separator for rejecting coarse (grit) particles from the
separator.
c) Inlet Vanes
Located below the top of the separator, separates the inner
and outer cone. The vanes are adjustable and can be
moved to increase or decrease the circular motion of air as
it enters the inner cone. Typically they are mechanically
linked and are adjusted using a single hand wheel.
d) Thimble
Placed through the separator top (center). Fines and (exit
gas) leaves the separator through this duct. Adjustable to
pick fines from the air flow inside the separator. However,
the thimble is set up at installation and not moved as a
regular procedure.
3 - 89
Static Grit Separator:

Cutaway Schematic Exit Gas and
Fines Product
Adjustable Guide Vanes Thimble
Inner Cone
Modification Idea:
Outer Cone
for high blaine
product mount a
fresh air bleed
valve in front of
each guide vane.
Inlet Duct Rejects Pipe

(from mill)
3 - 90
Control
a) Fineness Control
A change in the product fineness can result of either the
volume of air swept from the mill changed or the separator
vanes are adjusted. Keep in mind that either change will
have effect on the separator efficiency.
• Air Volume
An increase in volume of air swept from the mill results in a
higher percentage of coarse material going through the
separator to the collector - lower product Blaine.
• Vane Adjustment
Each static separator vane adjustment area is marked for
direction of movement. A coarser adjustment (less spin)
will result in the removal of fines to the collector with a
higher percentage of coarse particles - lower product
Blaine.
Modification Idea
Installed in Demopolis, fresh air bleed in valves were
mounted in front of each guide vane. By opening these
valves fresh air leaks in which reduces dust loading
(favouring the superfines), but maintains the same high
internal grit velocities in order to maximize separation.
Inspection requirements
No moving components or power source.
Basic inspection requirement is a visual inspection of the
separator and related intake and exhaust piping for leaks.
Depending on the material abrasiveness the vanes and the
inner cone should be checked for replacement at least
every 4 to 6 years.
3 - 91
3 - 92
4.2.2 Sturtevant Separators
Operation
• Material goes down the feed spout to the intake cone

and to the lower distributing plate hub.
• By centrifugal action (rotation) the particles are thrown

outward through the ports of the hub and onto the lower
distributing plate (protected by the upper distributing
plate).
• Particles are dispersed from the plate into the separating

zone. A curtain or umbrella of material is formed outside
the lower distributing plate.
• Forces acting on the particles are ascending air, gravity,

and centrifugal action.
• Coarse particles settle by gravity to the tailings or

rejects chamber.
• Finer particles are acted upon by the upward air flow

created by the main circulating fan and lifted to the
selective zone where final selection takes place.
• The selector blades impart additional centrifugal force.

Heavier particles are thrown outward underneath the
drum cover to the rejects cone. The finer, lighter
particles are drawn through the path of the selector
blades to the finished product area.
• More selector blades or fewer main fan blades will result

in a finer product.
3 - 93
• Control valves (or diaphragm) are located between

selector blades and main fan blades, and move in or out
to vary the size of the opening between the two.
• Return air vanes (between the fines cone and the inside
drum cone) allows fines to settle, while returning air to
the separating zone.
• Drying in the separator can also be achieved through

balanced hot air inlet(s) and outlet(s) to a dust collector
and fan.
Material Flow Pattern inside a Sturtevant
Sturtevant Separator Adjustment
The possible operating adjustments that can be made on a

separator are, in order of decreasing priority:
3 - 94
• Diaphragm if there is one

• Number of fan or selector blades
• Main fan diameter
or by means of the feed flow rate in terms of the output and

fineness.
3 - 95
The general rules under all circumstances are:
SEPARATOR ARROW DIAGRAM Direction Change of Variable
Bypass or Fines Blaine or

Direction Change of Adjustment
Rejects TPH 325Mesh
MAIN FAN
More Blades (or larger Diameter)
Fewer Blades (or smaller Diameter)
Higher Speed (sheave change)
SELECTOR BLADES
More Blades
*
Fewer Blades *
Higher Speed (for separate drives)
DIAPHRAGM or VALVES *
% Opening
SEPARATOR FEED RATE

(or mill circulating load)
SEPARATOR FEED
PRODUCT FINENESS
* MAIN FACTOR
In the case of a separator with a diaphragm, as a general rule, it should

be adjusted for the different modes of operation (selector fan, main fan)
3 - 96
in such a manner that, during normal operation, the diaphragm will be

near its maximum opening.
3 - 97
Sturtevant
12
17 18
21
11
24 13
9
19
5
10 15
23
22 7
20
16 14 8
6
4
3
1
1. Fines Chamber 9. Inside Drum Cover 17. Intake Cone

2. Tailings Cone 10. Inside Drum Cover Liner18. Intake Cone Line
3. Ring Liner 11. Valve & Valve Rod 19. Fan Cone
4. Air Vane 12. Gear Reducer 20. Packing Ring
5. Outside Casing 13. Main Shaft 21. Upper Dist. Plate
6. Outside Casing Liner 14. Distributing Hub 22. Upper Dist. Plate
Liner
7. Inside Drum 15. Dist. Hub Liner 23. Selector Blade
8. Inside Drum Liner 16. Lower Dist. Plate 24. Main Fan Blade
3 - 98
Air separator arrangement for air drying or air cooling
3 - 99
4.2.3 Raymond Separators
Operation
• Material is introduced into the top of the separator and

falls on the distribution plate. The material is slung
outward into an air stream, created by the fan, and lifted
to and through the separation blades.
• The double whizzers knock down the coarse particles

where they are collected and discharged from the inner
cone back to the mill inlet for regrinding.
• The finer particles are lifted through the whizzers

passing by the selector vanes and collected and
discharged from the outer cone to a transport devise.
• The selector vanes determine how much air and material

is directed through the double whizzer blades. If all the
material were to be directed through the whizzer blades
there would be too many returns. This means higher
production costs. If too little material is directed
through the whizzer blades there is not proper
segregation and the proper amount passing a 325 mesh
screen is not achieved.
• The selector vanes can be adjusted and serve the same

function as the adjustment plates in a Sturtevant
separator.
• Since the system is not of the air through type there is

no need for a large dust collector. The only air
generated is from the transportation system (air gravity
conveyor) and thermal displacements.
3 - 100
• Water cooling jackets are fixed on the exterior of the

outer cone in case the finished product is excessively
hot.
3 - 101
Raymond Double Whizzer Separator

Material Flow and Air Sweep Through The Air Separator
Fan blade
Control
damper
Detector
Selection Air sweep

zone Whizzer blades
Separation
zone Sweep
Guide vanes
Inner cone
Outer cone
Rejects returning to mill
Finish cement
3 - 102
4.2.4 High Efficiency Separators
Principle of Operation
High efficiency separators (HES) are sometimes called

CAGE ROTOR separators. One of the most obvious
differences between and HES and the older Sturtevant and
Raymond separators is that the older machines had internal
fans while the HES has no fan and relies on an external fan
to supply air for separation and transport of material. With
in Lafarge in North America there two types of HES
presently in service. O-SEPA's are built and sold by Fuller
under license from Onoda Cement in Japan. SEPOL's are
built and sold by Polysius. Note also that the major
difference between the O-Sepa and the Sepol is that air is
down-drafted in the Sepol where as it is up-drafted in the O-
Sepa. Operational installations are as follows:
O-SEPA: Bath FMA, FMB

Richmond FM1, FM2
Whitehall FM2
SEPOL: Alpena FM19, FM20, FM21
Classification Process
The only moving part within the separator is the cage rotor.
The rotor is driven by a variable speed motor controlled
from the control room. Clean air enters the volute housing
and is forced to travel in a circular path by the shape of the
housing. The air encounters the inlet vanes which are
arranges in a circle completely around the cage rotor. The
air next enters the classification zone which is the space
3 - 103
between the inlet vanes and the cage rotor vanes. The air
now enters the cage rotor through its vanes. The air then
exits through the bottom of the cage rotor in Sepol's; or
through the top of the cage rotor in O-Sepa's; and then out
of the separator through the exit elbow. Unclassified
material enters the separator at the top and falls by gravity
to the top of the cage rotor and lands on the distribution
table. Since the table is rotating, centrifugal force propels
the material outwards and off the rim of the distribution
table where it impacts a wear ring and begins to fall into
the classification zone.
O-Sepa Details
Classification Zone
Feed
Cage Rotation
Inlet
Plenum Uniform
Flow Dist'n
Guide Vane Direction
At this time, the material encounters the air entering

through the inlet vanes and begins to fall through the
classification zone in a circular path induced by the air
currents. The larger, heavier particles tend to fall through
the circular air currents while the smaller lighter particles
tend to flow with the air currents into the cage rotor and
out of the classifier as finished product. Particles which
3 - 104
are marginal in size may enter the cage rotor vanes but may
also be rejected if their velocity is not great enough to pass
between adjacent vanes without being struck by a vane.
One way to visualize this operation is to consider that there
is specific amount of time between the passing of one vane
and the next vane. If a particle can travel through the rotor
vanes in less time than this passing time, it can penetrate
the cage and flow out of the separator. If the particle is
traveling too slowly, it will be struck by a vane and be
rejected.
3 - 105
Fineness control
Understanding of the previous described classification

process leads to an understanding of the manner in which
the fineness is controlled. Two methods are available for
controlling the fineness:
A) By varying the volume of air flowing through the

separator, the velocity of the air entering the cage
is also varied. As the volume and velocity is
increased, the PRODUCT becomes coarser. As the
volume and velocity is decreased, the PRODUCT
becomes finer.
B) By varying the speed of the cage rotor, the blade

passing time is varied. As the rotor speed is
increased, the PRODUCT becomes finer. As the
rotor speed is decreased, the PRODUCT becomes
coarser.
Table speed refers also to cage rotor speed. The

circulating load for 3300 to 3800 cm2/g is in the order of 100
to 200%.
3 - 106
It is readily seen that product fineness control is much

easier with the HES than with the older types of separators.
In addition, there are no selector blades to change. All
fineness control can be done from the control room.
The finer material that leaves the HES in the air stream is
considered finished product. This product is carried in the
air stream and on to the dust collection system. It is here
that the dust laden air entering the collector is cleaned and
the finish product is collected and transported via air slides
and air lock feeders to the cement pump which pumps it to
the storage silos.
Rejects from the HES are returned back into the system. If
the ball mill circuit is equipped with a roller press, the
rejects are proportionality divided between the two.
Otherwise, all rejects go back to the ball mill to assist in its
grinding capability. These rejects are considered the
circulating load of the circuit.
ALPENA HES DESIGN DATA
FM 19 FM 20,21
Separator horsepower 200 500
Cage rotor diameter (m) 2.0 3.1
Fan HP 300 700
Draft rating (ACFM) 59,000 143,000
Rotor speed range (RPM) 47-280 31-190
3 - 107
Design feed rate (TPH) 165 514
3 - 108
High Efficiency Separator: Fuller O-SEPA
Breakaway View Fuller/O-SEPA Fine Product

(to Dust
Drive Shaft Collector)
CW Rotation
(looking down)
Feed Inlet
(1 of 2)
Dispersion Plate
(Buffer Plates
not shown)
Primary Air
Secondary Air
Rotor Cage Assembly

with Swirl Vanes and
Partition Plates Guide Vanes
Coarse Discharge (Reject)
3 - 109
High Efficiency Separator: O-SEPA Cross Section
Motor
Reducer
Reducer Support
Coupling
Drive Support
Shaft Assembly
Exit Duct (To Dust

Collector & Fan)
Feed
Air Seal
Buffer Plate
Dispersion Plate
Guide Vanes
Partition Plate
Swirl Blade
Secondary
Air Intake
Primary
Air Intake
Tertiary Air
Intake &
Damper
Hopper
Flap Gate
3 - 110
High Efficiency Separator: Polysius SEPOL
3 - 111
High Efficiency Separator: SEPOL Cross Section
Motor Gear unit
Coupling
Motor
support Gear distance
piece
Feed hopper
Distributing
plate Baffle ring
Air guide
plate
Guide
Rotor
vane ring
Upper section
of housing
Lower section of
housing Tailings cone
Flap valve
3 - 112
4.3 Separator Efficiency
Some suppliers and engineers have used concepts of

separator efficiency as related to % of #325 in the feed
which is selected. This single figure is an
oversimplification.
4.3.1 Tromp Curve
In Lafarge, we use a Tromp curve (also called partition or

Selectivity or Fractional Recovery Curve) to evaluate the
performance of a separator. This is simply a plot of the
probability of rejection of a particle versus its size.
A) Perfect Separator
For example for a perfect #325 (45 µm) screen the Tromp or
Selectivity or Partition Curve would look as follows;
3 - 113
Unfortunately this separator exists only in a process

engineer's dreams.
3 - 114
B) Bypass
Now let's assume the screener is lazy and installs a splitter

that automatically "bypasses" 20% of the feed. Instead of
screening it he simply throws it in the reject chute. Now
the partition curve will look like this;
The bypass corresponds to the minimum probability of

rejection. In a real separator it means that a certain
percentage of the feed which is of correct size to be
selected is rejected without a good chance to be selected.
Generally it is felt that these smaller particles get caught in
the boundary layer or wake of larger ones and carried
directly to the rejects.
Reducing bypass by design (high-efficiency separators)

involves increased retention times, distributing the feed
around a large classification zone and low material to air
concentrations in the classification zone to break up the
particle cloud and allow selection of the small sizes.
3 - 115
Reducing bypass in the Sturtevants involves more air

through the classification zone or spreading out the "cloud"
so air can get at these small particles and select them.
C) Imperfection
On top of this, we discover that the screen is worn in some

areas and has build-up on others. So some coarser
particles can get through and sometimes particles finer
than 45 µm are rejected. Now the partition looks like this;
In a real separator the imperfection is likely a result of

somewhat uneven flow velocities in the classification zone
and also non-uniform feed distribution in that airstream.
This means that at some locations due to different
velocities a particle may be rejected at one part of the
separator and selected at another.
3 - 116
D) Secondary Bypass
Finally this crazy guy throws some of the finest of the

selected fines back into the rejects .
The combination of the three "inefficiencies";
• bypass of a portion of the feed stream straight to

rejects,
• imperfection of the cut, and
• return of a portion of the fines back into the rejects,
creates the classic separator efficiency curve shown

below.
In a real separator the outer fines collection cone, or

external cyclones in a second generation machine, are
simply separators where the 'fines' or smallest particles are
3 - 117
'selected'. This selected product is be re-introduced into

the inner cone where much is trapped by the curtain of
rejects and carried back to the mill.
Therefore the Tromp curve of the outer cone works in

reverse where the rejects are actually going to finish
product and the fines are rejected back to the mill.
The three inefficiencies create the classic fishhook curve

so often seen.
E) Example Tromp Curves
For Sturtevant separators typically have high bypasses (30

to 60 %). Higher feedrates or circulating loads tend to raise
the bypass.
On the other hand High Efficiency separators will

significantly lower bypass and the imperfection or cut is
slightly better.
3 - 118
Depending on the Qf / Qa ratio used, bypass on H.E.S.'s vary

between 6 and 15 %.
Typical Imperfection values are as follows (guidelines only):

High Efficiency 0.40
Raymond 0.45 - 0.60
Sturtevant 0.60 - 0.70
4.3.2 Rosin- Rammler Number
Within Lafarge, one will encounter the Rosin Rammler

number. This term relates to the use of the Rosin-Rammler-
Bennett curve. The Rosin-Rammler curve is simply a
mathematical formula which can be made to approximate
most powder size distributions.
Where as the Tromp curve requires a lot of laborious

calculations in order for one to to evaluate the grinding and
separation efficiencies, the R.R. calculation produces one
3 - 119
value which reflects the efficiency of the whole circuit. In

this case one inputs the values for the particle size
distribution into the equation. The steeper the size
distribution; the higher the R.R. number; the more efficient
the grinding circuit is.
For finish cement here are some typical ranges:
0.75 - 0.85 open circuit grinding or very bad separator

performance
0.85 - 1.00 bad to mediocre performance for first

generation separators (Raymond, Sturtevant)
1.00 - 1.20 good performance for first or second

generation separators (Humbolt-Wedag)
1.10 - 1.40 high efficiency separators
The figures given above are for cement, raw mix R.R. Nos. are
usually lower.
4.4 Mill Circulating Loads
4.4.1 Definition
3 - 120
A mill's circulating load is the amount material rejected by

the separator (material that is too coarse) and is returned
back to the mill for regrinding. This flow represents the
tonnage rate of material that is "circulating" around inside
the circuit. Mathematically, it is the amount of separator
rejects expressed as a percentage of the mill's finished
production rate.
For example, if the production rate is 90 tph and the rejects

rate is 170 tph, then the circulating load is 188.9%. Note
that the mill throughput is the sum of fresh feed rate and
the rejects flowrate. In our example, it would be 260 tph.
3 - 121
4.4.2 Circulating Load and Production Rates
Production is maximized at a certain circulating load which

is a function of:
• product fineness
• components in product
• mill dimensions
• ball charge
• linings types
• internal designs
• separator type and size
• separator tuning (airsweeps)
• mill ventilation
• auxiliary constraints (elevators, dust collection
This best circulating load is unique to each circuit and can

only be found by experimentation.
A) Circulating Load Below Optimum
In this condition the mass flow rate to the mill is too low.
Therefore the mean residence time is too long and the
material is ground too fine in the mill. The overgrinding
means a lot of super fines are giving a high separator feed
blaine so the separator is opened up, allowing coarser
particles to product and the overall particle size distribution
is not as sharp as it should be. Therefore the average
particle size is smaller than desired, if #325 is maintained.
Therefore production is reduced as the average work being
done is higher.
3 - 122
In this condition underfilling of the first, and even the

second compartment, is likely so ball wear will be
exaggerated. If really low circulating loads are run then the
material filling ratio may fall below 0.6 and material
breakage rates will decrease.
However, back-spilling is not much of a risk and many mills

run in this condition.
B) Circulating Load Above Optimum
Here all situations are reversed. In this condition the mass

flow rate to the mill is high. Therefore the mean residence
time is short and the material is not overground. Therefore,
the ball charge is more efficient since the particle sizes are
well sized related to the balls.
The separator feed is coarse so the separator is tightened

up. However, the feed rate to the separator is now high.
Therefore, the bypass increases and a percentage of the
finished product is bypassed back to the mill.
In this condition overfilling of the second, and even the first

compartment, is likely so breakage rates will be reduced.
Eventually either the mill power drops, reducing production
dramatically, or the internal transport mechanisms are
overwhelmed and the mill back- spills.
However, if you are on manual control and your miller

believes that elevator amps equals production the mill
probably runs too high a circulating load.
3 - 123
C) High Optimum C.L.
The following will require/allow a higher optimum circ. load;

• short L/D
• coarser ball charge
• 2nd compartment lining with moderate lifting action
• higher-efficiency separator (according to Onoda)
• higher separator sweep
D) Low Optimum C.L.
The following will require/allow a lower optimum circ. load;

• long L/D mill
• finer ball charge
• classifying liners
• undersized or Sturtevant separator
• high mill ventilation
4.4.3 L/D Ratio and Circulating Load
The L/D Ratio, or length divided by the diameter, is a

fundamental characteristic of the mill. It is used to
describe the relative length or shortness of the mill or the
individual compartment. The appropriate L/D is selected by
the supplier to achieve a certain type of grinding. For
example, open circuit mills, (or mills grinding a harder than
average clinker), will tend to be very long to ensure size
reduction in one pass. Where as closed circuit raw mills
will tend to be short, since we generally do not need to
produce ultra fines. However it is important to realize that
the L/D will influence to a certain extent the optimization
3 - 124
range. For example, a high L/D ratio mill will prefer to grind
with a low circulating load and attempts to grind with too
much circulating load will result in a loss of production.
Fig. 1 is a plot of various mills' nominal L/D vs circulating
load. As a rule of thumb we can see that raw mills
generally have a low L/D, (1.4 - 2.9) and operate with higher
circulating loads than do finish mills which have a higher
L/D, (2.8 - 3.8).
Recognize that there are many other factors that influence

circulating loads such as grindability, ball charge design,
blaine/mesh targets, separator designs and whether or not
the circuit has been optimized. Due to the variability of
these factors, from mill to mill, will results in a wide scatter
of data shown above. CAUTION: the plot line represents
the average for data gathered and does not represent
optimum.
4.4.4 Circulating Load Calculations
3 - 125
Earlier we defined circulating load as the flow of rejects

divided by the flow of finish product (or mill fresh feed).
However there are a number of ways to calculate
circulating load.
3 - 126
Circulating load can be determined by:

• Mass flow (best method if you have an accurate
rejects flow meter)
Direct weight measurement.
• Sieve Analysis (will give a very rough answer)

325 (45 µm) mesh. [Using blaines are reported to give
better approximations. However some care must be
taken regarding correction factors used].
• Particles Size Distribution (long to get results, more

accurate than just a sieve)
Several sieve sizes
Sedigraphs/Lasers
Discrete samples
Feed rates
(See also section in Circuit Evaluation for how to sample)
3 - 127
Mass flow determination:
Where: A = Separator Feed

R = Separator
Rejects
F = Separator Fines
Circulating Load (C.L.) =

Sieve analysis determination:
K = F (fresh feed weight = finished product weight)
A = R+K
 Aa(x) = Rd(x) + Ft(x) where:a(x) = mass % of dimension x in
feed
r(x) = mass % of dimension x in
rejects
f(x) = mass % of dimension x in
fines
Substituting (R + F) for A gives:
(R + F) a(x) = Rr(x) + Ff(x)
R [a(x) - r(x)] = F [f(x) - a(X)]
and circulating load is defined as follows:
3 - 128
C.L. =
combining these two equations gives:

U se Word 6.0c or l ater to
vi ew Maci ntosh pi cture.
Circulating load = or
3 - 129
Factors affecting circulating load:
• Mill dimensions
• Ball charge gradation
• Size of the feed material
• Amount of feed material (fresh and tailings)
• Mill ventilation
• Type of partition(s)
• Linings
• Grinding aid
• Water spray
• Separator efficiency
• Mill bypass
• Product fineness
• Auxiliary constraints
3 - 130
4.5 Qf / Qa Principle
The fact that the separator runs in a circuit means that the
parameters of the feed may change. Primarily these are;
feedrate and feed fineness, of a secondary nature is the
particle size distribution shape and the mix in a multi-
component product.
Also there are other parameters which affect the separator

performance. Sizing, speed, main fan blade number and
position, air ventilation, the number of selectors and the
diaphragm position all affect efficiency, as well as typical
maintenance issues like; the gap between the selectors and
the drum cover and how far the blades extend under the
drum.
In reviewing separator performance one over-riding

relationship, feed concentration, affects efficiency.
Qf/Qa
This ratio neatly ties up the combined effect of; circulating

load, system output, separator and fan speed as well as
blade position and number and also separator sweep on a
high efficiency separator. The Qf/Qa ratio is simply;
Qf/Qa = kg feed to separator

m3 of air sweep through the classification zone
3 - 131
All air separators are predominantly affected by this ratio.
3 - 132
4.5.1 Bypass and Qf/Qa
The bypass is usually related to Qf/Qa in an exponential

way. This is shown below for both an O-Sepa and a
Sturtevant.
Qf/Qa vs Bypass - O-Sepa & Sturtevant

80
70 Sturtevant Bath
Sturtevant
60
50
Bypass (%)
40
30
O-Sepa
20
10
0
1.0 1.5 2.0 2.5 3.0 3.5
Qf/Qa (kg feed/m3 separator sweep)
What the above graph shows is that as separator feed is

increased bypass also increases provided the amount of
separator internal airsweep remains the same. Increasing
sweep through the separator with more fan, more speed,
more blades or a larger diameter fan should all reduce the
bypass. Therefore reducing sweep through closing the
diaphragm will reduce efficiency or increase bypass.
In general high efficiency separator operators have the

luxury of being able to adjust separator draft as separator
feed changes and adjust cage rotor speeds to control
blaines and 325 mesh. Sturtevant separator operators do
not have this on line ability except for the diaphragm,
3 - 133
(valves). They must instead make adjustments to fan and

selector blades, during shutdown intervals.
For Sturtevant separators, the general rule of thumb for

best separator performance is to find the combination of
selector and fan blades that maximizes the number of each
type. One or the other will be the limiting factor depending
on the cement type being ground. For example, on most
Type 1 cements one should be using the full complement of
fan blades and adjust the number of selector blades
accordingly. However on very high blaine products (7700
cm2/g) it could be the exact opposite. More importantly, if
the separator is operated with the valves or diaphragm
nearly always closed, one should be adding more selector
blades to control fineness. This will increase internal
airflow and reduce bypass.
For high efficiency separator operators the idea is similar.

One should maximize the fan flow and adjust the cage rotor
speed accordingly to maintain fineness.
In both cases the general rule is to strive for the largest

amount of internal airflow or the lowest Qf/Qa ratio that still
meets product quality constraints. At these settings the
separator will likely operate at its best efficiency.
3 - 134
Ball Mill Partition or Diaphragm
Purpose
1) Divides mill into chambers or compartments. Allows operators

to have different ball charges and liners in each and hence a
different type of grinding action in each.
2) Controls the material flow from one compartment to the other.

Regulates partially the retention time and the degree of
material filling in the grinding media voids. Adjustable
partitions thus allow some optimization of the grinding action
in each compartment.
Parts and Function
Figure 1: Feed End View of Partition with Grate Segments
Figure 1 shows the basic parts a typical double wall, fully liftered
partition, (the most common type). Note that large mills often
have rows or rings of grate segments. For convenience, they are
usually referred to as inner, outer of middle grates. In small
3 - 135
mills, these are often one piece pie shaped sections. From
inside, the mill the discharge grates or outlet diaphragm will
have a similar appearance.
3 - 136
Figures 2 and 3 shows the pumping action of a liftered partition.

Liftered outlet diaphragms work on the same principle. Lifters
can be straight as shown, curved or adjustable scoops
(Slegten/Magotteaux design).
Figure 2: Feed End View of Partition without Grate Segments
Figure 3 : Side View of a

Traditionally, the liftered Partition
partition (and outlet
diaphragm) tend to over
pump, leaving the grinding
media adjacent to the
diaphragm, on either side
empty of material. This
reduces the amount of actual
grinding done by the grinding
media. The higher steel to
3 - 137
steel contact also

accelerates wear in these
areas. view Macintosh picture.
3 - 138
Adjustable Partitions
To optimize the mill, adjustable partitions have been introduced

to the industry since the mid-eighties. Types include:
a) adjustable scoops (Slegten/Magotteaux) - most common
b) adjustable center hub openings (Pfeiffer)
Other (makeshift) methods tried to other plants include:
i. Smaller slot sizes

(Risks - plugs, peens over quickly, cutting or welding sharply
shortens the useful life of the casting)
ii. Cut holes in lifters

(Risks - difficult to fine tune, significantly weakens the
structural integrity)
iii. Replace slotted plates with blind plates

(Risks - difficult to fine tune, if no center screen exists then
it will sharply decrease mill sweep)
iv. Install an overflow type partition (Sonex)

(Risks - non adjustable, very prone to backspilling - not
recommended)
To get the full benefit of an adjustable partition, one must be

prepared to make trial and error changes and follow-up each
time. This is very time consuming but plants have had good
results.
3 - 139

Mill Grinding

Uploaded by

Copyright:

Available Formats

Mill Grinding

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mill Grinding

Uploaded by

Copyright:

Available Formats

Introduction to Process Engineering

3. MILL GRINDING THEORY

1. Introduction and Overview.............................................. 3

1.1 Introduction to Grinding......................................... 3

1.2 Grinding Mill Circuits.............................................. 4

1.3 Types of Mills.......................................................... 8

2. Internal Ball Mill Parts..................................................... 12

3. Ball Mill Grinding and Related Topics............................. 25

3.1 Ball Milling Process (Basics and Fundamentals)... 25

3.2 Grinding Media........................................................ 39

3.3 Ball Mill and Circuit Evaluations and Tests........... 52

3.4 Ball Mill Control...................................................... 58

4.3 Separator Efficiency............................................... 87

4.4 Mill Circulating Loads............................................. 93

4.5 Qf/Qa Principle........................................................ 101

3. MILL GRINDING THEORY

The text of the following section on Mill Grinding was adapted

Lafarge Mill Grinding Reference, Edition 1

Readers wishing to study the topic in greater depth and detail

Other volumes that will be available in the Lafarge Mill Grinding

Volume 2: Auxiliary Equipment in Mill Circuits

Volumes 1 & 2 were meant to be practical reference and idea

"Introduction to Process Engineering". Realistically we cannot

Sam Fujimoto Paul Ukrainetz

1. Introduction and Overview

1.1 Introduction to Grinding

In the cement industry we spend a lot of effort and money

In raw mix grinding, the particle size has an important role

different emphasis are used depending on the type of

In short how we grind materials at the plants at each stage

1.2 Grinding Mill Circuits

A mill circuit is a combination and proper arrangement of

Let us first look at the Ball Mill grinding circuits. Once a

1.2.1 Open-Circuit Grinding

Where the mill product is sent to storage silos without

Since size reduction must be accomplished in one pass,

1.2.2 Closed-Circuit Grinding

Where the mill product is sent to the separator and the

Sketch of an open circuit mill

Fig. 1.2.2 - Closed Circuit Grinding

Separator Feed (A)

Fresh Feed (K)

Fig. 1.2.3 - Closed Circuit Grinding

With Drying of Feed

Separator Feed (A)

1.2.3 Principles of Closed Circuit Grinding

For most raw mix and cement grinding we used closed-

In a closed circuit system, the mill product is conveyed to a

From the cost standpoint, it is less expensive to grind in

1.3 Types of Mills

The grinding mill is the main piece of equipment used in the

1.3.1 Ball Mills.

1.3.2 Roller Press Mills.

1.3.3 Roller or Bowl Mills.

1.3.1 Ball Mills

A ball mill is a cylindrical drum with varying ratio of

1.3.2 Roller Press Mills

A roller press as shown on the attached diagram is nothing