87-31-P

GRINDING: AN OVERVIEW OF
OPERATION AND DESIGN
by
John w. Schlanz
Mineral Processing Engineer

ABSTRACT

This paper was specifically prepared for presentation at a

mill operators symposium held in Spruce Pine, NC in October of
1987.
The purpose of this discussion was to review grinding
concepts and show how better understanding of these concepts are
leading to new ideas in the design of comminution circuits. This
paper covers topics relating to grinding mechanisms and laws,
energy relationships, types of mills, control factors and design
of such circuits.

TABLE OF CONTENTS

ABSTRACT

........................

INTRODUCTION

.......................

GENERAL REVIEW
Breakage Forces .
Grinding Laws

.....
...
.....
... ....
.....
.....
...

...
. .. . .
. .. ..
. ....
.....
. . ..

3:
4:

5:

..
.....
.....
...
...
...

8
9

10
12
15

16
17
18

.................
...
. ....
....
CIRCUIT DESIGN
......... .......
.......
Bond Method . . . . . . . . . .
.......
Step 1: Grindability Test
2:

CONTROL OF CIRCUITS
Variables
Automatic Control .

Step
Step
Step
Step

...........
........

TUMBLING MILLS .
Characteristics
Types
Rod Mills
Ball Mills
.
Autogenous Mills
Grinding Circuits
. .
Open Circuit Grinding .
Closed Circuit Grinding

Conversion to an 8 ft. Mill

Allowance for Size Reduction Ratio
Scale-up to Larger Mills
Corrections for Milling Conditions

19
19
21

...

22

...

23
23

...
. .

........
. ..

24

25
25
26

PSU GRINDING CIRCUIT SIMULATOR

...............

26

REFERENCES .

...............

31

TABLE OF CONTENTS (cont.)

Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.

1.

2.
3.

4.
5.
6.

7.
8.
9.

10.
11.
12.
13.
14.

.. .....

Breakage Mechanisms .
.
.
Grinding Laws

Patterns of Tumbling . . . .
Classification of Rod Mills
Determination of Media Charge . . . . . .
Grinding Action in Rod Mills . . .
Classification of Ball Mills . . . .
Open Circuit Grinding . .
. . . . . . . .
Closed Circuit Grinding . . . . . .
Grinding Circuit Variables .
Bond's Equations . . . .
Equations for size Mass Balance Method
Nomenclature Utilized in PSU Simulator
44 Grinding Circuits in PSU Grinding Circuit Similator

.....

32
33
34

35
36

37

38
39
40
41
42
43

45
46

87-31-P

GRINDING: AN OVERVIEW OF
OPERATION AND DESIGN
by
John w. Schlanz
Mineral Processing Engineer
North carolina State University
Minerals Research Laboratory
180 coxe Avenue
Asheville, North Carolina 28801

INTRODUCTION

The

importance

of

communition

circuits

in

any mineral

processing facility is often misunderstood or, at the very least,

not fully appreciated.

Grinding, because it is normally the

final stage of this communition, takes on a special importance

and is often described as the key to successful plant operation
and production.
The grinding circuits carry responsibilities that can have
an impact on all phases of the operation, including efficient
mill operation and successful production, with its main purpose
being to exercise close control on product size.

In other words,

the grinding mill must produce an optimum mesh of grind.

An optimum mesh of grind is important for several reasons,

with

the most

mineral

species

basic

being

contained

grinding fine enough,

achievement
in

the ore.

of

liberation of the

If the mill

is not

or undergrinding the ore, it produces a

product which is coarse and may have a low degree of liberation,

which will result in poor recoveries, poor concentration ratios
and uneconomical operation.

Other factors which contribute to

the importance of this optimum mesh of grind is consideration of

downstream processing and possible customer size specifications
imposed on the product.
Probably the most critical reason for good control over the

grinding

circuit

is

associated with

energy consumed in this unit process.

the

tremendous amount of

It is no great secret that

grinding circuits are the largest power consumer and most costly
phase of most operations and it has been estimated that SO% of
the 100 billion kilowatt hours of electricity used in

u. s.

per year can be attributed to the grinding circuits.

mills

An extra

19% of power must be supplied to grind the ore to one screen size
finer based on standard size scales.
Because

of

the

aforementioned

reasons,

responsible

and

efficient operat_ion of grinding circuits is critical and much of

this responsibility will fall on the personnel who operate these
circuits.
grinding

Recent
are

understanding
operation.

research is showing that the mechanisms of

very
of

complex
grinding

Therefore,

and

it

concepts

is
may

felt
lead

that
to

basic

improved

this discussion will focus on providing

some general knowledge concerning grinding, covering topics such

as grinding laws, types of mills and how they operate, critical
variables and how they can be controlled, and some comparisons of
design methods which show how grinding concepts have evolved.

GENERAL REVIEW
In order to appreciate what is actually taking place in a
grinding mill it may be helpful to review some of the mechanisms
of breakage, energy relations and laws that attempt to explain
these occurrences.

Breakage Forces
Basically materials can be classified into two catagories.
A material is either described as ductile or brittle.

A ductile

material, when stressed to failure, will normally break into two

pieces.

Stressing

of

brittle

material

will

result

in

shattering, or breakage into many pieces of different sizes of

which

the

fracture

paths

cannot

be

controlled.

Because

ore

behaves as a brittle material, the pattern of breakage presents

problems

in grinding

by

attempting

to

create fracture

specified limits without having any control over the

within

fracture

process.
There are several mechanisms through which breakage of
particle can occur (Fig. 1).

These mechanisms are:

1) Impact or Compression - Forces Applied Normally

2) Chipping - Forces Applied Obliquely
3) Abrasion - Forces Applied Parallel
Regardless

of

the

mechanism

involved,

in

any

fracture

process the particle must be raised to a state of strain which

will initiate the propagation of fracture cracks.

In order to

create this state of strain, energy greater than or equal to the

stored

strain

energy

of

the

particle must be supplied.

The

factors that decide how much energy is required to overcome this

stored energy are:
1) Presence of preexisting cracks or flaws
2) Degree of plastic flow in the solid vs. complete
brittleness

3) Geometry and rate of stress application

In most mills, the forces which act on a particle to create
fracture

can occur

thru one or more of several means, which

include:
1) Collision between particles
2) Pressure loading on the particles
3) Shear and abrasion by particles falling through the
grinding media
4) Impact of falling media on particles
5) Shock transmitted through a crop load, or packed bed
Therefore, when deciding upon the required energy, or the
power which must be supplied to a grinding mill, several factors
must be considered.
1) Efficient conversion of input energy to mechanical
action
2) Efficient transfer of the mechanical action to the
particle
3) Matching the stress produced by the mechanical
action to the failure stress of the particles
Unfortunately,

most of the energy supplied to a grinding

mill is lost primarily as heat, noise and deformation of rock and

equipment.

Only a small amount of the supplied energy actually

reports as energy of new surfaces created from broken products

although this is the true value of useful energy consumed in the
mill.

This is the direct reason why grinding circuits are so

cost intensive.

Grinding Laws
Over the years, many attempts have been made, in the form of
mathematical

laws,

to

explain

the

relationships

of

energy,

breakage and particle size (Fig. 2).

The first of these laws was Rittinger's Law, which states
that the energy required for fracture is proportional to the new
surface produced.

This relationship can be expressed

mathematically as:

ER

= CR

(l/d2 - l/d1)

Where CR

= Constant

d2 = Product Size
d1
This

law

requires

that

the

grinding

= Feed
rate

Size
function

is

proportional to particle size which would be the 6ase only for

very hard materials over limited size ranges.
Rittinger.'s Law was followed by Kick's Law, which formed the
basis that the energy required for fracture is proportional to
the volume broken.

This law can be expressed as:

Ek

This

law

implies

that

= ck

log <d2;di>

grinding

rate

is

independent

of

particle size because it is based on a constant and areduction

ratio, indicating equal amounts of energy would be required to
achieve size reductions regardless of particle size.

However, it

is well known that more energy is required for fine grinding as

compared to coarse grinding, thus it is felt this law is valid
for very short grinding times with particles confined in coarse
size ranges.
Probably the most accepted law was that of Bond, which tends

to fall between Kick and Rittinger.

Bond's basis was that work

input is proportional to the new crack tip length produced in

particle breakage and equals the work represented by the product
minus that represented by the feed.

Bond's law is characterized

by the well known Bond Work Index and can be expressed as:
10 Wi

w=
Where

10 Wi

w = work

input

P = 80% passing size of product

= work

index

Wi

= 80%

passing size of feed

Bond's Law has been found to apply reasonably well in the

range of conventional rod mill and ball mill grinding.
Other laws, such as those proposed by Charles and Holmes,
have been investigated and all seem to have similar flaws, which
include:
1) All are based on an applied energy to size reduction
relationship which is considered too simplistic for such a
complex process.
2) They do not apply to the entire spectrum of size ranges.
3) All were very liberal with assumptions.
4) They do not consider mechanical concepts such as recycle
ratios,

classifier

effectiveness,

optimum media charges,

lifter designs, changes in mill feed, etc.

5) They are all purely empirical relationships evolved from
fitting

laboratory

data

to

predictable

equations

and

relationships.
Because of these discrepancies in the older laws, a newer

method

is

being evolved which considers

factors

such as mill

size, mill power (specific grinding energy), efficient grinding

conditions, recycle and classification efficiencies, mill circuit
behavior

under

varying

conditions,

circuits and economic optimization.

mill selection for

complex

This new method is termed

the size-mass balance concept.

The size-mass balance method is based on knowing how quickly
a

certain size particle breaks and in what sizes its products

appear.

This

breakage,

method combines

residence

time

concepts

of

distributions

specific
and

rates

of

mathematical

descriptions of classification plus a set of relationships which

describe how each factor

in the size-mass balance varies with

mill conditions and mill size.

can

be

simulated,

compared

and

With such methodology circuits

optimized

for

technical

and

economic performance on paper before a final design is adopted.

This

method

will

be

described

in more

detail

later

in

this

discussion.

TUMBLING MILLS
Most grinding, whether done wet or dry, is usually performed
in rotating cylindrical steel vessels known as tumbling mills.
These mills usually contain a charge of loose crushing bodies, or
grinding medium,

which is

free

to move inside the drum.

The

grinding media normally employed is steel rods or balls, ceramic

pebbles, hard rock, or, in some instances, the ore itself.

Characteristics
Regardless of the grinding media utilized all tumbling mills
effect breakage according to the same basic principles.

As the

mill rotates the grinding charge is raised from a level surface

position with liners preventing slippage of the charge so that
the media moves with the shell until they fall and tumble down
over the mass of the charge.

The pattern in which the charge

tumbles is directly related to the speed of rotation of the mill

(Fig. 3)
The various tumbling patterns are described as cascading and
cataracting.
tends

to

Cascading describes the portion of the media that

roll

down

abrasive-type

to

the

toe of the mill which results

comminution,

increased liner wear.

leading

to

finer

grinding

in
and

Cateracting is experienced by the portion

of the charge that is raised high enough in the mill to actually

fall

back

down

comminution,

to

which

the

toe.

produces

This
a

action

coarser

leads

product.

to

impact

Most

mills

utilize a combination of the two in actual operation.

As

stated,

all

tumbling

rotational speed of the mill.

patterns

are

result

of

the

A critical speed does exist for

each mill that creates centrifuging of the charge and no tumbling

action occurs.

The media is theoretically carried around in a

fixed position against the shell.

This critical speed can be

calculated for any tumbling mill using the formula (Fig. 3):

42.3

J D-d
= Critical

Where Nc
D
d

speed

= Mill diameter
= Charge diameter

All tumbling mills

are operated at a percentage of this

critical speed to achieve the correct tumbling patterns.

Even though tumbling mills have been developed to a high
degree of mechanical efficiency and reliability they are still
extremely wasteful in terms of expended energy because, even with
correct tumbling action of the medium, ore is broken as a result
of repeated, random impacts, indicating that grinding is subject
to the laws of probability.

This indicates that the tumbling

mill merely provides a vehicle in which a multitude of fracture

opportunities exists, leading to perhaps a definition that it is
a device for converting electrical energy into mechanical energy
and incidentally causes fracture in the process.
Types
Tumbling mills
medium utilized.

are

categorized according to the grinding

There are three basic types:

1) Rod mills
2) Ball mills
3) Autogenous
Rod mills, obviously, are charged with steel rods while ball
mills employ steel or ceramic balls, or pebbles.

The autogenous

mills are charged with large pieces of the ore itself, which

10

serve as the actual grinding medium.

Rod Mills
Rod mills are often described as fine crushers or coarse
grinders as they do produce a relatively coarse product, taking
feed as coarse as 50 mm with products as fine as 300 mm (SO mesh)
with reduction ratios of 15 - 20 to 1.

A distinctive feature of

rod mills is that its length is 1.5 to 2.5 times the diameter.
Because rods longer than 6 m tend to bend, a maximum size for rod
mills does exist.
Rod mills are classified according to the type of discharge
(Fig. 4).

There are 3 main catagories:

1) Center Peripheral Discharge - This unit is fed at both

ends thru the trunnions and discharged thru ports spaced
along

the

center.

Normally

gives

coarse grind with

minimum fines and has a limited reduction ratio.

2)

End Peripheral Discharge - This type is fed thru one

trunnion end and discharges from the other end by means of

several peripheral apertures.

It is used mainly for dry and

damp grinding, producing a moderately coarse product.

3) Overflow Discharge - This rod mill is the most widely
used and is fed thru the trunnion at one end and discharged
thru the trunnion at the other end.

It is used for wet

grinding

function

only

with

its

principle

being

the

conversion of crushed product to ball mill feed, although it

is often used to produce a final grind.
Rod mills are initially charged with a selection of assorted

11
diameter rods calculated to provide a maximum grinding surface
and to approximate a seasoned load.

The maximum size media can

be calculated from the following equation (Fig. 5):

Where:

FW
:..!!.J..-

KCs

VD

= dia. of topsize media in inches

F = 80% passing size of the feed (microns)
W = work index
Cs = percent of critical speed
s = specific gravity of feed
D = inside mill diameter
K = constant, 200 for ball mills, 300 for rod
M

].

mills

Once the top size of rods has been determined, the initial
charge can be made up according to the chart shown in Fig.

These sizes will range from 5 in. to 1-1/2 in. in diameter (150
to 25 mm).

It should be noted that this data supplies a good

starting point but optimum charge is normally determined thru

trial and error practices.

The rod charge normally occupies 35%

of the mill volume with a void space of 22% of this.

The new

stacked

lb/ft3.

rod

charge

should

weigh

approximately

382

overcharging of the mill will result in inefficient grinding and

increased wear.
To maintain

optimum grinding conditions,

worn rods

(less

than 25 mrn in diameter) should be replaced with fresh rods of the

maximum diameter.

Rod consumption (wear) is usually in the range

of 0.1 to 1.0 kg of steel per ton of ore.

12
Rod mills are normally operated at SO to 65% of critical
speed.

This allows

the

rods

to tumble

in a

cascading-type

pattern instead of cataracting, which tends to tangle the rods.

Pulp densities usually range from 60 to 75% solids with finer
feeds requiring lower pulp densities, although optimum conditions
are ultimately dictated by the ore itself.
Actual grinding action results from line contact of rods on
particles (Fig. 6).
the

The coarse feed tends to spread the rods at

feed end producing a wedge or cone shaped array.

This

creates a tendency for grinding to take place preferentially on

larger particles,
gives

thus

producing minimum fines.

This action

product of relatively narrow size ranges with little

oversize or slimes.

Because of this controlled size reduction

rod mills are nearly always run in open circuit.

Rod mill grinding does present several cost advantages which
should be considered when selecting a mill.

These advantages

are:
1) The size distribution of the product is controlled by the
grinding action, virtually eliminating the need for closed
circuit equipment.
2) The grinding medium is low cost.
3) Rod mills generally have high grinding efficiencies with
low steel consumption as compared to other mills.

Ball Mills
Conventional ball mills

are generally characterized with

13
shorter length to diameter ratios of the magnitude of 1.5 to 1.0.
Ball mills with ratios of 3 to 5 are designated as tube mills,
which are divided into several compartments with each compartment
having

different

compartment are

ball

charge.

Tube

termed pebble mills.

mills

with only one

In industrial minerals

applications, pebble mills are normally used for fine grinding

and usually contain ceramic
avoid contamination.

liners

and ceramic-type media to

In some circumstances, coarse, hard rock

pebbles may constitute the grinding medium.

Ball mills, like rod mills, are classified according to the
type of discharge the mill is equipped with.

The two types of

discharge are (Fig. 7):

1) Grate Discharge - These mills are fitted with discharge
grates between the cylindrical mill shell and the discharge
trunnion.

This type discharge requires a lower pulp density

than others which reduces retention_times, resulting in very

little overgrinding but also discharges a large fraction of
coarse particles, necessitating closed circuit operation.
2) Overflow Discharge - This ball mill discharge is very
similar to the rod mill overflow discharge although many
times a scalping screen is needed to collect smaller or worn
grinding medium which may overflow with pulp.
the

simplest

to

consumption may

operate

and

is

be as much as

the
15%

most
less

This type is
used.

Power

than the grate

discharge mill although grinding efficiencies are nearly

equal.

14
Control of the product size thru the grinding action in a
ball mill is inferior

to that associated with the rod mills.

This is due to the grinding action being a purely random process

with the probability of a fine particle being struck the same as
that of a coarse particle being struck.

If run in open circuit,

this results in a product with a wide range of particle size and

overgrinding becomes a problem.

Closed circuit operation lessens

this effect and most ball mill circuits

are operated in this

mode.
Pulp

densities

in

ball mills

are

maintained

as

possible, usually in the range of 65 to 80% solids.

the balls to be coated with a layer of ore.
too

dilute

metal

to

metal

contact

high

as

This allows

If the pulp becomes

increases,

resulting

increased steel consumption and inefficient grinding.

in

Like the

rod mill, finer grinding requires a more dilute pulp.

The ball charge should contain as small balls as possible to
offer the greatest grinding area.

The largest balls should be

just heavy enough to break the coarsest particle.

An accurate

seasoned charge will consist of a wide range of ball sizes.

The

optimum ball charge can be determined the same as for rods, using
the equation shown in Fig.

5 to calculate the top ball size.

Once this top size is determined the proper distribution can be

chosen

from

replacement

the
balls

chart
should

in
be

Fig.
of

5.
the

As

with

largest

the

rod

required.

mill,
Steel

consumption is again in the range of 0.1 to 1.0 kg per ton of new

feed.

15
The media charge in a ball mill is normally larger than that
of the rod mill, occupying 40 to 50% of the total mill volume.

ball charge generally weighs out at 280 lbs. /ft3 giving a void
space of 44% of the volume occupied by the charge.
Ball mills are also normally operated at higher speeds than
rod mills, usually between 70 to 80% of critical.

This allows

more of a cataracting action of the charge, which results in more

impact on the particles.
Autogenous Mills
In autogenous mills comminution is achieved by the action of
the ore particles on each other.

Usually coarse rock is screened

from the crusher circuit and stored for grinding medium.

This

type of grinding is becoming increasingly popular because, if the

ore is susceptible to this grind, medium costs can be eliminated.
Autogenous mills generally have a large diameter relative to
length, such as the cascade mill and are run at high speeds of 80
to 85% of critical.

The load volume is in the range of 35 to 40%

of the total and these mills can effect very large reduction
ratios.

The product size is generally in the

range of the

natural grain size or crystal size of the ore.

This type of grinding is used most often on friable, grainy
material such as silica rock, asbestos, slag, bauxite, dolomite,
ferrosilicon,

limestone,

detailed testing is

taconite

and

cement

clinker.

More

also required initially because all ores

cannot be treated in this manner and, although medium costs are

eliminated, overall power consumption can be much higher than

16
conventional grinding.

Grinding Circuits
There

is

no

such

equipment for grinding.

thing

as

"best"

industry

today

is

of

The circuit which works well on one ore

may be totally unsatisfactory for another ore.

in

configuration

flowsheet

The general trend

simplification,

using

larger

parallel circuits instead of multiple-line circuits with smaller

. mills.

This change has been brought about by the savings of both

capital

and

operating

costs

that are associated with

larger

equipment.
In any grinding installation a choice must be made between
wet or dry grinding.
wet

grinding,

which

Most mineral processing facilities utilize

has

the

following

advantages

over

dry

grinding:
1) lower power consumption per ton of product
2) higher capacity per unit of mill volume
3) makes possible the use of wet screening or classification
for close product control
4) elimination of dust problems
5) allows the use of simple handling and transport methods
(i.e., pumps, pipes, launders, etc.)
A choice must also be made between open and closed circuit
operation.

While both have their own advantages,

the final

decision is usually based on factors such as mill selection,

product

requirements,

tonnage

requirements

and

ore

17
characteristics.

Open Circuit Grinding

Open
mills,

circuit

grinding

consists

of

either parallel or in series,

one or more

grinding

that discharges a final

ground product without classification equipment and no return of

coarse discharge back to the mill.

Some very simplistic examples

of open circuit grinding are (Fig. 8):

1) Rod mill
2) Ball Mill
3) Rod mill, ball mill combination
Not all ores
arrangement.

can be ground in an open circuit type of

Some

conditions

which

do

favor

open

circuit

grinding are:
1) small reduction ratios
2)

reduction of particles to a coarse, natural grain

size
3)

recirculation of

cleaner flotation middlings

for

regrinding
4) a non-critical size distribution of the final ground
product
Some advantages of operating in the open circuit mode vs
closed circuit are:
1) minimum equipment requirements

2) high pulp density discharge

3) simplicity of operation

18

Closed Circuit Grinding

Closed

circuit

grinding

consists

of

one

or

more

mills

discharging ground product to classifiers which in turn return

the coarse product from the size separation back to the mill for
further grinding.

In this circuit, grinding efficiency is very

dependent upon the size separation effected so care should be

exercised in selecting the type and size of classifier used to
close the system.
This type of grinding is the most common circuit found in
mineral processing facilities, mainly because a lot of ores and
product requirements are not suitable for open circuit grinding.
Some advantages presented by grinding in closed circuit are:
1) This arrangement usually results in higher mill capacity
and lower power consumption per ton of product.
2) It eliminates overgrinding by removing fines early.
3) It avoids coarse material in the final ground product by
returning this material to the mill.
Although closed circuit grinding offers many choices for
arrangement

of

the

equipment

as

well

as

combinations

equipment, some of the more common circuits are (Fig. 9):

1) Rod mill/Classifier
2) Ball mill/Classifier
3) Rod mill/Ball mill/Classifier
4) Rod mill/Classifier/Ball mill/Classifier

of

19

CONTROL OF CIRCUITS
The importance of the grinding circuit to overall production
in

any

facility

should

be obvious by now.

Because of

the

responsibilities assigned to grinding it becomes essential that a

grinding mill accepts a certain required tonnage of ore per day
while yielding a product that is of a known and controllable
particle size.

This leads to the conclusion that close control

over the grinding circuit is extremely important.

Variables
There are many factors which can contribute to fluctuations
in performance of a mill, but some of the most common found in
industrial practice are:
1) changes in ore taken from different parts of the mine
2) changes in crusher settings
3) wear in the crushers
4) screen damage in the crusher circuit
These are a few things that operators should look for when
changes
ahead

in mill performance are noticed.

of

the

mill

can

aid

in

Stockpiling of ore

smoothing

out

some

of

the

fluctuations although it must be stored in such a manner that no

segregation occurs.
In

operating

grinding

circuit,

like

any other

unit

process, variables key to the performance must be dealt with.

Some of the principle variables affecting control of grinding
mills are (Fig. 10):

20

1) changes in feed rate

2) changes in circulating loads
3) size distribution of the ore
4) hardness of the ore
5) rate of water addition
6)

interruptions

replacement

of

in

operation

grinding

medium,

(i.e.,

stoppage

clearing

of

for

choked

classifiers, etc.)
Of these variables, feed size distribution and ore hardness
are the two most significant because they can affect the actual
grinding mechanics.
The only two variables that can be independently controlled
by the operator are

feed rate and water addition.

All other

variables depend on and respond to changes in these two items,

thus these are used to control the grind.

Too dilute a pulp,

which can arise from a decrease in feed rate or an increase in

water

addition,

resulting

in

will

decrease

coarser

opposite circumstances,

retention

product

and

time

increased

in

the

wear.

mill,
Under

(high feed rate, decrease in water) pulp

densities can become too high.

This can lead to an increase in

retention time resulting in a finer product.

However, in most

situations concerning too thick of a pulp the pulp viscosity will

become so great that the grinding medium may begin to float and
actual grinding will cease, and if not corrected, the mill will
eventually choke.

21
Automatic Control
Since grinding is extremely energy intensive and the product
from

grinding

essential.

affects

It

is

subsequent

processes,

now generally accepted

close control
that

some

form

is
of

automatic control is required.

In

implementing

instrumentation

for

objective must first be clearly defined.

process

control

the

Some examples of these

objectives relating to grinding circuits may be:

1) to maintain a constant product size at maximum thruput
2) to maintain a constant feed rate within a limited range
of product size
3) To maximize production per unit time in conjunction with
downstream circuit performance.
Instrumentation to control process variables is receiving
more and more attention lately.

Some general methodology used to

control some of the more critical variables are as follows:

1)

Feed

controllers,

Rate

Variable

which are

speed

belt

feeders,

audio

sound sensitive devices that will

measure feedrates according to noise levels in the mill.

2)

Medium

Charge

Can

be controlled thru continuous

monitoring of mill power consumption.

3)

Flowrates and Densities - Can be measured with magnetic

flowmeters and nuclear density gauges.

4)

Sump Levels - Controlled by bubble tubes, capacitance

type devices or other electronic devices.

5)

Product Particle Size - Can be measured directly by the

22
use

of

on-line

monitors

or

inferred

thru

mathematical

modelling.
This

is

very general run down of types or methods

control although specific devices are

readily available.

of
The

best choice of controllers will depend on the characteristics of

individual circuits.

CIRCUIT DESIGN
There are many methods available to size grinding mills and
to design grinding circuits.

The majority of these methods have

been around for some time and are largely a matter of using data
from laboratory tests, applying empirical equations to this data,
and fine tuning with correction factors that are often based on
accumulated

experience.

Different

manufacturers

of

grinding

mills use different methods and it is at times difficult to check

on

the

validity

of

the

sizing estimates

when

estimates

from

different sources are widely divergent.

This part of
sizing mills.

the discussion will focus on two methods of

The first will be the Bond method, which has been

considered by most as the standard for a long time.

The second

will be a computer simulation developed at Penn State University

which

utilizes

the

relatively

new

concept

grinding) of the size-mass balance theory.

radically

different

in

their

(as

applied

to

These two methods are

approach

and

exemplify

the

differences between the older (empirical relationships) and more

recent theories.

23

Bond Method
The Bond method of sizing grinding mills has enjoyed wide
acceptance in the mineral industry.

It has two major engineering

advantages, which are its simplicity and the fact that experience
has shown that it does work for many (but not all) circumstances.
The Bond method can be broken down into 5 major components:
1) Standardized grindability tests on the material
2) An empirical equation which is designed to convert the
test results to observed results in an 8 ft. diameter wet
overflow mill operating in closed circuit with a circulating
load of 250%.
3)

An empirical equation to allow for overall reduction

ratios in closed circuit operation

4)

Scale-up

relationships

to predict

results

for

larger

mills
5)

A series

of

empirical

correction

factors,

based

on

experience, to allow for milling conditions

By looking at each of these 5 steps in detail, a good idea
of the methodology of the Bond method can be obtained (Fig. 11).
Step 1:

Grindability Test

The Bond method does utilize a standard laboratory test

procedure to supply the basic data.

The feed for this test is

crushed to 100% -6 mesh and approximately 80% finer than 2000

micron.
weight

The ore charge to be used is specified at 700 cm3, the

of

which is determined by packing the material in a

graduated cylinder.

This charge is then placed dry in a 12 X 12

24
in. mill rotating at 70 rpm (85% of critical) with a specified
ball charge of:
43 balls

1.75 in.

67 balls

1.17 in.

10 balls - 1.0 in.

71 balls

0.75 in.

95 balls

0.61 in.

This ball charge should equal a total of 20.1 kilograms.

This charge, with the above conditions, is ground for a set
number of revolutions,
desired screen size.

removed from the mill and sieved at the

The undersize is removed and weighed while

the oversize is returned to the mill with fresh feed added to

reestablish the initial material charge weight.

The new charge

is reground and the process is continued until a constant 350%

circulating load is achieved.

At this point the net grams of

screen

mill

undersize

produced

per

is

revolution

determined

( Gbp) .
Step 2:

Conversion to an 8 ft. Mill

From the results of the grindability test

index

(Wi)

that

is

unique

to

the

ore

(Step 1) a work

being

tested

can

be

determined using the following equation:

eq.1

Wi

Where Pi
Pao

c1. 10) c 4. 4 5)

(P10.23)(Gbp)

0.8280J;:: Pao

= opening of test sieve in microns

= 80% passing size of the test product

~kwh/t
0

(microns)

Faa = 80% passing size of the test feed (microns)

25

Step 3:

Allowance for Size Reduction Ratio

Equation
calculate

calculates

the

specific

the

work

grinding

index

energy

which
from

is

the

used

to

following

equation:

eq.2

To

10 ~

~Pao

= Wi

Pao = 80%
Faa = 80%

Where

Step 4:

(10

~'

kwh/t

passing size of circuit product

passing size of circuit feed

Scale-up to Larger Mills

convert

diameter,

the

to

value

larger
of

the

mill
work

than
index

one

having

an

must

be

scaled

(Wi)

ft.

accordingly using the following equation:

eq. 3

Wi

=
=

allows

Wi

This

(Wi) test (2.44/0)0.2, D<3.8lm (12.5 ft.)

(Wi) test (0.914),
new

specific

energy

D>3.8lm (12.5 ft.)

to

be

calculated

for

larger mills from eq. 2.

Using the correct specific energy value (E) for the desired
mill diameter, the required shaft power for a desired feed rate
can be determined from:
Mp = QE, kw

eq. 4
Where

MP = shaft power

Q
The

shaft power,

feed rate

Mp,

can be related to mill power as a

function of mill dimension using:

26

eq. 5

Mp

Where:

= 7.33J

0c (l-0.937J)(l- Ogl
c)(pb
2 -100

LD2.3)

pb = density of grinding medium (t/M3)

0c = fraction of critical speed
J = formal ball loading

L = mill length
D = mill diameter
All values in eq. 5 have either been determined or can be
specified,

therefore

the

necessary

values

of

mill

length

necessary to give this power can be determined.

Step 5:

Corrections for Milling Conditions

Correction

factors

for

specific

mill

conditions

can

be

calculated or taken from tables to adjust the work index, Wi.

This may not be necessary, although equations and/or values
are available for conditions such as allowances for oversized
feed, fineness of grind, low reduction ratios and converting from
closed circuit to open circuit.

PSU GRINDING CIRCUIT SIMULATOR

A simulation of any physical process is a mathematical model
which behaves on computation in a manner identical to that of the
real process.

Generally, a simulation is only an approximation

to the real behavior, especially for a process as complicated as

milling and the mathematical models can be more or less complex
depending

on

how

closely

one wishes

to

simulate

the

real

situation.
The

Penn

State

University Grinding

Circuit

Simulator

27

utilizes

the

size-mass

grinding circuit.

rate

balance concept

in simulating a

This method uses two breakage concepts as its

basis (Fig. 12).

1) Specific Rate of Breakage - fractional rate of breakage
per unit of time per mass of the size particles are broken
from.

This is expressed mathematically by eq. 1 shown in

Fig. 12.
2) Primary Breakage Distribution Function - Description of
Function - Material breaks and the fragments produced are
mixed back in with the general mass of powder in the mill.
If this distribution of fragments can be measured before any
of the fragments are reselected for further breakage, the
result

is

the

Primary

Breakage

Distribution.

The

mathematical expression for this function is also shown in

Fig. 12.
These two principles sum up to define the overall basis of
the program as the rate of some size i particles leaving the mill
is equal to the rate of size i particles entering the mill plus
the rate of production of size i

particles from breakage of

larger particles within the mill minus the rate of disappearance

by breakage of size i particles within the mill.
The

details

computations ,

of

the

program

itself

(i.e.,

flowsheet,

etc . ) and the labor a tory procedures required to

generate much of

the input data are much too complex to be

included within this discussion.

However, once the necessary

data has been compiled, the program becomes very "user friendly,"

28
allowing

easy

input

of

the

data

as

well

as

changing

of

conditions.
The program allows the selection of 5 basic mill circuits,
which is an indication of the flexibility of the system.

These

circuits are (Fig. 14):

1) Open Circuit
2) Normal Closed Circuit
3) Reverse Closed Circuit
4) Open Circuit/Scalped Feed
5) Combined Closed Circuit
This is the first choice offered by the program.

Once the

proper circuit has been selected, it is shown on the monitor for

verification.

Following this selection, the program asks for the

input data.
The
circuit

required
includes

selectivity values,
time of grind.

input
feed

data
size

for

simulation

distribution,

of

the

desired

classifier

size

mill selection values, breakage values and

To be more specific, examples of required input

are:

No. of size intervals desired in evaluation

No. of different ball diameters in both simulated and test
mill
. Diameter of both simulated and test mill
. Maximum feed size
. Scale-up constants which correspond to effects of mill and

29

ball diameter on specific breakage rates and large

particle size corrections
. Operating conditions such as media charges, powder
charges, and critical speeds utilized in both the
simulated and the test mills
. Whether grinding is wet or dry
Parameters for specific breakage rate calculations
(determined from laboratory testing)
. Parameters for primary breakage distribution calculations
(determined from laboratory testing)

Four values are

required for each ball diameter

. Feed size distribution
. Classifier data such as dso values, sharpness index and
apparent by-pass
. Time of grind
After all
calculate

the data has been entered,

the

specific

distribution for
function.
the

rate

of

breakage,

the computer will

a

residence

time

the simulated mill and the primary breakage

The output generated will use these values to predict

performance

conditions.

of

the

simulated mill

under

the

desired

The predicted performance includes flowrates of each

stream in the simulated circuit, predicted size distributions for

each

stream,

classifier performance based on predicated mill

discharge and % circulating loads, if applicable.

This simulation can be extremely helpful to operators of
existing

mills

because

it

will

automatically predict

the

30

variation in mill output for different feed makeups, changed mill

conditions

and changed classifier parameters.

Alternatively,

from a design viewpoint, the ef.fects of these same parameters on

mill size, for a given feed rate, can also be predicted.
In contrast to other methods of design, this technique sizes
the mill based on the rates of breakage occurring in the mill.
The energy input necessary to run the mill then follows from the
size and mechanics necessary to turn the device.

This does seem

to be a far more logical approach rather than estimating the

energy first followed by making the mill big enough to consume
this amount of energy.

31
REFERENCES

Prasher, c. L., Crushing and Grinding Process Handbook, Wiley and

Sons Limited, New York, 1987.
Mular, A. L., Bhappu, R. B., Mineral Processing Plant Design, 2nd
Edition, AIME, New York, 1980.
Austin, L. G., Klimpel, R. R., Luckie, P. T., Process Engineering
of Size Reduction, AIME, New York, 1984.
Pryor, E. J., Mineral Processing,
Publishing Company, New York, 1965.

3rd

Edition,

Wills, B. A., Mineral Processing Technology,

Pergamon Press, New York, 1985.
Thomas, R., E/MJ Operating
McGraw-Hill, New York, 1977.

Handbook

of

Mineral

3rd

Elsevier

Edition,

Processing,

Brown, J. H., Unit Operations in Mineral Processing,

International Academic Services Ltd., Ontario, 1979.

-32-

Fig. 1.

Breakage Mechanisms

(a

I bl

(c I

Mechanisms or breakage: ta11mpac:t or compression, lbl c:hippins. lei abrasion.

-33-

Fig. 2.

Grinding Laws

Rittinger's Law
ER = CR (l/d2 - 1/d1)

Where:

ER = Specific Energy
CR = Constant
d2 = Product Size
d1 = Feed Size

Where:

EK
CK
d2
d1

Kirk's Law

= Specific Energy
= Constant
= Product Size
= Feed Size

Bond's Law
W = 10 Wi

..r;-

10 Wi

,j

Where:

w = Work Input
Wi = Work Index
P = 80% Passing Size of
Product
F = 80% Passing Size of
Feed

-34-

Fig. 3.

Patterns of Tumbling

. Empty lone

..

Oeoel zo,.e

Rorahol'l \

Impact lO"e
Abrasan
zone

C.oscaelnq
meeloum

Motion of c:har~~ in a tumbling mill.

Nc

42.3
= ---
" 0-d

Where

Nc
D
d

= Critical
.
=
=

Speed
Mill Diameter
Charge Diameter

-35Fig. 4.

Classification of Rod Mills

DISCHAAGl

Central peripheral dtscharge mill.

Dsc:llorge

End peripheral discharge mill.

Overflow mill.

-36-

Fig. 5.

Determination of Media Charge

M~:~

= diameter of topsize media in 1nches

= size in microns of the screen opening

M
F

which 800fa of the feed will pass

Wi = work index
c. = percent of critical speed
S = specific gravity of feed
0
diameter of mill inside liners in feet
K = constant, the value of which is 200 for ball
mills and 300 for rod mills

TABLE OF INITIAL CHARGES

Rod
Ol.mtw
Cinch eo,

Rod 11. .
I

11

4%

17
18
15
13
10
10

4
J%

s
Z%

I
1%

.....

Dl.,...._
(IIIChft,
I
4%

S%
1

2%

Dltrtbudo~t

For Startup Charve

c.,. Wleht,

4'4

21
1!1
18
17
15
10

24
2l
20
18
15

J%

2&
2Z
20
17
t5

30
211

23
21

...111. . Dltrtllvtl- For Startup Charee

1"- Woieht,
I '4% I 4 J% : S
17.0.
25.0 111.0
20.0 30.0 20.0

2'16

1'4 1'4

15.0 21.5 32.0 22.0

---------10.0 14.0 21.0 35.~-~a.o- ,..,.--...;....--...;..._----

1.1 12.5 tJ.o""ii:o";.l;.:;o~,...,....-----~-

1.a u

a.a a a i2.c iio ll.o

3.4 5.3 9.2 16.535:'0:8.0

-..;.''.::11.~--.:..:1.:..1-;.:'-;.2....;:.:2.J -i.:: 6.1 13.0 li~ 30.0._,_......:,_..__
1
1.3 1.~ u
2.1 8.4 11.0 'jz.o .;:22:;.;.o:-=-=,__'1'16

2.1

2.4

_,,.;-_ _ _ _ _..;.'-'
,.

u
2.1

3.1 a.o t4.5 s2.o 24.0.

4.5 12.0 23.5 28.0 78 0 ,100

-37-

Fig. 6.

Grinding Action in Rod Mills

a :
B

I Wi I I I

:u ,,.., ....

::oiO~~ ....... ~

-========::=;:;;;;.;;,~?
I

2'

_]

Particle' Distributum '''

lirt'

R,,J .\Ill/

Fig. 7.

Classification of Ball Mills

(X)

F\
.

lirate

1.ll~harge

mill.

DIIChargt

-39-

Fig. 8.

Open Circuit Grinding

CRUDE ORE FEED

PRODUCT
ROD WILL
BALL MILL

CRUDE ORE FEED

~~ooue:
ROD MILL

CRUDE ORE FEED

b~
8ALL WILL

Open-Circuit Grinding

-40-

Fig. 9.

Closed Circuit Grinding

CRUD!

OIIE

PRODUCT

CYCLONE

IALL lULL
IALI. MILL

Ball Hill-Ball Hill,

Closed-Circuit Griodina

ROD MILL
PRODUCT
CYCLONE

PRODUCT

BALl. NiLL

IALL MILl.

Single-Stage
Closed-Circuit Grinding

Two Stages Grinding

and Classification

PRODUCT
CYCLONE

CRUDE ORE FEED

BALL MILL

ROD MILL

BALL MILL

Rod Hill-Sall Hill,

Closed-Circuit Grinding

Fig. 10.

Grinding Circuit Variables

Portcle s1re dslnbuhon/

densllv I flowrote
Water

. flowrole

.....
I

Pump
speed

Flowrole/
densotv .1
pressure

Orinding circuit control variables.

-42-

Fig. 11.

eq.1

Wi

Where:

= (1.10)(4.45)

= Work Index
= Net gms per

Wi
Gbp

Bond's Equations

(p 1 0.23)(Gbp)

P
revolution
F

0.82~0r---

VP80

= 80% Passing Size

test product
= 80% passing size

of
of

test feed

eq.2

~)

= Wi

Where P = 80% passing size of circuit product

F = 80% passing size of circuit feed

Wi
(Wi) test (2.44/D) 0 2, D<3.81m (12.5 ft.)
Wi = (Wi) test {0.914),
D>3.81m (12.5 ft.)

eq. 3

Mp = QE,

eq. 4

eq. 5

Mp = 7.33J
Where:

Pb
~c

L
D

~c

Where:

= shaft
=

diameter

power
feed rate

~(Pb LD2.~
~

(1-0.937J)!i- 0-1

= density of grinding medium

= fraction of critical speed
= formal ball loading
= mill length

= mill

Mp

(t/M3)

Fig. 12.

Equations for Size Mass Balance Method

SPECIFIC RATE OF SREAKAGE

FUNCTION - SCALE UP

FORMULAS

~he"Ce

l+(d /dT)

).*

( l+(d */d).\ *

~..,
D-<J . ....,._

(D/DT) 1

c3 =

3.8~1

~1-u

Li~J.

( DT) ( 3. 3)

CS

-0.1

) '

l+exp(l5.7(~

T-0.94) l

(~ cT-O.l 'l+exp[l5.7(~ c -0.34)

BREAKAGE DISTRIBUTION

FUNCTION

n~ 1~1

-43-

3m

-44FIGURE 13 - NCMENCLATURE llriLIZED IN PSU SINULAIDR

NOMENCLATURE
A

di/d
(-)
m n max

constant in

Eq. 1 which defines the effect of hold-up on breakage

rates (-)
d

ball diameter (L)

dr

ball diameter used in laboratory

make-up ball diameter (L)

max

t~st

mill (L)

d .
mln

minimum diameter of ball in equilibrium ball charge (L)

mean internal mill diameter

DT

D for the laboratory test mill (L)

volume fraction of mill filled by ball bed (-)

JT

J for the laboratory test (-)

mk

weight fraction of ball charge of ball size class k (-)

exponent in

(L)

Eq. 1, exponent of effect of ball diameter on S?ecific

breakage rates (-)

N
1

exponent in Eq. 1, exponent of effect of

~ill cia~ecer

on S?ecific

breakage rates (-)

N
2

exponent in Eq. 1, exponent of effect of mill

dia~eter

on large ?article

size correction (-)

s1

specific rate of breakage of particle size interval i, fraction

per unit time (1/T)

fractional interstitial filling of voids of ball bed by particles (-)

UT

U for the laboratory test mill (-)

-45x

upper size of particle size interval 1 (L)

standard size (L)

d/d

aT

pre-exponential factor in Eq. 1. proportionality factor for

max

(L)

specific breakage rates (1/T)

exponential factor in Eq. 1. defines variation of specific

breakage rates with particle size (-)

factor in Eq. l, defines the decrease in specific breakage ra:es

of large particles (-)

uT

the particle size for which the specific breakage rate is

o~e-~alf

chat ex'Pect.ed from S. = :.:.(x./x )~. in the laboratory test :=~ill ;.rich
~

ball diameter dT (L)

~c

rotational speed of mill as a fraction of critical speed (-)

~

for the laboratory test mill

(-)

~j

is t.he intercept of the breakage distribution function - eQn f2)

is the slope of t~e 16wer side of the ~reakage distri~utians

fun~tion - eon (2)

is the slope of the upper

function

sirl~

of

t~e

~reaka~e distri~ut~on

Fig. 14.

Grinding Circuits in PSU Grinding Circuit Simulator

lal 0111"1 Cin:uit

lbl Normal Closed Crcun

lei Reverse Closed Cn:uit

ldl Open Circuit With

Se~lpect

Feed

General single-stage mill circuit with pre- and post-classifier: the

combined closed circuit.

-46-