..J ._.
1111
su .
Js
11.
\~11
. .
S..
's l!ook tlo\o\' S n>w liglt on the internal eclHics of mills. lt covers among
others thc following aspects: Ball, tube,
rod, ebl>le, IHtcl, gratedischarge, trunnionoverfiow, larc]inage and airswept
mills; open and closedcircuit grinding;
dry and wet milling; contamination of
products. Single ball in mill; cascading;
cataracting; equilibrium surface of
charge, etc. The power to drive a mill.
Perfect crystals; heterogeneous boclies;
crushing a single particle; scale effects;
effects of impact velocity. Dirnensional
analysis; rate of g1inding. Effect of pulp
consistency preferential grinding; influence of shape of grinding media; size
distribution of grinding media: rate of
flow through continuous mills. Surging;
lifters; vib!'ation frequency spectrum: clestruction of geal'ing. Rose 's tleory of
metal \vear; factors affecting ball wear;
influence of coosion. The l'ole of aclditives in milling.
TR EA TI SE
IN TE RN AL ME CH AN IC S OF
BALL, TU BE AN D
R O D MILLS
by
. . RO SE
D.Sc .(Eng .) , Ph.D .
AND
R.
. .
SU LLI VA N
B.Sc.(Eng .)
~
'
1 9 5 8
CHE MIC AL PUB LIS HIN G CO., INC
.
212 Fijt h Ave nue
New York , . .
FOREWORD
by
PROFEssoR DR .ING. RoBERT MELDAU
First published
1958
Chemical Publisl1in g Co., lnc.
e1 York
wi l1. onsidera ble pleasure that write this Foreword
1 l oo k , written by Dr. Rose, d evoted to the analysis of
l 1 i 1(( n al p rocesses of the family of mills which h ave
. .
I l g grinding media.
I ) , I ose is wellknown to the F achgruppe "Staub I i l<" of the V erein D eutsch er Inge nieure, for his
l in Germ a ny on his researches into milling and
l 111 gen eral properties of p owdered m a terials and
1 y knowled ge of him and of the researches which are
111 ; d out under his direction, a m sure tha t Dr.
I ' : 1 d Mr. Sulliva n are well equipped to write on the
l j .
'I'IH d for a book d evoted to such an analysis is
1 \ l wl1en the great number of " tumbling" mills used
11 ls tr, their low efficien cy a nd the great resista nce
i 111 y have offered to a ny substantial improvement,
1 II < l mind. Thus, this book tackles a problem to
u h thou ght has b een d evoted during at least a
1 l : nd to the solution ofwhich every advance will b e
I 1 )( .
I ) '. I s uses, with convincing r esults, the m ethod of
l ' i n al analysis as a preferred approach to his probl r : d it a ppears tha t the statisti cal b eh aviour of fine
1 l nds itself to su ch treatment, to a rema rka ble
'
Prir ted in t l e Ur it e<l States of A 11 e rica
',
k in milling terms; may the mill of Dr. Rose's
l ; I S yield ever finer products for the use of those
5
BALL, TUBE AND ROD MILLS
..
interested in this important subject! believe th.a~ t~s
book will provide improved theory and als<? furns t e
means whereby those interested c.a; obtan a ?eeper
understanding of the internal mechancs of such mlls.
PREFACE
ROBERT MELDAU
PoL YTECHNISCHE HocscHULE,
HANNOVER.
\ w i l Ll1.c previous volumes in the same series, the object
/ l : >r sent work is to bring together the data relating
n part ofthe field ofscience dealing with powdered
11 1 \ ' ' i : l s , in this case to the internal mechanics ofthe ball,
I I r d mills, to analyse these data in the light of the
\ i ' ' xperience and researches and, as far as is pos i i w itl the present state ofknowledge ofthe subject, to
. 1 111 ll1 I esults in a systematized form.
1ti has been given to the ball, tube and rod mill
/ 1 , : , <lthough the results of numerous tests on these
l i H 'R have been published, it is believed that no cm
/ 11 1 analysis has yet been made. Thus, fol" example,
I I f.\' 11 many investigations have been carried out to
I ~ l influence of the various parameters. defining the
/ , ristics of the mill and the charge, as far as is known
l 1 ' : previously been offered no formula for the calcu ' h instantaneous rate of production of specific
11 I ' ' m a knowledge ofthese parameters. ln Chapter
, l w: : r, such a formula is provisionally presented.
I i :{g sted that the formula is, at present, subject to
111 I l' f, say, 50 % , and SO is to be regarded as giing
1 / roximation only to the desired result. Neverthel , i iN /) lieved that the reduction of these data to such
1 I l : is a considerable advance along the road to
l tl subject on a firm foundation and that the
ii of details, with consequent improvement in
. : y, is largely a matter of wellconducted experi1111 11 1, : v though the highest accuracy is lacking in the
.ll /) n ul a , the work is believed to be of some
; I alu e since the influence of nearly every variable
1.1 I ; n alysed in the light of the available data. Thus,
l 11 1 : l , the mill operator will find information as to
7
LS
BAL L, TUB E J\ND ROD MIL
ball s~ze upo n t~e
the prob able effect of vaation of, say,
umpton o.f a ~11
cons
er
grin ding char acte risti cs and pow
~e, the d1rect?
iden
~onf
e
som
and so h e may pred ict, with
m1ll para met ers IS
in whic h a cont emp lated chan ge 1n the
.
.
likely to alte r the mill perf orm ance
of
Cs
hani
mec
rnal
inte
the
to
n
The r easons for restr ictio
g:
the mill are seve ral; these reasons bein
ral desi gn and
(1) Seve ral books deal ing with the gene
tence, e.g.
oper atio n of the tum blin g mill are in exis
rung.
Her r Carl Mitt ag's Hartzerkleine
with the inte rnal
(2) As far as is know n no book deal ing
.
lable
processes in a mill is avai
of design and oper a(3) have trea ted the prob lems
pres ent field, wou ld
the
to
tion
addi
in
tion of mills,
the Auth ors,
have defe ated one of the objects of
at mod erat e
me
volu
ll
whic h was to prod uce a sma
~tur e on the
liter
the
i~
.gap
a
pric e whic h wou ld fill
e mor e on
wnt
to
n
nt
1nte
the
not
was
subj ect. It
uate ly
adeq
ady
alre
part s of the field whic h are
cove red.
chos en since
The title "Int erna l Mech~nics" has been
~ithin, the
ing
inat
orig
or
in,
rring
only phen ome na occu
1ls such as
deta
s,
shell of the mill have been trea ted. Thu
have not
ings
bear
and
classifiers, feed m echanisms, gear ing
been
has
ing
gear
of
n
kdow
brea
been cons ider ed. The
since
,
ever
how
",
ging
"sur
with
ion
unct
cons ider ed in conj
ofte n arise from
it is the Auth ors' belie f that gear failures
on (surg ing)
llati
osci
the
by
vibr atio ns whic h are initi ated
.
shell
the
in
of' the mill char ge with
a .mill has been
Sim ilarl y, the pow er requ ired to driv e
IS clea rly conand
stud ied in d etail since the pow er dem
in the mill
with
ge
char
the
of
n
troll ed by the co'nfiguratio
.
mics
dyna
rnal
inte
of
and is so a mat ter
intro duct ion of ~ sh~rt re apol ogy is offe red for the
of comr.rnnut~ of
view of the pres ent state of know ledg e
operatns tak1ng
the
of
e
ledg
solid bodies since a know
9
ied
appl
y
tivel
j wi tl1i11 the mill cann ot be mos t effec ess for whic h
proc
e
ofth
e
ledg
know
und
kgro
i : b
l 111 11 iR r quired.
revi ewin g the
l;: y,. t~e intr~d.ucti<;>n of a chap ter grou nds that
the
on
fied
justi
Is
tlves
addl
l1 1 / mill ing
ify the beha viou r of a
t I ~~ l Hl< nces can grea tly mod
so, even thou gh it is
and
"' 111 i,~ .turi ng com min utio n
actio n is phys ical,
this
her
~het
11 1 l ly kno~n
tted from a book
I 1 : l J' echanical, It cann ot be omi
atio ns ofth e
oper
rnal
/ rts to deal with the inte
mill.
'
(
.
l
give n to man y
1 , \' . !1 ugh final answers .can not y~t be
n and appl icati on
l l )I / 1 ms enco untere~ lll. t~e d es1g that usef ullin es
ed
hop
IS
l l l : /1 , ube and rod m1ll, It
r eade r.
l will be sugg este d to the
wor k is not a tet
this
that
ized
real
lly
\\! i I I i is fu
pted theo ries
lI i l s nse that wor ked out and acce
in whi ch is
ual
man
gn
'" ' 1l cl, nor is it a desi
ul data for
usef
and
ulae
form
of
; 111 I a ollec tion
ofin tere st
be
/ I : id oper ator , it is hop ed that it will ator , to the
oper
the
l \' l t .the rnill designer, to
read er with an
11111 / ng1neer and to the gene ral
ing.
neer
engi
and
"'',' i mcc ha?ics
ness to all the
I l1ors w1sh to express thei r inde bted
in part icul ar
and
,
i w l sc works have been cited
aney whose
Dev
and
hill
Cog
.
1 ~ / .Jol111 ross and M essrs
to the U.S.
l I . : .been quo ted at grea t leng th,
on of Min ituti
\ 111 11 1 111 Mn es and to the Ame rica n Inst nission to reperr
for
rs,
'''1' I al~urgical Eng inee
ous publ icati ons
1"" '''''' t ns1ve data from the vari
1111 I(IIIII '( / jn the text.
ture rs who have
I / l <r also due to all the man ufac
r nam es bein g
the!
,
d~ta
~ "''I l i.ll.ust.ratio ns and ot~er IllustratiOllS for thei r
ate
ropn
app
l/ l r,I.Jf1St the
e the mate rial
1
 t nnd perm issio n to repr oduc
repr oduc e Figs
to
"1111''''' ' he ... for perm issio n
reas en for per~
And
r
esso
Prof
I I, :. and 3.11, to
J. . Con nor
Dr.
to
and
4.11
Fig.
1111 '"" prod uce
PRE FAC E
BALL, TUBE AND ROD MILLS
10
and the University ofLondon for permission to reproduce
Figs. 2.4 and 2.5.
Finally, our best thanks are due to Mr.J. F. Hinsley and
Monsieur . Fobelets, who have kindly made available
the results of their researches which have extended over
many years and upon whose mathematical analyses much
of the treatment of the motion and configuration of the
mill charge is based, for their personal assistance and advice. Finally, thanks are due to all whose writings have
been consulted and to all who have contributed to the
work by advice and criticism.
H.E.R.
PowDER SciENCE LABORATORY
R.M.E.S.
KING's CoLLEGE, LoNDON
CONTENTS
I
I,
1 1 '1R
PAGE
NTRODUCTION
17
of millig;
Ball mill;
) finitio of millig; Objects
I I Jnill ; Rod mill; Pebble mill; Batch mills; Grate
l1 1 l: ge mills; rui verflw mills; Hardige
1111 11 ; i rswept mill ; Ope ad clsedcircuit grindig;
I 11 y ill ig ; Wet milling; Cotaminati of prducts
'1
'
OF
MILL CHARGE
35
i mill; Mass f balls; Iterrelatiship
w cefficiet of fricti ad umber f balls ;
< : 1 t li ng; C ataractig; Equilibrium surface f charge;
II ll ':j ctories; Theory f Davis; Iterferece betwee
I ll t rajectory ; Thery of Steiger; Magnitude
, 'i ).t l b all
I"
" IIy ng" ch a rge
ad "rllig"
POWER
charge.
DRIVE
MILL
69
I' : .s up which pwer deeds; Difficulties
1\' 111 :sts large mills; Tests smallscale mdels
l vstigati by Rose ad Evas; The ivestigatio~
ampbell ad Coghill; The equation ofDavis
l <:1w,
I i f d; Equatio ofHacock; The effect of
1l1' 1 l 1 : . teristics; Extensio to a rd mill Method of
'
' 11 1 l tairmad
<>
<
COMMINUTION OF SOLID
109
)I) [ES
I l 1 1 : e of slid bodies; Perfect crystals; Mosaic
1l ; I fc erogeeous bdies; Discussi of cm
i I ; Distributio of eergy a body durig
' shig a sigle particle; Crushig i a
I 111 I II ;
,, ' l"; Rittiger's Law; Kick's Law; Bod's
11
CONT ENTS
BALL , TUBE AND ROD MILLS
12
: D
l
theory ; The_ory of Carey and Stairm and; Geom etrica
t
impac
of
s
Effect
;
effects
Scale
ng;
crushi
in
rity
simi1a
velocity
5.
PROC ESS OF GRIN DING WITH IN
MILL
' ''bles of Mill Filling Param eters
140
OTH ER CON SIDE RAT IONS RELA TING
MILL
CHA RAC TERI STIC S OF
178
of
Prefer ential grindi ng in rod and ball mills; Effect
ball
of
Effect
ng;
grindi
ential
prefer
pulp consis tency
e
hardn ess the rate of grindi ng; Metal wear and relativ
of
shape
of
nce
Influe
les;
partic
and
hardn ess of balls
;
grindi ng media ; Size distrib ution of grindi ng media
s;
Lifter
;
media
ng
grindi
of
size
of
D avis distrib ution
R ate of flow throug h contin uous mills
7.
SUR GING AND VIBR ATIO N
MILL 203
Surgi ng; Tests of Hinsle y; Criterion of R ose and
the
B1unt ; Impo rtance of the coefficient of friction of
cDestru
um;
spectr
ncy
freque
tion
charg e; Lifter s; Vibra
tion of gearin g
8.
AL
WEA R
221
MILL S
s;
Rose's theory of metal wear; Obser vation of Engel
wear;
ball
ing
affect
s
Factor
es;
surfac
of
uring
Armo
Ana1ysis of the variou s factor s; Theor y of Hukk i; Influence of corros ion
9.
ROL E OF ADD ITIV ES
252
Facto rs upon which grindi ng depen ds; Treat ment by
dimen sional analys is; The influe nce of the differe nt paraof
meter s on the rate of grindi ng; Equat ion for rate
es
surfac
mill
of
"
grindi ng; 'rmouring
6.
13
MILL ING
Defin ition of additi ves; Funct ion of additi ves; Wet
ive;
millin g; Dry millin g; Dispe rsing agents effect
es ;
Theories of additi ves ; Clean ing of the mill surfac
ucing
introd
of
anger
D
n;
frictio
of
cient
Effect on coeffi
surgin g; Polar and nonp olar additi ves
236
i ; Index
r~ 1111 Index
253
257
LIST OF SYMBOLS
/
I
I
"'
11
I )i:n ter of particle
l )in n ter ofball (or rod)
I >i n ter of mill shell
\ . ' of naturallogarithms
( : ~e !Ii ient of restitution
: r~y necessary to change the surface of p owder by
amount
: 'f.\ necessary to change the surface of m etal ball by
1111 amount
( : 11 i .ient offriction
\ t I ation due to gravity
I I l ss [ material or Height of lifter
of ball: H ardness [
I ( ~ ive hardness (Hardness
le)
ll izo ntal displacement of centre of gravity of mill
: ge
J ' ti o nal
ball filling of mill
lume of balls + Volume of space between balls
Volume of mill
\ I ;meter or Specific rate of grinding
\ 1 : ' nme ter
l , )..\ 111
fmill
\ i . l [ product per revolution
: J t l [ rotation of shell in r.p.m. or Number of lifters
l
rotation of shell
speed of rotation
l 'w drive the mill
I . t iy ofballs in ftight
1\ l ll l 1o u g hput (mass per unit time)
'" I of element of charge from centre of rotation
tli ofmill shell or R eduction ratio
fi surface (Surface per unit volume)
t : 11 i n l
,,
15
16
BALL, TUBE AND ROD MILLS
Total surface
Time of grindin g
Time of revolut ion of mill
Time for elemen t of charge to .traverse circula r path
Time for elemen t of charge to traverse parabo lic path
Diame ter of exit from mill
Pulp consistency (volume of solid to volume of liq1.1id)
Velocity of projection of an elemen t of charge or Fractional filling of solid
_ Volum e of powder + Volum e of space betwee n particles
Volum e ofspac e betwee n the balls
Wear of a single ball
Total wear of balls
Ratio of circula ting load to totallo ad
Angle of displacement of centre f gravity or Angle of
project ion
Parame tric angle
Angle of repose of charge
Angle betwee n rays of the equian gular spiral or Mean
density of the mill charge
Semiangle subtend ed by the charge at the centre of the
mill
Parame tric angle
Angula r velocity of mill shell (radfsec)
Angle of displacement of centre of gravity of mill charge
from the vertica l
Coefficient of friction of charge
Absolute viscosity
Kinem atic viscosity
Weight ofball charge + Weight of powder charge
Ratio:
W eight of ball charge
Density of grindin g media or Radius of curvatu re of ball
path
Density of powder
Denotes "a function of"
parame ter
Angula r velocity of precession of the mill charge or element of charge
CHAPT ER
INTR ODU CTIO N
l r y industr ies the final produc t, or the raw materi al
1 11 11 s ~ge of the manuf acturin g process , is in powde red
l ,111 l conse~uen.ce the rapid and ~heap prepar ation
1, W< I r d matena ls 1s a matter of cons1d erable econom ic
.
; 1/111 1 : 111
prepar ed
11 cases the powde red materi al may be
n a
solutio
from
itation
I 11 ly; fl r examp le by precip
of
type~
certain
of
ation
prepar
the
in
j ' wl1i h is used
fine
a
of
drying
vacuum
the
by
or
dru~s,
and
:~
l
'
j
111 I h e ma.ten al, a. process which is widely adopte d
s
l1 : > ~ 1 arat10n ofm1lk powde r, soluble coffee extract
of
er
howev
are
es
process
Such
ts.
produc
1 l l a r
I I ' l . : pplica?ili~y and in by far the grea~est numb; r of
l I i rr: I pplicati~ns t?e powde red materi als are pre1 1111l Iy Lhe reduct1 0n, 1n some form of mill, of the grain
1 / .' m a terial having an initial size larger than that
j l rn tl e final produc t. These process cs for the reduc l l~ 1 article size of a granul ar materi al are known
i 11 i _g" or "grind ing" and it appear s that these names
l rtcrchangeably, there being no accept ed tech111
l l i ll ( r ntiatio n betwce n the two.
I s ns for the grindin g of industr ial materi als are
1 s 'b ut the princip al reasons may be summa rized
/ 1 I llowing headin gs :
I ) ' beration of an econom ically import ant
: 'n al from the unrequ ired constit uents of a
lure.
") '1 xposur e of a large surface per unit mass of
: rial in order to facilita te some ch emical process .
17
BAL L, TUB E AND ROD MILL S
18
redu ce the mate rial to the desired form of the
final prod uct.
(4) satisfy mark et requ irem ents.
(3)
ral dressing,
Exam ples of the first two classes occu r in mine
desired ore
the
ate
in whic h size redu ction is used to liber
a form in
to
ore
the
ce
redu
to
from the gang ue and also
reage nts.
ing
leach
the
to
ce
surfa
large
a
whic h it prese nts
cinal
medi
y
man
U nder the third head ing may be classed
in,
izers
fertil
tuffs,
and phar mace utica l prod ucts, foods
size
the
falls
ing
head
h
fourt
the
r
secticides, etc., and unde
often being
redu ction of mine ral ores, etc.; these mate rials
hand ling,
in
redu ced to parti cles of mod erate size for ease
ofshi ps.
holds
the
into
stori ng and loadi ng into trucks and
prosuch
to
cted
subje
be
to
er
The quan tity of powd
the
to
rding
acco
ly
wide
s
varie
ction
redu
cesses of size
l
utica
mace
phar
indu stries invol ved, for exam ple in the
be
can
.
m
annu
per
lved
indu stries the quan titite s invo
of certa in
meas ured in term s of a few tons, or in the case
the ceme nt
drugs, possibly a few poun ds; wher eas in
of millions
tens
into
indu stry the quan tities invo lved run
prod uced ,
ng
havi
alone
stry
of tons; the British ceme nt indu
ent in
Cem
and
Portl
of
tons
on
milli
in roun d figures, 12t
1955.
man y
For the prep arati on ofsm all quan titite s ofpo wder freis
mill
ball
the
so,
even
but,
types of mill are avail able
of
tities
quan
st
large
the
of
ing
grind
the
quen tly used. For
st
almo
used
is
mate rial howe ver, the ball, tube or rod mill
h
whic
mill
of
types
exclusively, since these are the only
.
itude
magn
ired
requ
the
of
possess throu ghpu t capa city
ns, and
Thus , it is the extre mely wide rang e of appli catio conh
whic
mills
the grea t indu stria l impo rtanc e, of the
justi the
s
form
h
whic
ly
stitu te the "tum bling mill" fami
.
ficat ion for the prese nt work
l ball mills
The grea t rang e of sizes cove red by indu stria
the first
ln
1.2.
is well exemplified by Fig. 1.1 and Fig.
t 1litre
abou
of
mill
h
batc
y
illus tratio n is show n a labo rator
in the
used
mill
tube
a
n
show
is
1.2
capa city, whilst in Fig.
F.
J.J
Labor atory Batch M iJJ
F.
1.2
I" a"ge lndus trial Tube Mill
(Pascal Engineering Co. Ltd. )
(fr Alen & Co. Ltd.)
BALL, TUBE AND ROD MILLS
20
cement industry the tube having a diameter of about
8 ft and 1ength of about 45 ft. In Fig. 1.3 is shown a 1arge
ball mill, designed for the dry. g~inding of ~imesto~e, do1omite, quartz, refractory and sm1ar m~ten<l:1s; ths type of
mill being made in a series of sizes havng dameters ra~g
ing from about 26 in. to 108 in., with t~e corres:ondng
1engths of drum ranging from about 15 . to 55 .
21
INTRODUCTION
tl1is point it is perhaps of value to study the
tH clature used in connection with the mills under
i I ration, but it must be emphasized that the 1ines of
l : ation between the types to which the names are
i i~ are not very definite.
' : broadest c1assification
appears to be:
( I ) Ball mills.
(' 1 'fub e mi1ls.
('I Rod mills.
( I bble mills.
' crm
"ball mill" is usually applied to a mill in
1 ;t,he gri~ding. media are bodies of spherical form
I ll ) and whch the length of the mill is of the same
l : ~ tl1e diameter ofthe mill body; in rough figures the
I 11 I , say, one to three times the diameter of the mill.
ll tt " ~b e mill': is a ~ill in which the grinding bodies
I i ii' t Cal but l whch the length of the mill body is
11
i proportion to the diameter than is the case of the
I ll 11 i11 "; in fact the length to diameter ratio is often of
l l
ften to one.
' l mill is a mill in which the grinding bodies are
I ' ds instead of balls, and, in order to avoid
, l " of the rods, the length to diameter ratio of
1 I i is usually within the range ofabout 15 to 1 and
"
ll 11 I" '' ble mill is the name associated with a mill in
1 l I ,;r.inding bodies are natural pebbles or possibly
l s ; the general proportions of the mill being
" l ; ll
FIG. 1.3
Industrial Ball Mill
mill".
1 11 I' noticed that the differentiation between ball
11 t l l u~ e ~11 arises only from the different length
l 11 lt' ' ats nvolved, and not from any difference
t l lal principles. The rod mill however differs
'
'
.
' ' I n t hat the grinding bodies are Ods instead of
1 , w lil ~ a pebble mill is a ball mill in which the
1 l li s are ofnatural stone or of ceramic material.
ll 1 t ' [ mill normally encountered can further be
MIL LS
BAL L, TUB E AND ROD
22
of feeding the raw
classified acco rdin g to the met hod l pro duc t from the
fina
mat eria l to, and the rem ova l of the
mill.
The se classes are:
Bat ch mills.
Gra te disc harg e mills.
Tru nni on overflow mills.
Air swe pt mills.
stra ted in Fig. 1.4 and
The prin cipl es of these mills are illu
Fig. 1.5.
mills, Fig . 1.4a, the
As the nam e implies, i~ the batc h
ed into the mill in a
load
is
cha rge of pow der to be gro und
ess is com plet ed, is rebatc h and , afte r the grin ding proc
mod e of ope rati on can
mov ed in a batc h. Cle arly suc h a
mod erat e sizes; say to
or
only be app lied to mills of sma ll
abo ut 7 ft long.
by
eter
mills of up to abo ut 7 ft diam
tinu ous sup ply and
con
a
mill
of
s
type
In the thre e othe r
, and these met hod s
rem ova l of the pow der is mai ntai ned s.
size
est
are app lica ble to mills of the larg
, a diap hra gm in
In the grat e disc harg e mill, Fig. 1.4b rge to one end of
cha
the form of a grat ing confines the ball
diap hra gm and the
the
een
betw
e
spac
the
the mill and
the rem ova l of the
fo"r
p
othe r end of the mill houses a scoo
fed in thro ugh a
is
l
eria
mat
raw
gro und mat eria l. The
ofth e mill and dur ing
holl ow trun nion at the entr ance end
afte r whi ch it passes
grin ding traverses the ball cha rge;
and rem ove d by the
up
ed
thro ugh the grat ing and is pick
ugh peri phe ral ports,
discharg e scoop or is disc harg ed thro
men tion that scoops ar
In this con nec tion , it is rele van t to
the lite ratu re. In the
in
sometimes refe rred to as "lif ters "
ter" will be confined
"lif
term
present wor k, the use of the
of mil llin er con stru cto the desc ript ion of a cert ain form
ord er to pro mot e th
tion, fitte d with "lif ter bar s" in
be desc ribe d in a late r
tum blin g of the cha rge, whi ch will
section.
1.4c the raw mat eria l
In the trun nion overflow mill, Fig.
at one end of the mill
is fed in thro ugh a holl ow trun nio n
(1)
(2)
(3)
(4)
()
Batc h Mill .
Grt~ .
I J)( .
Surt.c( Ot Powdv.r M~s
Surtc( Ot Bll Mas~ .
Discharge Scoo.
(b )Grte Dischrg~ Mill.
S urtc e 01 Pow drr s~
SurtceOtBa/1 Moss.
Oertlow
'~;;~~~~t$,~i~~F,..;;;j~Trunnion.
Trun nion Orrflow Mill . (c)
~DustHood.
r  ::.,0
Dischr9e Port s .
I
I
Disc harg e
Hopper.
P~riph~ ral Dischargfl' Mill
FG.
1.4
(d)
24
INTRODUC TION
BALL, TUBE AND ROD MILLS
25
1 i: l maintain the charge at the working level. The
I 11 11 such a closedcircu it plant is shown in Fig. 1.5.
I 1( t'inding in a batch mill, a grate discharge mill or
11 11 verflow mill may be p erformed either "wet" or
l " 1 l > l t clearly in an "airswept" mill the grinding
1111 l trried out in the dry state.
I " v' l" grinding the material to b e ground is fed to
t II the form of a "pulp" containing, p erhaps, 50 %
1 l l f water. In dry grinding, however, the granular
1 , l is fed into the mill in a sensibly dry condition; the
11! ~ e present being the few p ercent normally held
fEED
HDPPER
1111 1)Owcler . Both the grate mill and the trunnion overI
ll may be operated in either of two distinct ways,
1 11 'l'Sp ectively as ''o encircuit" operation and
1 I I r uit" oeration, and, furthermore , either of
I 1 t ' l l1 ds of operation m ay b e worked either "wet" or
",
1 1  ircuit grinding the mill product is not subject
\ l ~ssifcation, other than that which occurs witlin
I 1 II I) r eason of selective grinding of the larger parl , . I no part of it is, returned to the mill for further
1 . I n closedcircu it grinding, however, the prolt 1 l mill is conveyed to some form of classifier, and
l . l al coarser than the required m aximum size is
111 I the mill feed; suffiient n ew feed material be1
l i I to maintain the charge within the mill at the
1 vel. For wet grinding a hydraulic classifier is
l 1 i s for dry grinding an air separator, normally of
ELECTRIC R
"'IRICONE .. ILL
(Hardinge G_o. In c.)
Fc.
1.5
Layout of a typical closedcircuit plant
d
"overflows" at the other end. In
an_d the grouhnd pro u~~e grating and d~scharge scoop
this case, t erere,
are eliminated.
d. h arge mill is sh own in Fig.
variant of the grate sc .
.
. .
d b th .
.
h' h the disch ar ge .scoop s elmnate
l .4d,_ ~ w c . h l d' h arge ports  with a sutabl~
provsn of p enp er a
sc
. '
.
"d t" hood at the exit end of the mll._
d'fli
us
. '
t mill the body of whch does not cr
In an arswep
'
. f
trunnion overflow
significantly in general d~s~n . ro;ss~d through the mi\1
mill, a powerful stream ~ti~~~ss rodu ced by the grindin~
and removes _the finer pa owd!rladen air is then passc<l
process. Ths st:eam of "oversize" material, rej ectcl
through the_ class~fier and tdh~ the mill with sufficient fecd
by the classfier, s returne
I 1
type, is necessary.
m ethod of operation of a mill eliminl I n nveyor system which was often used in earlier
1 l ) Innt to transport the product to the classifier and
1 111 Hi zc material from the classifier back to the mill.
I . , lowever, that such a system can only b e used
1 lll y when the m aterial undergoing treatment is
I l y light, and the particles produced are sufficiently
111 ll , I ' le desired final product to b e entrained in the
l . Furthermor e, the removal of the powder from
I " : swept"
26
I.
BALL, TUBE AND ROD MILLS
the air stream, or its utilization whilst suspended in the
stream must be practicable and, for these reasons, a
typical application of systems of this type is in connection
with the grinding of coal for pulverized fuel firing, sincc,
in this case, the fuel may be burnt directly in the air stream
which cares it.
Within the classes of mills enumerated above there arc
a number ofvariations; for example there occur in practicc
mills in which the shell is divided into a number of chambers by means of perforated diaphragms and it is arranged
that the mean diameter ofthe balls in the various chambers
shall decrease towards the discharge end of the mill; such
an arrangement being shown in Fig. 1.6. The reason for
this distribution of ball size is that, for optimum grinding
conditions, the ratio of ball diameter to particle diameter
should be approximately constant. In consequence smaller
balls should be used for the later stages of the grinding
process, where the powder is finer, and by the adoption of
a number of chambers in each ofwhich the mean ball diameter is suitably chosen an approximation is made towards
the desired constancy in the ratio of the ball size to thc
particle size.
The problem of the optimum distribution of ball siz
within a mill will be dealt with more fully in a latcr
chapter, but at this point it is relevant to mention a mi\1
which the segregation ofthe balls is brought about by an
ingenious method; especially as the mill carries a distin tive name, even though no principles which place it out
side the classification given previously are involved.
The Hardinge mill, Fig. 1. 7, uses spheres as a grinding
' agent l;>ut the body is of cylindroconical form and usually
has a length to diame"ter ratio intermediate between thos
associated with the ball mill and the tube mill. The reason
for this form of construction is that it is found that, durin
the operation of the mill, the largest balls accumulate a
the large end of the cone and the smallest balls at the small
end; there b~ing a continuous gradation of size along th
cone. If then the raw material is fed in at the large end
<U
.
::
<U
....
"'
..
<U
<U
l1
<
28
INTRODUCTION
BALL, TUBE AND ROD MILLS
the mill and the ground product. removed at the sm~ll~r
end, the powder in its progresson t~rough the mll S
ound by progressively small balls and conseque~ce thc
~heoretical ideal of a constant ratio between ball sz~ and
article sizc during grinding is, to some exte~t, attaed.
The type of ball mill illustrated in Fig. 1.3, nc.oorates
a pepheral discharge through fine screens lng thc
cylindrical part of the mill. Heavy perforated plates pro
29
l1 11 l xcessive. At this point it will perhaps be useful
1\l S~ the factors upon which the choice between a
. ll , . tl c or a rod mill depends.
\' I ' mill is used as a batch mill, the capacity of the
ll l :rly limited to thc quantity which can be handled
111 tt y; furthermore the mill is, as far as useful work is
. tt ' , idle during the time required for loading and
t I 1,11 1i " ' the machine: the load factor thus being ad1I I I . r~ ted. Clearly then, there will be a considerable
111 t loughput, a saving in handling costs and im 1 1l l o <d factor, if the mill operation is made continuous
1 l111l l f.\' l1 e mateial into the mill through one trunnion
"I I \ I Ia wing i t ei ther through the other trunnion or
I l1 Iischarge ports a t the exi t end of the mill body.
', l1 wever, the flow of powder through the mill is
111 ttl inuous, it is necessary that the mill body is of such
I I lt tia t the powder is in the mill for a time sufficiently
111 It ' 111 grinding to be carried to the required degree
'SS . This, in general, demands a mill body of c
l i l I ngth, or continuous circulation with a classifier,
tl 11 iN ir r eased length which gives rise to the "tube
lll '
The Hardinge Mill
l 1 l tt metallurgical industries very large tonnages have
l I . \l ed and, furthermore, an excess offine material
is i ;ble since it often complicates subsequent treat" j ' csses. such applications a singlestage tube
1 l l tt uitwitha productclassifier, bymeansofwhich the
l . l which has reached optimum fineness is removed
1 j) rt to the subsequent processing and the oversize
tect the screens from injury and act as a linin~ for tl1'
tumbling charge; sometimes also the fine _screen ~s f~rth~r
rotect~d by coarse screens mounted drectly sde .t .
this type ofmill, which is often kno\vn as the Kruppm~\i
is of interest since it rep~esents a very ear~y type of r:
which with modifications, has retaned ts p~pu~any .
The Krupp mill is particularly suited to the gnnd~g <~I
soft materials since the rate ofwear ofthe perforated lnc
1111 11 d to the mill for further grinding, is an bvious
li.
nce continuous feed and a long mill body have
I l ' p ted, however, the overall grinding efficiency of
!11 1nay b e improved by fairly simple modifications.
: s a lrcady b een mentioned; for optimum grinding
i S tl1ere is a fairly d efinite ratio of ball size to
I I sizc and so the most effi cient grinding process ca
I l 1 . aine d when a product with a large size range is
I I ir the mill. If, however, a tube mill is divided into
(Harding< Co . lnc. )
Fra. 1.7
30
INTROD UCTION
BALL, TUBE AND ROD MILLS
a number of compart ments. and the mean b~ll size of thc
grinding media decrease s ~ each succeed~g compart ment; then the optimum ra~10 ~etween ball sze and partic1e size is more nearly mantaned, and a better overall
performa nce of the mill is achieved ; this giving ~ise to thc
compart ment mill shown in Fig. 1.6. The tube mll has thc
further advantag e that, to some extent, the grinding characteristic s of the mill are under control; for examp1e, an
increase in the size of the balls in the final chamber will
reduce the rate of grinding of the finer fracti.ons but ~ill
leave the rate of grinding of the coarser fracton~ se~sbly
unchang ed and so the amoun~ of coarse mate~al th~
final product will be reduced wthout any excessve overall
increase in fineness.
The principa l field of applicati on of the rod mi~l is probably as an intermed iate stage betwe~n t~e crus~ng plan
and the ball mills, in the metallurgcal ndustnes. Thus,
material between about 1in. and 2in. size may be rcduced to about  6 mesh for feeding to the ball mills. Rod
mills are, however, b eing used in closed circuit with a
classifier to produce a product of 1ess than about 48mesl
. .
size, but such applicati ons are unusua~. .
The choice between wet and dry mllng s, general,
unimpor tant in smallsca le mil~i~g b~t is a major techn~cal
problem when largescale mllg t~e metallurg~c~l
industrie s is involved . On purely mechacal grounds t
difficult to see any great diffe.r~nce i~ fundame.n~a1 principles b etween wet and dry mllng, ~nce ~ry mll:g m.ay
be regarded as wet milling wit? a f:ld ~a:rng the v~co~ty
and d ensity of air, whereas wth w~t. mllng the flud s a
liquid. slight difference of con<;i~on betwe~n t~e two
cases however is that with dry mllng the flud (ar) fill
the ~hole of ~he mill shell not occupied by the balls,
whereas with wet milling the liquid occupies only a part of
the mill volum e not occupied by the solid charge. hi
differenc e would be' expected to have some slight effect 011
the motion of the ball charge but the effect would b
expected to be so small as to be practical ly n egligible.
31
lll)~atio n Of the literatur e ShOWS, hOWCVer, that
"11 lfference between the results of wet and dr
l 11 . orten
' for example th e power to
.
. .very Iarge
II l I IH mll whch a given material is being wet
, ~y b. as much as 30% less than that of a similar
11 l y ~: nd~ng_. Furtheri? ore the ~hroughput of a c
(I !ls greate~ wet gnnding ; probably due
,,, / : now propertes ofa pulp in compari son with
I. , a.. It appea~s, however , that the rate of ball
11 I I wnr s greater wet milling so the decision be1 ~ d dry operatio n is, in cases where the choice
"' , a matter of balancin g the economi c gain
f r~duced power demand and increase d
I L agast the increase d cost of replacem ent f
I I 11 : I trg and the mill liners and the Ioss of prod u~ l , the shut down period n ecessar for such re
1'' ~ be m ade.
is not erchoice
this
I I r ertain applicati ons,
a
1! , r example ' the grind'ng ofth e raw m
the
1111 111 :.r may be carried out either wet or dry but
~\ l t~1e cem entcan, from thenatur e ofthe material
'
l l .r r 'dout dry.
I,I _i mportant point of differenc e between wet and
t l .~ s_that we~ milling allows the easy introduc tion
I 11 f ll I' ; llVC med~a for the reductio n of the energy ref 11 : J)' duce ut n ew surface or for the inhibitio n of
1 f the fine particles . t is theoretic ally possible
l such substanc es in gaseous or vapo ur r:rm
I t 1 I b ut,
pr~cti~e, su~h methods have not been
r:act
adoption , and ,
tojustfy wde
l
successfu
iy
1
ll
.
.. . bl
I 1 "" I ,:
tca e method of ntroducing such material S
11 I
I II g.
"" lation. can now well be given to some of the
' 11?1'1 wh~h the choice of the type of mill, or the
f
d'
l1l l eraton
. of a given mill , r:0 r th e gnn ng a
.
I
peris
it
l nnteral wll depend. In the first place
1' 1 f t r.est to note that the problem s encounte red
1 11
I 1 ;,t of powdere d materials are not necessar~
32
INTRODUCTION
BALL, TUBE AND ROD MILLS
reduced when the quantity of material to be processed is
small. As an example of this, in t?e preparatio~ '
materials such as Portland Cement, whiCh scll at rclatvcly
low prices, the problem is ma~nly one of o~t~ining a fi.n~l
product having the required s~ze c?aractenstcs .at a mn
mum total cost per ton and, th~s case, guestns of. th
the contamination of the matenal, dunng processng,
scarcely arise.
.
.
In the grinding of drugs and cosmet~ p~eparatns,
however, the question of the cost <?f t?e gnn?ng pr.ocess,
per pound of product, is scar.cely sgficant. n relatn to
the high value of these matenals but the a_vda~ce o;.contamination is a major problem. Thus, the nclusn oL ron,
from the baHs and the mill body, in "face powd~r" i~
highly undesirable since such iron imparts ~ brown !~ to
the powder, and for this re~son c?smet~ and sml~:
materials are usually ground a mll havng a c~rar~:n.
body and with smooth natura~ pe~bl~s as the gnnd?g
medium; that is, in a pebble mll. S~larly, pebble m~ll~
are used for grinding of feldspar for hghgr.ade p~rc~lan,
since, as the previous case, minute t:aces of r~n wl.ln~tro.
duce objectionable tinting. In ths connecton, t. lS
interest to note that there appears to be extant ~ lcen t'
for the removal of five tons of pebbles from Chesl Bea !1
per annum; these pebbl~s, which ar~ naturally graded for
size being ideal for use pebble mlls.
As an extreme example ofthe importance ofth~ el~mina.
tion of contamination may be quoted the gnndng
minute quantities of me~allic halides, used in. certai
classes ofinfrared absorptn spectroscopy, for whch pur
pose a small vibration mill with body and balls :n:~de '
agate is often used. In cases such as these, the dec~n H!l
to wet or dry grinding is rarely based on. economc considerations, but is usually made on the bass of a study ~
number of extraneous factors, such as some spe.CJ:l
property of the material being processe~,. s.ome desrc<l
condition of the final product or of the tal and fin:l
states of the materials involved.
33
ll . y rotech.nic materia~s
might well be ground dry,
wo uld, general, gve a product having superior
lt ~ i cs to one which has been ground wet, but if
l11 , ! al. were very reactiv.e ~t would probably be
1 l~ an nert atmosphere. Sinlarly, tungsten carbide
1 11 the prepar~tion ~ sintered carbide tooltips
ii i ' gound dry, snce mosture would complicate the
l n t proces~es and be harmful to the final product.
ll11 1 wder pgments would probably be ground dry
I y hi~ ~eans discoloration is minimized. The grind~
I l I :um for the preparation of paint is, however,
11 , cd out wet since by this means stearic acid or
i ' antiflocculent, may be introduced.
'
I lly, claysforthe preparationofceramic slips would
. l > ly b e expected to be ground wet, since the initial
11 11 ll of the clay is wet and it is required in a wet form
l H qu e~t processi~g. Clearly,. in such a case, dry
1 l ,(, whch would nvolve dryng, would be uec
, nd would only be adopted in very special circum 111 . appl.ic~tion of the. ball mil1 which can scarcely
l1 , ' d as mllng but whch should, neverheless be
11111 d is the dispersion of pigment in the var~ish
l l1 in connection with the manufacture of paint. The
t :n tion of solid I?aterial in the suspension is, in this
l1 . t , usually qute low, gener~lly being about 10%
l1 c, and, although appreable reduction of the
l l particles occurs, considerable ball wear is often
t cd . These observations are of some interest in
l ' ~ theory of the internal dynamics of the ball mill,
t lcd by one ofthe present writers, since that theory
"'' 1 ~ hat when the quantity of solid material is small in
I: the volume ofthe ball charge, the particle size
l might be small but that the ball wear would be
l
/ : b le .
I 1111 t le foregoing OUtline of the problems inoled
l1 1l : ~ ~f ~ mill for a given purpose and in the pera
I , m lls, t s apparent that the subject is very coml
r thermore,
much of the theory underlying a rigid
34
BALL, TUBE AND ROD MILLS
solution of any problem is in a very fluid state and, i
conseque ncc, th e solution of a practica l problem generll
involves tl1 e use of considcr able judgeme nt, based or
previous experien ce, to temper thc conclusions
from theoretic al analysis.
CHAPTER 2
' Jl,
OF CHAR GE
WITH IN MILL
I of the charge, that is the grinding media and
l . < al undergo ing grinding , within a mill is ofconl , l l ( heoretical interest and practica l importan ce,
l , l11 tl se r easons, has b een the subj ect ofconsid erable
l , l y
number ofworke rs, but, even so, no rigid and
'l l . heory, covering all the aspects ofthe dynamic s
I l ni ll ch arge, has yet been produce d. The practica l
I . of this subject clearly resides in the possibili ty
. 1 diction of the grinding behaviou r, and other
11 l l : racteristics, of a mill from the knowled ge of the
o ics of the elem ents of the mill charge. The
t i a L interest lies in the study of the dynamic s of
I
m and in the d erivation of equ ations to define the
!
th e elements of the mill charge in terms of
ti, :ntal qu antities such as the size and the speed of
\ \ of the mill. simple example of the practica l
111 11111 :1 e of this informat ion is in the use of the know1 l tle trajector ies followed by the balls in a mill to
l 11c the speed at which the mill must run in order
I t t I clescendi ng balls shall fall the toe of the charge,
1111 I upon the millliner. The impact of the balls upon
tl l' t' plates can lead to u'nduly r apid wear of the
I \ , : nd so to high mainten ance costs.
tl1is chapter tle motion of the particles constitut ing
l l ;11ge ofthe mill will b e consider ed; it being assumed
ll l is no slip between the mill shell and the charge.
l ists another type of motion, in which the charge
wl1 le slips relative to the shell, which gives rise to the
35
36
BALL, TUBE AND ROD MILLS
phenomenon of "surging", bu~ consideration of this typ<:
of motion will be deferred untl a later chapter.
study, from first principles, of the b~h~viou.r of a mi ll
charge is much simplified if the ch<7rge s m<7gned to b
composed of rods, instead of balls, snce by t~s ~ea.ns th
complication of any axial motion of th~ balls s ~lmn.at <I
and the problem is reduced to . one two d~mensonH.
Consider first the motion of a sngle rod, of dameter t/,
within a smooth shell of internal diameter D; when tl1t'
OF CHARGE WITHIN MILL
l tto ti .
ns of th_e surfaces of the rods at the point of conand free motion inside the shell
l a t e?. In ths case the angle between the vertical
i :adus . vector joining the centre ofthe mill and the
of gr_avty of the pair of rods is much greater than
I I /
sngle rod. But, again, equality exists between
l ~.k . done to rotate. the she~l and that dissipated in
't < t the contact ponts and distortion ofthe metal
/_ '. . further increase in the number of rods would
i hs effect until relative rotation between the rods
1
~ opposte ~enses
~
S t l
11.I

St
Bolls Only.
Bolls W thA TrocOf Quo,.tz .
I
10
20 %>
Fc.
FG.
2.1
FG .
37
30
40
50
2.3
2.2
shell rotates about a horizontal axis with an angul:.
velocity radians per second. In. such ~ ca~e the rod w~\1
lie near the lowest point of the mll, as Fg. 2.1 and w\1
rotate at such speed that the peripheral speed of th.e rod i
the same as that of the shell. Furthermore, the dspla<
ment ofthe radius vector joining the centre ofthe mll al
that of the rod would be such that the work done by reaso
of the couple formed by this displacement is equal to tl
energy dissipated in the distortion of the rod and shell
the line of contact.
Ifnow two rods are placed within the mill, as in Fig. 2.'
I , y precluded and, in this respect, the charge be ln ost as a solid body.
l o ectness ofthis view is supported by the work of
' / l present authors who, with a coworker Rose
l , . s, has shown that, for all other factors r~main
' : , the power to drive a mill increases with the
1 of tl1 . rato (D jd); that is for increasing numbers of
l l c l s ( the. exp~riments balls were used). This rela1 11 1 , sl1own 111 F1g. 2.3 indicates, at once that the
l l ' lli' 11t
.
h of the centre of gravity of th e c'h arge 111 w ~ 1ncreasing number of balls, which is in
l n wth the reasoning given above. T his figure
38
D RO D MI LL S
BA LL , TU BE AN
of the
tio n in the dis pla cem ent fricalso shows tha t the va ria
of
t
ien
ffic
coe
less wh en the
cen tre of gra vit y is mu ch hig h, wh ich ind ica tes tha t, for
is
es
fac
sur
wh en
tio n be tw een the
g of the ch arg e is effective
this con dit ion , the loc kin olv ed.
bu t a few ele me nts are inv of the mi ll is so low tha t th
If the spe ed of rot ati on
ted , the n
ele rat ion ma y be neg lec
e will
effects of cen tri pe tal acc
ch
tre of gra vit y of the arg
the dis pla cem ent of the cen possible lim itin g con dit ion s is
o
inc rea se un til eit he r of tw ing :
be
s
ion
dit
rea ch ed ; these con
to ma inforce at the shell, necessary of th
() Th e tan ge nti al
tre
cen
ss
ma
the
tai n the dis pla cem ent of
the fri cti on al force. Wh cn
cha rge , becomes eq ua l to
wh ole ch arg e slips bac k.
this sta te is rea ch ed the
wh en ,
cem ent , , rea che s 30
(2) Th e ang le of dis pla
n fall
the
s in eac h lay er
since the centres of the rodof rod s collapses.
on ver tic al lines, the he ap
eta l
e, since the effects of cen trip nts
In fac t this is a triv ial cas
me
ele
affect the mo tio n of the
acc ele rat ion pro fou nd ly
tio n /'
speeds of rot ati on the mo
w
giv cn
t
of the cha rge , bu t for slo
tha
to
s
ate
xim
mill ap pro
will
the ch arg e in a pra cti cal
s)
s mo tio n the balls (or rod mill,
by case (2) abo ve. In thi
he
oft
ll
cen tric wi th the she
tra ve l on cir cu lar arcs, con is rea ch ed , aft er wh ich th y
y
ilit
tab
th
un til the po int of ins
is inc lin ed at ab ou t 30 to
ich
wh
e,
fac
n
tio
roll do wn the sur
mo
pa ral lel layers. Th is
nt
ho riz on tal , in a series of
en thr ou gh the tra nsp are
tak
sho wn on a ph oto gra ph . 2.4. It will be no tic ed tha t
en d of a mo de l mi ll in Fig s the mi dd le of the ch argr,
ard
sm all "v ort ex " exists tow the balls no lon ger roll down
on
ati
rot
of
eds
spe
r
pro
At hig he
bu t, at a cer tai n po int , are
the sur fac e of the cha rge rea fte r describe approximatelg
the
th s
jec ted int o spa ce an d
ain me eti ng the ba ll mass;
ag
ore
bef
ths
to b
pa rab oli c pa
ar
pe
ap
in Fig. 2.5 . Th ere
ba ll pa ths be ing as sho wn me s for these two types
na
no un ive rsa lly ad op ted the evi den ce ap pe ars to be
t
bu
,
rge
cha
ing '
mo tio n of the
the first typ e an d "ca tar act
for
g"
din
sca
fav ou r of "ca
AR GE
OF CH
WI TH IN MI LL
(Dr.
y [ London)
J. . Connor and the Uniit
Fc.
(Dr. J.
of London)
Connor and le University
F.
2.5
39
2.4
~ I ' '
OF
CHARGE WITHIN MILL
41
BALL, TUBE AND ROD MILLS
40
for the second type. These names will be adopted for thc
present work.
As the speed of rotation of the mill is increased thc
partic1es are projected with progressively greater ve1ocitics
until the theoretical trajectory for a particle, which is in
fact 1ying against the mill shell, wou1d fall wholly outsid
the shell. Clearly, since the partic1e cannot pass through
the shell, it wou1d 1ie against the shell throughout the cycl
and so be carried around continuously with the mill shell.
This condition is known as centrifuging and the speed ('
rotation at which it occurs, for the outermost 1ayer ('
particles, is known as the "critica1 speed" of the mi\1.
This speed, the critica1 speed, is of considerab1e importan
in mill techno1ogy since, for example, other factors beinf.(
equa1, the equal performance of two mills of different siz
demands that the ratio of the actua1 running speed to th
critica1 speed should be the same for the two mills. An
expression for the critica1 speed of a mill will now b
derived.
By reference to Fig. 2.6, it is easi1y seen that if th
trajectory is not to fall inside the shell, the radius of curv<
ture of the path, , must be greater than R, that is tha
that of the mill she11. Furthermore, this must be true f
from wherever the trajectory might start.
From the dynamics of a partic1e it follows that for tl11
trajectory,
Fc.
l ' m the conditions
'fi d
I
('>. 1) into (2.2).
speci e ~ R, so, substituting
R
V3
gV cos
vz
gcos
' =
cos
" = }
y' = V sin  gt
y" = g
A1so, from e1ementary mathematics, the
curvature, , of any curve is given by
{1 + (dy fdx)2}3/2
=
d 2yfdx2
which by substitution (for examp1e, see Todhunter's
jerential Calculus, p. 174) gives
(y'2 + '2) 3/2
=
x'y" y'x"
2.6
cos
V2
~ 
gR
.... (2.3)
l ::> 10 so the limitin cond' .
.
.
'' iR to
g
It!On IS = wh1ch cor
DU
V2jgR
1 i (~
V=
which
'
= 1
= gfR
= y(gfR)
.... (2.4)
~~

~

OF CHARGE WITHIN MILL
42
BALL, TUBE AND ROD MILLS
When the diameter of the ball is taken into considera
tion, the radius of the outermost ball path is giyen by
!(Dd) which, when substituted for R in equation (2.4) ,
gives
and this expansion by the binomial theorem leads to
= y(2g/D ) . (1 +05d/D)
.. . . (2.r.)
If, as is usually the case in practice, the value of d/D iH
less than about 1/2 0, the second bracketed term may, i11
general, be taken as unity.
In this treatment is the angular velocity of precessio
of the mill charge and, because of slip between the mi\1
shell and the charge, this is not necessari1y equal to , thc
angular ve1ocity of the mill shell. If, as is often the cas~,
this slip is assumed to be negligible, then = and thcs
equations may be cast into the slightly more convenicn
practical form
, = (60/2)y(2g/D)
= 766y(l/D)
....
(2.)
where D is the mill diameter feet and , is the criti :\
speed of the mill in r.p.m.
The above formula is based on the assumption thl
there is no slip between the ball charge and mill shell anl,
to allow a margin of error, it has been common practi:
to increase the coefficient in the equation by as much :
20%.
That relative slip between the charge and shell is of n
portance in this connection is supported by the work
Rose and Evans<O who report that, using ground s I
balls in a smallsca1e model mill made ofsoliddrawn bra
tube, centfuging did not occur even at running speeds
120 % of the critica1 value. However, when the same mill
was fitted with effective lifters centrifuging occurred wh~
the actual speed exceeded the theoretical value by a fcw
per cent. This observation is confirmed by Grinder, (3) wl1
~
~
43
II ~ed that centrifuging in a mill with porcelain balls and
c , oc~~rred when the actua1 speed exceeded the ca1cul , d h
cntca1 va1ue by about 5
to 10
It s
. qu eston.
' , > .' owever, wh~~her with modern liners maintained
e a~onab1e condtn
= y{2g/(Dd)}
the increasing of the va1
f h
' cfficent of equation (2 6) by 20 .
ue t e
lcNirab1e.
s necessary or
I ~ttentio_n will now be turned to a more detailed analysis
' e motn. of t~e. ball charge. the first p1ace, the
; _~~ ~
1 ,~ ~
w
1 . e consdered, even though, as shown 1ater
s~mplc~y, the ana1ysis first developed by
_certan r~spects, an oversimplification.
'
. of
1 a pont ' Fig 2 7' at wh'ch th e proJ. ectn
: ::: partc e occurs a~d let this. point be at a distance r from
1 . e~tre of the mll. In ths connection the "
. t f
I'' c~ cton" may be considered to be the point at whf~~ t~e
I ' 1 ; Itory of the partic1e is influenced by gravity and is not
controlled by the packing of the
, re.
surroun d'ng
.onsder
11
I' .r _ proJecton
. .
to occur, it is necessary that the radius of
; vLtu[e
of the path under the influence of gravity should
Mma er than the ra?ius of curvature of the initial
'I lar{ath. An expressn for the radius of crvature of
: ' \fr~mu~~f~h gravity is given by equations (2 .2) and
cc
r ~
I ,
' n: th~ geometry,
: ; :cL~

'i
2rf g
r,
 _,
(2. 7)
however, it is c1ear that the line of
mduhst be at right angles to the radius vector
an
ence
r
Ic'
for the limiting case and since V =
cos
V2f (g cos
(gf 2 )
cos
.... (2.8)
quation is satisfied when point is distance gf 2
centre of the mill, a.nd furthermore, since t~e
he
' C
44
BALL, TUBE
AND ROD MILLS
OF CHARGE WITHIN MILL
l11 , is the radius of the Davis circle and n is in r.p.m.
i11, in this treatment, slip is usually neglected and so
11 1" r ts the speed of the mill in r. p.m.
l ::jectori es ofthe ball paths may, on the assump l r he ball paths are parabolic, be easily plotted.
l , l y he integration ofequations (2.1) and taking the
o fhe coordinate system at the point of proj ection.
= Vt cos
}
 gt 2
(2. 10)
= t sn
2
FG.
FIG. 2.7
.
nd from the theorem of gc<~
angle is a nght a!l~~e a circle is a1so a right anglc, !
metry that the a~gle wt ~ h projection takes plac
follows that the lne fro~ w /~c 2. the "Davis circle".
arc of a circle of radus g '
.
an
. .
his leads to 1he expresson
By a simple substtutn t
(2.< )
rc = 0408/n2
45
11111 l i ninating
2.8
t between these equations
gx2
y = xtan V
.... (2.11 )
2 2 cos 2
'''''" wl1ich the parabolic trajectories may be plotted.
important than a know1edge of the form of the
1 j 1 or i es is a know1edge of the form of the surface in
/ Ll1e parabo1ic traj ectories terminate.
1 Fig. 2.8, it is clear, from requirements of
eteri~g ay ~,
MI LL S
TU BE AN D RO D
out
BA LL
of materia1 quatity 1eavtg
'
46
qua1 to the
b
am
the
t
tha
t oss a
y,
cotiuit
stbo1eice tra jec tor ies ca cr '
mu
q'q
t
me
u1are1e ce the ara the sam e anu1ar e1eme~t as
it ad so, si
ied ? pts:
~face bR c is defers
iO ~f thc
tra jec tor y mu st termta~:
b . d by the it ect r IS ~h
it origiates. ~hus, the
ere
wh
r;
ius
tanl~c1e ofr ad
suc h as R, wh tch are . a c ori iates. represe~atit:.
h
WI~
y
tor
arabo1ic tra jec
the traJeCtbory. dgby the
. . as.
. tiO
padius at wh ich
R
r oit, suc h as ' m ay be . the oit of proJec o1 a wlth
of the pa rab
.Y't
ues, of
h
va1gi
fhe ori
p
the
of
the
s
ge re::.erred to these coor.di atc
ou1
th e circ1e of rad ius r th~
c
Cir
the
f
.
o
Th e equati
is
.  ... . (2.12)
2ry Sl
2 +y 2 + 2rx si 
determiatiO
taie
ad tafkil:t~rsectfo
by (2 .11 ). On
1a is give
ara bo12
ressn
.
exp
the
),
(2.
Pi
h)
oft
iOm (2 1
uat
.eq
the
d
a
fro
1g
tut
sti
sub
g2x4
4V2 cos4
is
obtaied;
~ + (sec2
 {; sin
V2 cos
h
fro m w h IC
g2
( 4V2 cos2
No w fro m
2 
 V2 ~~s )2
)xgrcos + V2
) x2(gta
equatio
(2. 7),
ing
. the immediate1y pre ced
wh ich sub sti. tut e d
1eads to
2
 tan =
;3
4r cos
ad
an d a1so
V2 = gr cos
Th e
so1utios f
t h'IS equatio are
= 4r si
cos 2
equatio,
... . (2.13)
~ '
CH AR GE
in the pr ese
o wh ils t the fou rth ,
<rcat zer
'i'1 ),ri.~ Ll1
req uir ed .
sti tut ed
l ' /1s value for is sub
t,
rl r f ' t/) la, it is foud tha
1
MI LL
47
cle i fou r
ral , a pa rab ola Cuts a cir qu art ic
lr; r Lhat, in genet
the
of
ts
roo
case, thr ee
OF
t . 1C/,
equatio
give
by
equatio for
the
2
. . .. (2.14)
y = 4rsi cos
t the
essary to plo
pu rpo ses it is t ecint
b
be
y
r r , t" most the
ma
R
po
reps:resetative
cto ry,
// r// r<~
.r me tri cal ly thu
the mill an d let
be the cetre ofcir
r ' ,. 2.8 let po intRP
cu lar pa th of
Q represet the
1 1rr: nt ric circle
2 , repregf
ius
rad
S is the arc , of
/ / /. Th e arc OP
t a ho rijec
Pro
.
urs
occ
projectio
tre Q
cen
' r rr ~ /1 1i e at wh ich
th
t the circle at Qad wi
Th e
rr rl 1/ l1rough to cuius
R.
at
cle
cir
t the
l / an arc ofaradpoitQPon, tothecusurface i which the
/ is the n
epe ati ng the
balls me et the ch arg e. R et rad ii, lea ds to the
fer
dif
mbe r of cirles, of
I ,, ru
I sur fac e bR C.
is sim ple :
r j)I"Oof of this costructionthe
, an d len gth QR is give by
I I ng th PQ = 2r sin
yR)2}1!2
QR = {(xQ xR )2 + (yQ
1/ r ~ v ',
xQ = 2rsi ;
YQ =
11
". = 4r si cos2 ;
YR =  4r si 2 cos
111
/ rr.r"
costruction
+ (4r sin2 cos )2}1!2
{(2r si + 4r sin cos2 )2
fic ati on leads to
n exp an sio n an d sim pli
QR = 2rs in
ati on shi p used
d thi s is pre cis ely the rel
an
QR
=
PQ
,
I
n.
tio
uc
str
on
l1
k fro m the cen tre of
1 '/: curve ab is a cir cu lar arc struc
/{
/ /1
48
tc
face wh ich does t oscula
the mil l ad represets a sur S, ad, i cosequece, fron
wit h the segmet of circle OP
pla ce. Th e rad ius of this arc
wh ich projectio takes
l{
oti
the mill cha rge ad, de
fixed by the ma gni tud e of
es the expressios
the rad ius by r;, Davis giv
... . (2.15)
r;/R =
app rox im ate ly; J
fillig of the mill.
thi s
beig, i
expressio,
the
OF
0 4)
= (36360
c
/ 11 ,
i / lim
' l y
par abo lic pat h is
e reqred to tr?erse the

... . (2.l(i
0 02 4+ 03 9y (7 10 J)
wh ere
49
MI LL
CH AR GE W
e [
l , ',, iH l /n ad thebytim ' c> or the ball to cover the
e
giv
is
11 I I I );) tl1
~ ~ '
RO D MIL LS
BA LL , TU BE AN D
1 '
2 si cos
fractioal
t the mill
tha
tha zer o, it follows
Sice caot be less
erece b~
erf
it
t
abo ut 7 wit hou
fillig caot exc eed
vis sta t
Da
lly
tua
Ac
takig pla ce.
twee the par ticl es
ur, r; should
erece sha ll t occ
tha t, i ord er tha t iterf
i revolu ed
28/n 2 , wh ere n is the spe
t be less tha 02
per second.
wh ich is i "fl igh t" ca br
Th e fractio of the cha rge
s.
cal cul ate d fairly sim ply thu
(2.14) ,
give, fro m equatios
is
2.8
.
Th e agle i Fig
by
2cosrcos)
=  n
( 1/r)(4rsi
(4 cos3
Fro m var iou s
si
ad
so,
cos
3
tites
trigoometrical ideti
r = JR 2;r ;2
= cos 3,
cos
bu t
= cos (18 0 3)
),
sice si =cos (90
cos
(90)
= cos
90 =
or
(1803)
1803
= 903
or
agle
4.
... . (2.17)
par tic le whilst i11
t~ughbyichtheis +
90 + 3 90 n
passed
Now, the
the par abo lic pat h is + 9
d to be 54 44'
/ I i agl~ wil l be fou
.
b
th
d
a
d
1e
ag
11 , ; f this
e a OVe expressios, l t lS fou
I I
wh
r.p .m. , the tim e per revolu
If the spe ed of the Inill is n
1, = 03/n
c = 03 9/n;
1
.
th
of
ti
spe
, / tim e
e Circu ar pat h ' as a fJract
is
1e
cyc
the
of
'
'
this leads to
(2.18)
.
ed tha t th
11 w t is ass um
oft he cha rge ma y
t
m
e
.
th
by
d
.. 11 cs nte
par tic le at the rad ius
mot~ofol of the
/ ;' of. the ch~ rge
essary to fi d t h e ag1e
, It IS ec
rad ius .
'[) dg to the
tios
si
V cos
= 1/n
'
039
0 56 =5 6 %
... . (2.19)
1 03 = '
f
b
tio is give by
/ , ( /
um er cyc1es p er revo1u
,; = 1/07 = 145
(2.20)
.
.
'I
t
firs
a
as
t
of
" ~, lt foll ows tha e .
1 approximat , 56 %
iss p etith' Circu ar p ath ad a so t h at the
d
mill sh 11
11 1 , I twee the
the cur ves dP a bR C is
ah
b
iede
u
occ
a
are
l ",, the
~ e cha rge wh e the mill
. s t. It also fol lo! s tha {' 1a. average, every ball
I ' ao the r 145 tim es p er rev o utiO of the Inill.
W MIL L
OF CHA RGE
MIL LS
BAL L, TUB E AND ROD
50
, is ope n to criticisn1
The analysis of Davis, just outl ined
the fric tion al cha re,
mpl
exa
on a num ber of grou nds . For
the effects of intc racteristics of the cha rge are igno red, cha rge in a givcn
the
actions betw een the elem ents of
een adja cen t traj e betw
e
enc
rfer
inte
the
and
traj ecto ry
etal forces acti ng on
trip
cen
the
tories are neg lect ed, as are
the equ atio ns involv d
the part icle s. Fur ther mor e, some of
wid e app roxi mat ions .
app ear to incl ude a num ber ofra ther atte mpt to deri ve
ous
Eve n so this wor k was the first seri
, whi lst form ing a good
rati ona l theo ry of the ball mill and
also pav ed the way for
basis for mill calc ulat ions , it had
late r studies.
the abo ve trea tme nt
The first imp orta nt mod ific atio n to
ted out that th
poin
who
5);
ger(
Stei
was sug gest ed by von
alon g the ascending
hyp othe sis of free flight of the part icle
lid since the con
inva
is
bra nch of the para boli c traj ecto ry
traj ecto ry result
any
g
alon
tinu ous pro ject ion of mat eria l
Thu s for the ascending
in con tact betw een the part icle s.
g the arc is con stan t ;
bra nch ofth e par abo la the spee d alon ht of particles.
flig
not vari able as is the case with free
tha t the intrinsi c
On refe rrin g to Fig. 2.9, it is seen al acce lera tion
radi
equ atio n to the nor mal is, since the
of the grav itati ona l
ent
pon
com
mal
nor
the
pro vide d by
acce lera tion , give n by
V2 f
whe re
poi nt .
= g cos
FIG. 2.9
lso from equ atio n (2.21),
V2
g cos
V 2 sin d
dy =
n
whi ch
2
V2
 dsfd = g cos
..
(2.21)
yy o = V
g
V2
or
from whi ch
V2
g
=  cos
ds =  dx
ds
ltt l sinc e dy= ds sin ,
ecto ry at tl1
is the radi us of curv atur e of the traj
Thu s
5}
cos
[log cos ]
cos
1og cos
. . . . (2.23)
equ atio n (2.22),
l'or sum mit = , and ,= , so from
V2
.,  = g
(2.24)
52
MI L L
O F CHAR GE WITH IN
BALL , TUBE AND ROD MILLS
and from equati on (2.23) ,
V2
y,  yo =  g log sec 0
.... (2.25)
The surfac e from which the projec tion of the partic lcH
occurs is unalte red by this treatm ent and is, as befor,
2
given by a circle of radius gf 2 drawn in the way pn
themselves are con
ories
traject
The
ned.
explai
y
viousl
sidera bly altere d, howev er, since for the same initial conditions, the horizo ntal veloci ty of a partic le, at the summ it ,
is greate r under the assum ption of von Steige r, and so thc
partic le travels a greater distan ce horizo ntally . Also tl1 r.
height of the summ it above the point of projec tion is dif~
ferent in the two cases. Furth ermore, since from contin uity
consid eration s, a trajec tory on termin ation must cut thr.
surfac e of the ball charge at the same radius as that a
which it comm ences, it follows that the equili brium surface of the ball charge is modif ied. These differences ar
clearly broug ht out by the diagra m of Fig. 2.10 in whi
traject ories startin g at the same point are plotte d
accord ance with the two treatm ents and the corres pondin H
surfaces of the charge are shown .
As alread y stated , the motio n of a mill charge is considera bly more compl icated :than the previo us treatm entH
would imply , and so far the variou s propo sed treatment
are not compl etely satisfa ctory.
One of the most compr ehensi ve analyses, and that up
which most of the remai nder of this chapte r is based,
that develo ped by Barth (6) and, indep enden tly ofBar th, by
Hinsle y and Fobele ts. (7) *
, was d evelop!'cl
* This work, which h as not previously been publishedwho
were appar
over a number of years by M essrs. Hinsley and Fobelets,
ently unawar e of the prior work of Barth .
their mater ir l
Messrs. H insley and Fobelets h ave very kindly m ade all on the bal
written,
was
apter
ch
this
of
uch
m
and
authors
availabl e to the
of the presc: rI
of their m a terial, before the work of Barth came to the notice
writers.
indebtedness
The present authors are pleased to acknow ledge their
tter.
a
m
this
in
Fobelets
and
M essrs. Hinsley
53
I he first place the equili brium surface of the mill
ve
. 'g~ is de ter~i_n ed and for this purpo se the effecti
r)
powde
plus
(balls
l 1ent of f~1ctn * of the c? arge
n
frictio
of
ent
i
coeffic
d
eral1ze
gen
t be establ1shed.
of
heap
l
conica
large
a
g
formin
by
ined
determ
y be
1
1
CuR  PARAOLIC RAJEC TORV.
 R JE C TORY O rVoN ScR
Bc c
RAJcov .
FIG. 2. 10
l ill ~harg e on a flat surfac e, as slown in Fig. 2.11; the
l ' ' be1ng formed by shovel ling the mater ial to the
the coefficie nt
. I 11. th is oveall coefficie nt there are compon ents such as
; a ''plough 
ll,: ! between th_e ba ll a nd the surface of the powd er particle
of the ba ll through
';' : ~~ representng the force req ~ired to move the tip
a factor which, for
'' ""' lace layers of the_ powder partcle q uestion, anddisplace
ment factor
'" I of a _better expresson, can be called a mecha nical
of th~
l rgntud e of wh ic~ is determi ned largely by the mode of packng
hich are rubbed
1'' 11 IJ l ., and th e coeffic_e n toffrict ion between the p articlesw e compon ents
'I te another dur~ngd splacement. T he m agnitud eofthes
l; for e
l ~ rmned by physca l cha racteris tics of the powder materia
of ploughi ng of
l , ~ the h ard ness, whch wll determi ne the degree
"''"' contact of the ball through the surface of a p article.
54
OF CHARGE WITHIN MILL
BALL, TUBE AND ROD MILLS
apex of the cone and allowing the stable angle of r~posc I
become established. The angle of repose, , s th
measured and the coefficient of friction is given by
= tan
55
<)
1
equating the forces acting on the particle the follow, xpression is obtained:
'/ 2
cos
(  )+ mg
sin
+ mr2
sin
( )
= mg cos
.... (2.2(1)
. . . (2.27)
Let the upper surface of the mill charge be given b">:' tl
curve  of Fig. 2.12 and take the centre of the mll :
the origin; al~o take the coordinate axes as shown. Let tl
F .
2.11
F.
mill shell rotate with an angular velocity and .also assun
that the particles of the charge describe circular a
around the origin with this velocity.
ow any particle, , of mass m, this surface, at
distance r from the origin, will be in equilibrium under tl
influence of three forces:
(1) The radial "centripetal" force: mr 2
l ttl i ply this equation by 1fm2 and rearrange.
( / ~ sin
+r
cos
"
cos
+r
gf 2
cos
sin sin )
r sin cos
I h en substitute = r cos ,
l y i which then gives
(gf2+
(2) The weight: mg
(3) Frictional forces acting betweent11epa:!_icle and i
neighbours.
2.12
I l \VCVCr, cot
cot
y = r sin
+y) = gf2
dyjdx,
cot
r cos
+y
(gf2
sin
and also divide
cot
SO
dyfdx(gf 2 +y )
+y)
56
OF CHAR GE WITHIN MILL
BALL, TUBE AND ROD MILLS
57
Now let z= (gf2 +y), which, on substitution, gives,
(zx)(dzfdx)
z
dz
z+
dx = z
This is the standard form of homogeneous equation '
first order and may be solved by putting z = kx, wherc
is a function of .
This, on integration, gives,
log
tan1 kt log (1 +k 2) +log C
z
Jx 2+z2
tan1 log
~
~:::25
5
z
x2+z2
tan 1 log  c
Now ifwe write: x2+z 2 =R 2, tan 1 (zfx) =, the exprcli
or
son
R =
Ce
(2.28)
is obtained; this expression being the equation of a
equiangular spiral.
.
. .
Also since R2 =2 + z2, t follows that the ongn of tl11
spiral (the pole) where R=O, . must ?e situat~d at = ,,
y =  gf2; that is on the vertcal axs at a dstance gf'r! ~
above the centre of the mill. (lt should be noted that thH
distance is also the diameter of the circle, the Davis circl<,
defining the surface at whic~ p~ojection occurs.) T?e co
efficient C is a parameter whch s related, but not smply ,
to the degree of filling of the rnill.
.
The configurations of the surface of t~e. charge w~
various values of speed coefficient of fncton and m 11
filling, as calculated on this basis, are shown in ~ig. 2.1 :
Examination of this figure shows that, for a gven ~ll
filling and coefficient offriction, the surf~ce. con~gurat
of the charge is not greatly altered by vanato.n~ ( Nf ,)
the ratio of the running speed to the cntcal speccl .
Similarly, for a given value of the ratio Nf Nc and a giv
c
F G.
2.13
r .ient offriction, the slope of the surface does not alter
. t ly with the rnill filling; furthermore, the line defining
1I rface is almost straight. ln fact the average slope of
I rface increases very slightly with d ecreasing filling.
c variable which has the greatest influence on the
1 iguration of the surface of the charge is clearly the
l llc.ient of friction and, all other variables remaining
58
BALL, TUBE AND ROD MILLS
unaltere d, the slope of the surface increases rapidly wit
increasin g coefficient of friction.
It is desirable that the meaning ofthis surface should l>c
borne clearly in mind. lf the derivatio n of the equation
studied it will be seen that the equation represents the su
face upon which a particle is in equilibri um under tiH
influence ofthe centripe tal force, based on the assumptio
that the particle is traversin g a circular path around tlc
centre of the mill with the same angular velocity as tiH
mill shell, ofthe weight ofthe particle and ofthe friction:l
force between the particle and the underlyi ng surfa .
Now, clearly, in a real mill the balls which come up fr 111
the body of the ball mass must either be projecte d or muHt
roll down to the toe of the charge and so, in cascadinH,
this mathem atical surface cannot represen t thefree surfa
of the ball charge. lt is probable , however, that a surfa
of approxim ately this form exists a few ball diametc H
below the free surface; that is, there exists an equilibriu
surface down which the cascadin g ball charge rolls. ThuH,
such a free surface cannot exist with cascadin g moti ,
although it is probable that e does, over a portion of tlc
charge, in the case of cataracti ng motion.
It is clear from the previous discussion that, ifprojectio
of the balls is to take place, the charge must cross the Davi
circle. Now, the equilibri um surface, as defined by tl
equiang ular spiral, can fall either below or above the DaviH
circle; as shown in Fig. 2.14. When the equilibri um su
face falls below the Davis circle, the particles, as , th(
emerge from the main mass of the ball charge, pile
until the Davis. circle is reached; as shown in Fig. 2. J!"> .
This is clearly possible since in the mass of balYs lyi H
above the main equilibri um surface but below the D avi
circle, there exists an infinite number of equilibri um su
faces such as ab.
The inner boundin g curve for this mass of balls is :
circle concentr ic with the mill shell and tangent to tlc
equilibri um surface. The position of this circular arc anl
the equilibri um surface is, of course, fixed by the conditio
OF CHARGE WITHIN MILL
59
I . t he "rolling" mass of balls and the "flying" mass are
~ c t l1er equal to. J, ~he static filling of the mill. Since, in
' ase, the prOJCCtn takes place from the Davis circle
l nstr_ucti?n fo~ the terminat ion of the trajectories:
1 :. e~rl1er, 1s val1d. ln the second case, in which the
projectio n
I I bnum surface _falls above the D avis circle,
reached.
is
surface
I ' not occur unt1l the equilibri um
I
F c .
2.14
Fc .
2. 15
11 1 jection does not now occur at the Davis circle
'
l n~vious construc tion must be modified.
11 asoning very sim~lar ~ that used previously, it can
I l own _ that the te.rminatiOn of a paraboli c trajector y,
nCing at a pnt ' above the Davis circle is ob1 c w~en, Fig. 2.16, distance P'Q.: = Q:R 'and the
/' 1 / C ' zs on the D avis circle. Thus the construc tion is
I : to the pre~ious except that point Q: is on the Davis
' 11 whereas pnt Q is on the mill circle. So for either



60
OF CHARGE WITHIN MILL
BALL, TUBE AND ROD MILLS
61
case the point of termination of the trajectories may IH
determined by simple graphical methods.
lt is interesting to have a knowledge of the proportio
of the charge which is in "flight", and of that whicl i
"rolling"; also to see how these quantities vary with tlc
speed of rotation of the mill and with the coefficient
friction of the charge material. simple construction (
this purpose will now be described; this construction bc.
due to Fobelets. ()
F.
F.
2.16
CASE 1. The point of projection on the Dav~ circle:
(a) As in Fig. 2.17, draw the mill shell circle and t\
Davis circle.
(b) With compasses determine position of point \{ ;
the termination of the trajectory originating at 1'.
(Use point Q on mill circle as centre.)
(c) From the pole C ( the Davis circle), draw radi
vectors at 10 intervals; taking vector CR as datu11 .
(d) From C set out along the successive rays a lengtl
Ca = CR. e(/l), where CR is the length '\{
( )
( ')
(~)
2.17
measured_from the figure, is the angle, in degrees,
fr<?m the lne CR to the ray under consideration and
~s the coe_fficient of friction of the charge.
Jon the ponts such as "a" to give a smooth curve
f~om the centre of the mill, , draw an arc of ~
c~rcle tangential to this curve and to cut the Davis
c1rcle at b.
Draw a line from thtough C to intersect a vertical
drawn thugl1 point R at point S.
R epea t (b ) and (f) using mill circles of various radii
and so establish the curve bS. (For most purpose~
62
OF CHARGE WITHIN MILL
BALL , TUBE AND ROD MILLS
this Step may be dispensed With and pointS I
joined by a straight line. There is then some los~
accuracy but this is not generally of great pract< l
importance.)
(h) Determine the area SbaRPCS. This when exprcsHccl
as a fraction of the area of the mill circle giv s ./ ,
the static mill filling.
(i) Determine the ratio of areas PCSbP and PbaR I' ;
this ratio being the ratio of the "flying" charg to
the "rolling" charge.
When the equiangular spiral 'enters the Davis circle tl
treatment is slightly more complicated and demands s
trial and error. Thus,
1
CAsE 2. The equiangular spiral enters the Davis circl
(a) As for Case 1.
(b) Assume a position for the point of projection, I'',
and using the construction of Fig. 2.16, determi
point R; the termination of a parabolic traject
which starts at '.
(c) and (d) As for Case 1.
.
.
(e) J oin points such as "a" and thc equangular sjnrl
should pass reasonably close to '. If it does not,. st: t
with a new point ' and repeat the constructon.
(f) to (i); As for Case 1.
The basis of this construction is as follows.
lf the flight trajectories are parabolic, then the ho i
zontal velocity ' is constant and given by
'
cos
... . (2 .2!1)
where is the angle of projection and the other symboiN
have the same meaning as before.
Also the time of flight, , is given by
= xfx'
= xfr
cos
. . . . (2.30)
63
l)uring this time, however, the rotation of the mill will
charge at radius r,
~ i 1 g, from the depth ofthe mass ofthe
column of balls of length r.
'1'11is is the quantity of balls projected in time Tp, from
Nlrip of width dr, and this may be written dq1 . Then
dq1 =
=
l 'm
r(xfr
xfcos
= PS
(2.32 )
from equations (2.31) and (2.32)
q1 =
PS dr
c the integration is carried out along
. This integral is, however, equal to
ll, r bPCS and hence the construction.
l1 to
(2.3 1)
Fig. 2.17, however, it is evident that
xfcos
'hen,
cos
the curve from
the area ofthe
lthough there is still an element of approximation in
111 iNtreatment, it appears to be superior to that of Davis in
l 1 the effects of frictional characteristics of the charge
included and the computation of the "flying" charge
l 'i not involve reference to some mean radius. the
i of this treatment it is possible to study the conditions
l
taracting or cascading in some detail and this will
I IIW be done.
.
I will be noticed that, so far, no rigid definition of cas' ~ l ing and cataracting has been given. ln fact such a
\ 1 l i ntion is probably impossible since it now appears that
l ' are not two types of motion of the charge involved,
I a single type of motion in which, in certain cases, some
1\1 i ls ofthe motion are not apparent. For the present pur ' it is probably sufficient to name the motion "cataract1 " when an appreciable ga:(> exists between the parabolic
11 j tory of the innermost particle and the upper surface
1l t \ rolling charge. Cascading then corresponds to the
1 t i in which this gap is so small as to give, in general,
l appearance of being nonexistent.
64
~~ O'I'JON OF CHARGE WITHIN MILL
BALL, TUBE AND ROD MILLS
1l ~; it follows that
I ll filling for which
This difference is c1ear1y shown in Fig. 2.18, which art
the trajectories for a mll runnng with a definite cha:~
at a definite speed but with different fillings . ln Fig. 2.18: ,
with the hgher filling, the space is very small a~d wh
allowance s made for the finite size ofballs the m1ll woulcl
appear to be casca~ing. In the sec~nd llu~tration therc
a definite space wh1ch, even allow1ng for 1nterference b tween the balls in the different trajectores, would not b<
obliterated, and hence the mill wou1d be described :
cataracting. For the present purpose it will be assum l
65
these va1ues are the maxi1num
cataracting will be apparent.
2.1
Maximum static mill filling for cataracting to occur
,.
Area of
circle
nill
( 2)
Area giving
"rolling"
charge
(in2)
Area giving
"flying"
charge
(in2 )
Jmax
(%)
Ratio Ratio
,
Mr
,
1111
.
75
11'1
111
075
() ()
11
IHi
075
O!J
(b )
F.
2.18
that when this gap is 1ess than 005 of the diameter of tiH
mill the charge will appear to be cascading; this gap bein
cho;en because, in view of the finite size of the ball in :
rea1 mill, it is improbab1e that a gap_ smaller than tl1 H
figure wou1d be discerned. By the p1ottlng of a number <!I
such figures the conditions for a g':'p of 1ess than the sp~c
fied size to occur are soon estab1Ished. The constructOJ
for the determination of the "flying" and "rolling" b;ll
charges is then app1ied to each ofthese 1i~iting ca_ses wh 11
the resu1ts given in Tab1e 2.1 are obta1ned. S1nce any
increase in the mill filling above these va1ues will reduce tl
3 14
314
314
314
050
081
128
015
050
09
205
410
698
023
038
042
10
077
062
058
314
314
314
046
090
126
015
05
09
190
450
691
025
036
042
075
064
058
3 14
314
3 14
038
082
115
015
050
09
171
415
650
028
038
042
072
062
058
I :; seen that a fourfo1d ariation in the coefficient of
11 i produces, for a given speed of rotation, but a
,li ib1e variation in the maximum mill filling which will
" cataracting. Thus, it is probably safe to say that
l sults, shown graphically in Fig. 2. 19, are app1icable
, y practica1 mill with an accuracy sufficient for any
1 ;l purpose. The genera1 va1idity of these resu1ts has
1 iv d an amount of indirect support in the following
l
t photographs ofthe motion ofthe ball charge, pub11 l by Rose and Evans, <9 1 have been questioned
I A'ounds that they show cataracting. study of
l orgina1 artic1e shows, however, that the three cases
.! = 025, J=05 and J=075, with N fNc=056 in
l ase. From Tab1e. 2.1, however, it is seen that each
1 l l se va1ues of J is above the maximum for which
66
BALL, TUBE AND ROD MILLS
'[Q OF CHARGE WITHIN MILL
cataracting will be apparent, and so the results shown
the photographs are to be expected.
The introduction of a coefficient of friction of constt
value, as in the extended theory, is not completely ad
quate, however, since a study of Fig. 2.13b suggests tl1 :
the average angle of slope of the surface of the cha!{ '
should decrease with increasing filling, whereas tiH
, , would bring about an even better agreement
l 1\ : tl1eory an~ practice. Unfortunately, the introduci / h a vanable quantity very much complicates
l . I matical analysis.
\ 1 , 1 1 and very important, use to which
'111 ns can be placed is to ensure that
10
81 <i> 
)(
OS
10
20
,
~ 06
z
Pol< .
1/
:::;
.J
t'
:
'
.J
v
v
CASCADING
Rc .
0 2
0.2

CAT ARACTNG
REGION
0 4
(ty'Nc)
FG.
2.19
measurements of Ros e and Evans(9) show that the sl
increases with mill filling. This difference is easily xplained since the charge consists of a limited number
balls of finite size and, in consequence, the energy di
sipaied in friction in the charge increases with the size
the charge. This has an effect equivalent to an incre;asc in
the coefficient of friction of the charge; and so brings abol
an increase in the slope of the surface of the charge, in tlr
same way as is shown in Fig. 2.13a. Thus, it appears th :t
the int.r oduction of a variable value for the coefficient '
67
FIG.
2.20
the foregoing
the condition
68
BALL, TUBE AND ROD MILLS
that the charge shall not slip the milllining is fulfill '(\ .
For this purpose,
(1) Use the previous constructi ons to de~ermine ~11 '
configura tion of the "rolling" charge the m\1 ;
shown shaded in Fig. 2.20a.
.
,
(2) Determin e the position of the centre of gravty, (,,
of this area.
.
2
3 Through G plot the radial acceler~ton r!:
( )
d' 11 and the gravitatio nal acceleraton g vc.rl!
ra lla ~Note. radial force = MrG2 and vert :\
~~rc~ = Mg, ~ these vectors are proportio nal to th
forces.)
(4 ) Obtain the resultant of thes~ vectors.
.
Draw a straight line connectng the centre of mll ~)
(S) to the point , where the resultant cuts the m\1
shell, and determine the angle. .
. .
, .
(6) Compare tan with the coefficent offncton . ~
safety must exceed tan by a reasonabl e marg.
The last condition is clear from Fig. 2.20b, ~ince th
resultant has radial and tangential compo.nents a~ showln .
The tangential componen t tends to gi":e nse to s~ a: t ~~:
surface and slip will, in fact, occur f the rato .t
tangential force ~. the radial force exceeds the coefficcn t
of friction: that s f tan > .
I
.
.
3.
4
5.
6.
7.
.
9.
REFERENC ES
VANS D. .
Proc. Instn mech. Engrs,Lond., 1956, 170,
RosE, . ., and E
'
773.
.
f 1
R
and S D. . Loc. ct., r e . .
.
GOSE, . w'' Lecture 'to the Fachausschuss fir Staubtechnk of tl
RUNDER,
V.D.I. Dec. 1955.
19 61 250
Trans.
Amer.
Jnst.
min.
(metall.)
Engrs,
19
,
,
D AVIS,
VoN STEIGER R. Zement , 1929, 18, I \06.
B ARTH w.
'ech. Mech. Thermodynam., 1930, , 321. . .
'
HINSLEY, J . F ., an d FoBELETS ' . Private Communcaton to tlr
authors.
FoBELETS, . Procds Henricot, 1955 . .
RosE, . ., and S, D. . Loc. ct., ref. ,
CHAPTER 3
WER REQUI RED
MILL
DRIVE
I 111 w required to drive a tumbling mill is of interest
I designer and to the mill operator: to the former
. i f design for the determina tion of the necessary
1
111 tl1
111 1 , ,1ll
l
l1
elements of the machine; and to the latter beother factors being equal, the most economic al
is that which demands a minimum power for
I l1 wer required to drive a mill depends, to some
1 every one of the physical dimension s defining
1!1 111 i 11 l1ell and,ball charge and many of those defin lc 1 roperties of the powder charge. Thus the number
l 11 i bl s involved is very large. Since even a moderatel y
\ l theory for the internal dynamics ofthe ball mill,
111 1 i all these variables are given due importanc e, has
1 I 11 propound ed, the calculatio n of the power requireII Il of a mill, from theoretica } COnsiderations, Cannot be
l , ,' imilarly, owing to the great number ofvariable s,
1111 tlete experimen tal investigat ion of the power det ti f ball, tube and rod mills has been made, the
t of work required for an analysis of all of the
l l s being prohibitiv e.
l11 ol~bly one of the most comprehe nsive experimen tal
t li :( tions of the power demands of the ball or tube
I iM that carried out by one ofthe present authors, with
1 1 1 w rker, Rose and Evans, by use of smallscal
e
i I ; the test apparatus being that shown in Fig. 3.1.
l\ 1 . [ smallscale models, a very large number of tests
. carried out in a reasonabl e time and, furthermo re,
I \)lcs WruCh WOU}d be difficu}t tO Change Oll a large69
70
scale test may easily be studied. valid major objection
to the use of smallscale models exists sice by use of th~
method of dimesioal aalysis the results may be geeral
ized to be applicable to mills of ay size. The valicl
geeralizatio of the results of the model test demads
that there should be a complete dyamical sirnilarity bctwee the systems, ad, i the case of the ball rnill, th
pressures i the bed ofmaterial i the mill caot be mad
ter of ball, d, the desity, , of the ball, the volume
piecby tle ch~rge (icludig voids), expressed as a
ti of the total mill volume, J, the speed of rotatio,
, l1 c acceleratio due to gravity, g, the coefficiet of
i uti of the material of the balls ad mill, e. The
~ r, , would also be expected to deped upo the
I l l owi g characteristics of the powder; the represetative
i , ~ter of the particles, b, the energy ecessary to bring
lt l uit increase 1 the specific surface of the powder,
, 1d the volume, V, occupied by the powder charge
Jig voids), expressed as a fractio of the volume
I w the balls i the mill. Furthermore, the power, ,
I be expected to deped upo the effective kiematic
i y of the mixture of powder ad fluid, , the effective
, ~ y of the mixture, , ad, i the case of wet milling,
I\ ratio of the volume of the solid material to the
l of the liquid i the pulp, U. Fially, whe the
l ior ofthe mill is fitted with lifters it would be expected
I l1 power, , would deped the umber oflifters, n,
I the height of the lifters, h.
l ' ltt s, the power ca be expressed symbolically:
i. n
111
I'
FG.
3.1
Apparatus for Tests on smallscale mills
strictly similiar i a large ad a small mill. e so the satis
factoy results obtained by use of model tests suggest that
errors arisig from this source are small. the pres 111
chapter the questio of the power to drive a mill will b
approached from the poit ofview ofthese riodel tests al
the results ofpeious workers studied by com~ariso wil
these data.
The power, , to drive the mill would be expected to
eter, D, th
deped upo the legth ofthe mill, L, the d/
71
POWER REQ,UIRED DRIVE MILL
BALL, TUBE AND ROD MILLS
/>(L,
D, d, , J, , g,j, e, b, , V, , , U, h, n)
(3.1)
' deotes some fuctio of each of the quatities
i the bracket.
' from the wellkow priciples of dimesioal
111 l yH is it is easy to derive the expressio i terms of
I ' ~ less groups,
( n' :)
{ (~), (~), (~), (ni2)' (i), (D3~2)'
(n~N ), (~),
(J), (j), (e), (V), (U) , (n)}
.... (3 .2 )
111 /> deotes some fuctio of each of the dimesio
ps upo
the R.H.S. of the
equatio.
72
I' ffl, PO WER REQ.UIRED DRIVE MILL
BALL, TUBE AND ROD MILLS
Theoretically, the dimensionless groups can be combined in any way and it is the function of experiment t
determine these relationshipso ln the present work, however, it has been found that, in most cases, the functions
of the dimensionless groups may be multiplied togethcr
and the results of the experimental analysis have been
arranged in this wayo
On this basis equation (32) can be written:
(n5~3)
(ntz) z(~) 3(J) 4(~) 5(n)
z(~ 3; 2 ) 0 3( U) 4(D:)
(33)
ln the work of these investigators, attention has becn
confined to dry grinding and, in this case, the groups
(fD 2 N) and ( U) are eliminatedo
The first variable to be studied was the effect of thc
length of the mill on the power requirements: that is, th
determination of 2 (L/D)o The experimental results of'
Rose and Evans fall on a series of straight lines passin~
through the origin and so it follows that 2 (LfD) is ;
linear functiono Thus,
(3.4)
The critical speed of a mill has already been discussecl
in Chapter 2 and it has been shown there, by equation
(205), that the critical speed is given by
(3.5)
, = y(2g/D)
which, by transposition, gives
gfD =
, 2
roup (g/DN2) is proportional to the square of
the actual speed ofrotabe written:
l .~ .i othe crit~cal sl?eed to
1 ''" I the relatnsh may
(38)
I 1 ~ ws that in order that dynamical similarity, as
l ~ group, shall hold between a number of mills of
l ll l li ameters, it is necessary that they shall all run
I ~ nc fraction of their respective critical speedso
I i . ~ cordance with the findings of Gow Campbell
I :11 11 <2> and other workerso
6(~) ,(f) os(e) 9(i) ) on(V)
(LjD) = K(LfD)
I\ l '
73
..
(36)
lf now is the actual speed of rotation, it follows th<
(30 7)
'
tionship between the P?Wer group ( PfD6N2)
I [ I : gr~up (,/ ), oaso determned by experiment, is
I ',',' 11 ! fg 3_.~, and t s seeon that for speeds less than
I tl 11
cntcal, the relatnship ' (,/ ) can be
,, 1 , ~~~ 1 by
'(,J') = K 2 (N,jN) 2
(39)
ll 11
11 1<
1\~ = 31 3 for lifter mills and 366 for mills without
, yYithin this r~nge these relationships can be used
I 11 I y the oequ atr:s for the determination of power
1 ' ' l dnve a ~ll. The deviation fr linearity
I I
'' "' ''
theose cuivesos exactly what would be predicted
since, ideally when the
I I ll charge is ~er:trifug~ng, the centie ~f gravity of
1111 1l1 11g would concde wth the axis of rotation and
I w r required to drive the mill would be zeroo
l11 I t' ry, the centrifuging of the charge takes place in
1 1 1 : d each layer form~ at a de~nite speedo prac , l w ver, the p:ocess s complcated by interaction
I I \ t ' ' II tl1e balls, lfters, etco, and centrifuging occurs
' less gra~u~lly over ~ wide range of speeds, this
I I ' the devatn from lnearity, shown in the figureo
111 (l se curves are plotted on logarithmic scales the
1 111 \'' H l1ou~d be asym~totic to the negative branch of the
1 , : nd fact the lnes show this trendo
, t: , ~s has already been mentioned, the critical speed
1 \ for the occurrence of many phenomena in a
I
I' retcal consderation
75
BALL, TUBE AND ROD MILLS
POWER REQ.U IRED DRIVE MILL
mill, it is perhaps of interest to mention that ~uring th<
course of tests at various speeds, the configuraton of th
balls within the shell was observed through a perspex enL
l ation exceeded the critical value " by a very small
ut . For a smooth mill, cataracting of the charge
1 Hally commenced at speeds of about 0 8" while,
1 l 1 lifters were fitted to the shell, cataracting cm
11 d at about 06Nc. This is roughly in accord with the
1lts of Chapter 2, illustrated in Fig. 2.19, that if a mill
\ \ a 30% to 40 % filling is taken as repres entative, the
74
s.ooo
Smo!,hJII.i
1,000
500
1./
I
I
/
/1~ Mill with lifters.
1/
10
oj_
5
10
I
F.
100
50
SMOOTH MILL Nc/ N
10
5
MILL WITH LIFERS Nc/N
10
FIG.
500
1,000
3.2
cap and in the case of the smooth mill, centrifuging
the ute; layer of the balls was not observed at speeds UJ
to 1 2" the highest speed reached during the tests, bt
when effective lifters were fitted to the shell, the ut
layers of b alls commenced to ((entrifuge when the spcrl
3.3
ll(lary between cataracting and
" '( I [ rotation is about 7 N c.
cascading is when the
\'Vitl1 a smooth mill surging of the entire charge was
imes observed for speeds below about 04Nc.
! 1 ing now to a study of the effect of the density of the
Iing media upon the power required to drive a mill it
l ll ws that since the mass of grinding bodies has a
l ltil ' configuration, as suggested by Fig. 3.3, then the
drive the mill will be r9J20rtional to the weight
l I harge; that is, without owder pro ortional to the
\1 i y of the material of' the al1s or rods. The correctness
~ ~ .~~ 
76
' ' 11 I~ P OWER REQ.UIRED D RIVE M I LL
BALL, TUBE AND ROD MILLS
of this view is established by Fig. 3.4a, based on the work
of Rose and Evans, ( 3) in which the measured power is
plotted against power calculated on the basis of this assu mption for charges of glass, steel and lead balls; th.
relative densities of these materials being given in Tab]:
10.000
5.000
~
~
~J(jl
"'
::
<(
::
1.)
1.000
::
u
I
::
500
I
/
4:ooth LD'=
"
:;'
"'"'
"'
mill;
steel balls;
100
100
500
,_/
10; D/d = 11 0; J = 05.
lead balls; + glass balls
1.000
CALCULA rED POWER  CM
5.000
TABLE 3.1
Materia1
7"
..
"""
10,000
GRAMMES PER SEC .
77
Specific gravity
Glass
265
Stee1
78
L ead
1 J.4
onfi guration as it has when balls only
111 l , Ll1 configuration is unaltered by the
are present,
inclusion of
l '' wl red material, then it follows that the torque to
! t l1 c mill will be proportional to the total weight of the
I ~, i 1 the mill. Thus it would be expected that, at least
~ approximation, the power to drive a mill con f', R powder can be calculated from the equation
'll l>l to a mill containing balls only, provided that
l l ity of the balls, in the latter case, is increased by an
t sufficient to include the effects of the additional
l to the powder charge. If this point of view is
l , it follows that the mill charged with powder
l s a special case of the mill containing only a ball
1 l11  , and effects arising from the particle size and other
1 I 11 : t ristics of the powder charge may be expressed by
''' i ns to the equations applicable to the mill without
I '"~ . On this basis, for a mill containing a powder
1
11
 :
(a)
I ' input
FIG. 3.4
rque due to ball space charge + Torque due to
I' wder) Speed Proportionality factor
3 .1. t is a t once seen tha t an extremely good linear r lationship holds between these variables and so it must b<
concluded that the suggested relationship is valid.
If now the material undergoing grinding occupies th
spaces between the balls of the ball charge, it follows that
the centre of gravity of the powder mass will coincide witl
that of the ball mass. Thus, if the mill charge retains th
(Wt. of ball charge + Wt. of powder charge)
' : these terms are simply additive is supported by
l s of Fig. 3.4b in which are plotted data given by
\ and Devaney,< 4 J for the power required to drive a
l , in which the rods are tubular and the effective
l i y of the rods is varied by "filling" the tubes with
BALL, TUBE AND ROD MILLS
' ' 11 ; POWE R REQ.UIRED DRIVE MILL
various mateals. For these tests the mill was grinding
chert, for which c::::. 30, and, since the void space between
the rods is about 15 % of the mill filling, it follows that, if
the simple addition is valid, the power is given by thc
expression of the form
y = mx+c
ll 11 tl1 functional relationship 10 ( /) is given by
l ltiplicative factor (1 +04/) .
I I ' , lidity of this relationship has been further e
11 ' I l>y tests carried out using powders consisting of
, i' powder and silica sand; the density of these
111 t: I ~ bcing given in Tab1e 3.2.
78
Furthermore, the power is given by
79
TABLE 3.2
= (15 30+085)
and it is seen that the curve given coincides very closely t
this expression.
5 r,,
Material
Iron
Emery
Silica sand
Density (gfcm3)
79
. 95
27
r:
I 111' sentative
duction to
I I) :bove, are
Q.
V>
r:
::
EFFECiVE
DENSIH OF RODS .
(b)
FIG. 3.4
If now is the ratio of ~,hL power tQ drive the mill
charged with balls and powder to the power to drive th<:
same mill witfi a a cfiarge only t hen:
Wt. of ball charge Wt. of powder charge
Wt of ball charge
=
lf also it is assumed that the porosity of the ball charg
and of the powder is 04 in each case, then:
(10 4)+04(104)
(104)
= (1
+04/)
1~ ,4 / 2,"1 / , ~ { 7 14
1 + 4 ,2 (;./~,& ~ i ) 13
( .... (3.10)
data from both smooth and lifter mills,
a consistent basis by the use of equation
shown in Fig. 3.4c.
~ ! nation of these data shows a scatter over a range
l l 5 % with no consistencies in the deviations. Thus,
' s that these differences arise from random errors
l () the functiona1 relationship 10 (/), given by
i (3.10), again appears to be satisfactory. It is
l 11 1ssible that some of the scatter arises from the actual
1 i y of the powder and ball charge differing from the
cl value of 40%, but such deviations are scarcely
l l :bl e . Thus it is concluded that the function 10 (f)
y , replaced by (1 +04/), with ah accuracy suffi r practical purposes. The remaining functional
l ionsl1ips cannot be represented by simple algebraic
't 'HS i OllS and SO are best presented by graphs.
first place it follows from Fig. 3.4a that, since
111 t i < ls having a wide range of values of coefficient of
11 l are used and yet the points suffer no significant
~t , the effect of the coefficient of restitution on the
l "'~ rcquirements ofthe mill is very small. Thus, at least
I 11 I 'l\ tical purposes, the function s (e) may be neglected.
80
' ' P OWER REQ.UIRED DRIVE MILL
BALL, TUBE AND ROD MILLS
This result would be expected since, although the coefficient of restitution of the metals would affect the conditions of rebound when clean solid particles make contact,
the effects of changes in the coefficient of restitution of th
metals are, in the case of a ball in a mill, masked by th
energy dissipating capacities of the loose bed of balls anr\
JSO r     .    , .      ,  
,,~,
"'>
81
' . l area of_the charge and the radius of the couple
1 , I h of whch depend the mill filling J. the
I 1 , l1 se assumptions it is a simple matter to calculate
l ' in terms of the mill filling and, when this is
l , " theoretical" line ofFig. 3.5 is obtained. This
1 I l
f treatment is a gross oversimplification how' , in c, as explained in Chapter 2, the curve 'repre' ' h surface <;>f ~he charge depe.nds the filling and
I ; l cl1aractenstcs of the mateal. Furthermore, the
I ll1 '"' not mathematical points and so there is mutual
~Joor~~~+~~++~~~~J4~
z
Smooth mill.
~2\0,L~~~~_L_ _J_J___j
1 :
.;?:
~J~',4~~~~~~
Th.oret ical Cure.
~rimental Curz.
8 Smooth Mills.
@  Lifter Mills.
Lifter mill; n
= 6.
200 r~~2S~~S0~~71~~,~00~,2LI_ _ _I\LO_L~7S
SPEEDR .P.M.
Quanz.
Emery.
Iron.
(c)
FIG.
~.4
powder upon which the ball falls and by powder betwec
t.h e contact surfaces.
If, on referring to Fig. 3.3, it is assumed that the curv<
representing the free surface ofthe charge is a straight lin<
and a~so that the angle is unaltered by variation of th
quantty of the charge, then it is clear that the power
drive the mill is proportional to the product of the cross
i__;~\1
06
08
10
FIG. 3.5
l1 1 \ cnce between the balls. There is also interference
I w ~ the balls and lifters, etc., and, when the relation1 iHdetermined by experiment, a vastly different curve
. :ls? shown in. Fig. 3.5, is obtained. The importanc~
l l ~ dfference wll bec?me apparent later, when various
l d f~mulae, whch. have been proposed for the
w requrements of a mll, are discussed.
11 l1 a~ already .been explained in Chapter 2, in which
I I t of a mll charge is discussed that the coefficient
l : between the ball surfaces ~nd the number of
l . ii H the charge are, from the point of view of motion of
, '' , not independent variables. For this reason the
82
BALL, TUBE AND ROD MILLS
functions 4 ( d/D ) and 7 ( f) must be considered toge tit: .
Fig. 3.6a are shown the results of tests on a smo t
mill containin g clean steel balls and in Fig. 3.6b of IH
a mill with lifters. Examinat ion of these curves showN
that, for the practical case of (D/d) greater than 20, tl11
value of the function does not differ significantly fro11
unity. For values of (D fd) appreciab ly less than 20, how
POWER REQ.UIRE D DRIVE MILL
83
r, the value of the function is not independ ent of the
of the mill.
to test the effects of variation of the coefficient
order
f ' riction tests have been carried out in which:
flling
( 1) The balls and mill are cleaned and tested dry.
(2) The balls and mill are cleaned and slightly oiled.
(3) The balls and mill are cleaned and treated with
molybden um disulphide .
(4) The balls and mill are contamin ated with a trace of
powder.
(5) full powder charge is used.
' l'he coefficient of friction between surfaces treated witl
olybden um disulphide is about 005 as against 01 to 02
I ' i1ed surfaces and 03 for dry steel on steel. The coeffi t of friction between the surface contamin ated with
wder is unknown but is probably very high, say 05
l t lO.
n. order to investigat e the effect of small quantities of
( twder, quartz powder from the sample passing through
' 00mesh sieve was added to the ball charge under
stigation and no~mal tests carried out. The quantity
l > wder used for th1s purpose was 50 % of that required
t fiiL the void spaces between the balls. This quantity of
, gc was sufficiently small to ensure that, for all prac . ! purposes, the change in torque, from the correspon dI , <:nse of the mill charged with clean dry balls only, was
l the change in the coefficient of friction and not to
I i reased densi ty of the charge . As a resul t of these
IH, no significan t difference was found between the
it for the cases (1), (2) and (3) above, and the cm
l < l esults for these tests can be presented by the full
l1 of?ig. 3.7, this curve, ofcourse, correspon ding to that
(. / ) = 05 on Fig. 3.6. It is probable that this drooping
ll': teristic arises from the comparati ve freedom of
, of the balls when the coefficient of friction is low
l wl1 cn the number of contact points between the sur1 11 s small; as is bound to be the case when (D / d) is small.
84
'' ; PO WER REQ.UIRED DRIVE MILL
BALL , TUBE AND ROD MILLS
The effect of adding a small quantity of powde~ ix
shown by the broken line on Fig. 3. 7, this curve beg.
adequate for all values of (J) and, furthermore, .a study
Fig. 3.6 suggests that this curve is adequate fo~ l1fte~ mi_lls.
Thus, it m ay be concluded that ~or smooth. ~11ls gnnd1n A"
materials having a low coeffiCient <?f fnct~on 4(D(tl)
should b e read from Fig. 3.6a. For l1~ter m1ll~, ?r miii H
grinding materials giving a high coeffiCient offnctn, tlc
functional relationships should be r ead from the dottccl
line of Fig. 3. 7.
111 11 l: r d with iron shot or glass balls, has "slumping"
i l :r e been observed during dry milling.
l , 1 cxplanation ofthis phenomenon can be offered
11 I i r bable that it is in some way bound up with a
11 i the frictional characteristics of the mill charge.
sul ts of tests carried ou t to determine the effect of
l l> r of lifters, n, on the power requirements to
J Q75 ,
.......
10
0 8
~;.::::
J s
v
/
I\ 11 J 0 25 ....
Bo lls Onl y.
StZc l Bolls With Trocc Of Quortz
\ .
20 . D
(!)
Fc .
30.
~~=
I li
,I
D;d 20
t?f I
0 7
40.
S tZcl

............_
(')
02
85
5
IO(D/d) 15
for(IJ.j)<20 Muily j1) <n> &y
5
20
25
50 ,
4
N umb r
3.7
Fc.
It has been suggested by
that for val~,es of tl,c ;
Fatio (Dfd) greater than about 80 the ball charge slump~
the curve of Fig. 3. 7 falls for values of the rato
an d ' thus
'
. also state d t h. at t h'
.
greater
than
this figure. It 1S
.~s.eHecl
particularly marked in wet mlling, but, s1~ce 1t 1s rall1c
unusual to use balls .so small that the rat. of (D fd) c
ceeds 80, it is probable that this effect 1s not of gn:t
practical importance.
In view of this comment the present authors havc, l1y
use of the smallscale apparatus, subseq~ently extend.ccl
the range ofFig. 3.7 to (D jd) :::200 and, 1n no case, wtl
6
L if t c:r s  n .
10
3.8
Bond, <5>
rp
: nill, that is to establish 5 (n), are plotted in Fig.
' : unction 5(n) is influenced by a large number of
, l> l s to a small extent, and the curves presented in
l ; are the lines best representing the whole of
\11
l1 l a.
rp
This method of presentation is adopted since it
l1 l a t the errors introduced by the use of a single line,
111 1 : many, are not significant but that the simplifica 111 oiJtained is worthwhile.
1 1' the figure it can be seen that, provided the number
I lll s is greater than about 6 and the ratio of (Djd)
86
'1'1 1 POWER REQ.UIRED DRIVE MILL
BALL, TUBE AND ROD MILLS
exceeds about 20 the function has a constant va1ue '
unity. Thus it fo1lows that, provided t11ere a.re suffici nt
1ifters to effective1y key the charge to the mll shell: tl1c
number of 1ifters is unimportant. For a number of 1ftcH
1ess than 6, but with (D fd) 1ess than 20, there is s
dependence upon the va1ue of mill filling as shown by thc
branching of the curve to the 1eft of th~ figure. For prac.
tica1 purposes the va1ue of the functon, for va1ues
(D fd) 1ess than 20, can be ca1c~1ated by m~1tip1ying tl
re1evant va1ue read from the man curve ofFg. 3.8 by thc
10
0 8
 
'4
""::.
e  D;d=20
@  DlJ = 10
Q4
Rot
0 1
Fa .
3.9
appropriate correction factor read off from the gral
inset in that figure.
The effect ofthe height ofthe 1ifters is given by the fu
tion 6 (h/D) and the resu1ts of tests t? de~e~mine this fun.c
tion are given in Fig. 3.9, from whch t s seen t~at
mills in which (D fd) is greater than about 20, the heght
the 1ifter has 1itt1e effect on power requirements. For t
1ess usua1 case, in which (D fd) is 1ess than 20, there js
considerab1e deviation for these 1imited cases, but
straight1ine interpo1ation between the t~o curves ~l
Fig. 3.9 is probab1y adequate. T?ese conc1~sons a!e val~cl
on1y for 1ifters of reasonab1e heght, that s, for 1fters
which h is not greater than, say, 02D and not 1ess th:
about d/2. If h exceeds 02D the 1ifters undu1y interfir
87
l ~: motion of the ball charge, whi1e if h is much 1ess
l
11
rl/2 the balls ride over the 1ifters and so 1ocking to
d.oes not occur. This at once suggests that "wave"
I  :p", and other such 1iners cannot act as effectiv~
lll , .
I'I 11:. .rect of the partic1e size on the power requirements
l . ll wou1d be expected to be small, provided that the
I' c~s we~e not so small that powerful aggregation takes
'''' ..lf ,t,hs were to occur, the charge cou1d become
I' ll y ll{~ and the J?Otion of the ball charge would be
s ly mpeded; wth a consequent effect on the power
l . ds ofthe mill.
l ' l c : results of tests, carried out on smooth and lifter
, .to ~etermine the form of the function 9 (Dfb) are
Fg. 3.10; these tests being such that the va1ue of
l ; r up (Dfd) exceeded 19, (J) varied from 01 to 075
l the case ofthe lifter mill, (hfD) =01 and (n) =6
INt the group (D fb) was varied from 77 to 740 thi~
11 <: f values ofthe group (Dfd) largely representin'g the
a1 range cyered by the opera ting condi tions of a
.. ~umber of mlls. The points relating to different test
l rts .are seen to suffer a considerab1e amount of
I ( . Ths sc.atter pro?ably arises from the interdepend1 . c :. f the varables whch has been discussed earlier, and
I I.t .that, un1ess ~ very great complication in the form
/ . c l dtJOnal correc~on factors is introduced, this must be
: ?ted. For .practcal purposes such complication is not
tl d, and, the present work, mean curves have been
l , w through the points ofFig. 3.10. When this is done
l r: rves for the two cases are seen to be identical and
I' the fu?ctional relationship 9 (D jb) shown in Fig. 3.11.
I ) creasng the va1ue of the (D fb) ra tio in a given mill is
"' to decreas.e the power necessary to drive the mill. This
. y be ex~1aned on the basis that the 1arger particles
I 11:11ate adJ.aCent balls to a greater extent than do the
. ll r partcles, and, therefore, increase the dilatation of
l h~rge . The centre of gravity of the combined charge is
l s dsplaced towards the vertica1 through the axis of
I''
' ~ PO WER REQ.U IRED DRIV E MILL
89
l 1l 1 nd so the torqu e is decre ased. In the small mills
/ hese tests, the prese nce of very fine powd er, such
ll l; ratio (D j b) great ly exceeded 800, result ed in
I' 1 ,1 i n of the powd er from the balls, bu t this only
. The powd er formed
' ' ' 1 c! at speeds in excess of OBNc
wall of the shell,
rical
cylind
the
d
r us layer aroun
decre ased by a
mill
the
drive
to
red
requi
l l power
. mount when this occur red. This pheno mena would
/ , l>/ y not occur with large indus trial mills, but data is
L+++t R
with
l v:il able this point . The quest ion is boun d up
ofBOO
jb)
(D
of
ratio
a
and
le
partic
ofthe
size
1/ 1 . / s lute
.__4++~:"~tli...., ~
(/
.._.f!ll',
.. , "
L+++ir ~ ~
,,
,
)4
2
200
400
600
(%)
800
FIG. 3. 11
/ls used in the tests corres ponds to a charg e consisting
l : icles of0003in. avera ge diam eter. Generally, howt in the case of very
' 1 , it would be expec ted that, excep
. L~~~~~~~~~~6~o
~L
Sll~W HlOOWS (qja)'
..:.
e>
CD
,.._
SlliW '131 (qja)'
"'
I ' Rrind ing, this effect is unim porta nt.
er
'l roughout the series of tests, the value of the powd
;
unity
at
ant
ll i , group (V) has mainl y been held const
is
e
charg
er
powd
the
I being the condi tion that
lr tically just sufficient to fill the void space betwee en
l>alls. In fact, the powd er increases the total volum of
e of
l l1arge and a layer ofbal ls may occur at the surfacwith
ed
unfill
is
which
en
l harge, the void space betwe
wder. This restri ction on the value ofthe param eter (V)
jl
, ,
' ' f POWER REQ.UIRED DRIVE MILL
90
BALL, TUBE AND ROD MILLS
has been made because, in practice, less powder than ~
necessary to fill the void space is rarely used since a pa
of the ball charge is then inoperative in the grinding pr cess; even though the power necessary to maintain tiH
centre of gravity of the inoperative portion of the ball
charge in the displaced position must still be suppli cl.
Conversely, if the powder is excessive, choking of the mill
occurs, and a portion of the powder charge is remote fr0111
the ball charge at any instant and so cannot be effectiv ly
ground. Thus the decision to restrict the value of the parnmeter (V) to unity is based upon the practical importan
of this case and the necessity, in view of the large numb
of variables involved in the problem, to liinit the amount
of experimental work.
Although this restriction is useful from the point ofvi w
of limiting the number of variables to be investigated, it
imposes the restriction that the results are strictly appli ~
able only to batch mills and not to the industrially i
portant grate mills and overflow mills. Thus the results
the present investigation, which has been carried out
the case of ( V) = 10, are strictly applicable only to a bat
mill. The results have, however, been extend ed direct1y
the cases ofthe trunnion overflow mill and by inferenc to
the grate discharge mill. Consider now the extension of t IH
work to these cases; the mills together with a batch mill
have already been illustrated in Fig. 1.4.
With a batch Inill, as previous1y suggested, the centrc
gravity of the ball charge and powder charge sensi])l
coincide and so the density correction ( 1 + 04/ ) pr
vious1y deduced is applicable.
In the grate mill the powder surface probably sl
from the in1et end of the mill to the discharge end in l
manner indicated in Figs. 1.4b and 1.4d. In this case tl
moment ofthe powder charge about the axis ofrotation l
the mill is probab1y much the same as in the ball Inill ; tl
greater moment at one end of the mill being rou g ll y
ba1anced by a smaller moment at the other end. Thus, l
method of the present work would be expected to l r
l : blc
91
to a grate mill with an accuracy sufficient for
11111 :>ses .
l case of the overflow mill, in order that the pro l N l a ll overflow through the hollow trunnion, it is
NH: ry that the genera1 leve1 ofthe powder surface must
' > nd to about a 45 50 % filling of the mill Then
tl1e ball filling is about 4550% the overflow Inill
''' : nds to a batch Inill, and so the power input shou1d
I l tly calculable from the present work. For small
14
~r
lron Powd~r p78
~
........
Silic Powd~r p:26:"
Not~!
02
'Stondrd' Mill ls Mill ln Which
Th~ Powd~r is Just Suffici~nt
I Fill Th~ Soc~ B~wv~n ~h~ Bll;
"
01
0.3
0 2
Boll
F c .
Charg~
0 4
0 5
J .
3.1 2
ls ball ~harge, however, a correction factor is
~ :;~ary an? th1s has been determined by experiment and
~, V('l1 F1g. 3.12 .. the use ~f this factor the power is
l 11l a~ed on the bas1s that the m1ll has a ball charge equal
: In the actual overflow Inill and the normal correc l> pow~er. charge is introduced. This calculated power
l n multlied by a factor, corresponding to the value
l l ' actual ball charge; the factor being read from Fig.
I I ' . Th~ shape of the curve of Fig. 3.12 would be ex1" d, Since for very small ball charges the powder
BALL, TUBE AND ROD MILLS
92
standing above the ball charge 1eve1 contributes a rel ative1y 1arge moment, and so increases the power requirrl
over that ca1cu1ated on the basis of a batch mill. F
intermediate fillings, however, the contribution to tiH
moment by this additiona1 powder is small, but there iN,
in addition, a 1oss ofmoment due to this additional powd
disp1acing the centre of gravity of the re1ative1y heavy ba ll
charge towards the centre of rotation.
Strict1y, there shou1d be a correction curve for ea l
va1ue ofthe density ofthe charge materia1, but, in fact, t11t'
correction is small and when regard is paid to the overall
accuracy which can be expected in ca1cu1ations of tl1i.
type, interpo1ation between the curves of Fig. 3.12 is ;ll
that is justified.
The parameter is the characteristic of the powcl<'l'
which is a measure of its resistance to grinding. Thus tlH
quantity (EfD3N2 ) wou1d not be expected to affect tiH
instantaneous power input to the mill, a1though it woull
be expected that the rate of grinding is dependent upon i t.
The question of rate of grinding will, however, be treatcl
in a 1ater chapter.
The tests carried out with different materia1s in the mill ,
in order to determine the effect of the density of tl
powder, show no significant variations which can l>c '
attributed to the difference in the va1ue of the param t'l'
between the materials. Thus it appears that in connt('
tion with the power input to the mill, the dimensionlNH
group ( EfD3N2) is not a significant variable, and c1
12 (fD 32) is e1iminated from the equation.
ln order to compute the power required to drive a mill
it is necessary to inc1ude the values of all the various r lc
vant factors previously discussed and on the basis of tli H
foregoing work, the equation for the power becomes
(n5~3 ) =(+ 0 ~4 )(~)'(~)3(J).4(~)()
6 (i) .g(i) 11(V)
.. . . (3.1
POWER REQ.UIRED DRIVE MILL
IH
'c'
93
4> 1'(NcfN) is read from Fig. 3.2,
3(J) is
4 ( dfD)
read from Fig. 3.5,
is read from Fig. 3.6 or Fig. 3.7 as required
by mill parameters,
5 ( n) is read from Fig. 3.8,
6 (hjD ) is read from Fig. 3.9,
9 ( b JD ) is read from Fig. 3.11,
(/>n( V) is read from Fig. 3.12 for a trunnion overflow mill or is taken as unity for a batch mill or
gra te discharge mi11.
app1ied to the computaof 1arge industria1 mills
I . I' on the published data given by Taggart, {6) and th~
1 111 ts of these ca1cu1ations are shown in Fig. 3.13. From
l l .rv~ it is seen that the points :re scattered a1ong good
. ht 1es, even though the rat of the horsepower to
I v the 1argest mill to that required to drive the mode1
l ' the order of 30,000 to 10. Thus it appears that there
systematic errors in the resu1ts of the mode1 tests.
11 c
connection with the scatter ofthe points, it is perhaps
l ~ 11terest to mention that the data quoted by Taggart,
\ ~ hav~ been apparent1y obtained by circu1ating a
~ tonnare to a large number of mill owners are for
l llowing reasons, subject to some uncertaincles. '
'I ' equation (3.11) has been
the power requirements
The power quoted is the power input at the motor
terminals. Furthermore, this figure is often estimated; or even the nominal installed power is
quoted.
(2) The overall efficiency from the motor terminals to
the mill body is unknown and can vary from mill to
mill. For purposes of ca1culations an overall efficiency of 85 % has been assumed for mills with concentric drives, and an overall efficiency of 75 % for
mills with girth ring drives .
(3) Th~ figures for diameter and length are quoted on
varous bases; for example some are the nominal
dimensions of the mill and others are inside new
()
POWER REQ,UIR ED DRIVE MILL
BALL, TUBE AND ROD MILLS
94
610
HJRENCI
600
ctischargc: mil1s . .......,
Gra~
HOL[GR
L_
SUIYINE
v
[{;'
t
ENGELS
....
~~
ENGH~
L
MAGNA
L.
!?
i'
2
V
O
LL
v/
F~NOH D 'ERDE
lL/
F'
NIHD
lEROE
1
"'...
Overfiow trunnion di scharge:
mill .
~
I
~OLI
v/
HI.&.MI
i/
'/
//
UrAH__,)
ARTHUR
}'
~ s
~
iL:=
iz ; that is to machines absorbin g even several thousand
losepower. The method of calcu1.ati on develope d in this
wo rk will be demonst rated by applicat ion to the data,
l a ting to a large mill, publishe d by Carey and
, ; irmand.<7)
lifter mill of6 ft diameter , revolvin g at 187 r.p.m., is
i rding coal. The body of the mill is divided into two
: s : one section, 7 5 ft long, being charged with 5 tons
! ' RLcel balls, of 2 in. average diameter , and the other sec111)11, 15ft. long, being charged with 6 tons of "Cy1peb s"
l ' a in. average diameter.
l'or the section 75 ft long and assuming 40% voids
l Lween the balls and in the coal, the above data lead to:
~/
L vv
odels may be applied with confiden ce to the calculati on
l ' Lhe power input to industria l mills of even the largest
ROAD
~/
~/
10
000
'"
JOO
'"
)00
OSER I/E O POWR DR V E
)10
GRAH BA.LL
'"
500
510
600
650
100
HlLSHORSEPOWER
~~so~oo~7..,.o,,~oo,>~so~Joo~~Jo~~oon~..~O~
OBSER VfD POW[R
95
Volume of balls = 38 cu. ft
cl Volume of chamber = /4 62 75 = 212 cu. ft
(J) = 018
'hus
' om equation (3 .5), the critical speed for both sections
I :11. r.p.m.
w from the appropri ate graphs,
= 92, 3(J) = 065, 4 (Dfd) = 10,
5 ( n) = 10, and 6 (h/D) = 10
(/> '(c!)
DkVE Q~.W L HLLSHORSEPOWER
Fro. 3.13
liners. Since the power input, , is related to tiH
diameter D by the expressio n = KD 2 5, the u
certainti es in the diameter of the mill can produ
consider able inaccura cy in the calculate d valuc '
the power input.
(4) The density of the material being ground is nol
known precisely.
In view of these uncertain ties, it is believed that tl11
agreeme nt between the calculate d and measur~d pow rH
is as close as can reasonab ly be expected , and It may \)
conclude d that the results of this work upon smallscal
I since the particle diameter, b, is probably less than
lt
'e
(,
in., 9 (D fb) = 10.
from the equation
~3 )
(+ ;~) . ()'(~)3( J).4(i)
0
5(n) 6(~) .g(~)
04 . 14) (75)
6 92 0605 l
79
10 10
( 1+
96
or
= 820,000 ft pdlfsec
= 495 h.p.
I 111
If, for the section 15 ft long, it is assumed that th :
"Cylpebs" pack with a voidage of 40%, then (J) = J :1
and, by similar calculations
= 66 h.p.
From the published data, the measured power to tl1.
motor terminals is 103 k W, and so the power demand
86 k W by the millleads to a combined efficiency of m
and transmission of 83%, which is reasonable. Thus tlc
agreement is as close as car. be expected .
Considerati on of equatron (3 .11) wll show that for
speeds 1ess than 80% of the critical,
D53.(1+ :) (~)(~. 3 (J)
so
=
=
tl1is CaSe,
=
KD Zo
(3 .14)
l i ~ a widely accepted relationship for the power
lo
(1.
mil1.
I . , it has been found by Gow, Guggenheim , Campj,, 11 111 Coghill,<B) Fahrenwald and Lee(9) and others
l , i
( '
IIJ
I"
ractice, the power demand is given more near1y
lationshi
= KD26
.... (3.15)
I~
4 '/d)
(
,. '!
.e
'''"''~'"'"'/
(%) 75
11
Kfy'D
D5. (K/D) 3/ 2
KDH
j ll
11 1
.... (3.12)
the remaining groups of the equation being neg1ected f
the time being.
Now, for a given mill filling and a ~iven. material bein l
ground, 3 (J) and (1 +~~f) are nvanant, . and for :
constant fraction ofthe cntcal speed 1 (,f) s constan.
Furthermor e, since the mill runs at a given fraction '
its critical speed, it follows from equation (3 .6) that
=
l l
Thus the power to drive the who1e mill
= 495 + 660 = 1155 h.p.
= 86kW
p =
97
' ' ~ POWER REQ.UIRED DRIVE MILL
BALL, TUBE AND ROD MILLS
. . . . (3. :)
This equation is true for complete similarity betwc
1engtl and diameter of the mill but, for a mill of constaf
length, it follows that
(L/D) oc (1/D)
10
F.
(%)
15
20
25
3.14
Fig. 3.7, however, it would be expected thq.t
varies approximat ely as (D fd)Ol and in Fig. 3.14
l o wn the curve (Dfd)0075 plotted the same axes qs
l l o tted curve of Fig. 3.7. It is seen that the fit is
b1e.
' s it follows that in tests in which the ball diameter is
lnt the ratio (D fd) varies direct1y as D, and so the
w the mill, which varies with 4 (Dfd), will vary as
: pproximately. Thus, it follows that, if the ball dia is maintained constant, the equation (3. 14) must be
Iilied by DOl, which brings about agreement with
( . t (3 .15).
' n
l ( f) jrl)
98
POWER REQ.UIRED DRIVE MILL
BALL, TUBE AND ROD MILLS
One of the earliest attempts to derive, from theoreti : l
considerations, a formula for the power to drive a mill i~
that of Davis, (lO) who starting from the basis of the b : 11
trajectories discussed in the previous chapter, calculated tiH
striking velocity and so the kinetic energy of the elem ni H
ofthe charge in a giveri trajectory and, by integration, tlc
kinetic energy to the whole charge; this kinetic energy
clearly being supplied to the mill from the external sour
99
' equation (3. 16) is not directly comparab1e with
n_ (3.11), since for a given composition of charge,
l1 w ght of the charge, W, cannot be varied inde l 'ntly ofthe diameter ofthe mill and similarity main1 i c d at the same time. Thus, the functional relationship
I 1(.1) of the previous work is absorbed in the form of
: (3.17). Furthermore, the equation of Davis
I,., i d for the conditions of maximum efficiency and
is
so
general.
equation developed by Bond(ll) for mills dryling is
1 11
=4~~4+~~~~~
I'
126{D04
~c (6 16 575J) 
01
210(N,/N06)l}
~N3J++~~~~I
.... (3.18)
Q
1
~~~~~++~
c(
(( ~~~~~+~~1
2
L~
nqth
FG.
F~~t .
3.15
On this basis the following formula for the power input to
the shell, when working under conditions ofbest theoreti : l
efficiency, was obtained:
(lK3)
(lK5)
= WR3 /2{ 0004467 (l2)1 8 0003700( 1 2 ) 318
( 17) }
+ 0000880 ( 1 _ 2 ) 518
where the value of
mate equation:
=
....
(3.1)
is given by the following approxi 
0024+039(7lQJ)I /2 .... (3 .1 7)
l c c is the power draft in kilowatts. per ton of grinding
ii . The equations of both Davis and Bond are based
l y on. the mill diameter, tht: mill filling and the speed
l otatn express_ed in ~~rms of the critical speed; the
l s of t~e equatns _ans~g from the empirical expresNII 111 ~ definng the relatnshs between several variables.
' the weight of the ball charge enters both these
tj! :ions directly, it follows that the power is proportional
l length ofthe mill; a conclusion which is in accord
1vi 111 the equation of Rose and Evans. (12)
l1as been suggested by Gow, Guggenheim Campbell
c l Coghill,<I3) however, that_the power do~s not vary
lt ' ly as the length of the null, but these workers give
111 pression which reduces to
2(LjD ) =
(05Ll)K+1
.... (3. 19)
L is the length ofthe mill and is a constant having
l value of09 for mills oflength less than 5 ft and 085 for
i of length greater than 5 ft. This function is plotted in
' .(. 3. ~5, and it is seen that. for long mills the power
c: ?s1derably smaller than s demanded by a linear
11 .
BALL, T U BE AND ROD MILLS
100
The higher power requirements of a short mill woul I,
p erhaps, be expected, since the end plates of the mill migl11
cause a "pilingup" of the charge at the ends of the mill .
Such an effect would, however, be expected to be 1
ligible in a long mill. Thus, this deviation for a long mill iM
surprising and it is believed that this formula should bc
accepted with caution.
Hancock<l4) proposed the formula
. . . . (3.20)
= 000078C3Ln
' ' POWER REQUIRED DRIVE MILL
101
11 ~S
' =
==
==
K 1D 5 N 3 D~
2 2 (J) t ( 1 +04~)
2 {D 3 3 (J)}L
00005 lNC3L
(3.2 1)
l , he general form of
n nt b~t the value
lc : s to be lll error.
where C is the length in feet of the chord defined by tl~e
surface of the mill charge, L is the length of the mill
feet, is the mean specific gravity of the charge and
the speed ofrotation in r.p.m. This formula approximatcH
to the formula of Rose and Evans insofar that the tcr
( 1 + 4 j) is proportiona1 to thc mean density ofthe pul) ,
and so to of cquation (3.20), and in that the powcr
proportional to the length of the mill.
Furthermore, the quantity C3 may bc written:
the two equations is in close
of the coefficient (0 00078)
'(C/D)3
ByExpHim~nt . ?. '\
C3 = D 3(C/D )3
but (CjD ) is the sinc ofthe semiangle, , subtended by tl
surfacc of the chargc at the centre of the mill and so
C3 = D3sin3
02
04
Examination of Fig. 3.16 shows, however, that for tl
practically useful range, of up to 50% filling, the curvc
ofsin3 and 3 ( J) practically coincide. Thus
C3 = D 3
3(J)
Also from equation (3.6) it follows that for constac
in the ratio of running speed to critical speed
n oc 1/y D
Then, on making these substitutions, equation (3.1] )
= D53 '( ~) 3 ( J) ( 1 +04~) (t)
F.
08
10
3.16
J 'rH values of_ mill filling greater than 50%, however,
I : <greement s not good but, in practice, high mill
ll s are_ perhaps not so frequently used as the lower
, \ ~s so t ~ay
concluded that, provided that the
' "(' t val:e s _assgn_ed to the coefficient, th.e equation of
11 , 11 ock, . spte ~f ts simpler form, is, over a practical
{ ofvanables, good agreement with those of other
\ ker~ . Thus this formula includes, approximately, the
\,. nsonless group 3 (J) of the previous work.
?e
I le dependence of the power demand of a mill on the
11 ure of the pulp does not appear to have received a great
102
BALL, TUBE AND ROD MILLS
deal of study, and, in genera l, observ ations on this matt
take the form of genera l statem ents. For examp le, T aggart05) states that, other things being equal, wet millin~
require s 6090 % of the power for dry milling . This vicw
is confirm ed by calcula tions of the presen t author s, basccl
on figures for two mills of about 8 ft diamet er and 45 1
long kindly made availab le by Messrs. Edgar Allen & C .
ln this case the power as calcula ted by the formul ae
Rose and Evans was in extrem ely good agreem ent witl
the observ ed power require ments of the mill when dry
grindin g (within 5 %). Howev er, the power calcula ted
the basis of drygr inding formul ae with modifi ed chargc
density was found to be about 20% higher than the actu~l
deman d of the mill when grindin g wet.
It would be expect ed that a fine "smoo th" fluid pul>
would reduce the power deman d, since such a pulp woulcl
act as a lubrica nt to the mill charge and so reduce thc
equilib rium displac ement of the centre of gravity of thc
charge . This reduct ion of the power require ment of tl1c
mill is probab ly bound up with the phenom enon
"slump ing" ofthe charge , mentio ned earlier. More viscou ~
pulps would, by increas ing the displac ement ofthe charg<',
lead to increas ed power consum ption. With increasin !
"tackin ess" of the pulp, a reduct ion of the power requircments would be expect ed, since, in such a case the baliH
might adhere so well to the shell that a part of the chargc
is spread over the shell and, thus, is balanc ed about tl1c
axis of rotatio n. This decreas e in power deman d witl
pulps of increas ing thickne ss is in accord with th
observations of Gow, Gugge nheim, Coghil l and Cam
bell (16) but these worker s did not carry the tests to tlc
poi~t where the balls adhere d to the shell of the mill. ThiH
effect is, howev er, also observ ed in mills which are used for
the dispers ion of pigmen ts in varnish , and similar dutic11,
and so is very real.
From what has already been said about the interactio
betwee n the coefficient of friction and the magni tude
the ball charge , it would be expect ed that the effect of th
. l POWE R REQ.U IRED DRIVE MILL
103
c: pulp conditi ons would be relativ ely more iml in mills for which the charge is small. This con in fact follows from the observ ations of Gow
' '' n leim, Campb ell and Coghill<t6). Thus it would b~
I that, for large fillings, the effects ofthe variati on
l l u lp charac teristic s would be, largely , in accord ance
l what would be expect ed to result from the conse'1'" linl changes in the coefficient offrict ion involve d.
2 5r.~
~ 1 5rt+"""7L_+;;;~1
V>
:
1 0
ri7''+:"'++~
02
0 4
0 8
1 0
Pup CoNSSHNCY
FIG. 3.17
iew of the lack of data on this subj ect, it would
'' ': r that the best correct ions for pulp charac teristic s
I an be made is to assume that, for very thin pulps
l derate values of J and (D jd) the power required is
ll % f that for dry milling , and then to use a linear inter l .ti for less liquid pulps. For Iarger mill fillings or
. l values of (D fd) the reduct ion in power would be
a for the previou s case and the multip licatio n
I
slould. b~ m~dified accord ingly. The genera l cor ss of this vew s suppor ted by Fig. 3.17 i which the
104
POWER REQ.UIRED DRIVE MILL
BALL, TUBE AND ROD MILLS
results of tests by Gow, Guggenheim, Campbell and Col
hill, (16) are plotted. Although these oints are small i
number and rather closely spaced they give lines whi
when extrapolated to zero pulp density give 022 and 015
horsepower as the powers to drive the mills without pul
charge ; these figures being in the ratio 205 to 10.
R eference to equation (3.9), however, shows that, ; /1
other variables being maintained constant, the ratio of tl
powers is proportional to the ratio of the square of th
speed; in this case (070/050) 2 , that is in the ratio 196 to
10, which is in good agreement with the figure 205 to
10. Furthermore, for a mill running at 70 % of the criti : I
speed the ratio of the power when grinding a pulp of 50 %
density to the power when dry grinding is about 06 to
10: which is in accord with the previous general statemcn t
that when wet grinding the power d emand is about 60 %
of that when dry grinding. Thus it appears that,
roughly, the linear interpolation previously suggested ili
adequate.
The possibility of calculation of the power required to
drive a mill from the knowledge ofthe horizonta1 displa
ment of the mill charge has been developed by Carey a nl
Stairmand. (17)
From the principles discussed earlier in this work, it iN
easily shown that
= WN!f, ( 1 / 314)
105
w in Fig. 3.18. These curves are stated to be applicable
ills having diameters ranging from a few inches to
f ' l 'c feet and, practice, tO gie reaSOably reliable
I
l s .
.nalysis of this method in the light of the previous disshows that the effects of the weight of the charge,
l so ( 1 +0 4f ) , are inc1uded. Also the effects of
 
Smooth
is .
  Ltfttr Mills
0 20
40
0 60
DBO
tyN,
. . . . (3. 22)
where is the power input to the shell in kilowatts, t
maintain the displacement of the charge, W is the wei g lt ,
in tons, of the grinding media plus powder, is the mill
speed in r.p.m. and 1, is the horizontal displacement
feet of the centre of gravity of the charge. Thus, the pw
input may be calculated provided th e val e of . can
obtained.
For this purpose Carey and Stairmand furnish a gra l
which gives the value ofthe ratio H1,/D (D is the mill di:
meter), in terms ofthe mill filling and the ratio ofthe sp l
ofrotation ofthe mill to the critical speed; this graph bcir ~
FIG.
3. 18
/ '(Nf Nc) , 3(J) and of lifters are included, and so it
, rs that only variables which have been shown to be
l omparatively minor importance are omitted. Thus, it
1 l d be expected that this treatment would have an
acy
sufficient for most practical purposes.
power requi~ed to dri~e a rod mill does not appear
: e b een extensvely studed. As a first approximation ,
wo uld be expected that the value of the function
/ (/) would be different to that for a ball mill but
106
BALL, TUBE AND RD MILLS
1': PWER REQ,UIRED DRIVE MILL
that the values f the remaiig fuctis wuld t b
seriusly altered.
Sice the "vidage" i a bed f circular rds, packed in
the tightest array, is abut 10 % (0095) ad assumig th
vidage f the pwder t be 40
it fllws that
t y high csts for the tarriage f replacemet balls
l s. f such lumps f ore be justified ecmic
IH scc the ~r~dig efficiecy is adversely affected
l 1.' l y ad the mllg.cst correspdigly icreased .
l s of such crcular r spherical materials
ll l ad t a "lockig" fthe elemets fthe charge
11l a icreased displacemet f the cetre f
1\ tt y [ the charge, with a relatively icreased power
1 111 I. The pwer demad wuld, furthermre be mdi
III l y. h~ tightess f packig f the gridig bdies,
lt l1 ~fferet t that fr balls, ad by the desity f
tl 111 . al. The pwer t drive mills charged with such
111 ;ls des t appear t have bee studied hwever
ti wil. the preset state fkwledge, little ~re tha~
"" l1 s mates f the pwer calculated the basis f
11 I III'(~VOUS equatis mdified the lieS Suggested,
1 111 ' nade.
IIJ fregig work deals etirely with the pwer
1 : t t keep the charge withi the mill shell i
l~. Clearly, i rder t determie the pwer t be
lt ~d by the drivig mtr, it is ecessary t add t this
I .thc pwer ~st i the bearigs su pprtig the mill
11 , th~ beangs su~prtig the itermediate speed
:d the reduct gearig. The estimati f
l I sses, is, h?wever, a ~atter f geeral egieerig
Iwldge, ad s t specfic t the tumblig mill ad
~ ~~~ l1 n a tters are specifically excluded frm the p;eset
%,
( ~)
_ ( 1l ) +0 1(10 4)
(1l)
l+0 067f
.... (3.23)
Als, fr a give value f mill fillig (J) the lad f rds
i the mill will exceed that f balls i the rati f th
vlume f rds t the vlume f balls, that is,
1  01 = 15
1  04
Thus, the pwer t drive a rd mill, withut a
charge, will be give apprximately by
PR = l5PB
pwd
.... (3.21)
where R is the pwer t drive the rd mill ad is tl1
pwer t drive the crrespdig ball mill, as calculat cl
frm the fregig frmulae.
It is prbable, hwever, that the acti f the pwd
t dilate the charge, ad s t reduce the hztal diHplacemet f the cetre f gravity f the charge, wuld b
mre marked with rds tha with balls. This wuld le<HI
t a reducti i the actual pwer t belw that whicl
wuld be calculated by the use f the frmulae f the ball
mill, as mdified by the use f equatis (3.23) r (3.24).
This effect wuld, hwever, prbably be small ad so
culd be eglected.
further case which requires csiderati is that
which the gridig media is f sme frm such ;~
"Cylpebs"; ff cuts frm scrap drill rds smetimes bein
used i the minig idustry. Als i sme cases, thugl
rather rarely, large pieces f re are used as a grindi
medium. It appears, hwever, that ly i the case
107
I ,
REFERENCES
I ( , , . ., and EVANS, D. ., Proc. Instn mech. En:grs Lond
17 .
:w, . ., CoGHILL, W. . , and CAMPBELL, . .
(etall.) Engrs, T ech. Pub. , . 326.
1 1 t 1 . ., and EVANS, D. .. Loc. cit., ref. .
l. , W. ., and D, F. D.
'
.,
1956 170
'
'
Amer. Inst. min.
Ball Mill Grinding, U.S. Bur. of
T ech. R ep., . 581, 1937.
ll , F. C. Proc. Instn mech. Engrs, Lond., 1956, 170, 800.
I UR, . F. Handbook of Mineral Dressing (New York), 1945.
I: ~v, W . F.,. an~ STAIRMAND, C . J. "Recent Developments in
Mneral Dressng , Instn Min. Metall., Symposium, 1952.
Mncs,
108
BALL, TUBE AND ROD MILLS
., CAMPBELL, . . , and CoGHILL, W. 11 .
8 . Gow' . ., GU GGENHEIM
.
,
87 51 112 24
Tran>. Amer. Inst. mzn. (metall. ) Engrs , '
. t ' . . (metall) .
9. FAHRENWALD, . W ., a nd L, . .
mer. ns. mzn.
.
.
T ech Paper, . 375 .
19 19 61 256
10 Ds, . W . Trans. Amer. Inst . nin. (netall. ) Engrs,
, '
11 . BOND, F. c. Proc. Instn mech. Egrs, Lond., 1956, 170, 800
12: Ros, . . , and S, D. . LCoc. czt., rLeLf.
1. CoGHILL W 11 .
13. Gow, . . , GucGENHEIM,
.,
,
. .,
'
CHAPTER
'
14. ~~"~~: :{::. Trans. Amer. Inst., min. (metall. ) Engrs, 1934, 112, 77
15. TAGGART, . F. Loc. czt ., r ef. 6
d C GHILL W. 11 .
16. Gow, . ., GuooNHEIM, ., CAMPBELL, . . , an
,
Loc. cit., ref. 8.
f 7
17. CAREY, W . F. , and SAIRMAND, C. J. Loc. czt ., r e
COMMINUTION OF SOLID
BODIES
bl
ms associated with comminution of solid bodies
and complex. In spite ofthe large amount
/ I\ i ation which has been undertaken, the present
knowledge is such that there is no theory by means
,f 11 the behaviour during fracture of even a single
11 under the simplest _Rossible mode of loading can
li ted in reasonable detail or with moderate
:rus
, ( .
l :
absence of an adequate theory applicable to even
ase, it is clearly unreasonable to hope that the
l . r of a comparatively complicated system, such as
111 11 , which numerous p articles are involved and the
" ~ l [ loading gives rise to both impact and abrasive
111l ' of the p article, can be predicted from theoretical
11 11 l< a ions. Because of this absence of adequate theo l 1 , \ l<nowledge much of the design of large industrial
II iHbased on previous experience with similar mills or
ll d from tests on pilot plant. In spite ofthe absence
11 ! l <uate theories, it is intended to give in this chapter a
I ll rvcy of the present knowledge of the subject of the
11 11 '<' of solid particles, since it is believed that such
I\' Ic dge is desirable if the problems of grinding are to
l l rstood.
1'1t ' naterials which are ground industrially may be
l'' l1 omogeneous or heterogeneous. Thus, graphite,
i y lc num disulphide, pigments and drugs are, when
li< a lly pure, homogeneous from the standpoint of
nty of crystalline makeup. particular crystalline
111 il : l will tend to exhibit characteristic mechanical
l<:
109
110
~
~~

proper ties which sprig from the crysta1 structu re acl
associated factors, such as crystal lite size ad arragemct
of the crystallites withi the macro crystal or powd
particl e. Such materi als are homogeeous.
Materi als like iro ore, however, will geerally be composed of particl es of almost pure iro oxide, itermix:l
with particl es of silica. Thus, metal ores i geeral, co<l
ls. F
ad cemet c1iker are heterogeeous materia
sounl
is
there
later,
ed
discuss
be
reasos which will
t
differcn
have
al
materi
of
types
two
these
that
evidece
fractur e charac teristic s. The lie of demarcatio betwc 11
the two types ofmate rial is t sharp, howev er; for althougl
a heterogeeous materi al is fairly obvious, homogeeoN
materi als have iteral defects which itroduce some
ieeharactenstics of hetero eeous materia ls.
Cosider, i t e rst p ace, t e acture a heteroor ir 11
geeous particl e; for examp le, a cube of graite
grair
the
at
m
mediu
ig
cemet
the
,
geeral
i
Sice,
ore.
bou'Crares suc a maten a may e cosClered to a
a ower stregth tha the grais themse ves the fractur
Wtl , ! the mai, OCCUf by failure of the grai boudarit'N ,
ter a while the particl e will be broke dow ito tl11
reachl
costituet grais. Whe this stage has bee
by fr~H'
about
t
brough
be
oly
ca
further size reductio
t
stated,
as
er,
howev
Sice
lves.
themse
grais
ture of the
is probab le that the crystals of such materi als are strog r
tli H
tha the materi al at the boudary, gridig beyod
pari
stage.
earlier
the
i
tha
lt
poit is more difficu
t l~
to
ratios
coside
these
applied
has
g
Hitti
cular,
gridig process.
Thus, a graph of fieess attaied agaist the expecli
ture of work ecessary to attai the fieess would be
pected to have a form somew hat as show in Fig. 4.1 : .
Clearly , i the case of metals, it would be highly improl
able that all the grai boudaries would fail before ay of
the crystals failed. Micro photog raphs, by Felix a11l
Geiger, (2 ) of samples of steel which have failed undt~ r
brittle conditi ons show clearly that the fractur e is pro_p
111
BALL, TUBE AND ROD MILLS
Enorgy lnput .
Htrrogrnrous t.
.
j~tl
En<rqy lnput .
Homogr nrous Matrial.
b.
Enorgy lnput.
R<al Crystal.
c.
FG.
4.1
ys tals would occur simultaeously with the failure
l l)ounda ry and th~s w~uld elimin ate the sharp corner
11 t l1 curye and so gve nse to a nonliear relatioship.
I I rushng of a homogeeous materi al howev er the
would be differet. If the crystal lattic'e were pe~fect
l plates of the press ad the polishe d faces of the
112
crystal were absolutely plae ad parallel, the theor ti
cally the stress would be uiform througho ut the parti cl .
Presuma bly, whe the stress at which failure occurrr<l
were reached, t e crysta would sitegrate t a Clust
which the particles woula be of the same oraer szc :
are t1ie crystallit e or elemetary blocks of the cryst:f
attice. e if such a ideal crushig process did nol
occur, it would be expected that the particles of a homo
geeous material would be of the same stregth, regardl
ofsize, ad so the curve offineess attaied plotted agaiNI
work required to attai that fieess would be a straigl
lie, as show i Fig. 4.1 b.
That a crystal of a homogeeous solid will ted to brcak
dow commiutio ad reproduc e the characteristi
geometr y of the larger particle, is idicated by the work
Stott, (3) which has show that, gridig, graphite , talc
ad other lamiar solids retai a platelik e form down to
the smallest sizes. this there exists a codition it
mediate betwee the idea1 case metioed above and tiH
case of heterogeeous material s such as iro ore or th
material s of mixed crystallie makeup . The fact remain ,
however, that real crystal is perfect, ad so the crushin~
of a truly homogeeous crystal is of theoretcal inter ~
oly.
The imperfectios i a real homogeeous article m:y
be roug Iy c assifie u er t ree leadigs . tfie fir t
class may be placed "micro" defects, which may be of tlc
form of a i missig from a lattce poit or a simil:
defect i the ultimate structure of the material. Defects
this ature are probably so small i relation to even tlc
smallest idustrial particle, ad so uiformly distributc l
througho ut the particle, that, although they reduce l
stregth of the material below that which the san
material would have if the lattice were perfect, they havc
no marked effect upo the fracture characte ristics of tl
material ; which remais, i this respect, homogeeous.
the secod class may be placed "macro" d efc ,
Defects of this nature may be flaws or crac s extenai
BALL, TUBE AND ROD MILLS
COMMIN UTION OF SOLID BODIES
113
. : apprecia ble fractio of the crosssectioal area
11/ l >:r ticle. They may arise from defects i the growth of
I '. ystal, t~ermal str~sses, shock loading, etc. Defects
1 l ature, fpreset ay apprecia ble umber are of
' ce i coectio with fracture sice clearly' they
'
'
11. ~ tc p 1aes of weakess, similar to the grai buI ~.~ ~ a .heterogeeo:s p~;ticle, ad so would give rise
11 1 d.g charactenstc curve as for a heterogeeous
/1 ' ; l , Fg. 4.1a.
1 l ~?ird c~a~,s may be placed :v~at, for brevi~y, may
l . ll: mos~nc defects. '!he ?asc dea uderlyg this
I l or de ect IS tfiat there lS evdece that material s are
l of elemetary "bricks" havig a size of the order
/ lnndred lattice dimesios. Thus, for examp1e
/ l (I) sugge~ts. that the micelle, which represen ts th~
1I ' L ' ut coal ad coke, has a size of about 35111 : llumphr eysOwe n<5 > states that mieral salt (NaCI)
I lemetary structure of about 1500 . One of the
6
' w~i ters, Rose, < > ~a.s show that. curves of both the
ll l nsty and reflectvty of graphte have a disc
i y, which probably idicates a chage ofstruct ure or
1 1l1 ~ 11 in the surface codi tio of the primary parti~les
l g' the aggregat e, at a particle size ofabout 500 .
I ()W the cr stal st~ucture is "P.erfect"' the elemetary
a ordered WaY. the adjective
II .~ :re arrag
g that the arrayfthe "bricks"
idicati
merely
"
:
f i
1~ ~ d ~?d.not,~hat there are defects, of the type
1 I ~d as mcro defects above, i the crystalla ttice. If.
I II 1
I 1 ? ! he crystal is "mosaic ", the elemeta:ry: block~
be
11 1 a ordered array a,nd such a crystal would
the corres1 i : d to be weaker i brittle fracture tha
I l i g crystal with ordered array. crystal structure of
l : ure wo:uld agai suggest that gridig would
I 1 11 111 more dfficult whe the particle size reached that
1l I . nosaic bloc~. ~tartig from a large particle size, the
i ng charac~enstc curve for a mosaic crystal would be
d to be ~ ~hree parts, as show i Fig. 4.1 c.
1 ;c, the gndg of material s to below the mosaic
1
114
COMMINUiON OF SOLID BODIES
BALL, TUBE AND ROD MILLS
block size is rarely carried out and so evidence as to l
behaviour of particles in this size range is lacking.
It is known that, as grinding proceeds, there is an
creasing degree of aggregation of the particles, whi
makes it increasingly difficult to separate the prima
particles of the aggregate as, for example, in the prepar:
tion of dispersions of the solid material in a liquid vehi l1.
The underlying reasons for this are not clearly understoo<l ,
but a factor is probabJy the increase in the magnitudc of
the surface forces . as the particle size is reduced, th
forces approximat ing to those which hold the material
solid form as the particle size approq.ches that of tlc
elementary units from which the material is constructccl .
The foregoing elementary discussion is relevant only to
the crushing of a single particle but in practice a numb r
particles, often in a bed of several particles thickness,
crushed and so it is probable that the behaviour duri ~
multiplepa rticle crushing will differ widely from that
lined. In addition, the foregoing discussion involves l y
the energy supplied to the particle and not that suppli I
to the crushing device. This difference is of considera\)l
importance , however, because the results obtained fru
certain types of crushng are sometimes interpreted
arising from the crushing characterist ics of the materinf,
s an example of this fallacy may be cited the use of tlc
results of tests carried out upon a ball mill. From Chapt
3, it is easily seen that the change in particle size has prn ,
tically no effect upon llie power to drive a ball mill and
the energy input to the mill is proportiona l to the tim~ l
milling, as in Fig. 4.2. :When, however, the mill content
sampled and the fineness of the product plotted aga
the time of gndi:ng, the result is a curve, as shown.
At first sight this curvature would a:rpear to, and prnl
ao y does in part, anse from the increased strength of /
fine partic1es. In fact, however, many other more probal>l
mechanisms which would lead to this result are operativ ;
for example the reduced area around the contact poit
between a pair of balls, within which crushing must tak
/.
Ii
111
I1
115
the fineI wi lead to a decreased rate of grinding as
ases. Another possibility is the carrying out of a
~:r ticle by the air expelled when the falling ball
'/''"' : :l1es the stationary one, or even welding together
l1 fin st particles by the blow would tend to produce a
I f ' thc type shown. Thus, there is certainly no ground
l l;ting, withoutfurt her evidence, that the curvature of
l11 i constitutes proof of increasing strength of the
I 11 Is with decreasing size. Again, the existence of fine
111 ~ l, along with the coarser particles, may exert a
1 i g effect during milling; the fine material interl 11!i wi th the mechanical action of the ball on the
,_
"'
!~~~~
...
:!
c
~
c
~
~~~Timc
Scac .
FJG . 4.2
material being ground by forming a resilient
in which the larger particles become embedded:*
l l this mechanism is ofimportan ce is supported by the
l tion that the rate of grinding in a batch mill is in' 11 1~1 d when the charge is removed from the mill, the
ved out, and the oversize material remaining again
11 ' d to the mill.
: t l1cr of these explanation s forms a more practical e
l , : for the decrease in the rate of sizereducti on of a
tw l r with increased milling time than does that based
hypothesis that the mechanical strength of the
11 . al increases with a decrease in particle size .
l1is point it is interesting to consider the distribution
l l11: energy supplied to a particle during crushing.
' IJi s cnergy would be distributed in:
t: ll ine
1111 1i
116
COMMINUTION OF SOLID BODIES
BALL, TUBE AND ROD MILLS
( 1) Elastic deformation of the particle.
(2) Plastic deformation of the particle.
(3) Gliding or slip on atomic planes, twinning and p
sibly ot.her lattice rearrangements within Jt
mineral.
(4) Increased surface energy of the body.
1 11 1 , (7)
117
who in 1867 stated, apparently axiomatically,
I 11 '' l energy necessary for crushing is directly prpr
i l the new surface produced". Continual reference
l 'law" is found in the Jiterature SO it is important
I i li'i i nplications should be carefully considered.
I ' Ly from the point of view of an energy balance, this
wou ld hold provided:
ermanent lattice m
rents, wfiicli w6e smalor ar , 5'1 e materials, anl
~ Incre~sed surfac~ energy, i'S"absoroea.. llri::.!KJhe crus l
!E,g; process and does not reap2ear as heat. The remaincln
ofthe energy supplie , however, will reapp"ear as heat anl
this difference at once gives rise to the idea of the enerAY
which is "necessary" for crushing and that which iM
"sufficient". Thus, it is essential that sufficient encrgy
should be supplied to bring about enou,g_h elastic deform :
tion for fracture to occur, and no less guantit;y will suffi
When fracture occurs 1 however, this energy, less that c
responding to the permanent Iattice rearrang_ements anl
to the increased surface ener.gy~ ears as eat and
is this difference which is measured when a mill is run in :
calorimeter. Since, however, the energy for e astic, al
possioly that for a part of the plastic, deformation
essential iffracture is to take place, the use ofthe absorb l
energy as a standard of comparison is grossly misleadi f.\
as to the efficiency of a crushing process. As an exampl
this discrepancy, it has been observed tlat the enc.\
supplied in free crushing is about five times that calculat : l
on the basis of surface energy, whilst for a ball mill l
value of the ratio is often between 500 and 5000.
When a bed ofparticles is crushed the energy expendl
in friction between the particles must be supplied, b ,
again, this ultimately reappears as heat and performs : 1
usefu1 wo.rk in grinding. Thus, friction in the bed ol
particles will, again, lower the apparent "efficiency" :
crushing process.
Turning now to the socalled ''laws of crushing" whi l
have been propounded, the earliest of these is that of Ri The
energ~orreseondin
to
( I) The whole of the energy considered is absorbed as
(' 1 )
Hurface energy.
'L'Le energy per unit surface is independent of
article size.
l l : w could not hold when strain energy and plastic
i l tion energy is included. It cannot therefore be
1 1 t d to apply to the whole comminution process or to
'l! nminution of ductile materials. It would be expected
" I ' <pplicable
to the comminution of brittle materials
l the energy expended in elastic deformation is de' I c L from the energy input to the particle.
, \ 11 , for spherical particles the specific surface, in area
p r unit mass, is given by
S = 6Jd
.... (4.1)
~ is the density ofthe materia1 ofthe particlt; and d
I )itrticle aiameter, It fol ows t at tlie increase in specific
11 I:H; during grinding is given by
S 1 S0 =
(6/)(l/d 1
l/d0 )
= S0 (Rl)
.... (4.2)
.... (4.2a)
: R ( = d0 (d 1 ) is the reduction ratio.
I 11 w W is the energy necessary for the creation of one
i nf' new surface the ratios of the energies necessary to
I : unit mass of a given material through the reduction
l os R 1 and R 2 is given by
WR
RI1
WR2
= R2l
.... (4.3)
118
COMMINUTION OF SOLID BODIES
BALL, TUBE AND ROD MILLS
and ifthe reduction ratios are so large, in comparison
unity, that it may be neglected, then
WR f WRz
R /R z
.... (4 .3 )
Furthermore, the energy necessary to reduce unit
of material from size d0 to size d1 is given by
W
=
=
(6/)
W(1fdo 1/d )
W(SoS)
witl
mn NM
.... (4)
.... (4.4:)
These expressions, the use of which involves a know
ledge of the results of sieve analyses or of the specific s
face determination, are easily applied and are widely usc l
for the control of fine grinding operations.
The law of Rittinger appears to have remained unch:l
1enged unti1 1885 when Kick<B> propounded the "Gcsct .
der proportionalen Widerstande", but since that timc.:
controversy has raged, concerning the validity of the two
laws, which has not yet been resolved.
The "law" as formulated by Kick is:
(1) The quantities ofwork necessary in order to prodc c
equivalent changes in shape in two geometricall
similar bodies of the same material are in the rat
of the volumes or the weights of the bodies.
(2) Bodies of given shape or form, in order to undcr.~o
a definite given subdivision or change in sha,
require an expenditure ofwork equal to the prod t
of the weight and a quantity which is the quantit y
of work necessary for unit weight of the same sh ;
to undergo the same deformation or subdivision.
(3) The forces necessary for corresponding changcs
shape of two geometrically similar bodies of ti
same material are in the ratio of the correspondi
crosssections or the surfaces of the bodies.
Actually the second statement has precisely the sa
information content as the first but is stated in an altern:
ti.ve form.
btained experimental
vefication
119
of this law
" 11 s s the crushing of simple bj ects and, by
'\11 i the law ta size reduction generally, was appar
' tl
. l>l c to refute*the law ofRittinger.
l{e case or
ittinger's law, an expression for the
11 i ut in order to bring about any specified reduction
11 11 y be deduced. Thus, Kick's law states that the
ssary to reduce a body ofunit size to a number of
1 l
I' t s of, say, half the size is constant. Each of these
t ls, however, requires only oneeighth (!3) of the
11 1 f the original body to reduce their size to onehalf,
lt , . h re are now 8 (23) ofthem, the energy to reduce
.. 3
.
~
..:;:;;;=""
2 1+~.r
::>
~~ I ~~~+~~
.,c
.....
2
Rducion
Raio
FIG. 4.3
i H it to onehalf their size is equal to that required to
1i : he original particle to particles of onehalf of the
11 ' the oginal body. Thus, to reduce unit mass of a
t al to a given fraction ofthe ori inal size re uires the
, l.ture of a constant amount of energy; regardless of
tl11 i al size of 12article.
l ' l i s can be illustrated graphically as in Fig. 4.3.
' . y, if the reduction ratio R is defined as d0 fd 1 , when
l t lerial has been reduced to onehalf the original
I = 2 and one unit of energy has been supplied.
l <' tle size to onehalf again changes the reduction
f'rom 2 to 4, but again one unit of energy is required;
I for further subdivision.
120
BALL, TUBE AND ROD MILLS
COMMINUTION OF SOLID BODIES
The relationship between W and R for such a case i.
clearly
W = KlogR
.... (4.5)
but this relationship can be obtained formally a3 follows:
I I f'orcgoing analysis brings out a number of interest 11 111 i ts in connection with the fields of applicability of
l1 wo laws. In the first place Kick's law states that the
111 ,y bring about a specified change in the size of a
1 11 1 1rss of mateal is constant and, furthermore, since
1l1 I 'r energy to induce a certain critical stress in a
1111 : ss of material is also constant, it would ap ear that
l 1 s involved in breaka e in accordance wt
c s
I 1 . c boa orces an furthermore, after fracture the
I fl.Y~loula appear as heat. us,
ick's law would be
1 l to ho1d or particles for which the surface energy
111 ~, li ible; that is for large particles during coarse
11 li g.
i rse1y, Rittinger's law, which states that the energy
I 111 1'(d for fracture is proportionaJ tO the lleW SUrface
' < I , would be expected to hold for very fine grinding
1 , 11 this case, the surface is large and the surface
\ requirements could swamp the elastic energy
11 111 cments. In fact there is evidence that, in practice,
, : rushing is roughly in agreement with Kick's law,
11 ''ftS fine grinding is more nearly in accord with
I g r's law. At this point the implication of these two
I '"' relation to the physical process of crushing will be
. red.
, ka}(9) has pointed out that although Rittinger's law
1 " ~:nts a logical interpretation of the process of frma
i new surface, the manner in which the process is
. t d is disregarded. ln technical size reduction, the
1 i ; l s are crushed by impact, pressure or abrasion. The
111 . energy of the resulting newly formed surface is
11 c l at the expense of the elastic deformation energy of
dy; the balance of the energy reappearing as sound
1111 l cat . Smekal has also pointed out that particles ca
I> considered as being ofhomogeneous structure, and
11 1 g( ) has shown clearly how the influence ofthe micro ture affects the validity of the laws of crushing. The
race of inhomogeneity in the particle is shown by
l nc t that it is generally accepted that the calculated
(/) dW =
Increase dR in the reduction ratio
expressed as a fraction of the
initial value of R
= (R+dRR)/R
= dR/R
or, on integration,
W = Klog R
Spqclic
F.
.... (4.())
Surlacc .
4.4
Alternatively, since R=d0 fd1 or R=S0 fS 1 , equation (4 ./)
may be written:
W = log (d0 /d 1 )
. (4.6:)
or
W = log (S1 /S0 )
... (4 . 6 \ )
The essential difference between the law of Rittinger a11
that of Kick is that according to Rittinger the energy c
quired to crush a particle depends upon the initial si~.c
of the particle, whereas the law of Kick states that tlr
energy required depends only upon the reduction ratio,
The relationships b etween energy input and total surfa r,
for various values of initial fineness, according to the law
of Rittinger and Kick are shown in Fig. 4.4.
121
122
BALL, TUBE AND ROD MILLS
chemical bond stren th of a material is of the order /
1 times t e actua mechanica strengt tl
matena . urtfi'e'rmore its generally 'Delieved 1:t
 e aiscr~pancy is due to points of inhomogeneity i
the crystal structure; the elastic energy for fracture b ci
concentrated at these points due to notch effects.
t<ND~r vov It is implicif in the Rittinger law of crush~ng that ~~~~
particles are completely homogeneous, and 1n the
law of crushing th ;J;,t the.2articles are either lomogeneo M,
or that the In ternal structure of externally geomefncaHy
s1milar part1c es is also geometrica ly simi1ar; that is, 1:
a!strioufion of faults in the internal Structure IS lllC ('
penaent f e volume of the P.artic e. Tli1s latter assuml!
'ton IS c1early, in a reai particle, of doubtful val1Ity:.
tl
the laws of crusiig discussed therefore contain assum
tions implicit in their statement which recent th~oreti : l
and experimental work on the fracture of part1cles ] :
shown to be of doubtful validity. ln fact, the process /
size reduction has been shown by Smekal, (9) Honig, ( )
Axelson and Piret, (11) and others to be so complex that i
existence of any general rigid law of crushing is doubtfl ,
Nevertheless, these laws have been established over tl r
last fifty years, as being useful tools for the predicting of
the performance of sizereduction machinery.
.
.
Within the last few years, a new theory of comminuto
has been developed by Bond. <17 > It is based on the analyNi~
of actual sizereduction machinery rather than 011
theoretical consideration of fracture. According to tl1i ~
the,ory, termed by Bond the "Third Theory of Co~mi~
tion", the effective work done in crushing and gnnd111 ~
resalts primarily in the formation of crack tips and it iH
directly proportional to the total crack length for~ed.
equidimensioned particles the total crack length IS stat<l
to be equivalent to the square root of onehalf the surfa<r
area ofthe particle. The total work input represented by
given weight of a crushed or ground product is t~erel r
inversely proportional to the square root of the diamcn
of the particles of the product.
COMMINUTION OF SOLID BODIES
Jj!
123
represents the work required to reduce, from a
F microns to a product with 80%
microns a unit weight of material, then
f, l w il 80 % passing
I' ,_ "' "
(4.7)
W . is the "work index" or the kilowatthours per
q u'i red to reduce from a theoretically infinite
I ' 1 I i l size
: : t tion
11 l
J'
to 80 % passing 100 microns.
(4. 7) can be transposed so that when the work
W; is known the work input W required by any feed
and product can be found. Thus,
W= W;
y'F  y'P 100
y'F p
(4.8)
11 represents the reduction ratio, or F J, then
y'F
y'R
y'Fy'P = \IR1
W;
y'R
y'P
= vR 1
.... (4.9)
' mesh size in microns, of the sieve through which
/11) ' 1 ~, th e mat~rial passes is selected as the criterion of
\ :nd product sizes since this is a figure which is readily
tt lt ible by plotting the results ofscreen analyses. lt IS
l , :cl to be more reproducible than the size of mesh
l !l which 100 % of the sample passes, but ot~er
<rs use the sieve which passes 100 % ofthe sample w1th
cess.
l1 as been pointed out by Walker, Lewis, McAdams
l
illil and<3) and more recently by Coulson and
l : rdson (14) that the various laws of comminution result
rgy relationships which can be derived from the
fll ' ~sion
dWJdd = Cd"
.... (4.10)
124
COMMINUTION OF SOLID BODIES
B A LL , T UBE AN D ROD MILLS
which states that the change in diameter, and hen 11 1
specific surface, brought about by a unit of energy, iH .
function of the particle diameter.
If the index n is put equal to 2, and the equaio
integrated between the limits d0 and d1, tle expression
W = C(l/d1 l /d0 )
. . . . (4. I I)
is obtained; this expression being identical with equ al i 1
(4.4), and so is Rittinger's law.
Kick's law may be derived by putting n equal to  J
equation (4.10) and integrating between the same lini
as above, whereupon the expression .
. . . . (4.1 J :)
which is identical with equation (4.6a), is obtained.
In the same way, insertion of n = 3/2 gives
dWfdd =  Cd31 2
which when integrated between the same limits as
viously, gives
. . . . (4. 1 '~)
or
2C d0 l /2  d1 I /2
.... (4.] : )
dol /2
W = d/2 .
Bond's law thus represents a condition between tl1o
corresponding to the laws of Kick and Rittinger in n l i
tion to the energy n ecessary for crushing through a g
reduction ratio.
It has been pointed out by Dobie< 15) that if the en c l(
WR necessary to crush a material through a redu
ratio R is written in the form.
. . . . (4. 1:i)
where d0 is the initial size of the material, then the
Rittinger and Kick are contained therein.
law~ l
125
' s,
WR = bfd0
(Rittinger's law)
WR =
(Kick's law)
l nnore
WR = bfy'd0
(Bond's law)
there is a reasonable amount of evidence that
r's law applies moderately well fo_r fine grinding
1l i e Bond's law is deduced from the results of
us test on real mills, it appears that, for fine grind , i is necessary to have a law in which the work input,
11 1 , bring about a given change in the reduction ratio,
l lcl be a function of the fineness of the material and,
I mre, WR should increase with increasing fineness.
I li H basis Carey and Stairmand(l6) have suggested that,
I 11 ' l 1' tO proide a criterion for the performance of
II l{ machinery, a crushing coefficient which increases
i lineness and is related to the size distribution of the
I' i l s produced by the crushing is required. In this case
l ushing coefficient is d efined as the energy cor" 1 dg to a given reduction ratio, and so co.;esponds
11' 11 bove.
ol'der to obtain such a "crushing coefficient", Carey
l Kait'mand made use of the similarity in frequency
I i l> tions of crushed products and so developed the
111 ile of "associated energy".
' characteristic grading produced by free crushing is
I 1111' l a "natural" for the substance, and it was found
I l il' the largest particle of a natural be crushed and
ll l to the remainder of the distribution, the resulting
li is again a "natural". By summating the energy
I'' i d to crush from one natural grading to another
. ll r one, it is possible, by laboratory experiments, to
~l, the energy required for free crushing between
11 i ;l distributions of given maxima sizes. Thus, for
lc, ifitis assumed that a "natural" with a maximum
ol' 1 metre represents zero associated energy, and it is
lI hat the energy required for successive crushings to
: ~
126
COMMINUJON OF SOLID BODIES
BALL, TUBE AND ROD MILLS
produce a "natura1 " ofmaxim umsize 1 cmis 05 kWh /t ,
and that a further addition of 02 kWhfton produ <'
"natura1 " wi1h a maximu m size of 1 mm; then t~e e?<',~
associated with a natural of 1 cm maxmum sze IS
kWhfton and the energy associated with a natura1 of 1
maximu m size is 7 k Whfton.
In order to derive va1ues for the associated energy,
crushing was effected by crushing between horizot l
p1atens with a 1imited reductio n ratio; a 1oad versus d_ l
tion curve being used in order to compute the energy 111JHI
to the crusher. By use of the figures so obtained for tl
associated energy, a figure for the mechani cal efficien y cl
any grinding process may be obtained from the relatio
ship.
Mechani cal efficiency of grinding
Energy associated with product
 Energy associated with
Energy used by the mill
c ~ cc I
In the absence of estab1ished laws of comminu tion, tl
methods of Bond and of Carey and Stairman d provicl 1
means of deriving a _crit~rion of performa nce for m~Jii ~
machine ry. The denvatn of values for the assoc:t cl
energy or work index is tedious and the use of siC'V
analyses for the size frequenc y determin ations limits tl
fineness of grinding to which the method can be appli l ,
but, nevertheless, these methods appear to form us/l
bases for the design of crushing machine ry.
Turning now to the efforts which have been mad tc
establish the validity of the varios laws of crushing
experim ent, it must be S!ated at O~Ce t!at much of I J.
work must be treated w1th reserve. ,The reason for tl
caution is that the ieasurement of both the energy inJHt
to the crushing system and of the fineness characteristic
of the product are problem s of the utmost difficulty.
Conside r first the difficulties involved in the determi 1
tion of the fineness of the crushed material . The us
127
extrapol ation ofthe grading curves into the
and, since very high specific surface is asso i Lcd with the smaller sizes, small errors in the extrapol a1 will give rise to large errors in the calculate d surface
this
111 he product. Theoreti cal difficulties also arise in
surface
external
ifthe
even
,
example
for
Thus,
nection.
1
ld be determin ed with absolute accuracy , the existence
111 internal surface within a particle of any real substanc e
ds, to some extent, to reduce the particle to an aggret ofsmalle r particles . Thus it would appear that the surI which should be measure d is the external surface lus
l surface of an crac s wl1c are su ae~ w e fo
/ olies1ve forces, between the two boun aries of the
.
1 . k, to be ~ffectively zero.
t first sight it would appear that a better approxma
i to this true surface would be btained by the use
/ Lhe adsorpti on method of surface measure ment. UnI l nately, however , the theory of the adsorpti on method
ill in an incompl ete state and in conseque nce no more
l i :_nce may be placed upon the adsorpti on method, as an
11c 1 rate measure of surface, than any other. Even if this
1 not the case, however , there still remains the guestion ,
l wle ther the adsorbed molecules wed e en narrow 'IF .. ..,.1.::.
~SCSrsurface ,W.:~
I I c: ks and 50 whether ad sorptlOll ffiCt oas
1 i: s not strictly free surface in the present sense. Thus,
11 us t be admitted that no known methods exist, for the
l 1 : ' mination of the sze or ~ecific surface of the smaller
l ns, which are free from criticism.
'c method of solution of tne particles , adopted by
1\ 1 (l7) and by Gross and Zimmerly<s) is also open t_o
11 i i sm since, for example , it is known that a crystal 1s
lved at different rates faces parallel to and per1'' l ular to the lattice planes. Hence, a figure for the
c area based on the average rate ofsolutio n, without
i t ain k~OWledge that the particleS are geometri cally
l l : r, is unreliab le.
supplied
' rning now to the measure ment ofthe ener
ted that
apprecia
e
once
at
I crus ng process. t wi
cs involves
l >sieve range
128
129
BALL, TUBE AND ROD MILLS
I ' COMMINUiON OF SOLID BODIES
the energy to which reference is made in the equation~ 1l
comminution is t e energy usefully expended in the fr: <
ton of the partic e and not tlat which is in the crushi B
system as a wfiole. onsiaer, for examp1e, the measu ' '
ment ofthe energy transferred to a single partic1e by a dc
weight. The initial potential energy of the weight may
determined with considerable accuracy and, by the us~ cl
modern laboratory methods, the height of rebound of I '
hammer may also be determined. Thus, the energy givcn
by the weight may be determined with adequate accura<' .
The energy given up by the hammer is, however, in J
converted to strain energy, and thence to heat, in part c l
to increase the surface energy of the powder, in par
give kinetic energy to some ofthe particle, in part to scl
energy and in part to "pelletiz~ some of ~he prodc. 11 ,
however, this heat energy coul<be deteruned, then, rtN
first approximation, the energy used to fracture l1
particle could be regarded as the difference between
energy given up by the weight and the heat ene~gy .' c
jected. It is easily shown, however, th~t the determii~a~o
of this quantity of heat is extremely d1fficult. Thus, 1f
assumed that the hammer weighs 10 lb and falls throu.~l ,
say, 3 ft, and also that the metal masses (ham:er, anvl ,
etc.) which will be heated weigh 20 lb, then a s1mplc ::l
culation shows that, if all the potential energy were
verted to heat, the average temperature rise wold lc
about 005. Thus it is almost impossib1e to make a l
balance with worthwhile accuracy. For this reason, lc
analysis of the results of crushing tests are usually b~ l
the assumption that the whole of the energy lost by ,
hammer is expended in fracture, but such an assumpt10
clearly of doubtful validity.
further complication is that it is possible tha
geometry of the syst~m and the rate. of loadiD:g of I
particle have, for a g1ven energy dens1ty, some flu c c
on the fineness of the product obtained .
With the drophammer apparatus, the energy, althol(l
available in the hammer, is apparently not effectivc~ ly
lc'l't' d to the particle. Thus it would appear that the
, y ransfer process is, or can be, a controlling factor.
ll i N vi w is supported to some extent by the statement
I 11.111 , ) that the Hadsel mill, in which ore is lifted and
I 11 11 I I' pped to steel plates, has a power consumption
I 11 I' kWhjton which "compares favourably with ball 11 1 onsumption for feed of i in."
'
40
30
20
10
Nft En;:rqy lnut. Kqm .Cm./Gn.
F.
50.
4.5
' : ts upon quartz, by use of a drop hammer, carried
II l>y Gross and Zimmerly<IS) show a linear relationship
l11 w ' Cn the new surface produced and the energy input
I ' ( on the energy lost by the hammer. These resu1ts are
l ttwn in Fig. 4.5.
wong and Piret, (20)
using a dropweight apparatus
i l ar to that of Gross and Zimmerly, also carried out
ling tests, bu~ in this case the air permeability method

130
~
   
BALL, TUBE AND ROD MILLS
COMMINUTION OF SOLID BODIES
was used for the determination of the specific surfa1 .
These workers also found a constant relationship betw<
work input and surface produced for glass, fluorite a l
calcite, Fig. 4.6a, but, for salt, which is a softer mineral, .
l l spectrum" to explain similar results obtained for
I i rding of sa1 t in a ball mill. These resul ts are also
131
ord ance with the work of Smekal(9) mentioned
I'' \ i o s !y.
o der to carry these investigations into still finer size
, Kwong and Piret< 21 > made use of an adsorption
~400~~~~74+~L_i
u
..
'' e for specific surface measurement, with ethane as
l . rbent. For the harder minerals, they found an
. in the energy input per unit increase in surface
111 increasing fineness (Fig. 4. 7) but claimed that
aled no discrepancy with their previous work
u
~
~200~~~~~+i
2ooorr+~L~
40
20
Nct Enrqy nput
CIO
Kgm.Cm.j Cm.
~~~+4
"'"
80
&
300
200
Energy Jnput Kgm.Cm./CCm.
..
FIG.
~250~~~4+i
"'.."
~
8 
Natural .
0 
Synotic.
40
20
t Enerqy lnput
CIO
Kgm .Cm./Gm.
80
b.
F G .
4. 6
nonlinear relationship (Fig. 4.6b) was found. This rcl
indicates a decrease in the amount of surface formecl, IH I
unit input of energy, for the larger values of specific
face. Plastic deformation of the sa1t crystals was suggcslccl
as the reason for this. The work ofHittig is ofinter st i
this connection, since he suggests the presence of'
4.7
.~ r meability techniques, since the results were e
l < l into much finer size ranges. This work also shows
l I ' r the size range studied, the ratio of area by ad i to the area by permeability is fairly constant at
i I 1920. Work was a1so carried out on the slow
ssion crushing of particles, and in this case, the
I input was measured by computing the area under
l r e obtained by plotting the displacement of the
111 ling surfaces against the applied load. The results
' ~.8) show clearly that the ratio of new surface to
y is greater for slow compression crushing than for
. crushing. This deviation between the results was
1i , in
d tentatively by changes in the usage of energy in
1 1 . d drops of the ball of the impact crusher.
132
COMMINUTIO N OF SOLID BODIES
BALL, TUBE AND ROD MILLS
If it is assurp.ed that the average stress concentratio 11
is a function ofthe energy input for each l1 'J
the
all, and that a critical stress concentrati 11
ne~essary for fracture to :;c, then initialiy, lll the c!'(
we1ght crusher, the stress concentration in the bed is 11 .~ l1 ,
and therefore the energy is used efficiently. However, wii
~essive drops the effective energy input per drop l
creases by 4050 0 due to the frictionallosses between I
par IC es, an so t e average stress concentration, and ('
sequently the num  er ot fractures, decreases.
th._material
3000r . 
was caused by high local stress concentrations at
" I l y ew points; the e~astic deformation energy, which
~ ; s
mostly as heat, IS thus small compared with the
l ?duced. When fracture occurs at high energy
""'
tats,
however, fracture starts at a number of
high level of en_exg;;
I I, andsince t~e fracture at most points will
/ : . only a small _d1stance before meeting a further
111 l rI'~Cture a relatively high..:eerce!ili}g e of the elastic
'y_ w1ll rea pear as heat. The effiCiency ol the process
~. ;~ nce more .:ents are at a
~00
"'
"''u
~2000~17~~
,;..
..
Snglr Crytl .
Multiplc Croin.
;;:
. I
t'"200
"'"
.
..
::>
~ 1000~or~~~~
n. 100
~
11'1
"'~ .
.
)<
100
Enrrgy lnput Kgm .Cm./n.
200
200
400
600
Enrrgy Drnity Kgm.Cm./Gm.
FIG. 4.8
FIG. 4.9
The work of Axelson and Piret(20) is of particular in t
in the study of crushing, since these workers have inv l
gated the slow crushing of single polished and unpo liHi l
quartz crystals . By us e of a technique for energy
measurement similar to that described above, it was ((l
that the necessary stress for fracture lay between 9780 : l
500 kg/cm 2 , the natural crystals tending to have thc : {
stress for fracture . It was observed that although th:
face created is greater the greater the energy concentt':ti
(kg cm fg), the surface createa _eer unit energ!J.'inf.J.ut de
or Increas1ng energy concentraton, as Shown in Fi . f
e lypoth esis used in order to explain these results is
when fracture occurred at low energy concentrations,
133
134
BALL, TUBE AND ROD MILLS
linear relationships were found to exist betwee!l cn )'
input and new surface area for Cumberland Baryl 'H,
North Wales limestone and Durham anhydrite.
adsorption technique, with krypton gas as the aciHo
bate, was also used to measure the total surface, b ""
evidence of a constant ratio between the surfac ' I\
adsorption and permeability was found; in fact the r: i
was found to decrease with increasing fineness .
The ~uestion of elleting_, of the roduct mentio l
~ier, is of some iterest since t e researclies of w1
a ~> sho.;t:hat impacted "sr.faces may bc
Sfiongiy wel eC'tO'g'Ctner. I'i: s eviaent that i this' h.~l)
t ere will e a consideraple ,ut surtace with
quent increase in particle size. ..
Reseafclies oy :M"acenz!e and Milne<23) show tha l
Xray diffraction pattern of muscovitep ecomes mor lil
:fi'Se as grinding proceeds until, after 8 hours, no s l ;
pattern remains. After further grinding the pattern
appears but has a different form to the original. This r l
suggests the possibility of recrystallizati on of tl
matenals an , a1thou gh the evidence is not conc1usiv , t
s suggestive that the same ,2 enomenon mi ht occu i
the fine rnilling ot other materia s.
In fne wor whcfi fias 6een ds cussed, in this chapt , t
has been assumed that there is a similarity in the crusli
attern of materials comminuted by slow crushing antl 1
blows of. various }!l<;gnitudes.
owever, none of t1
workers appears to have attempted to establish e i
mentally the truth of the assum ption. This raiscs
following question: If a given input of energy to a part I
produces a given increase in surface and assuming, ''
example, that an input of twice the energy produces twi
the surface increase, then are the size frequency curvcN
the two products similar, with the scale of particlc 11i1.
changed, or is there no relationship between the produ : t ,
other than a twotoone ratio of the increase in finencNH '
This question has been studied by
en (24) J1o
ever, who has s~q~:,. tl.le.t the cumulative curves of tl
COMMINUTIO N OF SOLID BODIES
I'' l
135
s obtaif..l, by g,ushi!.lgM,<,L~~callysirnilar bodies
' ' l1ogonally affiie, ~ aa"Lnence t fi5l ows a :
'
( I) The size distributions for the materials are geometrically similar.
(' .) :rhe specific surfaces. of a series of products vary
nversely as the absCissae of the cumulative curve
r any specified value of the ordinate.
'
rie~ of product curves for various porcelain cubes,
l t rmned by Andreason, are given in Fig. 4.10, and it
oor,,rr,
80
..
....... 60
...,
c
:;,
.,.
..
c 40
...u...
20
Pr ~ ssurq
lmact
Crushinq.
Crushinq.
40
Grain Slzq Mm.
FIG. 4.10
60
136
BALL, TUBE AND ROD MILLS
is seen at once that, for all practical purposes, thesc :
orthogonally affine. These curves are also representa t
of curves obtained by tests on feldspar cubes, gl : H
cylinders, etc., and so it may be stated that there is s
experimental proof that geometrical similarity holds
the fracture pattern. The present authors, Rose ;I
Sullivan, have, in the experiments on the crushing of r
spheres, also found that the product curves are, roug ll ,
orthogonally affine and, in this respect, confirm the w l
of Andreasen.
I
I
_ 
ldealized.
Real .
Fracture Patterns According to Andreasen.
Fa . 4 . 11
Andreasen has also put forward a logical explan a tio
for the existence of such similarity. Thus, in Fig. 4.1 J i
shown the pattern obtained by the actual fracture ' 1
cube and in Fig. 4.11 b, an idealized fracture pattern wlti
can be deduced from it. Ifit is assumed that the prin i I
plane of failure is parallel to one of the diagonal pl a 1
then it follows that the grain size of unity comprises 50 ,
by volume of the material, the grain size of! comp i N '
25 % , of t comprises 12! % and so . t is also clear t l1 11
the weight cumu1ative curve for such a product j M 1
straight 1ine passing through the origin. Thus, t l~
appears to be a fundamental r eason for the type of
obtained by Andreasen. His work is also interesting in tl.l
the treatment leads, at once, to Kick's law, and Andre N rll
( COMMINUTION OF SOLID BODIES
1 1l . i1 th at the c1oser a
137
roximation to Rittin er's law
lt1 } obtanea<:q.r,~ ne ,g:n 1~ anses rom a oss
1! 11 1 ncy of th e rmll wfl en _ri~In hner ~artc es.
\ interesting pnt wnic emerges f~ e anaiYs'i'; Qr
\ I s n's dropweight tests is that, for the larger cubes,
l l{ rcement between the product size obtained for
1 1n .rgy input, by slow crushing in a press and by drop
i\ t, s very close. For the smaller cubes however the
11 I t size from slow crushing is in very close agree~en t
1 l1 lat ca1culate~ from the s1ow crushing of the large
1i ' nd hence wth that obtained by impact crushing
l I : large cubes, but the product obtained by the impact
11 I i g of t~e small cubes is far too coarse, although the
1 ' are stll orthogonally affine. Since Andreasen used
<l height of drop for the hammer and hence a fixed
l y . at imp.act, this result suM;est~ that the crushing
111 1 ' :ss ~s veloaty de endent an'Ctt~mi1arity
11 ~hg t s necessary or t e sze an ot er cliaracter1 '. of the arti},e , to ,lie in a de,ti t: .,:e,l;:t.. to the
I ' l ofthe hammer.
ll fair success 'df researches of Martin (25) Rosin
111 nl er and Sperling, < 2 J> and Mller(27) andthers int~
l ' 1Jossible existence of a general law of size distrib~tion
l 11 wdered materials a1so suggests the existence of a
l : metal mechanism of comminution but "laws" of
1:~ '< l appli~ation do not, at the present' time, exist.
ll comp1cated process of fracture and the lack of
l10gcneity in the structure ofthe majority ofmaterials
I11 giund, particul~rly n.inerals, makes the develop of a genera1 relatnship between size distribution
, duction and energy input extremely difficu1t. Thu~
l t .~ocess of deducing energy relationships is largely
cal and r elated to particular conditions and types
l luction machinery.
' I'11 pres en~ po~ition regarding the investigation of the
I \V.H commnutn car; probably be fairly summed up
l ll ws. The problem nvolves extreme difficulty in the
I l ;mental measurements. In addition, apart from any
138
COMMINUTION OF SOLID BODIES
BALL, TUBE AND ROD MILLS
questions of the accuracy of the actual measurem ,
much of the existing work must be treated with res v
since it is generally impossible to differentiate betw
energy put in to the grinding system and that usefll
expended in comminution. Many of the published resl
are further confused by the crushing of deep beds l
particles; th,::re . b~ing po ~thereticall sound reason w
the crushing ofthe particles in a bed should obey the s11
ruf;,~ a,s ,tJi~~ <;rusfi[fOFasz:~<;..,Bar IC e.
ln gener:i, questns ansrng rrom imperfect natun ,,,
individual crystals and the inhomogeneous nature ofrn111~'
industrially important materials have not received
attention which the subject merits. The truth [ l
assumption of geometrical similarity between a parti l
and its progeny has not been adequately investigated. 11
laws of Rittinger and of Kick are possibly extremes,
tween which most ,practical ideal crushing processcs
It is scarcely possible at present, however, to statc l
conditions for one or other of these laws to be applical>l ,
nor is it possible to state, in any particular case, the law 111
which the actual grinding operation will approxin:l
most closely. It appears fairly safe to say, however, l.l
coarse jrinding ~JL.<:I?.E.~?i i,R~,.EJ.e closel to } i k '
law an fine fr~ rn~more near ~.19.. !{\~~i~~r) l_<~;w.
Tli.e extent to whrc the desrgn of a mifl, and. l
physical dimensions of the crushing elements influen l
crushing characteristics of a given material is unccrt:i ,
It ap_eears that1 in general, the efficiency of transf: 11l
energy to tle articrerans wrt
ecreasrn
ar rc c :.
an t rs actor a one ma es t e overa
e avrour
tl'.l?es n ar'ox'imafe to '1he"..~itlin~er" tyl w
even tliought e find ai.e'at......c'ising process m rg
more nearly in accordance with a law of the "Kick" ty) .
Finally, it may be said that the subject of cmmi
offers a wide field for useful, interesting and fundam t l
investigation.
139
REFERENCES
'
'
,
c, G . F. DechemaMonogr., 1952, 21, 245 and 96.
' .,
and GEIGER, . Sulzer tech. R ev., 1956, 38, 14 .
, F . D . Proc. phys. Soc. Lond. , 1949, 62, 418.
., D. . Proc. Conf. UltraFine Structure of Coal, 1943.
w.,
JMPHREYS0WEN, s. F. . Proc. Phys. Soc. Lond., 1955, 68, 325.
(s, . . Unpub1ish~d research, but see J . app. Chem., 1957,7, 244.
VoN RGR, . . Lehrb. d. Aujbereitungskunde (Ber1in, 1867).
CJK, F. Das Gesetz d. proportionalen Widerstinde u. seine Anwendung
(Leipzig, 1876), p. 14.
11 ~ L, . . Ver. dtsch. lng., 1937, Beiheft Verfahrenstechnik, Nr. 5.
Sce a1so: Ost. lngArch., 1953, 7, 49 /70; Glastech. Ber., 1950, 23,
"7/67; A cta phys. austr. , 1953, 7, 110 / 122.
111 ll.INIG, F. Forschungsl. Ver. dtsch. lng ., Heft 378, 1936.
I LS, J. w ., and R, . L . lndustr. Engng Chem., 1950,42, 665.
I 1 D, F . C. Chem Eng., 1952 , 59, 242.
WLER, W. ., LEws, W. . , McADAMS, W. ., and G...D,
1\. . Principles ofChemical Engineering (New York). 1937.
11 ( : ou.soN , j. ., and RcHARDSON,j. F. Chemical Engineering (London),
Vol. 22, p. 665.
' >, W. . Recent Developments in Mineral Dressing, 1952.
( 1 , W. F ., and SAIRMAND, C . .J. Recent Developments in Mineral
Dressing, 1952.
11 R", G. J. $oc . chem. Ind. , Lond., 1926, 45, Part 4, 160.
( :ss, J., and ZIMMERLEY, S. R. Bull. U.S. Bur. Mines, 1938, 402.
1'1 /\ . ., R. G . Trans. Amer. Inst. min. (metall. ) Engrs, 1935, 112, 20.
11 x,LSON, J . w., ADAMS, J. ., jOHNSON, J. F., KwONG, J. . s., and
> , . L.
Trans. Amer. Inst. min. (metall.) Engrs, 1951, 190, 1063.
' Rs, G. .
Trans. Amer. lnst . min. (metall. ) Engrs, 1953, 63, 211.
wD , F. . , and TABOR, D. Friction and Lubrication of Solids, Camb.
l] niv. Press.
', R. C., and ., . . Miner. Mag., 1953, 30, 178.
N JREAS,!!Nl . . . Trans . lnstn chem. Engrs, Lond., 1957 35, 104.
'', <J. Tra!U. ~~ram. 3'oc.,"'f92:f,' 2 ,
( , R., RAMMLER, ., and SPERLING, .
Krngrssenprobleme d.
Kollenstaubes u. ihre Bedeutung J. d. Vermahlung. Berlin, 1953. (V.D. I.
V lag . Bericht C52).
..R, R. Sprechsaal, 1935, 68, 61315, 62731.
PROCESS OF GRINDING WITHIN MILL
141
l d be analysed and an effrt made to establish the
t nt to which the performance of a mill may be pre' I d from theoretical considerations. The experimental
Iy of the performance of mills has, in general, been
l ~ the following lines:
CHAPTER 5
PROCESS OF GRINDING WITHI
MILL
IN the previous chapter, the various "laws" of cmm
tion which have been propounded have been studi d
rela.tion to the physical process of size reduction and to 11
avalable results of experiments into the fracture of h11
geneous and ~eterogeneous materials by impact ancl
slow compressn. ln this study it is seen that no sal
factory "law" of comminution has yet been deduced 11 t
tha~ a number of m~re or less empirical expressions : .
avalable for the gUidance of the designer of crus hi ~
machinery.
The presen~ c~apter. wl be devoted to a study of l
process of gnndng wlthn a real mill. Since within 1
"tumbling" type of mill the size reduction can result f 11
dire~t cr~shing, from shearing or abrasion between 111
partc!es ~he feed, between the particles in the feed ncl
the mll bo~es or between the particles and the milllini 11 1\,
t~e process s much more complicated than in the cas:
smple ."[ree" cr~shing c~ndi~ions considered previously.
In addtn to ~hs c.omplcatn, however, the moti f
the ball charge tself s extremely complex. Thus the pr i
mode of.transfer ofthe energy to the crushing surfaces :
not be smply represented.
For these. reasons, . a rigid theoretical analysis I
para~eters nvolved the milling process becomes
pract~cable, a?d recourse is frequently made to pilot sc: l
expe~~ents order to determine optimum milli l\
condt?ns. evertheless, considerable experimental
theoretcal work has been done on the crushing proc s i
the "tumbling" mill, and it is desirable that these c\t
140
( I) T he enegy relationships and theoretical efficiency
of the ball mill from surface energy and thermodynamic considerations.
(' ) The size distribution of the product in relation to
the mill parameters.
(3 ) The rate of crushing in relation to the mill parameters.
Lual.ly these three divisions are not independent since,
I ' i ll 'ly, f there is a definite relationship between the
y input to a particle and the change of surface area,
luction ratio, it follows that a given mill, to which
I <~nergy input is constant, willlead to a definite rate of
11 ~ ling ; the rate being a function of the relevant mill
' : meters. ln fact the inease in the surface area of the
, c~i~l, prod_uced by the in pu t of a uni t of energy to the
ll , s ncreasngly accepted as a criterion of performance
I ' e milling operations. Without doubt this is due to the
i nological importance of specific surface in the subM i range, and to the establishment of reasonably satisl t ry techniques for specific surface measurements. It is
I l1is reason that Rittinger's law, even though lacking a
n d theoretical basis, is widely accepted as a basis for
l alculation of the performance of such machines.
. ~ t us aE l these ideas to the derivation of an eguation
1
I ' Ll1e rate of p;rinding in a mill. lt follows that if,
~ars
as
reasonaole, tne cnange of SJ)eCific surface S
1 , 2f m 3 , within a mill d epends uEon tl1e dameter oftne
, upon the ail diameter and uEon the mean diameter
f' ' h arti cle, D:aandb ~spectivel>J. upon the densty of
I ' bal s and the ulE, and respectively on gravity, g,
" " bc speedofthe mill, ; upon u tfie diameter ofthe exit
the mill, upon , the energy necessary to change the
142
BALL, TUBE AND ROD MILLS
' I{OCESS OF G RINDING WITHIN MILL
surface ofthe material by unit amount and upon the pa : t
meters J, U, V and , where J is the fraction of the n1ill
volume occupied by the balls (or rods) and the spac
tween them, V the volume occupied by the powder (
pulp) charge, including oids expressed as a fraction [ ll11
volume between the balls (or rods), Uis the ratio ofsolil
volume to total volume of the pulp, and is the ratin l
circulating load to the total load when closed cir
operation is considered, and clearly, S will d epend on I
time of milling, t.
Then we may write symbolically :
S = (D, d, b, , ,
g,
, ,
t, J , u, U, V,
143
1 \ proceeding wi'th such an analysis of experimenta1
l ' w ver, a digression will be made to intro_duce
11 y theoretical treatment of the internal dynamcs of
l11 : 11 mill propounded by one of the present authors,
, . (I) This, as far as the writers know, is the on1y
to deduce the general behaviour, including such
; :s ball wear, fro m purely theoretica1 considerations.
l~ more this theory can best be studied in conjunc
ll
wi h th; analy:;is of the experimenta1 resu1ts which is
.... (r.. l )
where, as before, d enotes "a function of".
The application of the principles of dimensio r : l
analysis, already illustrated in Chapter 3, leads to l/11
following equation.
(S .D) =
{ (Di), (i), (i), ) , (;) , (n~2)'
' }
(Dgt2) '(J), (U), (V), ()
Fc.
which, as before, may be written
(Di) 2(i) s (i) 4(n:Vz) 5(~ )
2
(S.D) =
6). 1 (J). s( U). 9 ( V). (). (~)
.... (r:.)
This equation (5.2 ), is the dimensipnless equation gi v i
the rate of production of specific surface in a tumbli H
mill.
Purely theoretical considerations are insufficient to t;k
the ana1ysis further, but, in pr.actice, the various dim<
sionless relationships may be established by referen c t
experimental data.
sary
: .
5.1
for the evaluation of the above dimensionless
This theoretical treatment is of necessity e
n ly tentative since the dynamics of a mill are of ~uc~
l xity as to defy rigid analysis. Nevertheless, t s
l li vcd to form the basis of a more precise treatment.
'h basic idea underlying this treatment is that the
~ l o l behaviour of a ball mill depends on the probability
l . particle arriving into a "zone of operation" between
1~ /)alls, and on the probability of occurrence of some
I'' < ss when it arrives there. Outside this zone, no pera
, such as crushing or metal wear, which demands the
. g of a partic1e between two balls can take p1ac.e.
' m Fig. 5.1 it follows that the zone of operatn s a
i ular area of radius , where is given by
144
ROCESS
BALL, TUBE AND ROD MILLS
2 + )
= (
d; b)
it .
= /(dbf2J
(5 .3)
provided b is small compared with d; d and b being t]H
diameter of the ball and of the particle respectively.
Now the maximum number ofparticle s, Nmax> which ca
be in the operating zone are those which just form a rin ~
around the contact point. *
MILL
145
per
per
< :'' t : plausible probabilit y functions are derived for
l antities and so an expression for the equilibriu m
i [ particles embedded in the surface of a ball is
~ l , d ; this expression incl uding terms to cover the
I \s of the ball relative to the hardness of the particle.
I <y then be stated that a relatively very soft particle
ll I rushed when trapped between
Thus, from equation (5.3 ),
' w
bare metal surfaces,
ne bare surface and one surface "armoure d"
with embedded particles,
Two "armoure d" surfaces,
2'(db2) = dN max
Nmax
l11 ; is the number pressed in to the metal su.{[ace
i and is the number of particles torn out
~ +~ +~
or
or
OF GRINDIN G WITHIN
= 45(djb) 112
.... (5.1)
ln fact the probab e number of particles in the zonc
of operation is less than this; the maximum number occu
ring when there is an infinite number of particles in tl1c
system.
'"
t s suggested that a likely form of the probabilit y fun<:tion is given by
() = 1 ekv
.... (5.5)
and if it is assumed that about 40% of the maximu
number of particles are present when the void space b
tween the balls is just filled, that is when V = 1, then
= 45 (dfb )l/2(l  e05V)
.... (5.())
Thus, this equation gives the probable number
particles in the "zone of operation ", in terms of the b:ll
diameter and the quantity of powder present in the rnill ,
Considera tion is then given to the equilibriu m numbc
ofparticle s,N,, embedded in the surface of a ball and tli H
is clearly given by
dNfdt =
.... (5.7)
I
* The number would scarcely be expected
to be proportional to tl1c 'r
of the circle of radius , since a particle inside the circle automali r l y
relieves some of the others of load, but all of the particles in a ring coulcl
loaded.
as a relatively very hard particle will be crushed
l y I tween two "armoure d" surfaces (it will be pressed
: "unarmou red" one) and probabilit y functions to
1 11 hese possibilities are introduce d. On the
basis of
l ' unctions, expressions for the rates of milling, in
of the various mill parameter s, are deduced.
, i l ~rly it may be deduced that metal wear will occur
1 , during relative sliding motion between a pair of
l , ii N, tl1e "armoure d" portion of the surface of one ball
rses the "unarmou red" portion of the surface of the
l\ lbouring ball. From a knowledg e of the equilibriu m
I r of particles embedded in the surface of a ball the
I>:bili ty of such "armoure d" to "unarmou red" contact
computed and expressions tor metal wear deduced.
' I'11 analysis leads to extremely complex expressions for
l11 r e of crushing; expressions which do not lend them l v s to a general form ofsolution . For the two boundary
1 litions, in which the particles are either very hard
or
oft compared with the balls, however, the equations ,
l< r certain assumptio ns, become tractable and the
lt for these two conditions can be written in the
I '
S = Kt
.... (5.8)
146
BALL, TUBE AND ROD MILLS
>ROCE SS
where S is the sp_ecific surface of the particles,~ K is a :
stant, aid is ! for very soft materials and 45 for '
hard materials; all other cases falling b etween thes wc 1
conditions .
It is interesting to note that the case in which =!
responds to the " Bond Law of Comminution" dis c ss I
previously. Furthermore, for some intermediate hardH 'H
there is an pproximtely (but not strictly) linear relaio1
aor~..,
~~~r~.r~
;:.
=
~
~40r++~~~~
..
u
11'1
n/+ ~...
u20l_____
~==f==~~~fi~====~~:1
"'
10
2 0
Ball Diomot<r F .
5.2
ln. )
30
ship between the increase in specific surface and thc llil
ling time.
The analysis also reveals that the rate of grincli
increases w1th decreasg 6all s1ze; the Improveml
15rouglit about ya:gven reductn of size being gr <t
with soft material. This result is clearly confirmcd
Fig. 5.2, in which results of'tests by Coghill and Devan<'y('
on chert and dolomite, these materials having hardncss
in the ratio of about 3 to 1, are plotted.
Analysis of the fineness of grind in relation to the pw\
charge reveals that within limits, the fineness of grindi~
can be increased by reduction ofthe amount ofpowder in lc
OF GRINDING WITHIN
MILL
147
111 11 ;
: ll other variables being held constant; a maximum
l1 11 ached when the powder about fills onetenth of
l1 ;
between the balls. When a soft material is being
, however, this effect is much less marked and also
l1 l1 <ximum is reached when the powder occupies about
1 l ; lf of the space between the balls. The analysis also
l1~V hat the wear on the surfaces of the balls and mill is
1 / i ofthe quantity ofpowder present; this function
l1ing a maximum when about onetenth of the space
l1 w the balls is filled and thereafter decreasing rapidly.
( )' urther matter of interest arises in connection with
I once:et of a " zone of operation" in the b aH mill. It
I ll w; from ~ uatn (i.3), that'""'iJie"""radius of the zone
1 ; as the square root of the particle diameter. Other
l f.' being equal;tFie'refore, there is a defnitely reduced
i:blity of the smailer partiCle entering the zone of
: and, consequentiy, a preferent1a gniiag of the
I particles would 15e ~pected; rovided, of cours~
l. l1 e partic1es are neitl er s'Oi)Ig 1n relatn to the ball
, l1 a.!J~ are ex~ed from the we gesha ed s ace
l1 Iw tlie bails nor tlat the balls contain jnsufficient
1 ' for tlie rac ure t e RartiC es. fact, the
~ c ific gnn Ing charactenstics' ' deduced by Theimer
l Moser <3> show a marked tendency in agreement with
l : conclusions. This preferential grinding would also
" Kf.\'CSt that the size distribution of the material after
cling would not be geometrically similar to that
ili :lly, but that a "tail" initially on the righthand side
11 l curve would be reduced; this geometrical similarity
lt wcen mill feed and product would not be expected.
l'r m this theoretical treatment the following tentative
11 11 \Usions emerge.
( ) The processes occurring in a ball mill are controlled
Iy le probability of particles entering a "zone of pera
" between two balls and by the probability ofoccurt ' ' of some process after they have entered the zone.
(2) The rate of grinding of the powder is a function of
I; quantity ofpowder in the space between the balls and
148
BALL, TUBE AND ROD MILLS
this function reaches a maximum value when this spac
approximately full.
(3) The fineness of grinding attained by a l l
material in a given time of grinding increases rapiclly :1
the quantity of powder between the balls is decreascd :
maximum being r each ed when the powder fills appro i
mately onetenth of the space b etween the balls. Wh n .
soft material is being ground, however, this effect is m
less marked and also the maximum is reached when 1
powder occupies about onehalf of the space between tl
balls.
(4) The law giving the change of surface with tim l
grinding is of very complicated form and includes tcr
depending upon the h ardness of the material undergoi
grinding.
(5) The reduced probability of the smaller parti I1M
entering the grinding area would suggest that there .
preferential grinding of the larger particles. Thus it wolc I
follow that the size distribution curve of the material a
grinding would not be geometrically simi1ar to tl:
initially, but that any "tail" initially to the righth a I
end of the curve would be reduced; the n egative slop 1l
the curve thus being increased.
(6) The wear of the b all surfaces is a function of l
equilibrium number of particles pressed into the surfac't'
and this number depends upon the relative hardnes 1l
the balls and of the particle. Thus it follows from l1
treatment, that there exists the possibility that sojt b:ll
will wear less than hard ones; since soft balls m .~
acquire an artificial "hard" surface by reason of the : 1I
particles embedded in the surface of the metal. This is i
accordance with the practical observations of Engcls, l )
who states that the bodies ofmills for grinding soft m a L'l'i.l
should be hard whereas the bodies of mills .for grindi~
h ard materials should be soft.
(7) The wear on the surfaces of the balls and mill i~ .
function of the quantity of powder present, this f i
reaching a maximum when approximately onetenl1 l
I'ROC ESS OF GRINDING WITHIN MILL
149
I : between the balls is filled and thereafter de'
, i ~ rapidly.
will be disconclusions
thesc
of
validity
I ' ncral
1 ner in this chapter in connection with the study
1 11
/ l c su} ts of grinding tests.
I v~ ting ?ow to the study of the grinding tests by use
1
'' I '. lmensonless eqation (5.2), it is found that Gross<5J
I I' ~ figures f?r the values of "Rittinger's Constant"
f, 1\ lc:na, sphalente, pyrite and calcite, as determined by
FIG. 5.3
11
or a
laboratory ball rnill, r under specified cdi
I~, ~nd also figures for the Rittinger's constant as
l ; ed .by the use ofa dropweight apparatus.
( l : rly, snce, as follows from the treatment of Chapter
'', l ' po~er. to drive the miiis sens6'~ constant, tfe
rl [ Rttnger's constant for tle miiJ.s proportional
! t I ' ~crease of specific srfaces, S. Furthermore, the
1
11 we~ght. figures for energy p er unit increase in specific
s drectly proportional to , so when the Ritl <
II 'cr's constan_t for the mill is plotted against the results
/ l drop weght test a curve of S plotted against l jE
i ' (/1C mill is obtained.
'lis is done in Fig. 5.3 and a sensibly straight line is
150
PR OCESS OF GRINDING WITHIN MILL
BALL, TUBE AND ROD MILLS
obtained; thus it follows that the functiona1
(S.D) =
relationshi
l (D 2 gfE )
is, in fact, a direct proportiona lity or
(S. D ) =
K (D 2 gfE )
(5.9)
This relationship is for a batch mill grinding a singlc
material.
This expression can, however, be written as
(S. D ) =
K ( D4g. 1 fD 2E )
151
l t d in Fig. 5.5, and it is seen that the influence of
I .~ up is not large and it appears probable that the
lllc'l'c n e arises from the reduced efficiency of crushing
l~ nergy densities, which has been discussed in the
I"' vio us chapter. These relationships are still further c
t I, lowever, by the results of comparativ e tests, car l ou t by Coghill and Devaney, in which chert and
ll are ground by use of balls and pebbles (Table
' I ) . 'J se resul ts are summarized in Ta ble 5.1, and i t is
[ Jat
(5.10)
the agreement
good.
,r==::::::=,
~ t/1J?jfl
(~'}
FIG. 5.5
2
Sp~cilic Croity
Ot
TABLE 5.1
Crindinc, M~di.p,
FIG. 5.4
1\
where D 4g is the potential energy in the rods and D ~ /.'
is the energy associated with the powder, and a furl1
check made on this relationship since, in a given mill, tlc
increase in surface, SJ er unit time will be 12ro ortio l
to the density: of the grinding bodies, . _
Coghill and Devaney, (6) have given the results (abl< 1)
for the grinding of chert in a rod mill, in which the densi t
of the rods is varied and these results are plotted in Fig. 5.'1.
It is at once seen that the two quantities are almost exa tl
proportiona l, as required by equation (5.10). IC tl
deviation from linearity is assumed to arise by reason l
the existence of the group 6 ( f), then this relation s li
may be computed. The relationship 6 (f) so determinccl
al
Density
(g/cm3)
Increase in
surface per hour
Balls
Pebbles
Balls
Pebbles
78
78
27
27
154
460
55
169
Ratio of
surface
n crease
hou
Ratio for
=
78
27
and
From Fig.
5.4
I il r
ll l c
28
27
258
258
c function given in Fig. 5.5 is probably not highly
c ra te, since it has been derived from a combinatio n of
r ' ls for tests on dropweigh t crushing, on a ball mill
I a rod mill, but it is ofvalue in that it indicates the
11
'l
Js which may be expected to exist.
is treatment leads to the interesting idea that, for a
153
BALL, TUBE AND ROD MILLS
) ROCESS OF GRINDING WITHIN MILL
given material to be ground there pos~ibly exis~s :
optimum density of rriaterial for the gnndig matenal.
Thus, from Chapter 3, the power to rotate a ball ill
<:
is given by
h e same figure is also shown the correspondi ng
[ r :; rod mill; and it is seen that the same general
or relationship holdso
' s curves are not of great practical importance since
l y s of materials suitable for grinding bodies are very
l i < l , and so the range of densities available is re
152
= (l +04f )
~ +04 =J<:..("+04<:r
(5011 )
Also, from Figo 504
I.
curves do show, however, that the use of lum s of
1es 1s a practice which, at least on
;s grinding
1ds of ming effiCiency, cannot be recommende do
' next p air of groups which may well oe stu 1e are
1 I co
' I :
Sf t oc 50 2
(5o I
1
)
/........//
1/
.."'
c
4
Drnsity Of Powdt:r o' GmjCm3
05
<'!>
FGo
507
FIGo 56
/ (rl / D) and 3 (bfD) and these are most convenientl y c
__
Pt
50 
+4
On differentiati ng with respect to
simplifying,
(5o I 'I)
equating to zero
:cl
2+0820 =
or
=  4 y(0162+20)
l d together. From Table 32 of the work of Gross the
l of Rittinger's constant for various diameters of ball,
. nill of specified size, is given. Since the diameters of
11i ll and balls are known a curve of Rittinger's constant
,. inst (d/D ), Fig. 5.7, may be plotted; this curve being
i valent to a curve of 2 ( d fD) plotted against (dfD).
'l variation shown, however, arises in part from the
of (b /d) which occurs with the change ofball size;
arising from the use of the same feed material
I ' c~ach ball sizeo
(: ghill and Devaney, however, give sufficient data, in
', l l s 2 7 and 28 of their work, for these effects to be
;a ted, and the basic data from these tables are plotted
\ i : i
(5o ll)
From this equation the optimum values of densiy
balls for grinding materials ofvarious densities, are a lc
lated and these figures are plotted in Fig. 56.
I iHvariation
154
PROCESS OF GRINDING WITHIN
BALL, TUBE AND ROD MILLS
MILL
155
J ' i ~ . 5.8; a base of dfD being used since, again, the ball
l , tcr and the mill diameter are known. Since the time
l inding is comparative1y short, it is assumed that the
ll ive particle diameter is the average of the mean
111 11 i I size of the fe ed and of the product; the mean
( ' ( i lc size of the feed and product being assumed to be
l11 Hi ve size which passes (or retains) 50 % of the sample.
l11 tllis basis the data of Table 5.2 may be calculated.
now the curves of Fig. 5.8 are studied, it will be
11 d that the curves are similar in form and that they
. bc brought into coincidence by :
( I ) Multiplying the abscissae of each curve by a suitable constant, such that the maxima are brought on
to the same verticalline ; this constant being related
to the mean particle size of the material to which
the curve relates.
(' ) Multiplying the ordinates by a suitable factor.
e
65Muh .
( )  35M<sh.
<!> 20
M <sh.
8  14 M<sh.
  M <sh.
005
b.
FIG. 5.8
:xamination will show that the curves are brought into
l :. ntal coincidence by multiplying the value of (d fD)
l y (D/b)l/ 2 and this product, after squaring, becomes
(l ''j f)b) which may be treated as the new independent
. ;ble. This may be regarded as a correction to the (d fD)
to allow for the effect of the ratio (bfD ) upon the
ll of this term .
W lcn the data are plotted for the dolomite and chert
ltis way, as in Fig. 5.9, good horizontal coincidence is
li : incd. (The root scale ofthis diagram is used merely to
l , g the scale for low values.) It will also be noticed
l1.1 t he maxima for these materials occur at, as closely as
, I> expected, d2jDb = 200 ( = l42 2) for the dolomite
l 256 ( = 1602) for the chert.
Si11 e the mill is 19 in. diameter, these values give the
' J : tions :
d2 = 380b for dolomite
.111 I
. . .. (5.15)
d 2 = 485b for chert
156
BALL, TUBE AND ROD MILLS
e
PRO CESS OF GRIND1NG WITHIN MILL
5 <h .
( )  35 Mcsh .
@ 20M .. h.
 14 Mcsh .
50~~~~,_~rr+
&  10 Mh.
Up c: r Sit' Oo!omitc .
Lowcr
Sc:ris
 Chcrt
L
~
"'
~u 40~~~~~~~~~~~i~~
157
i l d value of (d2fDb), are plotted agai:st the cor o nding values of (bfD ), the curves of F1g. 5.10 are
l. d .
;~ it will be seen that the curves may be brought
almost perfect coincidence by multiplying the
l a c for chert by 28 and, as a confirmation of the
a y of the treatment, this is almost exa~tly the ratio
l l c rates of milling obtained from the Independent
IN, naking use of balls and p ebb1es, summarized in
11
"'
: 30~~~4~tt_,~~~r
d_'2
 = 1 4 2
Db
)2
} 42\
~
r:
3
FG .
""
.......
~ ~ .
/CHERT.
0001
/
F.
0002
0003
5.10
5.9
'.l> le 5. 1. It will, however, b e remembered that for
r is about 28 to 30 times that for dolomite, and it
d2 = 35b
d2 = 55b
(5 . ll )
given by Coghill and Devaney,(6) and it is seen tha l
agreement is satisfactory. These formu1ae are also sirnil:
to those developed by Starke. (7)
If now the ordinates of the curves of Fig,: 5.9, for a
ill
. ~~ ~
I :
IH' l
also be noticed that 3 1 /3 38=55 which is in fair
ment with equation (5.15 ). This will at once suggest
the curves of Fig. 5.10 will be brought into even
r agreement if the quanti~y d 2 / Db is multi:plied by
(/':) /3 . However, since in the present work chert IS taken
.~
f unit h ardness, this result is best expressed as
(l 2fDb)( /E ) ' I \ where Ec is the value of
for chert.
' quantity is then used as the complete llldependent
!i
ri able .
158
BALL, TUBE AND ROD MILLS
PR OCESS OF GRINDIN G WITHIN
TABLE 5.2
'.
Feed Ball
mesh size
Mean
feed
size
(in.)
pa~ticle
lJ
( . )
( . )
00098
00098
00098
00098
00098
00098
00089
00085
00079
00072
00071
00070
00094
00092
00089
00085
00085
00084
00034
00046
00065
00085
00113
00138
0145
0105
0072
0052
0039
0032
000050
000048
000047
000045
000045
000045
275
20
138
10
075
00210
00210
00210
002\0
00210
002 \0
00140
00138
00138
00134
00126
00146
00175
00174
00174
001 72
00168
00178
00063
00087
00127
00172
00225
00285
0145
0105
0072
0052
0039
0032
000093
000092
000092
000092
000089
000094
275
20
138
00400
00400
00400
00400
00400
00400
00156
00119
00126
00109
00145
00301
00278
00258
00263
00255
00272
00150
00101
00129
00191
00255
00362
00565
0145 000147
0105 000135
0072 000139
0052 000134
0039 000144
0032 000185
....:6 f(j7s
00480
00480
00480
00480
00480
00152
00151
00186
00166
00329
00316
00315
00333
00326
00405
00115
00158
00240
00326
00540
0 145
0105
0072
0052
1039
000167
000167
000175
000171
000210
275
20
00690
00690
10
00690
00226
00232
00230
00286
00406
00460
00460
OMs8
00167
00230
00330
00488
0145
0105
0072
0052
000240
000240
000240
000258
25
65
20
138
10
075
I 062
_,.
35
. s
__.
tm"
20
~ ~
075
062
14
 ~
0 048
I?
(in.)
CHERT
Mean
  
20 8
10
275
20
138
10
sze
nn
in ~
t'
5.
' 75
<JU
10
075
(). 12
42!1
51HI
57 1
pa~ticle
sze
Mean
product
size
( .)
( .)
( . )
00098
00098
00098
00098
00098
00098
00075
00075
00071
00070
00070
00074
00184
00184
00184
00184
00184
00184
Mean
Increase
surface
per
minute
JJ
J5
00086
00086
00084
00084
00084
00086
00031
00043
00061
00084
00 113
00139
0 145
0105
0072
0052
0039
0032
0000455
0000455
0000440
0000440
0000440
0000455
00 122
00116
00108
00109
00107
00132
00 153
00150
00146
00146
00145
00156
00056
00075
00106
00146
00193
00252
0145
0105
0072
0052
0039
0032
000081
000078
000077
000077
000076
000082
15 1
187
209
230
r
183
00374
00374
00374
00374
00374
00374
00142
00138
00138
00 142
00174
00285
00258
00256
00256
00258
00274
00329
00094
00128
00185
00258
00365
00530
0145
0105
0072
0052
0039
0032
000134
000135
000135
000136
000145
000171
144
164
193
184
174
121
00700
00700
00700
00700
00700
00328
00390
00512
00530
00622
00514
00545
00606
00615
0066 1
00187
00272
00441
00615
00885
0145 000272
0105 000286
0072 000320
0052 000325
0029 000350
134
153
eed
s z e
111
150
195
214
233
106
6(
:. 75
'
3SO
4:\/
I <18
570
10
075
67 r
m
3!
12
'11
54 !
57 !1
;,.,,
3(
47 11
4 7'
5().11
";ffi.
:. 75
:.!0
138
10
075
062
47 11
111
275
20
138
10
075
.12:.7~
157
114
3:111
37
4!
Thus, the_func~iona1 relationsh ips (d /D ) and (b/D ) m:
summanz ed 1n~o the two curves of Fig. 5.11 a l
F1g. 5.12; where F1g. 5.11 consists of the curves [ /\ .
5.10 brought into coinciden ce by multiplyin g the ordinzt H
ofthe curve for chert by 28 and then plotting the resu1ti~
curve on a scale such that the vertex gives 100 %, anl
~ ~ ~Q.f1o\e ~)
d...:::. l
'
..D~ ' ' I (
( i . )
I :  u
2r
3
b~
M ean
ml
1\
" ' '1.('
1111
pt""
~ ~r 0 0690
159
MILL
DOLOME
Mean
product
size
d.
J ' .
.
5.12 is the resu1ting curve when the data of Fig. 5.9
brought into coinciden ce by the use of the curve of
'. 5.1 1, the resulting curve being p1otted on a per~
ntage scale.
: nsider now the form of the curve Fig. 5.12 . FQr a
~ i size mill and ~artic1e, but with an extremely: smali
l : II the rate of rin ing :.1IlJ e 1ow; since tfie a w1ll not
111tain sufficient energy to proauce much new surface.
VVih increasing ball size, however, the rate of grinding
160
BALL, TUBE AND ROD MILLS
>ROCESS
('~
20
I
I
0001
<%)
1
0002
Sj t
W l.cn
I~
1/
06
~
0 4
02
~ J'..._
..........
I
I
!/
d ~(~ ) ~
FIG. 5.12
MILL
161
:.
d3 ( l jd3)
= constant
.... (5.17)
the balls are excessivel lar e, however the rate
a ter 4, the
111 oclu ctn of WILa.\eer unit erier ecreases w1t
y I1igli ener~ ,9epsities and also the efficiecy of the
I low is impaired When th e balls are unduly large in relaI to the mill size. Thus, the falling branch of the curve
ol)lained. (In fact this _reduction is overemphasized by
use of the root scale; a natural scale the decrease is
h more gradual.) Furthermore, as the ball diameter
' ' ases, the number of particles which may be trapped,
: point of contact, is increased and since an impact
\ l h contains enough energy to crush a single particle
l not necessarily contain enough to crush two, and so
11 11 ; his cause in itself can result in a reduced rate of
li ' i1ding.
Thus several causes operate to reduce the rate of grind .( with excessive ball size and this is exactly wha_
t has
n fo und by experiment. It is interesting to note that
l : results of Gross, Fig. 5.7, are in close agreement with
ls d educed from the work of Coghill and Devaney,
in a suitable ch ange of the scale ratios will bring th e
l n g ecreases Since, as ex a1ne 1n
1 0
08
' s
FIG. 5. 11
OF GRINDING WITHIN
1\ ill tt ease since the increased energy will bring about
:s d surface production per blow. This gain is in part
ll by the decreasing number of balls, and so blows,
1 i l ieasing ball size but, nevertheless, the result is an
: sc in rate of grinding. When the balls have reached
~ such that there is suffiet energy to crush p racl , ll y ~artic e w~en, theoreticaiJ:""
l ;e gnn ing wou
rema1n cons ant Since t e
'/.\ producea 15y a impact a al 1s proporfnalO
I b e of the ball diameter, d 3 , whilst the number of
: c ls, per unit time, varies as the number of balls, that
, : r s inversely as the cube ofthe ball diameter, that is,
l tly as l fd3 .
100
162
BALL, TUBE AND ROD MILLS
PR OCESS OF GRINDING WITHIN
curve of Fig. 5. 7 into extreme1y c1ose agreement with that
of Fig. 5.12.
Turning now to Fig. 5.11; ifthe ratio of (d/D) is constant then in fact (bjD ) is direct1y proportiona1 to (bjd)
the ratio of partic1e diameter to ball diameter. It follows
from equation (5.3 ) and the exp1anation given ear1ier
however, that the number of partic1es which wou1d b:
crushed between two balls 1s g1ven by t e expressn
cc
.... (5. 18)
when b is small compared with d.
Thus, for the initia1 part of the curve, the rate of grinding wou1d be expected to vary as the square root of th
partic1e diameter and, in fact, this part of the curve /'
Fig. 5.11 coincides c1ose1y with such a re1ationship. As thr
partic1e size increases, however, the ang1e , Fig. 5. J,
increases unti1, at a certain critica1 partic1e size th
partic1e is expelled from between the balls; because th
frictional forces are insufficient to ho1d it in p1ace. Clear1y,
when this condition is reached the partic1es are expelled,
instead of crushed, and the rate of grinding should ,
ideally, fall to zero. Since in practice, however, the pa
tic1es have a range of sizes and the ball surfaces are
p erfect1y smooth, the curve shows a steady d ecrease in
grinding rate, instead of a sudden drop to zero.
This view can, to some extent, be confirmed as follows.
For the maximum, at which proj ection commenc s,
(bfd) =000075. Assuming this condition a1so coincid ~
with the maximum of the curve of Fig. 5.12, (d2jDb ) = 2,
or (d/D )2 = 2 000075. Thus djD = 004 and so (bjd)
= 000075 /004~ 002.
Thus, cos = d j l02d = 098 or = ll 30' approx. Also
a study of the forces acting on the partic1e shows that th1
coefficient offriction, , is given by =tan or ~2 ;
this value being quite as close to the va1ue which wou1d b
expected as can reasonab1y b e hoped .
The infiuence of the ratio of rod diameter to mill dia meter on the rate of grinding for a rod mill may now w 11
MILL
163
<' nsidered, and in Fig. 5.13 are given curves ofthe rate
f race productia re1ated to (d/D ) ratio, based upon
f' lf 11 of the work of Coghill and Devaney. A1though
l l. ata on the point are 1imited, it apP.ears that the rate
f f.\ 'inding increases steadi1y with the diameter of the rod
, l r 1arge rods becomes sensib1y constant. This resu1t
t surprising, however, since the rods act as crushing
ii and all the partic1es have to
surfaces. Thus, the probabi1ity
such an important variab1e as
pass between the crushof trapping a partic1e is
with balls as the grind
100
~
;;;::.
50
,./
//
Rod  n Crcut
_/. 0 7
@ i  Q4
c
0 02
0 0 4
INc
Q08
0 10
FIG. 5.13
" '
media. Clear1y, thus, the rate of grinding will increase
11 i ially, after which the increased production per impact
offset by the reduced number of impacts per unit time;
li s
1eading to a constant rate of grinding, as suggested for
ball mi11.
In all cases the production of specific surface wou1d be
x. pccted to be direct1y proportiona1 to the time of grinding
. 11 so it follows that
.... (5.19)
Finally, there is the question ofhow the rate of grinding
with the group (gjDN2) ; that is, from Chapter 3
with (NfNc)
'
: ries
164
In Fig. 5.14 is given the values of Rittinger's constant,
at various speeds, as determined by Gross for batch
milling a dry powder. From this curve it is seen that the
value of Rittinger's constant, which is proportional to the
rate of increase of surface of the powder, varies almost
directly with the mill speed. Theoretically, the charge
would centrifuge at N f Nc = 10 and, since there can then
be no grinding, the curve should theoretically drop to
zero along a line coincident with the line f Nc = 10.
40
...
30
<:
~<:
u 20
...,.
~
<:
oc
10
MILL
165
~ll occurs. This decrease i the rate of gridig at
li ~ l sp eds ofrotatio, whe wetgridig, would be e
(' . Thus, for a VCl wet pulp or for a dry powder,
l'' would be appreciable cohesion of the powder
l , '.'{ ad so cetrifugig of the gridig media would
l o i nflu eced by the presece of the pulp ad would
r
at, practically, the critica1 speed.
nod erately dry pulp would, however, exhibit conN l ;bl e cohesive stregth ad so the balls would ted to
4 0rr.~,~
~
c
::
~2 0?&L~~~~
..
"
..
10
t~~+++___j\1
Dot OCoghill and ~q_<y .
Rod Millw<t Crndng ( 60~sol i4
Data Cross.
Ball Mill Dry Crinding .
a.
<!) DscharC)t: Open nq4 S" .
I
08
02
P H. OCESS OF GRINDING WITHIN
BALL, TUBE AND ROD MILLS
10
02
11
11
8'D
0 8
10
FIG. 5.14
practice, sin ce there is s1ip between the mill shell and
charge, th e charge does not centrifuge at N f Nc = 10, buL
the operation takes p1ace progressively over a range [
speed. Thus, the curve would be expected to gradually
fall to zero for speeds of rotation greater than the critical.
In Fig. 5.15 are give the results oftests, by Coghill ad
Devaey, a rod mill wet gridig chert ope circuit.
this case the rate of gridig falls rapidly for speeds i
excess of about 65 % of the critical ; this result beig i co
trast with that for dry millig, just quoted, i which
ll1 re to the mill shell. For
ll1 sio might b e sufficient to
very cohesive pulps, the
allow a cosiderable por1i of the charge to b e carried aroud wi th the shell, even
w l the actual speed of rotatio is cosiderably below
l ritical. This explaation is sufficiet to accout com l : Lcly for the observed results. fact this phenomeo is
! quently ?bs:rved i~ bal! mills used to mix the pigmet
t
the lqud vehcle the maufacture of pait.
,c. ; i ni~ar. results for tests ball mills are show i Fig. 5.16,
. 1 d rt s see that curves of the type show in Fig. 5.15
rI Lhe exp erimetal poits well.
166
PROCESS OF GRINDING WITHIN
BALL, TUBE AND ROD MILLS
These results can be summarized in a curve of the form
of Fig. 5.17; this curve giving 4 ( Nf ,) in terms of
(/ ,) and the pulp composition. Unfortunate1y there are
~
Brokt:n
,....._ \,.,..,.2~Lb. Cht .
Lin~s
Fq . S
15.
11
30
I/
20
1/
10
40
.,so b. Ch crt .
//
 /
........
OOLb . Cht:r .
'
0 2
F.
: ,
~200Lb . C
'~J
0 8
5.16
10
Pulp Eithcr Dry
or Vcry Wc .
y /
08
/___
/
/
/
',
\
Pulp About 70% Solids. \ .
By Volumc .
'\
\\
0 4
0 6
NfNc
F.
5.17
08
MILL
167
111 fficient data for the effect of pulp
y va1uated but 1inear interpo1ation
composition to be
between the curves
ttl J' ig. 5.17 is probably sufficient for most purposes;
i.ally as but few mills run at much above about 75%
ttl l critica1 speed.
will be noticed that in Fig. 5.15 the increase ofspecific
111 1 e per unit contact time is plotted for the continuous
1 11it operation and that when p1otted on this basis the
1 s correspond with those for batch mills. Thus it
ltt ll ws that if the surface produced per unit contact time
1 scd, a direct comparison between batch milling and
o Hnuous milling is obtained. Doubt1ess this is the basis
1 h e statement, often made without proof, that the perl mance of continuous rnills may be predicted from tests
1111 batch mills provided that, in the case ofthe continuous
ati, the actua1 contact time is used.
onsidering now the rnill filling parameter 7 ( J), results
d upon tests by Gross (Table 32) and Coghill and
l >vaney (able 19) are given in Fig. 5.18, the former for
ltitl h milling and the latter for opencircuit milling. It is
. nce seen that a simp1e change of ordinate will bring
IH:s independent curves into coincidence; and, since the
I11Ul surface, S 1, is given by S J, these results may be
ralized into the curve of Fig. 5.19, for 7 ( J). These
nves are of the form which would be expected since,
wi l no balls present (J =) there would be no appreci. l > l grinding. Similarly, when the mill is packed full with
l : lls (J = 100%) there can be no ball motion and so no
' iding . In between these values the rate of grinding
ust attain a maximum. Furthermore, for small ball
l i ll ngs the rate of grinding increases almost directly with
111 ball charge and this would be expected since the
' bability ofimpact increases with the number ofballs. *
T he next variable which may well be studied is the
ftct of the variation of the fraction of the void space
ll
ball can only collide with its immediate neighbours, and so, p er unit
, the probability of collision is not the probability of collision of a given
with every other ball in the mill.
168
P ROCESS OF GRINDING WITHIN
J:3ALL , TUBE AND R O D M I L L S
roor,.,,
801~~~,_~~~
::
=
...
l w
MILL
169
the balls which is occupied by so1id particles,
function 9 ( V). In Fig. 5.20 are plotted the
1 t'. 11l s of a number of tests by Coghill and D evaney
(': 1IJI s 9, 13 and 14) for ball and rod mills grinding
'' io u s materials at various speeds; the results being cor' <t d by the elimination of the effects of critical speed
I y t h use of a multiplicative factor, and again the agree" t ' t is satisfactory .
I , I
js
tle
:360~~f.~~~~;~~i~~_,
0:
,,,
,,,
11
I\
....
~
~
20
nr,~+::::::=:1::~
Continuous MillingBoll Lood Vori~d Chorg~
Constont .
20
40
Boll Lood
2 0 ~~+~r~~+~~~
Cf)
Curv~s 2&3 Botch MiU i nq Chorg~ Proportionol
1 1
~4~~~+~~~~~~~
CHARc{J)  PERCENT
FIG. 5.18
I 00
"'
I
(J)
0 2
04
0 6
0 8
10
12
R  V.
F.
Wlen
w ih the
3 0 5 1111""........::t1
er:
20
40
F .
( J)
5.19
80
5.20
these results are plotted on a basis comparable
theoretical curves as deduced from the theory of
l{ose, explained earlier, Fig. 5.21a, it is seen that the
rimental curves have the same form as the theoretical
0 11 s, shown in Fig. 5.21 b . The theoretical curves are caln i a ted for materials of zero and infinite hardness, and
11 1 curves for real materials show intermediate characteriNti s, as would be expected. Furthermore, when curves
/' lotal surface change per unit time, Fig. 5.22a, are plotted
170
P ROCESS OF GRINDING WITHIN MILL
BAL L , TUBE AND R O D M ILL S
1
"
VChcrt (Hard)
0 (1
<
08
Rat io
or tical treatment proposed by R ose; this treatment
l g, as far as is known, the only one so far proposed by
1 h th e effect of the mill parameters has been deduced
~ ~~lomi tc (s .)
0 2
mpared with those calculated theoretically, Fig.
, ' ' IJ, the agreement in form is again seen to be reasonably
I\ I.
.
' s these results support the general validity of the
I
I
I( I
171
10
12
14
0 2
04
06
0 8
Rat io V
10
1 2
[\... "+n
"'~Z t ro Hordnc: .
/
f
02
l
0 4
0 6
08
Rotio V
10
4
b.
Ro tlo V
b.
FIG. 5.22
Im theoretical considerations. U ntil a theory is suffi
\
\
,,
l s function from the basic physical data, the curve of
~Hord Orc: .
~' ..........._
I /.. . Sott O rc.
i ntly developed, however, to allow the calculation of
1
' g. 5.2lc probably gives the value of 9 (V) with an
; racy
08
Ratio  V
10
12
sufficient for design purposes.
The question of the effect of the consistency of the pulp
/( U) would appear to be, at least as a first approxima1i, one which is fairly easily treated. H owever, the
blished data relating to this variable are in fact so
r nflicting as to render impossible the drawing of firm
onclusions.
FIG. 5.21
172
BALL, TUBE AND ROD MILLS
When grinding a pulp, a mill ma~, i?~ally, be operat~d
in accordance with either of two lmltlng cases; one
which the pulp just fills the space between the balls and
the other in which the solid material in the pulp just fills that
spaceo In addition there are an unlimited numb~ro of other
possibilities, in which neither of these condtns oare
observed, and it is probably the grcat numbe~ of poss?lc
conditions of operation which lead to the ~conc~usvc
nature of the experim ental evidence on ths vanableo
From other evidence however, it is possible to draw
tentative conclusions 'which are possibly of value until
sound experimental evidence is available o
In the first place, if the mill is operated in such a way
that the space between the balls is just filled ~ith pulp,
then clearly a reduction in the solid content wll reducc
the quantity of powder present and, at lea~t ~ a first
approximation, have the same effect as the ovanat~n ofthe
parameter V discussed earliero Thus, for ths case, t would
be expected that an approximate allowanc~ could be ~a~e
by the use of Figo 5o2lco Clearly, since this calculatiOn s
based a knowledge of the volume of solid material present
it can be extended to cases in which the quantity of pulp
i~ either less than or greater than that required to fill t~e
void space between the ballso In the second case, that
which the solid material is just sufficient to fill the space
between the balls, it would be expected that the consistency of the pulp would have but little effect and, as a
first approximation, may .be neglectedo
In both cases the fluidity of the pulp will have s?me
effect in that with a more fluid pulp the finer partcles
will be washed from between the balls and the efficiency
of the crushing process improvedo This action will ~lso
lead to an amount of selective grinding, but this subjeCt
will be treated in a later chaptero Thus, an approximation
to the value of s ( U) may be madeo
The diameter u of the exit aperture from a mll would
be expected to h~v~ but little effect up; theorate ?f~illing
since, other things being equal, vanatns ths dmen
PROCESS OF GRINDING WITHIN MILL
}73
~ill m erely vary the flow rate through the mill. In
l c ltn
the effects of variation of this quantity are in
l t l d when the calculations are based on the actual
t <~:ct time of the material with the mill. There might
, howe~er, some small effects arising from variation
lc operfect10n of the mixing in the mill, with varying
. t dameters, and similar minor influences; but such
. rs would be e_xpected too be ~nimportanto The experi o ntal daota relatong oto ths vanable are sparse, but the
of Fgo 5023 ndcates the trends as far as they can be
luced from the published evidenceo It is at once seen
: t the effect of this variable is small. It is clear that the
l g th of the mill is also a significant variable in this
~ O St+++1
e.::
111
02
24" Long.
FIGo 5023
:~ ct; since this affects the angle of slope of the surface
/ the charge, but there are insufficient data for the rela iorships
to be evaluatedo
The question of the influence of the ratio of new feed to
: ulating load does not appear to have r eceived system. treatment, but general conclusions based on the
I"' '1ous treatment may be drawno
l "ro~ the previ~us treatment it is clear that, provided
l1 c rat of b all s1ze to average particle size is not too
~1 all,o th~ ratoe of grinding is not greatly influenced by the
z dstnbutns of the granular material, and the foreo ng work may be used for the estimation of the increase
surface in the mill. Thus, the combined feed may b e
l'('ated as a normal feed and the product size calculated
BALL, TUBE AND ROD MILLS
174
PROCESS OF GRINDING WITHIN
175
MILL
' .!I
Authority
Coghill and Devaney *
(ab1e 25)
Materia1
Chert
"
Do1omite
"
19 36
2 4 24
6848
19 36
10
10
036
036
060
060
080
080
Mill size
Diam. (in.)
Length
060
060
077
077
Gow, Guggenheim,
Coghillt (ab1 e 16)
Dolomite
78. 144
036
0715 078
Maxson, Cadena and
Bond:
Quartz 628
648
6100
6200
12 12
108
10
10
10
089
089
089
089
100
100
100
100
Coghill and Devaney,
Rod Mill (able 9)
Chert
19 36
19 36
10
10
040
070
060
060
3648
05
075
100
"
Cement
Clinker
Starke
(~))
(b)
065
065
()o(J5
()o(J5
005
005
000 10
00008
00003
00010
070
003
098
098
098
096
!HJ94
IHJ94
( )( ) 4
11094
040
070
087
4(~)
 
2(~) a(b)
10
10
075
097
(V)
g ( V )
(J)
? ( J)
(~)
          
100
100
100
100
100
100
100
100
050
050
050
050
070
070
070
070
30
30
30
30
06
06
06
06
0036
0030
0006
0030
000013 083
050
100
10
047
072
30
06
00082
085
085
085
085
060
060
060
060
036
036
036
036
40
40
40
40
02 13
0213
0213
0213
096
096
096
096
30
30
30
30
06
06
06
06
00079
00104
00121
00144
IHJ 6 00007
!10 6 00007
090
090
10
10
10
10
10
10
045
045
075
075
30
30
06
06
078
074
IHJ28 0001
100
060
03
50
048
07
20
08
0012
0 6 J
0005
0005
0005
0005
* Ball Mill Grinding. U.S. Bur. of Mines Tech. Paper,
. 581 .
t Trans . Amer. Inst. min. (metall .) Engrs, 1934, 112, 24.
: Ibid. , 112, 130.
Rock Prod., 1935, 38, 40.
w l1 re
from it by the use of the various functional relatiJnships
already discussed.
.
.
On the basis of the present analys~ equaton (5.,_ 2 ) for
the computation of the change of specfic surface, nay be
simplified.
It has been suggested that (D 2gf) = (DZgfl. and
5 (gt2D) = (gt2JD )ll2, so it follows that
s =
(D~t) 2(;)  3(i) 4(~) . 6 ) ; 7 (J)
9 ( V) (5)
(5.20)
095
100
073
095
068
. d
2 (d/D)
is read from Fig. 5.12,
3 (b /D ) is read from Fig. 5.11,
4 (J , ) is read from Fig. 5.17,
6 (f ) is read from Fig. 5.5,
7 (J) is read from Fig. 5.19,
9 ( V) is read from Fig. 5.21c,
11 (ufD ) is read from Fig. 5.23,
l graphs s( U) and 10 () being omitted since it
l> ~ shown that their effects are largely included by
has
the
rrect computation of the values of other variables.
he value of the constant must be determined by
rence to the results of experiment and there is an
. nount of data available for this purpose. ln Table 5.3
, given a number ofvalues; based C.G.S. units, but
176
PROCESS OF GRINDING WITHIN
BALL, TUBE AN D ROD MILLS
with taken as unity for ch ert. All these figures are seen
to be of the same order, there being about a 3 to 1 ratio
between the extreme values. This scatter probably ariscs
from difference in the bases of calculation of the chang
in specific suface; since the various workers have diffeen
methods. The agreement is such, however, as to suggest
that the genera1 method of approach is correct and tha
an accurate series of experiments for the eva1uation of thc
constant is the primary requirement. On the basis of thesc
1imited data, however, it appears safe to draw the provisiona1 conc1usion that the value of the constant , in
equation (5.20), may be taken as
= 002 for a ball mill l.
= 075 for a rod millf
(5.21)
with a deviation of, say, 50%, but these figures will
probably be modified as more data become availablc.
Thus, the rate of change of specific surface may be calculated from the :nill dimensions ; provided the grindability
of the material, r eJative to chert, is known.
There now remains the question of the grinding of a
mixed material; this case being practically important
since, for example, any ore contains at least two components. The problem h as been studied by Holmes and
Patching, (8) and some interesting tentative conclusions
have been reached.
The rate of grinding is written in the form:
kW
(5.22)
where is the weight of product per revolution, is thc
bulk density of the material, W the weight of material in
the mill and k is the specific rate of reduction, and it is
shown that the ratios (k 1' /k2 ') and (k 1 /k 2 ) are approximately equal and sensibly independent of the proportions
of the two components in the mill; k 1 ' and k 2 ' being th
specific rates of reduction for the two materials millecl
separately and k 1 and k 2 for them milled together.
177
MILL
Thus, it appears reasonable to conclude that when two
:aterials are g~ound in combination, the rate of produc of surface each of the components is in the same
ratio as it would be ifthe materials were milled separately.
rthermore, the rate of change of specific surface in each
omponent is the same as :when the component is milled
scl?arate1y. This view is also confirmed by Gow, Guggenl cm, Campbell and Coghill, who show that for tests on
hertdolomite ~ixtures, the rate of size reduction is, very
r ughly, proportonal to the grindabilities ofthe chert and
lolomite when milled separately.
REFERENCES
I . RosE, . .
Trans. Instn . chem. Engrs, Lond., 1957, 35, 87.
2. CoGHLL, W. ., and DEVANEY, F. D. Trans. Amer. Inst . min. (metall.)
Engrs, 1930, 87, 51.
3. R, 0., and MosER, F. Kolloidzschr., 1952, 128, 2, 6874.
4. ENGELS, . Metall, 1954, 3 4, 102.
Crushing and Grinding, U.S. Bur. of Mines T ech . R ept 402
r.. GRoss, ] .
(1938).
6. Coc:rLL, W. . , and DEVANEY, F. D . Ball Mill Grinding, U.S. Bur. of
.
Mn es Tech. Paper, . 581, 1937.
7. SR, . R. Rock Prod., 1935, 38, 40.
11. HoLMEs,J. . , and PATCHNG, S. W . F. Trans . Instn chem. Engrs, Lond.,
1957,35, 111.
179
MIL L
CHARAC T E RI S iC S OF
CHA PTE R 6
OT HE R CO NS ID ER AT IO NS
RE LA TIN G
LL
CH AR AC TE RIS TIC S OF MI
;..
~~~~+4
~ ~~+~~~
'
of the various
IN the previous cha pter the influ ence
char ge has
mill
and
rnill
the
ning
phy sica l qua ntiti es defi
orm ance of a rnill
bee n stud ied in con nect ion with the perf in the pow der.
ace
as a device for the crea tion ofn ew surf
desi rabl e that the
For some purposes, however, it is also size dist ribu tion
of
form
d
erre
pref
a
prod uct shou ld have
it is usua lly necescurve. For exam ple, in min eral dressing mesh shou ld be a
200
sary that the mat eria l belo w abo ut
ifications in the sizc
min imu m. some exte nt such mod
brou ght abo ut by
be
can
cs
risti
distbution char acte
mus t be emp hasized,
cha nges of the rnill dimensions. lt
ted scope but, even
lirni
however, that such chan ges are of
so, are often wor thw hile .
ed is the effec t of
The first rela tion ship to be consider diam eter , upo n
rod
the vari atio n of the ball diam eter , or
Fig. 6. 1 are shown
ln
.
ined
obta
tion
ribu
dist
size
the
by ball milling;
typi cal size freq uen cy curves obta ined of Coghill and
data
these curves bein g ded uced from the
(l>.
Devaney
of a coarse feed
As wou ld be expe cted , the grin ding
ctive grin ding
effe
the
to
rise
give
with sma ll balls does not
does prod uce a conof the coarse part icle s in the feed but
larg est ball sizes,
the
side rabl e amo unt offi ne prod uct. For
tion is atta ined .
frac
se
coar
however, rapi d grin ding of the
ever, it will be
how
,
feed
r
fine
the
Tur ning atte ntio n to
s a coar ser prod uct
obse rved that the larg est ball size give s it app ears that
Thu
than does the nex t size sma ller balls.
178
+4
..u 40 ~.+~+~~~~
c;
05
20
15
Port clll
Dlam t~r  Millintrli's .
400 r  .  r  . 
r  .  r ,
F~~d
(35 ~h
Ch~r .)
;.
.!
i 200
..
~~r~~~~~
t~~i~~f7~~
c:
~~~~~_~
~ r+~f
3~~o~~~o~~~~~aJ
o~~o~~~~0~2~o~
Part iciC'
Diamrt~r
FIG.
6.1
Millimtrrs .
BALL, TUBE AND ROD MILLS
180
the coarser fractions may be ground, without the production of excessive fine material, by the use of balls of larg
diameter; however, the particle size must not be ex siv
or this ben efit is 1ost.
From these rather limited data it would appear that th
best selective grinding is obtained whe the ball diameter
is about 100150 times that of the particle to be crushecl.
This figure is very roughly in accordance with that whic]
wou1d be d educed from Fig. 5.12, but close agreemen
CHARACTERISiCS OF MILL
ii
1 aded to give a constant force. Thus, the lar~er
111 , t i l s would be the first to be bro~en,. whatever th~ s.ze
rods. t is possible that the grn~ng chara~terstcs
W I be different when the charge lS cataractng, but
.
. .
,, i l nc e on this point is not available.
nmination of Fig. 6.3 shows that varatons the
l i y of the grinding media does not have great effect
lective grinding; although the use of the .more. de~se
IH does very slightly increase the. preferental g~~dng
l t 11 coarser fractions in the feed. Snce, however, t s not
11 1 t 1

 .........,_
,
~
'
'\ I
"
I""
11
11
PARTCLE
D . R  L L es .
Fc.
Q2
'
""
~
DN stvOF
Roos .= 289
 5 0 @
 7 3 
......
f",'
~+
Q8
RC.
6.2
cannot be expected since the two cases are not identical.
Thus one is for the crushing of a powder having a singlc
size of particle and the other is for the most rapid production of surface when a materia1 having a wide sizerangc
is comminuted .
For the rod mill, Fig. 6.2, however, marked selectiv
grinding is not evident; in fact the size distribution curves
of the products are very similar for the three sizes of rod
tested. This is to be expected since the results are for
NJNc= 04, which would suggest that the charge is cascading and, in such a case, the rods would act as crushing
181
""'
LFEEo .
"'< '.....
~ ::.:
I 2
14
).
DAMETER. ..RS .
FG.
6.3
""
to select gri~~ing .bodies ~ade from materia!s
of ths
should
IH noted that these tests have been run for times such that
1 work has been expe?ded in driving the mill. Thus,
t (' Jlows that the exp endture of equal amounts of work
J' l.uce practically identical pro?uc.ts from ~he same feed,
~ardless of the d ensity of the grndng m~da. ) Wh:n the
: ll clarge is fixed, variation of the quant~ty <;>f ore the
ll brings about no significant selective grndng between
t I different sizes of material in the feed.
~ : icable
l ~ wide range of denstes, nteres t th: effects
: i able are theoretical rather than practcal. (It
182
In Fig. 6.4 are plotted curves, again according to Coghill and Devaney, which show the effect of pulp density
upon selective grinding and it is at once evident that ;
more rapid reduction of the larger fractions is obtainccl
with a pulp of moderate thickness than with a very thi k
one. The probable explanation of this phenomenon jg
that with a fluid pulp the fine material is washed from
between the crushing surfaces, and so is ..not subject t
further grinding. In addition, cleaning the grinding zon c
of such materialleads to a higher efficiency of grinding /'
183
CHARACTERISTIC S OF MILL
BALL, TUBE AND ROD MILLS
~ fficiency being defined by the equation
Increase of surface funit time
Energy inputfunit time
. . . . (6.1)
(~ nsideration of the treatment of Chapter 3 will s~o7
t at, for practica1 purposes, the power nput to a mll s
i ven by
D2.5L( 1+ :) '(;) Ps(J) Pn(V)
(6.2)
~80~~~++~q,_rrr+~
u
~
~60 ~~~~+4~~rrr+~
w
effect of variations in the other dimensionless groups
h
h
ing small.
imilarly, consideration of Chapter ~ s~ows t at t e
1 : t of production of specific surface, S jt, s gven by
4 ~+~~d~~,_~~~r+~
D~S s4(:.) s7(J) s9(V)
63)
l vanatons of the other dimensionless groups being,
dcr normal conditions, negligible.
10
PAFH CL E DAMEfER
IN
RS
FIG. 6.4
the large particles remaining. Thus, both from the point
of view of rate of production of surface and of improved
grinding of the larger materials, it is desirable that the
pulp should be of moderate consistency.
Although there is not much evidence on the point, it
appears that the size distribution curves of the products
from mills with exit passages of lage and small diameter
are very similar. Thus selective grinding is but slightly
controlled by variation of this dimension.
At this point it is of interest to consider, briefly, the
conditions necessary for maximum efficiency in a mill; the
.
J!ere the presuffix or S attached to , de_notes ~hat
l . power or the surface product~on, r~spectvely, s a
tion ofthe dimensionless quantty whch follows ..
The total surface, S 1, produced is, however, proportonal
le product of the specific surface produce_d and the
vo lume of the mill. Thus the rate of producton of total
t r ace is given by
~~
nz;Ls4(:) s1(J) s9(V)
(6 4 )
Wl equations (6.2) and (6.4) are substitute_d into _(6.1)
l ~ quantities D , , L and cancel and the mll ,efficency
f.\' vcn
'1
by
{s4 (NJN,)} fs7(J)} {s9(V)}
'(f ,) .;,s(J) Pn(V)
....
(6.5)
184
CHARACTERISTICS OF MILL
BALL, TUBE AND ROD MILLS
From this equation it follows that, for given values
(N j Nc), (J), and ( V), the efficiency of the mill is ind pendent of size .
In Fig. 6.5 the curves of power input to a dry mill ,
~omputed from. the results of Chapter 3, and the rate I'
ncrease of specfic surface, from Fig. 5.1 7, are plotted on
..
. !:
0 2
'
12
( t;YNc)
F.
6.5
a ?~se of (N/Nc) From this figure it is
mllng of a dry powder, or a very fluid
seen that, for the
pulp, there is fair
agreement between the curve of power input and that of
the rate ~f.production of SJ?e~ific surface, for all speeds up
to the. cntcal. . For the mll'ng of moderately dry pulps
there s ~~so far agreement for speeds up to about 70 %
of the cntcal. Thus, for these two cases the mill efficiency
185
. y b considered to be sensibly independent ofthe speed
l t
mill.
lt is also probable that for speeds above 70 % of the
i t i:al the power demand would fall away when mode . t l y dry pulps are undergoing milling and so, in this
. <' ls, the mill efficiency would remain sensibly inde l nt of the mill speed. Experimental evidence this
i is lacking, however, so a final conclusion cannot be
. l ( ly drawn, for this case.
appears to be safe to state, however, that, for all
: ical milling conditions, the mill efficiency is sensibly
l pcndent ofthe speed ofthe machine.
.
.
'' relationship between mill efficiency and mll fillng
1 i ll now be studied.
Fig. 6.6a is shown the curve relating the rate of
l"'oduction of specific surface to the mill filling (see Fig.
' . <) and in Fig. 6.6b is the curve of total surface produc , per unit time, deduced fr?m it. '!h~ curve Fig. 6.6c
lat ofpower input to the mll and s, fact, the curve
l l' ig. 3.5.
.
. . .
'le mill efficiency s then deduced by the ds of
l: rdinate of curve Fig. 6.6b by that of Fig: 6.6c and
I urve so obtained is plotted in Fig. 6.6d. From this
'il h it is seen that the mill efficiency is sensibly cons!ant
I flllings ofup to about 50% and that for greater fillngs
I ' fficiency increases rapidly.Thus, it would appear that,
l1 le greatest efficiency, a large filling, say 75%~ should
used. This is only partially true, however, for wth such
. lilling the rate ofproduction of~pecific surface is re~u~ed
. l so to obtain a product of gven fineness the mllng
t must be increased. If the energy to cause comminu 1011 were all that is involved, this would be immaterial,
t hc en ergy lost in friction in gearing, bearings, etc., is
. ! . t independent of the charge in the mill and this loss
ninimized by the reduction of the milling time to a
1ti1imum. Furthermore, the reduction ofthe milling time
ll H'rates the plant for other duties and so is economically
li able.
186
BALL, TUBE AND ROD MILLS
CHARACTERISTIC S OF
Thus, on balance it appears that lower fillings, witl1
higher rates of specific surface production, are desirabl ,
even though the energy consumption is proportionately
larger, and the statement by Gow, Guggenheim, Campbc11
1 0
v !'>
""'..
Mi tt
"""\
0 4
F lli ng
~0 5
""' ""
10
1/
I
4
Mll
J,
b,
10
/ \
06
Flling
I
v
02
Mll
6
4
Fill ing  J .
~
0 5
'
02
04
v
6
Mi ll Filling 
c.
J.
d.
FG.
6.6
and Coghill (2> that: " commercial operation large
charges of 5560% of the mill volume have been found
to give lower capacities and efficiencies than loads of
4550 %", is probably well founded .
Finally, the relationship between mill efficiency and the
MILL
187
: ntity of solid material present in the pulp, the para r ( V), can be examined.
this case Fig. 6. 7a shows the curve for the rate of
oduction of speci.fic s.u rface, ~ terms of this parameter
(' l<'ig. 5.21) and Fg. 6. 7b s the curve of total surface
l luced from it.
'he curve of Fig.
~I\
6.7c gives the power demand of the
111il\ in terms ofthe variable V, so deduced from the treat ~t of Chapter 3. Although this curve is com~ute~ f~r
~ \: ific values of mill filling and powder densty, t s
obable that the form of the curve is appli~able to all
( \ r practical cases, with an accuracy sufficent for the
.
sc t purpose.
'fhe mill efficiency is then deduced ~ the same. way ~s
viously and it is seen that the maxmum efficency s
; hed when the value of V is about 30%.
ft must be pointed out, howev~r, that the red.uction of
\ quantity of powder in the mll leads to an ncreased
'L ofmetal wear. Thus, the quantity ofpowder cannot be
cJuced unduly and, on balance, it ~s probable that the
111 timum value ofthe parameter (V) s between about 06
d 10.
'fests by Coghill and Devaney<3) show that the. ha~dn~ss
of' he balls has some effect upon the rate of gnndng
\1. These results are plotted in Fig. 6.~a an~ it is se:n
: , both for dolomite and chert, there s an ncrease
l rate of grinding with increasing hardness of ball.
se data require correlating, however, and as a first
roximation this can be carried out as foll?ws: In ~he
~ place it is clear that the greater .rate ~f.gnndng. wth
\11 dolomite is due to the greater gnndablty, that lS the
1 all er value of , of this material. There is a.mple
vidence however that , is 275 to 30 times Ed so f the
linates ofthe curve for dolomite are divided by, say, 275
elation with respect to this vaabl_e is introduced. It
: lso clear however, that , the relatve hardness of the
: \l to the :aterial undergoing grinding, is an important
v. riable.
"'
l:r
... .: ~~
"'
>
....
.ci
~."..
(Q
.;
~~~
400
so
25
7S
Min<ralogical Hardnus Scalc .
b
Hardn<ss Of Ball.
I/
'
"' "'"
11 11
"
.....
>
~~
4~L~
0...
 6
500
"'
'
....
()
"'
. J)MOd
w!l }!unjJons =>!d
FIG.
6.7
soo
ISOO
FIG. 6.8
...
1000
~IlM I! ~rdn aaOr I ~~~;~~7~~;~::~:~O~l~;~t .
()
 ~
~
'J_
r l n<l l
S'
'#
f.....1/ ~
'!'
"'
200
11
I
I
 
,!
1+1
'
(JU ! J! ~ J3 I/!~
11... ~
..
"'
r,, ~
10 ~...
16
',, ::1_ "'
<:'
w!1 Hun..d :>ns: ~1
~ /.
189
MILL
k~~
.:
CHARACTERISTICS OF
'
<:'
~~
I
I
'
'!
Now a curve by Fahrenwald, <4 > shown in Fig. 6.8b,
l , t ng grinda bili ty to mineralogical hardness, suggests
l , ~ reduction in the value of from 10 to 1/275 cor o n ds to a reduction of mineralogical hardness from
( 05 and so the relative hardness of the ball to the
l l oni te is twice that of the ball to chert. Thus the
11l
issac values of the curve
ltli ed by 20. When this
relating to dolomite must be
is done the curve ofFig. 68c
190
is obtained and it is seen that there is fair agreemc nt
between the results for the two material s. From this it may
be deduced that the rate of grinding is approxim atcly
halved when changing from grinding the softest ore wit11
the hardest possible ball to the grinding of flint with ;
very soft castiron ball.
This result is reduced to the form of a multiplicativ
function in Fig. 6.8d. Unit value of (') has been taken
when '= 450, since this corresponds to chert undergo ing
grinding by moderat ely hard castiron balls, which is, a~
far as can be judged, the condit~on of the tests which hav<~
been examine d in the present work.
The rate of grinding is not the whole story, however
for it has been shown, from theoretic al consider ations by
Rose(5) and from tests on industria l mills by Engels, (6) tha
ball and liner wear is reduced by the use of sojt metal for
the balls and mill shell when very hard material s are to b
ground. Thus, although the rate of grinding is reduced,
the replacem ent costs of the balls, and also contami nation
of the product, is reduced by the use of soft metal for thr
grinding surfaces. The reduced rate of grinding is doub
less due to the energy absorbed by the plastic and elasti
distortions of the soft baJls reducing that available for
crushing the particles .
Whilst dealing with the question of the influence of thc
grinding media upon the grinding characte ristics of a mill,
it is of interest to consider the available data relating to
tests upon grinding bodies of various shapes. These data
are collected in Table 6.1, and it is seen that, all counts,
spherica l grinding bodies give the highest performa nc
figures . Thus, although it might be economi cal to use larg
lumps of ore, or "cylpebs " made from scrap drillrods,
when such are available at the plant, there appears to b
no economi c justifica tion for the use of grinding bodies
shape other than spherica l when the l&.dies have to bc
'
purchase d at compara ble prices.
about by thc
brought
demand
power
of
increase
The
use of bodies of nonsphe rical form would be expected ;
CHARAC TERISTI CS OF
BALL, TUBE AND ROD MILLS
...
"""
"'....~""'....
"""
. . .
"' "'
::g~
MILL
U")
""
      ..ou
ti
~ ~
"
p..,
*....
"'....
"
"8
~ :::5
..,
::g5
U")
"""
""
""
U")
~"
"'
~
"'
:::5
.... u
~g
c: ....
"""
U")
""
>
U")
U")
,_
;::::>..
~ c:
c:.:::.
.
.....
.;
>
~6
,_ I'

<> ..
~
CJJB
U")
.,
,_


U")
U")
"'""" "'"""
~~
u
"' "dY~~
~
::g"'
'5
<:
.;,
c
'<:>
""
<>:J 
..c:
u:>'7'
"'!"
"'~
"'
co
+ +
MU")
U")
....
::
 
 ....
~ ..C::!
+

~~
"!:::.
c:
~
c:
~
::::::>
..C:OJ
~~
..
"' ""'
~8
"'
~
....
..c:"..
CIJ
~u
::
....
"'
:.
:>
"'
""
~
"
:::5
"'
....
~lj
"..
OCIJ
~ ::g
u:::..o
;='
::g
192
CHA R ACTERISTICS OF
BALL , T U B E AND ROD MI L LS
since tl1e interlocking is tantamount to an increase in th:
coeffici ent of friction, which brings about an increasrcl
angle of slope of the free surface of the charge, and so a n
increased displacement of the centre of gravity, whi l1
leads to increased power demands.
The size distribution of the balls in the mill charg
appears to have no significant effect upon the size distribution of the product, nor upon the rate of grinding
provided that the size distribution of the balls is not to
unreasonably chosen. Thus, no significant difference i
the grinding characteristics can be d etected in the published data on the subject between grinding with balls /'
uniform size and with balls having a ratio of diameter
largest ball in the charge to the diameter of the smalles
b all of about 2 to 10 or even greater. This comparativc
insensitivity of the grinding characteristics to ballsiz<
distribution is fortunate since it means tha t any convenien\
method of ball rationing may be adopted.
This at once gives rise to the question ofthe equilibrium
size distribution of the balls in a mill; this distribution
being in equilibrium under the conditions of wear whicl1
hold during grinding. The problem has been studied by
Davis, ) and, on the assumption that the rate of wear i~
proportional to the weight of the ball, it has been shown
that the size of distribution of the equilibrium charge is
given by
. . .. (6.6)
where W is the p ercentage by weight of the balls falling in
the size interval d. to dh, d,n is the diameter of the balls fed
to the mill and d0 is the dia meter at which the balls arc
r ejected.
The results of sieve analysis of the b all charges of largc
mills, quoted by D avis, are in extremely close agreement
with those calculated from equation (6.6) and so th
general truth ofthe theory appears to b e established. From
MILL
193
l 1iH would appear that a definite equilibrium of the size
l bution of the balls exists. So all that can be done in
" . c is to feed replacement balls of diameter d," at the
, '" rate and to remove the balls smaller than diameter
, / 11 ithcr continuously or at fairly frequent intervals. The
l of Davis leads to an expression for the size distribulo
or the balls in a mill but does not give information as
rate of loss of m etal. This matter will be discussed
, lnter chapter.
' ' : factors upon which the rate of flow of the pulp
o 11 g l1 a mill depends appear not to have received
1 ~e investigation. In an article by Anselm,OI) transI , i c l by Pearson,<l2) a method for the calculation of the
f passage of cement through a mill was given. The
l 1 iH f the method is that the volume rate of throughput
wn and it is assumed that the surface of the powder
ident with the surface ofthe ball charge. It follows
l . h e average throughput velocity is the quantity of
ll w dvided by the area open to flo'v'.. From this average
\ 1 iy the time for a particle to traverse the mill_m<l:y be
tuted . This treatment does not, however1 ndcate
1 11 r, for a given rate offlow, the surfaces ofthe powder
l he ball charge will coincide or not. For example,
i i clischarge by peripheral ports, it would b e expected
l , he surface of the powder charge would slop e q ui te
. !y from the entrance to the exit end. With a tru_nnion
, ''(! w mill, however, by the nature of the desgn, a
slope is precluded and there would be but a small
" IH td" to maintain flow.
'l'lt ' small amount of evidence relating to the rate of
l w tl1rough a mill which is available can again best be
l d by the use of dimen sional analysis. It would appear
~ reasonable to expect that the mass rate of throughput
() ons per hour would dep e ~d upon the diameters <;>f the
, ofthe ball and the p artcle, D, d and b respe ctv_ely,
the length of the mill, L , and the speed of rotatn,
. would also be exp ected that the throughput would
l 'nd upon the kinematic viscosity of the pulp , the
1 I:
194
CHARACTERI STICS OF
BALL, TUBE AND ROD MILLS
density of the pulp , upon the exit diameter of the mill ,
and upon the mill filling J.
When the method of dimensional analysis is applied to
these variables it is found that
....
195
MILL
"
";;: 0 111b~=+f1
..."
(n52~g2) = { (n3~gz ), (} (~), (i), (i)
(];), (J)}
Mill
0 5
03
0.2
(6.7)
(JJ
Filling . 
FIG. 6.9
The kinematic viscosity of the pulp would doubtlcss
depend upon the consistency of the pulp, the size of th~
particles in the pulp and on the cohesive properties of th
solid materials when immersed in the liquid phase. Unfortunately there is not nearly sufficient published data to
allow a complete study of this equation, but the data
Coghill and Devaney, (13) plotted in Fig. 6.9, indicate tlt
the throughput varies directly as the mill filling, J, at least
for values of filling up to 50 %  Similarly, from Fig. 6.10, i
would appear that the throughput varies directly with th~
speed of rotation, for speeds up to about 70% of th
critical. At some point above this speed the curve woulcl
be expected to become horizontal, since at the critical
speed the charge would be spread around the rnill ancl
horizontal flow would then be unlikely. Finally, it woulcl
be expected that the ratio of the average size of the passages between the grinding media to the particle size woulcl
be a controlling factor in the flow of the pulp; since if th
particles are larger than the passage dimensions there js
m echanical sieving, whilst, if the particles are relativcly
small, the flow would be limited only by the effectiv
viscosity of the pulp.
Now, from the geometry of a bed of granular materials,
the mean pore size is directly proportional to the particl
size. Thus, the ratio (dj b) would be a controlling variablc.
This group is not inconsistent with equation (6.6) sincc
(djb) = (djD) 7 (b /D).
In Fig. 6.11 are plotted the quantities of flow through a
Sr...,
;;
Q.
.c
"
"e
,_
.c
10
Q6
0 4
( N/Nc)
FIG. 6.10
04 . . .         ;     . . . . . . , . ,      ,
++.!1Nc"0 .7
06
..:
"
2
':;;
.    05
04
..."
Rod
Damrt
Rotio
r.
lnch~s
50
FrG. 6.11
100
196
CHARACT ERISTICS OF
BALL, TUBE AND ROD MILLS
rod mill, in terms of the diameter of the rods. subsidiary scale also gives the correspon ding values of (d/b)
and it is seen that the rate of flow is substantia lly ind pendent ofthe ratio of (dj b) when the ratio is greater tl1a1
about 50. This result is in accord with a finding by one "
the present authors, Rose and Barnacle, 04) working 0 11
the flow of suspensions of solid materials through pip s,
that the hydraulic resistance is independe nt of the particlc
size provided the ratio of (D jd), where D is 'the pipe di ameter and d, the particle diameter, is greater than abol
30 to 40.
It is improbab le, however, that, in practice, the ratio
(dj b) would exceed about 40 50, since this would giv
rods of excessively 1arge size in relation to tl1e particl .
Thus, it is probable that the lower values of (djb) on th~
curves of Fig. 6.11 would be relevant and, for this casc,
Qocdj b.
Thus, insofar as there are the data available for tl1
analysis, it appears that equation (6. 7) can be reduced
.. .. (6.8)
Q = KD 2.5J. ( / ,) . (djb).
From equation (6. 7) it is seen that, all other things b eing
equal, the throughpu t should vary as D 25. Unfortuna tely
data for the verificatio n of this equation are practically
nonexiste nt, but those given in Table 6.2 suggest that tl1~
observed throughpu t is roughly in accord with this expression.
TABLE 6.2
Authority
Mill si ze
Di a m .
length (in. )
Coghill and
Devaneyl3 )
Norris\9)
19 36
68 48
050
050
90 68
0 30
Q/Ql 9 . 36 .
observed
(tonsfhr)
08
08
079
28
07
1520*
* Estima ted figures.
Calculated
10
25
26
Obse vrcl
10
35
19 25
MILL
197
cffect of liner design upon mill performan ce
': s to l1ave received 1ittle attention. Clearly, the main
I n of the liner is to form a removable surface to the
! I I> dy, which may be replaced when seriously worn.
Hi so clear however, that the metal plates which serve
11 rpose may have a surface which ranges from smooth
! : which carries an intricate pattern of raised bars or
11 depressions. The merits of the various types do not
1 ' 11 : r, however, to have b een studied.
~ first step in the analysis of the problem it would be
d that liners would fall into one or the other of two
1
: l lasses:
( I ) mooth liners,
(' ) " Lifter" liners,
11 l " smooth" liners are those which have projection s
111 1 / ient to give appreciab le keying between the liner
1 I h ball charge, whilst lifter liners are those which are
l nvil y ribbed as to give rise to appreciab le interlocki ng
l .<" the balls and the liners.
: i o us common types of liners are illustrated in Fig.
I' . lthough these liners have various patterns of pr
ions, or depression s, to give an amount of interactio n
I 1 the liner and the grinding medium, it would be
>< c d that wear would round the edges. It is doubtful
i I r, after some time in service, the performan ce ofa
111 ill with these lincrs differs appreciab ly from that of a
ill with a smooth surface. Liners furnished with heavy
ll bars are also sometimes used and in such a case the
I 1 king f the ball charge to the sl1ell must be very effective.
N v heless, although a few vague general statement s to
l cct that a lifter mill gives a product with different
. 1 racteristics to that of a smooth mill have appeared,
I int does not appear to h ave been widely investigat ed .
I! tH bable, howeer, that, the grounds of differences
111 111 size characteristics of the products, ther,e exists
1 cl reason for the use of lifters in preference to the
'al smooth liners.
(d )
(Hardinge G'o. Inc. )
Wave liners
( Usines Ernile
Henrico
S .A. )
H enricot studded liners
(Hardinge Co. Inc. )
(c)
' Vedgebar liners
( 1lardin.~e
Cornpat tnt
Co. Inc.)
grate
(f)
(Edgar Allen Ltd.)
Chrome steel step
FIG. 6.12
lines
( Usines
W orn solid liners
FrG. 6.1 2
/e
1/enricot S.A.)
BALL, TUBE AND ROD MILLS
200
It is possible that, when a materia l with a low coeffici n
offricti on is milled, the charge might slip on a smooth mill
shell, with consequ ent loss of grindin g capacity , and
such a case the use oflifter bars might well be the soluti .
It has also been suggested by one ofthe authors , Rse,< )
that the use oflifter bars might elimina te the "surgin g" I
the charge sometim es encoun tered in mill operatio n.
An entirely differen t concept ion ofthe duty ofthe mill
liner underlie s the design of the studded liner develop ed by
Usines Emile Henrico t of Count St. Etienne . These lincr~,
illustra ted in Fig. 6.12 and Fig. 6.13, consist of compa
tively thin plate liners with uniform ly spaced studs on tl1<
working surface; these studs being integra l with the pla '.
Provide d the spaces between the studs are not allowed
become choked with trampi ron, etc., the studs furnish ;
good key betwee n the shell and the charge which, it iH
claimed , leads to a greater power consum ption and lo
improv ed grindin g. Further more it would appear that th
studs impose a definite geomet rical arrange ment in tht
outer layer of balls which, in turn, brings about a closc
packing , through out the ball mass, than obtains witl
conven tional types ofliner. This effect would also lead to
improv ed perform ance. Evidenc e of this effect of the studH
upon the packing of the charge appears in Fig. 6.13b, f'
the balls are clearly seen to lie in rows in the mill insteacl
of in comple tely random array.
An inciden tal merit claimed for these liners is that tlH:
high bearing pressure between the balls and the studs
the liners leads to work hardeni ng of the studs; with ;
consequ ential reducti on of the rate of metal wear.
The Henrico t liners, which have been discussed in :
paper by Belwinkel, (6) appear to be the only attemp t so
far made to influenc e the grindin g charact eristics of ;
mill by means of correctl y designe d liners. It would ther fore appear that there is some room for develop ment in
this directio n.
(a)
liners
~l ll 2 4 m . d iam. 12 m. lon g. Fitted with H enricot classifying
( Usinrs Emile Henrico S.A.)
11 0 1"
(b)
of M ill fitted wi th H enricot studded liners (underfill ed to show
ordered a rray imp osed th e ba lls.
Fro. 6.1 3
202
BALL , TUBE AND ROD MILL S
REFE RENC ES
EY, F. D. Ball Mill Grinding, U.S. u.
DEVAN
and
I . CocHILL, W .
581 .
.
Paper,
of Mines, Tech.
and CoGHILL, W.] .
2. Gow, . ., GuGGENHEIM, ., CAMPBELL, . .,
5 1.
Trans. Amer. Inst. min. (metall.) Engrs, 1934, 112,
ref. I.
3. CoGHILL, W. ., and DEVANEY, F . D. L oc. cit.,
) Engrs, 1934, 112, 91.
4. FAHRENWALD, W . Trans. Amer. l nst. min. (metall.
5, 87.
1957,3
Lond.,
Engrs,
.
hem
c
.
nstn
l
.
Trans
5. RosE, . .
6. ENGELS, . Metall, 1954, 34, 102.
ref. I.
7. CocHILL, W. ., and DEVANEY, F . D . L oc. cit.,
855.
118/1,
1924,
.,
J
Min.
Engng
.
D.
ILD,
8. FAIRCH
) Engrs, IY54, 63, 197.
9. NoRRs, G . C . Trans . Amer. l nst. min. (metall.
Engrs, 19 19, 61, 250.
.)
(metall
min.
.
Inst
10. Ds, . W. T rans. Amer.
11. ANsELM, W . ,ZementKalkGips, 1950, 3, 16.
12. PEARSO N, . . Rock Prod., 1951, 54, 106.
cit., ref. I .
13. CocHILL, W . . , and DEVANEY, F . D. L oc.
r, Ld., 1957, 203, 898.
14. R osE, . ., and BARNACLE, . . Enginee
15. RosE, . . Engineer, Lond., 1957, 203, 522.
16. BELWI NKEL, . ,ZementKalkGip.< , 1953, 6, 439.
. ,
CHAP TER 7
,' URG ING AND VIB RAT ION
MIL L
a subje ct upon
is
r, ienomenon of "surg ing" in a mill
mabl y becau se it
presu
en;
writt
een
b
has
1 l I very little
ted in mill opera tion
, ordition whic h cann ot be tolera
tion of some or
varia
by
l wl1ich must be elimi nated
cteris tics of the
chara
or
s
nsion
dime
, y of the physi cal
impo rtanc e,
such
of
is
ct
subje
The
e.
charg
11 or mill
a
from book deallt~ wv r, that it cann ot well be omitt ed
An attem pt will
mill.
ling
, wh the subje ct of the tumb
a revie w of the
ter,
chap
this
in
l  re be made to give,
ct.
subje
the
of
' : nt knowledge
a
phen omen on know n as "surg ing" consists of
'
mill,
ofthe
e
charg
e
whol
llumlike oscill ation ofthe
al
the mill shell. Thus for one part of the cyclic
mill
ofthe
e
centr
the
d
aroun
ng
movi
is
e
charg
l the
and durin g the
l same direc tion as the mill shell,
direc tion. As a
site
oppo
the
1 d part of the motio n, in
charg e, the
the
of
n
motio
atory
oscill
H guence of this
in stead y
shell
mill
the
tain
main
to
sary
neces
mstan ces,
circu
in
certa
t1 fluctu ates wide ly and, in
r that the
moto
ng
drivi
the
' nake such dema nds upon
of the
out
ing
throw
This
out.
n
throw
are
11 t brea kers
er
dang
the
evil since
1 break ers is, however, a lesser
more
died.
reme
be
such a case, appa rent and must
torqu e whic h are
It dang er is hidden in fluctu ations of
devices into
ctive
prote
t
crcui
the
bring
ffi ient to
existence of such fluctu ation s
1 <ti . In this case the
ce of fluctu ation s may
appa rent, and the main tenan
ng geari ng throu gh
drivi
the
of
n
. l to the destr uctio
1 l 11
203
204
S U RGING AND VIBRATION MILL
BALL, T U BE AND ROD MILLS
fatigue failure. Although, for obvious reasons, bu t littlc ~
m ade pu blic concerning such failures, it is airly well know 1
that failue of the gearing of large mills is e ncounten l
with a frequ ency such as to suggest that factors ther th
mere wear and tear are operative.
The oscillatory motion of the charge, which has b c '
described above, is complicated by the collapse of t!<
leading toe of the charge when the charge comes to r ~ ~
at the end of the forward motion. This collapse impos\~
fluctuations of small amplitude but higher frequency up 11
the fluctuations arising from the m ain motion; this, i11
turn, gives rise to a complex motion which involves :
great number of harmonics in the frequency spectrum.
Consider now the motion of the charge in some d eLBil
and, since, in Chapter 2, it has been shown that t iH
equilibrium surface of the charge approximates to :
straight line, the charge may be considered to be a m ass
the shape shown in Fig. 7.1. The following m athemati l
treatment, which has been suggested by Hinsley( l), is th
applicable.
The charge in the mill is assumed to b e a coherent mass
with the centre of gravity at G. The gravitational force 11.f!,
and the centripetal forc e mr 2 have a resultant R whi !
meets the shell at . This resultant can, in turn, b e resolv !
into tangential and normal components at .
Rs.
Fa .
+ g sin )
.. .. (7.1)
since is small.
Also, from the equations of motion
mk 2
mg
k2
FR
gr
= k2 cos  mk2
gy'(r2+2R2 cos2 )
.
k2
(sin cos  cos sn )
k2
gives
. . ~ . (7 .3)
1 l ' '
or
.. .. (7.2)
"' I /( cos {j2
l i l
cos  FR
= gr cos  Rg cos sin  k rB 2 cos
k2
~ mR cos ( r 2
7.1
( :obining (7.1 ) and (7.2)
Then,
FR = mR (r2 cos + g sin )
205
rR
cos
= __k_2_ _
and
206
SU RGING AND VIBRATION
BALL, TUBE AND ROD MILLS
r ' tme,
Let =, so 0=  and = , so that
(7.4)
 2 =  2 sin
This equation can be integrated once by use of th:
integrating factor 2e2, whence
207
MILL
t, ofthe descent ofthe charge is given by
''' d
J,,,,
2
) , y [{2A 2/y( l +4 1 2 )} sin ( +) +C 1 e ]
(7 .6)
l :quation may be solved by graphical and numerical
l t llion.
.2
. (
22
y(l + 4 12 ) sn +) +ce2 1
(7.5)
where and 2 1 are related as shown in Fig. (7.2).
FIG.
7.2
FIG. 7.3
If now the angle , at which the charge just slips,
observed by experiment and since, for this condition ,
= the value of C1 the constant of integration, c, may b
obtained from equation (7.5). Since equation (7.5) is :
quadratic . there are two solutions, 1 and 2 , to th
equation 2 = which, in practice, are one negative an(l
the other positive.
The expression (7.5) is the equation for the velocity
motion of the charge and the solution can be taken no
further by formal methods. For practical purposes, however, it is desirable to know the time for a complct
oscillation, since the frequency of the oscillation may thc
be calculated thus.
W l1 the function on the righthand side of equation
( /, ()) is plotted, a curve of the form shown in Fig. (7.3)
1 i JLc ined. The portion of the curve between b and c may
111 cgrated by Weddles' rule, or any similar method,
1 owing to the infinite values, the strips above ab and cd
, 1 t be so treated.
nverse sine function is therefore fitted to these end
I 1JS so tha t
_ [ . _1 / ]
dx
P
. . . . (7. 7)
 Slll
y(22)
by the correct choice of , , and a very close fit
208
S URG ING AND VIBRATION MILL
BALL, TUBE AND ROD MILLS
may be obtained. Thus the time of descent of the load m: y
be computed.
After the load has gripped the shell, the ascent is m<I
at the same speed as the shell. Thus the time of asccn i
given by ( 1  2 )j ; where is the angular velocity
the mill shell.
By the above treatment the frequency of oscillation l
the charge may be calculated, provided that certain i
formation, which must be obtained by observation tl
rnill, is available. Furthermore, the treatment is based
t~e ~ssumption that the coefficients of static and dyna n1i
fnct10n have constant, although different, values; bu
fact it wou1d be expected that the coefficient of fricio
would depend on the slip ve1ocity.
Experiments by Hinsley on a mill of6 ft diameter wil
40% filling, show tha: s c:::035, dc:=:025, that there :
lag of about 0015 sec before the load "locks" on to l
mill shell, and that the frequency of oscillation is ab
58 per minute. On the basis of the above ca1culations i
following results were obtained.
Time for ascent
0233 sec
Time for descent = 787 sec
Time for locking* = 0015 sec
Time for cycle
= 1035 sec
This gives a frequency of 1/1035 60 =58 cfmin, whirl
is in close agreement with the observed value.
The above treatment is of great interest in that it giv 'N
the frequency of surge with considerable accuracy, bu
is not of such a nature as to decid e, whilst the machin~ iM
still on the drawing board, the question ofwhether a mill
of proposed design will surge or not.
a theoretical study of the problem, one of the pres 11
writers, Rose, <2 > attempted to analyse the motion of tiH
mill charge by use of the ma thema tica1 method of topo1ogy ;
* Determined by use of an oscillograph.
209
. i s of the coefficient of friction with rubbing
I I( i bcing included in this treatment. the basis of
l11 logical diagram relating to a mill of about 8 ft
li ' l 'r, together with an assumed but probably reasonI I l t ionship between the coefficient offriction and the
I i i g speed, the curve relating the torque to rotate the
I
II l
'll at constant speed, on a base of time, was
cm
< l ;nd this curve is showr_ in Fig. 7.4. From this curv:e
that violent surges In the torque occur; a
of about
25 times the m ean torque being attained
ma
' 11
11 , ' quency of about 30 cfmin. This ratio of maximum
25
20
OLJ
Timt
F.
7.4
I to mean torque is in fair accord with an observa1 11 1 l>y Hinsley, working the mill mentioned previously,
l .l during surging the current to the motor varied be about 300 and 700 amperes.
ll ;v ing regard to the theoretical nature of the treat f Rose, which involved a number of assumptions as
11 values of important parameters, the agreement
l11 w the observed and calculated magnitude of the
1 11 t surge is as close as can be expected. t appears
. Ho nable to conclude that peak values of the torque of
11 l ;s t two or three times the mean va1ue may be en1 t red during surging.
'/1 topological method is, however, not of a nature
210
such as to be readily used i the desig office. a effo
to obtai a simple criterio for the existece, or 11
existece, of surgig, a experimetal ivestigatio w:
carried out by e of the preset authors with a cowork,
Rose ad Blut<3) use beig made of the smalls ~tl
models ad the t~st apparatus already illustrated
Fig. 3.1.
.
The basis ofthis work is that the agle 2 through wh
the cetre of gravity of the mill charge oscates is ?ep !
det the "fluidity" of the charge ad this quatlty, 101'
reasos already discussed, depeds upo the mill diamct,
D the ball diameter, d, upo the coefficiet of frictio ,
of the mill charge (if = the mill char~~ would alw~yH
be i stable equilibrium at the lowest st ofthe mll)
ad upo the chage of the value of the coefficiet of
frictio with the slidigvelocity. Thus if v=(1 +v),
where deotes some fuctio, the amplitude woul<l
deped 0 , ad .
The amplitude would also be expected to deped ';! ."
the speed of rotation of the mill, , ad upo the penodc
time t ofthe oscillatio ofthe mill charge as a pedulum .
This' ~ould appear to be reasoable sice i~ ad t a1
such that the mill makes e or l f n revolutos, wher
is a iteger, durig the time that the charg.e makes onc
complete cycle, the ay roughess of t~e mll, such as :
fillig hatch, would produce forces whch are sy~ch~o
ized with the oscillatio of the charge ad the oscllato H
would build up. If, however, sychroizatio is
preset, then oscillations would be unlikely to build u .
Fially, the amplitude would be expected t<:> depenl
upon the mill filling, J, since bot~ ~he restong fo~cc
actig on the mill charge ad the fnctal forces, wh l1
depend the number of cotact points, depend up 11
this quatity.
Thus
.. .. (7.8)
= (D, d, , t, , , J)
where
SU RGING AND VIBRATION MILL
BALL, TUBE AND ROD 'MILLS
agai deotes
"some function of".
/ IIOW the methods of dimesional aalysis
, I ~ variables it is foud that
{ (D fd) , (Nt), (), (J), (BDN)}
211
are applied
.... (7.9)
11 : ralysis may be carried further thus;
I ,cl be some fraction of the critical speed , of the
110 tha t =kN,. Now it is easily shown that, for a
rill fillig, the periodic time of oscillation of the ball
I l 1'1\(' varies directly as D and also that the critic~l
I" cl , varies as y' ( l /D ) ad so the value of (Nt ) s
Iit > ndet ofthe size ofthe mill, thus it follows that, as
his group is cocered, the results of the model
directly applicable to the fullsize mill.
' rg attetio now to the group (BDN) it follows
l,/ , :s before Nocy'(1/D) so this group becomes (By'D).
l , t follows that there is some "scale effect", the value
l i ; group icreasing as the root of the mill diameter.
1\ r the other hand, the value of is usually smll ad
/ variations in the value of the group would be
1 1 c tcd to be small. Furthermore, the effects risig from
l locity dependece ofthe coefficient offrictio can be,
ne extet at least, absorbed by the
1w the coefficients of static ad
I l , t would be expected that, in ay
use of the average
dynamic friction.
practical mill, the
I I t f this group will t be large.
' lu s, the basic equation for surgig i
11 , 1 ast for a first approximation,
{(D fd) , (), (J)}
a mill reduced
.... (7.10)
accordance with this equation, tests were carried
111 1 n stablish how the existence, or otherwise, of surging
1 / 1H'nds the dimensionless quatities on the righthad
11 < the equation.
l as already been shown, i Chapters 2 and 3, that
' " (I uidity of the charge decreases with a icreasig
\) r of cotact points ad with icreasig coefficiet
l tion . Thus, the mill fillig, J, ad the coefficient
of the ch arge iH
of friction, , are, as far as the fluid ity
ld be expe ct cl
con cern ed inter chan geab le, and it wou
f~rm of tl~
the
in
that these' quan titie s wou ld o~cur
mental r:
expe
the
of
yss
prod uct J . Then , from anal
across tl1
n
draw
be
d
coul
line
a
that
sults it was foun d
nate a<l
ordi
th.e
field f a grap h, Fig. 7.5, whi~h (!) iswhc
of ~IH
part
~
(D fd) the abscissa, and accordng nto
nf.
surg
r
et~e
fell,
ts
poin
g
ati~
field of the grap h the oper
lsh wheth<:estab
to
r
orde
,
Thus
r.
occu
does or does not
O IS
0 10
::,
>(
....,
1\\
005
Non Surg inq
R:qion .
""
Surg1ng R:gion .
20
213
MILL
S URG I NG AND VIBR ATIO N
S
BAL L, TUB E AND ROD MILL
212
oper ating poin t
lll be varie d in orde r to brin g the
"non surging" zone .
of the
oing treat men t involves a know ledg e
oreg
'IH
ion of
frict
of
ts
icien
coeff
'"' the stati c and dyna mic
henc e
and
,
essed
proc
g
bein
rial
mate
l11 l: ll char ge and
lly.
idua
indiv
ts
1l static and dyna mic coefficien
e
ofth
es
valu
al
reve
not
r a ti ofth e liter atur e does
to
le
icab
appl
on
fricti
mic
dyna
and
1 111 111. nts of stati c
ofpo wde r. How of balls load ed with vario us types
ofth ese
auth ors have mad e roug h dete rmin ation s
1 1 , 111
icab le
appl
be
to
y
likel
are
l ~ unde r cond itions whic h
in an
ed
plac
were
der
pow
and
balls
I I. 11 mills; that is,
the
and
ce
surfa
 1 > ttom ed box on a smo oth steel
ns
mea
by
d
sure
mea
i v ' force necessary to mov e.the box
give
to
ved
belie
is
od
meth
l , sprin g bala nce. This
for the pres ent purp ose and a
11 l  of suffi cient accu racy
purp oses ofill ustra111  ofsu ch results are show n, for
ll, Tab le 7.1.
ll he
TABL E 7.1
I' 
4.0

150
l1re
80
Rat io Dl,
FIG. 7.5
necessary . t
surg ing is likely to occu r or not it is only of the mll,
ng
fill
osed
prop
the
J,
establish the prod uct of
the bal~ charl\< '
and , the effective coefficient offri ction of
plot ths a l
to
and material unde rgoin g milli ng, and
If the opcr; I
d.
/
D
ratio
agai nst the prop osed valu e of the
ing". z
surg
'_'non
.the
e
insid
well
ing poin t so obta ined is
[ tl
but
e
surg
then it is impr obab le that the mll wll
t.
near
sly
erou
dang
poin t is inside the "sur ging " zone, or
:.
s
the
and
r.
occu
to
y
likel
is
boun dary line, then surg ing
g, or bot ,
filln
mill
the
of
e
nitud
mag
of the balls or the
of
11 ~ ; system
10
11
I
Coeff icient of
dynam ic frictio n
Mean value
of coeffic ient
of frictio n
018
018
018
025
018
02 1
037
30
033
042
04 1
04 1
044
042
043
/ .rteel:
l l 1 t d with
111 y denum
l l ~ l l1id e
l1 i cd
"'
with
a nd
l ln l
Coeff icient of
static frictio n
r<"c
silica
I" w l rr
\ /lce
" wl
emery
SU RGING AND VIBRATION
BALL, TUBE AND ROD MILLS
214
It might be of interest to note in passihg that tl1
dependence of surging upon the coefficient of friction, :.
revealed by model tests, is as shown in Table 7.2, and th;l
215
MILL
I
.;
g
Observation
Conditions in mill
MillCieaned and dried
MillOiled
Mill Treated with molybdenum disulphide
MiiiWith about 1% of voidage filled with
silica powder
Balls and Mill With about 1% of voidage filled with
emery
Balls
Balls
Balls
Balls
Surging
Surging
Surging
and
and
and
and
.z=
:;
.!!!
Q.
...
..."' ...."' g~
 .z= "'
occur
occu
occu"
surging
I surging
1
this observation is in accord with experience, sincc it
is unusual, though not impossible, for surging to arisc i
mills grinding cement clinker, limestone or such simil :
materials which give rise to a high coefficient of frictin .
Confirmation of the general accuracy of this criterion i
contained in Fig. 7.6, in which the operating points for
large number of mills are plotted; these points being :\
culated from the data of Taggart(4) and relating to mill
between about 4ft and 10! ft diameter. Additional d:t
relating to mills between 2 ft and 8 ft diameter grindi
molybdenum disulphide and graphite are included; th 1
being supplied by Messrs. Acheson Colloids Ltd.
The data have been reduced on the basis that the cofl
cient of friction is, for the metalliferous ore, 04; a fig
which is reasonable for such materials. In fact, the d:t
also refer to mills fitted with various types of lin ,
"Shiplap", "Wave", etc., but it is probable that aJI of
these liners would approximate to "smoath" after s
wear has taken place and so, for safety, these mills should
operated under "nonsurging" conditions. Examinatio
of Fig. 7.6 shows that, although some closely appr :
the boundary, all the points, with one exception, t:ll
within the nonsurging zone. Thus, there is fair, thoul
...
.,::;,
TABLE 7.2
Q.
>
"' ()...
~
()
~
e
.,
48
v
/
()
 .., .
...v .
.:
11)
...,.
1/)
216
SU RGING AND VIBRAiON MILL
BALL, TUBE AND ROD MILLS
not abso1ute1y conc1usive, evidence that the resu1ts of tl~
smallsca1e tests app1y direct1y to 1arge mills,_and S? _tl:l
the effect of variations of the group (BDN) s neg1g1bl ,
Strict1y, the above criterion is app1ic~b1e <_>n~y for val
of the ratio (/ ,) greater than ~15 snce t s found t!:~
surging does not occur fo~ :un11:ng_ speeds ~ess than l
va1ue. This further condtn s, practce how ,
trivia1 since industria1 mills are rarely, if ever, operatl
.
.
at such 1ow speeds.
In the derivation of the curve of Fg. 7.5, the exst  ,
or nonexistence, of surging was determined by obscrv:
tion of the sound of the mill and of the fluctuatns of tl
dynamometer arm. It is improbab1e, therefo:e, that ~
surging was completely ~uppressed; although t was so I
practical purposes. It s probable, however, that t
boundary line on Fig. 7.5 corresponds to a small fluct:
tion oftorque; say 5% or 10% ofthe mean value. It woull ,
therefore, appear that further lines. could be. added to t I
graph, as shown in Fig. 7.7, in whch each lne represc.l
a fluctuation in the driving torque, equal to some defint
percentage of the mean value. If this i~ a tru~ pic_turc c l
the case, then it wou1d appear _that a gven mll mght IH
subject to considerable fluctuatns oftorque, eve~ tho~l
the gross manifestations of the occurrence of surgng, s
as the throwingout of the circuit breakers, are not at <
..
.
apparent.
From the tentative graph, Fg. 7.7, t s clear that sm:ll
changes of operating conditio~s cou1d bring a~out ap] c
ciable changes in the fluctuatn of torque. It s proballc
that it is these minor differ~nces in, for examp1e, t
coefficient of friction of the charge, which give rise to t
often inexp1icab1e failures which occ_ur in the gearing
some mills, whilst identica1 mills working under appar tI
identica1 conditions suffer no such breakdowns. As :
example of this the attention of the writ.ers ~as b
drawn Anon. (5) to the case ofa number ofdentcal t\C
mills, ~orkini under app~rently ide~tica_1 conditions,
one of which mills the man gear par fa1ed after abol
217
111 ,!10(} hours' service; whereas the other mills have given
l tcly satisfactory performance over a number of
< . T he reduction gearing of these mills, which were
l rc d and constructed by makers experienced in the
tion of gear trains such as those used in rolling
11 :nd in ship turbine installations, were generously
1" ''1 ed and apparently adequate for the heaviest
ll . r now it is assumed that the fundamental fluctua1 torque is of the order of 30 cfmin, which the tests
l 11 i ley and the calculations of Rose indicate as being
I'' l : y reasonable, and that the reduction ratio of the
..,
FrG. 7.7
l speed pair is 5 to 1, then 10,000 hours of service
t tl l correspond to about 3,000,000 cycles of stress on
1111 1{~ aring. It is significant that it is at this order ofnuml of stress cycles that fatigue failure would be expected
ur.
11 , i the absence of an exact knowledge of the com1"' i ion of the gear material and the heat treatment em~
i y d, it is assumed that the fatigue characteristics of the
, naterial are approximately the same as those for
1 l i : l1 the diagram Fig. 7.8, quoted by Morley,<6> was
l iv d, then this dia?;ram may be used for the further
c
SU RGIN G AND VIBR ATIO N
BALL , TUBE AND ROD MILL S
218
Fig. 7.I\
analysis of the prese nt probl em. The diagr am of
befo
sals
rever
of
er
numb
inite
is such that, for an indef
shac!l
the
n
withi
fall
must
es
stress
ing
failure, the work
area.
in~
Now suppose that the mini mum stress, correspond ,
thrl
s
follow
It
.
in2
tonsf
20
is
e,
torqu
to the mean drivi ng
er
if failure is not to occur withi n a limit ed numb not
must
stress
appli catio ns of the stress, the maxi mum
atio
exceed abou t 25 tonsfin 2; that is, the allow able fiuctu
al ,
mean
the
of
25%
t
abou
d
excee
of torqu e must not
7.7, tiHt
Howe ver, it has alrea dy been suggested by Fig.
20r~~~~~~
~ ...
~ ~
~
c
~
,_
Of;~~M~>L::.__
____ ___

0 1t~~_.:....
Vl
0..
U O'L~F. 7.8
qt
fiuctu ation s of this magn itude may well ase from
J 11 11
say,
by,
bly
possi
,
mode rate chang es in mill conditions
on of tl
varia tion in the value of the coefficient of fricti
brou~lt
be
well
could
es
chang
such
that
charg e, and
surfi
size,
le
abou t by differences in pulp density, partic
as
small
so
are
h
whic
finish ofthe balls and mill, etc.,
ap
d
woul
it
,
Thus
nt.
porta
unim
norm ally regar ded as
practi
in
ed
unter
enco
es
failur
gear
ed
plain
that the unex
tio;
migh t arise from differences in the plysical condi
iN,
terist
charac
er
powd
and
e
the mill liner, ball charg
bu l,
nt
porta
unim
as
ded
regar
be
to
as
Thes e are so small
atio N
in fact, have a profo und effect on the torqu e fiuctu
furtl
still
is
em
trans mitte d to the geari ng. The probl
MILL
219
ated by the fact that the stresses invol ved will
upon such factors as the rigid ity ofthe drive , the
moto r, etc. Thus , for
SJ eed chara cteris tics of the
are such that the mill
tics
cteris
, if the moto r chara
1
e increases, the load
torqu
the
as
speed
lose
l l w d to
the moto r main tains
I : gcari ng will be lowe r than if
const ant. For these
less
or
d of rotat ion more
e fiuctu ation s are
torqu
the
to
as
s
1 1 t s xact calcu lation
nable doub t
reaso
no
ins
rema
there
but
e,
11 1 cabl
n the "
withi
/ mill shoul d be opera ted as far
ble.
possi
is
f ~j" zone ofFig . 7.5 as
to the
11 l as alrea dy been point ed out that, owin g
n
withi
e
charg
the
of
n
motio
ng
surgi
any
of
l ity
the
e,
charg
the
of
toe
/ ll and to the collapse of the
d by the drivi ng
l ; s of the torqu e trans mitte
er ofhar moni cs.
numb
large
very
a
in
conta
1 /1 :1nism will
r at
, herefore, possible that reson ance migh t occu
ratus
appa
y
nearb
and
111t ' requency, betw een a mill
sensitive to vibra / , if the adjac ent equip ment were
origi n of whic h
the
,
ulties
diffic
to
rise
give
l , ould
natur e have
this
of
l , l not be easily trace d. Probl ems
whic h it is
in
,<7>
Anon.
e,
l liscussed in a recen t articl
mitte d
trans
be
may
tion
vibra
e
l 1w that appre ciabl
tras
being
tion
vibra
the
;
ation
found
piled
l ~h a
bedthe
gh
throu
mill,
< l down the piles bene ath the
the
ath
bene
piles
other
gh
1 /1 :nd then emer ging throu
exbe
not
d
woul
t
effec
This
.
ment
111 11\ l bouring equip
there
since
1 I < l to be so mark ed with friction piles,
trans mit the vibra l / then be no unde rlyin g rock to
duce consi derab le
intro
d
woul
<. nd the clay itself
that the corre ct
ar
appe
d
woul
it
s,
heles
I 1ing . evert
point of the trans / 1 /\ 11 of found ation s from the view
ir of vibra tion, is essential.
that some vibra tion
1 he same articl e, it is ment ioned
was suggested that
it
and
ng
geari
mill
the
from
. ; ted
use of fibre or
the
nditio n could be allev iated by
/
view of
train.
gear
the
in
. / tooth ed drivi ng pinio ns
howe ver,
ter,
chap
this
of
part
r
earlie
11 liscussion in the
l
l I" l
220
BALL , TUBE AND ROD MILL S
n th l\
it appe ars that atten tion to the cond itions withi
cati o H
i
modif
such
than
ive
shell migh t be moe effect
wll l
tion
viba
of
type
er
furth
the gear tain itself.
~ ll
mill
migh t well be ment ioned is the vibra tion of the
l
J>o
e.
charg
g
fallin
the
as a b eam, unde r the influe nce of
)
l
cl
b
absor
is
tion
vibra
this
ably the great er part of
11 I
etwe
b
oints
j
the
shell,
the
of
etal
m
damp ing in the
e ap
liner plates, etc., and, in a b all or tube mill the
i
ures
meas
al
speci
to be no n eed to intro duce any
ocl
idual
indiv
the
ver,
howe
mill,
suppr essio n. ln a rod
wil l
ct
conta
make
they
when
ct
impa
nt
viole
give a more
cm
the liner or the toe ofthe charge. It appea rs to be
: I
liners
the
een
betw
wood
pract ice to inser t billets of
l c
an<
shock
the
uate
atten
to
order
in
shell of such mills
.intro duce some damp ing.
fa lli
The p articu larly d estru ctive actio n ofhea vy rods
l
rtan
impo
the
ver,
howe
s,
on t~e milllin ~ rs emph asize
c
g<'
impin
rods
the
that
such
speed
a
at
runng rod mlls
.
the toe of the charge and not upon the liners
J.
ion to the Autho rs.
4.
5.
6.
7.
Instn mech . Engrs Lond. 170 1
ME TAL WE AR
MIL LS
from the work 
1111 <: st ofrep lacem ent ofthe meta l worn
of a mill often repre sents
1 :aces of tl1e b::lls and liners
of opera tion of the plant
cost
the
of
n
fract
ble
dera
N
rtanc e. In spite of
t l ~ is ofsom e r eal econo mica l impo
of infor matio n
nt
amou
the
ver,
.portance, howe
l
and, unfor tusmall
ingly
amaz
is
ct
subje
this
i l :b l e on
re
ofmo than a very
l l y_, is insuffi~ient to form the basis
far as is know n the
As
lem.
111: t ve analyss of the prob
etal wear is that of
m
for
y
theor
a
e
l y <ltempt to deriv
the relev ant parts
and
Rose,
rs,
autho
nt
prese
11 1 the
y
111 l treat ment will b e outli ned briefl here.
is that parti cles
I'11 basic idea unde rlying this treat ment
conta ct be
upon
will,
g
millin
g
rgoin
f I I : mate rial unde
nds
I v into the_ surfa ce of the b alls at a rate whic h dep~
the
cle,
parti
and
ball
c ~he relatve hard nesses of the
REFE RENC ES
F . Privat e Comm unicat
. HNSLEY,
2. RosE, . . Unpub lished R esearch.
3. R osE, . ., and BLUNT, G. D. Proc.
CHAP TER 8
'} '
'
'
1957.
~)
(New
Wiley
n
ng,Joh
Dressi
l
TAGGART, . F. Handbook of Minera
1948.
s.
Privat e Comm unicat ion to th e Author
.
reen & Co. (L l ),
MORLEY, . Strength of Materials, Longm ans, G
17.
68,4
1953,
Lond.,
Age,
. Chem .
other such factors.
, lty of powd er prese nt and upon
relati ve tangential
is
there
ded
provi
that,
ed
:l so assum
or other of the em ion betw een a pair of balls, one
Fig.
d partic les will,_ upon maki ng conta ct, as in
that
usion
concl
the
to
leads
then
11 , b e torn out. Ths
l surfaces at a rate
l c >artic les will be torn from the meta
ball and
I i is a funct ion of the relati ve hardn ess of the
relati ve
the
,
dded
embe
les
i le, of the numb er of partic
rs.
facto
ar
simil
I Ii g velocity, and
val of partic les
\V Ien the rates of embe dding and remo
er of partic les
numb
ibriu m
11 cq ual ther e will b e an equil
word s a cerother
in
or,
ces
I ><' lded in the m etal surfa
'
.
"arm oured "
be
will
ball
each
of
ce
surfa
the
1 11 ortn of
now, an embe dded partic le
1 I mbe dded partic les. If
part ofthe surfa ce
J ball trave rses the "una rmou red"
l c l l
22 1
222
BAL:L, TUBE AND ROD MILLS
of a neighbouring one, as in Fig. 8.1b, then metal w ,
will take place. ln the work cited the calculation of I
wear has, to a first approximation, been carried out l
it is suggested that the relationship between rate of w
and relative hardness of ball and particle has th i
sl;own in Fig. 8.2, whilst the relationship between 1 l
()
223
MILLS
iably greater hardness and so ~he _reduction <?f
1 ly .increasing the hardness of the gnndng surface s
" Hsible . ln such a case, however, the rate of w~ar can
l l d by the use ofery soft metal (by reducng )
l arrying the operating poin~ ~ell to the left <?f the
1 um of curve Fig. 8.2. Ths mprovement s, of
brought about by the increased "armouring" of
1 :race of the balls by the particles. Conversely when
1 1 y s ft material is being ground, the ratio will be
, (( than unity, even with soft metals, and so the
, ing point is the righthand side of the peak.
i ' s reduced wear is obtained by increasing the hard'" ol' the metal.
(b)
F c .
METAL WEAR
8.1
>
.._ 5 1 +      " t   +       i
10~+~+i
":"
Relatir
F c.
(VJ
FIG. 8.3
Hardnrss 
8.2
wear and quantity of powder in the mill has the for11
Fig. 8.3.
Although the derivation of these curves is compl t l
theoretical there is some qualitative evidence in su
oftheir validity. Thus for example, it is clear from fj g, 11.',
that ifa material is being ground by balls ofapproxim : l
eq ual hardness (= 1), then the ball wear may be re I I
either by increasing or by decreasing the value of 11. 11,
however, a hard mineral, say of Brinell Number ~lolll
500 to 600, is being milled, it is not practicable to us .\1
't' l <~sc conclusions are confirmed by the practical ob' ;1 ns of Engels(2) that for the grinding of soft materials
1 /l mill body andoalls should be used, but for grinding
111 ma terials a soft bu t tough metal should be used for
l 1 inding surfaces. Similarly, the c~r~e of Fig. 8.3, is
l med by Smith, (3) who states that t s well known to
op rators that a impr0~ed fineness, in_a given t~m~,
t t r.ined when the quantty of powder the mll s
l . d, but that this improvement is obtained at the
I " ' lS of increased iron wear.
I rder to analyse the various published data, advan nay well be taken of the simplification offered by thc
META L WEAR
BALL , TUBE AND ROD MILLS
224
metho d of dimen sional analysis and this will be uscd
basis of the subseq uent treatm ent.
of n1 , l ,
t would be expec ted that the total wear
. ll
ss
hardne
e
W 1 lb, would depen d upon the rel~tiv
.
pa~t
and
ball
mll,
of
size
the
and partic le, , upon
I,
m1ll!
the.
of
length
the
upon
tively,
respec
b
D, d and
I
y
denst
upon the speed of the mill, , and upon the
ball mater ial, . It would also be expec ted to depen d
the energy necessary to increase the surface or I
rr{~tal by unit amoun t, upon the mill filling, J, upon t
fractio nal volum e of the powde r, V, upon the pulp
sistency, , upon gravit y, g, and upon the time ofmillin ~o~,
This may be writte n symbo lically as
W = {D, d, b, L, , Em, , , g, t, J, V, U} .... (11. )
225
MILLS
the rate of ball wear, insofar as these
on the speed of rotatio n of the mill .
< 06, the
I 1 ; l so found that for slow speeds, ( I Nc)
able
reason
with
ed
onal relatio nship can be replac
,, a mill and
, cs depen d
and, after applyi ng the usual metho ds of
analysis, this leads to
Sccific
dimen s o l
(~~) = { (~), (i), (i), (~), (;;~g)' (D;
(J), ( V), ( U)}
4 t+r
Ball
War
( // ),
r.
. . . . (11. 'I)
(Nf ,),
First exami ne the functi onal relatio nship
5 > relatin g to
<
Gross,
and
Data by Coghi ll and Devaney<3J
metal wear, are plotte d in F~g. 8.4, and on the same gr:l
the curve relatin g the rate of produ ction of specific s
face to the group (N fNc), taken from Fig. 5.17, is sl1ow
for compa rison. It is at once seen that there is clo
agreem ent betwe en the rate of produ ction of specific s
Cross
>
(11. ~)
() .z(~~g) 3(~) 4(~) 5(i)
Data Of
Coqhill And Dvany
)'
2
This can be simplified, however: since (D jg) is,
2
equati on (3.7), equiva lent to (Nf Nc) .
Surfacc .
.
8
@
..
~ 0 21i"tt1
.
(~~)
lncrcas
Ratc Of
0 2
04
0 6
(N/Nc)
FIG. 8.4
racy by 1 (Nf Nc) =K(N fNc)l .30
J ' ' ds, ( fN c) > 06 by
(/c)
and for the highe r
= K(NfN c)OB ,
speeds
' decrea se in the rate of ball wear at higher
tangen tial
ld be expec ted since, whilst decrea sing the
ncy ~nd
freque
the
se
increa
t , the catara cting would
ng.
armour
te
promo
would
which
ts
tsity of impac
\ t l1 e centrif uging condit ion is appro ached the numb er
, l l>alls in action decreases, which again would reduce
227
e has a shar ply defi ned
ll .' :.nd it is seen that the curv
exp lana tion of the form of this
111 Jrnu m. The prob able
becomes less effective but
<: is that the "arm ouri ng"
easi ng part icle size unti l
incr
l :bra~ion in~reases with
r this, even thou gh the
~ !'' ' max~m~~ s . re~ched . .Aftethe
part icle s sepa rate the
nounng IS still neffectve,
bear ings , and so,
ball
like
. l. surfac~s, ac.ting rath er
wea r of the
sive
abra
the
ce
redu
l ncreasing sze,
righ tha nd
the
that , for
, 1. It shou ld be noti ced
e, wea r
rang
l
tica
prac
I . h of th~ curve, :hich is the
dire ctly
,
thus
or,
/D)
(b
as
y
:s approxmately nversel
ely, for
vers
Con
.
eter
diam
mill
the
(D/b), and so :ith
alm ost
es
vari
r
wea
the
l I fth and portn of the curv e
MET AL WEA R MIL LS
226
LS
BAL L, TUB E AND ROD MIL
8.4 app ears to I J ' l
ball wea r. Thu s the curv e of Fig.
reas ona ble form.
poin ts and
The close agre eme nt betw een the
met al wea r per I
curv e of Fig. 8.4 suggests that the
app roxi mat ely l
incr ease in surface atta ined shou ld be
y ofC hap ter : l 11
stud
.
pen den t ofth e spee d ofth e mill
for speeds of 11
and
size
n
give
of
shows that , for a mill
er inpu t to a mill '"
abo ut 70% of the critical, the pow
. Aga in insofar a~
prop orti ona l to the spee d of rota tion
a stra ight linc
by
aced
curv e of Fig. 8.4 may be repl
y gen eral ly accll
this rang e, it also leads to the fairl
prop orti ona l to tl
conclusion that the rate of wea r is
ears that th . ll
app
pow er abso rbed in a mill. Thu s it
rate ofpro<l
both
to
l
ona
orti
wea r is, very roug hly, prop
mill. This ; l
the
to
t
inpu
er
pow
to
tion of surf ace and
of surface is rou i ll y
implies that the rate of prod ucti on
also been dedH'I'l
has
ch
prop orti ona l to pow er inpu t; whi
that to atteml t
,
ever
how
ed,
in Cha pter 6. It is cons ider
to pow er inpu t
calc ulat e the met al wea r by refe renc e
the calcultl 111
that
surface incr ease is artificial and
lving its w
invo
lem
prob
shou ld be trea ted as a sepa rate
ers.
met
para
mill
the
rela tion ship s betw een
obvious is 6 (/,fl)
The nex t rela tion ship whi ch is fairly
dire ctly as
vary
t
since the tota l rate of ball wea r mus
aini ng
rem
es
ntiti
qua
r
leng th of the mill; all othe
alte red.
2
it wou ld ap
Tur ning now to the grou p (Em fD g)
mus t tak l
ship
tion
that , on logical grou nds, the rela
form
. . . . (11.1 )
the tota l wea r, W, ,
since, all othe r thin gs bein g equ al,
valu e of Em. U
the
to
mus t be inve rsel y prop orti ona l
whi ch this r l
nst
agai
ence
evid
fort una tely ther e is no
tion ship can be teste d directly.
ther e are limit
Tur ning nex t to the func tion 5 ( b/D) est the gcn r
sugg
ch
whi
rele van t data by Mor tsell (6)
are plot ted in I'i ,
form of this rela tion ship . The se data
~{] ?1
0
< >
FIG. 8.5
)10. The left han d
. tly . as (b /D ); or thus as (D/b
ding a fine pow der and
. ~h s clea rly for larg e balls grin
s with coarse mat eria l
l ngh tha nd side is for sma ll ball
nt.
hard ness on the
( : nsid erin g nex t the effect of rela tive has been preship
tion
rela
al
retic
theo
. c of _ball .wea r; a
king of this rela tion '. ' d Fg. 8.2. Dat a for the chec
but a few values of
ing,
l are alm ost com plet ely lack
es of ball , pub ness
hard
ell
Brin
I. 11 wea r, for diff eren t
Fig. 8.6. This curv e
l l d by McLeod<7J are plot ted in
definite corr elat ion
l ows the gen eral form of Fig. 8.2, but
eral s wer e mill ed
min
ed
1 n t be mad e sinc e very mix
is unk now n.
ness
hard
eral
min
: l t~e corr espo ndin g
plet e data it mus t be
l s, n the abse nce of mor e com
repr esen ts this rela tion . , me d that the curv e of Fig. 8.2
R
l .
228
METAL WEAR
BALL, TUBE AND ROD MILLS
The influence of the ratio of mill size the ratc w '''
4 (d/D ) can more conveniently be written in th
4 (D fd ), and in Fig. 8.7 are plotted the data of :i\ \1
and Devaney<4> and Gross, <5> relating to the wear or \ ) , \1
when the charge consists of balls of uniform size, ~I I
McLeod <7>, relating to the wear of balls of size d in a : 1
consisting of balls of mixed sizes. simple chang cl I~
vertical scale of these curves brings them into coin icl 1
and so they may be generalized into the curve of Fip;. 11 11.
this curve representing the function 4 (Dfd) insofa : 11
may be derived from these data. It is also seen fr l
figure that, over the practical range of lO < (D /d) ,
the function may be replaced by 4 ( Djd) = 0556 (D/ rf) 1 ,
with an error of less than about lO %
The relationship between rate of wear of the meta l
faces and the ball charge in the mill is shown by thc (I I
of Gross, (5) plotted in Fig. 8.9. This curve is seen to \11 ,
very roughly, ofthe same form as that ofFig. 5.18, r J : i
to the total surface prod uced . The agreemen t wi t l I
general form of Fig. 3.5 r elating the power input to l
mill, is not so good for the higher mill fillings; the t 1 l
metal wear being much higher tha the power input.
observation is in accordance with the hypothesis r "o ll
rhouring" since, with high mill fillings the balls an~
strained to roll, with considerable abrasion, but im 1 , ,
which leads to effective armouring, is suppressed.
The only data relating to the effect ofthe quantity 11l
consistency of the pulp on the ball wear appear to \
those of Gross.<5) Unfortunately, these are insuffici:l
form the basis of a complete analysis. Furthermor l
data appear to be defective in that they are based on t
of 21 minutes' duration and it appears that this pcriol I
insufficient to allow equilibrium to be reached; that iN
the armouring to have become established. Howe<'l' , 111
the absence of more complete data these must be ac < l
tentatively. It appears that, as a first approximatio , t
effects of both pulp quantity and pulp density c l11
stated in terms of the volume of solid material pres )(
.     ,  
010 r  
229
MILLS
,,
...J
0 05
~~/t'"""'=ti
;:
6 00
4 00
200
800
Br inr ll Hardn r ss Or al ls .
F.
;)
v
li
?.....
60
"
"'40
"
20
8.6

Cross
Cog hill
andD<on<y.
~ @
@
Mc l <od .
....
10
40
30
20
<% >
Fa .
8.7
v=
50
"'""'. 0556t%>0S
10
2%)
Fa.
8.8
40
50
 
230
METAL WEAR
BALL, TUBE AND ROD MILLS
use of the function ( V ) where, as before, the quantity
is defined by
V=
\ l
oz
t.
oa
is read from Fig. 8.2,
is read from Fig. 8.4,
4(Djd) is read from Fig. 8.8,
5 (bfD) is read from Fig. 8.5,
7 ( J) is read from Fig. 8.9,
( V) is read from Fig. 8.10.
3 (J ,)
(f now Em and are regarded as sensibly constant for
steel used in the manufacture ofballs and liners and if
. y
f 1 (Djd) is replaced by (D fd)Ol5 and 5 (b jD) by' (Djb),
w lich may be done over the practical
v: riables, then equation (8.6) becomes
J.
F.
231
MILLS
()
re
Volume of particles
+ Volume of space between the particles
Volume of space between the balls (or rods)
~::vrr JS
range of these
W 1 = K 1DH5dOl5bl DLt. 1 () . 3 (J,)
8.9
1(J). 10 ( V)
.. .. (8.7)
In. practical milling the ball diameter is usually ap1)I' XImately a constant multiple of the particle diameter
'
this equation becomes
W1
= K 2D465d1 15Lt. 1 (H). 3 (NjN,)
7(J).(V)
(8.8)
If the relative hardness, , the mill filling, J, and the
of solids, V, are maintained constant then from the
lationships 1 (J,) =K(NJN,)1 30 for iow speeds and
/ 1 (J, ) =K(NJN,) 08 for high speeds, this relationship
) comes:
For low speed,
<Li
10
20
F.
8.10
and the curve of 10 ( V) so d educed is given in Fig. 8 10.
This curve should, however, be taken as only the rugh
approximation to the actual relationship.
Collecting these data, the equation for ball w
becomes
W1
= D35m. 1 1 2 Lt.() a(~,) .4(~) .5(i)
1(J) (V)
(8.()
W 1 = K 3D465 (NJN,)1 3d115Lt
.nd
.... (8.9)
for high speeds,
W1 = K 4D465 (NjN,)DBd1 15Lt .... (8 .9a)
The number of balls in the mill oc D 2L jd3 so the wear
r ball wb is given by
For low speed,
Wb = K5D2 65dl B5 (NJN,)I 3t
....
( 8 .1)
232
METAL WEAR
BALL, TUBE AND ROD MILLS
and for high speed,
Wb = K 6D265dl B5(NfNc)OBt
. .. . (8. 10 )
These equations should be contrasted with
Hukki<B):
For low speed,
W1 = KD 25 (N fNc)d lOLt
thos~ l
wt
11c l
~ quation
( 8. . )
which lead to:
F or low speed
Wb
= KDH5(NfNc) dlOLt
KD0 5(NfNc) d2t
.... (8. 1' )
and for high speed,
Wb = KD0 5(N/Nc) d3t
... .
(8.13)
(8.14)
.. .. (13. 11 )
....
233
ll aving regard for the general uncertainties in the data
d, the equations can be simplified to, for all speeds,
; r ,
given by Mortsell <6J, for the rate of metal
IS
W/ =
KD 2 5(NJNc)Lt
MILLS
c l
and for high speed,
W =
(8 .1 2. )
For low speeds the present equation is in fair agreen~t l
with that of Hukki insofar as ball diameter is conccrJ l ,
but at high speed, and in all other respects the agre m
is not so good.
The formulae of Hukki are, however, derived from fi1
principles on the basis of clearly defined assumptio1 .
These assumptions are that, for slow speed, the ra l
wear is proportional to the rubbing velocity and, for ~
speed, the wear is proportional to the force of the blo'A'
The first assumption appears to b e quite reasonabl l
the second is not so free from objection. In the first plt~ ,
the question of the comminution of a material by a ~\'
is involved, and this has already b een shown to be a 0111
plicated question. Even, however, if the damag
assumed to be proportional to the intensity of a sjJ~I
blow, the "armouring" of the ball surface, and als : ,
work hardening ofthe metal, would be expected to rccl
the rate of ball wear at high speeds.
(8K:b)
... . (8.15)
is. the total weight ?f balls, W/ is the wear per
d IS as before and s a constant which generally
llls between 9 and 16 for Swedish ore but, in extreme
. s, between 5 and 23; W/ being in kilograms per hour
. d Mb in metric tons.
'ince Mb :.D 2 L it follows that, from this equation the
lotal. wear is proportional to D 2; not to D 465 as sugg~sted
1 . ' r~ Mb
111 tme,
1 :vusly .
It is .interes~ing now to exa~ine equations (8. 13) and
(11. 14) the lght of the publshed r esults of the various
. lhorities.
F'irstly, it has b een suggested by Davis, (9) that the rate
(' wear of an individual ball is proportional to d3. But
: s, <10 > Prentice, <10 Garms and Stevens, <t 2) Norman and
.
b, <13) and Norquist and Moeller(l 4) state that the rate
is proportional to d2. Furthermore Bond, (15) gives
figures, such as 20, 221 and 229 whilst Tag: rt (16) gives 20 for cascading balls. Thus, on this point,
111 results ofthe present analysis are roughly in agreement
wi tl1 the view of the majority of other workers but more
w rk is required before this question is settled.'
The question of the influence of the mill diameter does
t appear to have received more than passing study but
111 suggestion that the ball wear varies as D2 6 where D
ix the mill diameter, appears to be reasonable.' serious
iticism of all the work cited is that no effort appears to
l ave been made to maintain similarity between the
' ~ear
: nous
METAL WEAR MILLS
BALL , TUBE AND ROD MILLS
234
various tests. The most satisfactory type of equ~tio
wou1d appear to be equation (8.6) and, whenever possIJi ,
the resu1ts shou1d be presented in this, or simi1ar, f .
Since however the balls are invaab1y of iron or s
the va1~e of Em ~nd may be regarded as sensibly
stant, so the equation may be reduced to
W
1
KD3 5 Lt.(H) .3(;) .4(~) 5(~)
REFERENC ES
....
(8.1/)
An effort has been made to ca1cu1ate the va1ue of i
this equation, the basis of the various pub1ished w k,
augmente d with reasonab1~ ass~med va1ues where '(' .
sary, and these va1ues are gven Tab1e 8.1; all va1u s
C.G.S. units.
TABLE 8.1
Va1ue of
Authority
McLeodl71 test: ( 1) Chillcast balls
(2) Sandcast balls
(3) Annea1ed balls
16 10 8
1 6 10 8
23 10  8
Coghill and Devaney<41
06 10  8
Mortse!lll
26 10  8
(pot mill test)
Mortsetl (from eqn. (8 .15) for 6ft
6ft mill)
235
iman
I shou1d a1so be pointed out that corrosion has
o ant effect on the rate of liner wear. Thus, if corrosive
l)stances are present, the products of corrosion will be
noved from the wrking surfaces by abrasion and new
111 a1 will be constant1y exposed to chemica1 action. The
r arising from this cause will be proportion a1 to the
. of meta1 exposed, but there is insufficient data for an
ysis ofthis prob1em.
12 10 8
The va1ues ca1cu1ated from equation (8.16) show faily
wide disagreem ent but this is not surprising since al
for many ofthe parameter s hz.ve been assumed. lt app : ,
however that = 16 10 with an error of about 50 ;{,.
Thus.it ~ppears safe to conc1ude that the ~oregoing analyHi
inc1udes the major factors in the mechaca1 wear of b:ll
and 1iners and it is possib1e that it can form an adeq:
basis for an extended and improved treatment .
. tosE, . . "Symposium Gi:nding", Trans . Instn chem. Engrs., Lond.,
1957, 35, 87.
' , ENGELS, . Metall, 1954, 34, 102 .
: S, . . Trans. lnstn chem. Engrs, Lond. , 1957, 35, 144.
. OGHILL, W. . , and DEVANEY, F. D. Engng Min. J., 1937, 138,337.
ROss, G. Bull. U.S. Bur. qf Mines, . 402.
' ,
. MOR'SE LL, S. Engng Min. J ., 1948, 149, 91.
. McLEOD, . . Trans. Amer. Inst . min. (metall. ) Engrs, 187, 33.
11. ur, R. . Trans. Amer. Inst. min. (metall. ) Engrs, 1954, 190, 642.
'1 . DAvs, . W . Trans. Amer. Inst. (metall. ) Er,grs, 1917, 61, 250.
10. ELLIS, . W. Trans. Inst. Brit. Foundrym, 1937, 40, 137.
11 . PRENCE, . . Trans. Amer. Inst. min. (metall. ) Engrs, 1946, 169, 147.
I ~ . GARMs, W. 1., and Ss, J. L . Trans. Amer. Inst. min. (metall .)
Engrs, 1946, 169, 147.
I . NORMAN, . . , and LOEB, c. .. Trans. Amer. Inst. min. (metall.)
Engrs, 1949, 183, 330.
l . NORQ.Us, D. ., and MoELLER, J. . Trans. Amer. Inst. min. (metall.)
Engrs, 1950, 187, 712.
' . BoND, F. C. Disc. Trans. Amer. Inst. min. (metall.) Engrs, 1943, 153,
373; ibid., 1946, 169, 138.
lli. R, . F. Handbook qf Minral DreSJing (New York).
ROLE OF ADDITIVES MILLING
CHAPTER 9
ROLE OF ADDITIVES
MILLING
term "additive" is applied, in milling practic ,
various substances which when added to the content~
a mill bring about an improvement in the rate of grincli\
and, therefore, a reduction in the time of grinding req i I
to attain a specified degree of fineness. Additives could ; I
be introduced in order to attain some special chara
istic of the final product. However, such applications : 1
d ependent on the purpose of the final product ~nd s : '
outside the scope of the present c~apter. An nter s ,
feature of additives is that the quantity required is g
ally very small; usually being between about 001 1 t
.
.
01 % of the weight of the charge treated.
That additives are of sufficient importance ndustr ll '
to warrant consideration is shown by recent work 1l
Skaupy, in which it is sho.rn that the rate of millin ~ I
iron powder in isoamyl alcohol is no less t?an tw }
times the r ate in dry air. These tests were carned out
vibration mill, and so might not be directly comp at'II
with a ball mill but nevertheless the result is signifi :
In view of the arge energy requirements for g~incli, .,.
especially fine grinding, a saving ?f eve~ 20 {! In .)IJ.ll
ing time could make a substantal savng In mtll
costs.
In view of the potential importance of these subst
it is perhaps surprising that their mode of action i~
precisely known. It is impossible to. esta?lish ~rom . l
literature on the subj ect whether ther actn anscs f
improvement in the conditions of crushing by the pr:
tion of a coating of powder on the ball surface, from l
'
236
237
vention of agglom:r~tion in dry milling, or flocculation
~e case of wet mllng, or by some action within the
Ht~tcle, such as the promotion of the spread of the fissures
~ I Hch occur naturally in all solid materials. It is possible,
~ d ed probab~e, that. all these modes of operation are of
nortance, ether singly or in various combinations
~. ending on the circumstances and the controlling con~
I s presen t .
. For. dry milling, the additive may be present in solid
lo m, the form of a solution added to the dry feed, or in
. gaseou.s or vapour state. For wet milling, the additive is
sent lll the water; .the concentration generally being
ss than 1.% of.the .solid n:aterial present.
, ~efore dscussng In detai1 the use of additives in milling
lS perhaps relevant to recall what has been said earlier.
s shown 1n Chapter 4, for a subs~an.ce in which the yield
ress and the stress for fracture cncide, the strain energy
~entually r eappears as heat and the energy associated
~ th. the creatn of new s.urface re.presents a very small
I~ actn. of the total .e nergy Involved the crushing operal n i th1~ energy ?eing only about 5 % ofthe total when a
article s commnuted in a simple impact crusher. For
v ry ~ne powders, howev~r, th~ surf~ce energy becomes
re mportant and manfests Itself In high adsorbtivity
: ?~ the tendency to agglomeration observed in finely
l vided subs.tances. Even so, t~e energy associated with
le surface s a small proportn of the total involved
cause ofthis it is difficult to see how the reduction ofth~
N.rfac~ e~ergy can significantly affect the energy required
I r gnnd1ng but, nevertheless, surface active dispersing
:gents are often very effective grinding aids. The reduction
th e surface energy of the particle itself is not however
lhe o~ly possible explanation of the action' of thes~
na t enals .
possible direct mechanism for the operation of these
L bstances has been suggested by Rehbinder. ( 2) Assuming
.hat the molecu~es of the additive are adsorbed the surface of the partcles, then the cohesive forces which bond
238
BALL, TUBE AND ROD MILLS
the surface cetres of the crystal lattices, at the surface or
the particle, will, i accordace with Gibbs' equatio, b
lowered by a amout ecessary for the adsorptio to tak<:
place. Thus, the microscopic cracks ad fissures, which ar
always preset i brittle materials, will, after the adsorptio process, be mre easily wideed. Frm this hypothesi s
tw tetative cclusis, regardig the effect f such
agets the crushig process, may be draw.
Firstly, whe the particles are very small the physical
stregth would be expected t be affected. Also, witi
larger particles, the cditis existig at the crack tip
are mdified ad this culd lead t higher stress ccetra
tios i these regios: these higher stress cocetration s
leadig to reduced physical stregth f the particle. Thus,
the stregth of particles f all sizes wuld be expected to
be reduced but, the whole, the fier particles would bc~
expected to be mst affected.
Secodly, the dispersive effect of the additive would bc
most marked with the small sizes of particle ad so
agglomeratio would be reduced. This would lower thc
socalled "limit of fie gridig", observed by mi\1
operators, at which equilibriu m is established betwc 11
the rate of productio of fie material ad the rate of pr<>
ductio of larger particles resultig from agglomera ti 11
ad, possibly, from recrystallizati, of smaller oes. 11
icidetal effect would be that chesio betwee the ball
ad layers f particles surroudig it would be reduc cl
ad that this would, by reductio of the thickess of t!H
layers, reduce "cushioig" of the blow, with cosequ 111
loss of eergy i the form of heat.
Strog evidece supportig this hypothesis is give by
the work of Baker ad Presto, (3) the tesile streg t l1
f glass. Experimets show that the tesile stregth of gl a~
is 20 % greater whe the specime is dried i a ('
tha whe the surface is wet ad is as much as 200 %
greater whe baked i a vacuum tha i the wet conclitio. The tests also show that moisture is the chief factor
i the loss of stregth; althugh the presece of carbon
ROLE OF ADDiiVES MILLING
239
dioxide also has some effect. Thus the surface coditio f
le glass affects the average tesile stregth at failure eve
~ r the case of ro?s ?f comparati vely large diamete;. The
surface of glass s, fact, always covered with microracks, ad the hypothesis put forward was that the water
molecules ~re active at t~e crack tips ad, that there is a
local cheffilcal attack, wt~ the.formatio of a gel, which
weakes the structure at this pt. However, this may be,
the tests d:mos~rate that the surface coditios of eve
large partiCles fluece their stregth i a decisive
ma er.
It is clear that, irrespectiv e of the questio of additives
' s su~h, the ature of the fluid medium i which the
rushg process take~ place would be expected to iflu
~e the eergy reqUired to produce a specified fieess.
lt s commonly .stated that, all other things beig equal
the eergy reqUiremets for wet gridig are as much a~
20% less tha fo~ d~ gridi~. Practical experiece also
~ uggests t.hat the lJ~toffieg~d~~ is lower i wet grid
g th~ ~ dry .. Ths observato s geeral accord with
he pnncples dscussed above.
The ge~eral pictu.r: wh.ich teds to emerge cocerig
~he opera.to of addtves s therefore that this is primarily
1 r;duc~g the effective eergy at the particlefl uid
medum ~erface, due to the adsorptio f the additive
the partcle surface. This could result i decreasig the
s tregth of the larger particles, but a much more proouced effect would be expected i the very fie size
rages, whe~e the surface of the particle becomes very
much more ~portat i relatio to its volume. Also i
t~e very fie. sze rages, the reductio i the agglomera tlo tedece~ amog .the particles would be expected t
have. a beefi.cal effect reducig the eergy for gridig
ad lJ? reducg the gridig time. It can also be deduced
that, accordace with the above geeral priciples the
mo~t effective additives would be expected t be surface
actve substaces.
Such cosideratios do t, however, explai the
240
ROLE OF ADDITIVES MILLING
BALL, TUBE AND ROD MILLS
operation of solid additivesin dry ~rinding; for exampl i
improved mill performance obtaed when a.small qu:~1
tity of carbon black is adde~ to ~he. cement cl1nker uncl'l"
going grinding. Here aga1n, 1t IS su~gested th~t ~ I
function of the additive is in the preventn of cushnl\
since Schweitzer and Craig(4) have observed that adclitio
of carbon black in proportions as low as 008% is hi ll
effective as a cleaning agent for the surfaces of the b~ll
.
and mill.
rather different explanation of the operation of acl l
tives is put forward by Bond<5J; it again being suggcst l
that ball coating is an import~nt factor. Th~se t<'H IH
showed, in the first place, that w1th a soft m~tenal, s
as limestone excessive ball coating occurs w1th pow cl 'M
of much larger particle size than for hard mat.eria~s, su l
as quartz. Furthermore the occurren:ce of c~at1~g IS pr:.
tically independent ofthe character ofthe gnnd.Ing mccl~: ;
cast, forged and ballbearing steel balls show1ng no S J.I
nificant difference other than that, when new, the smooll
hard balls are not subject to coating. Once abrasion 1:
removed the surface, however, these balls show the sa
behaviour as the others.
According to Bond, the ball coating consist~ of a.n innn
most layer of particles, of about 0550 1n d1am tn,
wedged into the surface of t~e ball. The outer l~y~rs
particles making up the coat1ng are keyed to th1s 1nc
layer and are of much larger size. The effect of these layn
is to cushion the grinding surface and so to greatly ~ecr '
the grinding rate. It is possible that. other mecha.sms :c
operative; for example the benefiaal results obtan~ed . '
the use of small quantitites of carbon black as a gd.
aid with cement clinker suggests some analogy wl I
electrostatic precipitation. ln this application it is know
that the addition of a small amount of carbon, gencr<II
in the form of unburnt fuel, can bring about the precipi
tion of a flue dust which otherwise cannot be treated; t lc
function of the carbon apparently being to form a <!
ducting network through the layer of powder by wl
l cc trical charges leak away to
i not possible, therefore, that
es ent in this application?
241
the collector electrode. Is
some such mechanism is
T he question of additives will now be considered in
detail; commencing with dry grinding. As previously
Ntated, in dry grinding the additive may be present,
: least in theory, in solid, liquid, vapour or gaseous
rm and some of the most interesting results of the use of
these additives are reported from the cement industry.
According to Berry, (6) advantages have been shown by
he use of vinsol resin, cod oil, beef tallow, aluminium
s e arate and glycerol in the grinding of cement clinker. ln
articular colloidal carbon and two dispersing agents,
known as R.D.A. and T.D.A. have proved effective.
ccording to Kennedy, (7) R.D.A., which is arylalkylHulphonic acid, and T.D.A., which is a mixture of tri th anolamino salts and highly purified calcium salts of
li gnin sulphonic acid, were introduced as dispersing agents
:nd catalysts to improve the properties of cement. These
:clditives were introduced in the form of a solution in
water, to the raw clinker feed of the tube mill in the proortion of about 006% by weight of the charge. Apart
m the improvement in the setting properties of the
rcment, the output of the mill, for the same specific surace, was reported to be increased by 3040 % The
cse arches of Schweitzer and Craig<4J into the grinding of
: cment clinker, in a laboratory ball mill, with carbon
ack as a grinding aid show a continuous increase in the
a te of grind, or fineness attained in a given time, for
: dditions up to about 1 % by weight of the feed. The rate
f grinding thereafter fell with increasing amounts of the
nore
:dditive.
These results, which are summarized in the curve of
9.1, show that the time required to grind to 85 %
through 325 mesh is reduced by 35 % when compared
with that required when no additive is used. The curve
<lso shows that the optimum quantity of additive is fairly
lcfinite; small deviations from the optimum quantity
ig.
BALL , TUBE AND ROD MILL S
242
ars to bc
havin g appre ciabl e effect. This concl usion appe
true for most addit ives.
low as
As previously noted , quan titite s of addit ive as
effect
ed
mark
a
had
008 % by weig ht of the solid prese nt
cing
produ
ut
witho
es,
surfac
mill
on the cleanliness of the
of
rate
The
ct.
produ
the
of
on
iorati
deter
r
notic eable colou
100
..
.:
.r
..,
.:
C7'
"
.:;
...
~~
.....;:._
_., /
__...
"'<D
c
4:40
...
.!
..
320
i=
10
01
P~rcrntoq
F .
100
Colloid ol Corbon .
9.1
this sm:ll
grind ing is not, howe ver, mark edly impr oved by
quan tity of addit ive.
a sm:ll
The work of Bond(5) shows that the addit ion of
t
abou
ght
brou
t,
weigh
by
%
088
r,
wate
of
amou nt
.,
D.
ofT.
ion
solut
reduc tion in the coati ng of the balls.
lt
brou.
ver,
howe
ed,
fe
the
of
ht
at abou t 013 % by weig
thercf(
abou t comp lete inhib ition ofba ll coati ng. It was
ti
preven
in
concl uded that the actio n of the addit ive
t
eff<
than
rtant
impo
coati ng ofthe ball charg e was more
ROLE OF ADD ITIVE S
MILL ING
243
partic les.
nns1ng from adsor ption on the surface of the
amou nts
small
the
that
fact
the
by
orted
'his view is supp
mark ed
a
t
[ addit ive whic h are sufficient to bring abou
allow
to
ient
suffic
are
i provement in the rate of grind ing,
20%
than
less
on
layer
cular
l1e foration of a mono mole
that
fact
the
by
orted
supp
also
is
t
ce.
surfa
f the prod uct
ce
produ
d
woul
ven large chang es in the surface energ y
nded
dema
y
energ
the
in
but small propo rtion al chang es
energ y
by the whol e grind ing process (since the surface
nd
dema
y
energ
total
ofthe
on
fracti
Sall
repre sents a very
tives,
addi
solid
[ the process). t is also significan t tha t the
coati ng; proba bly
tlentioned earlie r, all tend toinh ibit ball
ing edia since
grind
the
to
ce
surfa
y
by iparting a greas
ting prope rties.
brica
lu
ed
mark
it
exhib
to
n
know
all
hey are
wate r in preof
s
The effect of very small quan titite
catio ns. For
appli
other
in
n
venti ng ball coati ng is know
n in the
know
well
is
it
that
states
xamp le, Manson<s>
enam els,
s
paint indus try that, in the grind ing of dryproces
effective
be
can
r
wate
he addit ion of as little as 00625 % of
ermo re,
Furth
ng.
coati
liner
and
! the preve ntion of ball
d balls,
coate
badly
from
ved
remo
be
also
Lhe coati ng can
wet
few
a
of
unde r runn ing condi tions , by the addit ion
tive
selec
that
sts
sugge
(5)
balls to the mill charg e. Bond
the
be
t
migh
ce
surfa
ball
the
at
les
l1ydra tion of the partic
of
tities
quan
small
by
ng
coati
of
ition
inhib
the
eason for
tive.
tenta
water. Such a suggestion is, at the most,
surThe reaso n for such prefe rentia l hydra tion at the
ed
mber
reine
be
must
it
but
,
clear
not
ace of the balls is
le
derab
consi
a
is
there
ct,
impa
of
ent
mom
the
that, at
this
of
some
tless,
moun t of energ y dissip ated and, doub
physicoould be used to bring abou t vario us chem ical, or
energ y.
of
input
an
nd
dema
h
whic
ions
hemi cal, react
n by
vatio
T his possibility is supp orted by an obser
to a
ged
chan
was
ore
Pryor, (9) who found that copp er
the
e
wher
s
point
at
level
y
energ
ompo und of a highe r
*
it.
ing
break
ut
witho
ore
the
ed
rubb
balls of the mill
has a piece of this ore, presen ted to him
this transfo rmatio n at a numbe r of points
shows
clearly
by Mr. Pryor, which
the balls.
with
made
been
has
where contac t
* One of the author s (H.E.R .)
244
BALL, TUBE AND ROD MILLS
There are other examples of the increase in the ra
grinding brought about by the use of additives in
milling. For example, WaeserOO) states that both R.D ..
and ammonium salts are effective in the grindin
graphite and that oleic acid is an effective additive for l11
grinding of zinc blende. Gotte and Ziegler<ll) qu
interesting results as to the milling of cement clink,
marble and glass in atmosphere s of various polar and
polar substances. The tests were carried out in a vibraio
mill, the additive being introduced in vapour form i1t
the previously evacuated mill body containing the fc l .
The polar additives investigated were acetone, nit
methane and water; the nonpolar additives being carbo
tetrachlorid e, benzol, hexane and petroleum ether.
These results can be summarized as follows:
the milling of clinker, the improveme nt brou r l
about by the use of nonpolar additives in placc
an atmosphere of dry air, decreases with increa
fineness and increasing time of grind; the avera11
increase inrelativefin eness being 10 20 % The pl :
additives show an increasing effectiveness with tinH
of grind and fineness of powder; the most effeci
being acetone which led to improvemen ts of I
50% in the rate of grinding, relative to that carricl
out in dry ai.
(2) Marble showed the same general characterisi1
but, in this case, water vapour, which led to a 60 / 11
increase in speed of grinding compared with a cl
atmosphere was the most effective additive.
(3) For the grinding of quartz, polar and nonp l;
atmospheres led to about the same increase in l~
rate of grind.
(4) Glass proved anomalous in that, whilst nonp :
additives effected some small increase in the rac
grind, polar additives produced a very definil
decrease in the grinding rate. definite explanaio
()
ROLE OF ADDITIVES
MILLING
245
of this anomaly is forthcoming but it is possible that
it might be connected with the strong adsorptive
properties of finely divided glass powder.
Whilst it. is realized that the tests of Gotte and Ziegler
:re not strictly relevant, in so far that they were carried
out in a vibration mill, it is considered that they are of
interest in that they are, as far as is known to the authors,
tl1e only tests carried out with a controlled atmosphere
onsisting of a surface active agent in vapour form. The
ne thod may well have applications for certain specialized
batch drymilling operations .
Turn now to the use of additives in wet milling. Accordi ng to Berry, <6 > silicates, phosphates and arylalkyls ulpho~ic acid are often used as additives in wet milling;
the pant and colour industry, especially, offering a very
wide field for the use of such agents (Anon. (1 2)).
Probably the most thorough investigatio n into the
ffects of additives in wet milling is that reported by von
S zanth~( l 3 ) In this work, the effect of varying percentages
of flotatn additives and of oleonic acid on the wet milling
f quartz and limestone, in a laboratory batch ball mill
was investigated . In each case the characterist ic produc~
article size, d, resulting from grinding for 30 minutes was
estimated fro.m the size frequency curves. The results,
showing this characterist ic particle size plotted against the
ercentage concentrati on by weight of the additive are
shown in Fig. 9.2 . It is seen that, in each case ' there is an
.
op tmum percentage beynd which the grinding rate
decreases with any increase in the quantity of additive.
Von Szantho considers that the effect of the additive is
pri~arily d~~ to the lowering of the surface energy of the
partcles, ansng from the adsorption of the additive with
consequent widening of the elementary cracks and fi~sures
?n. the .s~r~ace ~f the part_icles. As discussed previously, it
s mpl~t ths conceptn of the operation of additives
~hat th~r effect should ~ecome progressivel y greater with
ncreasng fineness of gnnd. The undersize curves shown
Fig. 9.3 do in fact show that the effect of the additive is

BALL, TUBE AND ROD MILLS
246
..,
l"i
.!::!
.n
~0 4
~
 ....
..
. "
6u"
....
; ;
;::
~02
"
u"
~ 0 04 ......=:::::::..'+:::::::+1
..
>
u"
Cur~ Ouortzit~
"v
Cur~2 Lim~s!on~ withfotiqon.
V1
Cur~ 3  Lim~ston~ with o~ic Acd ,
.~ 02
. 08 r\
Cr i ndnq Umnton~.
\\
O CI
\"~
"!'...
\/')
.::
~ 0 4
..
:2
~~
~~~
0~P~rc~ntoq~
a:
Ot Additi~ .
0 25
P<rcontoq<
Limstont
.____.._._
Qurtzt .
70r.~~~~~~~
60rr~~+~~~~
40rr~~~+~~~
..;
1 4
.2
~
.::
12
ue
"c
2 ~...,.__,.c.__+_~:_1
....
Friction .
r
0 5
Sodium
F.
FIG. 9.2
"
MILLING
....
....~~~~
greater at the smaller particle sizes (for particles below
about 60).
The d ecrease in the rate of grinding observed when the
0 6
ROLE OF ADDITIVES
Surtoc~ .
075
Ol~at~ .
9.4
Crindinq
1\
Quortzit<
~~
Surfac
............:::
~ 0 6
I
~r: ic t i on
:;
"'0.4
r~~~~+~~
>
"'
C7l
"'
..
u
..
F.
20?Lr~+1
0003
Chrctristic Prtlcl Siz~ Mill i mttrs .
F .
9.3
0.075
OOS
010
Pcrc4f'ntaqr Ot Ftot i qan
     With Additi .
Without Additivt .
9.5
concentration of additive exceeds a certain value is tentatively explained the basis of the lubricating effect of
these substances at high concentrations. This explanation
is supported by the data of Fig. 9.4 and Fig. 95, in which
curves of surface produced and coefficient of friction are
plotted against concentration of additive. These curves
.J'
..
248
BALL , TUBE AND ROD MILLS
ve fr which i
shw that fr ccetratins f additi
gridig is hi A I ,
f
rate
the
high
is
cefficiet f fricti
a t l
but whe the cefficiet f fricti decre ases the r
fr<~.,
f
ciet
ceffi
f
s
gridig dimiishes als; the curve
,
J
r.
simila
y
clsel
eig
b
ig
ti ad rate f grid
i ~
fricti
f
ciet
ceffi
the
that
udes
ccl
Szath
ad that, at hi A i
majr imprtace i the gridig rate,
f fricti t
ciet
ceffi
ccetratis, the decre ase i
ti i sur< c
reduc
the
frm
d
gaie
weighs the advatages
the additiv .
f
pti
adsr
e
surfac
the
frm
g
eergy arisi
ctets
the
f
This result is als f iterest i the light
i a nll
g
surgi
that
sted
Chap ter 7, i which it is sugge
i tf,
ti
reduc
small
a
by
abut
ht
migh t be brug
s l
curve
the
cefficiet f fricti f the charg e. Thus,
mgll
tives
faddi
use
Fig. 9.4 ad Fig. 9.5 suggest that the
tl
reduc e the cefficiet f fricti by 4050 % ad
mll ,
a
cause
t
iet
suffic
tha
mre
culd well prve
which therwise perfrms satisfactrily, t surge .
~
It is als iterestig t equire why the rate f millin
rl
Clea
.
fricti
f
ciet
ceffi
asig
shuld fall with decre
t
this must arise frm a reducti i the effective heigh
H
tl1i
ad
e
charg
fall f the cetre f gravi ty f the mill
tl
e
charg
the
f
g
wuld prbably arise frm slippi
l
t<
expe
be
wuld
it
the
s,
fact,
i
mill shell. If this is,
; l
that the use f adequ ate lifters wuld, t sme extet
aiHo
is
there
ly
least, restre the rate of gridig. Pssib
g ad the icreas l
sme relatiship betw ee the slippi
ig as cmpan l
grid
wet
with
wear,
ball
rate flier ad
with dry gridg, which is smetimes reprted.
It is f iterest that Szath cites the wrk
4) i supprt f li N
Rehbider<z> ad f Egelhardt(l
crrelati betwee tl
ad their resistace
was also show that th<:
hi .:
is a prgressive decre ase i the resistace t scratc
bed tl
adsr
ve
additi
of
es
ti
quati
sig
with icrea
c
S]Jrface fthe bdy, there beig a clse correlatio betw
n,
r
isthe
ption
adsor
the
the resistace to scratchig ad
cclusis. R ehbider shwed
surfac e hardess f mater ials
scratchig with a fie pit . It
ROLE OF ADDI TIVES
MILL ING
249
ed
with a minim um of resistace t abras ion beig reach
%
70
as
much
As
.
pti
adsor
ete
cmpl
of
under coditios
the
way,
this
in
red
measu
was
ess
hard
e
surfac
i
ase
decre
maxim um d ecreas e correspodig to the greate st differ
It
e.
surfac
the
ad
m
mediu
the
cce i polar ity betwee
would
is sugge sted by R ehbid er that plar additi ves
ad plar
ials,
mater
philic
hydro
fr
ive
effect
most
be the
a dditiv es shuld be used for hydro phobi c crystals.
mff, (15) who
Egelhardt reprts the work of Obrei
the lamia
tig
separa
in
doe
work
elastic
the
measu red
to the
equal
y
of mica and foud that this was nume ricall
d.
forme
e
surfac
ew
the
th eoreti cal surfac e energ y f
prethe
i
out
d
carrie
was
a
lami
When splittig of the
e exsence of surfac e active m edia, hwever, the surfac
i
ased
icre
,
posed, for a give amou nt of strain eergy
by
d
cause
y
eerg
e
surfac
the
proporti t the d ecreas e i
the
t he adsorpti. Egelhardt furthe r coducted tests
i
disc,
e
carbid
silic
a
with
z
quart
of
ig
grid
abrasive
polar
that
foud
th e presece of varius additi ves. It was
i reducig the
ad diplar additi ves were very effective
quatities of
small
ular
partic
In
.
abrasi
resistace t
e i
octan ol or butyr ic acid in bezol or sdium silicat
ace.
resist
wear
water were effective i reducig
outstadig
Skaupy has recetly repor ted some
r i varius
pwde
fir
g
millin
the
frm
ed
results btai
i a
fluid m edia. Altho ugh the tests were carrie d ut
f
ds
gru
the
here
vibratig mill, they are icluded
ig
grid
f
rate
the
that
fud
was
their imprtace. It
tha
of irn was five times as great i glycl, ad less
air.
dry
i
as
l
twety times as great i isamyl alch
ially
espec
were
ves
additi
The tests revea led that diplar
sureffective as gridig aids, as wuld be expec ted frm
ale.
s
erati
csid
y
eerg
face
effect the
Althugh, i view f the mre mark ed
energ y as
e
surfac
the
f
evide nce pits t a lwerig
ves i
additi
f
ti
pera
the
i
factr
being a imprtant
mre
is
g
millin
dry
f
case
the
in
ce
evide
the
g,
wet millin
pr with
ials
mater
cfused . The effectiveess f varius
16*
248
BALL , TUBE AND ROD MILLS
t/1c
show that for conce ntrati ons of additi ve for which
A" I ,
hi
is
ing
grind
of
rate
the
high
is
n
coefficient of frictio
/
but when the coefficient of frictio n decreases the rat
fri<
of
cient
coeffi
of
grind ing diminishes also; the curves
,
tion and rate of grind ing being closely similar. JL l
i$
n
frictio
of
cient
coeffi
the
that
udes
concl
von Szant ho
h~
major impo rtanc e in the grind ing rate, and that, at
t
n
frictio
of
cient
conce ntrati ons, the decre ase in coeffi
sur<<
in
tion
reduc
the
from
d
gaine
weighs the advan tages
additiv .
the
of
ption
adsor
e
surfac
the
from
g
energ y arisin
nts
This result is also of intere st in the light of the conte
n/1
a
in
g
surgin
that
sted
Chap ter 7, in which it is sugge
t/
in
tion
reduc
small
a
by
about
ht
migh t be broug
/
s
curv
the
Thus,
e.
charg
the
of
n
frictio
of
coefficient
ll
g
m
itives
ofadd
use
Fig. 9.4 and Fig. 9.5 suggest that the
tli
reduc e the coefficient of frictio n by 4050 % and
m/1 ,
a
cause
to
ient
suffic
than
more
could well prove
which otherw ise perfo rms satisfactorily, to surge.
~
It is also intere sting to enqui re why the rate of millin
l
Clear
n.
frictio
of
cient
coeffi
asing
shoul d fall with decre
/
this must arise from a reduc tion in the effective heigh
H
tli
and
e
charg
fall of the centre of gravi ty of the mill
t/
on
e
charg
the
of
ng
would proba bly arise from slippi
l<l
expe
be
would
it
then
so,
fact,
in
mill shell. If this is,
t ;l
that the use of adequ ate lifters would , to some exten
aiHo
is
there
ly
least, restor e the rate of grind ing. Possib
/
some relati onshi p betwe en the slippi ng and the increa
an l
comp
as
ing
grind
wet
with
wear,
ball
rate ofline r and
with dry grindng, which is somet imes repor ted.
It is of intere st that von Szant ho cites the work
of
rt
suppo
in
l4)
hardt(
Engel
Rehbinder<z> and of von
t/
en
betwe
lation
corre
a
d
showe
inder
conclusions. Rehb
surfac e hardn ess of mater ials and their resistance
'
th'l'
that
shown
also
scratc hing with a fine point. It was
i:
scratch
to
ance
resist
the
in
ase
is a progressive decre
l
with increa sing quant ities of additi ve adsor bed on
betwc
lation
s~rface ofthe body, there being a close corre
,
isother
ption
adsor
the
the resist ance to scratc hing and
ROLE OF ADDI TIVES
MILL ING
249
ed
with a minim um of resist ance to abras ion being reach
70%
as
much
As
ption.
adsor
lete
comp
of
under condi tions
the
decre ase in surface hardn ess was measu red in this way,
differ
st
greate
the
maxi mum decre ase corre spond ing to
It
e.
surfac
the
and
m
mediu
ence in polar ity betwe en the
is sugge sted by Rehb inder that nonp olar additi ves would
polar
be the most effective for hydro philic mater ials, and
ls.
crysta
c
phobi
additi ves shoul d be used for hydro
Enge lhard t repor ts the work of Obrei moff, (15) who
a
measu red the elastic work done in separ ating the lamin
the
to
equal
y
of mica and found that this was nume ricall
d.
theor etical surface energ y of the new surface forme
prethe
in
out
d
carrie
was
a
lamin
When splitti ng of the
e exsence of surfac e active media , howe ver, the surfac
in
sed
increa
y,
posed, for a given amou nt of strain energ
by
d
cause
y
energ
e
surfac
the
propo rtion to the decre ase in
the
the adsor ption . Enge lhard t furthe r condu cted tests on
in
disc,
de
carbi
n
silico
a
with
z
quart
of
ing
grind
abras ive
polar
that
found
the prese nce of variou s additi ves. It was
the
and dipol ar additi ves were very effective in reduc ing
of
ities
quant
small
ular
partic
In
ion.
resist ance to abras
e in
silicat
m
sodiu
or
l
benzo
in
acid
ic
butyr
or
ol
octan
water were effective in reduc ing wear resistance.
outsta nding
Skaupy has recen tly repor ted some
in variou s
er
powd
n
ofiro
g
millin
the
from
results obtai ned
in a
d
fiuid media . Altho ugh the tests were carrie out
of
ds
groun
the
here
vibra ting mill, they are inclu ded
ing
grind
of
rate
the
that
found
was
their impo rtance . It
than
of iron was five times as great in glycol, and no less
air.
dry
in
as
twent y times as great in isoam yl alcoh ol
ially
espec
were
ves
additi
The tests revea led that dipol ar
sureffective as grind ing aids, as would be expec ted from
.
alone
ons
derati
consi
y
face energ
the
Altho ugh, in view of the more mark ed effect
as
y
energ
e
surfac
the
of
evide nce points to a lower ing
in
ves
additi
of
tion
opera
the
in
factor
being an impo rtant
more
wet millin g, the evide nce in the case of dry millin g is
prowith
ials
mater
confused. The effectiveness of variou s
16*
250
BALL, TUBE AND ROD MILLS
nounced lubricatin g properties such as carbon black, as
an aid in the grinding of cement clinker, is much easier to
explain on the basis of the inhibition of ball coating by
means of the deposition of a layer of carbon on the balls.
Furtherm ore, the small quantities of water which have been
observed to remove the coating from a badly coated mill
lends support for the idea put forward by Bond (5) that, in
this case, the particles are removed from the surface by a
process of selective hydration , a surface phenomen on .
However, the small quantities of water which are effectiv
suggests that the adsorption occurs at the ball surfac !I
only. It thus appears that the surface phenomen a involved
are complex and that present knowledg e is insufficient t
give more than a very imperfect picture of the mechanism
which is operative.
In conclusion, it is fair to say that any process of crushing must be influenced to a large extent by the nature of'
the atmosphe re in which the operation is carried out.
There is ample experimen tal evidence that in many cases,
the energy or time of grind to a given fineness, and als
the ultimate fineness of grind attainable , can be favourabl y
~odified by the introducti on of a suitable additive into
the mill. In view of the very large energy requireme nts for
grinding, and in particular fine grinding, and also the low
theoretica l efficiency of the grinding operation as at prcsent carried out, it would appear that the use of milling
additives could effect substantia l economies. It is als
clear that the subject has not yet received the attention
which its importanc e warrants, and it appears to be a
field in which much interesting research remains to b
done.
REFERENC ES
I. SAUPY , F . Kolloidzschr. , 1948, 111 , 123.
2. REHBINDER, . . Phys ., 93, 72, 91.
3. R , . c., and PRESTON, F. w. J . app. Phys., 1949, 180, 382.
4. SCHWEITZER, C . W. , and CRAIG, . . J. industr. Engng Chem., 940,
32, 751.
5. BoND, F. C. Amer. Inst . min. (metall. ) Engrs, Tech . Paper, . 1160 (1940).
ROLE OF ADDITIVE S MILLI N G
6.
7.
8.
9.
.
2.
3.
4.
5.
251
BERRY, C. . Perry's Chemical Engineer's H andbook (New York) .
K ENNEDY, . L . J . I ndustr . E ngng Chem ., 936, 28, 963.
MANso N, . . J. Amer. ceram. Soc., 938, 21, 3 6.
P RYOR, .] . V erbai contribution to the discussion of "Symposium
Crushng and Grinding" , Instn Chem. Engrs, Lond . ( 956) .
W AEsER, . Dtsch. Farbenz tg., 952, 6, 434.
G, . , a nd ZIEGLER, . . Ver. dtsch. Ing. 956 98 9 373
' ' r. '
'
An on ., Paint Oil chem. R ev., 940, 102, 70.
VoN SZANTHO, . . Erz berg b. M etallhiittenw., 949, 2, 2 .
VoN ENGELHARDT , . N aturwi,senscha ften, 946, 33, 195.
0BREMOFF, . Proc. rl!JI. soc., (), 1930, 127, 290.
APPENDIX
SUBJECT INDEX
_1__
MILL FILLING PARAMETERS
15
30
45
60
75
90
105
120
135
150
165
180
J
0000
0003
0028
0090
0194
0335
0500
0665
0804
0910
0972
0997
1000
afR
C/R
1000
0966
0866
0707
0500
0259
0000
 0259
0500
0707
0866
0966
 1000
0000
0 259
0500
0707
0866
0966
1000
0966
0866
0707
0 500
0259
0000
252
bfR
0000
0034
01 34
0297
0500
0741
1000
1259
1 500
1707
1866
1966
1000
xfR

0960
09"0
0810
070()
0572
0430
0 2811
017 3
0091
002
0004
0000
Abasi on , grinding by, 133, 140
Add iti ves, mi lling:
Charge friction, effect on, 247
Cleaning of surfaces by, 242, 243
Common, 2404
Function of, 236
.
Surging, effect on, 248
Types ofGaseous, 237
Liquid, 237
Nonpolar, 244
Polar, 244
So1id, 240
Vapour, 237
Aggregation, 114, 237
Aids, milling. See Additives
Analysis, Dimensional :
Metal wear, 224
Power, 71
Rate of milling, 142
Surging, 210
Armouring, 144, 189
Associated energy, 125
F1ying, 48, 60
Profile, 5 1, 64,65
Rolling, 49
Total, 48
Circle, Davis, 44, 58
Classification, 24, 25
Contamination ofproduct, 31, 179,
187, 223
Coefficient of friction, 54, 213
Comminution:
Energy,distribution in, 115
Laws ofBond, 122
Carey and Stairmand, 125
Dobie, 124
Kick, 11 8
Rittinger, 117, 149
Modes ofAbrasion, 133, 140
Crushing, 137
Deep bed, 132, 133
lmpact, 130, 140
Sing1e particle, 133
of ca1cite, 130
Ball diameter, influence :
of cement clinker, 30, 240
Efficiency, 183, 186
of coa1, 110
M eta1 wear, 228
of coke, 11 3
Power, 81 ,97
ofdo1omite, 146, 154
Product characteristics, 178
of drugs, 32
Rate of milling, 153
of Durham anhydrite, 134
Surging, 210
of feldspar, 136
Ba\1 mi\1 , 2 1
of galena, ,130
Ball rationing, formula for, 192
of glass, 130
Ball wear. See under Wear, meta1
of graphite, 109
Balls:
of gran ite, 11
Cleaning, 242, 243
of moybdenum disulphid e, 109,
Particles embedded in, 144, 189
2 14
B1ow, cushioning of, 11 5
of pigmen ts, I 09
Bond's law, 122
of quartz, 109, 130, 146, 154
Breakage. See under Comminution
of rock salt, 130
of steel, 110
Charge:
Con tact:
Cascading, 38, 65, 75
Frequency of, 49, 161
Cataracting, 38, 65, 75
Time of, 163, 167
Coefficien t of friction of, 54, 2 13
Corrosion, 235
253
SUBJECT INDEX
SUBJECT INDEX
254
Pwer,
81
Millig, rate f (q.v.)
Surgig, 210,213
Critical speed, 42
.
.
Crushig. See under Cmmut
Crystallite, 112
Crystallizati, 134
Crystals :
Imperfect, 110
Perfect, 109
Davis circle, 44, 58
Desity
f gridig
bdes;
Griding:
i
Pwer, 76
Desity f particle; ifluece :
Mllig , efficiecy f (q. v.)
f, 150
Pwer , 76
Diameter (ball, mill, particle, rd);
 , rate
ifluece :
174
Wear, meta1 (q.v.)
Dim esial aa1ysis fr:
Millig , rate f, 142
 , efficiecy f, 183
Pwer,
71
Surgig, 210
W ear, metal, 224
Diffracti patter, 134
Drp hammer, 128
Efficiecy. See under Millig, efficiecy f
Eergy:
Assciated, 125
Ba1ace, 128
Elastic strai, 128
Necessary, 116
Plastic strai, 128
Sufficiet, 116
Surface, 116, 245
Flight, time f, 49, 62
F1yig charge, 48, 60
Fracture, gemetry f, 112, 125, 136
Cefficie t f, 54, 213
Effects f additives 247
Ifluece 
efrectfadditives thestr f<l
' f, 238
flu ece :
Fricti:
He tergeeus, 109,
Hmgeeus, 109
Gearig, destructi f, 204, 2 1
Glass cmminuti f, 130
Millig, efficiecy f (q. v.)
 , rate f, 150
P wer, 81, 97
Rate f millig ,
Surgig, 210
Materials:
Abrasive, 133, 140
Impact, 130, 140
..
Griding aids. See under ddt ~ .
millig
bdies :
Types f, 106, 190
M erit f, 190
Armurig f, 144, 18~ .
 , selective. See under llg
Hadse1 mill, 129
Hammer, drp, 128
Hardige mill, 26
Hardess (gridig meda, pwl,
relative) ; influece
Pwer (q. v.)
Millig, rate f (q. v.)
Wear, meta1 (q.v.)
Impact, cmmiuti by, 128, :10
Iterferece traJectres, 50
 betwee trajectries, 48
Kick's 1aw, 118
Krupp mill, 28
Laws f cmmiuti. See /
Cmmiuti
Lifters, 22
Lifters; ifluece :
Pwer , 81, 85, 86
Rate f milling, 197
Surging, 200,214
Liers :
Armurig f, 144,
Fucti f, 197
189
Types fStep, 198
Studded, 199201
Wave, 198
Wedgebar, 198
Wear f, 200
Wrkhardeig f,
200
121
Mixed, 110
Metal, wear f. See under Wear
Mill, types f:
Airswept, 22, 24
Batch, 22, 29, 90
Ball, 21
Grate discharge, 22, 90
Hadse1, 129
Hardinge, 26
Krupp, 28
Pebble, 21
Rd, 21, 30
ru verflw, 22, 90
Tube, 21
Mill prducts:
Ctamiati f, 31, 187, 223
Diffracti patte r f, 134
Mill surfaces, cleaig by additives,
242, 243
255
  particle diameter, 153
   desity, 150
   gridability, 150, 157,
169, 189
  pulp csistecy, 165,
171 , 182
   charge, 169, 172
  relative hardess, 169,
189
  rd diameter, 163, 178
  time fmillig, 163, 167,
174
  wet millig, 165
Reass fr, 17, 18
Recrystalliiati durig, 134
Selective; ifluece fBall diameter, 178
Millig bdy desity, 181
Particle diameter, 178
Rd diameter, 180
W et, 25, .30
Mdels, tests smallscale, 42, 69
Millig:
Chage f
chemica1 state
243
Clsed circuit, 24
Dry, 25, 30
durig,
Efficiecy fEquatis fr, 152, 183
Dimesia1 aalysis fr, 183
Depedece up mill speed,
183
   fillg, 183, 186
  pulp charge, 183, 188
   csistecy, 184
Mixed materials, 176
Ope crcuit, 25, 165
Rate fEquati fr, 142, 174
Dimesial aalysis fr, 142
Depedece up ball dam eter, 153
  dry millig, 164
  gridg bdy desity,
150
  1ifters, 197
..,. diameter, 174
   fillig 167
  ___.: speed: 164
   thrughput, 163, 167,
173
  millig ads, 236, 245
Particle, size f; ifluece :
Rate f milling, 153
Metal wear, 226
Pwer, 87
Particles embedded balls. See
under Surfaces, armuring f
Pebb1e mill, 21
Pelletig , 128, 133, 134
Pigmets, 33, 102
Piles:
Edbearig, 219
Fricti, 219
Pwer, depedece up:
Ball diameter, 81, 97
Charge ceffi ciet ffricti, 81
Lifters, 81, 85, 86
Mill diameter, 73, 96, 97
 filling, 81' 103
legth, 72, 99
 speed , 73
Milling additives, 247
 bdy desity, 75
Partic1e desity, 76
 diameter, 87
 gridability, 82
Pu1p charge , 90
 csistecy, 102
 viscsity, 102
 , dimesial aalysis fr, 71
256
SUBJEC T INDEX
Power, equation for, 71, 93, 98, 99,
100, 101, 104
Product:
Contarnina tion of, 31 , 187, 223
Size Distributio n of, 179
Production , rate of. See under
Throughpu t
Quartz, 129
Recrystalli zation, 134
Reduction ratio, 117
Rittinger's law, 117, 149
Rod rnill, 21, 30
Duty , 30
Power to, 106
Vibration , 220
Rolling, tirne , 49, 62
 charge, 49
Sirni1arity, dynarnica1, 69
Slip :
On rnill surface, 35, 67
Limiting condition, constructio n
f, 67
Speed:
Critical, 42
InAuence onArrnouring , 144, 189
Efficiency, 183
Meta1 wear, 225
Power, 73
Rate of milling, 164
Surging, 75,210
Surace:
Arrnouring of, 144, 189
Profi1e of charge, 51, 64, 65
Rate production specific.
See under Milling, rate of
Surging:
Criterion for, 212
Dependenc e onAdditives, 248
Ball diameter, 21
Charge friction, 2 10,2 13
Lifters, 200, 214
Mill diarneter, 210
 filling, 210
Speed, 75, 210
Dirnensional ana1ysis for, 21
Nature of, 36, 203
Tirne of, 208, 209
Throughpu t:
Dependenc e onBall diarneter, 194
Mill diarneter, 196
 filling, 194
Particle diameter, 194
Pu1p consistency, 193
Speed, 194
Equation for rate of, 194, 196
Tirne of Aight, 49, 62
 [ rolling, 49, 62
Torque, fluctuation of, 209, 217
Trajectorie s:
Constructio n for, 45
Equation of, 40, 50, 51
lnterferenc e between, 48
 in, 50
Velocity, irnpact, 137
Vibration:
Frequency spectrurn , 2 19
Origin of, 209,217,2 19
Transmissi on of, 2 19
Water as grinding aid, 242, 243
Wet rnilling:
Power dernand in, 101
Reasons for, 25, 30
Wear, rnetal:
Dimension al analysis for, 224
Equations for, 224, 230, 231, 232
lnf!uence ofArrnouring , 221 , 232
Ball diarneter, 228
Corrosion, 235
Geornetry of trajectory, 35
Hardness of rneta1, 190, 222
 of particle, 190, 222
Mill diarneter, 224
 filling, 223
Particle diarneter, 226
Pulp consistency, 228
Relative hardness, 190, 222,
227
Speed, 225
INDE X
Andreasen , . . ., 134
Anon., 216
Anon ., 219
Anon., 245
Anselm, W., 193
Axelson, ]. W., 122, 132
Baker, . C., 238
Barnacle, . ., 196
Barth, W., 52
Belwinkel, ., 200
Berry, C. ., 241 , 245
Blunt, G. D., 210
Bond, F. C., 84, 99, 122, 74, 233,
240, 242, 243, 250
Bowden, F. . , 134
Cadena, F. , 174
Carnpbell, . . , 73, 97, 99,
103, 4, 86
Carey, W. F., 95, 4, 125
Coghill, W. ., 73, 77, 97, 99,
103, 104, 46 , 5, !56, 156,
178, 186, 187, 9 , 194, 96,
228,234
Couson, ]. ., 23
Craig, . ., 240, 241
102,
102,
74,
224,
Davis, . W., 43, 98, 92, 233
Devaney, F . D ., 77, 46, 150, 156,
174, 78, 187, 9, 94, 196,224,
228, 234
Dobie, W. ., 24
Ellis, . W., 233
Engelhard t, ., 248
Engels, . , 48, 223
Evans, D. . , 37, 42, 65, 66, 69, 76,
99
Fahrenwal d, . W. , 97, 89
Fairchild, D. ., 9
Fais , G. ., 33
Felix, W ., 110
Fobelets, , 52, 60
Garrns, W. . , 233
Geiger, ., 10
Gilliand, . ., 123
Gotte, . , 244, 245
Gow, . ., 73, 97, 99, 102, 103,
104, 174, 86
Gross, ]. , 127, 129, 133, 149, 224,
228
Gr.inder, W., 42
Guggenhei m, . , 97, 99, 102, 103,
104, 174, 186
Hall, R . G., 129
Hancock, R. . , 100
Hinsley, J. F., 52, 204
Holrnes, J. ., 76
Honig, F. , 121 , 122
Hukki, R. ., 232
Hurnphrey sOwen, S. F. . , I 3
H.ittig , G . F., 110, 3 0
Kennedy, . L ., 24
Kick, F., 8
Kwong, J. . S., 129, 3
Lee, . . , 97
Lewis, W . . , 23
Loeb, C. ., 233
Mackenzie , R. C., 34
Manson,. . ., 243
Martin, G., 127, 137
Maxson, W. L. , 74
M cAdarns, W. ., 23
McLeod, . . , 227, 228, 234
Mi1ne, . ., 34
Moeller, J. ., 233
Mor1ey, ., 2 17
Mortsell, S., 226, 233, 234
Moser, F., 147
M.iller, R., 137
Nordquist, D . . , 233
Norrnan, . . , 233
Norris, G. C., 9, 96
257
258
Obreim off, 0., 249
Patchin g, S. W. F., 176
Pearson , . ., 193
Piret, . L., 122 , 129, 131, 132
Prentice, . ., 233
Preston , F. W., 238
Pryor, . J., 243
Ramml er, ., 137
Rehbin der, ., 237, 248, 249
Richard son, J. F., 123
Ri ey, D. ., 13
Rittinge r, . V., 117
R ose, . ., 37, 42, 65, 66, 69, 76,
99, 13, 143, 196, 200, 2s, 210,
22 1
R osin, R ., 137
Schweit zer, C. W., 240, 241
INDEX
Skaupy , F., 236, 249
Smekel, ., 121 , 122, 130
Smith, . ., 223
Sperling, ., 137
Stairma nd, C. J., 95, 104, 125
Starke, . R. , 156, 174
von Steiger, R ., 50
Stevens , J. L., 233
Stott, F. D. , 112
von Szantho , ., 245, 248
Tabor, D., 134
Taggart , . F., 93, 102, 214, 233
Theime r, ., 143
Waeser, ., 244
Walker, W. . , 12 3
Ziegler, ., 244, 245
Zimmer ley, S. R ., 127, 129, 133