Professional Documents
Culture Documents
MECHANICS OF CRUSHING
C. R. MURRY and J. E. HOLT
Defiartment of Mechanical Engineering, Urciversity of Queensland,
Brisbane, Australia
(Presented by Mr. J. L. Clayton)
INTRODUCTION
The effect of roll diameter on crushing mill performance has received a fair amount
of attention in the literature, but, to the authors' knowledge, no experimental results
under controlled conditions have been published. Historically the tendency in the
industry has been towards larger diameter rolls and one suspects that this is based
on sound observation in factory milling. On the other hand occasional reports of the
installation of very large diameter mills never seem to be followed by reports of their
performance and one wonders whether they were not satisfactory in service.
For the purposes of applying results from small experimental mills to factory
size mills, it is important to have some information on diameter effects. Since 3
different roll diameters have been used in the experimental two-roll mill at this
University3~ l4 some data are now available. If the results are to be applied.
49 6g
two dimensionless numbers which specify the milling conditions. The first is the
compression ratio of the mill. This indicates the severity of the crushing and is defined
by
where
C o = compression ratio
q, = mass rate of cane per foot width of mill
W o = worlc opening
S = roll surface speed
do = no void density of cane
The second is Bullock's capacity number which indicates the comparative rate
at which the mill is being fed. This is defined by
C N T 4 C (2)
DSd,
where
C N = capacity number
D = roll diameter
d , = bulk density of cane approaching the mill
That is, all mills having q,/SD the same may be compared. No further theory
is immediately available and we must decide whether to compare mills of different
diameters at the same crushing rate, the same surface speed or the same rotational
speed. The choice appears to be quite arbitrary and may be made for convenience
alone.
HUGOT~O believes that comparisons should be made a t the same rotational speed.
Since surface speeds are usually quoted by the Queensland industry one might infer
that they would make comparisons a t the same surface speeds. A case could be built
for comparison at the same crushing rate since a factory usually has a fairly definite
rate which must be maintained.
Of course, the choice of compression ratio and capacity number as criteria is
somewhat arbitrary, but has support from the dimensional analysis point of view.
For the purposes of the following discussion it will be accepted that mills should
be compared a t the same Co and CN and when crushing the same preparation. The
choice of speed will be investigated from experimental results.
I t has already been shown13that speed has no effect on the torque-load number *
and only slight effectson the roll load and roll torque. The major effect of speed is on
the extraction or, since we are considering constant compression ratio comparisons,
on the reabsorption factor.**
There are two other variables known to affect the milling results, namely, feed
depth and fibre-in-cane. Very little information is available about feed depth but the
indications are that this variable has little effect. The effect of fibre-in-cane has not
yet been completely investigated but, from the limited information so far available,l7
we will assume that:
(i) Fibre-in-cane has no simple systematic effect on reabsorption factor.
(ii) Fibre-in-cane has no effect on the torque-load number.
(iiq The torque and roll load increase linearly with increasing fibre-in-cane (at the
same crushing rate in terms of cane).
Attempts will be made to compare experiments with the same fibre-in-cane but
this will not always be possible since experiments have not been specifically conducted
At speeds above 60 ft./min the extraction drops suddenly for the 18 in, rolls, but it
is felt that this drop is due to juice being carried over in the juice grooves, trapped
by the integral scrapers a t the scraper plate and absorbed in the bagasse.
At speeds below 20 ft./min the extractions from the 18 in. rolls are solnewhat
higher than those from the 26 in. rolls, but this very low speed range is of little prac-
tical interest. I t may be said in general, that the results of the two experiments agree
very well over the practical operating range and i t is suspected that the extraction
is the same at the same surface speed.
The bulk densities* of the preparations indicate that the preparations for the
26 in. rolls are somewhat finer than those for the 18 in, rolls. (Corresponding hammer-
mill* treatments are tabulated on the same line in Table I). This is to be expected
since it has been shown that lower fibre canes give finer preparatioils and that greater
depths give somewhat greater bulk densities2>16, l7. I t was decided, therefore, to
select preparations for comparison with approximately constant ratio of bulk densities
as shown ill Table 11.
Since reabsorption factor has been shown to be an approximately linear function
of compression ratio2> 43 l4 the reabsorption factors for the 18 in. rolls have been
' 9
TABLE I1
SELECTION O F PREPARATIONS
a adjusted to the same compl:ession ratios as those for the 26 in. rolls by linear inter-
polation.
The tests at the same surface speeds were compared. The results on this basis are
* The hainmermill used has been previously described3. and preparation is soinetimes
designated by the speed and time of treatment (e.g. 750115 means treated a t 750 r.p.in, for 15 sec).
Bulk density a t a standard arbitrary test pressure has been shown to be a satisfactory measure
set out in Table I11 using the designations of the 26 in, experiment (Table I) and Dl
to denote 18 in. roll diameter and D z to denote 26 in. diameter.
Tables I11 and IV show that there is very little diameter effect for these experi-
ments. The significance level of diameter effect is only 0.20 and the diameter effect
is not significant when compared with its interactions with compression ratio or
speed. That is, all difference between diameters can be attributed to these interactions
which are themselves not very significant.
TABLE I11
REABSORPTION FACTOR
61 c 2 c3 c4
Dl Dz Dl Dz Dl Dz Dl Dz
P~ Sz 1.21 1.24 1.08 1.15 1.07 1.12 1.12 1.03
S4 1.51 1.49 1.19 1.18 1.26 1.15 1.25 1.13
TABLE IV
ANALYSIS O F VARIANCE
Degrees of Varzance Significance
Source
freedom ratzo level
C 3 363.51 0.001
P 2 76.22 0.001
S I 160.05 0.001
D I 2.92 0.20
CP 6 38.70 0.001
CS 3 18.37 0.01
PD
SD
CPS
CPD 6 - A
CSD 3 2.05 -
PSD 2 1.30 A
TVPSD 6
Mean square of error interaction = 0.00185
I t may be said with some confidence that the effect of diameter is small compared
with the effects of either compression ratio, preparation, or speed.
Referring to Table I, the five speeds on the 26 in. rolls, when compared with the
speeds for the 18 in. rolls, give, approximately
(i) the same crushing rates (S1 and S,) ;
(ii) the same surface speeds (Sz and S,) ;
(iii) the same rotational speeds (Sa and S 5 ) .
Since there are no major interactions with speeds in Table IV, it is not unreason-
C. R. MURRY, J. E . HOLT 1°79
able to compare the average reabsorption factors for the various speed combinations.
This comparison is set out in Table V.
Tablevshowsthat thebest agreement isbetweenthe testsat the same surface speed
which agrees with the result of Fig. I. I t must be remembered that there are differences
in fibre between the experiments, but it is thought that reabsorption factor is not
much affected by fibre-in-cane.
Some further evidence is available by comparing the reabsorption factor at
30 ft./min for the experiments of Fig. I with the results of an experiment using 22 in.
TABLE V
AVERAGE REABSORPTION FACTORS
Roll diameter
1 18 zn. 26 zn.
(i) Same crushing rate
(ii) Same surface speed 1.288 I ,267
(iii) Same rotational speed 1.288 , 1.332
diameter rolls. The cane used in this experiment was of 13.1% fibre and so compares
with the experiments of Fig. I. The experimental conditions were: fine preparation,
surface speed 30 ft./min and a range of compression ratios from 1.5 to 3.5. The
crushing rate was such as to compare with the experiments of Fig. I under condition
of Eq. (5). The reabsorption factor for compression ratio 3.0 was 1.24. That for
Fig. I at the same conditions is 1.23.
D~arneter( ~ )n
TABLE VI
ANALYSIS O F VARIANCE
Roll load per f t . (lb. x 1000)
PD 2
CPD 6
Table VI shows very little diameter effect, but this is almost certainly due to
the compensating effect of lower fibre on the 26 in. rolls. There appears to be a fairly
marked compression ratio-diameter interaction which examination of the experi-
mental results shows is due to a somewhat greater response to compression ratio
changes with the larger diameter rolls. The effect is illustrated by the averages given
in Table VII. I t should be remembered that this interaction might be a fibre effect.
The analysis of the torque results is given in Table VIII. As might be expected
it is similar to the roll-load analysis, but the compression ratio-diameter interaction
is much less marked.
Vevijcation of Eq.(6)fov 22 in. and 26 in.rolls
Before proceeding to discuss the variation in e with roll diameter we must
verify that the form of the function relating the torque-load number to the compres-
sion ratio does not change substantially with diameter. That is, that the variation
.is still proportional to COPas determined experimentally for the 18 in. rolls. For this
purpose the results of an experiment at constant preparation and surface speed on
the 26 in. rolls were selected. The crushing rate was constant and the work opening
was varied to give various compression ratios. The values of e from Eq. (6) as found
from this experiment are set out in Table IX and the analysis of variance of these
results is given in Table X.
There is no evidence in these results that e is affected by compression ratio and
so we may assume that the variation in torque-load number with compression ratio
has the same form. The grand mean value of Q for the experiments is 0.265.
Values of Q for the 22 in.-diameter experiment are given in Table XI which
shows no significant trend with compression ratio.
A further verification is given from the results of the factorial experiment of
Table I for the 26 in. rolls which shows no significant interactions for the values of Q,
TABLE IX
VALUES OF Q FOR 26 in. DIAMETER
" cO2 N
(W,/D):
Compression
ratio 1.5 1.9
TABLE X
ANALYSIS O F VARIANCE
Values of Q
TABLE XI
VALUES OF Q FOR 22 in. DIAMETER
no speed effects and a compression-ratio effect significant at only the 0.05 level.
The preparation effect is markedly significant (0.001 level). The mean values of Q
are 0.21, 0.24, and 0.30 for PI, Pz, and Pa respectively.
The variation between values at the same diameter are probably due to small
differences in preparation. I t will be seen that, approximately, Q is inversely pro-
portional to the roll diameter.
TABLE XI1
I M E A N VALUESOF Q
(Fine preparation)
Roll diameter Source e
18 in. Ref. 4 0.39
Varying speed experiment (Fig. 2) 0.36
2 2 in. 0.30
26 in. Table IX 0.26
Varying speed experiment (Fig. 2) 0.26
Table I factorial experiment 0.24
Hence from HUGOT'S information his mills operate a t a Q of 0.165. The authors
do not have information as to HUGOT'Sroll diameters, but, assuming they follow
g'eneral Queensland practice with which the authors are familiar, it may be assumed
that the diameter is about 36 in. for this result. On this basis, Fig. 4 has been drawn
, using the values of Table XII. I t will be seen that the value of Q deduced from HU-
GOT'S formula does not disagree with the hypothesis that Q is inversely proportional
to the roll diameter.
0
0 0 . 0 2 0.04 0.06
Rec~procalof Roll D~arneter
(In:')
TABLE XITI
M E A N VALUES OF Q
For voll dzalneter
Ratio
26 zn. 18 zn.
Fine preparation 0.21 0.36 1.71
Mediuin preparation 0.24 0.39 1.62
Coarse preparation 0.30 0.42 1.40
The variation in energy consumed per ton with roll diameter may immediately be
deduced as follows.
Froln the foregoing information, assume that the roll load is proportional to
the diameter and the torque-load number inversely proportional to the diameter at
the same coinpression ratio, work opening-diameter ratio, and preparation. That is:
Also both R and N are almost independent of speed. The power P is given by
P cc RSN
Hence
P oc S and is independent of D.
Now crushing rate is given by
Qc = C oL W o S d o
where
L = roll length
that is
Qc cc D S since W , cc D
C . R. MURRY, J . E. HOLT 1085
Therefore, the energy absorbed per unit weight of cane crushed is given by
w = P/QGcc I/D
This is the power for crushing only and allowance must be made for the friction
power of gearing, bearings, and trash plate (if any). Results obtained by SMART^^
suggest that the actual crushing power is only about 58% of the total engine power.
Hence the engine power required is about 170 h.p. which is a not unreasonable esti-
mate for a first mill at 116 toils cane/h.
I
SUMMARY AND CONCLUSIONS
411 attempt has been made to develop a logical method of comparing the performance of mills of
different diameters and it was suggested that mills should be compared a t the same compression
ratio and work opening-diameter ratio when crushing cane of the saine fineness of preparation
(this is a t the saine capacity number). I t was found that under these conditions the crushing rate
must be proportional to the product of roll speed and diameter and that, therefore, an arbitrary
qhoice o f speed could be made.
Analysis of the experimental results suggested t h a t the extraction was not dependent on
diameter a t the saine surface speed and, slnce roll load and torque are almost independent of
speed, it seems that null performances should be compared a t the saine surface speeds.
On this basis it appears that the reabsorption factor 1s illdependent of roll diameter, that the
roll load IS proportional to the roll diameter and t h a t the torque-load number IS Inversely pro-
portlonal to the roll diameter From this it may be deduced that the torque is proportional to the
roll diameter Also ~thas been shown that the energy absorbed per ton cane crushed is lilrely to
be Inversely proportional to the roll diameter and that, in this sense, larger rolls are more efficient.
ACKNOWLEDGEMENTS
The authors are indebted to Professor M. SHAWfor his encouragement and advice,
to Mr. F. P. SMITHfor assistance in the laboratory and in preparation of the results
for publication and to the Sugar Research Institute and Colonial Sugar Refining
Co. Ltd, for provision of the fellowships under which much of the work is being
carried out.
REFERENCES
S u m m a r y of Cane 1?.tillingand sugar manufacturing research work. Sug. Res. Inst. Mackay, 1960,
p. 12.
The influence of roller load and speed on the performance of a hydraulically loaded No. 2 mill.
Tech. Rep. Sug. Res. Inst. Mackay, N o . 68, p. 31.
3 BULLOCK, I<.J., 1956. Self-feeding characterlstlcs of an experimental sugar mill. Proc. ilzt. Soc.
Sug. Cane Tech., 9 [z] : 166-85.
BULLOCIZ, K. J., 1957. A n investigation into the cvushilzg and physical properties of sugar cane and
bagasse. Ph. D . Thesis, Univ. Queensland.
BULLOCIZ, I<. J., 1958 The capacity of sugar cane mills. Sugar J., N.Orleans, 21 [5], [GI, [7].
0 BULLOCIZ, I<. J , and MURRY,C. R., 1960. Extraction as a function of roll load in the experimental
mill. Proc. znt. Soc. Sug. Cane Tech., 10 : 111-7.1.
7 CRAWFORD, W. R., 1957. Improving extraction a t No. I mill. Proc. Qd Soc. Sug. Cane Tech.,
24 : 89-103.
CRAWFORD, W. R., 1960. Reabsorption - A limiting factor in mill performance. Pvoc, i n t . Soc.
Sug. Cane Tech., 10 : 100-11.
9 HUGOT, E., 1960. Handbook of Cane Sugar Engzneering. Elsevier Publ. Co., Amsterdam, p. 112.
lo HUGOT, E., IbZd , pp. 143-147.
11 HUGOT, E., Ibid., p. 187.
l2 HUGOT, E., Ibzd., p. 168.
13 MURRY, C. R., 1959. Mill torque as a function of roll loading. I n t . Sug. J.,61 : 46.
l4 MURRY,C. R., 1960. Sugar crushing research - A survey of the theory developed and experi-
mentation carried out in Queensland since 1952. Trans. Inst. Engrs. Aust., Elec. G. Mech.
Eng., EM2 No. 2 : 31
15 MURRY, C. R., 1960. A theoretzcal and exfierimelztal investzgation znto the mechanics of crushing
firefiared sugar cane. Ph. D. Thesis, Univ. Queensland.
16 MURRY,C. R., 1960 The pressure required to feed cane mills. I n t . Sug. J., 62 : 346-349;
63 : 7-10,
17 MURRY, C. R., 1961 Milling Studies - The effects of fibre-in-cane and roll diameter. Proc. Qd
Soc. Sug. Cane Tech., 28 : 109-115.
18 MURRY,C. R. and SHANN, D. S., 1960. Effects of degree of preparation on the Queensland
experimental mills. Proc. iint. Soc. Sug. Cane Tech., 10 : 122-31.
19 SMART, H. P . , 1961. A mill steam turbine drive investigation. Proc. Qd Soc. S u g . Cane Tech.,
28 : 163-172.
DISCUSSIONS
H. JENIZINS (Australia) : One would expect extraction for different size of rolls to be the same a t
similar rotational speed since time under pressure would appear to be an important factor in
juice drainage and therefore in extraction. This suggests that rotational speed rather than surface
speed would be the main factor influencing extraction. If Fig. I be replotted to give extraction
as a function of rotational speed the points for the two different roller diameters fall almost on
the same line.
J. CLAYTON (Australia) : I cannot accept that the governing factor is time because in one
case there is only a short way for the juice to travel whilst in the other the juice travels a con-
siderably longer distance.