Download as pdf or txt
Download as pdf or txt
You are on page 1of 39

Polymer Testing 5 (1985) 269-307

A Comparison of Tensile Results Obtained on


Rubber Test Pieces Prepared by Moulding,
Buffatg and SpHttlng

D . I. J a m e s a n d J. S. G i l d e r
RAPRA, Shawbury,Shrewsbury,Salop SY4 4NR, UK

SUMMARY

Buffing has been the accepted method of preparing tensile test pieces
for many decades; tensile results from a wide range of rubber com-
pounds have indicated that splitting with a leather splitting machine is
equally satisfactory. Comparison of the ratios of tensile strength to
elongation at break for orthogonally cut test pieces is a good indicator
of grain effect, while similar comparison of the products of TS and EB
gives an indication of the influence of surface texture. It has been
established that milling grain or likely variation in the degree of cure
has more in[luenee on tensile test results than surface texture induced
by the method of sample preparation. In general, buffing is the
preferred method for soft compounds and splitting for hard ones.

1. INTRODUCTION

Rubber test pieces for tensile testing are most conveniently prepared
from sheets moulded to give the required test piece thickness. When
testing rubber products, however, it is necessary to produce flat
sheets either by buffing or by splitting. The results obtained on
nominally identical rubbers prepared by moulding and by butfmg are
different and Morley and Scott 1 gave recommendations to keep this
difference to a minimum. They emphasised a point made earlier by
269
Polymer Tesang 0142-9418/85/$03.30 © Elsevier Applied Science Publishers Ltd,
England, 1985. Printed in Northern Ireland
270 D. L James, J. $. Gilder

Reece, 2 for example, that cut dumb-bells have a trapezoidal rather


than a rectangular cross-section and if test pieces are first cut and
then buffed to the required thickness the final mean width will
depend on whether the wide or the narrow side has been buffed
away. Morley and Scott recommended that buffing should precede
the cutting of the dumb-bells, and the smoothest possible surface
should be obtained. Both of these recommendations, long since
incorporated in a British Standard Specification,s have been followed
in the present work.
Following the development of an automatic buffing machine at
R A P R A (now marketed by H. W. Wallace of Croydon), Brown and
Jones 4 confirmed that in general buifmg led to lower tensile figures
than were obtained on test pieces cut from moulded sheets. They
reported also that in some cases buffing appeared to bring about a
considerable drop in elongation at break and a large increase in
modulus, although the present authors feel that this was possibly due
to a grain effect.
In a related study, Brown 5 compared results obtained on test pieces
from sheets taken from products by splitting with those obtained
from laboratory prepared sheets. In general the tensile results ob-
tained from the products were lower than those obtained from
moulded sheets, but it was not claimed that this was due to the
method of preparation, but rather to a different degree of cure.
Brown's work emphasises an underlying difficulty when attempting
to compare different methods of sample preparation, namely that test
pieces taken from different sheets may be at different states of cure.
Equally, sheets cut from the surface of a block will differ from those
cut from the interior and due allowance must be made for this. Add
the possibility that the sheets may be anisotropic and the difficulties
are compounded.
Because in the present work an attempt has been made to allow for
both these factors, the resulting experimental programme is necessar-
ily complex. The object was to compare tensile test pieces cut from
moulded sheets with those from sheets prepared by splitting or by
buffing or sheets in which different operations were used to prepare
each face. Complicating influences were:

(1) grain in the test pieces from milling or other process opera-
tions;
Tensile results obtained on rubber test pieces 271

(2) different degrees of cure b e c a u s e of different initial test piece


thicknesses;
(3) different degrees of cure t h r o u g h o u t the thickness of a block,
(4) grain induced in the test pieces by the m e t h o d s of sample
preparation, namely splitting or buffing;
(5) different degrees of carbon black distribution.
A n a t t e m p t was m a d e to t a k e s o m e account of all these factors
except (5). C a r b o n black distribution was assumed to b e identical in the
test sheets p r e p a r e d from each c o m p o u n d since they were all
m o u l d e d from the same stock.

2. SPECIMEN PREPARATION

T h e five different r u b b e r c o m p o u n d s used for this investigation are


listed in the Appendix. F r o m each c o m p o u n d 150 m m square sheets
of 2, 6 and 12 m m thickness w e r e m o u l d e d to nominally the same
d e g r e e of cure.
M o u l d e d sheets of a thickness specified for tensile testing are
unique, and it is rarely possible to p r e p a r e from t h e m buffed or split
sheets of a thickness which remains within the specified range of
2-0 + 0 - 2 mm. C o m p a r i s o n must b e m a d e with sheets p r e p a r e d from a
thicker block, which m a y perhaps have a different degree of cure.
H o w e v e r , if o n e m o u l d e d surface of the thicker sheet is left undis-
t u r b e d it is then possible to r e m o v e the opposite surface by splitting
or buffing and so c o m p a r e m o u l d e d / b u f f e d test pieces with
moulded/split test pieces.
F o r comparison of samples with b o t h surfaces either split or buffed
it is important to select areas taken from identical depths in the thick
block in o r d e r to ensure, as far as possible, the same degree of cure.
T h e complex cutting pattern set o u t b e l o w was designed to high-
light, as far as possible, the various influences (1) to (4) listed above.
Buffing was carried o u t according to the m e t h o d described in BS903,
Part A23 using a Wallace buffing machine type 55A. Split surfaces
were p r e p a r e d on a S A S L e a t h e r Splitting Machine type F O N 6 5 . The
cutting blade in this machine forms a continuous b a n d moving at high
speed. This blade is lubricated with a commercial lubricant believed
to b e b a s e d on polyethylene glycol. T h e t r e a t m e n t of each of the
three thicknesses of r u b b e r is set o u t below.
272 D. L James, Jr. S. Gilder

2.1. 2 mm moulded sheets

E a c h 150 m m square sheet was cut into two equal rectangles and test
pieces p r e p a r e d according to the cutting layout illustrated in Fig. 1.
Sheet 'p' with the axes of the dumb-bells parallel to the shorter
side of the rectangle was said to have b e e n cut in the X orientation
and sheet 'q' with the axes of the dumb-bells parallel to the longer
side of the rectangle was said to have been cut in the Y orientation.
T h e dumb-bells were labelled accordingly.
T h e direction of the grain in the rubber was not known, so that 'p'
and 'q' indicated a cutting layout only. However, because the blanks
for moulding are always cut orthogonally from the stock as it comes
off the mill, it can be assumed that the grain runs parallel to the
dumb-bell axes in o n e case and perpendicular in the other, although
it is not known which is which.

2.2. 6 mm sheets

T h e 6 m m sheets were used for comparing test pieces p r e p a r e d by


splitting with those p r e p a r e d by buffing. E a c h 6 m m sheet was split

X orientation Y orientation

Fig. L Cutting layout for 2 mm moulded sheets.


Tensile results obtained on rubber test pieces 273

into two nominally identical 3 m m sheets and these were labelled 1


and 2. The action of splitting left fine lines parallel with the blade, i.e.
perpendicular to the direction of feed. Each 3 m m sheet was divided
into two equal rectangles along a line perpendicular to the lines
created by splitting. These were labelled so that sheet 1 yielded
rectangles A and B and sheet 2 rectangles C and D. Labelling was
such that in the original 6 m m block A and C had been in contact as
had B and D. This is clear from the cutting layout shown in Fig. 2.
The moulded surfaces were removed from A and C by splitting,
the sheets being fed into the machine with their long axes perpen-
dicular to the splitting knife. Similarly B and D were buffed to
remove the moulded surface, the direction of travel of the buli~g
wheel being parallel to the long axis in each case. This produced
abrasion ridges perpendicular to the direction of buffing. The sheets
were then brought to 2 m m thickness in such a way that A and C
yielded split/split test pieces and B and D buffed/buffed, the splitting
and buffing directions previously described being preserved. From
sheets A and B dumb-bells were cut in the X configuration and

Direction of splittingl Direction of buffing

s/s/x btb/x
A B Sheet 1

I slslx lib b/b/x 12


I c sisly II D b/b/y I'

s/s/y b/b/y
C D Sheet 2

Fig. 2. Cuttinglayout for sheets cut from 6 mm blocks.


274 D. I. James, Z $. Gilder

similarly dumb-bells were p r o d u c e d from sheets C and D in the Y


configuration. This is illustrated in Fig. 2, w h e r e it is immediately
apparent that X and Y n o w indicate the orientation of splitting or
buffing on the surfaces of the dumb-bells.

2.3. 12 mm sheets

Originally it was the intention to split each 12 m m sheet into five


pieces labelled 3, 4, 5, 6 and 7 b u t to use only the four outside sheets,
namely 3, 4, 6 and 7. With s o m e m a t e r i a l s this p r o v e d to b e
exceedingly difficult and in these cases only four pieces w e r e o b -
tained. F o r continuity, the original concept of labelling was retained
and the sheets used were labelled 3, 4, 6 and 7 as before. Addition-
ally, in s o m e cases it was f o u n d necessary to cut the 150 m m square
12 m m thick sheets into two rectangles b e f o r e splitting b e c a u s e of the
large forces involved in cutting whole sheets. In all cases the direc-
tions of splitting and the labelling were in accordance with the
scheme set o u t in Fig. 3, although w h e n the 12 m m sheets were
initially cut into two rectangles each of the sheets L and M consisted
of two squares. T o avoid further confusion, sheets L and M are
referred to as complete rectangles in the description below.
E a c h of the four square sheets n u m b e r e d 3, 4, 6 and 7 yielded two
rectangular sheets measuring 150 × 75 mm. N u m b e r s 3, 4 and 7 were
cut normal to the splitting lines, and n u m b e r 6 was cut parallel to the
splitting lines.
T h e rectangular sheets were lettered according to the scheme

3 Direction of feed into the


splitting machine in each case,
including Sheet 6
4

Fig. 3. Labelling of sheets cut from 12 mm blocks.


Tensile results obtained on rubber test pieces 275

shown in Fig. 3. Because they w e r e likely to have the same degree of


cure, sheets E, F, G and H formed o n e h o m o l o g o u s set and J, K, L
and M formed another.
The m o u l d e d surfaces were left on E, F, G and H. Sheets E and G
were reduced to 2 m m thickness by splitting and F and H by b u t ~ g .
T h e direction of splitting was such that in each case the splitting
blade was normal to the major axis of the sheet. T h e direction of
buffing was parallel to the major axis of the sheet as in the specimens
prepared from the 6 m m sheet.
T h e split surfaces of sheets J, K, L and M which were adjacent to
E, F, G and H were retained and the opposite sides buffed to reduce
the sheets to 2 m m thickness. In all cases buffing was carried out
parallel to the major axis of each rectangle (as in every case so far

Sheet 3 Sheet 7

E F G H
m/s/x m/b/x m/s/y m/b/y

l Sheet 4 l Sheet 6

slbl× L

_ m
M

J K

s/b/x s/b/y s/b/y


Fig. 4. Cutting layout for sheets from 12 mm blocks. Sheets 3 and 7 are moulded
on the unseen face. The arrows indicate the direction of splitting or buffing on the
unseen faces of Sheets 4 and 6.
276 D. L James, J. S. Gilder

described). However, because sheet 6 had been cut parallel to the


splitting marks into sheets L and M, these two sheets yielded material
which had been buffed parallel to the long axis on the one side and
split perpendicular to the long axis on the other. This may be
contrasted with sheets J and K where buffing and splitting had been
carried out in each case parallel to the long axis of the rectangle.
When tensile test pieces were cut, the X layout of Fig. 1 was used
for sheets E, F, J and L and the Y layout for G, H, K and M.
The whole procedure for cutting the 12 m m sheets is summarised in
Fig. 4.

3. M I C R O S C O P I C A L SIGNIFICANCE OF
T H E LABELS X A N D Y

Sheets moulded to a mirror finish show no noticeable surface texture


(Fig. 5) and any directional effect due to the cutting orientation is
then due solely to grain effects induced by milling. In general the
grain direction on a moulded sheet is not known. The designations 'p'
and 'q' then merely indicate that test pieces were cut in orthogonal

Fig. $. Mouldedsurface (mirror finish), ×450.


Tensile results obtained on rubber test pieces 277

directions and differences in test results m a y be interpreted as being


d u e to anisotropy in the m o u l d e d sheet.
In cases w h e r e the m o u l d has b e e n ground, a surface texture is
imposed on the r u b b e r and 'p' and 'q' t h e n indicate that the grinding
marks run parallel to or perpendicular to the axis of each dumb-bell.
In addition there m a y be an underlying grain due to milling, the
direction of which is not known.
Microscopical examination of the m o u l d e d surfaces used for this
series of experiments showed that both types of mould had b e e n
used. A mirror surface has already b e e n shown in Fig. 5 and for
comparison a g r o u n d surface is shown in Fig. 6. A l t h o u g h this
difference in the two types of m o u l d was identified, no further
account was taken of it.
Splitting produces a surface w h e r e the lines run parallel to the
splitting knife (perpendicular to the direction of splitting). A n exam-
ple is shown in Fig. 7.
Buffing produces abrasion ridges perpendicular to the direction of
buffing. T h e surface texture is quite characteristic and is unlike any of
the others as m a y be seen f r o m Fig. 8.
Since the directions of splitting and buffing w e r e carefully laid out

l ~ . 6. Moulded surface (ground finish), x450.


278 D. L James, J. S. Gilder

Fig. 7. Split surface, x450.

i ' - . +

f .... _ . _ .

Fig. 8. Buffed surface, x450.


Tensile results obtained on rubber test pieces 279

in Figs. 2 and 4, the letters X and Y give a clear indication of the


direction of the ridges d u e to splitting or buffing. F o r symmetrically
faced test pieces the designation X implies that the lines d u e to
splitting, or the ridges caused by buffing run parallel with the axis of
the d u m b - b e l l and the designation Y that such lines are perpendicu-
lar to the axis. In the case wh.ere o n e surface is moulded, there m a y
b e s o m e insignificant orientation on the m o u l d e d surface if this bears
a ground finish, b u t b e c a u s e this is far less influential than the marks
p r o d u c e d by splitting or butfmg it has b e e n ignored in the analysis of
results given later.
In the case of sheets H , J and K w h e r e o n e surface is split and the
o t h e r buffed, the meaning of the designations X and Y remains
unchanged and samples labelled X will have lines or ridges parallel
with the axis as b e f o r e and those labelled Y will have lines p e r p e n -
dicular to the axis. Sheets 6 L and 6 M present something of a p r o b l e m
in that the lines d u e to splitting and buffing are normal to each other.
F o r sheet 6 L X the buffing ridges run parallel to the axes of the
d u m b - b e l l s and the splitting lines perpendicular to the axes, whereas
for sheet 6 M Y the splitting lines run parallel to the axes and the
buffing ridges perpendicular. Only in the case of m o u l d e d surfaces is
t h e r e any d o u b t a b o u t w h e t h e r there is any surface orientation b u t it
is p r o b a b l e that the effect on the experimental results is negligible.

4. SUMMARY OF THE CUTTING LAYOUT

T h e surface orientations of the 2 m m m o u l d e d sheets were different


for each mix used. F o r completeness they have all b e e n listed in
T a b l e 1. T h e designations have b e e n p u t in parentheses to indicate
that they are of lower significance than those applicable to the split or
buffed surfaces.
T h e patterning d u e to splitting or buffing was the same for all the
mixes used. H e n c e the s u m m a r y p r e s e n t e d in T a b l e 2 applies to each
of the five c o m p o u n d s .
A s a m a t t e r of general observation it should b e r e c o r d e d that
usually the softer c o m p o u n d s w e r e difficult to split and the harder
c o m p o u n d s difficult to buff, and to s o m e limited extent this is
reflected in the test figures q u o t e d later. It should not b e taken as a
280 D. L James, J. S. Gilder

TABLE 1
Surface Orientation of 2 mm Moulded Sheets

Cutting Orientation of grinding lines


Compound orientation with respect to axis of dumb-beU

Z75A p (Perpendicular)
Z75A q (Parallel)
Z75B p (Perpendicular)
Z75B q (Parallel)
X20A p (Parallel)
X20A q (Perpendicular)
X20B p (Perpendicular)
X20B q (Parallel)
W154 p (Glossy)
W154 q (Glossy)

TABLE 2
Surface Orientation of Prepared Sheets

Orientation of Orientation o[
lines or ridges lines or ridges
Cutting relative to axis relative to axis
Sheet orientation Side 1 of dumb-bell Side 2 of dumb-bell

1A X Split Parallel Split Parallel


2C Y Split Perpendicular Split Perpendicular
1B X Buffed Parallel Buffed Parallel
2D Y Buffed Perpendicular Buffed Perpendicular
3E X Moulded (Ground) Split Parallel
7G Y Moulded (Ground) Split Perpendicular
3F X Moulded (Ground) Buffed Parallel
7H Y Moulded (Ground) Buffed Perpendicular
4J X Split Parallel Buffed Parallel
4K Y Split Perpendicular Buffed Perpendicular
6L X Split Perpendicular Buffed Parallel
6M Y Split Parallel Buffed Perpendicular
Tensile results obtained on rubber test pieces 281

universal rule, however, as with C o m p o u n d W 1 5 4 , for example,


splitting gave results superior to those o b t a i n e d b y buffing.

5. T E S T I N G

T h e test conditions used were those set o u t in BS903, Part A2,


namely a test s p e e d of 500 m m / m i n and a gauge length of 20 mm. In
addition to tensile strength and elongation at break, it was possible to
n o t e tensile moduli at 100 and 2 0 0 % since automatic recording of
the elongation was used. W h e r e it was possible, 3 0 0 % m o d u l u s was
also noted. In s o m e cases this was not possible as tensile failure
occurred b e t w e e n 200 and 3 0 0 % extension.
T h e results are summarised in Tables 3 to 7 w h e r e b o t h m e a n s and
m o d e s are q u o t e d for tensile strength and elongation at break, and
m e a n s only for moduli. T h e m e t h o d b y which the m o d e s were
calculated and the reasons for doing so will b e discussed later.
T h e n u m b e r of test pieces used in each case was the m a x i m u m
n u m b e r that could b e cut from each p r e p a r e d strip. As can b e seen
f r o m Tables 3 to 7 this was never m o r e than 11 b u t in s o m e cases was
as low as 3 (zero if 3 E X of mix W 1 5 4 is included). T h e n u m b e r of
test pieces successfully cut from a strip p r e p a r e d by a particular
m e t h o d is s o m e m e a s u r e of the difficulty e n c o u n t e r e d with the
c o m p o u n d u n d e r consideration. In those cases w h e r e the n u m b e r s
are very small it implies either that the surface was very irregular or
that the thickness variation left patches outside the tolerances set b y
BS903, Part A2.

6. RESULTS

B e f o r e looking at the results in detail, it is w o r t h emphasising that


n o n e of the m e t h o d s of preparation used is so unreliable that it must
b e r e j e c t e d o u t of hand. A glance d o w n the figures in Tables 3 to 7 is
sufficient to indicate that all the results are within the limits of
variation that any experienced r u b b e r technologist w o u l d expect from
r e p e a t mixes (although the variation is rather greater than the s o m e -
what unrealistic limits set b y A S T M Specification D 4 1 2 - 8 0 ) . 6 It is
282 D. L James, J. S. Gilder

TABLE 3

Mean Coefficientof Mode Coefficientof Mean ~ t of


Number ~ variation TS variation E13 variation
Code Oriemaaon Surfaces tested tMPa) (%) (MPa) (%) (%) (%)

M p m/m 11 24.2 13 25-5 10 370 11


M q m/m 10 25.4 9 26.4 7 360 10
1A X s/s 10 24.3 4 24.7 3 370 8
2C Y s/s 10 23.0 14 24.4 10 370 12
1B X b/b 9 23-6 10 24.6 7 370 8
2D Y b/b 9 22"0 12 23.2 9 360 13
3E X m/s 10 24-5 6 25-1 5 370 5
7G Y m/s 9 23-6 7 24-3 5 390 9
3F X m/b 10 23.2 5 24.8 11 370 14
7H Y m/b 7 23.0 14 24-4 9 360 13
4J X s/b 9 23.9 10 24.9 7 390 11
4K Y s/b 9 22.3 15 23.8 11 360 14
6L X s*/b" 8 24"9 7 25"6 5 400 10
6M Y s*/b 10 24"9 6 25'5 4 410 4

a s * / b i n d i c a t e s t h a t t h e d i r e c t i o n o f s p l i t t i n g is n o r m a l to the direction of buffing.

TABLE 4

Mean Coef~ient of Mode Coe~ient of Mean ~ t of


Number "IS variation 1S variation EB variation
Code Orientaaon Surfaces tested (MPa) (%) (MPa) (%) (%) (%)

M p m/m 11 14.1 4 14.4 3 210 6


M q rrdm 11 15-9 7 16'4 6 200 8
1A X sis 9 15-8 9 16-5 7 200 11
2C Y sis 11 13-3 10 13"8 8 210 10
1B X b/b 10 15-0 9 15.7 7 200 10
21) Y b/b 10 12.7 8 13.2 6 200 10
3E X m/s 9 13.5 11 14.1 8 220 14
7G Y ntis 9 15-1 10 15.8 7 190 10
3F X ndb 11 13-7 8 14"1 6 220 11
7H Y m/b 10 15.5 3 15-6 2 190 5
4J X s/b 7 14.5 10 15.1 7 230 11
4K Y s/b 7 14.6 7 15.1 6 190 6
6L X s*/b a 9 15"2 12 16.1 9 200 8
6M Y s*/b 9 12.2 10 12.8 7 190 15

a s * / b i n d i c a t e s t h a t t h e d i r e c t i o n o f s p l i t t i n g is n o r m a l to the direction of buffing.


Tensile results obtained on rubber test pieces 283

Results for Compound Z75A

Coeff~ent of 100% C o e f ~ e m of 20O% C.oeff~/ent of 300% Coeff~c/ent of


EB variation Modulus variation Modulus variation Modulus variation
Code (%) (%) (MPa) (%) (MPa) (%) (MPa) (%)

M 390 9 3.6 9 10.7 10 18"9 7


M 380 8 4"0 8 11-6 9 20-0 9
1A 390 6 3.9 6 10.5 8 18.4 7
2C 390 9 3.6 9 10.1 6 18"1 5
1B 390 6 4-0 7 10.6 7 18"1 9
2D 380 10 4.0 9 10'2 8 17.3 5
3E 380 3 4.3 7 11.1 6 19.0 4
7G 410 7 3-9 8 10.2 8 17.6 7
3F 390 9 3-9 7 10-4 8 18-0 7
7H 380 9 3-9 4 10.5 7 18-2 7
4J 410 8 3-8 7 9-9 7 17-4 6
4K 390 10 3-6 5 10.1 7 17.7 7
6L 410 7 3.8 8 10.1 8 17-9 6
6M 420 3 3'5 4 9.5 4 17.0 3

Results for Compound Z75B

Mode C_.oeff-~em of 100% ~ t of 200% Coefftciemof Number


EB variation Modulus variation Modulus variation reaching
Code (%) (%) (MPa) (%) (MPa) (%) 200%

M 220 5 7.6 6 13-1 7


M 210 7 9.0 6 15'8 3
1A 210 8 9.0 8 15.6 6
2C 220 12 6"9 6 12.4 3
1B 200 13 8'6 7 15.2 6
2D 200 8 7.2 9 12.6 5
3E 240 10 6-8 4 12.1 4
7G 200 8 8.8 6 16"0 4
3F 230 9 6.7 11 12.2 9
7H 190 9 9.1 6 15-6 0
4J 240 8 6"5 11 12-0 6
4K 200 4 8-1 8 14.6 0
6L 200 7 8-3 5 15.8 5
6M 200 10 7.0 13 12.0 4
284 D . L J a m e s , J. S. Gilder

TABLE 5

Mean ~ of Mode Co~'ientof Mtean C.oelT-~of


Number "IS variation TS variation EB variation
Code Orientation Surfaces tested (1~'a) (%) (MPa) (%) (%) (%)

M p m/m 11 22-9 2 23.2 3 670 3


M q m/m 11 22.8 2 23.1 2 670 3
1A X s/s 4 22.1 2 22.3 2 670 5
2C Y s/s 7 22.3 4 22-8 3 660 2
1B X bfo 5 22.7 4 23.0 3 680 4
2D Y bfo 7 21.7 3 22-2 3 660 4
3E X m/s 6 22.7 3 23.0 2 680 2
7G Y m/s 8 22.5 2 22.7 2 720 5
3F X m/b 10 22.4 3 22.7 2 690 4
7H Y m/b 4 23"0 2 23"1 1 690 4
4J X s/b 8 22"8 4 23"2 3 720 5
41( Y s/b 11 22'5 3 22"7 3 690 5
6L X s*Fo" 10 22.6 4 22.9 3 730 3
6M Y s*Fo 8 22'3 5 22.8 4 670 4

a s*/b indicates that the direction of splitting is normal to the direction of buffing.

TABLE 6

Number "IS variation "IS variation EB variation


Code Or~taaon Sur[aces tested (l~a) (%) G~IPa) (%) (%) (%)

M p m/m 10 11.7 6 12.0 5 620 3


M q m/m 8 11.9 4 12.1 3 640 5
1A X s/s 10 11.8 8 12.3 6 630 3
2C Y s/s 6 11"6 3 11"7 2 620 3
1B X b/b 10 12.5 7 12-9 5 640 4
2D Y b/b 7 11.0 11 11.6 8 600 6
3E X m/s 9 11.5 13 12-2 9 600 7
7G Y m/s 9 11.0 11 11.6 8 620 6
3F X m/b 10 11-9 7 12-3 5 630 5
7H Y m/b 8 11-5 5 11-8 4 610 3
4J X s/b 8 12.3 3 12-5 3 650 3
4K Y s/b 10 10.9 7 11-3 11 620 6
6L X s*fo° 10 10.4 9 10.8 7 600 5
6M Y s*/b 9 12-0 5 12-3 4 620 3

s*/b indicates that the direction of splitting is normal to the direction of buffing.
Tensile results obtained on rubber test pieces 285

Results for Compound X20A

Coeffa~-m of 100% C o e f f ~ n t of 2O0% Coeff~/ent of 300% C o e f f ~ , m of


EB variation Modulus variation Modulus variation Modulus variation
Code (%) (%) (MPa) (%) (MPa) (%) (MPa) (%)

M 680 3 0-8 8 15.5 6 2.8 9


M 680 3 0.9 6 16.4 7 3-0 6
1A 690 3 0.7 10 15-7 7 2.8 7
2C 670 2 0-9 7 16.2 3 3.0 4
1B 690 3 0.9 7 16-2 10 2.9 10
2D 670 3 0-8 13 15.4 8 2-8 9
3E 690 2 0-7 10 15.6 5 2-8 3
7G 740 4 0-7 14 13.9 10 2.4 13
3F 700 3 0.8 11 14.4 10 2.6 8
7H 700 3 0.8 2 16-3 3 2-9 5
4J 730 4 0.9 7 15.4 9 2-6 9
4K 700 4 0.8 14 15.4 9 2.7 9
6L 740 2 0.8 9 14.8 2 2.5 3
6M 690 3 0-7 21 14.9 13 2-6 9

Results for Compound X20B

Mode Coefficient of 100% Coe/~ient of 200% Coe/~ient of 300% Coel~ient of


F_.B variation Modulus variation Modulus variation Modulus variation
Code (%) (%) (MPa) (%) (MPa) (%) (MPa) (%)

M 630 2 1-2 4 2-3 4 3.4 3


M 650 4 1.2 5 2.3 4 3.4 6
1A 630 2 1.2 4 2.3 5 3-4 2
2(2 620 2 1"2 10 2-3 9 3-4 9
1B 650 2 1.3 9 2.5 3 3-3 6
2D 620 5 1-2 9 2.4 5 3"5 5
3E 620 5 1.3 5 2-5 4 3.7 5
7G 630 5 1.2 4 2-3 4 3.2 8
3F 650 3 1-2 7 2-3 2 3-4 2
7H 620 3 1.2 11 2.3 7 3.4 6
4J 650 2 1.2 6 2-4 5 3"4 3
4K 630 5 1.1 9 2-3 6 3.3 4
6L 610 4 1.1 6 2.3 4 3.4 2
6M 630 2 1-2 8 2.4 3 3"5 3
286 D. L James, J. S. Gilder

TABLE 7
Mean Coe~iem of Mode ~ t of Mean ~ i e m of
Number TS variation "IS varialion F~ variation
Code Orientation Surfaces tested (li4Pa) (%) (MPa) (%) (%) (%)

M p mlm 9 8.2 4 8.3 3 600 4


M q rrdm 9 8-1 11 8.4 7 640 9
1A X s/s 9 8.4 3 8.5 2 600 3
2C Y s/s 9 8.0 5 8.1 4 620 6
1B X b/b 3 7-4 12 7-8 8 540 9
2D Y b/b 10 7.7 4 7.8 3 560 4
3E X m/s . . . . . . .
7G Y m/s 8 7-9 6 8.1 4 550 5
3F X m/b 10 7.2 9 7-5 6 520 9
7H Y m/b 10 7.9 6 8.1 4 560 4
4J X s/b 10 7-9 4 8'0 3 570 3
4K Y s/b 7 8.5 2 8.7 2 600 3
6L X s*/b * 11 8.1 8 8.4 7 560 6
6M Y s*/b 8 8.6 2 8.7 1 600 3

s*/b indicates that the direction of splitting is normal to the direction of b u f f i n g .

against this background that differences indicated by detailed statisti-


cal analysis must be judged.

7. T R E A T M E N T OF SKEWED D A T A

Before proceeding further, it is necessary to discuss whether when


Gaussian statistics are applied to tensile data the resulting analysis is
valid.
In any field of measurement which does not involve catastrophic
failure a high result is just as probable as a low one, and the results
are then likely to be symmetrically distributed about a mean. In this
situation mean and standard deviation are sufficient to specify the
data.
When considering ultimate tensile data, different criteria apply, for
the effect of minor flaws in the test pieces is always to produce lower
rather than higher test results. A symmetric distribution of test
figures is therefore unlikely and the results will be skewed towards
the lower figures. If the preparation is very bad, then the absence of
Tensile results obtained on rubber test pieces 287

Results for CompoundW154


Mode Coe~ient of 100% Coe~ient of 200% Coef~ient of 300% ~ient of
EB variation Modulus variation M o d u l u s variation M o d u l u s variation
Code (%) (%) (MPa) (%) (MPa) (%) (MPa) (%)

M 610 3 0"9 9 1"8 8 3.2 5


M 670 7 0-9 8 1"7 5 3.0 5
1A 600 2 0.9 8 2.1 5 3-6 6
2(3 630 4 0.9 11 1-7 10 3-3 5
1B 560 6 0.9 8 1"8 9 3.0 8
2D 570 3 0"9 5 1-9 10 3.4 4
3E . . . . . . . .
7G 560 4 1.0 3 2'1 4 3-5 3
3F 540 7 0.9 4 2.0 7 3.4 5
7H 570 3 0.9 4 2-0 6 3.5 5
4J 580 2 0.9 3 2.0 4 3-4 5
4K 610 2 0-9 2 1.9 4 3-3 4
6L 570 5 1"0 8 2.1 4 3.5 4
6M 600 2 0-9 2 2-0 4 3.4 3

high values coupled with an increase in the number of low results


may make the limited available data seem more normally distributed.
The mean values of these low figures, although truly representative of
the particular test method, will not, however, give a good representa-
tion of the material. The asymmetrical nature of the results must be
acknowledged and the data treated accordingly.
May, 7 Kase s and Heap 9 have demonstrated the double exponential
nature of tensile strength data and recommend that the mode should
be used rather than the mean. The various treatments advised by
these authors have been incorporated in BS5324:1976 l° and it is this
treatment that is used here. To find the mode of a set of test figures,
they are first arranged in decreasing order and then each value
multiplied by a weighting factor as set out in a table given in
BS 5324. This table relates to numbers of test pieces from five to
twelve. Because, in a few cases, lower numbers of test pieces have
been used in this report, the BS table has been extended by using
similar factors for three or four test pieces quoted by Heap. The
complete table of weighting factors is presented here as Table 8.
A similar table in BS 5324 lists weighting factors for calculating the
288 D . L J a m e s , J. S. Gilder

TABLE 8
Weighting Factors w for Calculating Mode*

N
n 3 4 5 6 7 8 9 10 11 12

1 (0-656) (0.511) 0.327 0.274 0"236 0.207 0.185 0'167 0-152 0.139
2 (0.256) (0.264) 0'269 0"233 0.205 0"182 0.164 0.150 0"137 0-126
3 (0.088) (0-154) 0.217 0.197 0-179 0.163 0.149 0"137 0"126 0.117
4 (0"071) 0'153 0.159 0.159 0.144 0.134 0.125 0"117 0-109
5 0.034 0"111 0-125 0.124 0-120 0"114 0.107 0.101
6 0.025 0.086 0.100 0"103 0-101 0.098 0"094
7 0.015 0.069 0-084 0-087 0.088 0.086
8 0-010 0.056 0-071 0"075 0-076
9 0.005 0.047 0.061 0-066
10 0.002 0"039 0-053
11 0.000 0.034
12 -0.001

a N is t h e t o t a l number of t e s t r e s u l t s ; n is t h e serial number of t h e t e s t r e s u l t


beginning w i t h 1 f o r t h e h i g h e s t .

standard deviation for five to twelve results from a known skew


population. Figures for three or four observations are not given, nor
are they quoted by Heap. With so few observations the method is not
entirely reliable but, for the sake of completeness, empirical esti-
mates have been included in Table 9 which also incorporates the
weighting factors given in BS 5324.
TABLE 9
Weighting Factors d for Calculating Standard Deviation °

N
n 3 4 5 6 7 8 9 10 I1 12

1 (0-35) (0-27) 0-213 0.187 0.162 0.142 0.128 0-116 0.106 0-097
2 (0.08) (0-12) 0.116 0-115 0.107 0-100 0-092 0-086 0.080 0-075
3 (-0-43) (-0.06) 0-028 0.053 0.063 0.066 0.065 0.063 0.062 0-059
4 (-0.33) -0-079 -0.013 0"019 0.034 0.040 0.044 0"045 0-045
5 -0-278 -0.096 -0.031 -0.001 0.015 0.024 0.028 0-032
6 -0.246 -0-098 -0.042 -0"013 0.003 0"019 0"018
7 -0.220 -0.098 -0-047 -0"022 -0.006 0-004
8 -0-200 -0.095 -0.051 -0.027 -0.012
9 -0.184 -0.092 -0.052 -0"031
10 -0.170 -0.089 -0.053
11 -0.158 -0.086
12 -0" 148

a N is t h e t o t a l number of t e s t r e s u l t s , n is t h e serial number of t h e t e s t r e s u l t ,


beginning with 1 for the highest.
Tensile results obtained on rubber test pieces 289

T h e standard deviation m a y b e expressed as a percentage of the


m o d e and the figure so obtained, b y analogy with Ganssian statistics,
is referred to here as a coefficient of variation.
A s can b e seen f r o m Tables 3 to 7, for any particular set of data
the BS 5 3 2 4 tables give a value for the m o d e which is slightly higher
than the mean, as expected, and a coefficient of variation which is
slightly lower than that estimated b y the traditional m e t h o d .
If the n u m b e r of test pieces used is small and occasional very low
test figures are r e c o r d e d then these can have a p r o f o u n d effect on the
calculated m e a n (less so on the mode). BS 5 3 2 4 acknowledges this
and gives criteria for the rejection of outlying low values. This is, in a
sense, an a t t e m p t to weight the m e a n so as to bring it nearer to the
m o d e and is a s o m e w h a t insensitive approach to the m e t h o d s de-
v e l o p e d b y May, Kase and H e a p .
Nevertheless it has b e e n c o m m o n to use m e a n (more recently
median) and standard deviation for tensile d a t a and in o r d e r that
comparisons with previous w o r k might b e made, b o t h m e a n and
m o d e for tensile strength and elongation at b r e a k have b e e n q u o t e d
in Tables 3 to 7 which summarise the experimental results.
Stress at a given strain does not follow the d o u b l y exponential
s k e w distribution function and m o d u l u s data should b e treated ac-
cording to the well-known Gaussian statistics. Thus only m e a n m o d -
uli are p r e s e n t e d in these tables.
W h e r e care is taken with the preparation, differences in the test
results tend to b e small as m a y b e seen f r o m Tables 3 to 7, and
traditionally in this situation the Student's 't' test is used to indicate
w h e t h e r o b s e r v e d differences are significant. D a t a o b t a i n e d by the
application of the 't' test are p r e s e n t e d in Tables 10 to 19. A n entry
'NS' in these tables indicates that there is no significant difference
b e t w e e n the sets of results being considered.
In s o m e cases, for example the elongation at b r e a k of samples type
6 L X listed in T a b l e 11, there is a very strong indication of a
significant difference. In general, however, the results for tensile
strength and elongation at b r e a k should b e a p p r o a c h e d with s o m e
caution, for traditional statistics applied to data from a s k e w e d
population m a y merely establish that each of two sets of results is an
equally b a d approximation to a notional Gaussian distribution which
the data have b e e n assumed to fit. In any case these tables do not
290 D. L James, J. S. Gilder

T A B L E 10
Student 't' test (TS and EB) Compound Z 7 5 A
Elongation at break

MX MY IAX 1BX 2CY 2DY 3EX 3FX 4.IX 4KY 6LX 6MY 7GY 7HY

NS NS NS NS NS NS NS NS NS 95 NS NS
MY NS NS NS NS NS NS NS NS 99 NS NS
lAX NS NS NS NS NS NS NS 99 NS NS
1BX NS NS NS NS NS NS 99 NS NS
2CY NS NS NS NS ~ N S NS NS NS NS NS 95 NS NS
2DY NS 95 95 NS NS ~ N S NS NS NS NS 99 NS NS
3EX NS NS NS NS NS 95 ~ N S NS NS NS 99 NS NS
3FX NS NS NS NS NS NS NS ~ N S NS NS 95 NS NS
4JX _U__S _N_S N_S NS NS NS NS NS ~ N S NS NS NS NS
4KY Ns 95 NS NS NS US NS NS US ~ N S 95 US NS
6LX NS NS NS NS NS 95 NS NS US NS ~ U S NS 95
6MY ~ _ S U_S NS 99 NS NS NS 95 US ~ N S NS
7GY ~ S NS NS NS NS NS NS NS NS ~ N S
7HY NS 95 95 NS NS NS 95 NS NS NS NS 99 NS

Tensile strength

TABLE 11
Student 't' Test ('IS and EB) Compound Z75B
Elongation at break

MX MY IAX 1BX 2CY 2DY 3EX 3FX 4.IX 4KY 6LX 6MY 7GY 7HY

~ 95 NS 95 NS 95 NS NS 95 99 >>99 95 99 99
MY 99 ~ NS NS NS NS 95 NS 99 NS >>99 NS NS 95
lAX 99 NS ~ N S NS NS NS NS 95 NS >>99 NS NS NS
1BX NS NS NS ~ N S NS 95 NS 99 NS >>99 NS NS NS
2CY NS 99 99 99 ~ N S NS NS NS NS >>99 NS 95 99
2DY 99 99 99 99 NS ~ 95 NS 99 NS >>99 NS NS NS
3EX NS 99 99 95 NS NS ~ N S NS 95 :~99 NS 95 99
3FX NS 99 99 95 NS 95 NS ~ N S NS >>99 NS 95 99
4JX NS 95 NS NS NS 99 NS NS ~ 9 9 ~'99 95 99 99
4KY NS 95 NS NS 95 99 NS NS NS -"-.~>99 NS NS NS
6LX NS NS NS NS 95 99 95 95 NS NS -",.-...~>99 >>99 >>99
6MY 99 99 99 99 NS NS NS 95 99 99 99 ~ N S NS
7GY NS NS NS NS 95 99 95 95 NS NS NS 99 ~ N S
7HY 99 NS NS NS 99 99 99 99 NS 95 NS 99 NS

Tensile strength
Tensile results obtained on rubber test pieces 291

T A B L E 12
Student 't' Test (TS and EB) Compound X 2 0 A
Elongation at break

IVIX MY lAX 1BX 2CY 2DY 3EX 3FX 4JX 4KY 6LX 6MY 7GY 7HY

MX NS NS NS NS NS NS 99 NS 99 NS 99 NS
MY NS NS NS NS NS 99 NS 99 NS 99 NS
lAX NS NS NS NS 95 NS 99 NS 95 NS
1BX NS NS NS 95 NS 99 NS NS NS
2CY 95 95 99 NS 99 NS 99 NS
2DY 99 99 NS NS NS ~ N S NS 99 NS 99 NS 99 NS
3EX NS NS NS NS NS 95 ~ N S 95 NS 99 NS 95 NS
3FX 95 NS NS NS NS NS NS ~ 9 5 NS 99 NS 95 NS
4JX NS NS NS NS NS 95 NS NS ~ 9 5 NS 95 NS NS
4KY NS NS NS NS NS 95 NS NS NS ~ 9 9 NS 95 NS
6LX NS NS NS NS NS 95 NS NS NS NS ~ 9 9 NS 95
6MY NS NS NS NS NS NS NS NS NS NS NS ~ 9 5 NS
7GY NS NS NS NS NS 95 NS NS NS NS NS NS NS
7HY NS NS 99 NS NS 99 NS NS NS NS NS NS NS

Tensile strength

T A B L E 13
Student 't' Test (TS and EB) Compound X20B

Elongation at break

MX MY lAX 1BX 2CY 2DY 3EX 3FX 41X 4KY 6LX 6MY 7GY 7HY

MX- NS NS 95 NS NS NS NS 99 NS NS NS NS NS
MY NS ~ N S NS NS NS NS NS NS NS 95 NS NS NS
lAX NS NS ~ N S NS NS NS NS 95 NS 95 NS NS NS
1BX 95 NS NS ~ 9 5 95 95 NS NS NS 99 NS NS 95
2CY NS NS NS 95 ~ N S NS NS 99 NS NS NS NS NS
2DY NS NS NS 99 NS ~ N S 95 99 NS NS NS NS NS
3EX NS NS NS NS NS NS ~ 9 5 99 NS NS NS NS NS
3FX NS NS NS US US NS US ~ S US 95 NS NS NS
4JX 95 NS US NS 99 95 NS NS ~ N S 99 95 NS 99
4KY 95 99 95 99 NS NS NS 95 99 ~ N S NS NS NS
6LX 99_99_ 99_ 99 99 NS NS 99 99 NS ~ 9 5 NS NS
6MY NS NS NS NS NS 95 NS NS NS 99 99 ~ N S NS
7GY r ~ ~ 9 ~s Ns Ns Ns 95 Ns Ns Ns ~ N S
7HY NS US NS 95 NS NS NS NS 99 NS 99 NS NS

Tensile strength
292 D. L James, J. S. Gilder

TABLE 14
Student 't' Test (TS and EB) Compound W154
Elongation at break

MX MY lAX IBX 2CY 2DY 3EX 3FX 4.IX 4KY 6LX 6MY 7GY 7HY

MX NS NS 95 NS 99 -- 99 99 NS 99 NS 99 99
MY NS ~ 9 5 95 NS 99 -- 99 99 NS 99 NS 99 99
lAX 95 NS ~ 9 5 NS 99 -- 99 95 NS 99 NS 99 99
1BX 95 NS 99 ~ 9 5 NS -- NS NS 95 NS 95 NS NS
2CY NS NS 99 NS ~ 9 9 -- 99 99 NS 99 NS 99 99
2DY 99 _. 99
3EX
3FX 99 95 99 NS 99 NS -- ~ 9 9 99 NS 99 NS NS
4JX NS NS_ 99_ NS NS NS -- 99 ~...99 NS 99 NS NS
4KY 95 NS NS 99 99 99 -- 9999 ~ 9 9 NS 99 99
6LX NS NS NS NS NS NS -- 99 NS NS ~ 9 9 NS NS
6MY 99 NS_ NS 99 99 99 -- 99 99 NS 95 ~ 9 9 99
7GY N ~ NS NS N_S -- 95 NS 99 NS 99 ~ N S
7HY NS NS 99 NS NS NS -- 95 NS 99 NS 99 NS

Tensile strength

TABLE 15
Student 't' Test (100 and 300% Modulus) Compound Z75A

300% Modulus

MX MY lAX 1BX 2CY 2DY 3EX 3FX 4dX 4KY 6LX 6MY 7GY 7HY

95~~95NS NS NS 95 NS NS 95 NS NS 99 95 NS
MY 95 99 99 NS 95 99 99 99 99 99 95
lAX NS NS ~ N S NS NS NS NS NS NS NS 99 NS NS
1BX 95 NS NS ~ N S NS NS NS NS NS NS NS NS NS
2CY NS 95 NS 95 ~ N S 95 NS NS NS NS 99 NS NS
2DY 95 NS NS NS NS ~ 9 9 NS NS NS NS NS NS NS
3EX 99 NS 99 NS 99 NS ~ N S 99 95 95 99 99 NS
3FX NS NS NS NS NS NS 99 ~ N S NS NS 95 NS NS
4JX NS NS NS NS NS NS 99 NS ~ N S NS NS NS NS
4KY NS 99 99 99 NS 99 99 99 NS ~ N S NS NS NS
6LX NS NS NS NS NS NS 99 NS NS NS ~ 9 5 NS NS
6MY NS 99 99 99 NS 99 99 99 95 NS 95 ~ N S 95
7GY NS NS NS NS NS NS 99 NS NS 95 NS 99 ~ N S
7HY NS NS NS NS NS NS 99 NS NS 99 NS 99 NS

lO0%Modul~
Tensile results obtained on rubber test pieces 293

TABLE 16
Student 't' Test (100% & 300% Modulus) Compound Z75B
300% Modulus

MX MY lAX 1BX 2CY 2DY 3EX 3FX 4.1X 4KY 6LX 6MY 7GY 7HY

MX
MY 99 --
1AX
IBX 99 NS NS
2CY 99 99 99 99 ~ - -
2DY NS 99 99 99 NS
3EX 99 99 99 99 NS NS
3FX 99 99 99 99 NS NS NS ~ - -
4JX 99 99 99 99 NS NS NS NS
4KY NS 99 95 NS 99 95 99 99 99
6LX 99 95 NS NS 99 99 99 99 99 NS ~ - -
6MY NS 99 99 99 NS NS NS NS NS 95 99 ~ - -
7GY 99 NS NS NS 99 99 99 99 99 95 NS 99
7HY 99 NS NS NS 99 99 99 99 99 99 95 99 NS

100% Modulus

TABLE 17
Student 't' Test (100 and 300% Modulus) Compound X20A

300% Modulus

MX MY lAX 1BX 2CY 2DY 3EX 3FX 437( 4KY 6LX 6MY 7GY 7HY

MX ~ N S NS NS NS NS NS 95 NS NS 99 NS 99 NS
MY NS ~ N S NS NS NS 95 99 99 95 99 99 99 NS
lAX 99 99 ~ N S NS NS NS NS NS NS 99 NS 95 NS
1BX NS NS 99 ~ N S NS NS 95 NS NS 99 NS 95 NS
2CY NS NS 99 NS ~ N S 99 99 99 95 99 99 99 NS
2DY NS NS NS NS NS ~ N S 95 NS NS 99 NS 95 NS
3EX 99 99 NS 99 99 NS ~ N S NS NS 99 NS 95 NS
3FX 95 99 NS 99 99 NS NS ~ N S NS NS NS NS 95
4JX NS NS 99 NS NS NS 99 99 ~ N S NS NS NS NS
4KY NS 95 NS 95 95 NS NS NS 95 ~ 9 5 NS 95 NS
6LX 95 99 NS 99 99 NS NS NS 99 NS ~ N S NS 99
6MY NS 99 NS 95 95 NS NS NS 95 NS NS ~ N S NS
7GY 95 99 NS 95 95 NS NS NS 99 NS NS NS ~ 9 5
7HY NS NS 95 NS NS NS 99 95 NS NS 95 NS NS

100% Modulus
294 D. L James, J. S. Gilder

TABLE 18
Student 't' Test (100 and 300% Modulus) Compound X20B
300% Modulus

MX MY lAX IBX 2CY 2DY 3EX 3FX 4JX 4KY 6LX 6MY 7GY 7HY

NS NS NS NS 99 NS NS 95 NS NS 95 NS
MY NS NS NS 95 NS NS NS NS NS 95 NS
lAX NS 95 99 NS NS 95 NS 95 95 NS
1 B X NS 99 NS NS NS NS 95 NS NS
2CY 99 NS NS NS ~ N S NS NS NS NS NS NS NS NS
2DY 99 NS NS NS NS ~ N S 95 NS 99 95 NS 99 NS
3EX >>99 99 99 NS 95 95 ~ 9 9 99 99 99 95 99 99
3FX >>99 NS 95 95 NS NS 99 ~ N S 95 NS 95 95 NS
4JX >>_9_9 NS 99 95 NS NS 99 NS ~ N S NS 95 NS NS
4KY >>99 99 99 99 95 99 99 95 95 ~ 9 5 99 NS NS
6LX )>99 >>99 >>99 ~99 ~99 ~99 >>99 >>99 >>99 >>99 ~ 9 5 95 NS
6MY Ns
7GY ~ S NS 99 NS NS 99 >>99 NS ~ N S
7HY 99 NS NS NS NS NS 99 NS NS 95 >>99 NS NS

100% Modulus

TABLE 19
Student 't' Test (100 and 300% Modulus) Compound W154
300% Modulus

MX MY lAX 1BX 2CY 2DY 3EX 3FX 4JX 4KY 6LX 6MY 7GY 7HY

MX 99 99 NS NS 99 -- 99 95 NS 99 99 99 99
MY NS ~ 9 9 95 NS 99 -- 99 99 99 99 99 99 99
lAX NS NS ~ N S 99 95 -- NS 95 99 NS NS NS NS
1 B X NS NS NS ~ N S NS - - NS NS NS 95 NS 95 NS
2CY NS NS NS NS ~ 9 9 - - 99 99 95 99 99 99 99
2DY
3EX
3FX NS NS NS 95 NS NS - - ~ N S 95 NS NS NS NS
4JX NS NS NS NS NS NS - - 99 ~ N S 95 NS 95 NS
4KY NS NS 95 NS NS 99 -- 99 95 ~ 9 9 95 99 95
6LX 99 99 99 99 NS 99 -- 99 99 99 ~ N S NS NS
6MY N$ NS NS 95 NS NS -- NS NS 99 99 ~ N S NS
7GY 99 99 99 99 NS 99 -- 99 99 99 NS 99 NS
7HY NS NS NS NS NS NS - - NS NS 99 99 NS 99

100% Modulus
Tensile results obtained on rubber test pieces 295

distinguish the effects of grain and degree of cure from the effects of
sample preparation.
Since figures for modulus are normally distributed, the 't' test is
valid for these results and there is a clear indication from Tables 15
to 19 that modulus is affected by the method of sample preparation, a
result which confirms the previous data published by Brown and
Jones. a However, since modulus can also be affected by orientation in
the test pieces as well as by differing degrees of cure, it is difficult to
be specific about the causes of the differences apparent in Tables 15
to 19. Before proceeding further it is necessary to attempt to disting-
uish the effect of grain in the rubber from the effect of surface
preparation.

8. I N T E R A C T I O N OF S A M P L E P R E P A R A T I O N
AND GRAIN

Morley and Scott, 1 recognising the significance of anisotropy on


tensile test results, eliminated this variable by always cutting their
dumb-bells parallel with the calender grain of the test sheets. Unfor-
tunately, it is not always possible to do this as the original direction of
milling may not be known. The influence of grain may then be mixed
with the effect of surface texture produced by sample preparation or
perhaps different degrees of cure, either in different sheets or at
different depths in the same sheet. These latter influences showed up
quite clearly in the comparison of test results from products and from
laboratory prepared sheets carried out by Brown. 5
In the present work, some attempt has been made to differentiate
these effects and we consider first the influence of grain.

9. E F F E C T OF G R A I N

Since dumb-bells were cut in orthogonal directions from all the test
sheets, then if in all cases the direction of milling were known, it
would be possible to classify test pieces as cut parallel to or perpen-
dicular to the direction of milling. This cannot be done, but a
classification which approaches it may be attempted. For example, in
the moulded sheets group 'p' test pieces will belong to one set and
296 D. L James, ]. S. Gilder

group 'q' test pieces will belong to the other in the above classifica-
tion, although it is not known for a particular sheet which is parallel
to or perpendicular to the grain of milling. Similarly, reference to the
cutting layout for the 6 m m sheets indicated in Fig. 2 will show that
groups l A X and 1BX will belong to one set and 2CY and 2DY to
the other. The groupings for 12 m m sheets deduced from Fig. 4 are
3EX, 3FX, 4JX and 6MY form one set and 7GY, 7HY, 4KY and
6LX form the other. (For simplicity in graphical presentation, p, C,
D, E, F, J and M are represented by squares and q, A, B, G, H, K
and L by circles.)
In order to understand the method by which the grain directions
may now be deduced it is necessary to consider the likely effects of
grain and/or surface texture on the stress/strain characteristics of a
given material.
Experience shows that test pieces cut parallel to the grain generally
exhibit a higher modulus and higher tensile strength than those cut
perpendicular to the grain. Similarly, the elongation at break of the
former group tends to be lower than that of the latter. Thus, in
general, if grain is a predominant factor in the variability of results, a
high tensile strength is accompanied by a low elongation at break and
vice versa.
The reverse is true where flaws introduced during sample prepara-
tion are the primary source of variability. A flaw cuts short the
degree of extension which would otherwise be possible in the test
piece, and thus a low tensile strength is accompanied by a low
elongation at break.
It follows then that if the ratios of tensile strength to elongation at
break are considered a set of results where grain is significant is likely
to produce two distinct values according to the orientation in which
test pieces were cut, whereas a set of results where flaws are signific-
ant will produce only one, irrespective of the orientation. Tensile
strength, elongation at break and the ratios of these are presented for
all five compounds in Table 20. The groupings of the test sheets are
those which it has already been deduced should differentiate any
grain effects.
It is at once apparent that compound Z75B (natural rubber 80-85
IRHD) exhibits a considerable grain effect, whereas the other com-
pounds do not. This may be expressed graphically by plotting tensile
strength against elongation at break (with a displaced zero) as has
T A B L E 20
Ratios of tensile strength to elongation at break

Mix number

Z75A Z75B X20A X20B W154

Group "r~ Group TS Group Group "IS Group


TS EB --×100 mean 'IS E.B "-×100 mean "IS EB - - x l O0 mean "IS EB --x100 mean "IS EB - - x l O0 mean
(Mea) (%) EB ratio (MPa) (%) EB ratio (MPa) (%) EB ratio (MPa) (%) UB ratio (MPa) (%) E.B ratio

Moulded 'p' 25-5 390 6.54 6.5 14.4 220 6.54 6-5 23.2 680 3-41 3.4 12-0 630 1-90 1-9 8-3 610 1.34 1'3
Moulded 'q' 26.4 380 6"95 7.0 16-4 210 7.81 7-8 23.1 680 3.40 3-4 12.1 650 1.86 1.9 8.4 670 1-25 1-3
lAX 24-7 390 6-33 6-3 16-5 210 7"86 7-9 22"3 690 3-23 3"3 12"3 630 1"95 2"0 8-5 600 1-34 1"4
1BX 24.6 390 6"31 15"7 200 7'85 23-0 690 3"33 12"9 650 1-98 7"8 560 1"37
2CY 24.4 390 6'26 6'2 13-8 220 6"27 6"4 22-8 670 3.40 3.4 11"6 620 1"87 1"9 8'1 630 1"28 1-3
2DY 23.2 380 6-11 13"2 200 6"60 22-2 670 3'31 12"2 620 1-98 7'8 570 1"37
3EX 25-1 380 6'61 6"3 14"1 240 5'87 6"2 23:0 690 3-33 3"3 12-2 620 1.97 1-9
3FX 24.8 390 6"36 14-1 230 6"13 22'7 700 3.24 12-3 650 1-89 7"5 540 1-39 1"4
4JX 24"9 410 6-07 15"1 240 6"29 23-2 730 3"18 12-5 650 1"92 8"0 580 1"38
6MY 25-5 420 6-07 12.8 200 6-40 22"8 690 3-30 12-3 630 1"95 8-7 600 1-45
7GY 24.3 410 5-93 6-1 15"8 200 7"90 7"9 22-7 740 3"07 3-2 11-6 630 1.84 1"8 8-1 560 1"45 1-4
7HY 24-4 380 6"11 15"6 190 8-21 23"1 700 3-30 11"8 620 1-90 8"1 570 1"42
4K'Y 23-8 390 6"10 15-1 200 7"55 22"7 700 3.24 11"3 630 1"84 8"7 610 1"43
6LX 25'6 410 6"24 16"1 200 8-05 22"9 740 3"09 10"8 610 1-77 8"4 570 1"47
298 D. L James, J. S. Gilder

been done in Figs 9 to 13. There is no doubt at all that when


c o m p o u n d Z75B is considered (Fig. 10), the figures for tensile
strength and elongation at break d e p e n d m o r e on the grain in the test
piece than on surface effects introduced by the preparative technique.
It is worth noting that the spread of results of the upper group in
Fig. 10 is smaller than it is for the group with the lower tensile
figures. Although it is probably valid to deduce that when dumb-bells
are cut with their axes parallel to the milling grain in the rubber the
technique used in preparing the test pieces has a smaller influence
than when the grain runs perpendicular to the axes of the dumb-bells,
some caution is needed, for the directions of splitting and bulling are
not entirely random. Table 2 shows that for c o m p o u n d Z75B, test
pieces l A X and 1BX both have lines parallel to the axes of the
dumb-bells while 7GY, 7 H Y and 4KY have lines perpendicular to
their axes. 6LX has both. The grouping of the test results is such that
it would appear that when the lines brought about by surface prep-
aration are parallel to the dumb-bell axes the surface preparation has
only a minor effect on the test results.

27

Oq
"~ 26

rnp OL liM
lie
[gA liJ
E
O" rlc OG

lid
23

I I I I I I
380 390 400 410 420 430

Elongation at break %

Fig. 9. Tensile strength versus elongation at break: Compound Z75A.


Tensile results obtained on rubber test pieces 299

17

q~A

O,
OK DJ
Dp
DF I--IE
Dc
~ID
13
DM
I , I I I i l
190 200 210 220 230 240 250

Elongation at break %
Fig. 10. Tensile strength versus elongation at break: Compound Z75B.

24

c
23
~!0 O ' DJ
¢,
Elc I-IMO~ 0 ~

Do
O A

I I I I I I I I I I I l I
660 680 700 720 740 760

Elongation at break %

Fig. 11. Tensile strength versus elongation at break: Compound X20A.


300 D. L James, J. S. Gilder

14

13 O B

EDD MO A q
12 Dp

11
Oi

10

I I I I I I
610 620 630 640 650 660

Elongation at break %

Fig. 12. Tensile strength versus elongation at break: Compound X20B.

g MD O K
O~
~E
O L Oq
rlp

oO O" I-lc
8 DJ
BO lid

DF

I I I | I | I I i I I

540 560 580 600 620 640 660

Elongation at break %

Fig. 13. Tensile strength versus elongation at break: Compound W154.


Tensile results obtained on rubber test pieces 301

There is a much greater spread of results when the milling grain


runs perpendicular to the axes of the dumb-bells and here it is
reasonable to suppose that tensile strength is likely to be at its lowest
when the grain and the lines due to surface preparation both run
perpendicular to the major axes of the test pieces.
Leaving aside sample 6MY, which cannot easily be classified in this
way, then of the group with the g r a i n running perpendicular to the
dumb-bell axes, samples 2CY and 2DY, additionally have s u r f a c e
t e x t u r e perpendicular to their axes and 3EX, 3FX and 4JX all have
t e x t u r e parallel with their axes. The groupings on the lower curve of
Fig. 10 fit in with the hypothesis that grain and surface texture both
perpendicular to the test piece axis give low tensile strength and
elongation at break.
It is interesting to observe that the only difference between groups
4JX and 6MY is that the buffing lines on the former run parallel to
the axis and on the latter, perpendicular, the split sides remaining
unaltered. This change was sufficient to transpose the modal tensile
values from the highest to the lowest position and emphasises the
importance of identical methods of sample preparation if compari-
sons are being made.
Figures 9 to 13 show that only compound Z75B shows a marked
effect due to milling grain. With all the other materials milling grain
produced no more noticeable an effect than change in surface tex-
ture. It was earlier .demonstrated that this observation could be
established numerically by considering the ratio of tensile strength to
elongation,at break and this would seem a useful routine procedure.

10. E F F E C T OF S U R F A C E T E X T U R E
As already discussed, the result of poor surface preparation is likely
to be a lowering of both tensile strength and elongation at break.
These differences can be emphasised by considering the product of
tensile strength and elongation at break (tensile product) as has been
done in Tables 21 and 22.
Test pieces with surfaces prepared by the same technique but cut in
orthogonal directions are grouped together in Table 21. Thus, for
example, test pieces with buffing ridges running parallel to the
dumb-bell axes (1BX) may be compared with those having buffing
ridges perpendicular to the dumb-bell axes (2DY).
to

T A B L E 21
Tensile Products Calculated Using Modes

Mix number

Z75A Z75B X20A X20B W154

TS x E B TS x E B TS × E B TS x E B TS x E B
100-----0 Ratio 1000 Ratio 1000 Ratio 1000 Ratio 1000 Ratio Prep
c••
Moulded 'p'
Moulded 'q'
9.95
10.03
0.99
or (1.01)
3.17
3-44
0-92
or (1-08)
15.78
15.71
1.00 7.56
7.87
0-96
or (1.04)
5.06
5-63
0.90
or (1.11)
m/m
m/m
lAX 9"63 1.01 3.47 1.14 15.39 1.01 7"75 1-08 5"10 1.00 s/s//e
2CY 9"52 3.04 15 '28 7-19 5-10 s/s ±"
1BX 9.59 1"09 3"14 1.19 15.87 1-07 8"39 1"11 4.37 0.98 b/b//e
2DY 8"82 2-64 14.87 7.56 4-45 b/b Z '
3EX 9.61 0"96 3-38 1-07 15-87 0-94 7.56 1"03 -- -- m/S//e
7GY 9-96 3.16 16.80 7.31 4-54 m/S'
3FX 9-67 1"10 3"24 1"09 15"89 0-98 8"00 1"09 4"05 0"88 m/b//e
7HY 8.82 2-96 16.17 7.32 4-62 m / b .1."
4JX 10-21 1.10 3-62 1-20 16.94 1-07 8.12 1-14 4.64 0-94 s/b//e
4KY 9.28 3.02 15.89 7-12 4-96 s/b _1_'
6LX 10-50 0.98 3.22 1.23 16.95 1-08 6.59 0.85 4.79 0.92 srb//e
6MY 10.71 2-56 15.73 7-75 5-22 s/fib_t.'
T A B L E 22
Tensile Products Calculated Using Modes

Mix number

Z75A Z75B X20A X20B W154

~perty _ _ TS × E B TS × E B TS × E B TS × E B
"IS × E B Ratio Ratio Ratio Ratio Ratio Prep
~r
Code ~ 1000 1000 1000 1000 1000

lAX 9.63 1.00 3-47 1-11 15-39 0.97 7.75 0.92 5.10 1.17 s/sH e ¢a.
1BX 9.59 3.14 15-87 8.39 4.37 b/b//e

2CY 9-52 1.08 3.04 1-15 15.28 1.03 7.19 0.95 5.10 1.15 s/s±'
2DY 8-82 2.64 14.87 7.56 4-45 b/b± '

3EX 9.61 0.99 3-38 1.04 15.87 1.00 7-56 0-95 -- -- m/sH e
3FX 9-67 3-24 15-89 8.00 4.05 m/b~~e

7GY 9.96 1-13 3.16 1.08 16-80 1.04 7.31 1.00 4-54 0-98 m/s±'
7HY 8-82 2.96 16-17 7-32 4.62 m/b.L '

6LX 10.50 0.98 3.22 1.23 16.95 1.08 6.59 0-85 4.79 0.92 s'/b//e
6MY 10.71 2-56 15.73 7-75 5.22 slle/b"

L~
k~
304 D. L James, 3". S. Gilder

This presentation minimises the effect of grain. Test pieces yielding


a high tensile strength but a low elongation at break (i.e. cut parallel
to the milling grain) give a tensile product (TS x EB) directly compar-
able with those having a lower tensile strength coupled with a larger
elongation at break. Just how effective this is may be judged by
looking at the results for c o m p o u n d Z75B, shown previously to have
a prominent milling grain.
T h e most important conclusion to be drawn from Table 21 is that
in most cases the test pieces with lines running parallel to the
dumb-bell axes give a slightly higher tensile product than those with
lines running perpendicular to the dumb-bell axes. The differences
are not great as may be judged by looking at the results for m o u l d e d
test pieces where the ratio should be 1.00.
It is interesting that Table 1 which gives the direction of the mould
marks could be used to explain why the m o u l d e d 'p' samples in
general give lower tensile products than m o u l d e d 'q', were it not for
c o m p o u n d W154, which with glossy surfaces should yield a ratio of
1.00.
The superiority of splitting is most apparent when c o m p o u n d
W154 is considered (a soft neoprene of hardness 45 IRHD). Test
pieces l A X , 2CY, 1BX and 2 D Y were all prepared from the same
6 m m sheet. The mean tensile product of the split sheets ( l A X and
2CY) is 5.1 whereas that for the buffed sheets (1BX and 2DY) is 4.4.
This gives a ratio of 1.16 indicating that splitting has here given a
tensile product some 16% greater than that obtained with buffed
samples and more closely in accord with that obtained from m o u l d e d
test pieces. The corresponding ratios of moulded to split and
m o u l d e d to buffed are 1-05 and 1-21, respectively.
Table 21 shows that this trend is not universally upheld and with
c o m p o u n d X20B, for example (a soft butyl of hardness 53 IRHD)
the buffed results more closely accord with the figures obtained from
moulded test pieces.
In order that test pieces prepared by splitting may be directly
compared with those prepared by buffing, Table 22 has been ar-
ranged somewhat differently. Here split test pieces with lines running
parallel to the dumb-bell axes are compared with buffed samples with
lines running similarly and so on. Again the differences are small, but
in general the split test pieces tend to give higher tensile products
than the buffed. C o m p o u n d X20B (a soft butyl) is an exception,
Tensile results obtained on rubber test pieces 305

bearing out the comment made earlier that some soft compounds
were difficult to split.
It may also be observed from Tables 21 and 22 that the practice of
bringing moulded test pieces to the thickness specified for tensile
testing by splitting or buffing one side only is not an unreasonable
procedure. The results tend to fall within the range obtained from
other methods of preparation. Again the figures show that buffing
would be the preferred method for the soft butyl rubber X20B.

11. CONCLUSIONS

Buffing has been the accepted method of preparing tensile test pieces
for many decades and the results obtained here confirm those of
previous workers, namely that when carded out carefully the results
compare reasonably with those obtained from moulded sheets. It
appears that grain and degree of cure are far more dominant than the
surface texture induced by buffing. For some softer rubber com-
pounds buffing would still be the preferred method.
Where a leather splitting machine is available this offers a useful
alternative to buffing. It is a method that is particularly suited to
compounds of moderate hardness (say 55 to 85 IRHD), although
with care it can be used, outside this range. Test pieces prepared by
splitting are in no way inferior to those prepared by buffing and
indeed there is some evidence that they may give a slightly higher
tensile product.
Whichever method of preparation is used it is best to arrange that
the surface lines induced by the preparative technique run parallel to
the dumb-ben axes.
If moulded test sheets are outside the specified limits of thickness
then buffing or splitting one side would appear to be as effective as
treating both surfaces.
However, if different materials are being compared, or different
batches of the same material, then the same preparative technique
should be used. As has been pointed out previously by Brown, 5
degree of cure is crucial and so is the direction of the milling grain.
Where the direction of the grain is not known, a numerical method
has been proposed which should indicate whether this feature is
likely to have a dominating influence on the test results.
306 D. L James, J. S. Gilder

Provided that due care is taken, either splitting or buffing may be


regarded as a satisfactory m e t h o d of specimen preparation. Some soft
c o m p o u n d s do not split easily and some hard c o m p o u n d s are not
easily buffed. Where both methods are available preliminary trials
should indicate which technique is preferable in a particular case.

A P P E N D I X : LIST O F F O R M U L A T I O N S

C o m p o u n d Z 7 5 A (Hardness 69-71 I R H D )
SBR 1500 100
H A F (N330) black 50
High aromatic oil 10
Zinc oxide 4
Stearic acid 2
Poly(trimethyldihydroquinoline) 1
CBS 0.9
DPG 0-8
Sulphur 2.7
C o m p o u n d Z75B (Hardness 80-85 I R H D )
SMR 5 100
H A F (N330) black 50
Naphthenic oil 10
Zinc oxide 5
Stearic acid 2
Mercaptobenzimidazole 1
CBS 0.5
DPG 0.1
Sulphur 2.5
C o m p o u n d X 2 0 A (Hardness 44 IRHD)
SMR 5 100
S R F (N774) black 10
Zinc oxide 20
Stearic acid 1
Octylated diphenylamine 1
MBTS 1
Sulphur 2.75
Tensile results obtained on rubber test pieces 307

C o m p o u n d X20B (Hardness 53 I R H D )
I I R (1.6 mol. % unsaturation) 100
SRF (N774) black 50
Paratiinic oil 10
Zinc oxide 5
Stearic acid 1
Paraffin wax 1
MBTS 0.5
TMTD 1
M C sulphur 2
C o m p o u n d W 1 5 4 (Hardness 45 I R H D )
C R (Neoprene WK) 100
High activity M g O 4
M T black 30
Surface-treated whiting 20
Dark brown tactice 30
Octylated diphenylamine 3
Stearic acid 0-5
Low M W polyethylene 2
Butyl carbitol formal 30
Zinc oxide 10
75 % active E T U dispersion 1
CBS 0-75

REFERENCES

1. Morley, J. F. and Scott, J. R. (1946). Journal of Rubber Research,


15(10), 199-207.
2. Reece, W. H. (1935). Trans. IRI, 11(3), T312-35.
3. BS903: Part A2 (1971). Methods of Testing Vulcanised Rubber; Deter-
ruination of Tensile Stress-Strain Properties.
4. Brown, R. P. and Jones, W. L. (1972). RAPRA Bulletin, February,
36--9.
5. Brown, R. P. (1980). RAPRA Members Report No. 42, February.
6. ASTM D412-80 (1980). Method A: Rubber Properties in Tension;
Dumb-bell and Straight Specimens.
7. May, W. (1964). Trans. IRI, 40(2), T109-22.
8. Kase, S., J. Polymer Science, 11 (1953), 425-31; 14 (1954), 497-501;
14 (1954), 579-82.
9. Heap, R. D. (1965). Trans. IRI, 41(3), T127-35.
10. BS5324 (1976). Guide to Application of Statistics to Rubber Testing.

You might also like