JOURNAL OF MATERIALS SCIENCE 20 (1985) 2045-2059

Fatigue behaviour of synthetic fibres,

yarns, and ropes
M.C. K E N N E Y * , J. F. M A N D E L L , F.J. M c G A R R Y
Department of Materials Science and Engineering, Massachusetts Institute of
Technology, Cambridge, Massachusetts 02139, USA

S - N fatigue and creep-rupture data have been obtained for nylon 6,6 single fibres,
interlaced yarns, and small ropes under a variety of loading conditions. The results show
a similar degradation rate at each level of structure, with no apparent influence of inter-
fibre effects. Cyclic lifetimes of single fibres of nylon 6,6 as well as polyester and aramid
can be predicted from a creep rupture model. Consistent with this model, the time to
failure is insensitive to frequency over a broad range. For each level of structure the strain
at failure is the same whether tested in simple tension or under cyclic or creep loading.
Failure modes were generally similar in creep rupture and cyclic fatigue tests; no effect of
a slack load on each cycle was evident either in the failure mode or specimen lifetime,

1. Introduction for an equivalent period of time, with failures

Oriented fibre structures are used in many applica- occurring at times and elongations similar to creep
tions where a combination of good axial mechani- loading [4, 5]. Finally, failure in bulk polymers
cal properties and light weight is required. The use may occur as a result of fatigue crack initiation
of nylon fibres and yarns in marine ropes is one and stable propagation, similar to that in metals.
application which has grown considerably over the For nylon in fibrous form, some evidence exists
past 15 years. As with many applications, marine for all three modes [6-8].
ropes in service are subjected to a complex history
of static and cyclic mechanical loading [1], which 2. Experimental procedure
may, in some cases, end in sudden and unexpected Fatigue testing was conducted using closed loop
failure. This work is concerned with determining selvohydraulic machines. Most tests were run in
the mechanisms by which cyclic fatigue loading the load control mode, using a sine wave function.
causes failure in oriented fibre structures, with Ultimate tensile load was evaluated using a ramp
emphasis on nylon single fibre and yarn properties, test (constant loading rate) to failure at a rate
and additional work on small rope structures and equivalent to the averaged loading portion of the
other materials. fatigue cycle. Fatigue testing was conducted to
The effects of fatigue loading in polymers as a various percentages of this ultimate tensile load.
general cla~s of materials is receiving increasing The specimens were cycled between the maximum
attention in the current literature (a recent review load and one-tenth of this value; i.e. the ratio
is given in [2] ). In bulk polymers, failure during arnin/Oma x (or stress ratio, R ) was equal to 0.1.
cycling can occur by a number of modes [3]. (Other R values have been used in some cases.) For
Some polymers such as nylon, with large hyster- single fibre fatigue testing, a specially designed
esis loops and poor thermal conductivity, may steel beam support frame on pneumatic shock
fail thermally as a result of internal hysteretic absorbers was constructed to minimize vibrations
heating. The material may also respond to cycling transmitted from the hydraulics and floor. A
as if a static load had been applied continuously 2000g capacity load cell suspended from this

*Present address: Albany International Research Co, Dedham, MA 02026, USA.

0022 2461/85 $03.00 + .12 9 1985 Chapman and Hall Ltd. 2045
frame registers approximately 1.5g noise. The points for single fibres and yarns was 5 in (127
number of cycles to failure was recorded for each mm); the gauge length between fittings for small
specimen. Various parameters, such as load, dis- ropes was 4 in. (102 ram).
placement and hysteresis energy were monitored Dead load creep rupture lifetimes for fibres and
using a minicomputer. yarns were also measured using tabbed specimens.
A master frequency of 1 Hz was used for most The top of the specimen was hung from a support
testing, master frequency referring to the fre- and weights were suspended from the bottom.
quency corresponding to cycling at 100% of ulti- Selected creep rupture tests were also run on the
mate tensile load. The actual test frequency at servohydraulic machines to allow better monitor-
each percentage of ultimate tensile load was then ing of data; failure times in these tests agreed well
adjusted upward to maintain a constant average with dead load data.
loading rate. Thus, the frequency was increased as The primary material was DttPont 707 nylon
the maximum load was decreased. Test frequency 6,6 yarn, a standard rope yarn used commercially.
was kept as close as possible to this level, except The yarn consists of 210 lightly interlaced fibres
where machine resonances required slightly lower producing a total of 1260 denier, taken from a
values. The master frequency is given on each single lot and merge; the single fibre diameter was
figure. 30#m. The small rope was a 3/16in diameter
A cardboard tabbing system was used for test- double braided based on DuPont 707 nylon yarn,
ing of yarns and single fibres. A flexible silicone purchased from Sampson Ocean Systems. Except
sealant was incorporated at the tab end to improve for single fibre creep rupture tests which were run
stress transfer, and protect the specimen from at 65%RH, the humidity was not controlled.
abrasion. Further into the tab the specimen was However, yarn creep rupture lifetimes obtained
bonded with a rigid adhesive (Eastman 910). Using throughout the year were indistinguishable from
this tabbing method, the majority of fibre breaks those at 65% RH.
occurred in the gauge section, away from the grips,
and were considered valid test results. For rope 3. Results and discussion
gripping, silicone and polyester adhesives were used 3.1. Fibre, yarn, and small rope fatigue
to encapsulate rope ends into tapered metal fit- Single fibre and yarn S - N data are plotted in
tings. The gauge length between rigid adhesive Fig. 1. The fatigue data are presented as normalized

I00-"
90-
C}
,<
0 80-
70- S n e. re
H
Z
60-

50.
H
40.
,50.
[] Single Fibre 1 Hz, ~= 0.i
20. O Yarn 1 Hz, R = 0.i

I0- /~ Rope 0.5 Hz, R = 0.i

0 I I I I I I
0 I 2 3 4 5 6
L O G C Y C L E S TO FAIL

Figure 1 S-N fatigue data for DuPont 707 nylon single fibres, yarns, and 3/16 in. double braided rope.

2046
T A B L E I Ultimate load and elongation in single cycle impose. This can be seen from Fig. 1, or by com-
ramp tests paring the regression equations, with 95% confid-
Ultimate* Ultimate* Ultimate* ence limits given in parentheses:
load stress (MPa) elongation (%) single fibre: P/Pvat = 1.00 (-+ 0.04)
Fibre 68.1 g 965 15.0
- 0.077 (+ 0.002) log N
(• 2.0) (• 0.75)
Yarn 11.61 kg 779 17,0 yarn: P/Pult = 1.02 (-+ 0.04)
(• (•
-- 0.09 (-+ 0.003) log N
Rope 605 kg 40.9
where P is the load and N is cycles to failure. Thus,
(-+21) (+-2.0)
the two levels of structure show essentially the
*Approximate failure time = 0.5 sec.
same fatigue resistance. This implies a common
factor in the fatigue degradation of each, which
plots based on the single cycle ultimate tensile does not appear sensitive to structural or interfibre
load for the conditions of that particular data set. effects at this level.
Table I gives values of ultimate tensile load at fail- Fatigue data for the small ropes are also given
ure for the specimens tested; these can be used to in Fig. 1. Tests were conducted at a frequency of
determine absolute load values for the S - N data. 0.5 Hz. A fan was used to cool the rope during
The lines on the graph show least square regres- cycling to reduce the hysteretic temperature rise,
sions of experimental data (not including single which was otherwise substantial. Failure in the
cycle ramp tests). Both sets of data are approxi- grips became a problem at lower loads and longer
mately linear, and show a steady decrease in life- times. Grip failure points are indicated by arrows
time with increasing maximum toad. The linearity in Fig. 1 and represent the lowest possible gauge
of these data and absence of a break in the curve section lifetimes; gauge section failure could have
imply that a single mechanism may operate occurred at longer times. The rope data points fall
throughout the fatigue range studied. These S - N close to those of both the single fibre and yarn.
curves are consistent with literature data [ 9 - 1 1 ] . Even if a slight adjustment were made for differ-
For example, at about 60% of ultimate load, life- ences in frequency, the points would fall at similar
times are approximately l0 s cycles. times, indicating that interfibre contact and other
It is important to note that the normalized structural differences have a minimal effect in the
graphs of the single fibre and yarn nearly super- small rope as well as in the yarn.

I00'

90'

80"
O
70'

H 60" o
03
Z
50" gauge length

40"
H
.50'
o\o O Parallel
20' (2ingauge length)
[] Interlaced
tO'

(3 I I I I I I I
0 f 2 3 4 5 B 7
LOG CYCLES TO F A I L 1 Hz, /?= 0.I
Figure 2 S - N fatigue data for parallel and interlaced sections of nylon yarn.

2047
 To further investigate possible interfibre or the absence of interfibre effects already demon-
structural effects several studies were conducted, strated.
including interlaced and parallel yarn sections, In a final study of interfibre effects, specimens
added lubricants, and specimens composed of were prepared with three fibres twisted together
three contacting fibres. The DuPont 707 yarn is (two turns per inch). Fatigue data for these are
lightly interlaced by a jet of air blown orthogon- compared to normalized single fibre lifetimes in
ally at the yarn during production, resulting Fig. 3. Lifetimes for specimens with three fibres
in a lightly entangled point every 2 in on average, in contact are clearly comparable to those for
so all standard 5 in yarn specimens probably con- single fibres at this low twist level.
tained at least one interlace. To investigate this All of the data reported in this section indicate
effect, short sections of yarn (2 in) were prepared that the single fibre fatigue behaviour determines
containing either an interlaced point or a com- the yarn and, apparently, small rope fatigue behav-
pletely parallel length. A standard fatigue S - N iour without any significant complications. While
curve was determined for each (Fig. 2). Fatigue the average single fibre tensile strength is not com-
data are identical for the parallel and interlaced pletely realized in the average yarn strength, due
specimens, as well as for specimens of standard to long recognized statistical bundle effects [12],
length. Apparently, the interlace has no significant the rate of loss of the initial tensile strength in
effect on fatigue performance. fatigue (slope of the normalized S - N curve) is
The effect of a lubricant on dry yarn perform- similar for all three structures. Interlace regions
ance was examined briefly using a silicone fibre and added lubricants do not affect yarn fatigue.
finish, Union Carbide LE-9300, applied as an However, in highly twisted yarns and in larger
emulsion and catalysed for full cure. A limited rope structures with more lateral pressure and
number of yarn fatigue tests at one load were relative movement, interfibre interactions and lub-
conducted, and showed that the lubricant had no ricant performance could become important.
significant effect on the cycles to failure. A few
tests were also run after the application of a 3.2. F r e q u e n c y e f f e c t s
standard silicone lubricating spray, and again no As interfibre effects at the yarn level have no sig-
effect was found. This is not surprising in view of nificant effect on fatigue performance under the

I00 "

90-
O
0%~176 []

8O ODD 9
C~
,< OO
0 70
:-1 l-i[] co []

H tO [] o',-
r~
Z
E~ 5O
E-,
40"
H
250'

20" 1 Fibre
O
I0' [] 3 Fibers in c o n t a c t

0 I I I I I I
0 I 2 .3 4 5 co
LOG CYCLES TO FAIL 1 HZl ~ = 0.i

Figure3S-N ~t~uedataforsinglenylonfibresandforthreefibresmcontact.

2048
 I00
o o o O
90 zx AaOm ol o

8O 9 O Aeoo O
O o
70 9 06 0O O9 O~
< A
0
60- O9 A9

H
Z
50-

40-

H
30-

O 0.i Hz
20-
O 1.0 Hz
A 6.2 Hz
I0
9 Creep Rupture Lifetime

0 I I I I I I
0 I 2 3 4 5 6
LOG TIME TO F A I L (sec)

Figure 4 Fatigue data for single nylon fibres at several frequencies, plotted against total test time for comparison with
creep rupture data.

conditions studied, the controlling mechanism These data indicate that single fibres fail accord-
must, therefore, be the individual response of the ing to a total time under load criterion.
fibres to the applied loading. The failure mode Corresponding fatigue data for yarns tested at
under the applied loading may be dominated frequencies of 0.1, 1.0 and 10Hz are given in
either by creep rupture (depending on time under Fig. 5. (Again, adjustment could not be made
load), traditional fatigue crack initiation and exactly for the 10Hz tests because of machine
growth (depending on the number of cycles [3]), resonance and performance limits.)Temperature
or hysteretic heating and the associated tempera- change from hysteretic heating was monitored
ture rise. The dominance of a particular effect may for the highest test frequency using a temperature
be determined by varying the fatigue loading par- sensitive lacquer (Omega Lac 100). No colour-
ameters and frequency. change was observed in the lacquer, indicating
Fig. 4 gives fatigue data for single fibres tested that the temperature did not exceed 100 ~ F. This
at frequencies of 0.1, 1.0, and 6.2Hz. The 0.1 and is consistent with the literature [7] and shows that
1.0 Hz tests were conducted in the usual manner complications of a thermal failure mode are not
with the frequency adjusted upward to maintain present. An intensive series of tests was also run
constant loading rate. The 6.2Hz tests, however, on yarns at these frequencies at two specific load
could not be adjusted upward because of machine levels (80% and 60% of the 1 Hz ramp strength).
resonances, and were run at a constant frequency; Specimens were prepared identically and tested
this should have little effect over this broad range in random order. Time and cycles to fail for these
of frequencies. Note that the data are plotted as individual specimens are given in Table II. Both
a function of log time to fail, rather than cycles sets of data show that yarns follow a time under
to fail. The total test lifetime is equivalent regard- load criterion, as do the single fibres. Within the
less of test frequency, and is also similar to con- range of experimental scatter, the time to fail is
stant force, creep rupture data also shown in Fig. 4. constant regardless of the frequency of testing

2049
 I00
m [3
90

80"
<
o 70" /Y%~- O O [] [] 9 9 Creep Rupture

M 60 e@
z [] ~v~z~l o
5O []
N
40
M
E~ 50- [] 0.i Hz
i
oko O 1.0 Hz

,oj ! /k
9
I0 Hz
Creep Rupture

O I I I I I I I
O I 2 3 4 5 6 7
LOG TIME TO FAIL (sec)

Figure 5 Fatigue data for nylon yarns at several frequencies compared with creep rupture data.

TAB LE l I Fatigue lifetime of yarn at various frequencies over the three frequencies used in this study
(single specimen data) (0.I, 1.0, 10Hz). The common time under load
Frequency Log ]Log Total failure dependence of both fibres and yarns is
(ttz) (cycles (time to elongation consistent with the equivalent S - N behaviour
to fail) fail, sec) (%) observed at 1 Hz. This type of frequency depend-
80% ultimate ence contrasts with that found in the literature for
load* bulk nylon where crack growth rate per cycle,
0.167 1.690 2.467 16.7 rather than total test time, remains similar as fre-
1.716 2.493 16.3 quency is changed [3]. The observed contrast in
1.67 2.846 2.624 > 16.3 behaviour may result either from the dimensional
2.723 2.501 17.8 characteristics or oriented structure of the fibres,
3.223 2.223 16.3
or from material differences. Studies on acrylic
10
3,625 2.625 18.7 and nylon fibres [6, 13] have shown decreasing
time to fail with increasing frequency, but the
60% ultimate
load*
extremely high frequencies used (50 to 90Hz)
could have affected these results.
0,167 3,897 4.674 17.t
3.312 4.089 18.4
4.152 4.930 15.4 3.3. Failure strain
3,685 4.462 17.5 Table Ill gives failure strains for single fibres~
1,67 4.983 4.760 17.4 yarns, and small ropes for various types of load-
4.527 4.305 17,5 ing. Under all loading conditions studied (fatigue,
4,522 4,300 17.6 creep, and single cycle ramp tests) the total cumu-
10 5.868 4.868 17.6 lative strain reached at failure is roughly constant
5.810 4,810 - for a given level of structure. The overall strain
5.649 4.649 17.5 reached is, of course, influenced by the structure;
5.549 4.549 17.0 it increases from fibre to rope as the structure
*As measured in a 1 Hz test. becomes more geometrically complex. Similar

2050
T A B L E I I 1 Comparison of single cycle ramp, fatigue,
and creep values of failure strain
Failure strain (%)
Ramp Fatigue* Creep*
Fibre 15.0 14.6 14.4
(_+0.75)
Yarn 17.0 18.9 18.9
(• 1.4)
Rope 40.9 41.9 41.4
(• 2.0)
*Single specimen at 70% ultimate load, typical data.

Figure 7 Fracture surface of nylon fibre failed in fatigue

failure strains are also reached regardless of test at 90% of ultimate load.
frequency, as is shown for yarns in Table II. Fig. 6
shows that the strain accumulating with total test dominance of simple creep rupture in controlling
time is remarkably insensitive to the frequency or failure. The creep behaviour of the single fibre,
number of cycles. Just prior to failure the strain as moderated by increasing structural complexity,
increases rapidly as some fibres fail. Tests with a may determine the limiting extension in each
combination of loading conditions, as in tests level of structure studied. Within the single fibre,
where cycling is followed by ramp testing, also on a molecular level, this could be interpreted
show a constant cumulative failure strain equiva- from various nylon models as either the accumu-
lent to the initial ramp value. lation of a critical number of taut tie molecules
The observation that failure occurs at a particu- in the amorphous region [14], or as reaching a
lar strain for any level of structure, for all loading limiting extension of the oriented interfibrillar
conditions, provides further strong support for the regions [15].

20-

19-

18-

17- X

16-
oko
15-
Z
0
14- +~
E~

O 13-

12-
+
II.

I0,

9.
9 10 Hz x 0,1 Hz + 1 Hz
I I I I I I
0 i 2 3 4 5 6
LOG TEST TIME (sec)
Figure 6 Elongation during fatigue for nylon yarns at three frequencies, tested at 60% of ultimate load.

2051
 splitting generally parallel to the fibre axis, in a
mode described as fibrillated [6].
Transverse failures were observed for all fatigue
breaks at high loads and for most fatigue breaks at
low loads. In approximately 5% of breaks at low
loads in yarns a fibrillated fracture was observed.
This behaviour is not unique to fatigue breaks, as
both types of fractures were also observed for
specimens from creep rupture tests at similar
loads. This finding is contrary to reports in the
literature [6] that in similar fibres the axial crack-
ing mode is indicative of a unique fatigue failure
mode and requires a slack load condition, which
has been suggested to reduce fatigue lifetimes.
Figure 8 Fracture surface of fibre from nylon yarn failed
in fatigue at 50% of ultimate load. The origin of this difference could be the higher
frequency (50 Hz) used in the tests in Bunsell and
Hearle [6], which could not be duplicated here.
3.4. Fractographic s t u d y Several fatigue tests were conducted on yarns
Two types of fracture surfaces were observed in using a slack loading condition (R = 0) at 1 Hz.
this study which are generally similar to those The data shown in Fig. 9 fall within the previously
reported in the literature [6]. Fig. 7 shows the determined fatigue lifetime scatter. Apparently, a
most typical surface, the classic scallop shell trans- period of slack load has no effect on yarn fatigue
verse fracture observed in many textile fibres. In behaviour under these conditions, compared with
this fracture, a crack initiates at the surface, then tests at the usual R = 0.1.
grows generally transverse to the fibre axis through Slack load fatigue tests were also run on single
a slow growth region. Finally, there is a fairly fibres, but at a frequency of 20Hz. Because of
sharp transition to the fast fracture region. Fig. 8 dynamic (recording) effects, the measured load
shows a second type of fracture which also appears peak values were adjusted downward by 10% to
to initiate at the surface, but then is dominated by give a better estimate of the actual loading. The

I00

90 00000

80. 0 0
0
70' 0 0 00

60. 0 0 0
Z o
50'
E-.
40.
H
E~
.50'
o~
20. o 1 HZ, R = 0.1

I0. 9 1 Hz, Slack Loading

0 I I i I I I
0 [ 2 3 4 5 ro
LOG CYCLES TO FAIL

Figure 9 Effect of slack condition during each cycle in fatigue of nylon yarns.

2052
 I00"
0 00
90 0 0 0

80' 9 OCmO
c~
< O0
O 70 0(3 Io

60- O~o
Z
M 50-
N
40-
H
.50-
i Hz R = 0.i
o~o 20- o
20 Hz, Slack Loading
IO- 9 peak resolution adjusted downward 10%

0 i I I I i
0 1 2 3 4 5
LOG TIME TO F A I L (sec)

Figure lO Effect of slack condition at high frequency on single nylon fibres.

adjusted data are plotted in Fig. 10, along with the breakdown [3], but rather represent a limiting
fatigue data at 1 Hz. Two features are apparent: rearrangement and extension of the fibre structure
the high frequency data follow quite closely the which is a consequence of creep.
fatigue behaviour at 1 Hz, and the presence of a
slack load does not reduce the fatigue lifetime 3.5. Residual strength
below the expected range. Fibre fracture surfaces Residual strength tests were conducted at various
from these tests also show a mixture of transverse fractions of specimen lifetimes, typically 25, 50,
and axial types. Our data, including observation of and 75N of the mean lifetime, at a variety of load
the axial mode in creep as well as cyclic tests, levels. Residual strength was determined within
suggest that the axial splits do not require cyclic or approximately 1 min of the end of cycling. The
slack loads, and that the mode of cracking does single fibre data in Fig. 11 show that the residual
not influence the lifetime over the frequency range strength is almost equal to the original ultimate
studied here. The occurrence of axial cracking strength under all conditions. Thus, failure must
clearly reflects the longitudinal structure of nylon occur in a sudden death fashion, with little decrease
with some preferred path between fibril or micro- in strength before the fibre fails. The residual
fibril units. Kevlar fibres, with a more nearly per- strain to failure shows a more gradual decrease as
fect orientation, were observed to fail in the axial creep occurs. For a single fibre tested at 80N of
mode under all loading conditions in this study. the UTS for 75N of its lifetime, the residual
Failures of the type described here suggest a fatigue strain to failure is approximately 75% of the initial
crack growth controlled failure. In bulk polymers value.
this would usually suggest a cyclic, rather than Higher levels of structure show somewhat
creep, dominated behaviour which could be con- more gradual property changes due to structural
sidered in a fracture mechanics context. However, rearrangements and single fibre failures. For exam-
previous results for the strains at failure, and the ple, a yarn tested at 70% of its UTS contained
similar times to fail under differing conditions are almost exactly 70% of the original fibres unbroken
more consistent with a creep rupture mode. While just prior, to failure. Thus, strength degradation in
cracks are clearly associated with eventual failure yarns and small ropes may be explained more by
of the fibres, they may not represent the typical accumulating broken fibres than by changes in
fatigue cracks which dominate bulk polymer unbroken fibres. Permanent strains accumulating
2053
 Single Cgcle

[] ,. X l t t,
B []m
Z
0 []
60- []
E-4
03

<

H
50- % Ultimate Cgcle Ratio (,,/, e,xpecetd^)~M
ul Load 0.25 0.50 0.75
c~
90 O ~ Q q

40-
80 [] [] []
70 A ' A 9

C I I I J
0 I 2 3 4
L O G C Y C L E S (N)

Figure 11 Residual strength of single nylon fibres after fatiguing various fractions of their lifetimes at three load levels.

during fatigue are also greater with yarns and dividing the sinusoidal fatigue wave into many
ropes. A small rope tested at 60% of its UTS for small increments and weighting each increment
75% of its lifetime has a permanent creep strain by its failure time under pure creep loading; when
of approximately 60% of its original failure strain. the fractional lifetimes sum to 1, failure occurs.
(This criterion is analogous to Miner's rule used
3.6. Creep rupture model for fatigue life in cycle dominated cases.) Failure then may be
Evidence for a simple creep rupture mechanism expressed as:
controlling failure in single fibres, yarns and small /~b d(t)
ropes appears quite strong. Normalized S - N
rbo(t) (1)
curves are similar for all three structures. Interfibre
effects are minimal in these yarns and small ropes, where ~'b is the time to fail under constant stress
and do not appear to influence failure. Time to fail (creep rupture), o is the applied stress, and t b is
at the same load remains constant over a range of the time to fail under sinusoidal stress.
frequencies for both single fibres and yarns. Also, Creep rupture is expressed in the form:
a constant strain is reached at failure, for a given
structure, regardless of loading pattern or fre- rb = A exp -Ba (2)
quency. All of these results point strongly to a where A and B are experimentally measured con-
simple creep rupture mechanism. The constant stants. The creep rupture data which have been
failure strain [16] and time to failure regardless collected for single fibres, yarns, and small ropes
of frequency [17] have been reported in earlier all follow this type of relationship. For the case of
fibre studies under more restrictive loading con- sinusoidally varying load, the appropriate expres-
ditions. sion for a is:
A creep rupture failure mechanism may be ana- a ( t ) = P + Q sin (cot). (3)
lysed using the concepts of reaction rate theory
and accumulation of damage [5]. This approach When these expressions are substituted i n t o
has been applied in the literature to both fibres Equation 1 the integral may be solved using a
and bulk materials [4, 5]. According to one zero order modified Bessel function giving the
typical approach [4], failure is approximated by time to break as:
2054
 I00 -

90" 97k Predlctlon from creep rupture

' '9 5 % " " its)

8o
<
0
70

~4 60
Z
50

40
H Experimental Fatigue Data
E~
,50-
9 1 Hz [] 0.1 Hz ~ 6.2 Hz R = 0.i
20
O 20 Hz R = 0 peak resolution adjusted
downward 10%
I0

0 I I I I I
0 I 2 3 4 5 6
LOG TOTAL TEST TIME (see)
Figure 12 Comparison of single nylon fibre cyclic fatigue data with prediction from creep rupture based model.

i (p)
t,o - (4) fibre prediction and yarn creep is used in the yarn
Io(BQ) prediction. Since it has been shown that the single
where [(p) is the time to fail under a constant fibres and yarns have the same normalized S-N
load equal to the mean load p, B is the slope behaviour under cyclic loading at different fre-
from Equation 2, and Q is the amplitude from quencies, it is clear that the difference in model
Equation 3. To summarize, creep rupture times accuracy between yarns and single fibres must
are integrated over a varying load to calculate derive from differences in their creep rupture
the fatigue failure time. behaviour. Fig. 16 compares the creep rupture
Calculated failure times based on yarn and curves for single fibres and yarns. (The ultimate
fibre creep rupture data are compared to the load used to normalize the creep data is determined
actual experimental fatigue data points in Figs. in a ramp test with a total failure time of approxi-
12 and 13. (Confidence limits on the graphs were mately 0.5 sec). The yarn creep rupture curve is
determined by finding the 95% limits on the clearly less steep. A comparison of Figs. 12 and
creep rupture pre-exponential terms, and then 13 shows that the yarn cyclic fatigue data are
using these bounds for the creep rupture equation predictable from the creep rupture model if the
in the integration procedure.) Single fibre behav- single fibre creep rupture curve is used in the
iour is very well predicted by the creep rupture prediction.
model. This behaviour is common to several other The less steep creep rupture slope of the yarn
highly oriented fibres which were also examined may be explained by the observation that, in a
briefly. Aramid (Kevlar 49) and polyester (DuPont yarn structure under constant dead load, indivi-
D608 PET) single fibre data, shown in Figs. 14 dual broken fibres may be caught within the struc-
and 15, both conform to the model. ture and reloaded along part of their length. In a
In contrast, yarn (Fig. 12) does not conform yarn undergoing cycling, on the other hand, the
as well; the data consistently fall below the model structure continually opens and closes, allowing
prediction. Note that different sets of creep rup- broken fibres to work free and become completely
ture data are used in the single fibre and yarn unloaded; this is, in fact, visually observed.
predictions; single fibre creep is used in the single The question of relative fibre movement within
2055
 I00'

90 ~ Prediction from Creep rupture

9 ~ ~ ~ ~ t s )
80
O 70

H 60

50

40

3o
*~ 20
Experimental fatigue data 1 Hz, R = 0.1
I0-

o I I I I I I
0 I 2 5 4 5 6
LOG TOTAL TEST TIME (sec)
Figure 13 Comparison of nylon yarn cyclic fatigue data with prediction.

the yarn during cycling was examined in a series approach the limit of constant loading or creep
of yarn fatigues at varying R ratios (0.1, 0.5, and rupture at R = 1.
0.8). Fig. 17 indicates that higher R ratios, result- A series of experiments was conducted to
ing in reduced opening and closing of the structure, directly observe the behaviour of yarns containing

I00"

90-
~ c t i o n from Creep Rupture
80
,<
O 70

H 60
u]
Z
50
95% Confidence Limits

40
H
50"
oko
20" O Experimental Fatigue Data
I0-

0 I I I ; I I
0 1 2 3 4 5 6
LOG TOTAL TEST TIME (sec) 1 Hz, •= 0.I
Figure 14 Comparison o f single aramid fibre cyclic fatigue data with prediction.

2056
 100'

90'

80'
r tion from Creep Rupture
0
70"
;,,,-,t
Z
60" ~-'--_ ~o- r
50" )--_%
4O 95% C o n f i d e n c e Limits

30
~o
20" O Experimental Fatigue Data

10"
9 ! ....|..
0 , , I '- , ; ......... '
o I 2 3 4 5 6
LOG TOTAL TEST TIME (see) 1. 5z, R = 0.i

Figure 15 Comparison of single polyester fibre cyclic fatigue data with prediction.

some cut fibres to simulate broken fibres. In the location of the fibre ends, the amount of strain
these experiments, several fibres within the struc- on the cut fibres, compared to the yarn as a
ture were cut, their ends and starting positions whole, may be calculated. The results given in
marked, and displacement of the ends examined Table IV show an increasing amount of average
after 1000 cycles at 70% of ultimate load. Fiom strain over the cut fibre length, and corresponding

100"

90" O

80' O OO
a~
Cl

70" O0 Q~O Aa&

P-1
~,0'
r.n
,Z
P..1
50' ~D O
F..T.J.
40' O
H
E~
50" O Single Fibre

20" 9 Yarn

10" A Rope
] I_ I _. | ....
0 I . . . . 1 -- -- r

0 I 2 3 4 5 G 7
LOG TIME TO FAIL (see)

Figure 16 Comparison of creep rupture data for nylon single fibres, yarns and 3/16 in. double braided rope.

2057
 100"

90'

80'
s
70'

CoO-
z

50"
E~

40"

50" 0 R=O.I
[] ~ = 0.5 Experimental Fatigue Data
20- 9 R =0.8
I0 9 Creep Rupture

0 I I I I I t
0 t 2 .5 4 5 co

LOG TIME TO FAIL (sec)

Figure 1 7 Effect of R ratio on fatigue of nylon yarns.

greater average loading in the cut fibres at higher depending on the total time under load, not the
R ratios, approaching a maximum during creep number of cycles. A cumulative time under load,
rupture. Thus, the fibres are only partially creep rupture model for fibres can accurately
unloaded, and the less steep creep rupture curve predict the failure in cyclic tests at different
of yarn may derive from this effect. Cycling at frequencies. Other oriented fibres (polyester and
the usual R =0.1 reduces this process, so aramid) also agree with this creep rupture model.
the fibre and yarn data agree in cyclic fatigue. Consistent with a creep rupture dominated mode,
Similar reloadings may occur in small ropes as the strain at failure for a given structure is inde-
well, although the amount will necessarily be pendent of the type of loading, including ramp
affected by the complex geometric structure. (tensile test), creep rupture, or cyclic fatigue. The
model is less accurate for yarns, as the baseline
4. Conclusions creep rupture curves become less steep in going
The fatigue resistance of nylon single fibres, yarns from fibre to yarn. This may arise from broken
and small ropes are all nearly identical on normal- fibres which are partially reloaded in the higher
ized S - N scales, at all frequencies tested. Yarns levels of structure. However, the yarn cyclic
and single fibres fail by a creep rupture mode data are accurately predicted when the single
fibre creep rupture data are used in the model
T A B L E IV Analysis of fibre movement in yarn after prediction.
1000 cycles at 70% ultimate load, 1 Hz
R ratio Number of Average % strain along Acknowledgement
cut fibres cut fibre length This research is part of a broad study of the
0.1 9 0 deterioration of synthetic marine rope supported
by the Naval Sea Systems Command through the
0.5 3 1.5
MIT Sea Grant Program. Mr George Prentice is the
6 0
Navy's technical liaison person on the project.
0.8 4 5.4
6 3.1
References
1.0 5 7.9 1. N. STARSMORE, M.G. HALLIDAY and W.A.
5 4.7 EWERE, "Barge Motions and Towline Tensions

2058
 Measured During a North Sea Tow", in Proceedings 10. A.R. BUNSELL and J.W.S. HEARLE, J. App L
of the International Symposium on Ocean Engr. - Polym. Sei. 18 (1974) 267.
Ship Handling, Gothenberg, Sweden (1980) paper 13. 11. W.T. KELLEY, Textile Res. J. 35 (1965) 852.
2. R.W. HERTZBERG and J. A. MANSON, "Fatigue 12. M.M. PLATT, W. G. KLEIN and W. J. HAMBUR-
of Engineering Plastics" (Academic Press, New York, GER, ibid. 22 (1952) 641.
1980). 13. W.J. LYONS, ibid. 32 (1962) 553.
3. J.A. MANSON and R. W. HERTZBERG, CRC Crit. 14. A. PETERLIN, J. Polym. Sei. 7 A - 2 (1969) 1151.
Rev. Macromol. ScL 1 (1973) 433. 15. D.C. PREVORSEK, E J. HARGET, R. K. SHARMA
4. G.B. McKENNA and R.W. PENN, Polymer 21 and A.C. REIMSCHUESSEL, J. MaeromoL Sei.
(1980) 213. Phys. 1-2B8 (1973) 127.
5. B.D. COLEMAN, J. Polym. Sci. 20 (1956) 447. 16. J.W.S. HEARLE and E. A. VAUGHN, Rheol. Aeta
6. A. R. BUNSELL and J.W.S. HEARLE, ar. Appl. 9 (1970).
Polym. Sci. 18 (1974) 267. 17. A.R. BUNSELL and J.W.S. HEARLE, ibid. 13
7. R.L. HAMILTON and C.H. MILLER Jr, Textile (1974) 711.
Res. Z 34 (1964) 20.
8. J.W.S. HEARLE, J. Mater Sci. 2 (1967) 474.
9. I. NARISAWA, M. ISKIKAWA and H. OGAWA, Received 24 May
J. Polym. Sci. 15 (1977) 1055. and accepted 19 July 1984

2059