Fatigue and Environmental Resistance of Polyester and Nylon Fibers J. F. MANDELL, M. G. STECKEL,* S.-S. CHUNG, and
M. C. KENNEY**
Department of Materials Science and Engineering Massachusetts Institute of Technology Cambridge,Massachusetts 02139 The fatigue resistance of individual synthetic fibers can govern the performance of complex fiber assemblies such as tire cord and marine rope under certain loading conditions. This paper explores the relative performance of polyester and nylon 6,6 fibers and yarns, both dry and in aqueous solutions, primarily synthetic seawater. Fiber failure over a range of loading conditions and frequencies was found to occur at a critical cumulative strain, governed by a creep rupture process; the cyclic lifetime for both fibers is predictable using a simple creep rupture based theory. Polyester is more resistant to creep rupture, and consequently outperforms nylon 6.6 in cyclic fatigue. The advantage of polyester is considerably greater in aqueous solutions, where the performance of the nylon is diminished. Other comparisons indicate that the particular polyester fibers studied have higher stiffness and strength, lower strain to failure, and much lower hysteresis energy absorption compared with the nylon. The actual fatigue performance of complex fiber assemblies such as ropes is also limited under many conditions by factors not present in single fiber or yarn fatigue, including hysteric heating and internal and external abrasion.
INTRODUCTION ighly oriented polymer fibers are used in many structural applications, such as tire cords, coated fabrics, and ropes, where resistance to cyclic fatigue loading is important. Marine ropes used in ship operations including towing and mooring are often large in diameter with millions of individual fibers twisted and combined into many levels of structure. They must withstand creep and fatigue loading, internal and external abrasion, and various environmental agents including seawater. Rope failures can be sudden and catastrophic. Of the fibers used in ropes, interest in marine applications currently centers on the relative merits of polyester [poly(ethylene terephthalate), PET] and nylon. Nylon has been used most widely in the past, and its resistance to fatigue and the marine environment has been studied in detail (1, 2). However, polyester has shown improved performance in a variety of rope test programs (3-6).This paper presents the results
H
Present address: CHEMFAB, Merrirnack, NH. ** Present address: Albany International Research Co., Dedham. MA.
POLYMER ENGINEERING AND SCIENCE, AUGUST, 7987, Vol. 27, No. 75
of a study of the resistance of polyester fibers to fatigue and seawater, as compared with earlier results for nylon 6,6.It must be emphasized that the results to date are for only one type each of nylon 6,6 and polyester out of the many fibers and finishes now available. The results are also limited to simple fiber and yarn behavior, avoiding many of the complexities present in full-scale ropes (see Ref. 6 ) . EXPERIMENTAL The experimental methods used in this study have been described in detail in Refs. 1 and 2. The nylon and polyester single fiber and yarn characteristics and properties are described in Table 1. Fatigue tests used a sinusoidal waveform in load control at a stress ratio, R (minimum force/maximum force), of 0.1. Testing was done on a servohydraulic machine with a special vibration isolation frame described in Ref. 1. The frequency was varied in each data set to maintain a constant average load rate; the master frequency given on each figure is the value in a single-cycle ultimate strength 1121
J . F. Mandell, M. G. Steckel, S.-S. C h u n g , a n d M. C. K e n n e y Table 1. Static Properties of Polyester and Nylon 6,6 Fibers and Yarns' Nylon 6.6
Polyester Single Fiber Type Denier Fiber Dia. (pm) Spec. Gravity (g/cm3) Dry Strength (GPa) (g/denier) Wet Strength (GPa) (g/denier) Dry Ult. Strain (YO) Wet Ult. Strain (YO) Dry Initial Young's Mod. (GPa) (g/denier) Wet Initial Young's Mod. (GPa) (g/denier)
Yarn
(DuPont D608) 5.2 23 1.39 1.14 9.37 1.19 9.78 16.3 14.9 15.2 126 15.4 133
1000
-
1.09 8.96 1.10 9.01 15.9 14.2 14.7 121 14.8 122
Single Fiber
Yarn
(DuPont 707) 6 30 1.14 1.01 10.3 ,923 9.43 19.9 18.5 5.86 59.8 1.93 19.7
I260 .97 (.92)** 9.96 (9.2)" .94 9.6 19.3 19.6 5.87 60.2 2.20 22.6
* A l l yarn tensile properties obtained at a load or stroke rate to give failure in approximately 0.5 s; single fibers were tested at a slower rate, giving failure times of 3 s. Dry indicates ambient laboratory air; wet is immersed in seawater with a 5-15 min preconditioning.All yarns were lightly interlaced (not twisted). Data are an average of 5 or more tests in most cases. **Values in ()obtained in load control with an initial minimum load (l), which gave lower strength; no differencewas found in wet tests.
test, while the frequency at lower loads is varied inversely in proportion to the ratio of the maximum cylic load to the one-cycle ultimate load. Creep rupture tests were run on the servohydraulic equipment and also on deadweight test stands. Cyclic data including hysteresis, creep, and strain range per cycle were recorded on a digital oscilloscope and computer system. Seawater tests were run immersed in synthetic seawater of composition specified by ASTM D 1141-75; specimens were preconditioned for 10 min before testing. All test specimens were prepared as described in Refs. 1 and 2 with adhesively bonded tabs, a silicone rubber transitional region, and a gage length of 12.5 cm. A t least 80% of the failures were in the gagesection; any tab failures were discarded.
LL1
0
LL
0
k.
SUMMARY OF NYLON 6, 6 BEHAVIOR
References 1 , 2, and 7 give detailed results and discussion for the effects of various loading, geometrical, and environmental parameters on the behavior of DuPont 707 nylon single fibers, lightly interlaced yarns, and small ropes. The findings are summarized as follows: 1. Single fibers, interlaced yarns, and small double-braided ropes all show similar fatigue sensitivity when the data are normalized by the initial strength; thus, there are no measurable interfiber effects on the lifetime in simple tensile fatigue up to the scale of small (4.8-mm diameter) rope. This finding appears also to apply to large nylon ropes at high load levels (6). 2. Failure occurs when a critical cumulative strain is reached. A s illustrated in Fig. 1 , the cumulative failure strain is unaffected by load history, including creep, cyclic fatigue at various load levels and frequencies, simple stress-strain loading, and residual strength tests after cycling (the effects of test interruptions with long recovery periods are presently being investigated). 1122
STRAIN
ilb~
cuMu Iv E FAILURE STRAIN, ALL CASES
Fig. 1. Schematic of high stress tensile behavior of single fibers and yarns showing the effects of load history.
3 . If we assume that the cyclic fatigue lifetime is dominated by a creep rupture mechanism, the cumulative time to failure in fatigue can be predicted by a simple model such as that reported by Coleman (8)using creep rupture of individual fibers as the only material property. Figure 2 (from Ref. 1 ) shows excellent agreement between this theory (including statistical variation from the creep rupture tests) and fatigue data at master frequencies varying from 0.1 to 20 Hz. The failure condition is defined in terms of the time under load, not the number of cycles. If plotted as a function of cycles, the data sets at different frequencies would be separated by more than two decades. 4. The creep rupture curve for yarns and small ropes is complicated by frictional reloading POLYMER ENGINEERING AND SCIENCE, AUGUST, 1987, Vol. 27, No. 15
Fatigue and Environmental Resistance of Polyester and Nylon Fibers
7 %
I
w GI H
m
z
w E w
3 40t H
B GI 3
30
t
01
0
Experimental Fatigue Data
1 Hz
0
0.1 Hz
I
R = 0.1
1 I
I
I
I
I
I
2
3
4
1 1 1
Hz
6.2
LOG TOTAL TEST TIME
I
I
5
6
(set)
Fig. 2. Cyclicfatigue d a t af o r nylon 6,6 single fibers compared with creep rupture based theoretical prediction From Ref. 1).
of broken fibers under dry conditions, so the prediction in ReJ. 3 works best using single fiber baseline creep rupture data. 5. Distilled water immersion reduces the strength of single nylon 6,6 fibers by about 9% compared with 65% relative humidity air; this effect varies at higher levels of structure, but a reduction in strength of 1020% for wet vs. dry conditions is usually reported for nylon 6,6. The initial (lowstrain) modulus is reduced by more than a factor of two, but the slope of the remainder of the stress-strain curve is less affected (Fig.3).Most rope usage is in the low-strain range. 6 . Single fibers and yarns give similar results. Salt water and more severe stress-cracking agents such as LiCl and LiBr, which have a strong effect on bulk nylon 6,6, show no worse effect on nylon fibers than does distilled water. The effect of stress-cracking agents is greatly reduced by orientation, so that no effect is found in highly drawn rope fibers in the axial direction.
POLYESTER RESULTS AND DISCUSSION The polyester yarn results given here are taken from Ref. 9, where more details are available. As shown in Table 1 and Fig. 3, the DuPont D 608 polyester yarn used in this study has a higher initial modulus and strength, but POLYMER ENGINEERING AND SCIENCE, AUGUST, 1987, Vol. 27, NO. 15
I000
-
800
L
600
200
n STPAIN
I81
Fig. 3. Tensile stress-strain curvesfor nylon 6,6 and PET single fibers under wet (sea water immersion) and dry (ambient air) conditions.
a lower strain to failure, as compared with DuPont 707 nylon 6,6. The polyester shows little effect of water on the stress-strain curve (Fig. 3 ) or the strength. Polyester does have a significantly higher density than nylon (Table I ) , so that the strength in terms of linear density (g/denier or g/tex) is lower in the dry condition. Figure 4 gives maximum stress vs. log cycles to failure ( S - N ) data for the polyester yarns in 1123
J . F. M a n d e l l , M . G. S t e c k e l , S.-S. C h u n g , and M. C. K e n n e y
ambient air and in seawater at a master frequency of 1 Hz. Under both conditions, the polyester performance is superior to the nylon 6,6 results presented in Refs. 1 and 2 [each data set is normalized by the initial strength at the respective condition). These results are in agreement with literature data for larger strands and ropes (3-6). When expressed in terms of cumulative time to fail, the polyester follows a similar criterion to that discussed for nylon. Figure 5 indicates that, at the same maximum stress, the creep rupture lifetime [ R = 1.0) is very similar to the cumulative test time to failure under cyclic loading. (Failure in creep is expected at a slightly shorter time at the same maximum load because of the reduced load during much of each fatigue cycle; this effect is then accounted for in the creep rupture based model.) Figure 6 gives the maximum cumulative strain during cycling at two load levels. At failure, the maximum cyclic strain approaches the creep rupture strain as well as the failure strain in a simple stress-strain test (about 16%).Thus, the polyester data conform to the schematic in Fig. 1 , which was based on nylon 6,6 results. Similarly, in Fig. 7, the residual strength at up to 80% of the mean S-N lifetime is unchanged from the initial strength. Figure 1 also includes a trend in the cyclic
stress-strain curve shape toward a steeper (stiffer) curve with less hysteresis as cycling progresses. Figure 8 gives typical data for the strain range (maximum to minimum) on particular cycles as cycling progressed to failure. The strain range at low cycles (note the log scale) dropped rapidly, followed by a very gradual de-
I UG CYCLLS
Fig. 6. Creep extension (strain at maximum loadJduring cyclicfatiguefor polyester yarns (R = 0.1)compared with failure strain in creep rupture.
~ ~ y o ~ l s , , l . O i l z
,
I
~
85 P I T S . 1.0 H Z
A
80 XUTS, 0.1 llZ
0.4
0.7
(I. b
TIME UNDER FATI6UL LOAD / AVG I l M L 10 Nvi ON
rAll
.
I I1
11.8
I T I I I
A l bIVLN I O A D
Fig. 7. Residual tensile strength of polyester yarns at various fractions of the average failure time for three cyclic load conditions (R= 0.1).
Fig. 4 . S - N cyclic fatigue data for polyester yarns as compared with trend lines for nylon 6.6 (R = 0.1. master frequency = 1 H z ) .
. 2 GU w
:!
;r" 7
7
50-
:: 3 .
? z
IIU-
z
30-
I
7
20-
1u-
0 CREtP
CYCLIC FATIGUL
I
I
1
I
1
2
3
I,
I
I b
I1 LO11 CYCLLS
(
4 I 1
Fig. 8. Cyclic extension (maximum strain - minimum strain on each cycle)of a polyester yarn specimen tested to failure at 75% UTS maximum load, R = 0.1, and 1.0 Hz. 1124
POLYMER ENGINEERING AND SCIENCE, AUGUST, 1987, Vol. 27, No. 15
Fatigue and Environmental Resistance of Polyester and Nylon Fibers
crease over most of the lifetime. Just before failure the strain range increased because of the failure of some indyvidual fibers in the yarn (under load limits). Figure 9 shows a similar pattern in the hysteresis energy for the same specimen. The change in strain range and hysteresis between the first 0.1% of the lifetime and failure is very small. The cyclic fatigue lifetime of polyester can be predicted using the creep rupture based model, as indicated in Fig. 10. The only significant difference between the model prediction and the experimental data at 0.1 and 1 .O Hz master frequencies is a slight overestimate of the 1-s intercept; the slope of the prediction is in good agreement with the data. Any inaccuracies in the model prediction were found with nylon (1, 2) to derive primarily from the creep rupture behavior of the yarns, which can be complicated because broken fibers only locally unload as a result of friction in the lightly interlaced yarn structure. Wet creep rupture conditions and cycling appeared to reduce this effect with nylon, and predictions based on individual fiber creep rupture data were very accurate, as shown in Fig. 2. The polyester results (Fig. 10) are in acceptable agreement even when the prediction is based on dry yarn creep rupture results.
A li A A A A
5
2
4
(1)
Nylon 6,6 Wet: P / P o = .98 - .lo08 log t
(2)
Polyester Dry: P / P o = .97 - .0436 log t
(3)
Polyester Wet: P / P o = 1.02 - .0570 log t
(4)
fi
1
IOG cvctEs (
Nylon 6,6 Dry: P / P o = .98 - .Of353 log t
where P is the maximum load, P o is the initial strength in a 1.O-s test, and t is the cumulative time to failure in seconds. The number of cycles to failure a t any frequency can be found by multiplying t by the frequency in Hz. There are a variety of qualifications for E q s 1-4, the most obvious being that they apply only to the yarns tested: DuPont 707, nylon 6,6,
A
i
COMPARISON OF NYLON 6, 6 AND POLYESTER A t the level of individual fibers and yarns, nylon 6,6 and polyester behave in a qualitatively similar manner. The trends depicted schematically in Fig. 1 apply to both materials, and the cyclic fatigue lifetime can be predicted in each-case by the creep rupture based model. Essentially, both fibers must be elongated to a critical strain to produce failure. It is the quantitative differences between the two fibers that must be considered in. potential fatigue-sensitive applications (along with other effects such as stiffness and abrasion). Figure 4 indicates that polyester is superior to nylon under load control in dry conditions, with a n even greater advantage in seawater. The applicability of the creep rupture model allows this comparison to be made in more general terms of time to failure under sinusoidal loading (without test interruptions), as shown in Fig. 11 and expressed in the following relationships from curve fits:
N I
Fig. 9. Hysteresis energy loss p e r cycle during cycling to f a i l u r e of a polyester y a r n specimen a t 75% UTS maximum load, R = 0.1, a n d 1 .0 H z .
Y
111ll
U
I
I
I
I
2
J
1
I
LOG iitii i o r n i i
I I,
II
I \it
Fig. 10. Cyclicf a t i g u e d a t af o r polyester yarns compared w i t h creep rupture b a s e d theoretical prediction. POLYMER ENGINEERING AND SCIENCE, AUGUST, 1987, Vol. 27, No. 15
LOG (TOTAL TIME TO FAILURE, SEC.) Fig. 1 1 . Cumulative time tofaiture of nylon 6,6and polyester yarns under cyclic sine-wavef a t i g u e loading (independent of frequency, R = 0.1). 1125
J . F. Mandell, M. G. Steckel, S.-S. Chung, and M. C.Kenney
and DuPont D 608 polyester. The greatest variation in other similar nylon and polyester fibers may be in the surface finish, which has little effect on these results (7),but may have a great effect on properties such as abrasion resistance, which is important in rope (as distinct from yarn) fatigue (6).Other fibers and finishes currently are being studied in this program. The dry condition is at typical ambient humidities, with similar data found for nylon 6,6 at a controlled 65%relative humidity (2).The wet condition appears to apply to a wide range of aqueous solutions for nylon (21, but does not include solutions such as strong acids (lo), which may attack the fibers. E q u a t i o n s 1-4 have been tested for frequencies of 0.1 to 20 Hz for the nylon and 0.1 to 1 .O Hz for the polyester; although a stress ratio of 0.1 was used in most experimental confirmations, the equations appear to be approximately correct for the entire tension-tension R range of 0.0 to 1.0, as tested for nylon ( 1 ) and suggested for polyester in Fig. 5. While these equations have been shown to correlate with large rope fatigue test results under certain loading conditions, particularly high relative loads (6),many potential complications that may occur in rope applications have been ignored, including fiber shrinkage, abrasion, transverse compression, long unloaded recovery periods, photochemical degradation, hysteretic heating (ropes have reduced heat transfer and increased hysteresis frum structural effects), and temperatures other than the 20-25°C test range. Studies of the failure modes of individual fibers indicate a tendency for some axial splitting at low stresses with the nylon 6,6 under both creep and cyclic loading, but no effect on the lifetime was evident; polyester failures were all transverse to the fiber axis (1).Other studies of nylon fibers at higher frequency (50 Hz) have reported a transition in mode from transverse to axial splitting under low load and low R value conditions (11 , 12). Quantitative differences between nylon and polyester are also evident in the cyclic stressstrain behavior. As shown in Fig. 3, the D 608 polyester has a higher modulus, particularly under wet conditions, and slightly lower strain to failure. The greater creep resistance of the polyester is reflected in a considerably lower hysteresis energy at a comparable relative load range, as shown in Table 2. While the hysteresis energy for both the nylon and the polyester decrease gradually as the frequency increases, the values for polyester are about 10 times lower than for the nylon. (The considerable scatter is due to limited precision in the data acquisition.) This significant difference between fibers does not carry over proportionally to typical ropes, where structural hysteresis can be the dominant factor (6). The results given here are for cycling between fixed load limits, with the yarn allowed to creep 1126
Table 2. Hysteresis Energy of Nylon and Polyester Yarns at Various Frequencies [from Ref. 71.’
Freq (Hz)
Period Is)
Polyester Hysteresis (10-3 J)
14.0 14.0 13.0 13.0
1
8.9
1.20 0.76 1.4 0.99 0.75
5.0
0.2
6.2
0.16
3.0 3.2
0.58
0.00028 0.0021 0.033 0.13 1.o
3600
Nylon Hysteresis J)
480 30
8
0.46
*Cycling from 1% to 20% of ultimate tensile load (as measured in 1-Hz ramp test) under dry conditions. Measurements made on one specimen of each material after preconditioning for several hundred cycles, with further preconditioning at each frequency before measurements were made.
during the test. Both the absolute and relative findings for nylon and polyester could change if tests were run under fixed extension limits, which may better represent some marine rope applications. Fixed extension limits would generate relatively higher stresses in the stiffer polyester fibers, so that the greater fatigue resistance and lower hysteresis of the polyester under load control might not be observed under strain control. This tendency is evident in the results given in Re$ 13. CONCLUSIONS Results of cyclic fatigue tests on DuPont 707, nylon 6,6, and DuPont D608 polyester yarns show qualitatively similar creep-dominated behavior for both. The polyester has superior fatigue resistance, particularly under wet conditions, because of its greater creep rupture resistance. The nylon shows about 10 times greater hysteresis energy absorption at comparable (low) cyclic load levels, lower stiffness, and slightly greater strain to failure. While the polyester is clearly more fatigue resistant in simple laboratory yarn tests, the proper choice of a rope fiber and surface finish also requires consideration of the many additional parameters and possible failure modes inherent in typical marine rope deployments. ACKNOWLEDGMENTS This research is part of a broad study of the deterioration of synthetic marine rope supported by the Naval Sea Systems Command through the MIT Sea Grant Program, Mr. George Prentice is the Navy’s technical liaison person on the project. Acknowledgment is also made of the considerable practical experience and advice that has been given to the MIT Synthetic Rope Program by the Navy’s Man-made Fiber Rope Technical Advisory Group. REFERENCES 1 . M. C. Kenney, J. F. Mandell. and F. J. McGarry, J . M u t e r . Sci., 20, 2045 ( 1 985). 2. M. C. Kenney, J. F. Mandell, and F. J. McGarry, J. M u t e r . Sci., 20, 2060 (1985). 3. H. Crawford and L. M. McTernan, in “Proc. Offshore
Technology Conference,” Paper 4635, pp. 455-466, Houston [ 1983). POLYMER ENGINEERING AND SCIENCE, AUGUST, 1987, Vol. 27, No. 15
Fatigue and Environmental Resistance of Polyester and Nylon Fibers 4. M. R. Parsey, A. Street, and S. Banfield, in “Proc. Offshore Technology Conference,” Paper 5009, pp. 429-
444, Houston (1985). 5. J. F. Flory, “OCIMF Hawser Standards Development Program, Trial Prototype Rope Tests,” Final Report, Oil Companies International Marine Forum (October 1983). 6. J. F. Mandell, “Modelingof Marine Rope Fatigue Behavior,” to be published. 7. M. C. Kenney, Ph.D. Thesis, Dept. Mat. Sci. and Engr., MIT (1983).
POLYMER ENGlNEERlNG AND SCIENCE, AUGUST, 1987, Vol. 27, No. 15
8. B. D. Coleman, J.Polyrn. Sci., 20,447 (1956). 9. M . G. Steckel, M . S . Thesis, Dept. Mat. Sci. and Engr.,
MIT (1984). 10. J. W. S. Hearle and B. S. Wong, J . Text. Inst., 4, 127 r 19771. 11. A.-R. Bunsell and J. W. S. Hearle, J . Appl. Polyrn. Sc i . , 18.267 (1974). 12. C. H. Oudet and A. R. Bunsell, J . Mater. Sci. Lett., 3, 295 (1984). 13. D. C. Prevorsek and Y . D. Kwon, J . Macrornol. Sci.Phys. B12,447 (1976).
1127