Construction and Building Materials 12 Ž1998.

303]310

Tension]tension axial fatigue of E-glass fiber-reinforced

polymeric composites: fatigue life diagram

Cornelia E. DemersU
Ci¨ il Engineering and Engineering Mechanics, Ci¨ il Engineering Bldg a72, Room 206B, Uni¨ ersity of Arizona, Tucson, AZ 85721, USA

Received 5 October 1996; revised 13 December 1997; accepted 20 December 1997

Abstract

Fiber-reinforced polymeric ŽFRP. composites have been under analytical and experimental investigations for approx. 40 years
in the context of aerospace, marine and mechanical applications. Now the use of FRP composites in civil engineering applications
is steadily increasing. For civil engineering structures subject to cyclic loading, one important limit state in design is fatigue. Thus,
a review of the existing data is necessary to determine the applicability to civil engineering structures. This review will focus on
E-glass FRP composites under tension]tension axial fatigue with test frequency of 5 Hz or less, without environmental concerns.
Test and material parameters, which define the data, are identified as R ratio, test frequency, load control, specimen shape, type
of reinforcement and resin. A fatigue life diagram summarizing the E-glass FRP composite tension]tension axial fatigue data is
plotted for normalized stress Žmaximum fatigue stress divided by ultimate tensile strength. vs. log of fatigue life. This plot reveals
a lower bound confidence level, regardless of test parameter combination, such as reinforcement, resin, or R ratio. This lower
bound confidence level was challenged via laboratory testing of an E-glassrvinylester composite in tension]tension axial fatigue
for R ratios 0.05, 0.1, 0.5 and 0.9, and test frequencies 1, 3 and 5 Hz. The laboratory test data support the lower bound confidence
level. This lower bound may be used in designing E-glass FRP composites for use in civil engineering structures conservatively
estimating tension]tension axial fatigue life with test frequency 5 Hz or less. This lower bound may be used until further studies
refine the effects of the individual test parameters on fatigue life. Q 1998 Elsevier Science Ltd. All rights reserved.

Keywords: E-glass fiber-reinforced polymeric composites; Composites; Fatigue; Axial fatigue

1. Introduction short-term as well as long-term behavior under various

loading and environmental conditions.
In recent years, an interest in utilizing fiber-rein- For composite structures subjected to cyclic loading,
forced polymeric ŽFRP. composites in civil engineering fatigue becomes one of the most important limit states
applications has been steadily increasing. This is pri- that needs to be considered by the designer. The sub-
marily due to the ever-increasing demand for materials, ject has been under analytical and experimental investi-
which are characterized by high strength-to-weight and gations for almost 40 years in the context of aerospace,
stiffness-to-weight ratios at an effective installed or life marine and mechanical applications. Over this period
cycle cost. However, the use of such advanced ma- of time, fatigue design data have been generated for a
terials in the construction industry may be hampered wide range of composite material systems under axial
due to the lack of adequate data representing their and flexural fatigue loading as well as environmental
conditions. The fatigue loading conditions model
tension]tension, tension]compression and compres-
U
Corresponding author. Tel.: q1 520 6216550; fax: q1 520 sion]compression state of stresses with frequencies of
6212550; e-mail: cdemers@engr.arizona.edu loading as high as 30 Hz, i.e. application to a helicopter

0950-0618r98r$19.00 Q 1998 Elsevier Science Ltd. All rights reserved.

PII S0950-0618Ž98.00007-5
304 C.E. Demers r Construction and Building Materials 12 (1998) 303]310

rotor blade w1x. Varying combinations of fibers, resin A tension]tension axial fatigue test program was un-
systems and reinforcing schemes comprise the com- dertaken to evaluate the effects of R ratio and test
posite material systems. As a result of the diversity of frequency on fatigue life of an E-glassrvinylester com-
the fatigue design data, this study focuses on posite of combination unidirectional fibers interspersed
tension]tension axial fatigue data with frequency of with five mat layers of reinforcement.
fatigue load 5 Hz or less for E-glass FRP composites
without environmental concerns. This summary of fa- 2.3. Experimental procedure
tigue design data, which is applicable to civil engineer-
ing structures, is augmented by laboratory testing of an The material is comprised of vinylester resin with
E-glassrvinylester composite in tension]tension axial E-glass reinforcement. The reinforcement is comprised
fatigue with frequency of fatigue load 5 Hz or less. The of continuous unidirectional fibers with five mat layers
objective of this survey is to assist in recognizing placed throughout the thickness. The combined fiber
the fatigue potential of E-glass FRP composite as an weight is 38%. The E-glassrvinylester composite was
independent structural material for infrastructure supplied in plate form. Coupons were dry cut with final
application. dimensions obtained using a water-lubricated diamond
saw. The coupon dimensions were 2.54 cm wide= 1.27
cm thick = 25.4 cm long Ž1 = 0.5= 10 inch. with the
2. E-glass FRP composite tension–tension axial fatigue clear distance between grips 15.24 cm Ž6 inch.. The
coupon shape remained bar form. Tabs were not
2.1. Pre¨ ious research utilized in this study.
All tests were performed on a servo-hydraulic MTS
The process of reviewing the fatigue data revealed machine. The actuator speed was the same for both the
significant test parameters, which defined the individual ramp-up to mean load and to ultimate load. A 222.4 kN
tests. These test parameters are fiber reinforcement, Ž50 kip. capacity load cell was used to monitor the
resin, frequency, load control, specimen shape, R ratio. fatigue load and a uniaxial extensometer was used to
Other parameters of interest are percent fiber volume measure the specimen strain over a 2.54 cm Ž1 inch.
or weight ratio, failure criteria, number of tests repre- gage length. A temperature voltmeter, sensitivity
senting each data point and failure strain. Table 1 "0.18C, was used to monitor the specimen surface
identifies the previous research fatigue design data temperature during testing. A computerized data ac-
utilized in this study in column 1 as tests 1]16. Test quisition system was used to record the loadrstrain
parameters, which are not reported in an original refer- data. The fatigue tests were conducted at room tem-
ence, are not reported in Table 1. All data are perature in stress control with R ratios of 0.05, 0.1, 0.5
from tests conducted at ambient room temperature and 0.9, and test frequencies of 1, 3 and 5 Hz.
Ž; 20]308C.. Varying combinations of the test par-
ameters exist such that only two tests Žtests 6 and 7. by
Agarwall and Dally w2x are shown in Table 1 for which
3. Test parameters
the singular varying test parameter is load control.
Thus, no true evaluation of the effect the individual
test parameters have on fatigue life may be ascertained The following sections discuss the test parameters
from the fatigue data due to the dependent nature of and their potential effect on fatigue life according to
the data on the test parameter combinations. The data the test data 1]16 in column Ž1. of Table 1. The
are, however, plotted according to fiber reinforcement, current laboratory testing of the E-glassrvinylester
frequency and resin, and are further discussed below. composite, test data 17, is included.
The data are graphically represented as normalized
stress Žmaximum stress divided by ultimate tensile 3.1. E-glass fiber reinforcement and resin
strength. vs. log of fatigue life which are referred to as
fatigue life diagrams. The fatigue tests Žtests 9]14., Examples of fiber reinforcement include continuous
column 1 of Table 1, by Owen and Rose w3x were fiber Žunidirectional, 0r90, "45, weave. and short fiber
periodically stopped, the specimens unloaded to ex- ŽCSM.. Examples of resins include epoxy, polyester and
amine damage, and then reloaded to resume fatigue vinylester. Various combinations of resin and fiber re-
testing. This action assumed no difference in fatigue inforcement are shown in Table 1.
life compared with uninterrupted fatigue testing. All fatigue design data of Table 1, presented accord-
ing to fiber reinforcement, are plotted in Fig. 1 as a
2.2. E-glassr ¨ inylester composite fatigue life diagram. Both weave and short fiber re-
inforcement Žisotropic properties. composites plot on
Test 17, column 1 of Table 1, identifies the current steep slopes. A conservative lower bound to the test
laboratory testing of an E-glassrvinylester composite. 1]16 data for fatigue life less than 10 000 cycles is
 C.E. Demers r Construction and Building Materials 12 (1998) 303]310 305

Table 1
Tension]tension axial fatigue of E-glass FRP composites

Reference Type of Resin Specimen Freq. R Load Percent Failure Testsr Failure
Ž1. reinforcement Ž3. shape ŽHz. Ž6. control fiber mode sample strain
Ž2. Ž4. Ž5. Ž7. Ž8. Ž9. Ž10. Ž11.

1. Mandell et al. w4x Unidirectionala Epoxy Dogbone 5 0.1 Stress 45 vol. 2k } }

2. Mandell et al. w4x Unidirectionala Epoxyf Dogbone 5 0.1 Stress 49 vol. 2k } }
3. Mandell et al. w4x Unidirectionala Polyester Dogbone 5 0.1 Stress 23 vol. 2k } }
4. Jessen and Plumtree w10x Unidirectionala Polyester Rodh 5 0.05 Stress 60 vol. 3l 4 min 1.9
5. Mandell et al. w4x 0r90 Ž5 ply.a Epoxy Dogbone 5 0.1 Stress 50 vol. 2k } }
6. Agarwall and Dally w2x 0r90 Ž15 ply.a Epoxy Dogbone 0.01]2i 0.05 Stress 70.4 wt. } } 2.1
7. Agarwall and Dally w2x 0r90 Ž15 ply.a Epoxy Dogbone 0.01]2i 0.05 Strain 70.4 wt. } } 2.1
8. Mandell et al. w11x "45r0r45 Epoxy Necked 5 } Stress } } Range }
Ž5 ply. a
9. Owen and Rose w3x Weaveb Polyester Necked 1.66 0 Stress 50.4 wt. 2k 10 1.85
10. Owen and Rose w3x Weaveb Polyester Necked 1.66 0 Stress 52.5 wt. 2k 10 2.05
Ž30.g
11. Owen and Rose w3x Weaveb Polyester Necked 1.66 0 Stress 52 wt. 2k 10 2
Ž50.g
12. Owen and Rose w3x CSMc Polyester Necked 1.66 0 Stress 24.5 wt. } 10 1.5
13. Owen and Rose w3x CSMc Polyester Necked 1.66 0 Stress 33.9 wt. } 10 2.2
Ž30.g
14. Owen and Rose w3x CSMc Polyester Necked 1.66 0 Stress 35.6 wt. } 10 2.3
Ž50.g
15. Mandell et al. w7x Short fiberd Nylon 6r6 Dogbone 1]2 0.1 1j 40 wt. } 1 3.85
16. Mandell et al. w7x Short fiberd Poly- Dogbone 1]2 0.1 1j 40 wt. } 1 1.75
carbonate
17. Demers Žthis study. Combinatione Vinylester Bar 1,3,5 0.1, Stress 38 wt. 2k 1]5 1.67
0.5,
0.9
a
Continuous.
b
Five-layer eight-end satin weave: aligned, continuous fiber representing orthotropic sheet laminate.
c
Chopped strand mat: random, discontinuous fiber representing isotropic sheet laminate.
d
Short fiber oriented with melt flow pattern, anisotropic sheet laminate.
e
Continuous unidirectional fiber with five mat layers reinforcement.
f
Rubber modified.
g
30% or 50% flexibilizer by wt.
h
Reduced gage.
i
1]10 000 cycles.
j
Constant strain rate.
k
Complete specimen separation.
l
20% decrease in material stiffness.

provided by the unidirectional continuous fiber re- rubber does not significantly affect fatigue life as com-
inforcement composites. For fatigue life between 10 000 pared to the unmodified resin for these E-glass FRP
and 1 000 000 cycles, a conservative lower bound to the composites.
data is provided by the short fiber reinforcement com- In addition to resin and fiber reinforcement, percent
posites Žanisotropic properties. and "45 continuous fiber composition varies as shown in Table 1. The effect
fiber reinforcement composites. The test data 17, com- the percent fiber composition has on fatigue life cannot
bination continuous unidirectional fibers interspersed be ascertained from the data.
with five mat layers reinforcement, plot within or near
the lower bound. 3.2. Frequency of fatigue load
The effect of resin on fatigue life was evaluated for
test data 1]16, column Ž1. of Table 1, as shown in Fig. The magnitude of test frequency affects the internal
2. A conservative lower bound to the data for fatigue heating of the composite in tension]tension axial fa-
life less than 10 000 cycles is provided by the epoxy tigue. As frequency of fatigue load increases, internal
resin composites. The nylon or polycarbonate resin heating of the composite increases and fatigue life of a
composites provide a conservative lower bound to the composite decreases. Frequencies of fatigue load under
data for fatigue life between 10 000 and 1 000 000 cy- 4 or 5 Hz have been reported to produce negligible
cles. Also, this evaluation suggested that modifying internal heating in glass FRP composites w4,5x. Thus,
polyester resin with flexibilizer or epoxy resin with only the data for which the frequency of fatigue load
306 C.E. Demers r Construction and Building Materials 12 (1998) 303]310

Fig. 1. Fatigue life diagram.

remains equal to or below 5 Hz is utilized in this study. The data for tests 1]16, column Ž1., Table 1, are
This test frequency limit comfortably encompasses civil replotted in Fig. 3 according to frequency of fatigue
engineering applications; for example, bridge members load. The data suggests greater fatigue life was ob-
are designed to accommodate a frequency of loading of tained with test frequencies 0]2 Hz than with test
approx. 1 Hz or less. frequency 5 Hz and the difference decreases as fatigue

Fig. 2. Normalized maximum stress vs. log N according to resin.

 C.E. Demers r Construction and Building Materials 12 (1998) 303]310 307

Fig. 3. Normalized maximum stress vs. log N according to frequency of fatigue load.

life increases. A conservative lower bound to the data Table 2

is provided by test frequency 5 Hz data over all fatigue Observed heating of E-glassrvinylester composite
life. Sma xrSult 0.8 0.8 0.8 0.8 0.6 0.6 0.6 0.6 0.6 0.4 0.4
The test data 17, E-glassrvinylester composite, was R 0.1 0.1 0.1 0.5 0.5 0.1 0.1 0.5 0.5 0.1 0.5
evaluated for the effect on fatigue life according to Frequency ŽHz. 1 3 5 3.5 1 3 5 3 5 5 5
frequency of fatigue load. For test frequencies of 1, 3, max dT Ž8C.a 8 11 11 3 8 24 21 5 4 12 1
or 5 Hz, no distinct trend in this data suggested a a
Difference between room temperature and specimen surface tem-
significant effect on fatigue life; however, heating of perature.
the specimen surface was observed. Table 2 shows the
combinations of normalized maximum stress, R ratio Žvarying frequency. although no significant difference
and test frequency, for which heating of the specimen in fatigue strength or fatigue life is apparent for fatigue
surface was observed. For other combinations not tests conducted with either constant frequency or vary-
shown, heating of the specimen surface was not ing frequency. Mandell et al. w4,7,8x conclude that con-
observed. For R ratios 0.1 and test frequency 3 or 5 Hz, stant or varying frequency does not significantly affect
the maximum difference between specimen surface fatigue strength or fatigue life. Agarwall and Dally w2x
temperature and room temperature was recorded as: concludes that no significant difference in fatigue
128C for normalized maximum stress of 0.4; 248C for strength or fatigue life is achieved for fatigue tests
normalized maximum stress of 0.6; and 118C for nor- conducted with either stress control or strain control
malized maximum stress of 0.8. For R ratio 0.1 and test Žboth at constant frequency.. This previous research
frequency 1 Hz, the maximum difference between suggests type of load control is not a critical test
specimen surface temperature and room temperature parameter, which affects fatigue life or fatigue strength.
was recorded as 88C for normalized maximum stress of Fig. 1 illustrates the Mandell et al. w7x test 15 and 16
0.6 or 0.8. The E-glassrvinylester composite specimens data from Table 1 with constant strain rate vs. the
for 1, 3 and 5 Hz testing frequency fall within or near remaining data with stress Žconstant frequency. control.
the lower bound. The Mandell et al. w7x data is identified in Fig. 1 as
short fiber reinforcement Žanisotropic properties. and
3.3. Type of load control in fatigue testing plots lower bound for fatigue life between 1000 and
1 000 000 cycles compared to all other data for stress
control. However, due to previous research results as
The concept of type of load control affecting fatigue discussed above, it would appear more reasonable that
strength is discussed by Agarwall and Dally w2x, Sims the lower bound nature of the Mandell et al. w7x data is
and Gladman w6x and Mandell et al. w4,7,8x. Sims and influenced by the type of reinforcement or the resin
Gladman w6x favor a constant rate of applied load rather than the type of load control.
308 C.E. Demers r Construction and Building Materials 12 (1998) 303]310

3.4. Specimen shape

The test 1]16 data, column Ž1., Table 1, were evalu-

ated for specimen shape of dogbone Žbar., necked Žbar.
and rod Žcircular . with reduced gage, on fatigue life.
The data suggests increasing fatigue life for dogbone
shaped specimens, necked shaped specimens and rod
shaped specimens. A conservative lower bound to the
data is provided by dogbone shaped specimens over all
fatigue life. The bar shape specimens of E-
glassrvinylester composite fall within or near the lower
bound data.

3.5. Tests per sample

The number of repetitions conducted per test set-up

vary as shown in Table 1. Reporting the average fatigue
life of the tests is not sufficient. All values should be
reported including the lowest fatigue life achieved per
test set-up. This is necessary to determine a conserva-
tive lower bound to all data.

3.6. R ratio

The R ratio is defined as minimum applied load Fig. 4. Typical cross-sections of test specimens at failure.
divided by maximum applied load. For all data in Table
1, the effect R ratio has on fatigue life is best given by cate tests, the grip failures achieved: greater fatigue
the test data 17, E-glassrvinylester composite. This lives than that for a uniaxial stress state failure; lower
data was evaluated for the effect of R ratio on fatigue fatigue lives than that for a uniaxial stress state failure;
life. Generally, as R ratio increased, regardless of test or fatigue lives within the span of the other data. All
frequency, fatigue life increased. the E-glassrvinylester composite data are utilized and
plot within or near the lower bound level of the fatigue
3.7. Failure criteria design data.
Echtermeyer et al. w9x observed the decay of tensile
The specimen condition or criteria defining fatigue fatigue modulus with fatigue life on Rs y1 axial fa-
failure varies among the fatigue tests utilized in the tigue tests of glassrneopentyl glycolriso polyester for
fatigue life diagrams. The criteria defining fatigue fail- constant load rate at frequency 2]5 Hz. The fatigue
ure is reported as complete specimen separation or maximum stress was 25% ultimate tensile strength to
20% decay of initial tensile fatigue modulus. Often the 100% ultimate tensile strength. This work demon-
failure criteria of test specimens is not clearly reported. strated that defining the failure criteria as 20% decay
The specimen condition defining fatigue failure for the of initial tensile fatigue modulus effectively defined the
E-glassrvinylester composite is complete specimen fatigue life of these specimens. Jessen and Plumtree
separation Žor significant separation such that load is w10x observed the decay of tensile fatigue modulus with
no longer maintained.. Fig. 4 shows three typical E- fatigue life on Rs 0.05 axial fatigue tests of unidirec-
glassrvinylester composite specimen cross sections at tional E-glassrpolyester for constant load at frequency
failure. The mat layers are easily seen. Of fifty speci- 5 Hz. The fatigue maximum stress was 50% ultimate
mens tested, six are considered grip failures, with the tensile strength to 88% ultimate tensile strength. This
majority of the test specimens failing near the grip work demonstrated that defining the failure criteria as
region. The grip region represents a constrained biaxial 10% decay of initial tensile fatigue modulus effectively
stress state while between the grips a uniaxial stress defined the fatigue life of these specimens.
state exists. Thus, failure initiates: in the grip region The fatigue life diagrams reflect the definition of
under biaxial stress state; at an inherent flaw in the specimen failure due to fatigue life being defined upon
microstructure under uniaxial stress state; or in combi- that failure. For the data plotted, a horizontal arrow on
nation via biaxial stress state in an outer mat layer and a data point indicates the specimen condition which
uniaxial stress state in the inner mat layers. In dupli- defined fatigue failure not yet achieved at the fatigue
 C.E. Demers r Construction and Building Materials 12 (1998) 303]310 309

life for which the test was stopped. Mandell et al. w4x test parameter combination. This lower bound level is
noted significant specimen damage Žsufficient length represented by the linear relation
and through-thickness cracking. when examining sev-
eral stopped fatigue test specimens. With further fa- SmaxrSult s y0.078 log N q 0.790 Ž1.
tigue testing, these specimens would have achieved
failure. and applies for fatigue life less than or equal to one
million cycles. The axial fatigue data may also be
represented by a lower bound 99% confidence level by
the linear relation
4. Discussion
SmaxrSult s y0.078 log N q 0.737 Ž2.
All the tension]tension axial fatigue E-glass FRP
composite data of Table 1 form an extended band as and applies for fatigue life less than or equal to one
shown in the fatigue life diagram of Fig. 1. The test million cycles. A ‘fatigue limit’, defined as the normal-
data for the E-glassrvinylester composite verifies that ized stress below which failure does not occur, is not
for frequency 1, 3, or 5 Hz, as R ratio increases, fatigue evident.
life increases. And no distinct trend in loss of fatigue
life was noticed using increased test frequencies 3 or 5
Hz compared to 1 Hz. However, as Table 2 shows, 5. Conclusion
specimen surface heating was observed. Low R ratios
ŽF 0.1. and test frequency 3 or 5 Hz achieved the Tension]tension axial fatigue data with frequency of
greatest surface heating, regardless of normalized max- fatigue load 5 Hz or less for E-glass FRP composites
imum stress. Surface heating decreased, respectively without environmental concerns are reviewed in this
for: R ratios ŽF 0.1. and test frequency 1 Hz; R ratios study. This frequency limit minimizes selecting fatigue
Žs 0.5. and test frequency 3 or 5 Hz. No surface life results adversely affected by internal heat genera-
heating was observed for R ratio 0.5 and test frequency tion. The fatigue design data include the effects of
1 Hz or R ratio 0.9 and test frequency 1 and 5 Hz. The various combinations of the test parameters: frequency
E-glassrvinylester composite specimens tested at nor- of applied load, R ratio, specimen shape, fiber rein-
malized maximum stress 0.4, and which fall slightly forcement, percent fiber, resin and failure criteria in
outside the 95% lower confidence level, were observed fatigue testing. Too many combinations of these test
to have surface heating. These two data points did not parameters exist among the fatigue tests to distinguish
significantly alter the lower confidence level achieved the effect of an individual parameter on fatigue life.
without the E-glassrvinylester composite data, thus, However, the E-glassrvinylester data did show increas-
the given 95% lower bound confidence level effectively ing fatigue life for increasing R ratio. A fatigue life
represents all data as shown in Table 1. The E- diagram of all the E-glass FRP composite data, regard-
glassrvinylester composite data falls within or near the less of test parameter combination, does present an
overall data band indicating the combinations of test extended data band for which a lower bound 95%
parameters not greatly changing the fatigue life charac- confidence level is established. The lower bound is a
teristics. conservative estimate of E-glass FRP composite fatigue
The lower R ratios represent large stress ranges and life in tension]tension axial fatigue for R ratios 0.0,
thus form a conservative data group. An upper bound 0.05, 0.1, 0.5, 0.9 and for fatigue life less than or equal
to the Rs 0.1 Žor less. data band is formed by the to one million cycles. This lower bound is represented
0r90 continuous fiber reinforcement Rs 0.05 dogbone by a linear relation. This lower bound can be used to
specimens for fatigue life less than 1000 cycles and by estimate E-glass FRP composite fatigue life in
the short fiber reinforcement Žisotropic. Rs 0 necked tension]tension axial fatigue, with R ratio and fatigue
specimens for fatigue life between 1000 and 1 000 000 life restrictions, until further fatigue testing refines the
cycles. A lower bound to this band is formed by the effect of the individual test parameters on fatigue life.
unidirectional continuous fiber reinforcement Rs 0.1
dogbone specimens at test frequency 5 Hz for fatigue
life less than 10,000 cycles and by the short fiber Acknowledgements
reinforcement Žanisotropic. Rs 0.1 dogbone specimens
and the "45 continuous fiber reinforcement necked This research was funded by the National Science
specimens at test frequency 5 Hz for fatigue life Foundation Visiting Professorships for Women Pro-
10 000]1 000 000 cycles. A lower bound 95% confidence gram. The host institution for this program was the
level is defined for all the E-glass FRP composite School of Civil and Environmental Engineering, Geor-
tension]tension axial fatigue data, regardless of varying gia Institute of Technology, Atlanta, Georgia. The ex-
310 C.E. Demers r Construction and Building Materials 12 (1998) 303]310

cellent laboratory work was conducted by Patrick Bair, w3x Owen MJ, Rose RG. Polyester flexibility versus fatigue behav-
graduate student. All the above are gratefully ac- ior of RP. Modern Plastics 1970;47:130]138.
w4x Mandell JF, Huang DD, McGarry FJ. Tensile fatigue perfor-
knowledged for support of this project. Special thanks
mance of glass fiber dominated composites. 36th Annual Tech-
also to Dennis Senal for his assistance with the graph- nical Conference. Reinforced PlasticsrComposites Institute,
ics. Society of Plastics Industry, Session 10A, 1981:1]6.
w5x Jones CJ, Dickson RF, Adam T, Reiter H, Harris B. The
environmental fatigue behaviour of reinforced plastics. Proc
Appendix 1: Notation Roy Soc London 1984;A 396:315]338.
w6x Sims GD, Gladman DG. Effect of test conditions on the
fatigue strength of a glass-fabric laminate: Part A } fre-
The following symbols are used in this paper: quency. Plastics and Rubber: Materials and Applications,
1978:41]48.
N s fatigue life measured in cycles; w7x Mandell JF, Huang DD, McGarry FJ. Fatigue of glass and
Smax s maximum tensile fatigue stress in fatigue load carbon fiber reinforced engineering thermoplastics. 35th An-
cycle; nual Technical Conference, Reinforced PlasticsrComposites
Institute, Society of Plastics Industry, Section 20D, 1980:1]11.
Sr s stress range; and
w8x Mandell JF. Fatigue Behaviour of fibre-resin composites. In:
Sult s ultimate tensile strength. Pritchard G, editors. Developments in reinforced plastics }
2, Properties of laminates. London, NY: Applied Science Pub-
lishers, 1982:67]107.
References w9x Echtermeyer AT, Buene L, Engh B, Sund OE. Significance of
damage caused by fatigue on mechanical properties of com-
w1x Davis JW, Sundsrud GJ. Fatigue data on a variety of nonwoven posite laminates. SAMPE, 1993:38-B-1-38-B-10.
glass composites for helicopter rotor blades. Composite Materi- w10x Jessen SM, Plumtree A. Fatigue damage accumulation in pul-
als: Testing and Design ŽFifth Conference.. ASTM STP truded glassrpolyester rods. Composites 1989;20:559]567.
1979;674:137]148. w11x Mandell JF, McGarry FJ, Im J, Meier U. Fiber orientation,
w2x Agarwall BD, Dally JW. Prediction of low-cycle fatigue be- crack velocity, and cyclic loading effects on the mode of crack
haviour of GFRP: an experimental approach. J Mat Sci extension in fiber reinforced plastics. Failure Modes in Com-
1975;10:193]199. posites, II, Metallurgical Soc. AIME, 1974:33.