Professional Documents
Culture Documents
www.elsevier.com/locate/ijfatigue
Received 15 October 2000; received in revised form 10 August 2001; accepted 20 November 2001
Abstract
Unidirectional glass fiber reinforced and glass–carbon fiber reinforced epoxy matrix composite specimens were subjected to
tension–tension fatigue in air and in distilled water at 25 °C. While no significant change in fatigue life was observed for both
types of specimens tested in air and in water when cyclically tested at 85% of average ultimate tensile strength (UTS), the detrimental
effect of water becomes apparent at lower stress levels of 65 and 45% UTS. Compared to specimens tested in air, cyclic loading
in water results in shorter fatigue lives for both glass and hybrid specimens. While all of the glass fiber specimens did not survive
to 106 cycles when cyclically loaded in water, hybrid specimens (with 25% carbon fiber (by volume), 75% glass fiber (by volume),
30% total fiber volume fraction) showed better retention in structural integrity under environmental fatigue, for fatigue lives up to
107 cycles, a consequence of the corrosion resistant of carbon fiber. Thus it is shown, by incorporating appropriate amount of
carbon fibers in glass fiber composite, a much better performance in fatigue can be achieved for glass–carbon hybrid composite.
A simple life prediction model for the hybrid composite is proposed. Model predictions are compared with experiments results
from both laminated interply and intermingled intraply hybrid composites. Results suggest that synergistic effect of the reinforcing
fibers is critical in governing the fatigue behavior of intraply hybrid composite. 2002 Elsevier Science Ltd. All rights reserved.
0142-1123/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 2 - 1 1 2 3 ( 0 1 ) 0 0 2 1 0 - 9
848 Y. Shan, K. Liao / International Journal of Fatigue 24 (2002) 847–859
reductions in fatigue life under fluid immersion were The total fiber volume fraction was kept at approxi-
observed. mately 30% for both all-glass and hybrid specimens.
It is indicated by majority of the available data that Hybrid composite panels of relative glass to carbon fiber
severe material degradation could be introduced to the volume fractions of 100:0, 95:5, 90:10, 75:25, 50:50, and
material under environmental fatigue of GFRP, even at 0:100 were fabricated to assess the effect of relative
low stress levels. In order to enhance the durability of glass to carbon volume ratio on the mechanical proper-
GFRP, hybrid composite systems using both carbon and ties of the hybrids. Bar specimens of dimension
glass fibers have been proposed [16,17]. As mentioned 100×12×1.5 mm were cut from the unidirectional panel
previously, hybrid composites seem to be an attractive after being released from the mold, and the two edges
material in resisting environmental attack. Both fatigue of each specimen were polished to avoid edge effects.
performance and environmental resistance of glass–car- Composite print circuit board (PCB) end-tabs of length
bon hybrid composite is enhanced compared to all-glass 25 mm were bonded to the specimen surface, leaving a
composites [9]. Compared to interply hybrid composites gauge length of 50 mm. Quasi-static tensile tests were
constructed by laminating together distinct layers of performed using an Instron 5567 universal test
glass-fiber-reinforced composite and carbon-fiber- machine to determine their ultimate tensile strength
reinforced composite, the intraply hybrid composite, in (UTS), at a cross-head speed of 1 mm/min. At least four
which bundles of carbon and glass fibers are mixed specimens were tested for each group.
within a single layer, is considered more desirable in Fatigue and environmental fatigue were carried out at
terms of reduced mismatches. However, very limited stress levels of 30, 45, 65, and 85% UTS using an
work has been carried out on intraply hybrid systems Instron 8516 servo-hydraulic test frame. Two groups
because they are more difficult to fabricate. of composites specimens were employed to characterize
In this paper, results of a study on fatigue and environ- the fatigue and environmental fatigue behavior: all-glass
mental fatigue behavior of unidirectional glass com- 100:0 and hybrid 75:25. Specimens were cyclically
posite and glass–carbon intraply hybrid composite are loaded using a sinusoidal wave function at 10 Hz for
presented, and thereby it is shown that the hybrid 65% and 85% UTS, and 20 Hz for 30 and 45% UTS,
approach is viable in enhancing long term durability. A at R ⫽ 0.1 (where R is the ratio of minimum to the
simple phenomenological fatigue life prediction model maximum cyclic stress) at 25 °C. Environmental fatigue
for unidirectional hybrid composites is also presented. was performed with specimens immersed in distilled
water, shown schematically in Fig. 2, where the gauge
section of a specimen was wrapped with a rubber bag
2. Materials and experiment filled with water-containing cotton with the lower end
sealed. Fatigue tests were interrupted periodically to
Unidirectional glass/epoxy and intraply glass– measure the tensile modulus of a specimen by means
carbon/epoxy specimens were fabricated by a wet wind- of an extensometer (Instron model 2620-601). Surface
ing and hot pressing method described by Peijis et al. related damage during fatigue was also visually exam-
[18]. E-glass fiber, PAN–HTA carbon fiber, and an Epic- ined. Specimens that have survived to 107 cycles were
ote 600 epoxy, a room temperature cured resin, are used
for this study. Fibers are used as-received without sur-
face treatments. Continuous glass fiber bundles were first
wound in parallel on a rectangular metallic frame. In
order to enhance the environmental resistance, carbon
fibers were wound after and on top of the glass fiber
bundles with an even spread. As a consequence, more
carbon fiber was near the surface of the resulting com-
posite plate. The epoxy matrix was prepared by first mix-
ing with the hardener at a specified ratio recommended
by the vendor, and subsequently degassed in a vacuum
oven for 5 min at a pressure of 25 mm Hg. Fibers were
wetted by pouring the degassed epoxy into a plastic bag
containing the fiber bundle-frame assembly. They were
then put into an open-ended mould pre-pasted with
release film to ensure ease of detachment of the com-
posite after curing. The whole assembly was subjected
to hot press for 2 h at 80 °C and a pressure of 20 t.
When curing was completed, the composite plate was
removed from the mold and post cured at 110 °C for 6 h. Fig. 2. Environmental fatigue specimen setup.
850 Y. Shan, K. Liao / International Journal of Fatigue 24 (2002) 847–859
Typical stress–strain curves for various hybrid compo- 3.2. Fatigue behavior
sitions are shown in Fig. 3. Tensile test results are sum-
marized in Fig. 4. All specimens (all-carbon, all-glass, Results of fatigue test of all-glass 100:0 and hybrid
and hybrid) show linear elastic behavior to final failure. 75:25 in air are summarized in Fig. 6. When calculating
Different tensile failure modes are seen, they varied from the slope of stress life (S–N) curves, test results of all-
dominant fiber roving splitting and bundle failure in all- glass 100:0 at 30%, and 45% UTS, and hybrid 75:25 at
glass specimens to dominant brittle-like failure mode in 45% UTS were excluded since at these load levels, all
all-carbon specimens where only few splitting between specimens survived 107 cycles and showed no sign of
fiber roving was seen. The tensile failure modes for damage as reflected in their residual strength (Figs. 7
hybrid specimens were in between these two extremes,
and 8). The S–N curve for the all-glass 100:0 specimens
has a slope of 18.8 MPa per decade of life, or about
5.7% of the static strength per decade. Our results do
not agree with the arguments of Mandell et al. [7,8],
who maintained that the slope of S–N curves for GRP
Fig. 14. Transverse matrix cracking and fiber debonding on specimen Fig. 16. Fatigue failure surface of a specimen tested in water.
surface.
冘i
ni
Ni
ⱖ1 (3) sr ⫽
s·Er
Ernfr
(r ⫽ 1 or 2) (11)
where ni is the number of cycles experienced at a certain And the number of cycles to failure under sr is
stress level, si, and Ni is the number of cycles to failure Nr ⫽ f⫺1
r (sr) (r ⫽ 1 or 2) (12)
at si.
Taking into account the damage accumulated during
The S–N curve of a composite is most simply and stage 1, the remaining cycles to failure, Nremain, is
conveniently described by the following simple linear Nremain ⫽ Nr(1⫺D) ⫽ f⫺1
r (sr)(1⫺D) (13)
function
By adding the cycles in stage 1 and stage 2 together,
s⫽sult⫺msultlogNf (4) the total number of cycles to failure, Ntotal, is
where sult stands for its ultimate strength, m is the Ntotal ⫽ min(N1,N2) ⫹ Nremain (14)
decrease in fatigue strength per decade of fatigue life,
and Nf is the number of cycles to failure at applied cyclic where min (N1,N2) denotes the smaller value among N1
stress, s. However, since Eq. (4) may not be applicable and N2. Therefore, Ntotal is a function of applied stress,
to certain types of composites such as aramid fiber s, and the volume fraction, nf, of the component.
reinforced composites, a more general form would be Similar to the constant strain model in the static ten-
sile test of hybrid composites, this simple fatigue model
s ⫽ f(Nf) (5) does not take into account the synergetic effects such as
and in reverse form bridging of cracks in the low elongation component by
high elongation counterpart, thus deviations of actual
Nf ⫽ f⫺1(s) (6) fatigue response from the model predictions can be
regarded as fatigue hybrid effects. Thus if the actual
The loading history of a hybrid composite is divided fatigue life is longer than the prediction, hybridization
into two stages. The first stage is the period from the has a desirable positive synergetic effect. If the opposite
start of the test to the moment when one of its load- is true, then hybridization only yields a detrimental syn-
carrying components fails. Stage two is the duration ergetic effect or negative effect. In the following, model
from the failure of one of the components to final failure prediction will be compared to fatigue data for two types
of the composite. of hybrids, namely, the interply and the intraply com-
In the first stage, before a component fails, both posite.
components are able to carry the applied load, thus the
stresses s1, and s2, carried by components 1 and 2 are
3.4.1. Interply hybrids
s·E1 s·E2 Both all-carbon and all-glass unidirectional com-
s1 ⫽ s2 ⫽ (7)
E1nf1 ⫹ E2nf2 E1nf1 ⫹ E2nf2 posites essentially follow a linear S–N curve [Eq. (4)].
Fatigue behavior of unidirectional laminated carbon and
where s is the applied stress, E is the Young’s modulus, glass composites (in which carbon-fiber and glass-fiber
nf is the fiber volume ratio, subscript 1 or 2 denotes laminae are arranged in an alternate manner) are
component 1 or 2, respectively. Thus the number of extracted from [6], and are tabulated in Table 1. Predic-
cycles to failure for each component are tion of the fatigue response of a family of glass–carbon
N1 ⫽ f⫺1 laminated hybrids is illustrated in Fig. 17 by a 3-dimen-
1 (s1) and N2 ⫽ f2 (s2)
⫺1
(8)
sional (3-D) plot. Cycles to failure, and carbon fiber con-
If N1⬎N2 then component 2 fails first. Damage, D, tent are represented by the Y- and X-axis, respectively,
accumulated in the surviving component 1 is and applied cyclic stress is expressed by the Z-axis. As
856 Y. Shan, K. Liao / International Journal of Fatigue 24 (2002) 847–859
Table 1
S–N curve parameters for all-glass and all-carbon unidirectional lami-
nated composites. From [6]
Fig. 20. Comparison between fatigue model prediction and the Fig. 21. Comparison between model predictions of environmental
experimental results of intraply hybrid composite. fatigue behavior of hybrid 75:25 and experimental results.
858 Y. Shan, K. Liao / International Journal of Fatigue 24 (2002) 847–859
taneous failure. However, at high cycles, it is the glass- [2] Chou TW, Kelly A. Mechanical properties of composites. Annual
fiber component that fails first, followed by carbon fib- Review of Materials Science 1980;10:229–59.
[3] Phillips LN. The hybrid effect—does it exist? Composites
ers. Thus the fatigue response of a hybrid is dominated 1976;7:7–8.
by glass fibers at low stress for high cycle environmental [4] Marom G, Harel H, Neumann S, Friedrich K, Schulte K, Wagner
fatigue. This may further explain why the moisture HD. Fatigue behavior and rate dependent properties of aramid
effects become obvious only at low stress levels. The fibre/carbon fibre hybrid composite. Composites
experiment data show a considerable positive deviation 1989;20(6):537–44.
[5] Fernando G, Dickson RF, Adam T, Reiter H, Harris B. Fatigue
from the prediction. As discussed earlier, the tensile behavior of hybrid composites: part 1 carbon/kevlar hybrids.
strength of the intraply hybrids is higher than predictions Journal of Materials Science 1988;23:3732–43.
from the constant strain model, thus the entire experi- [6] Dickson RF, Fernando G, Adam T, Reiter H, Harris B. Fatigue
mental S–N data of hybrid 75:25 is higher than the con- behavior of hybrid composites: part 2 carbon–glass hybrids. Jour-
stant strain based life prediction model. This is also true nal of Materials Science 1989;24:227–33.
[7] Mandell JF, Huang DD, McGarry FJ. Tensile fatigue perform-
under fluid immersion condition. Therefore the proposed ance of glass fiber dominated composites. Composites Tech-
model can also be used as a lower bound for intraply nology Review 1981;3(3):96–102.
environmental fatigue behavior. [8] Mandell JF, McGarry FJ, Hsieh AJY, Li CG. Tensile fatigue of
glass fibers and composites with conventional surface compressed
fibers. Polymer Composites 1985;6(3):168–74.
[9] Liao K, Schultheisz CR, Hunston DL, Brinson LC. Long-term
4. Conclusions durability of fiber-reinforced polymer-matrix composite materials
for infrastructure applications: a review. SAMPE, Journal of
Advanced Materials 1998;30(4):2–40.
Base on the results obtained in this study, the follow- [10] Vauthier E, Abry JC, Bailliez T, Chateauminois A. Interactions
ing major conclusions can be reached: between hygrothermal ageing and fatigue damage in unidirec-
tional glass/epoxy composites. Composites Science and Tech-
앫 Cyclic loading in distilled water significantly shortens nology 1998;58:687–92.
[11] Jones CJ, Dickson RF, Adam T, Reiter H, Harris B. The environ-
the fatigue life of both unidirectional glass and glass– mental fatigue behaviour of reinforced plastics. Proceedings of
carbon hybrid fiber reinforced epoxy matrix com- the Royal Society London Series A: Mathematical and Physical
posites compared to their fatigue lives in air, at lower Sciences 1984;396:315–38.
stress levels. [12] Sumsion HT, Williams DP. Effect of environment on the fatigue
앫 Glass–carbon hybrid specimens showed a better of graphite–epoxy composites. Fatigue of composite materials,
ASTM STP 569: American Society for Testing and Materials,
retention in fatigue life in water than that of all-glass 1975, pp. 226–47.
composite specimens, up to 107 cycles. Therefore by [13] Komai K, Minoshima K, Shiroshita S. Hygrothermal degradation
hybridization with an appropriate amount of carbon and fracture process of advanced fibre-reinforced plastics.
fibers, resistance to environmental fatigue degradation Materials Science and Engineering A 1991;143:155–66.
of GFRP can be enhanced significantly. [14] McBagonluri F, Garcia K, Hayers M, Verghese KNE, Lesko JJ.
Charaterization of fatigue and combined environment on dura-
앫 The proposed phenomenological fatigue model is cap- bility performance of glass/vinyl ester composite for infrastruc-
able of estimating the fatigue behavior of hybrid com- ture applications. International Journal of Fatigue 2000;22:53–64.
posites from their parent composites. The fact that this [15] Liao K, Schultheisz CR, Hunston DL. Long-term environmental
simple model can better describe the behavior of fatigue of pultruded glass-fiber-reinforced composites under
interply composite suggests that synergistic effects in flexural loading. International Journal of Fatigue 1999;21:485–
95.
the intraply composites are significant and a refined [16] Hofer KE Jr, Stander M, Bennett LC. Degradation and enhance-
model entails the incorporation of the bridging mech- ment of the fatigue behavior of glass/graphite/epoxy hybrid com-
anism. posites after accelerated aging. In: Proceedings of the 32nd
Annual Technical Conference, The Society of the Plastics Indus-
tries, Inc. 1977, section 11-F.
[17] Manders PW, Bader MG. The strength of hybrid glass/carbon
Acknowledgements fiber composites, part 2. Journal of Materials Science
1981;16:2246–56.
The authors would like to acknowledge the support [18] Peijis AAJM, De Kok JMM. Hybrid composites based on poly-
ethylene and carbon fibres: part 6: tensile and fatigue behavior.
of this research by Nanyang Technological University Composites 1993;24(1):19–32.
through AcRF grant RG7/98. [19] Reifsnider KL, Schulte K, Duke Jr JC. Long-term fatigue
behavior of composite materials. In: O’Brien TK, editor. Long-
term behavior of composites, ASTM STP 813. Philadelphia:
American Society for Testing and Materials; 1983, p. 136–59.
References [20] Hofer KE, Skaper GN, Bennett Effe LC. Effect of moisture on
fatigue and residue strength losses for various composites. Jour-
[1] Manders PW, Bader MG. The strength of hybrid glass/carbon nal of Reinforced Plastics and Composites 1987;6:53–65.
fibre composites: part 1: failure strain enhancement and failure [21] Watanabe M. Effect of water environment on fatigue behavior
mode. Journal of Materials Science 1981;16:2233–45. of fiberglass reinforced plastics. In: Tsai SW, editor. Composite
Y. Shan, K. Liao / International Journal of Fatigue 24 (2002) 847–859 859
materials: testing and design (5th Conference), ASTM STP 674. on the Fracture Mechanics of Ceramics, vol. 4, Crack Growth
Philadelphia: American Society for Testing and Materials; 1979, and Microstructure. New York: Plenum Press; 1973, p. 549–81.
p. 345–67. [26] Evans AG, Wiederhorn SM. Proof testing of ceramic materials—
[22] Fried N. Degradation of composite materials: the effect of water an analytical basis for failure prediction. International Journal of
on glass-reinforced plastics. Mechanics of composite materials. Fracture 1974;10:379–92.
In: Wendt FW, Liebowitz H, Perone N, editors. Proceedings of [27] Wiederhorn SM. Subcritical crack growth in ceramics. In: Bradt
the Fifth Symposium on Naval Structure Mechanics, 1967, pp. RC, Hasselman DPH, Lange FF, editors. Symposium on the Frac-
813–37. ture Mechanics of Ceramics, vol. 2, Microstructure, Materials and
[23] Hofer KE Jr, Bonnett LC, Stander M. Effect of various fiber sur- Applications. New York: Plenum Press; 1973, p. 613–46.
face treatments on fatigue behavior of glass fabric composite in [28] Ritter Jr JE. Mechanics of subcritical crack growth in glass. In:
high humidity environments. In: Proceedings of the 31st Annual Bradt RC, Hasselman DPH, Lange FF, editors. Symposium on
Technical Conference of SPI, Washington DC, February 1976, the Fracture Mechanics of Ceramics, vol. 4, Crack Growth and
section 6A. Microstructure. New York: Plenum Press; 1973, p. 667–86.
[24] Shih GC, Ebert LJ. The effect of the fiber/matrix interface on the [29] Gupta PK. Examination of the tensile strength of E-glass fiber in
flexural fatigue performance of unidirectional fiberglass com- the context of slow crack growth. In: Bradt RC, Hasselman DPH,
posites. Composites Science and Technology 1987;28:147–61. Lange FF, editors. Symposium on the Fracture Mechanics of Cer-
[25] Wiederhorn SM. Mechanics of subcritical crack growth in glass. amics, vol. 5, Surface Flaws, Statistics and Microstructures. New
In: Bradt RC, Hasselman DPH, Lange FF, editors. Symposium York: Plenum Press; 1973, p. 291–303.