Document
a) ?3-J
FATIGUE OF COMPOSITES* By Michael J. Salkind Sikorsky Aircraft Division, United Aircraft Corporation, United States INTRODUCTION During the past decade, the extensive development activity in composite struc- tures has indicated the promise of substantial improvements in the performance of aerospace systems. This experience, however, has shown that composite materials are substantially different from our former materials of construction, and new con- cepts in analysis, design, fabrication, quality assurance, and even systems manage- ment will be necessary.
A major difference between composites and metals exists in their respective behavior in a fatigue environment. Whereas metals usually fail by crack initiation and growth in a manner which has come to be predictable through fracture mechanics analysis, composites exhibit several modes of damage including delamination, matrix crazing, fiber failure, void growth, matrix cracking, and composite cracking. A par- ticular structure may exhibit any or all these damage modes, and it is difficult to pre- dict, a priori, which mode will dominate and cause failure. The selection of fiber and matrix can produce predictable fiber or matrix-dominated failure in simple unidirec- tional specimens. (See refs. 1 to 4.) In real structures, however, the complex multi- directional loadings and complex reaction of nonsimple laminates precludes easy prediction of failure modes. Also, joints and attachments in composite structures generally result in failure modes which are peculiar to a particular design.
A characteristic of composite materials which differs substantially from metals is the relative difference between low- and high-cycle fatigue behavior. Whereas most metals behave according to the so-called Coffin-Manson relationship (refs. 5 to 7) in low-cycle fatigue, composites have been shown to be more sensitive to strain range (ref. 4). This sensitivity results in the high-cycle fatigue strength of composites being high with respect to static- and low-cycle fatigue strength. Many structures designed for fatigue experience a spectrum of high as well as low stresses, and whereas the more numerous low stresses may be the critical design factor for a metal structure, the same structure made from a composite material may well be critical in low-cycle fatigue.
A problem arises in the design of composite structures for fatigue loading because of the lack of an adequate d.efinition of failure. A large part of the fatigue *Also published in ASTM STP 497.
data in the literature is based ontime to fracture. For high-cycle fatigue, metals are
generally structurally adequateto the point of crack initiation (andto some extent beyond
that), which is usually a large part of the time to fracture. Although it is preferable to
use fatigue data based on crack initiation, the error introduced by designing metal struc-
tures using fatigue data based on fracture is usually not significant. Using such an
approximation for composites, however, could lead to disastrous results. Composites
generally begin to exhibit changesin properties very early in the total life to fracture.
Suchchangesin elastic properties could lead to structural failure long before the struc-
ture is in danger of fracturing. Rotating airfoils such as helicopter rotor blades or gas
turbine blades are subject to aeroelastic instabilities if the fundamentalfrequency shifts
becauseof early fatigue damagethat causesstiffness changes. A composite spring whose
spring constant changesbeyondan acceptablevalue would be considered failed even
though it was in no danger of fracturing. The same composite usedfor a tension cable
application for which stiffness is not critical might have an acceptablefatigue life to fail-
ure several orders of magnitudehigher than the stiffness critical spring under the same
loading conditions. A further complication to this problem is the fact that composites
are anisotropic, andfor anynumber of cycles, the changein stiffness in onedirection
may be unrelated to the changein stiffness in a seconddirection.
The requirement for an adequatefailure criterion, coupledwith the challenge of
providing adequatedamagedetection schemesfor multiple damagemodes, clearly indi-
cates the requirement for a new approachto the design of fatigue-critical composite
structures. This paper includes a review of the fatigue behavior of composite materials
and structures anda proposedapproachfor design of fatigue-critical components.
FATIGUE OF COMPOSITEMATERIALS
A large bodyof small specimenfatigue data has been generatedover the past
10years. These data are primarily for unidirectional laminates and, as mentionedpre-
viously, are baseduponfracture as the definition of failure. Hence, much of it is of
limited value for use in design. A survey of pertinent observations is included in this
section.
Fiber-Reinforced Polymers
The most widely used class of composite materials, glass-fiber-reinforced poly-
mers, has been the subject of extensive fatigue testing (refs. 8 to 23). Although the data
in references 8 to 23 are based uponfracture as the failure criterion, they give us an
indication of the effect of significant variables, such as resin composition and content and fiber composition and orientation on fatigue. Boiler (ref. 11) has evaluated a variety
of matrix systems andfound epoxidesto be superior in fatigue. His data, seenin fig-
ure 1, is more than 15years old, andimproved surface treatments andprocessing have
since been developed;however, the relative ranking remains unchanged(ref. 19). As
seenin figure 2, varying the resin content between20 and 37 percent has a negligible effect
on fatigue behavior for _5 ° glass-fiber composites. The effect of fiber orientation is
rather complex. Althoughthe tensile strength of unidirectional composites is a maximum
at 0° to the fibers (ref. 24), in fatigue the unidirectional construction is not optimum, as
can be seenin figure 3. The best explanation for this phenomenon is the fact that unidi-
rectional material is subject to splitting and rapid crack propagation in the matrix parallel to the fibers. Fatigue data for +5 ° , 67% 0o/33% 90 ° , and Style 181 satin weave (0o/90 ° ) are compared in figure 4. In general, nonwoven materials are superior to woven mate- rials in fatigue. Note also the low notch sensitivity of these materials. A comparison of the behavior of S-glass and E-glass composites (ref. 15) is seen in figure 5. The higher modulus S-glass is consistently stronger in fatigue than E-glass. The effect of mean stress is seen in figure 6 for tension-compression and tension-tension behavior.
There have also been some measurements of the compression-compression fatigue behavior in low-cycle fatigue (ref. 20). These data indicate that the effect of mean stress is similar to that for metals, that is, the Goodman diagram is approximately linear.
In recent years, the realization that fracture was not an adequate design criterion for failure has led to studies of failure mechanisms in fatigue which provide a foundation for design. Broutman (ref. 25) studied the mechanisms of failure in glass-fiber- reinforced polymers subjected to fatigue loading. He noted cracks originating at fiber- matrix debonds propagating through the matrix and being deflected by fibers. Recent studies (refs. 26 to 30) have quantitatively described the changes in elastic and strength properties associated with this type of damage. Smith and Owen (ref. 26) evaluated eleven different composite systems with modulus values ranging from 0.5 × 106 to 6.5 × 106 psi and found that the initial damage debonding occurred at 0.3% strain as seen in figure 7.
Thus, the limiting factor in fatigue is not the fiber but the interface or the matrix. As described in the following section, studies with metal matrix composites confirm this behavior. The data seen in figure 8 for chopped-strand mat-reinforced polyester lami- nates confirm that initial damage occurs at stresses well below those required for frac- ture. The effect of such damage on structural capability will determine whether the part has failed. Based on the observation of a critical strain, Smith and Owen postulated a critical maximum stress independent of mean stress for any material. As seen in fig- ure 9, this postulation was found to be incorrect (ref. 26).
Cessna et al. (ref. 27) performed constant-deflection flexural fatigue tests on glass- reinforced polypropylene and monitored the load decay (proportional to modulus decay) with cycles as seen in figure 10. They also monitored the temperature rise, as seen in figure 10, due to viscoelastic energy dissipation, which is common for polymeric com- posites. (See refs. 31 to 37.) In addition to indicating progressive fatigue damage, the temperature rise also contributes to weakening the material and shortening its fatigue life. As seen in figure 11, by cooling their specimens to maintain isothermal conditions, Cessna et al. (ref. 27) were able to extend both the cycles to onset of stiffness change and the fracture life by an order of magnitude.
Broutman and Sahu (ref. 28) have related the changes in residual tensile strength and modulus to the development of cracks in 0o/90 ° crossplied material as seen in fig- ure 12. Although the decrease in residual strength and primary modulus is expected because of the increasing crack density, the initial increase in secondary modulus remains unexplained. A quantitative relationship between modulus change and crack den- sity (fig. 13) has been developed on the same material. Fujii and Mizukawa (ref. 30) have also determined the change in elastic and strength properties with cycling for laminates consisting of several combinations of roving cloth, chopped mat, and woven cloth.
The higher modulus composite materials, graphite and boron, have exhibited higher fatigue strengths than glass-reinforced polymers, as seen in figure 14 (refs. 15, 38 to 41). This difference is primarily attributed to the higher modulus resulting in less strain in the matrix and interface at the same cyclic stress level. The phenomenon of low fatigue strength at zero mean stress for the high modulus composites as seen in fig- ure 14 was first noted for boron-reinforced aluminum. It is thought to be the result of low transverse strength of unidirectional composites resulting in splitting under cyclic compressive loads. This behavior severely limits the use of unidirectional composites as discussed above.
A major variable which can affect the data obtained in a composite fatigue test is the specimen geometry. This variable is very much a function of the particular laminate orientation being tested, as interlaminar shear can be a primary controlling factor (refs. 42 and 43). Although this problem is negligible for a unidirectional material, it becomes a major factor in fatigue testing of _45 ° laminates. Figure 15 compares the axial tension-tension fatigue behavior (based on fracture) for +45 ° 1002 E-glass compos- ites for three different specinlen configurations (ref. 43). The straight-sided specimen is flat, and each fiber terminates at the edge, thus, high interlaminar shear stresses are created. The x-type specimen has all fibers continuous from grip to grip; thus, inter- laminar shear stresses at the edge are precluded. The latter type has the disadvantage of having a vanishingly small gage length and very uniformly loaded fibers, which are not representative of the types of loading experienced in most structures; thus, the resultant data are considered to be too optimistic for design. The tubular specimen has a uni- formly loaded gage section and no fiber edges; thus, interlaminar shear is precluded. It is felt that such a specimen comes closest to yielding representative material properties for design. Edges are usually handled as a separate factor in design. Although the dif- ference between the straight-sided and tubular specimens is only 5% for glass, it is approximately 50% for boron, as can be seen in figure 16. The reason for this effect is that the interlaminar shear stresses are higher for high modulus materials.
Fiber-Reinforced Metals Although fiber-reinforced metals considerably lag reinforced polymers in terms of development and usage, some of the earliest and best fundamental studies of fatigue have been accomplished with metal matrix composites. Forsyth, George, and Ryder (ref. 1) demonstrated that the inclusion of steel wires in an aluminum-alloy sheet reduces the rate of crack propagation substantially (fig. 17), although the improvement in total fatigue life to fracture is small (fig. 18). Baker and Cratchley (refs. 2, 44 to 48) performed extensive studies of the fatigue behavior of silica- and steel-fiber-reinforced aluminum alloys. Although silica-reinforced aluminum did not show promise as a fatigue-resistant material, Baker and Cratchley made important observations concerning failure modes and anelastic behavior. They identified the crack-diverting capability of strong fibers (ref. 2) and made important observations concerning the stress-strain behavior and damping capability of composites as seen in figure 19 (ref. 44). In addition, Baker quan- titatively defined the effect of fiber length on fatigue behavior (ref. 46) and evaluated the effects of fiber fatigue behavior and the interface (ref. 47).
Extensive studies have been made of the fatigue behavior of composites made by unidirectional solidification of eutectic alloys (refs. 4, 49 to 52). Because these com- posites have very regularly distributed fibers and well-bonded interfaces, they serve as excellent systems for studying the mechanical behavior of composites without the variable fabrication effects common to other composites. The A1-A13Ni (10% reinforcing A13Ni whiskers) and the A1-CuA12 (50% reinforcing CuA12 platelets) composite materials exhibit markedly different fatigue behavior as seen in figure 20. A comparison of the stress-strain behavior of the two materials showing fatigue failure at the same stress amplitude (ref. 49) reveals that A1-CuA12 appears to work-harden more rapidly with narrower hysteresis loops than A1-A13Ni. It can be speculated that the wide CuAI 2 platelets are more effective at blocking plastic flow than the A13Ni whiskers (which have a spacing at least an order of magnitude too large for optimum dispersion hardening). In addition, the greater volume fraction of CuA12 is more effective at blocking plastic flow in the matrix. If matrix strain is the controlling factor, then the behavior seen in fig- ure 20 would be an expected consequence of the difference in stress-strain behavior.
The fatigue behavior of metals has been found to obey a simple empirical relation- ship (refs. 5 to 7, 53 and 54)
G Ny
(1)
AeT = MNZ + _"
Plastic Elastic
where
total cyclic strain range elastic modulus N number of cycles to failure material constants M, G, z, 7 This relationship is depicted schematically in figure 21, and it is seen that the plastic component (first term of eq. (1)) dominates at low cycles (high strain amplitudes), and the elastic component dominates at high cycles. Although most metals exhibit values of z of -0.5 to -0.6 (refs. 53 and 54), A13Ni whisker-reinforced aluminum exhibits much lower values as seen in figure 22. It has been proposed (ref. 4) that the low-cycle fatigue behavior is not governed entirely by the plastic behavior of the aluminum (z = -0.5) but also by the elastic behavior of the A13Ni (7 < -0.1). This type of behavior would be expected for most composites having fibers which behave elastically in the stress range of use (glass, boron, graphite, highly cold-worked steel) and accounts for the relatively flat S-N (that is, stress S - number of cycles to failure N) curves discussed in the Introduction.
As seen in figure 23, the flexural fatigue behavior of A1-A13Ni is substantially higher than that for the matrix alone. The mode of failure is matrix cracking, the fibers serving to reduce matrix strain. Testing in a protective atmosphere such as argon results in higher fatigue strength as seen in figure 24, which further verifies the fact that failure is matrix dominated. Although the aluminum is susceptible to attack by moisture in the atmosphere, the A13Ni whiskers are not.
The fatigue behavior of boron-reinforced aluminum has been extensively studied (refs. 55 to 59). The very high modulus of the reinforcing fiber keeps the matrix strain low for any given stress level. This factor, coupled with the excellent fatigue resistance of boron fiber itself (ref. 60), provides extremely good fatigue resistance as seen in fig- ure 25. The data by Young and Carlson (ref. 56) is particularly valuable as it records changes in deflection for torsion, tension, and combined-load fatigue testing.
Gates and Wood (ref. 61) performed detailed studies of the microstructural changes which accompanied the torsional fatigue testing of copper reinforced circumferentiaily by tungsten or molybdenum wire. They noted that work hardening followed by crack initia- tion occurred in the matrix between fibers.
FATIGUE OF COMPOSITE STRUCTURES In order to define a design base for structures, it is necessary to determine the full-scale fatigue behavior of composite materials to identify the effects of size, manu- facturing variation, and combined loads. A limited amount of experience now exists with full-scale components (refs. 62 and 63); however, much of it relates to specific geome- tries and constructions and few generalizations may be drawn. Two things which are clear from this experience is the fact that composite structures are very fatigue resis- tant in aerospace applications relative to metal structures, and joints and attachments remain a major design problem, especially in fatigue.
The most substantial body of structural fatigue data exists for helicopter rotor blades(refs. 62 and 63). Jarosch and Stepan (ref. 62) have done fatigue testing of root end and outboard sections of the BO-105 fiberglass/epoxy rotor blade. The blade con- struction, seen in figure 26, consists of a C-spar of unidirectional E-glass-reinforced epoxy wound around a pin fitting at the root end, a skin of woven-glass cloth/epoxy oriented at +45 °, and a foam core. Measured values of the spar and skin elastic modulus in the spanwise direction were 6.0 × 106 and 2.6 x 106 psi, respectively. The root end was fatigue tested, as seen in figure 27, with equal flapping and lagging loads of 1900 + 2600 ft-lb and a steady centrifugal load of 24 000 lb. Fatigue lives varied to 13 × 106 cycles. Failure generally occurred in the unidirectional roving at the root-end pin attachment and was accompanied by considerable heating due to interlaminar friction.
Readily visible delamination and gradual changes in stiffness and damping also occurred.
Outboard specimens were tested in flapping resonance at a fatigue strain level of +0.8_ which is ten times the maximum strain in flight. Typical fatigue lives were more than 106 cycles_ and failure was preceded by obvious visible delaminations and accompanying changes in damping and stiffness.
Fatigue tests of boron/epoxy and glass/epoxy CH-47 rotor blades (ref. 63) have also indicated excellent fatigue resistance although failures occurred in the metal root- end fitting. Similarly, fatigue testing of the boron/epoxy F-111 horizontal tail resulted in failure associated with the attachment of the composite to the titanium root end.
Spectrum testing of components of a boron/epoxy wing box has shown excellent fatigue resistance, as has sonic fatigue testing of a boron/epoxy C-SA slat component. Both of these items have exhibited failure in fittings of composite bonded to metal.
COMPOSITEFATIGUE DESIGNCONSIDERATIONS
A successful design procedure for composite materials in fatigue applications will
not be a simple extrapolation of procedures usedfor metals. Metal parts exhibit cracks
when they begin to fail in fatigue, andthe cracks generally propagatein a predictable
manner to failure. Thus, the metal part may be inspected at specific intervals and
removed from service prior to failure. Composites, on the other hand, do not fail in the
same manner.
The difference betweenfatigue behavior of a composite andthat of a metal structure
is depicted schematically in figure 28. The primary modeof damagein a metal struc-
ture is cracking. Cracks propagatein a relatively well defined manner with respect to
the applied stress, and the critical crack size and rate of crack propagationcan be related
to specimendata through analytical fracture mechanics. In this discussion, the critical
damagesize is defined as that amountof damageat which the composite will be no longer
structurally adequate. In general, the crack initiation time (definedas the time to detec-
able cracking (inspection threshold)) occupiesa large part of the fatigue life of a metal
part (ref. 64). It shouldbe notedthat all structures have some initial damagein the form
of microcracks, surface imperfections, inclusions, and other stress risers andthat much
of the so-called crack initiation time involves propagationof this damageto detectable
size. With composite structures there is no single damagemodewhich dominates.
Matrix cracking, delamination, debonding,voids, fiber fracture, andcomposite cracking
can all occur separately andin combination, andthe predominanceof oneor more is
highly dependent on the laminate orientations and loading conditions. In addition, the
uniquejoints and attachmentsusedfor composite structures often introduce modesof
failure different from those typified by the laminate itself.
The composite damagepropagatesin a less regular manner anddamagemodescan
change. (Seefig. 28.) Present experiencewith composites, althoughlimited, indicates
that the rate of damagepropagation in composites doesnot exhibit the two distinct regions
of initiation and propagation. Although, as mentionedpreviously, the crack initiation
range in metals is actually propagation, there is a significant quantitative difference in
rate. This quantitative difference appears to be less apparentwith composites. This
observation is very subjective andapparently dependent uponthe observer's definition of
initiation. Some investigators have observed matrix crazing and other indications early in their tests but have reported short-time rapid propagation because they define the latterbased upon their experience with metals as crack propagation. Indeed, composite cracking may occupy only a small part of the fatigue lifeat the very end, but we can cer- tainly make use of allthe earlier indicationswhich are prevalent.
34O It is expectedthat composite materials will be more damagetolerant than metals.
Again, this expectation is based uponlimited experience andwill dependupon the lami-
nate orientation (unidirectional composites are subject to splitting) andloading conditions,
but, in general, it can be argued that eachfiber is a separate load path andthat a com-
posite is therefore highly redundant. Our present analytical fracture mechanics tool
must be supplementedfor use with composites before we have a better understandingof
this behavior. Several investigators have indicated that, in general, composites exhibit
goodfracture toughness(refs. 65 to 67) and, unlike metals, increase fracture toughness
with increasing strength. It is thus reasonable to predict the critical damagesize in
composites to be greater than that for metals (fig. 28), althoughthe multiple failure modes
make this value a bandfor composites. Similarly, the inspection threshold is depicted
as a band in figure 28 becausethere are multiple failure modes andmultiple inspection
methods.
The problem then is to determine the critical mode or modesof failure anddevelop
detection schemesin order to insure fail safety in critical components. One such pro-
cedure involves the determination of changesin the static or dynamic stiffness properties
of the component. A changein the resonant frequency or dampingbehavior of a part is
an indication of damage. Failure criteria can be developedfrom data such as those seen
in figures 10to 12 as substantial changesoccur early enoughin the fatigue life to allow
safe detection andremoval from service. This characteristic may provide excellent fail
safety for rotor blades in that the aeroelastic behavior may degradenoticeably long
before the part has sustained damageof critical size. Other detection schemessuch as
temperature rise measurements (fig. 10), embeddedconductingwires, radiography,
sonics, ultrasonics, holography, infrared inspection, dye penetrant, andvisual inspection
will probably be used separately or in conjunction with dynamic measurements.
As mentionedearlier, a major consideration for developinga valid design method-
ology is a definition of failure. A problem exists however, in that a single failure crite-
rion may be inadequatefor all applications. This situation is seen schematically in fig-
ure 29. The example considers two structures, a spring and a tension cable, and two
candidate materials, a metal and a composite, for each application. The failure criterion for the spring is a specified loss in stiffness, whereas that for the cable is fracture.
Since metal structures exhibit little change in stiffness until cracking is extensive, the metal spring and metal cable have approximately the same life, and a single criterion based on fracture is probably adequate for design. The composite material spring would lose sufficient stiffness to be considered failed at only a fraction of its fracture life, whereas, the tension cable made of the same material and subject to the same loading would have a much greater useful life. In order to provide for such design considerations, it will be desirable to record fatigue data as depicted schematically in figure 30.
CONCLUDING REMARKS
Composite materials appearto offer excellent resistance to fatigue loading and,
as such, will likely find use in dynamic components. A secondfactor which makes com-
posites attractive for these applications is the opportunity for tailoring of the stiffness
in different directions, thus the designer is given the capability of tuning dynamic com-
ponents. As composites find wider application, it will be necessary to provide more
precise definitions of failure andto couplethese definitions with proper damagedetection
schemes. New, sophisticated damagedetection methodswill probably not be necessary;
however, becauseof the multiple damagemodespossible, it will be necessary to utilize
multiple detection schemes. The apparenthigh damagetolerance of composites will
allow somewhatrelaxed inspection requirements andwill provide for improved repair-
ability procedures. At this writing, the cost of high modulus composites is still high and
must be further reduced to allow wider usage.
REFERENCES
1. Forsyth, P. J. E.; George, R. W.; andRyder, D.A.: SomePreliminary Tests on
Aluminum Alloy SheetsReinforced With StrongWires. Appl. Mater. Res., vol. 3, no. 4, Oct. 1964,pp. 223-228.
2. Baker, A. A.; andCratchley, D.: Metallographic Observations on the Behaviour of
Silica Reinforced Aluminum Under Fatigue Loading. Appl. Mater. Res., vol. 3, no. 4, Oct. 1964,pp. 215-222.
3. Morris, A. W. H.; and Steigerwald, E.A.: An Investigation of the Fatigue Behavior
of Tungsten-Reinforced andSteel-Reinforced Silver Composites. Trans. Met. Soc.
AIME, vol. 239, 1967,pp. 730-739.
4. Salkind, M. J.; andGeorge, F.D.: Fatigue and Bondingof A13NiWhisker Reinforced
Aluminum. Rep. G910583.4(Contract No. 0019-68-C-0016), United Aircraft Corp.,
July 31, 1968. (Available from DDC as AD 838871.)
5. Tavernelli, J. F.; andCoffin, L. F., Jr.: A Compilation andInterpretation of Cyclic
Strain Fatigue Tests on Metals. Trans. Amer. Soc.Metals, vol. LI, 1959, pp. 438-453.
6. Manson, S.S.: Behavior of Materials Under Conditions of Thermal Stress. NACA
Rep. 1170,1954. (Supersedes NACA TN 2933.)
7. Sachs,G.; Gerberich, W. W.; Weiss, V.; and Latorre, J.V.: Low-Cycle Fatigue of
Pressure-Vessel Materials. Proc. Amer. Soc. Testing Mater., vol. 60, 1960, pp. 512-529.
8. Boiler, K.H.: Fatigue Tests of Glass-Fabric-Base Laminates Subjectedto Axial
Loading. Rep. 1823,Forest Prod. Lab., U.S. Dep. Agr., May 1952.
9. Stevens,G. H.; and Boiler, K.H.: Effect of Type of Reinforcement on Fatigue Prop-
erties of Plastic Laminates. TR 59-27, U.S. Air Force, May 1959. (Available
from DDC as AD 213835.)
10. Kimball, KennethE.: Supplementto Fatigue Tests of Glass-Fabric-Base Laminates
Subjectedto Axial Loading - Effect of Notches. Rep. No. 1823-C, Forest Prod.
Lab., U.S. Dep. Agr., Oct. 1958.
11. Boiler, KennethH.: Resum_of Fatigue Characteristics of Reinforced Plastic
Laminates Subjectedto Axial Loading. ASD-TDR-63-768, U.S. Air Force, July
1963.
12. Boller, K.H.: Effect of Pre-Cyclic Stresses on Fatigue Life of RP Laminates.
Modern Plastics, vol. 42, no. 8, Apr. 1965,pp. 162, 165-166, 168, 171, 173.
13. Boller, Kenneth H.: Effect of Tensile Mean Stresses on Fatigue Properties of Plastic Laminates Reinforced With Unwoven Glass Fibers. ML-TDR-69-86, U.S. Air Force, Mar. 1964. (Available from DDC as AD 605 412.)
14. Boller, Kenneth H.: Effect of Pre-Cyclic Stresses on Fatigue Life of Plastic Laminates Reinforced With Unwoven Fibers. ML-TDR-64-168, U.S. Air Force, Sept. 1964. (Available from DDC as AD 606 769.)
15. Boller, Kenneth H.: Fatigue Strength of Plastic Laminates Reinforced With Unwoven "S" Glass Fibers. AFML-TR-64-403, U.S. Air Force, Dec. 1964.
16. Boller, Kenneth H.: Fatigue Characteristics of Two New Plastic Laminates Rein- forced With Unwoven "S" Glass Fibers Under Cyclic Axial or Shear Loading.
AFML-TR-66-54, U.S. Air Force, Mar. 1966.
17. Boller, Kenneth H.: Effect of Notches on Fatigue Strength of Composite Materials.
AFML-TR-69-6, U.S. Air Force, Apr. 1969. (Available from DDC as AD 853 045.)
18. Boller, K.H.: Fatigue Fundamentals for Composite Materials. Composite Materials: Testing and Design, Spec. Tech. Publ. No. 460, Amer. Soc. Testing Mater., c.1969, pp. 217-235.
19. Broutman, Lawrence J.: Fiber-Reinforced Plastics. Modern Composite Materials, Lawrence J. Broutman and Richard H. Krock, eds., Addison-Wesley Pub. Co., Inc., c.1967, pp. 337-411.
20. Cornish, R. H.; Nelson, H. R.; and Dally, J. W.: Compressive Fatigue and Stress Rupture Performance of Fiber Reinforced Plastics. Proceedings 19th Annual Technical and Management Conference, sec. 9-E, Soc. Plast. Ind., Inc., Feb. 1964.
21. Mettes, David G.; and Lockwood, Paul A.: The Mechanical Properties of Laminates Reinforced With High Performance Glass Fiber Fabric. Proceedings 21st Annual Technical and Management Conference, sec. 4-G, Soc. Plast. Ind., Inc., Feb. 1966.
22. Davis, J. W.; McCarthy, J. A.; and Schurb, J.N.: The Fatigue Resistance of Rein- forced Plastics. Mater. Des. Eng., vol. 60, no. 7, Dec. 1964, pp. 87-91.
23. Cutler, Martin B.; and Pinckney, Robert L.: Static and Fatigue Test Properties for Woven and Nonwoven S-Glass Fibers. USAAVLABS Tech. Rep. 69-9, U.S. Army, Apr. 1969. (Available from DDC as AD 688 971.)
24. Kelly, A.; and Davies, G.J.: The Principles of the Fibre Reinforcement of Metals.
Metallurgical Rev., vol. 10, no. 37, 1965, pp. 1-77.
25. Broutman, L.J.: Failure Mechanisms for Filament Reinforced Plastics. Modern Plastics, vol. 42, no. 8, Apr. 1965, pp. 143-145, 148, 150, 153, 214, 216.
26. Smith, T. R.; and Owen, M.J.: Fatigue Properties of RP-1. Modern Plastics, vol. 46, no. 4, Apr. 1969, pp. 124-125, 128-132.
27. Cessna, L. C.; Levens, J. A.; and Thomson, J.B.: Flexural Fatigue of Glass- Reinforced Thermoplastics. Proceedings 24th Annual Technical and Management Conference, sec. l-C, Soc. Plast. Ind., Inc., Feb. 1969.
28. Broutman, L. J.; and Sahu, S.: Progressive Damage of a Glass Reinforced Plastic During Fatigue. Proceedings 24th Annual Technical and Management Conference, sec. ll-D, Soc. Plast. Ind., Inc., Feb. 1969.
29. Lavengood, R. E.; and Anderson, R.M.: Matrix Properties Controlling Torsional Fatigue Life of Fiber Reinforced Composites. Proceedings 24th Annual Technical and Management Conference, sec. ll-E, Soc. Plast. Ind., Inc., Feb. 1969.
30. Fujii, Taichi; and Mizukawa, Kiyoshi: The Effect of the Combination of Roving Glass Cloth and Mat Upon the Fatigue Strength of Reinforced Polyester Laminates.
Proceedings 24th Annual Technical and Management Conference, sec. 14-D, Soc.
Plast. Ind., Inc., Feb. 1969.
31. Hagerup, E.: Flexural Fatigue Testing of Polyesters. J. Appl. Polymer Sci., vol. 7, no. 3, May 1963, pp. 1093-1116.
32. Lazan, B.J.: Some Mechanical Properties of Plastics and Metals Under Sustained Vibrations. Trans. ASME, vol. 65, no. 2, Feb. 1943, pp. 87-104.
33. Lazan, B. J.; and Yorgiadis, A.: Behavior of Plastics Under Repeated Stress.
Symposium on Plastics, Spec. Tech. Publ. No. 59, Amer. Soc. Testing Mater., Feb. 1944, pp. 66-94.
34. Boller, Kenneth H.: Fatigue Characteristics of RP Laminates Subjected to Axial Loading. Modern Plastics, vol. 41, no. 10, June 1964, pp. 145-150, 188.
35. Thompson, A.W.: Fatigue and Creep Properties of Reinforced Plastics. Plastics Inst.-Trans., vol. 30, no. 85, Feb. 1962, pp. 39-50.
36. Owen, M.J.: New Fatigue Testing Machine for Reinforced Plastics. Plastics Inst.-Trans., vol. 35, no. 115, Feb. 1967, pp. 353-357.
37. Dally, J. W.; and Broutman, L.J.: Frequency Effects on the Fatigue of Glass Rein- forced Plastics. J. Compos. Mater., vol. 1, no. 4, Oct. 1967, pp. 424-442.
38. Shockey, P. D.; Anderson, J. D.; and Hofer, K.E.: Structural Airframe Application of Advanced Composite Materials. Vol. Y - Mechanical Properties - Fatigue.
AFML-TR-69-101, Vol. V, U.S. Air Force, Mar. 1970. (Available from DDC as AD 867 017.)
39. Hayes, R.D.: Flightworthy Graphite Fiber Reinforced CompositeAircraft Primary
Structural Assemblies. Vols. 1 and2. FML-TR-70-207-Vols. 1 & 2, U.S. Air
Force, 1970. (Available from DDCas AD-877 234andAD-877 235.)
40. Holmes, R. D.; and Wright, D.W.: Creep and Fatigue Characteristics of
Graphite/Epoxy Composites. ASMEPaper 70-DE-32, Amer. Soc.Mech. Eng., May 1970.
41. Owen,M. J.; andMorris, S." An Assessmentof the Potential of Carbon Fibre Rein-
forced Plastics as Fatigue Resistant Materials. Proceedings 25th Annual Technical
and ManagementConference,sec. 8-E, Soc.Plast. Ind., Inc., Feb. 1970.
42. Puppo,A. H.; andEvensen,H. A.: Interlaminar Shearin Laminated Composites
Under Generalized Plane Stress. J. Compos.Mater., vol. 4, Apr. 1970, pp. 204-223.
43. Pipes, R. Byron; and Pagano, N.J.: Interlaminar Stresses in Composite Laminates Under Uniform Axial Extension. J. Compos. Mater., vol. 4, Oct. 1970, pp. 538-548.
44. Baker, A. A.; and Cratchley, D.: Stress-Strain Behaviour and Toughness of a Fibre- Reinforced Metal. Appl. Mater. Res., vol. 5, no. 2, Apr. 1966, pp. 92-103.
45. Baker, A.A.: The Effect of Fibre Volume Fraction and InterracialBond on the Fatigue of Aluminum Reinforced With Stainless Steel Wires. Appl. Mater. Res., vol. 5, no. 3, July 1966, pp. 143-153.
46. Baker, A.A.: The Effect of Fibre Diameter and Discontinuous Fibres on the Fatigue of a Fibre-Reinforced Metal. Appl. Mater. Res., vol. 5, no. 4, Oct. 1966, pp. 210-217.
47. Baker, A. A.; Mason, J. E.; and Cratchley, D.: High-Strain Fatigue Studies of a Composite Material. J. Mater. Sci.,vol. i, no. 3, Aug. 1966, pp. 229-237.
48. Baker, A.A.: The Fatigue of Fibre-Reinforced Aluminum. J. Mater. Sci.,vol. 3, no. 4, July 1968, pp. 412-423.
49. Salkind, M. J.; George, F. D.; Lemkey, F. D.; and Bayles, B.J.: Investigationof the Creep, Fatigue, and Transverse Properties of Al3Ni Whisker and CuA12 Platelet Reinforced Aluminum. E910344-4 (Contract NOw-65-0384d), United Aircraft Corp., May Ii, 1966. (Available from DDC as AD 633 241.)
50. Salkind, M. J.; Lemkey, F. D.; and George, F.D.: Whisker Composites by Eutectic Solidification. Whisker Technology, Albert P. Levitt, ed., Wiley-lnterscience, c.1970, pp. 343-401.
51. Thompson,E. R.; George, F. D.; andKraft, E.H.: Investigation To Developa High
Strength Eutectic Alloy With Controlled Microstructure. Rep.J910868-4 (Contract
No. 0019-70-C-0052),United Aircraft Corp., July 31, 1970. (Available from DDC
as AD 873832.)
The Fatigue Characteristics of Unidirectionally
52. Hoover, W. R.; and Hertzberg, R. W.: Trans. Amer. Soc.Metals, vol. LXI, 1968, Solidified A1-A!3Ni Eutectic Alloy.
pp. 769-776.
53. Manson,S. Stanford; andHirschberg, Marvin H.: Fatigue Behavior in Strain Cycling
in the Low- andIntermediate-Cycle Range. Fatigue - An Interdisciplinary
Approach, JohnJ. Burke, Norman L. Reed, and Volker Weiss, eds., Syracuse
Univ. Press, 1964, pp. 133-178.
54. Coffin, L. F.: A Study of the Effects of Cyclic Thermal Stresses on a Ductile Metal.
Trans. ASME, vol. 76, no. 6, Aug. 1954, pp. 931-950.
55. Forest, J. D.; and Christian, J. L.: Development and Application of Aluminum- Boron Composite Material. AIAA Paper No. 68-975, Oct. 1968.
56. Young, J. R.; and Carlson, R.G.: Advanced Composite Material Structural Hardware Development & Testing Program. AFML-TR-70-140-Vol. I, U.S. Air Force, July 1970. (Available from DDC as AD 872 264.)
57. Shimizu, H.; and Dolowy, J. F., Jr.: Fatigue Testing and Thermal-Mechanical Treatment Effects on Aluminum-Boron Composites. Composite Materials: Testing and Design, Spec. Tech. Publ. No. 460, Amer. Soc. Testing Mater., c.1969, pp. 192-202.
58. Kreider, K.G.: Mechanical Testing of Metal Matrix Composites. Composite Materials: Testing and Design, Spec. Tech. Publ. No. 460, Amer. Soc. Testing Mater., c.1969, pp. 203-214.
59. Toth, I.J.: Creep and Fatigue Behavior of Unidirectional and Cross-Plied Composites. Composite Materials: Testing and Design, Spec. Tech. Publ.
No. 460, Amer. Soc. Testing Mater., c.1969, pp. 236-253.
60. Salkind, M.; and Patarini, V.: Fatigue of Boron Filament. Trans. Met. Soc. AIME, vol. 239, 1967, pp. 1268-1270.
61. Gates, R. G.; and Wood, W.A.: Evaluating Potential Fatigue Performance of Composites (Cu/W and Cu/Mo) From Microstructural Behavior. Metal Matrix Composites, Spec. Tech. Publ. No. 438, Amer. Soc. Testing Mater., c.1968, pp. 218-228.
62. Jarosch, E.; and Stepan,A.: Fatigue Properties and Test Procedures of Glass
Reinforced Plastic Rotor Blades. J. Amer. Helicopter Soc., vol. 15, no. 1, Jan.
1970,pp. 33-41.
63. Stratton, Warren: The Potential of AdvancedCompositesfor V/STOL Propellers.
Proceedings of the V/STOL Technologyand Planning Conference, U.S. Air Force, Sept.23-25, 1969.
64. Manson,S. S.: Fatigue: A Complex Subject - SomeSimple Approximations. Exp.
Mech., vol. 5, no. 7, July 1965,pp. 193-226.
65. Salkind, M. J.; andGeorge, F.D.: The Charpy Impact Behavior of A13NiWhisker-
Reinforced Aluminum. Trans. Met. Soc.AIME, vol. 242, 1968,pp. 1237-1247.
66. Cooper, G. A.; and Kelly, A.: Tensile Properties of Fibre-Reinforced Metals: Fracture Mechanics. J. Mech. Phys. Solids, vol. 15, no. 4, July 1967,pp. 279-297.
67. Tetelman, A.S.: Fracture Processes in Fiber Composite Materials. Composite
Materials: Testing and Design, Spec. Tech. Publ. No. 460, Amer. Soc. Testing Mater., c.1969, pp. 473-502.
I I I I I i I Phenolic Heot-tesisk_nt elpoxide--._ \ IPolyeseer --,--'_ _ I I I I I I I
_o 8
,o ,o 2 lo3 ,0 4 ,o s Io 6 ,o 7
Cycles to failure Figure 1.- Effect of matrix material on fatigue of glass fabric composites (from ref. 18).
- 100
I I I I
I I I
"13 .b,- em
8O
26%
20%
32%
_o .6.-0
and
_0 "40
37%
u_ O
" 20
t_
1 1 1 1 ! i
< 0
lo lo? lo3 lo4 los lo6 _0 7
Number of cycles to failure Figure 2.- Effect of matrix content on fatigue of +5o glass-fiber-reinforced epoxy (from ref. 13).
lO 10 3 10 5 10 7 Cycles to Failure Figure 3.- Effect of fiber orientation (from ref. 18).
60 - Ill @ lb.
45 - _'_ %,.. Notched - "_- _._ c 4" cl L.
o 30 - < E D E X "" "_'_c h e d "_ O ._.,.
1 1 J l I lo _ ]04 ]0 5 ]06 lo z Cycles Io Failure Figure 4.- Fatigue of woven and nonwoven materials (from ref. 18).
1OO /"S" GLASS & _ 25.000 PSI _ _. MEANSTRESS 8O 'S' 6LASS'_.
¢L ZERO O _ D oo 7o; MEAN STRESS "_ &_.
.
,ir-- 6O "E" 6 L A SZ_'_'_'--...
25,000 PSI "_._. rt_ MEAN STRESS "_",_. n_ lu 50 1- 4O ZERO "_'_. O MEAN STRESS "_"_._ 30- 20- 10- I I I I I I , , O I 1 10 10 2 10 3 10 4 10 S 10 6 10 7 10 NUMBER OF CYCLES TO FAILURE Figure 5.- Fatigue of E- and S-glass composites (from ref. 15).
1 ! 1 1 ! 1 1 120 < _£th e) o u) E -I 25ks, E 2O 10 10 3 10 $ Cycles to Failure Figure 6.- Effect of mean stress level (from ref. 18).
8 i l I i i i i i i i
7 zj
w x"//
_2_- u_V/_
, I s_"'_ R
=_ 0 _" _, _ , l , ,. , l , i
0 2 4 6 8 10 t2 14 iG i8 20 22 Stress (]_ onset of debonding,lO 3 p s L Figure 7.- Composite modulus as a function of stress at onset of debonding (from ref. 26).
2O 1 I 1 1 1 [ l Cracking ,,o
o
_ E Oebooding __..
I.
1 l _ 1 1 1 .. 1 ,
0 i
10 10 z 103 104 t05 106 O1 t Cycles Figure 8.- Fatigue behavior of glass-mat-reinforced polyester [from ref. 26).
I I I I I I I
...., 12
•-' 10
o 8
c6
=" 6
E
4-
O ¢¢}
2-
I ,,I l I ! I _?"_
0-
1 10 102 103 104 105 106 10 7
0.1
Number of lood
repetitions
Figure g.- Effect of mean stress on debonding of glass-mat-reinforced polyester (from ref. 26).
7O
.-
10 50 _] r_
:_ e 4o,__'
X
6 30
" 4 20 _ _' 2 lo I00 I000 I0, O00 I00,000 IO 6 IO CYCLES ENDURED Figure ]0.- Apparent stress and temperature rise in glass-reinforced polypropylene Ifrom ref. 27).
,-/
i | .i Z LI.., r'1 SOOC p,a |
lOO
o ,®c ,,, ] I"I t'_ 0 0 AI_)C P_ _ |IItTTL| I--" _ 13) 0 0 0 0 ,oloc _'_, l;*,Lu.!
..,11 Z .. (30 m
o
!Or O t_INL _. NOt[ • - • _c, ,-'1 It .k L 4O
o
I,I.
o
. o,_oc , ,,J "i t&) #$OTM_I_AL FAILUffF _/_1_10 , I I I ,_ i 0 2 103 104 105 10 6 CYCLES ENDURED Figure 11.- Stiffness decayduring fatigue of glass-reinforced polypropylene (from ref. 271, E El 10 pERPENDICULAR PLIES Z< n, wE)
>
l--U- <0 PARALLEL PLIES _J ILl n, -E) i I I !
m PRIMARY i MODULUS 11.
k- _.5 ,o Ir 4.0 J :3.5 D MODULUS I I I I I I I 3.0 I0C)_ _- °-ult = 124/:)00 PSI =, O Iii re 2O
I I I
o I I -I I
5 6 7 0 1 2 3 4 CYCLES,IO Figure 12.- Changes in strength, modulus, and cracking of 6°/_0° glass-reinforced epoxy (from ref. 28).
0.20 0.15 - 47,500 psi & S-42,500 p=i/'_>_,_ 0.I0 _ Z IAi o_ 0.05 _ (r u bJ ¢3 0.00 I ! I ! I 1.2 1.4 1.6 1.8 2.0 2.2 CRACK PITCH TO PLY THICKNESS, pt Figure 13.- Change in modulus with density of cracking for 0o/90 o glass-reinforced epoxy.
150 200 Figure ]4.- Comparison of fatigue behavior of unidirectional boron, graphite, and glass-reinforced polymers.
3O R 01 v" W _ IU a, I- w TUBULAR _) x 14 m| STR A I __:_ H T_ S i D E D--_----_------_,__ ,L,_ .,5 ,I o I I I I 4 $ 10 10 10 106 10 _ 101 CYCLES TO FRACTURE Figure 15.- Effect of specimen configuration on axial fatigue behavior of _5 ° IO0_ E-glass/epoxy.
R=O_I 36- "'X" TYPE (/)'_ ILl IE I-- 2C- (/) TUBULAR 16- :E < STRAI6HT SIDED I I I I 10 ) 10 `= 10 s 106 10 10 a CYCLES TO FRACTURE Figure 16.- Effect of specimen configuration on axial fatigue behavior of :f.45 ° boron/epoxy.
35'7
Z 4.0
X_ 3.0
_ 2.O
J
_ 1.0
<
1 2 3 4 5 6
CYCLES XlO 5
Figure 17.- Effect of steel wire mesh on the crack growth behavior of aluminum alloy sheet _from ref. l).
LOADING P-t- P
o 16
_ 8
(/) 104. 105 106 101 108
ENDURANCE
-'_ (CYCLES)
Figure 18.- Fatigue behavior of aluminum alloy with and without 13.5 volume percent steel wire (from ref. 1).
0"_ 0-05 4 _]
j
O-01 0.005 /
Y
O001 2 3 4 56 7 8 9 10 11 Stress 1 : Composite 2: Grey cast iron 3: Scotch ply 1009 4:RR58 AI alloy 5: Silver steel Figure 19.- Vibrational damping capacity (6) as a function of stress for silica-reinforced aluminum compared with conventional engineering materials (from ref. 44).
5O "o & O• - -- Q J
o
a.
SOLIDIFICATION RATE 0 0.3 CM/HR w O O 8 CM/HR" n,, Z CM/HR oi* O" II CM/HR tO FILLED SYMBOLS - _-Ai3N_ OPEN SYMBOLS-- AI- C. &l 2 _ROW IN01CATES SJI_C=MEN 010 NOT FAIL ! I , 1 I tO I00 CYCLES TO FAILURE N ) Figure 20.- Comparison of fatigue behavior of lamellar /AI-CuAI2) and fiber (AI-AI3Ni) composites (from ref. 49).
10o Z n,, TIC _" _1_ ELASTIC Z n_ J_E f:L _STtC I I I l l I I J 10-3 10- I 100 101 102 103 104 105 106 107 CYCLES TO FAILURE Figure 21.- Schematic behavior of metals in strain cycling (from ref. 57).
2.0 MEAN STRAIN 0 I.S _SLOPE -0.111 uu MEA_I Sr RAIN 0.4,10 -2 $1.O _'E -0 22 0.9 l 0.8 -- SLOPI_ -0 _ 0.7 0.6 0.5 _2 10 _ CYCLES TO ICA_LURE Figure 22.- Low-cycle axial fatigue of AI3Ni whisker-reinforced aluminum (from ref. 4).
A AJEm(_IS DEP;OTE SPECPM[N DR) MOT FAll Jt-Atlm, - ] c_ m qm S -- _ATm_Xr o.-- o.--- f Q L IO s 10 7 CYCLES TO FAILURE Figure 23.- Flexural fatigue behavior of AI-AI3Ni in air (from ref. 4).
O _pl r.,GI, I al,, IE_ZlPIC, roOTED _ S- -- AIIIIOIS DE'NOT[ _ECIMEt_ DCD NOT IrA[L I I I 10 6 107 10 o CYCLES 10 FAILURE Figure 24.- Effect of environment on the flexural fatigue behavior of AI-AI3Ni (from ref. 4).
loo % AI-B OAT_A'_ ,,ol 1 -'°:- _,_ .... ! ?'.
k t
o l "N i \\
20 40 60 80 lOO MEAN STRESS, KSI Figure 25.- Fatigue behavior of 40 volume percent boron-reinforced 606] aluminum, compared with unreinforced aluminum alloys (from ref. 57).
GRE(Roving ) Foam Core G RE -Skin Nose Cover Endurance-Test specimen cut out of production Blade with Erosion and Roving De-icing protection 0 39xO59x2 40" 024 x0.59" ] 032x 0 59"_1 = 70" 0 47_ OZ._' j F or Bendlnq Tests For Shear Tests Figure 26.- Glass/epoxy BO-I05 rotor blade (from ref.62).
I I Full-Size Blade Root Sample I I !
!
a e :1 d Root Attachment ibr ted Measuring Flange Mflap : Mlag ---1900-2600 ft-lbs Tension (C.E) :- 12 t Figure27.- Root-end fatiguetest specimen of B0-105rotor blade(from ref. 62).
FRACTURE DAMAGE SIZE METALS CRACK LENGTH COMPOSITES BROKEN FIBERS _LAMINAIION MATRIX CRACKING COMPOSITE FRACTURE CRACKING DEBONDS CRITICAL VOIDS DAMAGE SIZE
\
INSPECTION I IffR _- F_cTLONS METAt )N __INITIATION.
FATIGUE CYCLES OR TIME Figure28.- Comparison of fatigue behaviorin metalsandcomposites.
. . ii
MPOSITE
i!I t \
t!l t \
iJl I \
S]IFFNESS
le '
II I t
!if!
METAL _COMPOSITE C OMP_OSIT E J MFTAL CABLE LIFE--JFRMIAEcTTAuLRIE SPRING LIFE CABLE LIFFcoMPOSIT E FRACTURE LOG CYCLES Figure 29.- Schematic fatigue behavior comparison of spring and tension cable made of metal or composite.
_oo
. 7 _0½ _ O STRESS o_ o,-,,,_% \-%, (OR STRAIN) I I I I 1 4 b 0 ? 0 a 10 10 10 s 10 1 1 LO0 CYCLES Figure 30.- Proposed method for reporting composite fatigue data.