You are on page 1of 39

STATE-OF-THE-ART REPORT on

FATIGUE AND FRACTURE MECHANISMS IN COMPOSITE MATERIALS

PREPARED FOR PERMANENT SCIENTIFIC COMMITTEE 58 FOR STRUCTURAL & CONSTRUCTION ENGINEERING

BY HOSSAM EL-DIN MOHAMMED SALLAM, Ph.D.

November, 1998

FATIGUE AND FRACTURE MECHANISMS IN COMPOSITE MATERIALS

BY HOSSAM EL-DIN MOHAMMED SALLAM, Ph.D.

CONTENTS
Glossary of Terms 1. Background and Brief Overview 1.1 Suggested Readings 2. Fracture and Fatigue Behavior of Fiber-Reinforced Composites 2.1 Introduction 2.2 Fiber Bridging Models 2.3 Effect of The Specimen Geometry 2.4 Laminated Plates 3. Fracture and Fatigue Behavior of Short Fiber-Reinforced Composites 3.1 Fracture Behavior of Fiber Reinforced Concrete 3.2 Fatigue Behavior of Fiber Reinforced Concrete 4. Fracture and Fatigue Behavior of Particulate Reinforced Composites 5. Bonded Composite Patch Repair 6. Summary 7. References iii 1 4 6 6 7 14 14 17 17 22 24 27 28 31

Glossary of Terms
angle-ply laminate. A laminate having fibers of adjacent plies oriented at alternating angles. cohesive failure. Failure of an adhesive joint occurring primarily in an adhesive layer. cross-ply laminate. A laminate with plies usually oriented at 0o and 90o only. crack driving force. In fracture mechanics terms, the extension of a crack is driven by the presence of a crack driving force and opposed by the resistance of the microstructure. Crack driving force is generally defined by some characterizing parameter, such as the stress intensity factor, K, or path independent integral, J-integral, which describes the dominant stress and deformation fields in the vicinity of the crack tip. crack tip shielding phenomena. They are means that, the effective crackdriving force actually experienced at the crack tip is locally reduced. damage tolerance. A design measure of crack growth rate. Crack in damage tolerance designed structures are not permitted to grow to a critical size during expected service life. delamination. Separation of the layers of material in a laminate, either local or covering a wide area. fatigue. The failure or decay of mechanical properties after repeated applications of stress. Fatigue tests give information on the ability of a material to resist the development of cracks, which eventually bring about failure as a result of a large number of cycles. da/dN-K curve. A plot of fatigue crack growth rate (da/dN) against the stress intensity factor range K in fatigue testing. Starting with a mechanically sharpened crack, cyclic loads are applied and the resulting change in crack length is recorded as a function of load cycles to obtain da/dN and is plotted against the instantaneous K, K 1.12 (a)0.5, is the applied stress range and a is the instantaneous crack length. S-N curve. A plot of stress amplitude (S) against the number of cycles to failure (N) in fatigue testing. fatigue crack closure. The faces of fatigue crack could make contact even though the component remained in tension due to plastic deformation existing in the wake of the crack, i.e. plasticity-induced crack closure, roughness of the

crack surfaces, roughness-induced crack closure, or corrosion debris, i.e. corrosion-induced crack closure. fatigue threshold. The minimum value of stress intensity factor range below which a fatigue crack does not grow. This value is denoted by the threshold stress intensity factor range, Kth. fiber. It is a general term for a filament with a finite length that is at least 100 times its diameter, which is typically 0.10 to 0.13 mm. A whisker, on the other hand, is a short single-crystal fiber or filament made from a wide variety of materials, with diameters ranging from 1 to 25 m and aspect ratio between 100 and 15000. fracture toughness. A measure of the damage tolerance of a material containing initial flaws or cracks. For mode I deformation and for small crack-tip plastic deformation (plane-strain condition), the critical-stress-intensity factor for fracture instability is designated KIC. KIC represents the inherent ability of a material to withstand a given stress-field intensity at the tip of a crack and to resist progressive tensile crack extension under plane-strain conditions. Thus, KIC represents the fracture toughness of the material and has units of MPa m0.5. glass transition temperature (Tg). The approximate midpoint of the temperature range over which the glass transition takes place; glass and silica fiber exhibit a phase change at approximately 955 oC and carbon/graphite fibers at 2205 to 2760 oC. J-integral. A parameter defines the fracture conditions in a component experiencing both elastic and plastic deformation. Path-independent integral which is an average measure stress-strain field ahead of a crack. For elastic plane-strain conditions, JIC = KIC2/E(1-2). lamina. A single ply or layer in a laminate made up of a series of layers (organic composites). A flat or curved surface containing unidirectional fibers or woven fibers embedded in a matrix (metal matrix composite). matrix. The essentially homogeneous resin or polymer material in which the reinforcement system of a composite is imbedded. Both thermoplastic and thermoset resins may be used, as well as metals, ceramics, and glasses. stress intensity factor (SIF, K). Stress intensity factor is the fundamental principle of linear elastic fracture mechanics (LEFM). In essence, SIF serves as a scale factor to define the magnitude of the crack-tip stress field. SIF is related to both the nominal stress, , level in the member and the size of the crack present, a, i.e. K = Y (a)o.5, where Y is the geometry correction factor.

he goal of this review article is to present a sound background regarding the development of the understanding of the fracture processes and fatigue damage in composite materials and show how they act as a foundation for

current models. This presentation shows the importance of matrix, reinforcement, and reinforcement/matrix interface in fatigue and fracture behavior of composite materials. Composites exhibit an unusual combination of brittle and ductile phenomena.

1. Background and Brief Overview Composites have been developed in order to reduce the weight of components in structural applications and to improve mechanical and thermal properties. Composite materials can be defined as a combination of two or more materials (reinforcing elements, fillers, and composite matrix binder), differing in form or composition on a macroscale. A strong interface bond between the reinforcement and the matrix is obviously desirable, so the matrix must be capable of developing a mechanical or chemical bond with the reinforcement. The reinforcement and the matrix materials should also be chemically compatible.

Polymers, ceramics, and metals are all used as matrix materials, depending on the particular requirements. The matrix holds the fibers together in structural unit and protects them from external damage, transfers and distributes the applied loads to the fibers, and in many cases contributes some needed property such as ductility, toughness, or electrical insulation. The structure of polymers consists of long molecules with a backbone of carbon atoms linked by covalent bonds. In noncrystalline or amorphous polymers the molecular chains have an entirely random orientation and are cross-linked occasionally by a few strong covalent bonds and numerous but weaker van der Waals bonds. These weaker bonds break as the temperature reaches a value known as the glass transition temperature, Tg, characteristic for each polymer. Below Tg the polymer behaves as a linear elastic solid. Creep becomes increasingly significant as the temperature increases and, above Tg, the polymer deforms in a viscous manner under load. In crystalline
1

polymers the molecules are oriented along preferred directions, bringing with them optical and mechanical anisotropy. Polymers are described as being either thermosets (e.g., epoxy, polyester, phenolic) or thermoplastics (e.g., polyimide, polysulfone, polyetheretherketone, polyphenylene sulfide).

Ceramics, such as glasses, cement & concrete, and engineering ceramics ( Al2O3, SiC, Si3N4, ZrO2), have a wide range of engineering applications. A simple definition of ceramics for practical purposes is that of solids which possess an ordered arrangement of atoms bonded together by covalent or ionic forces. The bonding in most useful ceramics is usually of a hybrid nature with extremes represented by silicon carbide, which is practically 100% covalent, and by magnesium oxide, which is almost completely ionic. The strong ionic and covalent nature of the bonding in most ceramics leads to a stable crystal structure with a high melting point and high stiffness. Many ceramic materials have very high elastic moduli and strengths, but the advantages these properties bestow are often outweighed by their highly brittle nature, which leads to low and unpredictable failure stress resulting from the presence of flaws. Therefore, in order to improve the strength of a ceramic component, two approaches are followed, in which either extreme care is taken to minimize the presence of flaws or the ceramic is toughened and made more resistant to cracks. Metal matrix composites typically comprise a light metallic alloy matrix, usually based on aluminum, magnesium or titanium alloys. The major reinforcing elements used in composites are glass, carbon/graphite, organic, and ceramic.

With polymer matrix composites, it is almost true to say that the properties of the composite are essentially those of the fibers, with little contribution from the properties of the matrix. Both metal and ceramic materials have properties closer to those of likely reinforcements and this leads to a different choice of properties for which these composite systems are optimized. The fabrication of metal matrix composites usually requires temperatures close to the matrix melting points and this precludes the use of glass and polymer fibers. The driving force for most of the applications of metal matrix composites is the potential for an increase in stiffness
2

over the matrix alloy with little or no increase in density. Such composites fabricated from long continuous ceramic fibers behave in a manner similar to polymer matrix composites with their strengths and stiffnesses predicted. They have different fracture behavior for two major reasons. First, there is generally a much stronger fiber/matrix bond than is found with polymer matrix composites. This is believed to be because these composites are usually fabricated at high temperatures and at these temperatures chemical reactions can occur between matrix and reinforcements, so promoting an increased adhesion. These reactions can sometimes form undesirable brittle intermetallic phases at fiber/matrix interface. Second, when the metal matrix fractures in these materials it will only do so after considerable plastic work, and therefor another term must be considered in calculating the total work of fracture.

The major difficulty with the use of ceramic components in design applications is their low toughness and high variability in strength. Polymer matrices, such as epoxy resins, have a very low fracture toughness, but when combined with brittle fibers an increase of many orders of magnitude of fracture energy occurs. Hence, it is hoped to achieve similar increases in the fracture toughness of ceramic matrices by the incorporation of long ceramic fibers. This has been achieved recently, with carbon and silicon carbide fibers being used to reinforce glass and glass-ceramic matrices. In this case the composite achieves toughnesses many times greater than those found in monolithic ceramics by fiber bridging, frictional delamination and pull-out, much as is found with polymer matrix composites. The reinforcement of ceramics by short fibers or whiskers (short single-crystal fibers of radius < 1 m) has also been investigated, but toughening less efficient than has been found with long fibers.

In recent years, composites have become more popular for critical structural applications. The fatigue life of an engineering component is composed of the progressive growth of an initiated or existing crack as it passes through the short and long crack regimes. One of the problem associated with designing against fatigue is the ability to predict the safe life of structures under practical load.
3

Obviously, there are some difficulties in using the concept of stress intensity factor to correlate the fatigue and fracture behaviors of composites. The main mechanisms controlling these behaviors in composites are 1- Crack trapping in particulate reinforced composites and 2- Crack bridging in fiber reinforced composites. Both mechanisms depend on the relative strength of the matrix and reinforcement and the interfacial shear strength. Composites may be notch sensitive or notch insensitive depending on their interfacial shear strength and the stress-strain field around the crack tip, i.e. specimen geometry. Furthermore, the effect of reinforcement on improving or degrading the fatigue life of composites is not absolutely clear. Thus, understanding the fatigue and fracture behavior of composites presents a new challenge. Composite materials can be divided into classes in various manners. One simple classification scheme is to separate them according to the reinforcement formsparticulate-reinforced, fiber-reinforced, or laminar composites. Fiber-reinforced composites, can be further divided into those containing discontinuous or continuous fibers. This classification is adopted in the present study as follows: Section 2 presents a review of fiber bridging models used in the continuous fiber reinforced composites. The effect of the specimen geometry on the fracture and fatigue behaviors of composites is discussed. Fatigue and fracture behaviors of laminated plates are introduced. In Section 3 the behaviors of short fiber-reinforced composites are introduced. The section focuses on the fracture and fatigue behaviors of fiber reinforced concrete. Section 4 provides the concept of particulate reinforced compositeshow composites can be improved or degraded by the incorporation of reinforcement particles. Section 5 presents a brief overview of the repair of cracks by using composite materials.

1.1 Suggested Readings The fundamental principles of fracture mechanics, fatigue mechanisms, and stress & failure analysis of composites, as well as manufacturing and applications of composites are discussed in the following text- and hand-books

M. W. Hyer (1998), Stress Analysis of Fiber-Reinforced Composite Materials, WCB/McGraw-Hill. ASM Handbook (1996), Volume 19 Fatigue and Fracture, ASM International. R. W. Hertzberg (1996), Deformation and Fracture Mechanics of Engineering Materials, 4th ed., John Wiley & Sons. R. F. Gibson (1994), Principles of Composite Material Mechanics, McGrawHill. S. Suresh, A. Mortensen, and A. Needleman (1993), Fundamentals of MetalMatrix Composites, Butterworth-Heinemann. P. Balaguru and S. P. Shah (1992), Fiber Reinforced Cement Composites, McGraw-Hill. B. Derby, D. A. Hills, and C. Ruis (1992), Materials for Engineering - A Fundamental Design Approach, Longman Scientific & Technical. Engineered Materials Handbook (1988), Volume 1 Composites, ASM International.
L. J. Boutman (1974), Composite Materials 5, Fracture and Fatigue,

Academic Press.

2. Fracture and Fatigue Behavior of Fiber-Reinforced Composites 2.1 Introduction The three fundamental constituents of fracture and fatigue models for unidirectional composites are schematically represented on Fig. 1. First, debonding occurs at fiber/matrix interface, requiring an understanding of interface fracture mechanics in mixed-mode. Second, fibers exert tractions on the crack surfaces, requiring a mechanics of large-scale bridging. Third, fiber fracture may occur, usually at locations away from the matrix crack plane, resulting in pull-out. The dominant dissipation mechanism that allows fibers to enhance the fracture and fatigue resistance is caused by frictional sliding along previously debonded interface. Such dissipation occurs at both intact and failed fibers. However, the extent of zone that provides dissipation is strongly influenced by the fiber failure site relative to the crack plane, which governs the pull-out length. Large pull-out lengths relative to the crack opening also lead to large-scale bridging, wherein the nominal crack growth resistance depends on crack size and specimen geometry.

Fiber Bridging Fiber Pull-out Crack Front Debonding

Matrix Crack Wake Debonding/ Sliding

Figure 1. The various mechanisms that accompany mode I matrix crack propagation in unidirectional composites

The bridging fibers carry part of the applied load and shield the crack tip. Consequently, crack bridging enhances the fracture and fatigue behavior of composites in comparison to their unreinforced matrix material. The mechanics of
6

crack bridging by frictionally constrained fibers in brittle matrix composites under monotonic tensile loading has been established, see e.g. Ref. 1. A fundamental assumption in the analysis is that the driving force for crack extension is the crack tip or the effective stress intensity factor, Keff, as governed by remote stress and the tractions acting in the crack wake. Equating Keff with the composite fracture toughness, which usually scales with the fracture toughness of the matrix itself, gives the stress required for matrix cracking in terms of the component geometry and various constituent properties, as follows:

Keff = Ka + Kb

(1)

where Ka and Kb are stress intensity factor due to applied stress and bridging tractions respectively and Kb < 0.

2.2 Fiber Bridging Models In order to successfully model the influence of fiber bridging on the composite crack growth behavior, the bridging tractions and their effect on the crack tip region stress and strain fields have to be calculated. The bridging tractions, T, can be represented by closure pressure function, c(x), acting in the direction opposite to the applied stress, as shown in Fig. 2. Only by properly formulating the closure function c(x) can the reduction of the crack tip driving force and crack opening displacements be predicted. Telesman et al [2] reviewed the three different methods used at NASA Lewis to account for the influence of fiber bridging on fatigue crack driving force and crack opening displacements. Two of these methods are based on analytical models, i.e. shear lag models and fiber pressure model, while the third is based on an experimental approach that uses direct measurement of crack opening displacements. Both shear lag models and fiber pressure model attempt to predict the fatigue and fracture behavior of the composites by calculating, through the use of very different formulations, the stresses carried by bridging fibers. The analytical models are based on two significantly different closure pressure formulations which reflect the differences in the approach used to model bridging tractions. Once the

closure pressure functions are known, both the crack tip driving force, i.e. Keff, and the crack opening displacements can be obtained numerically either by the use of the weight function method or by using the finite element method. (a) Shear Lag Models The shear lag models have been used frequently for the analysis of stress concentrations in composites, see e.g. Ref. 3. The shear lag models are based on the load transfer from the cracked matrix to the fibers through relative sliding between the fiber and the matrix over a region where the interface shear stresses exceed the strength of the interface. In the original formulation, developed by Marshall et al [4], the closure pressure is determined via a force equilibrium of a concentric cylinder model, as shown in Fig. 3. The fiber/matrix interface is treated as a purely frictional interface with a constant frictional shear stress, , and relative

fiber/matrix sliding occurs over a debond distance, l. Beyond the distance l, no debonding occurs and iso-strain conditions is assumed. In this formulation the closure pressure in bridged region is proportional to the square root of the opening displacement and the interfacial shear stress. The closure pressure is given by

u ( x ) Vf Ef Ec c( x) = 2 R (1Vf ) E m
2

0 .5

(2)

where u(x) = half crack opening displacement, = fiber/matrix interfacial frictional shear stress, Ef, Em, & Ec = elastic modulus of fiber, matrix, and composite respectively, Vf = fiber volume fraction, and R = fiber radius.

McCartney [5] derived the above closure pressure function by performing an energy balance calculation for a bridged fiber as opposed to the force balance approach used by Marshall et al [4]. He showed that the above equation violated energy balance principles. The corrected closure formulation according to his work is as follows

u ( x ) Vf Ef Ec 2 c( x) = 2 2 2 R (1Vf ) E m
2

0 .5

(3)

Closure Pressure, c(x)


Matrix

Fiber

Matrix

ao a

Figure 2. Closure pressure function simulating fiber bridging tractions

T = c(x)/vf Figure 3. Schematic of the concentric cylinde model used to determine the force-displacement relationship for the shear lag analysis

Since c(x) in both equations is a function of the unknown u(x), an iterative scheme is required to solve for these unknown displacements. In order to validate the shear lag models, the value of independently. It is questionable whether a single meaningful value of can be used to accurately describe the frictional load transfer between the matrix and the bridging fibers over the entire region of the crack wake, as is done in the shear lag models. The use of a single value of over the entire crack bridged region by the shear lag models neglects the interfacial wear that occurs during fatigue cycles. The interfaces of the bridging fibers near the mouth of the machined notch have been exposed to considerably more wear cycles than the interfaces of the bridging fibers near the crack tip. It has been shown that the interfacial wear results in a decrease of in the fatigued crack wake region in comparison to the virgin composite [6]. These findings agree with the work done by Marshall et al [7] who have also shown that decreases with an increase in the interfacial wear. The typical methods of estimating interfacial shear stress, , has to be determined

, such as the push-out test, are typically performed on material not subjected to
9

fatigue conditions. Thus the shear stress coefficient obtained from such tests does not take into effect the interfacial wear or other load history effects generated during cyclic loading. Kantzos et al [8] have shown that for individual bridging fibers subjected to cyclic loading the amount of interfacial wear is greatest near the crack faces and progressively decreases, and thus increases, along the debonded interface length, l (see Fig. 3). (b) Fiber Pressure Model The fiber pressure formulation originated from an analysis of fast fracture in steel [9] and was adopted by Ghosn et al [10] to the analysis of composites under cyclic loading conditions. The fiber pressure model considers the fiber bridged specimen as a structure whose members in the crack wake (i.e., bridging fibers) can carry tensile loads created by normal stresses or bending stresses, or both. The force balance is derived through an elementary strength of materials approach to calculate the stresses in a beam subjected to bending or tensile forces, or both. The actual load transfer mechanism through which the bridging members are loaded is unimportant as long as an accurate force balance can be written which fully describes the stresses in these members. After all, in order to model the effect of bridging on the crack driving forces and displacements, it is only necessary to correctly predict the closure pressure function. The closure pressure in the fiber pressure model is assumed to be equal to the stress carried by the fibers in the bridged region averaged out over the total bridged area. The closure pressure c(x) for a single edge notch geometry, see Fig. 2, is given by
6wao(0.5( w ao) ( x ao)) c( x) = w + for ao xa ( w ao) 3 w ao

(4)

where is the applied remote stress, w is the width of the specimen, ao and a are the initial notch length and the total crack length respectively, and x is the distance along the crack measured from the free surface. This formulation is applicable to a composite system with very stiff fibers and overcomes many of the difficulties
10

inherent in the shear lag formulation, such as avoiding the need to determine the shear strength coefficient and does not require an iterative solution since c(x) is not a function of the crack opening displacements.

The size of the fiber bridging region depends on the applied stress and the mechanical properties of the composite constituents and can vary from zero to the entire crack length. Therefore, Nayeb-Hashemi et al [11] modified the fiber pressure model by considering a partial crack bridging zone of length l, i.e. l (a - ao). Crack Driving Forces and Crack opening profiles The closure pressure functions obtained by analytical models can be used to obtain crack driving forces and crack opening displacements of the bridged specimens through the use of the weight function method or finite element method.

The weight function used is based on the Bueckner formulation [12] for the stress intensity factor calculation of a single edge notch specimen with a finite geometry. The homogenized composite stress intensity factor, Kc, for a partially bridged specimen is given by
ao a ( c ( x )) H (a, x) 2 H ( a, x ) dx+ dx ax ax 0 ao

Kc =

(5)

where
H (a, x) =1+ m1 (a x) (a x) 2 + m2 a a2

(6)

and where m1 and m2 are functions of the ratio of the crack length to the width of the specimen given by

a a m1 =0.6147+17.1844 ( w ) 2 + 8.7822 ( w ) 6 a a m2 =0.2502+3.2889 ( w ) 2 +70.0444 ( w ) 6

(7) (8)

11

The Bueckner weight function method can be extended to calculate the crack opening displacements. It is assumed that the isotropic displacement equation can be applied to the composite since the difference between displacements calculated by orthotropic and isotropic formulations is minimal [3]. The crack opening displacements are calculated at any location xo for a crack length of (a - xo). By incrementing the location of xo over the entire crack length, the full crack opening profile is obtained. The bridged displacement at a location xo for an isotropic material is given by

u( x o ) = where

a l 2(1 2 ) H ( l , x o ) P( x ' )) H ( a, x ' ) c ( dx ' )dl Ec xo l x o 0 l x'

(9 )

P( x ' ) = c( x ' )

for 0 < x ' < a o for a o < x ' < a

(10 )

and where Ec and c are the homogenized composite elastic modulus and Poissons ratio respectively in the loading direction.

By a substitution of the appropriate closure function, c(x), for the shear lag and fiber pressure models into Eqs. 5 and 9, the bridged crack driving force and crack opening displacements are obtained. For the fiber pressure model, a direct numerical integration of the equations gives the solution for the opening profile and stress intensity factor. For shear lag model, an iterative scheme is required with a small damping factor to guarantee convergence.

The fiber pressure model is used in the finite element method as a nonuniform pressure applied in the bridged region [10,11]. For the shear lag model, the closure pressure was applied as a nonlinear foundation pressure [10]. Thus the closure pressure for the shear lag model is given by

12

c( x ) = K st ( u) u( x ) K st ( u) = 2(

(11)

V f2 E f Ec
u( x ) R(1 V f ) Em

0.5

(12)

where Kst(u) is a nonlinear foundation constant, Em, Ef, and Ec are the elastic modulus of matrix, fiber, and composite, respectively, R is fiber radius, and Vf is fiber volume fraction. Finally, the composite stress intensity factor is determined from displacement field near the crack tip. The advantage of the finite element method over the Bueckner weight function method is the convergence speed for the shear lag model.

Since the cracking observed in the composite material tested was limited to the matrix only, the crack driving force is assumed to be the effective stress intensity factor in the matrix of the composite, Keff. Assuming a condition of iso-strain between the composite constituents ahead of the crack tip, Marshall et al[4] postulated that the effective crack driving force in the composite is related to the Kc of the homogenized composite (obtained from Eq. 5) by
Em ) Ec

K eff = K c (

(13)

In contrast, the following relationship was obtained by McCartney [5] based on his energy balance formulations
K eff = K c ( Em ) (1 V f ) Ec

(14)

The validity of the previous models to predict the fatigue crack growth behavior in metal matrix composite was examined by several researchers [2, 10-11, 13-14]. They concluded that, there is an agreement between the experimental results and the prediction obtained by the fiber pressure model. In the case of shear lag model, the ability to select a value of that accurately predicts crack opening profile may still not adequate to fully correct the crack driving forces for the influence of bridging.
13

Thus, shear lag model overestimates the fatigue crack driving force, especially with an increase in the applied stress. This may be due to a higher in the crack tip region than the rest of the crack wake. Bakuckas and Johnson [15] concluded that, the dependency of on so many factors, such as the applied stress, wear of the debonded surfaces and crack extension length, poses restrictions on using the fiber bridging models in a damage tolerance methodology.

2.3 Effect of The Specimen Geometry It is worth to note that, cracks grow rapidly in the composites if the interface is either too strong (low fiber bridging) or too weak (cracks grow parallel to the fiber and applied load). Ghosn et al [13] found that, single edge notch specimens had resistance to fatigue crack growth more than center crack panel specimens. On the other hand, Kantzos et al [16] found that fatigue crack grew parallel to the loading and fiber direction in compact tension specimen while it grew normal to the loading direction in single edge notch specimen. These observations can be explained by examining the stress-strain field around the crack tip in each specimen geometry. Sallam and Hashemi [17] found that, at the same stress intensity factor, T-stress or T-strain, i.e. stress or strain parallel to the crack surfaces, at the crack tip is not only a function of a specimen type and thickness but also on the shape and the angle of the notch. Since fatigue behavior in fiber reinforced composites depends on the fiber bridging zone, the extent of the bridging zone is a function of fiber/matrix bond strength and the shear stress and T-stress around the crack tip. Therefore they concluded that, the effect of shear and T-stress on the reinforcement/matrix interface must be taken into consideration to predict the closure pressure, c(x), and consequently the fatigue behavior in composites.

2.4 Laminated Plates Unidirectional composite materials typically have exceptional properties in the direction of the reinforcing fibers, but poor to mediocre properties transverse to the fibers. Thus, with the exception of one-directionally loaded members (for example, truss members), unidirectional composite materials would be expected to poorly
14

compared to conventional materials. In practice, structures made from composites contain a series of layers of unidirectional fibers such that each layer has some predetermined orientation with respect to the defined dimensions of the structure to overcome the material anisotropy. Furthermore, hybrid composites (Sandwich Laminates) showed superior fatigue crack growth resistance as compared with conventional materials. Fiber-reinforced aluminum laminates (FRALL) is a new class of hybrid composites, which consists of alternating layers of thin aluminum sheets bonded by an adhesive impregnated with high strength fiber-epoxy composites, such as aramid fiber-reinforced aluminum laminates (ARALL) and carbon fiber-reinforced aluminum laminates (CARALL). On the other hand, when viewed at fiber-matrix scale, fiber-reinforced composites are seen as a complex structure rather than a basic homogeneous substance. Depending on the dimensional scale of interest, a composite laminate is an even more complex structure. Strength and fracture behavior of laminates depends on a host of geometrical factors, such as fiber diameter and volume fraction, thickness and number of plies, ply angle orientation and sequence, size and content of microvoids, and laminate thickness.

The effect of matrix resin on the mechanisms of delamination fatigue crack growth in unidirectional carbon fiber reinforced plastics laminates under mode I loading at different stress ratios was studied by Hojo et al [18]. They concluded that, Laminates with toughened matrices are most resistant to fatigue crack growth. On the other hand, the ratio of fatigue threshold to the fracture toughness in the toughened laminates decreased in comparison with the brittle laminates. Furthermore, the increase in fracture toughness by matrix toughness is not really translated into the improvement in delamination resistance under fatigue loading. Harris et al [19] found that the addition of transverse (90o) plies exerts little effect on the fatigue response of the main load-bearing 0o plies. However, in composite containing 45o plies, shear cracks will occur in these off-axis plies, the crossing point of a pair of such crack may lead to delamination. Three different methods to determine the effect of crack wake bridging on mode I delamination toughness were analyzed by Jain and Mai [20]. It was shown that both Griffith energy release rate approach and the J-integral approach underestimate the effect of bridging on crack
15

growth resistance when compared to the stress intensity factor approach though the difference may not be large. Moreover, the energy release rate method becomes difficult to use when the extensibility of the reinforcing thread is taken into consideration, as the displacement profile and hence the various energy terms cannot be determined simply. Recently, the strain energy release rate of nonhomogeneous delaminated laminates is derived by Sheinman and Kardomateas [21] based on J-integral. They decomposed the strain energy release rate into mode I (tensile mode) and mode II (in-plane shear mode) based on the assumption of equivalent orthotropic properties through the laminate thickness.

The double cantilever beam (DCB) test specimen, as shown in Fig. 4, has been used for measurement of mode I interlaminar fracture toughness, GIC , of composite laminates since 1960s. Since then there has been much interest in the pursuit of a mode I standard test method and the progress towards standard fracture and fatigue test methods, ISO standard, for mode I delamination toughness testing of fiber reinforced polymer composites were recently reviewed by Blackman et al [22]. A mode I standard for the delamination toughness of composites, GIC, was proposed as a new work item at ISO in 1994. Because the Japanese Industrial standards (JIS), American Society for Testing and Materials (ASTM), and European Structural Integrity Society (ESIS) had all been investigating the test method, effectively three working drafts were prepared, one by each committee, see Ref. 22. The technical differences between these drafts reflected partly the differing experiences of the technical committees and partly the different motivating factors which had driven each committee to pursue a standard.
(a) End-blocks P P (b) Piano Hinges

ao P

2h

ao P

2h

Figure 4. The double cantilever beam test specimen with loading, P, via (a) en blocks and (b) piano hinges. Initial film length is ao, width B and thickness 2h.
16

3. Fracture and Fatigue Behavior of short Fiber-Reinforced Composites Many experts claim, however, that the big future for composites will be in discontinuous fibers, which exhibit essentially isotropic properties and can be shaped, machined, drilled, etc. using conventional fabrication facilities.

The fatigue and fracture properties of short fibers reinforced composites are controlled by fiber bridging and influenced by the volume fraction, size, aspect ratio, and distribution of the reinforcement [23]. Improvements can be gained by use of short fibers or whisker that have aspect ratios of > 10:1, even though cracks are initiated at whisker or fiber ends. On the other hand, Short fibers may enhance or reduce the fracture toughness and fatigue resistance of the matrix based on the strength of the matrix and the interface, see for example Refs. 24 & 25.

3.1 Fracture Behavior of Fiber Reinforced Concrete The application of the concept of fiber-reinforced composites to concrete or mortar is by no means new. In 1910 Harry Porter claimed dramatic increases in the physical properties of concrete by adding cut nails and spikes to the mixes, see Ref. 26. Fibers in general and polypropylene fibers in particular have gained popularity in recent years for use in concrete, mainly to enhance the shrinkage cracking resistance and toughness of plain concrete [27]. It is generally accepted that fracture toughness of fiber reinforced concrete (FRC) cannot be evaluated using linear elastic fracture mechanics without modifications because of a nonlinear zone ahead of the crack tip often termed the fracture process zone. The nonlinearity of the process zone arises from heterogeneity inherent in concrete, i.e. microcracking ahead of the crack tip (zone shielding), and from fiber bridging in FRC. The presence and the important influence of the fracture process zone in concrete has been recognized since the late 1970s [28]. Most of the fracture mechanics models in FRC simulate the bridging effect of fibers with a closing pressure on the crack surface, as mentioned in sec. 2.2. Hilleborg [28,29] extended the fictitious crack model to FRC by proposing that the closing pressure to be a function of fiber
17

length, fiber diameter and interface bond strength. Wecharatana and Shah [30] assumed a parabolic closing pressure for fiber toughening.

The following mechanism is proposed in the case of a matrix reinforced by an identical volume and aspect ratio of large fibers and microfibers. For the volume of fibers normally used for cementitious composites, i.e. large fibers, only a small improvement in tensile strength is observed, as sketched in Fig. 5. This is probably due to the fact that matrix cracking first occurs at the micro level [31]. If fibers far apart, they have no ability to arrest microcracks. However, once the microcracks condense into macrocracks, the large fibers can not arrest propagation of macrocracks and substantially improve the toughness of the composite. If microfibers are used, they can bridge microcracks, since for a given volume these fibers are much closer together. Microfibers can thus significantly enhance the tensile strength of the composite, Fig. 5. However, for the same aspect ratio, microfibers are shorter and therefore may be pulled out after macrocracks are formed, thus providing little improvement in post peak toughness. By combining fibers of varying size into the matrix, improvement in both the peak stress and post peak toughening can be expected.

Microcracks Stress Microfibers Plain matrix

large fibers Strain

Figure 5. Illustration of different sizes fibers on crack bridging

Several approaches to the study of the fracture of cementitious materials have been proposed recently, see e.g. Refs. [32-34]. These approaches can be categorized as either cohesive crack models or effective crack models. In cohesive crack models, the fracture process zone is modeled by applying traction forces across the surfaces
18

of newly formed cracks. A basic requirement of the cohesive crack model is the softening curve, sometimes known as the stress-separation curve, which relates the stresses across the crack surfaces, the cohesive stresses, to corresponding crack openings.

An analytical model based on Bueckner weight function and an iterative procedure to match the experimentally obtained load vs. crack mouth opening displacement curves was developed by Eissa and Baston [32]. The model simulates the behavior of the fracture process zone as well as the length traction free crack. The model also allows the calculation of the crack tip opening displacement and J-integral at any load level. Hamoush et al [33] proposed a fracture model based on the superposition technique in fracture mechanics in conjunction with an existing pullout model to predict the stress intensity factor of FRC. The model assumes that the final slip distance of the fibers equals the final crack opening displacement. In this model, two basic steps are used in the solution procedure. The first step ignores the contribution of the fiber and finds the crack opening displacement at each fiber location. The second step finds the crack opening displacement due to one unit of force at each fiber location. The compatibility condition, including fiber pullout displacement, is employed to find the final pullout force in each fiber. The forces in pulled-out fibers are restricted to the capacity of the fiber.

The effect of the specimen size, fiber volume content, fiber type, and the presence of the notch on the fracture toughness of FRC was studied by first phase of sixuniversity study funded by the Concrete Materials Research Council-American Concrete Institute (CMRC-ACI) and the National Science Foundation (NSF) [3536]. The variables of experimental program is shown in Table 1 . Two different methods of measuring beam midpoint deflection were used in their project. In the first method, Method I, the deflection of the tension face of the beam midpoint in relation to the machine cross-head was measured. This method includes, in addition to the true beam deflection, extraneous deformations such that the elastic and inelastic deformation of the loading fixtures/supports and local deformation of the specimen at its supports. A more accurate method is to measure the beam midpoint
19

deflection in relation to the neutral axis of the beam at its support. They found that, the error in deflection measurements as a percentage of true beam deflection prior to matrix first-crack can be unacceptably large if these extraneous deformations are not excluded from the deflection measurements. However, after first-crack, these extraneous deformations constitute only a small fraction of the overall beam deflection. The main conclusions of that investigation are (i) the ASTM C 1018 toughness indexes (I5, I10, and I30) are observed to be relatively insensitive to fiber type, fiber volume fraction, and specimen size, (ii) toughness as a measure of absolute energy, like TJCI, is capable of distinguishing among composites with different fiber types, different volume fractions, and different specimen sizes, and (iii) the load-crack mouth opening displacement approach to characterizing toughness of FRC appears to offer some promise. Table 1. Variables of experimental program found in Ref. 35*
Specimen size, mm **
Small Large 102 x 102 x 356 152 x 152 x 533 Notched Unnotched Crimped steel 0.5% 1% 50 50 Two types of fibrillated polypropylene 0.1% 0.5% 50

Hooked-end steel Fiber volume fraction 0.5% 1% Fiber length, mm 50 Aspect ratio 100

Specimen type *** Fiber type

All six participating universities conducted all of the tests. Either three or four specimens were tested for each of the 32 series at each participating location. ** Span lengths of the small and large beams are equal to 305 and 457 mm, respectively. *** The notch-to-beam depth ratio is equal to 0.125 for both small and large specimens.

The main objective of their investigation is to compare between the available test standards and guidelines for measuring toughness indexes, such that ASTM C 1018, JCI SF4, JSCE SF4, and ACI 544, see Ref. 35, therefore, a lack of the discussion on the effect of the above variables on the fracture toughness of FRC is observed in their paper. Thus, another analysis for their experimental results will be made in the next paragraph.

20

The stress at first crack increased with decreasing the specimen depth for both notched and unnotched specimens. However, the opposite trend is observed for toughness, TJCI. At the same notch-to-beam depth ratio, equals 0.125, the crack mouth opening displacement at first crack is little affected by the specimen size, while the deflection at first crack increased by increasing the specimen depth. In the case of small beams, stress and TJCI at first crack are not affected by the presence of the notch, notch depth equals 12.75 mm and may be considered as a non damage notch. In contrast, stress and TJCI at first crack of large beams are significantly affected by the presence of the notch, notch depth = 19 mm. The effects of the fiber type and the fiber volume fraction on the fracture behavior of FRC cannot be distinguished due to the difference in the aspect ratio, shape, and the mechanical properties of the fibers and the difference in the amount of increase of the fiber contents in each type, i.e. two times in steel fibers and five times in polypropylene fibers. At the same fiber contents, the steel fibers increase the toughness of plain concrete more than polypropylene fibers and steel-hooked fiber reinforced concrete specimens have the highest toughness. The energy-absorption capacity of the composite increased with increasing fiber volume content for all types of fibers. However, In the case of polypropylene fibers, both the toughness and the deflection at first crack decreased by increasing the fiber volume fraction.

The effects of steel fiber type, i.e. hooked-end, deformed-end, and corrugated fibers, fiber length of hooked-end fibers, cement content, and the presence of silica fume on flexural toughness of FRC were studied by Balaguru et al [37]. They found that, Toughness indexes I5 and I10 computed using the ASTM C 1018 procedure do not provide a good indication of the variations that are present in load-deflection responses. For a given fiber content, toughness indexes are smaller for high-strength concrete (containing 564 kg/m3 of cement and w/c ratio = 0.26) compared to normal strength concrete(containing 335 kg/m3 of cement and w/c ratio = 0.45). Highstrength concrete sustained higher first crack load, but the post-peak drop is steeper. Therefore, it is advisable to use higher fiber volume fractions for high-strength concrete. Typically, the addition of silica fume to high strength concrete makes the material little more brittle. However, they found [37] the differences are not
21

significant. Higher fiber contents result in much higher load-retaining capacity at large deflection. Hooked-end fiber geometry provides better results than other geometries. Since the mechanical anchorages, provided by deformations, are a significant factor in hooked-end fibers, the effect of fiber length is not as pronounced as in concrete reinforced with straight fibers.

Recently, some of the fracture mechanics approaches to the prediction of failure of FRC structures, methods used to rank the toughness of FRC, toughness optimization, and the properties of concrete reinforced with selected synthetic and recycled fibers are reviewed by Wang [38]. He concluded that, methods of linear elastic fracture mechanics and elastic-plastic fracture mechanics are not applicable to laboratory-sized specimens of FRC. Numerical methods may be used to predict the failure of FRC structures based on the materials stress versus crack opening curve. For routine quality control and toughness comparisons, flexure toughness indices and residual strength ratios are useful. However, direct comparisons should only be made among tests under the same conditions. Since the elastic modulus and matrix cracking strength of FRC are generally not strongly affected by the presence of fibers, it is desirable to normalize the indices and ratios with representative quantities, rather than with actual cracking energy and load, in order to reduce data scattering. He found that, carpet waste fibers can effectively improve the shatter resistance, toughness and ductility of concrete.

3.2 Fatigue Behavior of Fiber Reinforced Concrete Hsu [39] categorized fatigue applications as follows. Low-cycle is the term applied to structures exposed to earthquakes and loads less than 1000 cycles of load. The high-cycle category starts with airport pavements and bridges expected to withstand up to 100,000 load cycles, and extends to highway bridges and pavements, railway bridges, and ties subjected to up to 10 million cycles. FRC has been employed in some way in almost all of the applications termed high-cycle fatigue by Hsu [39]. Fatigue behavior in flexure, or even in compression, of FRC is not completely understood in terms of all the influential variables, such as type and
22

configuration of loading, frequency, effect of rest periods, matrix composition, durability of concrete, and, perhaps most important of all, fiber parameters [40].

Johnston and Zemp [40] studied the effect of fiber content, aspect ratio, and fiber type on the fatigue behavior of FRC. They concluded that, the S-N relationships depend primarily on fiber content and aspect ratio. The 100,000 cycle endurance limits are 84 to 89 percent of the first-crack strength under static loading for the better combinations of fiber type and amount characterized by at least 1.0 volume of fibers of aspect ratio of 70 or greater. However, fiber type is secondary in importance, they used four types of steel fibers, i.e. smooth uniform wire, surfacedeformed wire, melt extract, and slit sheet. All are straight and uniform in cross section without hooked or enlarged ends. On the other hand, Ramakrishnan et al [41] found that, Hooked-end fibers provide better resistance to flexural fatigue than other types of fiber. Wei et al [42] used X-ray diffraction to analyze and calculate the orientation of Ca(OH)2 crystal at the interfacial zone and they also used microhardness to determine the thickness of the interfacial layer, forming, vanishing, and restrengthening, and the effect of superimposing-strengthening of interfacial layers, to understand the fatigue damage mechanism of FRC. They found that, the fatigue process is relative to the space of fibers, performance of cement matrix, and forming, strengthening, and vanishing (which means that the orientation index, orientation range, average crystal size, crystal curve range of Ca(OH)2, and the regularity of variation in microhardness in the interfacial zone are the same as that in the cement matrix), and restrengthening of the interface. The key to increase in fatigue resistance for high strength concrete is increase of crack-arresting ability. Resistance to crack arrest comes in two ways: 1) reduction of size and amount of original crack sources; and 2) capacity of inhibiting initiation and extension of crack. Owing to the addition of silica fume and steel fiber, interfacial structure is improved effectively, interfacial weakness is removed, and the effective range of both interface and fiber is extended. Both silica fume and steel fiber reduce the number and size of crack sources from different angles and resist the initiation and extension of crack. By double or multiple effects, they increase various performance indexes of fatigue resisting.
23

4. Fracture and Fatigue Behavior of Particulate Reinforced Composites Spectacular properties obtainable in a continuous-fiber-reinforced composites are not expected for particulate reinforced composites; however, particulate reinforcement can provide reasonable improvements in strength together with the additional advantages of isotropic properties and allowing fabrication. Although particulate reinforcement is less efficient than short fibers and whiskers, significant improvements in specific strength and stiffness, compared with matrix materials, can be still obtained. The low cost of particles is attractive. Recently, Velasco et al [43] studied the effect of particle size on the fracture behavior of polymer matrix composites, i.e. aluminum hydroxide polypropylene. They found that no significant differences between the fracture toughness, KIC, value of unfilled matrix and those of the composites. Nevertheless, the fracture energy, GIC, of the composites was clearly lower than that of the unfilled matrix. Such reduction depends on the particle size, i.e. the composites filled with the finest grade Al(OH)3 showed higher stiffness, tensile yield strength, fracture toughness and fracture energy than the composite filled with coarser particles. On the other hand, ductile particles, i.e. rubber or metallic particles, were used to enhance the toughness of brittle matrix, such as glasses, ceramics, and polymers [44-46]. An increase in the volume

fraction of metallic particles results in an increase of the fracture resistance and the measured fracture toughness level of composites [44]. The observations of crack propagation in epoxy resin containing dispersed rubber particles are summarized schematically in Fig. 6. A notched sample is loaded in tension, Fig. 6.a; at the fracture stress of the brittle matrix, a crack extends, Fig. 6.b, by-passing the rubbery particles without penetrating them (crack bridging); as the crack propagates, Fig. 6.c, the particles bonded to the matrix are stretched between the opening crack and fail when they reach a critical, large extension.

24

(a)

(b)

(c)

Figure 6. Schematic diagram showing the stages of crack propag epoxy resin containing a dispersion of rubber particles, Ref. 46.

In metal matrix composites, ceramic particles, such as SiC and Al2O3, may enhance the resistance of the matrix to fatigue crack initiation and early growth. Fatigue life behavior of metal matrix composites can be improved or degraded by the incorporation of reinforcement particles, depending on many factors such as fabrication processes, particle type, size, volume fraction and distribution, and matrix properties [47]. It is difficult to make generalizations about the effects of these factors on fatigue and fracture behavior of particulate reinforced metal matrix composites [47-54]. Hung et al [47] concluded that, large particle size and, aging defects, and machine-induced defects reduce fracture toughness of composites. Beck et al [48] found that, the fracture toughness of all tested composites was affected by changes in the matrix microstructure produced by aging. The fracture toughness was adversely affected by increases in particle volume fraction. The relation between particle size and fracture toughness is not clear.

Ogarevic and Stephens [49] found that fatigue crack growth rates were higher and Kth values were lower in the composite compared to those of the unreinforced material. Increasing the volume fraction of reinforcement has been observed to increase the fatigue life under stress-controlled conditions in wrought aluminum alloys as well as a magnesium alloy. This can be attributed to the decreases in elastic and plastic strains that result from the increasing modulus and apparent work hardening, both which increase with increasing volume fraction. Experimentation on wrought aluminum alloys clearly indicates that particle size does indeed have a
25

significant influence on fatigue life of metal matrix composites. Near crack initiation, particles are unsuitable obstacles for short crack growth, but brittle fracture of reinforcing component may occur [50]. Kumai et al [51] found that, the fatigue crack avoids SiC particles at low stress intensity factor, K, ranges, but at high K ranges the crack appears to proceed by linking fracture SiC particles ahead of the main crack front. Kumai et al [52] found that, fatigue cracks initiate mainly at matrix-particle interface in molten-metal-processed composite because of degraded interfacial strength. On the other hand, applications relating to the shot peening of metal matrix composites are, until now, rare, and it is not known if this treatment can be successfully applied to these materials in terms of actual improvements to the fatigue limit. Recently, Baragetti and Guagliano [53] found that, the effect of shot peening is more evident if a steep stress gradient (e.g. due to notch) is present. An improvement in the fatigue limit can be over 22%. For smooth specimens, the improvement is less evident even if shot peening shifts the crack initiation point from the surface to an internal defect, thus improving the surface fatigue strength.

Shang and Ritchie [54] concluded that, after allowing for crack closure, the effective threshold stress intensity factor range, Keff,th, is intrinsic thresholds and intrinsic thresholds are solely a function of the effective mean particle size and to be independent of volume fraction. In contrast, to measured low load-ratio Kth thresholds, intrinsic thresholds are found to be somewhat higher in fine particles composites. Reinforcing particles which interact with the crack path are considered to impede near-threshold crack extension in two ways: (i) by promoting crack deflection (in avoiding the particles) and hence enhanced roughness-induced crack closure at low load ratios, and (ii) by crack trapping, i.e. arrest of a crack at a reinforcing particle. Since the process of crack trapping must involve plastic flow at the crack tip, crack extension cannot be completed unless the plastic zone engulfs the particle. Accordingly, a limiting condition for fatigue crack advance, i.e. the intrinsic fatigue-threshold condition, can be represented by the assertion that the maximum plastic-zone size at the crack tip must at least exceed the average particle size. Therefore, crack-tip shielding phenomena such as crack deflection, roughness26

induced crack closure and crack trapping have a direct and significant impact on fatigue behavior. This has provided an insight into otherwise contradictory results related to particle size and volume fraction effects.

Finally, the most recent work in particulate reinforced composites involves systematic investigations of the effects of particle size and volume fraction on the fatigue crack growth behavior using composites and matrices produced by the same fabrication routes. However, in the most of the previous studies, various kinds of composite with different matrix compositions and different fabrication routs have been examined, with the result that it is difficult to compare the reported behavior. This is because of the choice of materials was controlled by availability rather than by the requirements of a planned research program. Thus it has been difficult to clarify the general role of microstructural factors in fatigue in these materials. In some studies, models have been proposed for the effect of particles on fatigue crack growth, but still they are generally poor because of a lack of adequate data. Accordingly there is a need of further intensive research in this area.

5. Bonded Composite Patch Repair The repair of cracks or post-strengthening existing structures by advanced materials have attracted the interest of scientists and engineers, see for example Refs. [55-57]. Based on the world-wide research and development work carbon fiber reinforced plastics strips to rehabilitate structures is already routine for many firms in North America, Europe, and Japan. However, even in the future, fiber composites will not replace traditional construction materials, such as steel, concrete, and wood, but will be used instead to supplement them as needed [56,57]. In metallic aircraft structures, the repair of fatigue cracks has been received much attention in recent years [55].

Although extensive theoretical and experimental research on bonded composite patch repairs has been carried out, and service experience with such repairs has been good, it is evident that further work is required to assess the full potential and
27

limitations of this method of repair, and to develop optimum repair schemes for a wide range of applications. For example, further research is required to establish the effects of impact damage, service temperature and long term exposure to hotwet environments on the efficiency and durability of repairs carried out with various types of patch. Various models have been developed for predicting the efficiency of bonded patches in retarding the growth of fatigue cracks. In general, the efficiency of repairs to thin flat sheet can be predicted accurately using analytical closed form expressions, but for complex or thick section repairs threedimensional analyses are necessary. Unfortunately, a suitable model is not yet available for predicting the development of debonding [55], and therefore measured levels of debonding have to be used in patch efficiency predictions.

6. Summary A brief review of the fatigue and fracture mechanisms in composites has been given. Here, all the probable crack-tip shielding mechanisms in composites are summarized. Sources of shielding are described in terms of mechanisms relying on the production of elasticity constrained zones which envelop the crack (zone shielding), on the generation of wedging, bridging, or sliding forces between the crack surfaces (contact shielding) and on crack deflection and meandering. Under small-scale yielding condition, the crack-driving force in composites can be expressed as

Ktip = KI - Ks

(15)

where Ktip, KI, and Ks are the local near-tip stress intensity, the applied or nominal stress intensity, and stress intensity due to shielding respectively. The objective of extrinsic toughening is thus to enhance Ks. It is possible to categorize mechanisms of extrinsic toughening into several distinct classes, as illustrated schematically in Fig. 7. These classes involve crack tip shielding from: (a) crack deflection and meandering, whereby the mode I crack-driving force is locally reduced by deviations of the crack path from the surface of maximum
28

tensile stress, has been shown to play a significant role in governing the fracture and fatigue behavior in particulate reinforced Composites, (b) inelastic or dilated zones surrounding the wake of the crack, termed zone shielding. Zone shielding mechanisms include (i) microcrack toughening & crack field void formation, which serve to relax crack-tip triaxiality stress and diffuse the intensity of crack-tip stress singularity, and (ii) crack wake plasticity & residual stress field, which develop a favorable compressive residual stress, (c) wedging, bridging and/or sliding between crack surfaces, termed contact shielding. A more general source of contact shielding during cyclic crack growth arises from the wedging action of fracture surface asperities an/or corrosion debris, where the crack tip opening displacements are small. Crack bridging is most prominent in whisker- and fiber-reinforced composites. A prominent characteristic of non-mode I crack growth is interaction between sliding surfaces, i.e. rubbing. This phenomenon has been shown to be a very potent shielding mechanism during fatigue crack growth in shear modes, and (d) Plasticity-induced crack closure and phase-transformed-induced crack closure are considered as a combined zone and contact shielding. Plasticity-induced crack closure is generally considered to be more prevalent under plane stress conditions and is thus more significant at higher stress intensity levels in metal matrix composites. On the other hand, Analogous to transformation toughening in ceramic matrix composites, an additional crack closure mechanism can result in materials which undergo a stress-or strain-induced phase transformation. All the above models based on linear elastic fracture mechanics (LEFM). However, LEFM sometimes conflicts in some details with experimentally observed phenomena of fatigue crack growth rate due to the failure of stress intensity factors to provide an adequate representation of crack stress or strain fields. Thus, any of the above models, i.e. based on LEFM, can not be generalized to explain or predict the fatigue failure in composites. As mentioned by Sallam and Hashemi [17], the correlation of fatigue crack growth behavior in composites by a single parameter, such as effective stress intensity factor or crack tip opening displacement, may not be appropriate. Further work is required to improve the above shielding
29

mechanisms by using two- or three-dimensional nonlinear micro-analysis taking into consideration the strength of matrix, reinforcement, and interface.

(a) Crack deflection and meanderi

(b) Zone shieldin


- microcrack toughenin

(c)
- wedging:
* corrosion-induced crack closur * roughness-induced crack closur

Conta
- bridgin

(d) combined zone and contact shieldin


- plasticity-induced crack closur

Figure 7. Schematic representation of mechanisms o shielding in composite

30

7. REFERENCES 1. Walls, D. P., Bao, G., and Zok, F.W.(1993), Mode I fatigue cracking in a fiber reinforced metal matrix composite, Acta metall. mater., Vol. 41, p 2061. 2. Telesman, J., Ghosn, L.J., and Kantzos, P.( 1993), Methodology for prediction of fiber bridging effects in composites, J. of Composites Technology & Research, Vol. 15, p 234. 3. Nairn, J. A.(1988), Fracture mechanics of unidirectional composites using the shear-lag mode I: Theory, J. of composite materials, Vol. 22, p 561. 4. Marshall, D. B., Cox, B. N., and Evans, A. G.(1985), The mechanics of matrix cracking in brittle-matrix fiber composites, Acta Metallurgica, Vol. 33, p 2013. 5. McCartney, L. N.(1987), Mechanics of matrix cracking in brittle-matrix fiber reinforced composites, Proceedings, Royal Society of London, Vol. A409, p 329. 6. Kantzos, P., Eldridge, J. I., Koss, D., and Ghosn, L. J.(1991), The effect of fatigue on the interfacial friction properties of SCS-6/Ti-15-3 composite, HITEMP review, NASA CP 10082. 7. Marshall, D. B., Shaw, M.C., and Morris, W. L.(1991), Measurement of interfacial properties in intermetallic composites, Titanium Aluminide Composite Workshop, P. R. Smith, S. J. Balsone, and T. Nicholas, Eds., WL-TR91-4020. 8. Kantzos, P., Eldridge, J. I., Koss, D., and Ghosn, L. J.(1992), The effect of fatigue loading on the interfacial shear properties of SCS-6/Ti-Based MMC, Proceedings of MRS spring meeting, San Francisco, CA, USA. 9. Hoagland, R. G., Rosenfield, A. R., and Hahn, G. T.(1972), Mechanisms of fast fracture and arrest in steels, Metallurgical Transactions, Vol. 3, p 123. 10.Ghosn, L.J., Kantzos, P., and Telesman, J.(1992), Modeling of crack bridging in a unidirectional metal matrix composite, Int. J. of fracture, Vol. 54, p 345. 11.Nayeb-Hashemi, H., Yang, Y., and Murphy, R. J.(1996), Mode I fatigue crack growth in continuous alumina fiber-reinforced magnesium matrix composite, Int. J. of Fatigue, Vol. 18, p 287. 12.Bueckner, H. F.(1971), Weight functions for notched bar, Zeitschrift Angewandtle Mathematic Mechematic und Mechanik, Vol. 51, p 97. 13.Ghosn, L. J., Kantzos, P., and Telesman, J.(1994), Fatigue Crack Growth and Crack Bridging in SCS-6/Ti-24-11, ASTM STP 1184, p 64. 14.Covey, S. J., Lerch, B. A., and Jayaraman, N.(1995), Fiber volume fraction effects on fatigue crack growth in SiC/Ti-15-3 composite, Materials Science and Engineering, Vol. A200, p 68. 15.Bakuckas, J. G., and Johnson, W. S.(1996), Thermal residual stresses in the analysis of fiber-bridged matrix crack growth in titanium metal matrix composites, J. of Composites Technology & Research, Vol. 18, p 67. 16.Kantzos, P., Telesman, J., and Ghosn, L.(1991), Fatigue Crack Growth in a Unidirectional SCS-6/Ti-15-3 Composite, ASTM STP 1110, p 711. 17.Sallam, H.E.M. and Nayeb-Hashemi, H.(1996), Effect of specimen geometry on fatigue crack growth behavior of metal matrix composite, PVP-Vol.

31

342/MD-Vol. 72, Recent Advances in Solid/Structures and Application of Metallic Materials, ASME 1996, p 17. 18.Hojo M., Ochiai, S., Gustafson, C., and Tanaka, K.(1994), Effect of matrix resin on delamination fatigue crack growth in CFRP laminates, Eng. Fract. Mech., Vol. 49, p 35. 19.Harris, B., Reiter, H., Adam, T., Dickson, R.F., and Fernando, G.(1990), Fatigue behaviour of carbon fibre reinforced plastics, Composites, Vol. 21, p 232. 20.Jain, L. K. and Mai, Y.-W.(1994), On the equivalence of stress intensity and energy approaches in bridging analysis, Fatigue Fract. Engng Mater. Struct., Vol. 17, p 339. 21.Sheinman, I. and Kardomateas, G. A.(1997), Energy release rate and stress intensity factors for delaminated composite laminates, Int. J. Solids Structures, Vol. 34, p451. 22.Blackman, B. R. K., Brunner, A. J., and Davis, P.(1998), The evaluation of an ISO standard for the mode I delamination toughness of laminates, Proceedings of ECF 12 Conf., Sheffield, UK, 14-18, September 1998, Emas Publishing, Vol. III, p 1369. 23.Harris, S. J.(1994), Fatigue resistance of Fiber-reinforced alloys, in Mechanical Properties of Metallic Composites (edited by Shojiro Ochiai, Marcel Dekker Inc), p 612. 24.Preston, S., Melander, A., Groth, H., and Blom, A. F.(1990), Fatigue crack growth in an Al2O3 short fiber reinforced Al-2Mg alloy, ASTM STP 1080, p 219. 25.Tvergaard, V.(1990), Micromechanical modelling of fibre debonding in a metal reinforced by short fibres, in Inelastic Deformation of Composite Materials, IUTAM Symposium, Springer-Verlag, p 99. 26.Dixon, J. and Mayfield, B.(1971), Concrete reinforced with fibrous wire, Concrete March 1971, p. 73. 27.Alhozaimy, A. M., Soroushian, P., and Mirza, F.(1996), Mechanical properties of polypropylene fiber reinforced concrete and the effects of pozzolanic materials, Cement and Concrete Composites, Vol. 18, p 85. 28.Hilleborg, A., Modeer, M., and Patersson, P. E.(1976), Analysis of crack formation and crack growth in concrete by means fracture mechanics and finite elements, Cement and Concrete Research, Vol. 6, p 773. 29.Hilleborg, A.(1980), Analysis of fracture by means of the fictitious crack model, particularly for fiber reinforced concrete, The Int. J. of Cement Composites, Vol. 2, p 177. 30.Wecharatana, M. and Shah, S. P.(1983), A model for predicting fracture resistance of fiber reinforced concrete, Cement and Concrete Research, Vol. 13, p 819. 31.Betterman, L. R., Ouyang, C., and Shah, S. P.(1995), Fiber-mortar interaction in microfiber-reinforced mortar, Advn. Cem. Bas. Mat., Vol. 2, p 53. 32.Eissa, A.-B. and Batson, G.(1996), Model for predicting the fracture process zone and R-curve for high strength FRC , Cement and Concrete Composites, Vol. 18, p 125.

32

33.Hamoush, S. A., Salami, R., and Abu-Saba, E. G.(1991), Fracture model to predict stress intensity in fiber reinforced concrete, ACI Materials Journal, Vol. 88, p 504. 34.Hamoush, S. A., Terro, M. J., and Ahmad, S. H.(1997), Shear modulus of the interfacial bond between steel fibers and concrete, Magazine of Concrete Research, Vol. 49, p 277. 35.Gopalaratnam, V. S., Shah, S. P., Batson, G. B., Criswell, M. E., Pamakrishnan, V., and Wecharatana, M.(1991), Fracture toughness of fiber reinforced concrete, ACI materials Journal, Vol. 88, p 339. 36.Johnston, C. D.(1992), Discussion on the above Ref., i.e. Ref. No. 35, ACI materials Journal, Vol. 89, p 304. 37.Balaguru, P., Narahari, R., and Patel, M.(1992), Flexural toughness of steel fiber reinforced concrete, ACI materials Journal, Vol. 89, p 541. 38.Wang, Y.(1998), Toughness characteristics of synthetic fiber-reinforced cementitious composites, Fatigue Fract. Engng Mater. Struct., Vol. 21, p 521. 39.Hsu, T.(1981), Fatigue of plain concrete, ACI materials Journal, Vol. 78, p 292. 40.Johnston, C. D. and Zemp, R. W.(1991), Flexural fatigue performance of steel fiber reinforced concrete-Influence of fiber content, aspect ratio, and type, ACI materials Journal, Vol. 88, p 374. 41.Ramakrishnan, V., Wu, G. Y., and Hosalli, G.(1989), Flexural fatigue strength of fiber-reinforced concrete, Endurance limit and impact strength, Transportation Research Record No. 1226, Transportation Research Board, Washington D.C., USA, p 17. 42.Wei, S., Jianming, G., and Yun, Y.(1996), Study of fatigue performance and damage mechanism of steel fiber reinforced concrete, ACI materials Journal, Vol. 93, p 206. 43.Velasco, J. I., Morhain, C., Maspoch, M. L., and Santana, O. O.(1998), Effect of particle size on the fracture behaviour of aluminium hydroxide filled polypropylene, Proceedings of ECF 12 Conf., Sheffield, UK, 14-18, September 1998, Emas Publishing, Vol. III, p 1381. 44.Dlouhy, I., Reinisch, M., Boccaccini, A. R., and Knott, J. F.(1997), Fracture characteristics of borosilicate glasses reinforced by metallic particles, Fatigue Fract. Engng Mater. Struct., Vol. 20, p 1235. 45.Ostlund, S.(1995), Fracture modelling of brittle-matrix composites with spatially dependent crack bridging, Fatigue Fract. Engng Mater. Struct., Vol. 18, p 1213. 46.Kunz-Douglass, S., Beaumont, P., and Ashby, M. (1980), A model for toughness of epoxy-rubber particulate composites, J. of Material Science, Vol. 15, p 1109. 47.Hung, N. P., Zhou, W., Peh, E. T.( 1995), and Chan, C.S. Fracture toughness and low cycle fatigue of 6061/Al2O3p composites, Composites Engng., Vol. 5, p 509. 48.Beck, N. C., Aikin, Jr., R. M., and Briber, R. M.(1994), " Fracture toughness of discontinuously reinforced Al-4Cu-1.5Mg/TiB2 composites ", Metallur. and Mater. Trans., Vol. 25A, p 2461.

33

49.Ogarevic, V. V. and Stephens, R. I.(1994), Fatigue of a particle-reinforced cast aluminum matrix composite at room and elevated temperature, ASTM STP 1184, p 134. 50.Papakyriacou, M., Mayer, H. R., Stanzi-Tschegg, S. E., and Groschl, M.(1996), Fatigue properties of Al2O3-particle-reinforced 6061 aluminium alloy in highcycle regime, Int. J. of Fatigue, Vol. 18, p 475. 51.Kumai, S., King, J. E., and Knott, J. F.(1990), " Short and long fatigue crack growth in a SiC reinforced aluminum alloy", Fatigue fract. Engng. Mater. Struct., Vol. 13, p 511. 52.Kumai, S., King, J., and Knott, J.(1991), " Fatigue in SiC-particulate-reinforced aluminum alloy composites", Material Science and Engineering, Vol. A146, p 317. 53.Baragetti, S. and Guagliano, M.(1998), Influence of low-intensity shot peening on the fatigue strength of an Al/Al2O3 composite material, Fatigue Fract. Engng Mater. Struct., Vol. 21, p 717. 54.Shang, J. K., and Ritchie, R. O.(1989), " On the particle-size dependence of fatigue-crack propagation thresholds in SiC-particulate-reinforced aluminumalloy composites: role of crack closure and crack trapping", Acta metall, Vol. 37, p 2267. 55.Poole, P. and Cook, R.(1998), Current airframe fatigue problems, Proceedings of ECF 12 Conf., Sheffield, UK, 14-18, September 1998, Emas Publ., Vol. I, p 23. 56.The proceedings of The Arabian Conference on Repair and Rehabilitation of Structures, Cairo, 16-19 September 1998, Three Volumes. 57.The proceeding of The First International Civil Engineering Egypt-ChinaCanada Symposium 1997, Evaluation, Repair and Retrofit of Structures Using Advanced Methods and Materials, Cairo, 18-20 December 1997, 281 pp.

34

You might also like