Destructive techniques in the measurement of residual stresses 29 2.8 The ply sectioning method The ply sectioning method is a destructive technique used to study residual stresses in laminated composites (Joh et al.,1993).This method uses Moire interferometry,and measures the deformation due to sectioning and releasing a layer from the constraints imposed by an adjacent layer.The deformation strains are used to calculate the resulting release of stress.This technique overcomes the problems resulting from the cutting method by obtaining a small strip specimen from the edge of the larger specimen,which leads to creation of a plane stress state.Ply sectioning can also be used to study the nature of warping observed in an unbalanced composite laminate.The outside layers of a laminate can be machined away,resulting in an unbalanced,warped structure (Chapman et al., 1990;Manson and Seferis,1992).The resulting warpage is measured and can be used as an input in classical lamination theory(CLT)to calculate the corresponding residual stresses(Myers,2004). In the sectioning method,removing sections of interest is performed in such a way that it can be reasonably assumed that the final stress state is zero.The total change in strain from the original state to the unstressed state corresponds to the negative of the strain in the component prior to testing (Reid,2009).In one variation of this method,strain gage rosettes are used.In this case,the section is cut as close as possible to the edges of the rosette,so that it is secluded from the neighboring parts to make sure that the final stress state is negligible.If the variation of the through-thickness stress is needed,more gages are attached to the surfaces that have been newly exposed.Further cuts can then be made to part sub- sections,each with its own gage.This approach is known as the 'slice-and-dice' method(Reid,2009). Strain gages cannot be used when the stiffness of the removed section is low.Under such circumstances,Moire interferometry can be used to measure the released strains,because this measurement system applies no loading to the removed section.Moire interferometry has consequently been used to measure the residual stresses in individual plies of cross-ply laminates (Gascoigne,1994;Joh et al.,1993)and thick-walled cross-ply cylinders.The sectioning method is not considered suitable for measuring micro-scale residual stresses in uni-directional FRPs.This is because the thickness of each removed section is much larger than the fiber diameters.Therefore,this method fails to achieve a resolution sufficient to give a clear understanding of the stresses at the micro-scale(Reid,2009). The sectioning method has been used extensively to measure residual stresses in metallic components.The residual stress variations across a welded steel H section,(Tebedge et al.,1973)in cold-bent steel plate (Weng and White,1990), welded seams(Kovac,1995)and in filament wound tubes(Casari et al.,2006) have all been measured using this technique.A finite element method can also be used to predict the tri-axial stresses at points away from strain gages.This approach has been simulated in the residual stress analysis of a welded joint(Hill and Nelson 1995,1996). Woodhead Publishing Limited,2014
Destructive techniques in the measurement of residual stresses 29 © Woodhead Publishing Limited, 2014 2.8 The ply sectioning method The ply sectioning method is a destructive technique used to study residual stresses in laminated composites (Joh et al ., 1993). This method uses Moiré interferometry, and measures the deformation due to sectioning and releasing a layer from the constraints imposed by an adjacent layer. The deformation strains are used to calculate the resulting release of stress. This technique overcomes the problems resulting from the cutting method by obtaining a small strip specimen from the edge of the larger specimen, which leads to creation of a plane stress state. Ply sectioning can also be used to study the nature of warping observed in an unbalanced composite laminate. The outside layers of a laminate can be machined away, resulting in an unbalanced, warped structure (Chapman et al ., 1990; Manson and Seferis, 1992). The resulting warpage is measured and can be used as an input in classical lamination theory (CLT) to calculate the corresponding residual stresses (Myers, 2004). In the sectioning method, removing sections of interest is performed in such a way that it can be reasonably assumed that the fi nal stress state is zero. The total change in strain from the original state to the unstressed state corresponds to the negative of the strain in the component prior to testing (Reid, 2009). In one variation of this method, strain gage rosettes are used. In this case, the section is cut as close as possible to the edges of the rosette, so that it is secluded from the neighboring parts to make sure that the fi nal stress state is negligible. If the variation of the through- thickness stress is needed, more gages are attached to the surfaces that have been newly exposed. Further cuts can then be made to part subsections, each with its own gage. This approach is known as the ‘slice- and-dice’ method (Reid, 2009). Strain gages cannot be used when the stiffness of the removed section is low. Under such circumstances, Moiré interferometry can be used to measure the released strains, because this measurement system applies no loading to the removed section. Moiré interferometry has consequently been used to measure the residual stresses in individual plies of cross- ply laminates (Gascoigne, 1994; Joh et al ., 1993) and thick- walled cross- ply cylinders. The sectioning method is not considered suitable for measuring micro- scale residual stresses in uni- directional FRPs. This is because the thickness of each removed section is much larger than the fi ber diameters. Therefore, this method fails to achieve a resolution suffi cient to give a clear understanding of the stresses at the micro- scale (Reid, 2009). The sectioning method has been used extensively to measure residual stresses in metallic components. The residual stress variations across a welded steel H section, (Tebedge et al ., 1973) in cold- bent steel plate (Weng and White, 1990), welded seams (Kovač, 1995) and in fi lament wound tubes (Casari et al ., 2006) have all been measured using this technique. A fi nite element method can also be used to predict the tri- axial stresses at points away from strain gages. This approach has been simulated in the residual stress analysis of a welded joint (Hill and Nelson 1995, 1996)
30 Residual stresses in composite materials Gascoigne (1994)used high-sensitivity Moire interferometry and linearized strain-displacement equations to measure released displacements and residual strains.Joh et al.(1993)developed a novel concept of layer separation to measure quantitatively and precisely the tensile residual stresses in thick plates with layered distribution of residual stresses.Chapman et al.(1990)presented a model to predict the macroscopic in-plane residual stress state of semi-crystalline thermoplastic composite laminates induced by process cooling.Their model predictions were in good agreement with experimental residual stress measurements for uni-directional graphite (AS4)reinforced PEEK laminates. 2.9 The radial cutting method The radial cutting method is a different version of the sectioning method.It has been used extensively to determine the residual stresses in fiber reinforced tubes (Cohen,1997;Ganley et al.,2000:Seif and Short,2002,Seif et al.,2006)and rings (Aleong and Munro,1991;Roy,1991).Based on the assumption that the residual stresses in such structures are invariant with axial and circumferential position,a single axial cut can result in their release.The extent of a laminate's opening or closure can be considered as ameasure ofthe variation in circumferential and radial residual stresses through the laminate(Reid,2009). In the radial cutting method,the amount of relative deformation is measured when the cylindrical structure is cut radially.Fourney (1968),Dewey and Knight (1969)and Aleong and Munro (1991)used elasticity equations to transform the measured relative strain either on the inner surface or on the outer surface during the radial cut into the residual stresses and residual strains in filament-wound composite rings.When material properties in the thick fabric composite cylinder are homogeneous,both the Sachs'boring method and the radial cutting method give similar results (Lee,2004),and consequently the radial cutting method is preferred over the Sachs method,because it is simpler and less expensive than the boring method. Current analytical methods related to this method are confined to laminates, which are balanced with respect to the cylindrical coordinate system.As a result,the residual shear stresses are taken to be zero.According to Kaddour et al.(2003),this assumption does not necessarily hold true.Thin filament-wound laminates exhibit a change in axial displacement across the cut.This implies the existence of a built-in twist with corresponding residual shear stresses(Reid,2009).The radial cutting method is based on measuring the elastic response of the laminate as a whole and also on the assumption of homogeneous materials.Both of these conditions prevent the resolution of residual stresses at the micro-scale(Reid,2009). Kaddour et al.(2003)conducted a preliminary study to investigate the residual stresses developed in hot cured thin-walled angle-ply filament wound tubes made of E-glass/epoxy,Kevlar/epoxy and carbon/epoxy materials.Kim et al.(2006) developed a smart cure method with cooling and reheating to reduce residual Woodhead Publishing Limited,2014
30 Residual stresses in composite materials © Woodhead Publishing Limited, 2014 Gascoigne (1994) used high- sensitivity Moiré interferometry and linearized strain- displacement equations to measure released displacements and residual strains. Joh et al . (1993) developed a novel concept of layer separation to measure quantitatively and precisely the tensile residual stresses in thick plates with layered distribution of residual stresses. Chapman et al . (1990) presented a model to predict the macroscopic in- plane residual stress state of semi- crystalline thermoplastic composite laminates induced by process cooling. Their model predictions were in good agreement with experimental residual stress measurements for uni- directional graphite (AS4) reinforced PEEK laminates. 2.9 The radial cutting method The radial cutting method is a different version of the sectioning method. It has been used extensively to determine the residual stresses in fi ber reinforced tubes (Cohen, 1997; Ganley et al ., 2000; Seif and Short, 2002, Seif et al ., 2006) and rings (Aleong and Munro, 1991; Roy, 1991). Based on the assumption that the residual stresses in such structures are invariant with axial and circumferential position, a single axial cut can result in their release. The extent of a laminate’s opening or closure can be considered as a measure of the variation in circumferential and radial residual stresses through the laminate (Reid, 2009). In the radial cutting method, the amount of relative deformation is measured when the cylindrical structure is cut radially. Fourney (1968), Dewey and Knight (1969) and Aleong and Munro (1991) used elasticity equations to transform the measured relative strain either on the inner surface or on the outer surface during the radial cut into the residual stresses and residual strains in fi lament- wound composite rings. When material properties in the thick fabric composite cylinder are homogeneous, both the Sachs’ boring method and the radial cutting method give similar results (Lee, 2004), and consequently the radial cutting method is preferred over the Sachs method, because it is simpler and less expensive than the boring method. Current analytical methods related to this method are confi ned to laminates, which are balanced with respect to the cylindrical coordinate system. As a result, the residual shear stresses are taken to be zero. According to Kaddour et al . (2003), this assumption does not necessarily hold true. Thin fi lament- wound laminates exhibit a change in axial displacement across the cut. This implies the existence of a built- in twist with corresponding residual shear stresses (Reid, 2009). The radial cutting method is based on measuring the elastic response of the laminate as a whole and also on the assumption of homogeneous materials. Both of these conditions prevent the resolution of residual stresses at the micro- scale (Reid, 2009). Kaddour et al . (2003) conducted a preliminary study to investigate the residual stresses developed in hot cured thin- walled angle- ply fi lament wound tubes made of E-glass/epoxy, Kevlar/epoxy and carbon/epoxy materials. Kim et al . (2006) developed a smart cure method with cooling and reheating to reduce residual
Destructive techniques in the measurement of residual stresses 31 stresses in thick-wound composite cylinders made of carbon phenolic woven composite.Roy (1991)presented thermal stress analysis of a thick laminated ring. Kim and Lee (2007)measured the residual stresses in thick cylinders made of carbon fabric phenolic composites by a new radial-cut cylinder-bending method. 2.10 Matrix removal methods Matrix removal methods are based on the fact that residual stresses are mutually self-equilibrated.In uni-directional laminates,stresses within the fibers are opposed by stresses in the matrix.If the matrix material is removed,the stresses within the fibers are released.The consequent elastic response of the fiber allows the residual stresses to be measured(Reid,2009). As long as the fibers are not damaged,the matrix material can be removed using a variety of techniques,and it is worth noting that the technique selected depends greatly on the type of composite material.Strong acids can be used to etch away metal matrices,acid digestion is utilized for the removal of polymer matrices surrounding carbon and aramid fibers,while polymer matrices surrounding glass fibers are vaporized with the aid of high temperatures(American Society for Testing and Materials,2002). Another method that employs the matrix removal procedure relies on the micro-buckling of the fibers(Zong and Marcus,1991).In metal matrix composites, the stress in reinforcement fibers is compressive at room temperature,thus the matrix material around the fibers provides them with enough support to prevent buckling.In absence of the matrix,the fibers can buckle freely.In this method,the lengths of buckled and unbuckled fibers are determined after a small portion of metal matrix is etched away from the surface of a composite plate and the underlying fibers are exposed.Using these lengths,the'clamped-clamped'Euler buckling stress is determined,which corresponds to the residual fiber stress.The major shortcoming is that a parameter known as the 'knock-down factor'needs to be introduced to take initial fiber imperfections and misalignment into account, and the use of this method is greatly restricted by the uncertainty concerning this parameter (Reid,2009). Another complication that is sometimes troublesome relates to the bending of the fibers after matrix dissolution(Ramamurty et al.,1996).Since a portion of the fiber length is inclined to the length direction,the extension of the fiber appears shorter than it actually is.In addition,this bending phenomenon alters their obvious location on the reference plane.It is not possible to align the reference plane completely normal to the measurement axis;the change in position alters the apparent change in length relative to the reference plane.Regardless of these issues,the method has been used to investigate the residual stresses within silicon- carbide reinforced titanium alloys(Fang et al.,2000;Gungor,2002). Fiber reinforced plastics have not been subjected to the etching technique as a method of residual stress measurement.This is most likely due to two main Woodhead Publishing Limited,2014
Destructive techniques in the measurement of residual stresses 31 © Woodhead Publishing Limited, 2014 stresses in thick- wound composite cylinders made of carbon phenolic woven composite. Roy (1991) presented thermal stress analysis of a thick laminated ring. Kim and Lee (2007) measured the residual stresses in thick cylinders made of carbon fabric phenolic composites by a new radial- cut cylinder- bending method. 2.10 Matrix removal methods Matrix removal methods are based on the fact that residual stresses are mutually self- equilibrated. In uni- directional laminates, stresses within the fi bers are opposed by stresses in the matrix. If the matrix material is removed, the stresses within the fi bers are released. The consequent elastic response of the fi ber allows the residual stresses to be measured (Reid, 2009). As long as the fi bers are not damaged, the matrix material can be removed using a variety of techniques, and it is worth noting that the technique selected depends greatly on the type of composite material. Strong acids can be used to etch away metal matrices, acid digestion is utilized for the removal of polymer matrices surrounding carbon and aramid fi bers, while polymer matrices surrounding glass fi bers are vaporized with the aid of high temperatures (American Society for Testing and Materials, 2002). Another method that employs the matrix removal procedure relies on the micro- buckling of the fi bers (Zong and Marcus, 1991). In metal matrix composites, the stress in reinforcement fi bers is compressive at room temperature, thus the matrix material around the fi bers provides them with enough support to prevent buckling. In absence of the matrix, the fi bers can buckle freely. In this method, the lengths of buckled and unbuckled fi bers are determined after a small portion of metal matrix is etched away from the surface of a composite plate and the underlying fi bers are exposed. Using these lengths, the ‘clamped- clamped’ Euler buckling stress is determined, which corresponds to the residual fi ber stress. The major shortcoming is that a parameter known as the ‘knock- down factor’ needs to be introduced to take initial fi ber imperfections and misalignment into account, and the use of this method is greatly restricted by the uncertainty concerning this parameter (Reid, 2009). Another complication that is sometimes troublesome relates to the bending of the fi bers after matrix dissolution (Ramamurty et al ., 1996). Since a portion of the fi ber length is inclined to the length direction, the extension of the fi ber appears shorter than it actually is. In addition, this bending phenomenon alters their obvious location on the reference plane. It is not possible to align the reference plane completely normal to the measurement axis; the change in position alters the apparent change in length relative to the reference plane. Regardless of these issues, the method has been used to investigate the residual stresses within siliconcarbide reinforced titanium alloys (Fang et al ., 2000; Güngör, 2002). Fiber reinforced plastics have not been subjected to the etching technique as a method of residual stress measurement. This is most likely due to two main
32 Residual stresses in composite materials factors;the low elastic modulus of polymers and the small diameter of the fibers used with polymer matrices (Reid,2009).Since the modulus of polymers is notably lower than that of metals,polymer matrix composites retain lower residual strain in the fibers compared to metal matrix composites.Thus in order to obtain acceptable resolution in the displacement of the fiber ends,the matrix must be removed over a greater fiber length.This increases the chances of bending in exposed fibers,the effect of which is greatly exacerbated by their small diameter in comparison to those used in the metal matrix composites studied previously (Fang et al.,2000;Gungor,2002;Ramamurty et al.,1996).Significant bending of the fibers reduces the accuracy of the measured change in fiber length,and consequently this method is of limited helpfulness for fiber reinforced plastics. Another problem that limits the use of this method with GFRP is that if strong acids are used to remove the matrix,the fibers might corrode or crack (Jones and Chandler,1985).Using high temperatures to vaporize or burn off the polymer matrix are possible alternatives (American Society for Testing and Materials, 2002).However,the residual stress state in the material will be altered,because the high temperatures employed to vaporize the matrix between the slits will affect the neighboring material.Therefore,if this method must be used for GFRPs,the entire matrix should be burned off simultaneously.Achieving a measure of the original residual stress in the material is possible through a comparison between the length of the glass fiber prior to and after matrix removal (Reid,2009).Modifications to the etching process could thus allow the measurement of the longitudinal fiber stresses within GFRP.However,bending of fibers and the need to vaporize rather than etch the matrix away,causes the measurement to be less accurate.The practicality of this method is therefore questionable (Reid,2009). The etching or dissolution method has been used extensively to quantify fiber stress in silicon-carbide reinforced titanium alloys.The method was first used by Cox et al.(1990),who dissolved the matrix from the central part oflong rectangular specimens and subsequently measured their change in length.The change in length was then related back to the average fiber strain resulting from the release of residual stresses. The etching technique has been enhanced since it was first proposed.Pickard et al.(1995)presented a simplified experimental technique to determine the axial fiber residual strain in continuously-reinforced metal matrix composites.Kendig et al.(1995)measured residual stresses in Ti-15-3/SCS-9 composites with controlled matrix and interfacial microstructures.The FEMUR test was used to measure axial residual strains in the reinforcing fibers.Ramamurty et al.(1996) modified the method by preparing a flat face perpendicular to the fiber direction, at the end of a composite specimen,which acted as a reference surface.This method was accurate and simple to implement.In addition,it completely released the strains in every fiber,thereby allowing the residual stress in individual fibers to be determined. Woodhead Publishing Limited,2014
32 Residual stresses in composite materials © Woodhead Publishing Limited, 2014 factors; the low elastic modulus of polymers and the small diameter of the fi bers used with polymer matrices (Reid, 2009). Since the modulus of polymers is notably lower than that of metals, polymer matrix composites retain lower residual strain in the fi bers compared to metal matrix composites. Thus in order to obtain acceptable resolution in the displacement of the fi ber ends, the matrix must be removed over a greater fi ber length. This increases the chances of bending in exposed fi bers, the effect of which is greatly exacerbated by their small diameter in comparison to those used in the metal matrix composites studied previously (Fang et al ., 2000; Güngör, 2002; Ramamurty et al ., 1996). Signifi cant bending of the fi bers reduces the accuracy of the measured change in fi ber length, and consequently this method is of limited helpfulness for fi ber reinforced plastics. Another problem that limits the use of this method with GFRP is that if strong acids are used to remove the matrix, the fi bers might corrode or crack (Jones and Chandler, 1985). Using high temperatures to vaporize or burn off the polymer matrix are possible alternatives (American Society for Testing and Materials, 2002). However, the residual stress state in the material will be altered, because the high temperatures employed to vaporize the matrix between the slits will affect the neighboring material. Therefore, if this method must be used for GFRPs, the entire matrix should be burned off simultaneously. Achieving a measure of the original residual stress in the material is possible through a comparison between the length of the glass fi ber prior to and after matrix removal (Reid, 2009). Modifi cations to the etching process could thus allow the measurement of the longitudinal fi ber stresses within GFRP. However, bending of fi bers and the need to vaporize rather than etch the matrix away, causes the measurement to be less accurate. The practicality of this method is therefore questionable (Reid, 2009). The etching or dissolution method has been used extensively to quantify fi ber stress in silicon- carbide reinforced titanium alloys. The method was fi rst used by Cox et al . (1990), who dissolved the matrix from the central part of long rectangular specimens and subsequently measured their change in length. The change in length was then related back to the average fi ber strain resulting from the release of residual stresses. The etching technique has been enhanced since it was fi rst proposed. Pickard et al . (1995) presented a simplifi ed experimental technique to determine the axial fi ber residual strain in continuously- reinforced metal matrix composites. Kendig et al . (1995) measured residual stresses in Ti-15-3/SCS-9 composites with controlled matrix and interfacial microstructures. The FEMUR test was used to measure axial residual strains in the reinforcing fi bers. Ramamurty et al . (1996) modifi ed the method by preparing a fl at face perpendicular to the fi ber direction, at the end of a composite specimen, which acted as a reference surface. This method was accurate and simple to implement. In addition, it completely released the strains in every fi ber, thereby allowing the residual stress in individual fi bers to be determined
Destructive techniques in the measurement of residual stresses 33 Gungor(2002)measured the residual stresses in two Ti/SiC uni-directional composite panels with thick cladding using two experimental methods,crack compliance to measure the variation of in-plane residual stresses in the cladding of the materials,and matrix etching to measure the longitudinal fiber strains.By combining the results of both methods,the out-of-plane stresses were also determined,so that the full stress state in the reinforced section of the material could be obtained.Fang et al.(2000)applied a method based on matrix etching for calculating residual stresses in continuous fiber reinforced titanium matrix composites. 2.11 Micro-indentation methods Micro-indentation techniques are micro-mechanical methods that are used to analyze the interfacial characteristics of composites (Kalton et al.1998; Parthasarathy et al.,1991).Micro-indentation techniques do not require the use of model composite systems(Ramanathan et al.,2001).Single fiber push-in and single fiber push-out are two micro-indentation methods.Both of these methods include exerting a compressive longitudinal load at the end of a single fiber using a small indenter.The end of the fiber is exposed by cutting the composite perpendicular to the direction of the fibers and then polishing the cut surface.The test relies inherently on the heterogeneous nature of a composite material,but allows measurements of micro-scale residual stresses (Lara-Curzio and Ferber, 1994;Ramanathan et al.,2001). In push-in tests,a backing plate is positioned on the reverse face of the specimen, and the compressive load on the fiber causes an increase in interface stress between the fiber and matrix.The loading is increased until the interface finally fails and debonding occurs at the fiber end.As the load is further increased,the debonding extends progressively along the fiber.However,since the fiber is long,the debonding length never extends beyond a small fraction of the length of the embedded fiber. In push-out tests,the specimen is less thick,and the debonding length can consequently extend over the complete length of the fiber.As a result,the fiber is pushed out of the rear side of the specimen(Lara-Curzio and Ferber,1994). Several mechanical properties of the fiber and matrix affect the beginning and propagation of the debonding,including the strength and fracture toughness of the interface,and the friction between fiber and matrix.Residual stresses that appear in the form of radial clamping stress and longitudinal stresses need to be considered.Analytical predictions are generally used to characterize the interface properties(Zhou et al.,2001).It is simple to visualize the influence of longitudinal residual stresses on the measurements.The release of longitudinal residual compression in the fiber with increasing debond length causes the fiber to extend. The extension and the applied compressive loading are in opposition,therefore the displacement of the fiber end is less for a given applied load than if longitudinal residual stresses were not present(Reid,2009). Woodhead Publishing Limited,2014
Destructive techniques in the measurement of residual stresses 33 © Woodhead Publishing Limited, 2014 Güngör (2002) measured the residual stresses in two Ti/SiC uni- directional composite panels with thick cladding using two experimental methods, crack compliance to measure the variation of in- plane residual stresses in the cladding of the materials, and matrix etching to measure the longitudinal fi ber strains. By combining the results of both methods, the out- of-plane stresses were also determined, so that the full stress state in the reinforced section of the material could be obtained. Fang et al . (2000) applied a method based on matrix etching for calculating residual stresses in continuous fi ber reinforced titanium matrix composites. 2.11Micro- indentation methods Micro- indentation techniques are micro- mechanical methods that are used to analyze the interfacial characteristics of composites (Kalton et al . 1998; Parthasarathy et al ., 1991). Micro- indentation techniques do not require the use of model composite systems (Ramanathan et al ., 2001). Single fi ber push- in and single fi ber push- out are two micro- indentation methods. Both of these methods include exerting a compressive longitudinal load at the end of a single fi ber using a small indenter. The end of the fi ber is exposed by cutting the composite perpendicular to the direction of the fi bers and then polishing the cut surface. The test relies inherently on the heterogeneous nature of a composite material, but allows measurements of micro- scale residual stresses (Lara-Curzio and Ferber, 1994; Ramanathan et al ., 2001). In push- in tests, a backing plate is positioned on the reverse face of the specimen, and the compressive load on the fi ber causes an increase in interface stress between the fi ber and matrix. The loading is increased until the interface fi nally fails and debonding occurs at the fi ber end. As the load is further increased, the debonding extends progressively along the fi ber. However, since the fi ber is long, the debonding length never extends beyond a small fraction of the length of the embedded fi ber. In push- out tests, the specimen is less thick, and the debonding length can consequently extend over the complete length of the fi ber. As a result, the fi ber is pushed out of the rear side of the specimen (Lara-Curzio and Ferber, 1994). Several mechanical properties of the fi ber and matrix affect the beginning and propagation of the debonding, including the strength and fracture toughness of the interface, and the friction between fi ber and matrix. Residual stresses that appear in the form of radial clamping stress and longitudinal stresses need to be considered. Analytical predictions are generally used to characterize the interface properties (Zhou et al ., 2001). It is simple to visualize the infl uence of longitudinal residual stresses on the measurements. The release of longitudinal residual compression in the fi ber with increasing debond length causes the fi ber to extend. The extension and the applied compressive loading are in opposition, therefore the displacement of the fi ber end is less for a given applied load than if longitudinal residual stresses were not present (Reid, 2009)