The effects of shear on Mode II delamination : a critical review

The paper focuses on the effects of shear deformation and shear forces on the mode II contribution to the energy release rate in delaminated beams. A critical review of the relevant literature is presented, starting from the end-notched flexure test as the prototype of delaminated laminates subjected to pure mode II fracture. Several models of the literature are recalled from simple beam theory to more refined models. The role of first-order shear deformation in line with the Timoshenko beam theory is investigated as distinct from the local crack-tip deformation related to the shear modulus of the material. Then, attention is moved on to a general delaminated beam with an arbitrarily located through-the-width delamination, subjected to mixedmode fracture. Several fracture mode partition methods of the literature are reviewed with specific attention on the effects of shear on the mode II contribution to the energy release rate.


INTRODUCTION
elamination, or interlaminar fracture, is a major failure mode for composite laminates, which still attracts researchers' attention despite the huge number of dedicated studies during the last decades [1][2][3][4][5].Since the earlier works, it has been recognised that the structural behaviour of a laminate affected by delamination can be analysed by schematising the laminate as an appropriate assemblage of sublaminates [6,7].These are in turn modelled as plates or beams, depending on the geometry, loads, and boundary conditions of the particular problem at hand.Proper homogenisation techniques are used to obtain the overall laminate stiffnesses [8].The Euler-Bernoulli beam theory -also referred to in the literature as classical or simple beam theory (SBT) -is the most elementary structural theory that can be used in this context.SBT assumes that the plane cross sections of a beam remain plane and orthogonal to the centreline after deformation.Thus, SBT completely neglects shear deformation, which indeed can be relevant for composite laminates because of the orthotropic material behaviour.Shear deformation is taken into account at first order by the Timoshenko beam theory (TBT), which admits relative rotations between the plane cross sections and the centreline of a beam [9].Simple and Timoshenko beam theories correspond to the Kirchhoff-Love and Mindlin-Reissner theories for thin and thick plates, respectively [10].Higher-order shear-deformation theories (HSDT) for D beams and plates assume that plane cross sections may not remain plane, which enables better agreement between the structural models and three-dimensional elastic analyses.After the pioneering work by Reddy [11], several modified HSDTs have been developed (see Ref. [12] for a recent review), including specialised versions for the analysis of delaminated plates [13,14].Besides the adoption of a more or less refined structural theory for the sublaminates, the connection between them has to be suitably described.To this aim, models of growing complexity can be chosen, ranging from rigid connections [15][16][17][18][19][20][21][22][23][24] to elastic interfaces [25][26][27][28][29][30][31][32][33][34][35][36][37][38] and cohesive zone models [39][40][41][42][43][44][45][46][47].This choice is relevant not only for the accurate prediction of the laminate structural response -e.g. in terms of displacements, stresses, etc. -but also for the evaluation of crack growth.In fact, the type of connection model determines the type of description of the stress field at the delamination crack front.With a rigid connection, the stress field will be represented by concentrated forces and couples, whereas with deformable interfaces there will be a distribution of peeling and shearing stresses [23].Based on such quantities, typical fracture mechanics parameters can be evaluated.Within linear elastic fracture mechanics (LEFM), the energy release rate, G, is commonly considered to predict the initiation and growth of delamination cracks [48].However, delamination cracks usually propagate under a mix of the three basic fracture modes (I or opening, II or sliding, and III or tearing).Therefore, appropriate mixed-mode partition methods are adopted to decompose G into the sum of three modal contributions, GI, G II , and G III [49].Besides, suitable experimental procedures are used to characterise the delamination toughness in pure and mixed fracture mode conditions [50].In particular, for pure mode II, the end-notched flexure (ENF) test is the current standard [51].This paper focuses on delaminated laminates modelled according to simple or Timoshenko beam theories with rigid connections or elastic interfaces.The aim is to shed light on the effects of shear deformation and shear forces on the mode II contribution, GII, to the energy release rate.Literature on this topic is contradictory with some Authors asserting [19,21,31,46] and others negating [16,20,24,30,36] this effect.As will be illustrated, the origin of this controversy can be dated back to a series of papers on the analysis of the ENF test [52][53][54][55][56][57][58][59][60][61][62][63][64][65].In particular, Carlsson et al. [53] modelled the ENF test by using the Timoshenko beam theory and found correction terms depending on shear deformation for both the specimen compliance, C, and energy release rate, G II .Their formulas have been widely used in the later literature for the interpretation of experimental results [66][67][68] and comparison purposes [69][70][71].Unfortunately, as pointed out by Fan et al. [64], the derivation of such formulas was biased by a wrong boundary condition.Actually, Silva et al. [63] had obtained the correct expressions for C and G II according to the TBT: the compliance does have a term depending on shear deformation, which however does not depend on the crack length, a.Hence, this term vanishes when differentiating C with respect to a to obtain GII according to the well-know Irwin-Kies formula [72].Notwithstanding the above, the wrong formulas from [53] are still used even in the more recent literature [73][74][75][76][77] and reported in the latest edition of an otherwise excellent book [50].This paper extends some preliminary considerations by the Author [75], also in the light of new findings in the recent literature.The outline is as follows.First, a review on the analysis of the ENF test is given and some preliminary conclusions are drawn for laminates with symmetric -i.e.placed on the mid-plane -delamination.Then, attention is moved on to a general delaminated beam with a through-the-width delamination arbitrarily located in the thickness.In this case, mixedmode fracture conditions generally occur with G = GI + GII.Several mixed-mode partition methods of the literature are reviewed with specific attention on the effects of shear forces and shear deformation on G II .Lastly, a quantitative comparison is pursued between the predictions for G I and G II stemming from (i) a rigid-connection model [24], (ii) an elastic-interface model [37], and (iii) a solution based on the theory of elasticity [21].

END-NOTCHED FLEXURE TEST
he end-notched flexure (ENF) test has been recently standardised by ASTM International as the method for the characterisation of mode II delamination toughness of unidirectional fibre-reinforced composite laminates [51].In the test, a laminated specimen with rectangular cross section is loaded by a force P in a three-point bending configuration.Let L = 2l denote the specimen length, B and H = 2h the cross-section width and thickness, respectively.Specimens are prepared with a mid-plane delamination crack at one of their ends.Let a be the delamination length.A Cartesian reference system Oxyz is fixed with the origin O at the centre of the crack-tip cross section, the x-axis aligned with the specimen's longitudinal direction, the z-axis pointing downwards, and the y-axis completing the right-handed reference frame (Fig. 1).The ENF test specimen can be considered as subject to an antisymmetric loading condition with respect to its mid-plane (Fig. 2).In this case, if also material properties have a symmetric distribution with respect to the mid-plane, then normal stresses on the delamination plane will be null.Hence, pure mode II fracture conditions will be attained.This is the case of T homogeneous and orthotropic specimens, as well as unidirectional laminated specimens and multidirectional laminated specimens with symmetric lay-ups.It is hardly necessary to remember that in the asymmetric end-notched flexure (AENF) test, the delamination crack will generally be subjected to mixed-mode fracture conditions [79][80][81][82].In the following, focus will be on a homogeneous and orthotropic specimen.The elasticity moduli in the material reference will be denoted as Ex, E z , G zx , and  xz [8].For the following analysis, it is useful to introduce the specimen compliance, defined as the ratio between the displacement,, of the load application poin and the load intensity, P. Furthermore, the energy release rate, G, is defined as the decrease of potential energy of the system spent in the crack growth process, per unit area of new surface created.The energy release rate can be obtained by differentiating the compliance with respect to delamination length according to the Irwin-Kies formula [72]: Depending on the adopted structural model, various expressions for the specimen compliance have been obtained in the literature.Correspondingly, various expressions for the energy release rate are deduced through Eq. ( 2).

Simple and Timoshenko beam-theory models
The ENF test was first used for composite materials by Russell and Street [52], who applied simple beam theory (SBT) to determine the specimen compliance, and energy release rate, To consider shear deformation, Carlsson et al. [53] modelled the ENF test via the Timoshenko beam theory (TBT) [9] and obtained the following expressions: for the compliance, and 9 2 1 10 16 for the energy release rate.However, Whitney [55] promptly observed that «continuity of displacement at the crack tip is not attained with the approach that yields» Eqs. ( 5) and (6).Later, also Fan et al. [64] realised that «a false assumption was made in the derivation», leading to «inconsistency (…) for the expressions that are used to calculate the energy release rate».Actually, in Ref. [53] the cross-section rotation at the crack tip is taken equal to the slope of the deflected beam, but this assumption manifestly contradicts the hypotheses on which Timoshenko's first-order shear-deformation beam theory is based.As a matter of fact, by applying the TBT without the above unnecessary approximations, Silva et al. [63] obtained: for the compliance, and for the energy release rate.Eq. (7) shows that shear deformation at first order modifies the compliance with an additional term with respect to simple beam theory.This correction terms is however constant with respect to a, hence it vanishes when applying Eq. ( 2) to deduce the energy release rate.Thus, as Eq. ( 8) shows, G II turns out to have the same expression according to both SBT and TBT.To confirm this, it is instructive to examine qualitatively the deformed shapes due to shear only of two ENF test specimens with shorter (Fig. 3a) and longer (Fig. 3b) delamination cracks.If the specimen is modelled as an assemblage of rigidly connected sublaminates, the deformed shape due to shear is clearly independent of the delamination length, a.In this respect, it can be mentioned that Ozdil et al. [59] analysed the ENF test using shear-deformation laminated plate theory and found that «there is no (…) contribution from shear deformation to the energy release rate for unidirectional (…) laminates with mid-plane cracks», while a «very small contribution» emerges for angle-ply laminates.Also, Chatterjee [56] used shear-deformation laminated plate theory to study the ENF test and found that «the energy release rate (…) will not be affected if shear deformation effects (in the context of beam theory) are neglected».

Elasticity-theory models
Authors who modelled the ENF test within the theory of elasticity generally obtained numerical results showing a dependence of the energy release rate on the shear modulus of the material [54,56,57].
Chatterjee [56] suggested the following semi-empirical expression to match the elasticity theory solution: Wang and Williams [57] used finite element analysis and found a dependence of both the compliance and energy release rate on the shear modulus of the material.They observed that their numerical results for the energy release rate could be approximated by introducing an increased delamination length into the SBT expression, Eq. ( 4): A similar crack-length correction parameter, , was later adopted by many Authors [36,62].Today, it is also suggested in the standard method for the mixed-mode bending (MMB) test [83].Andrews and Massabò [21], based on finite element analyses, gave the following expression for the energy release rate of an orthotropic ENF test specimen: where 1 2 1 2 , , ,and a a a a are compliance coefficients given in a tabular form as functions of the material elastic moduli, including the shear modulus.Andrews and Massabò's solution [21] highlights the fact that local deformation -in particular, deformation related to root rotations, i.e. the rotations of the sublaminate cross sections at the crack tip -plays a relevant role in delamination fracture problems.It is also noteworthy that for isotropic materials, Eq. ( 11) degenerates into Eq.( 9).

Higher-order shear-deformation theory models
Whitney [55] used second-order shear-deformation beam theory (SOBT) and obtained the following approximate polynomial solution for the energy release rate: , where 4 75 5 16 Pavan Kumar and Raghu Prasad [61,65] compared simple beam theory with first-order (i.e.TBT), second-order (SOBT), and third-order shear-deformation beam theories (TOBT).Through numerical computation, they found that the simple and Timoshenko beam theory models of the ENF test yield the same values of the energy release rate, while the higher-order beam theory models furnish larger values.For short crack lengths, they reported that the TOBT model gives corrections to G II up to 30% of SBT ENF G for unidirectional laminated specimens and nearly 50% for multidirectional laminated specimens.They also found that TOBT was in good agreement with the results of finite element analyses.

Elastic-interface models
Corleto and Hogan [58] modelled the ENF test by considering the upper half laminate as a Timoshenko beam on a generalised elastic foundation consisting of extensional and rotational distributed springs.They found that the energy release rate is independent of the shear stiffness of the sublaminate, 5/6 GzxBh, but not of the shear modulus of the material, Gzx, which enters the expressions of the elastic constants of the foundation.Based on this result, Ding and Kortschot [60] used SBT to deduce a simplified model of the ENF test, where the foundation consists of tangential springs only.Wang and Qiao [62] used TBT and considered an elastic foundation made of distributed tangential springs.By neglecting some numerically small terms in the solution, they obtained the following expression: for the compliance, and 9 ( ) , where and 5 12 16 for the energy release rate.Eq. (13) shows that the shear stiffness influences the specimen's compliance, but not the energy release rate.The latter, however, is dependent on the shear modulus of the material through the crack-length correction parameter, .
A similar result was obtained by Bennati et al. [35,36], who developed an enhanced beam-theory (EBT) model of the MMB test, where the sublaminates are flexible, extensible, and shear-deformable laminated beams, partly connected by an elastic interface consisting of normal and tangential springs.As a special case, they obtained the solution for an orthotropic ENF test specimen in terms of compliance, and energy release rate,

(
) , where 8 16 The third addend in Eq. ( 15) comes from the deformability of the elastic interface.Eq. ( 15) differs slightly from Eq. ( 13), but furnishes quite similar numerical values.Also according to the EBT model, the shear deformability does not influence the mode II contribution to the energy release rate.However, it should be noted that, albeit the shear modulus, Gzx, does not enter explicitly Eq. ( 16), it is related to the elastic interface constant, k x [27].
As for models with rigidly connected sublaminates, this result is perfectly intuitive for elastic-interface models if the deformed shapes of two specimens with different delamination lengths are considered (Fig. 4).

Preliminary conclusions
From the above literature review, the following preliminary conclusions can be drawn with reference to the ENF test: -three-dimensional finite element analyses show that both the specimen compliance, C, and energy release rate, G II , exhibit a dependence on the shear modulus of the material, Gzx; -the Timoshenko beam theory adds a correction term with respect to simple beam theory into the expression of C, depending on the (half) specimen shear stiffness, 5/6 G zx Bh; -the above correction term is constant with respect to delamination length, a; hence, it does not influence the energy release rate, GII, which turns out to be independent of shear deformation at first order; -some widely used expressions of the literature for C and G II accounting for shear deformation turned out to be wrong; -the dependence of GII on Gzx seems to be related to local deformation occurring in the neighbourhood of the delamination front because of high stress concentration, e.g.strain in the laminate thickness direction, Poisson's effect, and root rotations; -to catch such effects, it is necessary to use complex models, such as those based on the theory of elasticity or higherorder shear-deformation beam theories; alternatively, enhanced beam theory models with deformable -elastic or cohesive -interfaces between the delaminating sublaminates can be effectively used.The above considerations can be extended from the ENF test to similar mode II delamination tests -e.g. the end-loaded split (ELS) test, the four-point end-notched flexure (4ENF) test, etc. [69] -where the specimen has a mid-plane delamination.In this case, fracture modes I and II are related to the symmetric and antisymmetric external forces acting on the specimen, respectively.Instead, the analysis of laminates with delamination cracks arbitrarily placed in the thickness requires the adoption of more complex mixed-mode partition methods.These will be the subject of the next section.

GENERAL DELAMINATED LAMINATES
ttention is now moved on to a general delaminated beam with an arbitrarily located through-the-width delamination.Let L be the laminate length, B and H = 2h the cross-section width and height, respectively.A Cartesian reference system Oxyz is fixed with the origin O at the geometric centre of one of the end sections, the x-axis aligned with the laminate longitudinal direction, the y-and z-axes aligned with the cross-section width and height directions, respectively (Fig. 5a).The analysis can be limited to an infinitesimal beam segment included between two cross sections located immediately behind and ahead of the delamination front.The delamination is not necessarily placed on the mid-plane, but divides the laminate into two sublaminates with thicknesses H1 = 2h1 and H2 = 2h2 (Fig. 5b) [24].

Rigid-connection models
If the delaminated beam is modelled as an assemblage of rigidly connected sublaminates, the total energy release rate can be evaluated based on beam theory [21][22][23][24].Thus, an analytical expression for G is determined depending on the values of the internal forces acting on the crack-tip segment: the axial forces, N i , shear forces, Q i , and bending moments, M i (i = 1, 2, 3).However, to distinguish the single contributions stemming from fracture modes, GI and GII, it is necessary to introduce additional assumptions.To this aim, several fracture mode partition methods have been proposed in the literature.
Williams' global method [16] is based on the analysis of the external forces globally acting on the laminate.The method assumes that fracture mode I is produced when opposite bending moments act on the two sublaminates into which the laminate is split; besides, mode II is obtained when the sublaminates have equal curvatures.As a consequence, the following expressions for the modal contributions to G can be obtained: According to Eqs. ( 17)-( 19), axial forces produce only mode II, shear forces (and shear deformation) give only mode I, while bending moments contribute to both modes I and II.Schapery and Davidson [18] observed that Williams' assumptions on the partition of fracture modes are not generally fulfilled if the delamination crack is not placed on the laminate mid-plane.Hence, they proposed a method based on classical laminated plate theory, where the mode I and II contributions to the energy release rate depend on the stress resultantsan axial force, N C , and a bending moment, M C -exchanged between the upper and lower sublaminates at the crack tip (Fig. 6a, b).Schapery and Davidson did not consider shear forces and shear deformation in their method.Wang and Qiao [20] considered two perfectly bonded, shear-deformable laminated beams.They introduced a shear force at the crack tip, QC, but neglected the bending moment, MC (Fig. 6c).Their expressions for the modal contributions are: where, for orthotropic specimens, the coefficients take the following form: According to Eqs. ( 20) and ( 21), shear forces and shear deformation contribute only to mode I.
Valvo [24] extended Williams' method from homogeneous beams to general laminated beams.He determined the modal contributions to G based on a modified virtual crack closure technique.By considering all the three stress resultants -NC, Q C , and M C -at the crack tip (Fig. 6d), he obtained: where, for orthotropic specimens, Comparison of Eqs. ( 22) and ( 23) with ( 20) and (21) shows that Valvo's method [24] reduces to Wang and Qiao's method [20], if the crack-tip bending moment, MC, is disregarded.In general, however, MC gives a contribution to both modes I and II.Instead, shear forces and shear deformation contribute only to mode I.By substituting Eqs. ( 23) into ( 22) and expressing the crack-tip forces in terms of the internal forces on the delaminated sublaminates, the expressions for the modal contributions to the energy release rate become According to Eqs. (24), axial forces may contribute also to mode I.This contribution -neglected by both Williams' [16] and Wang and Qiao's [20] methods -passes through the crack-tip bending moment, M C , and depends on the contribution of axial forces on the moment balance on the laminate cross section [24].

Elasticity-theory models
As opposite to Williams' global method [16], Suo and Hutchinson [17] developed a local method based on the analysis of the singular stress field at the delamination crack tip.They analysed the problem of a semi-infinite crack between two homogeneous and isotropic elastic layers subjected to axial forces and bending moments.First, they computed the energy release rate based on simple beam theory.Then, within linear elastic fracture mechanics, they solved numerically a plane elasticity problem to obtain the mode I and II stress intensity factors, KI and KII [48].Li et al. [19] extended the local method to include the effects of shear forces.Andrews and Massabò [21] further extended the method to include the effects of root rotations.They computed the total energy release rate based on the J-integral and then determined the stress intensity factors.Their analytical expressions depend on numerical coefficients obtained through finite element analyses and given in a tabular form.If no axial forces and bending moment are present, but only shear forces, their expression for the energy release rate takes the form and the stress intensity factors can be written as , , , a n d 2 The expressions for the parameters , , , ,and  are given in the cited paper [21].The modal contributions to the energy release rate can then be computed as follows: Inspection of Eqs. ( 26)- (28) shows that, according to elasticity theory, shear forces and the material shear modulus will generally affect both modes I and II.

Elastic-interface models
Bruno and Greco [30] specifically addressed the influence of shear deformation on delamination in the setting of elasticinterface models.They considered a two-layer plate with an elastic interface consisting of normal and tangential distributed springs and obtained an analytical solution by treating the spring constants as penalty parameters approaching infinity.They concluded that, for symmetric delamination, shear forces influence only the mode I contribution.Instead, for asymmetric delamination, shear forces are expected to affect both fracture modes.Furthermore, they found an interaction between shear forces and normal stresses coming from axial forces and bending moments.Qiao and Wang [31] determined an analytical solution for a bilayer beam with an elastic interface.They computed the energy release rate through the J-integral [48] and determined the mode-mixity angle via an adaptation of Suo and Hutchinson's method [17].In general, they found an influence of shear forces and shear deformation on both fracture modes.Wang et al. [46] compared several mixed-mode partition methods based on SBT and TBT with rigid and deformable interfaces.In general, they found that shear forces cause additional contributions to both modes I and II.Liu et al. [37] gave a general solution for adhesively bonded joints, where the adherends are modelled as Timoshenko beams and the adhesive layer is represented as an elastic interface.The interface consists of a continuous distribution of linearly elastic springs with constants kz and kx, respectively acting in the normal and tangential directions with respect to the interface plane (Fig. 7).Their solution can be used also for the analysis of delamination in composite laminates, if the interface is interpreted as a conventional means to account for the laminate transverse deformability and not as representative of a physical layer of adhesive.The modal contributions to the energy release rate can be expressed as [34]: where  0 and  0 respectively are the values of the normal and tangential interfacial stresses at the crack tip.For a homogeneous delaminated beam, such stresses turn out to have the following expressions [37]: and for symmetric delamination, i.e. when H1 = H2, and for general asymmetric delamination.In Eqs. ( 30) and ( 31), f1, f2, …, f7 and g1, g2, …, g7 are integration constants depending on the problem boundary conditions; furthermore, are problem parameters, while  1 ,  2 , …,  6 are the roots of the characteristic equation of the differential problem for the interfacial stresses [37].Inspection of Eqs. ( 29) and (30) shows that, for symmetric delamination, it is possible to separate fracture modes, i.e. to give suitable boundary conditions producing , hence G I = 0 (pure mode II), or f 5 = f 6 = f 7 =0, hence GII = 0 (pure mode I).In such cases, the analytical solution reported in Ref. [37] also shows that the sublaminate shear stiffnesses influence only the mode I contribution to the energy release rate.Instead, for asymmetric delamination, pure fracture modes cannot be obtained in general, since the normal and tangential stress components given by Eqs.(31) depend on the same integration constants, g1, g2, …, g6 (the last constant, g7, corresponds to the Jourawski shear stress occurring in a unbroken laminate and therefore is irrelevant in fracture problems).Hence, for asymmetric delamination, mixed-mode fracture conditions are generally present.In this case, both the mode I and II contributions to G depend on shear deformation and shear forces.

Numerical example
To complete the above discussion with some quantitative results, a comparison will be made between the predictions for G I and G II stemming from (i) the rigid-connection model [24], (ii) the elastic-interface model [37], and (iii) the local method in the version by Andrews and Massabò [21].The latter can be used for reference as it represents an exact solution obtained within the theory of elasticity.
For illustration, a homogeneous and orthotropic laminate is considered with cross-section sizes B = 25 mm and H = 4 mm and elastic moduli Ex = 100 GPa, Ez = 10 GPa, Gzx = 5 GPa, and xz = 0.3.Such properties are intended to be representative of a typical laminated specimen used in delamination toughness tests.
A beam segment at the delamination crack-tip is considered (Fig. 8a), subjected to shear forces Q1 and Q2 on the delaminated sublaminates and Q3 = Q1 + Q2 on the unbroken part.To investigate all possible load conditions, the acting forces are decomposed into the sum of a symmetric system (Fig. 8b) and an antisymmetric system (Fig. 8c) with With the above assumptions, for symmetric delamination, the symmetric and antisymmetric parts of the acting forces give rise to pure fracture modes I and II, respectively.But, if the delamination is not placed on the mid-plane, mixed-mode fracture conditions are expected.For the mode I contribution (Fig. 9a), all compared methods predict similar qualitative trends.The predictions of the elasticinterface model approach those of the rigid-connection model as the values of the elastic-interface constants increase.For 10  For the mode II contribution (Fig. 9b), the rigid-connection model predicts a null value over the whole range of thickness ratios.Instead, the elastic-interface model and local method predict non-zero mode II contributions.However, the computed GII values turn out to be two orders of magnitude lower than the corresponding GI values, as can be noticed from the different scales on the ordinate axes.Thus, it can be concluded that the application of symmetric shear forces produces prevailing mode I fracture conditions, regardless of the asymmetry in the position of the delamination crack.For symmetric delamination ( 1   ), all compared methods predict II 0 G  , i.e. pure mode I fracture conditions.Fig. 10 shows the mode I and II contributions to the energy release rate produced by two antisymmetric shear forces of intensity Qa = 100 N applied at the crack tip, as functions of the sublaminate thickness ratio.The same graphical conventions of Fig. 9 are used to denote the predictions of the rigid-connection model [24], the elastic-interface model [37], and the local method [21].Also, similar comments apply.For the mode I contribution (Fig. 10a), the elastic-interface model approaches the rigid-connection model as x  and z  increase.Again, the predictions of the elastic-interface model match closely those of the local method for 10 For the mode II contribution (Fig. 10b), the rigid-connection model predicts a null value over the whole range of thickness ratios.Instead, the elastic-interface model and local method predict non-zero mode II contributions.The computed G II values are lower than the corresponding GI values for smaller thickness ratios.Mixed-mode fracture conditions generally occur, with the mode II contribution increasing as  increases and G I correspondingly decreases.For symmetric delamination ( 1  ), all compared methods predict I 0 G  , i.e. pure mode II fracture conditions.Lastly, it should be noticed that the above numerical results correspond to the application of only shear forces at the crack tip.In general, however, bending moments and axial forces will be present as well.Indeed, their contributions to the energy release rate may be quite larger than that due to shear.However, a quantitative assessment of those effects requires referring to specific cases in terms of geometry, loads, boundary conditions, etc.This assessment is not pursued within this paper, as focus here is on shear forces and their effects on mode II delamination.

CONCLUSIONS
hear deformation is relevant for composite materials because of their anisotropic elastic behaviour, regardless of the slenderness of beam-like structural elements.As such, shear deformation increases the compliance of laminated beams affected by delamination and, consequently, may influence the energy release rate associated to delamination growth.In this paper, the effects of shear deformation and shear forces on the mode II contribution to the energy release rate have been examined in the light of different structural theories and fracture mode partition methods of the literature.S First, attention has been devoted to the analysis of the ENF test as the prototype of symmetrically delaminated laminates subjected to pure mode II fracture conditions.Comparison of the available models reveals that shear deformation as accounted for at first order by the Timoshenko beam theory affects the specimen compliance, but not the mode II energy release rate.The shear-deformation correction terms given in Ref. [53] and used in much of the later literature turned out to be incorrect.Hence, their use for experimental test interpretation and comparison of models should be avoided in the future.Indeed, three-dimensional finite element analyses of the ENF test show a dependence of the energy release rate on the shear modulus of the material.This behaviour may be related to local deformation occurring at the delamination crack tip because of high stress concentration, e.g.strain in the laminate thickness direction, Poisson's effect, and root rotations.Such effects can be captured also with beam theory models, however based on higher-order shear-deformation theories or through the introduction of deformable interfaces connecting the delaminated sublaminates.Similar considerations hold in general for laminates with symmetric, i.e. mid-plane, delamination.Next, a general delaminated beam has been considered with an arbitrarily located through-the-width delamination.In this case, mixed-mode fracture conditions generally occur and G II is only a part of the total energy release rate.In the paper, several mixed-mode partition methods of the literature have been reviewed with specific attention on the effects of shear forces and shear deformation on GII.Lastly, a quantitative comparison has been made between the predictions of the rigidconnection model [24], elastic-interface model [37], and local method [21] for a homogeneous and orthotropic laminate with shear forces applied at the crack tip.The results show a quite limited influence of the shear forces on the mode II contribution to the energy release rate, except for nearly symmetric delamination and antisymmetric shear forces.For the sake of simplicity, in the paper, attention has been limited to homogeneous and orthotropic beams.Further studies will be necessary to extend the above considerations to more complex structural elements, such as bi-material and multidirectional laminated beams and plates.

Figure 2 :
Figure 2: Free-body diagram of the ENF test specimen.

Figure 3 :
Figure 3: Rigid-connection model of the ENF test: deformed shapes due to shear for (a) shorter and (b) longer delamination cracks.

Figure 4 :
Figure 4: Elastic-interface model of the ENF test: deformed shapes due to shear for (a) shorter and (b) longer delamination cracks.

Figure 7 :
Figure 7: Elastic-interface model of a delaminated beam.

Figure 8 :
Figure 8: Decomposition of (a) the shear forces acting on a crack-tip segment into (b) symmetric and (c) antisymmetric systems.

Fig. 9 .
Fig.9shows the mode I and II contributions to the energy release rate produced by two symmetric shear forces of intensity Q s = 100 N applied at the crack tip, as functions of the sublaminate thickness ratio, 1 2 H H   .Without loss of generality, attention is restricted to the case 1 2 H H  , corresponding to good matching between the elastic-interface model and the local method.

Figure 9 :
Figure 9: (a) Mode I and (b) mode II contributions to the energy release rate due to symmetric shear forces.

Figure 10 :
Figure 10: (a) Mode I and (b) mode II contributions to the energy release rate due to antisymmetric shear forces.