# Complete analytical solutions for double cantilever beam specimens with bi-linear quasi-brittle and brittle interfaces

- 400 Downloads

## Abstract

In this work we develop a complete analytical solution for a double cantilever beam (DCB) where the arms are modelled as Timoshenko beams, and a bi-linear cohesive-zone model (CZM) is embedded at the interface. The solution is given for two types of DCB; one with prescribed rotations (with steady-state crack propagation) and one with prescribed displacement (where the crack propagation is not steady state). Because the CZM is bi-linear, the analytical solutions are given separately in three phases, namely (i) linear-elastic behaviour before crack propagation, (ii) damage growth before crack propagation and (iii) crack propagation. These solutions are then used to derive the solutions for the case when the interface is linear-elastic with brittle failure (i.e. no damage growth before crack propagation) and the case with infinitely stiff interface with brittle failure (corresponding to linear-elastic fracture mechanics (LEFM) solutions). If the DCB arms are shear-deformable, our solution correctly captures the fact that they will rotate at the crack tip and in front of it even if the interface is infinitely stiff. Expressions defining the distribution of contact tractions at the interface, as well as shear forces, bending moments and cross-sectional rotations of the arms, at and in front of the crack tip, are derived for a linear-elastic interface with brittle failure and in the LEFM limit. For a DCB with prescribed displacement in the LEFM limit we also derive a closed-form expression for the critical energy release rate, \(G_c\). This formula, compared to the so-called ‘standard beam theory’ formula based on the assumptions that the DCB arms are clamped at the crack tip (and also used in standards for determining fracture toughness in mode-I delamination), has an additional term which takes into account the rotation at the crack tip. Additionally, we provide all the mentioned analytical solutions for the case when the shear stiffness of the arms is infinitely high, which corresponds to Euler–Bernoulli beam theory. In the numerical examples we compare results for Euler–Beronulli and Timoshenko beam theory and analyse the influence of the CZM parameters.

## Keywords

DCB test Mode-I delamination Analytical solution Timoshenko beam theory Cohesive-zone model Linear-elastic fracture mechanics## List of symbols

*A*Cross-sectional area of a single DCB arm

*a*Crack length

- \(a_0\)
Initial crack length

*b*Width of a DCB

- \(C_i\)
Integration constants for the undamaged part (\(i=1,\ldots ,6\))

- \(C^j_i\)
Constants depending on \(C_3\), \(C_4\), \(\xi _3\) and \(\xi _4\) (\(i=3,4\) and \(j=M,T\))

- \({\overline{C}}_i\)
Integration constants for the undamaged part that are zero because of the boundary conditions for a semi-infinite DCB (\(i=1,\ldots ,6\))

- \(c_2\)
function \(\cos (L_{cz}\kappa \xi _2)\)

- \({\overline{c}_2}\)
function \(\cos (\overline{L}_{cz} \kappa \xi _2)\)

- \(ch_1\)
function \(\cosh (L_{cz}\kappa \xi _1)\)

- \({\overline{ch}_1}\)
function \(\cosh (\overline{L}_{cz}\kappa \xi _1)\)

- \({{\mathcal {D}}}\)
Dirac delta

- \(D_0\)
A constant depending on \(\psi \)

- \(D_i\)
Integration constants for the damaged part (\(i=1,\ldots ,4\))

- \(D_{ij} \)
Constants for the damaged part depending on the value of \(\omega \) (\(i=1,\ldots ,4\) and \(j=1,2\))

- \(E\)
Young’s modulus of DCB arms

- \(F\)
Vertical force applied on the upper layer of a DCB with prescribed displacement

- \(F_0\)
A tensile stress resultant (concentrated force) at the crack tip for the LEFM limit case

- \(F_E\)
Value of the applied force computed using Euler–Bernoulli beam theory

- \(F_L\)
Maximum value of the applied force in the linear-elastic phase for a DCB with prescribed displacement

- \(F_{max}\)
Maximum value of the applied load for a DCB with prescribed displacement

- \(F_T\)
Value of the applied force computed using Timoshenko beam theory

- \(F_{\sigma _{max}}\)
Value of the applied force computed for a finite value of \(\sigma _{max}\)

- \(F_{\infty }\)
Value of the applied force for the LEFM limit

- \(G_c\)
Critical energy release rate

- \({{\mathcal {H}}}\)
Heaviside function

- \(h\)
Depth of a single DCB arm

*I*Second moment of area of the cross-section of a single DCB arm

- \(J_c\)
Critical value of the J integral

- \(k_s\)
Shear correction coefficient

*L*Total length of a DCB specimen

- \(L_{cz}\)
Length of the cohesive or damage-process zone

- \({\overline{L}}_{cz}\)
Maximum value of \(L_{cz}\) for a DCB with prescribed rotations

- \(L_{cz}^{max}\)
Maximum value of \(L_{cz}\) for a DCB with prescribed displacement

- \(L_{cz}^{min}\)
Minimum value of \(L_{cz}\) during crack propagation for a DCB with prescribed displacement

- \(M\)
Concentrated moment applied on a DCB arm

- \({{\mathcal {M}}}_1\)
Bending moment in the upper DCB arm on the part where the interface is undamaged

- \({{\mathcal {M}}}_1^L\)
LEFM limit value of \({{\mathcal {M}}}_1\)

- \({{\mathcal {M}}}_2\)
Bending moment in the upper DCB arm on the part where the interface is damaged

- \(M_L\)
Maximum value of the applied moment in the linear-elastic phase

- \({M}_{max}\)
Maximum value of the applied moment

- \(q\)
Distributed transverse loading along the upper DCB arm

- \(r_i\)
Roots of the characteristic equation for the undamaged part (\(i=1,\ldots ,4\))

- \(s_2\)
function \(\sin (L_{cz}\kappa \xi _2)\)

- \({\overline{s}_2}\)
function \(\sin (\overline{L}_{cz}\kappa \xi _2)\)

- \(sh_1\)
function \(\sinh (L_{cz}\kappa \xi _1)\)

- \({\overline{sh}_1}\)
function \(\sinh (\overline{L}_{cz} \kappa \xi _1)\)

- \({{\mathcal {T}}}\)
Shear force in the upper DCB arm

- \({{\mathcal {T}}}_1\)
Shear force in the upper DCB arm on the part where the interface is undamaged

- \({{\mathcal {T}}}_1^L\)
LEFM limit value of \({{\mathcal {T}}}_1\)

- \({{\mathcal {T}}}_2\)
Shear force in the upper DCB arm on the part where the interface is damaged

- \(t_i\)
Roots of the characteristic equation for the damaged part (\(i=1,\ldots ,4\))

- \(v\)
Transversal displacement of the upper DCB arm

- \(v_1\)
Transversal displacement of the upper DCB arm on the part where the interface is undamaged

- \(v_1^L\)
LEFM limit value of \(v_1\)

- \(v_2\)
Transversal displacement of the upper DCB arm on the part where the interface is damaged

*x*Co-ordinate along the interface

- \(x_0\)
Co-ordinate

*x*corresponding to zero stress at the interface for the case of a linear-elastic interface with brittle failure- \(x_{min}\)
Co-ordinate

*x*corresponding to the minimum (maximum compressive) stress at the interface for the case of a linear-elastic interface with brittle failure- \(\alpha \)
A constant defined as \(\alpha ={\delta }_0/{\delta }_c\)

- \(\beta _i\)
Constants depending on \(\xi _j\), \(D_{kl}\), \(\kappa \) and \({a}_0\) (\(i=1,\ldots ,8\), \(j=1,2\), \(k=1,\ldots ,4\) and \(l=1,2\))

- \(\gamma \)
Shear strain in the upper DCB arm

- \(\Delta \)
Prescribed crack mouth opening displacement of a DCB

- \(\Delta _{a} \)
Part of the crack mouth opening displacement due to bending of the arms assuming that they are clamped at the crack tip

- \(\Delta _{CT}^{\delta }\)
Part of the crack mouth opening displacement due to the opening at the crack tip

- \(\Delta _{CT}^{\varphi }\)
Part of the crack mouth opening displacement due to the rotation at the crack tip

- \(\delta \)
Relative displacement at the interface in mode I

- \({\delta }_0\)
Linear-elastic limit value of the relative displacement at the interface in mode I

- \({\delta }_c\)
Relative displacement at the interface corresponding to the total loss of interconnection in mode I

- \(\varepsilon _r\)
Relative error due to using Euler–Bernoulli instead of Timoshenko beam theory

- \({\overline{\varepsilon }_r}\)
Relative difference between CCM and LEFM solutions

- \(\zeta _i\)
Constants depending on \(\omega \) (\(i=1,\ldots ,4\))

- \({\overline{\zeta }_i}\)
Constants depending on \(\omega \) (\(i=1,\ldots ,4\))

- \(\eta \)
A constant defined as \(\eta =\lambda /\kappa \)

- \(\eta _E\)
A constant defined as \(\eta _E=\lambda _E/\kappa \)

- \(\theta \)
Prescribed rotation on a DCB arm

- \(\kappa \)
A constant depending on the bending stiffness of DCB arms and the softening part of the \(\sigma -\delta \) traction-separation law of the interface

- \(\lambda \)
A constant depending on the bending stiffness of DCB arms and the linear-elastic stiffness of the interface

- \(\lambda _E\)
A constant defined as \(\lambda _E=\sqrt{2}\lambda /2\)

- \(\mu \)
Shear modulus of DCB arms

- \(\nu \)
Poisson’s ratio of DCB arms

- \(\xi _i\)
Constants depending on \(\psi \) (\(i=1,2\))

- \(\rho _i\)
Constants defined as \(\rho _i=\zeta _i/\lambda \) (\(i=1,2\))

- \(\sigma \)
Contact traction at the interface in mode I

- \(\sigma ^L\)
LEFM limit value of \(\sigma \)

- \(\sigma _{max}\)
Maximum value of contact tractions at the interface in mode I

- \(\varphi \)
Cross-sectional rotation of the upper DCB arm

- \(\varphi _1\)
Cross-sectional rotation of the upper DCB arm on the part where the interface is undamaged

- \(\varphi _2\)
Cross-sectional rotation of the upper DCB arm on the part where the interface is damaged

- \(\varphi _1^L\)
LEFM limit value of \(\varphi _1\)

- \(\chi \)
A constant defined as \(\chi =EI/(2{\mu }{A}_s)\)

- \(\psi \)
A constant depending on bending and shear stiffness of DCB arms, and \(\kappa \)

- \(\omega \)
A constant depending on bending and shear stiffness of DCB arms, and \(\lambda \)

## List of Abbreviations

- ASTM
American Society for Testing and Materials

- BS
British Standard

- CBBM
Compliance-based beam method

- CBT
Corrected beam theory

- CCM
Cohesive crack model

- CZM
Cohesive zone model

- DCB
Double cantilever beam

- EBT
Enhanced beam theory

- ESBT
Enhanced simple beam theory

- FE
Finite element

- FEA
Finite element analysis

- ISO
International Organization for Standardization

- LEFM
Linear elastic fracture mechanics

- SBT
Simple beam theory

- TDCB
Tapered double cantilever beam

## 1 Introduction

Since introduced in 1960s by Dugdale (1960) and Barenblatt (1959), the use of cohesive-zone models (CZM) has become one of the most popular ways of describing fracture processes within the research community (Hillerborg et al. 1976; Alfano and Crisfield 2001; Park and Paulino 2011). Nowadays CZMs are widely implemented within the framework of finite-element analysis (FEA) and interface elements, based on CZMs, can be found in element libraries of many commercial softwares for FEA used to solve delamination/debonding problems in 2D (Alfano and Crisfield 2001) and 3D (Park and Paulino 2011). However, because the accuracy of such computations is always dependent on the size of the FE mesh, they can be computationally demanding and suffer from convergence problems.

One of the ways to reduce the computational cost of the analysis was already proposed by the first and third author and consists of using beam finite elements instead of plane solids to model the bulk material of the specimens in 2D analysis of delamination. The results were presented for single-mode (I and II) and mixed-mode delamination problems in geometrically linear (Škec et al. 2015) and non-linear analysis (Škec and Jelenić 2017). Compared to models which use plane solid FEs, the computational efficiency of the beam model was improved (the reduction of total number of degrees of freedom can go up to 40 %) without any significant loss in the accuracy. However, the beam model still suffered from convergence problems and spurious oscillations around the exact solution for cases of brittle interfaces when the mesh is not sufficiently refined. For this reason, it is always very useful when an analytical or semi-analytical solution can be found for cases of engineering interest.

In this paper we focus on mode-I delamination and the double cantilever beam (DCB) test, which is the standard test for determining fracture toughness in mode I. We use quasi-static and geometrically linear analysis where the arms of the DCB are modelled as Timoshenko beams and at the interface, a bi-linear CZM is used. Since the traction–separation law at the interface is composed of two linear parts and a final part with zero tractions, the solution of the problem can be obtained analytically, separately for each part of the interface whose state falls within one of the linear parts of the traction separation law. Because all the quantities of the problem will be expressed exactly with no need for discretisation, the computational cost is negligible and the obtained solutions have no spurious oscillations, which typically occur when delamination problems are solved using FE analysis (Alfano and Crisfield 2001; Blackman et al. 2003b; Škec et al. 2015). However, the idea of using analytical solutions for a DCB is not new and many researchers have contributed to the field in the last 50 years.

The simplest way to analytically model a DCB would be to assume that the arms of the specimen act as if they were cantilever beams clamped at the crack tip. Under this assumption, Benbow and Roesler (1957) used Euler–Bernoulli beam theory to establish the Griffith’s energy balance for a flat-strip specimen where the crack gradually propagates down the middle by holding the specimen in a state of lengthwise compression. During 1960s and early 1970s, Ripling et al. (1971) introduced the DCB and tapered double cantilever beam (TDCB) tests and specimens. They also provided analytical formulae based on Irwin’s energy approach (Irwin 1956) and Timoshenko beam theory (assuming that the arms are clamped at the crack tip) to compute the fracture toughness of the adhesive, which in 1974 became part of the American standard for determining fracture toughness of adhesive joints in mode I. The current version of that standard, ASTM D3433-99 (reapproved in 2012) (ASTM D3433-99 2012), still exclusively uses the same formulae proposed in Ripling et al. (1971). The formula for the DCB used in ASTM D3433-99 (2012) is also used in BS ISO 25217:2009 (2009), where it is called ‘simple beam theory’ (SBT). We will adopt this terminology and use the term ‘simple beam theory’ (SBT) for formulations based on simple beam theories (Euler–Bernoulli or Timoshenko) and the assumption that the DCB arms act as if they were clamped at the crack tip.

However, Ripling et al. (1971) noticed that the SBT formula gives smaller deflections than those obtained from the experiments and they attributed it to not taking into account the rotations of the arms which take place at the crack tip. They also suggested that the results could be simply corrected by increasing the measured crack length by a fixed amount. Although they did not propose a method to obtain this crack length correction, this concept was further developed by other researchers (Blackman et al. 2003a) and became the basis for a data reduction scheme in BS ISO 25217:2009 (2009) called ‘corrected beam theory’ (CBT) based on Euler–Bernoulli beam theory. de Moura et al. (2008) developed the so-called ‘compliance-based beam method’ (CBBM) where the corrected-crack-length concept was used with Timoshenko beam theory.

Kanninen (1973) presented an analytical model for a DCB where the upper arm was modelled as a Euler–Bernoulli beam on elastic foundation (Winkler model), allowing for relative displacements and rotations at the crack tip and ahead of it. The very next year (Kanninen 1974) extended his formulation to account for the shear deformability of the arms and rotational stiffness of the interface, which was accomplished by using Timoshenko beam theory and Pasternak elastic foundation. However, Gehlen et al. (1979) (with Kanninen as the third author) showed that the rotational stiffness of the foundation springs is not relevant in a symmetrical DCB configuration. Kanninen’s model (Kanninen 1974) was later extended by Williams (1989) to account for orthotropic material behaviour. Shahani and Forqani (2004), Shahani and Amini Fasakhodi (2010) developed solutions for a DCB model of finite length consisting of a Timoshenko beam lying on an elastic Winkler foundation for the conditions of fixed force and fixed displacement. Most of the mentioned papers (Kanninen 1973, 1974; Gehlen et al. 1979; Shahani and Forqani 2004; Shahani and Amini Fasakhodi 2010) also investigate the dynamic analysis of unstable crack propagation and arrest in a DCB test. We will refer to the beam-on-elastic-foundation DCB models as ‘enhanced beam theory’ (EBT) models.

In EBT models the interface acts elastically linear up to a certain point where brittle failure occurs. However, fracture processes can usually introduce a certain level of quasi-brittle behaviour, which cannot be captured using EBT models. It is, however, well know that the quasi-brittle behaviour of the interface in a DCB test has an important influence on the structural response before the crack starts to propagate, whereas during the crack propagation the influence of the interface ductility is negligible. One way of introducing a quasi-brittle behaviour of the interface in the model is to use CZMs which account for progressive softening/damage after a certain critical value of the traction at the interface has been reached.

Stigh (1988) developed an analytical solution for a DCB where the arms are modelled as Euler–Bernoulli beams and a bi-linear CZM is embedded at the interface. This solution was revisited and extended to account for a finite length of the specimen and a trapezoidal CZM by Dimitri et al. (2017). de Morais (2013) proposed a solution for a DCB with prescribed rotations (loaded with moments) where the arms are modelled as Timoshenko beams and a bi-linear CZM is embedded at the interface. Unlike Williams (1989) who gave the complete solution for the linear-elastic phase of the interface behaviour, de Morais (2013) took into account only real roots of the characteristic equation of the differential equation of the problem. We discuss this issue in detail in Sect. 2.2. We will refer to the models with quasi-brittle crack as ‘cohesive crack models’ (CCM) (Dimitri et al. 2017).

To the best of authors’ knowledge, a complete analytical solution for a DCB with arms modelled as Timoshenko beams and the interface modelled using a bi-linear CZM, which covers any of the two cases of prescribed rotations and displacement, is not available in the literature. Therefore, one aim of this paper is to fill this gap and to show a clear relationship between SBT, EBT and CCM solutions for a DCB for both Euler–Bernoulli and Timoshenko beam theory.

Furthermore, reducing the general CCM solution down to the SBT solution shows that even in the limit case of LEFM, rotations at the crack tip and in front of it still occur when Timoshenko beam theory is used. Thus, the assumption made in SBT that the arms act as if they were clamped at the crack tip is not applicable even for an infinitely stiff perfectly brittle interface, which is assumed in LEFM. This is somehow an expected result since even an infinitely stiff interface cannot prevent the bulk material of the arms to deform (rotate) around the interface. However, the fact that we can capture this behaviour using Timoshenko beam theory, to the best of authors’ knowledge, has not been addressed in the literature so far. More generally, a comprehensive investigation of the analytical solutions for the LEFM limit is lacking in the literature. Thus, we call this novel approach ‘enhanced simple beam theory’ (ESBT). However, we will show that when the shear deformability of the arms is excluded from the model, which corresponds to Euler–Bernoulli beam theory, ESBT is equivalent to SBT. A second aim of this paper is to determine the interface stresses and the stress resultant profiles in the LEFM cases. We will also show that a novel LEFM-based formula for the determination of the critical energy release rate, \(G_c\), can be derived. This formula takes into account the rotation at the crack tip, unlike those currently available in the standards (ASTM D3433-99 2012; BS ISO 25217:2009 2009).

The outline of the paper is as follows. In Sect. 2, we define the problem and derive the general solutions of differential equations of the problem. In Sects. 3 and 4, we determine the integration constants for the cases of DCB with prescribed rotations and DCB with prescribed displacement, respectively. The solutions are given in a unified and compact form, thus avoiding cumbersome expressions which can be often encountered in such analytical solutions.

The presented general CCM solutions (based on Timoshenko beam theory and a bi-linear CZM at the interface) are then used to derive solutions for different particular cases. EBT solutions presented in Sect. 5 are obtained from CCM solutions by making the interface brittle (removing the softening branch in the traction–separation law of the interface), whereas ESBT (LEFM-based) solutions presented in Sect. 6 are obtained from EBT solutions by letting the interface stiffness go to infinity. The solutions for Euler–Bernoulli beam theory (including CCM, EBT and SBT solutions) given in “Appendix B” are obtained from the Timoshenko beam theory solutions by letting the shear stiffness become infinite.

Results obtained by means of the presented analytical solutions are studied for a number of cases, including sensitivity analyses on the interface strength. In particular, results for a linear interface with brittle behaviour and in the LEFM limit are presented and discussed in Sect. 7.3.

## 2 Problem description

*x*-axis is the centroidal axis of the arm (reference axis), while

*y*and

*z*axes are the principal axes of the arm’s cross section (see Fig. 1).

*x*and

*y*directions. However, for our problem, which is symmetric with respect to the mid-plane of the interface, there is no relative displacement at the interface responsible for mode-II delamination. This is because the upper and the bottom arm of a DCB at the same co-ordinate

*x*have the same, but opposite cross-sectional rotation. Since Timoshenko beam theory is a geometrically linear theory, all points in a single cross-section of the arm experience the same displacement in the

*y*-direction, i.e. \(v(x,y)=v(x)\). Therefore, for a DCB with symmetrical arm deformations, opening (mode-I) relative displacement at the interface, \(\delta (x)\), corresponds to the sum of transverse displacements of both arms. This can be written as

The two types of tests we investigate in this paper are the DCB with prescribed rotations, \(\theta \), and the DCB with prescribed displacement, \(\Delta \), as shown in Fig. 1. In the first case the crack propagates with a constant cohesive zone length (i.e. crack propagation is steady-state) (Suo et al. 1992; Škec et al. 2018), the cohesive zone being where softening/damage of the interface takes place. In the second case the crack propagation is not steady-state, but it tends to being so in the limit of infinitely long cracks, i.e. as \(a\rightarrow \infty \), where \(a\) denotes the crack length. Complete solutions for both cases are given in Sects. 3 and 4, respectively.

For the interface we use a bi-linear CZM consisting of a linear-elastic branch and a linear softening branch, followed by zero tractions for relative displacements greater than the critical value \({\delta }_c\), as shown on the right-hand side in Fig. 2. Here we emphasise that CCM solutions presented in this paper are not general solutions valid for any shape of the traction–separation law of the CZM, but are only valid when the interface behaviour can be assumed as bi-linear with progressive damage. Thus, for a bi-linear CZM law the solution will be obtained for three different phases in the crack propagation process, which is explained in detail in the following subsection.

### 2.1 Three-phase solution

#### 2.1.1 Linear-elastic behaviour

In this phase the entire DCB acts as a linear-elastic body, which means that besides the DCB arms (with a linear-elastic constitutive law), the interface behaviour is linear-elastic, too. For all points at the interface undergoing separation (including points A and B in Fig. 2a), \(\delta (x)<{\delta }_0\) and \(\sigma (x)<\sigma _{max}\), where \({\delta }_0\) and \(\sigma _{max}\) are linear-elastic limit values of the separation \(\delta \) and the traction \(\sigma \) on the interface, respectively. This also means that, as long as \(\delta (x)<{\delta }_0\) and \(\sigma (x)<\sigma _{max}\), no energy dissipation (damage) at the interface occurs and the initial (undeformed and undamaged) configuration can be recovered if the load is removed. Once at the crack tip \(\delta (0)={\delta }_0\), we enter a new phase where damage at the interface starts to develop. This phase is explained in the following section.

It is well known that a zone of compressive stresses (\(\sigma (x)<0\), resulting in \(\delta (x)<0\)) ahead of the crack tip exists in a DCB (Alfano and Crisfield 2001; de Morais 2015; Dimitri et al. 2017). However, our CZM assumes that in compression no softening (or damage) occurs, and the behaviour is still linear elastic. In Fig. 2a we can see that point C experiences negative (compressive) contact tractions and negative relative displacements at the interface. For a zero-thickness adhesive layer this results in a non-physical overlapping of the arms, which in our model is allowed and does not create any additional stresses. But because the interface thickness in our model does not influence the results (although it is well know that in reality a different thickness of the same adhesive results in a different structural behaviour) and we can assume that the interface is sufficiently thick to prevent the arms coming in direct contact.

#### 2.1.2 Damage growth before crack propagation

In this phase, the interface can be divided in two zones. In the first zone, just ahead of the crack tip, damage is developing (although the crack is still not propagating) and \(\delta (x)\in [{\delta }_0,{\delta }_c)\), where \(x\in (-L_{cz},0]\). In the second zone interface behaviour is again linear-elastic, i.e. \(\delta (x)<{\delta }_0\), where \(x>0\). Note that according to Fig. 2b, \(x=-L_{cz}\) corresponds to the initial crack tip and \(x=0\) corresponds to the point where the first and the second zone meet. In this phase, the damage builds up with loading and \(L_{cz}\) increases from 0 (corresponding to \(\delta (-L_{cz})=\delta (0)={\delta }_0\)) to a certain limit value (corresponding to \(\delta (-L_{cz})={\delta }_c\)) at which the crack begins to propagate. Note that, according to Fig. 2b, the \(\sigma -\delta \) relationship in the first zone is also linear, but with softening (damage).

#### 2.1.3 Crack propagation

This phase is similar to the previous phase (Sect. 2.1.2) with the difference that here the relative displacement at the current crack tip, \(\delta (-L_{cz})={\delta }_c\), remains unchanged during the whole phase. For a DCB with prescribed rotations the cohesive zone length remains constant for any position of the crack (steady-state crack propagation), while for a DCB with prescribed displacement the cohesive zone length will change (non-steady-state crack propagation). Thus, when the crack propagation is steady state the deformed shape of the interface shown in Fig. 2c, and the contact traction distribution over the interface, \(\sigma (x)\), simply translate to the right-hand side. This is why point A in Fig. 2c is the point at the interface where the crack tip is currently located. When the crack propagation is not steady-state, we still have \(\delta (-L_{cz})={\delta }_c\) and \(\delta (0)={\delta }_0\) for any position of the crack tip, but the deformed shape and the contact traction distribution at the interface (including \(L_{cz}\)) change as the crack propagates. The part of the interface which has been completely damaged (\(\sigma (x)=0\)) is excluded from the domain of the solution for \(v_2(x)\) and becomes a part of the DCB arm.

### 2.2 Solutions of differential equations of the problem

#### 2.2.1 Solution on the undamaged part of the interface

*r*is a constant, results in a characteristic equation with four roots, namely

- 1.The solution of Eq. (7) for \(\omega >1\):where \(C_1\), \(C_2\), \({\overline{C}}_1\) and \({\overline{C}}_2\) are integration constants.$$\begin{aligned} v_1(x)= & {} e^{-\lambda \zeta _1 x}C_1+e^{-\lambda \zeta _2 x}C_2+e^{\lambda \zeta _1 x}{\overline{C}}_1\nonumber \\&+\,e^{\lambda \zeta _2 x}{\overline{C}}_2, \quad x\ge 0, \end{aligned}$$(13)
- 2.The solution of Eq. (7) for \(\omega <1\):where$$\begin{aligned}&v_1(x) = e^{-\lambda \zeta _3 x}\left[ \sin (\lambda \ \zeta _4\ x)C_3+\cos (\lambda \ \zeta _4\ x)C_4\right] \nonumber \\&\quad +\,e^{\lambda \zeta _3 x}\left[ \sin (\lambda \ \zeta _4\ x){\overline{C}}_3+\cos (\lambda \ \zeta _4\ x){\overline{C}}_4\right] , \quad x\ge 0, \end{aligned}$$(14)and \(C_3\), \(C_4\), \({\overline{C}}_3\) and \({\overline{C}}_4\) are integration constants. Note that because of$$\begin{aligned} \zeta _3= \sqrt{\frac{1+\omega }{2}}, \quad \zeta _4=\sqrt{\frac{1-\omega }{2}}, \end{aligned}$$(15)we have$$\begin{aligned} \zeta _1\ \zeta _2=1, \quad \text {and} \quad (\zeta _1+\zeta _2)^2=2(\omega +1), \end{aligned}$$(16)$$\begin{aligned} \zeta _3=\frac{\zeta _1+\zeta _2}{2}, \quad \zeta _4=i\frac{\zeta _2-\zeta _1}{2}. \end{aligned}$$(17)
- 3.The solution of Eq. (7) for \(\omega =1\):where \(C_5\), \(C_6\), \({\overline{C}}_5\) and \({\overline{C}}_6\) are integration constants.$$\begin{aligned} v_1(x)= & {} e^{-\lambda x}(C_5+C_6\ x)+e^{\lambda x}({\overline{C}}_5\nonumber \\&\quad +\,{\overline{C}}_6\ x), \quad x\ge 0, \end{aligned}$$(18)

### Remark 2.1

For the sake of simplicity and because of the extreme unlikelihood that the value \(\omega =1\) occurs in a real case, the solutions in the following sections are given only for the cases when \(\omega >1\) and \(\omega <1\). However, the results for \(\omega =1\) are given separately in “Appendix A” for completeness. In the numerical examples presented in Sect. 7 we will show that, unlike stated in de Morais (2015), all solutions from Eq. (20) are possible for realistic values of geometrical and material properties of a DCB. \(\square \)

#### 2.2.2 Solution on the damaged part of the interface

*t*is a constant, results in a characteristic equation with four roots, namely

In the following sections, the problem is solved and the integration constants are determined for each phase for a DCB with either prescribed rotations or prescribed displacement.

## 3 DCB with prescribed rotations

Consider a DCB with prescribed rotations as illustrated in Fig. 1a. At the left-hand end of each arm an equal, but opposite rotation \(\theta \) is prescribed, causing the opening of the DCB along the interface due to equal, but opposite concentrated moments, \(M\) acting at the point of the prescribed rotation. This implies that the disconnected parts of the DCB arms are in pure bending and the shear force at the crack tip is zero during all phases. Each of the three solution phases is derived in detail in the following sections.

### 3.1 Linear-elastic phase

We will use Eq. (36) as a general solution for the crack mouth opening displacement of a DCB, which is valid for all three solutions phases, not only for a DCB with prescribed rotations. However, in general \(\Delta _a\), \(\Delta _{CT}^{\varphi }\) and \(\Delta _{CT}^{\delta }\) are computed differently each time.

The final forms of functions \(v_1(x)\), \(v_1^{\prime }(x)\), \(\varphi _1(x)\), \({{\mathcal {M}}}_1(x)\), \({{\mathcal {T}}}_1(x)\) and \(\sigma (x)\) for this phase are given in “Appendix E.1”.

### 3.2 Phase of damage growth before crack propagation

In this phase, a cohesive (or damage-process) zone is developing in front of the crack tip. As already mentioned, for the cohesive zone (\(x\in [-L_{cz},0]\)) we use the solution (23), whereas for the zone of linear-elastic behaviour (\(x\ge 0\)) we use the solution (20).

### 3.3 Crack propagation phase

In the previous phase (damage growth before crack propagation) the applied moment is increasing from \(M_L\) to \(M({\overline{L}}_{cz})={M}_{max}\). Since for a DCB with prescribed rotations there is no shear force at the crack tip during all phases, it means that only the applied moment *M* (same at the crack tip as at the point of application) is responsible for crack propagation. Obviously, the crack will propagate when the applied moment reaches the value \({M}_{max}\) and this value will not change as the crack propagates. A constant value of \({M}_{max}\) during crack propagation implies that the boundary conditions at the crack tip remain constant and that \(L_{cz}={\overline{L}}_{cz}\), during crack propagation. This kind of behaviour is known as ‘steady-state crack propagation’. Thus, unlike in the previous phase, in this phase \(L_{cz}\) is not a variable.

*a*can go to infinity. Note also that \(v_2(-{\overline{L}}_{cz})\) and \(v_2^{\prime }(-{\overline{L}}_{cz})\) are constants and thus the function \(\Delta (a)\) in this phase is quadratic. Because in this phase \(M\) does not change, \(\Delta \) does not depend on \(M\) and we cannot define \(\Delta (M)\).

## 4 DCB with prescribed displacement

In this section we consider a DCB with prescribed displacement where, according to Fig. 1b, at the left-hand end the bottom arm is pinned, whereas the upper arm is pulled upwards. In order to prescribe a displacement \(\Delta \) at the left-hand side of the upper arm, a vertical force \(F\) must be applied at the same place and in the same direction. Thus, unlike in the case of a DCB with prescribed rotations, at the cracked portion of a DCB with prescribed displacement there is bending and shear in the arms, which will make the problem slightly more complex. Furthermore, because the crack propagation in the case of a DCB with prescribed displacement is not steady-state (\(L_{cz}\) changes during crack propagation), the solution for the third phase (crack propagation) will be also more complex, compared to the case of DCB with a prescribed rotations where \(L_{cz}={\overline{L}}_{cz}\) is constant during crack propagation. Each phase of the solution is explained in detail in following sections.

### 4.1 Linear-elastic phase

### 4.2 Phase of damage growth before crack propagation

As explained in Sect. 3.2, in this phase \(L_{cz}\) grows from 0 to a value corresponding to the initiation of crack propagation, i.e. transition to the third phase. This is a maximum value for \(L_{cz}\), because during crack propagation \(L_{cz}\) decreases and asymptotically tends to a minimum value when \(a\rightarrow \infty \), as is discussed in the next section. Therefore, this initial maximum value of \(L_{cz}\) at the initiation of crack propagation will be denoted by \(L_{cz}^{max}\). In order to obtain this value, the same approach as for a DCB with prescribed rotations is followed, i.e. condition (58) is imposed. This is again a highly non-linear equation in terms of \(L_{cz}\), which is solved numerically (in our approach Newton-Raphson procedure is used).

### 4.3 Crack propagation phase

As previously mentioned, in the case of a DCB with prescribed displacement, the cohesive zone length decreases during crack propagation from \(L_{cz}^{max}\) asymptotically approaching a lower limit value \(L_{cz}^{min}\). This means that the crack propagation is not steady state, but it approaches steady state for infinitely long cracks (Dimitri et al. 2017). Because \(L_{cz}^{min}\) corresponds to a steady-state crack propagation, it must have the same value as \({\overline{L}}_{cz}\) found in the identical DCB loaded with prescribed rotations, where crack propagation is always steady state, i.e. \(L_{cz}^{min}={\overline{L}}_{cz}\). Note that both \(L_{cz}^{max}\) and \(L_{cz}^{min}\) are obtained by numerically solving Eq. (58), where constants \(D_i\) (\(i=1,\ldots ,4\)) in (23) in the former case are computed using (81) in (50), whereas in the latter case they are computed using (55) in (50).

### Remark 4.1

- 1.
Solutions for Euler–Bernoulli beam theory. These solutions are obtained by simply letting the shear modulus \(\mu \rightarrow \infty \) and are presented in “Appendix B”.

- 2.
Solutions for a linear-elastic interface with brittle failure. These solutions, presented in Sect. 5 for Timoshenko beam theory and in “Appendix C” for Euler–Bernoulli beam theory, are obtained by letting \({\delta }_c\rightarrow {\delta }_0\), which means that there is no damage before crack propagation, i.e. only the first and the third solution phases remain.

- 3.
LEFM solutions. These solutions, given both for Timoshenko beam theory in Sect. 6 and Euler–Bernoulli beam theory in “Appendix C”, are obtained from the solutions for a linear-elastic CZM with brittle failure by letting \({\delta }_0\rightarrow 0\).

## 5 Analytical solutions for a DCB with linear-elastic interface with brittle failure (EBT)

Here we assume that the interface behaves as linear-elastic up to a certain point after which brittle failure (instantaneous loss of cohesion) occurs. In our model, this is achieved by setting \({\delta }_0={\delta }_c\), which removes the softening branch in \(\sigma -\delta \) diagram shown in Fig. 2. This also means that the second part of the solution (damage growth before crack propagation) does not exist and that \(L_{cz}=0\) during crack propagation.

We will now derive the solution for a DCB with a brittle interface, first for a DCB with prescribed rotations and then for a DCB with prescribed displacement.

### 5.1 DCB with prescribed rotations

### 5.2 DCB with prescribed displacement

### Remark 5.1

Note that presented solutions in terms of the applied load and the crack mouth opening displacement of a DCB with either prescribed rotations or prescribed displacement for the case of a linear-elastic interface with a brittle failure (EBT) are valid for all values of \(\omega \) (\(\omega <1\), \(\omega =1\) and \(\omega >1\)). \(\square \)

## 6 LEFM solutions for a DCB with prescribed rotations and a DCB with prescribed displacement: enhanced simple beam theory(ESBT)

The presented model for a linear-elastic interface with a brittle crack still allows some opening at the interface in the linear-elastic range before crack starts to propagate (\(\delta (x)<{\delta }_0\)). If this initial linear-elastic behaviour is excluded from the model by letting \({\delta }_0\rightarrow 0\), while keeping \(G_c\) constant (which also results in \(\sigma _{max}\rightarrow \infty \)), we obtain solutions equivalent to those given by linear-elastic fracture mechanics (LEFM). However, we will not a-priori assume that the DCB arms act as if they were clamped at the crack tip, which is usually done in the SBT approach. In this way we will show that in the limit case of LEFM the arms rotate at the crack tip and in front of it (even though their centre-lines there remain straight), which means that the clamped conditions at the crack tip cannot be obtained even for an infinitely stiff perfectly brittle interface. This is due exclusively to the shear deformability of the arms, which is accounted for in Timoshenko beam theory. For Euler–Bernoulli beam theory, the limit case of LEFM indeed corresponds to SBT and the arms do not rotate at the crack tip and in front of it. Thus, for our LEFM-limit solution for Timoshenko beam theory we will adopt the term ‘enhanced simple beam theory’ (ESBT). In the following subsections we will present only the final results for a DCB with prescribed rotations and a DCB with prescribed displacement, respectively, while the complete derivation is given in “Appendix E”.

### 6.1 DCB with prescribed rotations

Moreover, because the DCB arms deform (rotate) in front of crack tip, contact tractions at the interface, as well as the shear forces and bending moments in the arms, appear in front of the crack tip, with an exponential decay as \(x\rightarrow \infty \). We can notice that at the crack tip there is a jump in the shear force (from 0 to \(-M\sqrt{{\mu }{A}_s/EI}\)) which corresponds to a transition in the bending moment diagram from the constant value \(M\) in the cracked portion of the arms to the function (103) in front of the crack tip. This implies that in the limit case of LEFM there is a concentrated transversal cohesive force exchanged at the crack tip, so that the interface stress is the sum of a compressive smooth part and the Dirac distribution centred at zero. Because at the crack tip the cross-sectional rotations of the arms must be continuous and there is a jump in the shear force in the arms, the function \(v_1^{\prime }(x)\) is also discontinuous due to \(\varphi _1(x)=v_1^{\prime }(x)+{{\mathcal {T}}}_1(x)/{\mu }{A}_s\).

### 6.2 DCB with prescribed displacement

Note that expressions (107)–(112) are valid only for the phase of linear-elastic behaviour before crack propagation. However, analogous expressions for the crack propagation phase can be obtained by substituting \(F\) with \(F(a)\) (defined in (116)) and \({a}_0\) with \(a\).

### Remark 6.1

In “Appendix E.3” we show that if Euler–Bernoulli beam theory is used, there are no cross-sectional rotations, shear forces or bending moments of the arms in front of the crack tip. However, singularity of contact tractions at the interface, as well as discontinuity of the shear stresses and bending moments in the arms, take place at the crack tip. These conditions are equivalent to clamping DCB arms at the crack tip and explain why the formulae obtained for the limit case of LEFM in “Appendix D” indeed correspond to widely used formulae in SBT. \(\square \)

## 7 Numerical examples

In this section, for a DCB with a bi-linear CZM at the interface, the analytical solutions derived in this paper using Timoshenko beam theory will be compared to the numerical results obtained with an equivalent finite-element (FE) model, in which the same beam theory and CZM are used (Škec et al. 2015), and to the Euler–Bernoulli beam theory analytical solutions derived in “Appendix B”. The latter will allow us to investigate the influence of shear deformability of DCB arms on the results. LEFM solutions obtained in Sect. 6 will also be presented as limit cases for a brittle interface.

Material data used in the numerical examples. Except in the last example, \(\alpha =0.01\)

| \(\nu \) (–) | \(k_s\) (–) | \(\Omega \) (N/mm) | \(\sigma _{max}\) (MPa) | \({\delta }_c\) (mm) | \({\delta }_0\) (mm) |
---|---|---|---|---|---|---|

70 | 1/3 | 5/6 | 1 | \(\{7.5, 15, 30, 60, 120\}\) | \(2\ \Omega /\sigma _{max}\) | \(\alpha \ {\delta }_c\) |

The numerical model used is the multi-layer beam model presented in Škec et al. (2015) where we assume a total length of the specimen \(L=200\) mm. A total number of 2000 2-node Timoshenko beam elements are distributed evenly over the upper half of the DCB, meaning that the element length is 0.1 mm. Such a fine mesh is used to eliminate or at least minimise the influence of discretisation-caused spurious oscillations on the results (Alfano and Crisfield 2001; Škec et al. 2015). A 4-node interface element is attached to every beam FE from \(x={a}_0\) to \(x=L\) making a total of 1700 interface elements. The solution is obtained using displacement control and Newton-Raphson iterative procedure. Because our numerical model has 4002 degrees of freedom (one transverse displacement and one cross-sectional rotation per node), in each iteration of each increment, 4002 linear equations are solved in order to obtain the cross-head displacement. In our analytical solution, we obtain the cross-head displacement from a single closed-form solution. The same applies to any other quantity we want to obtain. Furthermore, all analytical solutions, unlike the numerical ones, are perfectly smooth.

It is worth noting that the values reported in Table 1 according to (9)\(_2\) for \(\sigma _{max}=\{7.5,15,30,60,120\}\) MPa give \(\omega =\{0.32,0.64,1.28,2.57,5.13\}\). Thus, we can deduce that in real-life applications both \(\omega <1\) and \(\omega >1\) are possible and therefore the analytical solution should account for both cases.

In the following sections we will present the results for the DCB first with prescribed rotations and then with prescribed displacement.

### 7.1 DCB with prescribed rotations

From Fig. 4a it can be clearly seen that the crack will start to propagate sooner (i.e. for smaller crack mouth opening displacements) when the interface is more brittle. The numerical model again agrees perfectly with the analytical solution, and there are some differences between Euler–Bernoulli and Timoshenko beam theory, which again become more significant as \(\sigma _{max}\) increases.

In Fig. 4b we can finally compare the cohesive zone lengths, \(L_{cz}\), for Timoshenko and Euler–Bernoulli beam theories. As expected, \(L_{cz}\), which is highly influenced by the value of \(\sigma _{max}\), remains constant during crack propagation. Differences between Timoshenko and Euler–Bernoulli beam theory solutions are now even more pronounced, especially for more brittle cases. Again, the numerical results match perfectly with those obtained from the analytical solution for Timoshenko beam theory.

### 7.2 DCB with prescribed displacement

In Fig. 5, the reaction force, \(F\), is plotted against the prescribed crack mouth opening displacement, \(\Delta \), for different values of \(\sigma _{max}\). Both Timoshenko and Euler–Bernoulli beam theories are used. We can see that \(\sigma _{max}\) has a considerable influence on the results in the first two phases (before crack starts to propagate), especially in the second phase, when the stiffness of the DCB progressively decreases as damage is developing in front of the crack tip. This also results in a reduction of the peak force. In Fig. 5b it can again be seen that differences between Timoshenko and Euler–Bernoulli beam theory become more pronounced as \(\sigma _{max}\) increases, i.e. the interface becomes more brittle. Results from the FE model based on the Timoshenko beam theory perfectly match the results obtained from the analytical solution. In the third phase (crack propagation) all curves are extremely close, but they do not coincide perfectly. This is better shown in Fig. 6a, where it can be noted that the differences of the results in the crack propagation phase indeed exist, but they are too small to be appreciated on a normal scale.

In fact, a simple argument explains why the curve corresponding to a finite value of \(\sigma _{max}\) must lie above the LEFM solution after a point which coincides with the start of crack propagation. For finite values of \(\sigma _{max}\), the gradual development of the cohesive zone ahead of the initial crack tip, before the crack starts propagating, is responsible for the nonlinear deviation of the load-displacement curve from the initial straight line of the LEFM solution, culminating in the rounded part so that, the lower \(\sigma _{max}\), the lesser the peak load with respect to the theoretical peak load predicted by LEFM. In terms of energy, this means that, in the case of a finite value of \(\sigma _{max}\), before the crack starts propagating, less external work is performed by the external force, \(F\), than for the LEFM case. However, once the DCB is completely separated, the total amount of external work must be equal to the interface area times the area under the traction–separation law, that is the work of separation, \(\Omega \). Therefore, for a DCB with an infinite length, the only way that the total external work spent is the same for the two cases is that, for a finite value of \(\sigma _{max}\), the curve lies above the case for \(\sigma _{max}\rightarrow \infty \) during crack propagation, so that the increase in external work in this part of the curve compensates the lower amount of external work before crack propagation.

In Fig. 5 we noted that \(\sigma _{max}\) significantly influences the peak load reached before the crack starts to propagate and dictates how far from brittle behaviour (LEFM) the considered CZM solution is. However, changing the ratio, \(\alpha \), between \({\delta }_0\) and \({\delta }_c\) has a noticeable influence on the results, too. We will assume that \(\alpha \) can vary between a value very close to 0 (meaning that \({\delta }_0\) is almost negligible compared to \({\delta }_c\)) and 1 (meaning that \({\delta }_0={\delta }_c\)). The latter case, which is covered in Sect. 5.2, implies that we have a linear-elastic behaviour at the interface up to \({\delta }_0\) (or \({\delta }_c\)) followed by brittle failure (leading to \(L_{cz}=0\)).

### 7.3 Behaviour in front of the crack tip for a DCB with linear-elastic interface with brittle crack and in the limit case of LEFM

In Sect. 6 (see also “Appendix E”) we showed that in the limit case of LEFM stresses and strains are found in front of the crack tip when Timoshenko beam theory is used to model the arms. In “Appendix E.3” we show that this is not the case when Euler–Bernoulli beam theory is used to model the arms. In this section, using the same geometrical and material data for the bulk material as in the previous examples, we will show that the behaviour of a DCB with a linear-elastic interface with brittle crack approaches the behaviour described in Sect. 6 for the limit case of LEFM as the stiffness of the interface increases. We will consider only the case of a DCB with prescribed displacement and investigate the case when the crack starts to propagate, which means that in Eqs. (107)–(112) we use \(F(a)\) (defined in (116)) and \(a\) instead of \(F\) and \({a}_0\). However, in this example we will assume that \(a={a}_0=30\) mm, which means that we will investigate the case when the crack starts to propagate from its initial position. Note that, because the crack propagation for a DCB with prescribed displacement is not steady state, the presented results would change for \(a>{a}_0\), eventually approaching the steady-state solutions for \(a\rightarrow \infty \). These solutions are given by Eqs. (98)–(103) for a DCB with prescribed rotations (where \({M}_{max}\) should be used instead of \(M\) for the crack propagation phase).

In this example we are again assuming that \(\Omega =1\) N/mm, where \(\Omega =\sigma _{max}\ {\delta }_0/2\). The values of \(\sigma _{max}\) are varied according to \(\sigma _{max}=10^i\) MPa, where \(i=0,1,2,3,4\), and the values of \({\delta }_0\) follow from \({\delta }_0=2\ \Omega /\sigma _{max}\).

It is worth noting that using Euler–Bernoulli beam theory in the limit case of LEFM we have \(\varphi _1(x)=\sigma (x)={{\mathcal {T}}}_1(x)={{\mathcal {M}}}_1(x)=0\) for \(x>0\), but \(\varphi _1(0)=0\), \(\sigma (0)=\infty \), \({{\mathcal {T}}}_1(0)=F(a)\) and \({{\mathcal {M}}}_1(0)=F(a)a\) at the crack tip (\(x=0\)).

## 8 Conclusions

In this paper we have derived complete analytical solutions for DCB specimens where the arms are modelled using simple beam theories (Timoshenko or Euler–Bernoulli). At the interface, three different models have been assumed: (i) a quasi-brittle bi-linear CZM, (ii) a liner-elastic CZM with brittle failure and (iii) a perfectly brittle and infinitely stiff interface (corresponding to LEFM solutions). The models obtained for each mentioned type of interface are called ‘cohesive crack model’ (CCM), ‘enhanced beam theory’ (EBT) and ‘enhanced simple beam theory’ (ESBT), respectively. In our approach EBT solutions are obtained from CCM solutions by removing the softening branch (responsible for progressive damage) from the CZM. From there, ESBT solutions can be obtained by letting the interface stiffness go to infinity. We have introduced a new term ESBT because we show that in the limit case of LEFM, EBT model does not correspond to ‘standard beam theory’ (SBT) model where the DCB arms act as if they were clamped at the crack tip. In ESBT, the arms are allowed to rotate at and in front of the crack tip, which is due exclusively to the shear deformability of the arms. This is why for the case when the arms are not shear-deformable (Euler–Bernoulli beam theory), ESBT corresponds to SBT. We have also derived the CCM, EBT and SBT solutions for the case when Euler–Bernoulli beam theory is used to model the arms by simply letting the shear stiffness of the arms go to infinity. All the mentioned solutions are derived for two different types of DCB: one with prescribed rotations and one with prescribed displacement. The presented solutions allow us to easily compute the crack mouth opening displacement, applied load (moment or force), contact tractions at the interface, displacement and rotations of the arms ahead of the crack tip, and shear forces and bending moments in the arms for different DCB models.

- 1.
The complete analytical solutions given in a unified and compact form for DCBs (either with prescribed rotations or prescribed displacement) with a bi-linear CZM at the interface.

- 2.
The complete analytical solutions given in a unified and compact form for DCBs (either with prescribed rotations or prescribed displacement) with linear-elastic interface with brittle failure. It is shown that such EBT solutions, compared to CMM solutions, are very accurate in the phase of crack propagation. However, they are not able to capture the quasi-brittle behaviour before crack propagation.

- 3.
The solutions for the limit case of LEFM which, in the case of Timoshenko beam theory (ESBT), show that at and in front of the crack tip the arms are allowed to rotate even if the interface is infinitely stiff. This implies that Timoshenko beam theory is, in a way, capable of capturing the realistic behaviour of the DCB in front of the crack tip and that the widely used assumption that in LEFM the arms act as if they were clamped at the crack is not necessary. However, we show that this assumption is still valid when the arms are modelled as Euler–Bernoulli beams.

- 4.
An expression for the critical energy release rate, \(G_c\), for a DCB with prescribed displacement is proposed, which, to the best of authors’ knowledge, is original. This expression, compared to the formula derived under the assumption that the arms are clamped at the crack tip, has an additional term which is dependent on the shear deformability of the arms and accounts for the rotations of the arms at the crack tip. If the arms are non-deformable in shear (Euler–Bernoulli beam theory), the expression for \(G_c\) corresponds to the one obtained under the assumptions that the arms are clamped at the crack tip.

It is also worth noting that, using the approach presented in this paper, obtaining the analytical solution for a DCB with trapezoidal CZM at the interface using the Timoshenko beam theory to model the arms should be straight-forward and will also be covered in future work.

## 9 Free software made available

All the results are implemented in a software application with a user-friendly graphic interface where Euler–Bernoulli or Timoshenko beam theory for the arms and CCM, EBT or ESBT models for the interface can be selected. Results of the analysis can be plotted and exported. Because the computations are based on the presented analytical solutions, the results in our software are obtained instantaneously, even on a regular laptop computer. The software is free and can be downloaded at http://dx.doi.org/10.17633/rd.brunel.7223795.

## 10 Supplementary data

Supplementary material related to this article can be found online at http://dx.doi.org/10.17633/rd.brunel.7212218.

## Notes

### Acknowledgements

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie Grant Agreement No. 701032. The third author wishes to acknowledge the financial support of the Croatian Science Foundation (Research Project IP-2013-11-1631).

## References

- Alfano G, Crisfield MA (2001) Finite element interface models for the delamination analysis of laminated composites: mechanical and computational issues. Int J Numer Methods Eng 50(7):1701–1736CrossRefGoogle Scholar
- ASTM D3433-99 (2012) Standard test method for fracture strength in cleavage of adhesives in bonded metal joints. ASTM International (2012)Google Scholar
- Barenblatt GI (1959) The formation of equilibrium cracks during brittle fracture—general ideas and hypothesis, axially symmetric cracks. J Appl Math Mech 23(3):622–636CrossRefGoogle Scholar
- Benbow JJ, Roesler FC (1957) Experiments on controlled fractures. Proc Phys Soc 70(B):201–211CrossRefGoogle Scholar
- Biel A, Stigh U (2008) Effects of constitutive parameters on the accuracy of measured fracture energy using the DCB-specimen. Eng Fract Mech 75:2968–2983CrossRefGoogle Scholar
- Blackman BRK, Kinloch AJ, Paraschi M, Teo WS (2003a) Measuring the mode I adhesive fracture energy, \({G}_{IC}\), of structural adhesive joints: the results of an international round-robin. Int J Adhes Adhes 23:293–305CrossRefGoogle Scholar
- Blackman BRK, Hadavinia H, Kinloch AJ, Williams JG (2003b) The use of a cohesive zone model to study the fracture of fibre composites and adhesively-bonded joints. Int J Fract 119:25–46CrossRefGoogle Scholar
- BS ISO 25217:2009 (2009) Adhesives—determination of the mode I adhesive fracture energy of structural adhesive joints using double cantilever beam and tapered double cantilever beam specimens. British StandardGoogle Scholar
- de Morais AB (2013) Mode I cohesive zone model for delamination in composite beams. Eng Fract Mech 109:236–245CrossRefGoogle Scholar
- de Morais AB (2015) Analysis of the metal adhesively bonded double cantilever beam specimen. Int J Adhes Adhes 61:8–14CrossRefGoogle Scholar
- de Moura MFSF, Campilho RDSG, Gonçalves JPM (2008) Crack equivalent concept applied to the fracture characterization of bonded joints under pure mode I loading. Compos Sci Technol 68:2224–2230CrossRefGoogle Scholar
- Dimitri R, Cornetti P, Mantič V, Trullo M, De Lorenzis L (2017) Mode-I debonding of a double cantilever beam: a comparison between cohesive crack modeling and finite fracture mechanics. Int J Solids Struct 124:57–72CrossRefGoogle Scholar
- Dugdale DS (1960) Yielding of steel sheets containing slits. J Mech Phys Solids 8:100–104CrossRefGoogle Scholar
- Freiman SW, Mulville DR, Mast PW (1973) Crack propagation studies in brittle materials. J Mater Sci 8:1527–1533CrossRefGoogle Scholar
- Gehlen PC, Popelar CH, Kanninen MF (1979) Modeling of dynamic crack propagation: I. validation of one-dimensional analysis. Int J Fract 15(3):281–294Google Scholar
- Hillerborg A, Modéer M, Petersson PE (1976) Analysis of crack formation and crack growth in concrete by means of fracture mechanics and finite elements. Cem Concr Res (Sec. 12.2) 6:773–782CrossRefGoogle Scholar
- Irwin GR (1956) Onset of fast crack propagation in high strength steel and aluminuim alloys. Sagmore Res Conf Proc 2:289–305Google Scholar
- Kanninen MF (1973) An augmented double cantilever beam model for studying crack propagation and arrest. Int J Fract 9(1):83–92Google Scholar
- Kanninen MF (1974) A dynamic analysis of unstable crack propagation and arrest in the DCB test specimen. Int J Fract 10(3):415–430CrossRefGoogle Scholar
- Park K, Paulino GH (2011) Cohesive zone models: a critical review of traction-separation relationships across fracture surfaces. Appl Mech Rev 64:060802-1–060802-20Google Scholar
- Rice JR (1968) A path independent integral and the approximate analysis of strain concentration by notches and cracks. J Appl Mech 35:379–386CrossRefGoogle Scholar
- Ripling EJ, Mostovoy S, Corten HT (1971) Fracture mechanics: a tool for evaluating structural adhesives. J Adhes 3(2):107–123CrossRefGoogle Scholar
- Shahani AR, Amini Fasakhodi MR (2010) Analytical modelling of dynamic fracture and crack arrest in DCB specimens under fixed displacement conditions. Fatigue Fract Eng Mater Struct 33:436–451CrossRefGoogle Scholar
- Shahani AR, Forqani M (2004) Static and dynamic fracture mechanics analysis of a DCB specimen considering shear deformation effects. Int J Solids Struct 41:3793–3807CrossRefGoogle Scholar
- Škec L, Jelenić G (2017) Geometrically non-linear multi-layer beam with interconnection allowing for mixed-mode delamination. Eng Fract Mech 169:1–17CrossRefGoogle Scholar
- Škec L, Jelenić G, Lustig N (2015) Mixed-mode delamination in 2D layered beam finite elements. Int J Numer Methods Eng 104:767–788CrossRefGoogle Scholar
- Škec L, Alfano G, Jelenić G (2018) On \({G}_c\), \({J}_c\) and the characterisation of the mode-I fracture resistance in delamination or adhesive debonding. Int J Solids Struct 144–145:100–122Google Scholar
- Sørensen BF, Brethe P, Skov-Hansen P (1996) Controlled crack growth in ceramics: the DCB specimen loaded with pure moments. J Eur Ceram Soc 16:1021–1025CrossRefGoogle Scholar
- Stigh U (1988) Damage and crack growth anlysis of the double cantilever beam specimen. Int J Fract 37:R13–R18CrossRefGoogle Scholar
- Suo Z, Bao G, Fan B (1992) Delamination R-curve phenomena due to damage. J Mech Phys Solids 40(1):1–16CrossRefGoogle Scholar
- Williams JG (1989) End corrections for orthotropic DCB specimens. Compos Sci Technol 35:367–376CrossRefGoogle Scholar

## Copyright information

**Open Access**This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.