Accepted Manuscript

Displacement discontinuity method for cohesive crack propagation

E. Chen, Christopher K.Y. Leung

PII: S0013-7944(17)30611-2
DOI: https://doi.org/10.1016/j.engfracmech.2017.11.009
Reference: EFM 5747

To appear in: Engineering Fracture Mechanics

Received Date: 10 June 2017

Revised Date: 1 November 2017
Accepted Date: 3 November 2017

Please cite this article as: Chen, E., Leung, C.K.Y., Displacement discontinuity method for cohesive crack
propagation, Engineering Fracture Mechanics (2017), doi: https://doi.org/10.1016/j.engfracmech.2017.11.009

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
 Displacement discontinuity method for cohesive crack propagation

E Chen1,*; Christopher K.Y. Leung2

1
Post-doc, Dept. of Civil and Environmental Engineering, Hong Kong University of

Science and Technology, Hong Kong, China

2
Professor, Dept. of Civil and Environmental Engineering, Hong Kong University of

Science and Technology, Hong Kong, China

Abstract: Displacement discontinuity method (DDM) can conveniently solve fracture

mechanics problems because the discontinuous displacement field across the fracture

surfaces can be explicitly captured by the displacement discontinuity elements.

However, the potential of DDM on solving crack propagation problems in quasi-

brittle materials has not been covered in the past. In this paper, the DDM is extended

to take into account the fracture process zone in quasi-brittle materials such as

concrete by incorporating an iterative algorithm to account for the cohesive stress vs

crack opening relation. Tensile strength criterion and fracture toughness criterion for

crack propagation are used respectively for two corresponding types of crack tip

element-parabolic displacement discontinuity element (PDDE) and root-r

displacement discontinuity element (RRE). Numerical of the wedge splitting test and

three-point bending of notched beam are performed. The results are found to compare

well with experimental and finite element results simulations.

Keywords: Displacement discontinuity method; Cohesive crack propagation; Tensile

1
strength criterion; Fracture toughness criterion

*Corresponding author.

Email address: teresachan@connect.ust.hk (E Chen)

2
1. Introduction

The boundary element approach uses only the boundary values of the problem to

establish the integral equations, which can provide the complete solution of the

problem [1]. In the finite element method, differential equations are set up in terms of

the unknown displacements of elements’ nodes of the entire domain [2]. Therefore,

the boundary element method (BEM) reduces the dimensionality of the problem [1].

In BEM, for a boundary value problem, some parameters such as stresses or

displacements at parts of the boundary are defined as boundary conditions, and others

are not, which emerge as part of the solution to the problem. In the direct BEM,

boundary unknown parameters are solved directly, and once all the boundary

unknowns are obtained, the stresses and displacements inside the domain of the

problem can be calculated by applying the governing equations to the interior points

[1]. Besides solving for the unknown boundary displacements or stresses directly, an

alternative approach to establish the governing equations is to apply fictitious

variables on each boundary element, and matching the resultant stress or displacement

at each element contributed by all fictitious variables to the prescribed boundary

conditions [3]. This approach is referred to as the indirect boundary element method,

because the fictitious variables are first solved from the governing equations and then

boundary unknowns are calculated by summing up the contributions of all fictitious

variables to the stresses or displacements. In the fictitious stress method (FSM), the

fictitious variables are tractions applied on the boundary elements whereas in the

displacement discontinuity method (DDM), the fictitious variables are displacement

3
discontinuities on the boundary elements [3, 4]. The DDM is especially suitable for

problems involving cracks as it is based on the analytical solution to the problem of a

constant discontinuity in displacement over a finite line segment of an infinite elastic

solid [3]. In problems involving cracks, displacement discontinuities represent real

relative displacements across the discontinuous surfaces (i.e. cracks) and are no

longer fictitious variables [3].

Since the early work of Crouch [5] on displacement discontinuity method, it has been

successfully applied in rock fracture (such as geological processes in discontinuous

rock masses, slope stability) [6-8], mining and petroleum engineering including

hydraulic fracture [9, 10]. This method is well suited for fracture propagation because

no re-meshing is necessary as new cracks are formed. However, one of its limitation

is that it can only be used in elastic media [11]. Due to the strong stress singularity at

the ends of displacement discontinuity elements, it has been pointed out that the DDM

is undesirable to model smooth boundaries. Therefore, DDM has been often

combined with fictitious stress method or direct boundary element method to be a

hybridized scheme [7, 11-13] in which the FSM or direct BEM model the smooth

boundary and DDM models the cracks. Actually, DDM has been shown in [3] as an

indirect boundary element to be capable of obtaining accurate stress distribution when

modeling smooth boundaries. Although there is stress singularity at the ends of

displacement discontinuity elements, the stresses at collocation points are accurate

and can be used. This study uses the DDM to model both smooth boundary and cracks

to verify its accuracy.

4
According to the interpolation function of displacement discontinuities along the

element, constant displacement discontinuity elements (CDDE), linear displacement

discontinuity elements (LDDE), quadratic displacement discontinuity elements and

other kinds of higher order elements have been developed by previous researchers [3,

12, 14]. For cracks intersecting with the member surface, constant displacement

discontinuity surface elements (CDDSE) and linear displacement discontinuity

surface elements (LDDSE) can be used [12]. For the crack tip, it must be modeled

with special displacement discontinuity elements to improve the accuracy [3]. If linear

elastic fracture mechanics (LEFM) applies, the relative slip and separation near a

crack tip is proportional to the square root of the distance r from the tip, so a root-r

displacement discontinuity element (RRE) can be used [3, 12]. A parabolic

displacement discontinuity element (PDDE) has been constructed in [12] to model

crack tip with finite stress. This element is suitable for modeling tension-softening or

slip-weakening processes at crack tips [12].

Despite its potential, the DDM has found little application in the fracture analysis of

quasi-brittle materials with tension-softening behavior. To the authors’ knowledge,

there have only a few trial attempts [13, 15] to discuss the potential of the DDM for

cohesive crack propagation. [15] proposed the idea of implementing the DDM for

cohesive crack propagation problems and [13] performed a simple academic test of

rectangular plate under tension for preliminary study, but neither complete simulation

on structural test of member has been reported nor comparison with experimental

results or finite element analysis has been made.

5
Numerous numerical procedures for cohesive crack propagation in quasi-brittle

materials have been developed by many researchers with the methods of finite

element and boundary element. Since Hillerborg’s pioneering work [16] on mode-I

cohesive crack propagation in concrete, the finite element method has been widely

used with the cohesive crack model [17-20]. However, in FEM, re-meshing of the

domain is required at each crack length increment to follow the crack patterns which

makes FEM computationally costly [21]. To overcome this drawback of FEM, the

extended finite element method (XFEM) [22] was developed by introducing

appropriate enrichment functions to model the discontinuity without re-meshing.

Boundary element method (BEM) is an alternative numerical method widely applied

in crack propagation simulation. As BEM requires only to discretize the boundary and

the crack surfaces, it requires less computational effort to generate new elements

when modeling crack growth. The direct application of BEM to crack problems gives

rise to singular system of algebraic equations due to the coincidence of points

coordinates on two sides of the crack surfaces producing the same integral equations

[23], so various special techniques have been developed by previous researchers to

overcome this difficulty, such as the sub-regions method [24], dual boundary element

method (DBEM) [23, 25] and displacement discontinuity method (DDM). The sub-

regions method introduces artificial boundaries to divide the regions that are not

unique and it is not easy to implement it for multiple cracks and mixed mode cracking

[23]. DBEM adopts singular and hyper-singular integral equations for collocation

points at the opposite crack surfaces, which is an effective method to solve various

6
crack problems including cohesive crack propagation [25]. The DDM, as a special

boundary element technique, is convenient to solve crack propagation problems

which has been introduced previously. However, its application combined with

tension-softening behavior is not fully covered in the literature.

The objective of this study is to develop the DDM for analyzing the propagation of

cohesive crack with tension-softening behavior. In the next section, the formulation of

DDM for a member with smooth boundary and cracks will be described. Sections 3

and 4 will present the cohesive crack model and its implementation in DDM. Section

5 will give numerical examples on wedge splitting test and three-point bending of

notched beam to verify the performance of the modeling approach.

2. Formulation of the displacement discontinuity method

An arbitrary problem involving smooth boundary and cracks shown in Fig. 1(a) can

be resolved into two sub-problems as in Fig. 1(b) and (c) [12]. The sub-problem 1 is

the body without any crack subjecting to fictitious variables on the boundary.

The influence coefficients at the smooth boundary are given by , in which the

superscript 1 refers to the sub-problem 1 and the superscript b refers to the smooth

boundary, representing the solution of stresses or displacements caused by fictitious

variables with unit magnitude. The boundary stresses or displacements for the sub-

problem 1, , are expressed as influence coefficients multiplying the fictitious

variables, as in Equation (1). The influence coefficients at the imaginary cracks

caused by are (with the superscript 1 referring to the sub-problem 1, and the

superscript c referring to the boundary of the cracks). Hence the stresses at the

7
imaginary cracks produced by , noted as , can be expressed as Equation (2).

(1)

(2)

Sub-problem 2 is an infinite domain with internal cracks subjecting to displacement

discontinuity . The stress or displacement at smooth boundary and stresses

on the cracks are given by

(3)

(4)

By superposing the two sub-problems, the original problem is recovered, which

means the boundary conditions of the original problem should be satisfied. Hence, we

have

(5)

(6)

where are the boundary conditions (stresses or displacements) on the smooth

boundary and are the stress boundary conditions on the cracks.

From Equations (1)-(6), we can get the following equation system for the original

problem in Fig. 1 (a),

(7)

Since the cracks are modeled by displacement discontinuities, the present approach is

a kind of discrete crack method. In the discrete crack approach, the un-cracked

material stays linear elastic and crack propagation is treated as additional geometric

discontinuity [21]. In the DDM proposed herein, with crack propagation, the only

8
change is to add a new displacement discontinuity element along the crack

propagation direction once the fracture criterion is satisfied, while the element number

along the smooth boundary does not need to change. For a given problem, the size of

, and is fixed, and the values in only need to be computed once.

The size of the other submatrices in Equation (7) changes with the increasing of crack

element number due to crack propagation. On the traction free crack elements, in

Equation (7) are zero. However, for cohesive mode-I cracks, the normal stresses on

the crack elements σn are not zero but dependent on the value of the crack separation

w (equal to the negative normal displacement discontinuities -Dn as positive

displacement discontinuity is normally defined as inwards [3]), which has to be

determined through an iterative process. The following two sections will cover the

method to solve the nonlinear governing equations resulting from the cohesive crack

model with the method of displacement discontinuity developed in this study.

3. Cohesive crack model

3.1 Introduction to cohesive crack model

Quasi-brittle materials under tension exhibit strain-softening behavior due to the

existence of the fracture process zone [26, 27]. The softening traction-separation law,

relating the cohesive force σn due to aggregate bridging and the crack opening w,

describes the material behavior within the fracture process zone (Fig. 2). A model

with cohesive force on the crack is also called the fictitious crack model or cohesive

crack model [16]. It assumes that the crack can still carry stress as long as the crack

opening w is less than or equal to the critical crack width wc, and it becomes traction

9
free after wc is exceeded. Various expressions for the softening traction-separation law

that are linear, bilinear or nonlinear ones have been proposed in the literature [26, 27].

As linear and bilinear softening laws (i.e. σn-w relation) (Fig. 3(a) and (b)) are most

widely used for plain concrete [26], both of them will be formulated in the iterative

algorithm for cohesive crack propagation. Other nonlinear cohesive laws can also be

included directly by modifying the iterative algorithm.

3.2 Fracture criteria of cohesive crack propagation

For the cohesive crack model, crack propagation is often determined by the tensile

strength criterion which states the following: when the maximum principal stress at

the fictitious crack tip is equal to the maximum tensile strength of concrete , the

fictitious crack will propagate perpendicular to the maximum principal stress [26].

The strength criterion is based on the assumption that the stress singularity at the

crack tip is eliminated by the fictitious stresses in the fracture process zone around the

crack tip [28]. The parabolic displacement discontinuity element (PDDE) constructed

by Chan [4] is suitable to model crack tip with finite stress.

The other fracture criterion is the fracture toughness (or critical stress intensity factor)

criterion. It assumes that the stress singularity at the crack tip still exists and the

fictitious stress in the process zone reduces the stress intensity at the crack tip [28].

When the stress intensity factor at the crack tip exceeds the fracture toughness of

the brittle matrix , a root-r displacement discontinuity element (RRE) is to extend

the crack. The stress intensity factor at the crack tip can be expressed in terms of the

displacement discontinuity of the crack tip element through [12].

10
In [29], the numerical stress intensity factor KI for cracks under various conditions

were determined from the DDM by modeling the crack body with LDDE and crack

tip with RRE. The results were found to be in good agreement with solutions given in

the handbook [30].

3.3 Boundary conditions on cracks with linear and bilinear cohesive laws

The linear and bilinear softening relation shown in Fig. 3 (a) and (b) are expressed in

Equations (8) and (9) respectively [26].

(8)

(9)

Through the σn-w relation given in Equations (8) or (9), the normal stresses on the

crack elements are related to the value of normal displacement discontinuities. As the

present study focuses on mode-I crack, shear stress on the crack is assumed to be zero.

As the governing equations are set up at the collocation points of boundary elements,

their boundary conditions are specified. The position of collocation point to achieved

highest accuracy was suggested in [12]. For CDDE, the collocation point is at the

middle point of the element. For LDDE and LDDSE, the collocation points are two

third points. For PDDE and RRE, the collocation point is at ¾ of the element length

away from the crack tip. In this study, LDDE is used for the smooth boundary and the

crack except the crack tip which is modeled by either PDDE or RRE. In addition, the

part of the crack which intersects with the external boundary is modelled with

11
LDDSE.

For LDDE and LDDSE, the normal displacement discontinuity of the first and second

collocation point of the element is

(10)

(11)

where and are normal displacement discontinuity at the beginning point and

end point of the element, as shown in Fig. 4.

For PDDE, the distribution of normal displacement discontinuity (Fig. 5) is expressed

in Equations (12) [12].

(12)

For RRE, the distribution of normal displacement discontinuity (Fig. 6) is expressed

in Equations (13) [12].

(13)

The normal displacement discontinuity at the collocation point of PDDE and RRE is

then given by Equation (14) and (15) respectively.

(14)

(15)

If the i-th collocation point on the crack elements has a displacement discontinuity ,

the traction boundary conditions for linear softening law are given by:

(16)

(17)

If bilinear softening law is used, the traction boundary conditions are:

12
 (18)

(19)

(20)

The normal stress on the crack element can also be expressed as

(21)

where M is the number of all collocation points including those on the smooth

boundary and cracks boundary, and are normal stress influence functions of

displacement discontinuity variables and on the i-th collocation point.

For a crack with linear softening behavior and before wc is reached, by equating

Equations (16) and (21), the following equation is obtained,

(22)

Similar equations can be written for other boundary conditions listed in Equations

(17)-(20).

In the computational process, the displacement discontinuities are unknowns to be

solved with Equation (7), the branch of the σn-w relation employed for each element

has to be first assumed according to the crack width obtained in the previous step. For

the newly added crack element, the first branch (i.e. Equation (16) for linear law or

Equation (18) for bilinear law) is first assumed. The calculated crack width is then

checked if it lies within the assumed range. If not, another branch of the σ-w relation

will be taken to impose the stress boundary conditions on the crack. Iterations are

performed until the obtained σ-w relation on each crack element agrees with the

assumed relation in the iteration step. Fig. 7 shows schematically the process of

13
cohesive crack extension under strength criterion for crack growth.

4. Loading process

Two different loading processes have been developed: incremental loading method

and inverse calculation method. In both loading schemes, the crack length increment

for every step of crack propagation is fixed. Actually, all crack elements are taken to

have the same element length, which will improve accuracy.

4.1 Incremental loading method

To trace the crack propagation process, loading (in terms of force or displacement) is

applied incrementally. At every load level, the crack propagation criterion (either the

strength criterion or the fracture toughness criterion) at the initial notch or current

crack tip is checked. If the criterion is satisfied, a new displacement discontinuity

element is added along the crack propagation direction. As the propagation criterion

at the crack tip is unlikely to be exactly satisfied with the prescribed load increment,

the incremental load needs to be adjusted until the maximum principal stress (or

the stress intensity factor ) is very close to the tensile strength (or the fracture

toughness ), i.e.,

or (23)

where and is the error tolerance.

4.2 Inverse calculation method

As an alternative approach, the fracture criterion or are imposed

at the crack tip. Mathematically, the equation for fracture criterion is included as an

additional equation to the governing Equation (7). By doing so, the external load F

14
(displacement or force) becomes an additional unknown in the new governing

equation that can be directly solved.

4.2.1 Stress strength criterion

When using tensile strength criterion, the additional equation is .

(24)

where and are normal stress influence functions of displacement

discontinuity variables and on the crack tip point T.

To re-write the governing equation after including Equation (24), Equation (7) is

simply re-written as

(25)

where includes the and as the fictitious variables applied on the smooth

boundary and cracks on the discontinuous boundary are both modeled by

displacement discontinuity elements, and includes and .

By combining Equations (24) and (25), the new governing equation is formulated as

(26)

where except at elements for which F is involved in the boundary

condition expression. Also, and (j=1~M).

4.2.2 Fracture toughness criterion

When using fracture toughness criterion, the additional equation is . The

stress intensity factor at the crack tip is given by

(27)

where is the normal displacement discontinuity of the crack tip element RRE.

15
By combining Equations (25) and (27), the new governing equation is formulated as

(28)

where except at elements for which F is involved in the boundary

condition expression, and except .

5. Numerical examples

In order to show the capability of the developed boundary element model with

displacement discontinuity method for solving cohesive crack propagation problems,

two examples with mode-I cracking are simulated: the wedge splitting test and three

point bending of notched beam. As the results from the incremental loading method

and inverse calculation method are the same, only the results obtained from the

inverse calculation method are plotted in the following.

5.1 Wedge splitting test

The concrete specimen under wedge splitting in [31] is modeled. The geometry of the

specimen is shown in Fig. 8. The measured material properties of the concrete were

Young’s modulus E=25200MPa and fracture energy Gf=101J/m2. The load-

displacement curves at the loading points for two specimens were given together with

numerical results from finite element analysis with the code Merlin. The bilinear

softening parameters adopted in the finite element analysis in [31] were ft=3.3 MPa,

fs=0.825MPa, ws=0.023mm, and wc=0.153mm which produced Gf (i.e. the area of the

softening curve) of 101J/m2. These bilinear softening parameters are adopted herein

for the boundary element analysis programmed with Fortran 90 and finite element

simulation with ATENA [32]. In addition, a linear softening law with parameters

16
ft=3.3MPa and wc=2Gf/ft=0.0612mm is also input in the boundary element program

and ATENA to examine the effect of softening law shape on the load-displacement

curve under the same fracture energy and tensile strength.

Finite element analysis adopted the discrete crack method and used the interface

material to model the crack, so the crack path is a pre-defined interface in front of the

initial notch. In the boundary element model, the length of elements on the smooth

boundaries and initial notch has the same value of 5mm. Crack length at every

increment is also 5mm. Fracture criterion in DDM adopts the tensile strength criteria,

so PDDE is used to model the crack tip.

The load-displacement curves calculated from the FE analysis and BE analysis with

DDM are plotted in Fig. 9. The figure shows that the results from the two methods are

in good agreement for both linear softening law and bilinear softening law. Using

bilinear softening law in the numerical models can predict the experimental load-

displacement curve with higher accuracy. From the figure, with the same tensile

strength and fracture energy, linear softening predicts higher peak load and steeper

descending slope. For the peak load, the initial softening slope affects its value. The

descending curve of the load-displacement depends on the later softening slope. The

gentler slope of the linear softening curve than the initial slope of the bilinear

softening curve results in a higher peak load, while the steeper slope of the linear

softening curve (relative to the second slope of the bilinear softening curve) leads to

the steeper descending slope of p-u.

5.2 Three point bending of notched beam

17
A beam with dimension 400mmx100mmx100mm and initial notch 10mm in [28] is

studied. The authors of [28] used inverse analysis approach to evaluate the fracture

parameters of plain concrete from the measured load-CMOD (crack mouth opening

displacement) curve under three-point bending. Two different criteria (the cracking

strength criterion and fracture toughness criterion) were considered in their study. The

determined fracture parameters include the cracking strength σ fc, critical stress

intensity factors KIc of cement matrix and σn-w relation corresponding to each

criterion. The softening curves (or σn-w relation) under both fracture criteria obtained

in [28] are nonlinear, as shown in Fig. 10. These curves are first fitted by bilinear

curves (Fig. 11) for the preparation of input parameters in the finite element analysis

and boundary element analysis on the beam. The obtained fitting bilinear softening

parameters are ft=2.21MPa, fs=0.584MPa, ws=0.0646mm, wc=0.589mm, for cracking

strength criterion, and ft=2.68MPa, fs=0.506MPa, ws=0.0665mm, wc=0.461mm for

fracture toughness criterion. In [28], the cracking strength σfc instead of the tensile

strength ft was suggested for strength criterion, where the cracking strength represents

the bond strength between cement matrix and aggregate while the tensile strength is

the maximum tensile stress that can be carried, which is slightly higher than the

cracking strength. If the cracking strength is used instead of the tensile strength, ft in

Equation (24) is replaced by σ fc. The results obtained from the present study show that

cracking strength criterion and tensile strength criterion give almost the same results.

The load-CMOD curves obtained from the DDM with the two fracture criteria are

compared with FE result (adopting tensile strength criteria) and experimental result in

18
Fig. 12. It should be mentioned that the fracture toughness criterion cannot be adopted

in the finite element software ATENA as stress singularity element is not available.

The numerical load-CMOD curves from DDM with the tensile strength criterion is

close to the experimental curve in [28] and finite element result. The small difference

may be due to the bilinear fitting of the nonlinear softening relation. The load-CMOD

curve obtained from DDM with the fracture toughness criterion is also close to the

experimental curve. The slight difference may be caused by the difference in the

stress intensity factor calculated by the weight function method in [28] and by the

DDM.

6. Conclusions

In this paper, the displacement discontinuity method has been extended for its

application in problems with cohesive crack propagation. The introduction of the

cohesive crack model requires an iterative algorithm to solve the nonlinear equations

resulting from the cohesive law on the crack elements. Tensile strength criterion and

fracture toughness criterion can both be used by adopting PDDE or RRE to model the

crack tip correspondingly. Two loading processes-the incremental loading method and

inverse calculation method have been implemented, and they give the same results. As

shown in this paper, the DDM with the cohesive crack model can predict load-

displacement and load-CMOD curves close to the experimental and finite element

results for the wedge splitting test and the beam with initial notch under three-point

bending. In ongoing work, the present method is further developed to consider mixed

mode crack propagation. Results will be reported in a separate publication.

19
Acknowledgment

Financial support of the work by the Hong Kong Research Grant Council, under

GRF615513, is gratefully acknowledged.

References
[1] Katsikadelis JT. The boundary element method for engineers and scientists. 2nd ed. Oxford:
Academic Press; 2016.
[2] Zienkiewicz OC, Taylor RL, Zhu JZ. The finite element method: its basis and fundamentals. Oxford:
Butterworth-Heinemann; 2013.
[3] Crouch SL, Starfield AM. Boundary element methods in solid mechanics: with applications in rock
mechanics and geological engineering. London: Allen & Unwin; 1983.
[4] Chan HCM. Automatic two-dimensional multi-fracture propagation modelling of brittle solids with
particular application to rock [Ph.D. Thesis]: Massachusetts Institute of Technology; 1986.
[5] Crouch SL. Solution of plane elasticity problems by the displacement discontinuity method. I.
Infinite body solution. International Journal for Numerical Methods in Engineering. 1976;10:301-43.
[6] Bobet A, Einstein HH. Numerical modeling of fracture coalescence in a model rock material.
International Journal of Fracture. 1998;92:221-52.
[7] Shou K-J. A three-dimensional hybrid boundary element method for non-linear analysis of a weak
plane near an underground excavation. Tunnelling and Underground Space Technology. 2000;15:215-
26.
[8] Bobet A. A hybridized displacement discontinuity method for mixed mode I–II–III loading.
International Journal of Rock Mechanics and Mining Sciences. 2001;38:1121-34.
[9] Dong C, De Pater C. Numerical implementation of displacement discontinuity method and its
application in hydraulic fracturing. Computer Methods in Applied Mechanics and Engineering.
2001;191:745-60.
[10] Behnia M, Goshtasbi K, Fatehi Marji M, Golshani A. On the crack propagation modeling of
hydraulic fracturing by a hybridized displacement discontinuity/boundary collocation method. Journal
of Mining and Environment. 2012;2:1-16.
[11] Bobet A, García Marín V. A stress and displacement discontinuity element method for elastic
transversely anisotropic rock. International Journal for Numerical and Analytical Methods in
Geomechanics. 2014;38:1898-922.
[12] Chan HCM, Li V, Einstein HH. A hybridized displacement discontinuity and indirect boundary
element method to model fracture propagation. International Journal of Fracture. 1990;45:263-82.
[13] Gospodinov G. Boundary element modelling of cohesive cracks using displacement discontinuity
method. Advances in Boundary Element Techniques. 2003:19-26.
[14] Shou K, Crouch S. A higher order displacement discontinuity method for analysis of crack
problems. International Journal of Rock Mechanics and Mining Sciences & Geomechanics Abstracts:
Elsevier; 1995. p. 49-55.

20
[15] Harder NA. Computer simulated crack propagation in concrete. 8th European Biennial Conference
on Fracture (efc8). 1993:706-14.
[16] Hillerborg A, Modéer M, Petersson P-E. Analysis of crack formation and crack growth in concrete
by means of fracture mechanics and finite elements. Cement and concrete research. 1976;6:773-81.
[17] Gálvez J, Červenka J, Cendon D, Saouma V. A discrete crack approach to normal/shear cracking
of concrete. Cement and Concrete Research. 2002;32:1567-85.
[18] Elices M, Guinea G, Gomez J, Planas J. The cohesive zone model: advantages, limitations and
challenges. Engineering Fracture Mechanics. 2002;69:137-63.
[19] Rots JG. Computational modeling of concrete fracture [Ph.D. Thesis]: Delft University of
Technology; 1988.
[20] Bocca P, Carpinteri A, Valente S. Mixed mode fracture of concrete. International Journal of Solids
and Structures. 1991;27:1139-53.
[21] Borst Rd, Remmers JJ, Needleman A, Abellan MA. Discrete vs smeared crack models for concrete
fracture: bridging the gap. International Journal for Numerical and Analytical Methods in
Geomechanics. 2004;28:583-607.
[22] Moës N, Belytschko T. Extended finite element method for cohesive crack growth. Engineering
Fracture Mechanics. 2002;69:813-33.
[23] Portela A, Aliabadi M, Rooke D. The dual boundary element method: effective implementation for
crack problems. International Journal for Numerical Methods in Engineering. 1992;33:1269-87.
[24] Blandford GE, Ingraffea AR, Liggett JA. Two‐dimensional stress intensity factor computations
using the boundary element method. International Journal for Numerical Methods in Engineering.
1981;17:387-404.
[25] Saleh A, Aliabadi M. Crack growth analysis in concrete using boundary element method.
Engineering Fracture Mechanics. 1995;51:533-45.
[26] Shah SP, Swartz SE, Ouyang C. Fracture mechanics of concrete: applications of fracture
mechanics to concrete, rock and other quasi-brittle materials. New York, NY: John Wiley & Sons; 1995.
[27] Bazant ZP, Planas J. Fracture and size effect in concrete and other quasibrittle materials. Boca
Raton: CRC Press; 1998.
[28] Zhang J, Leung CK, Xu S. Evaluation of fracture parameters of concrete from bending test using
inverse analysis approach. Materials and Structures. 2010;43:857-74.
[29] Chen E. Computational modeling of concrete cracking due to non-uniform steel corrosion in
reinforced concrete structures [Ph.D. Thesis]: The Hong Kong University of Science and Technology;
2015.
[30] Murakami Y. Stress intensity factors handbook. Oxford; New York: Pergamon; 1990.
[31] Denarie E, Saouma V, Iocco A, Varelas D. Concrete fracture process zone characterization with
fiber optics. Journal of Engineering Mechanics. 2001;127:494-502.
[32] Červenka V, Jendele L. ATENA Program Documentation, Part 1, Theory. Prague: Červenka
Consulting; 2016.http://www.cervenka.cz/assets/files/atena-pdf/ATENA_Theory.pdf

21
Figure Caption List:

Fig. 1 (a) A problem with smooth boundary and cracks (b) Sub-problem1 (c) Sub-

problem 2

Fig. 2 Fracture process zone in concrete

Fig. 3 (a) Linear softening law (b) Bilinear softening law

Fig.4 Normal displacement discontinuity distribution for LDDE and LDDSE

Fig. 5 Normal displacement discontinuity distribution for PDDE

Fig. 6 Normal displacement discontinuity distribution for RRE

Fig. 7 Illustration the process of cohesive crack extension (a) Crack tip satisfying the

tensile strength criterion (b) Adding a displacement discontinuity element with

cohesive stress on it (c) some displacement discontinuity elements traction free and

some with cohesive stress

Fig. 8 Specimen configuration of the wedge splitting test in [31]

Fig. 9 Load-displacement curves obtained from the wedge splitting test, FEM and

DDM

Fig. 10 σn-w relation given in [28]

Fig. 11 Bilinear fitting of tension-softening curve given in [28]

Fig. 12 Load-CMOD curves obtained from DDM with ft criterion and KIc criterion,

ATENA and experiment

22
23
 b

σ
σ

(a)

Pn
Ps
b1 b2
Ds
σ1 σ2 Ds
1
Dn
σ
Dn σ2
Ps
Pn b1 b2

+
(b) (c)

Fig. 1 (a) A problem with smooth boundary and cracks (b) Sub-problem1 (c) Sub-

problem 2

24
σn
ft

σn(w)

wc w
Fracture Process Zone Real crack

Fig. 2 Fracture process zone in concrete

25
σn σn
nf t ft

fs

wc w ws wc w

Fig. 3 (a) Linear softening law (b) Bilinear softening law

26
 n
Dncol2
col1
Dn
Dd

Db s
( a, 0) ( a, 0) (2a, 0)

Fig.4 Normal displacement discontinuity distribution for LDDE and LDDSE

27
 n Dd
col
Dn

s
(1.5a,0)
(2a, 0)

Fig. 5 Normal displacement discontinuity distribution for PDDE

28
 n
Dncol Dd

s
(1.5a,0)
(2a, 0)

Fig. 6 Normal displacement discontinuity distribution for RRE

29
 (a) (b) (c)
Fig. 7 Illustration the process of cohesive crack extension (a) Crack tip satisfying the

tensile strength criterion (b) Adding a displacement discontinuity element with

cohesive stress on it (c) some displacement discontinuity elements traction free and

some with cohesive stress

30
 70 60 70
P, u
30

85

200

200 mm

Fig. 8 Specimen configuration of the wedge splitting test in [31]

31
 4
Test1
3 FEM_linear
FEM_bilinear
P (kN)

2 DDM_linear
DDM_bilinear
1

0
0 0.2 0.4 0.6 u (mm) 0.8

Fig. 9 Load-displacement curves obtained from the wedge splitting test, FEM and

DDM

32
Fig. 10 σn-w relation given in [28]

33
 6
fracture toughness criterion_first
fracture toughness criterion_second
tensile strength criterion_first
tensile strength criterion_second
Linear (fracture toughness criterion_first)
4 Linear (fracture toughness criterion_second)
Linear (tensile strength criterion_first)
Stress (MPa)

Linear (tensile strength criterion_second)

y = -32.728x + 2.6821
2
y = -25.197x + 2.2111

y = -1.1131x + 0.6555
y = -1.2811x + 0.5912
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Crack width (mm)

Fig. 11 Bilinear fitting of tension-softening curve given in [28]

34
 8000

7000 Zhang et al.(2010)

6000 FEM_tensile strength criterion

5000
Load P (N)

DDM_tensile strength criterion

4000 DDM_fracture toughness criterion
3000

2000

1000

0
0 0.1 0.2 0.3 0.4 0.5 0.6
CMOD (mm)

Fig. 12 Load-CMOD curves obtained from DDM with ft criterion and KIc criterion,

ATENA and experiment

35
Highlights:

 The iterative algorithm of combining DDM and the cohesive crack model is developed to

simulate cohesive crack propagation.

 Two loading schemes (incremental loading method and inverse calculation method) are

proposed to accurately predict the load level satisfying the crack tip condition.

 The proposed method with DDM provides a convenient way to model nonlinear crack

propagation.

36