You are on page 1of 13

Computer methods

in applied
mechanics and
engineering
ELSWIER Comput. Methods Appl. Mech. Engrg. 167 (1998) 139-151

Cracks in three dimensions:


A dynamic dual boundary element analysis
P.H. Wen”“, M.H. Aliabadi”‘*, D.P. Rookeb
“Engineering Department, Queen Mary and Westfield College, University of London, London, UK
hStructural Materials Centre, DRA, Furnborough, Hunts, UK

Received 4 April 1997

Abstract

In this paper the dual boundary integral equation for three-dimensional dynamic problems in Laplace space is presented for the tirst time.
The dual boundary element method is used to calculate dynamic stress intensity factors for three-dimensional cracked structures. The
application of Laplace transforms to the time-dependent equations of elasticity reduce the problem to a static one in Laplace space. The
displacement and traction boundary integral equations for the transformed variables are established by Somigliana’n identity. By applying
the displacement integral equation on one of the crack surfaces and the traction boundary integral equation on the other, a general
mixed-mode crack problem can be solved in a single region, in the same way as a static problem. The transformed mixed-mode stress
intensity factors are calculated from the transformed displacement discontinuities near the crack tip. The dynamic stress intensity factors in
the time domain are obtained by the use of Durbin’s inversion method. The accuracy of this method is demonstrated by its application to
embedded penny-shaped and elliptical cracks, and edge cracks. 0 1998 Elsevier Science S.A. All rights reserved.

1. Introduction

In recent years there has been a considerable interest in understanding the mechanics of dynamic fracture
mechanics. Progress towards a better understanding of dynamic fracture phenomena relies heavily on the
advancements of numerical models.
One of the earliest studies of the transient problem is due to Baker [l]. He investigated the problem of a
semi-infinite crack suddenly appearing and propagating at a constant speed. Baker used the Laplace transform
and Wiener-Hopf technique to obtain the distribution of normal tractions ahead of the moving crack-tip. Later,
Achenbach and Nuismer [2] extended Baker’s work to include incident waves of arbitrary stress profile. They
also included the case of oblique incidence. Freund [3] removed the restriction of instant crack growth in the
earlier works and introduced a finite delay time between the instant stress waves arrival at the crack tip and the
onset of crack extension. He used a superposition procedure to obtain the expression for the dynamic stress
intensity factor at the running crack tip. More recent studies of transient effects for dynamically initiating and
growing cracks can be found in [4].
One of the earliest applications of numerical methods to dynamic fracture problems is due to Chen [5]. He
used a Finite Difference Method (FDM) to obtain dynamic stress intensity factors for an edge crack in a
rectangular crack subjected to impact loading. Later, the Finite Element Method (FEM) became the preferred
numerical tool for the solution of crack problems. A comprehensive review of finite element methods is given
by Nishioka and Atluri [6].

* Corresponding author.
’ On leave from the Central-South University of Technology, Changsha, P.R. China.

0045-7825/98/$19.00 0 1998 Elsevier Science S.A. All rights reserved.


PII: SOO45-7825(98)001 16-9
140 P.H. Wm ~‘t al. I Comprrt. Mrrhods Appl. Muh. Engr,y. I67 (1998) 1.39- ISI

Solutions in elastodynamics using the Boundary Element Method (BEM) are usually obtained by either the
time domain method, Laplace or Fourier transform methods or the dual reciprocity method (see [7,8]). The time
domain was used by Nishumura et al. [9] to solve elastodynamic crack problems. They solved both two- and
three-dimensional crack problems in an infinite domain using a hypersingular boundary integral equation. Later,
Zhang and Gross [lo] used a two-state conservation integral of elastodynamics to obtain a non-hypersingular
traction integral equation. They used this method to obtain stress intensity factors for a penny-shaped crack and
square crack in an infinite domain. Dominguez and Gallego [ 1 11 developed a BEM subregion formulation for
analysis of two-dimensional crack problems in an intinite domain. They later extended that formulation to study
crack growth by using moving singular elements [ 121.
The Laplace transform method was used by Sladek and Sladek [ 1.31 to study penny-shaped cracks in an
infinite domain subjected to harmonic and impact loading. The Laplace transform and subregion method was
used by Polyzo et al. [14] to study two-dimensional linear viscoelastic crack problems. The subregion method
was also used by Chirino and Dominguez [ 15 1 to analyse 2D problems using BEM and Fourier transform
method. Zhang and Gross [16] developed a non-hypersingular traction integral formulation using the
elastodynamic conservation integrals and studied problems of elliptical and square cracks in infinite domains
subject to time-harmonic loading. The boundary element method does not appear to have been previously
applied to three-dimensional elastodynamic crack problems in linite domain.
Recently, Fedelinski et al. [ 17- 191 have developed a general formulation for modelling elastodynamic crack
problems in a single region by the dual boundary element method. A Laplace transform technique, a time
domain analysis and a dual reciprocity technique were studied and detailed comparisons made. Several
applications were described 1201 using these techniques: good agreement was obtained with other workers.
In this paper, the dual boundary element method is further extended to analyse a three-dimensional cracked
structure subjected to dynamic loading. The displacement and traction boundary integral equations are
formulated in the Laplace transform domain. The spatial singularities, both for the displacement and the traction
fundamental solutions, are exactly the same as in the static fundamental solutions; thus the evaluation technique
for singular integrals used in [21] can also be used for the dynamic problem.
In this paper, mixed-mode stress intensity factors are calculated from the opening displacement on the
elements adjacent to the crack front for each fixed value of the Laplace parameter. The choice of a special set of
Laplace transform parameters in the transform domain allows the displacements and the stress intensity factors
in the time domain to be obtained by Durbin’s inversion method [22].

2. The dual boundary element method in Laplace transform variables

Consider a cracked isotropic elastic body fJ enclosed by an outer boundary 4 + $,, where 4 and $, are,
respectively, the boundaries on which the traction and displacement are given. The body contains a crack,
boundary r+ + r-, where r’ and r- are, respectively, the crack upper and lower surfaces, see Fig. 1. The
boundary r is subjected to dynamic loading.
If the body forces are zero, then the Navier-Cauchy equation in the time domain is given [23] by

Fig. I. Cracked three-dimensional body


P.H. Wen et al. I Comput. Methods Appl. Mech. Engrg. 167 (1998) 139-151 141

a2u
2
CZU,,IL + cc:
- &J,.,, = (at2 ’ i= 1,2,3,

where c, and c2 are the velocities of longitudinal and shear waves, respectively. Taking Laplace transforms, Eq.
( 1) becomes

c;g& + (c; - c:)l&; = p2u;


where the initial (t = 0) displacements and velocities are assumed to be zero. The Laplace transform of a
function f(x, t) is defined as
*
J?[f(x, t)] = fix,p) = Io f(x, t) e -“I dt (3)

In the transform domain, the solution of (2) should satisfy the boundary conditions

LI;(x, p) = iI’(x, p) x E c
(4)
t;(x, p) = Fj’(x, p) x E c

where 2:(x, p) and iy(x, p) respectively denote transformed displacement and traction boundary conditions. The
constitutive relationship is unchanged in a Laplace transform domain, that is, the stress is given, in terms of the
strain, by

&,(X? P) = C,,L,&.,(X, P) ; (5)

and the traction by

t;(X> P) = c,,~,nj(x)&J(-K P) ; (6)

where n,(x) is the normal unit vector on the boundary and c,~~, are material constants given by

C,,k/ = A&@,, + /-@j$j, + s,J,,) > (7)

where A and ,u are the Lam6 constants. The displacement at a point X’ in the domain 0 (called a collocation
point) can be determined from the boundary values of displacement and traction through Somigliana’s identity,
as in a straightforward static case. It is given by

i,(X’, P) = G,,(X). x, p)<(x) d0) - ?JX’,x, p)U,(x) dT(x) i, j = 1, 2, 3 ; (8)


I I‘ Ir

where fiJX’, x, p) and f,,(X’, x, p) are the Laplace transformed fundamental solutions of elastodynamic
displacement and traction, respectively. The integration boundary includes both internal and external boundaries,
that is r = 4 + rj’, + f + + rP. The differentiation of (8), with respect to X’, coupled with Eq. (5) allows the
stress components at point X’ to be expressed as boundary integrals:

4,(X’, P) = &(X’, x, p)t;l(x) dT(x) - fki,(X’, x, p)&(x) dT(x) .

The functions ck,,(X’, x, p) and f,,,(X’, x, p) contain derivatives of the fundamental solutions.
The displacement boundary integral equation can be obtained from Eq. (8) by considering the limit as the
domain point X’ +x’ on the boundary. The same procedure as illustrated in [24] can be used to obtain the
displacement boundary integral in the Laplace domain as

c,,(x’)17,(x’) +
f I‘
f,,(x’,x, p%,(x) dT(x) =
J-‘,j(X’,
I-
X, p)t;(X>
d’(X) ; (10)

where f denotes a Cauchy principal-value integral and c,(x’) is a function of the geometric shape at the
boundary point x’; c,,(x’) = a,;/2 for a point on a smooth boundary.
If x’ is assumed to be on a smooth boundary, the stress boundary integral equation can be written as
142 P.H. Wet? et al. I Comput. Method,v Appl. Mech. Engr~. I67 (1998) 139%IS1

1
5 4,(x’) + &,,(x’, x, pn$(x) dT(x) = fik,,(x’> x, p)t;(x) dT(x) (11)
$ I‘ I‘

where # denotes a Hadamard principal-value integral 1251. Multiplication of both sides of Eq. (11) by n,(x’)
gives the traction boundary integral equation:

; f,(x’) + n,(x’) f,,,(x’, x, p)uIk(x) dT(x) = n@‘) fi&C x, j?)fk(x) d&) (12)
f I‘ I‘

3. Transformed fundamental solutions

The Laplace transforms of the fundamental solutions C?,,(x’, x, p) and ?,,(x’, n, p) in Eq. (8), which represent
the displacement and traction caused by a unit force 1 . 6(t), are given [26,27] by

(13)

(14)

where

(15)

-,“I’ 2
(.; e-lJ’l’ ,
x=?ll,-2+-,;~ (16)

and

r = Ix’ -XI

and the kernel functions in Eq. (9) can be written as

&,(X’J> PI = -/NL., + q,.,,- a,,.,S,,

(17)

and

C,,(-G, PI = -/-a, + q,,,,


- e,,,q,

-5$+85),:,,:,,,-($,,-$- 3~+6~)(4,r,+q,l,~

( ))
Q1r,,
1
r,!r.,nl
( ))

_ + _&-3&+6x (r,,n, + y,n,)r,k +(4f$+i (4++ 8$-4$)


II r r r2 )
P.H. Wen et al. I Comput. Methods Appl. Mech. Engrg. 167 (1998) 139-151 143

The derivatives of the functions I+IJand x with respect to r are given by

( )

(20)

&=4$+(3+3~+(y)~-$(l+~)Jy (21)

2 e -pr-/<,, 2 -prl’z
C2 2 e
X ,lr =12x+ 5+5:+ 5 --1 5+5$+ $ ___ (22)
r2 ( ( . >> 2 Cl ( ( >> r3

In order to analyse the singularities of the transformed fundamental solutions in Eq. (8) and the functions in Eq.
(9), the following limits need to be considered:
2

bin (r@j =i
( 1 1 +2
Cl
2
, (23)

lii”(YX) = -;
( 1 l-2

2
(24)

liiy(r2JIT)=-+
( > I+%

2
Cl
, (25)

iii) (r’x,,) = +
( > 1- 2 , (26)

(27)

h-i (r3x,I,-) = 2 - 1. (28)

Substituting these values into Eqs. (13) (14), (17) and (18), the fundamental solutions can be obtained thus:

and
I
f,,(x’.x, p) = 7-:,(x’. x>= - d’ [( 1 - 2v)S, + 3 y,‘,,l - (1 - 2dr.pi - r,jn,l 3
87r( 1 - v)r2 an

1
Q(x’, x, p) = lJ;Jx’, x) = [(I - 2vj(6Jiyk + &r,, - q,r,,) + 3r,,r,jr,l (31)
8~( 1 - v)r’

and
144 P.H. Wm et (11. I Comput. Mrthods Appl. Mrch. Engrg. 167 (1998) 139-1.51

p
f&(x’, x, PI = T;;,w, XI = 3 $ L(1 - 2d&,r., f dac,r, + S,kr,,)
47r(l - v)r3 1

- 5r.,r,,r,kl + 3v(n,r,,r,8 + n,r,r.,) + (1 - w(3 ntr,,r,, + ~z~S,~ + n,S,,) - (1 - 4v)n,S,,

(32)

These are fundamental solutions for static problem. So from these limits, it is clear that the singularities in the
fundamental solutions are exactly the same as in static problems. Thus, the transformed fundamental solutions
and kernel functions can be written in two parts as

U,,(X’J, p) = I/:,(x’,x) + u::‘(x’J, P) ,


(33)
TJX’J, p) = 7y,(x’,x) + T:;(x’,x, p)

and

fi&“X’ p) = U&‘,x) + u;:,(x’J, p) 3


(34)
~~,,(X’,X, p) = T;,,(x’,x) + TZ:,(x’,x, P)

where U”, T’ denote the solutions of static problem with singularities (as r -+O) and i?“‘, T”’ denote the
modified terms, which are functions of the Laplace transformed parameter and contain only weak singularities.

4. Solution of integral equations

The displacement and traction boundary integral equations (10) and (12) are discretized with, in general, three
different types of elements, as shown in Fig. 2
l Quadratic continuous elements on the outer boundary except the element at the junction with an edge crack.
l Quadratic discontinuous elements on the crack surfaces.
l A semi-discontinuous element on the outer boundary at the junction with an edge crack.
The displacement and traction on the element n are approximated as

7 6 5

4 F 8

(4 (b) (4

0 element node

collocation node

Fig. 2. Different types of element.


P.H. Wen et al. I Comput. Methods Appl. Mech. Engrg. 167 (1998) 139-151 145

(35)

and the integration variable as

x,,(c5‘I) = c
IX=,
M”(5, W” (37)

where llra and ,:a are displacement and traction values at the node (Y, N”( 6,~) are shape functions which
depend on the type of element; M”( 5,~) are geometric shape functions and ?,a are element nodal positions. The
use of these approximations enables the displacement and traction boundary integral equations to be discretized
as

and

where x” is the collocation point; M, is the number of collocation on the boundary c + < f f +, M? is the
number of collocation on the boundary f _. J”( 5,~) is the Jacobian of the transformation to 5, 7 coordinates
(151 c 1, 1~)c 1). Th e set of discretized boundary integral equations can be written in matrix form as

&=(jf (40)

where the matrices a and G depend on integrals of the fundamental solutions in Eqs. (38) and (39). It is clear
that when r = Ix’ -xl tends to zero, the fundamental solutions c!,(x’,x, p), f,,(x’,x, p) and the functions
ck,,(x’, x, p), fk,j(x’, X, p) have the same form as in the static case. Thus, the transformed fundamental solutions
and functions can be written in two parts, the first part is independent of the Laplace transform parameter
(solutions as in the static case) and the second part is a function of transform parameter p. This behaviour
enables the matrices H and G to be written as [7]

and

e=G”+d”,
(41)

where H” and G” are the matrices for a static problem, I?” and 6’” are modification terms. During the integral
procedure, there is no singularity in fiM and a weak singularity in 6, which can be removed by a polar
coordinate system transform. Together with the displacement and traction boundary conditions, the unknowns of
displacement and traction can now be determined.
146 P.H. Wen et cd. I Comput. Methods Appl. Mech. Eqy,q. 167 (1998) l-39-151

5. Cauchy and Hadamard principal-value integrals

There are singularities 0( I /I-?) and 0( 1 /r’), respectively, in the integrands in the following integrals, which
appear in H’,

I I
II-I
T:,b’ ‘>4 C>v),plN”( 5, dJ”( 5, rl) d5 dv (42)
-I

and

(43)

if m = II, that is the collocation point is on the integration element. If the collocation point, in the coordinate
system (5, v) is ([,, vp), then the distance r is defined as

In the polar coordinate system (p, 13) shown in Fig. 3, the singular terms are obtained as (see [28])

in integral equation (42) and

T_,(B) T- ,(8)
I2 = T;,, Lx“3 x( 5, rl)> plN”( ~5rl)J”( 5,~) = /73 + ___
PZ
in integral equation (43), where T_,(B) and T_,(B) are functions of H only in the static case (see [21]). In the
same way as in the static analysis, the Cauchy principal-value integral and Hadamard principal-value integral
can be written as

and

Fig. 3. Transformed coordinate system (.$, 7)


P.H. Wen et al. I Comput. Methods Appl. Mech. Engrg. 167 (1998) 139-1.51 147

(48)

where 80) is the polar coordinate equation of the boundary shown in Fig. 2 and p(0), ~(0) are defined in [21].

6. Dynamic stress intensity factors

The transformed dynamic stress intensity factors can be calculated directly from the transformed displace-
ments on the crack surface. Some techniques were introduced in [ 181 and [ 191 to improve the accuracy of these
procedures in static problems. On the crack front, a local coordinate system (.?j?y^;),as shown in Fig. 4, can be
established to determine the stress intensity factors. In this figure, e is the nearest point on the crack front to the
collocation point X”; f, y^ and z^are unit vectors. The crack openings at the collocation point, can be written as

A&,(X”, p) = <C: - i,-)i; ,


(40)
Au&“, p) = (i: - U-)2,

AU,&“, PI = G: - u,->j,,
where p is any one of the set of Laplace parameters chosen for the solution of the integral equations. The stress
intensity factors can be obtained approximately from
E
i,(p) = ~-
4(1 - V2)
E
K,,( PI = (50)
4(1 - V’)

&,(P) = E
4(1 + V2)

where d is the distance between x” and e.


The time-dependent values of any of the transformed variables must be obtained by an inverse transform.
There are many Laplace transformation inversion methods; here, the method put forward by Durbin [22] is used.
For the Heaviside function, this method can obtain very accurate inversion results for long durations. The
calculation formula used is as follows

f(t) = 2 $ [ - i M{f(s)} + i (!lii(,S + i ZjF) COS F - Z,f( s + i F) sin F) ] (51)


k=O

where f(p,) stands for the value in Laplace space at the sample point

crack front

Fig. 4. Element in front of the crack and local coordinate system.


148 P.H. Wm et al. I Comput. Methods Appl. Mrch. Eqr<q. 167 (1998) 139%151

p,=s+2k~i/T.

The sample points are chosen for k = 0, 1, . , L,. Good results have been obtained for ST = 5 and T/t,, = 20,
where t, is unit time. The results of many test examples, showed that the sample number L should be larger than
or equal to 25. In the following numerical examples, L is chosen as 25.

7. Applications

7.1. Rectangular bar with one central penny-shaped crack

A square bar (2w, X 2w, X 2h), containing a central circular crack, is loaded on the ends by a Heaviside load
in the direction perpendicular to the crack plane, as shown in Fig. 5. The radius of the penny-shaped crack is a,
with a/w = 0.5, w , / w2 = 1 and h/w = 2; Poisson’s ratio v = 0.2. There are 40 elements on the outer boundary
and 20 elements on one crack surface. The displacement u,(t)/ u0 at the center of rectangular bar top are plotted
in Fig. 6 and compared with the result in the absence of a crack, where u” = u,,h/E and E is Young’s modulus.
It is evident that when t d 2h /c,, these results are almost identical; the presence of the crack should have no
influence until t = 2h/c,. The values of the normalized dynamic stress intensity factor K,/K” at point A have
been calc,ulated and are plotted in Fig. 7, where K” = 2a,m. Good agreement is obtained when compared
with the results obtained [30] by using the displacement discontinuity method alone. The maximum value of the
stress intensity factor by DBEM is slightly lower and the variation as a function of t is smoother than that
obtained by DDM.

7.2. Rectangular bar containing one elliptical crack

A rectangular bar of cross-section 2w, X 2w2 (as shown in Fig. 8) and height 2h contains a centrally located
elliptical crack. The dimensions of the bar are h = 15 cm, w, = 9 cm, wZ = 6 cm; the two principal axes of the
crack are a = 3.5 cm and b = 2 cm. The material constants are: the bulk modulus K = 165 GPa; the shear
modulus G = 77 Gpa and the density p. = 7.9 mg/m’. The bar is subjected to a Heaviside load on the ends. The
normalized dynamic stress intensity factor at the end of the minor axis is plotted, as &K,(t) /uo, against time in

ooHf t)
t t

‘I::-,
I-
I
- with crack
I ---~~~ without crack

I
2h

c,t/h

Fig. 5. Circular or elliptical crack embedded in a rectangular Fig. 6. Normalized dynamic displacement on the top of the bar.
parallelepiped.
P.H. Wen et al. I Comput. Methods Appl. Mech. Engrg. 167 (1998) 139-151 149

s DEEM
------ Wen [15]

a
T-
ZW,

I
Fig. 7. Normalized stress intensity factor for rectangular bar Fig. 8. Section of cracked bar.
containing a circular crack.

Fig. 9. The numerical results given by Chen [31] and Nishioka [32] are plotted for comparison. Before the
longitudinal waves arrive, t < 25 ps, the stress intensity factor is almost zero. There is a kink at about t = 40 ~JA,
which is caused by surface waves coming from the opposite crack front (see [33]). The maximum value and the
shape of the stress intensity factor curve are close to both of the previous solutions, although there are
differences in detail.

7.3. Single edge cracked bar

The single edge crack in a rectangular bar is shown in Fig. 10, the thickness of the bar is denoted by w, , the
width by 2w, and the height by 2h. The dimensions chosen were a/w, = 0.4, h/w, = 1.5, h/w, = 2.0, and
Poisson’s ratio Y = 0.3. The dynamic stress intensity at two points on the crack tip are plotted in Fig. 11, both of
them are normalized to the static stress intensity factors. There is a local maximum in the stress intensity factor
for x2/w? = 0.875 at about c,t/h = 3; the maximum value occurs at c,t/h = 6.5. This behaviour is similar to

0.60
08880 DEEM

c
-- 0.40

I
5
I
I
k
lfl 0.20 2h
.u
E
:
0"
-0.00
1
-0.20
0.00 50.00 100.00 150.

t (usl

Fig. 9. Normalized stress intensity factor for rectangular bar Fig. IO. Edge cracked rectangular bar.
containing a elliptical crack.
150 P.H. Werr et al. I Comput. Methods Appl. Mech. En,qrg. I67 (1998) 139-151

Fig. 1 I. Dynamic stress intensity factor for edge cracked bar.

that illustrated by Balas et al. [23] for a two-dimensional problem. The solution for x2/w? = 0.125 is similar,
but no comparisons are available.

8. Conclusions

The dual boundary element method has been extended to the analysis of three-dimensional cracked structures
subjected to dynamic load. The time-dependent problem is reduced to a static one by the use of Laplace
transforms. The form of the spatial singularities in the transformed Kelvin’s fundamental solutions is exactly the
same as that in static problems. Hence the existing treatment of the singular integrals, required in the
determination of matrix coefficients in the static case, can be used directly in the transform domain. The
transformed stress intensity factors are calculated from the crack opening near the tip. The dynamic values in the
time domain are determined by Durbin’s inversion method. Several examples were analysed with this dual
boundary element method and accurate results were obtained, even for a relatively small number of Laplace
parameters.

Acknowledgement

This work has been carried out with the support of the Ministry of Defence, Defence Evaluation and Research
Agency, Farnborough, Hants, UK.

References

[ I] B.R. Baker, Dynamic stresses created by a moving crack, J. Appl. Mech. 39 ( 1962) 449-458.
[2] J.D. Achenbach and R. Nuismer, Fracture generated by a dilatation wave, Int. J. Fract. 7 (1971) 77-88.
[3] L.B. Freund, Crack propagation in an elastic solid subjected to general loading. III. Stress wave loading, J. Mech. Phy\. Solids 21
(1973) 47-61.
[4] Cheng Liu and A.S. Rosakis, Investigation of transient effects for dynamically initiating and growing cracks under stress waveloading
conditions, in: M.H. Aliabadi, ed.. Dynamic Fracture Mechanics (Computational Mechanics Publications, Southampton, 1995)
137-202.
[5] Y.M. Chen, Numerical computation of dynamic stress intensity factors by a Lagrangian finite-difference method (the HEMP code),
Engrg. Fract. Mech. 7 (1975) 653-660.
P.H. Wen et al. / Comput. Methods Appl. Mech. Engrg. 167 (1998) 139-1.51 151

161 T. Nishioka and S.N. Athni, Computational methods in dynamic fracture, in: S.N. Atluri, ed., Computational Methods in Mechanics of
Fracture (Elsevier Science Publishers, 1986) 335-338.
[7] J. Dominguez, Boundary Elements in Dynamics (Computational Mechanics Publications, Southampton, 1993).
[8] P. Fedelinski, M.H. Aliabadi and D.P. Rooke, Boundary element formulations for the dynamic analysis of cracked structures, in: M.H.
Aliabadi, ed., Dynamic Fracture Mechanics (Computational Mechanics Publications, Southampton, 1995) 61-100.
[9] N. Nishimura, Q.C. Guo and S. Kobayashi, Elastodynamic crack analysis by BIEM, in: M. Tanaka and T. Cruse, eds., Proc. Boundary
Element Methods in Applied Mechanics (Pergamon Press, 1988) 245-254.
[IO] Ch. Zhang and D. Gross, A non-hypersingular time-domain BIEM for 3-D transient elastodynamic crack analysis, Int. J. Numer.
Methods Engrg. 36 (1993) 2997-3017.
[I l] J. Dominguez and R. Gallego, Time domain boundary element method for dynamic stress intensity factor computations, Int. J. Numer.
Methods Engrg 33(3) (1992) 635-647.
[ 121 R. Gallego and J. Dominguez, Dynamic crack propagation analysis by moving singular boundary elements, J. Appl. Mech. 59 ( 1992)
158-162.
[ 131 J. Sladek and V Sladek, Dynamic stress intensity factors studied by boundary integro-differential equations, Int. J. Numer. Methods
Engrg. 23 (1986) 919-928.
[14] D. Polyzo, A.A. Stamos and D.E. Beskos, BEM computation of DSIF in cracked viscoelastic plates, Comm. Numer. Methods Engrg.
IO (1994) 81-87.
[ 151 F. Chirino and J. Dominguez, Dynamic analysis of cracks using boundary element method, Engrg. Fract. Mech. 34 (1989) 1051-1061.
[ 161 Ch. Zhang and D. Gross, 3D elastodynamic crack analysis by a non-hypersingular BIEM, Comp. Mech. 9 (1992) 137-152.
[17] P. Fedelinski, M.H. Aliabadi and D.P. Rooke, The dual boundary element method: J-integral for dynamic stress intensity factors, Int. J.
Fract. 65 ( 1994) 369-38 1.
[ 181 P. Fedelinski, M.H. Aliabadi and D.P. Rooke, A single-region time domain BEM for dynamic crack problems. Int. J. Solids Struct. 32
(1995) 3555-3571.
[ 191 P. Fedelinski, M.H. Aliabadi and D.P. Rooke, The Laplace transform DBEM method for mixed-mode dynamic crack analysis, Comput.
Struct. 59 (1996) 1021-1031.
[20] P. Fedelinski, M.H. Aliabadi and D.P. Rooke, Boundary element formulations for the dynamic analysis of cracked structures, Engrg.
Anal. Bound. Elem. 17( 1) (1996) 45-56.
[21] Y. Mi and M.H. Aliabadi, Dual boundary element for three-dimensional fracture mechanics analysis, Engrg. Anal. Bound. Elem. lO(2)
(1992) 161-171.
[22] F. Durbin, Numerical inversion of Laplace transforms: an efficient improvement to Dubner and Abate’s method, Comput. J. 17(4)
(1974) 371-376.
[23] J. Balas, J. Sladek and V. Sladek, Stress Analysis by Boundary Element Method (Elsevier, 1989).
[24] M.H. Aliabadi and D.P. Rooke, Numerical Fracture Mechanics (Computational Mechanics Publications and Kluwer Academic
Publishers, 1991).
[25] J. Hadamard, Lectures on Cauchy’s Problem in Linear Partial Differential Equations (Yale University Press, New Haven, 1923).
[26] T.A. Cruse and F.J. Riazo, A direct formulation and numerical solution of the general transient elastodynamic problem, I, J. Math.
Anal. Appl. 22 (1968) 2444259.
(271 L. Gaul and C. Fiedler, Improved calculation of field variables in the domain based on BEM, Engrg. Anal. Bound. Elem. I l( 1) (1993)
257-264.
[28] M. Guiggiani, G. Krishnasamy, T.J. Rudolphi and F.J. Rizzo, A general algorithm for the numerical solution of hypersingular boundary
integral equations, ASME J. Appl. Mech. 59 (1992) 604-614.
[29] Y. Mi, Three-Dimensional Analysis of Crack Growth (Computational Mechanics Publications, Southampton and Boston, 1996).
[30] P.H. Wen, Dynamic Fracture Mechanics: Displacement Discontinuity Method (Computational Mechanics Publications, Southampton
and Boston, 1996).
[31] E.P. Chen and G.C. Sih, Transient response of cracks to impact loads, in: Elastodynamic Crack Problems (Noordhoof, 1977).
[32] T. Nishioka, Recent developments in computational dynamic fracture mechanics, in: M.H. Aliabadi, ed., Dynamic Fracture Mechanics
(Computational Mechanics Publications, Southampton, 1995)
[33] P.H. Wen, M.H. Aliabadi and D.P. Rooke, The influence of elastic waves on dynamic stress intensity factors (three-dimensional
problems), Archive Appl. Mech. 66 (1966) 385-394.

You might also like