Professional Documents
Culture Documents
Summer School on
Fracture Modelling and Assessment
Wolfgang Brocks
Christian Albrecht University Kiel
Wolfgang Brocks
LEFM
Fracture modes
Seite 1
LEFM: Stress Field at a Crack Tip
IRWIN [1957]:
KI
σ ij ( r ,ϑ ) = f ijI (ϑ ) + T δ i1δ1 j mode I
2π r
K I = σ ∞ 2π a Y ( geometry )
1 r
displacements ui ( r , ϑ ) = K I giI (ϑ ) + K II giII (ϑ ) + K III giIII (ϑ )
2G 2π
1,5
Modus I f xx
fij f yy
1
f xy
0,5
-0,5 2
-180 -120 -60 0 60 120 180 Modus II f xx
1,5
θ[] fij 1
f yy
f xy
0,5
-0,5
-1
-1,5
-2
-180 -120 -60 0 60 120 180
θ[ ]
CompFractMech_01 - WB ECF20 Trondheim 2014 4
Seite 2
LEFM: Energy Approach
K I2 ∂ ∂
IRWIN [1957]: G el = =
E′ ∂Acrack B∂ 2a
Mixed mode: G el =
1
E′
( K I2 + K II2 ) + 21G K III2
Singularity
∫ ϕ n dS = ∫ ϕ n dS + ∫ ϕ n dS + ∫ ϕ n dS + ∫ ϕ n dS = 0
i i i i i
∂B0 ∂B ∂B − ∂BS ∂B +
Seite 3
J-Vector as Material Force
• Equilibrium: σ ij , j = 0 in B
∂w
• Hyperelastic material: σ ij = in B
∂ε ij
J-Integral (1)
∂U
RICE [1968, 1973]: J = G = − energy release rate
h ∂a vL
LEFM: mixed mode
J = J1 = GI + GII + GIII =
1
E′
( K I2 + K II2 ) +
1 2
2G
K III
K I K II
J 2 = −2
E′
Seite 4
J-Integral (2)
J ( sc ) = ∫Γ w n − σ
1 ij n j ui ,1 ds
∂
− ∫∫ wn1 − σ ij n j ui ,1 dS
S−
∂ξ3
1
2
1 KI plane stress
rp =
2π R0 (1 − 2ν )
2
plane strain
effective SIF
a
K Ieff = K I (aeff ) = σ ∞ π aeff Y eff
W no singularity
Seite 5
DUGDALE Model
strip
yield
model
2 c = 2a + 2 d p
CTOD
KI r 1 plane stress
u y ( r,π ) = 4
E 2π 1 − ν 2 plane strain
4 K I2 1 plane stress
δ tIrwin =
π ER0 (1 − ν ) (1 − 2ν ) plane strain
2
8 R0 π σ∞
δ tDugdale = a ln sec no dependence on geometry!
π E 2 R0
CompFractMech_01 - WB ECF20 Trondheim 2014 12
Seite 6
BARENBLATT Model
Idea:
Singularity at the crack tip is unphysical
• GRIFFITH [1920]: Energy approach
• IRWIN [1964]: Effective crack length
• DUGDALE [1960]: Strip yield model
• BARENBLATT [1959]: Cohesive zone
∂U ∂Urel
G = − =
B∂a vL B∂a
2
KIeff a + rp KI2
= Gssy = ≈ = Ge + Gp
E′ a E′
∂Usep
= Γ c > Γ ce = 2γ
B∂a
Fracture criterion: G = Γc
δc
local criterion!
Seite 7
EPFM
∂w t t
σ ij = w= ∫ σ ijεɺij dτ ≈ ∫ σ εɺ dτ
p
∂ε ij τ =0 τ =0
ij ij
K I2 ∂U p
U = U e + U p = ∫ F dvLe + ∫ F dvLp J = Je + Jp = −
E ′ ∂Acrack v
L
n
ε εe εp σ σ
uniaxial = + = +α
ε0 ε0 ε0 σ 0 σ0
1− n
ε ijp 3 σ σ ij′
3D = α
ε0 2 σ 0 σ0
HUTCHINSON [1968],
RICE & ROSENGREN [1968]
singular stress and strain fields at the crack tip (HRR field) – mode I
1
− 1
σ ij = Kσ r σɶ ij (θ )
n +1
J n +1
Kσ = σ 0
ασ 0 ε 0 I n
n
Kσ − nn+1
ε ij ≈ ε ijp = αε 0 r εɶij (θ ) σ ijε ijp = O ( r −1 )
σ0
Seite 8
HRR Angular Functions
σɶ xx
σɶ yy
EPFM: CTOD
n
J n+1 n1+1
HRR displacement field ui = αε 0 r uɶi (θ )
ασ 0 ε 0 I n
Seite 9
Path Dependence of J
FE simulation
C(T) specimen
plane strain
stationary crack
incremental theory
of plasticity
σ 0 J σ 0
≃ ≃
δt
Seite 10
Literature
Seite 11
Computational Fracture Mechanics
Part II: Determination of Fracture Parameters
Wolfgang Brocks
FE Models - General
Cracks and crack-like defects induce high stress and strain gradients
which require a fine discretisation → large number of elements and
degrees of freedom.
Reduce number of DOF by
2D models whenever possible,
Coarsening of the mesh remote from the defect,
Account for symmetry conditions,
Apply singular elements with special shape functions.
Seite 1
FE Models – Example C(T)
symmetry line
BC: u2 = 0, x1≥ a
2D model:
plane strain
stationary crack:
collapsed (singular) elements
α ij(0) (ϑ ) α ij(1) (ϑ )
ε ij ( r ,ϑ ) = + + α ij(2) (ϑ )
r r
extending crack:
regular array of elements
Seite 2
Biaxially Loaded Centre-Cracked Panel
two-fold symmetry
2D plane stress analysis
1 symmetry plane
Seite 3
3D Models: Railway Axle
full 3D model,
no symmetry plane
loading by tension,
bending, torsion
CTOD , CTOA
• crack-tip blunting of a
stationary crack
• collapsed elements with 1/r
singularity
• elasto-plastic, large-strain
(geometrically nonlinear)
analysis
Seite 4
Stresses at the Crack Tip
10
Steady state crack extension R-curve
CTOA [ ]
Newman [1999] 8
ψ ( ∆a ) = ψ c , ∆a > ∆ass
6
0
0 5 10 15
∆a [mm]
Seite 5
CTOA in FEM
• ABAQUS:
*DEBOND
*FRACTURE CRIT, TYPE=COD 1/2
• CTOA = COD / BL CTODFE
1/2 CTOAFE
• Base length BL = 1mm
base 0,25m
• Linear 2D Plane Stress Elements lengthFE m
1/2 CTOAFE
Results M(T)
Validation of ψR-curve
Seite 6
Stress Intensity Factors (LEFM)
3 −ν
plane stress
κ = 1 +ν
3 − 4ν plane strain
E plane stress
′
E = E
plane strain
1 −ν 2
Gϕ =
1
h∆a ∫∫
(σ iju j ,k − wδ ik ) ∆ xk ,i dS
B0
K I2 + K II2
G ϕ = 0 = J1 = J =
E′
KK
G ϕ =π = J 2 = −2 I II
2 E′
KI4 − E ′J1KI2 + 41 E ′2J 22 = 0
(K ) 2
I 1,2 = 21 E ′J1 ± 1
2 (
E ′2 J12 − J22 ) more than one solution!
Seite 7
Contour / Domain Integral
Using the divergence theorem, the contour integral can be expanded (?)
into an area integral in two dimensions or a volume integral in three
dimensions, over a finite domain surrounding the crack. This domain
integral method is used to evaluate contour integrals in Abaqus/Standard.”
K
III E ′ 0 0
for isotropic material: {B} = 0 E ′ 0
0 0 2G
−1 −1 −1
general J = KIB11 KI + 2KIB12 KII + 2KIB13 KIII + ...
−1
mode I “auxiliary” crack-tip field JIaux = kIB11 kI superimposed
Seite 8
Determination of SIFs in ABAQUS (II)
JIint
kα assigned as unit values {K} = {B}{J }
1
2
int
{J }
int
= JIIint
J int
III
∫ Γ n (σ ε δ − σ ik uk(α, j) − σ ik(α )uk , j ) q j ds
(α )
Jαint = lim i kl kl ij
Γ →0
Path Dependence
???
Seite 9
Example: Railway Axle - Geometry
cont1
cont2 10
cont3
60 cont4
50
KII [MPa sqrt(m)]
0
30
cont1
cont2 a/c=0.5, a/t=0.1, FV1
40 -5
-10 -5 0 5 10
cont3 20 cont1
cont4
cont2
crack front
cont3
KIII [MPa sqrt(m)]
-10 10
cont4
a/c=0.5, a/t=0.1: FV1
strong domain -15
0
dependence at free -10 -5 0 5 10
surface, crack front -10
no r -1/2 singularity!
-20
-30
-10 -5 0 5 10
results by courtesy of MANFRED SCHÖDEL crack front
Seite 10
Railway Axle: SIFs FV2
5
cont1
cont2
cont3
4,5 cont4
K I [MPa sqrt(m)]
1,5
-4
-10 -5 0 5 10
crack front
100
a/c=0.5, a/t=0.1: F H1
90
KI [MPa sqrt(m)
15
80
10
70 5
KII [MPa sqrt(m)]
cont1 0
60 30
cont2
cont3 -5 cont1
cont4 cont2 20
50 cont1
-10 cont3
-10 -5 0 5 10 cont2
cont4
K III [MPa sqrt(m)]
10 cont3
crack front cont4
-15
a/c=0.5, a/t=0.1: FH1
0
-20
-10 -5 0 5 10
crack front -10
-30
-10 -5 0 5 10
crack front
Seite 11
Railway Axle: SIFs FH2
-50
a/c=0.5, a/t=0.1: FH2
K I [MPa sqrt(m)]
-75
40
a/c=0.5, a/t=0.1: FH2
-100 cont1
30
cont2
cont3 cont1
-30
-10 -5 0 5 10
crack front
J-Integral in ABAQUS
Seite 12
Comments
domain: J =
1
∆ Acrack ∫∫
(σ iju j ,1 − wδ i1 ) ∆ x1,i dS
B0
Comments ctd.
ABAQUS 11.4.2:
“For elastic material behavior w is the elastic strain energy (3); for elastic-
plastic or elasto-viscoplastic material behavior w is defined as the elastic
strain energy density plus the plastic dissipation, thus representing the
strain energy in an equivalent elastic (4) material. Therefore, the J-
integral calculated is suitable only for monotonic loading (5) of elastic-
plastic materials.”
(3) density
(4) hyperelastic σ ij = ∂w ∂ε ij
(5) necessary but not sufficient!
Seite 13
Domain Dependence (1)
“... Since it is not always possible to include the plastic zone in three
dimensions (6), a finer mesh may be the only solution (7). If the first
contour integral is defined by specifying the nodes at the crack tip, the
first few contours may be inaccurate (8). To check the accuracy (8) of
these contours, you can request more contours and determine the
value of the contour integral that appears approximately constant from
one contour to the next. The contour integral values that are not
approximately equal to this constant should be discarded. (9) “
(6) This may also occur in 2D under conditions of full yielding.
(7) This is rubbish: (i) a fine mesh does not change the size of the plastic
zone; (ii) mesh refinement is commonly counterproductive in plasticity
and may make things worse!
(8) This is not a question of (numerical) accuracy but of physics, i.e. energy
dissipation in plasticity!
(9) Absolutely! It indicates that a “far-field” value of J has been reached
which corresponds to the value determined from the load-displacement
curve.
Seite 14
J Calculation
The user must specify the crack front, i.e., the region that defines the
first contour. ABAQUS/Standard uses this region and one layer of
elements surrounding it to compute the first contour integral. An additional
layer of elements is used to compute each subsequent contour. The crack
front can be equal to the crack tip in two dimensions or it can be a larger
region surrounding the crack tip, in which case it must include the crack tip.
By default ABAQUS defines the crack tip as the node specified for the
crack front in the so called CRACK TIP NODES option of the contour integral
command:
*CONTOUR INTEGRAL, CONTOURS=n, CRACK TIP NODES
Specify the crack front node set name and the crack tip node number or
node set name.
Alternatively, a user-defined node set which include the crack tip can
be provided to ABAQUS manually by omitting the crack tip nodes
option:
*CONTOUR INTEGRAL, CONTOURS=n
Specify the crack front node set name and the crack tip node number or
node set name.
Seite 15
VCE Direction (1)
Alternatively, the normal option can be omitted and the virtual crack
extension direction, q, can be defined directly:
*CONTOUR INTEGRAL, CONTOURS=n
crack front node set name, qx-direction cosine, qy-direction cosine, qz-direction
cosine (or blank)
Problem
Penetration point of a surface flaw to the free surface of the structure
No HRR singularity!
What is the „correct“ vce direction, q1 or q2?
See results of SIFs above.
The symmetry option must be used to indicate that the crack front is
defined on a symmetry plane:
*CONTOUR INTEGRAL, CONTOURS=n, SYMM
The change in potential energy calculated from the virtual crack front
advance is doubled to compute the correct contour integral values.
CompFractMech_02 - WB ECF20 Trondheim 2014 32
Seite 16
Example: 2D FE Analysis of a C(T)
Automatic option :
1st contour = crack tip,
last contour #24
Contour Definition
Plastic zone at VLL = 1 mm and 11th Plastic zone at VLL = 1 mm and 3rd
J-contour of automatic option J-contour of manual option
CompFractMech_02 - WB ECF20 Trondheim 2014 34
Seite 17
Contour Dependence (1)
800
J [kJ/m2]
J 24
"automatic"
J 12
Contour dependence of J:
600 J 10
“automatic option”,
J8
J5
200 J4
J3
J2
0
J 1 (crack tip)
0 0,5 1 1,5 2 2,5 3
800
VLL [mm]
J [kJ/m2]
J 12
"manual"
J9
600 J7
J5
J3
400
J 1 (node set)
800
Contour dependence of J
J [kJ/m2]
600
automatic
manual
400 800
J [kJ/m2]
J_ASTM
200 600 J 24 (automatic)
VLL = 3.0 mm
J 12 (manual)
0 400
0 5 10 15 20 25
contour
200
0
0 0,5 1 1,5 2 2,5 3
VLL [mm]
Seite 18
Surface Flaw in a Pressure Vessel
6
p = 160 [MPa]
p = 165 [MPa]
5
p = 170 [MPa]
normalised J-integral
0
0 0,25 0,5 0,75 1
normalised crack front coordinate
References
Seite 19
Computational Fracture Mechanics
Part III: The Cohesive Model
Wolfgang Brocks
Morphological
Node release + fracture mechanics criterion (R-curve)
no splitting of dissipation into global plasticity and local separation
Cohesive surface
Interface elements with traction-separation law responsible for local
separation
Seite 1
BARENBLATT Model
Barenblatt [1962]
δ ( x ) = u y separation
local fracture (separation)
criterion = u y ( x,0 + ) − u y ( x,0− )
Cohesive Model
Phenomenological representation
of various failure mechanisms
by cohesiveOverview
interfaces
Seite 2
Cohesive Law
0,5
Γc
mode I: normal stresses σn
normal separation δn
0
δ / δc
0 0,5 1
δc
σ /σ c
σ /σ c
0 0 0
0 0,5 δ / δc 1 0 0,5 δ /δ c 1 0 0,5 δ / δc 1
σ /σ c
σ / σc
0 0 0
0 0,5 δ /δ c 1 0 0,5 δ / δc 1 0 0,5 δ / δc 1
Seite 3
Cohesive Laws (II)
δ
HILLERBORG [1976]: linear σ (δ ) = σ c 1 −
δc
concrete, brittle fracture
1
separation energy Γ c = σ cδ c
2
BAZANT [1993]: bilinear δ δ
σ c 1 − + σ 1 für 0 ≤ δ ≤ δ1
δ 1 δ
σ =
1
two additional shape
parameters δ1, σ1 σ 1 − δ − δ1 für δ1 ≤ δ ≤ δ c
1
δ c − δ1 n
concrete, e.g. σ 1 = 0.2 σ c , δ1 = 0.2 δ c
MAIER et al. [2006]
ductile tearing, e.g. σ 1 = σ c , δ1 = 0.5δ c
δ1 σ1
Γ c = 12 σ cδ c +
σ c
separation energy
δc
• Infinite initial stiffness is numerically detrimental
16e 2 δ 16e δ
NEEDLEMAN [1987]: exponential σ (δ ) = σ c exp −
9 δc 9 δc
various materials and
applications e = exp(1)
9
separation energy Γc = σ cδ c
16
∂σ 16e 2 σ c
initial stiffness K coh = =
∂δ δ →0 9 δc
• Initial stiffness depends on σc , δc
Seite 4
Cohesive Laws (IV)
Seite 5
Micromechanisms: Void Growth
Ductile tearing:
• void nucleation,
• growth
• coalescence
3D Generalisation
δt Mixed Mode
Crack δn Interaction σ n = f n (δ n , δ t )
σ t = f t (δ n , δ t )
δn
Seite 6
Initial Compliance of TSL
σ / σc
numerical artefacts in the simulations. 0,5
∂σ
σ /σ c
*) stiffness = (compliance)-1 K coh =
∂δ δ →0
0,5
Scheider [2001]
1
σ / σc
0,5
0
0 0,5 δ / δc 1
R0 E = 1.7 ⋅ 10 −3
Modelling with triangular σ nc R0 = 2.8
continuum elements hel δ nc = 0.3
Mixed-mode TSL Meshing of the
δ nc δ1 = 100
necking region
K ncoh hel σ c R h δ c
= 2 n 0 elc n = 2 ⋅ (1.4 ⋅10−3 ) ⋅100 = 0.28
E R0 E δ n δ1
SCHEIDER & BROCKS [2003]
CompFractMech_02 - WB ECF20 Trondheim 2014 14
Seite 7
Effect of Initial Compliance
0,5
0
0 0,5 δ / δc 1 Necking prior to load
maximum due to
K coh hel 27 σ c R0 hel
= tangential separation
E 4 R0 E δ c
27
= ⋅1.4 ⋅10−3 = 0.009
4
Additional elongation of
18% at R0 due to normal
separation
e2 Plane 3D FE-
FE Models Models
e1 e3 e2
e1
linear 4 nodes element, 8-18 nodes element
2 integration points
4 or 9 integration points
5
Shell models
2
e2 4
e3
1 IP 2
e1
IP 1
3
linear 5 node element
2 integration points
Seite 8
Cohesive Elements (II)
Large Deformation
Rotation of local
coordinate system
Loading / Unloading
Seite 9
Crack Extension in FM Specimens
cohesive elements at
symmetry line: constraint
equations
u2( − ) = −u2( + )
u1( − ) = u1( + )
50
F [kN]
0 3
VLL [mm]
500 0
0 0 0,5 1 1,5 2 2,5 3
0 0,025 0,05 0,075 0,1
δn [mm] ∆a [mm]
CompFractMech_02 - WB ECF20 Trondheim 2014 20
Seite 10
Parameter Identification
50
F [kN]
W. BROCKS, D. ARAFAH:
40
Computational Fracture Mechanics
Students Project, Final report
30
January 2014
YYB-19-3
20
FE par-set (4)
10
3
0
VLL [mm]
0 0,5 1 1,5 2 2,5 3
2,5
VLL [mm]
2
σc = 2130 MPa = 3.28 R0
1,5
Γc = 94 kJ/m2 2500 YYB-19-3
σn [MPa]
FE par-set (4)
δ1/δc = 0.02 2000
(4)
1
1500
0
0
0 0,01 0,02 0,03 0,04 0,05
0 0,5 1 1,5 2 2,5 3
δn [mm]
∆a[mm]
Sensitivity: Effect of σc
50
σc / R0
F [kN]
40 2.31
30
3.08
YYB-19-1
sigma_c=1500 3.23
20 sigma_c=2000
sigma_c=2100 3.28
sigma_c=2130
10 sigma_c=3000 4.62
3
VLL [mm]
0
0 0,5 1 1,5 2 2,5 3 2,5
VLL [mm]
2
δc = 0.1 mm 1,5
δ1/δc = 0.02 1
YYB-19-1
sigma_c=1500
Seite 11
Sensitivity: Effect of δc
50
F [kN]
δc [mm] Γc [kJ/m2]
40
0.01 14.6
30 0.02 29.2
YYB-19-1 0.05 73
20 delta_c=0.01
delta_c=0.02 0.10 146
delta_c=0.05
10 delta_c=0.1
3
0
VLL [mm]
0 0,5 1 1,5 2 2,5 3
2,5
VLL [mm]
2
0
0 0,5 1 1,5 2 2,5
∆a [mm]
50
F [kN]
1
σ /σc
0,8
40
0,6
30 0,2
0,4
0,3
YYB-19-1 0,5
0,2
0,7
20 d2=0.2
d2=0.3 0
d2=0.5 0 0,2 0,4 0,6 0,8 1
10 δ / δc
d2=0.7
0 3
VLL [mm]
δc = 0.1 mm 1
YYB-19-1
d2=0.2
δ1/δc = 0.02 d2=0.3
0,5 d2=0.5
d2=0.7
0
0 0,5 1 1,5 2 2,5
∆a [mm]
Seite 12
Plane Stress Models
Problem:
Plastic localisation in continuum
elements under plane stress
conditions instead of separation of
cohesive elements Solution:
Transfer of thickness
information from continuum
element to cohesive element
Continuum
element
t
Sheet metal Al 5083 Cohesive
element
t = 3 mm
SCHEIDER & BROCKS [2006]
C(T): W = 50 mm
Seite 13
M(T): 2W = 300 mm
200 4
150 3
100 2
50 1
0 0
0 1 2 3 4 5 0 20 40 60
V LL [mm] a [mm]
CTOD R-Curves
3
δ5-Technique
C(T) + M(T) , W =50mm
C(T) test
C(T) FE pl. stress
M(T) test
2
M(T) FE pl. stress
0
0 5 10 15
a [mm]
Cohesive model captures
differences between C(T) and M(T)
Seite 14
ABAQUS
Continuum-Based Model
In 3D problems In 2D problems
• one direct (through-thickness) • one direct (through-thickness)
strain, strain,
• two transverse shear strains, • one transverse shear strain,
• six stress components. • four stress components.
Seite 15
C Model: Constitutive Response
Traction-Separation-Based Model
In 3D problems In 2D problems
three components of separations two components of separations and
and stresses, namely stresses, namely
• one normal and • one normal and
• two parallel • one parallel
to the interface to the interface
Seite 16
TS Model: Elastic
Initial Compliance
h0
that is K nn ≫ E
hel
Seite 17
TS Model: “Damage”
Damage:
• Once a damage initiation criterion
is met, material damage can occur
according to a user-defined
damage evolution law.
• A scalar damage variable, D,
represents the overall damage in
the material and captures the
combined effects of all the active
mechanisms.
TSLs
Traction-separation laws: 1
σ / σc
TSL
• linear softening, 0,8 linear
• exponential softening 0,6
tabular
exp: alf=10
• tabular (multilinear) function
0,4
0,2
0
0 0,2 0,4 0,6 0,8 1
δ / δc
Mixed-mode definition:
ABAQUS uses two measures of mode mix, one based on energies
and the other based on tractions.
Seite 18
”Damage“
D
described by functions σ(δ) and not
by evolution equations of a damage 0,8 "damage"
linear
variable. 0,6
tabular
exp: alf=10
0,4
• The variable D is not even capable of
significantly differentiate between the 0,2
different TSLs. 0
0 0,2 0,4 0,6 0,8 1
δ / δc
References
Seite 19