MLR Session 10 Ductile Brittle

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 172

1

Ductile fracture
Examples of applications
Benot Tanguy
Department of Nuclear Materials
Commisariat lEnergie Atomique
benoit.tanguy@cea.fr
With the acknowledged contribution of
J. Besson
W. Brocks
S. Marie
2
Motivations
3
Motivations
R
e
s
i
d
u
a
l
s
t
r
e
n
g
t
h
?
!
4
Ductile fracture
Ductile fracture mechanisms
5
Ductile fracture features
Fracture surface : presence of dimples
[Luu, PhD, 2006]
inclusions
6
Ductile micromechanisms
Industrial materials : presence of inclusions, precipitates
Several scales
7
Ductile micromechanisms
8
Ductile micromechanisms
9
Ductile micromechanisms
Nucleation : cavities creation (mechanisms to be defined!) related to plasticity
Mechanisms:
Voids issued form particules:
Carbides, precipitates, (micron to nanometer scale)
Iron carbides (Fe
3
C)
[Al-Li] (Turck, 2007)
10
Ductile micromechanisms
Dislocations pile-ups
Deformation localisation (involved in coalescence process)
[Zener mechanim]
[Tanguy, 2005] [Luu, PhD, 2007]
Void nucleation : Energetic criteria + stress criteria
(Surface creation)
(interface separation
Particle frcature)
11
Ductile micromechanisms
Void Coalescence : final stage of the failure process
Internal necking
Coalescence in shear (Void sheet : 2
nd
particules population at lower scale
involved)
Internal necking
12
Ductile micromechanisms
Void sheet
13
Short introduction to ductile fracture
Main features of ductile fracture
The three main steps of ductile fracture :
1. Formation of microvoids through decohesion of the metallic matrix around
inclusions or fracture of the particle,
2. Growth of these voids through plastic flow of the matrix
3. Colaescence of the voids and/or formation of the shear bands
The plasticity is in general, not confined to the vicinity of the crack tip front
Focus on macroscopic monotonic loading
14
Approaches for simulating crack extension
1. Morphological (Global approach)
no analysis of microscopic mechanisms mere transposition of the methods and
reasonings of LEFM (SSY assumption)
Node release + fracture mechanics criterion (R-curve)
No splitting of dissipation into global plasticity and local separation
Approach widely used in the industry because it is simple, but the
transposition of the results from one situation to another is somewhat
problematic (size effect, material heterogeneities, anisotropic behavior)
2. Cohesive surface
Interface elements with traction-separation law responsible for local separation
3. Continuum damage mechanics (Local approach)
Unified constitutive equations for deformation and damage, e.g. porous metal
plasticity
Relies on micro-macro analyses more reliable but still relatively little used
because of its complexity
S
i
m
p
l
i
c
i
t
y
C
o
m
p
l
e
x
c
a
s
e
s

d
e
s
c
r
i
p
t
i
o
n
15
Fracture of brittle material
el
c
( ) 2 a = = G
el el
el
rel
(2 ) (2 )
U U
B a B a

= =

G
sep
c
2
(2 )
U
B a

= =

( )
el
rel sep
crack
0 U U
A

crack extends if
energy release rate
separation energy (SE)
(energy per area)
fracture criterion
fracture stress c
2E
a

=
A.A. Griffith (1920)
(to create surfaces)
(more details in session 1)
16
Short introduction to ductile fracture
Typical load-displacement curve of a cracked specimen :
Small scale yielding
Large small yielding
Linear fracture
mechanics
Non-linear fracture
mechanics
R
p
<<B,W
17
Non linear elasticity: the J-integral
Assumptions:
Homogeneous, isotropic material
Small strains
Non linear elastic behavior:
n
B
0
=
Radial loading : no variation of the load trajectory
No direct extension to cyclic loadings
x

y
n
T(n)
ds


| || |

| || |

\ \\ \
| || |


= == = ds
x
u
). n ( T Wdy J
W : elastic energy density
T(n) : stress vector (stress acting on the contour)
x
u


: partial derivative of the displacement vector
ds : path increment on
J-integral independent on
J-integral: link between local parameters (near to crack tip) and global loading ->fracture
parameter
J-integral = G, energy release rate (surface energy: unit J/m
2
) (Budiansky and Rice,73)
Path-independant integral
(Cherapanov, 1967), (Rice, 1968)
Crack tip

=
ij ij
d w
(more details in session 5)
Stationnary crack
18
The J-integral
x

y
n
T(n)
ds


| || |

| || |

\ \\ \
| || |


= == = ds
x
u
). n ( T Wdy J
HRR stress field
(non linear elastic material)
) n , (
~
r I B
J
) n , (
~
r I B
J
ij
1 n
n
n 0
ij
ij
1 n
1
n 0
ij

| || |
| || |

| || |


\ \\ \
| || |
= == =

| || |
| || |

| || |


\ \\ \
| || |
= == =
+ ++ +
+ ++ +

+ ++ +


| || |

| || |

\ \\ \
| || |


= == = rd
x
u
T cos W J
r
1
~ d
x
u
T cos W

+ ++ +


| || |

| || |

\ \\ \
| || |



lorsque r 0
r
1
~
kl ij

1 n
n
ij
1 n
1
ij
n
0
r
1
~
r
1
~ B
+ ++ + + ++ +
| || |

| || |

\ \\ \
| || |

| || |

| || |

\ \\ \
| || |
= == =
Stress singularity controlled by J-integral
n
B
0
=
plastic behavior only valid for monotonic loading
J plays the role of an intensity factor like K in the case of LEFM
Small strain deformation theory
Elastic case: n=1
19
The J-integral
Fracture criterion : J=J
c
, J
c
material characteristic
Practical determination: ASTM standard E1820
0
1000
2000
3000
4000
5000
6000
7000
8000
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Ouverture (microns)
F
o
r
c
e

(

N

)
304L-1
Size criterion reached for little propagation
Value of J at initiation: J
i
Value of J after 0.2 mm crack propagation: J
Ic
20
The J-integral: limitation
For extending crack: -path dependence of J
-geometry dependence of J-Da curves
( )
el pl
ex
sep
crack crack
W
U U U
A A

= + +

pl
sep pl
diss
c
crack crack crack
U
U U
R R
A A A


= = + = +

energy balance
dissipation rate
commonly:
R
pl
causes geometry dependence
Problem: separation of R
pl
and
c
c
pl
R >>
21
Global approach: the J-integral
Size dependance of J
J increases due to the plastic
dissipation taking place in the
crack wake due to the
progressive elastic unloading,
due to the nonradial loadings
in the active PZ of a
propagating crack, and due to
change in the crack tip
geometry
J
R
very sentitive to sligth
changes of constraint
which affect a lot the
extrinsic plastic
dissipation -> Quantified
by the so-called Q stress
(2 parameters approach,
needs FE calculations)
22
Global approach: the J-integral
Size dependance of J
E36 Steel (-50C)
0
50
100
150
200
250
300
0.00 0.20 0.40 0.60 0.80
a/W
J
I
C
(
K
J
/
m

)
SENB 0.03 < a/W < 0.77
CCP 0.63 < a/W < 0.77
Sumpter
23
Ductile fracture
Continuum modelling of ductile damage
The so-called local approach
24
Application : fracture toughness fromCharpy tests
Tanguy and al.
25
Application : Charpy test modelling
Objective : simulation of the Charpy transition curve
26
Application : Charpy test modelling
27
Application : Charpy test modelling
28
Application : Charpy test modelling
29
Ductile failure model
30
Parameters adjustement
)]
2
1 ln( 1 [
2 2
aR
r a
zz

+ + =
31
Parameters adjustement
32
Mesh size dependance
In classical approach, mesh size is a compromise:
-description of the stress-strain gradients
-size of the FE calculations
-fracture energy related to mesh size
Notched specimen
[Mealor, 2004]
33
Mesh size dependance
Compact tension specimen
34
Mesh size dependance
Compact tension specimen
35
Mesh size dependance
0

a bB V W
R R
= =
0 0

b a B a bB S W
R R 0 0 0
) /( ) ( / = = =
Interpretation
Failure is a consequence of localisation in a strip of thickness b
Estimation of the apparent surface energy
Dissipated energy to fracture
Characteristic length: b =
In standard FE : b~~h
So that
h
0 0
=
h has to be fixed related to microstructure
36
Modelling of Charpy test
FE Mesh
37
Modelling of Charpy test
Ductile fracture simulation : Load-displacement curve
38
Charpy test modelling
39
Charpy test modelling
40
Charpy test modelling
DYN
VI
QS
Loading rate effect
Exp.
41
Towards ductile to brittle transition
42
Towards ductile to brittle transition
Constant value of the peak of the maximal principal stress with
increasing loading
CVN and CT
without damage
modelling:
43
Towards ductile to brittle transition
44
Towards ductile to brittle transition
45
Towards ductile to brittle transition
On-going work
46
Local approach- application 2
Anisotropic plastic models for sheets materials
Besson and al.
47
Sheet materials
48
Test specimens
49
Test specimens
50
Plastic anisotropy
51
Modelling plastic anisotropy
52
New model for plastic anisotropy
53
New model for plastic anisotropy
54
Yield surface shape
(biaxial loading)
1 E

55
Yield surface shape
56
New model for plastic anisotropy
57
Introducing plastic anisotropy in the GTN model
58
Ductile tearing of pipeline steel
59
Motivations
60
Motivations
61
Ductile tearing of pipeline steel
62
Outline
63
Sheet materials
64
Test specimens
65
Large facility at Arcelor-Mittal
66
Plastic anisotropy
67
Modelling plastic anisotropy
68
Modelling plastic anisotropy
69
Example : Fit for X100 steel
70
Application : limit load of a pressurized pipe under bending
compression
buckling
71
Ductile rupture modified GTN model
parameters
72
X70 steel
73
Wide plate test
74
Wide plate test: measurements
sudden propagation
75
Parameters adjustement
76
Characteristic length, mesh size
77
Simulation
Plate thickness
78
Crack tunneling
79
Plate thickness effect on Z
80
Through-thickness hardness gradient
81
X100 Europipe -- Plate
82
X100 ?
83
Plastic anisotropy
84
Fracture surface : two void populations
85
Same observations on Charpy specimens
86
Delamination on CT specimens
87
Results of mechanicals tests
88
Local approach
89
Local approach
90
Model
91
FE meshes
92
Elastoplastic anisotropic behevior
93
FE results for notched bars
94
FE results for plane strain specimens
95
FE results for Charpy specimens (static and dynamic)
96
FE results for CT specimens
97
Parametric study : yield surface
98
X100 NSC Plate & Prestrain
99
Prestrain at NSC
100
Application : toughness after UOE forming or ground movement
101
Similar characterization program as for the Europipe material
102
Cyclic behavior
103
Rupture anisotropy : notched bars
104
Rupture anisotropy : CT specimens
105
Prestraining reduces ductility
106
and crack growth resistance
107
CODLoad for 0% and 6% prestrain
108
Damage model with kinematic hardening
109
FE simulation of crack advance
110
Simulation of rupture anisotropy
111
Conclusions : main characteristics for ductile rupture of pipeline steels
112
Local approach
On-going work
113
Cohesive model
W. Brocks and al.
Use of cohesive elements for the
simulation of ductile tearing
114
Cohesive model
a
crack
Ligament
Separation of the
cohesive elements
Phenomenological representation
of various failure mechanisms
by cohesive interfaces
Implementation:
structure is divided into
material with elastic-plastic
properties
(continuum elements)
interface with damage
properties
(cohesive elements)
W. Brocks and al.
115
Cohesive law
Crack
Ligament
c
c n n n
0
( )d

two material parameters:


Cohesive stress,
c
(MPa) (maximum
stress carrying capacity of the
interface)
Critical separation,
c
(mm)
Traction Traction- -Separation Law (TSL) Separation Law (TSL)
Separation energy,
c
(J/mm
2
), defined by
( ) ( )
( ) ( )
n n
1 1
n 2 n 2
0 2 0 2
2
n 1
n n c n c 1 n 2
3 2
2 n c
2
( ) ( ) 1
2 3 1








<

= = < <

+ < <

f
116
Cohesive law
( )
( )
( )
( )
* act
n c n t
* act
t c t n


=
=
f g
f g
TSL for unloading TSL for unloading
( )
3 2
0 0
2 3 1


| | | |
= +
| |
\ \
g
( )
( )
act
*
act act
act
1 c
2

=
| |
+ <
|
\
f
f

n
TSL for mixed mode TSL for mixed mode
0.00 0.25 0.50 0.75 1.00
0.00
0.25
0.50
0.75
1.00
50.000
150.000
250.000
350.000
450.000
550.000
650.000
750.000
850.000
950.000

t
117
Implementation in FE model
Cohesive elements Cohesive elements
Plane
FE Models
e
1
e
2
linear 4 node element
2 integration points
3D
FE-Models
e
1
e
2
e
3
Variable Element (8-18 nodes)
4 or 9 integration points
No area or volume, resp., in the
unloaded/undamaged state
Implementation
Input: Nodal displacements
Calculation of stresses at the integration points
Output: Nodal forces,
element stiffness matrix
118
Micromechanical Mechanisms
void nucleation, growth
and coalescence
RVE unit cell
119
Parameter identification
Strategies:
Trial and error
Monte Carlo search
Minimisation algorithms
Evolutionary algorithms
Neural networks
Starting values
Comparison between
test results and simulations
of fracture tests
Increased dissipation
120
Trial and error
SE(B): BW s=2 4.5 18 mm
3
effect of
c
[kJ/mm
2
]

c
= 1000 MPa
121
Fracture mechanics specimens
122
Simulation of R-Curves
side-grooved C(T) and M(T) plane strain simulation StE460
123
Separation energy
results obtained with GTN model
Steady state
124
Simulation of Thin-walled specimens
Problem: Plastic localisation of plane stress
elements, if thickness reduction is not transferred
to the cohesive elements
1 Layer of plane strain elements
2 Layers of plane strain elements
Workaround: Definition of plane strain layers
along the ligament
Centre cracked panel M(T)
localisation
125
Simulation of Thin-walled specimens
Solution: Transfer of thickness information from
continuum element to cohesive element
Continuum
element
Cohesive
element
t
0 1 2 3
0
20
40
60
M(T) , W=50mm
V
LL
[mm]
test
pl stress const thickness
1 row pl strain
2 rows pl strain
pl stress w thickness red
zz

126
Slant fracture in Metal sheets
Mode I Mode II Mode III
Combined mode I+III fracture
projected to the structural plane
Cohesive parameters for mode I (normal
separation) in 2D simulation are
effective parameters of the mixed
mode case I+III.
Parameters differ from pure mode I fracture.
127
Crack extension in metal sheets: Al5083
a
0
=W/2
W
=

5
0

m
m
Static
displacement
controlled
tensile loading
Cohesive model with linear plane
stress elements CPS4
M(T)
C(T)
thickness 3mm
W = 50 mm
a
0
/W = 0.3
128
C(T)
material Al 5083
cohesive parameters:

c
= 560 MPa

c
= 10 kJ/m

c
= 0.024 mm
0 1 2 3 4 5
0
1
2
3
V
LL
[mm]
C(T), W=50mm
test
FE coarse
FE fine
FE extra
element size between
0.25 mm and 0.0625 mm
Results converge for decreasing element size
129
Transferability of cohesive parameters
0 1 2 3 4 5
0
50
100
150
200
V
LL
[mm]
0 20 40 60
0
1
2
3
4
a [mm]
test 1.1.3
test 2.1.8
FE pl stress
Simulation of M(T) specimen with 2W = 300 mm,
using cohesive parameters from C(T) tests
Good agreement for specimens with different size:
Transferability approved
130
Structural application: stiffened panel
131
Structural application: stiffened panel
Example for crack branching: Example for crack branching:
Structural integrity of a stiffened shell with crack between two stiffeners
c
r
a
c
k
Al 2024

Y
= 320 MPa

c
= 550 MPa

c
= 0.15 mm
Double symmetry
Static displacement controlled tensile loading
Cohesive elements along skin and stiffener
Finite element mesh
1980 Nodes
160 Cohesive elements
2225 Linear shell elements S3/S4R
Cohesive elements
132
Structural application: stiffened plate
Crack opening stress
133
Structural application: stiffened plate
R-curves for skin and stringer show that CTOD concept is not
applicable to branching cracks neither is J concept
134
Structural application: stiffened cylinder
p
cohesive
elements
0
.
2

m
Cylinder : 0.2 m
Wall thickness: 1 mm
Possible crack paths:
1. Through the stringer and along the panel
continuing in axial direction
2. Along the stringer
3. Debonding of the stringer
Depending on stringer characteristics
135
Structural application: stiffened cylinder
Symmetry: rotational (30) and longitudinal
Loading: Pressure controlled using RIKS algorithm (static)
4 Crack paths
Finite element mesh:
6542 Nodes
328 Cohesive elements
9193 Linear shell elements
Cohesive element length: 0.25 mm
136
Structural application: stiffened cylinder
Case 1 (Stringer thickness: 1,3 mm):
Crack runs through skin and stringer
Case 2 (Stringer thickness: 1.8 mm):
Crack propagates along the stringer
Max. principal stress
Debonding of the welded stringer did not occur in any case
crack tips
137
Structural application: stiffened cylinder
Stringer thickness: 1,3 mm
138
Structural application: stiffened cylinder
Stringer thickness: 1,8 mm
139
Structural application: stiffened cylinder
0
0,2
0,4
0,6
0,8
1
1,2
0 0,05 0,1 0,15 0,2 0,25 0,3
CTOD (mm)
P
r
e
s
s
u
r
e

(
M
P
a
)
t=1,3
t=1,8
t=2,4
Load-CTOD curves for varying stringer thickness
stringer thickness
Larger stringer thickness can prevent axial rupture and increase residual
strength but at the same time have negative effect on stability
Further increasing thickness has no effect on residual strength
140
Microscale application: Fibre debonding/breaking
e
x
is
t
in
g

c
r
a
c
k
f
ib
r
e

d
e
b
o
n
d
in
g
o
r

b
r
e
a
k
in
g
Material:
- Matrix Ti 6Al4V
(el.-pl.,
Y
=900MPa)
- Fibre SiC
(elastic, E=469GPa)
Crack will extend towards centre
At the fibre-matrix interface the crack
may
- penetrate the fibre, or
- run along the interface
Results depend on
- damage properties, and
- geometry
1
m
m
1mm
Unit cell containing a single fibre oriented in loading directio Unit cell containing a single fibre oriented in loading direction n
141
Microscale application: Fibre debonding/breaking
Damage properties:
Interface shear strength: 450 MPa
Interface normal strength: 1000 MPa
Fibre normal strength: 4400 MPa
Simulations with different fibre geometries: r
fibre
= 50 .. 150 m
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.05 0.1 0.15
E (.)
F
/
F
Y

(
.
)
0.75
0.8
0.85
0.9
0.95
0.1 0.12 0.14 0.16
E (.)
F
/
F
Y

(
.
)
rf=50m
rf=60m
rf=75m
rf=100m
rf=150m
Mesoscopic stress-strain curves
142
Microscale application: Fibre debonding/breaking
fully broken fibre
debonding at the interface
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.05 0.1 0.15
E (.)
Simulations with r
fibre
= 60 m
debonding at the top
crack tip
143
Conclusions
The cohesive interface is a phenomenological model with a sound
physical background
It does not need fracture mechanics parameters for establishing
crack extension criteria
Can be used with or without initial cracks (important e.g.
for debonding problems
Suitable meshing is crucial to realise the actual crack path
Model can also be used for branching cracks and for bifurcation
problems
Commonly 2 material parameters: cohesive strength and
separation energy (or critical separation)
Arising on the structural and on the material (micro-) scale
Cohesive interfaces cannot be used when
Location of cracking is not known at all
The crack path changes continuously
144
Applications
Ductile tearing test on a cracked ferritic pipe submitted to a four-point-bending
moment
Models under consideration :
J-Da approach (usual approach)
energetic approach : Ji-Gfr approach
local approach : Rousselier model
Material P : Tu52B ferritic steel
Element C Si P Mn S
wt % 0.19 0.45 0.016 1.29 0.027
Initial pipe dimensions :
outer diameter = 220 mm
inner diameter = 150 mm
Marie and al.
145
Applications (1/)
- 2 cracked Pipes : circumferential throughwall defect and semi-elliptical outer
defect
- Loading : 4 points bending moment
Defect angle 2.
146
Applications (1/)
550
1000 1000
1000
Sleeves (XC38 ferritic steel)
Intermediate pipe (Tu52b ferritic steel)
cracked pipe (Tu52b ferritic steel)
147
Application: cracked pipe
cracked pipe with a circumferential throughwall crack 2.=90
148
Application: cracked pipe
Strain gage
Ovalisation
Deflection
Rotation
Opening displacement
PDmeasurement
O-4
O-3 O-1
O-2
D-2
D-1
R-2 R-3
J-3
J-1 J-2
C-2
C-1
C-3
C-4
DDP-1
DDP-2
x
y
z

Instrumentation
149
Application: cracked pipe
results of the test
maximum ductile tearing of 28 mm
a - maximum a inner wall a outer wall maximum ()
Notch at +45 26,4 mm 20 mm 21,6 mm +64,7
Notch at 45 28,8 mm 26,4 mm 26,4 mm -67,5
0
20
40
60
80
100
120
140
0 10 20 30 40 50
LLD (mm)
M
o
m
e
n
t

(
k
N
.
m
)
150
Application: cracked pipe
0
100
200
300
400
500
600
0 2 4 6 8
a (mm)
J

(
k
J
/
m

)
J- a approach :
J-a curve from CT20 test up to 7.4 mm - linear extension up to 15 mm
Model parameters
151
Energetic approach: the J
i
-G
fr
approach
Simplified energy balance, no description of the mechanisms involved in ductile damage
Aim: descrition of large scale propagation
Load
a
a + da
Displacement
energy
dissipated
D
C
B
A
Energy dissipated from state A to C
Crack initiation : Ji
Energy dissipation rate R (Turner, 90)
R
dU
dA
d U U
dA
dU
dA
G
diss tot el
pl
fr
= =

= +
( )
Global plasticity energy
Calculation of a local energy release rate (Gl) ?
Energy dissipated by the fracture process
dA: crack extension
Separation of the two terms?
Assumption: crack growth is a discrete
sequential process
-Imposed displacement:increasing damage at
constant crack length
-crack extension at a critical damage at
constant load line displacement
152
Energetic approach: the J
i
-G
fr
approach
Load
a
a + da
Displacement
energy
dissipated
D
C
B
A
Energy dissipated from state A to C
(b) crack tip after the extension
Current plastic zone
Previous plastic zone : unloading area
Elastic domain
Discontinuous process (to avoid unloading)
Calculation of energy release rate associated
to this configuration
Crack extension to avoid interaction with
interaction with the initial plastic zone
(simultaneous node release technique)
Existence of a path independant value: Gl
Fracture criterion:
G
fr
(kJ/m
2
) dissipated energy
needed by the fracture process to
extend the crack area of a unit.
( )
G G G G G
l el c el fr
= + = + .

Dimensionless measure of crack


extension
153
Energetic approach: the J
i
-G
fr
approach
Load
LLD

2
=
1
+

4
=
1
+ 3.

3
=
1
+ 2.

5
=
1
+ 4.
(A
1
)
(D
1
)
(A
2
)
(D
2
)
(D
3
)
(D
4
)
(A
3
)
(A
4
)
(A
5
)
fr 1 pl 1 pl
G . ) A ( J ) D ( J =
fr 2 pl 2 pl
G . ) A ( J ) D ( J =
CMOD
Moment
Algorithm of the method from several stationary F.E. calculations
A
1
: crack initiation
D
1
determined from :
A
2
: point of the curve
2
for the imposed CMOD D
1
D
2
obtained from :
A
3
: ......
154
Application: cracked pipe
Model parameters
Energetic approach : Ji-Gfr approach (1/7)
From CT12.5 tests :
Crack initiation controlled by Ji = 100 kJ/m
Crack growth controlled by Gfr = 84 MPa
(slope of the J

,
pl
-a curve)
155
Application: cracked pipe
Model parameters
Local approach model parameters calibration
Rousselier model :
D: integration constant (between 1.5 and 2)
1: matrix strength,
f: void volumic fraction.
f0=initial void volumic fraction,
fc= critical void volumic fraction
other parameter :
mesh size Lc to simulate the crack growth.
( ) ( )
0 1
f 1
.
2
3
exp . D . f
f 1
1
m
*
1
*
eq
=
|
|

\
|

=
156
Application: cracked pipe
Model parameters
Several sets are tested to evaluate the best procedure for the calibration
D f
0
f
c
1
(MPa) (mm)
Set A 2 F.E. optimisation 0,05
2/3.
flow
F.E. optimisation
Set B 2 Franklin relation 0,05 F.E. optimisation F.E. optimisation
Set C 1.5 Franklin relation 0,05 F.E. optimisation F.E. optimisation
157
Application: cracked pipe
Model parameters
D f
0
f
c
1
(MPa) (mm)
Set A 2 5.10-5 0,05 300
Set B 2 1,4.10
-3
0,03 450
Set C 1.5 1,4.10
-3
0,03 375
0,15
0
5000
10000
15000
20000
25000
30000
0 1 2 3 4 5
CMOD (mm)
L
o
a
d

(
N
)
CT2 test
Set A
Set C
0
0.5
1
1.5
2
2.5
3
0 1 2 3 4
a (mm)
C
M
O
D

(
m
m
)
CT2 test
Set A
Set C
Same results with the 3 sets but different numerical stability
Set B shows early problem of convergence < > set A
SET A will be used for pipe tests FE calculations
158
Application: cracked pipe
Common assumptions to F.E. analyses
Geometry definition : 1 quarter of the pipe, straight crack front
Initial crack size : =50(correspond to the fatigue defect)
At the end of the pipe, thin rigid ring to take into account the embedded section
Modelling
159
Application: cracked pipe
0
20
40
60
80
100
120
140
0 1 2 3 4 5 6
CMOD (mm)
M

(
k
N
.
m
)
Test 1
J-Da approach
0
20
40
60
80
100
120
140
0 1 2 3 4 5 6
rotation ()
M

(
k
N
.
m
)
Test 1
J-Da approach
J- a approach :
exp max, M a J max, M
exp max, M a J max, M
exp a J
rotation rotation
CMOD CMOD
a a
>>
>>
>



& &
0
20
40
60
80
100
120
140
0 5 10 15 20 25 30
a (mm)
M

(
k
N
.
m
)
Test 1
J-Da approach
c
Maximum moment well predicted but
conservative prediction of the global
behaviour :
160
Application: cracked pipe
0
20
40
60
80
100
120
140
0 2 4 6 8 10
CMOD (mm)
M

(
k
N
.
m
)
Test 1
J-Da approach
Ji-Gfr approach
0
20
40
60
80
100
120
140
0 1 2 3 4 5
rotation ()
M

(
k
N
.
m
)
Test 1
J-Da approach
Ji-Gfr approach
Energetic approach :
J
i
-G
fr
approach
0
20
40
60
80
100
120
140
0 5 10 15 20 25 30
a (mm)
M

(
k
N
.
m
)
Test 1
J-Da approach
Ji-Gfr approach
c
161
Application: cracked pipe
0
20
40
60
80
100
120
140
-1.2 -1 -0.8 -0.6 -0.4 -0.2 0
strain gage J2 (%)
M

(
k
N
.
m
)
Test 1
Ji-Gfr approach
0
20
40
60
80
100
120
140
0 2 4 6 8
ovalisation ov2 (mm)
M

(
k
N
.
m
)
Test 1
Ji-Gfr approach
O-2
J-3
J-2
Energetic approach :
J
i
-G
fr
approach
162
Application: cracked pipe
Local approach : Rousselier model
Geometry definition : 1 quarter of the pipe, straight crack front
Initial crack size : =50(correspond to the fatigue defect)
27 elements along the crack front : average length = 0.55 mm
163
Application: cracked pipe
Local approach : Rousselier model
Early convergence problems, located at the inner surface crack front element due to
necking
none influence of 1) the element length in the trough-thickness direction 2) a pre-
damaged crack front elements
0
20
40
60
80
100
120
140
0 1 2 3 4
CMOD (mm)
M

(
k
N
.
m
)
Test 1
Rousselier model
164
Application: cracked pipe
Local approach : Rousselier model
Early convergence problems, located at the inner surface crack front element due to
necking
none influence of 1) the element length in the trough-thickness direction 2) a pre-
damaged crack front elements
0
20
40
60
80
100
120
140
0 1 2 3 4
CMOD (mm)
M

(
k
N
.
m
)
Test 1
Rousselier model
165
Application: cracked pipe
Local approach : Rousselier model
mesh size influence of the convergence study : meshes with = 0.3 and 0.45 mm
convergence problems appears later when the mesh size increase.
0
20
40
60
80
100
120
140
0 1 2 3 4
CMOD (mm)
M

(
k
N
.
m
)
Test 1
l = 0.45mm
l = 0.3 mm
l = 0.15 mm
166
Application: cracked pipe
Local approach : Rousselier model
crack initiation prediction : which criterion ?
1) initiation when one integration point reaches the critical voids fraction f
c
2) initiation when at least one element is fully failed
Test1 (from DDP) : Mini = 74.8 kN.m & Jini = 97.5 kJ/m
From CT12.5 tests : Jini = 100 kJ/m
criterion 1 criterion 2
M (kN.m) J (kJ/m) M (kN.m) J (kJ/m)
= 0.15 mm
61.0 58.3 75.7 100.7
= 0.3 mm
60.4 56.8 86.9 146.2
= 0.45 mm 101.8 230.4 109.1 284.3
= 0.3 mm + notch
102.4 234.2 106.3 262.4
Remark : J values are deduced from the F.E. calculation conducted for the J- a analysis
167
Application: cracked pipe
Choice of the new element shape :
The convergence of the calculation could be improved by decreasing the height h of the
elements in the crack growth area :
h = Lc / 2
168
Application: cracked pipe
New calibration of the parameters :
pre-identification :
D = 2 f
0
= 1,4.10
-3
( Franklin relation)
f
c
= 0,05
1
=
flow
= 450 MPa
CT test calculation : L
c
= 0,4 mm
0
0.5
1
1.5
2
2.5
3
3.5
4
0 1 2 3 4 5
Da (mm)
C
M
O
D

(
m
m
)
CT2 test
Rousselier model
0
5000
10000
15000
20000
25000
30000
0 1 2 3 4 5
CMOD (mm)
F

(
N
)
CT2 test
Rousselier model
169
Application: cracked pipe
Pipe test analysis
Geometry definition : 1 quarter of the pipe, straight crack front
Initial crack size : =50(correspond to the fatigue defect)
27 elements along the crack front : average length = 0.55 mm
170
Application: cracked pipe
Pipe test analysis
0
20
40
60
80
100
120
140
0 2 4 6 8 10
CMOD (mm)
M

(
k
N
.
m
)
Test 1
J-Da approach
Ji-Gfr approach
Rousselier model
0
20
40
60
80
100
120
140
-1 0 1 2 3 4 5 6 7
ov2 (mm)
M

(
k
N
.
m
)
Test 1
Ji-Gfr approach
Rousselier model
0
20
40
60
80
100
120
140
-1.2 -1 -0.8 -0.6 -0.4 -0.2 0
J2 (%)
M

(
k
N
.
m
)
Test 1
Ji-Gfr approach
Rousselier model
O-2
J-3
J-2
171
Application: cracked pipe
Pipe test analysis
Crack initiation : 1 element fully damaged (all integration points closed to the crack plan have reached f
c
)
Crack growth calculation : a = Area of fully damaged elements / Thickness
0
20
40
60
80
100
120
140
0 5 10 15 20 25 30
a (mm)
M

(
k
N
.
m
)
Test 1
J-Da approach
Ji-Gfr approach
Rousselier model
172
Application: cracked pipe
Conclusions
3 approaches have been considered : J-a, J
i
-G
fr
and Rousselier model
The 3 approaches give good initiation prediction
J-a approach leads to a good estimation of the maximum endurable moment but underestimate displacements (rotation,
CMOD,.)
J
i
-G
fr
approach gives a good prediction of the complete behaviour of the pipe
initiation Values at the maximum bending moment
M (kN.m) J (kJ/m) M (kN.m) Rotation () CMOD (mm)
a (mm)
Test 1 74.8 97.5 124.6 2.47 >4 6
J-

a
75.5 100 119.5 1.08 2.87 6
Ji -Gfr 75.5 100 126.8 2.25 6 11
Rousselier 75.7 100.7 - - - -

You might also like