You are on page 1of 11

Proceedings of the ASME 2013 Pressure Vessels and Piping Conference

PVP2013
July 14-18, 2013, Paris, France

PVP2013-97177

BENCHMARK ON RESIDUAL STRESS MODELING IN FRACTURE MECHANICS


ASSESSMENT

S. Marie H. Deschanels
CEA DEN/DM2S/SEMT/LISN Areva NP, 10 rue Juliette Recamier
CE Saclay 91191 Gif-sur-Yvette Cedex, France 69006 Lyon, France
stephane.marie@cea.fr hubert.deschanels@areva.com

S. Chapuliot P. Le Delliou
AREVA NP PEER-F EDF R&D
Tour AREVA, 92084 Paris La Défense Cedex, France Site des Renardières
stephane.chapuliot@areva.com Route de Sens, Ecuelles
77250 Moret sur Loing Cedex, France

ABSTRACT
1 INTRODUCTION
In the frame of development in analytical defect assessment Consideration of residual stresses in defect assessments may be
methods for the RSE-M and RCC-MRx codes, new work on the necessary in some cases where the material has a limited plasticity
consideration of residual stresses is initiated by AREVA, CEA and when the fracture toughness is achieved as in the case for brittle
EDF. The first step of this work is the realization of a database of F.E. materials or low ductility materials. Indeed, in other circumstances, the
reference cases. materials used for nuclear applications are sufficiently ductile and
To validate assumptions and develop a good practice guideline for tough, so residual stresses are 'erased' by plasticity.
the consideration of residual stresses in finite element calculations, a In the frame of development in analytical defect assessment
benchmark between AREVA, CEA and EDF is going-on. A first methods for the RSE-M and RCC-MRx codes, new work on the
application presented in this paper focuses on the analysis of the crack consideration of residual stresses is initiated by AREVA, CEA and
initiation of aged duplex stainless steel pipes submitted to an EDF. This work is first based on the realization of a database of F.E.
increasing pressure loading. Residual stresses are related to pipe reference cases.
fabrication process and act as shell bending condition. Two tests were To validate assumptions and develop a good practice guideline for
performed: the first with an internal longitudinal semi-elliptical crack the consideration of residual stresses in finite element calculations, a
and the second with an external crack. benchmark between AREVA, CEA and EDF is achieved. A first
The analysis first focuses on the ability to accurately estimate the application presented in this paper focuses on the analysis of the crack
measured pressure at the crack initiation of the two tests. For that initiation of aged duplex stainless steel pipes submitted to an
purpose, the comparison of results obtained with different methods of increasing pressure loading. Residual stresses are related to pipe
taking into account the residual stresses (i.e. thermal fields or initial fabrication process and act as shell bending condition. Two tests were
strain field). It then validates post-treatment procedures for J or G performed: the first with an internal longitudinal semi-elliptical crack
determination, and finally compares of the results obtained by the and the second with an external crack.
different partners. The analysis focuses on the ability to accurately estimate the
measured pressure at the crack initiation of the two tests, the
It is then shown that the numerical models can integrate properly comparison of results obtained with different methods of taking into
the impact of residual stresses on the crack initiation pressure. Then, account the residual stresses (i.e. thermal fields or initial strain field).
an excellent agreement is obtained between the different numerical It then validates the post-treatment procedures for J or G determination
evaluations of G provided by the participants to the benchmark so that and finally compares the results available today within the 3 partners:
best practice and reference F.E. solutions for residual stresses this paper presents the results obtained by CEA on the two reference
consideration can be provided based on that work. cases using CASTEM. A first comparison with calculations provided
by AREVA (and SYSTUS code) is also presented.

1 Copyright © 2013 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/24/2015 Terms of Use: http://asme.org/terms


2 PRESENTATION OF THE REFERENCE CASES semi-elliptical longitudinal defect was machined by EDM at the
external surface, without pre-cracking by fatigue.
2.1 REFERENCE CASE N°1: LONGITUDINAL INTERNAL SEMI-
ELLIPTICAL DEFECT IN DUPLEX STYLE PIPE 2.2.2 Geometry

2.1.1 Introduction The pipe geometry is similar to that of the case 1. The external semi-
The reference case #1 is a cast pipe in duplex stainless steel aged elliptical defect has a depth a = 26.8 mm and a surface half-length
8000 hours at 400 °C and pressure-loaded at room temperature. A c = 149 mm.
semi-elliptical longitudinal defect was machined by EDM in surface,
without pre-cracking by fatigue. The residual stress field induced by 2.2.3 Residual stresses
the manufacturing process of the tube, has been characterized by TWI Residual stresses are identical to those measured for the test case 1.
at the surface by the hole method and through the thickness by the a
destructive method.
2.2.4 Material data at room temperature
2.1.2 Geometry The following table summarizes the tensile properties of the
material at room temperature. Figure 2 presents the tensile curve.
The pipe has an internal diameter of 300 mm and a thickness of 50
mm. The total length of the pipe is 1 m. The ends are closed with flat Material Aged 1000h at 400°C
bottoms. The internal semi-elliptical defect has a depth a = 26.8 mm Young Modulus E (MPa) 195000
and a surface half-length c = 149 mm. 0.3
Poisson coefficient
Yield stress Rp0.2 (MPa) 375
Ultimate stress Rm (MPa) 670
2.1.3 Residual stresses
Residual stresses were measured in the English laboratories of Toughness tests were performed at 20°C with CT specimen, in the
TWI at the surface by the ‘hole method’ and through the thickness by a CR direction (direction of propagation of the defect). The toughness
destructive method. For these measurements, a pipe section with a value J0,2 was estimated between 10 and 15 kJ/m².
length of 870 mm was provided. Figure 1 summarizes the results of
the measurements of the axial and circumferential stresses. 2.2.5 Burst test
.
The burst test was conducted at room temperature. The pressure at
tearing initiation was estimated between 510 and 540 bars (using the
2.1.4 Material data at room temperature
electric potential method) and the burst occurred at 720 bars. The
The following table summarizes the tensile properties of the crack then propagated longitudinally up to the circumferential welds at
material at room temperature. Figure 2 presents the tensile curve. each ends of the pipe (unstable propagation phenomenon).

Material Aged 8000h at 400°C 3 RESIDUAL STRESSES MODELLING USING AN


Young Modulus E (MPa) 197500 EQUIVALENT THERMAL FIELD
Poisson coefficient 0.3
Yield stress Rp0.2 (MPa) 351 3.1 RESIDUAL STRESSES FIELD BUILDING
Ultimate stress Rm (MPa) 713
In the analysis of both test cases, the residual stresses are
introduced into the mechanical calculation via an equivalent thermal
Toughness tests were performed at 20°C using 5 precracked
field. This is particularly relevant in this case, because the residual
specimens CTJ22.5, taken in the CR direction (direction of
stresses field is practically biaxial as shown in Figure 3.
propagation of the defect). The toughness value was estimated at
26 kJ/m².
The developed thermal field in this analysis presents a shape close
to the residual stresses distribution (figure 4). This field is optimized
2.1.5 Burst test directly from thermal finite element calculations on the pipe mesh
The burst test was conducted at room temperature. The pressure at (Figure 5, for these calculations, the crack is blocked as the ligament),
tearing initiation was estimated at 224 bars (using the electric potential using quadratic elements. The optimization allows to obtain the final
method) and the burst occurred at 324 bars. The crack then propagated residual stress distribution shown in Figure 4 (identical distributions
longitudinally up to the circumferential welds at each ends of the pipe for the axial direction and the circumferential direction). Figure 6
(unstable propagation phenomenon). shows the final distribution of residual stresses in the pipe. For this
optimization, the coefficient of thermal expansion is fixed to
10.10-6 mm/mm/°C.
2.2 REFERENCE CASE N°2: LONGITUDINAL EXTERNAL SEMI-
ELLIPTICAL DEFECT IN DUPLEX STYLE PIPE 3.2 REFERENCE CASE 1 ANALYSIS
The finite element analysis of the cracked pipe is made first with a
2.2.1 Introduction purely elastic behavior and then with an elastic-plastic behavior.
The reference case #2 is a cast pipe in duplex stainless steel aged Figure 2 compares the stress-strain curves for the two materials of test
1000 hours at 400 °C and pressure-loaded at room temperature. A cases (having a slightly different aging). It appears that in the end these
two curves are very close and all analyzes are conducted with the same

2 Copyright © 2013 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/24/2015 Terms of Use: http://asme.org/terms


strain curve, namely the curve corresponding to the case 2. Young's Figure 14 shows the J variation at the deepest point with the
modulus for the calculation is 200 GPa. pressure, taking into account or not the residual stresses. In this figure
are shown the material toughness (between 10 and 15 kJ/m²) and the
The calculations take into account the crack: only the ligament is interval corresponding to the pressures at burst, estimated during the
blocked (using a symmetry condition). The lips of the crack are left test. More precisely:
free to move. The loading sequence is the following: - For the test, crack initiation occurs at a pressure between 51
- Imposition of the virtual thermal field (the crack is closed); and 54 MPa,
- Opening of the crack (in one step) - For the calculation under pressure only, the pressure at
- Imposition of the external loading initiation is between 32 and 37.5 MPa,
- For the calculation combining residual stress and pressure,
Figure 7 compares the evolution of J along the crack front in the the initiation pressure is between 49 and 53 MPa.
case of elastic residual stress and a pressure of 50 MPa. We note that
the residual stresses lead to a limited crack loading. Therefore, the 4 RESIDUAL STRESSES MODELING USING AN INITIAL
elastic-plastic calculation leads to a distribution comparable to the STRAIN FIELD
residual stress field (Figure 8). Conversely, for a pressure of 50 MPa,
the elastic-plastic amplification is significant (Figure 9). 4.1 RESIDUAL STRESSES FIELD BUILDING
We now consider the interaction between pressure loading and
residual stresses. For the internal defect, the residual stresses open the Previous analyzes are carried out by modeling the residual stresses
crack lips and therefore act in concert with the load pressure. Figure 10 field using an equivalent thermal field. This method becomes difficult
shows the profile of J obtained for the combined loading residual to use when the residual stress field is not biaxial or when the external
stress plus pressure 50 MPa. It appears from this figure that this loading is coming from a thermal transient.
variation is comparable to the variation obtained by the simple
summation on individual stress intensity factors (deduced directly It is possible to perform a mechanical analysis with the CEA F.E.
from J values) obtain for each loading component (the residual stress software CASTEM introducing in the loading conditions the initial
and pressure of 50 MPa for elastic –plastic), which leads to write: strains field. To test this solution, we conducted a new analysis of
reference case 1 with this method. The initial deformation field is
2 simply deducted from the temperature field considered in the chapter
JP res
JP J res
3:

In other words, this result implies that the kth* term, which represents 0 . T.Id
the elastic-plastic correction applied on the secondary elastic stress
intensity factor, is equal to 1. Figure 12 shows the evolution of J along the crack front for residual
stresses alone in the elastic case, for both methods of residual stresses
Elastic-plastic calculations were performed considering an modeling when using a conventional G( ) [2] formulation. A large
increasing pressure with or without residual stresses. J variation at the difference is observed at the crack center, whereas the mechanical
deepest point of the defect (where the value of J is maximal) are built fields are strictly identical. This result illustrates the need to modify
and compared to the toughness of the material and the pressure
the G( ) procedure for imposed strain field loading, because like for
detected during burst test in Figure 11. We note that the consideration
thermal loading, additional terms are needed.
of residual stresses is necessary to correctly predict the initiation
Both analyzes are in principle identical. This is what we illustrate in
pressure. Specifically, we obtain:
figure 15 where the opening displacement and opening stress are
identical for the two loading models.
- Pinitiation_F.E,,Ponly = 39.2 MPa (J0.2 = 26 kJ/m²)
- Pinitiation_F.E.,P+sig_res = 18.4 MPa
- Pinitiation,experimental_value_DDP = 22.4 MPa 4.2 J CALCULATION – ADDITIONAL TERM TO DEAL WITH THE
IMPOSED STRAINS LOADING CONDITIONS

3.3 REFERENCE CASE 2 ANALYSIS 4.2.1 Recall of the G( ) approach principle


The same approach is applied to the case 2. Figure 12 shows the
distributions of elastic J along the crack front for loading residual We recall that the energy release rate G is defined as the decrease
stresses, a pressure of 50 MPa alone and both combined loadings. For of the total potential energy with an increase in crack area dA:
external defect, the residual stresses close the crack lips and therefore
oppose the opening induced by the pressure, which explains the lower G=-d / dA.
values of J for the combined loading case than for pressure only.
The G( ) method introduced by Destuynder [1] consists in the
For the elastic-plastic case (Figure 13), we note that the evolution calculation of the derivatives of quantities versus crack extension
of the elastic-plastic combined loading is still equivalent to that using the Lagrangian method. A field is introduced to represent the
constructed by summation of the individual stress intensity factors position of each point of the perturbed body, according to its
obtained for the pressure alone and residual stresses. It thus confirms a coordinates, with the infinitesimal extension of the crack :
value close to 1 for the kth* parameter, assuming a negative value for M M
the stress intensity factor due to residual stresses.

3 Copyright © 2013 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/24/2015 Terms of Use: http://asme.org/terms


Mechanical fields in the perturbed configuration can be approximated
by: 0
. 1
0 0 . k
xk
U U .U1
The potential energy in the initial state and perturbed state are
Where 1 et U1 are the first order variations of U and when the written:
body is slightly disturbed.
Similarly, changes in load field and temperature can be expressed 1 1
in function of their gradient: Tr ( . U).d Tr ( . 0 ).d
2 2
F F .. F 1 1
Tr ( .gradU ).d Tr ( . 0 ).d
p p .. p 2 2
T T . . T
The development of these equations in the original undisturbed
system and the passage to the limit allows the formulation of the term
The energy release rate G is deduced by:
G and the extraction of an additional term to the original formulation
related to the initial deformation field:
G lim
0
0
G0 Tr . k d
In presence of external mechanical stresses alone, the potential xk
energy in the initial state and perturbed state is:
The implementation in CAST3M of this term allows to get exactly
1 same results when the J calculation is made using the imposed strain
Tr ( . U).d
2 loading condition compared to the values obtained with the residual
1 stresses are modeled using an equivalent thermal field.
Tr ( .gradU ).d
2
5 COMPARISON TO AREVA ANALYSES
The perturbation being infinitesimal, operations of derivation and AREVA conducted a similar analysis on the reference case 2 on the
integration on the disturbed body can be expressed with a linearized pipe with a longitudinal external semi-elliptical defect. Residual
approximation considering the same operations on the undisturbed stresses are modeled by equivalent thermal field identified by EDF:
body:
T(r) = 1985,298-25,30867.r+0,080489.r²
grad( ) .(Id )
This thermal field is associated with a thermal Young's modulus of
d .(1 .div ).d 190 GPa and a thermal expansion coefficient of 12.105.10-6
mm/mm/°C.
From these equations and introducing the perturbed fields by their AREVA provided J variations calculated for the reference case 2
linearized equivalent, it is possible to express the potential energy taking into account or not residual stresses (Figure 16): for the loading
difference between the initial state and the perturbed state relying only pressure only, CEA and AREVA results are identical.
on the initial state information.
When residual stresses are considered, AREVA predicted a crack
The details of the analysis can be found in reference [2]. Finally, we initiation for a pressure between 55 and 58 MPa, results close to the
can show that: experimental results. But, the evolution of J provided by AREVA when
residual stresses are taken into account is shifted towards higher
G Tr ( . U. ).d w.div .d pressures compared to the CEA results (about 5 MPa).

1 The main difference in the two analyzes is the thermal field


w Tr ( . U)
2 equation used to calculate residual stresses. CEA performed then a
new complete analysis of the reference case 2 using the thermal field
provided by EDF. Figure 16 compares the new evolution of J obtained
4.2.2 Consideration for an initial strains field by the CEA with the evolution AREVA: For the same thermal field,
As for the consideration of a thermal field, it is necessary to the variations are identical validating both the mechanical calculation,
incorporate changes of initial strains field. Compared to the and therefore how to integrate the residual stresses and the calculation
temperature that is a scalar, the difficulty here is to adapt the procedure of J.
demonstration to a tensor.

This variation of the initial strains field is expressed:

4 Copyright © 2013 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/24/2015 Terms of Use: http://asme.org/terms


CONCLUSION Finally, comparing the results with those provided by AREVA on
case 2 for an identical thermal field show that the two F.E. codes used
This paper presents the prediction of two tests consisting of two by both partners give the same results.
burst tests on pipes made in aged duplex stainless steel, with an
internal (test case 1) or external (test case 2) longitudinal semi-
elliptical defect. REFERENCES
A first modeling of these tests is performed representing the
residual stresses, associated with the manufacturing process and [1] Ph. Destuynder, M. Djaoua, M. et S. Lescure, ‘Quelques
measured by TWI, using an equivalent thermal field. The analyses Remarques sur la Mécanique de la Rupture Elastique’,
show the importance of taking into account the residual stresses to Journal de Mécanique Théorique et Appliquée, Vol. 2,
predict correctly the pressures at crack initiation in that case of low N°1, 1983, pp. 113-135.
ductility material, and validate the physical meaning of the J parameter
through the calculation methodology. [2] J. Brochard, T. Charras, A. Combescure, X.Z. Suo et D.
The modeling of residual stresses via an initial strains field is then Uhlmann, ‘Numerical evaluation of the energy release rate
discussed, this field is directly deducted from the thermal field used in for nonlinear fracture mechanics formulation of G-
the initial analyzes. Comparison of mechanical fields near the crack tip method and application with CASTEM 2000’, SISSI 94,
confirms the equivalence of the two approaches. However, it is shown Saclay, Vol. 2, 1994, pp. 211-215.
that an additional term in the G calculation procedure is necessary in
and successfully implemented in order to obtain a good evaluation of
the G parameter..

Figure 1: Residual stresses profiles measured by TWI.

5 Copyright © 2013 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/24/2015 Terms of Use: http://asme.org/terms


800 200
Axial stresses
700 150
Circumferential stresses
600 100
Calculated stresses
Stress (MPa)

500 50

Stresses (MPa)
400 reference case 1 material 0
0 10 20 30 40 50
300 reference case 2 material -50
Distance / inner skin (mm)
200 -100

100 -150

0 -200
0 0.5 1 1.5 2 2.5 3
-250
Strain (%)
Figure 2: Comparison of stress-strain curves of two reference cases.
Figure 3: Final results of the optimization of the initial thermal field:
comparison of the distribution of residual stresses with the
measurements made by TWI.

Figure 4: Schematic representation of the radial variation of the


temperature field imposed in the tube. Figure 5: Pipe mesh for the reference case 1.

6 Copyright © 2013 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/24/2015 Terms of Use: http://asme.org/terms


Figure 6: zz stress distribution (axial) and yy (normal to the plane of crack) the final residual stress field.

45 8
41 kJ/m²
40 7
35
6
30
J (kJ/m²)

5
25
J (kJ/m²)

4
20 Residual Residual stresses - elastic behaviour
stresses 3
15
Residual stresses - elastic-plastic behaviour
10 6.8 kJ/m²
2

5 1
0
0
0 50 100 150 0 30 60 90 120 150
Deep point z (mm) Surface point
Deep point z (mm) Surface point
Figure 7: Profile of J variation for an elastic behavior for the reference
case 1, where z is the curvilinear abscissa on the crack front (z=0 at the Figure 8: Impact of the elastic-plastic behavior on the J variation due to
deepest point). the residual stresses field for the reference case 1.

7 Copyright © 2013 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/24/2015 Terms of Use: http://asme.org/terms


60

50

40
J (kJ/m²)
30

Residual stresses - elastic-plastic behaviour


20 Pressure only - 50 MPa - elastic behaviour

Pressure only - 50 MPa - elastic-plastic behaviour


10

0
0 30 60 90 120 150
Deep point z (mm) Surface point
Figure 9: Impact of elastic-plastic behavior on the J variation due to a pressure of 50 MPa for the reference case 1.

140 Residual stresses - elastic-plastic behaviour


Pressure only - 50 MPa - elastic-plastic behaviour
Residual stresses + Pressure (50 Mpa)- F.E. - elastic-plastic behaviour
120 Residual stresses + Pressure (50 Mpa)- F.E. - elastic behaviour
Residual stresses + Pressure (50 Mpa)- summation of individual elastic-plastic KI
100

80
J (kJ/m²)

60

40

20

0
0 30 60 90 120 150
Deep point z (mm) Surface point
Figure 10: J variation for the combined loading residual stresses + internal pressure of 50 MPa for the reference case 1.

8 Copyright © 2013 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/24/2015 Terms of Use: http://asme.org/terms


200 30 Residual stresses - elastic behaviour
Jmax - residual stresses + Pressure
180 Jmax - Pressure only Pressure only - 50 MPa - elastic behaviour
J0.2 25
160
Experimental pressure at initiation Residual stresses + Pressure (50 Mpa)- F.E. - elastic behaviour
140
20
120
J (kJ/m²)

100

J (kJ/m²)
15
80
60
10
40
20
5
Pressure (MPa)
0
0 10 20 30 40 50 60 70
0
Figure 11: J variation with pressure, taking into account or not 0 30 60 90 120 150
residual stresses for the reference case 1. Deep point z (mm) Surface point
Figure 12: J variation of an elastic behavior for reference case 2.

45 Residual stresses - elastic-plastic behaviour


Pressure only - 50 MPa - elastic-plastic behaviour
40 Residual stresses + Pressure (50 Mpa)- F.E. - elastic-plastic behaviour
Pressure only - 50 MPa - elastic behaviour
35 Residual stresses + Pressure (50 Mpa)- F.E. - elastic behaviour
Residual stresses + Pressure (50 Mpa)- summation of individual elastic-plastic KI
30

25
J (kJ/m²)

20

15

10

0
0 30 60 90 120 150
Deep point z (mm) Surface point
Figure 13: J variation of an elastic-plastic behavior for reference case 2.

9 Copyright © 2013 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/24/2015 Terms of Use: http://asme.org/terms


80
Jmax - residual stresses + Pressure
70
Jmax - Pressure only
60
J0.2 - between 10 and 15 kJ/m²
50
J (kJ/m²)

Experimental pressure at initiation


40

30

20

10
Pressure (MPa)
0
0 20 40 60 80
Figure 14: J variation with pressure, taking into account or not residual stresses for the reference case 2.

0.25 Residual stresses + Pressure of 50 Mpa - imposed strains model 1400 Residual stresses + Pressure of 50 Mpa - imposed strains model
Residual stresses + Pressure of 50 Mpa - equivalent thermal field model
Opeining displacement (mm)

1200 Residual stresses + Pressure of 50 Mpa - equivalent thermal field model

0.2
Opening stress (MPa)

1000

800
0.15
600

400
0.1
200

0
0.05
0 10 20 30 40 50 60
-200
Thickness position (mm) Thickness position (mm)
0 -400
0 10 20 30 40 50 60

Figure 15: Opening displacement and opening stress near the defect - comparison of calculated fields with residual stresses modeled using a thermal
field (solution 1) or using an initial strains field (solution 2).

10 Copyright © 2013 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/24/2015 Terms of Use: http://asme.org/terms


80 Jmax - Pressure only - CEA
Jmax - Pressure only - AREVA
70 Jmax - Residual stresses & pressure - CEA - CEA thermal field
Residual stresses & pressure - AREVA - EDF thermal field
60 Jmax - Residual stresses & pressure - CEA - EDF thermal field
J0.2 - between 10 and 15 kJ/m²
50 Pressure at experimental initiation
J (kJ/m²)

40

30

20

10
Pressure (MPa)
0
0 10 20 30 40 50 60 70 80

Figure 16: Comparison of results obtained by AREVA and CEA for the reference case 2 - taking into account the EDF thermal field in the CEA
analysis.

11 Copyright © 2013 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 05/24/2015 Terms of Use: http://asme.org/terms

You might also like