You are on page 1of 8

Proceedings of PVP2008

2008 ASME Pressure Vessels and Piping Division Conference


July 27-31, 2008, 2008,
July 27-31 Chicago, Illinois,
Chicago, USA
IL, USA

Paper No. PVP2008-61369


PVP2008-61369
ALIGNMENT CRITERIA FOR
THROUGH-WALL FLAWS IN PLATES AND PIPES

B. Bezensek K. Miyazaki
Mechanical Engineering Department Hitachi Research Laboratory
University of Glasgow Hitachi Ltd.
Glasgow G12 8QQ Hitachi-shi, Ibaraki-ken
Scotland, UK Japan

b.bezensek@eng.gla.ac.uk katsumasa.miyazaki.xs@hitachi.com

ABSTRACT plane for the purpose of assessment. Aligned flaws


Multiple flaws in vessels and pipes are frequently are further considered as combined into one larger
assessed as a larger single flaw in accordance with flaw or assessed as independent planar flaws
the flaw alignment and combination rules. In this according to the flaw combination rules, updated in
paper the alignment of two through-wall flaws is the 2003 addenda to the code. Recent experimental
examined for plates in tension and pipes in bending data on through-wall flaws [2] indicates a possibility
using detailed finite element (FE) modelling. The FE of a non-conservative assessment of multiple non-
model is developed using the Gurson-Tvergaard aligned flaws when using the 12.4 mm alignment
damage model and accurately describes the flaw rule.
interaction and load capacity of a random
experimental sample. Results suggests that two Following the series of papers presented in the PVP
flaws should be aligned onto the same plane for the 2005 [3] and PVP 2006 [4] meetings this paper
purpose of assessment when the separation presents further studies on the development of
between the parallel planes containing flaws equals alignment criteria for through-wall flaws in plates
the flaw length for both, plates in tension and pipes in and pipes. The present results are obtained for
bending. strain-hardening behaviour by using stress-strain
data of a ferritic steel in conjunction with the
Gurson-Tvergaard damage model. The stress-strain
INTRODUCTION data was measured on the tensile specimens
Multiple flaws in vessels and piping are frequently extracted from plates used in the experimental
combined into a larger flaw or assessed as programme at Hitachi Research Laboratory [2].
independent flaws depending on the flaw alignment
and combination rules specified in fitness for service To ensure a conservative assessment of a multiple
codes such as Section XI of the ASME Boiler and flaw geometry as a geometry with a single flaw (i.e.
pressure vessel (BPV) code [1]. Flaws on multiple considering no interaction effects) the maximum
planes are first aligned (projected) on the same load capacity of the multiple flaw geometry must
plane or considered independent, depending on the equal or exceed that of a single flaw geometry. On
flaw alignment rule. In the ASME BPV code, Section this basis the recommended alignment distance
XI the alignment rule distinguishes between the in- between the planes containing parallel through-wall
plane and out-of-plane flaw separation and the flaws is determined by comparing the maximum
criteria for the out-of-plane separation (the alignment load for twin flaw geometry obtained from the FEA
rule) is currently a fixed number of 12.4 mm (1/2”). with the collapse load for a single flaw using the
Flaws that have the out-of-plane separation less than Appendix C [1].
this value are projected (aligned) onto the same

1 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/11/2013 Terms of Use: http://asme.org/terms


Experimental data: plate with two flaws [2]
Test 2w L S H Fmax,exp Fc,code Flaws
[mm] [mm] [mm] [mm] [kN] [kN] coalesce?

DN-02 108 10 10 12.4 221 192 Yes


DN-06 108 10 10 17 227 214 Yes
DN-07 108 10 10 22 232 214 No
DN-09 108 10 17 17 226 214 Yes
DN-10 108 10 22 22 227 214 Yes
DNW-02 180 30 0 10 298 263 Yes
DNW-03 180 30 0 20 307 328 No
DNW-04 180 30 5 20 303 328 Yes

Table 1a: Test results for non-aligned twin flaw geometries

Experimental data: plate with three flaws [2]


Test 2w L1 L2 S H Fmax,exp Fc,code
[mm] [mm] [mm] [mm] [mm] [kN] [kN]

TSH-6 108 29 30 10 10 71.7 43.7


TSH-7 108 19 30 20 10 123.1 87.4
TSH-8 108 29 30 10 20 96.8 109.3
TSH-9 108 19 30 20 20 131.9 153.1
TSH-10 108 29 30 10 30 114.7 109.3
TSH-11 108 19 30 20 30 144.9 153.1

Table 1b: Test results for non-aligned three flaw geometries

In the paper the finite element model is first


described. The model is based on the Gurson-
Tvergaard damage model with the void nucleation
option. The model is verified by modelling a random
experimental sample of plates containing through-
wall flaws [2]. The model is then applied to
configurations with twin non-aligned flaws in a plate
in tension and a pipe in bending.

EXPERIMENTAL DATA
Flat plates containing two or three non-aligned flaws
have been tested to failure in tension [2]. Material
was ferritic steel for welded structures JIS SM 400A.
A representative photograph of a test configuration is
shown in Figure 1 and nomenclature is defined in
Figure 2. Flaw separation parameters S (in-plane)
and H (out-of-plane) have been varied together with
the flaw lengths L1, L2 and the plate width 2w. Plate Figure 1: Photograph of a test configuration
thickness was 5.7 mm. Maximum load and mode of containing two non-aligned through-wall
failure were recorded and are listed in Table 1 flaws.
together with the flaw geometry. Each geometry is

2 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/11/2013 Terms of Use: http://asme.org/terms


Test ID σy,up σy,low σTS EL
[MPa] [MPa] [MPa] [%]
SL1 321.8 304.9 447.1 37.9
SL2 327.9 306.5 446.3 37.3
ST1 370.9 310.0 448.1 34.2
ST2 386.2 310.5 445.5 35.7
Approximate
engineering 315 315 447 36
curve
True stress- 315 315 546 20
Figure 2a: Illustration of a twin-flaw geometry.
strain curve

Table 2: Results of tensile tests on SM 400A steel.

Figure 2b: Illustration of a three-flaw geometry.

assessed in accordance with the ASME BPV code,


Section XI, IWA-3300 and Appendix C by
determining the collapse load from the material’s flow
stress and net section ligament. This value (Fc,code) is Figure 3: Stress-strain curves for the JIS SM400A
superimposed in Table 1 for each geometry. Majority carbon steel at ambient temperature.
of the tested geometries had the out-of-plane
separation H greater than 12.4 mm and are
according to current guidelines assessed as
Figure 3. Four specimens were tested. Two were
independent flaws. Test results identified a selection
extracted from the longitudinal direction and two
of geometries where multiple flaws have joined
from the transverse direction of the plate. The upper
(coalesced) before final failure. This in itself may not
and lower yield stress (σy,up, σy,low), tensile strength
be critical to the integrity of the component. Tests
have however shown that the maximum load (σTS) and elongation (EL) are shown in Table 2. For
(Fmax,exp) is in several geometries lower than the modelling purposes the engineering curve was
collapse load of a single flaw. Thus a possibility exist simplified by removing the upper yield point and
for a non-conservative assessment of non-aligned fitting a constant stress curve across the Lüder’s
flaws using the current flaw alignment rule of IWA- zone. The curve was then merged with the average
3300 and Appendix C collapse load solution. hardening behaviour of the four tensile tests, as
shown in Figure 3. This curve was then converted
into the true stress - true strain curve used in the
MATERIAL STRESS-STRAIN DATA finite element model.
Tensile tests were conducted on specimens
extracted from the remote ends of the test plate and
the measured stress-strain curves are shown in

3 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/11/2013 Terms of Use: http://asme.org/terms


DEVELOPMENT OF THE FINITE ELEMENT
MODEL
Experimental geometries were modelled using
ABAQUS/Standard v6.4 [5] finite element software.
Three-dimensional model was created using linear
solid elements and the mesh was focused towards
the central segment containing the through-wall
flaws. The dimensions of the finite elements in the
central tip-to-tip ligament were 0.5x0.5x0.7 mm to
allow accurate strain localisation to develop.

By implementing the large strain formulation and


using the true stress - strain curve both, geometric
and material non-linearities were included in the
model. To capture strain localisation and softening
behaviour in the tip-to-tip ligament the Gurson-
Tvergaard [6,7] damage model with the void Figure 4a: Force-extension plot for the
nucleation option was used. The Gurson-Tvergaard specimen TSH-6.
yield surface is given by the equation:

2
 q   3p 
Φ =   + 2q1f cosh − q2 − (1 + q3f 2 ) = 0
 σy   2σ 
   y
(1)
where q and p are the effective von Mises stress and
hydrostatic pressure respectively and f is the void
volume fraction. The initial void volume fraction was
set to 0. The remaining parameters are q1=1, q2=1.5,
q3=q12, consistent with the values in literature for
metallic materials [7]. The rate of nucleation of new
voids is proportional to the rate of equivalent plastic
strain εɺ pl
m , through a proportionality parameter, which
is a normally distributed function of plastic strain with
a mean value and a standard deviation:
Figure 4b: Force-extension plot for the
specimen TSH-8.
fɺ = Aεɺ pl
m
 2
fN  1  ε pl − ε N  
A= exp −  m 
 (2)
sN 2π  2  s N 
 
 

The mean value, ε N and the standard deviation,


sN were 0.3 and 0.1 respectively, consistent with the
values used in the literature [8]. The volume fraction
of nucleating voids, fN was determined by matching
the FEA model with the force-extension curves of the
three experimental geometries, as discussed below.
No additional failure criteria was implemented.

Verification of the FEA model

Force - extension curves are shown in Figure 4 for the Figure 4c: Force-extension plot for the
three-flaw geometries: TSH-6, TSH-8 and TSH-10. specimen TSH-10.
Superimposed are predicted force – extension

4 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/11/2013 Terms of Use: http://asme.org/terms


Figure 5: Comparison of the maximum forces Figure 6: Flaw alignment distance for two
from the experiments and the FEA model through-wall flaws in a plate in
utilising the Gurson-Tvergaard damage tension.
model.

curves obtained form the finite element model with of the material according to the Appendix C of BPV,
and without the Gurson-Tvergaard model and Section XI:
associated strain nucleation option. By setting the
volume fraction of nucleating voids, fN to 0.12 the Fc,single = σ flow ⋅ t ⋅ (2w − L) (4)
maximum force and the softening behaviour was
closely matched across the three geometries. where σ flow is the flow stress, defined as the
Thus calibrated FE model was applied to the arithmetic mean of the average lower yield and the
remaining two and three flaw geometries used in the average tensile strength measured on the tensile
experimental work (see Table 1). The predicted tests, t is the thickness, 2w is the plate width and L
maximum forces are plotted against the measured the flaw length.
maximum forces in Figure 5. An excellent agreement
can be observed for all plate geometries. The critical flaw alignment distance H was
determined from the criteria Fmax,two = Fc,single for a
range of flaw geometries, defined by systematically
ALIGNMENT OF THROUGH-WALL FLAWS IN A varying the flaw separation distances S and H, flaw
PLATE IN TENSION length, L and plate width, 2w. The results are shown
in Figure 6 in a form of a critical flaw alignment
To ensure a conservative assessment the two flaws distance H plotted against the plate width, 2w, with
on parallel offset planes are aligned onto the same both values non-dimensionalised by the flaw length,
plane for the purpose of engineering assessment if L. Results in Figure 6 apply to two flaws of equal
the maximum load capacity of the two flaw length which is the most pessimistic configuration.
configuration (Fmax,two) becomes less or equal to the To fulfil the above criteria, the flaws have to be
collapse load of a single flaw: aligned when the flaw separation H becomes less
than the flaw length L for all flaw length to plate
width ratios. An influence of the in-plane separation
Fmax,two ≤ Fc,single ⇒ flaws aligned
S is also observed for the larger flaws (small w/L)
Fmax,two > Fc,single ⇒ flaws independent with the largest separation required for the S=H
(3) geometry. This is consistent with the limit load
The collapse load of a single flaw is determined by solutions of Bezensek [4] for a non-hardening
multiplying the net cross-section with the flow stress material.

5 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/11/2013 Terms of Use: http://asme.org/terms


Figure 7: Illustration of a pipe containing two
through-wall flaws.

Figure 8: Flaw alignment distance for twin through-


wall flaws in a pipe subject to bending.
ALIGNMENT OF THROUGH-WALL FLAWS IN A
Pipe wall thickness is 6 mm.
PIPE IN BENDING

Geometry
A pipe containing two non-aligned through-wall flaws
is illustrated in Figure 7. Pipe outer diameter D(=2R)
and flaw length Lo measured at the outer diameter
are systematically varied to examine a range of flaw
geometries. Two wall thicknesses are considered
spanning a typical range of pipe wall thicknesses: t=6
mm and 12 mm. Pipe is subject to remote bending
moment such that the tip-to-tip ligament is under the
action of tensile stresses.

Procedure
The alignment distance at which two non-aligned
flaws are aligned on the same plane is determined
by comparing the maximum bending moment from
the FEA twin-flaw model with the collapse moment
for a single flaw. The later is obtained from the critical
stress for a single flaw using Section XI, Appendix C,
Section C-5321: Figure 9: Flaw alignment distance for twin through-
wall flaws in a pipe subject to bending.
Pipe wall thickness is 12 mm.
2σ flow  a 
σ b,c = 2 sin β − t sin θ
π  
(5)
1 a σ  If the calculated bending moment for a given flaw
β =  π − θ − π m 

2 t σ flow  separation (S and H) is greater than the collapse
moment of a single flaw, the interaction between the
flaws is not critical and the flaws will be
Where σb,c is the bending collapse stress, a is the
conservatively assessed as single flaws (i.e. non-
flaw depth (here equal to thickness) and θ is the flaw
aligned). Conversely, when the flaws are sufficiently
half span. As there is no externally applied
close, the calculated bending moment will be less
membrane stress the σm is zero. Although this than the collapse moment of a single flaw. In such
equation is derived for surface flaws with a/t ≤ 0.75, it configurations the flaws should be aligned (i.e.
is applied here as a proxy for a collapse stress of projected) onto the same plane for the purpose of
through-wall flaws. The collapse moment is then assessment.
derived from equations of Section C-2500.

6 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/11/2013 Terms of Use: http://asme.org/terms


Figure 11: Illustration of the flaw interaction vs
ovalization of the pipe by comparing
Figure 10: Plots of equivalent plastic strain illustrating bending moments for a two flaw and
the competition between the flaw no flaw configurations.
interaction and pipe ovalization.

In addition to the bending moment based criteria a geometry a significant strain localisation develops
visual check was also made to ensure a significant between the tip-to-tip ligament at the maximum
strain localisation develops in the tip-to-tip ligament bending moment. This is identical to the strain
as confirmation of the flaw coalescence. localisation observed in modelling a test plate with
through-wall flaws. By reducing flaw size such as to
Results give D/Lo=14, a minimal strain localisation is
Results in Figures 8 and 9 show the flaw alignment observed between the tip-to-tip ligament at the
distance H as a function of pipe outer diameter to maximum moment Mmax and the pipe ovalizes.
flaw length ratio, D/Lo. Values are shown for three in-
plane separations: S=H, S=0.5H and S=0.25H. Calculations were also performed on these
Figure 8 shows results for the pipe wall thickness of geometries without any flaws to determine the
6 mm and Figure 9 for the pipe wall thickness of 12 ovalization moment of an uncracked pipe. All other
mm. parameters were kept identical. The maximum
bending moments (or ovalization moments) are
Clear trends between the flaw alignment distance H superimposed with the results of the twin-flaw
and the flaw length L can be observed for relatively configurations in Figure 11. For long flaws
long flaws compared to the pipe diameter – up to (D/Lo=3.3) interaction precedes the pipe ovalization
D/Lo=9.5 for t=6 mm and D/Lo=14 for t=12 mm. In as can be seen by a lower Mmax for twin flaws
both instances and for all investigated in-plane compared to the value for no flaws. However for
separation the maximum required flaw alignment shorter flaws (D/Lo=22) Mmax for two flaws is
distance H approaches the flaw length Lo but does identical to the Mmax for a no flaw case. In
not exceed this value. conjunction with the above observation of minimal
strain localisation in the ligament in a short flaw
For shorter flaws (D/Lo>10 for t=6 mm and D/Lo>15 geometry, a conclusion can be made that for shorter
for t=12 mm) pipe ovalizes before a notable tip-to-tip flaws the flaw interaction is secondary to pipe
interaction can develop. This behaviour is illustrated ovalization. In such cases pipe will ovalize at a
in Figure 10 on a two flaw configuration, both being moment broadly corresponding to the value derived
at a nearly equal distance H/Lo and with flaw lengths by using equations of C-5321.
such to give D/Lo=9.5 and D/Lo=14. Wall thickness is
6 mm in both configurations. For the D/Lo=9.5

7 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/11/2013 Terms of Use: http://asme.org/terms


CONCLUSIONS [4] Bezensek B., “Flaw alignment criteria based on
Plates and pipes containing twin through-wall flaws limit load solutions for non-aligned through-wall
subject to tension or bending have been examined flaws", Proc. Int. Conf. ASME 2007 Pressure
using a detailed finite element modelling. By Vessels and Piping , San Antonio, TX, USA,
including the Gurson-Tvergaard damage model with July 2007; Paper No. PVP2007-26041
void nucleation option the experimentally observed [5] ABAQUS/Standard v.6.4, Abaqus Inc,
flaw coalescence and force-extension curves were Providence, Rhode Island, USA.
accurately modelled for selected plates in tension.
[6] Gurson AL, “Continuum Theory of Ductile
Thus calibrated model was applied to a range of
Rupture by Void Nucleation and Growth:
plate and pipe configurations, to determine the
Part I—Yield Criteria and Flow Rules for Porous
minimal acceptable separation between the parallel
Ductile Materials,” J. Eng. Materials and
planes containing through-wall flaws at which flaws
Technology, 99, 1977, 2–15
may be conservatively assessed as single (non-
aligned) flaws. It is shown that the alignment [7] Tvergaard V, “Influence of Voids on Shear
distance H is marginally less than the flaw length, L, Band Instabilities under Plane Strain Condition,”
for both, plates in tension and pipes in bending. Int. J. Fracture Mechanics, 17,1981, 389–407
[8] Aravas N, “On the numerical integration of a
For convenience it is suggested to consider flaws class of pressure-dependent plasticity models,
aligned onto the same plane when the separation Int. J. Num. Methods in Eng. , 24 (7), 1987,
between the parallel planes H equals the flaw length 1395-1416
L:
H ≤ L ⇒ flaws aligned
(6)
H > L ⇒ flaws independen t

FUTURE WORK
The future work shall examine alignment of surface
flaws in pipes in bending under plastic collapse
conditions.

ACKNOWLEDGEMENTS
B Bezensek acknowledges access to Abaqus at
University of Glasgow under the academic licence.
Contributions and discussions with Dr. K. Saito of
Hitachi GE Nuclear Energy Ltd and Dr. K. Hasegawa
of Japan Nuclear Energy Safety Organisation (JNES)
are gratefully acknowledged.

REFERENCES
[1] ASME, Boiler and pressure vessel code, Section
XI, Document IWA-3300 and Appendix C,
American Society of Mechanical Engineers,
Philadelphia, USA.
[2] Hasegawa K, Miyazaki K and Saito K,
“Development of Flaw Interaction Rules for
Ductile Behaviour”, Presentation for the ASME
Code Section XI, WG on Pipe Flaw Evaluation,
Orlando, FL, May 2005.
[3] Bezensek B. and McCulloch E., “Interaction of
twin non-aligned through-wall flaws under elastic-
plastic conditions ", Proc. Int. Conf. ASME 2006
Pressure Vessels and Piping , Vancouver, BC,
Canada, July 2006; Paper No. PVP06-ICPVT11-
93373

8 Copyright © 2008 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 10/11/2013 Terms of Use: http://asme.org/terms

You might also like