Вы находитесь на странице: 1из 9

ECA BY FAILURE ASSESSMENT DIAGRAM: CASE STUDIES

W. R. Tyson
CANMET/MTL, Natural Resources Canada
555 Booth Street, Ottawa, ON, Canada
S. Xu
CANMET/MTL, Natural Resources Canada
183 Longwood Road South, Hamilton, ON,
Canada

I. Ward
Shell Global Solutions
A Div. of Shell Chemicals
Americas nc.
3655 36 Street NW, Calgary,
Alberta T2L 1Y8, Canada

D.-M. Duan
TransCanada PipeLines Ltd.
450 - 1st Street S.W., Calgary,
Alberta, Canada

D. HorsIey
BP Upstream Engineering
Centre
240 - 4 Avenue, Calgary,
Alberta T2P 2H8, Canada




ABSTRACT
State-oI-the-art Engineering Critical Assessment (ECA)
employs the methodology oI the Failure Assessment Diagram
(FAD) Ior stress-based design. Current versions oI FADs used
in leading standards are reviewed and applied to girth weld
Ilaws in pipelines. The intent oI the work is to provide
background inIormation as a contribution to a proposed
upgrade oI the Canadian Standards Association standard Z662,
'Oil and gas pipeline systems.

Key Words: Engineering critical assessment, ECA, Iailure
assessment diagram, FAD, deIect assessment, pipeline girth
welds


INTRODUCTION
The traditional approach to dealing with Ilaws in welds
identiIied during construction is to assess them against
perIormance expected oI good workmanship, and to repair them
iI they do not meet this standard. Since the early 1980s it has
been possible to assess the eIIect oI Ilaws on integrity using the
concept oI 'Iitness-Ior-purpose and the methods oI
Engineering Critical Assessment (ECA). Various ECA
methodologies have been developed, and codiIied in a plethora
oI standards in use around the world, such as FITNET |1|. For
pipelines, ECA methods are provided as inIormative (non-
mandatory) annexes to the pipeline standards oI the Canadian
Standards Association (CSA) |2| and the American Petroleum
Institute (API) |3|. The current trend is to provide assessment at
various levels oI accuracy with, oI course, the complexity and
expense increasing with the level oI accuracy, as exempliIied by
the various levels oI analysis oIIered in the British Standards
Institution document BS 7910 |4| on assessment oI Ilaws in
metallic structures. The 'normal assessment route
recommended in BS 7910 is based on the Failure Assessment
Diagram (FAD). This procedure has been adopted recently by
the API.
It is the objective oI this paper to compare the
methodologies Ior assessment oI pipeline girth weld
imperIections in these three standards, principally by applying
them to case studies. On the basis oI this comparison, an
optimized procedure Ior ECA by FAD is proposed and
validated by application to the available database oI pipe and
plate tests.

NOMENCLATURE
a: deIect height in API 1104 Appendix A
c: halI imperIection length
d: imperIection depth
CTOD or o: crack-tip opening displacement
CTOD
e
or o
e
: elastic part oI crack-tip opening displacement
D: pipe outside diameter
K
I
: mode I stress intensity Iactor (abbreviated as K)
K
r
: toughness ratio in FAD Iormat; K
r
K(elastic)/K(material)
Proceedings of the 2012 9th International Pipeline Conference
IPC2012
September 24-28, 2012, Calgary, Alberta, Canada
IPC2012-90140
1
Copyright 2012 by ASME and
Her Majesty the Queen in Right of Canada


L
1
: imperIection length Ior Iracture (2c)
L
2
: imperIection length Ior plastic collapse (2c)
L
r
: stress ratio in FAD; L
r
o
a
/o
c

L
r
cutoII
L
rmax
: cutoII stress ratio in FAD
SMYS: speciIied minimum yield strength
t: pipe wall thickness
Y: dimensionless constant relating stress intensity Iactor to
crack size and loading
o
a
: applied stress
o
I
: Ilow stress
o
reI
: reIerence stress in BS 7910
o
t
, T: ultimate tensile strength oI pipe material
o
y
, YS: yield stress oI pipe material
u: D/t
q: a/t
: 2c/(aD)


DEFECT ASSESSMENT: CURRENT STANDARDS
The ECA approach in the Canadian pipeline standard |2| is
based on the COD design curve, modiIied to take into account
experimental results Irom tests on Iull-scale pipe perIormed at
the Welding Institute oI Canada (WIC). These tests showed that
predictions oI the unmodiIied design curve could be non-
conservative Ior long, shallow deIects and the curves Ior the
'eIIective imperIection size (see below) were adjusted to take
this into account. In addition to 'Iracture, 'plastic collapse is
dealt with as a separate Iailure mechanism requiring comparison
with an estimated plastic collapse stress.
In the API standard |3|, an approach based essentially on
the FAD oI BS 7910 is used. The methodology is presented in
two options, a simpliIied 'Option 1 and a more precise
'Option 2. Option 2 is more accurate and reduces the
conservatism implicit in Option 1, but requires more eIIort in
calculation.
The British standard |4| is the most general oI the three
considered in this paper, as it is applicable to a wide range oI
engineering structures (including, oI course, circumIerential
Ilaws in pipe). This standard has evolved Irom early editions
based on the COD design curve, and now includes three levels
oI analysis Irom a simple screening approach (level 1)
reminiscent oI the COD design curve to a sophisticated level 3.
Level 3 includes a tearing analysis allowing Ior crack growth
and enables determination oI an instability condition as an
alternative to deIining Iracture as attainment oI a 'critical
toughness.
All three standards require calculation oI a stress intensity
Iactor, based on linear elastic Iracture mechanics, and a plastic
collapse stress. The methods oI calculation oI these parameters
are reviewed in the next two sections.

STRESS INTENSITY FACTOR
In general, the stress intensity Iactor can be expressed as
aa Yo K
a I
=

where Y is a dimensionless constant depending on geometry (,
q, and u) and loading (membrane stress and/or bending stress).
In CSA Z662, an 'eIIective imperIection size is
determined Irom tables in the Standard as a Iunction oI
imperIection size and used to calculate K Irom the equations Ior
a centre-cracked inIinite plate containing a crack oI the
eIIective imperIection size. The resulting K value is appropriate
Ior a surIace Ilaw in a plate, with some modiIication Ior long
surIace Ilaws to correspond with experiments perIormed at the
Welding Institute oI Canada (WIC) in the early 1980s.
In API 1104, K is calculated Irom equations Iitted to Iinite
element calculations Ior surIace imperIections in pipes under
large-scale bending and are similar to those reported by Wang et
al. |5| based on work by Chapuliot et al. |6|.
BS 7910 recommends equations Ior surIace Ilaws in plates
based on the Iinite element results oI Raju and Newman |7|.
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14
(mperfection length)/(Circumference)
Y
CSA Z662
AP 1104
BS 7910 (plate)
D/t=90
d/t=0.3
Fig. 1. Dimensionless constant Y Ior circumIerential Ilaws oI
d/t0.3 in pipe oI D/t90.
Figure 1 compares values oI Y Ior circumIerential Ilaws oI
d/t0.3 in pipe oI D/t90 according to the three standards. In
general, the stress intensity Iactors are in good agreement Ior
small values oI Ilaw length. However, Ior long Ilaws the CSA
values are very conservative.

LIMIT LOAD
The 'limit load as used here is the load at which a
structure made oI elastic-perIectly plastic material will deIorm
Ireely. The 'collapse load may be deIined in relation to several
2
Copyright 2012 by ASME and
Her Majesty the Queen in Right of Canada


diIIerent conditions. According to BS 7910 (Annex P.1), 'Three
potential collapse modes can be identiIied: collapse oI the
remaining ligament local collapse; collapse oI the structural
section containing the Ilaw net section collapse; collapse oI
the structure by gross straining gross section collapse.
BS 7910 provides equations Ior o
reI
; Ior calculation oI L
r
in the
FAD the limit load o
reI
is divided by the yield stress. The
'collapse load (used as the cut-oII load in the FAD) is
evaluated as the limit load aIter some work hardening, in which
the yield stress is replaced by the Ilow stress. The 'Ilow stress
is normally deIined as the average oI the yield and ultimate
tensile strength, but sometimes as the yield strength modiIied by
an allowance Ior work hardening.
Miller |8| has given an equation Ior the stress at plastic
collapse that is widely used and underlies the collapse equations
in the CSA and API standards.
In the CSA standard the maximum allowable size oI
imperIection to prevent plastic Iailure Ior circumIerentially
oriented semi-elliptical imperIections is based on equations
proposed by Wang et al. |5|. These are essentially modiIications
to the Miller equation to account Ior the observation oI lower
Iailure loads Ior long Ilaws Irom experiments at WIC. For
determination oI allowable imperIection lengths, the Wang
equations are Iurther modiIied in the CSA standard by inclusion
oI a Iactor oI 2 on imperIection length.
In API 1104 an equation is given Ior the limit load
(identiIied in the standard as the 'plastic collapse stress o
c
)
which is proportional to the yield stress. At the collapse cut-oII
in the FAD, L
r
is equal to o
I
/o
y
, which is equivalent to replacing
the yield stress by the Ilow stress in the equation Ior o
c
. A saIety
Iactor oI 1.5 is applied to the imperIection length at plastic
collapse to obtain an allowable length, i.e. the length at collapse
is divided by 1.5 to obtain allowable length.
In BS 7910, a 'reIerence stress o
reI
is calculated and
divided by the yield strength to obtain L
r
. Equations are given
Ior o
reI
Ior internal circumIerential surIace Ilaws in cylinders
which are based on Kastner`s plastic collapse solution |9|.
Plastic collapse loads are compared in Fig. 2. For purposes
oI comparison, the saIety Iactors on length are not taken into
account in calculating the CSA and API collapse loads. The
CSA and API collapse loads are similar, as would be expected
because they are both based on similar modiIications oI the
Miller solution. The Kastner solution (used in BS 7910)
predicts higher collapse loads, but is more conservative than the
Miller equation.

0.0
0.2
0.4
0.6
0.8
1.0
1.2
0.00 0.05 0.10 0.15 0.20 0.25
(mperfection length)/(pipe circumference)
o
a

/

o
y

Miller
CSA Z662 Annex K
AP 1104 Append. A
BS 7910
Grade 550
D=914.4 mm, t=13.8 mm
d/t=0.3
Fig. 2. Comparison oI plastic collapse loads.

CASE STUDIES
To compare allowable Ilaw sizes according to the three
standards, ten typical cases were studied. The range oI variables
encompassed, chosen to be representative oI current pipeline
construction, is reported in Table 1. Cases include Gr 414 to
Gr 690 (X60 to X100), with pipes oI generally large diameter
635 to 1219 mm (25 to 48), at stress levels oI 0.7 and 0.8
SMYS and CTOD toughnesses oI 0.08 to 0.162 mm.

Table 1. Range oI variables Ior case studies.
Variable Range
D (mm) 635 - 1219.2
t (mm) 10 - 19.1
D/t 61.0 - 88.8
SMYS (MPa) 414 - 690
o
a
/SMYS 0.7 - 0.8
CTOD (mm) 0.08 - 0.162
a/t (d/t) (mm) ~0.002 to 0.50
2c/(aD) ~0.002 to 0.20

Allowable Ilaw sizes were calculated as a Iunction oI
imperIection size (length and depth). Typical results are shown
in Fig. 3. The Iigure shows results Irom the normal assessment
routes oI BS 7910 Level 2 and API 1104 Appendix A Option 2,
but includes also the results oI the more conservative and
simpler analyses Level 1 and Option 1 oI these two standards.
For BS 7910 Level 2, the same equations were used Ior the
stress intensity Iactor as those in API 1104, i.e. the Chapuliot
solutions as Iitted by Wang, because these are considered to be
more appropriate Ior pipe than the plate solutions given in the
BS standard. (In any case, there is little diIIerence between the
plate and pipe solutions as shown in Fig. 1; the Chapuliot
solutions are slightly more conservative Ior long Ilaws). Also, a
3
Copyright 2012 by ASME and
Her Majesty the Queen in Right of Canada


saIety Iactor oI 1.5 on Ilaw length was imposed on the BS 7910
Ilaw lengths so that comparison at a similar level oI
conservatism could be made with the results oI the CSA and
API standards. Note that BS 7910 requires that saIety Iactors be
determined according to the application.
0
100
200
300
400
500
600
700
800
0 1 2 3 4 5 6 7 8
Depth of imperfection (mm)

m
p
e
r
f
e
c
t
i
o
n

l
e
n
g
t
h

(
m
m
)






BS 7910 Level 2A with a safety factor of 1.5
BS 7910 Level 1A
AP Append. A Option 2
AP Append. A Option 1
CSA Annex K Brittle Fracture or max. length
CSA Annex K Plastic Failure or max. length
Case 1A
Pipe: D=914.4 mm, t=13.8 mm, Grade 550
CTOD=0.162 mm
Stress: 0.8 SMYS
max.
depth
CSA max. length
AP max. length
Fig. 3. Allowable Ilaw sizes according to various standards Ior a
typical case (1A).
Allowable imperIection sizes Irom the diIIerent procedures
are generally in the order oI BS 7910 Level 2A ~ API 1104
Option 2 ~ CSA Z662 Annex K. There were cases where CSA
Z662 Annex K was much more conservative than even API
Option 1; an example is shown in Fig. 4.
0
100
200
300
400
500
600
700
800
0 1 2 3 4 5 6
Depth of imperfection (mm)

m
p
e
r
f
e
c
t
i
o
n

l
e
n
g
t
h

(
m
m
)






BS 7910 Level 2A with a safety factor of 1.5
BS 7910 Level 1A
AP Append. A Option 2
AP Append. A Option 1
CSA Annex K Brittle Fracture or max. length
CSA Annex K Plastic Failure or max. length
Case 3
Pipe: D=914.4 mm, t=10.3 mm, Grade 550
CTOD=0.143 mm
Stress: 0.7 SMYS
max.
depth
CSA max. length
AP max. length
Fig. 4. Allowable imperIection sizes Ior Case 3
In comparing BS 7910 Level 2A and API 1104 Option 2
procedures, it seems at Iirst surprising that the Iormer lead to
substantially larger allowable imperIection sizes than those oI
the latter since both standards use the FAD approach with
identical Iailure curves Ior toughness-dependent Iracture.
However, the limit load equations in the two standards are
diIIerent and so the plotting positions along the abscissa (the L
r

axis) diIIer Ior the same stress and Ilaw size. Also, the diIIerent
deIinitions oI depth (BS 7910 contains no allowance Ior
measurement error, while API 1104 does) have the eIIect oI
increasing the apparent conservatism oI the API procedure.

OPTIMIZED FAD
Based on the results oI the case studies reported above, it is
evident that current CSA procedures lead to highly conservative
results compared to state-oI-the-art FAD procedures. Using the
more accurate FAD methodology an increase oI the allowable
imperIection length can be expected over current CSA
procedures, typically by a Iactor oI two or more. The procedure
with the least conservatism is that oI BS 7910 Level 2,
incorporating the K equations oI Chapuliot as parameterized in
API 1104, using the Kastner equation Ior the limit load as
recommended in BS 7910, and incorporating a Iactor oI saIety
oI 1.5 on length. This is the procedure that has been used above
as the 'BS 7910 level 2A with a saIety Iactor oI 1.5 and Ior
brevity will be reIerred to as the 'optimized FAD procedure.
The procedure is 'optimized in the sense that it incorporates
the best Ieatures oI existing procedures, and it is described in
detail in the Annex. The increase in allowable Ilaw length is
signiIicantly more than would be achieved using the FAD
procedures oI API 1104. This seems strange, since both
procedures use an identical FAD. To reiterate the argument in
the preceding section, the major reason Ior the diIIerence is a
diIIerent deIinition oI Ilaw depth; while BS 7910 contains no
allowance Ior measurement error and requires that a
conservative measure be used, API 1104 contains an embedded
Iactor on Ilaw depth. II a measurement error is assumed Ior the
depth variable Ior the BS 7910 curves, translating them to the
leIt in the case study Iigures, the apparent conservatism in the
API approach becomes much smaller. However, the BS 7910
allowable Ilaw length remains larger than the API allowable
length even when the diIIerence in depth deIinition is removed,
because the BS 7910 estimate oI the limit load is considerably
less conservative than that oI API (see Fig. 2). The preIerred
limit load solution is that oI Kastner as recommended in
BS 7910; the more conservative API equations derive Irom a
lower-bound Iit to experimental data that probably includes
some Iracture-dependent results.
It should be pointed out that some Iurther adjustments
recommended in the BS 7910 standard have not been taken into
account. For example, residual stresses have been ignored. This
is Ielt to be justiIied because in pipeline applications the
material CTOD is generally high enough and wall thickness is
small enough that ligament strains oI the order oI the yield
strain or more are achieved at Iracture, hence residual stresses
are relieved by plastic Ilow. Also, stress concentrations such as
those that occur at high-low oIIsets are ignored, and this will be
discussed below.

4
Copyright 2012 by ASME and
Her Majesty the Queen in Right of Canada


COMPARISON WITH EXPERIMENTAL DATABASE
The optimized FAD methodology proposed above is
summarized in step-by-step Iorm in the Annex. However, beIore
promoting a less conservative approach to ECA, it is necessary
to validate the method by comparison with experimental data.
This was done by comparison with databases Ior pipe and plate.
In the validation, input data on size was the measured size,
known to be accurate as the values were directly measured
rather than estimated by non-destructive evaluation (NDE). All
welds were over-matched or nominally even-matched. The
parameter ranges in the data banks are summarized in Table 2.
The database Ior pipe was taken Irom a compilation by
Wang et al. |10|. It includes a total oI 56 tests: 52 by Glover
and co-workers perIormed at WIC to support the original
introduction oI ECA to the CSA standard, and 4 by Erdogan
|11|. The results oI the analysis are shown in Fig. 5. The
analysis is conservative Ior all 56 cases in the database.
Table 2. Parameter ranges in pipe and wide-plate databases
Full-scale tests Wide-plate tests
Grade
414-483
(X60-X70)
448-483
(X65-X70)
YS (MPa) 441-689 470-498
c/a 3.8-48.3 5.1-50.6
a/t 0.06-0.91 0.16-0.50
D (mm) 508-1066.8 1066.8-1219.2
D/t 28.0-90.2 35.0-72.1
CTOD (mm) 0.03-0.55 0.093-0.448

Fig. 5. Validation oI optimized FAD procedure using Iull-scale
pipe database.
To assess the degree oI conservatism using the optimized
FAD procedure and to compare it with the degree oI
conservatism using the existing CSA procedure, a 'Iailure
reserve, deIined here as the observed Iailure stress compared
with the calculated Iailure stress, was deIined. This was
calculated Ior the 33 points in the pipe database within CSA
size limits
1
; the results are shown in Fig. 6. The average Iailure
reserve was 1.530.26 (min. 1.06) Ior the optimized procedure
and 2.120.73 (min. 1.21) Ior the CSA procedure. The latter is
more conservative and has larger standard deviation than the
Iormer. This is as expected; the more soundly-based FAD
procedure should give more accurate predictions oI Iailure
stress, with less scatter.
0
1
2
3
4
5
6
0 1 2 3 4 5 6
Failure reserve factor in stress
(BS 7910 Level 2A with a safety factor of 1.5)
F
a
i
l
u
r
e

r
e
s
e
r
v
e

f
a
c
t
o
r

i
n

s
t
r
e
s
s

(
C
S
A

Z
6
6
2
)
Weldments are even-matched
SMYS used as input

Fig. 6. Comparison oI Iailure reserve Iactors: optimized
procedure (BS 7910 Level 2A with a saIety Iactor oI 1.5 on
length) . CSA Z662 Annex K.
For plate, the database was taken Irom Smith and Pisarski
|12|. Results are shown in Fig. 7. Again, the analysis is
conservative Ior all 15 cases.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
L
r
K
r
Weldments are even-matched
Measured yield strength used as input

Fig. 7. Validation oI optimized FAD procedure using wide-plate
database.


1
CSA Annex K automatically rejects imperIections oI length larger than
0.1 times the pipe circumIerence
0.0
0.5
1.0
1.5
2.0
2.5
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
L
r
K
r
Filled data are within CSA Z662 size limits
Weldments are even-matched
Measured yield strength used as input
5
Copyright 2012 by ASME and
Her Majesty the Queen in Right of Canada

DISCUSSION
The optimized FAD procedure does not take explicit
account oI a number oI variables that are relevant in aIIecting
the Iailure load. These include mismatch, high/low, constraint
eIIects, variations in strength and toughness, residual stress,
location oI the Ilaw with respect to the location oI maximum
stress, and others. Some oI these Iactors are deleterious (e.g.
residual stresses) and some are beneIicial (e.g. weld metal
overmatch). In general, in the current state oI the art the most
cautious procedures use conservative estimates oI all the input
variables in making the assessment. It is unlikely that the
'worst-case scenario would be realized Ior all oI these
variables in any particular case. The proper way to deal with
actual circumstances is to use probabilistic methods (e.g.
'Monte Carlo simulations) in which the relevant variables are
distributed according to the corresponding probability
distribution and the resultant probability oI Iailure is evaluated.
In this way, the distribution oI Iailure probabilities can be
compared with a target reliability. However, this approach has
not yet been developed to a point where it can be considered as
a practical engineering tool. In the meantime, it is a matter oI
engineering judgment to select a reasonably straightIorward yet
acceptably saIe methodology. A temporizing approach is
mentioned below.
The current methodology oI CSA Z662 Annex K has
served the industry well, but is not particularly simple to use
and is based on an approach (the 'COD design curve) that has
been superseded by more accurate methods oI assessment,
notably the Failure Assessment Diagram. The current CSA
procedure (i.e. based on the COD design curve) generates saIe
acceptance limits, but with a rather large degree oI
conservatism. The accuracy oI the assessment can be improved
by using state-oI-the-art procedures such as the one presented
here.
By comparison with pipe and wide-plate databases, the
optimized procedure has been shown to give conservative
assessments. It can reasonably be presumed that these databases
contain results oI tests that were perIormed on materials with
the normal range oI relevant variables (residual stress, high/low,
etc.). This variability is most likely the source oI scatter in the
assessment points on the FAD. UnIortunately, most oI the
results in the database correspond to collapse rather than
Iracture, so the validation with respect to the latter mode is
limited. More data in the 'Iracture range would be desirable,
but none is available and its acquisition is very expensive.
The current CSA procedure has proven reliable,
demonstrating its conservatism and inherent ability to account
Ior the various uncertainties. However, without a more
extensive database oI Iailure stresses in situations covering the
Iull range oI allowable variables such as high/low, sizing error,
control oI lowering stresses, CTOD values at low temperatures
etc. we cannot comment on how close we are to a Iailure
condition and how large is the saIety allowance. This is true Ior
the optimized FAD as well as the current CSA procedure. A
possible approach to allowing Ior this uncertainty is to impose
Iurther saIety Iactors on Ilaw size while perIorming ECA. Data
is currently being reviewed on the inIluence oI allowable
high/low on Iailure stress to assess this option.

CONCLUSIONS
1. The methodologies in three standards (CSA Z662,
BS 7910, API 1104) normally used Ior assessment oI girth
weld imperIections in pipelines have been reviewed and
compared, principally by the use oI case studies.
2. Allowable imperIection sizes using the diIIerent
procedures are generally, in the order oI decreasing size
(increasing conservatism), BS 7910, API 1104, and CSA
Z662.
3. An optimized procedure Ior assessment by Failure
Assessment Diagram (FAD) is proposed in this paper and
validated by comparison with databases Ior pipe and plate.

ACKNOWLEDGMENTS
The authors are grateIul to Marc Spencer and Rory
Belanger Ior helpIul discussions.

REFERENCES
|1| FITNET, 'FITNET Fitness-Ior-Service (FFS) Procedure,
Revision MK8, 2008, and 'IIW FFS Recommendations Ior
Fracture Assessment oI Weld Flaws, under development, 2009,
M. Kocak, GKSS Research Centre, Geestacht, Germany.
|2| CSA Standard Z662-11, 'Oil and Gas Pipeline Systems, Annex
K: Standards oI acceptability Ior circumIerential pipe butt
welds based upon Iracture mechanics principles, June 2011,
Canadian Standards Association, Mississauga, Canada.
|3| API Standard 1104, 'Welding oI Pipelines and Related
Facilities, Twentieth Edition, October 2005 (Errata July 2007,
December 2008), API Publishing Services, Washington, D.C.
|4| BS 7910:2005, 'Guide to methods Ior assessing the acceptability
oI Ilaws in metallic structures, July 2005, Amended September
2007, British Standards Institution, London, UK.
|5| Wang, Y.-Y., D. Rudland and D. Horsley, 'Development
oI a FAD-Based Girth Weld ECA Procedure. Part I,
Theoretical Framework, Proc. 4
th
Intl. Pipeline ConI.
(IPC2002), Sept. 29-Oct. 3, 2002, Calgary, Alberta, Canada,
ASME, IPC2002-27171002.
|6| Chapuliot, S., Lacire, M. H., and Le Delliou, P., 'Stress Intensity
Factors Ior Internal CircumIerential Cracks in Tubes over a
Wide Range oI Radius over Thickness Ratios, PVP Vol. 365,
ASME 1998, pp. 95-106.
6
Copyright 2012 by ASME and
Her Majesty the Queen in Right of Canada


|7| I.S. Raju and J.C. Newman, 'Stress Intensity Factors Ior a Wide
Range oI Semi-elliptical SurIace Cracks in Finite Thickness
Plates, Eng. Fract. Mech., Vol. 11, 1979, pp. 817-829; I.C.
Newman, Jr. and I.S. Raju, 'Stress-Intensity Iactor Equations
Ior Cracks in Three-Dimensional Finite Bodies, STP 791,
ASTM, 1983, pp. I-238 to I-265.
|8| A.G. Miller, 'Review oI Limit Loads oI Structures Containing
DeIects, Int. J. Pres. Ves. & Piping, Vol. 32, 1988, pp. 197-
327.
|9| W. Kastner, E. Rohrich, W. Schmitt and R. Steinbuch, 'Critical
Crack Sizes in Ductile Piping, International Journal oI
Pressure Vessels and Piping, Vol. 9, No. 3, 1981, pp. 197-219.
|10| Wang, Y.-Y., D. Rudland and D. Horsley, 'Development oI a
FAD-Based Girth Weld ECA Procedure. Part II, Experimental
VeriIication, Proc. 4th Intl. Pipeline ConI. (IPC2002), Sept.
29-Oct. 3, 2002, Calgary, Alberta, Canada, ASME, IPC2002-
27173.
|11| Erdogan, F., 'Theoretical and Experimental Study oI Fracture in
Pipelines Containing CircumIerential Flaws, DOT-RSPA-
DMA-50/83/3, Contract DOT-RC-82007 Final Report to
USDOT, September 1982.
|12| Smith, S.E. and H.G. Pisarski, 'Comparison oI API 1104
Appendix A and BS 7910 Procedures Ior the Assessment oI
Girth Weld Flaws, Proc. Intl. ConI. Pipeline Technology 2009,
R. Denys, ed., Ostend, Belgium, October 12-14, 2009, Paper
No: Ostend2009-027.
7
Copyright 2012 by ASME and
Her Majesty the Queen in Right of Canada


ANNEX

PROPOSED OPTIMIZED PROCEDURE FOR ECA BY FAD

Maximum size of imperfection to prevent elastic/plastic
fracture
This procedure describes the use oI the FAD to assess
surIace Ilaws in pipe welds.
The maximum size oI imperIection to prevent
elastic/plastic Iracture shall be determined Irom the coincidence
oI the 'assessment point (K
r
, L
r
) with the Failure Assessment
Curve (FAC). The equation Ior the FAC (illustrated in Fig. A1)
is:
K
r
(1-0.14L
r
2
)0.30.7exp(-0.65L
r
6
)}
where K
r
and L
r
are deIined as:
K
r
(o
e
/o)
1/2

and L
r
o
reI
/o
y

where
o
e
elastic component oI the applied CTOD
o CTOD Iracture toughness
o
reI
reIerence stress
o
y
yield stress
Relevant geometrical parameters are:
d imperIection size (height)
L imperIection length
t pipe wall thickness
D pipe outside diameter
u D/t
L/(aD)
q d/t
The applied CTOD o
e
is obtained Irom the elastic stress
intensity Iactor K, using the Iollowing relation between o and K:
( )
I
2 2
e
o m
1
E
v 1 K
o

=

where
u Poisson`s ratio 0.3 Ior steels
E Young`s modulus 207 GPa Ior steels
o
I
Ilow stress (o
y


o
UTS
)/2


o
UTS
ultimate tensile stress
and
m A
0
-A
1
*(o
y
/o
UTS
)A
2
*(o
y
/o
UTS
)
2
-A
3
*(o
y
/o
UTS
)
3

with:
A
0
3.18-0.22*(q)
A
1
4.32-2.23*(q)
A
2
4.44-2.29*(q)
A
3
2.05-1.06*(q)
The stress intensity Iactor K (conventionally written as
Yo\(aa) where a is crack size) is given by
d F K
a b
t o =

with
o
a
applied tensile stress
and
( )
( )

< |
.
|

\
|
=
> > |
.
|

\
|
=
s >
= =
0.1 q ,0.1
u
0.1
a
80
u, F
u
q
a
80
and 0.1 q q ,
u
q
a
80
u, F
u
q
a
80
and 0.1 q q , u, F
q , u, F Y
bo
b0
bo
b
where
( )
|
|
.
|

\
|
+ + + =
2
0.806 1 0.983 0.906 0.791
bo
m u
u
m
q 2.31u 1.09 q , u, F

2
1
0.345q 0.163q 0.00985 m =

2
2
0.155q 2.18q 0.00416 m + =

The reIerence stress o
reI
is given by
( ) ( ) | |
( ) ( ) | | q D / L a q 1
D / L sin q 2 q 1 a o
o
a
reI

+
=

The imperIection height d shall be conservatively
determined, and shall include an allowance Ior measurement
error.
Values oI K
r
and L
r
may be Iound Irom the equations above
and plotted as an assessment point on the FAD as shown in
Fig. A1. II the assessment point lies between the FAC and the
origin, there is no risk oI Iracture and the conditions are
'acceptable (although subject to imposition oI a saIety Iactor
as described below). Outside oI the FAC, conditions are
'unacceptable (Fig. A1). Prevention oI plastic Iailure is
discussed below.
To estimate the imperIection length Ior Iracture Irom a
given Ilaw, the variables are adjusted iteratively until the point
(K
r
, L
r
) Ialls on the assessment curve oI the FAD. At this point,
8
Copyright 2012 by ASME and
Her Majesty the Queen in Right of Canada


the measured Iracture toughness o equals the plasticity-adjusted
crack driving Iorce and the variables are 'critical values.
The allowable imperIection length L
1max
is then calculated
by dividing the critical length by 1.5.
This iterative procedure may be used to generate a curve oI
allowable imperIection length as a Iunction oI Ilaw height, or to
determine acceptability oI a particular imperIection.

Maximum size of imperfection to prevent plastic failure
To prevent excessive plastic deIormation in the vicinity oI
the imperIection, the value oI o
reI
should not exceed the Ilow
stress o
I
. Then at the cut-oII condition to prevent plastic Iailure
Lr o
I
/o
y
(see Fig. A1) and the values oI the relevant variables
(o
a
, d, L) at plastic Iracture Iollow Irom the equation Ior o
reI

given above. The allowable imperIection length L
2max
is then
calculated by dividing the critical length by 1.5.

Maximum length of imperfection
The maximum acceptable length Ior each type and depth oI
imperIection shall be the least oI L
1max,
, L
2max
, and 0.1 times the
nominal pipe circumIerence.

0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4
Lr
K
r
Kr = (1-0.14L
r
2
){0.3+0.7exp(-0.65L
r
6
)}
Cut-off at L
r
= o
f
/o
y
FAC
assessment point
x
acceptable unacceptable
Fig. A1. Failure Assessment Diagram (FAD).

9
Copyright 2012 by ASME and
Her Majesty the Queen in Right of Canada

Вам также может понравиться