You are on page 1of 53

.

CSM Report n. 10423R


Final Report

JOINT INDUSTRY PROJECT


DEFECTS ASSESSMENT MANUAL (VERSION TWO)

Contract N. 10408429

Methods governing and preventing long running


ductile and brittle propagating fast fracture
in gas transmission pipelines

G. Demofonti
G. Mannucci

Final Report

Roma, September 2001


CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Methods governing and preventing long running ductile and brittle propagating
fast fracture in gas transmission pipelines.
G. Demofonti, G. Mannucci

Introduction.

The design of gas transportation lines needs an optimal compromise between the choice of the
operating pressure, linepipe geometry (diameter and thickness) and steel mechanical properties
(tensile and toughness characteristics). The wide variety of possible solutions requires the
development of specific tools, able to estimate the safety margin on events responsible for structural
failure.

The initiation and propagation of ductile or brittle longitudinal cracks are most serious events. The
crack can affect a long part of the line, thus causing a long and costly gas delivery service
breakdown. Fracture propagation control is therefore an extremely important aspect of gas
transmission design.

The philosophy followed by the gas companies to avoid brittle fracture initiation is to ensure that
the operating pipeline temperature is higher than the ductile-brittle transition temperature of the
pipeline steel, and then control the ductile fracture propagation by ensuring that the linepipe steel
has the capacity to absorb sufficient energy to arrest a fast propagating ductile crack.

This document summarises the literature relating to ductile and brittle fracture propagation. The
literature review and comparison with the full scale test data describe in this document, form the
basis of recommendations and guidance for the control of ductile and brittle propagating fast
fracture in the pipeline defect assessment manual.

This review has been conducted by CSM for Andrew Palmer and Associates as part of the Pipeline
Defect Assessment Manual Joint Industry Project.

2
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

CONTENTS

Introduction. 2

Part I: Methods governing and preventing long running ductile propagating fast fracture in
transmission pipelines. 4
1. Background. .....................................................................................................................................4
2. Methods to prevent ductile fracture propagation in a gas pipeline. .................................................5
2.1 The Full Scale Burst Test. 5
2.2 Available Predictive methods. 7
2.2.1 Predictive equations of fracture propagation event. 7
2.2.1 Battelle Two Curve Method. 11
3. Full Scale Ductile Fracture Propagation Burst Tests Database......................................................15
3.1 Comparison of test results with predictions. 22
4. EPRG Recommendations ...............................................................................................................27
5. The effect of deep water marine external constraint on ductile fracture propagation in gas
pipelines..............................................................................................................................................29
5.1 Background 29
5.2 Predictive Method for DFPC for offshore gas pipeline. 30
5.3 Recommendation for DFPC for offshore gas pipeline. 31
6. Effect of decompression behaviour of transported gas on minimum values of Charpy V shelf
energy requirement.............................................................................................................................31
6.1 Background. 31
6.2 Available gas decompression models for rich gas mixtures. 34
6.3 Predictive Method for DFPC for rich gas mixtures 34
7. Alternatives to Charpy V for DFPC ...............................................................................................35
7.1 Measurement of total absorbed energy in a DWTT test. 35
7.2 Measurement of propagation absorbed energy in a DWTT test. 37
7.3 39
Measurement of propagation absorbed energy in a Charpy V test. Leis’s correction factor. 39
7.4 Measurement of specific total absorbed energy by Two Parameters Approach 40
7.5 Measurement of the Crack Tip Opening Angle, CTOA, by Two Specimen Method. 41
7.6 Considerations about the alternatives to Charpy V for DFPC 43

Part II: Methods preventing the brittle fracture propagation in gas pipelines. 45
1. Background. ...................................................................................................................................45
2. On the influence of high wall thickness. ........................................................................................48

References 49

3
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Part I: Methods governing and preventing long running ductile propagating fast
fracture in transmission pipelines.

1. Background.

A crack or defect with critical dimensions, can cause a fracture to propagate along a pipeline, in a
ductile mode if the temperature is above the transition temperature of the linepipe steel, also named
Fracture Propagation Transition Temperature, FPTT /1/. In practice FPTT is the temperature at
which the fracture propagation mode changes from brittle to ductile. The FPTT transition is a
property of a steel and strongly varies with the manufacturing processing applied to the steel;
typical values for line pipe steels lies in a wide range from approximately –130 °C to 90 °C /1/.
After the fracture initiation, decompression waves at various pressure levels propagate in the gas
going back up the line.
If the fracture is ductile (i.e. the pipeline operating temperature is higher than FPTT) and in steady
state propagation conditions, the crack runs at constant speed corresponding to a balance between
the local pressure level and the steel toughness; if the steel toughness is higher, the fracture speed
will decrease to a lower pressure level, and new steady state propagation conditions will be reached.
If the steel toughness is high enough, no steady state propagation conditions can be reached and the
fracture will continuously slow, turn in a spiral direction and arrest.
When gases with heavy components (“rich gases”) are used, the Ductile Fracture Propagation
Control (DFPC) requires significantly higher levels of fracture propagation resistance than for
pipeline carrying pure methane. This is because rich gas decompresses more slowly and so the
driving force acting at the crack tip remains high. To limit the length of ductile fracture propagation,
gas pipeline steels having an appropriate toughness are used.
Summarising, ductile fracture propagation in gas pipelines is characterised by the following aspects:
• main factors are the line pipe geometry (diameter and wall thickness) and the initial line
pressure (hoop stress);
• the driving force comes from the energy release from gas expansion during escape acting at
the crack tip;
• generally, the crack path follows the top pipe generatrix;
• there is a high plastic deformation field associated with the near-the-crack-tip zone;
• the ductile crack velocity is usually up to 300 m/s (in practice it is limited by the acoustic
velocity of the gas in the pipeline at the initial temperature and pressure)1;
• the mechanical properties of the pipeline steel and the backfill type (soil or water) have a
prominent effect on the fracture propagation process;
• the gas nature and composition have a strong effect on the fracture propagation process;
• propagation control or arrest is essentially achieved by choosing an adequate level of pipe
toughness; as a second line of control, suitable mechanical devices (crack arrestors) can be
used2.

1
The acoustic velocity of the gas in the pipeline represents in practice the upper bound level of the crack velocity.
2
It is not in the scope of the present work to present the mechanical solutions usually adopted for arresting ductile or
brittle fracture propagation in a gas steel linepipe. A good literature review can be found in Ref. /2, 3/.

4
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

2. Methods to prevent ductile fracture propagation in a gas pipeline.

2.1 The Full Scale Burst Test.

The toughness required to control ductile fracture propagation is generally measured in terms of
fracture energy by the Charpy-V impact test on transverse specimens, and the required value has
been historically based on the use of empirical data from ductile fracture propagation full scale
burst tests.

The experimental set up of a full scale burst test is typically as follows. The pipeline under
investigation (typically 70 m long and consisting of several pipes welded together having an
increasing toughness from the centre) is located in the middle of an existing permanent test line
between two reservoirs (Figure 1). The whole line (test section plus two reservoirs) must be long
enough to guarantee, during the whole crack propagation in the central test pipes, the same driving
pressure, which would act on a pipeline of indefinite length.
If the line to be tested is an on-shore line, the whole test section is fully covered by soil (1 m depth
or more) that is lightly compacted to reproduce the actual operating conditions. In the case of an
off-shore line, all the test section has to be immersed in several meters of water3.
Under controlled temperature conditions (to assure fully ductile fracture propagation) the test
pipeline (fully instrumented) is pressurised at the desired hoop stress, by means of air or natural gas
according to the requirements. The crack initiation is usually caused either by using an explosive
charge, able to create a through thickness cut long enough to guarantee the break conditions in the
pipe, or a machined part wall slot in the pipe wall thickness connected to mechanical devices that
create a through thickness critical defect4.

Conventional lay-out for a full-scale


ductile fracture propagation test
Anchor Anchor
Initiation block
block
pipe
Reservoir H I I L I I H Reservoir

Test Line

L= Low level of touhgness, to assure an high speed crack initiation


I = Intermediate level of toughness, to have a steady fracture propagation
H = High level of toughness to arrest the fast propagating ductile fracture
Figure 1 - Conventional layout of full scale burst test.

By the 70’s, when ductile fracture propagation became a key issue in gas pipeline design in terms of
both safety and economic loss, full scale burst testing was recognised as the only reliable test to

3
Experimental full scale tests carried out by Battelle and CSM in underwater conditions /10, 15/ demonstrated that
shallow water (12-30m in depth) is sufficient to reproduce the fluid-dynamic constraint effect (overpressure wave
produced in the water surrounding the pipe) acting at the crack tip due to the marine backfill.
4
The length of the through thickness critical defect for the linepipe under investigation can be calculated using for
example the historical Maxey’s formula /4/ and then adopting a strong safety factor (>1.5)

5
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

assess the fracture behaviour for gas pipelines. The full scale burst test can be considered as an
indispensable tool for:
• assessing the minimum toughness level to arrest the ductile fracture propagation in a defined
short length for a specific pipeline in a specific project;
• providing the experimental data concerning the fracture behaviour, in relation to the design
parameters (line pipe diameter and thickness, steel grade, hoop stress level, gas nature, etc.),
for the predictive models development and validation.

In fact a full scale test provides a direct indication of the level of toughness necessary for crack
arrest and, when suitably instrumented, provides also a large amount of extremely useful data on
crack propagation speed, pressure decay, elasto-plastic strain field associated with the crack tip, etc.
These data may be compared with those predicted by the various models, providing both
confirmation of reliability and the basis for further refinements.

Despite the high cost necessary to perform a full scale burst test, the technical and economical
importance of the results coming from these tests, meant a large number of full scale tests have been
conducted in the last 25 years (Figure 2). These tests were carried out by several institutes, as for
example Battelle Memorial Institute (BMI), American Iron and Steel Institute (AISI), Centro
Sviluppo Materiali (CSM), High Strength Line Pipe Research Committee (HLP) of the Japanese
iron and Steel Institute (JISI) and British Gas (BG), on behalf of several pipe makers, gas
companies and international organisations interested in funding the increase of knowledge in this
field; see for example the activities sponsored by the Pipeline Research Committee [formerly
associated with American Gas Association (AGA)] in the USA, the European Pipeline Research
Group (EPRG) in Europe and the Japanese Iron and Steel Institute (JISI) in Japan.
As a consequence of these testing activities, a full scale burst test database has been developed
(PRCI, EPRG, JISI, BGC, CSM, etc.) where the main test parameters (pipeline geometry, test
temperature and pressure, pipes toughness, etc.) and results (propagation and arrest pipes, crack
speed, etc.) are collected. The cost of this huge amount of testing is more than 60 million US
dollars.

Full scale burst tests carried out in the last 25 years

60
Others
JISI
50 Mann
Foothills
EPRG
40 CSM
Test frequency

BMI-Athens
Batt
30 BG
AISI

20

10

0
1970 1975 1975 1980 1980 1985 1985 1990 1990 1995 1995 >1995
Year

Figure 2 - Number of full scale burst tests performed in the last 25 years5 (CSM database)

5
With respect to the data depicted in the graph four more tests were performed and their results were recently
published: two tests performed by British Gas Technology on 36”, X70 pipes with rich gas (Alliance tests no. 1 and 2
/5/) and two tests performed by CSM on 56” and 36” X100 pipes /6/.

6
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

As it is evident in Figure 2, the main effort in terms of experimental activities was done in the 70’s
and in the first half of the 80’s. At the end of the 80’s, the industry’s understanding and the relevant
predictive models were good enough to justify a decrease in the full scale burst testing. The full
scale burst testing revival in the last years is strictly related with the development of new high grade
pipeline steels (≥ API X70) for new applications (higher pressure, richer gas). Under these
conditions the straight extrapolation of the existing predictive models is not adviseable, and
therefore it is necessary to investigate the full scale behaviour of these new pipeline steels in terms
of ductile fracture propagation.

Required Charpy-V impact energy predicted by


several empirical models
(Calculated for a 56'' diameter and 25mm thickness pipeline)
500

Battelle

AISI New
Charpy-V impact energy [J]

400
applications
CSM
300
Mannesmann

Validation
200
data

100

0
200 300 400 500

Hoop stress [MPa]

Figure 3 -Limit of reliability of existing empirical models

2.2 Available Predictive methods.

2.2.1 Predictive equations of fracture propagation event.

Several models in the form of predictive equations, which state the minimum required value of the
Charpy energy as a function of both pipe geometry and applied hoop stress, have been developed by
various institutes using data from more than one hundred full scale burst tests /1, 4, 7, 8, 9/.

Note that all the equations were developed using the upper shelf energy obtained by breaking the
2/3 Charpy V specimen (extracted in transverse direction with respect to the pipe axis); the
equations have converted 2/3 thickness Charpy V energies to standard 10mmx10mm Charpy V
energies historically by assuming constancy of the energy value per unit fracture-surface.

1) BATTELLE (also known as PRCI-BMI Simplified Equation or Short Formula6)


CV10 x10 = 3.5766 ⋅ 10 −5 ⋅ σ 2H ⋅ ( R ⋅ t ) 3 [Joule]
1
1a
CV2 / 3 = 2.382 ⋅ 10 −5 ⋅ σ 2H ⋅ ( R ⋅ t )
1
3
[Joule] 1b
CV10 x10 Spec. = CV2 / 3 Spec. = 4.4707 ⋅ 10 −5 ⋅ σ 2H ⋅ ( R ⋅ t )
1
3
[Joule / cm 2 ] 1c

6
The PRCI-BMI equation was developed by curve fitting predicted Charpy V energy values using the Battelle Two
Curve Model presented hereinafter.

7
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

2) AISI
CV10 x10 = 3.569 ⋅ 10 −4 ⋅ σ 1H.5 ⋅ (2 R ) 2 [Joule]
1
2a
CV2 / 3 = 2.377 ⋅ 10 −4 ⋅ σ 1H.5 ⋅ (2 R )
1
2
[Joule] 2b
⋅ (2 R)
1
−4 1.5 2
CV10 x10 Spec. = CV2 / 3 Spec. = 4.4614 ⋅ 10 ⋅ σ 2
[Joule / cm ]
H
2c

3) BGC50
 R
−3 R 1.25 
CV10 x10 = 15015
. ⋅ 10 ⋅ σ H  185
. − ν 0.75  [Joule] 3a
 t 
1
t2
 R R1.25 
CV2 / 3 = σ H ⋅ 10−3 ⋅  185
. 1 − ν  [Joule] 3b
 t2 t 0.75 
 R R 1.25 
CV10 x10 Spec. = CV2 / 3 Spec. = 1.877 ⋅ 10 −3 ⋅ σ H ⋅ 1.85 1 − ν 0.75  [Joule/cm 2 ] 3c
 t 2 t 
ν represents the sound speed (0.396 for natural gas; 0.36 for air).

4) BGC95 (also known as BGC)


 R R 1.25 
CV10 x10 = 15015
. ⋅ 10 −3 ⋅ σ H  2.08 1 − ν 0.75  [Joule] 4a
 t2 t 

−3 R R1.25 
CV2 / 3 = σ H ⋅ 10 ⋅  2.08 12 − ν 0.75  [Joule] 4b
 t t 
 R R1.25 
CV10 x10 Spec. = CV2 / 3 = 1877
. ⋅ 10−3 ⋅ σ H ⋅  2.08 12 − ν 0.75  [Joule / cm2 ] 4c
 t t 
ν represents the sound speed (0.396 for natural gas; 0.36 for air).

5) ITALSIDER/CSM

 R2 R ⋅ σ 2H ⋅ t 
CV10 x10 = 7.848 ⋅  −011996
. . ⋅ 10 −8
⋅ t − 1301 H + 4.8214 ⋅ 10 −5 ⋅ R ⋅ σ H + 2.382 ⋅ 10 −6  [Joule] 5a
 t H 
2
−8 R2 −4 −5 R ⋅ σ H ⋅ t
CV2 / 3 = −0.627 ⋅ t − 6.8 ⋅ 10 H + 2.52 ⋅ 10 ⋅ R ⋅ σ H + 1245
. ⋅ 10 [Joule] 5b
t H
 R2 R ⋅ σ H2 ⋅ t  5c
CV10 x10 Spec. = 9.81 ⋅  − 0.11996 ⋅ t − 1.301 ⋅ 10 −8 H + 4.8214 ⋅ 10 −5 ⋅ R ⋅ σ H + 2.382 ⋅ 10 −6  [Joule/cm2 ]
 t H 

6) MANNESMANN

CV10 x10 = 30.0155 ⋅ e


( 0.287⋅10 −8
H (2 R)
⋅σ1.76
1.09
⋅ t 0 .585 ) [Joule] 6a
CV2 / 3 = 19.99 ⋅ e
( H (2 R )
0.287⋅10 −8 ⋅σ 1.76
1.09
⋅ t 0 .585 ) [Joule] 6b
CV10 x10 Spec. = CV2 / 3 Spec. = 37.52 ⋅ e
( H (2R)
0.287⋅10 −8 ⋅σ1.76
1.09
⋅ t 0 .585 ) [Joule / cm 2 ]
6c

8
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

7) JAPAN 82 (also known as JAPANESE)

CV10 x10 = 3.7497 ⋅ 10 −6 ⋅ σ 2H.33 ⋅ (2 R )


0 .3
⋅ t 0.47 [Joule] 7a
CV2 / 3 = 2.498 ⋅ 10 −6 ⋅ σ 2H.33 ⋅ ( 2 R )
0.3
⋅ t 0.47 [Joule] 7b
CV10 x10 Spec. = CV2 / 3 Spec. = 4.688 ⋅ 10 −6 ⋅ σ 2H.33 ⋅ (2 R )
0 .3
⋅ t 0.47 [Joule / cm 2 ]
7c

8) KAWASAKI STEEL

(
CV10 x10 = 120 ⋅ 2.62 ⋅ 10 −5 ⋅ σ H ⋅ R + 2.86 ⋅ 10 −6 ⋅ R 2 − 9.9 ⋅ 10 −3 ⋅ R − 4.42 ⋅ 10 −3 ⋅ σ H + 2.29 ) [Joule] 8a

(
CV2 / 3 = 80 ⋅ 2.62 ⋅ 10 −5 ⋅ σ H ⋅ R + 2.86 ⋅ 10 −6 ⋅ R 2 − 9.9 ⋅ 10 −3 ⋅ R − 4.42 ⋅ 10 −3 ⋅ σ H + 2.29 ) [Joule] 8b

( )
CV10 x10 Spec. = 150 ⋅ 2.62 ⋅ 10 −5 ⋅ σ H ⋅ R + 2.86 ⋅ 10 −6 ⋅ R 2 − 9.9 ⋅ 10 −3 ⋅ R − 4.42 ⋅ 10 −3 ⋅ σ H + 2.29 [J/cm 2 ]
8c

Notation:
CV 2/3 : 2/3 thickness Charpy energy, transverse specimen (J)
CV 10X10: Charpy energy, transverse specimen (J)
CV 10X10spec.=CV 2/3spec.: Specific Charpy energy, transverse specimen (J/cm2)
t: Wall thickness (mm)
D: Diameter (mm)
R: Pipe radius (mm)
d: Depth of backfill (mm)
σH: Hoop stress (MPa)
P: Line pressure (MPa)

All the equations predict that if the hoop stress applied is increased, the Charpy V energy level to
arrest a fracture needs to increase. The same is trend applies to the effect of the diameter; larger
diameters, require higher toughnesses for arresting the fracture. On the contrary the effect of the
wall thickness is not uniform in the equations: the AISI equation has no thickness effect and the
CSM one presents a very limited influence of the thickness while for the others the general trend is
if the thickness increases, the toughness needs to increase (with the only exception being the BGC
equations that reduce the energy required as the thickness increases).

The PRCI-BMI equation was developed for hoop stresses between 60 and 80 % of the SMYS of the
steel, lean gas and backfilled conditions; at lower stresses the equation is claimed to provide a
conservative estimate of the arrest toughness /1/.
All the other equations have the same limitations; this occurs because most of the test result data
used for their development were obtained for about 72% of SMYS, lean gas and backfilled
conditions. For the same reason these equations are claimed to be applicable for pipes of
conventional steels (up to grade API X80).
It can be noted that generally the Battelle equation results are the most widely quoted. The range of
validity of the equation, historically limited by the original test data /4/ at about 70 Joule (full size
specimen) was recently upwards to 90-100 Joule /48/.

In Figure 4 a comparison of the proposed formulae is shown for two selected pipeline
configurations. The first chart shows Hoop stress versus Predicted Charpy V energy for a 36”
diameter, 16 mm wall thickness, API X70 steel grade pipeline.

9
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Predicted CharpyV energy to arrest ductile fracture propagation

350
Battelle CV10x10 (J)
AISI CV10x10 (J)
300
British Gas BGC50 CV10x10 (J)
British Gas BGC95 CV10x10 (J)

72% SMYS

80% SMYS
250 Italsider/CSM CV10x10 (J)
Mannesmann CV10x10 (J)
CV10x10 Energy (J)

Japan 82 CV10x10 (J)


200 Kawasaki steel CV10x10 (J)
60% SMYS

150

100

50
Grade: X70 Diameter: 36.0 ''
Thickness: 16.0 mm
0
200 300 400 500 600
Hoop stress (Mpa)

Figure 4 – Predicted Charpy V energy versus Hoop Stress for a X70, 36”x16mm steel pipeline

The second chart shows the same for a 56”diameter, 16mm wall thickness, API X70 steel grade
pipeline .

Predicted CharpyV energy to arrest ductile fracture propagation

350
72% SMYS

80% SMYS

Battelle CV10x10 (J)


AISI CV10x10 (J)
300
British Gas BGC50 CV10x10 (J)
British Gas BGC95 CV10x10 (J)

250 Italsider/CSM CV10x10 (J)


Mannesmann CV10x10 (J)
CV10x10 Energy (J)

Japan 82 CV10x10 (J)


200 Kawasaki steel CV10x10 (J)
60% SMYS

150

100

50
Grade: X70 Diameter: 56.0 ''
Thickness: 16.0 mm
0
200 300 400 500 600
Hoop stress (Mpa)

Figure 5 – Predicted Charpy V energy versus Hoop Stress for a X70, 56”x16mm steel pipeline

10
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Although the range of hoop stress values is plotted up to 600 MPa, it needs to be underlined that all
these prediction formulae were validated using results from burst tests carried out with hoop stress
values less than 400 MPa (API X80, D.F.72%).

As evident from Figures 4 and 5, in the range 200 to 300 MPa of hoop stress applied all the curves
(with few exceptions) are similar. Once again the reason can be found in the set of experimental
data used for their development and validation, data lying just in correspondence of this range of
stresses.

2.2.1 Battelle Two Curve Method.

The first semi empirical physical model for DFPC was proposed by Battelle researcher Maxey:
“Battelle Two Curve Model” /4/. This method is based upon the comparison of the force driving the
ductile fracture propagation to that resisting it. For clarification, it should be noted that Equation 1
was a statistical fit to data developed from the “Battelle Two Curve Model”.
• The first (Driving Curve) is represented by the gas decompression curve and therefore is
dependent on the initial gas pressure, temperature and chemical composition.
• The second (Resistance Curve) is dependent on the linepipe geometry and the resistance of a
fracture to ductile propagation, characteristic of the steel under consideration.
In practice the approach, also known as “Battelle Two Step Analysis”, assumes that the gas
decompression behaviour and the dynamic crack propagation behaviour are uncoupled processes
that can be related through the fracture propagation speed in the pipeline.
The approach permits the method to be applied to rich gases involving two phase decompression
behaviour. The method was finalized for single phase (as air, nitrogen or essentially pure methane)
medium. However the validity of the method is general, and it can be applied to rich gas cases,
providing that suitable gas decompression models are available.
The approach gives a toughness value that ensures a propagating shear fracture will be stopped,
without giving an arrest distance; in the case of a stable propagation event this model can also
provide the value of the stable propagation crack velocity.

Driving Curve

For ideal gases, it is possible to determine the force driving the ductile fracture propagation, using
this analytical solution:

 2  γ − 1  V  γ −1
Pd = Pi  +    9
 γ + 1  γ + 1  Va 

where
Pd : decompressed pressure
Pi: initial pressure
V: decompression velocity associated with Pd
Va: acoustic velocity at the original temperature and pressure
γ: specific heat ratio at the original temperature and pressure

The hypotheses used for deriving equation (9) are based on the ideal gas law and they imply that:
• the process is isentropic;
• a sudden full pipe cross sectional opening occurs;
• the fluid mixture remains homogenous.
11
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

The gas decompression curve calculated with equation (9) is valid for an ideal gas; experiments
with air, nitrogen and natural gas with high methane content proved this equation works well. For
gases exhibiting liquid or two-phase behaviour during decompression (as for gas rich in heavy
components) this equation can not be used, and more complete equations of state are needed.

Resistance Curve

Maxey et al. /4/ has shown that the upper shelf energy of the Charpy V specimen can be used as an
estimate of the resistance of a fracture to ductile propagation. Correlation between the material’s
resistance fracture Kc2 and the Charpy V energy was found

Kc2 = 12 Cv E / Ac 10

Where
E = Young’s modulus
Cv = Charpy V upper shelf energy (2/3 Charpy V specimen)
Ac = fracture surface area of the 2/3 Charpy V specimen

The units are in in-lbs/in2. Equation (10) is considered valid for most material except quenched and
tempered steels /1/.

Maxey demonstrated that the ductile fracture arrest stress can be determined knowing the
propagation toughness resistance, the material’s yield strength and pipe geometry. The idea was to
assume the fracture could be treated as a static fracture initiation problem (given the slow fracture
velocity near arrest) for a through wall defect. Therefore the arrest stress could be determined with
Equation 11, an equation similar to those used for fracture initiation /4/.

  πK c2  
2 ⋅ σ flow
⋅ acos e  flow  
 8 cσ 2 
σ arrest =
  11
M Tπ
 
where
Kc2 : fracture resistance parameter;
t: wall thickness;
σarrest: arrest stress;
σflow: flow stress (assumed equal to yield strength + 10 ksi, i.e. yield strength+68.95 MPa);
MT: Folias correction factor, a function of 2c/(Rt)1/2.
c: half crack length.

To use this approach, both the Folias factor and the half crack length had to be determined.
In the chart in Figure 6 are plotted a large number of data obtained in experiments conducted at
Battelle /4/ for determining the arrest stress level using the Charpy V energy per unit area as the
material’s fracture resistance.
The abscissa of this chart is the normalized toughness parameter, whereas the ordinate is the ratio of
the decompressed stress level in the vicinity of the crack tip to the σflow value.

12
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Figure 6: Determination of the arrest stress level (Ref. /4/)

After iterative calculations using the above mentioned equation, the curve separating propagate
zone from arrest one was found, and the effective half-crack length and the corresponding Folias
factor were identified. Finally the equation for determining the arrest stress level was developed as:

2 ⋅ σ flow  24⋅ A− C⋅σ ⋅ E ⋅π V



⋅ arccos e 
2
σ arrest
Rt
= c flow 12
3.33 ⋅ π  
 
The basic concept governing the ductile fracture velocity is believed to be covered by the theory of
plastic wave propagation. This concept considers the ductile fracture and the plastic zone ahead of
the fracture to be equivalent to some large value of plastic strain that can be described by small-
scale material properties. The concept also considers that this high level of plastic strain can only
propagate as fast as the plastic wave.

On the basis of curve fitting of experimental data /4/ the relationship of fracture velocity to pressure
was taken as:
1
σ flow  P 6
Vf = K  − 1 13
Cv  Pa 
Ac

where
Vf: fracture velocity corresponding to pressure P
P: pressure at the crack tip
Pa: arrest pressure, equal to (σarrest t) / R, with R=pipe radius and t=wall thickness
K: constant depending on backfill.
Using American Units (as in the original Maxey’s paper /4/) of ksi (pressure and
stress), ft⋅s-1(crack velocity) and in⋅lb/in2 (Cv/Ac), the numerical value for backfill
constant are:

13
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Soil Backfill = 0.47


Sea Backfill = 0.4
No Backfill = 0.648
Using SI Units of MPa (pressure and stress), m⋅s-1(crack velocity) and Joule/cm2
(Cv/Ac), the numerical value for backfill constant are:
Soil Backfill = 2.46
Sea Backfill = 2.09
No Backfill = 3.39

As mentioned previously, the parameter K represents the external constraint and changes according
to the environment: No backfill, Soil backfill and Sea backfill.
Proposed numerical values for both No Backfill and Soil Backfill conditions come from the work of
Maxey /10/ where two sets of full scale burst tests data were used to calculate the value of K, one
for pipe having approximately 30 inches (1 meter) of soil and sand cover (backfill) and the second
for a pipe lying in an open ditch with no soil around it. Finally, about the value of K proposed for
Sea (or Water) Backfill a short background and the full scale burst test data used for its evaluation
are reported in paragraph n. 5.

The concept of the Battelle Two Curve Method provides that from a combination of equations (12)
and (13) the Resistance Curve can be obtained and finally compared with the Driving Curve shown
earlier through the fracture propagation speed in the pipeline.
Figure 7 shows the typical diagrams associated with the Battelle Two Curve Model; for a given
pipeline at its operating condition the minimum Charpy V energy value for having an arrest
corresponds to the tangency between the gas decompression curve (Driving Curve) and the pressure
/crack velocity plot (Resistance Curve). The value of the speed corresponding to the tangency point
represents the minimum crack speed sustainable during stable crack propagation.

Maxey et al /50/ state that for newer materials (high toughness, separations, higher yield strength
etc.) the simple fracture resistance determination (Equation 10) was not applicable. However the
individual decompression curve and the fracture curve as determined by Equations (12) and (13) are
still applicable.

14
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Battelle 2 Curves Approach

120

Natural gas
100
Pressure (bar)

80

60
Charpy V 90J
Charpy V 70J
40
Charpy V 50J
Gas decompression curve
20

0
0 50 100 150 200 250 300 350 400 450
Decompression speed (m/s)
Figure 7 – Battelle Two Curve Model concept

3. Full Scale Ductile Fracture Propagation Burst Tests Database.

Table 1 contains a total of 281 burst test results on longitudinal and spiral welded pipes; all these
results were published in technical magazines and/or in proceedings of international published
conferences. This database includes the EPRG one. Data on tests carried out using rich gas are also
included. At the end of the Table the list of references used to collect the data is reported.
Data from tests showing partially brittle fracture, tests designed without backfill or with frozen
backfill are not included.
These data cover a diameter range from 16” to 56” (406mm to 1422mm), a wall thickness range
from 6.4 to 30.5mm, API grades from X52 to X100 with most of the data being for grades X60 to
X80, and Charpy V energies from 24 Joule to 278 Joule (actual 1/1 specimen or equivalent).

No. Diameter Wall thickness API Pressure Hoop stress Test Actual CV 1/1
No. Test Pipe Weld Gas Ref.
Test (mm) (mm) Grade (bar) (MPa) results energy (J)

1 AISI SF-1 SF1-1 L 1219 18.3 X65 90 A 301 P 45 (2)


2 AISI SF-10 2411 L 1067 9.5 X65 45 A 250 A 61 (2)
3 AISI SF-10 2937 L 1067 9.5 X65 45 A 250 A 69 (2)
4 AISI SF-11 5 L 610 8.7 X65 112 A 391 P 60 (1)
5 AISI SF-11 9 L 610 8.7 X65 112 A 391 P 68 (1)
6 AISI SF-11 34 L 610 8.7 X65 112 A 391 A 84 (1)
7 AISI SF-2 2M5 L 1219 15.9 X75 97 A 371 A 61 (2)
8 AISI SF-2 3M3 L 1219 18.3 X65 97 A 322 A 87 (2)
9 AISI SF-2 6M2 L 1219 18.3 X65 97 A 322 P 57 (2)
10 AISI SF-3 C1001 L 762 8.0 X60 69 A 330 A 75 (2)

15
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

11 AISI SF-3 C1002 L 762 8.0 X60 69 A 330 P 49 (2)


12 AISI SF-3 C1003 L 762 8.0 X60 69 A 330 P 49 (2)
13 AISI SF-3 C1004 L 762 8.0 X60 69 A 330 A 71 (2)
14 AISI SF-4 2M1/M2 L 1219 11.9 X90 87 A 444 P 110 (1)
15 AISI SF-4 2M6/2M7 L 1219 15.9 X75 87 A 334 A 136 (2)
16 AISI SF-6 5-1 L 1067 15.1 X60 69 A 244 A 61 (2)
17 AISI SF-8 79574 L 1067 12.4 ** 94 A 402 P 89 (1)
18 AISI SF-8 79576 L 1067 12.4 ** 94 A 402 P 145 (1)
19 AISI SF-8 79577 L 1067 12.4 ** 94 A 402 A 167 (1)
20 AISI SF-8 79578 L 1067 12.4 ** 94 A 402 P 115 (1)
21 AISI SF-9 1401 L 1067 9.5 X65 65 A 362 P 62 (1)
22 AISI SF-9 1939 L 1067 9.5 X65 65 A 362 A 90 (1)
23 AISI SF-9 2941 L 1067 9.5 X65 65 A 362 P 65 (1)
24 AISI GL(1) L 914 9.5 X65 64 N 308 P 35 (2)
25 AISI NO L 610 6.4 X52 54 N 258 A 59 (2)
26 Alaska EEG L 1219 18.3 X70 86.9 B 289 P 90 (5)
27 Alaska EFA L 1219 18.3 X70 86.9 B 289 A 105 (5)
28 Alliance 1 2E S 914 14.2 X70 120.2 B 387 A 223 (5)
29 Alliance 1 1E L 914 14.2 X70 120.2 B 387 P 185 (5)
30 Alliance 1 2W L 914 14.2 X70 120.2 B 387 A 237 (5)
31 Alliance 1 1W L 914 14.2 X70 120.2 B 387 P 180 (5)
32 Alliance 2 2E L 914 14.2 X70 120.8 B 389 A 226 (5)
33 Alliance 2 1E L 914 14.2 X70 120.8 B 389 P 194 (5)
34 Alliance 2 2W L 914 14.2 X70 120.8 B 389 A 217 (5)
35 Alliance 2 1W L 914 14.2 X70 120.8 B 389 P 194 (5)
36 Batt 1 A-3 L 1220 18.3 X70 115.8 B 386 P 89 (5)
37 Batt 1 A-9 L 1220 18.3 X70 115.8 B 386 P 190 (5)
38 Batt 1 A-20 L 1220 18.3 X70 115.8 B 386 P 206 (5)
39 Batt 1 A-23 L 1220 18.3 X70 115.8 B 386 P 203 (5)
40 Batt 1 A-16 L 1220 18.3 X70 115.8 B 386 P 123 (5)
41 Batt 1 A-8 L 1220 18.3 X70 115.8 B 386 P 96 (5)
42 Batt 1 A-18 L 1220 18.3 X70 115.8 B 386 P 150 (5)
43 Batt 1 A-22 L 1220 18.3 X70 115.8 B 386 P 222 (5)
44 Batt 2 A-15 L 1220 18.3 X70 116.2 B 387 P 206 (5)
45 Batt 2 B-4 L 1220 18.3 X70 116.2 B 387 P 193 (5)
46 Batt 2 A-21 L 1220 18.3 X70 116.2 B 387 P 201 (5)
47 Batt 2 B-7 L 1220 18.3 X70 116.2 B 387 P 179 (5)
48 Batt 2 B-5 L 1220 18.3 X70 116.2 B 387 P 198 (5)
49 Batt 2 B-6 L 1220 18.3 X70 116.2 B 387 A 176 (5)
50 BGC 80 L 762 12.7 X60 68 A 205 A 49 (2)
51 BGC 80 L 762 12.7 X60 68 A 205 P 33 (2)
52 BGC 88 L 914 15.9 X60 48 A 139 A 47 (2)
53 BGC 100 L 914 12.7 X60 74 N 266 P 27 (2)
54 BGC 100 L 914 12.7 X60 74 N 266 P 26 (2)
55 BGC 100 L 914 12.7 X60 74 N 266 P 28 (2)
56 BGC 114 L 610 9.5 X52 70 N 223 P 26 (2)

16
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

57 BGC 114 L 610 9.5 X52 70 N 223 P 28 (2)


58 BGC 116 L 610 9.5 X52 68 N 217 P 33 (2)
59 BGC 118 L 610 9.5 X52 59 N 188 P 24 (2)
60 BGC 118 L 610 9.5 X52 59 N 188 P 24 (2)
61 BGC 119 L 610 9.5 X52 49 N 156 A 24 (2)
62 BGC 120 L 914 12.7 X60 69 N 247 A 49 (2)
63 BGC 120 L 914 12.7 X60 69 N 247 A 41 (2)
64 BGC 121 L 914 12.7 X60 54 N 196 A 28 (2)
65 BGC 122 L 914 12.7 X60 79 N 283 A 56 (2)
66 BGC 122 L 914 12.7 X60 79 N 283 A 62 (2)
67 BGC 141 L 610 9.5 X52 69 N 221 A 43 (2)
68 BGC 141 L 610 9.5 X52 69 N 221 A 37 (2)
69 BGC GL(2) L 914 9.5 X65 59 N 283 P 35 (2)
70 BGC Y L 762 12.7 X60 45 N 134 A 28 (2)
71 BGC Y L 762 12.7 X60 45 N 134 A 30 (2)
72 BMI AT20 FF2-1 L 762 9.5 X52 64 N 255 A 45 (2)
73 BMI AT21 CC1-1 L 762 9.5 X100 75 N 301 P 34 (2)
74 BMI AT21 T2 L 914 9.5 X60 75 N 361 A 65 (2)
75 BMI AT22 HM3-1 L 762 9.5 X52 62 N 248 A 47 (2)
76 BMI AT25 RR6-1 L 762 9.5 X52 63 N 251 A 61 (2)
77 BMI AT25 SS1-1 L 762 7.9 X60 63 N 302 A 53 (2)
78 BMI AT27 BF-2 L 914 9.5 X65 63 N 301 P 40 (2)
79 BMI AT27 BF-3 L 914 9.5 X65 63 N 301 P 34 (2)
80 BMI AT27 BF-4 L 914 9.5 X65 63 N 301 A 53 (2)
81 BMI AT27 BJ-1 L 914 9.9 X65 63 N 291 P 45 (2)
82 BMI AT27 BJ-2 L 914 9.9 X65 63 N 291 A 47 (2)
83 BMI AT28 BF-7 L 914 9.5 X65 55 N 265 P 28 (2)
84 BMI AT28 BL-3 L 914 8.4 X65 55 N 301 P 45 (2)
85 BMI AT28 BO-2 L 914 9.1 X65 55 N 276 A 54 (2)
86 BMI AT29 BT-1 L 1067 9.9 X60 56 N 300 P 51 (2)
87 BMI AT29 BT-2 L 1067 9.9 X60 56 N 300 P 44 (2)
88 BMI AT29 BT-3 L 1067 9.9 X60 56 N 300 P 51 (2)
89 BMI AT29 BT-5 L 1067 9.9 X60 56 N 300 P 49 (2)
90 BMI AT29 BT-8 L 1067 9.9 X60 56 N 300 P 51 (2)
91 BMI AT31 CA1 -2 L 914 8.4 X65 79 N 433 P 44 (1)
92 BMI AT31 CA3 L 914 8.4 X65 79 N 433 P 42 (1)
93 BMI AT32 AJ4 L 914 9.9 X65 59 N 270 A 45 (2)
94 BMI AT32 BL4 L 914 8.4 X65 59 N 320 P 43 (2)
95 BMI AT32 CA6 L 1067 8.4 X65 59 N 373 P 48 (2)
96 BMI AT33 CF5 S 1067 9.4 X65 73 N 411 A 156 (1)
97 BMI AT33 CF6 S 1067 9.4 X65 73 N 411 A 150 (1)
98 BMI AT34 CM1 L 914 11.4 X100 104 N 416 P 69 (1)
99 BMI AT35 CQ3 L 406 7.9 X52 80 N 206 A 54 (2)
100 BMI AT35 CQ5 L 914 7.9 X52 80 N 463 A 59 (2)
101 BMI AT35 CQ6 L 660 7.9 X52 80 N 335 P 47 (2)
102 BMI AT36 CV1 L 406 7.9 X60 150 N 385 P 35 (2)

17
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

103 BMI AT36 CV3 L 406 7.9 X60 150 N 386 P 41 (1)
104 BMI AT36 CV4 L 406 7.9 X60 150 N 386 P 39 (1)
105 BMI AT37 CS6 L 914 11.1 X60 89 N 366 A 53 (1)
106 BMI AT37 CS7 L 914 11.1 X60 89 N 366 P 51 (1)
107 BMI AT37 CS9 L 914 11.1 X60 89 N 366 P 53 (1)
108 BMI AT38 CW1 L 508 7.4 X60 106 N 363 P 39 (1)
109 BMI AT38 CW2 L 508 7.4 X60 106 N 363 P 39 (1)
110 BMI AT38 CW4 L 508 7.4 X60 106 N 363 P 39 (1)
111 BMI GA1 8 L 1219 15.7 X70 99 N 387 P 121 (1)
112 BMI GA1 10 L 1219 15.7 X70 99 N 387 A 126 (1)
113 BMI GA1 11 L 1219 15.7 X70 99 N 387 A 153 (1)
114 BMI GA1 12 L 1219 15.7 X70 99 N 387 P 90 (1)
115 BMI GA1 14 L 1219 15.7 X70 99 N 387 P 86 (1)
116 BMI GA2 10-S L 1219 16.9 X65 100 N 361 P 114 (1)
117 BMI GA2 12-S L 1219 16.9 X65 100 N 361 P 103 (1)
118 BMI GA2 2-S L 1219 16.9 X65 100 N 361 A 162 (1)
119 BMI NW1 KL17 L 1219 11.7 X65 70 N 364 A 102 (1)
120 BMI NW1 KL19 L 1219 11.7 X65 70 N 364 P 63 (1)
121 BMI NW1 KL5 L 1219 11.7 X65 70 N 364 A 78 (1)
122 BMI NW2 TS11 S 1219 11.7 X65 71 N 367 A 159 (1)
123 BMI NW2 TS6 S 1219 11.7 X65 71 N 367 A 223 (1)
124 BP 83 3-1 L 914 24.0 X65 172 B 328 P 102 (5)
125 BP 83 4.2 L 914 24.0 X65 172 B 328 P 126 (5)
126 CSM 2 1 L 1422 16.9 X70 87 A 366 P 85 (1)
127 CSM 2 18334 L 1422 17.0 X70 87 A 364 A 133 (1)
128 CSM 2 1 bis L 1422 17.5 X70 87 A 353 P 112 (1)
129 CSM 3 3 L 1422 20.0 X70 103 A 366 P 62 (1)
130 CSM 3 429 L 1422 20.0 X70 103 A 366 P 91 (1)
131 CSM 3 4341 L 1422 20.0 X70 103 A 366 P 100 (1)
132 CSM 4 9 L 1220 20.0 X70 118 A 360 P 80 (1)
133 CSM 4 11 L 1220 21.0 X70 118 A 343 A 97 (1)
134 CSM 5 2260 L 1220 17.0 X70 99 A 355 A 157 (1)
135 CSM 5 2484 L 1220 17.0 X70 99 A 355 P 97 (1)
136 CSM 5 5326 L 1220 17.0 X70 99 A 355 P 39 (1)
137 CSM 5 8578 L 1220 17.0 X70 99 A 355 P 39 (1)
138 CSM 7 58 L 1422 17.0 X70 92 A 385 P 102 (1)
139 CSM 7 68 L 1422 17.0 X70 92 A 385 A 116 (1)
140 CSM 9 64 S 1422 17.0 X70 90 A 375 A 92 (1)
141 CSM 12 2bis L 1422 17,0 X75 89 N 373 P 61 (1)
142 CSM 12 4bis L 1422 17,0 X75 89 N 373 P 57 (1)
143 CSM 12 4IT L 1422 17,0 X75 89 N 373 P 89 (1)
144 CSM 12 5IT L 1422 17,0 X75 89 N 373 P 165 (1)
145 CSM 12 7V L 1422 17,0 X75 89 N 373 P 96 (1)
146 CSM 12 8V L 1422 17,0 X75 89 N 373 P 147 (1)
147 CSM 14 A L 1422 17,0 X75 94 N 391 P 137 (1)
148 CSM 14 B L 1422 17,0 X75 94 N 391 P 141 (1)

18
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

149 CSM 14 C L 1422 17,0 X75 94 N 391 P 95 (1)


150 CSM 14 D L 1422 17,0 X75 94 N 391 P 137 (1)
151 CSM 14 F L 1422 17,0 X75 94 N 391 P 95 (1)
152 CSM 14 G L 1422 17,0 X75 94 N 391 A 175 (1)
153 CSM 15 14 L 1422 20,0 X75 104 A 371 P 134 (1)
154 CSM 15 21 L 1422 18,0 X75 104 A 413 P 194 (1)
155 CSM 15 24 L 1422 18,0 X75 104 A 413 A 210 (1)
156 CSM 15 26 L 1422 18,0 X75 104 A 413 P 139 (1)
157 CSM 15 27 L 1422 18,0 X75 104 A 413 P 147 (1)
158 CSM 15 36 L 1422 20,0 X75 104 A 371 A 187 (1)
*
159 CSM R60 25 L 1422 18.7 R60 92 A 350 P 76 (1)
*
160 CSM R60 30 L 1422 18.7 R60 92 A 350 A 107 (1)
*
161 CSM R60 31 L 1422 18.7 R60 92 A 350 A 124 (1)
*
162 CSM R65 13A L 1422 18.0 R65 100 A 395 P 163 (1)
*
163 CSM R65 2B L 1422 20.0 R65 100 A 355 A 138 (1)
*
164 CSM R65 9A L 1422 18.0 R65 100 A 395 A 195 (1)
165 CSM 23 1-H L 1422 30.5 X70 175 A 408 A 172 (3,6)
(3,6)
166 CSM 23 4-8501 L 1422 30.5 X70 175 A 408 A 228
(3,6)
167 CSM 23 9-8496 L 1422 30.5 X70 175 A 408 P 168
(3,6)
168 CSM 26 13-67273 L 1422 26.5 X80 161 A 432 P 88
(3,6)
169 CSM 26 14-67276 L 1422 26.0 X80 161 A 440 P 133
(3,6)
170 CSM 26 19-67272 L 1422 25.7 X80 161 A 446 P 229
(3,6)
171 CSM 26 21-67279 L 1422 26.1 X80 161 A 439 A 278
(3,6)
172 CSM 26 22-67277 L 1422 26.1 X80 161 A 439 P 144
(3,6)
173 CSM 26 24-67274 L 1422 26.1 X80 161 A 439 P 206
(3,6)
174 CSM 26 9-67284 L 1422 25.7 X80 161 A 446 A 272
175 CSM 27 846058 L 1422 19.1 X100 126 A 469 P 170 (4,6)
176 CSM 27 846113 L 1422 19.1 X100 126 A 469 P 151 (4,6)
177 CSM 27 846157 L 1422 19.1 X100 126 A 469 A 263 (4,6)
178 EPRG 1 AC 4460 L 914 12.7 X60 103 N 371 P 70 (1)
179 EPRG 1 AC 7770 L 914 12.7 X60 103 N 371 A 144 (1)
180 EPRG 2 AC4406 L 914 12.7 X65 81 N 290 A 61 (2)
181 EPRG 2 RD503 L 914 12.7 X65 81 N 290 A 41 (2)
182 EPRG 3 567/39 S 914 12.7 X60 103 N 371 A 77 (1)
183 EPRG 3 AC 3742 L 914 12.7 X60 103 N 371 A 57 (1)
184 EPRG 4 AC 18278 L 914 12.7 X60 115 N 414 A 144 (1)
185 EPRG 4 AC 2775 L 914 12.7 X60 115 N 414 A 58 (1)
186 EPRG 7 1114 L 1219 15.9 X65 97 N 372 P 72 (1)
187 EPRG 7 1135 L 1219 15.9 X65 97 N 372 P 84 (1)
188 EPRG 7 9952 L 1219 15.9 X65 97 N 372 P 128 (1)
189 EPRG 8 1123 L 1219 15.9 X65 97 N 372 P 80 (1)
190 EPRG 9 A 2793 S 1219 15.9 X65 92 N 353 A 105 (1)
191 EPRG 9 A 2796 S 1219 15.9 X65 92 N 353 A 144 (1)
192 EPRG 11 10951 L 1219 15.9 X65 93 N 358 P 89 (1)
193 EPRG 11 10954 L 1219 15.9 X65 93 N 358 A 82 (1)
194 EPRG 12 4751 S 1219 15.9 X65 93 N 358 A 102 (1)

19
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

195 EPRG 12 9857 S 1219 15.9 X65 93 N 358 A 95 (1)


196 EPRG III.2 2002 L 1219 17.0 X80 112 N 402 P 88 (1)
197 EPRG III.2 2003 L 1219 17.0 X80 112 N 402 P 134 (1)
198 EPRG III.2 2016 L 1219 17.0 X80 112 N 402 P 221 (1)
199 EPRG III.2 2017 L 1219 17.0 X80 112 N 402 A 169 (1)
200 EPRG III.2 2018 L 1219 17.0 X80 112 N 402 P 159 (1)
201 EPRG III.3 2 L 1219 19.0 X80 118 N 379 A 155 (1)
202 EPRG III.3 10066 L 1219 17.5 X80 118 N 411 A 189 (1)
203 EPRG III.3 10070 L 1219 16.5 X80 118 N 436 P 121 (1)
204 EPRG III.3 10076 L 1219 17.7 X80 118 N 406 P 143 (1)
205 EPRG IV.1 20-1 L 1422 18.8 X65 107 N 407 A 160 (1)
206 EPRG IV.1 20-2 L 1422 18.3 X65 107 N 418 P 133 (1)
207 EPRG IV.1 20-4 L 1422 18.5 X65 107 N 413 P 111 (1)
208 EPRG IV.1 20-5 L 1422 18.3 X65 107 N 418 P 126 (1)
209 EPRG IV.1 20-9 L 1422 19.0 X65 107 N 402 A 238 (1)
210 EPRG 6 738A S 1220 15.9 X65 96.8 N 371 A 81 (5)
211 EPRG 6 737A S 1220 15.9 X65 96.8 N 371 A 84 (5)
212Foothills 3 2-W L 1219 15.2 X70 86.9 B 348 P 107 (5)
213Foothills 3 3-W L 1219 15.2 X70 86.9 B 348 P 147 (5)
214Foothills 3 4-W L 1219 15.2 X70 86.9 B 348 P 159 (5)
215Foothills 3 5-W L 1219 15.2 X70 86.9 B 348 A 182 (5)
216Foothills 3 2-E L 1219 15.2 X70 86.9 B 348 P 110 (5)
217Foothills 3 3-E L 1219 15.2 X70 86.9 B 348 P 141 (5)
218Foothills 3 4-E L 1219 15.2 X70 86.9 B 348 P 151 (5)
219Foothills 4 2-E S 1219 13.7 X70 86.9 B 386 P 157 (5)
220Foothills 4 3-E S 1219 13.7 X70 86.9 B 387 A 166 (5)
221Foothills 4 2-W S 1219 13.7 X70 86.9 B 386 A 157 (5)
222Foothills 5 2-W S 1422 13.7 X70 74.5 B 386 P 105 (5)
223Foothills 5 3-W S 1422 13.7 X70 74.5 B 386 P 143 (5)
224Foothills 5 4-W S 1422 13.7 X70 74.5 B 386 A 147 (5)
225Foothills 5 2-E S 1422 13.7 X70 74.5 B 386 P 115 (5)
226Foothills 5 3-E S 1422 13.7 X70 74.5 B 386 P 140 (5)
227Foothills 5 4-E S 1422 13.7 X70 74.5 B 386 P 166 (5)
228Foothills 5 5-E S 1422 13.7 X70 74.5 B 386 P 190 (5)
229Foothills 6 2-W L 1219 15.2 X70 86.9 B 348 P 125 (5)
230Foothills 6 3-W L 1219 15.2 X70 86.9 B 348 P 149 (5)
231Foothills 6 4-W L 1219 15.2 X70 86.9 B 348 P 165 (5)
232Foothills 6 5-W L 1219 15.2 X70 86.9 B 348 A 173 (5)
233Foothills 6 2-E L 1219 15.2 X70 86.9 B 349 P 147 (5)
234Foothills 6 3-E L 1219 15.2 X70 86.9 B 348 P 155 (5)
235Foothills 6 4-E L 1219 15.2 X70 86.9 B 348 P 190 (5)
236Foothills 6 5-E L 1219 15.2 X70 86.9 B 348 P 112 (5)
237Foothills 7 2-W S 1219 13.7 X70 81.4 B 362 A 180 (5)
238Foothills 7 4-E S 1219 13.7 X70 81.4 B 362 A 215 (5)
239Foothills 7 2-E S 1219 13.7 X70 81.4 B 362 P 176 (5)
240Foothills 7 3-E S 1219 13.7 X70 81.4 B 362 P 214 (5)

20
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

241 JISI A1 N1 L 1219 18.3 X70 116 A 386 P 92 (1)


242 JISI A1 N2 L 1219 18.3 X70 116 A 386 A 188 (1)
243 JISI A1 S1 L 1219 18.3 X70 116 A 386 P 83 (1)
244 JISI A1 S2 L 1219 18.3 X70 116 A 386 P 140 (1)
245 JISI A1 S3 L 1219 18.3 X70 116 A 386 A 202 (1)
246 JISI A2 N1 L 1219 18.3 X70 116 A 386 P 108 (1)
247 JISI A2 N2 L 1219 18.3 X70 116 A 386 P 169 (1)
248 JISI A2 N3 L 1219 18.3 X70 116 A 386 A 275 (1)
249 JISI A2 S1 L 1219 18.3 X70 116 A 386 P 119 (1)
250 JISI A2 S2 L 1219 18.3 X70 116 A 386 P 139 (1)
251 JISI A2 S3 L 1219 18.3 X70 116 A 386 A 196 (1)
252 JISI A3 N1 L 1219 18.3 X70 116 A 386 P 122 (1)
253 JISI A3 N2 L 1219 18.3 X70 116 A 386 P 153 (1)
254 JISI A3 N3 L 1219 18.3 X70 116 A 386 A 202 (1)
255 JISI A3 S1 L 1219 18.3 X70 116 A 386 P 127 (1)
256 JISI A3 S2 L 1219 18.3 X70 116 A 386 P 103 (1)
257 JISI A3 S3 L 1219 18.3 X70 116 A 386 A 209 (1)
258 JISI B1 N1 L 1219 18.3 X70 116 A 386 P 155 (1)
259 JISI B1 N2 L 1219 18.3 X70 116 A 386 A 156 (1)
260 JISI B1 S1 L 1219 18.3 X70 116 A 386 P 124 (1)
261 JISI B1 S2 L 1219 18.3 X70 116 A 386 A 126 (1)
262 JISI B2 N1 L 1219 18.3 X70 116 A 386 P 125 (1)
263 JISI B2 N2 L 1219 18.3 X70 116 A 386 A 129 (1)
264 JISI B2 S1 L 1219 18.3 X70 116 A 386 P 118 (1)
265 JISI B2 S2 L 1219 18.3 X70 116 A 386 A 145 (1)
266 MW 1 20085 L 1219 15.9 X75 93 N 356 A 94 (1)
267 MW 1 20087 L 1219 15.9 X75 93 N 356 A 188 (1)
268 MW 2 67473 L 1219 25.4 X75 162 N 388 P 124 (1)
269 MW 2 67474 L 1219 25.4 X75 162 N 388 P 132 (1)
270 MW 2 67479 L 1219 25.4 X75 162 N 388 A 138 (1)
271 MW 2 67493 L 1219 25.4 X75 162 N 388 A 173 (1)
272 USS1 USS1 I-1 L 914 14.2 X65 92 A 298 P 37 (2)
273 USS1 I-2 L 914 14.2 X65 92 A 298 A 98 (2)
274 USS1 I-3 L 914 14.2 X65 92 A 298 A 51 (2)
275 USS1 I-4 L 914 14.2 X65 92 A 298 A 87 (2)
276 1 L 1422 17.5 X70 83 N 337 A 106 (2)
277 2 L 1422 17.5 X70 90 N 365 A 134 (2)
278 L 1422 17.5 X70 83 N 337 P 59 (2)
279 L 1422 17.5 X70 90 N 365 P 71 (2)
280 L 914 12.7 X60 60 N 216 A 33 (2)
281 L 914 12.7 X60 60 N 216 A 39 (2)
Table 1 – Full scale burst tests database7

7
Generally for the majority of the test data the actual steel properties (yield and tensile strength) were not published.
The version of this Table in Excel includes measured material properties for few cases (some CSM tests) where these
data were published.

21
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Notation:
• Gas: N= natural gas (methane or lean gas), A = air, B = rich gas;
• CV (J): full size (1/1) specimen Charpy V energy or converted from original 2/3 CV
specimen data by area ratio.
• Test result: P=propagation, A=Arrest;
• Weld: L=longitudinal welded pipe, S=Spiral welded pipe.
• *: Steel grade expressed according to GOST 1497 “Metals, method of tensile
tests”, 1988 (ex-URSS Standard):
- “R60” means steel with minimum ultimate tensile strength of 60 kg/mm2,
- “R65” means steel with minimum ultimate tensile strength of 65 kg/mm2.
• ** : Steel grade unknown

References to Table 1:
(1) G. Re, V. Pistone, G. Demofonti, and D.G. Jones "EPRG Recommendation for crack arrest
toughness for high strength line pipe steels" 3R International - 34, 1995.
Information from published data on burst test at hoop stress exceeding 350 MPa. Data from tests showing partially
brittle fracture, tests designed without backfill or with frozen backfill or containing rich gas were excluded. All results
are for pipe tested with single phase expanding gas, i.e. air or dry natural gas.

(2) G.D: Fearnehough, D.J. Jones “Toughness specification for shear fracture arrest in pipelines"
Intern. Conf. on Analytical and Experimental Fracture Mechanics, June 23-27, Rome, 1980.
Data are restricted to backfilled tests on longitudinally welded pipes pressurised with methane with two exceptions: one
non-backfilled result (BMI test A35) was included because it showed fracture arrest (conservative result). Tests using
air as pressurising medium (19 AISI tests and 3 BGC tests) were included.

(3) G. Mannucci, G. Demofonti, M.R. Galli, C.M. Spinelli "Structural integrity of API 5L X70-X80
steel grade pipeline for high pressure long distance transmission gas lines" International Gas
Research Conference, 8-11 November, San Diego, USA, 1998.
All tests included

(4) G. Demofonti, G. Mannucci, C.M. Spinelli, L. Barsanti and H.G. Hillenbrand “Large diameter
X100 gas linepipes: fracture propagation evaluation by full-scale burst test” Proced. Of the 3rd
International Pipeline Technology Conf., Brugge, Belgium, May 21-24, 2000.
All tests included

(5) V. Pistone, G. Mannucci “Fracture arrest criteria for spiral welded pipes” Proced. Of the 3rd
International Pipeline Technology Conf., Brugge, Belgium, May 21-24, 2000.
All tests included

(6) G. Buzzichelli “Designing against Ductile Fracture Propagation in Very High Strength Steel Gas
Pipelines: A Review” 13th European Conference on Fracture (ECF13), 6 – 9 September 2000, San
Sebastian, Spain.
All tests included

3.1 Comparison of test results with predictions.

In the following Figures the predictive capability of each equations is shown comparing the actual
Charpy V energy with the predicted one for both arrest and propagation pipes. The data used for
those Figures come from Table 1 and refer to tests where no rich gas was involved.

22
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Battelle equation
300

Arrest
Propagate 1:1
250
Predicted Charpy V Energy 1/1 (J)
Propagation

200

150

100

50
Arrest
0
0 50 100 150 200 250 300
Actual Charpy V Energy 1/1 (J)

Figure 8

AISI equation
300

Arrest
Propagate 1:1
250
Predicted Charpy V Energy 1/1 (J)

Propagation

200

150

100

50
Arrest
0
0 50 100 150 200 250 300
Actual Charpy V Energy 1/1 (J)

Figure 9

23
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

British Gas 95 equation


300
Propagate 1:1

Predicted Charpy V Energy 1/1 (J)


Arres
250
Propagati
200

150

100

50
Arrest
0
0 50 100 150 200 250 300
Actual Charpy V Energy 1/1 (J)

Figure 10

Mannesmann equation
300

Arrest 1:1
Predicted Charpy V Energy 1/1 (J)

250
Propagation
Propagate
200

150

100

50
Arrest
0
0 50 100 150 200 250 300
Actual Charpy V Energy 1/1 (J)

Figure 11

24
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Japan 82 equation
300

Propagate 1:1
250
Predicted Charpy V Energy 1/1 (J)

200

150

100

50 Arrest
Arrest Propagation
0
0 50 100 150 200 250 300
Actual Charpy V Energy 1/1 (J)

Figure 12

Kawasaki Steel equation


300
Propagate 1:1
Predicted Charpy V Energy 1/1 (J)

250

200

150

100
Arrest
Arrest
50
Propagation

0
0 50 100 150 200 250 300
Actual Charpy V Energy 1/1 (J)

Figure 13

25
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Italsider / CSM equation


300
Propagate 1:1

Predicted Charpy V Energy 1/1 (J)


250

200

150

100

50 Arrest

Arrest Propagation

0
0 50 100 150 200 250 300
Actual Charpy V Energy 1/1 (J)

Figure 14

From this series of charts it is easy to note that each formula has a range of Charpy V energy values
over which is better representative for the ductile fracture propagation phenomenon. The Battelle
and AISI equations for example, give better results at Charpy V energy value up to about 100 Joule
(this is due to this data range being used at that time of their development and validation).

Table 2 summarizes the results in terms of capability of each equation to correctly predict the arrest
or propagation event.

Equation Predicted Arrest / Actual Predicted Propagate / Total


Propagate, percent Actual Arrest, percent error,
(non-conservative (conservative percent
misprediction) misprediction)
Battelle (PRCI-BMI) 16.8 8.4 25.2
AISI 23.9 8.8 32.7
British Gas 95 20.8 15.9 36.7
Mannesmann 21.7 11.5 33.2
Japan 82 (Japanese) 10.6 10.6 21.2
Kawasaki Steel 0.4 33.6 34.1
Italsider CSM 27.4 5.3 32.7
Table 2 – Summary of fracture arrest equations accuracy

The result indicates that the Japanese equation has the smallest error with the PRCI-BMI equation
very close, but in general the Table indicates that all of the equations have a broad error band. This
is not surprising because most of the data in the Table were not available when the equations were
developed and validated (for examples data referring to high grade and/or high pressure involved).

Generally the Battelle equation is by far the most widely quoted, given that it was developed by
curve fitting of a (semi empirical) physical model of the ductile fracture propagation phenomenon.
Moreover comparison performed by Pistone et al /12/ between the tabulated test results collected by

26
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

EPRG and the arrest or propagate outcomes indicated that the Battelle equation gave the most
satisfactory predictions in terms of mis-predicted data and scatter.

Together with the Battelle equation also the Japanese one demonstrates to give satisfactory results.
No relevant information is available concerning its range of applicability; the good behavior of this
equation for high value of Charpy V energy can be however justified considering that this equation
shows the highest hoop stress exponent (equal to 2.33). High Charpy V energies usually correspond
to tests carried out at high hoop stress: since the general trend of all equations is to underestimate
the Charpy V energy required to arrest for high Charpy V energies, having the highest hoop stress
exponent the Japanese equation tends to underestimate the results less than the others.

It was generally noticed and it is also evident from the Figures above, that as the toughness of the
steels increased, the predicted toughness levels at which an experimental arrest occurred tends to be
lower than that actually required (non conservative prediction). In general for all equations this
begins for toughness levels around 100 Joule, that is about the upper limit of the toughness levels
represented by the lower-toughness steels embedded in their empirical calibration. This level of
Charpy V energy is therefore frequently associated with the border value between “old”
conventional steels (with low toughness) and “new” TMCP steels (high toughness).

As evident from the analysis performed, it is not possible to identify the “best equation” in general;
what can be recommended is to extract from the database a restricted range of the data as close as
possible to the case under evaluation, and on this selected data to perform a sensitivity analysis
using all the equations above with the aim to identify the better one (i.e. that with the smallest
error).

4. EPRG Recommendations

In 1983 the European Pipeline Research Group (EPRG) published a recommendation /11/ for
toughness requirements (as measured by Charpy V shelf energy) for the arrest of a long running full
shear fracture; this recommendation was applicable to longitudinal welded gas transmission
pipelines of grade not exceeding API 5LX70 (EN 10208-2, L480).

In 1995 EPRG extended the recommendation /12/ to API 5L X80 (EN 10208-2, L550) grade steel.

The EPRG toughness recommendations are reported in Table 3, for a line operating with hoop
stress of 62.5% SMYS, and in Table 4 for hoop stress of 72% of SMYS. Both recommendations
refer to full size (10x10 mm) Charpy V transverse specimen in Joules and are based on the average
of three samples (according to ISO 83 -1976 or EN 1010 045-1).

The EPRG recommendations are based on the following considerations:


• The full ductile fracture behaviour of the crack is ensured, i.e. 85 % shear fracture area will
be reached at the minimum operation temperature in a DWT-Test according to the
recommended practice in API RP 5L3 (as shown in a later part of this document).
• The proposed Charpy V shelf energy are a result of a statistical analysis of full scale burst
tests results; the database used by EPRG to carry out the 1993 extension included 137
results on pipe up to grade API X80.
• For steel grades below API X70 the recommended toughness corresponds to 0.75 times the
AISI prediction.
• For the steel grade API X70 the recommended toughness corresponds to 0.90 times the AISI
prediction.

27
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

• The recommended minimum toughness requirements for API X80 steel are based on the
Battelle short formulae for an operating pressure of 80 bar.
• The lowest toughness level recommended is 40 J, as recommended by the European group
for standardisation (CEN).
• The EPRG recommendation is applicable to natural gas transportation pipelines. It is not
applicable for pipelines designed without backfill, with frozen backfill, or when the
transportation involves “rich gas”8.

Finally it is important to note that the EPRG recommendations are not a lower bound to the full
scale test results, they are a judgement based on the assumption that 50% of the pipes in an order
exceed 1.3 times the minimum specified value. This is supported by toughness data collected by
EPRG; in particular a statistical analysis of these data has shown that in typical “modern” pipe
where the EPRG minimum recommended Charpy V level is exceeded by a “large margin” there is a
95% probability that a fracture will arrest within 3 pipe lengths /13/.

In the event that pipes were supplied with a mean Charpy V toughness equivalent or close to the
EPRG recommended toughness, there is a 95% probability that a fracture could propagate through
many pipe lengths (i.e. > 10 pipe lengths). So a dedicated statistical analysis should be made when
it is important to evaluate the risk of the length of fracture propagation in existing old natural gas
pipelines.

Pipe outside diameter OD (mm)


Grade >510 >610 >720 >820 >920 >1020 >1120 >1220
≤510 ≤610 ≤720 ≤820 ≤920 ≤1020 ≤1120 ≤1220 ≤1430
L240 (X35)
40
L290 (X42)
40
L360 (X52) 40
42
L415 (X60)
L445 (X65) 40 42 43 47
L480 (X70) 40 41 45 48 51 53 56 58 63
L550 (X80) 48 55 61 66 72 77 82 87 96
Table 3. EPRG Recommended minimum Charpy V energies in Joules for ductile fracture arrest in pipelines operative
with a safety factor of 1.6 (62.5% SMYS)

8
The term “rich gas” is used to indicate a gas rich of heavy components, but no specific percent composition of various
higher ends can be found in literature for defining this limit. From the “Battelle Two Curve Method” point of view, a
gas can be defined rich when its decompression curve exhibits a plateau as in Figure 16. Recently Eiber /46/, basing
upon a combination of expert opinions and calculations, suggests to define a rich gas as one where the calorific value
exceeded 1050 btu/scf (39.1 MJ/m3).

28
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Pipe outside diameter OD (mm)


Grade >510 >610 >720 >820 >920 >1020 >1120 >1220
≤510 ≤610 ≤720 ≤820 ≤920 ≤1020 ≤1120 ≤1220 ≤1430
L240 (X35) 40
L290 (X42) 40
40 42
L360 (X52)
L415 (X60) 40 41 44 46 48 51
L445 (X65) 40 40 41 43 46 48 51 53 57
L480 (X70) 46 50 55 58 62 65 68 71 77
L550 (X80) 61 68 76 83 90 96 102 108 120
Table 4. EPRG Recommended minimum Charpy V energies in Joules for ductile fracture arrest in pipelines operative
with a safety factor of 1.4 (72% SMYS)

Essentially for steel grades below X70 the recommended toughness is 0.75 times the AISI
prediction; however the low toughness levels have been raised by EPRG to a minimum of 40 Joule
as recommended by the European Group for Standardization (CEN) /12/.

No data below X52 are contained in the full scale burst test database; however given that lower
grade steels are operated at lower pressure level (and hence at lower hoop stress), and being the
Charpy V energy required to arrest the fracture roughly function of the hoop stress applied (higher
hoop stresses correspond to higher Charpy V energies required), the results of the EPRG analysis
could be extrapolated towards lower steel grades.

Recently the applicability of the EPRG Recommendations to spiral welded pipelines has been
verified by analysis of spiral published data /14/.
The main results were:
• In all full scale burst tests performed with methane or air the spiral data were in good
agreement with the prediction of the relevant equation suggested by the EPRG
Recommendation (that is the AISI one, given the range of steel grade was API X60 ÷70 for
all the spiral data) for tests carried out on longitudinal welded pipes.
• For the full scale burst tests carried out using rich gas as pressurising medium the analysis
performed using the “Battelle Two Curve Model” has shown no difference of behaviour
between longitudinal and spiral pipes exists.
The analysis concluded that available literature shows that spiral welded linepipe, in terms of
fracture arrest predictability, behaves as consistently as longitudinal seam pipe.

5. The effect of deep water marine external constraint on ductile fracture propagation in gas
pipelines

5.1 Background

A quantitative description of the effect of deep water marine external constraint on ductile fracture
propagation in gas pipeline was given by Maxey’s “Battelle Two Curve Model” /4/; as mentioned
before (see paragraph n.2.2.1), the backfill constraint was introduced in the model as a K factor in
the Resistance Force term (written in terms of crack velocity, Vf).

The numerical value of K proposed by Battelle /10/ was 0.40 for the sea backfill (American Units).
This value was obtained from three full scale tests carried out in shallow water (about 10 m. deep),
Table 5. In the same years (1981 to 1987) CSM carried out two full scale tests in open sea (water
depth about 30 m) to quantify the marine external constrain in ductile fracture propagation /15/;
Table 4. The CSM results substantially confirmed the Battelle data.

29
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

The K value proposed by Battelle for soil backfill is 0.47 against 0.40 for sea backfill. This means
that for the same operation conditions the Charpy energy requirements to arrest a running crack in
an offshore gas pipeline are lower that those for soil burial (about 10%). Figure 15 shows the
“Battelle Two Curve Model” applied on the same gas pipelines (X65, 36”x19.8mm steel pipeline
pressurized with methane at 140 bar). In the same Figure the case of no-backfilled condition is also
reported.

Diameter Thickness Test line Water Test Medium CV arrest


length deep pressure pipe
(in.) (mm) (M) (m) (bar) (J/ cm2)
Battelle test 22 6.5 45 12 100 Nitrogen 45 ÷ 48
Battelle test 42 15.0 53 12 116 Nitrogen NA
Battelle test 30 9.5 57 12 63 Nitrogen NA
CSM test 20 20.0 100 30 255 Air 50 ÷ 60
CSM test 36 20.8 108 30 157 Air 60
Table 5. Summary of underwater full scale tests data (NA = Not Available data)

120
95J
6J
70J

Cv=
Cv=
Cv=

100
Pressure (bar)

80

60

40
No backfill
Soil backfill
20 Sea backfill
Gas decompression curve

0
0 50 100 150 200 250 300
Decompression speed (m/s)

Figure 15 – Backfill constraint influence according to the Battelle Two Curve Model

5.2 Predictive Method for DFPC for offshore gas pipeline.

The only available predictive method for DFPC on offshore gas pipelines is the “Two Curve
Model” /1, 4/. Using this method the following considerations have to be taken into account:
• in offshore pipelines where the transported gas can be “rich” it is essential to use an actual
gas decompression curve, starting from the actual gas chemical composition of gas and
transportation conditions (pressure9. and temperature)

9
The internal pressure of the pipeline is used for the Battelle Two Curve Method calculation.

30
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

• The “Two Curve Model” is based on full scale burst tests on pipes with a maximum
thickness of about 25 mm; in this context the very high thick offshore pipes (≥ 30 mm) for
deep water gas pipeline are outside of the range of validation and therefore a supplementary
analysis should be done.

5.3 Recommendation for DFPC for offshore gas pipeline.

The most recent available recommendation for the fracture arrest requirement on submarine
pipelines is contained in the DNV 2000 /16/. It gives the requirement values of Charpy V shelf
energy on transverse full size specimens to control the ductile fracture propagation. These values
are shown in Table 6 and they are valid for gas pipelines carrying essentially pure methane up to an
80% usage factor, up to a pressure of 15 MPa, and to 30 mm wall thickness. For offshore line pipe
with fracture arrest properties outside these limitations, the DNV suggests an extra analysis.

WT ≤ 30 mm OD (mm) Notes
SMYS (MPa) ≤ 610 ≤ 820 ≤ 1120 Minimum individual results to
exceed 75% of these values
245 40 40 40 Fracture arrest properties for
290 40 43 52 larger wall thicknesses and
360 50 61 75 diameters shall be subject to
415 64 77 95 agreement
(Joules; Transverse values;
450 73 89 109
Average value of three specimens)
485 82 100 124
555 103 126 155
Table 6: Charpy V-notch Impact test requirements for fracture arrest properties tested at Tmin (DNV Off-shore Standard
OS-F101)

Short background to the toughness limits in Table 6-9 of DNV-OS-F101 /49/:


1. The values were calculated using Battelle equation (short formula) assuming a design pressure of
150 bars and a usage factor of 80%.
2. No account was taken for the shift between the minimum requirement and the average of
production pipes.
3. The values are given without taking into account a reduction of toughness levels because of the
water backfill effect.
The values of Table 6-9 were proposed by EPRG in a general comment in connection with the
external hearing for DNV-OS-F101.

6. Effect of decompression behaviour of transported gas on minimum values of Charpy V


shelf energy requirement.

6.1 Background.

Fracture control design for pipelines carrying rich gas mixtures often requires significantly higher
levels of fracture propagation resistance than for gas pipelines carrying nearly pure methane; this
trend has been noted in several full scale burst tests.

31
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

This is because rich gases are likely to initially decompress more slowly and, in addition, liquid
phase formation, leading to even slower decompression rates, can occur at significantly high
pressures, relative to those required for fracture arrest10.

In other words the rich gas mixtures may show two-phase behaviour which results in a discontinuity
in the gas decompression curve; such discontinuities may be longer or shorter and results in an
interruption in any decrease in pressure which may be taking place at the crack tip. The effect of
such a discontinuity on the fracture arrest toughness requirements can be appreciated by reference
to the schematic diagram shown in Figure 16. Using the “Battelle Two Curve Model”, pipe having
fracture resistance represented by curve n. 1 would be sufficient to arrest the fracture in the case of
lean gas being transported; for a hypothetical rich gas which, from the same operating conditions,
showed two- phase decompression, use of the same material would lead to stable propagation, since
an intersection exists between the gas decompression and fracture velocity curves. To have arrest
conditions, an increase in toughness or reduction of operating hoop stress (i.e.: higher thickness) is
needed, as represented by resistance curve number 2.

120

100

80
Pressure (bar)

60

40 Resistance curve 1
Resistance curve 2
Pure methane
20 Rich gas mixture

0
0 50 100 150 200 250 300
Decompression speed (m/s)
Figure 16 – Influence of rich gas on toughness required to arrest

Figure 17 gives a quantitative indication of the effects of relatively small quantities of higher
hydrocarbons in two component mixtures on the behaviour of lean gas (chemical compositions in
Table 7); it can be seen that only with a relatively high percentage of ethane does an effect on the
decompression curve occur, while propane and butane have a stronger influence.

Finally in Figure 18 the gas decompression curves of some examples of rich gas mixtures are
shown. The relevant chemical composition is reported in Table 8.

10
For a possible definition of “rich gas” see note 8

32
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

80
Mixture 1
70 Mixture 2
Mixture 3
60
Pressure (bar)

Mixture 4
Mixture 5
50
Mixture 6
Pure methane
40

30

20
0 50 100 150 200 250 300 350 400
Crack speed (m/s)
Figure 17

Composition (%): Mix 1 Mix 2 Mix 3 Mix 4 Mix 5 Mix 6


Methane 96.0 92.0 96.0 92.0 96.0 92.0
Ethane 4.0 8.0 0.0 0.0 0.0 0.0
Propane 0.0 0.0 0.0 0.0 4.0 8.0
Butane 0.0 0.0 4.0 8.0 0.0 0.0
Other 0.0 0.0 0.0 0.0 0.0 0.0
Table 7 - Chemical composition of mixtures of Figure 17

80

70 Mixture 1

Mixture 2
60
Pressure (bar)

Pure methane

50

40

30

20
0 50 100 150 200 250 300 350 400
Crack speed (m/s)
Figure 18

33
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Composition (%): Mix 1 Mix 2 Mix 3


Methane 100.0 83.8 92.6
Ethane 0.0 7.5 3.0
Propane 0.0 1.8 0.8
Butane 0.0 0.6 0.3
Pentane 0.0 0.2 0.1
Exane 0.0 0.1 0.0
CO2 0.0 0.2 0.9
Nitrogen 0.0 5.6 2.3
NO2 0.0 0.2 0.0
Table 8 - Chemical composition of mixtures of Figure 18

6.2 Available gas decompression models for rich gas mixtures.

A number of decompression models, based on several different equations of state and on the
numerical solution of conservation and continuity equations (/17, 18, 19, 20/), have been developed
and have been found to predict accurately the decompression behaviour observed both in full scale
burst test on rich gas pipelines or in shock tube gas decompression tests. Nevertheless it is
important to note that all equations of state are empirical and should be examined to make sure the
critical constants are appropriate for the temperature, pressure and gas composition being
considered. Pipeline decompression occurs over ranges of temperature and pressure which are not
usually encountered in other sections of industry; consequently equations of state, and in particular
their parameters, should be carefully examined as to their suitability for this application outside of
their ranges of specific validations.

Despite the above cautionary remarks, it should be noted that several analytical gas decompression
models from the literature are reported; one of the most used is the code “GASDECOM” developed
as a part of a research programme conducted for PRCI /19/.

6.3 Predictive Method for DFPC for rich gas mixtures

To estimate the crack arrest / propagation conditions the only available predictive method, when a
rich gas mixture is transported and a “two phase” decompression behaviour is expected, is the
“Battelle Two Curve Model” with an “ad hoc” decompression curve of the gas of interest. This
should be calculated using a dedicated gas decompression model validated on the range of chemical
analysis of gas, temperature and pressure of interest for the specific application.

34
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

7. Alternatives to Charpy V for DFPC

Historically the required level of toughness for DFPC has been specified by correlating full scale
burst test results with Charpy V impact tests via a series of predictive formulae. These formulae, as
commentated in the previous paragraphs, have proved to be non conservative under conditions
outside those used for their development, as for example high applied hoop stress and / or where
high grade-high toughness line pipe is involved; on the other hand the semi empirical (or empirical)
approach does not provide a physical meaningful description of either the driving force for crack
advance or the material resistance.

Attention has thus been focussed on alternative methods /21/ involving a more physical description
of the Driving Force acting at the crack tip and a measurement of the toughness by laboratory tests
alternative to Charpy V specimen and more representative of the energy spent by the crack during
the propagation in full thickness conditions. In fact the Charpy V test, although it is inexpensive and
easy to carry out, has a number of additional “disadvantages” with respect to the full scale fracture
conditions. Firstly the thickness of the specimen is less than that of the pipe, secondly the ligament
is only 8 mm (so the specimen can not replicate the steady state conditions occurring during a full
scale event), and third the standard Charpy V test provides a total energy to failure and does not
distinguish between initiation and propagation phase of the fracture. In addition high strength
materials that have been heavily controlled rolled in order to achieve the required levels of strength
and fracture appearance transition temperature could exhibit a “rising shelf"11 behaviour which is
associated with a considerable incidence of “separation” or “splitting”12 in the fracture surface. This
phenomenon is further raised as the pipe wall thickness increases. The extent and position of the
rising shelf region on the temperature axis varies strongly between Charpy V and full thickness
tests. It was suggested to use other points on the Charpy V curve, such as the lowest temperature
giving 100% shear area (CV100 or CVN100), but it was difficult to determine unequivocally and it
did not give good results in terms of full scale crack arrest predictions.

In this contest a few new parameters (and consequently geometry of specimens and test methods) to
quantify the real toughness of steel pipes have been proposed in the last twenty years.
The most relevant are:
• Measurement of total absorbed energy in a DWTT test.
• Measurement of propagation absorbed energy in a DWTT test.
• Measurement of propagation absorbed energy in a Charpy V test (Leis’s correction factor).
• Measurement of specific total absorbed energy by Two Parameters Approach, Rc and Sc.
• Measurement of the Crack Tip Opening Angle, CTOA, by Two Specimen Method.

7.1 Measurement of total absorbed energy in a DWTT test.

To solve the problem of the small dimensions of the Charpy V specimen several researchers have
considered whether larger specimens with a similar geometry would be appropriate, the Battelle
Drop Weight Tear Test specimen (BDWTT) was considered the more rated candidate /22/.

11
The absorbed energy continued to increase with rising temperature above the temperature for which 100% shear
fracture was first observed.
12
The terms refer to the presence of multiple fissures lying in planes parallel to the external surfaces of the material. It
is generally associated with material production routes involving extensive rolling at temperature around or below the
transformation temperature.

35
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

The Battelle DWTT specimen was devised first as a full thickness specimen to identify accurately
the ductile / brittle transition temperature of steel (see the paragraph dedicate to brittle fracture), and
was historically the first candidate to replace the Charpy V one for fracture propagation resistance
measurement.
Standard dimension of the BDWTT specimen are shown in Figure 22, in particular the thickness is
the same as the pipe (full thickness specimen) and the notch is produced by pressing.

Whilst for traditional and conventional steels a linear relationship was found between the BDWTT
specific energy and the Charpy V specific energy via the following equation (Wilkowski, /22/):

(E A) DWTT
=3E ( A)
CharpyV
+ 63.04 (14)

where:
E = Total fracture energy (Joule)
A = Fracture area (cm2), i.e. pipe wall thickness times 7.1 cm for DWTT specimen and 0.8 cm2 for
full size Charpy V specimen.

It appears from more recent experimental results performed in the last years that the correlation
tends to deviate from the linearity for modern, high grade / high toughness carbon steel /23/. In
Figure 18 a diagram of DWTT total specific energy vs Charpy V specific energy has been reported
and the deviation from the equation (14) as the toughness increase (typical for the more recent
TMCP steels) is evident. Once again can be noted that the deviation is more evident for Charpy V
energy values higher than about 100 Joule (i.e. 125 Joule/cm2).

Actual Charpy V vs. DWTT energy

1200

1000

800
BDWTT energy (J/cm2)

600

400

CSM database
DWTT=63+3CV
200

0
0 50 100 150 200 250 300 350 400
Charpy V energy (J/cm2)

Figure 18 – Correlation between Charpy and Battelle DWTT specific energy (CSM database)

On the base of the capacity of the DWTT specimen to better represent the full scale fracture
conditions on pipe, Wilkowski et al /22, 23/ indicated that an improvement in the predictions of
arrest / propagation event in full scale burst test exists when the DWTT specific energy is used as an
alternative parameter to Charpy V as a measure of the propagation fracture resistance of the
material.

36
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Wilkowski proposes to calculate a “corrected” Charpy V energy for the material under investigation
via the equation (14) above (having as input data the DWTT energy value of the steel). Figure 19
explains the concept of the Wilkowski’s method for determining the equivalent Charpy V energy
from DWTT energy data. This corrected Charpy V value should then be used as the “effective”
Charpy V energy of the steel in the Battelle Two Curve Model explained earlier. The concept of this
method is based upon the consideration that the Battelle Two Curve method was developed and
works well for steels for which the also the equation (14) was representative of the DWTT-Charpy
V correlation; for new steels in practice the Wilkowski’s method strongly derates the actual Charpy
V of the material.

Determining equivalent Charpy V energy from DWTT energy according to Wilkowski's method

1200

1000

DWTT=63+3CV
800
BDWTT energy (J/cm2)

High grade-High
toughness steel

600

400

Equivalent
CharpyV energy Actual CharpyV
200 from DWTT energy

0
0 50 100 150 200 250 300 350 400
Charpy V energy (J/cm2)

Figure 19 – Wilkowski’s method for determining the equivalent Charpy V energy from DWTT

7.2 Measurement of propagation absorbed energy in a DWTT test.

As shown above the standard DWTT specimen has the advantage that it is a full thickness specimen
and that it has a ligament of sufficient length to include the propagation properties of line pipe steel.
However for high toughness steels, the initiation energy can be a significant fraction of the
measured impact energy, so instrumented DWTT is frequently used in research to separate the
impact energy into initiation and propagation energies. However, it is not a technique that is readily
suitable for rapid product and quality control testing in a mill (the instrumented test tends to be
expensive, requires rapid data acquisition equipment and demands an element of sophistication in
interpreting the data).
In this context many works have been focussed on notch configurations of the DWTT specimen for
which the initiation energy is a small fraction of the impact energy so that the latter becomes an
adequate approximation of the propagation energy. A number of different techniques have been
investigated to minimize the initiation energy in the DWTT so that the total measured impact
energy can represent an adequate estimate of the propagation energy. The statically precracked
DWTT (SPC-DWTT) is the promising method and it is typically the standard against which the
performance of other modified DWTT specimens is assessed and has been the subject of
considerable investigation. The preparation of this specimen has the advantage that the standard
DWTT specimen can be easily precracked in a hydraulic press; however, it does require
measurement and control of the applied load. Because of potential strain aging arising from the
37
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

plastic deformation associated with the precracking, the specimen must be tested shortly after
precracking. It has been reported that because of this requirement, some steel mills are reluctant to
use this specimen /24/.

The Chevron notched DWTT specimen /25/ represents an alternative that requires minimal
machining. There are essentially two types of Chevron notches that have been proposed. One
requires that the notch be machined with a cutter having a well define tip radius and a 60 degree
notch that complies with ASTM E399. Whereas, the Japanese HLP Research Committee proposes a
90-degree notch having a 1 mm slit that may have a flat or rounded bottom and can be cut into the
specimen in a number of different ways. While the crack initiates at the point, there is still a degree
of initiation energy required to produce a full thickness crack from the Chevron that is included in
the measured impact energy. The slit notch requires less machining and, hence, is less expensive.
/25, 26, 27/.

The preparation of DWTT with embrittled notches requires greater preparation than the statically
precracked DWTT and the Slit Chevron Notched DWTT and, perhaps, the Chevron Notched
DWTT specimens. These approaches typically require the deposition of weld metal in the notched
region. Defects in the weld can arrest the brittle fracture so that good quality control is important in
this process. Moreover, the weld material tends to soften at elevated temperatures. The TIG welding
method does not consistently embrittle the notch area for heavy-walled, high toughness steels /28/.

Fatigue cracking of a standard DWTT specimen tends to be expensive and there have been some
problems with producing a straight crack front in this specimen. The laterally compressed notch
holds some promise, but experience with this specimen is limited.

Finally, a new interesting proposal for the propagation energy measurement using a back slotted
DWTT specimen comes from the work of Pussegoda et al. /29/; the method appears promising, but
also in this case the high number of specimens necessary and a particular specimen preparation
required could limit its application in a mill.

In Figure 20 a typical experimental diagram Load vs Displacement curves as a function of different


notches is reported.

38
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Influence of notch type on the Load vs. Displacement curve of a TPB dynamic impact
1,2 test (pipe material : 279)

1,0 Pressed notch 5mm

Machined notch 10mm, r=0,1mm


Normalized impact Force (kN)

Machined notch 10mm, r=0,3mm


0,8
Chevron with equiv. notch 16mm

Pressed notch 5mm and statically


0,6 precracked

0,4

0,2

0,0
0 20 40 60 80 100 120 140
Displacement (mm)

Figure 20 - Experimental diagram Load vs Displacement curves as a function of different notch types for an high
toughness – high grade steel. The load is normalized for comparison. (CSM data)

Since these studies are the subject of ongoing research, no reliable and applicable methods to
predict the toughness required to arrest a running ductile fracture are currently available.

7.3 Measurement of propagation absorbed energy in a Charpy V test. Leis’s correction


factor.

Through the measurement of the propagation energy in instrumented Charpy V tests, Leis /30/
postulated that Charpy V based empirical models fail to correctly predict arrest toughness in
applications involving modern high toughness steels because of significant differences in their
fracture behaviour as compared to that of the much lower toughness steels used in their empirical
calibration. Basing upon his experimental results Leis defined a correction factor for the total
energy absorbed by a standard Charpy V test to be used in conjunction with the Battelle Two Curve
Model.

The idea is that for the relatively low toughness steels used to calibrate the Charpy V based models
in its early stages of development the fracture process shows little resistance to the initiation of
ductile fracture propagation. Therefore the total Charpy V absorbed energy well represents the
fracture propagation resistance of the material. On the contrary for the modern high toughness steels
a larger portion of the energy absorbed during impact is spent for the initial deformation phase and
the initiation of the crack. Leis called “apparent” the total absorbed energy in a Charpy V test, and
he identified the part of it that contributed to fracture resistance. An expression was developed
between the linear behaviour of the lower toughness steels and the apparent toughness of much
higher toughness steels. This relationship provides the means to adjust the predicted Charpy V
arrest toughness based on Battelle Two Curve Models and denoted as CVNBMI. Apparent toughness
measured by Charpy V test was denoted as CVNTest. The corrected apparent fracture resistance was
denoted as CVNArrest to reflect that this was the toughness needed to arrest ductile fracture
propagation. The concept of the method is to compare CVNArrest with CVNTest in order to predict
arrest (CVNArrest < CVNTest) or propagation (CVNArrest > CVNTest) event.

39
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

The equations developed are:

2.04
CVN Arrest = CVN BMI + 0.002 ⋅ CVN BMI − 21.18 (15)
for CVNTest ≥ 94 Joule

CVN Arrest = CVN BMI (16)


for CVNTest < 94 Joule

with Charpy V energy expressed in Joule (standard 10x10mm specimen).

Leis affirmed that below 9413 Joule any minor nonlinear effects were accounted for in the original
calibration of the Battelle Two Curve Model, and in addition below this toughness level the original
Battelle model works well.

Finally because the Battelle two Curve Model uncoupled the decompression and the fracture
behaviour, the equations above can be used also for cases involving rich gases.

Leis demonstrated that an improvement in the prediction capability of the Battelle Two Curve
Model with respect to arrest data of modern high toughness steels occurred if his correction factor is
applied, even if a number of non conservative mispredictions remain for the very high toughness
steels arrest data. Recently Leis /31/ produced new data from instrumented Charpy V tests showing
that for very high toughness steels the mathematical expression in equation (15) can underestimate
the rate of correction. In fact equation (15) indicates that the rate of correction drops as toughness
increases, whilst there are experimental evidences of the contrary /31, 32/. Therefore he suggested
that the use of equation (15) will be limited in applications to very high toughness steels.

7.4 Measurement of specific total absorbed energy by Two Parameters Approach

A question remains as to how the specific energy absorbed in a DWTT test should be used
in a quantitative way, in place of relying on correlation with full scale tests. Extensive investigation
on this subject have been carried out by researchers at British Steel /33/. In particular a large
amount of experimental laboratory data (obtained using Three Point Bend (TPB) specimens with
different ligament lengths) indicates that, fixing the specimen geometry, the specific fracture energy
is linearly dependent on test specimen ligament length and can hence be expressed as a function of
two terms Rc and Sc as follows:
Et
= R c + S c (W − a o ) (17)
A

where Et is the total energy absorbed by the test specimen, A is the initial fracture area (ligament x
initial thickness), W is the specimen width and a0 the notch depth; as a consequence, the ligament
length is equal to (W-a0). A typical example of the linear variation of the specific fracture energy
with the ligament length is depicted in Figure 5-1 of reference /34/. From the physical point of
view, Rc represents the energy necessary for the formation of the two new fracture surfaces,
measured in terms of energy per unit area, while Sc is the energy expended for plastic deformation
of the zones adjacent to the fracture surfaces, per unit volume.

13
Leis in his paper /31/ quoted 94 Joule; but in the original work /30/ he introduced 70 ft-lb (i.e. 94.9 Joule) as the
border line value. Eiber also suggests /51/ to consider 70 ft-lb and 95 J are the effective controlling values.

40
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

This approach is known as “Two Parameters Approach”; it is proposed to evaluate the toughness of
steel using the Rc and Sc parameters measured by two sets of TPB specimens (with a geometry
substantially similar to a BDWTT) with two different ligament lengths broken using a dynamic
loading condition.
The authors also assert that the energy per unit area required for fracture in a pipeline, where the
plastic deformation field is limited not by the ligament size, but by the diameter, can be expressed
as /38/:

Et
= 2 ⋅ Rc + Z ⋅ S c ⋅ D (18)
A

where the parameters RC and SC are calculated from laboratory tests, D is the pipe diameter and Z a
constant equal to the proportion of the diameter which is plastically deformed adjacent to the crack.
The limit of applicability of this method is in the relatively poor database about the experimental
dimension of the plastic zone associated with the crack in propagating test pipes, dimension needed
for the evaluation of the constant Z in Equation (18). Approximate relationship for Z is Z≈0.3D, but
it was derived for 56” diameter pipe only. The factor 2 on Rc is derived from the observation that
for pure tension Rc is approximately twice that in bend tests.

7.5 Measurement of the Crack Tip Opening Angle, CTOA, by Two Specimen Method.

The use of CTOA as the toughness parameter derives from post-yielding fracture mechanics
concept. In practice this approach is based upon the balance between the capacity of the steel to
resist the crack propagation (expressed by the CTOA critical value of the material, named CTOAc)
and the driving force applied by the external loads (expressed by the CTOA applied, named CTOAa
or CTOAmax) /34/.

The CTOA based fracture criterion states that steady state propagation of a ductile fracture is
impossible if
(CTOA) max ≤ (CTOA) c (19)

In practice CTOAmax can be calculated with a suitable Finite Element Code, as for example the
PFRAC code developed by SwRI in collaboration with CSM /34/ or the CSM’s PICPRO finite
element code recently developed /35/.
For the measurement of the CTOA critical value of the material CTOAc, a suitable lab
testing procedure has to be used.
In the above mentioned AGA research project performed for PRCI /34/, the way to measure the
Crack Tip Opening Angle was investigated directly by a single BDWTT specimen, both under
static and dynamic test conditions, starting from measurements of both the angle at the crack tip
during the test (by means of photographs) and the propagation energy Ep involved in the fracture
process.

This latter CTOA value, called CTOAEp, is calculated by the following formula

 r*Ep 
CTOAEp = 2arctg   (20)
 K σ flow B (W − a o ) 2


41
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

with:
• B = specimen thickness
• K = constant depending on specimen geometry (0.35 for SENB)
• σflow = material flow stress, equal to 0.65*(yield strength + tensile strength)
• W = specimen width
• a0 = length of crack (i.e. notch depth)
• W-a0 = specimen ligament
• r* = dimensionless constant during the propagation phase, which depends on specimen
geometry and the external load, equal to 0.45 for SENB specimen.

From these and other similar tests (inside other projects carried out by CSM for ECSC /36/) it was
verified that, if the ligament is long enough, the Crack Tip Opening Angle from single specimen
CTOAEp is substantially constant during the test but its value is influenced by the ligament length.
As the ligament increases, the CTOA decreases tending to a limit value representing the CTOA
existing on pipe during the fracture propagation, where the ligament is longer than in a test
specimen (CTOA measurements carried out by CSM on pipes after a propagation full scale burst
test, by means of the elaboration of strain gauges data, also confirmed this). So to have a good
CTOA value by a single specimen test the ligament has to be long enough (greater than one
hundred centimeters). This behavior is depicted in Figure 5-2 of Reference /34/ and here reported as
Figure 21.

Figure 21 (Ref. /34/)

42
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

From the laboratory point of view, this highlighted the necessity to develop a simpler test technique
avoiding a specimen with a ligament so long; the Two-Specimen CTOA Test (TSC Test)
Methodology was the answer. With this test CTOA is not measured directly by a single specimen,
but it is evaluated by means of the measure of the Sc parameter based on the Two Parameter
Approach above described. The way followed at that time to develop the Two-Specimen CTOA
Test was essentially the following.
The total fracture energy Et can be expressed as the sum of two terms.
Et = Ei + E p (21)

where Ei represents the initiation energy and Ep is the propagation energy.


According to this approach, Ei is assumed to be independent of the ligament length, while Ep can be
expressed as (rewriting the (20)):
Kσ o B  CTOA 
 (W − a o )
2
Ep = tg  (22)
r*  2 

hence the (21) becomes:


E t E i Kσ o  CTOAC 
= + tg  (W − a o ) (23)
A A r*  2 

Comparing the (17) and the (23) and assuming Ei constant with the ligament length, we have:
Ei
RC =
A
(24)
Kσ o  CTOAC 
Sc = tg  
r*  2 

 CTOA  CTOA
assuming tg  ≅ we finally have:
 2  2
2S c r *
CTOAC = (25)
σoK
that is the formula currently used in the Two Specimen CTOA Tests.
The validity of the assumption made concerning the independence of initiation energy from the
ligament length was verified at the time of /37/ using steels with Charpy V energy up to about 200
Joule and grade in the range X60-X80. For higher grade steels this trend has to be confirmed, and it
is object of ongoing research programs.

Finally, correlations of CTOAc with CVN and BDWTT total energies were also investigated in the
above projects for AGA in order to simplify the Two-Specimen CTOA Test Methodology developed,
but no reliable correlation was identified, as expected.

7.6 Considerations about the alternatives to Charpy V for DFPC

As evident a number of alternatives to the “conventional” Charpy V test have been explored and
developed.

43
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

It has to be noted, however, that after several years of activity a simple widely validated DWTT-
based predictive method still does not exist.
The CTOA approach is considered one of the most promising alternative to Charpy V specimen, but
it implies utility for the numerical solution of the CTOA applied (even if using modern PC the run
time is strongly reduced with respect to the past and it is now in the range of some hours). In
addition the experimental aspects are not easy to use on a production basis in steel and pipe mills,
and also in this case the high number of specimens necessary and a particular specimen preparation
required could limit its application in a mill.

44
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Part II: Methods preventing the brittle fracture propagation in gas pipelines.

1. Background.

The driving force for brittle fracture propagation is given by the elastic energy stored in the pipeline
wall, as a function of the applied hoop stress. Since the brittle fracture could be arrested only at very
low stress levels, the only acceptable approach used by the gas companies for controlling brittle
fracture propagation is based on the specification of a minimum toughness requirement to ensure
that brittle fracture does not occur. In practice brittle fracture propagation is prevented by specifying
a minimum toughness to ensure that the linepipe steel is on the upper shelf of the toughness
transition curve at the minimum operating temperature; in other words the steel should be above its
Fracture Propagation Transition Temperature FPTT, that is the temperature that assures that the
steel will propagate in a ductile manner. The FPTT is a property of the steel and strongly varies
with the manufacturing processing applied to the steel /1/.

Burst tests carried out on a short pipe length, filled approximately 90 % with liquid and then
pressurised with air, proved able to reproduce the full scale ductile to brittle transition behaviour in
a correct way. The test, called West Jefferson from the location of the Battelle facilities where the
first tests were executed, is made at controlled temperature and the amount of ductile and brittle
fracture appearance on the fracture surface is measured. However, comparison between West
Jefferson test results and the Charpy V were not very satisfactory: the transition temperature of the
Charpy V specimen was at lower temperatures that the actual pipe behaviour.

Different laboratory test methods were tried in order to overcome the problem and the Battelle
DWTT emerged as the one with the best correlation with full scale behaviour /39/14. The DWTT
specimen is 76 mm high by full pipe thickness by 254 mm span length, and has a pressed notch 5
mm deep, as shown in Figure 22. The specimen is broken completely in one impact and the percent
of shear area is evaluated in the central part of ligament, as shown in Figure 22. The fracture
propagation transition temperature of pipe is normally taken to correspond to the temperature at
which a DWTT specimen exhibits at least an 85% shear area fracture, also known as “Battelle 85%
SA criterion” /1/. This requirement ensures that the linepipe steel cannot sustain brittle fracture
propagation in full-scale conditions.

14
Also the Chevron notched specimen /25/, together with the statically precracked DWTT one /24/, has been identified
as a good candidate for the evaluation of the transition temperature. The Chevron notch eliminates excess initiation
energy associated with higher toughness line pipe steels, which can result in invalid specimen as defined by API RP
5L3 /44/. For this reason the API RP 5L3 /44/ suggests to use DWTT specimen for lower toughness line pipe steels and
to use the Chevron notch specimen for higher toughness ones.

45
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Figure 22 – Drop Weight Tear Test specimen dimensions

This approach has been confirmed in recent years /40, 41, 42/ on large diameter pipes with
relatively high tensile properties (steel grade from API X65 to X80) and high thickness (26 mm to
38.1 mm).
The 85% shear area Battelle DWTT criteria was recently introduced in the EN 10208-2 standard
/45/ and since then has been widely used throughout the world. Therefore, the temperature
corresponding to ductile fracture behaviour of a pipeline is fixed to be the one at which the DWTT
percent shear area is 85% of the evaluated fracture surface.

In Figure 23 the comparison of PN-DWTT and West Jefferson test results presented in /40, 41, 42/
at West Jefferson test temperature for wall thickness above 19 mm is shown, where:
= Grade X80, D=56”, wt=26 mm (D/wt =54.7);
= Grade X70, D=56”, wt=30.5 mm (D/wt =46.6);
= Grade X65, D=36”, wt=27.5 (D/wt =33.3);
∆ = Grade X65, D=36”, wt=38.1 (D/wt =24.0).

Figure 23, Ref. /40, 41, 42/

46
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

Finally to complete this review about the brittle fracture propagation it is useful to mention the
“brittle fracture arrest” concept based upon the energy balance between the elastic energy stored in
the linepipe wall as a function of the applied hoop stress and the line pipe material toughness; this
approach has been investigated in the past but has not been used practically.
The main available work concerning this matter is the historical one of the Battelle /4/ where the
energy balance approach, linepipe elastic energy versus Charpy V energy measured at the same
Charpy V percent area as the observed full-scale percent shear area, was proposed and some
experimental results concerning pipe steels of that period were shown (Figure 24).

The hypothesis is that the strain energy in the pipe is equal to the energy required to create the
fracture.

The proposed equation for the brittle fracture arrest was:

σ 2πRAc (26)
CVS =
E
where:
- Cvs = Charpy V energy at full-scale percent shear area (Joule);
- σ = hoop stress applied (MPa);
- R = pipe radius (mm);
- Ac = area of fracture surface of a Charpy V-notch specimen (mm2);
- E = Young modulus

In practice the resistance to fracture is taken as the Charpy V energy at the pipe’s expected percent
shear expressed as energy per unit area of the Charpy V specimen fracture surface (Joule/mm2). In
practice the data plotted on Figure 24 utilize the full scale percent shear area of the fractured surface
and the measured Charpy V energy at that same percent shear area on a Charpy V specimen.

Figure 24, Ref. /4/

47
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

2. On the influence of high wall thickness.

To reduce the energy (and max load) needed to break DWTT specimen when the pipe wall
thickness is very high, the API recommendation RP5L3 requires that pipes with a wall thickness up
to 19.1 mm, should be tested using full thickness specimens, over which it is allowed to reduce the
thickness (up to 19 mm), with a corresponding reduction of test temperature as shown in Table 9.

Specified wall thickness Test temperature reduction


(mm) (°C)
19.0< t ≤ 22.2 6
22.2 < t ≤ 28.6 11
28.6 < t ≤ 39.7 17
Table 9. Test temperature reduction according to API RP 5L3 when testing reduced thickness specimens of 19.0 mm
wall thickness.

A recent work /40/ carried out on API X65 thick wall line pipes (up to 38.1 mm) has shown that
there is a reasonable agreement between the transition temperature obtained by full thickness
DWTT specimens and calculated transition temperature from 19 mm thickness specimens,
according to API RP 5L3. The API relations, reported in Table 9, appear good enough to be used
when the definition of the DWTT transition curve on full thickness specimens is limited by the
testing capacity15.

Finally it is important to note that the API relations are also limited to about 40 mm pipe wall
thickness, over which no indication is given.

15
In round robin Drop Weight Tests /47/ the ratio energy capacity of machine / energy absorbed during the test was
found to be satisfactory at a minimum level of 1.25.

48
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

References

1. Eiber, R. J., Bubenik, T. A. “Fracture Propagation Control Plan Methodology“ 9th PRCI/EPRG
Joint Technical Meeting, Houston, Texas, May 1993, paper 20.

2. Wilkowski, G.M, Scott, P.M. “Design and Optimization of Mechanical Crack Arrestors for
Pipelines” NG-18 / EPRG 5th Joint Technical Meeting, San Francisco, 1983, paper 20.

3. Mannucci, G., Landi, M., Demofonti, G. “State of the Art on Mechanical Crack Arrest Devices
for Large Diameter Gas Pipeline” 12th EPRG/PRCI Biennial Joint Technical Meeting on
Pipeline Research, Groningen, The Netherlands, May 1999, paper 14.

4. Maxey, W.A. “Fracture Initiation, Propagation and Arrest” 5th Symposium on Line Pipe
Research, Houston, Texas, November 1974, paper J.

5. Johnson, D.M., Horner, N., Carlson, L. Eiber, R. “Full Scale Validation of the Fracture Control
of a Pipeline Designed to Transport Rich Natural Gas” 3rd International Pipeline Technology
Conference, Brugge, Belgium, May 2000.

6. Mannucci, G., Demofonti, G., Barsanti, L., Spinelli, C.M. “High Pressure Long Distance
Transmission Gas Lines: Structural Integrity of High Strength Steel Grade Pipeline” 3rd
International Pipeline Technology Conference, Brugge, Belgium, May 2000.

7. AISI Technical Report “Running Shear Fractures in Line Pipe” Subcommittee of Large
Diameter Line Pipe Producers, September 1974.

8. Fearnehough, G.D., Jones, D.G. “Toughness Specification for Shear Fracture Arrest in
Pipelines” International Conference on Analytical and Experimental Fracture Mechanics,
Rome, Italy, June 1980.

9. Bonomo, F. et al. “A Survey and Tentative Revision of Ductile Arrest Criteria in Pipelines for
Gas Transmission” International Conference on Analytical and Experimental Fracture
Mechanics, Rome, Italy, June 1980.

10. Maxey, W.A. “Fracture Propagation Behaviour for Offshore Pipelines based on Underwater
Burst Tests” NG-18/EPRG 6th Biennial Joint Technical Meeting, Camogli, Italy, September
1985, Paper 11.

11. Vogt, Bramante, Jones, Koch, Hugler, Pero, Re “EPRG Report on Toughness for Crack Arrest
in Gas Transmission Pipeline” 3R International 22, 1983.

12. Re, G., Pistone, V., Demofonti, G., and Jones, D.G. “EPRG Recommendation for Crack Arrest
Toughness for High Strength Line Pipe Steels” 3R International 34, 1995.

13. Dawson, S.J., Pistone, V. “Probabilistic Evaluation of the Safety Embodied in the EPRG
Recommendations for Shear Fracture Arrest Toughness” 11th Biennial Joint Technical
Meeting, Arlington, Virginia, April 1997, Paper No. 27.

49
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

14. Pistone, V., Mannucci, G. “Fracture Arrest Criteria for Spiral Welded Pipes” 3rd International
Pipeline Technology Conference, Brugge, Belgium, May 2000.

15. Demofonti, G., Hadley, I. "The Effect of External Constraint on Ductile Fracture Propagation
in Gas Pipelines" International Conference on Pipeline Reliability, Calgary, Canada, June 1992.

16. DNV offshore standard OS - F101, Submarine Pipeline Systems - 2000 , Section 6, D200.

17. Groves, T.K., Bishnoi, P.R., Wallbridge, J.M.E.: Can. J. Chem. Eng., 1978, 56, 664.

18. Maxey, W.A. “Gas Expansion Studies” American Gas Association, Catalogne No. L51435,
1983.

19. Eiber, R., Bubenik, T.A., Maxey, W.A. “Fracture Control Technology for Gas Pipelines” AGA
Catalog No. L51691, 1993.

20. Demofonti, G., D’Anna, R. “Criteri per la valutazione della tenacità necessaria per l’arresto
della frattura duttile su di un gasdotto in presenza di un gas bifasico in condizioni on-shore ed
off-shore (Methods for determining the toughness required to arrest a ductile fracture
propagation in pipelines carrying gas with two-phase decompression behaviour both in on-
shore and off-shore conditions)” Internal CSM Report No. 7624R, Roma, October 1992.

21. Rothwell, A.B. “Fracture Propagation Control for Gas Pipelines – Past, Present and Future” 3rd
International Pipeline Technology Conference, Brugge, Belgium, May 2000.

22. Wilkowski, G. “Fracture Propagation Toughness Measurements” paper K, 6th Symposium on


line Pipe research, American Gas Association, Arlington, VA, Catalogue No. L30175, 1979.

23. Wilkowski, G., Wang, Y.Y, Rudland, D. “Recent Efforts on Characterizing Propagating
Ductile Fracture Resistance of Linepipe Steels” 3rd International Pipeline Technology
Conference, Brugge, Belgium, May 2000.

24. Wilkowski, G.W. “Studies on the Precracked DWTT Specimen to Predict Ductile Fracture
Arrest” IV Seminar AGA-EPRG, September 1981

25. Maxey W.A and Barnes C.R. “The Chevron Notched Drop-Weight Test Specimen” NG-18
Report No. 190, Sept. 1990.

26. Vogt G. and Junker G. “Comparison of test results from Chevron and Weld Notch DWTT
specimens with those from Pressed Notch specimens” 3R International, 33, October 1994, pp.
571 – 575.

27. Kubo, T. et al. “Quality Evaluation of Line Pipes by the Chevron Notched Drop Weight Tear
Test “ ISIJ Institute, Vol.35, (4), pp. 426 - 432, 1995.

28. Wilkowski, G.M., et al. “Use of a Brittle Notch DWTT Specimen to Predict Fracture
Characteristics of line Pipe Steels” Proc. of Energy Technology Conf., ASME 1977, paper
n.77.

50
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

29. Pussegoda, L. N., et al “Interim Approach to Ductile Fracture Arrest Toughness - Progress”
Proc, of EPRG/PRCI 12th Joint Biennial Technical Meeting on Pipeline Research, Groningen,
The Netherlands, May 1999, paper 15.

30. Leis, B.N. “Relationship Between Apparent Charpy Vee-Notch Toughness and the
Corresponding Dynamic Crack-Propagation Resistance” Battelle Report to R.J. Eiber
Consultant Inc., 1997.

31. Leis, B.N. “Predicting Fracture Arrest Based on a Relationship Between Charpy Vee-Notch
Toughness ad Dynamic Crack-Propagation Resistance” 3rd International Pipeline Technology
Conference, Brugge, Belgium, May 2000.

32. Mannucci, G., Demofonti, G., Harris, D., Barsanti, L., Hillenbrand, H.G. “Fracture Properties
of API X100 Gas Pipeline Steels” 13th Biennial Joint Technical Meeting on Pipeline Research,
New Orleans, Louisiana, April-May 2001.

33. Priest, A.H. et al, International Journal of Fracture, 17, pp. 277-299, 1981

34. Kanninen, M.F., Leung, C.P., O’Donoghue, P.E., Morrow, T.B., Popelar, C.F., Buzzichelli, G.,
Demofonti, G., Rizzi, L., Venzi, S. and Cinquetti, S. “The Development of A Ductile Pipe
Fracture Model” Joint Final Report by SwRI, CSM and Snam to the Pipeline Research
Committee, American Gas Association, August 1991.

35. Berardo, G., Salvini, P., Mannucci, G., Demofonti, G. “On Longitudinal Propagation of a
Ductile Fracture in a Gas Line Pipe: Numerical and Experimental Analysis” International
Pipeline Conference IPC2000, Calgary, October 2000.

36. Buzzichelli, G., Demofonti, G., Maresca, A. “Prediction of Ductile Fracture Propagation
Behaviour of Gas Pipelines by Laboratory Tests” Final Report for European Community of
Steel and Carbon, ECSC Agreement n. 7210-KE/409, 1985.

37. Kanninen, M.F., Morrow, T.B., Grant, T.S., Demofonti, G. “The Development and Validation
of a Ductile Fracture Analysis Model” Final Report, Pipeline Research Committee, American
Gas Association, May 1994.

38. Priest, A.H., Holmes, B. “The Characterisation of Crack Arrest Toughness in Gas Transmission
Pipelines in terms of Shear Fracture Propagation Energy”, Final Report ECSC Agreement No.
7210.KE/808(F5.5/82), December 1985.

39. Eiber, R. J. “Correlation of Full Scale Tests with Laboratory Tests”, 3rd symposium on line pipe
Research, American Gas Association, Catalogue N°. L30000, November 1965, pages 83 - 118.

40. G. Demofonti et al. “Transition Temperature Determination for thick Wall Line Pipe” Proc. Int.
Pipeline Conference, 1998, ASME, pp. 699 - 704.

41. G. Mannucci et al. “Structural Integrity of API 5L X70 - X80 steel grade Pipeline for High
Pressure Transition Gas Lines” Proc. of Gas Research Conference, 1998, GRI Ed., pp. 459 -
469.

42. Demofonti, G. et al. “Transition Temperature Determination for Thick Wall Line Pipes”, 3R
international, March 2000 (39), pp.199 - 204.

51
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

43. API 5L, “Specification for Line Pipe”, 1995 - 04.

44. API RP 5L3, “Recommended Practice for Conducting Drop-Weight Tear Test on Line Pipe”
1996 - 05.

45. EN 10208-2, “Steel pipes for pipelines for combustible fluids – Technical delivery conditions –
Part 2: Pipes of requirement class B”, June 1996.

46. Bob Eiber’s private communication.

47. . Fearnehough G. D, et al. “A Round Robin Test Programme within EPRG” NG-18/EPRG 4th
JTM 1981, paper 1.

48. Rothwell, A.B., “The Application of the Battelle “Short Formula” to the Determination of
Ductile Fracture Arrest Toughness in Gas Pipelines” 2000 International Pipeline Conference
(IPC2000), Calgary, Alberta, Canada, October 1-5 2000, pp. 233-238.

49. Agnar Karlsen, Det Norske Veritas AS, private communication.

50. Maxey, W.A., Eiber, R.J “Fractures in Pipeline – Main Influencing Factors” Fractures in Gas
Pipelines, Proceedings of an International Seminar held at the I.P. Bardin Institute
(Cniichermet), Moscow, USSR, March 1984.

51. Bob Eiber’s private communication.

52
CSM Final Report No. 10423R “Methods governing and preventing long
running ductile and brittle propagating fast fracture in gas transmission pipelines”

PROGETTO N. Progressivo per commessa

COMMESSA N. avanzamento
N.
RAPPORTO N. 10423R Finale x

RELATORI:

G. Demofonti

G. Mannucci

COLLABORATORI ALLA SPERIMENTAZIONE:

CAPO COMMESSA:

G. Mannucci

CAPO FUNZIONE: CAPO DIPARTIMENTO:

G. Mannucci M. Pontremoli

PRECEDENTI RIFERIMENTI SULLA COMMESSA:

Rapporto N. Data Commessa N.

Centro Sviluppo Materiali S.p.A.


Via di Castel Romano 100 - 00129 Rome, Italy
Tel. +39-6-50551, Fax +39-6-5055452

53

You might also like