Professional Documents
Culture Documents
C Course Notes Fitness For Purpose
C Course Notes Fitness For Purpose
C Course Notes Fitness For Purpose
Phil Hopkins
Penspen Ltd., UK
(p.hopkins@penspen.com)
Contents
1. INTRODUCTION.....................................................................................................2
7. CONCLUSIONS......................................................................................................6
8. REFERENCES.....................................................................................................10
These course notes explain how to conduct ‘fitness-for-purpose’ assessments (sometimes called
‘Engineering Critical Assessments (ECA)’) of defects in a transmission pipeline, and present a summary
of the methods available. The methods cover a wide range of defects, and can be applied to both onshore
and offshore oil and gas pipelines. The following types of defect are discussed:
Gouges, Dents, Dents and Gouges
Corrosion
Girth Weld Defects
Stress Corrosion Cracking
Material Defects
Construction Defects
There are millions of kilometres of transmission pipelines around the world. The oil and gas transmission
system in Western Europe alone is over 150,000 km in length. A pipeline, and all its associated plant,
must be operated safely and efficiently. There are four key issues ('risks') in the operation of these
systems:
1. Safety - the system must pose an acceptably low risk to the surrounding population,
2. Security of Supply - the system must deliver its product in a continuous manner, to satisfy the
owners of the product (the 'shippers') and the shippers' customers (the 'end users'), and have low
risk of supply failure,
3. Cost Effectiveness - the system must deliver the product at an attractive market price, and minimise
risk of losing business,
4. Regulations - the operation of the system must satisfy all legislation and regulations, and this
usually requires the operator to assess, and mitigate, pipeline-associated risks.
2.1 General
Pipelines are the safest form of transporting energy[1]. A review[2] on failure rates in pipelines in Western
Europe and North America concluded:
T ABLE 1 - Failure rates in transmission pipelines [2]
The risk resulting from pipeline failures are very low when compared with the risks from other forms of
failures/disasters[3]. This can be illustrated by considering the 450,000 mile long North American gas
transmission system where there have been 30 fatalities, and 106 casualties between 1987 and 1993.
These casualty figures are very low compared with other risks to the general public in the USA (Table
2).
T ABLE 2 - Relative risks of fatality in USA In 1992 [1]
Pipeline failures can also cause environmental damage. Figure 1 shows the total amount of oil spills into
the marine environment between 1990 and 1999. Nearly 1,000 million gallons (3 million tonnes) were
spilled in this period, but over 50% was due to tanker spills: pipelines accounted for only 6.4% of total
spills. Sub-sea pipelines can and do fail (see Figure 2, later), but they tend to spill less product that a sea
tanker. The tankers at sea today are massive: the Exxon Valdez that spilled oil in 1989 released 11
million gallons (257,000 barrels or 38,800 metric tonnes) of its 53 million gallon cargo. 11 million
gallons would fill 125 Olympic swimming pools.
50
40
30
20
10
(68)
Figure 1. Amount of Oil Spilled in Marine Waters Worldwide (1990-99)
Fitness-for-purpose methods for assessing defects in pipelines use fracture mechanics as their technical
basis[4]. Therefore, they are applied mechanics methods.
However, because they require an engineer to work outside existing codes, and because the defect may
be large and a potential safety hazard, they must be used in a structured and systematic manner, by
suitably qualified engineers.
When a defect is assessed using fitness-for-purpose methods, it should not decrease the overall integrity
of the pipeline. As the name suggests, a fitness-for-purpose assessment will always ensure that the
pipeline is left in a fit and safe state.
The above table shows various causes of failure and failure rates. There are more detailed breakdowns
of failure data: Figure 2 shows a detailed breakdown of failures in offshore pipelines in the UK sector of
the North Sea. This is taken from the Pipeline and Riser Loss of Containment 2001 Report, Energy
Institute, UK.
The differing failure rates are probably due to differing definitions of failure (see Reference 5 and Table
1). The variety of failure causes is not surprising as pipelines operate in a variety of hostile
environments. Eastern Europe has problems with construction and material defects, and also with girth
weld defects. However, problems with girth weld defects are not confined to Eastern Europe; one
Western European operator has reported the following defects in girth welds[4]:
T ABLE 4 - Older Girth Welds in a Transmission System [4]
Clearly, older girth welds may contain unacceptable defects. These welds are unacceptable to
workmanship standards, but they are not necessarily unfit-for-purpose. Clearly, there is a need for
detailed fracture mechanics analyses to determine the scale of these type of problems, and avoid
unnecessary repairs (see Section 5.3.9, later).
Any engineer with a potential defect problem should structure their assessment as follows:
5.1 Introduction
The first consideration when applying fitness-for-purpose methods to any defect detected or expected in
a transportation system is the predominant micromechanism of fracture.
Transmission pipelines usually have relatively thin wall thicknesses (most are less than 25 mm), and the
steels have been produced for many years to recognised international standards (e.g. API 5L). This
ensures that our line pipe is ductile, and will not exhibit low toughness or brittle behaviour. This means
that the line pipe we use in pipelines is ductile.
Transmission pipelines are usually buried or subsea, and therefore they are not generally subjected to
low temperatures. Accordingly, the failure of part wall defects (such as corrosion pits) in (parent)
transmission pipeline material, is by tensile necking of the ligament remaining below the defect. Sharper
defects in parent material, such as scratches or gouges, fail by progressive ductile crack growth
(microvoid coalescence).
Defects in older pipeline steels (e.g. those produced in the 1950s), and the thicker walled pipelines, will
generally show ductile fracture initiation (although they may eventually fail by dynamic brittle crack
propagation). Consequently, defects in transmission pipelines are generally assessed using solutions
based on limit loads, or plastic collapse[6-8].
DATA STAGES
STAGE 1
DEFECT QUALITATIVE - ACCEPT
REJECT
REJECT
STAGE 4b ACCEPT
REJECT
ACCEPT
REJECT
hoop
axial
A part wall defect in ductile line pipe under internal pressure fails as follows:
1
Sometimes called a ‘break’.
d
t
d. The through Wall Defect causes a Leak e. The Through Wall Defect causes a Rupture
if the defect is Short, or if the pressure is Low. if the defect is Long, or if the pressure is High.
Pipe Dimensions
The simplest and, in general, the most conservative formula for the range of transmission pipeline D/t
ratios is given by using u in the simple Barlow equation:
2t u
Pf (1)
D
where:
Pf = failure pressure (MPa)
t = pipewall thickness (mm)
D = outside diameter of pipe (mm)
u = ultimate tensile strength (MPa)
K 2 1
Pf for K > 1.5 (3)
K 2 1 u
where:
Ro
K (4)
Ri
Ro = outside radius of pipe D/2 (mm)
Ri = inside radius of pipe, Ro-t (mm)
Denting
Even
if
Hardened
layer Pipe
the
gouge Cracking
is not
associated with a dent, there may be a work hardened layer at the base of the gouge which may reduce
the local ductility and it may contain cracking.
It is good practice to remove any surface hardening by grinding, although special procedures are
required for dressing damage on a live pipeline. Following dressing, the gouge can be assessed using its
dressed dimensions.
2
See Section 5.3.10.2 for methods for assessing the fatigue life of these type of defects.
Defect Dimensions
A d
1 1
f Ao f t
or (5)
A 1 d 1
1 M 1 M
Ao t
2
2c
M 1 0.26 (6)
Rt
where:
t
Pf f
R
f = failure stress (MPa)
M = bulging factor (Eq. 6 is an example – there are many other formulations)
= flow stress (MPa)
d = maximum or average depth of part wall defect (mm)
2c = defect axial length (mm)
A = area of metal loss in the axial plane through the wall thickness (mm2)
Ao = original area (d2c)
R = outside radius of pipe (mm)
Equation 6 can also be used to determine the failure of a through wall defect: this failure of a through
wall defect will be a ‘rupture’ as the through wall defect is already ‘leaking’ (see Figure 3). The stress to
fail (rupture) a through wall defect is
Assessments of part wall defects are more accurate if the area ratio (A/Ao) is used. Using the maximum
defect depth will be conservative (using the average defect depth is equivalent to using the area ratio).
Figure 5 gives a simple representation of Equations 5 and 6a.
Gouges can be assessed using the above equations, providing your pipeline has a toughness >20J (19).
Note that a gouge needs to be checked for possible fatigue crack growth in some pipelines (e.g. some
liquid lines).
Allowance (e.g. adding 0.5mm to defect depth) for the hard layer or sub-surface cracking is advisable, if
they are to be left in the pipeline, but you should ensure there is no risk of environmental cracking, and
no residual denting, and no problems from cyclic loading.
Note that the assessment of gouges in pipeline is often not allowed by codes and regulations [64-66] due to
concern over the difficulty of detecting associated denting and cracking. Gouges in these
codes/regulations require repair.
Defect Dimensions
z 1
21 sin
c d
1
R t
where:
z = axial failure stress (MPa)
= flow strength (MPa)
d 2c
M 1 0.40
g
Rt
en
t
t
Flow strength = 1.15y
S
1
e
Ydr
t
0.8
l
1 - (d/t) = 0.6
e
Ss
s
/i
0.6
0.5
uer
t
0.4
a
Fr
0.4
0.3
l
i
0.2
0.2
0.1
0
0.05
0 1 2 3 4
2c/(Rt)^0.5
5 6 7
1.2
8
f This boundary is not sensitive to pressurising medium
2c or l
M 1
t
h
1
S
te
rng
t
0.8
sYe
ild
RUPTURE
/
0.6
S
te
rs
LEAK
e
0.4
a
Fl
iur
0.2
2c
0
0 1 2 3 4 5 6 7 8
2c/(Rt)^0.5
Figure 5. Equations 5 (left) and 6a (right) as Universal Graphs
N = fatigue life
D d
Defect Dimensions
f 2 1.5E 1.8 D0 R D0
2
ln(C v ) K 1
cos exp 2
1
Y1 1 Y2 10.2 exp (8)
Ad 2R t 2R K2
where:
K1 1.9
K 2 0.57
(K1 and K2 are non-linear regression parameters)
f = hoop stress at failure (lbf/in2)
= plastic collapse stress of infinitely long gouge (lbf/in2)
A = fracture Area of Charpy (0.083 in2 for a 2/3 Charpy specimen)
E = Young’s Modulus (30,000,000 lbf/in2)
Cv = 2/3 Charpy toughness (ftlbf)
d= maximum or average depth of part wall defect (in)
D0 = dent depth measured at zero pressure (in) (See Reference 17)
t = pipe wall thickness (in)
R = outside radius of pipe (in)
This failure criterion for a dent containing a metal loss defect does not give a lower bound failure stress.
It is a mean predictive model. Additionally, the model is semi-empirical and therefore limited by the
bounds of the original test data[16-22].
The model is well known to give large scatter in its predictions[63], and it is considered more of a research
tool than a practical model, hence, its use in the field is no recommended without expert help.
Finally, an engineer should always think carefully of the consequences of ‘getting things wrong’. If your
damage is in a pipeline in a ‘high consequence’ area, you should inspect the damage closely before
assessment, and include appropriate safety factors in your assessment.
2d
1
3t
f (9)
2d 1
1
3 t M
The flow strength ( ) is defined by 1.1xSMYS (specified minimum yield strength). In more recent
years a flow strength of 1.15xSMYS has been used to accommodate the added ductility in modern steels.
The parabolic shape of the projected area is used as an approximation to the actual defect, and a
modified bulging factor is used:
2
2c
M 1 0.8 (10)
Dt
It is stated in the code that the above equations should only be applied to corrosion defects which have a
maximum depth greater than 10% of the nominal wall thickness, and less than 80% of the nominal wall
thickness. Furthermore, the relative longitudinal extent should satisfy the following equation:
2
2c
M 1 0.8 4.12 (11)
Dt
3
Usually, the most difficult data to obtain when assessing corrosion, is the expected corrosion growth rate. This
is important, because most assessments of corrosion are based on intelligent pig data, where the defect must be
assessed over its ‘whole life’, and its size at the end of the pipeline’s design life needs to be used in the
calculations.
B M 2 1
The above equation limits the use of the parabolic shape formulation because when M is greater than
4.12 (i.e. long corrosion), the approximation of a parabolic shape is no longer adequate. Instead a
rectangular shape is used. Accordingly the failure equation is replaced by the following equation:
d
f 1 (12)
t
d
1 0.85
f t
(9a)
d 1
1 0.85
t M
where
= SMYS + 68.95 MPa (10 ksi)
2 4 2
2c 2c 2c
M= 1 0.6275 0.003375 for 50 (13)
Dt Dt Dt
2 2
2c 2c
M = 0.032 3.3 for 50 (14)
Dt Dt
Both the original B31G criterion and the modified B31G criterion define simple approximations to the
exact corroded area, based on the maximum length and the maximum depth of the defect. In addition, the
modified B31G criterion describes a method based on the effective area and effective length of the defect,
incorporated into software known as RSTRENG. The RSTRENG method, based on an iterative
algorithm, was developed to allow the actual profile of the corrosion defect to be considered, thereby
giving more accurate predictions of the failure pressure of the corrosion defect. The modified B31G
method, including RSTRENG, has been validated against 86 burst tests on pipe containing real
corrosion defects.
d
1 t
f (5)
1 d 1
t M
where
2
l
M 1 0 .31 (15)
Dt
= 0.9UTS
The method for taking into account the actual profile of the corrosion defect is, like RSTERNG, an
iterative procedure, but based on the principle of considering the actual profile as a collection of ‘pits’
within ‘patches’. The corrosion defect is divided into a number of increments, based on depth, and
1 0.6275 0.003375
B31.G (9a) 69MPa [0.85(d/t)] Dt Dt
The main methods (ASMEB31G, modified B31G and DNV are given graphically in Figure 6.
d 1
Original
1
Battelle
0.6
d 1 (‘NG18’)
1
0.4
t M 0.2
Equation
0
1.0 0 1 2 3 4 5 6 7 8
2c/(Rt)^0.5
0.9
D e fe c t D e p th /W a ll T h ic k n e s s
0.8
0.7
1.0
0.6
0.9
0.5
0.8
D e fe c t D e p th /W a ll T h ic k n e s s
0.4 REJECT 1.0
0.7
0.3 0.9
0.6
0.2 0.8
ACCEPT 0.5
D e fe c t D e p th /W a ll T h ic k n e s s
0.1 0.7
0.4
0.0
0.6
0 1 2 3 4 5 6 7 0.3
8
Defect Length, 2c/(Rt)^0.5 0.2
0.5
0.1 0.4
0.2
2c/(Rt)^0.5
0.1
DNV
Figure 6. Schematics of the Main Corrosion Assessment Methods
0.8
0.6
REPAIR
FITNESS-FOR-PURPOSE LINE
BASED ON MAOP
0.4
REASSESS
FITNESS-FOR-PURPOSE LINE
0.2 BASED ON HYDROTEST
ACCEPT
0
0 100 200 300 400 500
DEFECT LENGTH, mm
A fitness-for-purpose assessment based on a deterministic approach does not quantify the safety of the
pipeline. Instead, it needs to impose arbitrary safety factors, and use conservative methods, to ensure
that a 'safe' pipeline will emerge. A major problem with fitness-for-purpose methods is setting the safety
factor, and agreeing this factor with the operator and regulatory authorities. In many ways, the safety
factor is better termed an ‘ignorance factor’; the larger the safety factor used, the greater the
ignorance[11].
It is now possible to quantify the safety of a pipeline using reliability-based limit state design (also
known as load and resistance factor design, LRFD [11, 12, 46-51]).
The term ‘Limit State Design’ used in the pipeline industry is misleading as it describes the application
of two distinct concepts to a pipeline design, i.e. limit state (structural engineering) analysis and
reliability theory. This Section introduces the reader to this leading-edge technology.
RESISTANCE (R)
DISTRIBUTION
L = R
PROBABILITY
LOAD (L)
DISTRIBUTION Partial Safety Partial Safety
Factor, Factor,
Nominal Resistance
Mean Resistance
e.g. 80% SMYS
Nominal Load
Mean Load
e.g. SMYS
Overlap is
failure
Nominal Safety probability
Margin,
The nominal load and resistance values are selected using probabilistic criteria,
e.g. the nominal resistance is defined as the strength below which not more than
5% of all possible test values would fall (referred to as the 5% fractile). The nominal load is
often defined as the mean load (for fixed loading) or a specified annual probability of
exceedance (for variable loads), e.g. 100 years.
Once the nominal loads and resistances are defined (they will be dependent on
the variability of loads and resistances and the model uncertainties), the partial
safety factors are used to achieve a sufficient separation between the load and resistance
probability distribution, to keep the probability of failure below an acceptable target value.
7. BEST PRACTICES
The pipeline industry has used their fitness for purpose methods to produce generic guidelines for the
assessment of defects in pipelines. These methods and guidelines are based on pioneering work at
Battelle Memorial Institute in the USA on behalf of the Pipeline Research Council International (6,53,54)
with the more recent additions of ad hoc guidelines for the assessment of girth weld defects, mechanical
damage and ductile fracture propagation produced by the European Pipeline Research Group (36,55,56).
‘Best practices’ in structural assessments of defects in pipelines are now emerging (e.g. 57-59), although
the recognised generic publications (9, 60) are still applicable.
A Joint Industry Project sponsored by 14 major oil and gas companies produced a state-of-the-art
Pipeline Defect Assessment Manual in 2003 (59, 61-63).
1. Fitness-for-purpose methods are available for a wide variety of defects in transmission pipelines.
Some of these methods have been available, and tried-and-tested, for many years, and their use
should be encouraged.
2. There are five levels (or ‘stages’) of defect assessment, ranging from simple methods detailed in
current codes, to highly sophisticated risk analyses, and limit state methods. A very high level of
expertise is required on the higher levels.
3. The defect assessment will only be as reliable and as accurate as the method used, and the data that
is available.
4. Corrosion in pipelines can be reliably assessed, but account should be taken of the accuracy and
reliability of the inspection data, and any future corrosion growth.
5. External interference can result in severe defects such as combined dents and gouges. These type of
defects must be assessed with caution, as their behaviour with time (i.e. fatigue) is not well-
understood, and dents associated with welds are very difficult to assess.
6. Defect assessment in pipelines is moving into reliability-based limit state design and risk analysis.
These methods will allow even more sophisticated assessment, providing accurate data is available.
9. RECOMMENDATIONS
This paper will end by giving engineers general advice on assessing defects in pipelines (67):
1. ALWAYS THINK SAFETY – Pipeline codes are safety standards, and an engineer’s
prime role in any industry is to ensure safety. Structural assessments are safety
assessments.
2. PIPELINE DESIGN CODES ARE ‘DAY 1’ CODES – Pipeline design codes give us
minimum structural integrity requirements – they are the starting point for integrity, not
the end point. A pipeline designed and built to code will be very safe on its first day in
service. However, after day 1 the pipeline’s management dictates its safety, not the
design code. Therefore, good managers and good management systems4 are the key to
pipeline structural integrity. This means a continual appraisal of technical issues such as
smart pigging, risk management programmes, correct routeing etc., to achieve high
integrity. But remember - only good management will guarantee integrity
3. PIPELINES ARE SAFE IF WE ADOPT A HOLISTIC APPROACH – Pipelines fail for
a reason, and most of these reasons can either be avoided or mitigated by good pipeline
management. Structural assessments must address all elements of the problem including
consequences of failure and the management systems that may affect the assessment or
need to be in place to implement any recommendation. Therefore, pipeline integrity
management must consider all aspects of our pipeline system, as it is an integrated
process, where all elements affect safety.
4. DO NOT DO THE ‘MINIMUM’ - Codes. Regulations, etc., are minimum
requirements. Aim to do more than your peers. It makes sense… if you do only one
thing more than your peers, his/her pipeline has a higher probability of failing first. You
4
Management Systems are covered References 64 and 65.
ACKNOWLEDGEMENTS
The author would like to thank colleagues at Penspen and Andrew Palmer & Associates.
1. Anon., 'Interstate natural gas pipelines - Delivering energy safely', Interstate Natural Gas
Association of America Report, USA, 1994.
2. Hopkins, P., 'Transmission pipelines: How to improve their integrity and prevent failures', 2nd Int.
Conference. on Pipeline Technology, Ostende, Belgium, 1995.
3. Fearnehough, G.D., 'The control of risk in gas transmission pipelines', Institution of Chemical
Engineers, Symposium Series No. 93, Manchester, UK, 1995.
4. Hopkins, P., Haswell, J., ‘The Practical Assessment Methods for Application to UK Gas
Transmission Pipelines’, The Institute of Materials 2 nd Griffith Conference, Sheffield, UK,
September 1995.
5. Hopkins, P., ‘Ensuring the Safe Operation of Older Pipelines’, Int. Pipeline and Offshore
Contractors Association, 28th Convention, Acapulco, Mexico, Sept. 1994.
6. Kiefner, J. F. et al., 'Failure Stress Levels of Flaws in Pressurised Cylinders', ASTM STP 536, pp
461-481, 1973.
7. Shannon, R. W. E., 1974, 'The Failure Behaviour of Linepipe Defects', Int. J Press Vessel & Piping,
(2), pp 243-255.
8. Miller, A. G., 'Review of Limit Loads of Structures Containing Defects', Int. J of Pressure Vessels
& Piping, (32), nos1-4, p195, 1988.
9. Anon., 'Guidance on Methods for the Derivation of Defect Acceptance Levels in Fusion Welds', BSI
7910, British Standards Institution, London, 1999. Formerly: Anon., 'Guidance on Methods for the
Derivation of Defect Acceptance Levels in Fusion Welds', BSI PD6493, British Standards
Institution, London, 1991
10. Hopkins, P., Hopkins, H., Corder, I., ‘The Design and Location of Gas Transmission Pipelines
Using Risk Analysis Techniques’, Risk and Reliability Conference Aberdeen, UK, May 1996.
11. Hopkins, P., ‘New Design Methods for Quantifying and Reducing the Number of Leaks in Offshore
and Onshore Transmission Pipelines’, European Pipeline Leakage Prevention Conference, IChemE,
London, May 1997.
12. Hopkins, P., Haswell, J., ‘The Practical Application of Structural Reliability Theory and Limit State
Concepts to New and In-service Transmission Pipelines’, Int. Seminar on Industrial Applications of
Structural Reliability Theory, ESReDA, Paris, France, October 1997.
13. Anon., ASME Boiler and Pressure Vessel Code Section VIII, Unfired Pressure Vessel Division 2,
1980.
14. Stewart, G., Klever, F., Ritchie, D., ‘An Analytical Model to predict the Burst capacity of
Pipelines’, Offshore Mechanics and Arctic Engineering, Houston, Texas, 1994.
15. Kastner,W., Rohrich,E., Schmitt,W. and Steinbuch,R.; ‘Critical Crack Sizes In Ductile Piping’, Int.
Journal Press. Vess. and Piping, (9), 1981, pp197-219.
16. Hopkins, P., Jones, D. G., Clyne, A. C., 'The Significance of Dents and Defects in Transmission
Pipelines', Conference. on 'Pipework Engineering and Operation', IMechE, London, Paper
C376/049, 1989.