Journal of Sustainability Science and Management ISSN: 1823-8556

Volume 6 Number 1, June 2011: 69-78 Universiti Malaysia Terengganu Publisher

DETERMINISTIC PREDICTION OF CORRODING PIPELINE REMAINING

STRENGTH IN MARINE ENVIRONMENT USING DNV RP F101 (PART A)
NORHAZILAN MD NOOR*, NOR APAZIAH NOR OZMAN AND NORDIN YAHAYA
Reliability Engineering & Safety Assessment Research Group (RESA)
Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310 Skudai,Johor, Malaysia.

*Corresponding author: [email protected]

Abstract: Internal corrosion has long been acknowledged as one of the dominant forms of
deterioration process that contribute to the containment loss for marine steel pipelines. Aging and
deteriorating pipelines under influence of corrosion threat may experience a serious reduction of
their structural integrity and can lead to eventual failure. To secure pipeline safety for a long and
profitable life, the operators need to develop their own risk-based inspection schedule for future
inspection and maintenance activities. A deterministic methodology for predicting the remaining
strength of submarine pipelines subjected to internal corrosion using a capacity equation as included
in the DNV RP-F101 (Part A) code is described in this paper. The equation can be used to estimate
the maximum allowable operating pressure of the corroding pipelines based on a series of pigging
data, which represents the corrosion pit location and dimension. The introduction of partial safety
factors in the Part A of the DNV RP-F101 code is tailored to minimise the effect of uncertainties
due to defect sizing. The authors have added prediction capabilities to the capacity equation by
introducing a standard deviation model of future predicted defect depth. By doing so, the variation
of safety factors of the capacity equation has been manipulated to that extend where prediction of
future pipeline remaining life-time becomes feasible. The paper demonstrates derivation of time-
function standard deviation equation, Std(d/t) of tool error, calculation and prediction of pipeline
remaining lifetime subject to internal corrosion. The increment of standard deviation of corrosion
depth, Std(d/t) was addressed since it can affect the value of partial safety factor as corrosion
progresses, hence amplifying the conservatism of time to failure. The prediction results shows that
the dynamic of safety factors has successfully downgraded the structure resistance as corrosion
progresses to reflect the actual condition of the pipeline on site. The technique to evaluate future
pipeline remaining lifetime can effectively assist pipeline operators to evaluate future safe operating
strategies including re-inspection and appropriate maintenance schedule. As a result, it can minimise
the likelihood of pipeline failures until it reaches its designed lifetime.

KEYWORDS: Pipeline, DNV RP-F101, corrosion, reliability, deterministic

Introduction [Teixera et al., 2008]. As a pipeline ages, it can

Transporting crude oil and gas by the means of be affected by a range of corrosion mechanisms,
pipelines systems has become the safest and most which may lead to a reduction in its structural
economical method in the world as compared integrity and eventual failure [Ahammed, 1997;
to tankers, provided that the long-term integrity Netto et al, 2005; Teixera et al., 2008]. Corrosion
of the line is well-secured. Unfortunately, the is an important form of pipeline deterioration
increasing number of aging pipelines in operation due to aggressive environments [Ahammed and
has significantly increased the number of accidents Melchers, 1996&1997]. Without a practical and
effectual corrosion-prevention strategy, corrosion
Received: 15 July 2010 / Accepted: 15 September 2010 will continue to progress and the cost of repairing
Norhazilan Md Noor et al. 70

a deteriorating pipeline will escalate. Significant dependent upon inspection tool accuracy which
savings are possible by optimising the inspection is defined by the dispersion of corrosion growth-
and corrosion-prevention strategies [Ainouche, rate value and metal-loss data. The authors have
2006]. manipulated the polynomial equation of safety
factors in the DNV RP-F101 to make the capacity
Pipeline Inspection equation capable of predicting the future growth
In line inspection (ILI) tools, also commonly of defects. This is done by deriving a time-
called pipeline inspection gauge or pig, are function standard deviation equation, Std(d/t) of
devices used by the pipeline industry to survey the inspection tool. The predicted metal-loss data
mainly the internal condition of the pipeline in the future is supposed to pose higher variation
wall. Intelligent Pig, a tool with the capability from its central tendency value compared to actual
of mapping anomalies, is widely deployed to metal-loss data. Hence, a higher safety factor to
detect, locate and measure the size of a corrosion cater for defect depth is necessary to increase the
defect in a pipeline using high-resolution conservatism of assessment as well as to have a
Magnetic Flux Leakage (MFL) or Ultrasonic more realistic assessment due to rapid reduction
Testing (UT) techniques. The past 40 years has of structure capacity.
seen the development of a number of methods
for assessing the significance of defects. Some Metal Loss Information
of these have been incorporated into industry In this case study, an extensive amount of pigging
guidance [Cosham et al., 2007]. The nucleation of data has been gathered through repeated in-line
defects in the pipeline can result in a serious wall inspection activities using MFL intelligent pig
thinning in a pipeline. Whereas MFL provides a on the same pipelines at different points of time.
versatile and reliable method for determining the The transmission pipelines located in the North
geometry of metal loss in pipelines, UT allows Sea region used to convey crude oil and gas
direct and fairly accurate measurements of (multiphase line) from central offshore platform
pipeline wall thickness. However, the UT tool is to onshore terminal. The data provides valuable
limited in terms of usability on gas pipeline since information on the internal corrosion defect
the tool requires medium to transmit and receive geometry, such as defect location, depth and
back the ultrasonic signal during ILI inspection. length, orientation and types of corrosion regions.
The data were used to evaluate the current state
Research Problem and Methodology of the pipe under corrosion attack using the DNV
The inherent uncertainties embedded within RP-F101 equation. The authors have incorporated
metal-loss data play significant roles in reducing a statistical concept into the pipeline evaluation
the accuracy of pipeline future assessment. procedure so that future prediction of pipeline
These uncertainties are related to imperfect remaining capacity can becomes feasible. A new
tool measurement, uncontrolled environment equation of standard deviation, Std [d/t] reflects
and variation of operational data [Yahaya et. al, the defect sizing by the inspection tool has been
2011]. Furthermore, the complexity of corrosion derived from linear metal-loss rate equation based
mechanism involving numerous unknown factors on the statistical principle. This equation is meant
and limited resolution by the inspection tool can to recalculate the new safety factor and fractile
jeopardize the integrity of structure assessment value for future state whereby the factors are
practice [Din et. al, 2009]. To cater to the required by the DNV RP-F101 capacity equation
uncertainties, the DNV RP-F101 [DNV, 2004] to estimate the remaining life of a corroding
has incorporated safety factors into the capacity pipeline.
equation which are specially tailored to account
for uncertainties associated with defect depth.
Unlike conventional safety factors, the value is

J. Sustain. Sci. Manage. Volume 6 (1) 2011: 69-78

DETERMINISTIC PREDICTION OF CORRODING PIPELINE 71

Assessment of Corroding Pipeline DNV Recommended Practice (RP-F101)Criteria

Theoretically, the safety factors introduced to The DNV Recommended Practice for the
represent uncertainties associated with load assessment of corroded pipelines was first issued
system and structure resistance (material strength) in 1999. RP-F101 describes two alternative
is commonly found in all codes of assessment for approaches with different safety philosophies.
corroded pipelines. Nonetheless, the safety factor The equations in the DNV RP-F101 were derived
has led to a certain level of degree of conservatism by a probabilistic calibration taking into account
in regard to structural assessment causing over the uncertainties in defect measurements and
prediction of deterioration intensity [Ozman et. burst capacity [Bjornoy and Marley, 2001]. The
al, 2010]. The uncertainties subjected to structural equations account directly for the accuracy in
properties, loading condition, environmental sizing the corrosion defect. The DNV RP F101
behaviour and construction performance are recommends the assessment of corroded pipelines
always neglected in the calculation due to the subject to internal pressure and internal pressure
employment of a safety factor. In pipeline combined with longitudinal compressive stresses
assessment, deterministic assessment is a straight- [Bjornoy and Marley, 2001]. Moreover, this new
forward approach based on codes or developed criterion also provides an assessment procedure
capacity equation. Generally, the deterministic for single defect, interacting defects and complex
methods use lower bound data; for instance, shaped defects.
peak depth of corrosion, maximum corrosion
rate and minimum wall thickness without Capacity Equation
considering the existing uncertainties [Yahaya, The capacity equation is a mathematical model
2000]. Consequently, it can be over conservative used to estimate the remaining pressure capacity
in terms of safety when being implemented to of the line after the initiation of corrosion defect.
pipelines containing extensive corrosion defects. The maximum allowed operating pressure in
For example, the prediction of future growth pipelines for a single defect presented by the
of corrosion defects located in the pipelines DNV RP-F101 [DNV, 2004] is given as:
normally is based on averaged single rate value
without considering the possibility that not all
defects will grow at the same rate. The averaged (1)
rate is used for the sake of simplicity owing to
lack of information pertaining to environmental where;
and material properties [Noor et. al, 2008].
Assessment method is required to determine the (2)
severity of such defects when they are detected
in pipelines [Cosham and Hopkins, 2003]. The
assessment of the condition of existing oil and
gas pipelines is necessary in order to protect the (3)
public, financial investment and environment and;
from such failures. Systematic and optimised D = outer diameter (mm)
regular inspections of pipelines with state-of-the d = depth of corrosion defect (mm)
art tools and procedures can reduce significantly t = nominal pipe wall thickness (mm)
the risk of any undue accident caused by a lack of L = measured length of corrosion defect (mm)
unawareness of the integrity of the line [Cosham d/t = ratio of corrosion depth to pipe wall
et al., 2007]. thickness
(d/t)meas = measured relative corrosion depth
(d/t)* = actual relative corrosion depth to cover
uncertainties related to inspection tool

J. Sustain. Sci. Manage. Volume 6 (1) 2011: 69-78

Norhazilan Md Noor et al. 72

gm = partial safety factor for prediction of material properties, pipe geometries and
model and safety class corrosion defects configurations.
gd = partial safety factor for corrosion
depth Partial Safety Factors
ed = factor for defining a fractile value for The concept of partial safety factors is different
the corrosion depth from sole safety factors in that both strength and
Pmao = maximum allowable operating load system have to be multiplied with multiple
pressure (MPa) safety factors. In pipeline assessment, the partial
StD[d/t]= standard deviation for measurement safety factors gm and gd, and the fractile value
(d/t) ratio ed are determined from tables which depend on
SMTS = specified minimum tensile strength the safety class classification, the pipe quality,
(N/mm2) inspection method and sizing accuracy of the
Fundamentally, equation (1) is similar inspection tool [DNV, 2004]. They were given as
to ASME B31G [B31G, 1991]. However, functions of the sizing accuracy of the measured
the difference between these two criteria is defect depth for inspections based on relative
that partial safety factors are included in the depth measurements and for inspections based on
DNV RP-F101 equation to ensure a consistent absolute depth. The safety class is specified based
reliability level for various combinations on Tables 1 to 3.

Table 1. Partial safety factor gm, [DNV, 2004].

Table 2. Standard deviation, StD [d/t], for MFL inspection tool [DNV, 2004].

Table 3. Partial safety factor, gd and fractile value factor, ed [DNV, 2004].

J. Sustain. Sci. Manage. Volume 6 (1) 2011: 69-78

DETERMINISTIC PREDICTION OF CORRODING PIPELINE 73

Pipeline Remaining Lifetime The augmentation of prediction capabilities

The steel pipeline has a 14.2 mm wall thickness of equation (1) by increasing the standard
with outside diameter given as 914.4 mm. The deviation of d/t can be explained mathematically
allowable defect size of the pipeline is indicated by by referring to the linear growth rate model. The
the uppermost curve, i.e. the acceptance line (refer corrosion rate equation can be written as:
to Figures 1 to 5). Once the corrosion point exceeds
the acceptance line, the pipeline is considered to (6)
be in a critical condition. Hence, inspection and
repair are recommended to commence. The mean where T= ti+1 ti and is a constant value.
value and standard deviation of the corrosion rate
If corrosion depth d is assumed statistically
used in this assessment are 0.0405 mm/year and
varied, the variation of corrosion rate can be
0.08 mm/year respectively. The acceptance line
expressed as:
was constructed using equation (1). From this
equation, the maximum corrosion defect length
(7)
was estimated by fixing the ratio of corrosion
depth in percentage, d/t between 10% wt to
100%wt (wt = pipe wall thickness) and with the Since the time interval, T is a single value with no
different working pressures of 8MPa, 9MPa and variation, Equation 7 can be rewritten as:
10MPa. By taking out the defect length parameter,
an equation of length correction factor, Q can be (8)
written as:
and simplified into:
(9)

Therefore, the relationship between inspection

(4) time interval and the variation in corrosion
growth rate can be presented as:

(10)
Since Q can also be represented by equation
Since s=Std, therefore
(2), the maximum allowable defect length for a
given defect depth and working pressure can be
(11)
calculated as:
Inspection data of metal loss from MFL pig tools
(5) usually is represented as a ratio of defect depth to
where: wall thickness, d/t. By replacing the exact metal-
Lmax = maximum allowable defect length (mm) loss value, d with metal-loss ratio, d/t equation
(11) can be rewritten as follows;
The Q expression in equation ( 5) can be
calculated by inserting Equation 4 into Equation 5
so the allowable line/ acceptance line to evaluate (12)
pipeline condition can be constructed. The
increment value of Std [d/t], gd and ed in the future
can be estimated using Table 4 as given in the Equation (12) now can be reshuffled to make
DNV RP-F101 code. it as a standard deviation model of predicted
depth. The new form of equation as a function
of variation of defect from previous inspection,

J. Sustain. Sci. Manage. Volume 6 (1) 2011: 69-78

Norhazilan Md Noor et al. 74

inspection time interval and variation of corrosion Results and Discussion

rate is as follows; Pipeline time to failure was determined using the
DNV RP-F101 capacity equation (Part A) [DNV,
(13) 2004]. Figures 1 to 5 show the prediction result of
pipeline assessment subject to internal corrosion
from year t0 to t10. These predictions were based on
(14) gathered corrosion data from pigging inspection
done in year t0. From the assessment result in
year t0 and t2, the measured corrosion defect is
By assuming the wall thickness, t as a fixed value within the acceptance criteria for all operating
with no variation, the conclusive equation can be pressures where there are no defects exceeding
presented as the acceptance line as shown in Figures 1 and 2.
Further prediction was made from year t5 until t10.
(15) As can be seen in Figures 3 to 5, the corrosion
depth begins to exceed the acceptance line
where: starting from year t5 when the pipeline is operated
cr = corrosion rate (mm/year) under 9MPa and 10MPa of operating pressure.
Std[d/t]o = Standard deviation of inspection tool The pipeline is considered critical condition due
in first year assessment. to bursting in year t5 when the acceptance criteria
Std[d/t]T = Standard deviation of inspection tool for all operating pressures were exceeded by the
in the future. projected defects. Based on the result, it can be
Std[cr] = Standard deviation of corrosion rate. concluded that the pipeline should be inspected no
T = prediction interval in year later than year t5 for every condition of operating
ti = year of inspection pressure. The acceptance line predicted by the
2cr = variance of corrosion rate (mm/year)2 corrosion defect in year t until t was found to
5 10
The equation depicts relationship between be lower than the one estimated for the t0 and t2
deviation of predicted data and the interval of predictions. This is hypothecally owing to the
prediction. The longer the prediction interval, the increment of uncertainties related to the averaged
higher the variation of future metal loss, hence corrosion growth rate [Noor et. al, 2007]. Figures
the higher the partial safety factors for metal loss. 6 to 8 show the increment of StD[d/t], fractile
Table 4 shows the equations required to estimate value, ed and safety factor, gd over the year, tn.
the partial safety factors for metal loss and fractile The increment of StD[d/t] over time exhibits
value according to the range of metal loss standard the exponential shape indicating the growth of
deviation, Std[d/t]. uncertainties along with the projection of future

Table 4. Polynomial equation for partial safety factor (defect depth) and fractile value [DNV, 2004].

J. Sustain. Sci. Manage. Volume 6 (1) 2011: 69-78

DETERMINISTIC PREDICTION OF CORRODING PIPELINE 75

defect growth . Fractile value, ed and safety Conclusion

factor, gd follow the polynomial law and become The great advantages of the deterministic
constant when projection exceeded period of 10 approach are due to its simplicity and the
years. This gives an indication that the projection capability of being applied to an entire pipeline
reliability is not meant for long-term prediction or collection of pipelines straightforwardly
of pipeline future state due to extreme unforeseen [Lawson, 2005]. The inability to deal with
events. Equation (15) and Table 4 were used uncertainties in the input data is the primary
to recalculate the abovementioned values as weakness of deterministic approach which leads
corrosion progresses in time. to inaccurate interpretation of pipeline condition

Figure 1. Projection of corrosion depth in year t0.

Figure 2. Projection of corrosion depth in year t2.

Figure 3. Projection of corrosion depth in year t5.

J. Sustain. Sci. Manage. Volume 6 (1) 2011: 69-78

Norhazilan Md Noor et al. 76

Figure 4. Projection of corrosion depth in year t7 using DNV RP-F101 Assessment code.

Figure 5. Projection of corrosion depth in year t10 using DNV RP-F101 Assessment code.

Figure 6. The increment of StD[d/t] over time.

and not entirely due to error of inspection sizing. This is a result of the requirement that the
data [Lawson, 2005]. Therefore, a prediction operators have to inspect their pipelines frequently
of pipeline integrity by using a deterministic in order to obtain better information on pipeline
assessment methodology cannot effectively fulfill condition. The introduction of partial safety
the cost-saving requirement by the operators. A factors to minimise the effect of uncertainties due
semi-probabilistic theory was introduced in the to the defect sizing and standard deviation model
DNV RP-F101 code by estimating the standard of future metal loss has improved the capability of
deviation of inspection tool error and defect predicting the future growth of corrosion defects

J. Sustain. Sci. Manage. Volume 6 (1) 2011: 69-78

DETERMINISTIC PREDICTION OF CORRODING PIPELINE 77

Figure 7. The increment of fractile value, ed over time.

Figure 8: The increment of safety factor, gd over time.

deterministically. In addition, a new equation Ahammed, M. and Melchers, R. E. (1996). Reliability
was derived so that the partial safety factors can Estimation of Pressurised Pipelines Subject to
be recalculated so as to take into account the Localised Corrosion Defects. International
growth of uncertainties as corrosion progresses. Journal of Pressure Vessels and Piping. 69. 267-
272.
This technique ensures that the resistance system
will be naturally downgraded as the load system Ahammed, M. and Melchers, R. E. (1997).
Probabilistic Analysis of Underground Pipelines
approaches the limit state due to corrosion to
Subject to Combined Stresses and Corrosion.
reflect the actual load-resistance interference law.
Engineering Structures. 19(12). 988-994.
Ainouche, A. (2006). Future Integrity Management
Acknowledgements
Strategy of a Gas Pipeline Uisng Bayesian Risk
The first author is pleased to acknowledge the Analysis. 23rd World GAS Conference.5 9 June.
Ministry of Science, Technology and Innovation, Amsterdam, Netherlands: 3.3EF.05.
Malaysia (MOSTI) and the Ministry of Higher B31G (1991). ANSI/ASME B31G-1991-Manual for
Education (MOHE) for the support by providing Determining the Remaining Strength of Corroded
the funds and scholarship (NSF). Pipelines A Supplement to ASME B31 Code For
Pressure Piping, Revision of ANSI/ASME B31G-
References 1984, New York: ASME.
Ahammed, M. (1998). Probabilistic Estimation of Bjornoy, O. H. and Marley, M. J. (2001). Assessment
Remaining Life of a Pipeline in the Presence of of Corroded Pipelines , Past, Present and Future.
Active Corrosion Defects. International Journal Stavanger, Norway: ISOPE 2001, Proceedings
of Pressure Vessels and Piping. 75. 321-329. of the 11th International Offshore and Polar

J. Sustain. Sci. Manage. Volume 6 (1) 2011: 69-78

Norhazilan Md Noor et al. 78

Engineering Conference. June 17-22. Vol II. 93- Noor, N. M, Smith, G. H., Yahaya, N. (2007).
101. Probabilistic Time-Dependent Growth Model
Cosham A. and Hopkins P. (2003). The assessment of Marine Corrosion In Seawater Ballast Tank,
of corrosion in pipelines Guidance in the Malaysian Journal of Civil Engineering, 19,
pipeline defect assessment manual (PDAM), 2:142-155.
International Colloquium Reliability of High Noor, N. M, Yahaya, N., Othman, S. R. (2008). The
Pressure Steel Pipelines, Effect of Extreme Corrosion Defect on Pipeline
Cosham, A., Hopkins, P., Macdonald, K. A. (2007). Remaining Life-Time, Malaysian Journal of
Best practice for the assessment of defects in Civil Engineering. 20, 1: 47-57.
pipelines Corrosion. Engineering Failure pp 1-30, 2003
Analysis. 14. 1245 1265. Ozman, N. A. N., Noor, N. M., Yahaya, N., Developing
Din. M. M., Ngadi, M. A, Razak, K. A., Noor, N. Seamless Integrated Approach of Assessment
M., Improved Inspection Data Quality Using for Corroding Pipeline, The 2nd International
Modified Corrosion Rate Method for Offshore Conference on Engineering and ICT , 10th
Pipeline Assessment, ICCTD, vol. 1, , 2009 February 2010, Malacca, Malaysia.
International Conference on Computer Technology Teixera, A. P., Guedes Soares, C., Netto, T. A., Estefen
and Development, 2009 Kota Kinabalu, Malaysia, S. F. (2008). Reliability of pipeline with corrosion
November 13-November 15, pp.340-344. defects. International Journal of Pressure Vessels
DNV. (2004). Recommended Practice RP-F101 for and Piping. 85. 228 - 237.
Corroded Pipelines 2004., Norway: Det Norske Yahaya, N. (2000). Risk Based Method in
Veritas. Pipeline Maintenance Optimisation. Hanover,
Lawson, K. (2005). Pipeline corrosion risk analysis Germany: Proceeding of the Global Dialogue
an assessment of deterministic and probabilistic World Exposition (EXPO2000). Science and
methods, Anti-Corrosion Methods and Materials. Technology, Thinking the future. July 11-13
52: 1, 3 10 Yahaya, N., Noor, N. M, Othman, S. R., Lim, K. S., Din,
Netto, T. A., Ferraz U. S., Estefen S. F. (2005). The M. M. (2011). New Technique in Soil-Corrosion
effect of corrosion defects on the burst pressure Study of Underground Pipeline, Journal of
of pipelines. Journal of Constructional Steel Applied Sciences, Vol. 11, Online First.
Research. 61: 1185-1204.

J. Sustain. Sci. Manage. Volume 6 (1) 2011: 69-78