Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
+ =
=
1
] [
] )
365
365
365
[(
2
(1)
where N, m
Loss
, T
shut-in
, CR, F
[Cl-]
, CR
O2
, F
pitting
, and t
i
represent
the total operating intervals, cumulative metal loss, shut-in
period, predicted uniform corrosion rate, pitting factor due to
chloride ion concentration, oxygen corrosion rate, localized
pitting factor, and time at each operating stage, respectively.
Initially, the localized pitting factor was fixed in each
subregion, which was determined based on gravimetric weight-
loss coupon data measured from the second scenario of the
CR088 pipeline. Table 1 summarizes the predicted metal loss
for the first scenario of three gravity pipelines in each
subregion. Figure 5 predicts the metal loss for the second
scenario of CR088 crude oil pipeline.
Table 1 Metal losses predicted by the old model, measured by
UT, and predicted by the new model at the same chainages
Region Site
Distance
(m)
UT
Average
Metal
Loss
(%)
Old
enpICDA
TM
Metal Loss
(%)
Difference
between
Old
enpICDA
TM
and UT
(%)
Updated
enpICDA
TM
Metal Loss
(%)
Difference
Between
Updated
enpICDA
TM
and UT
(%)
1
1 4,408 4.44 20.99 16.55 7.64 3.20
2 211 5.72
11.02 5.30 2.90 -2.82
4 275 3.08
11.38 8.30 3.08 0.00
5 489 3.32
13.17 9.85 3.93 0.61
6 4,868 4.99
21.67 16.68 7.96 2.97
2
1 6,633 5.00 25.07 20.07 9.58 4.58
2 6,601 4.28
25.12 20.84 9.61 5.33
3 5,144.84 7.16
22.48 15.32 8.36 1.20
4 6,282 4.28
24.95 20.67 9.54 5.26
5 Copyright 2012 by ASME
6 7,703.27 6.12
26.73 20.61 10.38 4.26
0 5000 10000 15000 20000 25000
0%
20%
40%
60%
M
e
t
a
l
L
o
s
s
Distance, m
ILI Measured
ILI Measured Average
BCE Predicted
Fig. 5 Metal loss predicted by the old model and measured by
ILI tools
As well, the solids deposition behaviours were characterized
during the Indirect Inspection stage. Firstly, solids deposition
sites were qualitatively identified through comparing the solids
depositional velocities with average liquid velocity locally. If
the average liquid velocity is less than the solids depositional
velocity, solids are anticipated to settle at or near that location.
Subsequently, the deposited solids amount was quantitatively
estimated. For the first scenario of three gravity pipelines,
comparison between the solids depositional velocity and
average liquid velocity reveals that, during normal operating
conditions, the average liquid velocity is greater than the solids
depositional velocity. Hence, no solids are anticipated to be
deposited along three gravity pipeline. Contrarily, during the
shut-in period, the majority of solids inside the pipeline will
settle as a result of stagnant flow. The solids entrained by the
crude oil eventually accumulated at the lowest point of Region
3 which is situated at the very end of the deadleg. For the
second scenario of CR088 pipeline, Region 1 experienced
solids deposition on a continuous basis since the average liquid
velocity is lower than that of solids depositional velocity. In
Region 3, solids deposition was predicted to only occur at
certain locations.
Furthermore, the Indirect Inspection step characterizes the oil
wettability behaviours in both scenarios. To comprehend the
wettability of the crude oil pipeline under various operating
conditions, the maximum water droplet size and critical water
droplet size were firstly calculated and compared using NACE
SP0208-2008. To quantify the water accumulation level, water
layer thicknesses were calculated under an approximated
stratified flow condition in compliance with the requirements of
NACE SP0208-2008. Results demonstrate that, for three
gravity pipelines, the entire pipeline will be oil-wet if the crude
flow rate is higher than 7,200 t/h which leads to a zero water
layer thickness. If the crude flow rate is below 5,400 t/h, the
pipeline will be water-wetted. During the normal operating
condition with a crude flow rate of 6,000 t/h, the predicted
thickness of the water layer shows that, both water-wet and oil-
wet conditions take place inside the pipelines. The maximum
water layer thickness could be up to 5.5 mm. As the BS&W
amount increases from 0.015% to 0.025%, water layer thickness
could be increased further. For the second scenario of crude oil
pipeline CR088, the maximum water droplet size is always
greater than the critical droplet size under all inclination angles.
That is, water will not be entrained in the hydrocarbon phase.
Hence water and oil are not emulsified. The critical velocity for
water accumulation, that is to avoid water drop out in the
pipeline, could be 2.106 m/s, leading to a critical flow rate of
119,240 m
3
/d (750,000 bbls/d). Since the current nominal
crude oil flow rate ranges from 12,719 m
3
/d (80,000 bbls/d) to
19,078 m
3
/d (120,000 bbls/d), the crude oil pipeline will remain
water-wet. Despite a very low BS&W, the over sizing or low
line-pack production discharge rates render the pipeline to have
a continuous water-phase.
In addition, the threshold for chlorides to trigger autocatalytic
pitting corrosion was predicted to be 10,000 mg/L (i.e., 1% by
weight) using the newly developed localized pitting corrosion
model which has been experimentally verified [8]. Since the
chloride ion concentrations in both scenarios exceed this value,
once a defect is generated due to either mechanical or chemical
reasons, the pitting corrosion could be autocatalytic, having an
incubation time of approximate 3 weeks under a stagnant
condition.
Based on the predicted solids deposition, water accumulation
and metal losses, six assessment sites per region were
recommended for excavation to facilitate the Detailed
Examinations, i.e., 2 locations with the greatest probability of
internal corrosion due to accumulation of water and solids, 2
consecutive locations, 1 location where indirect inspection did
not identify internal corrosion, and 1 other site where a
fixture can promote water or solids accumulation and hence
corrosion.
Detailed Examinations
To determine whether internal corrosion exists at selected
locations and to use the field findings to assess the overall
condition of the ICDA region, the Detailed Examinations were
conducted. For the three gravity pipelines, the ultrasonic testing
(UT) technique was utilized as a primary assessment tool for
both wall loss (compression wave) and flaw detection (angle
beam). The measurements of the remaining wall thicknesses
and detection of internal corrosion were recorded and
conducted by certified ASNT Level 2 & 3 inspectors with the
Level 3 UT inspector having considerable experience in flaw
detection in the nuclear industry. A total of 10 sites were
excavated per pipeline (i.e., 5 excavation sites in Region 1 and
5 excavation sites in Region 2). For the second LP-ICDA
scenario of the CR088 crude oil pipeline, the Detailed
Examinations were carried out using the results of the ultrasonic
In-Line Inspection (UT-ILI) tool to measure remaining wall
thickness. Based on the measured wall thickness, the measured
metal loss could be determined. Namely,
6 Copyright 2012 by ASME
% 100
min
min
=
al No
measured al No
loss
WT
WT WT
m
(2)
where m
loss
, WT
Nominal
, and WT
Measured
represent the measured
metal loss, nominal wall thickness, and measured wall
thickness, respectively.
Post Assessment
To assess the effectiveness of the performed LP-ICDA, a
validation of the proprietary predictive model was conducted
during the Post Assessment step. Figure 6 compares the
predicted and measured general corrosion rates in the gravity
pipeline CR102. It is noted that, both the actual measured
metal losses and the predicted values did not remain constant,
having a tendency to increase from the upstream to downstream
due to an increase in hydrostatic pressure as a result of a
decrease in elevation toward the downstream. Nonetheless,
there is a gap between the predicted metal losses and the UT
measured metal loss although the predicted average general
corrosion rates agree well. For the second scenario, as
displayed in Figure 5, the predicted metal loss could not match
the fluctuation of metal loss albeit the predicted average metal
loss agrees quite well with the UT-ILI measured data along the
entire CR088 crude oil pipeline.
0 1000 2000 3000 4000 5000 6000 7000 8000
0%
10%
20%
30%
40%
50%
0
20
40
60
80
100
120
E
l
e
v
a
t
i
o
n
,
m
M
e
t
a
l
L
o
s
s
Distance, m
enpIcDA enpIcDA enpIcDA enpIcDA
1M 1M 1M 1M
with a pitting factor
UT
Fig.6 Comparison of metal loss predicted with a uniform pitting
factor and measured by UT for gravity pipelines
GAP ANALYSIS
To identify the root cause leading to the discrepancy between
the measured and predicted metal loss, a gap analysis for both
scenarios was conducted. Based on Equation (1), the main
feature leading to the difference between the predicted values
and the conducted tests values was due to the pitting factor that
was a used as a fixed value and was uniformly applied for each
region.
For the first scenario of three gravity pipelines, there was no
corrosion coupon or UT data available for all three gravity
pipelines before the completion of the Indirect Inspection.
Hence, the pitting factor was approximated based on the only
available coupon data of nearby but independent and
aggressively corroding CR088 crude oil pipeline. This is based
on the assumption the fluids and corrosive environments in
three gravity pipelines and CR088 were likely similar since they
come from the same oil reservoir with the same-level CO
2
and
H
2
S concentrations and chloride ion concentration. In
hindsight, using the pitting factor from CR088 for CR102 (GL-
10), CR103 (GL-11), and CR104 (GL-12) was not a proper
approach although there was no other plausible option available
while performing Indirect Inspection. CR088 and three gravity
pipelines in reality have very different corrosive environments
and operating conditions. For the three gravity pipelines, there
was no water until the late 1980s or about the time of the
immediate post-Gulf War I era. Since then, they started to
realize water in their raw crude oil production which resulted in
having a subsequent Basic Sediment & Water (BS&W) amount
between 0.015% and 0.025%. However, for CR088, the
BS&W amount is over 20 X greater with 0.5% in Regions 1 &
2 and about 10X more with 0.2% in Region 3. This almost a
ten-fold difference has a profoundly different impact on pitting
behaviour between the two scenarios. Meanwhile, the BS&W
amount of the gravity pipelines was measured using a sample
from the bottom of the storage tank, which differs with the
actual amount inside the pipeline. The crude flow in three
gravity pipelines has an average rate of 5,000 t/h, leading to an
average liquid velocity of 2.73 m/s, whereas the average 90,000
bbl/day in CR088 leads to an average velocity of 0.3 m/s in
Region 1. Under the higher flow rates in the gravity lines, the
micro-solids and water droplets moving inside the gravity lines
would likely not settle inside the pipeline when the fluids are in
motion. Hence, solids deposition and water accumulation in the
gravity pipelines will be totally different from that of CR088.
In addition, the chloride ion concentrations in the gravity lines
are as high as 1.7 wt%. However, this value is believed to be
below that found in CR088 pipeline because of the observed
solid halite in the pigged solids collected from CR088 which
had a maximum saturated chloride ion concentration up to 36
wt%! In other words, the severities of pitting corrosion in the
gravity lines and CR088 are vastly different from each other.
Furthermore, no formation of elemental sulphur had been
hitherto reported for the three gravity pipelines. Nevertheless,
as characterized by the XRD analysis, the extracted pig receiver
scale sample from the CR088 crude oil pipeline does include
Rosickyite, i.e., elemental sulphur, S
8
. Under the synergistic
effect of chloride ion concentration and elemental sulphur, the
pitting factor in three gravity lines should be much less than that
found in CR088. Some of these facts were known during the
Pre-Assessment step and others were reported late and only
7 Copyright 2012 by ASME
after the completion of Indirect Inspection, but the inclusion of
a similar pitting factor was not warranted. Consequently, the
usage of an upstream pitting factor from CR088 on the
downstream gravity lines was not correct and ought to have
been avoided.
For Scenario 2 of CR088 crude oil pipeline, the pitting factor
was a fixed value at each region. As demonstrated in Figure 5,
the ILI measured metal loss fluctuates more than the predicted
metal loss. The maximum metal loss discrepancy at some
segments could be as high as 30% albeit the average metal loss
discrepancy is only 3%. Undoubtedly, usage of constant pitting
factor during the Indirect Inspection is the reason, which leads
to a relatively uniform metal loss over the entire pipeline.
Hence, the approach with a uniform pitting factor is unable to
catch the peaks and predict the point by point metal loss along
the entire pipeline.
DEVELOPMENT OF DYNAMIC PITTING FACTOR
In reality, a robust pitting behavior should be dynamic, which
varies both along the crude pipeline length and at different
stages in time for the pipeline regions, and subregions. In 2011,
the authors developed a mechanistic model to predict localized
pitting corrosion through solving a variety of governing
equations from the first principle
[8]
, which has been validated
through the laboratory tests.
Based on the developed model, the
effects of pressure, temperature, convection, boundary layer
thickness and chloride ion concentration on localized pitting
corrosion could be quantified. As well, theoretical threshold for
each parameter on pitting corrosion was obtained. Below that
threshold, pitting corrosion will be passivated whereas above
the threshold, pitting corrosion will be autocatalytic. Hence, the
dynamic pitting factor was as a function of water accumulation,
solids deposition, pressure, temperature, deposition of
elemental sulphur, chloride ion concentration, oxygen ingress,
and bacteria activities. The updated equation for the dynamic
pitting corrosion can be expressed as :
i T i p i Solids i O H i ptting
F F F F C F
, , , , ,
2
= (3)
where F
pitting
, C, F
H2O
, F
Solids
, F
P
, and F
T
represent the dynamic
pitting factor, an empirical constant, water accumulation factor,
solids deposition factor, pressure factor, and temperature factor,
respectively. The empirical constant includes the effects of
deposition of elemental sulphur, chloride ion concentration,
oxygen ingress, and bacteriological activities. When the
dynamic pitting factor at each specific segment was obtained,
the fluctuating metal loss was determined.
Apparently, to obtain the dynamic pitting factors, the local
pressure, temperature, water flow rate, oil flow rate, solids size
and density, cation and anion concentrations, mole fraction of
CO
2
, H
2
S and O
2
, as well as bacterial counts need to be given,
which can be obtained through conducting water analysis, solids
analysis, gas analysis, and bacterial analysis. To ascertain which
variables were most likely responsible for the appreciable gap, a
point-to-point comparison between the UT-ILI measured and
the proprietary model predicted metal losses was carried out in
this study. Specifically, the dynamic pitting factor ranges from
2.0 to 33.5 for Region 1 and 1.5 to 22.5 for Region 3 in CR088
pipeline. But Broadsword used a constant value of 10.0 for
Region 1 and 7.0 for Region 3 along the entire region during
the conducted Indirect Inspection step for the previous
simulation of the metal loss in CR088 since the original pitting
factors based on the coupon data provided by Kuwait Oil
Company range from 6.6 to 14.3 for CR088 Region 1 only.
Hence, the pitting factor could be as low as 1 and as high as
33.5 for CR088 in Region 1. Through a comparison between
In-line Inspection (ILI) and Ultrasonic Testing (UT) data, the
wall loss with respect to the predicted general corrosion rates
could be correlated.
Based on Tafels law, the anodic corrosion current density has
an exponential tendency to anodic potential. In terms of the
Nernst Equation, the anodic potential is directly governed by
the above environmental parameters. Hence, a logarithm
relation was constructed for the variables of water, solids,
pressure and temperature. The following equations were
generated to describe those factors in Equation (3):
100
) 100 log(
,
2
B W A
F
i
i O H
+
= (4)
m i
S i S
S
S
F )] 1 log( 1 [
,
+ + = (5)
] log 1 [
,
P
P
F
i
P i P
+ = (6)
] log 1 [
,
T
T
F
i
T i T
+ = (7)
where A, B,
S
, m,
P
, and
T
are constants;
T P S , , represent
the average solids amount, average pressure, and average
temperature, respectively. Through a comparison with ILI data
for CR088, the empirical constants in Equations 4-7 could be
obtained based on the statistic theory since thousands of data
points are involved. To validate those obtained factors, the
measured data by Ultrasonic Testing (UT) for the first scenario
of three gravity pipelines were applied. Hence, the dynamic
pitting factors were obtained using the first-hand field ILI/UT
data for two extremely scenarios, i.e., one for CR088 using ILI,
and another for three gravity pipeline using UT data. Since
ICDA is a continuously iterative process, this study updated the
simulation by replacing the uniform pitting factor with a
dynamic pitting factor for the purposes of reducing local
measurement discrepancies as opposed to errors and having a
more accurate series of predictive metal losses when compared
with the UT and UT-ILI data.
8 Copyright 2012 by ASME
0 1000 2000 3000 4000 5000 6000 7000 8000
0%
10%
20%
30%
40%
50%
0
20
40
60
80
100
120
E
l
e
v
a
t
i
o
n
,
m
M
e
t
a
l
L
o
s
s
Distance, m
Lpdated enpIcDA Lpdated enpIcDA Lpdated enpIcDA Lpdated enpIcDA
1M 1M 1M 1M
UT
Fig.7 Metal loss of gravity pipeline CR102 predicted with a
dynamic pitting factor
Figure 7 shows the measured metal loss as compared with the
predicted metal loss with the enhanced and improved dynamic
pitting factor for the first scenario of the three gravity pipelines,
CR102. It is noted that application of dynamic pitting factor
sees the smart dig UT-measured metal losses agreeing well
with the local predicted metal loss. With respect to the second
scenario of CR088 crude oil pipeline, Figures 8-10 manifest the
comparison between the measured metal loss and predicted
metal loss in three subregions of Region 1. It is indicated that
fluctuations and their relative magnitudes of the predicted metal
loss agree reasonably well with the fluctuations recorded by the
UT-ILI measurements.
0 1000 2000 3000 4000 5000 6000
0%
10%
20%
30%
40%
50%
60%
70%
50
60
70
80
90
100
E
l
e
v
a
t
i
o
n
,
m
M
e
t
a
l
L
o
s
s
Distance, m
enpIcDA enpIcDA enpIcDA enpIcDA
1M 1M 1M 1M
ILI
Fig. 8 Metal loss predicted with a dynamic pitting factor and
measured by ILI in CR088 Region 1, Subregion 1
7000 8000 9000 10000 11000 12000 13000
0%
10%
20%
30%
40%
50%
60%
70%
50
60
70
80
E
l
e
v
a
t
i
o
n
,
m
M
e
t
a
l
L
o
s
s
Distance, m
enpIcDA enpIcDA enpIcDA enpIcDA
1M 1M 1M 1M
ILI
Fig. 9 Metal loss predicted with a dynamic pitting factor and
measured by ILI in CR088 Region 1, Subregion 2
14000 15000 16000 17000 18000 19000
0%
10%
20%
30%
40%
50%
60%
70%
50
60
70
80
90
100
E
l
e
v
a
t
i
o
n
,
m
M
e
t
a
l
L
o
s
s
Distance, m
enpIcDA enpIcDA enpIcDA enpIcDA
1M 1M 1M 1M
ILI
Fig. 10 Metal loss predicted with a dynamic pitting factor and
measured by ILI in CR088 Region 1, Subregion 3
To assess the effectiveness of the updated model with a dynamic
pitting factor, statistics theory had to be applied due to
thousands of ILI data points for CR088. As described below, the
standard deviation was selected to compare the fluctuation level
between the predicted metal loss and the ILI measured metal
loss.
=
=
N
i
i i
m m
N
1
2
) (
1
(8)
where , m, and m represent the standard deviation, metal loss
and average metal loss in a specific subregion, respectively. A
low standard deviation means the data points tend to be very
close to the mean whereas high standard deviation indicates that
the data are spread out over a large range of values. Table 2
compares the population standard deviation of the metal loss
between the measured data and the predicted values in
Subregion 1, Subregion 2, and Subregion 3 of both Region 1
9 Copyright 2012 by ASME
and Region 3 in Scenario 2. It is noted that the standard
deviation of metal loss measured by the ILI and predicted by
proprietary model correlates. Based on the performed deviation
analysis, the effectiveness of the developed model could be
ensured.
Table 2 Population standard deviation of metal loss between ILI
measurements and predictions with a dynamic pitting factor
Region
Deviation of metal loss
New model with a
dynamic pitting factor
enpICDA
ILI
ILI
Region 1 12.030 10.642
Subregion 1 12.968 9.803
Subregion 2 12.603 11.608
Subregion 3 10.904 9.401
Region 3 5.182 6.153
Subregion 1 7.144 8.336
Subregion 2 2.510 2.870
Subregion 3 2.889 5.382
Consequently, the updated model not only predicts average
metal losses but it is also capable of tracking the peaks and
troughs or valleys of the UT-ILI measured metal loss, which can
be utilized to assess the long-term effectiveness of the LP-ICDA
process.
Since this study was the first application of ICDA with a
dynamic pitting factor on the KOC crude oil pipeline systems,
the updated model can monitor the reliability or repeatability,
track the evolution of metal degradation, and establish absolute
criteria such as minimum performance requirements. Through
this approach, the crude oil pipeline operation can identify
geographic locations at which corrosion activity has occurred, is
occurring, or may occur. In addition, based on the modeling
and field testing results, it is noted that three gravity pipelines
which have an unblemished operating history of 59 years had a
low metal loss whereas a CR088 having a relatively short 6.5-
year operating history suffered from severe localized pitting
corrosion throughout Region 1.
Through above case studies, the proposed dynamic pitting
factor will qualitatively predict the fluctuating metal loss, which
demonstrates the predictive performance of the model and can
guide ICDA methodology towards a higher level of accuracy
and precision. Nonetheless, the developed dynamic pitting
factor can be utilized to predict the metal loss less than 60% of
metal loss limited by the provided ILI data. Although the
developed model can be extrapolated to predict the metal loss
over 60% metal loss, further validation is required.
CONCLSIONS
Localized pitting corrosion behaviour changes from pipeline to
pipeline, region to region, and subregion to subregion as a
consequence of variation of local pressure, temperature, solids
deposition, and water accumulation. Hence, pitting factors
which address the metal loss due to localized pitting corrosion
must also be dynamic rather than employing an assumed fixed
value. This paper conducted an innovative development of
dynamic pitting factor through two contrasting internal
corrosion scenarios assessed by Liquid Petroleum - Internal
Corrosion Direct Assessment (LP-ICDA). The predicted metal
loss updated with dynamic pitting factors is able to track the
fluctuation of metal loss along the entire pipeline which
enhances ICDA methodology towards a higher level of accuracy
and precision regardless of the severity of the internal corrosion
environment.
The contrasting pipeline scenarios discussed herein and
employing the NACE International SP0208-2008 LP-ICDA
methodology was highly successful. The two very contrasting
scenarios; one on a 36 piggable pipeline commissioned in
2005 with severe internal corrosion and the second with three
40 X 38 telescoped, export quality crude gravity fed non-
piggable pipelines commissioned in 1959 with minor to nil
internal corrosion were successfully inspected through the LP-
ICDA process. This proves that with the inclusion of dynamic
pitting models, LP-ICDA technique can be effectively applied
to crude oil pipelines regardless of their heritage, pressure,
internal corrosion severity or being piggable or not. The added
benefit of applying LP-ICDA especially on a piggable line is
that susceptible internal corrosion locations can be found prior
to being corroded along with the root-cause analysis and
forecasting internal corrosion rates based on future pipeline
operations.
REFERENCES
1. Liquid Petroleum International Corrosion Direct
Assessment, NACE International Standard Practice, NACE
SP0208-2008, 2008.
2. E.W. Klechka, Special issues related to the application of
Direct Assessment, NACE International corrosion
conference, Paper # 09147, 2009.
3. L.A. Bensman, Use of an internal corrosion threat
assessment to identify location to conduct an integrity
assessment, NACE International corrosion conference,
Paper # 09148, 2009.
4. B. Singh, K. Krishnathasan, Pragmatic effects fo flow on
corrosion prediction, NACE International corrosion
conference, Paper # 09275, 2009.
5. A.P. Pillai, Direct assessment pipeline integrity
management, NACE International corrosion conference,
Paper # 11126, 2011.
10 Copyright 2012 by ASME
6. Q. Lu, G. Huang, Internal corrosion direct assessment of
buried steel pipeline in Chinese oil industry, NACE
International corrosion conference, Paper # 08130, 2008.
7. V.V. Lagad, S. Srinivasan, R.D. Kane, Facilitating internal
corrosion direct assessment using advanced flow and
corrosion prediction models, NACE International corrosion
conference, Paper # 08131, 2008.
8. Z. Zhu, N. Tajallipour, P.J. Teevens, H. Xue, F.Y.F.
Cheng, A mechanistic model for predicting localized
pitting corrosion in a brine water-CO
2
system, NACE,
Corrosion 2011, Paper # 11256, Huston, TX, 2011.
9. Pipeline and Hazardous Materials Safety Administration
(PHMSA) Significant Incidents Files, October 2010.
10. American Petroleum Institute (API) Standard 1160 on
Managing System Integrity for Hazardous Liquid Pipelines.
11. ASME B31.8S Managing System Integrity of Gas
Pipelines.
12. P. Luo, L. Liu, W. Wang, H. Chen, Y. Zhang, F. Song, J.
Li, Experience with Dry Gas Internal Corrosion Direct
Assessment Methodology for a Gas Pipeline Segment
Located in Southern China, NACE International corrosion
conference, Paper # 11308, 2011.
13. C. de Waard, U. Lotz, Prediction of CO
2
Corrosion of
carbon steel, NACE International corrosion conference,
Paper # 69, 1993.
14. N. Brauner, D. Moalem Maron, Flow pattern transitions in
two phase liquidliquid horizontal tubes, International
Journal of Multiphase Flow, 1992.
15. P.W. Jepson, H. Shi, et al., Predicting of water film
thickness and velocity for corrosion rate calculation in oil-
water flows, NACE International corrosion conference
Paper # 02500, 2002.
16. S. Nesic, X. Tang, et al., Effect of oil type on phase wetting
transition and corrosion in oil water flow, NACE
International corrosion conference, Paper # 07170, 2007.