Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

NACE Paper No.

MECCOCT18- 12598

Novel Method for the Automated Alignment of In-Line Inspection (ILI) Features

Ameen AL Obedan
Saudi Aramco/R&DC Ahmad Saif
Saudi Arabia Saudi Aramco/P&DT
Saudi Arabia
Raed Sammanodi
Saudi Aramco/R&DC Rafael Mora
Saudi Arabia Saudi Aramco/P&DT
Saudi Arabia
Tarik Hoshan
Saudi Aramco/R&DC
Saudi Arabia

ABSTRACT

The measurement of corrosion growth rate is an essential step to assess the integrity of transmission
pipelines in the oil and gas industry. In-line inspection (ILI) is a ubiquitous tool in the industry to inspect
corrosion defects. To do this, at least two ILI runs from different points in time are necessary. The features
of each run are matched (in terms of location and orientation), and their sizes used for statistical analysis,
to determine their uncertainty, which is factored into the calculation for corrosion growth rate. The process
of aligning the anomalies is manually intensive, and is both time consuming and subject to human error.
In addition, a single run may contain hundreds of thousands of features, and manual matching means
calculations are made using only a small fraction of the ILI dataset.
This work presents a newly developed software, which can automate ILI feature alignment and validation
of the results. The program uses developed algorithms, visualization, and also incorporates user inputs
to account for situations such as rerouted pipelines, new segments, rehabilitated segments, reversed
flow, and pig orientation. Different assumptions were imbedded in the program based on field experience.
The performance of the software was assessed by comparing its predictions to that obtained from
historical data sets.

Key words: Pipeline, Corrosion, In-line inspection, Feature matching, Integrity assessment

INTRODUCTION

Oil and gas industry possess a large network of carbon steel pipelines that require frequent corrosion
inspection. In-line Inspection (ILI) tools is considered as one of the common non-destructive technologies
for detecting corrosion features and it is conducted in every few years to evaluate the integrity of the
pipelines and to estimate corrosion growth rate. Estimating corrosion growth rate is crucial to predict the
time of pipeline failures. Therefore, it is important to match the corrosion features of ILI consecutive runs

Saudi Aramco: Company General Use


with an accurate and defined approach. Currently, the corrosion features are matched by screening out
the data manually which is susceptible to human errors and considered time intensive when it is
performed for several pipeline spots. Therefore, an opportunity for an automated, reliable and time
efficient ILI matching tool with a clear described methodology is identified in this project.

ILI corrosion features can be matched by locating either the raw signals received from the ILI tools or the
boxed features which is originally post-processed from the raw signals. The adopted approach focuses
on the latter due to its practicality for engineers to perform their analysis and it can still be applied
unaffected of changing the ILI tool or vendor.

REPEATED ILI MATCHING

In order to estimate the corrosion growth rate in scrapable pipelines, at least two ILI datasets need to be
compared properly. The corrosion growth can be determined then by calculating the difference between
recent and previous corrosion values of the pipeline. Currently, there are various comparison techniques
are being used and four of them are listed below:

Segment Growth Estimation


One of the methodologies used to estimate the corrosion growth rate is by measuring the increase in the
average corrosion depth for a pipeline segment between two ILI inspections. This method does not
concentrate on a single feature growth rate; it is rather focuses on the distribution of the whole group of
features within a pipeline segment. This approach may be used for risk ranking between segments,
setting ILI intervals and estimation of pipe remnant life. 2

Manual Feature Matching


The current practice of matching corrosion features is performed manually. First, girth welds are used as
reference points to determine the degree of transformation of one of the in-line inspection coordinate
system. Second, based on the location, a few aligned features are identified to minimize the matching
error. Third, the degree of transformation of the coordinate system of the first in-line inspection is identified
from the result of the few matched features in previous step and then used to align with the second in-
line inspection coordinate system. Fourth, more matches from the first in-line inspection are identified
with the applied transformation of the coordinate system. This procedure of alternating between finding
features matching and refining the coordinate system transformation continues until the matched features
reached a satisfactory number to allow performing corrosion growth assessment.
Manual feature matching is easily achievable if the number of corrosion features is low in a single pipeline.
However, matching corroded pipelines with high densities of corrosion defects is a time exhaustive
method. Furthermore, the process of manual matching is error-prone and mismatching of features is
possible due to other features’ close proximity. 1

Features Box Matching


In box matching, log distances are used as a reference to align two in-line inspection runs. Corrosion
features are then aligned between the two in-line inspections in order to calculate the depth difference
based on the detected depth of every single feature in the two inspection. A linear growth over the
inspection cycle is assumed to calculate the corrosion growth rate. Features without corresponding
feature in the first in-line inspection may represent new corrosion defects. However, this case is often
associated with the failure to report the feature previously due to depth reporting threshold or detection
capabilities of the inspection tool. The corrosion growth rate accuracy is influenced by the accuracy of
the feature matching algorithm as well as the reported data accuracy of the two inspections. 3

Features Signal Matching


Saudi Aramco: Company General Use

2
In signal matching, the signal raw data of two inspections are compared and the change in signal
amplitude between two inspections is used to directly determine the corrosion growth rates. The depth
change can be calculated utilizing the matched signals. The calculation of corrosion growth rate of new
features using signal matching approach is much reliable, since these newly developed features may
have been detected previously but was not reported due to the fact of their sizes were below the reporting
threshold. 3

FEATURES MATCHING CASES

The representation of corrosion features may differ between two ILI. Both conditions of the inspected
pipeline and accuracy of the inspection instrument datasets are contributing factors to the detected
corrosion features transformation in terms of existence, size and location. Seven different matching cases
shown in figure 1 were taken in consideration throughout the matching process of misaligned features.
These cases focus in the size transformation of corrosion features and are discussed in this section.

One to One (A)


In this case, a single corrosion feature needs to be matched with its correspondence in the second ILI
data. The matched feature most likely will differ in size (metal loss) and dimensions (length, width). Both
size and dimensions could increase, decrease or remain unchanged depending on the pipeline health in
addition to the inspection data accuracy and quality.

One to None (B)


In this case, a single corrosion feature has no match in the second ILI data. The reason behind this case
can be one of the following:
1- Failure to detect the same feature in the second ILI.
2- The original reported feature was a false positive.
3- The feature is close to the reporting threshold.

One to Multiple (C)


In this case, a single corrosion feature found as multiple features in the second ILI data. This clearly
means that initiation of new corrosion features have occurred between the two inspections.

None to One (D)


In this case, a single corrosion feature appears in the second ILI data whereas it did not exist in the first
ILI data. This outcome is as a result of one of the following:
1- Initiation of new corrosion feature.
2- Failure to detect this feature in the first ILI.
3- The feature was under the reporting or detection threshold in the first ILI.

None to Multiple (E)


This case is exactly same as the previous case. The only difference is the appearance of multiple features
in the second ILI instead of a single feature.

Multiple to None (F)


In this case, multiple corrosion features have no match in the second ILI data. The reason behind this
case can be one of the following:
1- Failure to detect the same features in the second ILI.
2- The original reported features were a result of false positives.
3- The features are close to the reporting threshold.

Multiple to One (G)

Saudi Aramco: Company General Use


In this case, multiple corrosion features are found to be grouped (clustered) as a single feature in the
second ILI data. The growth of a localized distribution of multiple corrosion features in the duration
between two ILI can result in merging them into single feature sized to enclose all of the clustered original
corrosion features.

A B C D E F G

A B C D E F G
Figure 1: Features matching cases

FEATURES MATCHING METHODOLOGY

Girth Welds and Features Matching


A customize algorithm was used for matching girth welds in the first and second inspections. Girth welds
are joining method for small segments of a pipeline, the length of a single pipeline segment between two
girth welds is approximately between 10 to 24 meters. To match the exact girth weld for the two
inspection, a specified tolerance have been taken in consideration to overcome the challenge of different
starting point of inspection process.

After identifying and matching girth welds, Iterative Closest Point (ICP) is responsible for matching
corrosion features with the nearest neighbor. A modified ICP algorithm was developed using custom
mathematical equations to minimize the translation of points.

Mathematical Modeling
To start features matching, proper match for pipeline girth welds needs to be met first. In this work,
stretching technique (Equation 1) explained in figure 2 were utilized for the alignment of girth welds for
two ILI data.

𝑒 𝑠
𝑠 𝐺𝑖2 −𝐺𝑖2 𝑠
[𝑥𝑖1 ] = ([𝑥𝑖1 ] − 𝐺𝑖1 )∗ 𝑒 𝑠 + 𝐺𝑖2 (1)
𝐺𝑖1 −𝐺𝑖1

i1: first inspection


i2: second inspection
G: girth weld
s: start
e: end

Saudi Aramco: Company General Use

4
Figure 2: Example of the Stretching and Alignment Equation

For features matching, modified ICP algorithm was combined with an additional Sequential Quadratic
Programming (SQP) to minimize point translation. Objective function Equations 2 and Gradient functions
Equations 3-4 were used in the developed framework. A resulted point translation before and after ICP
implementation are shown in figure 3.

2 2 2
𝑓(𝑡𝑥 , 𝑡𝑦 , 𝑡𝑧 ) = ∑𝑛𝑖=1 [(𝑝𝑥𝑖 − (𝑞𝑥𝑖 + 𝑡𝑥 )) + (𝑝𝑦𝑖 − (𝑞𝑦𝑖 + 𝑡𝑦 )) + (𝑝𝑧𝑖 − (𝑞𝑧𝑖 + 𝑡𝑧 )) ] (2)

𝐺 (𝑡𝑥) = [−2 ∗ ∑𝑛𝑖=1(𝑞𝑥𝑖 )] + [2 ∗ ∑𝑛𝑖=1(𝑝𝑥𝑖 )] + [2 ∗ 𝑛 ∗ 𝑡𝑥] (3)

𝐺 (𝑡𝑦) = [−2 ∗ ∑𝑛𝑖=1(𝑞𝑦𝑖 )] + [2 ∗ ∑𝑛𝑖=1(𝑝𝑦𝑖 )] + [2 ∗ 𝑛 ∗ 𝑡𝑦] (4)

Figure 3: Modified ICP results with minimized point translation

Features Matching Results


Two ILI data sets for a single pipeline section were tested by the discussed matching methodology. The
results are displayed in figure 4 before the matching algorithm implementation, and figure 5 after the girth
welds stretching & modified ICP algorithm implementation.

Saudi Aramco: Company General Use


Figure 4: Results before matching

Figure 5: Results after matching

CONCLUSION

Comparing two ILI datasets is well-known approach in pipeline industry to determine the corrosion growth
rate. However, to have accurate calculation for the corrosion growth, both ILI features should be aligned
properly. Manual matching is time intensive and subjected to human error, instead box matching and
signal matching of ILI corrosion features is used. In this paper, a modified Iterative Closest Point (ICP)
algorithm have been used to develop a corrosion features matching software. The results show high
percentage of matched features and further development is being done to improve this work further.

Saudi Aramco: Company General Use

6
REFERENCES

1. Dann, Markus & Dann, Christoph. (2017). Automated Matching of Pipeline Corrosion Features from
In-line Inspection Data. Reliability Engineering & System Safety. 162. 40-50. 10.1016/j.ress.2017.01.008.

2. Nessim, Maher & Dawson, Jane & Mora, Rafael & Hassanein, Sherif. (2008). Obtaining Corrosion
Growth Rates From Repeat In-Line Inspection Runs and Dealing With the Measurement Uncertainties.

3. Russell, A., Smith, M., Sandana, D., & Knudsen, J. (2016). ADVANCES IN MAGNETIC FLUX
LEAKAGE SIGNAL MATCHING AND CORROSION GROWTH RATE SELECTION.

4. Shi, Y., Zhang, C., Li, R., Cai, M., & Jia, G. (2015). Theory and Application of Magnetic Flux Leakage
Pipeline Detection. Sensors (Basel, Switzerland), 15(12), 31036–31055.

5. Smith M, Argent C, Wilde A. Corrosion Growth and Remnant Life Assessment: How to Pick the Right
Approach for Your Pipeline. ASME. International Pipeline Conference, Volume 1: Pipelines and Facilities
Integrity.

Saudi Aramco: Company General Use

You might also like