Tracy Thorleifson, ENTRUST, USA, looks at the role that integrating inline inspection data into a pipeline digital twin can play in optimizing asset performance and risk management in the pipeline industry lifecycle.
Pipeline asset performance optimisation relative to the asset’s risk profile is critical for operators. Remaining asset life prediction within ‘as-low-as-reasonably practicable’ (ALARP) risk constraints is crucial to maximising pipeline asset performance. Remaining life prediction relies upon understanding the time-dependent threats to the pipeline, including corrosion. US regulations stipulate that pipeline segments affecting a high consequence area (HCA) are subject to periodic assessment. Most operators choose to perform inline inspection (ILI) assessments where feasible, because of the wealth of data produced by modern smart pigs. Many operators have performed multiple generations of assessments. Tracking corrosion anomalies across multiple assessments to estimate corrosion growth and remaining asset life is challenging. Constantly evolving ILI technology introduces complexity. Correlating coarse features in old assessments to high resolution features in new assessments can be problematic. Operators often utilise multiple vendors over time, introducing complications inherent to using multiple proprietary technologies. Despite tremendous volumes of data, anomaly identification and tracking across multiple assessments remains largely manual and is subject to inconsistency and unintentional bias. To address these problems, ENTRUST employs a novel application of geographic information system (GIS) technology via its PIMS ILI Data Integration Tool. This software applies GIS proximity analysis to anomaly tracking across multiple assessments. Process automation facilitates analysis of all anomalies, resulting in statistically robust corrosion growth estimates and better understanding of remaining asset life.
Figure 1 is a depiction of ENTRUST’s ILI data integration workflow. Orange boxes represent ENTRUST software elements used to support process steps. The orange software element boxes with dashed outlines near process steps (3) and (4) are separate from the PIMS ILI Data Integration Tool. The ILI data integration workflow is sandwiched between the assessments and confirmatory digs and repairs workflows. The ENTRUST ILI data integration workflow is a collaborative effort between ENTRUST’s level III ILI experts and GIS analysts.
Focus here is on process steps (3) to (8), executed by ENTRUST GIS analysts. Steps (1) and (2) are completed by an ILI expert. In step (1), the pipe tally data sheet and ILI assessment final report are received from the vendor. The final report contains detailed metadata about the assessment, including tool specifications, assessment findings, and other information. In step (2), the ILI expert reviews data to check for indications of abnormal smart pig tool behaviour. Abnormal tool travel velocity is important because it affects performance and accuracy, complicating ILI data calibration. The pipe tally data sheet, delivered in Microsoft Excel format, lists all features detected by the smart pig, including corrosion anomalies and other defects, pipe deformations (dents), pipe fittings and features. Pipe fittings and features include girth welds between individual pipe joints, bends, valves, taps, tees, flanges, pipe wall thickness changes, repair sleeves, pipe casings, cathodic protection cadwelds, etc. Some modern ILI tools can also detect longitudinal seam orientation for individual pipe joints. Feature locations in the pipe tally data sheet are reported using a ‘coordinate system’ specific to ILI data: odometer and clock position. Odometer is travel distance recorded along the interior of the pipeline. Odometer position may not be accurate; velocity departures can negatively impact odometer accuracy. Clock position records the axial position of features 12 o’clock is the top of the pipe; 6 o’clock position is the bottom of the pipe. Corrosion anomalies are listed in the pipe tally data sheet by odometer and clock position. Additional attribution includes anomaly length and width (in inches), anomaly depth (expressed as percentage of pipe wall loss), and location on the pipe (internal or external). Individual anomalies may be located within an encompassing anomaly ‘cluster’.
The odometer/clock position ‘coordinate system’ does not locate ILI features in the real world, so some method of ‘calibrating’ ILI data to real world coordinates is necessary. Many modern smart pigs are equipped with inertial measurement units (IMUs) providing real world coordinates for ILI features, but IMUs can be subject to inertial drift and lag, so some degree of location calibration is still required. Step (3) in Figure 1 comprises calibrating ILI data to real world coordinates, facilitated by the Calibratorator tool. The Calibratorator converts odometer values to pipeline stationing values via linear interpolation between features common to both the GIS and the ILI assessment. Each set of matched features in the GIS and assessment constitutes a ‘calibration point’. Larger numbers of calibration points result in higher quality calibration. The end goal of calibration is to use all girth welds as calibration points, referred to as the ‘weld map’. Once the weld map is established, accurate calibration of all ILI data is readily achieved. The weld map evolves over time, due to pipe replacements and pipeline reroutes. In performing ILI data calibration, it is vital for the GIS analyst to be aware of all pipe replacements and reroutes. Depending on data quality in the GIS and ILI assessments, calibrating ILI data can be extremely complex. However, once accomplished, the GIS becomes a highly detailed, temporally aware ‘digital twin’ of the physical pipeline. This enables the downstream steps in the ILI data integration workflow, setting the stage for remaining asset life determination.
For anomaly matching, relative accuracy trumps absolute spatial accuracy. Anomaly matching can be performed without real world coordinates. Establishing the weld map is critical. Weld-to-weld mapping across assessments reduces the scale of anomaly matching operations to a single joint of pipe. Absolute spatial accuracy is crucial for accurate dig locations and to understanding ILI data in full spatial context.
Step (4) in Figure 1 comprises loading calibrated ILI data into the GIS. ENTRUST practice is to store all ILI data in the enterprise GIS. The software is agnostic regarding data storage; local data stores are acceptable.
Step (5) in Figure 1 encompasses calculation of ASME B31G safety and pressure related attributes on ILI anomalies. The Pressure Calculator tool inputs include anomaly length and depth, recorded wall thickness and other physical pipe properties. Output is common ASME B31G remaining strength attributes, which are used in anomaly prioritisation and serve as a check on attribution provided by the ILI vendor.
In Figure 1, step (6), ILI data is converted into an artificial coordinate system by the ILI Data Converter tool, enabling GIS-based anomaly matching and tracking.
Concepts
The GIS does not understand clock position and odometer distance. To apply GIS tools to ILI data, it is necessary to convert to an artificial GIS X/Y coordinate system. This is done by ‘unzipping’ the pipe and laying it out flat. The X axis of the artificial coordinate system is pipeline stationing. The Y axis is pipe circumference, i.e. clock position. GIS proximity analysis can be used for anomaly matching in this artificial coordinate system.
Process
Converted ILI features are depicted in Figure 2. The ILI Data Converter tool can apply various shape transformations to input ILI features. Here, clock position grid lines are black horizontal lines labelled with clock position. Blue horizontal lines represent the longitudinal pipe joint seams. Vertical red lines represent girth welds, labelled above with girth weld number and below with pipeline stationing. Green rectangles represent pipe bends; width of these features is 1 ft. Brown rectangles represent pipe supports on this above ground pipeline. The pink background indicates this pipe could affect an HCA. Corrosion anomalies are visible at this scale as point features, labelled by anomaly depth. The dashed outline around the anomaly at the centre shows the display extent of Figure 3. Pipeline stationing, clock position, length and width are the key attributes for conversion of ILI anomaly features. Anomalies are converted as point, box (envelope) and ellipse features. At the scale of Figure 3, the ellipse and box representations of the anomaly feature are visible. Ellipses are most useful for anomaly matching.
Anomaly matching
Step (7) in Figure 1 comprises anomaly matching. The Anomaly Matching tool uses GIS spatial proximity analysis to detect potential matches. The tool is directional. Generally, it is appropriate to use an older ILI assessment as the first (parent) dataset, and a newer ILI assessment as the second (child) dataset. The tool attempts to find one or more matches in the child dataset for each parent anomaly, within a specified search radius. In Table 1, five anomalies from a parent assessment are matched to seven anomalies in a child assessment. There are seven records in total in the table because parent anomalies four and five are each matched to two child anomalies. The ‘near distance’ column displays parent/child separation distance; a zero value indicates spatial overlap between parent and child. Figure 6 illustrates parent anomaly five and its two matching child anomalies. There has been significant corrosion growth between parent and child ILI assessments.
Anomaly growth calculation
Step (8) in Figure 1 addresses anomaly growth calculation. The growth calculator tool calculates anomaly growth between two assessments, calculating annual corrosion growth for matched and unmatched anomalies. For parent anomalies with multiple children, growth statistics are calculated. ILI tool uncertainties are incorporated into worst case growth rates. Overall corrosion growth rates for the pipeline are evaluated using frequency distribution plots. Figure 4 depicts corrosion growth for the matched anomalies in the example data. All five parent anomalies appear in the frequency distribution plot, which is broken into five sample bins. Mean and median worst-case annual corrosion growth rates for the sample set are shown, as is the standard deviation. Little is to be gleaned from this simple plot, but when hundreds of samples are included, interesting patterns can emerge. Nongaussian and multi-modal frequency distributions may indicate that multiple corrosion modalities are operative on the pipeline. Figure 5 shows the worst-case annual corrosion growth frequency distribution plot for unmatched child anomalies. There are 72 unmatched corrosion anomalies in the child assessment, divided here into eight bins. Unmatched anomalies corrosion growth is calculated on a ‘halflife’ basis, assuming anomaly growth starts halfway between the parent and child assessments. This frequency distribution is skewed toward rapid corrosion growth. This suggests some portions of the pipeline may be subject to accelerated corrosion, meriting immediate further investigation (and repair action), because the extreme outliers suggest a failure could occur before the next scheduled assessment.
Figure 7 illustrates the pipeline’s digital twin, showing corrosion anomalies in spatial context. The close association of most corrosion anomalies with pipe supports is obvious. Most anomalies are coincident with pipe supports and are at the 6 o’clock position, suggesting a course of mitigative action to maximise remaining asset life.
ILI data is like any other data – messy. Proper calibration of ILI data to the GIS is the key to truly understanding the data’s message. Conversion to an artificial coordinate system facilitates GIS-enabled anomaly matching. Process automation enables analysis of all data, not just a subset. Anomaly growth estimates are statistically backed. Visualisation of ILI data in the GIS digital twin leads to insight and understanding, allowing transformation of data into actionable knowledge that can be used to maximise remaining asset life.
Please Sign Up to get Our Latest Newsletter
myENTRUST