Improving interpretation of SCC anomalies found by ultrasonic ILI.
For crack-like features, an amplitude-based depth model was developed to profile the crack, calculating the depth along the longitudinal direction This provides a more accurate representation of the crack's details and can reduce conservatism. The profiles have been verified in the field although some tolerances are run-specific so verification digs are recommended to quantify their effect on the pipeline's integrity.
For crack-field features, actual stress corrosion cracking (SCC) colonies were examined and measured using in-the-ditch NDE to produce accurate maps of the location. Subsequent Australian Pipeline Industry Association Research Standards Committee (APIA RSC) burst tests of these colonies then provided information on how the various crack-field alignments interact and their failure mode. It was found that the Canadian Energy Pipeline Association (CEPA) interaction rules were the most appropriate for modeling interaction while there was little variation in the assessment method. By applying different filtering techniques to the ILI data it was possible to produce accurate maps of the SCC colonies. When applying CEPA interaction rules to these filtered signals the upper bound conservatism in the failure pressure was reduced from 19% to 14%. Further work is being undertaken to increase the sample size used and the confidence associated with failure pressures calculated from USCD ILI data.
Both methods represent major advancements in the interpretation of ultrasonic ILI data. The reduction in unnecessary conservatism in crack evaluations can reduce the number of excavations and repairs to ultimately provide better informed integrity-and maintenance-planning decisions for pipeline operators.
SCC is an environmental assisted form of cracking that has plagued the pipeline industry since at least 1965. Much research has taken place and as this time-dependent threat is better understood, SCC management strategies become ever more sophisticated. However, for pipeline segments deemed susceptible to SCC, the traditional integrity assessment methods of hydrotesting, direct assessment (SCCDA) or ILI still apply.
[FIGURE 1 OMITTED]
Hydrotesting was introduced in the 1960s to combat the consequence of failure experienced as a result of pipeline testing using air or product (Kirkwood et al, 2002). In recent years the purpose has moved from being a leak-tightness test to one which benefits the pipeline as a structural system, guaranteeing the immediate integrity of the pipeline. However, the associated costs, logistics and lack of data regarding sub-critical flaws are major disadvantages to the assessment method.
SCCDA is widely thought of as a cheaper assessment method but that depends on the length of the segment and the quality and quantity of data readily available to the operator. Certainly for short segments the cost is relatively low but changes as the length of the susceptible pipeline segment and number of direct exanimations increases. Although ILI data is a codified element (NACE, 2004) of the SCCDA process it is often thought of as an ILI-exclusive assessment method. Without crack detection ILI data, the actual detection capability of the SCCDA method is difficult to quantify and there is often an uncertainty of when to terminate the direct examination stage. However, SCCDA is relatively new and the process will become more refined as usage increases.
ILI with an appropriate inspection vehicle is a reliable assessment method with a detection rate up to 95%. The vast amount of data generated by intelligent pigs is invaluable when establishing the extent of an SCC threat and making quantitative decisions regarding future operation of the pipeline. However, while ILI tools designed to detect cracks in liquids pipelines are established, crack-detection ILI technology for gas pipelines remains relatively new. The techniques discussed here apply to data collected by the ultrasonic shear wave inline inspection vehicle.
Ultrasonic Crack Detection
Since its introduction in 1994, the shear wave ultrasonic crack detection tool (USCD) has inspected more than 32,000 km of pipelines (Marreck et al, 1999, Uzlec 2000, Wolf 2001). Detection, identification and sizing capabilities of the tool have been verified and subsequently refined. Figure 1 shows a typical signal recorded by the USCD ILI tool as it passed by an SCC colony.
USCD tools report crack fields and features classified as crack-like, weld anomaly, metal loss, inclusion, geometry, and installations. For crack-like features, the length and maximum depth category are reported and for crack fields, field length, width, longest indication, and maximum depth category are reported.
The detection capabilities specified give thresholds below in which defects are not detected and reported. These are minimum crack length of 30 mm and minimum crack depth of 1 mm. The depth is categorized in four ranges, namely < 12.5% , 12.5-25%, 25-40% and > 40% of wall thickness.
As with all detection and measurement systems and as required by API 1163 (API, 2005), the USCD tool has a probability of detection, classification, and measurement error that has to be accounted for when calculating the remaining strength of the defects in the line.
During its 12-year history, the reliability and accuracy of the USCD tool has been verified through the excavation of over 1,100 anomalies. As the tool's abilities became quantified they were used in assessment methods to make decisions about the immediate and future integrity of the inspected pipeline. It was found that generally the assessment methods tended to be conservative. It was assumed that some of this conservatism was intrinsic to the way in which cracking indications were reported. As confidence in the technology had been proven, it was decided to investigate whether improvements could be made to the ways in which the anomalies were analyzed and reported.
When an individual cracking indication is recorded by the USCD inspection tool it is categorized as "crack-like." This will include such anomalies as fatigue cracks, toe cracks or hook cracks. The method of reporting this defect will include a maximum depth indication and an overall length. The correct method to assess the significance of this defect in relation to the immediate integrity of the pipeline is to utilize a fracture-mechanics-based Failure Assessment Diagram method presented in recommended practice API 579-2000 (API 2000, PDAM 2003).
[FIGURE 2 OMITTED]
While it is assumed that tool-specific tolerances will be considered in the calculation by some method, the inputs required to describe the size of the anomaly are crack depth and crack length. The depth will be reported within a depth range so it is conservatively recommended to assume the upper hound of that range.
For a part-wall surface flaw, a semi-elliptical idealization is assumed (Figure 2) and the defect assessed accordingly. From experience, it has been concluded that this assessment method poses two problems. The semi-ellipse idealization is only applicable to some flaws. The other problem stems from the fact that the cracks are all reported within a depth band, so for a particular pipeline with a cracking problem, there can be many cracks reported as having the same depth. This makes prioritization of remediation a difficult task.
Crack Field Indications
A crack field is reported when two or more cracking indications are detected within close proximity of each other. They are generally found to be associated with SCC. Traditionally, a crack field would be reported with the following dimensions, a maximum depth observed within the colony, the overall length and width of the colony and the length of the longest indication.
The correct method to assess the significance of this defect in relation to the immediate integrity of the pipeline is to utilize a fracture-mechanics-based Failure Assessment Diagram method (API 579, BS7910, CEPA 1997). As for crack-like indications, the depth is reported within a depth range and it is reasonable to conservatively assume the upper bound depth value as the maximum depth of the colony. The length is a critical part. The most conservative method is to assume the full length of the colony and idealize as a semi-elliptical flaw. This will produce very conservative results as SCC colonies are generally large in extent (Spencer et al, 2006) and using overall lengths can produce conservative failure pressures.
Using the length of the longest indication and the maximum depth of the overall colony is another method to try to eliminate some conservatism. It is well known that cracks within an SCC colony often coalesce to form [Parkins et al, 1993, Leis and Mohan 1993, Leis et al, 1995] even longer cracks which can lead to failure or reduced safe working pressures. By not taking this well-documented phenomenon into account then an appropriate safety factor should be employed.
It was clear that the present method of reporting cracking indications was not sufficient for modern standards. This prompted two initiatives to address unnecessary conservatism in crack assessments. For crack-likes it was decided that a method to determine the true profile should be studied and for SCC colonies some account of crack-density and interaction be considered.
Idealizing a crack as a semi-ellipse and then using an upper bound depth value can produce conservative results. Jaske et al  provide a model to consider the effective area of a crack. The model assumes that the equivalent crack has a semi-elliptical profile. This is either based on the maximum length and depth of the crack or on a measured crack-profile. In the latter case, an effective semi-elliptical flaw is based on the effective length and the area of the worst flaw identified by an RSTRENG-type analysis, recalculated to give an effective depth for a semi-elliptical flaw of the same effective area and failure stress [Jaske and Beavers 2002, Jaske et al. 1996].
[FIGURE 3 OMITTED]
[FIGURE 4 OMITTED]
An amplitude-based sizing model was developed to predict actual crack profiles from the ILI data. The sizing model uses various variables as inputs and includes orientation and location of the flaw, product medium and the distance from the sensor. With this improved analysis technique a detailed profile can be determined with a depth prediction given along the longitudinal direction. Once the profile is known it can be idealized as a semi-ellipse using the area to calculate the "effective dimensions". An illustration of the profiling and effective area is shown in Figure 3.
Crack-profiling and the amplitude-based depth sizing model has been developed over recent years and verifications performed with several pipeline operators. Figure 4 shows an example of such a verification. The black line represents the predicted profile while the blue line shows the actual profile as determined in the field.
While the crack-profile predictions accurately mirror actual profiles, there is still a need to account for tolerances. As mentioned, some of the inputs to the sizing model can be run-specific, e.g., medium, so it is recommended that a series of verification digs be performed. It is not recommended to provide a failure pressure at this stage, rather use the effective dimensions in a FAD calculation to relatively rank the cracks against each other.
SCC Colony Interaction
SCC colonies can consist of several hundred individual defects aligned perpendicular to the principal stress. The behavior of SCC colonies has been studied extensively and several phenomena are known to exist. Dense colonies, where the cracks are within close proximity of each other circumferentially, can shield each other from stress, whereas cracks within close proximity axially can join or coalesce to form longer cracks, which could lower their failure pressures. Although modeling of these behaviors can be done within a laboratory, it is unclear if this knowledge can be applied to ILI data.
[FIGURE 5 OMITTED]
APIA RSC Burst Tests
The APIA RSC performed burst tests on three sections of X65 pipeline containing high-pH SCC. The SCC had been detected by the USCD ILI tool during inspection, verified in the ditch and the section removed. The sections were subjected to more detailed NDE using magnetic particle inspection (MPI) and ultrasonic testing (UT). Figure 5 shows an example of a SCC colony prior to burst testing.
Comparison Of Rules
To enable a comprehensive examination of the effects of any improvements in analysis of the ILI data it is first necessary to investigate the effect of the other variable, i.e., interaction rules and assessment methods. Once the extensive NDE was completed, various interaction rules were applied to the colony and a failure pressure calculated. The interaction rules considered were API579, CEPA, BS7910 and Leis and Mohan (1993). The failure pressures were calculated using the API 579 Level II and with CorLAS software package. Material data was obtained from tangential coupons (Table 1).
Referring to Table 1, a negative percentage error represents under-conservatism and positive errors conservatism. It can be seen that the most accurate failure pressures are calculated from the deepest crack within the colony. However, two of the three readings are under-conservative and--since cracks are known to coalesce--it was the opinion of the authors that interaction must be accounted for in some form. Of the interaction methods CEPA gave the most accurate results. There was only a very small variation in the predicted failure pressure based upon the different assessment methods.
CEPA Interaction Rules
CEPA interaction rules were used in the remainder of the project which focused on improving the method in which the ILI data is reported and assessed. CEPA  considers two cracks to interact under the following conditions:
sy [less than or equal to] 0.14(c1 + c2) or
sx [less than or equal to] 0.25 (c1 + c2)
sy = circumferential separation (normal to the length direction) between cracks
sx = axial separation between cracks 2cl, 2c2 are the lengths of the two cracks
By applying signal-filtering techniques, it was attempted to gain an accurate crack map of the SCC colony. These were refined and optimized until the maps produced were accurate representations of the actual colonies. CEPA interaction rules were applied to the USCD data and a failure pressure was calculated, based upon the interacting lengths and maximum depths (Table 2).
As can be seen from Table 2 the signal-filtering techniques enabled the conservatism of the predicted failure pressure to be reduced significantly with the upper bound error from actual failure pressure reducing from 39% to 14%. It should be noted that only three samples were used which may not be representative of all types of SCC colonies. Work is being undertaken to increase this sample size and the confidence associated with failure pressures calculated from USCD ILI data.
As for all defect dimensions reported by the USCD tool, some of the inputs to the sizing model can be run-specific, e.g., medium, so it is recommended that a series of verification digs he performed at this stage. This method does prioritize the SCC colonies against each other; therefore, the more critical features can be selected for verification.
Relative Ranking Case Study
The methods in this article do not take into account the tool tolerances. For this reason, it is not recommended to calculate a predicted failure pressure until some field verification has taken place. The methods described here can be used to relatively rank the pipeline anomalies against each other for that pipeline. This will highlight the more critical features in relation to the immediate integrity of the pipeline and provide guidance in selecting a sample of features for field verifications. Once these results are correlated to the field results, a mitigation and remediation schedule can be determined.
To relatively rank the features against each other an estimated severity factor (ESF) is used. This is calculated using an FAD Level II fracture assessment, the MAOP and the pipeline properties. The defect with the highest ESF is considered the most critical.
The following example is taken from a 26-inch, x52 pipeline. The ILI data was used to calculate an estimated severity factor for each feature using both the traditional method and the advanced analysis technique described here, (Figure 6).
[FIGURE 6 OMITTED]
Although the ESF is non-dimensional, using the advanced analysis techniques results in a lower estimated severity. And by providing more information about the cracks, a better prioritization is enabled. Both methods identified the same two flaws as the most critical to the immediate integrity of the pipeline. The flaw ranked third by the traditional analysis was ranked thirteenth by the advanced analysis.
After studying this colony, the longest interacting length was measured at 6 inches as opposed to the reported 16 inches of the overall length of the colony. This resulted in the decrease in severity when compared to other flaws in the pipeline. Once field verifications are completed, results are incorporated into the engineering assessment and the process is repeated.
This article is based on a presentation at the Rio Pipeline Conference & Exposition 2007, held Oct 2-4, 2007, in Rio de Janeiro.
By Kevin Spencer, PII Pipeline Solutions; Shahani Kariyawasam, TransCanada PipeLines; and Udayasankar Arumugam, PII Pipeline Solutions
Authors: Kevin Spencer is senior pipeline integrity specialist for PII Pipeline Solutions Business of GE Oil & Gas.
Shahani Kariyawasam is principal engineer, TransCanada Pipelines.
Udayasankar Arumugam is integrity engineer, PII Pipeline Solutions Business of GE Oil & Gas.
Table 1: Results of the interaction and assessment comparison. Interaction Rules Deepest Crack within the Colony Assessment Method API 579 Experimental Predicted Failure Failure Identification Pressure Length Depth Pressure Percentage Number (Mpa) (mm) (mm) (Mpa) Error 0092-08702 10.9 26 4 10.84 0.3% 0092-08694 11.1 28 4 10.77 3% 0092-08717 10.2 59 4 9.86 4% Interaction Deepest Crack within Rules the Colony Assessment CorLAS Method Experimental Predicted Failure Failure Identification Pressure Pressure Percentage Number (Mpa) (Mpa) Error 0092-08702 10.9 11 -1% 0092-08694 11.1 11 1% 0092-08717 10.2 10.6 -3% Interaction Rules API 579 Interaction Criteria Assessment API 579 Method Experimental Predicted Failure Failure Identification Pressure Length Depth Pressure Percentage Number (Mpa) (mm) (mm) (Mpa) Error 0092-08702 10.9 90 4 9.22 15% 0092-08694 11.1 115 4 8.86 20% 0092-08717 10.2 195 4 8.2 20% Interaction Rules CEPA Interaction Criteria Assessment API 579 Method Experimental Predicted Failure Failure Identification Pressure Length Depth Pressure Percentage Number (Mpa) (mm) (mm) (Mpa) Error 0092-08702 10.9 36 4 10.5 3% 0092-08694 11.1 65 4 9.71 12% 0092-08717 10.2 64 9.74 5% Interaction CEPA Interaction Rules Criteria Assessment Method CorLAS Experimental Predicted Failure Failure Identification Pressure Pressure Percentage Number (Mpa) (Mpa) Error 0092-08702 10.9 10.4 5% 0092-08694 11.1 9.86% 11% 0092-08717 10.2 9.65 6% Interaction Rules B57910 Intraction Criteria Assessment API 579 Method Experimental Predicted Failure Failure Identification Pressure Length Depth Pressure Percentage Number (Mpa) (mm) (mm) (Mpa) Error 0092-08702 10.9 41 3 10.5 3% 0092-08694 11.1 n/a n/a n/a n/a 0092-08717 10.2 n/a n/a n/a n/a Interaction Rules Leis and Mohan (1993) Assessment Method API 579 Experimental Predicted Failure Failure Identification Pressure Length Depth Pressure Percentage Number (Mpa) (mm) (mm) (Mpa) Error 0092-08702 10.9 n/a n/a n/a n/a 0092-08694 11.1 115 4 8.86 20% 0092-08717 10.2 195 4 8.2 20% Table 2: Results of the different filtering techniques applied to the ILI data. Defect Size Measurement Original ILI Crack Field Length and and Interaction Max. Depth Dimensions Experimental Predicted Failure Failure Identification Pressure Length Depth Pressure Percentage Number (Mpa) (mm) (mm) (Mpa) Error 0092-08702 10.87 334 3.7 7.81 28% 0092-08694 11.09 190 4.6 7.7 31% 0092-08717 10.24 1274 4.6 6.29 39% Defect Size Measurement ILI Crack Field Details with CEPA and Interaction Interaction--1st Iteration Experimental Predicted Failure Failure Identification Pressure Length Depth Pressure Percentage Number (Mpa) (mm) (mm) (Mpa) Error 0092-08702 10.87 162 3.7 8.62 21% 0092-08694 11.09 190 4.6 7.7 31% 0092-08717 10.24 1080 4.6 6.3 38% Defect Size Measurement ILI Crack Field Details with CEPA and Interaction Interaction--2nd Iteration Experimental Predicted Failure Failure Identification Pressure Length Depth Pressure Percentage Number (Mpa) (mm) (mm) (Mpa) Error 0092-08702 10.87 152 3.7 8.69 20% 0092-08694 11.09 168 4.6 7.89 29% 0092-08717 10.24 1080 4.6 6.3 38% Defect Size Measurement ILI Crack Field Details withe CEPA and Interaction Interaction--optimized Experimental Predicted Failure Failure Identification Pressure Length Depth Pressure Percentage Number (Mpa) (mm) (mm) (Mpa) Error 0092-08702 10.87 51 3.7 10.14 7% 0092-08694 11.09 46 4.6 10.08 9% 0092-08717 10.24 100 4.6 8.77 14% Defect Size Measurement and Interaction NDE Results with CEPA Interaction Experimental Predicted Failure Failure Identification Pressure Length Depth Pressure Percentage Number (Mpa) (mm) (mm) (Mpa) Error 0092-08702 10.87 37 3 10.65 2% 0092-08694 11.09 54 4 9.98 10% 0092-08717 41.5 n/a 69 n/a 10.24 175 * 4 8.32 19%
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||ultrasonic crack detection inline inspection|
|Comment:||Improving interpretation of SCC anomalies found by ultrasonic ILI.(ultrasonic crack detection inline inspection)|
|Author:||Spencer, Kevin; Kariyawasam, Shahani; Arumugam, Udayasankar|
|Publication:||Pipeline & Gas Journal|
|Date:||Jun 1, 2008|
|Previous Article:||Improved methodology for guided wave inspections of pipelines.|
|Next Article:||What do we mean by integrity management?|