6 th Pipeline Technology Conference 2011 TOPIC Advanced Assessment of Pipeline Integrity Using ILI Data AUTHORS Dr. Ted L. Anderson Daniel J. Revelle Quest Integrity Group 2465 Central Ave., Suite 110 Boulder, CO 80301 USA ABSTRACT Improvements in in-line inspection (ILI) and computing technology, coupled with the emergence of fitness-for-service standards, have created an opportunity to advance pipeline integrity assessment. This paper describes innovative approaches for assessing cracks, wall loss and dents in pipelines using data from ILI tools. Crack detection ILI tools that rely on shear wave UT have improved significantly in both detection probability and sizing accuracy. Quest Integrity Group uses realistic fracture mechanics models that utilize 3D elastic-plastic finite element analysis. The combination of advanced modeling and reliable in-line inspection provides a superior alternative to hydrostatic testing for ensuring pipeline integrity. Inline inspection tools that measure wall loss with compression wave UT provide superior results compared to other methodologies. The former outputs a digital map of individual thickness readings, which is ideally suited to effective area assessment methods such as RSTRENG and the API 579 Level 2 Remaining Strength Factor (RSF) calculation. Quest has developed software that can rapidly process large quantities of ILI wall loss data and evaluate the maximum allowable operation pressure (MAOP) at discrete locations. The ranking of these MAOP values serves as a rational and rapid means for prioritizing the severity of corrosion throughout the line. Traditional dent assessments are based on a simplistic characterization of the dent (e.g. the ratio of the dent depth to the pipe diameter), combined with a simple empirical equation. Quest Integrity has developed an advanced dent assessment that combines a detailed mapping of the dent from ILI data (either UT or a caliper pig) with
3D elastic-plastic finite element analysis. This advanced methodology can be applied to interacting anomalies such as dent/gouge and dent/crack combinations, and can be used to demonstrate the fitness-for-service of difficult-to-repair pipelines. 1. OVERVIEW Advances in in-line inspection (ILI) technology have led to enhancements in both the quality and quantity of pipeline inspection data. Corresponding improvements in fitness-for-service assessment methods and technology are necessary to take full advantage of inspection data with higher resolution and higher accuracy. Assuming that integrity management resources are finite, assessing individual pipeline defects with an overly conservative methodology can lead to an unconservative integrity management program. For example, if an overly conservative assessment methodology leads to a significant number of unnecessary repairs and remediation in a particular line, there is less budget available to inspect other lines that may have more serious issues. Accurate assessment methodologies enable operators to optimize their expenditures on integrity management. The fitness-for-service standard API 579-1/ASME FFS-1 [ 1] provides a comprehensive guideline for assessing various flaw types and damage mechanisms in all pressure equipment including pipelines. This standard incorporates three levels of assessment: Level 1. This is a basic assessment that can be performed by properly trained inspectors or plant engineers. A Level 1 assessment may involve simple hand calculations. Level 2. This assessment level is more complex than Level 1, and should be performed only by engineers trained in the API/ASME FFS standard. Most Level 2 calculations can be performed with a spreadsheet. Level 3. This is the most advanced assessment level, which should be performed only by engineers with a high level of expertise and experience. A Level 3 assessment may include computer simulation, such as finite element analysis (FEA) or computational fluid dynamics (CFD). This division of assessment level methodology is not unique to API 579. Similar tiers can be found in ASME B31G, DNV-RP-F101 and other codes and practices. These three assessment levels represent a trade-off between simplicity and accuracy. The simplified assessment procedures are necessarily more conservative than more sophisticated engineering analyses. With Level 1 assessments, the specified procedures must be followed exactly, and there is little or no room for interpretation. Level 2 procedures provide some latitude to exercise sound engineering judgment. For
Level 3 assessments, the API/ASME standard provides a few overall guidelines, but the details of the assessment are left to the user. The lack of specificity in Level 3 is by design. There is no practical way to codify step-by-step procedures for advanced engineering analyses because every situation is different, and there a wide range of approaches that may be suitable for a given situation. The combination of Level 3 fitness-for-service technology and high-fidelity ILI data makes accurate predictions of burst pressure and remaining life feasible. In certain instances, simplified assessments are not sufficient. In the case of crack assessments, for example, supposedly conservative analyses have led to unconservative predictions in some cases. Quest Integrity Group has recently developed advanced assessment techniques for cracks, wall loss, and dents. Level 3 assessments that incorporate elastic-plastic finite element analysis are used for cracks and dents. We have adapted the API/ASME Level 2 assessment for wall loss in order to process large quantities of ILI compression wave UT data. Each of these advanced assessments is described below. 2. LEVEL 3 CRACK ASSESSMENT AS AN ALTERNATIVE TO HYDROSTATIC TESTING Traditional models for crack assessment are considered conservative because they tend to underestimate burst pressure and critical crack size. One such approach is the NG-18 method [ 2], which dates back to the early 1970s and is still widely used today. So-called conservative methods such as NG-18 can actually be unconservative in some instances, as described below. Hydrostatic testing has traditionally been used to protect pipelines against unexpected failures from cracks or other planar flaws. The hydrostatic test is designed to detect critical flaws by causing leaks and ruptures under controlled conditions. In many cases, the NG-18 equation has been used to estimate the critical flaw dimensions at the test pressure. If the pipe passes the hydrostatic test, it is assumed that no flaws larger than the calculated critical dimensions are present. However, this assumption is not justified because the NG-18 equation and other simplified models typically underestimate the critical flaw size. Figure 1. Schematic comparison of predicted and actual critical flaw size for a hydrostatic test. The conservative analysis under-predicts the maximum flaw sizes that survive the hydrostatic test.shows a bell curve that represents the population of crack-like flaws in a pipeline. If a hydrostatic test is performed on this line, cracks on the upper tail of the bell curve will be identified, as indicated by the area shaded in red. The NG-18 equation significantly
under-predicts the critical crack size. The yellow shaded area in the plot represents the population of flaws that were predicted to fail the test but did not. In other words, largerthan-predicted cracks are left in the pipe following a hydrostatic test. The scenario that is schematically illustrated in Figure 1 is demonstrated with actual data below. A 16-inch Schedule 10 pipeline, which was installed in 1955, has experienced hook cracking in Electric Resistance Welded (ERW) seams. These cracks have grown over time by fatigue due to pressure cycling. As a result of several in-service failures, the operator instituted a hydrostatic testing program in 1991. The NG-18 equation was used to predict the critical flaw dimensions at the test pressure. A fatigue crack propagation analysis was then performed on the calculated critical flaw sizes in order to infer an appropriate retest interval. The most recent full-line hydrostatic test on this pipeline was performed in 1999. The corresponding critical flaw calculation from the NG-18 equation is represented by the blue curve in Figure 2. This pipe was inspected by a shear wave UT ILI tool in 2008. A total of 139 cracks were reported, 62 of which were sized by manual UT. The measured crack dimensions for these 62 flaws are plotted in Figure 2. The NG-18 equation predicts that 10 of these 62 flaws would have failed a follow-up full-line hydrostatic test, had it been conducted at the time of the 2008 inspection. We performed a reverse fatigue analysis of these 10 flaws in order to estimate their dimensions at the time of the 1999 test. The results of this exercise are plotted in Figure 3. Although the flaw dimensions were smaller in 1999 than in 2008 (red data points versus green data points), 8 out of 10 of the red data points lie above the critical flaw curve, as computed from the NG-18 relationship. The other two flaws fall on the curve. This analysis demonstrates that larger-than-predicted flaws survived the 1999 hydrostatic test, which is consistent with the schematic in Figure 1. Underestimating the critical flaw size for a hydrostatic test is potentially unconservative. Large flaws grow faster than small flaws, so an underestimate of the maximum flaw sizes that survived the test can result in an overestimate of the safe operating interval between tests. The 1970s vintage NG-18 equation is incapable of accurate predictions of critical flaw size or burst pressure. A state-of-the-art Level 3 crack analysis provides a much more accurate reflection of reality. Quest Integrity has applied a Level 3 assessment to the 16-inch pipeline described above. Our assessment procedure contains the following features: Three-dimensional elastic-plastic finite element models of cracks in ERW seams. Fracture toughness inferred from laboratory tests on samples extracted from the pipe of interest. Weld residual stress computed from a finite element simulation of the ERW process.
Figure 4 shows a typical 3D model of a crack in an ERW seam. A total of 35 such analyses were run for the 16-inch ERW pipe, which encompassed a wide range of crack dimensions. Figure 5 is a repeat of the comparison between predicted and measured flaws in Figure 2, but with predictions based on the Level 3 assessment. This analysis indicates that 4 flaws were marginal at the time of the 1999 full-line hydrostatic test. That is, they barely survived the 1999 test. These flaws would almost certainly have failed a follow-up test. It is fortunate that these flaws did not lead to in-service failures in the 9-year period between the full-line hydro and the ILI tool run. The 16-inch line discussed above was due for a full-line hydrostatic test in September 2009, but the operator received a temporary deferment from the US Department of Transportation (DOT). Quest Integrity is working with the operator to validate an alternative to hydrostatic testing that is based on a combination of ILI and Level 3 crack assessment. Pending the results of this study, the DOT may permit the operator to permanently replace the existing hydrostatic testing program with the alternative strategy. Hydrostatic testing is a very expensive but ineffective means for identifying cracks and other planar flaws in pipelines. Figure 6 schematically compares the relative effectiveness of hydrostatic testing versus ILI. The former identifies only the largest flaws, while the current generation of shear wave ILI tools can detect very small flaws. For example, of the 139 reported cracks from the 2008 ILI of the 16-inch pipe, only 4 or 5 of these cracks would have failed a full-line hydrostatic test. Given the ILI data, a Level 3 analysis can be used to establish repair criteria and re-inspection intervals. This alternative strategy provides a greater degree of reliability at a significantly lower cost compared to the traditional hydrostatic testing approach. The shear wave UT ILI tool used to inspect the 16-inch ERW pipe has a 90% probability of detection for cracks greater than 40 mils (1 mm) in depth. Thus, this tool is far more sensitive at detecting flaws compared to hydrostatic testing. However, there is still room for improvement on flaw sizing accuracy with shear wave ILI data. In the case of the inspection on the aforementioned 16-inch pipe, flaw depths were reported in ranges: 40-80 mils (1-2 mm), 80-160 mills (2-4 mm), and > 160 mils. While flaws shallower than 40 mils (1 mm) can be detected, such indications were not reported because it is difficult to distinguish cracks from extraneous reflections from the ERW seam. Figure 7 and Figure 8 are plots of the measured flaw depths for cracks reported in the 40-80 and 80-160 mil ranges, respectively. For flaws reported in the 40-80 mil range, the manual UT measurements exhibit a significantly wider range of crack depths
compared to the reported range. Note that two 20-mil (0.5 mm) deep cracks were reported, which is an indication of the high sensitivity of the ILI tool. For the 80-160 mil depth range, the measured flaw depths generally fall within the reported range. This indicates that sizing accuracy with ILI shear wave UT data is better for deeper cracks. Both populations of flaws (40-80 and 80-160-mil reported ranges) follow Weibull statistical distributions. Given the uncertainty between the actual depth of a given flaw and the reported range from the ILI data, a probabilistic analysis is recommended. 3. RAPID ASSESSMENT OF METAL LOSS WITH COMPRESSION WAVE UT ILI DATA Metal loss in pipelines has traditionally been assessed with the ASME B31G Modified and Effective Area [ 3] methods. Given an ILI dataset covering several hundred kilometers of pipe, a manual data analysis taking up to 3 months is typically performed prior to assessing the wall loss and applying acceptance criteria. A primary purpose of this initial analysis is to identify and size discrete corrosion flaws. In addition to the time and cost associated with this painstaking process, a major problem with this approach is that reality seldom conforms to the ideal of discrete areas of wall thinning surrounded by uncorroded metal. Instead, wall thickness in a corroded pipe can vary continuously over the surface and a set of interaction criteria can be used to determine whether or not these flaws can be evaluated independently. This reality is evident in high-resolution compression wave UT data, which, unlike MFL data, results from a direct measurement of small portions of the larger corroded area. Figure 9 and Figure 10 compare the ideal of discrete flaws with a color map of actual UT wall thickness data. Part of the UT data analyst s job is to take the non-ideal wall thickness data and force-fit it to the discrete flaw ideal. The process is often referred to as flaw boxing, as the analyst defines the length and width of the flaw with a rectangle that contains the corresponding wall loss data. When applying the B31G Modified acceptance criteria, the only measurements that are used in the assessment are the length and width of the boxed flaw, along with the minimum measured wall within the box. In such cases, over 99% of the wall thickness data is discarded, and a key advantage of high-resolution UT data relative to MFL is lost. The Level 2 assessment of metal loss in API 579-1/ASME FFS-1 2007 [ 1] is an effective-area method that is similar to the B31G Effective Area methodology [ 3]. Flaw boxing is not required with the API/ASME method, however. A river-bottom profile is constructed from the thickness data, and a remaining strength factor (RSF) is computed, which can be used to compute a maximum allowable operating pressure (MAOP). These calculations can be performed over a short segment of pipe, or a single
MAOP can be computed for an entire pipe joint between girth welds. All valid wall thickness readings are considered with this assessment method. Any interaction between neighboring defects is determined explicitly during the iterative calculations. This approach is not only less labor intensive than flaw boxing, it is much less subjective, and results in a more technically sound MAOP. Long shallow corrosion defects are particularly suited for this type of assessment as reductions in burst pressures do not always track with reportable defect criteria. Being able to accurately map the load carrying capabilities of the pipe can eliminate unnecessary repairs, freeing up resources to detect and repair more critical defects elsewhere in a piping system. Quest Integrity has developed the LifeQuest TM Pipeline software to process and visualize data from high-resolution compression-wave UT ILI tools, including our InVista TM intelligent pigs [ 4]. LifeQuest TM performs a Level 2 API/ASME wall loss assessment over an entire ILI dataset, and computes the RSF and MAOP for each pipe section. The areas of highest corrosion damage can be quickly identified by ranking the calculated RSF and MAOP values. Figure 11 is a screen shot from LifeQuest TM Pipeline. 4. LEVEL 3 DENT ASSESSMENT Pipe denting is a sufficiently complex phenomenon for which Level 3 assessment technology is warranted. Significant plastic strain occurs when the dent first forms. The pipe tends to re-round upon pressure cycling, such that the observed deformation understates the true damage that has accumulated in the pipe. The size, shape, and location of the original dent affect the remaining life, as do external factors such as the constraint provided by the surrounding soil. In order to handle the complexities associated with dents, Quest Integrity has developed a Level 3 assessment methodology that relies on elastic-plastic finite element simulation. The formation of the dent is simulated, along with the subsequent pressure cycling. The support of the surrounding soil is incorporated as appropriate. The remaining life is computed through a proprietary low-cycle fatigue damage model that has been incorporated into the elastic-plastic finite element simulation. Dimensional data from ILI can be used to build 3D finite element models of dented pipes. The prior damage created during the initial denting, however, must be taken into account. We have performed parametric studies to infer the relationship between the current dimensions and the as-dented configuration. Elastic-plastic finite element simulation can also be used to model interacting anomalies, such as a crack in a dent.
Figure 12 shows a typical 3D finite element model of a pipe after the formation of a dent. Figure 13 shows the same model after 10 pressure cycles. Note that the pipe has re-rounded. This effect can be modeled using a finite element analysis (FEA). As part of the model validation, FEA models were run in ABAQUS and results compared to experimental results presented in API publication 1156 [ 5, 6], as shown in Figure 14. 5. CONCLUSION Improvements in inline inspection (ILI) and computing technology, coupled with the emergence of fitness-for-service standards, have created an opportunity to advance the state-of-the-art in pipeline integrity assessment. High resolution in-line inspection data can be used to produce more accurate answers for assessment of corrosion, cracks and dents. Although it may seem that using a more conservative assessment methodology would produce a more conservative integrity management program, the opposite is more likely to be true. Integrity management budgets are finite, and unnecessary repairs divert resources that could be better used for inspection or assessment of other assets. Accurate knowledge of the condition of a piping system allows for more efficient and safer operation of these key assets. Acknowledgements Much of the work described in this paper was funded by Koch Pipeline Company. The authors would like to acknowledge the contributions of colleagues at Quest Integrity Group who have participated in the development of the advanced pipeline assessment technology described herein. These colleagues include Devon Brendecke, Chris Tipple, Greg Brown, Jim Rowe, and Greg Thorwald. 6. REFERENCES 1. API 579-1/ASME FFS-1, Fitness-for-Service, jointly published by the American Petroleum Institute and the American Society for Mechanical Engineers, June 2007.
2. Kiefner, J. F., Maxey, W. A., Eiber, R. J., and Duffy, A. R., Failure Stress Levels of Flaws in Pressurized Cylinders. ASTM STP 536, American Society for Testing and Materials, 1973. 3. "A Modified Criterion for Evaluating the Remaining Strength of Corroded Pipeline", Pipeline Research Council International (PRCI)/AGA, Contract Number: PR-3-805, Catalog Number: L51688. 4. Papenfuss, S., Pigging the UNPIGGABLE : New Technology Enables In-Line Inspection and Analysis for Non-Traditional Pipelines 5 th MENDT Conference, Bahrain, November 2009. 5. Kiefner JR., Alexander CR., Effects of Smooth and Rock Dents on Liquid Petroleum Pipelines, API Publication 1156, The American Petroleum Institute, November 1997. 6. Kiefner JR., Alexander CR., Effects of Smooth and Rock Dents on Liquid Petroleum Pipelines (Phase II), API Publication 1156 Addendum, The American Petroleum Institute, October 1999.
7. FIGURES Figure 1. Schematic comparison of predicted and actual critical flaw size for a hydrostatic test. The conservative analysis under-predicts the maximum flaw sizes that survive the hydrostatic test.
Crack Depth, in 0.25 0.2 Comparison of Predicted Critical Flaw Size with Actual Detected Flaws Based on the NG-18 Methodology Computed Critical Flaw Size (1999 Full Line Hydro) Actual Detected Flaws (2008) 0.15 The NG-18 equation predicts that 10 out of 62 flaws would have failed a 2008 full-line hydrostatic test. 0.1 0.05 0 0 2 4 6 8 10 12 14 16 Crack Length, in Figure 2. Comparison of predicted maximum flaw sizes that survived the 1999 hydrostatic test with actual measured flaws following a 2008 ILI tool run. The NG-18 equation was used for critical flaw predictions.
Crack Depth, in 0.25 0.2 Comparison of Predicted Critical Flaw Size with Actual Detected Flaws Based on the NG-18 Methodology Computed Critical Flaw Size (1999 Full Line Hydro) Measured Flaw Size in 2008 Estimated Flaw Size in 1999 0.15 The NG-18 equation predicts that 8 to 10 flaws should have failed the 1999 hydrostatic test. 0.1 0.05 0 0 2 4 6 8 10 12 14 16 Crack Length, in Figure 3. Comparison of predicted critical flaw size for the 1999 hydrostatic test with the calculated dimensions of the 10 worst flaws in 1999.
Figure 4. Finite element model of a crack in an ERW seam. The model is ¼ symmetric.
Crack Depth, in 0.25 0.2 Comparison of Predicted Critical Flaw Size with Actual Detected Flaws Based on the Level 3 Methodology Computed Critical Flaw Size (1999 Full Line Hydro) Measured Flaw Size in 2008 Estimated Flaw Size in 1999 0.15 0.1 0.05 0 Four borderline cracks survived the 1999 full-line hydrostatic test. 0 2 4 6 8 10 12 14 16 Crack Length, in Figure 5. Repeat of Fig. 3, but with critical flaw size predictions based on the Quest Integrity Level 3 assessment.
Cumulative Probability Figure 6. Comparison of ILI crack detection capabilities with the ability of hydrostatic testing to identify cracks. 1 0.9 Actual versus Reported Flaw Depth 40-80 mil Reporting Range Reported Depth Range 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 20-mil deep cracks detected by ILI Inspection Data Weibull Fit 0 0 20 40 60 80 100 120 140 Measured Flaw Depth, mils Figure 7. Measured depths (with manual UT) of flaws reported to be within the 40-80 mil (1-2 mm) range based on ILI UT data.
Cumulative Probability 1 0.9 Actual versus Reported Flaw Depth 80-160 mil Reporting Range Reported Depth Range 0.8 0.7 0.6 0.5 0.4 Inspection Data Weibull Fit 0.3 0.2 0.1 0 0 20 40 60 80 100 120 140 160 180 200 Measured Flaw Depth, mils Figure 8. Measured depths (with manual UT) of flaws reported to be within the 80-160 mil (2-4 mm) range based on the ILI UT data.
Figure 9. Idealized case with discrete flaws surrounded by uncorroded material. Figure 10. Actual ultrasonic inspection data. This is a 2D unwrapped plot of wall thickness.
Figure 11 External corrosion viewed in LifeQuest Pipeline software
Figure 12. 3D Elastic-Plastic FEA: Dent formation
Figure 13. 3D Elastic-plastic FEA: dent re-rounding (10 cycles).
Radial Displacement (in) 0 Dent Profile, Axial Cross Section (Final Step of Analysis) -0.2-0.4-0.6-0.8-1 Figure B-13, API 1156 Phase II -1.2-1.4-1.6 0 2 4 6 8 10 12 14 16 18 Axial Length (in) 6%, Modified σ-ε curve 12%, Modified σ-ε curve 18%, Modified σ-ε curve 24%, Modified σ-ε curve Figure 14. Dent Re-Rounding: Simulation vs. Experiment.