Author: Chris Owens
Create One Comprehensive Approach for Dealing with Coinciding Stresses
Combining assessment methods
In a Nutshell:
Pipelines can be subject to significant amounts of stress, which may affect the failure pressures of anomalies such as corrosion. Although there are a variety of assessments available, they often have limitations. In this article, Chris Owens discusses the possibility of combining various existing methods to provide one efficient and reliable assessment, as well as the validation of this approach.
We know that pipelines are typically assessed considering internal pressure loading, but pipelines can also be subject to additional axial compressive stresses, for example from thermal expansion, bending or ground movement. If these compressive stresses become significant, then they can interact with the internal pressure and lower the failure pressure of an anomaly. Therefore, unless these stresses are accounted for, some assessments may provide nonconservative results.
Methods have been developed to account for corrosion subject to combined loading, such as approaches developed by the Southwest Research Institute or the RPA-PLLC method developed by Petrobras. The most widely used method, however, is the DNVGL-RP-F101 compressive method.
This method does have some applicability limitations, though. Because it uses an area approximation based on the total corrosion length and peak depth of each anomaly being assessed, it does not account for the profile of the corrosion. When anomalies have complex profiles, this simple area approximation can result in an overly conservative estimate of the failure pressure, depending on the profile.
Conversely, the widely recognized assessment methods Modified ASME B31G and Detailed RStreng allow for the corrosion profile to be accounted for, but these assessments only evaluate failure due to internal pressure loading. In fact, in these cases, ASME B31G directs the user to “refer to a more comprehensive fitness-for-purpose guidance document.”
To solve both of these issues, we proposed applying the logic within the DNVGL assessment to the Modified B31G and Detailed RStreng assessments, so they could consider anomalies subject to combined loading while also taking into account the profile of the corrosion.
The technical bit
The DNVGL method assumes that the failure surface follows the shape of the Tresca failure criterion, and failure is deemed to occur when the effective stress equals the flow stress. The Tresca approach is based on engineering principles and is therefore not related specifically to the DNVGL failure pressure equation. Logically, we therefore believe it can also be applied to other methods.
The DNVGL compressive method initially estimates the failure pressure using the equation for internal pressure loading only. To account for the influence of additional compressive stresses, it applies a linear correction factor (H1) to the initial failure pressure estimate. When H1 is less than 1, the anomaly’s failure pressure is reduced due to the influence of the compressive stress. If the stress is not significant enough, then there is no reduction of the initial failure pressure estimate.
If you compare the DNVGL and Modified B31G failure pressure equations when considering internal pressure loading only, they are actually very similar. Three key differences are the flow stress, bulging factor and area assumption. When altering the Modified B31G equation, we adopted the Tresca based correction factor used by DNVGL.
Combining the standard Modified B31G equation for internal pressure loading only with the Tresca-based correction factor, we have a proposed new Modified B31G equation to use. The same logic can be applied to other methods, such as the Original B31G method, ensuring that the appropriate definitions for bulging factor and profile factor are used in accordance with those defined for the specific assessment methods. However, the method of most interest is the Detailed RStreng approach, which allows for the corrosion profile to be taken in to account.
While the initial comparison appeared promising, the method needed validation.
Validating the approach
To validate the results of the newly modified equations, we used the results of 27 full scale burst tests, with further support from a parametric study based on finite element analysis (FEA) that included 28 FEA simulations.
The parametric FEA study was particularly beneficial because full-scale tests typically only consider single flat-bottomed anomalies and the FEA allowed us to consider a much wider range of corrosion profile. This was an important consideration for the Detailed RStreng compressive method, as it allowed us to consider the corrosion profile.
For each simulation, an API 579 Level 3 strength assessment was conducted to estimate the failure pressure under combined loading (i.e., internal pressure and axial compression). The comparisons for each compressive method considering the full-scale test data and FEA simulations are shown in the below plots.
In the plots, the x-axis represents the failure pressure obtained from the full-scale tests or FEA, and the y-axis presents the failure pressure predicted by the different compressive assessment equations. Points below the red line show that the predicted failure stress from the equations is conservative with respect to the FEA or full-scale test results.
The predicted failure stresses using simple area approximations (DNVGL and Modified B31G) are comparable, and both have a relatively high level of conservatism when compared to the FEA simulations. This is expected, as the FEA cases were specifically designed to consider more complex corrosion profiles. When comparing these to the failure stress predictions using the Detailed RStreng method, it is clear the Detailed RStreng results were notably less conservative than to the FEA simulations, as the corrosion profiles were able to be taken into account.
Based on these findings, the proposed modified combined loading methods were considered to give safe predictions.
Case Study
To demonstrate the potential benefit of using a profile-based assessment when there is significant compressive loading, a case study was considered for a 20” pipeline operating with a maximum temperature of 90°C. This line was inspected using an ultrasonic wall thickness measurement (UT) tool, which identified extensive external corrosion around the pipe circumference, with a large number of complex areas of corrosion that were typically axially long and with significant circumferential extents and varying depth profiles. The example image shows a representative corrosion cluster reported by the UT ILI tool; the UT C-scan is shown above, and the associated river bottom profile is below.
The number of predicted repairs/investigations was calculated using the DNVGL and Detailed RStreng compressive loading methods. The results displayed in the below table show a clear potential benefit using the newly proposed methods.
Conclusion
To assess corrosion subject to combined internal pressure and compressive loading while still considering the corrosion profile, modifications have been made to the Modified B31G and Detailed RStreng approaches. The results of these modifications were validated against full-scale test data and FEA simulations, and both modified results were found to provide safe predictions. The Detailed RStreng approach, in particular, was found to give safe predictions, but it also resulted in less-conservative results than the simple area approximations in DNVGL or Modified B31G. This enables an effective area method to be used when compressive loading is significant.