Crack Assessment: Are you Ready for Your ILI Results?
Prior to running a crack ILI it is very important to be prepared to perform your post ILI assessment. Of course this involves determining the burst pressure of crack-like anomalies identified but also much more.
ILI tools are able to detect crack and crack-like anomalies. The most common are weld anomalies such as hook cracks and lack of fusion as well as pipe body cracks such as SCC. While detection and sizing capabilities have improved there is still significant uncertainty associated with both sizing and characterization of crack-like anomalies.
First a quick primer on crack assessment. A crack or crack-like anomaly is a non-blunt flaw that can fail through flow-stress or toughness controlled modes. In the flow-stress controlled failure the anomaly will behave similarly to metal loss and failure is based upon strength properties. Toughness controlled failures will have burst pressures lower than a metal loss anomaly of the same dimensions and failure occurs when the crack driving force is greater than the material resistance or toughness.
Choosing the Fracture Model
One element that should not be underestimated is the importance of selecting a fracture model as it provides the prediction of failure pressures and critical flaw sizes. There is a wide variety of models in use and they vary from older empirical or semi-empirical models such as NG-18 (aka modified log secant) or CORLASTM to failure assessment diagrams such as API579 or BS7910 to a more rigorous physics based approach such as PRCI MAT-8.
The effectiveness of the fracture model needs to be evaluated for both flow stress and toughness controlled regimes as empirical models tend not to be as effective in the toughness controlled regime.
There is a trade off between the accuracy of the model and the complexity of the equation. More rigorous models, not surprisingly, may require fitness-for-service software to perform the burst pressure calculation.
Material property inputs, specifically toughness, are different between empirical and FAD and rigorous models. Empirical models rely upon Charpy impact values and the more rigorous models use fracture toughness although correlations between Charpy impact and fracture toughness can allow for Charpy values to be used.
To perform an assessment material properties including toughness are required. Most of the values can simply be based upon material specifications. Toughness is often not included in those specifications and is typically more representative of steel manufacture quality rather than grade itself. Toughness values can be obtained from material testing but industry databases can also be used.
In toughness-controlled failures the toughness values have a large impact upon burst pressure failures and careful consideration should be given to matching toughness values with the desired conservatism. Toughness values tend to vary more than strength values and toughness can be represented as a distribution in probabilistic assessments.
ILI Tool Accuracy
Both sizing and characterizing cracks by ILI tools is difficult and that needs to be considered in the assessment. Tool sizing accuracy, especially depth, should be considered in the assessment to ensure that conservatism is maintained at the appropriate level.
It should be noted that multiple levels of conservatism, such as considering both lower bound toughness plus adding depth sizing tolerance, may lead to overly conservative assessments. Sensitivity studies can assist in developing an assessment methodology that is in accordance with desired levels of conservatism.
It is best practice to perform validation of the ILI predictions using dig results on an on-going basis. This allows for assumptions used in the assessment to be evaluated and the assessment to be modified based upon actual accuracy of the run.
Flaw Acceptance Curves
A family of acceptance curves based upon a series of failure pressure ratios can be developed in anticipation of receiving the ILI results. The acceptance curves simply uses the fracture model to determine critical flaw curves using the material properties and pipeline M(A)OP.
The advantage of flaw acceptance curves is that flaw dimensions can simply be plotted on the curve and the correct response then determined.
If flaw acceptance curves are not used the burst pressure would be determined for each crack-like anomaly to determine the appropriate response.
Crack growth in the form of pressure cycle fatigue for seam flaws and SCC growth needs to be considered for anomalies not addressed through site investigation. The growth assessment needs to ensure that the crack does not reach critical size before the next inspection occurs (including the appropriate factor of safety).
Crack growth is another area where tool performance needs to be considered. Based upon the results of the ILI validation the largest remaining flaw size(s) should be estimated and the fatigue life determined and compared to the next scheduled inspection.