A Review of Methods for Assessing the Remaining Strength of Corroded Pipelines

When corrosion features in pipelines are detected by in-line inspection (ILI), a decision to replace, repair, or accept and monitor each feature must be made. This decision is based on the prediction of the failure pressure of the corroded pipe and must be accurate without being overly conservative. The US Code of Federal Regulations (CFR) Title 49, Parts 192 and 195 stipulate that ASME B31G or RSTRENG be used to assess the remaining strength of corroded pipe. A review of burst test data by Pipeline Research Council International, Inc. (PRCI) in 2000 (PRCI Report Catalog Number L51878) raised concerns that use of these methods can, in some instances, result in predicted failure pressures that are greater than the recorded burst pressures from actual tests. For example, predicted failure pressures obtained on tests conducted on line pipe of strength grades X60 and above with machined defects of depths 50% to 80% of the pipe wall were greater than the actual values derived from burst tests.

Germanischer Lloyd (hereafter GL, formerly Advantica ) has been active in developing methods for assessing corrosion damage in pipelines for PRCI and more recently for the Pipeline and Hazardous Materials Safety Administration (PHMSA). GL also led a Group Sponsored Project (GSP) in the late 1990’s funded by 8 operators and 2 regulators that ultimately resulted in the development of assessment methods that are now embodied into internationally recognized fitness for service standards. PHMSA has requested GL to develop an improved understanding of the performance of methods such as ASME B31G, Modified ASME B31G, and RSTRENG in predicting failure pressures of corroded line pipe against a database of burst test results.

Key Results

Failure pressure predictions made by common assessment methods used by the pipeline industry, such as ASME B31G, Modified ASME B31G, RSTRENG and others, were compared against recorded burst test pressures. The assessment methods were applied to an extensive database of burst test results which includes the results of various tests conducted by GL on higher strength line pipe material (grade X80 and X100), which have previously not been published. The analysis of the performance of pipeline remaining strength assessment methods consisted of six (6) separate sensitivity studies or ‘cases’, representing different flow stress definitions and material properties. Sensitivity studies were undertaken for the six cases defined below:

The findings of this study conclude that when these methods are used in the application Case 2, the predicted maximum safe operating pressure values are lower than the actual burst pressures from all the tests. Other key findings include:

1. Case 1 shows that RSTRENG is the most accurate method for predicting the failure pressure of corroded pipelines because it results in the least amount of spread between model failure predictions and recorded burst pressures.
2. In the application Case 2 (with a factor of safety equal to 1.25 or 1.39), the ASME B31G, Modified ASME B31G, and RSTRENG methods give maximum safe operating pressure values lower than the actual burst pressure for all defects (both machined and real). This shows that these methods, when applied in practice, are conservative. A similar conclusion can also be drawn when using the LPC-1, SHELL92, and PCORRC methods.
3. In general, the results of tests conducted on pipe with machined defects exhibit less scatter than on pipe with real defects. This is as expected because machined defects are well-defined and regular compared to real corrosion defects. Tests on pipe with machined defects tend to give lower bound failure pressures.
4. When the shape factor is modified to account for differences between real and machined defects, the actual recorded failure pressures are greater than the calculated/predicted safe operating pressures (i.e., a Case 2 assessment) for line pipe up to grade X100.
5. The actual yield strength of grade X65 line pipe for five burst tests with real corrosion defects is greater than the specified minimum yield strength (SMYS) of grade X70 line pipe. For all these tests, the actual recorded burst pressure is greater than the predicted failure pressure using ASME B31G, Modified ASME B31G, and RSTRENG (Case 2). These results provide confidence in the use of existing assessment methods for line pipe up to grade X70.

Industry experts from PRCI member companies involved in reviewing this report recommended including a description of the conservatisms built in when assessing the remaining strength of corroded pipelines. When calculating the maximum safe operating pressure of a corroded pipeline, an appropriate level of conservatism is achieved through a three step process that involves 1). establishing the accuracy of the prediction model; 2). using specified minimum material properties and nominal values for pipe diameter and wall thickness; and 3). use of an additional factor of safety.

The research completed did not include analysis of burst test data on line pipe with real corrosion defects in strength grades above X65, as the data were not available.  To address this gap, a focused program of full-scale tests is recommended on higher strength line pipe of strength grades above X65 with electro-chemically induced, simulated corrosion defects. These defects can be produced using electrochemical means to approximate real corrosion in the field, as opposed to flat-bottomed rectangular machined patches. Failure pressure predictions using ASME B31G, Modified ASME B31G and RSTRENG should then be compared to the recorded burst test pressures to confirm that these methods are applicable for higher strength pipelines.