Method for Establishing Hydrostatic Re-Test Intervals for Pipelines With Stress-Corrosion Cracking

Hydrostatic testing is one way to demonstrate the integrity of a pipeline that may contain stress-corrosion cracks. In order to establish appropriate intervals for such tests, it is necessary to make a reasonable assumption about the probable maximum growth rate of cracks that might exist in the pipeline. Although growth rates have been measured in laboratory experiments, those rates are not meaningful for a buried pipeline, because the growth rate depends upon many unknown factors, such as the condition of the coating, the composition of any liquid in contact with the pipe, the susceptibility of the steel, and the temperature. However, it is possible to infer what a probable maximum growth rate is, from the hydrostatic-test history of a portion of a pipeline. This paper describes a method for establishing hydrostatic-test intervals based upon the assumption that cracks that already led to a service failure or hydrostatic-test failure had a higher growth rate than surviving cracks. That assumption is reasonable, because the cracking conditions at the failed cracks must have been more severe than the conditions around any surviving or future crack. The method does not require any knowledge about the nature of the chemical environment at the surface of the pipe, the susceptibility of the steel, or whether the cracks are high-pH stress-corrosion cracks or near-neutral-pH stress-corrosion cracks. The only data that are required are probable maximum values for the actual yield strength and the actual ultimate tensile strength, which usually can be determined from mill records. Using this method, it can be shown that the interval lengths are strongly affected by the test pressure. It also can be shown that uniform test intervals are less effective than graduated intervals. In fact, subsequent intervals may be longer than previous ones, even if hydrotest failures occurred in the previous tests. The validity of this method was demonstrated by applying it, in principle, to actual historical data from over a dozen valve sections that have been subjected to multiple hydrostatic tests and showing that, if this method had been used, more failures would have been prevented with fewer tests.