Analysis of Full-Scale Burst Test Data by Examining Effects of Insulation and Using Fully-Instrumented Standard Pressed-Notch and Modified Backslot DWTT Specimens

An intensive effort was undertaken to understand the fracture behavior in a recent TCPL pipe burst test. The 48-inch diameter X80 pipe was buried in soil at the Spadeadam Test Site, but since it was desired to have the gas and pipe cooled to the minimum service conditions of −5°C, a 50-mm thick polyurethane foam (PUF) insulation was sprayed on the entire pipe test section. This was a reasonable precaution since freezing of a high water content soil around a large-diameter pipe burst test can require significant cooling capacity or a much longer duration to get to the burst test conditions. In this burst test, the crack propagated much farther than anticipated by traditional predictive approaches such as using Charpy energy to predict the minimum ductile fracture arrest toughness, assuming that soil backfill conditions existed. To explore this burst test behavior, two aspects were examined. The first was an assessment of the properties of the PUF insulation relative to the soil properties, and the second was the toughness evaluated by instrumented DWTT testing. The 50-mm thickness of the PUF insulation corresponded to about 8.3% of the pipe radius. In past loose-fitting steel sleeve crack arrestor burst tests, if the radial clearance was greater than 5.5% of the pipe, then a ductile fracture propagated under the arrestor regardless of its length with no change in speed. Hence if the PUF could be easily compressed, then the pipe in this burst test would behave as if it was in a non-backfilled condition. Non-backfilled pipe requires much higher toughness to arrest a ductile fracture. So perhaps the pipe in this burst test condition acted somewhere between non-backfilled and backfilled conditions — an aspect that might need a much more comprehensive computational model to better assess. Additionally to assess how material toughness played a role in this burst test, detailed instrumented toughness testing was conducted on material taken from several of the pipe lengths in the burst test. The evaluations of the Charpy energy to the DWTT energy suggested that one of the pipe materials may have behaved more like an X100 steel than an X80 steel. Correction factors on the predicted arrest toughness are well known to be needed for the Battelle Two Curve method (BTCM) when applied to X80 and X100 pipes. However, even with these corrections on the Charpy energy, arrest was predicted with soil backfill in several cases where in the actual test, the crack propagated through the pipes. Hence toughness corrections by themselves did not explain the test results. Additional calculations were then done assuming a non-backfilled condition (as suggested from the PUF property evaluation) along with the appropriate grade effect correction from the DWTT testing, and propagation was properly predicted in each case consistently with the burst test. So the fracture behavior in this burst test was somewhere between those of backfilled and non-backfilled pipe. As a result of this investigation it appears that the PUF insulation played an important role in crack arrest behavior, and because of its presence may have required much higher toughness than was actually needed for the actual service conditions of pipe buried in actual soil backfill.