Helms Project penstock

Tom's most-developed case in the session. Eight-and-a-half-month delayed hydrogen crack on a 22-foot-diameter PG&E penstock at 6,000 ft elevation. Used to illustrate (a) that hydrogen crack delay can extend far beyond the standard "first couple of weeks" rule if the weld is frozen, (b) that boiler-and-pressure-vessel-code exceptions for inch-and-a-quarter material with fillet welds can be load-bearing in failure analysis, and (c) the litigation politics of expert witnesses ("the guy whose wife drove around in a Rolls-Royce").

There are some other cases to show that this still occurs on a fairly regular basis. One of the stories I was involved in, in the early 1990s, is Pacific Gas and Electric decided to build a pumped water storage project, because the nuclear reactors like to run flat out. They don't like to cycle up and down. They like to run continuously twenty-four hours a day, nice and steady. But that's not the way people use their electricity. The peak electricity time is between five and seven PM, when people get home from work, it's still hot, they want their air conditioner running during the summer.

Think of the Venn diagram I did. I relieve the stresses, I've reduced the hardness, and I've reduced the hydrogen. So this is hardness HV, stress, and H2. By post-weld heat treatment I do lots of good things. If I properly post-weld heat-treat, I probably won't get a hydrogen crack. And I told you about the Helms project. Now, how do you do it? What procedures do you use? Mike Boomforth [Bumforth?] was talking about the structural welding code yesterday, and here are some of the welding procedures per the structural welding code.

Full forensic case study. PG&E pumped-storage project in the Sierra Nevada; American Bridge (division of U.S. Steel) field-fabricated a 22-foot diameter, inch-and-a-quarter wall penstock across a granite canyon. Welders skipped preheat in winter (welding inch-and-a-quarter steel with three feet of snow inside the pipe), skipped post weld heat treatment, and after frozen storage and spring pressurization the vessel failed by delayed hydrogen cracking 20 hours after reaching 188 psi service pressure. Released the upper lake, scoured 100 feet of granite, $100M loss on $800M project. Six-week trial in Fresno (early 1990s); jury found for American Bridge on a foundational-failure theory; Tom testified as the last witness.