Documents released Tuesday also revealed that it was a misinterpretation of old records - not just a lack of documents - that led PG&E to classify the doomed pipe incorrectly as seamless, meaning that it had been manufactured without a seam weld.
PG&E didn't know the pipe had such welds, and the company never conducted an inspection that could find faulty ones.
Blast mysteriesHow such a badly welded pipe was buried in the ground in San Bruno in 1956, at a time when PG&E crews were relocating a 1948 vintage pipeline to accommodate a new neighborhood, is part of the mystery unfolding as the federal safety board holds three days of hearings this week in Washington.
Experts who have reviewed the federal metallurgy report have questioned why the San Bruno pipe was made with an unusual number of short pieces, known as pups, which were joined together. There were six pups on the line.
Pups are sometimes pieces of good pipe salvaged from longer sections of pipe that failed pressure tests at the factory, Edward Salas, PG&E vice president of engineering and operations, testified Tuesday.
One veteran metallurgist in the hearing room audience said the pups Salas described aren't safe.
"As a practice, I think any (pipe section) that fails a (pressure) test should be scrapped altogether," Doug Chisholm, a Virginia consultant with 30 years' experience in the pipeline industry, said in an interview.
Pipe was running lowThe company's stock of 30-inch transmission pipe "was getting low" when the San Bruno relocation project occurred in 1956, PG&E told the federal safety board in answers to written questions.
"Surplus pipe and potentially salvaged pipe from several previous pipe purchases were likely used on the 1956 construction," the company said.
The company said the surplus pipe may have been stored in the back of a storage yard in Oakland. PG&E couched its written responses with terms such as "we believe" because the company is still researching records.
Bad records entryAnother mystery being addressed in this week's hearings is why PG&E's computerized records showed that the transmission pipe was seamless.
Seamless 30-inch pipe wasn't manufactured back then, records show. When PG&E inspected the pipe in 2009, it used a technique ill-suited for discovering faulty seam welds.
The mistake was made in 1998, according to transcripts of interviews with PG&E officials and records, and remained in the company's computer system despite a fact-checking process designed to eliminate such errors.
The problem was that someone made an entry into PG&E's all-important computerized database that said the San Bruno pipe was seamless. It appears the error was a misinterpretation of the letters "sml," which had been noted on a paper record known as a journal voucher.
Such vouchers are "created by the accounting department to reflect the transfer of materials from one project to another," the company told federal investigators.
Journal vouchers are not meant to be pipeline specification records, PG&E said.
It was fixableHad the company checked its original engineering records, it would have linked the coded description of the pipe on the journal voucher to a "material code description table," which would have indicated the pipe was manufactured with a seam weld, Robert Fassett, PG&E director of integrity management and technical services, told investigators in January.
"If someone had actually taken that document, looked up that material code, they would then have said this is a (welded) pipe," Fassett said.
A Chronicle investigation published in February revealed that PG&E didn't properly oversee computerized data entry in the mid-1990s.
Other undamaged pipe near the blast site on the San Bruno pipeline was also mislabeled as seamless in the database, Brian Daubin, a PG&E engineer involved with system, told federal investigators in January.
"The company knows it must improve its system of records," PG&E said in a statement Tuesday, adding that the company has 300 employees and contractors "hard at work confirming the quality of its data."