As federal investigators examine last month's deadly natural gas pipeline explosion in San Bruno, engineering experts already have a strong sense of what went wrong and say the evidence calls into question widely used industry estimates of pipeline safety.
A 28-foot segment of ruptured pipe shows signs that its steel had become brittle over the decades. The blast point also occurred at a dip in the landscape that left the underground pipe subject to corrosion from accumulating water and sewage. And that bend in the line meant more welds were used during construction, creating another potential vulnerability.
When the 30-inch-diameter line finally failed, it released an explosive force that exceeded the thrust of a space shuttle at launch, ejecting tons of asphalt and dirt into surrounding streets. A massive cloud of pressurized gas ignited moments later into a fireball that torched 55 homes and killed eight people.
Such pipelines exist under homes and neighborhoods throughout the country. California has 3,600 miles of pipe under densely populated areas.
Nearly half the high-pressure gas transmission lines in the country were installed from about 1950 to 1970. The industry has developed mathematical models to predict which pipelines are at risk and how often they need to be replaced. But a deep divide exists among experts about whether those models are accurate.
Now, the San Bruno disaster has bolstered the arguments of those who say the industry has put far too much faith in technology and inspection methods.
"The industry worships these models," said Robert Bea, a professor of civil engineering at UC Berkeley who is a member of the National Academy of Engineering because of his pioneering work in risk analysis. "They are treated as theology."
The inspections and analysis that PG&E conducted before the accident failed to identify the San Bruno pipeline segment as one of the 100 top safety priorities in the utility's system. That failure points to an analytical breakdown, Bea said. PG&E officials say they won't make any conclusions until after investigators determine the cause of the accident.
Because the industry thought its system was safer than ever before, the accident has triggered some serious soul-searching.
"Every time I see photographs of it, it makes me sick," said Theodore Willke, former chief of the federal Office of Pipeline Safety and now a technical consultant. "People don't know these huge pipelines are in their backyards. They don't know the energy that is contained in these pipes."
The failure of the San Bruno line has much in common with an explosion in 2000 that killed a dozen people camping along the Pecos River in rural New Mexico. That disaster triggered a major overhaul in federal law and regulation. Until the San Bruno explosion, those changes had led to a growing level of confidence about aging pipeline infrastructure.
The National Transportation Safety Board has shipped three sections of pipe from the explosion site to its laboratory in Ashburn, Va., where metallurgists and engineers will examine welds, corrosion coatings and so-called chevron marks, which are microscopic ridges in the tear surfaces of the pipe that can disclose which way it ripped apart. Although the investigation will probably continue into next year, a number of factors have become obvious to engineering experts.
Like the pipeline in the New Mexico explosion, the 28-foot segment of San Bruno pipe that ruptured at Earl Avenue and Glenview Drive was at a dip in the topography, a vulnerability that almost certainly played a key role. When the pipeline was built in 1956, the dip forced construction crews to weld the line together in small segments. Normally, a pipeline is supposed to flex. But the large number of welds could have created spots that were rigid and others that were weak.
"I would pay attention to the welds, because there are many, many welds," said Erez Allouche, a professor of civil engineering at Louisiana Tech University and research director for the school's center for trenchless pipeline technology. "There were so many welds in such a short area, maybe the pipe was rigid."
The dip would also allow water — among the compounds that can contaminate natural gas — to pool inside the pipe. PG&E only recently began installing collectors and filters to resolve the water problem in portions of the long segments that ran from Milpitas, just north of San Jose, to San Francisco, according to regulatory filings.
The problems relating to the dip continued to multiply when, in recent decades, sewers frequently overflowed in the neighborhood. Sewage can be extremely acidic, owing to bacteria that excrete acid. The potent brew accelerates corrosion and could have posed serious problems if the mix saturated the soil around the gas pipe.