San Bruno: NTSB's Docket Page, and More

A few updates on the San Bruno pipeline-blast investigation.

Here's my tally list of factors that still look like candidates for the root cause analysis, based on news reports and press conferences: overpressure in the pipeline; tied in with this, possible SCADA malfunction or fatigue-induced operator error (see this wrap-up from the Mercury News about a possible power glitch that could have affected readouts and controls at the PG&E Milpitas control center; unrepaired corrosion that could have weakened the pipe wall due to moisture outside the pipe or some kind of liquid that gathered at low spots inside the pipe; collateral damage from nearby construction, possibly the earlier pipe-bursting work on a sewer line nearby; or some problem caused by modifications to "Line 132" after it was laid down in 1956.

For those who don't follow the progress of NTSB work but would like to, the NTSB sometimes puts up a docket management page showing the progress of its work on a major incident. A tweet from NTSB pointed to the San Bruno docket page . There's not a lot on there right now -- only seven photos -- but in general dockets are where we technology-watchers find PDF versions of contractor studies, blueprints, pre-disaster inspection reports, metallurgical studies, and many other pieces of worthy information. While these materials are not official Board findings they are the raw material on which the Board will build its final report.

These are very important for reporters preparing for a press conference. I first learned about docket pages when reading up on the I-35W bridge collapse. Here's the docket page for the Minneapolis bridge disaster in 2007. The list of resources can grow to be quite long, and the indexing isn't necessarily easy to use. But nobody said research was supposed to be easy!

From the San Bruno docket, here's the best view I've seen of the full section that flew out of the crater, with a close-up view of one end. The pipe was originally entrenched with about three feet of dirt cover.

I did a bit of reading on pipeline construction of the 1950s and surmise that the pipe was fabricated from rolled sheets of carbon-manganese alloy with a yield strength between 50,000 and 55,000 pounds per square inch. Just a guess!

We'll have to wait for reports on whether corrosion played a role. NTSB metal experts can tell a great deal from studying broken metal, such as whether stress corrosion cracking played a role. They turn their microscopes on all the edges, looking at where ductile fracture changed over to brittle fracture. (An important subject but one for another post.)

Note in the photo that a long line of weld bead on the left of the photo, on the inner wall of the pipe, shows as intact. This weld most likely would have been by arc-welding machine at the pipe factory. Factory welds are ideal for quality: If an arc weld has been well executed it can be stronger than the plates to be joined. Welders call a good bead a "stack of dimes" (particularly in TIG and MIG work) and this bead looks like one of those, from a distance.

But arc welds at field joints, where ends of pipe are joined at the job site, are a different matter. Most difficult to execute is the welded section at the bottom of a pipe: were they done by machine, or by a guy with a welding rod, crouching in a trench and working overhand? Overhand welding is the most challenging work-setting for an arc welder, even an expert. (One of the Kaiser innovations leading to record-setting Liberty ship manufacture during WWII was to revolve huge sections to minimize overhand welding - Henry J. is on my short list of industrial heroes, but that's something for another post.)

Here's an aerial view of the site. This shows where the principal plume of fire moved, toward the north-northeast. Here's an overview map from the LA Times showing street names.

Take a look at this photo on the NTSB docket, a street-level view of the crater on the eastern side of Glenview Drive. In the foreground you can see where the pipe section landed after shooting out of the crater. Most likely it would have tumbled in mid-air, so it could have landed upside-down. Note also that the crater (28 feet by about 70 feet) seems to be where the street is crossing a shallow valley. This would suggest corrosion is somewhat more likely than if the section had been located at the top of a hill.

In other words, corrosion is still on my list of possible contributing causes, along with collateral damage from nearby sewer work. An alarm-flagged overpressure of 11 psi over the upper end rating of 375 psi doesn't sound like enough on its own to cause a catastrophic failure -- heavy iron like this, even 54 years old, should have enough safety margin to handle that pulse. It's possible, of course, that actual pressure spikes were considerably higher than 386 psi. SCADA records should reveal more detail than the fact that "hi-hi" alarms were triggered.

But as outsiders like me continue to speculate, it's reassuring to know that Board experts are on the job. They're not dismissing any possible root cause without good reason.