Comments about technological history, system fractures, and human resilience from James R. Chiles, the author of Inviting Disaster: Lessons from the Edge of Technology (HarperBusiness 2001; paperback 2002) and The God Machine: From Boomerangs to Black Hawks, the Story of the Helicopter (Random House, 2007, paperback 2008)

Saturday, October 16, 2010

San Bruno Gas Explosion: NTSB's Preliminary Report

In my previous post on the NTSB's docket page for the San Bruno gas-pipeline blast, I mentioned that the docket was light on data, but the seven photos were worth a look and the Board would be adding more information as it developed.

Three days ago the Board added a three-page report with new detail about pipeline steel, wall thicknesses, segments, corrosion protection, and fracture locations. Now we know this was an above-average grade of 30-inch-diameter steel pipe (0.375-inch, grade X-42, minimum yield strength of 40,000 psi). Thicknesses reported by the NTSB have been close to the nominal, between 0.36 and 0.38 inches. There's no indication in this report of large corroded areas due to pooled moisture, but highly localized stress corrosion cracking is probably something they're still looking into. SCC can happen when corrosion gets a grip on a little flaw in the metal. The combination of corrosion at the tip of a micro-crack, fed with pulses of energy, can cause these tiny cracks to slowly get longer until they reach a critical length, at which they fail at supersonic speed.

The loss of two Comet jet airliners over the Mediterranean in 1954 was probably due to slow, incremental fatigue cracking as the fuselage flexed with pressurization: that's because the cabin inflates, then relaxes slightly as pressure-relief valves balance it out. Remember, in the world of disasters, little things add up!

Yet to come are micro-level test reports about fracture edges and the most likely starting point. The report mentions that "key fracture surfaces" have been cut away for lab analysis and I'd bet these are suspected starting points.

Some info is provided on the 28-foot-long stretch of pipe that, about halfway down its length, includes the rupture that is the special focus of NTSB interest. Right now the description of pipe sections are in text and therefore it's hard to visualize the joints in the length of pipe removed from the excavation. The NTSB writes about the stretch of pipe as two segments: the south segment, about twelve feet long, and the north segment, about sixteen feet long. The "rupture" is between those two.

The north segment includes two short sections (called pups) that were joined with girth welds. Girth welds are where pipes are joined end to end; the other kind of weld is the longitudinal weld, which is done at the factory -- this is where steel plates are rolled into a tube and welded along a seam. Unstated is why the pups are there: were they part of a rebuild since the line went down in 1956? Were they necessary from the first so that the pipe could make a bend at this location, say because it's at the bottom of a hill or it was making a turn? There have been press reports that this area of pipe had enough of an odd shape that the standard testing pigs couldn't slip through; the pups might be part of the reason for this.

In preliminary reports the Board leaps to no conclusions (recall, there is no probable cause finding until the investigation is winding up). Still, a close reading of a preliminary report can shed a little more light about which suspected contributory causes are still "of interest." This list is usually quite a bit shorter than the list of hypothetical ones bounced around by press reports (and commentators like me) in the first two weeks.

One of the live issues continues to be a pressure pulse. Though not one that "pegged the needle," it came just minutes before the rupture. While it could be a coincidence, and likely isn't the whole cause, it's getting the NTSB's attention and rightfully so.

The pressure pulse arose out of a chain of events (in general I call these system fractures).

The timeline started with maintenance work at the Milpitas Terminal 40 miles southeast of ground zero, which is upstream in flow terms. Milpitas has SCADA controls that are part of the pressure regulation. Some cause, presumably human error but not conclusively, caused the 24-volt DC so-called "uninterruptible power" at Milpitas to fall to 7 volts, which dropped all or part of the terminal's SCADA system offline. Without its link to Milpitas, a remotely controlled valve on Line 132 moved to a pre-set uncontrolled position, from partially open to full open. This allowed the transmission-line pressure as measured at a point downstream of ground zero to climb above the maximum operating pressure of 375 psi. The location of these measurements was Martin Station in Daly City.

About 6:00 pm, eleven minutes before the rupture, pressure at Martin Station peaked at 390 psi. Then eight minutes later it dropped slightly to 386 psi, probably due to automatic pressure relief valves. At 6:11 pm it fell rapidly to 290 psi due to the rupture at San Bruno.

At 6:45 PG&E mobilized an emergency team to isolate Line 132. This delay has attracted much criticism considering that the flames were visible 10 miles away and 911 calls began flooding in at 6:11 and 59 seconds. PG&E closed an upstream valve at 7:20 and the downstream valve at 7:40. The total volume of gas released was about 48 million cubic feet. Firefighting work continued at individual homes due to gas-fed flames from the broken house lines. PG&E shut down the distribution lines to the neighborhood at 11:30 pm and the last of the house fires quickly came under control.

Is the pressure pulse a smoking gun, in CSI terms? Maybe part of one. As far as we know from the instrument readings at Martin Station, the measured peak of 390 psi was still 10 pounds below the maximum allowable operating pressure of 400 psi. Just browsing the Web, I see burst-pressures a good deal higher than that for pipes made out of 0.375-inch X-42 steel. But cracks in the wrong places can bring about anything down. If the nearby sewer line work two years earlier played a part, there might not be any marks on the gas pipe at all; a lateral shifting of the soil due to the sewer-line work could have raised the stress on some gas-line welds. The fracture analysis at the Board's materials lab in DC will tell us a good deal.

Here's an example of an NTSB materials-lab report, prepared for the Flight 587 crash investigation. So my advice is to stay tuned, and trust the process.

No comments:

Post a Comment