Metro hasn’t said what caused last week’s derailment. Here are some possibilities.

Last Thursday, a train came off the rails between Smithsonian and Federal Triangle, delaying tens of thousands of commuters. Metro still hasn’t said what caused the derailment, but we can surmise a few of the likely causes.

Derailments are generally caused by a fairly finite set of circumstances. This particular train was empty and in the process of changing from running north to running south so it could start morning Blue Line service at Smithsonian. Since this happened at a switch, the set of possibilities for what caused it is probably even smaller.

We’ll have to wait for Metro to list an official cause, which they absolutely must do in the name of transparency. For now, here are some of the probable causes.

Wheel climb

One of the likely causes of the derailment was wheel climb. Wheel climb happens when a wheel with a rough surface grabs the outside rail in a sharp turn (like at a switch) and climbs over the rail and falls over the other side.

WMATA has had trouble with wheel climb, especially with the 5000 series, over the years. In January 2007, a Green Line train derailed at Mount Vernon Square due to wheel climb, injuring several passengers. In 2010, NTSB found that improper milling procedures contributed to the derailment by creating a rough wheel surface. Milling, which is a type of truing, is a routine procedure that keeps wheels round.

In addition to recommending better truing procedures, the NTSB recommended that WMATA install guarded turnouts throughout the system. Guarded turnouts have additional rails that pin trains’ wheels to the running rail, and they should completely eliminate the risk of a wheel climb derailment. And as it happens, guarded turnouts were installed at the Smithsonian interlocking in 2011.

Mechanical problem with the switch

Another possibility is that there was some mechanical problem with the switch. If the switch points weren’t properly locked, the train passing through could have forced them open, causing a derailment. But the signal system should have been able to detect whether the switch was set properly, and if it wasn’t it should have displayed a red signal.

A problem at the Smithsonian interlocking led to the January 1982 derailment that killed three passengers. In that case, the switch was out of correspondence, and Metro was trying to recover from single-tracking around a disabled train. The operations center couldn’t verify the position of the switches, and a rail supervisor was manually setting them when a train was mis-routed through the interlocking. When personnel tried to back the train through the switch, it derailed, crushing one car against the wall.

Broken wheel

The train could also have had a broken wheel, though that’s unlikely since it’s extremely rare in the transit industry. A broken wheel could have caused the derailment either by damaging or forcing the switch or by causing wheel climb.

This is similar to what happened to a German high speed train at Eschede in 1998. In that case, the rim of the wheel became detached and eventually caused the entire train to derail and crash into a bridge support at 125 miles per hour, killing 101 people.

Excessive speed

Crossovers on WMATA are designed to be taken at slow speed, generally around 20 miles per hour. If the train were to take the crossover at high speed, it would very likely derail.

A crash like this happened in New York in 1991 when the drunk motorman of a downtown 4 train took a 10 mile per hour switch at over 50 miles per hour, derailing near Union Square and killing five.

However, on Metro, the signal system limits speeds through interlockings automatically, so as long as it was functioning correctly, it’s unlikely that the train would have derailed due to speed. Additionally, the train would have been starting from a dead stop at Federal Triangle, only a few hundred feet up the tunnel, so that makes this even less likely.

Broken rail

A broken rail can also cause a derailment if it’s severe enough. Broken rails are often caused during extremely cold weather, when the metal contracts more than it was designed to. Those conditions did not exist on Thursday.

However, broken rails can also be caused by flat spots on wheels, since there’s more pounding on the rail. It’s unlikely that this would happen at an interlocking, though, since the actual switch tracks only experience low-speed trains and are trafficked relatively infrequently.

Splitting the switch

Another possibility is that the train split the switch due to a problem with the switch. A split switch occurs when the first truck (wheelset) of a train car goes down one track, but the second truck takes the other branch. Above ground, this can be corrected fairly easily. But underground, the results can be catastrophic because eventually the car will collide with a support pillar or wall.

This is essentially what happened in 1990 in the 30th Street crash in Philadelphia. In that case, a traction motor in the rear car fell from its mounting and was being dragged behind the train. When the Market-Frankford train went through the 30th Street interlocking, the traction motor forced the switch into the diverging position, so while the front of the train went straight, the last truck of the train went left and was smashed against a support pillar, killing four people.

Technically, this isn’t a derailment, because all the wheels stay on the rails. And it sounds like in this case that there was a derailment, so a split switch is not likely to be the cause.

We know what it was not

We can almost certainly name some of the things that did not cause the derailment.

Heat kinks, which is when the metal of the tracks bends outward, happen in the summer heat, and have happened on Metro. But they don’t happen underground because they generally need direct exposure to sunlight to form.

Gauge spread, when the track’s rails move apart from one another, is also not possible here. Gauge spread generally happens because of a failure of the crossties or the clips binding the rails to them. But in the underground parts of the Metro system, WMATA uses direct-fixation track, where the rails are mounted directly to the concrete, so gauge spread can’t happen.

Automatic derails are located in certain parts of the system to prevent trains or other equipment from accidentally rolling onto the mainline. Each of WMATA’s pocket tracks have these. But regular interlockings do not have an automatic derail, so that can’t have caused this derailment, the way it caused a 2010 derailment at Farragut North.

Conclusions

Without knowing more, it’s hard to speculate on the cause of Thursday’s derailment. The risk of a derailment on mainline track is very low because WMATA has taken so many steps to mitigate it.

Guarded turnouts should have prevented the wheel climb possibility, and excessive speed shouldn’t have been possible. Yet the train came off the rails. And so far, Metro hasn’t said why. It’s possible they don’t know yet, which would be very concerning.

But it’s also possible the agency has come to at least a preliminary conclusion. And if that’s the case, Metro should report what they know to the public, and also report what steps they are taking to prevent this from happening again.

Update: Metro announced on August 12 that the cause was wide gauge detected over a month before the derailment which was not fixed. We had listed that possibility in the “things we know it wasn’t” because wide gauge (where the rails spread too far apart) is very rare in direct fixation track. Apparently, it did happen in this case.