[tw]

Originally Posted by glatt

The mother load of detailed public documents about this crash investigation is here at the NTSB site.
I was looking for a transcript of the hearings, but don't see one.

This document says why the failure happened.

Critical is something called bobbing. In simple terms, a transmitter at one end of two rails sends an audio signal to a receiver at the other end. Each interlock has its own frequency. If the receiver 'hears' the signal, then a relay closes. If anything appears on that track, the rails are 'shunted' together. No transmitter signal gets to the receiver. The relay opens. A train must be on that track when no signal is heard.

But the system has been defective (intermittent) for two years. Bobbing constantly. No useful solution implemented by identifying the problem. IOW repairs rarely identified a problem before 'fixing' it. They did what so many (probably a majority of) untrained A+ Computer Techs do. They just keep replacing parts until something worked.

And yes, this summary in Toyota stop sale applies. In both cases, we are talking about a logic one condition, a logic zero condition, and the always existing third state. The world is not binary. The world is ternary. That interlock system had noise that caused 'bobbing'. Caused constant failures (an intermittent failure is a complete failure).

In one case, while trying to fix bobbing on one track, suddenly bobbing was observed on an adjacent track. It is not clear if or why the adjacent track was also not fixed.

Other factors are discussed implying significance. The phone system was inoperable or unreliable even though conversations essential to human safety were reportedly conducted on that system. Rains had completely flooded phone and signalling wires days previous. But the reason for bobbing appears (may be) completely different. And something that no track maintenance people could possibly discover. They needed serious and better trained assistance. Instead they kept shotgunning.

A failure exists. So one keeps replacing parts until failures go away? Nonsense. Unfortunately so many never learn how to analyze before solving a problem. So they shotgun.

Bottom line: from what I understand, that entire section of track was completely unreliable for almost two years. At minimum, every engineer should have been trained to approach Fort Totten as if the system was always defective. We know this new (three month) engineer did what she was trained to do and what the signals told her to do - went through a blind curve at normal speed.

That report has numbers that the reporter should have been reporting. Bobbing was even reported via the remote data units (RDU) that receive that information every one second. But (apparently) no system existed to alarm on bobbing so that network operators could take immediate emergency action. Bobbing at the highest levels of management was considered all but normal operation. 85% of all problems are directly traceable to top management.

When bobbing occurs, transmitter power is changed (ie increased) in some desperate hope to solve the problem using the Tim Allen joke: "More Power!". Those new and 40 year old signalling devices are compatible. But require adjustments that are always done anyway on every track.

If the reporter was reading facts, she would not hype about mixing 40 year old equipment with new equipment. She would have read the report. Noted (by my definition of reliability) that entire section of signalling was completely defectve - intermittent. And that nobody had identified the problem. Just kept trying to fix it - for two years.

Shotgunning (except in a few instances) was the only repair technique. If your auto mechanic used shotgunning, he should be quickly unemployeed. And yet so many Americans so little learn how to analyze (solve) problems that shotgunning (especially in computer repair) is normal. It appears shotgunning was relavent for causing nine deaths. And it appears the reporter also did not identify the actual problem. Her report without the word 'bobbing' is woefully negligent.

Also discovered were supervisor errors:

During postaccident interviews, TSSM, ATC mechanics working on the Red Line stated using PMI 11000 – High Frequency Track Circuits adjustments procedures for GRS modules. They further stated having no procedures to adjust or to verify track circuits when US&S impedance bonds were installed with GRS ATP modules. Neither the CIT crew leader nor the ATC mechanics interviewed mentioned the October 6, 2006 engineering bulletin during their interviews. The CIT supervisor mentioned he was aware of an engineering bulletin, but understood the engineering bulletin only applied to high current substation return impedance bonds and did not apply to regular impedance bonds.

WMATA provides form PM-1, Track Circuit Adjustment to record all track circuit adjustments and verifications performed under the PMI 11000 procedure. Column seven of the form is labeled Shunt Test and provides two columns to place a check to indicate if a shunt was used on the transmitter end and/or the receiver end of the track circuit. The form is contradictory to the October 2006, Engineering Bulletin since it does not provide the option to indicate if three shunts were used for track circuit shunt verification as specified in the bulletin.

IOW after upgrading the signalling system, they are suppose to duplicate a train (shunt the rails) at each end and somewhere in the middle. Employees were only doing what management told them. Management failed to provide the necessary 'attitude and knowledge'.

Postaccident train control historical data were reviewed and indicate that at approximately 1:33 am on December 12, 2007 track circuit B2-304 was down. The data correspond to the date and time the high current substation return impedance bond at chain marker B2-311+71 was replaced. This is the B2-304 track circuit receive impedance bond. The data further indicate that about five hours later, track circuit B2-304 began bobbing between train movements. The bobbing continued intermittently until the day of the accident.

The crash was in 2009.

Postaccident data from the morning of June 17, 2009 indicate track circuit B2-304 was performing irregularly during the time the track circuit adjustment and verification process was conducted. Because of the frequent bobbing of track circuit B2-304, the shunt verification tests could not be verified to confirm the CIT crew leader statements made in the postaccident interview. According to the postaccident data, the performance of track circuit B2-304 changed significantly just prior to the arrival of the first train. From the time the impedance bond was replaced, the track circuit was bobbing and the track relay was seldom energized for more than 30 seconds between drop outs. Nine minutes before the arrival of the first train, the track circuit began staying energized for minutes at a time and was only bobbing for a second or two. The data further indicate that train detection failed for the first and nearly every train during the entire occupancy of track circuit B2-304 after the impedance bond was replaced on June 17, 2009 until the time of the accident.

We should be discussing murder charges.