A wrong-side failure describes a failure condition in a piece of railway signalling equipment that results in an unsafe state. A typical example would be a signal showing a 'proceed' aspect when it should be showing a 'stop' or 'danger' aspect, resulting in a "false clear".
Example of how a wrong-side failure may occur
Consider a relay that has to energize to show a green light. If a wire breaks, or the battery is drained, then the relay will de-energize and the signal will show a red light, which is fail-safe. If a stray wire from another circuit touches the wire connected to that same relay, then that would be a wrong side failure, which is potentially dangerous. This stray wire can be guarded against by ensuring that the insulation on the relay wire is of good quality, and that all terminals are locked away. In addition, the relay may be double-switched, also known as double cut, that is to say that it only energizes if a positive circuit and a negative circuit are both complete. That would then require two stray wires to cause a wrong-side failure, which is much less likely than a single stray wire.
Accidents
While accidents from the problem are rare, they do occur:
1915 – Rockledge, Tennessee - Relocation of a Western Union telegraph line from old alignment to new alignment causes a false clear signal indication when the Western Union wire comes in contact with the signal wire for the automatic block signal at the south end of the siding at Rockledge. A southbound freight train departed Rockledge on the false clear indication and collided with a northbound passenger train that had entered the block at Tantallon on a proper clear indication two minutes before the freight train departed Rockledge. A contributing factor was that the tower operator at Rockledge only controlled the signals, the track switches being operated by the train crews themselves.
1927 – Marrar, NSW, lightning damage meant that the single line train tablet system incorrectly released two tablets simultaneously putting two trains on a collision course; fortunately they saw each other and were able to stop in time.
1938 – Charing Cross tube crash - signal linesman rewires a faulty signal incorrectly, so that it shows green too soon; two Underground trains collide as a result, injuring twelve. Also Charing Cross tube crash, signal linesman makes wiring error in cabin, two Underground trains collide, killing six.
1976 – Glenbrook, New South Wales. Automatic signal in rear of a stationary EMU at "caution" when it should have been at "danger". Following goods train enters occupied signal section and collides with EMU; 1 passenger killed. Cause attributable to deteriorated wiring which allowed a false feed.
1979 - Invergowrie rail crash - signal was at "wrong" yet taken by driver to be a 'proceed' signal.
1990 - Cowan rail disaster, which occurred when sand on the rails insulated the wheels from the rails, causing a failure to shunt that allowed a trailing block signal to improperly display a clear aspect, resulting in a rear-end collision.
2005 - Deelfontein - solder splatter bridges a relay contact. Most testing is done with the relays unplugged which would not detect the splatter, while the solder splatter is likely to lead to a difficult to trace phantom problem.
2009 – On March 26, two trains collided near Arévalo, Spain while engineers were testing a new signalling system. Lack of safeguards and testing on an active line led to the collision of two freight trains. One of the train drivers died.
2009 – Ft Totten crashWashington DC Metro. An inbound Red Line train stopped short of Ft. Totten station; the following train failed to stop and impacted at ~42 mph. Nine dead, 80 injured. Preliminary NTSB reports were that the AC track circuit transmitter was breaking into parasitic oscillation, and that coupled across the rack cabinet to the receiver.
False alarms
Railway authorities usually give the drivers and signalmen the and investigate whether a wrong-side failure is the cause of the accident. This occurred with the Hinton train collision, but investigations soon showed that a wrong-side failure was not the cause. The 2008 Chatsworth train collision was determined to be human error, not wrong side failure as a few initially said.