SINGAPORE: It was a rogue train with faulty signalling hardware that was the cause of the wireless signal interference on the Circle Line, revealed the Land Transport Authority (LTA) and train operator SMRT on Friday (Nov 11).
The intermittent hardware failure between Nov 2 and Nov 6 caused about 100 occurrences of loss of signalling communications on trains travelling in the proximity of the train, identified as Passenger Vehicle 46 (PV46).
"The loss of communications on these trains resulted in their emergency brakes being automatically activated, which is a safety feature," both parties said in a press release.
Ongoing investigations also indicate a "strong likelihood” that PV46 was the cause for about 200 occurrences of loss of signalling communications on Circle Line trains between Aug 28 and Sep 2, they added.
Intermittent signal interference caused disruption and multiple delays on the Circle Line for nearly three hours on Nov 2, which resulted in mobile signals being shut off temporarily over the following two days.
The same signal interference problem dogged the line for five consecutive days in September.
Transport Minister Khaw Boon Wan said in Parliament on Wednesday (Nov 9) that the Circle Line disruption was "interesting" and has been challenging for his team, which had to eliminate possible sources of the signal disruption one by one.
ALMOST ALL FAULTS OCCURRED IN VICINITY OF ONE TRAIN
The joint team of LTA, SMRT, signalling system manufacturer Alstom, the Infocomm Media Development Authority (IMDA), the Defence Science and Technology Agency, DSO National Laboratories, the Government Technology Agency (GovTech) and electro-magnetic interference specialist Rohde & Schwarz had spent several weeks investigating the incidents.
After studying data from train logs and incident reports, they found that almost all the faults had occurred in the vicinity of one train, PV46.
The train was tested during off-peak hours on Nov 6 and found that it did indeed cause a loss of communications between nearby trains and the trackside signalling system.
Upon further tests, it was found that PV46 was emitting erroneous signals in addition to the ones it was supposed to emit. These erroneous signals occasionally prevented trains travelling in the vicinity of PV46, including those in the opposite direction, from properly communicating with the trackside signalling system.
This loss of communications led to the activation of the emergency brakes, they said.
PV46, which has been deployed since July last year, has been pulled out of service. At this point, it is unclear what caused the intermittent failure in the train’s signalling hardware, but so far, no other train has been found to have the same hardware issue. LTA and SMRT will also conduct a thorough examination of the signalling hardware for the rest of the Circle Line trains.
INTERFERENCE FROM TELCO SIGNALS RULED OUT
In investigating the issue, LTA and SMRT noted one possibility: That interference from telecommunications signals may have caused the repeated activation of the train’s emergency brakes.
To test this, LTA, IMDA and the three telcos suspended mobile telecom services across the entire line for two hours on Nov 3, and then for a full day on Nov 4. But during the shutdown, trains operating on the Circle Line still continued to see a loss of communications. “These tests showed that telecommunications signals were not the source of interference,” LTA and SMRT said.
Another possible source of the problem explored: The signalling modifications performed in the months leading to the incidents. After an independent hardware and software review however, it was determined that the changes were minor and should not have caused the problem.
Authorities added that it took some time to attribute the source of the problem to a defective train, as the train itself did not lose communications regularly. Instead, it was affecting trains adjacent to it and on the opposite track. Hence, PV46 was not mentioned in the incident reports, and investigators were not able to identify it from initial data analysis.
One difficulty the investigators faced was the fact that the problem was intermittent, and the issues stopped before they could complete testing after the first signal interference incident in September.
PV46 was identified as the problem only after extensive data mining revealed a cluster of incidents occurring when trains were travelling within its vicinity. This was later verified by analysing days of train timetables and operations data.
Channel NewsAsia understands that while deliberate interference has not been ruled out at this juncture, maintenance records show that it is highly unlikely and does not appear to be the case.
TIMELINE OF EVENTS
Aug 29 to Sep 2: Intermittent signalling issues plagued the Circle Line, causing disruptions for five consecutive days.
Sep 2: Mobile signals were switched off for about two hours at four Circle Line stations while authorities investigated the source of signalling problems.
Sep 3: SMRT and LTA said the tests were “inconclusive” as there were no incidents the night mobile signals were turned off.
Sep 21: Authorities said use of electromagnetic shields on trains are being explored to minimise the impact of interfering signals, as well as the possibility of changing the signal system frequency and/or modifying the system to provide redundancy in the event of signal interference.
Nov 2: "Intermittent signal interference" caused a disruption and multiple delays on the Circle Line for nearly three hours during morning peak period.
Nov 2: Authorities said they will turn off mobile signals along parts of the Circle Line for investigation should another signal interference episode occur.
Nov 3 to 4: LTA instructed telcos to shut off mobile network access along the line for about two hours in the evening of Nov 3, and a full day on Nov 4, as SMRT engineers sought to determine the cause of intermittent signal interference. As trains operating on the Circle Line still saw a loss of communications, telco signals was ruled out as the source of interference.
Nov 6: Data collected from train logs and incident reports showed almost all faults occurred in the vicinity of one train, PV46. The train was launched during off-peak hours of Nov 6 to test the hypothesis.
Nov 6 to 8: PV46 was identified as the defective train causing the problem. Tests were carried out to confirm this, and the train was subsequently pulled from service.