The Site Admin attended the first session of the Committee of Inquiry public hearing into the SMRT North-South Line disruption which occurred on 15th and 17th December 2011. The session was held 16th April 2012 at Court 1, Subordinate Courts Singapore.
The session was attended mainly by members from Land Transport Authority, SMRT Corporation Limited and the press. The Site Admin did not notice any members of the public attending the session, or at least they are undercover like himself?! Some members from LTA too commented that there is too little ‘juice’ for this case; one which the Site Admin would agree.
Three public figures hosted the Committee, Chief District Judge Subordinate Courts, Mr Tan Siong Thye, School of Mechanical & Aerospace Engineering Nanyang Technological University, Professor Lim Mong King and Director of Prisons, Mr Soh Wai Wah. Attorney General, Land Transport Authority represented by Allen & Gledhill LLP and SMRT Corporation represented by Drew & Napier gave their opening addresses. The first day of hearing also included part one of an investigation report by the Criminal Investigation Department. The Site Admin would highlight the points presented by each party present on 16th April and hope to give you, my readers, a first hand look at how the proceedings will likely go. Please be informed that these findings are not conclusive.
Before starting, the Site Admin will provide a link to a general map of Singapore’s MRT system today, for the benefit of international readers who may not be aware of the station locations.
Opening Address by Attorney General of the State
- 108 witnesses will be called forward to present their accounts to the Committee, of which they include,
- 11 Train Officers, 9 from the stalled trains on 15th and 17th December 2011, 1 from T151 which was the preceding train of all accident trains on 15th December. An additional Train Officer who saw the sagging third rail will also testify.
- 34 Station staff. All these staff were from affected stations on 15th and 17th December, in particular the Station Managers, all of whom were involved in both incidences.
- 6 service operations managers and 4 passenger service managers. They were in-charge of the crowd control that day.
- 5 Operations Control Center (OCC) personnel. They too were on duty on the 15th and 17th December.
- 3 commuters, one of whom fainted during the ordeal on board T134.
- 30 officers from SMRT who were involved in the maintenance and particularly those who operated the Multi-Function Vehicle (MFV) during non-revenue hour checks on 17th December 2011 before the start of operations, which suffered damaged during the process.
- 12 senior staff from SMRT and 8 from LTA will also testify.
- In addition to the above, 11 experts from various fields will also present their reports and testify.
The broad view of the failure on 15th December 2011 could be determined as follows,
- Cracked fastener of the 3rd rail. Impurities of the material was suspected.
- Excessive vibration that caused the ‘claw’ component of the 3rd Rail Support Assembly (TRSA) to dislodge.
- This could be due to trains in operations that has ‘flat wheels’, a condition which the Site Admin will explain later, or,
- Vibration by the broken fastener to the insulation, or,
- 3rd rail not fully weighted down, causing the claw to pop upwards and fall off. This could be due to stiff Current Collection Devices (CCDs) or the occurrence areas as a curved section.
The broad view of the failure on 17th December 2011 could be determined as follows,
- Excessive vibration to the 3rd rail ‘claw’ assembly, which could be caused by trains with ‘flat wheels’, or,
- Crack on the 3rd rail segment caused by impurities in the aluminium rail, or,
- 3rd rail not fully weighted down (refer to previous).
- There is also consideration of the train’s kinematic envelope interfering with the 3rd rail insulation cover, a possibility the Site Admin explored on his previous post. Similarly, flat wheels, or,
- Location being a tight curve, or,
- Train involved on the 15th December has gone unnoticed with a damage, continue to run service on the 16th, and finally causing damage on the 17th itself. This train was identified later in LTA’s report.
At this juncture, the Attorney General also expressed that the following two components were determined by experts appointed by SMRT and reviewed by other parties to be non-critical in the major disruption, in other words, they are not the main/decisive factor that caused the 3rd rail to be damaged. These two components were;
- Floating Slab Track, or FST, installed in certain urban sectors of the tunnel, and,
- The latest fourth generation rolling stock, the C151A or ‘KSF’ by SMRT.
Opening Address by Land Transport Authority
- It was not disputed that the train services were disrupted due to the sagging of the power conducting rail (3rd rail) which damaged the current collector shoes of the affected trains, such that these trains could no longer draw power from the power conductor rail.
- LTA’s review show that SMRT’s maintenance scheme is generally comprehensive and satisfactory. Maintenance of safety-critical items also appears adequate. Selected areas however are shown to be requiring improvement and/or further review.
- With regards to incident management, the procedures can be improved.
The subsequent proceedings were of the technical background of the structure of the NSEWL. The Site Admin will only single out interesting figures for your future reference.
- There are three types of trackbeds found in different sections of the network, the Ballasted Track (BT), Concrete Slab Track (CST) and Floating Slab Track (FST). BT could be found on viaduct and roadbed designed sections, CST commonly in underground sections, while FST are in urban and areas where there is a need to minimize noise and vibration emanating from trains underground.
- The 3rd rail is used to transmit electricity to power the trains. The aluminium and stainless-steel composite rail’s underside is positioned normally 170 mm above the height of the running rails, and elevated in this position by a Third Rail Support Assembly (TRSA), spaced approximately 5.6 or 6.3 meters apart from each other. The support bracket for the TRSA is fastened to a concrete sleeper in tunnels or timber sleepers on BT sections. It is usually located on the right side of the direction of travel, but switches to the opposite side of the platform at stations so it is further away from boarding and alighting passengers.
- The Current Collector Device (CCD) is used by the train to collect traction and auxiliary power, and each bogie has two of these, one on each side. It is bolted to the bogie and fitted with a spring-loaded collector shoe. It exerts upwards pressure onto the underside of the 3rd rail so that the carbon pad on the conductive shoe makes contact. The spring mechanism is such that a maximum defection of 65 mm is possible, more than which the frangible link between the shoe and the frame is designed to break away, limiting damage to the 3rd rail and other CCDs.
The immediate cause of the incidents on 15th and 17th December 2011 is highlighted as follows,
- For the 15th, the 3rd rail had sagged and resting on the trackbed as a result of 6 adjacent TRSA having failed along the affected section of the track, and,
- For the 17th, 3 ‘claws’ of the TRSA had dislodged and the 3rd rail of the section was almost touching the trackbed.
Trains passing the affected sections had had their current collector shoes either twisted badly or sheared off. There were also some CCDs which exhibited cracks. As a result the affected trains could no longer pick up electric power from the 3rd rail.
Correlation of wheel defects, which is termed for ‘flat wheels’, and high vibration was also explored. It was found that generally trains with such defects, whether new or old, produce significantly higher levels of vibration than healthy trains with properly re-profiled wheels.
Wheel flatting is characterized by flat patches along the circumference of the wheel such that the wheel is no longer round. This can be remedied by wheel profiling, which is done by a wheel lathe machine which cut around the wheel to make it smooth again. When a train with wheel flats travel along running rails, a transient high impulse vibration will be produced at points where the flat patches on the wheel interact with the running rail.
It is contrary to the previous speculations that the new generation trains (C151A or KSF) were the cause of the dislodged ‘claws’, the on-site and laboratory tests confirm that the trains are not to be blamed. In the report submitted to SMRT by Sinclair Knight Merz (SKM), it was noted that
‘Based on the tests conducted in this investigation, there was no significant difference between the kinematic or dynamic performance of any of the rolling stock [i.e. trains] types, including the new 4th generation trains which could have led to the December incidents in which the claws were dislodged.’
‘The new rolling stock [i.e. trains] under test behaved very similarly to the test trains drawn from existing fleet in terms of bogie kinematics (bounce and roll) and bogie accelerations (vertical and lateral).
There was no evidence of significant differences between the 4th generation trains and the existing rolling stock as tested that could have led to the December incidents.’
However the Site Admin would like to ask; are the KSF stocks compared to the KHI and SIE stocks in terms of this? Results would be very different if all above stocks are compared with the KNS stocks (in good order), which generally on a commuter’s experience have significantly lesser ‘bounce’. This may be highlighted in the future.
SMRT’s Internal Investigation Team too pointed out that there is a possibility of CCD(s) which were damaged during the incident on 15th December 2011 went undetected and continued to transverse the system on the 16th and caused misalignment of the third rail on other sections, leading to ‘claws’ dislodging and recurrence of the incident on 17th. This train was identified by the LTA as KNS 335F (335/336). It was the first train to be involved on 17th December, and also involved on the 15th of December.
- However this was contrary to the reports of trains by the Criminal Investigation Department later. On 17th December, the first train encountering problem after passing the section after Newton was T136 operated by KNS 319F (319/320), and the first non-revenue service train passed the sector, then turned around and served the Northbound to Yishun, turned back to Toa Payoh where it was then turned around again towards Yishun and finally stalling at Bishan was T123 operated by KHI 091F (091/092). 335F (335/336) was T119 which was out of the picture at that time, but was found to have damaged CCDs later.
The LTA also looked into the maintenance regime of SMRT, and find them in overall ‘comprehensive and satisfactory’. However there are areas which LTA pointed out in the address that SMRT lacked, or deviated, from standards.
- The maintenance of TRSA, it appears that SMRT only require preventive maintenance to be carried out on the ‘high speed ramp’ sections. This refer to the section where a CCD would enter or leave the third rail, at places where the third rail changes relative position, from the left of the train to the right, or at turn-outs (switches). Under the Mass Rapid Transit Corporation (MRTC) Maintenance Manual (Part 2 Section IX) it was noted that every aspect of the 3rd rail fastener assemblies ought to be inspected once every year, without differentiating the ‘high speed ramp’ or normal sections of the 3rd rail. Statistics submitted by SMRT showed that in 2006 to 2011, 79% of the dropped ‘claws’ happened to normal sections of the third rail. This raised some concerns to LTA.
- On the alignment of the 3rd rail, SMRT uses a Multi Function Vehicle (MFV) to check on this aspect. SMRT requires MFV runs to be conducted in 6 monthly cycles (what the Site Admin understands here is that a particular section of track/3rd rail will be inspected every 6 months). Based on SMRT’s Work Instructions (ETB/WI/PM/MC05 REV 01 ‘Measurement of Track Gauge, Superelevation and Versine by Track Geometry Measurement System), the tolerance is +/- 15 mm for vertical alignment of the 3rd rail, and +/- 15 mm for horizontal alignment of the 3rd rail. This defers from the limits set by MRTC Permanent Way Maintenance Manual Part 2 Section III, which recommends vertical alignment as +/- 5 mm, and horizontal alignment as +15 mm/-5 mm for the 3rd rail. LTA intends to review with SMRT in regards to this.
- Repeated out of tolerance readings. Three sections of the track along the NSL showed repeated lateral out of tolerance readings, namely,
- Newton – Orchard Southbound, between CH63925 and CH63970, a distance of about 65 meters, readings on 28th April 2011, 1st July 2011 and 11th October 2011. This location corresponds with the incident site of 17th December 2011.
- Marina Bay – Raffles Place Southbound, between CH59750 and CH59840, a distance of about 90 meters, readings on 5th September 2010, 29th November 2010 and 5th July 2011.
- Somerset – Orchard Northbound, between CH63060 and CH63180, a distance of about 120 meters, readings on 1st September 2010, 7th December 2010 and 8th April 2011.
- Paper work for the rectification following up the out of tolerance readings was also found not to be in good order. Some were not generated immediately after the readings were taken. There are also cases where the ‘notification’ date is later than the ‘actual’ and/or ‘scheduled’ start date of work, among other reasons.
Maintenance of Wheels. Records from SMRT indicate that there has been a general reduction in wheel-profiling works between 2009 and 2011. For the same period, there had also been an increase in average time to complete wheel-profiling corrective maintenance. This is in-spite of higher ridership and increased mileage ran by the trains.
On Incident Management, it was found that disruptions exceeding 30 minutes did not occur for more than 50 times in a period of 5 years between 2007 to 2011. LTA took charge of monitoring the bus operations, contacting taxi companies to broadcast the areas affected to the taxi drivers. LTA’s Intelligent Transport Systems Centre (located at River Valley Road) also participated in regulating traffic light changes to facilitate bus movements and liaise with the Traffic Police.
LTA recognizes that the current SMRT Rail Incident Management Plan (RIMP) focuses on recovering train service and safety of passengers, it should also take into account the welfare and comfort of commuters and the transport system at large.
A number of aspects in SMRT’s incident management was executed in accordance to the RIMP and Standard Operating Procedures (SOP), such as provision for an alternate MRT route (thanks to the structure of the initial phrase of MRT operations, the Raffles Place Cross-overs). However there are areas which were performed not as well, for example;
- Communications with the commuters affected, as well as the public, could have been better.
- Bus bridging drivers could have been better briefed on the routes to take.
- Crowd management procedures could have been better thought out and implemented.
In addition to the above, LTA feels that the key areas listed below could be further improved;
- Public communications within the station premises, on board trains, and to the general public with regards to service delays, service disruptions or emergencies.
- Provision of free bus bridging services.
- Detrainment of commuters from immobile trains.
- Appointment of a key contact person by operators.
- Notification of LTA and other government agencies.
CP 7 of 2012 on incident management during train service disruptions was issued by LTA and sets the minimum standards and/or response times for the above areas of improvements.
- CP 7 specifies the relevant timelines of which operators have to notify LTA of certain incidents that may occur within their train systems. Incidents such as outbreak of fire and require the help of the SCDF, operators have to inform SCDF immediately so that SCDF can respond in a timely and robust manner.
- Public communications wise, announcements during the disruption could have included information such as the cause as well as when normal train service could be expected to resume. This is also to avoid overcrowding at stations as experienced during the disruptions. [The Site Admin however would like to add in that according to surveys done in the past in overseas systems, the time which train service is expected to resume is very often a question which a commuter will ask, however to crew rectifying the problem, up to even to the controllers, no one will have a specific time to provide. This is one point LTA should take note when implementing this.]
- Detrainment of commuters, LTA views that it is a solution not to be taken lightly as this places passengers in an unfamiliar environment with much trackside equipment (a wonderful surprise to an enthusiast but not to an average passenger!). To assist the operator, LTA formalized the decision making process and setting out the relevant considerations, namely,
Whether the emergency lighting and ventilation and communication systems are functioning properly within the train,
Safety and health of commuters on-board the train, including without limitation whether any commuter has fainted or require medical care; and
The time taken to push and/or pull the train to the nearest station for detrainment directly onto the platform.
LTA too noted that on 15th December 2011, two trains (SIE 215F and KHI 067F) encountered a situation where back-up battery power of the train has failed. There was therefore no air-conditioning on the both trains and emergency lighting was also dim. Addressing this, operators should in the future, as far as possible, turn on tunnel lighting and tunnel ventilation in the immediate vicinity of the immobile train provided it is safe to do so. This will alleviate the lack of air-conditioning and bolster the limited lighting, more importantly add comfort and assurance to the affected passengers. [The Site Admin would like to add on that the 'unsafe' scenarios include fire in the tunnel, where ventilation cannot be turned on to minimize the spreading of flames.]
- Provision of bus bridging services, LTA has since set out the minimum standards and response times for bus bridging services in CP 7. LTA also implemented free travel on public buses at designated bus stops along an affected sector, so commuters can continue their journey without waiting for bus bridging services. The fares of these free bus travel will be borne by the operator of the disrupted MRT line.
Opening Address by SMRT
SMRT takes the incidents very seriously and approach it in threefold, first to learn from what happened, secondly to guard against it happening again, and thirdly to be much better prepared to handle future incidents. With that line of approach the Internal Investigation Team (IIT) was set up on 3rd January 2012. Over 10 weeks SMRT interviewed numerous staff, solicited public feedback and suggestions and gathered inputs from Subject Matter Experts (SMEs).
Separately SMRT also approached its peer, the Taipei Rapid Transit Corporation (TRTC) to review its operations. It also engaged vibration experts from Sinclair Knight Merz Australia (SKM) and ST Engineering to investigate the root cause. Industry experts such as WS Atkins (founded in 1938 as a multinational engineering, design, planning, project management and consulting services company) was also engaged to review its maintenance regime and incident management procedures.
Background of the system
The predecessor of Land Transport Authority (LTA) is the Mass Rapid Transit Corporation (MRTC), which was established in 1983 under the MRTC Act, vested with necessary powers to build and operate the MRT system in Singapore.
MRTC was responsible for the design of the MRT infrastructure and operating assets. Brecknell Willis for example, was engaged by MRTC in 1985 to design and develop the third rail network.
SMRT Ltd was incorporated in 1987 and granted a license to operate the network from MRTC. In 1995, roles and function of MRTC was transferred to the new statutory board, the LTA. LTA owns the MRT infrastructure (ie the stations, the tracks, including the third rail). The LTA also plays a role of a regulator.
SMRT is required to comply with LTA Operating Performance Standards, technical maintenance standards, replacement and renewal prescribed by LTA from time to time. SMRT is also responsible for formulating maintenance plans, but these plans must be submitted to the LTA on a regular basis. Clearance and approvals must also be obtained from LTA in order for SMRT to implement modifications, in particular any modifications to assets and infrastructure owned by LTA. RIMP and SOPs are also submitted to LTA.
LTA also procure the trains for SMRT. They take charge of the tender process and testing of new trains. LTA determines the specifications of new trains and leads acceptance and commissioning tests.
SMRT is consulted during the preparation of tender to submit a wish list of tender specifications to LTA, however these inputs are mainly in areas of operations and maintenance, not in design. The wish list will be based on experience as an operator, and also suggest on improvement to train specifications. LTA makes the final decision whether to accommodate or not.
In 2000, SMRT Corporation was formed and hold 100% of SMRT Ltd, subsequently its name was changed to SMRT Trains Ltd in 2004.
Chronology of events 15th December 2011, Combined with Criminal Investigation Department Chronology
Four trains stalled on that day. About 647 pm, Train 134 (SIE 215F, instead of KHI 067F as previously thought of) stalled just before Dhoby Ghaut station. One minute later, T139 (KHI 067F) stalled just before Orchard station. There was no immediate decision to detrain passengers onto the tracks, as general orthodoxy sees that this is a last resort. There are fewer risks involved in sending a rescue train to push the defective train to a station and let passengers out to the platform. That is what OCC started to plan and execute for T134 and T139.
At 655 pm, T133 (KHI 081F) stalled at Braddell station. Passengers were able to detrain to the platform. At the same time, OCC instructed a following train, T118, to do a ‘line-clear’ from City Hall northbound. At 707 pm, the Train Officer (TO) of T118 saw the sagged third rail.
At this point, OCC realized that a rescue train cannot be sent to T134 (and same to T139 as the whole section between City Hall and Orchard is down, and the closest cross-over is at Newton). A change of plan was required, and detrainment exercise was put into motion. A station manager was sent to the tunnel with a volt meter after power within the sector was switched off, third rail short circuiting device was placed, and then he walks to the stalled train. The TO then lowers the detrainment ramp. All these took some time. Detrainment for T134/SIE 215F started at 726 pm (39 minutes from start) and was completed at 805 pm (78 minutes from start, detrainment took 39 minutes, estimated 1200 passengers). During the detrainment at 739 pm, SCDF received a call from a passenger on-board, saying ‘someone fainting’. An ambulance (112 from Central Fire Station) was dispatched and arrived at 743 pm.
Meanwhile T112 (running towards Marina Bay) was identified as the rescue train for T139. T112 bumped off its passengers at Newton station at 704 pm, changed to T912, crossed over against the direction of travel (on Restricted Manual mode) and coupled to T139 at 731 pm. Detrainment started at Orchard for T139/KHI 067F at 733 pm. It was after the detrainment that the TO discovered someone has smashed open a window on car 3067 (6th car in the direction of travel).
The fourth train that stalled was T128 (KHI 111F). It bumped off its passengers at Toa Payoh station 656 pm after finding fault lights illuminated. It was then sent as T928 to push T133 which was stalled at Braddell. Shortly after leaving Toa Payoh, T928 lost traction power.
By 730 pm, teams from Permanent Way and engineering arrived on site. Southbound services were resumed at 819 pm, and at 856 pm the sagged third rail was restored. Line clear was done between Dhoby Ghaut and Braddell at 900 pm. At 952 pm the engineering team reported that the third rail between City Hall and Dhoby Ghaut stations had been gauged and rectified. Traction power was restored at 1042 pm, stalled trains were hauled back to depot starting 1113 pm, speed restriction (controlled from OCC) was implemented on the affected sectors at 1126 pm as precaution, and services resumed at 1140 pm, a total disruption duration of about 294 minutes.
After the close of service, post-incident checks were performed on tracks and trains to ensure that it is safe to run operations on 16th December.
Prior to operations on 16th, third rail cover and spacer for damaged third rail components were repaired at incident area. Manual third rail gauging was done on the Northbound track from Marina Bay to Orchard. The Multi Function Vehicle (MFV) was also used to do gauging from Jurong East to Ang Mo Kio in both directions. Track inspection was done in both directions from Marina Bay to Bishan save for two sectors; Bishan to Braddell (Southbound) and Newton to Orchard (Southbound). This two sectors could not be completed due to late track access as 12-car push-out of affected trains had to be done after revenue hours.
The conditions of all CCDs on all affected trains were inspected and checked, and three of the trains had their CCDs repaired on the same day.
During normal service on 16th December 2011, a permanent way team conducted track inspection from train cabs from Ang Mo Kio to Marina Bay (both directions) from 1420 to 1700 hrs, with special attention given to Ang Mo Kio to Bishan both directions, Bishan to Braddell Southbound and Newton to Orchard Southbound.
After close of operations on 16th December, the remaining two trains with damaged CCDs were repaired, and the MFV was scheduled to do a track geometry scan on third rail and running rail between Clementi to Raffles Place Eastbound and Newton to Marina Bay (both directions).
The plan was for the MFV to perform track geometry measurements (TGM) from Newton to Marina Bay station, then the same from Marina Bay to Bishan Depot.
During the passage of the MFV from Newton to Dhoby Ghaut, TGM images were lost on two occasions. At 210 am MFV personnel found that the left detection mirror of the MFV had cracked. The mirror was replaced and broken one discarded.
The personnel then did a check on the MFV and accompanying Schoma electric locomotive, at 226 am they found that the mirror was the only item that was damaged. MFV personnel then told Permanent Way to check on the chainages within 100 meters that TGM images were lost. A team left Bishan Depot and arrived at Newton station at 345 am, inspected the two zones for any signs that may have caused the mirror to crack. They did not find any abnormalities, and ended their inspection at 420 am, 10 minutes before the close of engineering hours at 430 am. To-date the cause of the cracked mirror is still unknown.
At this point of time the Site Admin would like to highlight that a picture of the cracked mirror was presented later with the CID report, and does show signs of dragging damages, with the leading side of mirror cracked and rear side totally removed by impact. According to his basic engineering knowledge, it is very likely that the mirror was damaged by the third rail and/or the third rail insulation cover.
At material time, SMRT feared the worst that a train somehow exceeded the kinematic envelope and infringed the third rail, knocking off the claws and causing the third rail to sag. [This theory that the Kinematic Envelope was exceeded was presented by the Site Admin before the release of this report in December 2011] The inspection of the third rail was only as a precautionary measure. The fact that a small portion of the track had not been checked before train service on 17th December due to the cracked mirror is not a major cause for concern. All this while the focus had been to inspect trains to make sure there are no foreign objects sticking out, that can infringe onto the third rail.
Chronology of events 17th December 2011, Combined with Criminal Investigation Department’s Chronology
Five trains stalled on the North-South Line on this day, on both directions between City Hall and Ang Mo Kio. With experience however, the controllers were able to begin detrainment more quickly, before the third rail was discovered to have sagged.
At 644 am, T137 (KHI 017F) reported no line voltage. It had previously lost power and fault lights illuminated along the way. The train coasted, then underran the station platform, and the TO was instructed to detrain passengers through the driver’s side cab door. Detrainment started at 651 am and was completed 658 am (7 minutes).
The following video is taken by a commuter on T137/KHI 017F.
At 645 am, T132 (KSF 507F) left Orchard station, and was instructed to bump off passengers at Somerset station. It arrived at 646 am, and disembarking proceeded. He was then instructed to do ‘line clear’ to Dhoby Ghaut, left Somerset at 703 am but lost traction power soon after, but still managed to coast to Dhoby Ghaut station, reaching at 704 am and train was shut down.
At 652 am, T113 (KSF 519F) arrived at Orchard station. It was the following train of T132, and had previously been held at Newton station for several minutes. The TO of this train made delay announcements, and finally T113 left Orchard at 701 am (delay of 9 minutes). At 702 am, T113 reached one signal block away from T132 which only moved off the next minute, but after T132 left Somerset and when T113 is commanded to power up, the train rolled backwards. The TO applied emergency brakes and informed OCC. At 717 am, detrainment onto tracks was started for T113, affecting about 500 passengers. At 733 am detrainment was completed for T113, taking 16 minutes and was 30 minutes after onset (the application of emergency brakes to prevent roll-back).
The following was a little perculiar to trains 136 and 123, hence a little detail will be required to further the understanding.
Train 136 (KNS 319F) first met with trouble at 628 am, when entering Orchard station towards Marina Bay, the TO felt a jerk and air con fault light was illuminated. Fault lights came on and off intermittently, it arrived at Marina Bay at 639 am, leaving the station for Jurong East at 643 am. At 647 am T136 experienced air-con fault lights while stopped at City Hall station. At this point of time, T137/KHI 017F was detrained at the same station, but opposite direction. Interestingly on the way to City Hall, T137 had also experienced similar faults. T136 left City Hall at 648 am with sluggish movement (typical of certain trains with ’3M3T’ problems, explained later). At 703 am, T136 arrived at Bishan. All this while the train suffered from intermittent air-con fault light. 706 am, T136 arrived at Ang Mo Kio station and TO found that his train was running on ‘Emergency mode’, detrainment at station was carried out and it was then the TO found that the line voltage light was off.
Train 123 (KHI 091F) ‘enjoyed’ a relatively long duration through this incident, starting along the timeline at Raffles Place at 626 am, experiencing no problems even as T132 bumped off its passengers at Somerset at 646 am, at the same time it arrived at Ang Mo Kio station. It continued its journey towards Yishun, where it was turned around sometime around 712 am when the RIMP was activated. As mentioned previously, SMRT’s procedures are to recover train service, this was probably done according to the plan and actually sensible to maintain service along sectors which are capable of doing so. T123, now travelling towards Marina Bay, reached Toa Payoh at 834 am, when it was instructed to bump off passengers, pick up passengers heading towards Yishun/Woodlands/Jurong East on the same platform and ‘change ends’, joining the Northbound track using the ‘Toa Payoh North’ cross-over. At 835 am the train left Toa Payoh and entered the Northbound tunnel. About 100 meters from Bishan later on at 838 am, the TO heard flickering sounds and realize that the line voltage indicator was flickering. The moment he arrived at Bishan station, the train lost power and air-con, compressor and converter fault lights illuminated. OCC was informed and passengers detrained.
Other trains were found out to have damaged CCDs later on, but did not stall on the tracks. Here is a partial list of them (part of the CID Report).
15th December 2011
Direction of travel indicates City Hall to Dhoby Ghaut Northbound tunnel.
T151, operated by KSF 503F. On the direction of travel, the 6th car, 3504, had two of its right side (direction of travel) CCDs damaged. This was peculiar to this train (ie only rear-most car CCDs damaged).
17th December 2011
Direction of travel indicated Newton to Orchard Southbound tunnel.
T119, operated by KNS 335F. On the direction of travel, the 1st car 3335 had one damaged CCD (right rear), 5th car 1336 had both right side CCDs damaged.
T134, operated by KNS 329F. On the direction of travel, the 1st car 3329 had one CCD damaged (right front), 1329 right rear, 2329 right both, 2330 right both, 1330 right front and 3330 right rear.
T133, operated by KHI 105F. On the direction of travel, the 2nd car had one CCD damaged (right rear), 2105 right rear, 2106 right front, 1106 right rear and 3106 right front.
T135, operated by KNS 321F. On the direction of travel, all right side CCDs are damaged.
T111, operated by KSF 515F. On the direction of travel, 4th car 2515 had one CCD damaged on the right rear, same goes to 5th car 1515.
T127/927 operated by KHI 065F. On the direction of travel all of the CCDs on the right side are damaged.
The incident on 17th December occurred during off-peak Saturday and relatively lesser commuters were affected (about 90000, 30000 less than Thursday). Sagged rail was discovered at 840 am by the Pway crew, and restored by 925 am. Service resumed at 150 pm.
After the incident, at 1 pm Rolling Stock staff began to check CCDs at all three depots operating the NSEWL. At 330 pm SMRT and LTA met to discuss a system-wide inspection of the third rail on night of 17th December 2011.
At 5 pm, Rolling Stock staff were deployed to stations to conduct visual checks on CCDs of passing trains. Engineering hours was also extended to 18th December noon to facilitate joint inspection by SMRT and LTA. Damaged CCDs were replaced, and a system-wide programme to secure claws using cable ties was put in place. The Kawasaki-Sifang (KSF), newest trains of the system, was also confined to operate on the East-West Line as it has lesser Floating Slab Track (FST) sections than NSL. A speed restriction was put in place for sections with FST.
The IIT and panel of experts in vibration found that the incidents were triggered when two or more adjacent claws dropped and caused the third rail to sag towards the trackbed. The sagging was then aggravated by the next train. On 15th December 2011, Train 151 (KSF 503F) was identified to be the leading train. More claws were dislodged when the springs built on the CCDs spring the collector shoes upwards following the sag. Subsequent trains’ CCDs had their collector shoes sheared off by design, or twisted badly that power can no longer be transmitted. Traction power was lost and train controls indicated a power fault due to loss of line voltage, thus stalling the train.
The IIT concluded that the incidents are not independent of each other. In other words, it is probable that CCDs shoes on some trains affected became twisted and continued to transverse the MRT system, affecting other claws (most of which perhaps, occurred on the 16th of December).
The Root Cause
A rare confluence of factors may have well caused the incidents, which is highlighted below.
The curvature of the track where a cant (tilt) has been provided. When the third rail is situated on the ‘high’ side of the tilt, ie on the outside of the curve, the bracket supporting the third rail (Third Rail Support Assembly/TRSA) is also tilted, leading to a higher probability of dropped claws.
The FST sections, both incidents occurred at FST sections. SMRT’s expert from SKM will explain that FST sections is likely to experience higher vibration levels at low frequencies than ballasted or concrete slab tracks.
The claws, part of the TRSA. SKM analyzed that it is possible for a claw to be relieved of a significant percentage of the vertical load from the power rail even when operating within normal third rail gauging and CCD contact tolerance. Common wheel defects (ie flat wheels) would further reduce the restraint on the claw and make it more prone to dislodge. SKM considers the design not sufficiently robust now.
Dropping claws had been classified as Severity Index – Urgent (SIU) and not a major problem in the past. It requires rectification within 24 hours. Such incident were dealt with and not escalated to senior management.
The number of dropped claws had been on a downward trend since 2007. The number of dropped claws for the year in 2009 was 29, 2010 was 29, and in 2011 up to the two incidents it had been 16. On ballasted track it had been 15, 26 and 16. On FSTs, 14, 3 and 0.
Higher vibration levels. SKM found the elastomeric bearings of the FSTs at the incident site to be good. SKM’s view is that the increased number of dropped claws in December 2011 may be due to a traveling train defect, such as wheel defects, rather than deterioration of FSTs. SKM found that vibration can be four times higher on trains that have non well-profiled wheels and this may have contributed to actions where the CCD shoe could have been forced upwards significantly and lessen the static load that the third rail place on the claws under normal circumstances.
Higher upward forces by CCDs. A new CCD may exert as much as 2 to 3 kg force than a seasoned one, as found by deflective tests conducted by ST Engineering. This may have combined with other factors to dislodge claws.
Material defects. Experts from TUV SUD found that a defective fastener and insulators of the TRSA attributed to the incidents as a main cause. The crack on the fastener was due to non-uniform structure, casting defects and residual stresses inside the material. The cracking of the insulator was due to thermal plastic contamination of the raw material of the insulator. These may be manufacturing defects.
TUV SUD report also mentioned that 17th December incident is due to a cracked third rail. The material is seriously embrittled over a long period of operation and attributed to abnormally high impurities of the third rail material and the segregation of such impurities on grain boundaries. These are inherent and latent defects.
A summary of these root causes would be that the incidents were caused by a rare confluence of factors, none of which would have caused the incident on its own.
Three carriages of a train experienced blackout
It was found that the last three cars of T139 (KHI 067F, subformation 067 with cars DT 3067, M1 1067 and M2 2067) experienced a blackout and loss of ventilation, leading to a commuter breaking a window with a fire extinguisher.
Picture taken by a commuter on KHI 067F. Picture reproduced from Victor Lim’s blog
The battery was found to be still within its useful lifespan of 7 years and also underwent three-weekly voltage measurements without any fault found. The IIT recommended that the batteries be changed out before its expiry date, or other measures to ensure a more satisfactory performance of batteries.
According to the CID report, each KHI and SIE train is equipped with 80 Ah batteries, which is expected to last KHI: 45 minutes at 80% charge and SIE: at least 45 minutes. Newer KNS and KSF stocks have 110 Ah batteries, but no data as to how long the batteries can last. The batteries are charged as long as the train has access to third rail power, by means of a converter to 110 VDC and a battery charger system.
Enhancing safety and reliability of SMRT’s Train Operations
A few steps in the following months will aim to achieve this, in the form of,
- System-wide replacement of claws,
- Installation of wheel impact load detection system,
- Purchase of an additional MFV, and
- More frequent maintenance checks.
Design of the RIMP
The RIMP spells out tasks and responsibilities of various SMRT departments managing an incident and classifies the severity of an incident into the following color coded levels:
- GREEN-2 Incident: An incident that has minimum disruption to train service schedule and where the issue can be resolved in a short time.
- GREEN-1 Incident: An incident that results in a longer recovery time to status quo for example an accident in the train depot that leads to a reduction in the train service frequency.
- ORANGE-2 Incident: And incident that disrupts train service by up to 15 minutes during peak hour travelling and 30 minutes during off-peak hour travelling. The cause of an ORANGE-2 incident is typically technical failure that can be resolved easily by the SMRT staff.
- ORANGE-1 Incident: An incident that disrupts train service by more than 15 minutes during peak hour travelling and more than 30 minutes during off-peak hour travelling. The RIMP will be activated for an ORANGE-1 incident.
- RED-2 Incident: An incident that disrupts train service by up to one hour but without mass casualty/fatality ie commuters injured/killed by a train.
- RED-1 Incident: An incident that involves mass casualty or fatality ie many injured or killed as a result of a major accident or terrorist attack. Uniformed and emergency public agencies such as the Police and SCDF will also be activated accordingly.
As a side note, in recent years the RED-1 level has not been met other than exercises simulating terrorist attacks such as Exercise Northstar. The only case that the Site Admin can draw reference to is the 1993 collision of two MRT trains at Clementi Station, however then MRTC had still been in existence. The 2008 Simei Accident involving a rail grinder and locomotive during engineering hours is barely RED-2.
The IIT and experts found that the RIMP needs to be changed in terms of classification, for example, the incident on 15th and 17th December 2011 well deserve a RED-1, however due to the absence of mass casualties, it has been kept at RED-2 and the Police/SCDF were not activated automatically. External assistance would be vital for management of overwhelming crowds.
Secondly, the RIMP needs to be fundamentally reviewed so that it can handle multiple train, multiple station disruptions where the cause may be unknown. The plan may be useful to address single train, single station or known-causes disruptions.
Bus bridging on 15th December were aggravated by festive season traffic congestion, while 17th December had been easier to manage, and initial concerns were better managed.
IIT commented that while SMRT’s Bus Bridging Plan is effective in handling single station incidents, it cannot handle incidents such as the one on 15th December. This is due to the vast difference in capacity of a bus and a train (80 versus 1500). Channeling commuters to other working parts of the public transport network (other rail lines) is the only tenable option. On 15th December, staff urged commuters at City Hall to join the East-West Line or take a short walk to take the Circle Line at Esplanade.
The Site Admin would like to note here that due to the extreme constraint of our system, this principle actually doesn’t work as well as major cities with parallel rail lines such as London and Tokyo. For a commuter heading to the north ie Woodlands on 15th December evening, he can take the East-West Line to Jurong East, transfer to the operational part of North-South Line and reach Woodlands. With a disruption, he may not be able to get on-board any East-West Line train! Alternatively, he may choose to take the Circle Line to Serangoon, transfer to the North-East Line to Hougang and take service 161 direct to Woodlands, a long detour which many common commuters does not know of. Detours would be easy if a commuter knows of them, however again noted by overseas experts, its always hard to provide detour option(s), especially the best ones. A commuter has to come up with his own, this is where backstage support such as websites that provide travel directions, becomes essential. LTA and SMRT cannot be taken for granted to provide a detour for commuters, as they have their hands full to manage the incident.
SMRT staff are always trained to reserve detrainment-to-track as a procedure of last resort. That is why on 15th December, passengers on T139 (KHI 067F) was not to detrain, but wait for T112/912 to couple onto and haul T139 into Orchard station. While waiting for T112 to arrive however, a passenger fainted in T139. The rationale is simple; platform detrainment is safer than detrainment-to-track.
IIT however commented that push-out operation (12-car) can no longer be preferred in isolation of commuters’ level of tolerance in a stalled and crowded train. The tolerable time commuters can remain within a stalled train, subject to ventilation and lighting, should be used as the primary determinant for the expedient method to be used to recover the stalled train before a detrainment-to-track is undertaken.
Following IIT’s work, SMRT has decided to reduce the stipulated time for detrainment of passengers from stalled trains from 45 minutes to 30 minutes since the onset of incident.
It is apparent that train drivers involved in the 15th December incident had informed passengers on stalled trains that a technical fault has occurred and repeated statements of apology, however SMRT accepts that this is insufficient to allay the anxiety, distress and anger among commuters.
SMRT has first, refined procedures towards making timely and regular announcements and service updates to passengers in stalled trains and affected commuters. Additional manpower has been allocated. An OCC communication team is also instituted to make public announcement to passengers.
Second, SMRT now provides service updates to general public via a comprehensive range of channels including both new and old media. The first announcement will be made within 15 minutes of a delay and updated every 20 minutes.
At this juncture the Site Admin would like to note the new media, being Twitter itself, is updated by SMRT sporadically through a minor disruption, with up to one hour gaps in-between. This is rather reasonable as information needs to be confirmed from the ground (ie trains) before being broadcasted. However, the time needed must be reduced. Updates from tweets on the ground may appear much faster as they are not filtered before broadcasting. Whether these updates are to be followed, it depends on yourself as a commuter. A five minute delay may sound like eons to a non-seasoned commuter, but it may sound perfectly tolerable for the Site Admin.
Third, SMRT has also set up information counter at stations during service disruptions. Information on alternative bus services and MRT system map will be provided.
Managing increased ridership and aging infrastructure
SMRt took steps to try and deal with the increasing passenger loads. Although train are built to hold 1400 passengers, SMRT set its own target to keep passenger loads below 1000. At trigger point of 800 passengers, additional trains are added to the train schedule until a minimum of 2 minutes interval was reached. IIT observed that the daily train trips increased from 1293 per day in 2008 to 1496 in 2011.
Pre-incident decision to purchase an additional MFV are examples that SMRT has not hesitated to spend money. It is important to stress that while SMRT cannot control issues such as population growth which has increased the strain on the MRT system, cash is not saved either. Another example would be budgeting ‘bottom-up’ ie driven by the needs of different departments in SMRT rather than from the top management.
The Site Admin too noted several questions raised by the panel of the Committee, especially from the Chief District Judge. Here are a few points raised on that day.
How many shoes can a train afford to lose before losing traction totally?
Are there any gauges/lights to show which shoe is lost?
The existing covers for the third rail is opaque. Is it possible for the covers to be transparent to facilitate quick inspections?
Should there be a failure, an incident like the 15th, would 45 minutes for the backup battery be sufficient?
The Site Admin would answer these questions to the best of his knowledge as an enthusiast. Bear in mind the Site Admin is just an enthusiast. If you are really interested to find the (true?) answer, you can always refer to materials and deduce your own, or wait for the official COI report.
Q: How many shoes can a train afford to lose before losing traction totally?
A: According to standing specifications, a typical 6-car train is in the power, or M:T ratio, of 4M2T. Each M car is approximately 35 tons, newer trains more heavier. According to M:T ratio theory and adhesion aspects, it would be possible to move the train off on level tangent track with just one motor car in operation. That means at least two shoes on one side of one motor car cannot be damaged. And this side has to be the one in contact with the third rail, which is normally located at the inner side of inter-station tracks, and away from a platform at a station. A train in theory would not be able to climb a positive gradient (upslope) with more than 20 ‰ (or 2 %) with two motor cars out. Traction would be equally hard for a train with one motor car out navigating a 30 ‰ upslope, a maximum for NSEWL.
On a side note, it has occurred to the Site Admin that many trains are operated with one motor car offline, without any particular car in direction of travel. This results in degraded performance for the train, evident in lower acceleration from the usual 1.0 m/s (~3.6 km/h/s).
Q: Are there any gauges/lights to show which shoe is lost?
A: Currently the main circuit boards and fault indication lights on SMRT stocks are simplified to the extent that knowing exact fault location is hard. Newer stocks such as the 2000 KNS and 2011 KSF have a digital display on the control console for vital components such as Static Inverter (SIV), ECB and door status. Older stocks do not have such luxury. As highlighted by the CID investigator, the line voltage indicator in the driving cabs of DTs only show the line voltage for the DT itself. For individual cars’ line voltage the driver will have to walk from car to car and see the board mounted at ends facing the M2 Car. For the last DT car he will have to enter the cab to see the indicator, a journey of more than 130 meters from end to end!
Adding on to this, the Site Admin would suggest SMRT to work with LTA to design a Train Information Monitoring System akin to those found on new stocks. These comprehensive TIMS system will enable the train driver to receive information about all cars in a train formation at one go. This will require the adaptation of electrical signals for passing of information from one 3-car formation to another 3-car formation.
An example of TIMS on Japan Railways (JR) East E233 EMU stocks. What is indicated include the time, speed, mileage, outside temperature, door statuses (inclusive of opened, closed, fault), announcement system (on/off), interior lights (on/off), air-conditioning statuses, cabin temperature, humidity, passenger load factor (with reference to JIS E7103) and passenger count (by average weight). These are actually very basic information which the Site Admin believes is critical for a train driver to know, especially when he is at one end of the train.
Q: The existing covers for the third rail is opaque. Is it possible for the covers to be transparent to facilitate quick inspections?
A: This is possible however the material sourced must be able to sustain the temperature and sudden weather changes in Singapore as well. Not to mention transparent covers face problems of rain/water stains, of which additional cleaning will be required.
Q: Should there be a failure, an incident like the 15th, would 45 minutes for the backup battery be sufficient?
A: As stipulated in the new guidelines, the decision making to detrain passengers to tracks is shortened from 45 to 30 minutes. This may sound sufficient for the batteries to last, however a single detrainment would take anywhere near an hour, true for the case of T134/SIE 215F on 15th.
Carrying more batteries would not be a good option, given that the underfloor space for equipment is limited. Same goes to bigger capacity batteries. A solution could be found by reducing fan speeds while increasing tunnel ventilation velocities. Installation of hopper windows would also be feasible, however it could be argued that how often this would be used (hopefully never again?). Constraints being so, it is even more critical to work on plans for multiple train/station failures, and incidents with no known cause. Only then a train can be evacuated ‘quickly’. The Site Admin acknowledges that the current plan for OCC to formulate a detrainment plan is 8 minutes as mentioned in the session, for train-to-platform, train-to-train or train-to-track detrainment options.
Answering the questions is one thing, the Site Admin would like to post questions as well, whether for the public to think about or for parties involved and whom may come across this post to consider.
Rolling Stock Issues
- Why are the trains, after losing line voltage to multiple cars, still be allowed to coast and Emergency Brake-Valve Relay (EBR) not tripped by loss of line voltage (a possible indication that a train has derailed, for example)?
-Should a minimum limit be set for the system to trip when more than a number of cars have no access to third-rail voltage/line voltage?
- Under circumstances where an entire train has no access to line voltage, would it be safe to do let the train continue to coast, or allow the safety mechanism if designed, to stop the train where it is?
- This is critical for example that line voltage is lost and the driver is incapacitated.
- Why are diesel locomotives not prioritized to recover the stalled trains, instead using 12-car push/haul method, when it is apparent that the third rail had been damaged (although recovered before trains are hauled)?
- The Site Admin envisioned that SMRT would utilize diesel locomotives in this case as third rail power is obviously in doubt. One train stalled is always better than having an additional rescue one complicated in the scenario.
Multi-Function Vehicle and Permanent Way Crew Issues
- Why operations is resumed on 17th December with earlier damage to the MFV’s inspection mirror apparent that it may be caused by third rail sagging?
- It is said that the damage to the mirror cannot be established. That leads to the next question.
- Why is the third rail not thoroughly inspected from Newton to Orchard/Dhoby Ghaut after the MFV’s readings were lost at that sector? Was starting service for the day a pressure factor?
- Experience from 15th December, it must be said, should have contributed to the added awareness of the MFV crew and Pway personnel. Why was there no added emphasis on the inspection, when it was apparently something went very wrong and broke the mirror of the MFV?
Train Driver Issues
- Does a train driver have adequate training to deal with not only passenger emergencies, but technical emergencies? Is he taught to reset the system using basic reset buttons as shown in photos at the session, or he can open the main controller panels and identify faults? Should train drivers be trained on that perspective?
- Doubling a train driver as a technician has been a long standing issue of railway companies. More training is needed, but lesser staff will need to be called to rectify every fault a train throws up. More importantly, lesser inconvenience in terms of disrupted time can be achieved with a technician on every train.
- Should the train driver slide open the emergency ramp to improve ventilation? (Raised by LTA as well)
- Should a train driver carry a GSM (mobile) phone used only during duty for communication with the OCC?
-This company phone will alleviate the stress on individual controllers and also provide an alternative beyond radio sets, one of which developed faults during the incidents.
Station Management Issues
- Are station staff passing information in-time for OCC to deploy bus bridging to the most needed station?
- Upon disruption, should station staff be given the authority to notify Police for crowd control instead of waiting for instructions from OCC, when they determine that an overcrowding condition is preparing to manifest?
As the Site Admin highlighted in the past, it is difficult for any operator in the world to manage a disruption well, keep commuters safe and sound and happy, and resume train services quickly. They can only work towards this goal. As seen from the investigations, communications, technical know-how have become focuses on fields to improve. No doubt we may have chance to see a better system emerge after the Committee publish its findings and improvements made to the system. Meanwhile we have to make-do with this rather aged rapid transit system, or perhaps join in to make it possible? Your call.
The Site Admin would like to make clear that this article is made from excerpts taken from the documents prepared for the Press and audience at the hearing on 16th April 2012. Certain wordings have been swapped out however, does not affect the definition of terms or scenario. It would be of great interest to obtain the final findings by the Committee, as these documents presented are not exhaustive neither conclusive of what happened on the 15th and 17th December 2011. These documents are also unclassified in nature.