TSB Laboratory Report LP136/2013

1.0 Introduction

1.1 Description of occurrence

1.1.1

On 6 July 2013, at approximately 0115 Eastern Daylight Time (EDT), Montreal, Maine & Atlantic (MMA) freight train MMA-002 rolled uncontrolled and derailed loaded tank cars of crude oil in the town of Lac-Mégantic, Quebec. Most of the derailed tank cars lost their contents as a result of the accident, which contributed to a large fire that burned for several days. There were 47 fatalities, and approximately 32 buildings were destroyed. The crude oil leaked into the river and caused serious environmental pollution.

1.1.2

Preliminary examination of the derailment site determined that buffer box car CIBX 172032, immediately behind the locomotive consist, and the following 63 loaded tank cars derailed on the main track of a 4° 7' right-hand curve in the direction of travel (eastward), covering a No. 11 turnout. The locomotive consist separated from the derailed cars and split into 2 portions, with each travelling different distances before they came to a stop. After a significant time, the front portion of the locomotive consist moved backward (westward) and collided with the second portion, both moving a short distance further (westward) and coming to a final stop together.

1.1.3

The derailed buffer box car struck a stationary cut of cars on the siding track. The following 8 tank cars were scattered in separated jackknifed positions. The next 2 tank cars lay in the direction of the turnout siding, ahead of the main jackknifed pile-up of the rest of the derailed tank cars among which the fire and explosions occurred. The last 9 tank cars in the train did not derail. They were disconnected and removed back and away from the derailment and fire by the locomotive engineer and emergency responders. An aerial-view photograph of the accident site is shown in Figure 1.

Image

1.2 Background

1.2.1

Train MMA-002 consisted of 5 locomotives, 1 operation control car VB-1, 1 loaded buffer box car, and 72 tank cars loaded with petroleum crude oil. The train weighed 10 287 tons and was 4701 feet long.

1.2.2

The train was operated by a 1-person crew. Before midnight, it came to a stop on the main track of Station Nantes with an automatic application of the air brakes. The locomotive engineer applied hand brakes on the locomotive consist and the buffer car, and then released the automatic brake, but kept the independent brake (IND) of the locomotive consist in the applied position. The engine of the lead locomotive, MMA 5017, was kept running at idle to maintain the air brake supply. The locomotive engineer left the train and went to a hotel for rest, as indicated in his schedule.

1.2.3

A fire was detected on the lead locomotive sometime after the locomotive engineer left (LP181/2013). Local firefighters came and put out the fire. A local MMA engineering employee was called to attend to the fire site. The engine of the locomotive was shut down, and the train was left unattended again. Approximately 59 minutes later, the train started to move down the descending grades, and accelerated all the way until it reached the town of Lac-Mégantic, where it derailed.

1.2.4

The lead locomotive, MMA 5017, was equipped with a Quantum Engineering Incorporated (QEI) locomotive event recorder (LER) version no. S45E, serial no. 0204100033. The recorded data in the “extend log” was downloaded from MMA 5017 by a MMA staff member soon after the accident. The Sureté du Québec (SQ, Quebec police) then seized the locomotive consist for their criminal investigation. When the SQ released the locomotives to the Transportation Safety Board (TSB), the LER records in the locomotives from before and during the accident had been overwritten by later recordings, so no meaningful data could be obtained from the other locomotives. The LER download from MMA 5017 was the only recorded data available for TSB analysis.

1.2.5

The train was also equipped with an end-of-train (EOT) sense and brake unit (SBU). The SBU was sent to the TSB Engineering Laboratory for examination (LP 132/2013). The records in the DataFlash were extracted and converted into Excel spreadsheets. The EOT SBU download data were provided to this LP officer (Appendix A) for a comprehensive analysis of LER and SBU data together.

1.2.6

The TSB investigation team also obtained a copy of the standard report of the public crossing at Mile 117.11, Moosehead Subdivision, that indicated the activation of the crossing signal and protection. The time record was calibrated by an independent crossing company. This record (Appendix B) was used as a reference in the synchronization and calibration of the downloaded LER time records.

1.3 Engineering services requested

1.3.1

The preliminary investigation found that the train experienced a number of events, including unattended parking, fire, engine shutdown, runaway, derailment, and explosion. There were several broken knuckles, including those between the second and third locomotives, indicating that the locomotive consist had separated and rejoined. The derailed cars scattered in several small, jackknifed groups as well as in a main pile-up. The downloaded raw LER data file was sent to the TSB Engineering Laboratory, and a complete analysis was requested to assist the investigation in determination of the times, locations, and related parameters of the sequence of events leading to the derailment. A laboratory project was opened for this LER data-analysis work.

1.3.2

The QEI Quantum Desktop Playback software was installed, and the raw download file was uploaded successfully. When first reviewed, the data was found to be not calibrated. In order to calibrate the data, measurements were taken for a number of reference locations, wheel sizes, travel distances, and times. The SBU and crossing records were reviewed, and an attempt was made to synchronize and analyze these data with the LER data. These data were further correlated with those of the on-site examinations and other investigation findings. A number of events of interest were identified with precise parameters. The sequence of events leading to the derailment was provided to help determine the most likely derailment scenario as well as potential causes and contributing factors.

1.3.3

This LP report describes the analysis of the downloaded LER data from the lead locomotive, MMA5017, including calibration of records, identified events of interest, calculation of the locations of vehicles in the train at the times of significance, and discussion of the likely derailment scenario.

2.0 Calibration

2.1 Wheel size and travel distance references

2.1.1

The wheels of the lead axle of locomotive 5017 were measured by the LP officer for the locomotive wheel examination project (LP182/2013). The circumference measurements were 126.75 inches for the left wheel and 126.25 inches for the right wheel. The average circumference for the 2 wheels of the lead axle was 126.50 inches, equivalent to a diameter of 40.266 inches, which was a little more than the labeled nominal value of 40 inches.

2.1.2

The site survey team (LP167/2013) determined that, before the runaway, the lead locomotive was located at Mile 6.718, Sherbrooke Subdivision, and that the final stop of the lead locomotive after the accident was at Mile 116.412, Moosehead Subdivision. The travel distance of the lead locomotive between these 2 locations included the return segment of the front portion of the locomotive consist during the last backward (westward) movement. The TSB engineering team also learned that the Moosehead Subdivision ends at Mile 117.14, where the Sherbrooke Subdivision begins at Mile 0.0 (Figure 2).

Image

2.1.3

The measured average wheel size was entered into the LER data analysis software to adjust the mileage record channel and the speed channel accordingly. The resultant LER travel distance matched (within the resolution level of the LER mileage records) the measured distance between the stop location of the locomotive before the runaway and the final stop after the derailment. This finding confirmed that the mile and distance data could be considered to be properly calibrated.

2.2 Time calibration references

2.2.1

The data downloaded from the crossing at Mile 117.11 on the MMA Moosehead Subdivision (Appendix B) showed that the alternating current (AC) power was cut off at 01:19:29.45, and the event identified as gcp1 detected the train at 01:19:31.04. A calibration of the time was performed by an independent company (X Rail), and it was determined that the data needed a time correction of −4 minutes 1 second. In other words, gcp1 recorded its detection of the locomotive consist as 01:19:31.04, but it was actually 01:15:30.04.

2.2.2

The location mile of the sensor/detector gcp1 that detected the locomotive consist at 01:15:30.04 was the location of the West approach limit, located 650 feet plus half of the 130 feet island, or 715 feet west of the centre of the crossing located at Mile 117.11. Therefore, the detector/sensor was located at Mile 0.11 of the Sherbrooke Subdivision.

2.2.3

The time channel of the LER download from MMA 5017 was calibrated using the detection of the lead locomotive at Mile 0.11 Sherbrooke Subdivision at 01:15:30 as a reference point. An adjustment of +01:01:31 was applied to the raw LER time channel.

2.3 Verification among LER, SBU and crossing signal downloads

2.3.1

The SBU recorded the SBU status, EOT brake pipeline pressure (BP), and MOVE/STOP activities, but its GPS position channels were empty. All parameters except the EOT BP were sampled at 60-second time intervals. The MOVE/STOP parameter did not capture detailed speed data. The 60-second time interval made it difficult to synchronize the SBU data with the LER data.

2.3.2

The EOT BP was sampled at shorter intervals. However, changes in EOT BP lagged behind changes in lead locomotive BP by an uncertain time, as a result of the propagation time of the brake air along the train pipe from the locomotive to the end of the train. Therefore, the EOT BP could not be synchronized with the LER BP events.

2.3.3

The SBU data did, however, provide a record of the EOT BP that linked the brake air state along the train brake pipe, so the automatic brake application and release could be confirmed. The MOVE/STOP channel also provided a simplified overview of the train’s motion. These 2 channels were used to complement the LER data analysis, as presented in the analysis below.

2.3.4

The downloaded crossing signal data recorded mainly the crossing activations and crossing state. The detection of the train arrival was utilized for time calibration and synchronization reference between the LER records and the crossing records. Another record was the speed of the train while passing the crossing. However, there was a difference between the LER recorded speed and the crossing record. Considering that the LER mileage, travel distance, and time had been accurately calibrated, the derived LER speed records were deemed more reliable than the crossing speed records. The LER recorded speeds are used in this report as the official train speeds, unless stated otherwise.

2.3.5

The recorded data from the LER, SBU and crossing downloads generally corroborated each other with minor differences. The LER data was used in the following analysis and discussions in this report.

3.0 Identified events of interest

3.1 Review of LER Records

3.1.1

The downloaded LER records covered 5 days since 2 July, and totalled 2191 pages of data. A compressed graphic of all of the records is shown in Figure 3. By placing tags at designated moments, the period between the train coming to a stop at Nantes and the final stop after the accident was zoomed in (Figure 4). Further zoomed graphics for the period of engine shutdown, runaway, and last backward (westward) movement are shown in Figures 5−7.

Image

Image

Image

Image

Image

3.1.2

The records of most interest were the following: speed, automatic brake, brake pipeline pressure (BP), independent brake (IND), brake cylinder pressure (BC), and the positions of the throttle\dynamic brake and the reverser handle. The time was recorded at a resolution level of 1 second, and the distance\mileage was recorded up to 0.01 miles, or 53 feet, which means a maximum round-up of 26.5 feet.

3.1.3

The recorded speed was increasing smoothly during the runaway, and there was only one abrupt turning point at the highest speed of 65 mph. After the turning point, the speed decreased nearly linearly until speed was recorded as zero. The speed records indicated that the lead locomotive suffered only one big backward tension force that changed its movement from acceleration into steep deceleration. The afterward deceleration did not show significant impulses that would appear if secondary derailments or surge/breakage occurred. The speed turning point most likely was the moment of the first derailment and the cause of the knuckle separations, and the separations of the locomotive consist in two parts from the derailed train.

3.1.4

The last separated speed loop recorded the backward (westward) movement of the front portion of the locomotive consist that was on a descending grade of 1% toward the downtown area. The speed climbed to a peak of 8 mph, and then decreased quickly until the final stop. This speed turning point corresponded to the recontact between the two portions of the locomotive consist, and the deceleration period indicated that the entire buffed locomotive consist moved backward (westward) to the final stop location. The travel distances of the acceleration and the deceleration periods of this speed loop provided the backward movement distance of the front portion alone and the re-coupled entire locomotive consist until the entire locomotive consist came to a final stop.

3.2 Events of interest

3.2.1

A number of significant events of interest were identified and are listed in Table 1. These events are described and explained in the following paragraphs. The corresponding events in the SBU records are listed in Table 2. The times of the corresponding events in the LER records and the SBU records could not be meaningfully synchronized due to the large interval of the SBU samplings and various air brake propagation delays.

3.2.2

The locomotive engineer started to apply the automatic brake in Event no.1 at 22:43:26, when the lead locomotive was located at Mile 8.01 Sherbrooke Subdivision. The speed was 19 mph, and throttle was moved down from position 4.

3.2.3

The locomotive took 5 minutes 46 seconds to slow down to a stop in Event no. 2 at 22:49:12, and the lead locomotive was located at Mile 6.72 (6.718 measured on site) with a BP reduction of 13 psi, from 95 to 82 psi. The SBU recorded the corresponding stop with a BP reduction of 12 psi, from 89 to 77 psi (the SBU time interval from the automatic brake initiation to the stop was 7 minutes 20 seconds, longer than the LER record, which was most likely due to the SBU 60‑second record interval and the propagation delay of air brake from train head to end).

3.2.4

The locomotive engineer applied IND in Event no. 3 at 22:49:37, 25 seconds after the train stopped. There was no record in SBU data corresponding to this event.

3.2.5

The automatic brake was released, and the BP increased to 94 psi in Event no. 4 after about 14 minutes 11 seconds, while the IND remained at the same level. There was no record in SBU data corresponding to this event.

3.2.6

Then 54 minutes and 54 seconds later, the lead locomotive engine was shut down after the fire was put out in Event no. 5 at 23:58:42, as indicated by the throttle movement from Idle to D and by the reverser position change from 0 to 1. No corresponding event could be identified in the SBU record.

3.2.7

BP started to drop in Event no. 6 at 00:05:55, 7 minute 13 seconds after the locomotive engine was shut down, but the IND BC remained. The SBU recorded a corresponding BP drop event.

3.2.8

IND BC started to drop at 00:13:55 in Event no. 7, 8 minutes after the BP started to drop. The BP dropped 15 psi in the 8 minutes, at a rate of about 2 psi/minutes. No corresponding event could be identified in the SBU record.

3.2.9

The train began to run away at 00:58:21 in Event no. 8, when the BP dropped to 32 psi and IND BC dropped to 27 psi. A corresponding MOVE event was recorded in the SBU data, which could be a synchronization reference but the 60‑second sampling interval prevented it from being a precise reference.

3.2.10

The train accelerated to the highest speed of 65 mph at 01:15:30 in Event no. 9, after running for 17 minutes 9 seconds and travelling 6.61 miles (34901 feet). The speed turned down abruptly into steep deceleration. The remaining BP of 16 psi dropped to zero suddenly in just 1 second. The lead locomotive was at Mile 0.11, where the detector of crossing 117.11 concurrently detected the train and activated the crossing signal. The LER time and the crossing time record were synchronized by this activation event. The SBU recorded a similar BP dropping from 16 psi to zero, 16 minutes 38 seconds after the start-to-move event. The slightly different records of the running times were caused by the SBU record interval.

3.2.11

Analysis of the speed record around this turning point and extending to the final stop showed that this moment was the only event when dramatic speed change was observed. The speed turning point indicated most likely the time of the derailment. As no further impulse event was observed afterwards but the locomotive consist was separated before the final stop, it is reasonable to conclude that the separation very likely occurred at the same time of the derailment or at an indistinguishable time difference.

3.2.12

The lead locomotive continued travelling for 1 minute 42 seconds over 5016 feet, and stopped at 01:17:12 and Mile 116.30 in Event no. 10, on an ascending grade of 1%. The residual IND BC was 6 psi. The SBU recorded a 1-minute interval from the sudden BP pressure decrease to the stop, shorter than the travelling time of the locomotives. The SBU at the end of train was decelerated quicker by the derailed cars. After the derailment moment, EOT SBU separated from the locomotive consist, and no more corresponding events could be compared.

3.2.13

The IND BC dropped to zero at 01:17:39 in Event no. 11.

3.2.14

The front portion of the locomotive consist, MMA 5017, VB-1 and MMA5026, started to move backward at 02:45:06 in Event no. 12, 1 hour 27 minutes and 27 seconds after its first stop. In the later hand brake testing, it was discovered that the hand brake on MMA 5026 was not functional (LP187/2013). The residual hand brake retarding force from MMA 5017 and VB-1 could not hold the entire portion on the 1% grade anymore, so the front portion of the locomotive consist  started to move backward.

3.2.15

The front portion of the locomotive consist accelerated to a speed of 8 mph and abruptly decelerate at 02:46:23 in Event no. 13, after travelling for 1 minute 17 seconds and 475 feet. The front portion recontacted the rear portion of the locomotive consist and pushed it, moving both backward together.

3.2.16

The recoupled entire locomotive consist continued travelling for 19 seconds over 106 feet to the final stop at 02:46:42 in Event no. 14. The lead locomotive, MMA 5017, was located at Mile 116.41 (116.412 measured on site).

3.2.17

The standard report of Crossing 117.11 recorded a speed of 58 mph approximately 2 seconds after the derailment/separation/crossing activation moment. The corresponding LER speed record at the time was 63 mph. The crossing record of the train speed was not calibrated, while the LER speed records had been calibrated perfectly with the measured wheel size and travel distance between the departure and final stop locations. Therefore, the LER speed is taken as the official reference speed in this report.

4.0 Calculation of vehicle locations

4.1 Rest position before runaway

4.1.1

Upon request from the investigation team, the locations of vehicles in the train were calculated at the times of interest after the LER mileage channel was calibrated. With the location of the lead locomotive and the length of each vehicle in the train consist, an Excel template was developed to calculate the locations of vehicles, and this data was compared with the on-site measured locations and distances of interest. A number of cases were calculated (Appendix C).

4.1.2

The lead locomotive, MMA 5017, was located at Mile 8.01, Sherbrooke Subdivision, when the locomotive engineer started to apply automatic braking to stop the train at 22:43:26. The rear end of the train was located at Mile 8.89.

4.1.3

The train came to a stop at 22:49:12 and rested until it ran away. During this period, the train covered the track section from Mile 6.72 to Mile 7.60 (measured on site at 6.718 to Mile 7.596), Sherbrooke Subdivision. The average grade under the train was calculated to be 0.918% from the on-site survey data (LP167/2013).

4.2 Derailment/Separation moment

4.2.1

The train accelerated to the highest speed and abruptly decelerated at 01:15:30 when the lead locomotive was located at Mile 0.11, Sherbrooke Subdivision. This deceleration was identified to be the derailment moment, and very likely the separation occurred simultaneously or within an indistinguishable time difference. The rear end of the train was located at Mile 0.986. The train occupied the track section with crossing, switch and curves. The front portion of the train was on flat track while the rear portion was on a descending grade of 1.2% and 1.3%.

4.2.2

The group of cars no. 6 to no. 11 at this derailment/separation moment covered the area of a crossing, switch, and curves that require a speed limit of 10 mph. These derailed cars finally came to rest scattered along a curved line tangential to the main track (Figure 1). The final position of the derailed cars indicated that the derailment very likely occurred among the group of the cars at this section of vulnerable track (Figure 8).

4.2.3

The derailment occurred suddenly at a high speed of 65 mph on the curve under a compression in-train force, causing a sudden change from compression into tension that probably broke the knuckles and separated the cars and locomotive consist. The cars in front of the first derailed car were likely swung out laterally, tangential to the curved track, landing in separated small groups in jackknifed positions. More analysis of the rollover due to over-speed and contribution of in-train force is presented in LP 188/2013.

4.3 Positions of locomotives after separation

4.3.1

After the derailment and the separation between the second and third locomotives, the LER records showed only the locations of the front portion of the locomotives up to the first stop at 01:17:12. The lead locomotive stopped at Mile 116.30, Moosehead Subdivision. The LER contained no data on the location of other separated locomotives.

4.3.2

Almost 1.5 hours later, the front portion of the locomotive consist started to move backward, and it recontacted the second part of the locomotive consist at 2:46:23. The recontact data showed that the third locomotive was stopped at Mile 116.412, 475 feet away from the first stop location of the second locomotive. The lead locomotive was at Mile 116.39 when the recontact occurred.

4.3.3

The recoupled locomotive consist moved backward (westward) together for another 106 feet and came to the final stop at 2:46:42. The lead locomotive came to rest at Mile 116.41.

5.0 Conclusions

5.1

The LER data was downloaded from the lead locomotive, MMA 5017, without calibration, and the LER records on the other locomotives not in possession of the TSB were not downloaded in time and were lost.

5.2

The LER downloaded data was calibrated with the measured wheel size and verified time reference of the detection of the train at the Crossing Mile 117.11. The adjusted LER travel distance perfectly matched the on-site survey of the distance between the departure and final stop locations of the lead locomotive. The LER speed records were calibrated accordingly and are considered very reliable.

5.3

The locomotive engineer applied the automatic brake and brought the train to a stop on the main track at Station Nantes. The automatic brake was released and the IND brake was applied on the locomotives together with hand brakes on the locomotive consist and the rail cars.

5.4

The engine of the lead locomotive was kept running until it was shut down after a fire was put out. The LER recorded that the throttle was moved from Idle to D position and the reverser record changed from 0 to 1 at 23:58:42.

5.5

The brake pipeline pressure (BP) began to drop about 7 minutes after the engine was shut down, but the IND BC pressure remained unchanged until 8 minutes later.

5.6

The BP dropped 15 psi in the first 8 minutes and continued dropping at an average rate of about 1 psi/min. The IND BC decreased at a similar rate and in a parallel delayed phase.

5.7

The train started to roll at 00:58:21 when the BP dropped to 32 psi and the IND BC dropped to 27 psi.

5.8

The train accelerated for 17 minutes 9 seconds to the highest speed of 65 mph, traveled 6.61 miles, then abruptly turned to steep deceleration at 01:15:30 with the lead locomotive at Mile 0.11, Sherbrooke Subdivision, indicating the time of the derailment. The BP dropped from 16 psi to zero in 1 second, indicating the separation of the train line.

5.9

The front part of the locomotive consist continued moving 5016 feet in 1 minute 42 seconds to the first stop at 1:17:12 at Mile 116.30 on a grade of 1%, and the IND BC decreased to 6 psi.

5.10

IND BC dropped to zero at 1:17:39.

5.11

The front part of the locomotive consist started to move backward at 2:45:06 after it had rested for 1 hour 27 minutes and 27 seconds on the 1% grade.

5.12

The front part of the locomotive consist moved backward 475 feet in 1 minute 17 seconds, accelerated to 8 mph, and recontacted the second part of the locomotive consist at 2:46:23.

5.13

The recoupled locomotive consist continued moving backward at a deceleration for 19 seconds and 106 feet before coming to the final stop on the grade at Mile 116.41, Moosehead Subdivision.

5.14

The SBU downloaded data verified that the EOT BP changed parallel with the train automatic brake application, but could not be synchronized due to the 60-second sampling interval and various propagation delays of the air brake along the train brake pipe.

5.15

The standard report at Crossing Mile 117.11 provided a verified reference of the detection of the train at the sensor location to calibrate the LER time channel, and allowed the synchronization of the downloaded LER and the crossing data.

5.16

The group of cars no. 6 to no. 11 at the derailment/separation moment covered the area of crossing, switch, and curves that require a speed limit of 10 mph. The final positions of these derailed cars indicated that the derailment very likely occurred among this group of cars at this section of vulnerable track.

5.17

The derailment occurred suddenly at a high speed of 65 mph on the curve under a compression in-train force, causing a sudden change from compression into tension that probably broke the knuckles and separated the cars and locomotive consist.

5.18

The cars in front of the first derailed car were likely swung out laterally, tangential to the curved track, landing in separated small groups in jackknifed positions.

5.19

The cars behind the first derailed car rushed into the pile-up one after another, at a decreasing rate of speed, until the last few cars in the train stopped before reaching the derailment.

Appendix A - Edited SBU download data

R13D0054 Lac-Mégantic runway and explosion EOT SBU download data

Edited portion for print since the train stopped and unattended

See original titled sheet for complete data