Derailment of Amtrak Train No. 58, City of New Orleans, Near Flora, Mississippi April 6, 2004

8

Derailment of Amtrak Train No. 58, City of New Orleans, Near Flora, Mississippi April 6, 2004

〔图〕插图

Railroad Accident Report

NTSB/RAR-05/02

PB2005-916302

Notation 7642A

References

References

Factual Information

2

Railroad Accident Report

〔图〕MemphisLITTLE ROCK★ARMSALFloraJACKSONLANew Orleane\n

Figure 1. Map showing train route and accident location.

According to the conductor, after leaving Jackson the train passed a wayside train defect detector at MP 209.2, and no defects were detected. He also recalled that a railroad employee observing the passing Amtrak train notified him that the portable marker at the rear of the train was functioning, although he did not remember the employee or the employee's location at the time of the observation.

Near MP 197, while the train passed a block signal displaying a clear indication (green aspect), the engineer said he looked at the track ahead of the train and observed something that was “not right.” (The train was traveling about 78 mph at the time.) Ahead of the train were two bridges over the Big Black River flood plain, with a filled levee between the bridges. According to the engineer, the track between the bridges looked “pretty bad.” He immediately started a service application of the brakes. The engineer described what happened next as follows:

I get up to it and I can see that the rail is kinked, and my engine goes through. I think the baggage car went through it too, and I started to feel everything moving and shaking. I knew something wasn't right. And my engine is already through it, and I'm not looking in the mirror, I'm looking ahead and I can see the right, or east rail rolling over, [so] I shot it.$^{3}$

Factual Information

3

Railroad Accident Report

About 30 minutes after departing Jackson (6:33 p.m.) the conductor felt the train start to derail, and then, he said, “... things got worse and worse.” He stated that the car he was in turned over on its side. Once the car came to rest, he became aware that he was still breathing, and then he heard a lot of moaning from the passengers.

The entire train derailed beginning on the filled levee about halfway between the two bridges over the Big Black River. The locomotive and the baggage car reached the second bridge and came to rest upright, although the rear of the baggage car hung over the right side of the bridge. The remaining eight cars derailed to the right of the track. (See figure 2 for a photograph of the derailed train and figure 3 for a diagram of the accident aftermath.) The first passenger car, a transition sleeper car,$^{4}$ separated from the baggage car and went off the fill to the right of the track before it reached the second bridge. The transition sleeper and the seven cars following it remained coupled. The transition sleeper, the dining car, the lounge, and the first coach car rolled to the right 90° and came to rest on their sides on the flood plain below the bridge. The last four cars in the train did not reach the second bridge and were on the filled area of the track between the bridges when they derailed. The first of these cars rolled on its side on the downward slope of the raised fill and came to rest at about a 110° angle from upright. The second car tipped over 90°, the third car leaned at about a 45° angle, and the last car came to rest near the tracks in an upright position.

〔图〕插图

Figure 2. Wreckage of derailed train.

Factual Information

4

Railroad Accident Report

〔图〕Locomotive口00Baggage CarConcrete Span Bridgeover Big Black Riveflood plainTransition SleeperDining Car::Lounge::"CoachCo

Figure 3. Diagram of wreckage.

The engineer told investigators that after the locomotive stopped he looked back and all he could see was dust on both sides of the train. He announced “emergency” on the radio three times. He then contacted the train dispatcher and requested help for the passengers and crew of the train. The engineer shut down the electrical system that provides high voltage electricity through the train. Then he walked back through the locomotive toward the derailed cars.

The conductor could hear the engineer talking on the radio to the CN train dispatcher. He also heard a car attendant calling 911. The conductor heard the train

〔图〕Command PostTriage AreaStaging AreaHelicopter Landing ZoneBig Black RiverPoint of derailmentRatliff Road49Livingston-Ver

Factual Information

8

Railroad Accident Report

Mississippi, at 6:35 p.m. The notification further stated that some derailed cars were on their sides, injuries were reported, and local police and rescue personnel and CN representatives were en route to the accident scene. At 6:51 p.m., the operations center notified the Amtrak police about the derailment. Between 6:51 p.m. and 7:00 p.m., the operations center dispatchers continued internal notifications.

Amtrak's Customer Services Group provided assistance to passengers after the accident. On the day of the accident, group representatives went to a shelter the Red Cross had established for passengers in the Flora accident. The group arranged for hotel accommodations, needed clothing and medicine, and continuing transportation. Counselors from Amtrak's Employee Assistance Program arrived on site the day after the accident to provide mental health services for passengers and Amtrak employees.

Injuries

A total of 58 persons (passengers and crew) were transported to six Jackson, Mississippi, area hospitals.$^{10}$ One Amtrak employee was subsequently transferred to North Oaks Medical Center in Hammond, Louisiana. Two emergency responders transported themselves to area hospitals where they were treated and released.

The derailment resulted in 1 passenger fatality, due to a traumatic impact injury, 1 passenger with serious injuries, and 34 passengers with minor injuries. Amtrak employees sustained two serious and seven minor injuries. Two emergency responders received minor injuries. (See table 1.)

Table 1. Injuries.

Factual Information

9

Railroad Accident Report

Damage

As a result of the derailment, about 273 feet of the main track were destroyed, which required the installation of 7 track panels $^{11}$ and about 1,100 feet of rail. A multispan concrete bridge spanning the Big Black River flood plain sustained damage to the concrete structure on the east side wall. The track damage was approximately $100,000.

Two cars, the transition sleeper and the dining car, were damaged beyond repair. Amtrak estimated monetary damage to the equipment as shown in table 2.

Table 2. Estimated damage to rail equipment.

Personnel Information

Amtrak Train-Operating Crew

Engineer. The engineer was hired by Amtrak on December 22, 1998. He was promoted to engineer after he completed the Amtrak engineer class on March 10, 2000. He was assigned to on-the-job training as a student engineer from March 11, 2000, to January 30, 2001, when he was certified as a train service engineer. He was recertified as a train service engineer on January 23, 2003. He qualified on the CN operating rules and territorial characteristics in 2003. During the month before the accident he had transited the accident territory about 21 times. He received Amtrak's PREPARE (Passenger Railroad Emergency Preparedness and Response Education) training on January 7, 2001.

°°

Factual Information

12

Railroad Accident Report

Table 3. Weather conditions in accident area (Jackson, MS) on selected dates.ª

ª Source: National Climatic Data Center, Asheville, North Carolina.

$^{b}$ Date rail plug was installed.

c Date track condition was inspected and reported.

d Dates crossties were installed and area of repaired rail was tamped.

Operations Information

Train movements on the CN Central Division, Yazoo Subdivision, are governed by the CN/IC Railroad U.S. Operating Rules, Second Edition, the August 10, 2003, revision. The trains were authorized by signals using a centralized traffic control system operated by a train dispatcher in Homewood, Illinois.

CN train crews used the Central Division Timetable/Special Instructions, effective August 10, 2003, for specific instructions about the territory. The timetable/special instructions indicated that the maximum speed for passenger trains in the area of the accident was 79 mph.

Signal Information

The derailment occurred on a single main track in centralized traffic control system territory that was signaled for movement in both directions, between the north end of control point Ragin, MP 198.0, and the south end of control point South Anding, MP 191.8. The traffic control system consisted of color-light wayside signals, power operated switch machines, and coded electronic track circuits controlled by the train dispatcher in Homewood, Illinois.

Factual Information

15

Railroad Accident Report

〔图〕o01\n

Figure 5. Abrasion marks on west rail.

〔图〕插图

Figure 6. Elastic metal clip anchors.

North of the 200-foot approach to the Big Black River Bridge, in the direction of the derailment site, was evidence that rail anchors had been applied to every other tie from the gage side. (See figure 7.) Investigators found 281 crossties in the area and 295 anchors placed against the sides of the ties. If the ties were box anchored on every other tie, that is, two anchors on each rail on every other tie, there would have been 281 anchors on each rail. The west rail had 144 anchors (51.25 percent), and the east rail had 151 anchors (53.74 percent). Additional anchors were lying on the ballast near by or away from the sides of the crossties.

Factual Information

16

Railroad Accident Report

〔图〕插图

Figure 7. Ties with few anchors, indicated by arrows.

Ballast. The crossties were surrounded by limestone ballast approximately 24 inches deep and 12 inches wide on the outside for shoulder support. The areas between the ties (the tie cribs) were filled with ballast. Near MP 196.6 at the north approach to the Big Black River Bridge were signs of improper drainage from fouled ballast.$^{18}$ (See figure 8.) Title 49 CFR 213.103, the FRA regulation regarding ballast, states that

all track shall be supported by material which will restrain the track laterally, longitudinally, and vertically under dynamic loads imposed by railroad rolling equipment and thermal stress exerted by the rails, provide adequate drainage for the track, and maintain proper track crosslevel, surface, and alignment.

Tie Plates. The CWR was seated in 14- by 7 3/4-inch double-shoulder tie plates.$^{19}$ The tie plates were designed with eight holes punched to accommodate two rail-holding spikes on both the field and the gage sides of the rails and two holes to secure the plate to the ties on each side of the rails. The rails were fastened to the crossties with 6-inch cut track spikes. Each tie plate had at least two spikes: a rail-holding spike on both the gage side and the field side. Additional spikes anchoring the plates to the ties were present at irregular intervals, particularly on the gage side.

Factual Information

17

Railroad Accident Report

〔图〕插图

Figure 8. Fouled ballast.

Crossties. The wooden crossties, or ties, were 8 feet 6 inches long with an average spacing of 19 1/2 inches between the centerlines of the crossties. Federal regulations, at 49 CFR 213.109, specify the condition of crossties required to provide support for the rail: they must not be “... broken through, split or otherwise impaired to the extent the crossties will allow the ballast to work through, or will not hold spikes or rail fasteners.” Each class of track requires a minimum number of ties that do not display the above defects for each 39-foot section of track. Class 4 track must have a minimum of 12 crossties in good condition for each 39 feet of rail on tangent (straight) rail and curves of equal to or less than $2^\circ$.

Several crossties near the point of derailment were broken through and split and showed signs of not holding the spikes. (See figure 9.) Neither the number nor the pattern of defective ties exceeded those allowed by the regulations. The crossties and track surface had been rehabilitated in 1998. CN had planned to do the next rehabilitation project for this area in the first quarter of 2005, but after the Flora accident, it accelerated the schedule. (See Postaccident Actions in the Other Information section of this report.)

Continuous Welded Rail

CWR is subject to mechanically induced outside forces when loaded and to thermally induced internal forces from expansion or contraction. CWR must be longitudinally restrained to prevent rail movement caused by these mechanical and thermal forces. Longitudinal restraint in CWR generally takes the form of additional rail anchors, solid crossties, and compacted ballast.

Factual Information

18

Railroad Accident Report

〔图〕Ties notholding spikes\n

Figure 9. Ties that are broken and not holding spikes.

According to the American Railway Engineering and Maintenance-of-Way Association (AREMA),$^{20}$ the magnitude of force that may be developed within CWR, independent of its length, is directly proportional to the difference in temperature between the “neutral rail temperature”$^{21}$ and the current rail temperature. On a sunny, cloudless day, rail temperature can increase up to 50° F above the ambient temperature.

Neutral rail temperature is the temperature at which there are no thermally induced forces in the rail. When the temperature of rail is different from the neutral temperature, internal forces develop within the rail. Rail anchors are an integral part of the track restraint system for CWR. Rail anchors prevent or restrict the longitudinal movement of the rail, localizing the compressive or expansive forces in the rail and allowing these localized forces to be uniformly distributed throughout the whole track structure. Without the restraint of sufficient, effective, and evenly distributed rail anchors, if temperatures are higher than the neutral temperature, the rail will expand. The expected amount of expansion or contraction of an unrestrained rail can be calculated using a standard formula that multiplies the rail length by the change in temperature and by the standard coefficient of expansion for steel rail.$^{22}$ For example, a 1-mile section of unrestrained CWR can expand as much as 18.5 inches if it experiences a 45° F change in temperature. CN's CWR program contains a CWR thermal expansion table that lists for given lengths of CWR how

Factual Information

19

Railroad Accident Report

much the length of the rail changes at various changes in temperature. (An abbreviated portion of CN's thermal expansion table is in table 4.) In the absence of proper anchoring, compressive forces within the rail may build up and overcome the lateral restraint at the weakest point in the track, which may create a track buckle. These forces may lift the rail out of the tie plates. Because “rail tends to expand in the direction of least resistance, it often expands upward sufficiently to free it from lateral restraint,” which allows the rail to move laterally.$^{23}$ To prevent such problems, rail anchors must be in good condition and properly placed. They must also retain their clamping strength and be tight against the side of the tie.

Table 4. CN CWR Thermal Expansion Table.

Railroads determine the temperature at which CWR will be under the least amount of stress throughout seasonal temperature changes, and this temperature is a railroad's preferred rail laying temperature. The intent is to lay rail at a temperature at which the rail is subject to no internal forces and that closely matches the rail temperatures found for the rail for the majority of the year. For example, CN's preferred rail laying temperature for most of Canada is 90° F, and the preferred rail laying temperature for the Central Division is 105°. In CN's "Policy on Continuous Welded Rail," dated April 2003, the preferred rail laying temperature is defined as the target temperature for CWR installation and anchoring.

Track Maintenance History

Since the beginning of 2004, several track maintenance procedures had been performed in the area near the point of derailment (MP 196.5). On January 29, welders cut out a 12-foot 11 1/2-inch portion of the east rail with a crushed rail head. When the piece of rail was cut out, the remaining rail pulled apart, and the replacement rail plug had to be 2 1/2 inches longer to fill the opening. The replacement rail plug was 13 feet 2 inches long. At that time, the rail plug was bolted rather than welded into place. The welder told investigators that the temperature was cold when the rail was repaired. The meteorological report for that day showed overcast conditions with a high of 56° F. After using a rail thermometer on the rail plug, the welder entered 60° F on the repair record, called a Non-

Factual Information

21

Railroad Accident Report

A few days later, the track foreman returned to the area to replace ties. Inserting the ties in two locations reinforced the correction of the gage problem and repaired the profile problem at the north end of the bridge. According to testimony by the track foreman, he “put the [existing] anchors back on” the area where the plug was inserted. As shown in figure 10, the new ties were not box anchored on every other tie. He maintained the speed restriction set earlier by the track inspector.

〔图〕插图

Figure 10. New ties with few anchors. Bracket indicates new ties. Arrows indicate anchors.

The track supervisor visited the area where the inspector and the foreman had performed the track work. He told investigators that he thought it was around March 16. He noted that “they had left a couple anchors out on a tie.” Other than the missing anchors, however, he said, “… everything was to my satisfaction that day ….”

CN had a table depicting the number of trains and the tonnage necessary to compact the track area that had been disturbed by the track work. Following CN railroad guidance for the number of trains and tonnage necessary to compact track after track work, train speeds were restricted to 25 mph for freight trains and 30 mph for passenger trains between MP 196.6 and MP 196.2 from March 6 to March 24. On March 24, the track supervisor determined that the necessary number of trains and tonnage had traversed the track, and he increased the train speeds to 60 mph for freight trains and 79 mph for passenger trains.

The track supervisor told investigators that he entered the data from the Non-Compliance Work Report into a computerized rail defect tracking system (RDTS). He stated that the RDTS does not have a function that highlights anything that needs immediate attention when he accesses the system. He said that he believed that a system that alerted him to conditions that needed to be addressed would be an improvement.

Factual Information

31

Railroad Accident Report

〔图〕PL\n

Figure 11. Emergency window handle that partially pulled free from rubber grommet.

Since the accident, Amtrak has informed Safety Board investigators that it will increase to four the minimum number of emergency windows tested per car during each 3-month cycle for all Superliner, Surfliner, and Viewliner passenger cars. If a failure is detected during the inspection, the same number of additional windows will be tested to ensure compliance and operability.

Amtrak has stated that it will also record in its maintenance database the location of each window tested on each car, and its Standards and Compliance Group will randomly audit the database. According to Amtrak, this new procedure will ensure that every emergency window on each passenger car will be tested at least once every 4 years. Additionally, at Amtrak's request, as of July 2004, both emergency window handle manufacturers began stamping the date of manufacture on the handle area of the window gasket assembly.

Other Information

The last train over the accident location before the arrival of train No. 58 was a southbound freight train, CN G89291-04, at 4:30 p.m. on April 6. It consisted of 2 locomotive units, 132 loaded freight cars, and 3 empty freight cars. The empty cars were the 21st, 24th, and 30th from the head end. The train weighed 17,118 tons and was 8,200 feet long. Thirty-two cars were set out at Jackson, Mississippi. After the derailment, those 32 cars were mechanically inspected at Jackson, and the remainder of the train was inspected at Collins, Mississippi. No defective cars were found during those inspections.

44

Railroad Accident Report

Recommendations

As a result of its investigation of the April 6, 2004, derailment of Amtrak's train No. 58, the City of New Orleans, near Flora, Mississippi, the National Transportation Safety Board makes the following safety recommendations:

To the Federal Railroad Administration:

Emphasize to your track inspectors the importance of enforcing a railroad's continuous welded rail program as a part of the Federal Track Safety Standards, and verify that inspectors are documenting noncompliance with the railroad's program. (R-05-05)

Establish an audit and enforcement program to verify that Amtrak complies with initial and periodic emergency preparedness training for all crewmembers systemwide, as required by 49 Code of Federal Regulations Part 239. (R-05-06)

To the Canadian National Railway Company:

Establish an audit program to verify that employees follow the current written track maintenance and inspection procedures, including rail anchoring requirements and specifically maintaining the preferred rail laying temperature. (R-05-07)

To the National Railroad Passenger Corporation (Amtrak):

Report to the Board within 90 days a schedule for training your employees who have not received emergency preparedness training as required by 49 Code of Federal Regulations Part 239. (R-05-08)

BY THE NATIONAL TRANSPORTATION SAFETY BOARD

MARK V. ROSENKER

Acting Chairman

RICHARD F. HEALING

Member

ELLEN ENGLEMAN CONNERS

Member

DEBORAH A. P. HERSMAN

Member

Adopted: July 26, 2005

Recommendations

45

Railroad Accident Report

Deborah A. P. Hersman, Member, filed the following dissenting opinion on July 28, 2005.

Notation 7642A

Member HERSMAN, dissenting:

I voted against the recommendations made in this report because I believe we have missed the mark -- the probable cause of this accident has to do with CN's lack of compliance with their own CWR program. In order to make a positive change, we must rely on the actions of the railroad and appropriate FRA oversight to prevent similar accidents from occurring in the future. What did not contribute to the cause or the severity of this accident was the currency of Amtrak crew's PREPARE training.

RECS #2 and #4 - Emergency preparedness training

I cannot reconcile the fact that 50% of the recommendations issued in the report are focused on the emergency preparedness training of the Amtrak crew whose actions were appropriate during the accident. Moreover, there is no finding in the report that such training or lack of training was causal to the accident or even a contributing factor. Nor is there any discussion in the report about any inappropriate action or inaction by the Amtrak crew during the accident sequence. Amtrak crew members notified CN dispatch, Amtrak's CNOC, and local 911 via radio or cell phones as soon as possible after the derailment. A call made to local 911 by a member of the crew provided detailed information about the train's location, the number of passengers aboard and the general conditions of the accident. The fact that local responders arrived on scene within 16 minutes of the derailment is a testament to the accurate information provided, even though the crew member called from his cell phone in the overturned dining car and the derailed train was approximately 1 1/2 miles from a paved road. I agree that Amtrak was not in compliance with the emergency preparedness provisions of 49 CFR Part 239. However, FRA has the responsibility to enforce this regulation. Our role is to identify the probable cause and make recommendations to prevent similar accidents from occurring. Neither the training of the crew nor their actions post accident were a factor in this accident; in fact, their actions were entirely appropriate. However, by devoting two of four recommendations to the issue of Amtrak's emergency preparedness training, the report implies otherwise.

RECS #1 and #3 - Continuous Welded Rail (CWR)

The two recommendations to FRA and CN regarding CWR are consistent with the probable cause, but I would have preferred to see more specificity with respect to certain areas of CWR programs. With only 119 track inspectors at the state and federal level overseeing the safety of more than 200,000 miles of track, we must rely on the railroads to develop robust CWR programs and follow through on implementing them.

While we ask FRA to enforce CWR programs and verify that inspectors are documenting noncompliance, we do not ask FRA to review all CWR programs to

Recommendations

46

Railroad Accident Report

ensure that railroads have in effect (as per the CWR regulations) procedures which specifically address maintaining a desired rail installation temperature range when making rail repairs (49 CFR 213.119). It is unacceptable that CN has not responded to FRA's request from over a year ago to clarify what it means by the “onset of warm weather” in its CWR program.

As early as 1971, the FRA recognized that installation temperature and anchoring were fundamental CWR issues. Disturbing CWR beyond the installation temperature is widely understood to affect the rail. The failure of CN to effect a more permanent repair to the CWR in the vicinity of the accident is identified in our findings as a factor in this accident. Furthermore, it was not an isolated occurrence on the CN, according to the track supervisor who stated that there were 15 other locations where 2 to 3 inches of rail had been added during the winter but had not yet been welded at the time of the accident.

I believe that there are other areas that we might have addressed to improve the safety of CWR. For example, we could ask FRA to: 1) review best practices (several Class I railroads have not had any reportable accidents caused by buckled rail in a year) and draft a model CWR program; or 2) compare submitted proposals to the AREMA manual's section on CWR (1988) which provides basic guidance. These recommendations could constitute improvements to the existing program without being overly prescriptive.

Additionally, I remain concerned about the many situations in which “temporary” repairs become long-term or permanent. We could recommended that, in light of this accident and others involving derailments due to temporary repairs of CWR, FRA review the nature and duration of “temporary” repairs to CWR and determine when delays in making such temporary repairs more permanent is acceptable. (One Class I railroad achieves permanent repairs on CWR in a much more expedited fashion because it makes temporary repairs operationally inconvenient).

Finally, in the report, there is some discussion of a CN computerized rail defect tracking system (RDTS) which appears to be underutilized or ineffective. Yet another major Class I railroad uses a rail tracking program that readily identifies those locations in CWR that may be vulnerable to buckling because more permanent repairs have not been made to rail cuts or plugs inserted and may be subjected to excessive longitudinal forces. Why wouldn't we recommend that CN improve their data collection and tracking systems?

The draft report was timely and thorough, and I commend the staff and the parties on their work. However, I believe we missed an opportunity to address the root cause of derailment through our recommendations. We shouldn't wait for another accident to develop a more complete record of weaknesses in CWR maintenance. We already possess enough information to make needed recommendations to improve the safety of CWR.