Derailment of Amtrak Auto Train P052-18 on the CSXT Railroad Near Crescent City, Florida April 18, 2002

8

Derailment of Amtrak Auto Train P052-18 on the CSXT Railroad Near Crescent City, Florida April 18, 2002

〔图〕插图

Railroad Accident Report

NTSB/RAR-03/02

PB2003-916302

Notation 7487A

National Transportation Safety Board Washington, D.C.

References

References

1

Railroad Accident Report

Factual Information

Accident Synopsis

About 5:08 p.m. eastern daylight time$^{1}$ on April 18, 2002, northbound National Railroad Passenger Corporation (Amtrak) train P052-18$^{2}$ the Auto Train, derailed 21 of 40 cars on CSX Transportation (CSXT) track near Crescent City, Florida. (See figure 1 for a map showing the accident location.) The train derailed in a 3-degree, 11-minute left-hand curve while traveling about 56 mph. The train was carrying 413 passengers and 33 Amtrak employees.$^{3}$ The derailment resulted in 4 fatalities, 36 serious injuries, and 106 minor injuries. The equipment and track costs associated with the accident totaled about $8.3 million. (See figures 2 through 4 showing the accident aftermath.)

〔图〕ALABAMAGEORGIAATLANTICJacksonvilleFLORIDAOCEANGulf of MexicoOrlandoTampaPalatkaCrescent CityMiamiDe Land2040060802040San

Figure 1. Map showing accident location on CSXT Sanford subdivision. (Insert shows CSXT rail line running north from Sanford. Star marks approximate accident site.)

Factual Information

2

Railroad Accident Report

〔图〕插图

Figure 2. Aerial view of accident site.

〔图〕插图

Figure 3. View of two cars after the accident.

Factual Information

3

Railroad Accident Report

〔图〕插图

Figure 4. Figure 4. Panoramic/composite image view of derailed cars. (Photo source: Putnam County Sheriff's Department. Panoramic/composite view developed by the Safety Board.)

Accident Narrative

Northbound Amtrak Auto Train P052-18 originated at Sanford, Florida, on April 18, 2002, bound for Lorton, Virginia. The train consisted of a 2-unit diesel-electric locomotive and 16 Amtrak Superliner passenger cars, followed by 24 loaded Amtrak automobile-carrying “autorack” cars.

According to the Auto Train crewmembers, the train had been assembled before they boarded. Before departure, carmen $^{4}$ successfully performed the required Federal Railroad Administration (FRA) initial terminal equipment inspection, a Class I air brake test,$^{5}$ and a two-way end-of-train (EOT) device test. The Auto Train operating crew held a job briefing before departure.

The CSXT train dispatcher granted the Auto Train permission to come out of the loading facility and occupy the main track after southbound Amtrak train 97 had cleared the Sanford Station. The Auto Train departed Sanford at 4:08 p.m., 8 minutes behind schedule. The engineer performed a mandatory running air brake test $^{6}$ after the train departed the station and noted that the brakes worked as expected. The train-operating crew said that the trip from Sanford until the accident was uneventful and that all equipment functioned normally.

Factual Information

9

Railroad Accident Report

improve emergency planning and response to future accidents and incidents. Although the debriefing minutes prepared by Amtrak do not discuss the issue of passenger accountability, some of the emergency responders involved in the on-scene command structure stated during the debriefing that the emergency responders found it difficult to account for the passengers because of the different lists Amtrak provided.

Injuries

The derailment resulted in 4 passenger fatalities, 35 serious passenger injuries, and 104 minor passenger injuries. Amtrak on-board employees sustained one serious and one minor injury. One emergency responder received a minor injury. (See table 1.)

Table 1. Injuries*

• Title 49 CFR 830.2 defines fatal injury and serious injury as follows: "Fatal injury means any injury which results in death within 30 days of the accident." "Serious injury means any injury which: (1) Requires hospitalization for more than 48 hours, commencing within 7 days from the date the injury was received; (2) results in a fracture of any bone (except simple fractures of fingers toes, or nose); (3) causes severe hemorrhages, nerve, muscle, or tendon damage; (4) involves any internal organ; or (5) involves second- or third-degree burns, or any burns affecting more than 5 percent of the body surface. The minor injury category includes all other persons, not cited in the other injury categories, who were reported treated by area hospitals within 24 hours following the incident.

Damage

The 3rd through 23rd cars derailed during the accident. The first 16 cars in the train behind the locomotive units were Amtrak Superliner passenger cars; these were followed by 24 autorack cars. After the derailment, the locomotive units and first two passenger cars remained on the rails. The next 14 passenger cars and the 7 succeeding autorack cars derailed. The 7 derailed autorack cars remained generally in-line. The remaining 17 autorack cars stayed on the rails. (See figure 5 and table 2 for the postaccident disposition of the 2 locomotive units and the first 23 cars of the train.)

Factual Information

10

Railroad Accident Report

〔图〕No.25No.24No.23M.P.722.24No.22-2No.20No.18No.19No.17No.16APPROXIMATE STARTNo.15OF DAMAGED TRACKNo.1413NoNo.12POINT OF DE

Figure 5. Postaccident disposition of 2 locomotive units and first 23 cars in accident train. (See table 2 for specific information concerning each locomotive unit and car.)

Table 2. Information on postaccident disposition of the locomotive units and first 23 cars of the Auto Train.

Factual Information

11

Railroad Accident Report

The disposition of the derailed cars, particularly the passenger cars, reflected the accordion effect of railcars when they derail, alternately folding together in parallel fashion, perpendicular to the track. (See figure 6.)

〔图〕插图

Figure 6. Aerial view of derailment showing accordion effect of derailed cars.



Factual Information

18

Railroad Accident Report

〔图〕RAILRAIL ANCHORSPIKETIE PLATETIE\n

Figure 7. Diagram showing the relationship between ties, rails, tie plates, spikes, and rail anchors. The insert is a cross section showing how the rail anchor grips the rail.

The FRA reviews each plan for compliance with various general principles, which are detailed in the regulations for numerous aspects of CWR use, including installation, adjustment, anchoring, rail temperatures (during installation, repair, and maintenance), monitoring, appropriate speeds, inspection, training for employees working on CWR, and record-keeping, among other issues. The FRA enforces the railroad's own CWR program, as detailed in the railroad's written CWR plan and standards (if any). An FRA violation concerning CWR generally reflects the failure of a railroad to comply with its own CWR standards.

Rail Anchors

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 ambient temperatures are high, the rail will expand and compressive forces may build up and overcome the lateral



Factual Information

22

Railroad Accident Report

〔图〕插图 〔图〕插图

Figure 8. Ballast conditions in accident curve.

°

° ° °

° °°

°°

°° °°

°

Factual Information

27

Railroad Accident Report

On March 6, 2002, the track inspector had the local track maintenance personnel construct a ballast-retaining wall of used concrete ties along the inside of the curve. Regional engineering managers were aware of this operation. Used concrete crossties were placed on the inside of the curve about 4 feet below the track level (the embankment was about 10 feet high at this location) to form a shallow, ballast-retaining wall. Concrete crossties were stacked 2 ties high (creating a wall about 16 inches high) end-to-end on the inside of the curve down from the toe of the shoulder ballast slope for about 240 feet along the track. The concrete ties were not secured into the embankment. (The point of derailment was near the south end of the ballast-retaining wall.) A number of CSXT maintenance and management personnel familiar with this section of track indicated that the ballast-retaining wall appeared to prevent further ballast migration down the slope. They also indicated that the ballast section on the curve was holding to standard after the wall was constructed. (See figure 9 for a representation of the relative location of the ballast-retaining wall, showing the general relationship between various roadbed and track elements.)

〔图〕SuperelevationCSX standardoutside shoulder ballast widthRAIL126"CSX standard6"CSX standardinside shoulder ballast width8

Figure 9. Conceptual drawing of cross section of the roadbed in the accident curve showing the estimated location of the ballast-retaining wall, as well as some roadbed, track, and ballast section elements.

On March 8, 2002, three-quarters of a hopper car of ballast, about 70 tons, was spread on the north end of the accident curve. (The section foreman's personal notes documented this ballast application.)

On March 11, 2002, the regional surfacing gang conducted machine spot-surfacing from MP 722.3 to MP 722.2.$^{59}$ During this operation, the surfacing gang used the ballast

Factual Information

28

Railroad Accident Report

that had been deposited on March 8 to help establish the track. The liner-tamper machine started in the body of the curve and moved northward through the spiral of the curve to create a 1-inch lift. Because the surfacing gang foreman “forgot to make out a Track Disturbance Report,” he did not record any rail temperature. (See figure 10 for a visual representation of the relative locations of the surfacing work carried out on the curve between October 9, 2001, and March 11, 2002.)

〔图〕N10/9/0110/11/0111/19/012/26/023/11/02MP 722.1.1.1.1.2.2PODMP 722.24.3.3MP 722.5\n

Figure 10. Relative locations of machine surfacing operations conducted on the accident curve between October 9, 2001, and March 11, 2002.

The Florida Department of Transportation railroad track safety inspector walked the accident curve on April 8, 2002, before conducting a Hy-Rail track inspection on April 9, which was his last inspection in the accident area before the derailment. He said that during his walking inspection, he found an area $^{60}$ where crossties needed to be replaced, and he contacted the CSXT track inspector. The CSXT track inspector placed speed restrictions of 30 mph for passenger trains and 25 mph for freight trains over the affected track. The Palatka track maintenance crew repaired the track by installing five crossties the same day at MP 722.4. After replacing the ties, the maintenance employees spot-surfaced the area, hand-tamped the ballast, and dressed it with a backhoe and by hand. The section foreman recorded the rail temperature as 92° F. During his April 8 and 9 inspections, the Florida Department of Transportation railroad track safety inspector noted no exceptions to the track ballast or rail anchor conditions, and he did not note any rail movement.

°

°

°

° °

° °

° °

° ° ° ° ° °

°° °° °°





Factual Information

33

Railroad Accident Report

Table 3. High and low temperatures recorded in the Crescent City area for the days when surfacing work took place on the accident curve (in boldface) and the 3 days following each surfacing operation. (Source: Record of River and Climatological Observations, National Climatic Data Center, Asheville, North Carolina.)

• Asterisk indicates that temperatures were recorded at Daytona Beach, Florida, rather than Crescent City, Florida.

Toxicological Information

Complete medical records for the Amtrak Auto Train engineer and assistant engineer were obtained from Amtrak and provided to the Safety Board's Medical Officer for review. No medical condition or use of prescription medication likely to cause performance degradation was found.

The four members of the train's operating crew provided blood and urine samples for toxicological testing pursuant to 49 CFR Part 219, Subpart C. Northwest Drug Testing of Salt Lake City, Utah, conducted the testing. Samples were screened for cannabinoids, cocaine, opiates, amphetamines, methamphetamine, phencyclidine, barbiturates, benzodiazepines, and ethyl alcohol. All results were negative for the presence of alcohol and cited drugs.

Factual Information

37

Railroad Accident Report

〔图〕9Amtrak\n 〔图〕6028\n

Figure 11. Postaccident Superliner sleeper cars. Top photo shows car 32503 with damaged lower level sidewall windows and largely undamaged upper level windows. Bottom photo shows car 32100 with compromised sidewall windows on both levels.

The typical service location of the folding armchair is an open space of the deluxe bedroom, facing a sofa, in an area measuring about 4 feet by 4 feet. The armchair can be repositioned to accommodate the needs of the user or moved elsewhere in the railcar. When not needed, it can be folded and stowed, although it is usually in use during daytime operations and positioned in its normal service area. The armchair is constructed of a welded stainless steel frame structure, tubular stainless steel legs, an upholstered seat-

Factual Information

38

Railroad Accident Report

bottom cushion, an upholstered seatback cushion, and folding upholstered armrests. The chair weighs about 55 pounds. In its open configuration, it measures about 25 inches wide by about 28 inches deep by about 25 1/2 inches high. When folded, it measures about 23 inches wide by about 25 inches deep by about 10 inches high. The folding armchair cannot be readily moved through the sliding main access door of the deluxe bedroom without first folding it. (See figure 12 showing the folding armchair.)

〔图〕插图

Figure 12. Exemplar of Superliner sleeper car folding armchair.

Investigators could find no regulatory requirements that specifically address provisions for folding armchairs in passenger railcars. Regulatory requirements addressing the category “interior fittings and surfaces,” which includes passenger railcar seating, are at 49 CFR 238.233. The regulations apply to passenger railcar equipment

° ° ° °

° °

°

Factual Information

41

Railroad Accident Report

Minimum neutral rail temperature requirements for buckling prevention (Rail temperature = 115° F; Class 3 line defect)

〔图〕插图

PARAMETERS:

Rail: 132 pound

Ties: wood

Tie spacing: 20 inches

Ballast: granite

Lateral resistance: 75 ppi (recently surfaced)

Tie ballast friction coefficient: 1.2

Torsional resistance: 60 in-kips/rad/in

Vertical modulus: 4000 psi

Longitudinal resistance: 200 psi

Line defect: 1.75 in. (Class 3)

Rail temperature: 115° F

Figure 13. Graph representing CWR-SAFE track buckling analysis (given a rail temperature of 115° F and Class 3 track).

° ° ° ° °

Analysis

46

Railroad Accident Report

Based on the known tie dimensions and CSXT track standards for existing track, Safety Board investigators calculated that the minimum roadbed width for the track on the accident curve, assuming it met the established CSXT standards, was 15 feet.$^{71}$ Visual evidence indicated that the roadbed in the accident curve was likely narrower than this minimum.$^{72}$

The CSXT chief regional engineer for the Jacksonville District stated that the area of the derailment had a relatively narrow roadbed. He also stated that in some cases, CSXT “roadbeds become narrow because we continue to elevate the track by surfacing our track and raising the track structure.” Because the roadbed in the accident curve had narrowed so that it could not adequately retain the normal ballast profile, ballast migrated down the inside slope of the embankment in the months before the accident, necessitating repeated resurfacing operations.

Despite the narrow roadbed, it appears that the ballast had not begun to migrate significantly until early in 2001. In fall 2000, about 18 months before the accident, CSXT had conducted shoulder ballast cleaning operations through the accident curve, in an effort to improve track drainage. The shoulder ballast cleaning removed fouled ballast from the shoulder area of the track, cleaned the ballast, and returned the cleaned ballast to the shoulder. However, according to the local CSXT track supervisors and inspector, in the months after the cleaning, ballast began migrating down the inside of the curve embankment. Evidently, by breaking up the hardpan soil supporting the ballast, the cleaning process had freed the ballast and allowed it to move down the inside of the curve embankment.

Adequate ballast is crucial to ensuring the track's lateral support and restraint. Because of the track's narrow roadbed and the consequent ballast migration, CSXT could not maintain sufficient ballast around the ties and rail at the roadbed track level at the top of the embankment to provide adequate support and restraint. In an attempt to halt ballast migration down the inside slope of the accident curve, CSXT constructed a ballast-retaining wall of used concrete ties on March 6, 2002, about 6 weeks before the accident occurred. The CSXT track inspector for this area acknowledged that before the installation of the ballast-retaining wall, the ballast levels supporting the track at the accident curve were not sufficient to meet CSXT standards. In particular, with respect to the shoulder ballast, he stated, "It wasn't maintained to [CSXT] standards with the 12 inches of ballast on the high side and 6 inches on the low, two-to-one."

°

° ° ° ° °

° ° °°

° °°

°° ° ° °°

°

°

° °°

°

°

°

°

°

°

°

°

Analysis

51

Railroad Accident Report

ballast in the track at the time of the February 26, 2002, surfacing to prevent track movement as the track stabilized.

On March 11, the accident curve was resurfaced, this time between MP 722.3 and MP 722.2, a segment containing the point of derailment. The liner-tamper machine started in the body of the curve and moved northward to create a 1-inch lift. The high and low daily temperatures in the days after this surfacing were as follows:

These ambient temperatures were somewhat higher than those in effect during the February 26 through March 1 period but were still well below those needed to achieve the desired neutral rail temperature standard of 100° F. Although less likely than the track disturbed during February 26 to fall to the 60° to 65° F neutral rail temperature range necessary for buckling, the track involved in this surfacing would still have developed significant compressive forces that would tend to further stress a buckle-prone weak point in the curve.

As has been stated, with the decline in ambient temperature that occurred after the February 26 surfacing operation (coupled with the lack of ballast on the curve from February 26 until March 8), the rail contracted and the track chorded inward. To detect track movement, CSXT requires curves over 1 degree in CWR track to be monumented if the temperature falls below 50° F during surfacing. However, CSXT procedures do not require monumenting if the temperature falls below 50° F during the stabilization period following surfacing work. (Monumenting consists of establishing known reference points against which any subsequent movement of the track can be measured.)

The temperature during the February 26 surfacing work was well above 50° F, so the track was not monumented. During the post-surfacing stabilization period, however, temperatures fell below 50° F. When the track was resurfaced on March 11, there was no way to determine whether the track had moved since the February 26 work, because the track had not been monumented on February 26. Therefore, the Safety Board concludes that because the curve was not monumented when it was surfaced on February 26, 2002, the crew resurfacing the track on March 11, 2002, did not know whether the track had moved since February 26, 2002. Had the crew members known the track had moved, they could have adjusted the curve appropriately. Therefore, the Safety Board believes that CSXT should modify its procedures to ensure that when a curve is surfaced, it is also monumented, if local temperatures are predicted to fall below 50° F during the post-surfacing stabilization period.

° ° °

°



°

66

Railroad Accident Report

Appendix B

Passenger On-Board Record Procedures from Amtrak Service Standards Manual for Management Employees

D. Passenger On-Board Record Procedures

This section covers the procedures that must be followed on long-distance, overnight, reserved trains to ensure that the Conductor's Ticket Collections Pouch provides an accurate list of everyone on-board who may not be tick-eted or appear on the manifest.

The National Transportation Safety Board (NTSB) has further recommended that designated reserved overnight trains have a passenger manifest that is maintained and updated at all staffed, en route boarding points.

This standard outlines procedures for these trains. The Conductor's Ticket Collections Pouch will provide an accurate list of all individuals on-board who may not be ticketed, or who may not appear on the manifest. To comply with the NTSB recommendation the following policy is effective immediately:

1. Requirements and Responsibilities

a. All Conductor(s) and Assistant Conductor(s) working designated trains must pick up an updated manifest/Passenger Name List before reporting to their train.

b. While the Conductor is responsible for the Train Collections Pouch, they will also brief other crew members on the location of the pouch.

c. The Train Collections Pouch must be kept in a place that provides protection from heat, water, or impact damage in the event of a train emergency.

d. In addition the Conductor will ensure Assistant Conductor(s) keep an ample supply of Form NRPC 3085 Passenger On-Board Record with them while on duty.

e. Conductor(s) and Assistant Conductor(s) must also provide all open, staffed, stations with a copy of any NRPC 3085 information received from unticketed passengers and/or crew members while en route.

f. This information is to be given to station personnel for entry into the ARROW system. Providing this information allows Conductor(s) to receive an updated Passenger Name List at the next open, staffed, station.

g. Conductor(s) can pull the passengers name list information from any ARROW terminal by entering SOLN followed by the train number (e.g., SOLN92).

2. When to Use the Form

a. The following are examples of persons who may not be ticketed or listed on the Passenger Name List manifest:

● Infants and small children

● Passengers purchasing tickets on-board

● Railroad officials

● Government officials

● Medical personnel

● Vendors and/or contractors

● Host railroad employees

b. All Amtrak employees, whether on business, deadheading (to or from work) or personal travel should have a printed ticket or must complete Form NRPC 3085. For those employees boarding at stations that are closed, the Conductor(s) or Assistant Conductor(s) must complete a Form NRPC 3085.

c. All Amtrak employees and pass riders must provide valid Amtrak photo identification.

d. In the case of an unticketed small child and/or infant, an individual Form NRPC 3085 is also to be completed.

e. All private car owners are required to have a Form NRPC 3085 completed for persons traveling providing valid photo identification.

f. Conductor(s) and Assistant Conductor(s) must complete Form NRPC 3085 for any passenger or employee who has no associated transportation to lift or does not appear on the Passenger Name List, including passengers purchasing their tickets on-board the train and passengers traveling in a group whose name does not appear on the manifest/Passenger Name List.

Appendix B

67

Railroad Accident Report

3. Documenting Travel Using the Passenger On-Board Record

a. A specimen of Form NRPC 3085 is reproduced below. Supplies of these forms are available at all crew base locations.

b. These are one-part, unnumbered forms, bound in booklets of twenty-five (25). Conductor(s) and Assistant Conductor(s) must ensure that all of the required information is provided on the forms, and must punch the form upon completion in the lower right hand corner to certify the record.

Specimen Form NRPC 3085

〔图〕Passenger Onboard RecordAmtrakConoviir giace compivied fom in Dan pouch Br each jassenger ho is not otherwioe lated onma c. The following is a list of Amtrak trains for which these procedures apply:

〔图〕Train NumberTrain Name1&2Sunset Limited3&4Southwest Chie5&6California Zephyr7&8/27&28Empire Builder11&14Coast Starlight1 〔图〕AMTRAK\n

On-Board Passenger Record

Conductor must place this completed form in train pouch for each passenger who is not listed on the manifest, group or crew list or does not have transportation lifted between city pairs shown.

NRPC 3085 (10/02) Amtrak is a registered service mark of the National Railroad Passenger Corporation.

71

Railroad Accident Report

Appendix D

Chronological List of Track Maintenance Operations on Accident Curve Following Shoulder Ballast Cleaning

Appendix D

72

Railroad Accident Report

73

Railroad Accident Report

Appendix E

Summary of Key Event Recorder Data