Accident Synopsis
About 4:17 p.m., central standard
time,1 on January 18, 2006, eastbound Norfolk Southern Railway (NS) freight train No.
226A117 (226), while traveling about 50 mph near Lincoln, Alabama, diverted
from the main track onto a siding track where it struck the rear of eastbound
NS train No. 22RA116 (22R), which was stopped in the siding. The collision
derailed the three locomotives and the first seven cars of train 226 and the
rear three cars of train 22R. The three crewmembers of train 226 were injured.
Property damage was estimated to be about $5.2 million.
Accident Sequence
Train
22R originated in New Orleans, Louisiana. Train 226 originated in
Los Angeles, California. Both trains were destined for Atlanta, Georgia, and
both trains were re-crewed at the NS Birmingham, Alabama, terminal at Norris
Yard, milepost (MP) 791, before continuing to Atlanta. The crew of train
226 2 (engineer, conductor, and conductor trainee) went on duty at Norris Yard at
1:15 p.m. on January 18, 2006. The train 22R crew 3 (engineer, conductor, and engineer trainee) went on duty at 1:35 p.m.
Train 22R departed Norris Yard at 2:30 p.m. 4 The train had 2 locomotives pulling 81 cars and
was trailing 6,046 tons. It was 8,276 feet long. Train 226 departed the yard at
2:50 p.m. with 3 locomotives pulling 23 cars (all of which were platform cars
loaded with intermodal containers 5) and trailing 3,582 tons. The train was 4,580
feet long. Train 226 left the yard behind train 198, which was following train
22R. About 9 miles from
Norris Yard, train 198 diverged from the east-west main line and started
southward toward Columbus, Georgia. At that point, train 226 was following train
22R on signal indication. 6
About
3:49 p.m., the NS dispatcher radioed the crew of train 22R to inform them that
they would be taking their train into the siding at Coosa (MP 758) to
allow train 226 to pass on the main line. 7 The crew of train 226 said that they overheard
this conversation and were therefore expecting to run around train 22R at
Coosa. The train 226 engineer (who had also worked as an engineer on train 22R)
said it was common for train 226 to be following train 22R. And because train
22R often had to work a local automobile plant en route while train 226 was a
“hot shot,” or priority train, 8 the dispatcher would routinely direct train 22R
into a siding to allow the through train to pass.
As
it followed train 22R, train 226 passed several clear signals 9 before encountering consecutive approach signals 10 at Holt (MP 767.8), 11 Eden (MP 765.4), and Pell City
(MP 762.8). These signals were consistent with the presence of train 22R
ahead. The approach signal at Pell City meant that
the crew had to be prepared to stop before passing the next signal, at
Riverside (MP 760.4), which was the last signal before the siding at
Coosa. (A map of these signals is shown in figure 1.)
Figure 1. Map
displaying selected signals that train 226 encountered before reaching Coosa.
About
4:03 p.m., as train 226 was passing the approach signal at Pell City, train 22R was diverting from the main track and entering
the siding at Coosa. The train 226 conductor said he had heard the train 22R
crew announce (on the radio) a diverging approach indication 12 at Coosa and was therefore aware that the train
was entering the siding and would thus be clear of the main line before the
arrival of train 226.
While
train 22R was moving from the main track into the siding, the signal just west
of the siding was displaying a stop indication to
following trains. This indication triggered the signal at Riverside, about 2
miles west of Coosa, to display a restricting indication 13 for train 226.
The
dispatcher stacked consecutive train movement requests into the centralized
traffic control equipment and lined train 22R into the siding. In this
situation, the switch would line for the main track after the rear car cleared
the siding switch track circuit. The signal system
would then light the corresponding signal aspect after the switch was lined for
the main track. In this case, however, when train 22R stopped in the
siding, about 84 feet of the rear of the train¾though
physically well clear of the main line¾remained
within the switch circuit. As a result, the switch remained lined for the
siding, and the signals continued to display stop and restricting indications.
A student engineer was at the operating
controls of train 22R when it entered the siding. The student engineer did not
activate the footage counter when the head of the train passed over the
insulated joint at the west end of the siding. 14 Train
22R was stopped at a familiar landmark near the east end of the siding. The
crew of 22R were not aware that the rear car was occupying the circuit of the
west switch.
Meanwhile,
according to interviews and event recorder data, the engineer of train 226 had
reduced the speed of his train to between 2 and 5 mph as the train came around
the 2° curve at the approach to the signal at Riverside. 15 According to signal and dispatcher data, the
Riverside signal, because of the switch alignment at Coosa, was displaying a restricting indication; that is, a single red aspect on
both the high and low signal heads. 16 (See figure 2.) According to the train 226
engineer, however, he saw the signal as “green over red.” He said he observed
that signal for “probably a minute.”
Figure 2. Eastbound
signal at Riverside as seen from a locomotive cab during postaccident sight
distance testing. The high signal head has three aspects: green over yellow
over red. The lower, offset, signal head has two aspects: green over red. This configuration should be considered an improperly or
imperfectly displayed signal.
|
| |
The train 226 engineer stated:
I looked at it [the signal
at Riverside] for the last time, and we still had a clear signal. I called it
again in the cab, still clear, and at that time I called it over the radio.
The
train 22R crew told investigators that they heard train 226 call “clear” at
Riverside. 17 The dispatcher told investigators that she did
not hear this clear call because she was tending to other duties. The conductor
and conductor trainee, who were in the 226 locomotive cab at the time, also
remembered calling the signal as clear.
Event
recorder data indicated that train 226 passed the Riverside signal at about 18
mph. The train then proceeded toward Coosa while increasing its speed, in
accordance with the clear indication the crew
believed they had seen at Riverside. As the train exited a 3° curve (restricted
to 55 mph) west of Coosa at 53 mph, the crew were able to observe the stop indication at Coosa and could see the switch
alignment. The engineer said he was preparing to announce the signal indication
for Coosa on the radio when he saw the stop signal
and realized that the switch was lined for the siding. Instead of announcing
the indication, he simply transmitted, “Oh Lord!”
The engineer placed the train’s braking
system into emergency, but the train only slowed to about 45 mph before
striking the rear of train 22R. 18 The
three crewmembers braced for the impact and were able to exit the cab after the
collision. Emergency responders examined the crewmembers and transported them
to a local hospital where the conductor trainee was treated and released. The
engineer and conductor were treated and held for observation. The crew of train
22R were not injured. 19
The
collision derailed the three locomotives of train 226 and its first seven cars.
(These were platform cars carrying a total of 34 intermodal containers and one
semi-trailer.) The three rear cars of train 22R, loaded automobile carriers,
also derailed. The collision and derailment occurred during bright daylight
under sunny, clear skies. The temperature was 53° F, and winds were from the west-southwest at 6 to 8
mph.
Sight Distance Tests
On
January 25 and 26, 2006, investigators, using a locomotive and locomotive
configuration (short hood forward) similar to those of train 226 on the day of
the accident, conducted sight distance tests of the NS main line signals at
Pell City, Riverside, and Coosa. The testing was conducted under sunny, clear
skies and at a time when the sun approximated its position at the time train
226 had passed the signals on the day of the accident. For the tests, the
siding switch at Coosa was lined and locked for the siding, and the Coosa
signal was energized to display stop. This caused
the signal at Riverside to display restricting, as
it had when train 226 passed the signal on the day of the accident.
About
1,230 feet before the Riverside signal, investigators observed two red aspects
(one on the upper signal head and one on the lower signal head), which is the
correct aspect for a restricting indication.
The
train then slowly moved forward by 40 feet, until it was within 1,189 feet of
the signal. At that point, the green aspect at the top of the upper signal head
appeared to be illuminated, even though the green aspect was not energized. As
the locomotive pulled slowly forward, the green aspect became even more
prominent. The red aspect in the same signal head remained energized and
continued to be visible to those in the locomotive cab. The upper signal head’s
top aspect continued to appear to display green as the locomotive approached
and then passed the signal. During a subsequent test run at accident train
speed, the Riverside upper signal head again appeared to display two
aspects—green over red—even though the green was not energized.
The
next day, investigators again performed sight distance tests at Riverside. The
results were the same as on the previous day, and investigators determined that
the period of time the green aspect appeared to be illuminated was about 1
hour. Although the green light was not energized, it appeared to be illuminated
because of sunlight reflecting from its lens. After the sight distance tests,
the railroad installed devices at Riverside to better shield the lights from
sun reflection.
NS Signal Recognition
Signal A in figure 3 illustrates
the signal as energized at the time of the accident, displaying a restricting indication. Note that the restricting indication is displayed as a
single red light illuminated in each of the upper and lower signal heads.
Signal B in figure 3 illustrates the signal as it appeared in postaccident
sight distance tests. Signal B is somewhat similar to signal A, but it has one
critical difference: in addition to the single red light on each of the signal
heads, a green light appears to be illuminated in the upper head. Because no
legitimate NS signal indication uses this combination of illuminated aspects,
this configuration should be considered an improperly or imperfectly displayed
signal.
NS operating rule No. 27 had instructions for
employees who encountered an imperfectly displayed signal. The rule stated, in
part:
A signal imperfectly
displayed, a signal functioning erratically, the absence of a light at a place
where a signal is usually shown, must be regarded as the most restrictive
indication that can be given by that signal and must be promptly reported to
the Dispatcher, Control Station, or Yardmaster.
Figure 3. Signal at Riverside at time of accident as
it was energized (A) and as it was observed during postaccident tests (B).
|
Neither NS operating rule
No. 27 nor any of the rule book’s signal illustrations or accompanying text
informed employees that extra lighted aspects 20 in a signal head indicate an anomaly in the observed signal, that such a signal
should be considered an imperfectly displayed signal, and that it should be
treated as if it were displaying its most restrictive indication.
The NS provides its employees
with classroom and practical training at the company’s training facility in McDonough, Georgia. Operating employees are
taught to recognize and respond appropriately to signal indications. To prepare
trainees to recognize improperly displayed signals, training staff set signals
with signal lights out or flickering.
During classroom work, because of the variety of signal types employees might
encounter on the company’s widespread territories, trainees use a computer
system that displays signals as depicted in the NS rule book. Employees are
expected to learn the various signal configurations used on their specific
territories while they are undergoing on-the-job training with regular crews.
Probable Cause
The National Transportation
Safety Board determines that the probable cause of the January 18, 2006,
collision of Norfolk Southern Railway train 226 with the rear of Norfolk
Southern Railway train 22R at Lincoln, Alabama, was the failure by the crew of
train 226 to recognize an extra lighted aspect (caused by reflected sunlight)
as an imperfectly displayed signal and to treat it as a most restrictive
indication. Contributing to the accident was Norfolk Southern
Railway’s inadequate illustrations and text in the rulebook and inadequate
training to prepare crews to recognize a signal displaying an extra lighted
aspect as an imperfectly displayed signal. Also contributing to the accident was
the lack of a positive train control system that would have intervened when the
crew did not respond appropriately to the signal.
Recommendations
As a result of its investigation
of the January 18, 2006, collision of Norfolk Southern train 226 with the rear
of Norfolk Southern train 22R at Lincoln, Alabama, the National Transportation
Safety Board made the safety recommendations listed below. (For more
information about these recommendations, see the safety recommendation letters
issued to the recipients. 21)
To the Norfolk Southern Railway Company:
Modify
your initial and recurrent training and operating rules to emphasize to your
employees and the crews of other railroads operating on your territory that any
signal that appears to display extra lighted aspects in a signal head should be
treated as an imperfectly displayed signal. (R-07-29)
To the Class I Railroads:
After reviewing the
circumstances of the January 18, 2006, railroad collision near Lincoln,
Alabama, modify, as necessary, your initial and recurrent training and
operating rules to emphasize to your employees and the crews of other railroads
operating on your territory that any signal that appears to display extra
lighted aspects in a signal head should be treated as an improperly or
imperfectly displayed signal. (R-07-30)
To the Association of American Railroads and the American Short
Line and Regional Railroad Association:
Using the circumstances of the January 18,
2006, railroad collision near Lincoln, Alabama, inform your members
through your publications, web site, and conferences of the need to enhance
their signal training to emphasize that extra lighted aspects in a signal head
should be treated as an improperly displayed signal. (R-07-31)
BY THE
NATIONAL TRANSPORTATION SAFETY BOARD
|
Mark V.
Rosenker
Chairman
Deborah
A. P. Hersman
Member
Steven
R. Chealander
Member
|
Robert
L. Sumwalt
Vice
Chairman
Kathryn
O’Leary Higgins
Member
|
Adopted: October 26, 2007
|
|
Chairman Rosenker and Member Chealander disapproved
adoption of this brief.
|
|
|
|
