Costos variables

The OCC was designed to monitor overall train operations, to tailor system operation in response to unusual traffic conditions, and to provide a recovery capability in the event of equipment failure or other unanticipated events. OCC controllers are able to see train movements on their display screens; however, this information is not, nor was it designed to be, sufficient or timely enough to allow controllers to provide warnings of imminent train collisions. OCC controllers can affect train movements at certain locations through the use of wayside signals and switches, but they must rely on the ATP subsystem to maintain separation between trains operating in revenue service.

In performing their respective functions, both the OCC and the ATP subsystem rely on accurate track occupancy information. The OCC AIM controller display is incapable of accurately tracking and displaying the instantaneous position of a train that is operating through an area with malfunctioning track circuits. The algorithm used by the ATP subsystem to compute the speed commands does not take into account information concerning the past status of track occupancies or speed commands, which makes the subsystem incapable of tracking train movements and limits its ability to recognize and appropriately respond to anomalies in track circuit occupancy detection.

The AIM display was designed to provide OCC controllers with the information they need to monitor traffic flow around the railway and to respond to events such as isolated equipment failures. The displays are not optimized, nor are they intended, for direct, individual, real-time train control. The fidelity of the information is dependent on the ability of the display to accurately update the information based on changes in wayside status. A review of the AIM controller display around the time and in the location of the Fort Totten accident indicated that, in areas where the wayside signaling system was functioning normally, the display consistently

and accurately depicted changes in wayside occupancy data within about 1 second. Conversely, in areas where the signaling system was functioning abnormally, the AIM display failed to accurately depict the presence and location of trains.

WMATA had implemented OCC AIM computer algorithms to identify malfunctioning track circuits and alert the controller via alarms under certain conditions. The NTSB considered whether OCC controllers could have prevented this accident if they had acted on the multiple ARB and NRB alarms that occurred in the 5 days before the accident near Fort Totten. However, it is not clear what actions OCC controllers were required to take in response to these alarms, and the programmed behavior of the AIM software—returning a track circuit graphic display status icon to “normal” without requiring controller acknowledgment—masked the presence and severity of potential track circuit failures. Further, the extremely high incidence of track-circuit alarms (that is, about 5,000 ARB and 3,000 NRB alarms per week) would have thoroughly desensitized OCC controllers to the track circuit malfunctions occurring across the Metrorail system. Therefore, the NTSB concludes that, because of the design of the WMATA OCC information management system and the high number of track circuit failure alarms routinely generated by that system, OCC controllers could not have been expected to be aware of the impending collision or to warn either train operator.

Urgent Safety Recommendation R-09-6, issued by the NTSB to WMATA on July 13, 2009, recommended that WMATA evaluate “track occupancy data on a real-time basis in order to detect losses in track occupancy and automatically generate alerts.” WMATA has reported that it is working with the provider of the current AIM software to develop a system designed to provide real-time alerts for loss-of-shunt events. The first step, according to WMATA, is to develop an improved loss-of-shunt detection algorithm that will reduce the false-alarm rate for reported track circuit failures. A reduced rate of suspect alarms, coupled with greater engineering attention to reported loss-of-shunt events, should allow OCC controllers to better monitor the operating conditions across the system and to respond more effectively to anomalies. The NTSB supports these efforts and will continue to monitor WMATA’s progress toward meeting its goals.

Emergency Response

The operator of train 214 reported the collision to the OCC, giving the location as chain marker 311+00. The collision caused third-rail power to go down immediately on track B2 (the accident track); about 10 minutes after the report of the accident, the OCC controller deenergized third-rail power on track B1.

The first transit police officer arrived on the scene about 5 minutes after the collision. The first medic arrived within 10 minutes of the collision. The assistant chief of operations for District of Columbia Fire and Emergency Medical Services acted as the incident commander and established a unified command system with the responding agencies and the railroads. Mutual aid resources were requested and received from surrounding areas in Maryland and Virginia. During the course of the response, the incident commander established an evacuation group, a rescue and extrication group, and a medical group. The battalion chief in charge of the medical

group estimated that all patients were treated and transported within about 90 minutes after the accident.

As a result of the collision, the rear car of train 214 telescoped into the first car of train 112, making the recovery of occupants from the lead car of the striking train a difficult and dangerous extrication operation that required extensive manpower, resources, and time. The NTSB concludes that considering the challenges of the recovery operations, the emergency response was well coordinated and effectively managed.