Why Bergen Street’s New Subway Signals Cause Worse Delays Than the Old Ones

It’s the computers, man


On April 5, the F and G trains stopped running in Brooklyn because of a signal problem near the Bergen Street station. The service disruption, which diverted the F and halted G service altogether, began at 5:44 p.m. and didn’t get fixed until 7:16 p.m., ruining rush-hour commutes for thousands of New Yorkers.

This particular disruption is not from the recent spate of signal issues at Bergen Street reported on yesterday by the New York Times. It is exactly the same kind of signal issue at exactly the same place, but it occurred in 2016, before the subway was in a recognized state of crisis. In fact, Bergen Street has been a trouble spot ever since a new signaling system was installed there a decade earlier, raising questions about just how many of the MTA’s delay problems can be solved with technological upgrades.

The Bergen Street upgrades in 2006 installed a type of signaling technology new to the New York City subway that used computers to control the interlockings, the series of signals and switches that govern train movement. Before this, subway interlockings used basic relay technology — think magnets and tiny electrical currents — to manage the signals and switches. But there was newer technology out there called solid-state interlocking (SSI) that substituted computers for the old electromechanical interlockings.

In March 1999, the control room at Bergen Street for the F and G lines caught fire, destroying key equipment for the nearby interlocking. Service was restored within months using the same century-old technology as before, but the MTA decided to make the Bergen Street interlocking the pilot for the system’s first SSI, which it awarded to Alcatel, Safetran, and L.K. Comstock. Since then the authority has installed SSI at several other points, including along the Dyre Avenue line.

Switching to SSI is a little like replacing your computer’s old mechanical hard drive with an internal solid-state drive, which likewise uses semiconductors rather than electromagnetism. This makes SSI interlockings more reliable from a technical standpoint — electrical currents can be fickle, especially in wet conditions, which the subway very often is — but also a bigger problem when something goes wrong.

The older signals may be more fickle, but they’re also easier to fix — often the problem can be diagnosed with a quick visual inspection, and it doesn’t have to cause a complete halt of service. When a signal malfunctions, a device called a stop arm rises from the tracks and prevents trains from moving past the broken signal. With older electromechanical signals, train operators can bypass these stop arms by doing what’s called a key-by, crawling past the signals at extremely slow speeds and either pressing a button or using a more automatic method to lower the stop arm.

But with the Bergen Street SSI, there is no key-by mechanism because the interlocking is controlled by the computer. The stop arm goes up, and it stays up. And if the computer is malfunctioning, it can’t be diagnosed visually; it’s often the software that’s malfunctioning, which requires a programmer to troubleshoot. That’s why when there are signal problems at other points in the system, there will be delays — but at Bergen Street, service is shut down entirely.

Modernizing the subway is an important part of restoring the city’s subway service, but it must be implemented properly. As the Bergen Street interlocking demonstrates on an all-too-regular basis, implementing new technology poorly can result in even worse service than that with the old technology. The lesson from the Bergen Street interlocking is not to force modernization on unwilling administrators, but to replace them with the type of people who know what they’re doing. Like the trains themselves, we’re getting there, but much too slowly.