Meet the Century-Old Technology That Is Causing Your Subway Delays

London and Paris have solved their cities’ subway signal problems. Why can’t the MTA?


If you check the MTA website before heading home from work, it probably hasn’t been too long since your line last had delays due to “ongoing signal problems.” The only thing less surprising than the fact that your commute will take twice as long is why the commute will take twice as long: those pesky signals.

For all intents and purposes, the subway uses the same signaling system that was installed when the first subway line opened in 1904. Most of the actual signals and related equipment have been replaced in the last forty to fifty years, so it’s not like the physical switches are older than the Model T. But the way the system operates hasn’t appreciably changed since its inception.

Meanwhile, over the last few decades, even older systems like the London Underground and Paris Metro have made significant strides in modernizing their signaling systems at a fraction of the projected cost to do the same for the New York subway. According to the New York Times, it will take the MTA some fifty years and $20 billion to install a modern signal system. Both figures are roughly double what London and Paris project. How we got here, and why the MTA and its customers are both literally and figuratively stuck between stations, is not a simple question to answer.

New York’s subway signals operate in what’s called a fixed-block system: Each signal indicates the status of one segment of track, generally around one thousand feet long. If a train is anywhere in that stretch — even if it’s just the last wheel of the final car — the block is marked as occupied and any trains in the block behind must halt.

Although this signaling system works — through all of this crisis talk, the subway still manages to get roughly 5.6 million people around every weekday, with varying degrees of efficiency — it’s slow, expensive, and inefficient.

The most obvious problem with this signaling system is that, when something goes wrong, it goes very wrong. Unlike in modern signaling systems, if one signal stops functioning, that portion of the track shuts down until a crew physically goes to the signal and fixes it. Depending on where this broken signal is, it can either result in a minor inconvenience, as trains are routed around it, or a major bottleneck that cripples a commute.

If the MTA discovers, upon diagnosing the issue, that a part needs to be replaced, it will cost the agency. Because the signals still rely on ancient, prewar technologies, many of the parts are no longer manufactured; the new part either has to be made specifically for the MTA or pieced together from parts of older units, which is both expensive and time-consuming. Because a single broken or malfunctioning signal can cause such problems, the MTA must conduct proactive maintenance to keep them in working order. According to a 2014 report by the Regional Plan Association, the MTA spends $106 million annually on signal maintenance, inspection, and repairs. That comes to $168,000 per track mile ($161,000 of which is spent on labor).

In addition, fixed-block signals don’t convey very much information. For train operators, they work like regular traffic lights, only worse: If the block ahead is occupied, operators get a red light on the trackside indicator. A yellow light signals to proceed with caution at a slower speed because a train was just in that block. A green light is an all clear ahead.

What the train operator doesn’t know based on this simple system is the location within that block of the train ahead, or how fast it’s moving, or if it’s moving at all. Any car driver knows how critical that information is — it helps you judge how fast you ought to be going — and the same is true for trains. Without that information, train operators must exercise an abundance of caution and maintain more distance and slower speeds than would otherwise be necessary. This, in turn, results in trains getting backed up, bunched, and unevenly spaced. A traffic jam might lead to three or four trains arriving within minutes, only to leave a large gap behind them.

This bunching may not sound like a serious problem, but it is a major contributor to the “overcrowding” issues the MTA cites as the biggest source of delays. Think of what happens when three trains arrive one after another: The front train becomes miserably packed with agitated commuters who waited too long and don’t want to risk another long wait for another train. Because that train and the platform is stuffed with people, loading and unloading at every stop takes longer than scheduled. The next train, which will be much less crowded, will pick up the people who didn’t fit in the first train; the next train will then remain under capacity despite running at rush hour because it’s right behind another train. After the clump passes, the station stuffs full with waiting commuters again, another bunch of trains approaches, and the entire process repeats.

There must be a better way, you’re probably saying to yourself now. Well, there is, says Rich Barone, vice president for transportation at the Regional Plan Association and the main author of the organization’s report on the MTA’s signals.

A decade ago, Barone wrote his graduate thesis on the MTA and a fairly new signaling system called Communications-Based Train Control, or CBTC. He later expanded that thesis into the 2014 RPA report, which argues that many of the subway’s issues can be mitigated with a heavy investment in CBTC and complementary structural improvements.

CBTC is the 21st century’s answer to the 19th century’s fixed-block system. It uses transponders, RFID chips, and other pieces of equipment invented during the digital revolution to determine the precise location and speed of every train. That information is relayed to a control center that utilizes computer models to determine how the trains run.

Why is CBTC better? For one, it has built-in redundancy, which means no more proactive maintenance costs: If one part fails, another can pick up the slack until the damaged equipment is replaced. The things that make CBTC work — the transponders, RFID chips, and communication lines — are far, far less likely to fail, and when they do, the system tells the control center precisely what’s wrong and where, as opposed to the current system, in which all the MTA knows is that one particular signal isn’t functioning for whatever reason. A broken piece of equipment will no longer cripple an entire line.

CBTC also includes all kinds of safety improvements. It’s far less reliant on drivers and can even override their errors. (Indeed, many CBTC systems in other cities operate driverless trains.) Even if the MTA were still to employ drivers to meet union contracts, it could save the MTA millions in energy costs, by Barone’s estimate, by letting a computer accelerate and brake as opposed to a human.

Most important, CBTC alleviates the train-bunching issue by automatically adjusting train speeds to maximize line capacity (getting more trains on the tracks) and car capacity (spacing trains properly to keep them from getting overloaded and thus delayed). Under ideal conditions, Barone’s report concluded, CBTC can run 40 trains per hour per line, or a train every 90 seconds. Due to physical limitations in the system, the best the subway can hope to see with a CBTC system is probably around a train every 120 seconds, or 30 trains per hour; still, that’s an improvement over today’s subway, which typically runs only 20 to 25 trains per hour.

Which brings us to the $20 billion question: Why hasn’t CBTC been widely implemented in New York?

There’s no single simple answer. CBTC has been installed and implemented on the L line, but it took a decade to complete the project, and it ran over budget. The exact cost, Barone says, is hard to pin down because the MTA included the installation and implementation of other systems in the cost — more on that later — even though the L is one of the only truly independent lines in the system. The 7 line is currently being fitted with CBTC, a project that was slated to be completed this year. (Spoiler: It won’t be.) At the MTA’s current pace, we can expect a fully CBTC-implemented New York subway sometime around the turn of the next century.

During roughly that same time frame and despite two canceled contracts with Bombardier and Metronet, the London Underground has implemented CBTC on four of its ten lines, with work underway on four additional lines. The Paris Metro has installed CBTC on a number of its lines, one of which is now fully driverless. Both of those systems are older than the New York subway.

Barone concedes that implementing CBTC on the New York subway would be more complicated than on those two systems. Our lines overlap and share tracks, tunnels, and critical junctures more frequently than do London’s or Paris’s, meaning work on, say, the Eighth Avenue corridor would affect several lines at once. And the London and Paris systems shut down at night, which makes it easier and quicker to do the work.

In addition, CBTC itself is not a cure-all. The MTA would also need to redo tunnels, tracks, and stations to permit a greater flow of trains and people. If trains still need to slow down to make a curve, or the terminus at a certain line can only accommodate so many trains, or a station can’t deal with the more rapid passenger flow, the MTA won’t be getting as much out of CBTC.

But these aren’t deal breakers, and Barone doesn’t buy that as an excuse for the glacial pace of CBTC implementation. “The reality is, nothing’s too complicated that we couldn’t have done this already,” he says.

Instead, Barone says the CBTC issue gets at something more fundamental: the MTA itself. Although he says the MTA has “made progress” with projects like the Second Avenue and 7 line extensions, “it’s far, far behind when it comes to modernizing and upgrading its system.” The MTA has spent more than $100 billion since the 1980s attempting to get the system in what it calls a “state of good repair.” But, according to the MTA’s twenty-year capital needs assessment issued in 2013, notes MTA board member Veronica Vanterpool, about a quarter of all stations and signals fall short of that distinction.

Instead of going all in on CBTC, the MTA decided in the 1990s to choose a half-measure by installing two partially redundant systems that each do a portion of what CBTC does but combined fall far short of CBTC’s benefits. One, called the Auxiliary Wayside System (AWS), is a redundant signaling system that allows non-CBTC-equipped vehicles to run on a CBTC-equipped track. The MTA bills this as a necessary fail-safe, even though many systems around the world run solely on CBTC without issue thanks to CBTC’s built-in redundancies. What AWS does do is allow the MTA to take trains from other lines that aren’t equipped with CBTC and run them on the L line — which is necessary because the MTA didn’t order enough CBTC-equipped cars. However, not only does this defeat the purpose of installing CBTC in the first place, but, according to a Federal Transit Administration study, installing AWS increases the cost of CBTC implementation by 30 percent.

The other system, Automatic Train Supervision (ATS), is a passive monitoring system that, like CBTC, provides more precise train location to a centralized control room. But unlike CBTC, it cannot control the trains. The main benefits of ATS are twofold: rerouting trains in case of issues, and providing information to run countdown clocks — both of which could be covered by CBTC, at significantly lower maintenance and upkeep costs.

Installing two redundant but inferior systems to CBTC has left the MTA having to maintain not one outdated system, but three. “The idea of eliminating is very hard for them,” says Barone, recalling that an MTA official he worked on the RPA report with didn’t seem to care for CBTC or understand the benefits it could provide New Yorkers. But, Barone says, “the idea of adding is natural.”

With our subways’ 24-hour service, fully upgrading the system would either take half a century or be a massive inconvenience. Until now, the MTA chose the path of least resistance, as bureaucracies tend to do, and opted for the former. But now that the subway is a massive inconvenience anyway, the door is open to speed things up. Barone advocates for a widespread adoption of the FastTrack program — the MTA’s subway maintenance and repair program that doesn’t shut down lines completely, but significantly alters service during off-peak hours — to include CBTC upgrades. Even then, an expedited plan might take twenty years. Let’s just hope we’re not all underwater by then.