A ride on the subway speed limit test train

Philip Dominguez wanted to show me something.

We were squeezed into the subway operator’s cab of a test train sitting at the Kings Highway station in Brooklyn. Ahead of us was nothing but perfectly straight, unobstructed express track. To our right, across the platform, was a local Q train heading in the same direction.

Dominguez called across to the other train operator. “I’m Phil Dominguez, train service supervisor,” he introduced himself. For the past several months, he has been tasked with finding ways to make trains go faster. Fitting with that mission, he issued a challenge: Let’s race to the next stop.

“I just want to see if my train is faster,” Dominguez explained as he patted the ancient levers he uses to control his machine, “or your train is faster.”

Our ride was an old pro just shy of retirement, the venerable R42 model that dates back to the Nixon administration. On the other track, the operator was running a newer model, a gleaming R160 about a decade into its career, still well within its operating prime.

As far as drag races go, this one didn’t quite live up to the high-octane, rubber-burning affairs you might briefly glimpse on ESPN2. It took 15 seconds for both trains to ease forward and hit 20 miles per hour. But then, the development Dominguez was waiting for: Next to us, the R160 kept gaining speed, but we did not. It was like we hit an invisible speed limit. The R160 steamed ahead and reached the next station a full seven seconds before us. During that time, we barely gained any speed at all.

No surprise, right? Just another reason the MTA needs to upgrade those antiquated subway cars?

Not so fast. Like many of the other reasons for slower trains Dominguez is investigating, the ol’ R42 hits a speed wall because of a management decision made long ago, not technological limitations.

In the train cab, Dominguez pointed to a switch about halfway up the front panel. One setting was labeled “Local” and the other “Express.” Local mode, he said, is “slow and sluggish.” Express mode, on the other hand, had a top speed of 55 or 60 miles per hour, which this R42 used to hit on a long, straight track, just like any other car.

But after the 1995 Williamsburg Bridge crash—the same one that spurred the installation of thousands of speed limiting-devices across the system that Dominguez is now testing—the MTA permanently disabled “Express” mode on every train. Every older train is permanently slow and sluggish.

About 2,800 trains with deactivated “Express” switches are still in service, or about 42 percent of the overall fleet. Newer trains, such as the R160s, have software that eases the train between local and express mode automatically and achieve a top speed of a wholly respectable 55 miles per hour.

As we plodded along at about 33 miles per hour on a perfectly straight track, Dominguez impotently flipped the switch. “What’s the top speed on this train now?” I asked.

Dominguez shrugged. “You’re looking at it.”

These are exactly the types of problems Dominguez and his team, dubbed the SPEED Unit—an acronym for Subway Performance Evaluation, Education and Development—have been tasked with solving under New York City Transit’s long-overdue “Save Safe Seconds” program. For his part, Dominguez has been aware of the subway’s slow speeds for about 20 years, since he was an 18-year-old intern with the MTA. His first job was translating the signal charts for software developers making a simulator to teach operators how to drive the trains. After stints with Amtrak and the Boston subway, he came back to his native city in 2012 as a train operator. In the 10 years he was gone, he noticed the subway got a lot slower.

Since charged with saving safe seconds earlier this year, Dominguez and his team have focused on two main initiatives: fixing malfunctioning signal timers and raising speed limits. Thus far, they have identified 267 signal timers—out of the roughly 2,000 systemwide, or about 13 percent—that don’t work properly, forcing train operators to go much slower than the posted speed limit (which can be as slow as six MPH) out of an abundance of caution. Otherwise, they risk having their emergency brakes activated, which results in punishment. Each time they violate a speed restriction, the punishment gets harsher.

These punishments, and the clear incentives they give train operators to go slower and slower, is why fixing the timers is not merely about a few seconds saved here and there. After a while, operators lose faith in all the timers, regardless of whether they work or not, because there’s only one way to find out which ones are working and which ones aren’t—and they don’t intend to do so.

This is how a minority of broken timers evolves into systematically slower trains throughout the entire subway system. Dominguez’s team identifies the broken timers by doing exactly what operators of in-service trains shouldn’t do: by going the posted speed limit and seeing what happens.

However, his team can’t actually fix them. That requires sending out a separate work crew, which won’t know what the problem actually is, only that the timer isn’t working. It could be a whole host of issues ranging from a simple mechanical error to more complex calculation issues. Once they think they’ve fixed it, Dominguez and his team need to go back with a test train and run through at the posted speed once again. Because of manpower and track access limitations, the whole process for just a single timer can take months.

Even then, a fixed timer is only as good as the train operator who trusts it. This is where the “Education” in the SPEED Unit acronym comes in. Operators need to be told the timers have been fixed in a way they can remember, rather than, say, a bulletin reading signal E2-549 is now operational. To that end, Dominguez prioritizes fixing all the timers in a certain corridor—say, by way of example, DeKalb Avenue to Canal Street on the R—before putting out a bulletin to operators on that line. This way, they can easily remember the territory that has been tested and fixed.

Adjusting speed limits requires a totally different approach. Just outside of the Coney Island-Stillwell Avenue terminal on the N line, Dominguez showed a speed limit his team has eyed for raising the current limit of 20 miles per hour. The track gently and gradually eases to the left, with the outside rail ever so slightly elevated to negotiate the physics of a vehicle in movement. Dominguez said they’ve tested this portion of the track at 35 and it feels “smooth and comfortable,” but because there were workers on the track we couldn’t ride it that day.

Aaron Gordon

Once Dominguez’s team identifies a speed limit that could be higher, the track is surveyed and, using a series of mathematical formulas accounting for the track geometry, an NYCT safety committee determines whether or not to approve the increase. At a switch just south of the Prospect Park B/Q stop, Dominguez took us through one change that had been approved, this one from 15 to 25 miles per hour. The train smoothly glided through at about 24 miles per hour without a hitch.

I was curious how these speed limits were determined in the first place. Unfortunately, nobody seems to precisely know. NYCT president Andy Byford, who has led the Save Safe Seconds initiative, said they tend to be relics from when trains didn’t brake as well or from before the track design was modified to allow higher speeds. Considering the age of the train we were riding on, this means nobody has seriously undertaken a review of the system’s speed limits in decades.

Similarly, Dominguez and his team constitute the first systematic effort to test and fix signal timers since they were installed. He largely credits this to Byford’s willingness to breathe fresh air into the bureaucracy. Previous managers were overly concerned about liability at the expense of speed, whereas Byford, to Dominguez’s mind, understands “that at the end of the day our mission is to move people quickly and safely, not move people without getting sued.”

The key element of Byford’s $40 billion Fast Forward plan is to upgrade the majority of the system to a computerized signaling system called communications-based train control, or CBTC, by 2030. Dominguez’s effort simultaneously highlights the need for this new system, and its limitations as a cure-all for what ails the subway. On the one hand, CBTC trains, which already run on the L and will soon be fully operational on the 7, eliminates the need for signal timers at all. Trains move literally at the push of a button. The software knows how fast to accelerate and how hard to brake. The entire issue of human trust in trackside mechanisms is moot. On the other hand, the machines won’t be determining the speed limits; humans will. CBTC, just like any other software, will only be as good as the humans programming and maintaining it.

Furthermore, even if Byford gets his money, it will be years until the CBTC project is complete. In the meantime, Dominguez and his team can help the subway run faster in a matter of months; the MTA estimates speed limits will be increased at more than 100 locations by the spring. He views his project as wholly separate from the CBTC conversation, a chance to “purely get the most out of what we have while we have it.”

At another point on our trip, we came to a stop in the R tunnel near Prospect Avenue in Brooklyn. We had a red light, but I couldn’t understand why; the track ahead of us was perfectly straight and clear. But then, far in the distance, I could barely make out the tiny yellow circle of the “R” sign in the rear of another train. Under the current signaling system, called “fixed-block,” that train was in the next signal “block,”which prevented us from advancing even though there was at least a thousand feet of empty track ahead of us. But under CBTC, the software calculates whether trains can move based on the speed and location of every other train, allowing them to run more efficiently but just as safely.

Back in the R42 cab, I asked Dominguez if NYCT can do anything about the disabled express mode and how much it would cost. He wasn’t willing to put a dollar figure on it, but that it would be less than a new subway car. Earlier this year, the MTA ordered 535 new cars for $2.5 million per car, with the option of ordering up to 1,077 additional cars. That made me wonder: Is it even worth doing?

“I think this is worth doing yesterday,” Dominguez answered, just like maintaining the timers and raising speed limits was also worth doing yesterday. To his mind, if NYCT disabled all the switches, they can find a way to gain that speed back. It was impossible to miss how excited he was at all this, at finally being able to speed the subway up again, to solve hard problems. There are possibilities now. He knows it won’t be easy, but the solution is somewhere in there, even if it won’t happen at the flip of a switch.