Most of those who will read this post know the financial issues with New York’s Metropolitan Transportation Authority. Decades ago, funding for the agency’s capital plan was cut, and some capital funds were diverted to the operating budget as “reimbursable expenditures.” At the same time, effective fares were cut when the Metrocard was introduced, and current and – in particular former – MTA workers benefitted from a huge pension increase in 2000. The cost of capital construction contracts soared, due to pension increases for union construction workers and managers, and construction union pension underfunding, at about the same time. It was a political win for everyone – who is no longer around. The cost of all of this was borrowed or deferred, and today much of the money being paid to the MTA, in taxes, tolls and fares, is being sucked into the past.
As the years pass, meanwhile, major systems and components of the subway and commuter rail network continue to age, and eventually reach the point where they will either have to be replaced or start to fail and disrupt service more and more frequently. Perhaps to the point where entire lines have to be abandoned for years or decades, as two tracks on the Manhattan Bridge were. Or permanently, as the West Side Highway was. One of those systems is the signal system on the New York City subway, which is aging even as the cost of replacing the signaling on a line has exploded. The plan had been to gradually replace conventional railroad signaling with Communications Based Train Control (CBTC). But after decades of borrowing the MTA Capital plan came to a near halt after the Great Recession, and compared with the plans in place 17 years ago the MTA is way behind, without any hope, at recent prices, of ever being able to catch up. Can technology provide a way out?
Two subway lines, the Canarsie Line (L) and the Flushing Line (7), have already had their signals replaced with CBTC technology. The Culver Line (F in Brooklyn) was to be the third “pilot project,” and was planned to be completed in 2012. The Crosstown (G) was also supposed to be done by now. The 6thAvenue, Queens Boulevard, 8thAvenue, Fulton Street, Liberty Avenue and Rockaway lines were supposed to at least be underway and nearing completion. The signal systems on most of these lines are 84-91 years old. Back when I worked for NYCT I was told that the drop dead date was 75 years old, and for much its history the agency had been replacing its signals at an every-60-years pace.
Also expected to be underway by now are projects to replace the signals on the BMT Broadway Line and the DeKalb Interlocking, now 63 and 62 years old. And coming up behind that is most of the IRT in Manhattan, the 1,2,3 and 4,5,6, at 50 to 60 years old. These are the lines that were rebuilt with money from a bond issue in 1951 – a bond issue that New Yorkers were promised would pay for a new Second Avenue Subway.
This is an onrushing disaster that just about everyone in charge only has one solution for – postpone as much of the damage as possible until everyone who matters can retire to Florida, and leave this place in ruins. The MTA is planning to replace subway cars for which 50 years worth of bonds were issued, before they reach 50 years old, because that is easier than signal projects.
Meanwhile, the big picture is the cost of services for seniors is going to eat up every government budget in sight – federal, state and local. Over and above the cost of the debts they ran up while in charge. If there has been little actual cash, as opposed to debt, to maintain the transit system for the past 30 years, what is going to happen over the next 30 years as the costs of Generation Greed explode? To someone familiar with what new signal systems have cost, what is going on with the overall federal, state and city budgets, with all the other interests grabbing, grabbing, grabbing all they can by taking more out or putting less in, it seems hopeless.
Meanwhile, however, a bunch of private information technology companies, without ties to the railroad industry, the transit industry, or the MTA, are working to do something far more difficult than create signal control systems for railroads and rail transit. To create control systems for motor vehicles that ride on streets with lots of intersections, pedestrians, and bicycles, with traffic signals and signs in all kinds of locations, and with pavement markings in various states of fading away. The autonomous vehicle (AV) industry. Compared with what they are trying to do, creating a new signaling system for the subway is easy.
Doesn’t that raise some questions about why re-signaling the subway has to take so long and cost so much?
Key firms pursuing the AV Holy Grail include Waymo, a unit of Alphabet (Google), Tesla, NIVIDA, GM Cruise and Argo AI, backed by Ford and Volkswagon. These firms have poured $billions in self-driving vehicle technology, thus far with little or nothing in revenues. The advent of AVs on local streets now seems quite a few years away. Perhaps they would like to get some revenues for the work they have done so far.
If someone were to ask me what I would do as last throw of the dice for a situation that seems all but lost, I would try to get these organizations interested in railroad signaling, starting with the MTA. And also perhaps Jack Dorsey, founder of Twitter and Square, who is already plenty busy, but was fascinated by railroad signaling as a teenager, and once said he’d like to be Mayor of New York City.
Subway signaling is something I happen to know a little bit about. You can find out more here.
When you drive a motor vehicle down the road, you can see not only where the motor vehicle in front of you is located, but also how fast it is moving. As long as they continue to move at the same speed, the leading and following vehicles get no closer, and the following vehicle need only be far enough away to stop in time if the leading vehicle were to stop suddenly. For drivers, that is generally taught as one vehicle length behind for every 10 miles per hour of speed.
Conventional “fixed block” railroad signaling does not work like that. The general location of trains is identified by current running through one of the rails, which is disrupted if a train is in the area and its steel wheels are contacting the track. That rail, the signal rail, is divided into a number of “blocks” by gaps in the metal. All the signal system knows is that there is a train somewhere in a “block.” A following train is signaled to remain a few blocks behind, to avoid hitting the train in front. It might have to stop even if the train in front is moving further away.
This limits both allowable train speeds and track capacity. Because the very factor that makes railroads more energy efficient than rubber-tired vehicles – the limited friction between the steel wheels and steel rails – makes trains harder to stop. In general, and in particular on the underground portion of the subway where the train ahead might be around a curve, slowing or stopping based on what is seen ahead is just not good enough.
CBTC does two things.
First,it replaces the existing signals and signs with electronic measurements and instructions that allow a more accurate measurement of the location and speed of other trains. The measurements and instructions are communicated to the trains not by traffic signals and signs observed by a train operator along the track, but rather the radio. Thus, as when you drive down the highway, the following train would be able to keep moving forward at the same rate of speed as the train ahead of it. The “blocks” that keep the trains separated would move along with them.
And second, it automates the control of the trains based on those instructions. In an interim step the New York City subway never took, “cab signaling,” instead of having traffic signals and signs by the side of the track, radio signals provided instructions to a human train operator, by signals inside the train cab. CBTC not only sends instructions to the train cab by radio, but also does away with the human train operator entirely.
Even in the best-case scenario, there will be odd-school, fixed-block, traffic light-type signal systems in parts of the subway system for decades. The last fixed-block system was installed about a decade ago, and will not come due for replacement for 50 years. Given that, and given the rise of AV technology, it might make sense to separate those two aspects of CBTC, and do the second thing first.
AV technology could be used to control the trains on both fixed-block andCBTC lines. In fixed block territory, as on the street, it would operate the trains based on viewing the existing signals and signs — and perhaps scanning for people on the tracks. This would be much easier than what the AV companies are trying to do now. On the subway system there are no pedestrians, no bicycles, no cross traffic, no parallel parking, etc. etc. etc. There is no need to steer a train, because there are interlockings with switches instead of intersections. Only stopping, starting, and obeying the signs and signals, which are located in a limited number of locations relative to the front of the train. An AV system on the subway wouldn’t have to look everywhere for everything, just some places for some things.
Based on work they are already doing, I suspect, the AV companies could have the entire system wired up for automated operation in relatively short order.
With that system in place, perhaps the alternative radio-based moving block signaling could be installed more aggressively – for large parts of the system at once, instead of one line at a time in four-to-seven year projects.
Today, after a new signal system is installed, it is tested on weekends until it is found to be safe. Then there is finally a 55 hour GO (general orders) over a weekend, where the old system is turned off and the actual signals covered, and the new system put in service.
Instead, the existing signals could remain as a temporary fail-safe fallback. Basically, if CBTC told the train to go through an existing red light, the AV system would override and stop it. The accuracy of the radio-based system could be evaluated while in use, thanks to that fail-safe, and then the older signals could be gradually covered and then removed, with safety checked at each stage, to take full advantage of CBTC. The existing fixed block system could serve as a more reliable fallback than the human testers riding around in AVs right now, and trying to keep paying attention.
There is also technology now available that creates a detailed 3D image of a series of rooms, with exact measurements, I saw in a documentary about archeology. Perhaps that technology, described as expensive at the time, could be used to scan the subway tunnels and viaducts for places where new equipment could be installed without being struck by trains, allowing rapid design and easy and accurate installation.
What we have to hope for is something like what has suddenly happened to security systems, where expensive hard wired systems are being replaced by much cheaper wireless systems – albeit with a risk of hacking. This video from This Old House shows the change.
Even in a hard-wired system, however, if one of the AV companies were writing the software and providing the sensors, perhaps the same transit workers who would later maintain the system could install it.
So I’ve provided a reason why the MTA should be interested in the AV providers. But why should the AV providers be interested in the MTA? After all, a new signal system for the rest of the IND and IRT, and some of the BMT, is a one-and-done, followed by (hopefully) a return to a normal pace of ongoing replacement once Generation Greed’s debts are finally paid off (in 2060?). Meanwhile there are, what? 100 million motor vehicles on the road, perhaps to be replaced by new motor vehicles with AV every 15 years.
But there is a reason that AV firms should be interested in railroad signaling. Freight.
Today railroads haul bulk products such as coal, ores, grain and chemicals and bulky products such as motor vehicles, construction materials, and garbage. For this they compete with barges. This freight rolls across the country at 15 miles per hour.
And railroads also haul “intermodal” cargo in trailers and containers. It moves across the country at 30 miles per hour, and then sits in a rail yard somewhere for an average of 24 hours, before being transferred to a truck. For this the trains compete with long distance, cross-country trucking. Containers are usually double-stacked on long trains, running on single-track rights of way, stopping at sidings to allow trains going in the other direction to pass. Tracks were removed to save money, in part because (unlike the roads the trucks use) the railroads have to pay property taxes.
There has already been talk of AV trucks rolling across the interstate highway system, with no driver, next to your car. Electric vehicles to combat global warming, perhaps fed electricity wirelessly from a power source buried in the road, because if the trucks were powered by batteries, all that additional weight would require even more energy. This sounds very hard, very dangerous, very expensive, very long term.
Now think about a New York City subway train. Every single car has its own traction motors, and gets power directly from the third rail independently of every other car. Although new models consist of two five-car sets, old school subway cars are each capable of riding the rails independently. There is no need for a locomotive to pull the train.
Now instead of subway cars, imagine independently powered electric flat cars, with containers and trailers on them. With AV technology on CBTC freight lines, as part of a national conveyer belt system. Instead of being pulled by a locomotive, and controlled by an engineer, those flat cars – by themselves or temporarily latched into platoons – could roll across the country at 50 to 60 miles per hour, on double or triple (to also accommodate bulk freight) track lines. And be single-stacked to be taken off and put on a truck for final delivery within an hour of arrival at their destination, with transfer points perhaps every 150 to 250 miles. So every truck driver could sleep in their own bed almost every night.
This would require investment too. Tracks would have to be added. Third rail power systems would have to be installed, along with the new train control system, which would also have to be able to accommodate old school trains. And grade crossings would have to be eliminated and fencing and other security measures installed, to protect against terrorists, and (even scarier) teenagers.
But that would be much, much easier than creating an AV system safe enough for trucks to exit the Interstates and then roll down local streets with no truck driver on board. Once created, moreover, such a “national conveyer belt” system could move goods around the country using fewer manhours, and much less energy, than existing trucks and trains. It would be a win for companies that wished to claim they were helping the environment, by designing and providing it, or using it for it’s freight. And with the difference in market capitalization between tech companies and railroads as of the moment I’m writing this, each of the tech companies could probably buy a railroad of their own.
That could be the hook. Get experience with railroad train control at the MTA, and then use that knowledge across the entire country in a more electric, energy efficient future.
In the short term politics makes things impossible. The federal, state and local governments are controlled by generations and interests that see anyone and everything else as a source of money to meet their ever expanding expectations of what they need and deserve. In the long term, however, technology may provide a way out, perhaps once the aging self dealers out of the way and later-born generations try to find a way to rebuild with whatever is left.