KOMANOFF: Investing Congestion Revenues in Better Train Signals Could Save New Yorkers Lots of Time

A timer signal on the Williamsburg Bridge.
A timer signal on the Williamsburg Bridge.

Gov. Cuomo and legislative leaders are looking to congestion pricing to generate $1 billion a year to support $15 billion in capital spending on transit, mostly on subways. What if that revenue stream went to install Communications-Based Train Control on as many lines as possible?

I’ve taken a shot at quantifying the benefits, and they appear huge. By my estimates, $15 billion could pay for NYC Transit to install CBTC on around a dozen, or nearly half, of the city’s subway lines, cutting passenger travel time on those lines by 20 percent. Averaged across the entire system — a calculation that necessarily dilutes the per-trip savings — a typical trip on the subway would go 10-11 percent faster. That speed-up will save riders a million and a half hours a week, or nearly 80 million hours a year.

Not just straphangers but drivers too stand to save considerable time with the investments in CBTC, as the switch of some trips to more-reliable transit thins out traffic on city streets, highways and bridges.

Before I detail my calculations, it’s worth asking how plausible is a scenario investing $15 billion in CBTC. Needless to say, it’s unlikely that CBTC will grab every dollar of congestion revenues, notwithstanding that signal issues account for half of subway delays, according to the Regional Plan Association, the transit research and advocacy group whose insistent and expert research helped make CBTC “a thing” in recent years:

RPA collected delay data published by the MTA daily from September 2016 to June 2017. Signal-related delays accounted for over half of the delays (54 percent) during this period. Second was the MTA’s generic category “incidents” — a catch-all that also likely includes more signal-related issues. Other mechanical systems accounted for a much lower share of overall delays — tracks (16 percent), switches (4 percent), rolling stock (3 percent) and power (less than 1 percent).

(Save Our Subways: A Plan to Transform New York City’s Rapid Transit System, June 2018, p. 23.)

But it’s possible that the budget bill that will pass in Albany by March 31 could include more than the default billion a year in new monies for the MTA, especially with new taxes on weed and pricey pieds-a-terre on the agenda. And if congestion tolls do pass, the legislature could further pad the revenue by retooling the for-hire vehicle surcharges that went into effect last month.

Moreover, replacing the subways’ ancient “fixed-block” analog signals with digital CBTC is widely seen — as in this MTA video and the Regional Plan Association’s ”Moving Forward” report — as the city’s most-burning transportation infrastructure priority. So even as a hypothetical, it’s worth seeing what New Yorkers could get from a decision to dedicate a billion bucks a year to CBTC. NYC Transit president Andy Byford’s 2018 “Fast Forward” pledge to compress from four decades to just 10 years the time to equip all 26 subway lines with CBTC remains a linchpin of hopes for 21st-century transit in New York.

Nevertheless, this being NYC and the MTA being the MTA, 100 percent CBTC has never gotten an official price tag; nor have the potential travel time savings from CBTC been estimated. Recently, with funding from an anonymous transit advocate, I set about filling the data gaps. Here are my findings:

  • Based on experience installing CBTC on the Canarsie Line (L train) and Flushing Line (7 train), equipping a mile of track with CBTC costs $45 to $50 million (in constant 2019 dollars).
  • Excluding lines shorter than 10 miles, which account for a very small fraction of total track-miles, the average subway line is 26 miles, a length that will require nearly $1.25 billion each to install CBTC.
  • $15 billion invested in CBTC — roughly the capital that can be secured with a new billion dollar a year revenue stream — could pay for installing CBTC on a dozen lines totaling 315 track-miles, or roughly 45 percent of the system’s 696 track-miles.
  • CBTC can cut 20 percent from the duration of a typical subway trip; I calculated that figure taking into account lower headways and better spacing between trains, fewer signal-caused delays, enhanced flexibility to mitigate issues as they occur, higher rolling speeds and shorter station dwell times from reduced crowding.
  • Prorating those savings over the entire system, the $15-billion investment in CBTC enabled by the $1 billion a year revenue stream will shorten the duration of an average subway trip by 10 percent.
  • Traveling by subway consumes an estimated 833 million hours a year of New Yorkers’ time. Dedicating $1 billion a year to install CBTC can shorten that time by 80 million hours.
  • Per day, the time saved is 216,000 hours, a daily savings worth $3.5 to $4 million to riders, based on a wage-based average value of time of $17.23.

To be clear, these hours of time savings are for the dozen lines that can be converted to CBTC with $15 billion. Making CBTC system-wide will slightly more than double them, but of course will cost well more than the initial $15 billion in capital that congestion pricing is intended to make available for transit.

If you’re familiar with my quantitative work on driving, transit and pricing, you know that I obsessively break travel costs and data into their component parts which I combine in my ever-evolving BTA spreadsheet (8.5 MB Excel file, downloadable here; see especially the new CBTC tab). My most important methods and assumptions follow:

CBTC costs to date

  • According to MTA records, the authority opened CBTC-install budget lines in 1996 for the Canarsie line and in 2004 for the Flushing line. Adjusting each year’s expenditures to constant dollars — a must for extrapolating from empirical data — the installations cost $56.4 million a mile (Canarsie) and $38.7 million a mile (Flushing) in 2019 dollars.
  • The marked drop in CBTC costs from the Canarsie install to Flushing cries out for explanation. While it’s tempting to discount the Canarsie data on the premise that it was bloated by first-time costs, it’s also possible that the financials for Flushing are incomplete. Lacking more info, I conservatively averaged the costs to project that future CBTC installations will cost $47.6 million a mile (in 2019 dollars).

Smoother running with CBTC

  • CBTC builds an interconnected network of trains and signals in which real-time tracking of trains is possible and accessible for riders. This central information network will enable NYC Transit to react in real time to both breakdowns and increased demand by rerouting trains. (With antiquated fixed-block signaling, a train’s final location is set beforehand and all tracks are unidirectional.) Individual issues can be localized without snowballing.
  • I assume that on lines with CBTC, during the 7 a.m.-7 p.m. period accounting for three-quarters of all subway trips, station “dwell times” will drop by ten seconds, rolling speeds will rise 25 percent, and trains per hour will increase to an average of 28, from 23 at present.
  • These assumptions translate mathematically to 20.4 percent quicker trips on the lines that will get CBTC (paid for by the $15 billion in capital backed by the $1 billion a year revenue stream). That equates to a 9.6-percent system-average shortening in trips and, in turn, a 10- to 11-percent increase in subway trip speeds (recall that velocity = distance/time, so that percentage shortenings of time give rise to slightly higher percentage gains in speed).

Other assumptions along with my calculations and sources are detailed in the BTA spreadsheet. One result not shown there — it takes a bit of tweaking with the model — is that drivers will save 40 million hours a year from lighter traffic as improved subway service reduces use of for-hire vehicles and private cars.

But the bigger picture is that CBTC will enable NYC Transit and the MTA to begin to unify the subway under a single network of connections allowing better communication and efficiency. Even for me, the ultimate benefits of that are hard to quantify.

Transit advocate and enthusiast Michael Kendall greatly assisted in the research underlying this post.

  • Larry Littlefield

    If it costs the same as the Canarsie Line we can’t afford it. And if the revenues are bonded again those likes not included will probably have to be shut down eventually, because there will be no money to replace THEIR signals.

    The real comparison isn’t between fixed block and moving block signals. It is between systems that are more than 75 years old and subject to deferred maintenance every time attention is directed away from where transit dollars are going, and new system. The old system are going to fail.

  • sbauman

    $15 billion invested in CBTC — roughly the capital that can be secured with a new billion dollar a year revenue stream — could pay for installing CBTC on a dozen lines totaling 315 track-miles, or roughly 45 percent of the system’s 696 track-miles.

    Here is one of many errors in this article.

    CBTC requires modifications to both the tracks and the rolling stock. If an alien (non-CBTC modified) train appears on CBTC track, the service levels decrease dramatically to roughly 3 tph. The NYC subway lines are interconnected. This provides operational flexibility to divert trains, when blockages occur. Re-routes are legendary. An F may substitute for an A in Brooklyn, while the D and G may cover for the diverted F and some N’s may cover for the diverted D, etc. Should the diverted tracks be CBTC and one of the diverted trains be an alien, then there will be additional blockages due to the diversion.

    The workaround is to make sure that alien trainsets are replaced with those that are CBTC equipped before individual the track conversions become operational. Only the NTT trainsets can be easily converted to CBTC; the rest of the fleet must be replaced.

    Their replacement is part of Mr. Byford’s Fast Forward plan. The equivalent of 2800 pre-NTT rail cars need to be replaced before CBTC can be implemented. This comes to approximately $7B @ $2.5M per rail car. This will bring the number of CBTC compatible trainsets to approximately 600. Each of these trainsets must be become CBTC equipped for an additional $600M @ $1M per trainset.

    Only $7.4B from the $15B will be actually available for implementing CBTC on the tracks. This will mean that the $15B installment will be able to implement CBTC on only half of the projected 315 miles or 45% of the tracks .

    It’s the initial cost of equipping existing rolling stock plus NYC’s interconnected track architecture that makes me wonder whether CBTC is the appropriate technological approach for NYC. The other cited retrofits have been along isolated lines like those in Toronto, London and Paris, as well as the L and 7 lines in NYC.

  • Larry Littlefield

    Most of the recent subway car purchases have been designed to allow CBTC to be added later.

    “Each of these trainsets must be become CBTC equipped for an additional $600M @ $1M per trainset.”

    So how much more will each motor vehicle with Google or GM’s self-driving system cost? $2 million extra?

    The fact that self-driving vehicles are being pursued has reduced my tolerance for high CBTC cost. The Canarsie, Flushing and Culver line projects were supposed to be pilot projects, with extensive development costs built it. After that, CBTC was supposed to be cheap.

  • sbauman

    Most of the recent subway car purchases have been designed to allow CBTC to be added later…After that, CBTC was supposed to be cheap. Meanwhile, technology has moved along to the point where it should be even cheaper.

    One reason why the conversion from CBTC compatible to CBTC equipped won’t come down has to do with the NTT design. The MTA car designers chose to make the NTT train’s fly-by-wire signals mimic the train operator’s commands rather than what the operator commands were supposed to accomplish. This means the the conversion for each NTT car series must include the translation from what the signal system wants to operator commands. Differences between car models means that this must be developed each conversion.

  • AMH

    “trains per hour will increase to an average of 28”

    This assumes that NYCT is willing to operate that much service! TPH is usually kept at a bare minimum to ensure adequate crowding.

  • Joe R.

    That’s a product of the increase in rolling speeds. If the train can get across the line faster, you can increase the number of trains per hour without needing more trainsets or TOs.

  • Joe R.

    Without bonding, you have enough money to replace 100% of the signals with CBTC in about 33 years. Probably less actually as once you standardize on a system the costs per mile should drop drastically. Bonding is stupid. You don’t need $15 billion in one shot because you’re not replacing everything in one shot. You get $1 billion per year. If it’s dedicated solely to CBTC you can replace signals on at least 21 miles of track per year–forever. Once CBTC is installed systemwide you can start using the congestion pricing revenues for something else. This is much better than getting a one shot cash infusion of $15 billion, and then never seeing another dime. Who exactly runs the finances here because these are people I would avoid using for my own personal finances? They’re doing the stupidest thing in the book you can do financially, which is spending a new revenue source all at once. While we’re at it, why not start doing the same with people’s salaries? Give them 15 years of pay in one shot, then have them pay back the loan over the next 50 years of their working life, plus continue paying it during their retirement. Or let’s starting bonding against income and sales taxes. Spend the next 15 years of tax receipts now, then use income tax receipts after that forever solely to pay off the bond holders.

    Bonding is monumentally stupid, period. Those in charge have to know this.

  • Joe R.

    This is the case for NOT bonding. Once those monies are spent, you’ll never get another extra dime. Whatever can’t be fixed with the initial $15 billion, or whatever the number is if we bond against other new taxes, won’t be fixed. We might as well just stop running about half the subway now because that’s what will happen eventually.

  • Larry Littlefield

    If they enacted congestion pricing when it was first proposed, all that money would already be gone for the next 25 years.

    As it is, whey can’t they fund the MTA with all that payroll tax money? Because it’s gone.

  • Larry Littlefield

    Greedy, selfish, and dishonest, not stupid.

    They’ve already sold the future down the river. Their only goal is to paddling furiously so they are all of the boat before it goes over the falls.

    I refuse to give New York’s special interests and the politicians they control, and the richest generations in U.S. history, the easy out of having made “mistakes.” That’s part of the propaganda, in places where the cover up is breaking down.

  • Joe R.
  • sbauman

    This wasn’t a good day for extolling the benefits of Thale’s CBTC system.

    https://www.scmp.com/news/hong-kong/article/3002089/hong-kong-mtr-trains-collide-near-central-station

    This collision occurred on a switch, which the CBTC paradigm does not handle. It must make use of conventional track circuits to locate trains and prevent collisions. This and the previously mentioned Singapore collision indicate that the Thales system has difficulty processing track circuit and CBTC messages simultaneously.

  • sbauman

    CBTC can cut 20 percent from the duration of a typical subway trip; I calculated that figure taking into account lower headways and better spacing between trains, fewer signal-caused delays, enhanced flexibility to mitigate issues as they occur, higher rolling speeds and shorter station dwell times from reduced crowding.

    There’s a big different between “can” and “does”. Here’s the schedule information regarding the 14th St Express that ran between 1936 and 1956.

    https://farm8.staticflickr.com/7880/46634725124_a66704d0a5_b.jpg

    It ran between Lefferts Blv and 8th Aves along a combination of the Fulton St El and the 14th St Line. Trains left Lefferts Blv at 10 minute intervals between 7:33 and 8:03 and took 43 minutes. Today, there only 4 A-trains departed Lefferts Blv during this time interval. They arrived at the 14th St station 46 and 47 minutes after their departure.

    The 1936 schedule permits running time comparison with today’s performance on sections of the A and L lines over the same track.

    The section between Lefferts and Hudson (80th St) was scheduled to take 5 minutes in 1936. Today’s trains making the same stop over the same tracks are scheduled to take 8 minutes.

    The section between Atlantic Ave Myrtle Aves was scheduled to take 7 minutes in 1936. Today’s L trains operating making the same stops over the same tracks and aided by CBTC are scheduled to take 8 minutes.

    A comparison between Myrtle Ave and Lorimer St cannot be made because the 1936 14th St Express skipped the intermediate stops.

    The section between Lorimer St and 8th Ave was schedule to take 10 minutes in 1936. Today’s L trains operating making the same stops at the same time over the same tracks and aided by CBTC are scheduled to take 11 minutes.

  • AMH

    Sure, but I wouldn’t put it past them to cut service.

  • Jensen Schram

    Can anyone shed some light on the way the 20.4% decrease in trip time was calculated? I’m researching the system for a project, and being able to give the calculations behind that figure would be really helpful

ALSO ON STREETSBLOG

Off-Peak Discounts for NYC Transit: An Intriguing Idea

|
Photo: cunningsue/Flickr Discounting off-peak transit service could be a boon to New York City’s transportation and quality of life, so long as revenues can be found to make up for the likely farebox shortfall. MTA chief Jay Walder floated the idea of off-peak discounts in an interview in today’s New York Times. While Walder didn’t […]