Today’s Headlines

  • It Just Got Harder to Clean Up Albany (NYT 1, 2; News 1, 2)
  • SUV Driver Kills Man Crossing East Tremont Ave; NYPD: Crash Appears to Be “Accidental” (News)
  • Richard Rojas Pleads Not Guilty to Murder for Times Square Motor Rampage (NYT, News, Post)
  • MTA Tells Queens Pols CBTC on the 7 Line Still on Schedule to Go Live This Year (QChron)
  • Bay Ridge Council Candidate John Quaglione — He’ll Go to the Mat for Free Parking! (Bklyn Eagle)
  • Review of Penn Station Concourses Will Proceed Without Tom Prendergast (Politico)
  • What WNYC Learned From Week 1 of the Summer of Hell
  • The Data in This Bike-Lanes-Boost-Property-Values Story Is Very Thin (DNA)
  • Driver Crashes Into West Brighton Nail Salon, One Injured (Advance)
  • FXFOWLE and SSE’s Vision for the Driverless Future Looks a Lot Like Streetopia (Crain’s)

More headlines at Streetsblog USA

  • Larry Littlefield

    The re-signaling of the Flushing line began some years before 2000.

    It was supposed to be the second pilot project, after which the installation of CBTC was supposed to be fast and cheap.

    The world has changed since then, thanks to the advance of information technology.

    When you consider the possibility of driverless motor vehicles in mixed transit on zillions of roads with all kinds of random elements, EZ-Pass, Google Maps and its competitors, how hard could it be to have safe, driverless transit in a limited, fixed environment with limited points of intersection and one kind of vehicle?

  • Joe R.

    The MTA seems to be expert at making what should be relatively easy, boilerplate projects into interminable, overly complex disasters.

  • bolwerk

    How hard? Driverless trains were possible in the 1960s.

  • Fool

    But when labor lights them on fire…

  • Larry Littlefield

    Long memories, eh?

  • sbauman

    It was supposed to be the second pilot project, after which the installation of CBTC was supposed to be fast and cheap.

    The basis for CBTC being cheap and fast is that there was supposed to be no wayside equipment in the tunnels. The wayside equipment was supposed to be in the stations and communications to/from trains were to be via radio with base stations located in the stations.

    The first problem was that the radios did not work in the tunnels. The second problem is that the trains needed beacons placed in the tunnels to determine where they were. These facts became known before they decided to proceed with the Canarsie Line Project.

    When you consider the possibility of driverless motor vehicles…

    The advances in driverless motor vehicles has been for autonomous vehicles. The model for smart highways or vehcile to vehicle communications is also being pursued. That’s not where progress is being made. Tesla, Uber, Google, Apple, et al are pursuing autonomous vehicles. If the MTA pursued this model, then driverless trains would not need an expensive infrastructure in place before the first driverless train could operate. The autonomous trains would have vision systems that would detect foreign objects on the tracks.

    Platforms doors would not be required to guard against people on the tracks. Unfortunately, transit is a very small market and the companies in the vanguard of driverless vehicles are looking for bigger fish than the NYC Subway system. Besides, dealing with the MTA is enough to make any rational company look for another customer.

  • cjstephens

    The DNA story on Bike-Lanes-Boost-Property-Values is indeed weak. However, can anyone point me in the direction of a study that does support the proposition that bike lanes boost (residential) property values? DOT was talking about putting a bike lane in front of my building, and many of my NIMBY-er neighbors were opposed. I would love to be able to counter their skepticism with some data.

  • Andrew

    The re-signaling of the Flushing line began some years before 2000.

    No it didn’t. Flushing CBTC project was awarded in 2010. (Perhaps you’re thinking of modernizations of some of the interlockings, which were prerequisites for CBTC but were fundamentally separate projects.)

    When you consider the possibility of driverless motor vehicles in mixed transit on zillions of roads with all kinds of random elements, EZ-Pass, Google Maps and its competitors, how hard could it be to have safe, driverless transit in a limited, fixed environment with limited points of intersection and one kind of vehicle?

    The Flushing line needs more than a possibility. It needs to actually provide subway service, typically at least the 27 trains per hour it delivers today, and preferably more.

  • Larry Littlefield

    Yes, I’m thinking of the prior projects. The whole thing has taken nearly 20 years.

    At that pace/cost obviously this can’t be done. What can be?

  • Larry Littlefield

    Autonomous subways wouldn’t be able to see around corners, and if radios from stations wouldn’t work (I’m not sure that was ever the idea — it was never mentioned to me) train to train radio communications wouldn’t work either.

    “Unfortunately, transit is a very small market and the companies in the vanguard of driverless vehicles are looking for bigger fish than the NYC Subway system.”

    Why not try this? As telecom shifts to 5G and wire line telecom employment falls, hire some workers into the MTA. And then offer a company, say, $5 billion to resignal the whole thing over five years (rail portion), with MTA workers installing all the wayside equipment?

    Not much for Google and the like? It would help burnish their green bona fides, however. In fact, free runs for GoogleFiber through the subway tunnels, and free rooms for GoogleFiber in many of the stations, could be part of the deal.

  • sbauman

    Autonomous subways wouldn’t be able to see around corners

    Autonomous ATO would use the existing block system with wayside signals, just as autonomous vehicles use existing traffic signals and signs. There already are repeater signals, where the tripper might be around the corner.

    As telecom shifts to 5G and wire line telecom employment falls, hire some workers into the MTA. And then offer a company, say, $5 billion to resignal the whole thing over five years (rail portion), with MTA workers installing all the wayside equipment?

    There is no wayside equipment with autonomous ATO. All the smarts to see and decode what’s ahead is embedded in the vehicle. That’s why it’s called autonomous. How many trains have to be equipped? NYCT operates approximately 600 trains in maximum service. Suppose it costs $1M dollars to equip a train, that’s an expenditure of $600M not your $5B. BTW, I believe the cost will be lower than $100K per train. Tesla is implementing it on a car that retails for $50K

    Not much for Google and the like?

    NYCT accounts for approximately 50% of the peak subway traffic in the US. That means the total US market is 1200 units. That’s a boutique business. Nobody makes money on that small a market at that price.

  • Joe R.

    Aren’t the long-term plans with autonomous vehicles to get rid of traffic signs and signals? Can’t see why they’ll be needed once human driving is banned and vehicles will communicate with each other.

    Doesn’t CBTC get rid of the existing wayside signals entirely? I thought that was the point. The signals are getting too old to cost-effectively maintain, so they would be replaced with CBTC transponders.

    I’m not seeing anything which looks like conventional wayside signals here:

  • bolwerk

    Continued learned helplessness?

  • bolwerk

    That kind of autonomy seems completely undesirable with rapid transit. The beauty of rapid transit is you can move large vehicles quickly and close together. Autonomous vehicles would only retard that over off-the-shelf railroad signaling/dispatching systems that work the world over.

  • sbauman

    The Flushing Line used to provide 36 trains per hour, with its current block system.

  • sbauman

    Autonomous vehicles would only retard that over off-the-shelf railroad signaling/dispatching systems that work the world over.

    Perhaps a more rigorous definition of an autonomous driverless vehicle is in order.

    An autonomous self-driving vehicle and a human driven vehicle use the same sensory inputs: namely vision and audio. The computer in the car or the driver react to these stimulii to control the vehicle’s motion. Whether an autonomous driverless vehicle would be superior to a human driven vehicle depends on the accuracy/speed of the visual/audio recognition software and its cost. There have been simultaneous advances in vision recognition software and reductions in the size and cost of embedded computers that now make this paradigm practical. These developments mean there should be no trains per hour penalty for an autonomous driverless vehicle vs. a human driver with the same stimulii.

    The alternative driverless vehicle paradigm requires continuous radio communications between the vehicle and either a smart highway, vehicle to infrastructure (V2I), or between adjacent vehicles (V2V). Such implementations foresee a closed infrastructure, where unequipped vehicles (or pedestrians) are not present. The benefit for this approach is higher capacity (vehicles per hour). However, there have been no real world demonstrations to date. It’s been only simulations.

  • sbauman

    Doesn’t CBTC get rid of the existing wayside signals entirely?

    What do you mean by wayside signals?

    If you mean the signal lights, then CBTC isn’t required to remove them. Cab signalling systems have been around since the 1920’s. A continuously updated allowable speed indication is displayed in the cab, instead of on wayside signal lights. The LIRR installed cab signals after the Richmond Hill disaster.

    If you mean the absence of any wayside equipment, then CBTC doesn’t achieve that goal. There’s a ton of wayside computers, radio transmitters, radio receivers, position beacons, etc. Much of that equipment is located between stations, in tunnels.

  • Joe R.

    I meant conventional wayside signals. Obviously CBTC requires wayside equipment but once installed the need to maintain 75+ year old wayside signals should no longer need to exist.

  • Joe R.

    My understanding here is that autonomous road vehicles will need both types of inputs (radio and visual/audio). The radio inputs will be sufficient to keep autonomous vehicles from colliding with each other (hence no need for traffic signs or signals) but you’ll need audio and visual inputs any place there are pedestrians and cyclists. I’m assuming the cars will be programmed to detect pedestrians or cyclists in time to slow or stop for them while they’re crossing. That basically means anything human-powered can just stay in motion with the assumption any autonomous vehicles will automatically grant them the right-of-way. Cyclists and pedestrians will still obviously need to watch out for each other but at least they no longer need to worry about motor vehicles.

  • sbauman

    once installed the need to maintain 75+ year old wayside signals should no longer need to exist.

    That depends on what the capacity of the line should be, when either a train’s radio fails or non-CBTC equipped rolling stock is detected.

    CBTC requires an auxiliary wayside system (AWS) to handle these two contingencies plus interlockings and broken rail detection. The AWS is a standard fixed block system with insulated joints, etc.

    The Canarsie Line’s AWS block lengths are long, essentially between switches. Consequently, following trains cannot enter a block until it has cleared. The Canarsie Line’s AWS capacity is 3 tph because of the long block length.

    If greater capacity is required, then more and shorter AWS blocks are required. If the AWS capacity specified to match the performance of the 75+ year old wayside signals, then the AWS would require as many blocks and associated equipment.

    Why would such a duplicate system be suggested? The Canarsie and Flushing CBTC installations did not provide a sufficient number of CBTC-equipped rolling stock to permit operation at historic service levels. That’s 24 and 36 tph for the Canarsie and Flushing lines, respectively.

    Should there be a demand for increased service levels on either of these lines, additional rolling stock would have to be converted with a long lead time and considerable expense. Without CBTC, rolling stock from over served lines could be transferred with no conversion expense and short lead time.

  • sbauman

    My understanding here is that autonomous road vehicles will need both types of inputs (radio and visual/audio).

    You are mistaken. There are some driverless vehicles that use both autonomous and V2I or V2V paradigms.

    The driverless vehicles being developed by Tesla, Uber, Google, Apple, and many car makers are strictly autonomous – no V2I or V2V radio communications.

    The military is also interested in strictly autonomous driverless vehicles. There are illegal, low cost, wide band jammers available on the Internet that will disrupt V2I or V2V communications. Imagine a tank column being halted by a $25 purchase from Alibaba. That’s the next IED.

  • Joe R.

    I think that issue is moot. Larry has mentioned a number of times that the MTA has no money for either additional trainsets or train operators. Long term it’s in the city’s best interest to give employers incentives to stagger work hours so we don’t spend a ton of money on equipment which is only needed during the peak hour or two.

  • Joe R.

    Why not have both? If the V2I or V2V communication is disrupted then the vehicle defaults to strictly autonomous mode. That’s the best of both worlds.

  • sbauman

    Larry has mentioned a number of times that the MTA has no money for either additional trainsets or train operators.

    They will have even less money, if they blow $20B to implement CBTC systemwide.

    Long term it’s in the city’s best interest to give employers incentives to stagger work hours so we don’t spend a ton of money on equipment which is only needed during the peak hour or two.

    Why don’t you examine the hourly tables for subway use in NYCMTC’s Hub Bound Studies through the years.

    https://www.nymtc.org/Utility-Menu/Archive/Data-and-Modeling-Archive/Hub-Bound-Travel-Archive

    You will discover that peak hour demand has decreased substantially over the last 50 years. The daily ridership increase is due entirely to increases during the off peak hours. It’s the cure you are suggesting that’s the MTA’s current problem.

    This raises two questions.

    First, if peak demand was greater 50 years ago, how did they handle the demand? That’s easy, they ran a lot more trains during the peak hour with the same signal system currently in use.

    Second, if peak demand has fallen, why are they claiming they need to increase peak service level capacity to handle the reduced peak demand?

  • sbauman

    Why not have both? If the V2I or V2V communication is disrupted then the vehicle defaults to strictly autonomous mode.

    If the autonomous mode can solve the problem, what additional benefit does V2I or V2V provide?

    That’s the best of both worlds.

    It’s certainly the most expensive.

  • Joe R.

    $20B versus how much to replace or upgrade existing wayside signals? To me the benefits of CBTC are two-fold. One, trains can operate at maximum capability immediately. That may or may not happen with wayside signal upgrades. When trains get across the line faster you can increase tph without adding trainsets. Two, eventually you can get rid of the TO entirely, and you can get rid of the conductor right away. The TWU is an impediment to doing that for now, but as the system continues running out of money, the TWU will have less and less leverage to keep people in obsolete positions.

    As for the MTA’s claim that they need to increase peak service level capacity, that’s probably mostly the TWU talking. More jobs for TOs, conductors, track workers, signal maintainers,and so forth.

    The problem with the MTA is that it’s too big for its own good. The left hand doesn’t know what the right hand is doing. Each part of it operates in a manner beneficial only to that part, regardless of the effect on the whole.

  • Joe R.

    Isn’t V2I and V2V the only thing which can safely enable much higher travel speeds on highways, and intersections where cars on perpendicular streets seamlessly glide past each other? Those are two oft-touted benefits of autonomous vehicles.

    Without radio communication you’re stuck using lines of sight. With human drivers this limits speeds through intersections to 15 or 20 mph. Autonomous vehicles remove reaction times from the equation, so maybe that gets speeds up to 30, perhaps 40 mph. To go any faster on streets with intersections vehicles will need to know where other vehicles are before they’re in the line of sight.

  • sbauman

    Isn’t V2I and V2V the only thing which can safely enable much higher travel speeds on highways, and intersections…

    If you will note in my definition of the V2I and V2V paradigms, I mentioned that their claimed benefit is higher capacity (vehicles per hour). I repeat this because I don’t want you to believe that I’ve been misleading you.

    The “claimed” qualifier to benefit is added because there has been no experimental verification to date. Mathematical simulations have been the only basis for these claims. These simulations have ignored radio transmission errors, traffic saturation, processor overload, and pedestrians to name a few. They are also subject to malevolent forces, which may limit their implementation, even if the problems omitted in the mathematical simulations are solved.

    Those are two oft-touted benefits of autonomous vehicles.

    Let’s be precise. I believe you are referring to V2I and V2V vehicles, not autonomous vehicles as I defined them.

    Without radio communication you’re stuck using lines of sight…To go any faster on streets with intersections vehicles will need to know where other vehicles are before they’re in the line of sight.

    You’re assuming that radio communications is continuous and instantaneous. It’s a sampled data system. Status snapshots are sent at regular intervals. Not all the snapshot transmissions are received correctly. Any driverless system that relies on such data inputs must make allowance for these real world occurrences and many others. The strategy to deal with these real world occurrences is to provide wider safety margins that could be interpreted as longer reaction times. The result is that theoretical increased speeds cannot be safely achieved.

    The desired benefit is increased capacity (vehicles per hour) not increased vehicle speed. Only the GM Futurama exhibits at the 1939 and 1964 Worlds Fairs envisioned a driverless system where cars traveled bumper to bumper at 80+ mph. That vision is no closer to reality.

  • sbauman

    $20B versus how much to replace or upgrade existing wayside signals?

    There are approximately 12,000 signals in the subway system. At the time CBTC was considered for the Canarsie Line, the cost to replace/upgrade a wayside signal was around $500K. That works out to $6B using the obsolete technology that the MTA preferred. It would be much less using technology that is used in other countries.

    trains can operate at maximum capability immediately…When trains get across the line faster you can increase tph without adding trainsets.

    The increased electrical power costs for operating at faster accelerations and higher cruising speed would more than swamp out any gain from fewer operating trainsets to provide the same service level.

  • Joe R.

    That works out to $6B using the obsolete technology that the MTA preferred. It would be much less using technology that is used in other countries.

    And in the end though you have a system which still requires a human to operate it. At least with CBTC in theory you can go with trains which operate themselves. Getting the TWU on board is of course the main obstacle to that, but my point is the MTA should set the groundwork for driverless trains so it can eventually get rid of TOs.

    The increased electrical power costs for operating at faster accelerations and higher cruising speed would more than swamp out any gain from fewer operating trainsets to provide the same service level.

    First off, cost isn’t the only thing which is important here. Other metros can and do operate at much higher average and cruising speeds. Both are important if the MTA wants to attract and retain customers. Second, with start/stop service most of the energy used goes for acceleration. The new trains can recover a large percentage of this energy while braking. Third, labor by far is the largest line item in a transit system. Faster trains in essence make labor more productive, decreasing the cost per train mile, and likely offsetting any increase in power costs. Fourth, power to light the tunnels and stations accounts for a fairly large percentage of the MTA’s energy use. When all is said and done, I doubt faster trains would increase energy use by more than a few percent, if that. Five, the MTA will save on track maintenance. Most railway equipment tends to have oscillations and rocking behavior between speeds of 25 to 45 mph. This is of course damped out by the running gear to some extent but it nevertheless causes more track wear which faster or slower speeds would avoid. Obviously we don’t want to run trains at less than 25 mph, so we should run them faster than ~45 mph. Right now MTA speed patterns largely fall into the critical range. It’s no wonder it seems so much track needs work.

    There’s never been any good reason for the MTA to have slowed down the trains. To say we shouldn’t return to previous speeds (or faster whenever possible) makes no sense. Transit is a service business. Speed is one of its key selling points. Faster trains attract customers. The extra revenue in turn generally offsets any increase in energy costs. If it didn’t, then why are metros the world over running at average speeds which are often 1.5 to 2 times that of the MTA?

  • Joe R.

    The “claimed” qualifier to benefit is added because there has been no experimental verification to date. Mathematical simulations have been the only basis for these claims. These simulations have ignored radio transmission errors, traffic saturation, processor overload, and pedestrians to name a few.

    There’s been no experimental verification because my understanding is we’re still developing the standards! You can’t test something which doesn’t yet exist. We do however have real world data from loads of other wireless transmission schemes like Wifi and cellular. These transmit data at far higher rates than would be needed for driverless vehicles and already have robust error correction mechanisms. Those problems are largely solved. Jamming or other interference is certainly a potential issue, which is why the vehicles would have to be programmed to default to visual and audio inputs in the event data transmission was compromised.

    The desired benefit is increased capacity (vehicles per hour) not increased vehicle speed.

    It’s both. Once you remove human reaction time from the equation (and V2I or V2V aren’t prerequisites for that), “safe” following distance is practically zero regardless of speed, at least in environments which are free of unpredictable obstacles like pedestrians or cyclists. The vision of autonomous vehicles safely meshing with each other through urban intersections at speeds of 150 mph probably will never be a reality unless we prohibit walking or cycling but no reason highway speeds and capacities won’t greatly increase once humans are out of the equation. In the end most long trips are mostly done on highways anyway. Even if speeds on urban streets mostly stay the same, autonomous vehicles can greatly boost average travel speeds except on strictly local trips.

    My guess is autonomous vehicles are going to have and need much stricter maintenance requirements in order to avoid the problem of crashes due to mechanical failure. That’s no big deal given that we’re moving towards shared fleet vehicles where the extra maintenance cost is amortized among many users.

  • sbauman

    There’s been no experimental verification because my understanding is we’re still developing the standards! You can’t test something which doesn’t yet exist

    The usual practice has been to get something to work and then set the working model as the standard. Before the administration change, the USDOT was sponsoring V2I tests and NYC was one of the test areas. I haven’t heard what that program’s fate has been.

    We do however have real world data from loads of other wireless transmission schemes like Wifi and cellular. These transmit data at far higher rates than would be needed for driverless vehicles and already have robust error correction mechanisms.

    There is an intrinsic capacity for all communications channels. Reliable, error free, communications is possible so long as the transmission rate is below the channel capacity. The lower the transmission rate, the more reliable the communications. If the transmission rate exceeds the channel capacity, then nothing gets through. Telcos historically worked on the assumption that a maximum 10% of the phones are active at at any given instant. When that percentage was exceeded, like on Mother’s Day, there were delays in getting a dial tone. What would the effect on vehicle safety, if there were a 1 second delay in sending or receiving a status packet?

    Jamming or other interference is certainly a potential issue

    Traditional jamming increases the noise level, which in turn reduces channel capacity. The most effective jamming technique is to mimic the signal that is trying to be received. Such “jamming” can occur naturally, when too many vehicles try to reach the same receiver. A similar occurrence happens, when single messages reach the receiver through multiple paths because of reflections due to buildings, other vehicles or rain drops. These are all real world considerations that cellphones and wi-fi handle by blocking out communications until these temporary conditions pass. That’s not an acceptable solution for the V2I or V2C applications.

    Once you remove human reaction time from the equation…

    Human reaction time is 1 second in the equation – look at the ITE’s equation for the yellow time for traffic signals. That’s only 16 to 33% of stopping time for speeds from 68 down to 27 mph. Human reaction time just does not play that big a part in establishing the safe distance between vehicles.

    The vision of autonomous [you mean V2I or V2V] vehicles safely meshing with each other through urban intersections at speeds of 150 mph probably will never be a reality unless we prohibit walking or cycling but no reason highway speeds and capacities won’t greatly increase once human drivers are out of the equation.

    We already experience 200 car pileups on interstates, during snowfalls. Think of how much larger such pileups will be, when V2I and V2C vehicles are traveling 150 mph. Snow isn’t required. A tire blowout, due to road debris, should be sufficient to create a giant chain reaction pileup.

  • sbauman

    And in the end though you have a system which still requires a human to operate it.

    The current and future CBTC systems being considered by the MTA still require human operation. The human’s task is to press a button to start the train at each stop. As I noted to Mr. Littlefield, the cost for implementing autonomous driver technology, should be less than $60M. That brings the cost of upgrading/replacing the current block system with similar technology to $6.06B for a driverless system vs. $20B for a system that still requires a driver to push a button.

    Other metros can and do operate at much higher average and cruising speeds.

    Most other metros do not have stations spaced as closely together (0.5 mi). Therefore, they spend more time traveling between stations and can derive more benefit from higher cruising speeds. The few metros that have stations as closely spaced as NYC, e.g. the RATP in Paris, exhibit the same cruising speeds.

    Second, with start/stop service most of the energy used goes for acceleration. The new trains can recover a large percentage of this energy while braking.

    That experiment has been a failure. The failure was due to not being able to store the energy recovered with regenerative braking. The idea was that there would be an adjacent, accelerating train able receive the energy without such energy storage. Such trains were too few and far between. The result was that standard dynamic/air brakes were used for most stops. Unfortunately, the MTA sized the dynamic and air brakes, assuming that regenerative braking would be used for most braking. The result was that the resistor banks for the dynamic brakes overheated and the brake shoes for the air brakes exhibited excessive wear.

    Third, labor by far is the largest line item in a transit system. Faster trains in essence make labor more productive, decreasing the cost per train mile, and likely offsetting any increase in power costs.

    There are 1.5 employees associated with maintenance for every employee actually operating a subway train. The breakdown is 1 employee for facility maintenance and 0.5 for vehicle maintenance. One should expect the need for vehicle maintenance to increase faster than operating speed.

    Fourth, power to light the tunnels and stations accounts for a fairly large percentage of the MTA’s energy use. When all is said and done, I doubt faster trains would increase energy use by more than a few percent, if that.

    NYCT consumed 1.7B kw-h at a cost of $352M for propulsion in 2015. That $352M exceeds the labor cost for subway car maintenance.

    The power to propel a subway train on level track varies as the cube of its velocity. Increasing the cruising speed by only 20% (from 30 to 36 mph) would increase the power requirements by 73%.

    Five, the MTA will save on track maintenance. Most railway equipment tends to have oscillations and rocking behavior between speeds of 25 to 45 mph…

    Most railway equipment is designed for freight operations. Any track oscillations due to railway rolling stock motion would most likely reflect freight train dynamics. Passenger only operations permit designing both rolling stock and right of way to move such oscillations well above travel speeds. On such designed rolling stock and right of way, speed is what determines how quickly rails wear out. The tests conducted in the 1960’s with the PRR’s Metroliners, showed that it was uneconomical to operate these trains above 125 mph, due to track wear. That’s why the Metroliner service maxed out at 125 mph, even though the rolling stock was capable of 150 mph.

    There’s never been any good reason for the MTA to have slowed down the trains.

    It may not be a good reason but the emergency brakes don’t work very well. They don’t even meet their design specifications. Poorly performing emergency brakes have been responsible for at least two collisions – including the Williamsburg Bridge collision that finally led the MTA to take remedial action. That action wasn’t to fix the emergency brakes, although the NYS PTSB recommended such action following the 1991 Union Square derailment. The MTA’s response has been to slow down the trains.

  • Andrew

    No signal system is cheap. Yes, CBTC is expensive – but how expensive is it compared to the alternative of a brand new wayside signal system comparable in achievable capacity and safety?

    This is the first I’ve heard that there was ever a serious plan for wayside equipment to only be in the stations. Although there is certainly some wayside equipment in a CBTC system, there is far less than in a traditional wayside signal system.

  • Andrew

    If I replace the stove in 2000 and replace the fridge in 2010 and paint the kitchen in 2020, has it taken me 20 years to renovate my kitchen? Flushing CBTC was awarded seven years ago and is nearing completion. That’s not “nearly 20 years.”

    As Mark Wild points out, experience brings faster implementation.

  • Andrew

    Autonomous ATO would use the existing block system with wayside signals

    So you’d replace the existing block signal system with a brand new block signal system? How much does that cost in comparison to a brand new CBTC system?

    Or you’d leave the existing block signal systems – over 80 years old in significant portions of the subway – and just pretend that there are no reliability concerns that come with age?

  • Andrew

    Doesn’t CBTC get rid of the existing wayside signals entirely?

    Yes and no.

    First – a signal modernization replaces all of the old signals. Whatever wayside signals you see out there on a modernized signal system are brand new. So, in that sense, yes, it gets rid of all of the existing wayside signals, even if it needs to put in new ones.

    Second – CBTC, at least as implemented in New York, still requires home signals at interlockings. It also may, optionally, include additional wayside signals, to better accommodate non-CBTC trains. The end of the Canarsie line between Broadway Junction and Rockaway Parkway has enough new wayside signals for a 5 minute non-CBTC headway. At least the two-track section of the Flushing line, and I think the local tracks on the three-track section, have a similar wayside system. I’m not sure of the plans for Queens Boulevard and the rest of the subway system.

    I’m not seeing anything which looks like conventional wayside signals here:

    There are relatively few of them, but they do exist. They blink green when a CBTC-equipped train approaches.

  • Andrew

    You mean that the then-new Transit Authority claimed to schedule 36 trains per hour in a promotional document in 1954.

  • Andrew

    The primary reason to modernize a signal system is to remove the old equipment. If CBTC relies on retention of the old wayside signals (as more than a temporary stopgap measure), the signal system hasn’t been modernized.

  • Andrew

    After Flushing, all NYCT CBTC installations will be to a common standard. All new cars will be equipped for this new CBTC standard. The challenge in transferring cars from one line to another is a temporary challenge.

    In the meantime, it is far less costly to equip additional cars with CBTC as needed than to outfit the entire line with a high-capacity wayside signal system in addition to a CBTC system.

    Today, there are enough CBTC-equipped cars to increase service on the L – some CBTC-equipped trains regularly run on the J/Z. The current constraint is power, not cars. The power constraint will soon be raised from 20 tph to 22 tph during the Canarsie Tube outage, and I believe the R179 order includes enough cars to go to the J/Z to permit more of the CBTC-equipped R160’s to go to the L.

  • sbauman

    The document in question was part of the Transit Authority’s annual report. It was a legal document and was a reprise of a similar document printed by the Board of Transportation in 1949.

    There was quite a discussion regarding 36 tph operation around 2000 on a now defunct forum. NYCT conducted a Saturday experiment to operate 30 tph on the Flushing Line. The statement was that it would have set a new service level record for the Flushing Line.

    I mentioned the TA’s 1954 report. The 36 tph schedule was backed up by someone who had an early 1970’s TDI timetable. He listed the Times Square arrival times.

    Another forum participant worked for NYCT. He did not believe it was possible. He researched old internal schedules from that time period. He reported back that those schedules did indeed show they operated 36 tph, however he was still skeptical it had ever happened.

  • Andrew

    There’s never been any good reason for the MTA to have slowed down the trains.

    Well, aside from the minor detail that, at the higher speeds that the cars used to be capable of, the NTSB reports that “it was determined that most signal blocks in the NYCT system [did] not have adequate emergency braking distance.”

    This wasn’t a matter of one or two noncompliant signals. This was a matter of a major mismatch between design and reality that rendered the signal system plainly unsafe. Ignoring it wasn’t an option.

  • Andrew

    I have no doubt that 36 tph were scheduled on paper, but I share your friend’s skepticism that it ever happened on a regular basis in practice.

  • Joe R.

    The tire issue is already on the way to being solved. There are lots of competing candidates for airless car tires. I’ve been using airless bike tires for the last 8 years. As an alternative, the car could even have a set of rail wheels and run on rails when on highways. That avoids lots of problems by restricting movement solely to the guideway. And there’s lots of real-world data on train behavior at speeds of 150 mph or more.

  • Joe R.

    Right, but that issue could have been mostly fixed by increasing the emergency braking rate (which had been decreased prior to the WB incident). Yes, this would have increased maintenance costs but at least the trains wouldn’t have been neutered. After that, you could have restricted speeds in the remaining problem areas.

  • Joe R.

    I wasn’t sure about the blinking green signals. They obviously don’t look like conventional wayside signals in terms of signal aspects but if that’s their function with non-CBTC trains then at least we got rid of the old, unreliable, wayside signal system.

  • Andrew

    The standard emergency braking rate hadn’t been decreased, but the standard wasn’t being properly upheld, as the brake cylinder pressure was too low. Increasing the brake cylinder pressure wasn’t in and of itself sufficient to render the signal system safe – the signal system was designed to be safe under assumptions of both braking and acceleration capabilities, and bringing only the braking capabilities back into line isn’t enough to do the job.

    Sorry. I don’t like it either, but I’m not going to ignore the facts just because they’re inconvenient.

    The good news is that none of this applies to CBTC. Not only is CBTC a brand new signal system designed from scratch to be safe at the higher speeds, but the cars know when they’re operating in CBTC. All of the post-1995 cars are capable of the higher acceleration rate but were designed to cap themselves at the lower acceleration rate when not in CBTC mode, to ensure safety.

  • Andrew

    They still have the standard red, yellow, and green aspects, but they revert to blinking green when CBTC is in control.

  • Joe R.

    Most other metros do not have stations spaced as closely together (0.5 mi). Therefore, they spend more time traveling between stations and can derive more benefit from higher cruising speeds.

    What about express trains? For example, the QB express is scheduled for 12 to 14 minutes from Queens Plaza to Forest Hills. If the R160s ran unfettered you could do that in 8 minutes. Knocking 4 to 6 minutes off the schedule just between two express stops is huge. And on average you’ll knock off about 10 seconds between local stops compared to running the trains with their acceleration rates retarded. As we’ve seen on the L, the MTA appears to be running the equipment to full capability when in CBTC mode. No reason it won’t do that on every CBTC-equipped line. In the case of the L, it only knocked two minutes off the schedule but that’s on a line with no express runs and some areas with speed restrictions due to curves.

    The power to propel a subway train on level track varies as the cube of its velocity. Increasing the cruising speed by only 20% (from 30 to 36 mph) would increase the power requirements by 73%.

    Not true. The cube root equation only applies to vehicles which only have aerodynamic drag like aeroplanes. A considerable percent of train drag at lower speeds is wheel resistance. The resistance equation (in pounds) for EMUs is 1.3T + 29N + 0.03TV + CV² where C depends upon the number of cars and drag coefficient. T is the weight in tons and N is the number of axles. V is the speed in mph. A subway train weighs roughly 40 tons per car. That’s 400 tons and 40 axles for a 10-car subway train. For aerodynamics the first car contributes much of the load. Maybe for a 10-car train C is roughly 0.7. So that gives us R = 1680 + 12V + 0.7V². At 30 mph resistance is 2670 lbs and the power is 213 HP. At 36 mph resistance is 3019 pounds and power is 290 HP. Both figures pale next to the acceleration power of at least 4000 HP (for DC trains with 100 HP motors). Moreover, in local trainset service between local stops you’re generally running at full power (unless the equipment is neutered) until you hit the brakes for the next stop. Sometimes you might start coasting after you hit some given speed if it’s determined that further acceleration will only shave some fractions of a second off the schedule.

    That experiment has been a failure. The failure was due to not being able to store the energy recovered with regenerative braking. The idea was that there would be an adjacent, accelerating train able receive the energy without such energy storage. Such trains were too few and far between. The result was that standard dynamic/air brakes were used for most stops.

    You can also used the power generated during braking for train lighting and HVAC, or even feed it back into the grid to power station lights. NJTransit does this with their Arrow IIIs. If there is still surplus energy, then you use batteries. This is already being tried out on some transit systems:

    http://www.sciencealert.com/london-is-now-recycling-energy-from-train-brakes-to-power-their-stations

    https://www.extremetech.com/extreme/180636-philadelphia-unveils-new-hybrid-subway-system-that-uses-prius-like-regenerative-breaking-to-feed-energy-back-into-the-grid

    https://www.extremetech.com/extreme/180636-philadelphia-unveils-new-hybrid-subway-system-that-uses-prius-like-regenerative-breaking-to-feed-energy-back-into-the-grid

    Most railway equipment is designed for freight operations. Any track oscillations due to railway rolling stock motion would most likely reflect freight train dynamics. Passenger only operations permit designing both rolling stock and right of way to move such oscillations well above travel speeds.

    You’re talking about critical oscillations at high speeds which can derail trains. I’m talking about oscillations which occur in the 25 to 45 mph range in nearly all passenger equipment I’ve ridden in. That includes the subway, NJ Transit (their Arrow IIIs rock noticeably when going through 30 mph zones), and Amtrak.

  • Andrew

    For example, the QB express is scheduled for 12 to 14 minutes from Queens Plaza to Forest Hills. If the R160s ran unfettered you could do that in 8 minutes.

    I’m looking forward as much as anyone to the higher acceleration rates under CBTC, but you pulled that number out of thin air.