16 December 2010

Progress at Transbay?

The many failures of the Transbay Transit Center design, as a train station, have previously been pointed out (see here and here). A November CAC briefing on the latest design iterations of the platform levels finally shows some incremental progress. After digging that $400 million hole, we may actually find signs of intelligent life down there.
  • Caltrain tracks are moved to the northwestern edge of the train box, closest to downtown. While this change was advertised to avoid the foundations of the high-rise at 201 Mission St, (TJPA headquarters!) a better reason is that Caltrain ridership at Transbay will likely exceed anything HSR can muster. In this case, two wrongs have inadvertently made a right.

  • With Caltrain tracks switched around, the future plans for tail tracks are now infeasible, and rightly so because underground tail tracks are the most expensive way you could possibly think of to park trains while they are not producing revenue. Again, two wrongs (the desire for tail tracks, and the competing desire for ramrod-straight platforms) make a right (no more tail tracks!)

  • A new entrance is provided on Beale Street, with an escalator bank perpendicular to the train box, reaching towards downtown. Caltrain passengers can directly access the Caltrain platform without any scenic detours through the rabbit warren of sequential, airport-like functional spaces intended for HSR, which make absolutely no sense for the daily commuter. This new entrance is a big step forward because it suggests for the first time a concern with minimizing the overall travel time of Caltrain passengers. If it's not too much to ask, two more such perpendicular access points should be added at Fremont and 1st streets (better yet: emerging north of Mission, so pedestrians don't have to wait to cross those busy intersections.)

  • Additional curved turnouts seem to have sprouted in the station throat, thankfully not labeled "emergency crossover", suggesting the possibility of improvements in operational efficiency and flexibility--although once again, nothing would beat a common platform height.
Happy Holidays to everyone.

67 comments:

  1. Is it just me, or does platform access appear to be spaced evenly and frequently throughout the mezzanine, without any designated "waiting" or "security" impeding access to any of the platforms?

    ReplyDelete
  2. Though it looks like those elements may be reinstated at some point in the future.

    ReplyDelete
  3. Why is lack of tail tracks such a good thing because the "bang for the buck" on them is low?

    I'd think they would still help increase train throughput a bit given that Caltrain only has two platforms down there.

    I thought I heard that there would be an underground passageway to connect with the BART station. Is that still the case, or am I confusing this passageway with one planned for Central Subway?

    ReplyDelete
  4. Why is lack of tail tracks such a good thing because the "bang for the buck" on them is low?

    Because as taxpayers we actually care about spend scarce funds on things that are relatively useful.

    ReplyDelete
  5. Tail tracks are relatively useless because they don't give you significantly more platform capacity - that is, rather than push a train onto the tail tracks in order to get another train into the station, why not just reverse the first train out of the station in revenue service?

    ReplyDelete
  6. Beale Street is also the provision for a connection to Muni/BART at Embarcadero. It looks like it would be built as part of DTX or Phase 2. It is conceivable that it could portal out on the north side of Mission Street or continue two city blocks further north to Market Street. All of that is unfunded at this point.

    ReplyDelete
  7. Tail tracks have only one use: they allow trains to enter the station at full throat speed, letting it slow down later, which reduces turnaround time. However, for this there's no need for full-length tail tracks - a few tens of meters suffice. The Chuo Line turns 28 tph at Tokyo on two tracks with about 30 meters of tail tracks, and the 7 train in New York turns 24 tph at Times Square on two tracks with about 80 meters.

    ReplyDelete
  8. One benefit I can think of is (perhaps) if a train develops a mechanical difficulty or is otherwise temporarily removed from service, it might be useful to have a spot to get it out of the way. But it's probably just as good to send it down to 4th St. or something.

    ReplyDelete
  9. Tail tracks also let you store trains, which means you don't have to send those trains back out to cause more conflicts in the station throat, which can help in the peak-of-the-peak (and not really beyond that, since there would only be two tracks). Also, with MU trains, they might want to drop cars after the morning peak and couple them back on for the afternoon. Tail tracks would be convenient for this, though splitting every other train in half and running the rest back to a yard at 4th/King would accomplish more or less the same thing too. And we don't really know what Caltrain's operations will look like anyway.

    ReplyDelete
  10. which means you don't have to send those trains back out to cause more conflicts in the station throat

    (1) 'sending those trains back out' is exactly what ought to happen, and with bums in seats. Doing lots of servicing and non-revenue moves at your most highly constrained terminal is poor operations planning.

    (2) This might not be such a problem in the first place if you could assign any train to any platform, since the throat conflict (crossing from one side of the approach to the other) could be selected to occur either inbound OR outbound, as best suits traffic conditions. When you cast incompatible platform interfaces in concrete, you don't have that degree of freedom... which creates the "requirement" for those silly tail tracks.

    Use. The. Same. Platform. Height.

    ReplyDelete
  11. Well, reverse commute patronage isn't going to be as big as the primary commute direction. And hey, maybe it'll be worth the money to have those extra 2 tph full of paying customers in the peak direction. And it's not like sending those trains back out would make you any more money, because that would just shift demand from other reverse-commute trains, unless you're assuming that the reverse commute is also at capacity. Which strikes me as a bit implausible.

    ReplyDelete
  12. Actually, arcady, the reverse commute patronage heading out of San Francisco is increasingly substantial. Many people now choose to live in San Francisco and work along the Peninsula. Look at how SoMa has changed with all the new condos. The Peninsula used to be mostly residential, but now it's job-rich.

    Planners should definitely be thinking about developing the reverse commute trend, because it is efficient for the system.

    The basic point is that tail tracks and parking trains in expensive and constrained real estate is enormously wasteful. The DTX tunnel is not funded and is a much higher priority, although they should only use two tracks, not three.

    ReplyDelete
  13. I might add that the transit connections between the East Bay and the Peninsula are particularly inefficient and awkward (thank MTC for decades of futility). I suppose there's BART. but that's a long, circuitous ride to transfer in Millbrae.

    The Transbay Terminal is supposed to improve regional transit connections, particularly between the East Bay and the Peninsula. In theory, one can take an express bus on multiple routes from the East Bay or Marin to the Transbay Terminal. Transferring from BART in downtown SF shouldn't be too difficult either. From the Transbay Terminal, an efficient transfer to a Caltrain local or express can be made. East Bay commuters headed to the Peninsula will make the reverse commute out of the TBT even more significant.

    Mission Bay will increasingly be a major regional employment hub, but how are all these regional workers going to access the area?

    ReplyDelete
  14. What does moving the HSR platform to the inside curve do to the minimum curve radius on the HSR platform tracks?

    ReplyDelete
  15. Arcady, the cost of providing extra reverse peak service is very small. The capital costs are all sunk, and the operating costs boil down to a little bit of extra energy and maintenance. The drivers by and large get paid regardless of whether they work or sit at San Francisco and play pool until the p.m. peak.

    In contrast, the cost of tunneling a few more hundred meters in the CBD is substantial, and competent transit agencies don't do it. If they tunnel more under the CBD, they add stations, or connect the line to a line going in the opposite direction of the city.

    ReplyDelete
  16. Alon, I'm conjecturing that tail track could allow something like 12 tph peak and 10 tph reverse peak, instead of 10 tph peak and 10 tph reverse peak, with no way of fitting 12 tph going both ways through the throat and the platform tracks. And it could be worthwhile to provide the facilities for those extra peak trains. If providing them didn't involve digging very expensive tunnels, it could be worth it.

    ReplyDelete
  17. Adirondacker1280022 December, 2010 15:09

    Arcady, railroads all over the world, including some in the US manage much much more than 10 trains an hour in and out of two track stub terminal. Some of them manage more than twice that.

    ReplyDelete
  18. BART has no tail tracks at Fremont and runs 16 to 20 tph at that terminal.

    ReplyDelete
  19. Are the trains at Fremont filled to capacity both coming and going? I would assume that they're not. That would have a pretty significant impact on dwell time. And, as far as I know, BART runs service every 15 minutes on each of the lines going to Fremont, so that's 8 tph terminating. So, that's not a good example. Anyway, metro examples aren't as relevant here, because the throughput is going to be dominated by dwell time, which is a matter of how quickly you can empty the train and fill it back up again before sending it on its way. Anyway, since we're talking about Transbay Terminal capacity, how many platforms does HSR really need? With 3 trains per hour per platform, they could make do with 3 platforms even for a fairly high tph, which would leave an extra platform for Caltrain to use during peak hours, albeit at the cost of having no choice but to turn back trains at 4th/King if a train decides to die in the platform. Of course, this would imply that Caltrain and HSR could use the same platforms, but it would be pretty stupid to have it otherwise, especially at Transbay, and especially if reliability is a concern.

    ReplyDelete
  20. Adirondacker1280022 December, 2010 17:51

    26 trains per hour, unfortunately they canceled it.

    http://www.arctunnel.com/

    Scary electric trains going to high platforms that are the same height as the scary almost-high-speed trains running at the same time on the same tracks...

    ReplyDelete
  21. I don't know about capacity at Fremont, but the Chuo Line turns 28 tph at Tokyo Station (which has 450,000 boardings per day), and the 7 turns 24 tph at Times Square (which has 160,000).

    If you have concerns about the throat, then advocate fixing the throat instead of wasting money on tail tracks.

    ReplyDelete
  22. I agree with Clem’s outline on what makes a good station: “Operational flexibility” which might be defined as: How quickly can schedules recover when, not if, unforeseen delays scramble any carefully devised train choreography? Comparing maximum train throughput at a station vs. scheduled train frequency would be a good start for estimating a schedule recovery period.
    . Looking into the TBT from the throat entering and leaving trains can operate without mutual interference as long as arriving trains are filling TBT tracks to the right of leaving trains. After all trains to the left of the slots being filled have gone the flow of leaving trains must stop while the TBT track on the far left is being filled. Placing the far left terminal track cross-over switches as close as possible to the terminal platforms will minimize this crossover delay . The latest TBT throat layout does that. The inbound track crossover switch for the far-left terminal track is as close as possible to the terminal platform.
    Outbound delays due to inbound traffic can be eliminated if inbound traffic is held one train length short of the TBT entrance until the outbound train from the far right hand terminal track clears the right-hand in-bound track. Using the currently planned 3 track line to the TBT vicinity an intermittent in-bound train capacity on two tracks is likely to be at least equal to an un-interrupted single track out-bound capacity.
    If the following headway equation is true single track outbound train headways (H) can be held to less than a minute if their minimum speed is at least 21 mph for 1320 ft trains and converging outbound tracks are equipped with spring-frog switches that allow following trains to proceed without waiting for switch re-alignment before proceeding behind the leading train.
    If the train position detector, such as used in a “Moving Block System”, determines all train positions within a few feet the following minimum safe headway equation will be true:
    H = (L + s)/v + v/2b plus one second to allow for smooth braking where L = train length, s = minimum distance allowed between trains (50 m), and b = the maximum safety braking rate permitted (usually close to 2.2 feet per second squared).
    H = (L + s)/v + v/2b
    H = (1320 + 164)/30.8 + 30.8/2(2.2) = 64 trains per hour
    A 21 mph minimum speed (30.8 ft/sec) would allow 64 trains per hour.
    To determine the maximum line (no stations) through-put capacity speed compute the first derivative of the headway equation with respect to speed(v) and set the result equal to zero.
    H = (L + s)/v + v/2b
    dH/dv = − (L + s)/v2 + 1/2b = 0
    v = square root of 2b(L + s)
    v = square-root of 2(2.2)(1320 + 164) = 80.8 ft/sec = 55 mph
    H = 1484/80.8 + 80.8/4.4 + 1 = 37.7 seconds
    (3600 seconds/hour)/(37.7 seconds/train) = 95 trains/hour

    ReplyDelete
  23. 12 - 15 min headways on two lines are 8 - 10 trains terminating, but movements in and out count. No, not full, but more than enough time to clear. Time through the interlocking is what matters after a few minutes dwell on the platform.

    ReplyDelete
  24. Clem, I would write this directly to you, but don’t know your email address. Despite our fundamental differences in point of view, I do follow your blog faithfully.

    So, what I’m asking for is putting your attention to detail to work based on the hypothesis that the Caltrain corridor will not see any HSR construction for ten years.

    There certainly are enough indicators in the new Washington D.C. as well as the decision to put all their eggs in the Central Valley basket, that make that a plausible hypothesis. In any case, indulge me with an analysis of Caltrain’s options without HSR.

    1. Electrification
    2. Grade Separations
    3. Subsidy funding for operational costs
    4. Caltrain/JPB/SamTrans management

    There may be other issues that you consider significant. Without HSR, what could the next ten years look like?

    Martin

    ReplyDelete
  25. Fremont is not a good example.

    While 8tph is generally not a problem at other transit agencies, BART is just unable to do it at the Fremont terminal. 3-5 minute delays are common for inbound trains terminating at Fremont. I say that based on very extensive experience.

    There are various reasons for this, but one problem is that they split trains there.

    ReplyDelete
  26. Are they just planning the basement of the Terminal itself at this point, with no thought given to anything beyond that, like the connecting tunnels and station at 4th/King? Or what about a storage yard for Caltrain (or even HSR) on the existing property between 4th and 7th street? It seems like it ought to be possible to put in a storage yard for EMUs accessible from Transbay at the 7th street end where the old Caltrain maintenance facilities used to be, which would be useful for nighttime and off-peak train storage on the north end. It might even be possible to fit a couple of tracks capable of holding HSR trains for light overnight servicing.

    ReplyDelete
  27. Martin Engel:

    Re "the hypothesis that the Caltrain corridor will not see any HSR construction for ten years."

    That sort of thing is called "being in possession of a strategic plan" and is equivalent to "employing a staff with the collective intelligence of a slime mould.

    Unfortunately for us, Caltrain fails on both counts.

    In a better world, one in which public agencies were staffed by public service professionals possessing rudimentary levels of professionalism, competence and/or ethics, we'd have found that the plan for the Caltrain corridor without HSR is exactly the same as the plan with HSR ... the only differences would be the rate and scale of funding.

    The key is that the exact same projects that benefit regional (local, regional express) transportation service in the publicly-owned Caltrain corridor are those that set the stage for and can be used by the first stages of inter-regional high speed service.

    The highest priority, of course, would have been for a staff with skills above that of slime mould to have spent the last 15 years constructing stations suitable for level boarding instead of wasting hundreds of millions of your tax dollars not doing so.

    The second highest priority would have been to have been fully prepared when the inevitable federal ruling came down (I have in my mail archives memos to Caltrain staff saying just this, from nearly a decade ago), to be ready to implement a global standard Positive Train Control system and at the same time finally release transportation service on the public's right of way from the dead hand of Federal Railroad Administration 19th-century freight regulation. Instead, Caltrain less than slime mould intelligence staff have actively worked in favour of FRA regulation and of being run as a freight railroad (with all the massive unnecessary costs and backwards technology that entails), and have actively promoted a unique, insider-consultant-rewarding, guaranteed catastrophe home-grown PTC fiasco.

    The third highest priority would have been to have have a prioritised and step-by-step staged series of capital investment (ie track build-out) tranches ready to go, all of which are clearly in support of clearly defined incremental steps of increasing levels of rail service provided to the public.

    Well, you know where we are with that. (Instead we have the impossibly stupid San Bruno grade separation as the only capital priority of the agency! Death really is too kind a fate for any of those involved in either designing or promoting or approving this catastrophe.)

    Electrification comes in next, implemented in stages if necessary for funding or for transition reasons.

    Note that all of these priorities, each of which can be and should have been broken down into incrementally fundable but clearly progressive (not doing the wrong thing and then knocking it down and doing it over, Caltrain style) stages, would all be directly useful for HSR at each stage, and would benefit both Caltrain riders and Caltrain neighbours and HSR riders at each stage. There's no conflict at all, if you have are in possession of a strategy. If money comes in slowly without the HSR tooth fairy sprinkling funding dust, then less stuff is done, but all the stuff that is done is useful. If an avalanche of cash descends, then there is an intelligent and well-defined plan for undertaking useful projects in an adult fashion.

    As for Caltrain staff's "plan", well, it was basically to do nothing, bend over, offer up the entire corridor to HSR, pray that money would come from somewhere or other, have no plans of any type for improving Caltrain service, and in fact to actively work for the worst possible technical and service delivery outcome.

    Happy New Year.

    And death to the Peninsula Rail Program.

    ReplyDelete
  28. "Looking into the TBT from the throat entering and leaving trains can operate without mutual interference as long as arriving trains are filling TBT tracks to the right of leaving trains. After all trains to the left of the slots being filled have gone the flow of leaving trains must stop while the TBT track on the far left is being filled."

    Of course, all HSR platforms are to the right of the two Caltrain platforms, so no arriving HSR would conflict with an exiting Caltrain.

    Its seems that maximum operational flexibility would require more egress than access capacity, which would call for two egress tracks and one (therefore central) access track, with crossovers from the wrong way egress track to the main egress track at two points, and then narrowing down to a primarily two track tunnel. If the fastest that the HSR system at San Jose can accept HSR services is five minute headways, and a pair of terminal platform is not going to have operational flexibility trying to turn more than eight tph, then two minute tunnel headways would allow 30 slots per hour, with at most 20 that can be scheduled to be filled so 1 open slot each way for each two filled slots.

    It still seems that with the bi-directional express track at 4th and Townsend, there's no benefit to running three tracks all the way through the tunnel ~ that seems to force the egressing trains to run HSR / Caltrain when the accessing HSR possesses the Express track, and the accessing trains to run HSR / Caltrain when the egressing HSR possesses the Express track. The biggest increase in flexibility would be dedicated access and egress express tracks at 4th and Townsend, switching with the inbound and outbound platform tracks resp. into a two track tunnel.

    ReplyDelete
  29. `There are several reasons why inbound capacity to the TBT should be enhanced with two inbound tracks in order to balance the train per hour capacity of a single outbound track with spring frog equipped converging switches that are reset by the flanges of approaching trains. Therefore outbound trains do not need to wait for converging switches to be reset before following an earlier train. (1) On the other hand a series of inbound trains must separate with diverging switches that must be reset for every route diverging from the leader’s route. The following train must delay its diverging switch approach until the system is assured the leading train has completely cleared the diverging switch and that switch is subsequently thrown. (2) Any entering train should approach their TBT birth at a braking rate significantly below leaving train acceleration rates in order to avoid slamming into the end-of-track barrier if there is oil on the track or brakes on some cars are less effective than usual. (3) At the last minute a train scheduled to leave their TBT track may be delayed preventing an inbound train from using that slot. If two inbound tracks are available following trains can go around the now stopped train without its scheduled TBT slot. (4) And finally spare trains could be close-at-hand on the second inbound track in order to substitute for late inbound trains scheduled for outbound runs.

    ReplyDelete
  30. The direction in which a train travels through a switch is more or less irrelevant if we are dealing with low speeds and there is some nonzero time interval between trains passing over the same switch.

    ReplyDelete
  31. @John Bacon
    Every switch will almost certainly be used in both directions, and spring switches and the like are not designed for high speeds and tend to not be very reliable. A spring switch on a siding on a line used by 10 trains a day is rather different from one in the throat of a station used by 20-30 train movements per hour. And modern switches take about 3-5 seconds to throw and detect. This isn't a metro system, you're not terminating 40 trains per hour, and the throughput is ultimately going to be limited not by the layout of the throat (assuming it's something reasonable, of course) but by the dwell time at the limited number of platforms.

    But having three tracks for at least a train length or so on the approach to the terminal does sound like a good idea, for pretty much the reasons you mention. And if a train is slightly late leaving, you can hold two inbound trains on the two inbound tracks, and once the track is clear, they can move to their platforms in parallel, thus reducing the potential to further delay trains behind them. The existing terminal at 4th/King has this sort of layout now, and it seems to work pretty well.

    I'm still interested in what they're going to do for storage yards though: it would be good to have a location fairly near the terminal to store Caltrains and maybe even HSR trains outside the peaks, and the existing yard is a pretty good location for it.

    ReplyDelete
  32. There's no need at all for barriers right at the end of the tracks. If there's literally no room for 30 meters of tail tracks, then they should run 15-car trains instead of 16-car trains, and could still come out ahead on capacity.

    ReplyDelete
  33. From Arcady: “I'm still interested in what they're going to do for storage yards though: it would be good to have a location fairly near the terminal to store Caltrains and maybe even HSR trains outside the peaks, and the existing yard is a pretty good location for it.”
    I essentially agree with Arcady’s comment and would extend his excellent idea to store rolling stock close to where many train runs will be initiated and terminated. Modern Control System Theory effectively says a quickly applied countervailing force to a system disturbance will tend to efficiently maintain system stability. Using this idea for the subject at hand a substitute train showing up from nearby storage less than 5 minutes after a scheduled outbound train fails to materialize or function will mean on-time scheduled outbound service can be maintained in spite of rolling stock failures. Dedication of the present Caltrain right of way to EMU storage and light maintenance infrastructure between 4th & King and 18th street would be the best use of this partly below a freeway of little alternative value real-estate. The tracks are already present and this location is less than two rail miles from the largest rail transit destination in the San Francisco Bay Area.
    The currently planned two mile 35 mph on the curves approach to the TBT will slow rail service at a point that will annoy the most passengers possible where the trains will be at their peak loading. The rebuilt Shoreline Route should take a short-cut across the China Basin Development between the present Caltrain 18th Street crossing and an elevated station next to the baseball park immediately north of the 3rd Street Bridge across Mission Creek. This shortcut would greatly ease three sharp curves and would make a set of 3rd & King Elevated Station walkways directly from the baseball stadium seats feasible. Quickly dispersing crowds immediately after a game is the type of service where rail transit technology can really shine. This approach should give a favorable impression that would generate new riders from a large population many of whom rarely use rail transit. An added advantage would be to increase the available near terminal train storage capacity by using the current mainline tracks between 18th Street and 4th & King for MU parking.
    The most important constraint on rail transit operations is the need for a cautious design for avoiding derailment or collision. In the situation where one train is following another across a diverging switch which cannot begin to reset until the entire leading train has gone beyond any moving switch part. The following train must be able to stop short of the diverging switch at a braking rate one-half the normal service braking rate until the switch reset has been completed. Contrast that with outbound trains converging on spring operated switches. In order to avoid excessive ware on flanges and rails spring switches can and should be reset. But even if a reset does not take place or is only half completed a converging spring switch will admit the following train safely. Therefore converging trains are not delayed waiting for switch reset completion.
    A survey of scheduled stub end station approach and exit periods to and from the nearest station on the line showed the following pattern: All entrance times exceeded exit periods except in one case where they were equal. The following schedule differences between terminal entrance and exit periods were observed: Caltrain BB express to and from 22nd street, 2 minutes, Penn. Station New York to Secaucus Junction 5 minutes, Chicago Metra Diesel Service 1 to 2 minutes. For the Chicago Metra Electric there is no difference between arriving and leaving periods to and from the downtown terminal. In the CME case there are two tracks inbound to the terminal and one track outbound.

    ReplyDelete
  34. @John: you say,

    A survey of scheduled stub end station approach and exit periods to and from the nearest station on the line showed the following pattern: All entrance times exceeded exit periods except in one case where they were equal.

    Could this be schedule padding in front of the last station on the run?

    ReplyDelete
  35. Adirondacker1280029 December, 2010 19:09

    best use of this partly below a freeway of little alternative value real-estate.

    Once they tear it down to get rid of the forest of support columns under it.

    ReplyDelete
  36. "(3) At the last minute a train scheduled to leave their TBT track may be delayed preventing an inbound train from using that slot. If two inbound tracks are available following trains can go around the now stopped train without its scheduled TBT slot."

    Decoding this, a train is held at the platform at the last minute. There is no other platform clear for the inbound train to arrive at. There is a platform clear for the following train to arrive at, so it goes around.

    OK, that's a following Caltrain train. Since it can go around, there is one Caltrain platform clear. Hold the other Caltrain train, switch the inbound Caltrain onto the normal Caltrain egress track, when its cleared the platform track switch, the other Caltrain is free to egress.

    No, wait, that's a following HSR train, and there's an open HSR platform. Hold HSR trains from departing, and bring the HSR behind on the normal HSR access track. When its cleared the platform track switch, the HSR are free to egress.

    Whichever has been reversed, the other remained in normal use, and with more egress than access capacity, its possible to catch up for the holds during the reversal.

    On the speed, due to the single bi-directional HSR passing track at 4th and Townsend, an HSR ought to be normally leading out a Caltrain, and since the curve radius is tight for the HSR, it won't be able to accelerate out of the station as fast as the Caltrain behind it could have done. So that sounds more of a factor if there are access and egress Express tracks at 4th and Townsend so that Caltrains can lead out through the tunnel.

    ReplyDelete
  37. Adirondacker1280031 December, 2010 17:14

    Decoding this, a train is held at the platform at the last minute. There is no other platform clear for the inbound train to arrive at. There is a platform clear for the following train to arrive at, so it goes around.

    Goes around what? Out in the real world where trains arrive and depart more than once a day the arriving train goes to the next available platform in the vain hope that the delayed departing train will have departed with such a small delay that the second arriving train hasn't gotten there yet by the time the delayed departing trains....departs. Otherwise the second arriving train is the one that is delayed. It's why railroads have arrival and departure boards, dispatchers and if you are lucky someone announcing that the 5:47 to Willoughby is departing on track 106....

    ReplyDelete
  38. For stub-end terminals, such as the TBT, maximum through-put combined with maximum dwell periods for every train occurs when, looking into the terminal entrance, entering trains are filling slots just vacated by leaving trains in order from right to left. Because of terminal schedule recovery techniques departing trains are far more able to achieve this optimal ordering for maximum throughput than arriving trains with a limited ability to reorder their approach sequence. Note: Substituting nearby stored trains for outbound rolling stock instead of waiting for scheduled but late arriving trains is a particularly effective terminal schedule recovery approach.
    The minimum stub-end terminal dwell period: The February 2001, May 2001, and January 2003 BART schedules all show that all Fremont weekday runs reverse direction at Daly City one minute after arrival. Considering likely passenger alighting and boarding periods The Federal Transit Administration’s TCRP Report 13 on Rail Transit Capacity on page 42 Figure 4.7 shows train door mixed flow level boarding and alighting average time per passenger stream as equal to 2.5 seconds. Thus for 150 passengers exiting a car and the same number entering through two double stream doors might average 188 seconds. Dwell time data from the same chapter shows that passenger flow periods and waiting doors to close without passenger flow periods varied by a factor of 2 to3. Passenger flow periods varied from 38% to 64% of total dwell. Given that Caltrain has a bi-directional peak Caltrain should allow at least six minutes dwell time at the TBT but not their requested 18 minutes.

    ReplyDelete
  39. Adirondacker1280002 January, 2011 17:48

    In their wildest dreams they will have 24 trains per hour. with 6 platforms that means a train has to depart a platform every 15 minutes.

    ReplyDelete
  40. @Adirondacker12800:

    If the HSR corridor has 5 minute headways as the design envelope, 24tph would be 12tph max at 4 HSR platforms, out of the 24tph total would be 12tph max at 2 Caltrain platforms, so departing every 10 minutes from the Caltrain platforms and every 20 minutes from the HSR platforms.

    The train behind going around the train ahead was John Bacon's scenario, which I'm trying to follow, not my scenario. I'm guessing it follows from the segregation between Caltrain and HSR platform heights.

    ReplyDelete
  41. This is all rather academic, since (a) HSR in the real world is unlikely to every operate 12 tph on the peninsula... more like half that; and (b) Caltrain won't realistically operate more than 6 tph, and shows little interest in terminating all those trains at Transbay (most would still divert to 4th & King). Figure about 1.5 tph per platform when all's said and done. Maximum.

    ReplyDelete
  42. Clem: I suspect in the real world, Caltrain would want to terminate most of their trains at 4th/King. It's HSRA that thinks that Caltrain should think that they have little interest in terminating their trains at Transbay, despite Transbay's obviously superior location for commuting. Also, I think even 6 tph is overkill, and not likely to happen, at least without people commuting from SF to Sacramento via Los Banos. For Caltrain, I think 6-8 tph is a reasonable upper bound for the near to mid term. As they switch to MUs, they'll have more flexibility to run longer peak trains, which is a cheaper way to add capacity than running more tph.
    And based on the data John Bacon posted, looks like it's probably reasonable to have at least 5 times per hour, even accounting for passengers getting on and off and recovery time. So, two platforms is about right for Caltrain, HSR's realistic traffic level of 4 tph needs two more platforms, and TBT capacity isn't really a problem, at least if you assume a reasonable track layout.

    ReplyDelete
  43. "This is all rather academic, since (a) HSR in the real world is unlikely to every operate 12 tph on the peninsula... more like half that; and (b) Caltrain won't realistically operate more than 6 tph, and shows little interest in terminating all those trains at Transbay (most would still divert to 4th & King). Figure about 1.5 tph per platform when all's said and done. Maximum."

    If 12tph is a design maximum its not an average hourly rate, and there won't be demand for six trains in the peak value half hour in the first five years, but this is permanent infrastructure, and its better if unnecessary bottlenecks can be avoided.

    Anything more than 6tph for Caltrain assumes more than just all-stations services going to the TBT, but assuming that commuter rail into the TBT will always be limited to Caltrain all-station services is a bottleneck you'd not design in if there was an alternative.

    ReplyDelete
  44. Adirondacker1280003 January, 2011 08:54

    emits better if unnecessary bottlenecks can be avoided.

    ...the "how do we get to Oakland and points beyond" bottleneck is going to be a doozy.

    ReplyDelete
  45. Yes, an excellent example of an unnecessary bottleneck, but not one that looks like it can be avoided with the design of the train box and tunnel track to platform track layout.

    ReplyDelete
  46. Here is one possible Trans bay Terminal throat design: The three TBT center platform track pairs should, as is now planned, turn left immediately after exiting the TBT platform area. Each track pair should have two crossovers immediately after leaving the platform section. In order to allow rail vehicles to smoothly traverse crossovers within a curve any super-elevated section should have the following property. The elevation of all crossover section top surfaces should allow a nearly horizontal straight edge placed perpendicular to each crossover section main line track pair to simultaneously touch the top of all track pair and crossover rails.

    Immediately beyond the throat curve’s two crossovers each track will have been turned into either an exclusively arriving track (A) or a departing track (D) only. Thus looking out of the TBT structure at the main-line tracks starting from left to right beyond the TBT entrance crossovers but still within the TBT entrance throat curve arriving trains use only main-line tracks A1, A3, and A5. Interleaved with these arrival tracks are departure tracks D2, D4, and D6. As soon as the departure tracks extend beyond the TBT entrance crossovers departure tracks D2 and D4 should, independent of their arrival counterparts, begin to increase altitude enough to bridge over the arriving track A3 and A5 levels to their right. Departure tracks D2 and D4 should become one departure track at a converging switch on a bridge over A5. After crossing A5 the combined D2 and D4 tracks will connect with departure track D6 immediately beyond the departure line’s A5 crossing.
    Assuming steady 24 mph arriving and departing speeds using generally accepted transit industry separation standards the minimum 1320 foot train close-up time (one train replaced by another on the same track at the first train’s position) should be under 2 minutes. At the same speed more than 60 trains per hour each way could traverse the TBT throat. This design’s reduction in mutual interference may permit a TBT connection to only two main line tracks to be sufficient.

    ReplyDelete
  47. John: you've found the optimal layout of the throat, but there's no need for all the cleverness, because you're going to hit other bottlenecks first. For one thing, you won't be able to get dwell time down to something like 4 minutes for Caltrain and especially for HSR trains. For another, how do you propose to push 60 tph through the approach? Even with four tracks all the way down the Peninsula to the TBT, that's 30 tph per track, when even 24 tph seems to be just on the edge of what is achievable for mainline rail on a single track.
    I think in the case of the TBT, what you probably want is the ability to make parallel moves from any pair of tracks, and a set of crossovers between the two main tracks some distance out from the terminal, to give a bit more opportunity for conflicting moves to be separated. And given realistic traffic levels, six platform tracks and two approach tracks should be enough.

    ReplyDelete
  48. On a parallel move for any pair of tracks, would you get that from an access track connected to all platforms, an egress track connected to all platforms, and a bi-directional central track connected to fix those pairs that are blocked in the simple fan-out?

    The bi-directional track could switch out before the second turn outbound from the TBT, leaving two tracks for most of the DTX.

    I just can't see digging a three track tunnel to gain no more system capacity than ought to be able to be handled by a two track tunnel.

    ReplyDelete
  49. I think a 3-track approach is good when you realize that by the time a train reaches its destination, it'll likely be +-5 off the schedule. Yes, 60tph is unrealistic, it might occur for two or three trains, so two inbound tracks will help prevent delays.

    Since departures can easily be scheduled on time, only 1 outbound track will suffice.

    Caltrain already pulls off that trick with trains 281 and 383. At Bayshore, 383 switches to the SB track before the tunnel and continues to SF (with a stop on 22nd street) on the SB track. This means that 383 doesn't have to slow down for 281 and 281 doesn't need to delay at bayshore to wait for 383 to pass. At times, both trains sometimes catch up and go side-by-side.

    You could imagine the same happening where 383=HSR and 281=Caltrain, and both trains going to Transbay.

    ReplyDelete
  50. A good portion of the world's HSR operators design trains to run on schedule. Five minutes off-schedule is acceptable to SNCF, maybe, but not to anyone else.

    ReplyDelete
  51. If the scheduling is really that tight at the terminal end and you're really that worried about trains being late, you can always put in a bit of padding into the schedule, with more padding during rush hour when the OTP is much more critical to keeping the terminal running smoothly. Except in this case you can't, because the end to end run time is dictated by law, and each minute of padding means more expense to save that minute of travel time somewhere else.

    ReplyDelete
  52. Adirondacker1280014 January, 2011 20:30

    Hold the late train at Bayshore and or pad the Caltrain locals more than the Caltrain expresses and the Caltrain Expresses more than the HSR trains. Padding the schedule a bit is a lot cheaper than a billion dollars or so of downtown tunnel.

    Railroads all over the world manage to get 20, 25 train an hour into and out of a terminal on two tracks reliably.

    ReplyDelete
  53. Dear everyone,

    The way you arrange to have maximum throughput through a short and massively expensive piece of underground track is to schedule (and dynamically reschedule, as necessary) the traffic entering it so that it flows uniformly and freely and without conflict.

    Trains don't enter unless they can progress without slowdown and exit without delay.

    It's really very very very simple and trivial and obvious and Not Rocket Science.

    The way this is done is to build a "buffering zone" of extra track outside the expensive constrained section, where (a) it is less expensive to do so and (b) it serves some purpose other than parking stopped or nearly-stopped trains.

    The way this can and should have (and theoretically still could be) done with Transbay/DTX is to build a station with three through platform tracks and an island platform at Mission Bay, in a less-expensive and massively more customer-friendly cut-and-mostly-uncovered trench. (More through platform tracks would be ideal, but somebody or other stole the right of way. Don't blame me: I was jumping up and down about the problem when it could have been avoided in 1995 and nobody cared.)

    With 3+ through tracks it is possible for trains entering the terminal area to dwell at an open air station platform (which is massively more passenger-friendly, as well as far safer than parking in s tunnel, as well as infinitely less expensive) until the approach route can be cleared for free running.

    It's possible to do limited reordering of trains (ie overtakes) in order to reduce non-parallel blocking moves at the critical terminalk throat.

    It's possible (and desirable) to allow in-service terminal outbound trains to use either track and still be able to stop at the Mission Bay station due to island platform. Without this configuration -- ie as "designed" by PTG/TJPA/PCJPB cretins with only an outside platform -- outbound trains must use only the one outbound platform track, since boarding passengers cannot quickly switch to another track if traffic flow demands that.

    In the outbound direction, of course, trains wait at any of the six completely flexible, identical-length, identical platform height platform tracks (unless you're a PTG/CHSRA/PCP/PCJPB/TJPA sub-human) and do not depart until a freely-flowing conflict-free route is available: hold trains full of passengers safely in the station, don't park them in a nasty tunnel!

    The same sort of argument of applies at Bayshore, where four tracks neck down to two for the approach in existing tunnels to SF (I've just saved you a couple billion dollars right there), except that the cost of building extra platforms for the fast tracks (which of course ARE ON THE OUTSIDE -- FSSF -- unless you're a cretin) isn't justified: slow (inside track) trains can wait to be overtaken at a two platform track, four through track, one island platform station, while any fast trains that need to be overtaken by a slow train (in exceptional circumstances) can do so at the Mission Bay station as described above.

    There's really nothing to it.

    All this nonsense about 3+ tracks in a station-free tunnel in the most expensive part of the entire line is limitlessly insane: if you think about it you will understand that the only way you can have 3 tracks occupied by trains is if at least one of them is stopped or close to stopped or if nobody anywhere is engaged in train dispatching. Building tunnels with tracks in order to, effectively, park trains is about the stupidest possible thing one could possibly do.

    ReplyDelete
  54. The way to arrange to have maximum throughput through a short and massively expensive piece of underground track is to schedule (and dynamically reschedule, as necessary) the traffic entering it so that it flows uniformly and freely and without conflict.

    WORK SMARTER, NOT HARDER.

    Trains don't enter unless they can progress without slowdown and exit without delay.

    It's really very very very simple and trivial and obvious and Not Rocket Science.

    The key is to have a "buffering zone" of extra track outside the expensive constrained section, where (a) it is less expensive to do so and (b) it serves some purpose other than parking stopped or nearly-stopped trains.

    The way this can and should have (and theoretically still could be) done with Transbay/DTX is to have a station with 3 through platform tracks and an island platform at Mission Bay, built in a less-expensive and massively more customer-friendly cut-and-mostly-uncovered trench. (More through platform tracks would be ideal, but somebody or other stole the right of way. Don't blame me: I was jumping up and down about the problem when it could have been avoided in 1995 and nobody cared.)

    With 3+ through tracks it is possible for trains entering the terminal area to dwell at an open air station platform (massively more passenger-friendly, far safer than parking in s tunnel, infinitely less expensive) until the approach route can be cleared for free running.

    It's possible to do some reordering of trains (ie overtakes) here to reduce non-parallel blocking moves at the critical terminal throat. Overtakes require miles less (hundreds of millions less) infrastructure at stations than on open track between stations.

    It's possible and desirable to allow in-service outbound trains to use either track and still be able to stop at the Mission Bay station due to island platform. Without this configuration -- ie as "designed" by PTG/TJPA/PCJPB cretins with only an outside platform -- outbound trains have to use only one outbound platform track. Insane!

    Naturally terminal-outbound trains wait at any of the six completely flexible, identical-length, identical platform height platform tracks (which any room temperature IQ planner would provide at the terminal) until a freely-flowing route is available: hold trains full of passengers safely in the station, don't park them in a nasty tunnel!

    The same sort of argument of works at Bayshore, where four tracks neck down to two for the approach in existing tunnels to SF (I've just saved you a couple billion dollars right there), except that the cost of building extra platforms for the fast tracks (which of course ARE ON THE OUTSIDE -- FSSF -- unless you're a cretin) isn't justified: slow trains can wait to be overtaken on either of the central island platform slow tracks, while any fast trains that need to be overtaken (in exceptional circumstances) can do so at the Mission Bay station as described above.

    There's really nothing to it.

    All this nonsense about 3+ tracks in a station-free tunnel in the most expensive part of the entire line is limitlessly insane: if you think about it you will see the only way you can have 3 tracks occupied is if at least one trsain is stopped or close to stopped, or if nobody anywhere is engaged in train dispatching. Digging tunnels in order to effectively park trains is about the stupidest possible thing one could possibly do.

    ReplyDelete
  55. You do have to admit that the New Austrian Public-Private Wealth Transfer Tunneling Method is pretty snazzy, though.

    ReplyDelete
  56. Arcady, it took you only 44 minutes to read, compose, and send an interesting response to respond to my comment. You would be the perfect individual to organize short Trans Bay Terminal turnarounds. As to whether a pair of main line tracks could have the capacity to accommodate 60 trains per hour each way there are precedents. William D. Middleton’s book Metropolitan Railways on page 38 says “each of the two tracks on the [Chicago] Loop were handling upwards of 60 to 65 trains an hour”. Levinson and Hoey’s paper from Reflections on Transit Capacity from Proceedings of the International Symposium on Highway Capacity, July 1991 “cites the historic high train throughput on the Chicago Loop with visual rules (70 trains per hour)...” The Federal Transit Administration TCRP Report No. 13 on page 149 quotes Professor Vukan R. Vuchic from his textbook Urban Public Transportation and Technology Prentice-Hall Inc. 1981: “Vuchic calculates the way capacity for BART at 185 trains per hour”.
    The quote that for a single track, “the way capacity for BART at 185 trains per hour” cannot be true. In order to find the speed that will produce the shortest headway compute the first derivative of the minimum headway formula and set the result equal to zero:
    H = (L + s)/v + v/2b
    dH/dv = − (L + s)/v(squared) + 1/2b = 0
    (L + s)/v2 = 1/2b, v = the square-root of 2b(L + s) =the square root of 2(2.97)(700 + 164) = 71.2 ft/sec = 48.5 mph
    Therefore: H = (700 + 164)/72 + 72/2(2.97) = 12.14 + 12.14 = 24.3 sec. Add 1 second in order to allow for a comfortable braking rate change and using current BART’s in tunnel braking rate safety standards the minimum possible headway of 25.3 seconds allows 142 trains per hour. With v = 80 mph (117.3 ft/sec) the same system would permit 127 tph.
    Way capacity for 1320 foot trains v = 125 mph (183.3 ft/sec):
    C = 3600/{(1320 + 164)/183.3 + 183.3/2(2.2) + 1} = 71 trains per hour
    Note that the safety braking rate used in the last equation was only 2,2 ft/second squared. Track running surfaces not in a tunnel are open to contamination from leaves and rain. Wheel-track adhesion can be sharply reduced under these conditions. This slippery condition should be compensated for by requiring following trains enough space behind their leader to safely stop at a lower braking rate than the safety braking margin required for tunnel track not subject leaf and rain contamination.
    An interesting relation between headway and train throughput: If BART reduced their present 150 second minimum headway by 6 seconds they could add one train per hour to their schedule. If BART were operating a 33 second headway through their Trans-Bay Tunnel and then they reduced their allowed headway by 6 seconds to 27 seconds (by adding track brakes or lowering speed for example) they could add 24 trains to their schedule.
    Note: C = 3600/H, The slope of the capacity curve is: dC/dH = − 3600/H(squared)

    ReplyDelete
  57. The next question: What is the minimum number of trains per hour capacity for traversing the TBT throat needed to produce a cost-effective benefit? This discussion group agreed that the German Federal Railway reversed High Speed trains quickly at stub end terminals. It is also true only that system is where a HSR accident caused a serious loss of life. A wheel tread failed on a car that had been repeatedly written up for excessive noise for several weeks before the accident. Could a series of dispatchers have brushed aside possible serious defects when given praise and bonuses for quick train turnaround times? Both American Space Shuttle disasters definitely occurred after program managers elected to proceed in spite of well-reasoned arguments presented stating that given current conditions risks were exceptionally high. HSR trains could also potentially crash with catastrophic results. The option for a thorough checkout and necessary repairs should be available at any major terminal without significantly disrupting service. One can no doubt run safely without frequent maintenance if when in the shop any part with the slightest wear is replaced at high cost. Therefore for safety at a reasonable cost the number of HSR movements in and out of the TBT to an auxiliary maintenance facility would increase TBT CHSR traffic above in-service movements with doubling as an upper limit.
    Way capacity in excess of required to accommodate any currently envisioned peak scheduled service can aid recovery when delays occur. As I have implied in my Dec. 22nd comment 400m train way capacity declines sharply at speeds below 55 mph. At least two of those three 35 mph curves the last two miles to the TBT have got to be straightened. Track capacity like money should never be squandered needlessly even if you have a great deal of both. A Moving Block Signal System rather than now ubiquitous 3 or 4 aspect block signals would permit closer headways. One estimate for the New York Subways indicates their passenger carrying capacity could increase 15% with such a communication based train separation system. Their station close-up times would drop to 34 seconds instead of their current 55 seconds for 600ft trains. Building parallel tunnels in order to reduce congestion would cost 50 times more.
    The construction of an entirely grade separated high speed railway presents an exceptional opportunity to dramatically improve Caltrain service while simultaneously reducing operating costs. Automatic operation where no paid employee need ride every train would allow an economical increase in train frequency where the total weight of trains in motion would be the most significant variable cost driver. If single 86,000lb EMU cars were used the total mass of 6 single car locals and 12 single car expresses per hour would be 10% less than present diesel trains on a 30 minute headway. If those EMU’s could accelerate to 125 mph and sustain that speed between stations 16.3 miles or more apart they could attain a 100mph+ average speed. A high quality service regional service that now has 40,000 riders per day seems likely to attract a significant portion of the 650,000 people per day now using the I280 and 101 freeways between San Jose and San Francisco.
    There are other co-lateral benefits for automatic operation. Platform edge barriers required for automatic operation would allow nearly the full width of center island station platforms to be usable for waiting passengers and therefore could be exceptionally narrow. The main line track center space at or between stations could remain constant allowing station location, distance between, and length to be changed in the future without shifting main-line tracks or increasing adjacent property encroachment. Between stations spare rolling stock on a center track could serve as gap trains ready to move less than a second after their need becomes apparent.

    ReplyDelete
  58. John: I don't know if you know this, but right now the main bottleneck in the BART system is... the doors of the train at Embarcadero. The most heavily loaded segment is the Transbay one, and the limiting factor there is the maximum dwell time at the stations on Market Street, so even if you have trains with infinite acceleration and zero variance in dwell time, you're not going to get above 90 tph until BART buys train cars with three doors per side instead of two. Key lesson: look at the system as a whole, because it's only going to perform as well as the bottleneck, and that bottleneck can be anywhere.

    Looking at how to make Caltrain more attractive, you have to remember that people aren't travelling from the Mountain View Palo Alto station to the 4th and King station or Transbay Terminal: they're going from home to work, or to some entertainment venue in the city, or any number of other journeys that have segments beyond the Caltrain line. And what these people care about is end-to-end journey time, as well as the reliability of that time. One huge disadvantage of the current Caltrain setup is that the San Francisco station is not within walking distance of the massive concentration of employment in the Financial District, and does not provide easy transfers to most transit connecting to the rest of the city and for that matter to other regions of the Bay Area. Extending the line to Transbay would remove one transfer from most people's rides, which could easily make Caltrain some 15 minutes faster for an end-to-end trip, and help it make considerably more competitive against driving. It would also make for easier connections to the East Bay and North Bay and to Amtrak's shuttle bus to Emeryville.

    ReplyDelete
  59. > right now the main bottleneck in the BART system is... the doors of the train at Embarcadero

    This can be dealt with in the short term by having some trains skip Embarcadero and make their first stop at Montgomery.

    In addition to the lack of train doors, Embarcadero is limited by infrastructure - more escalators and stairs are needed to get people off the platform.

    A more expensive solution would be to extend Montgomery platform east to the crossover, and have a double length platform. Or build double platforms at Embarcadero, so that passengers board and exit both sides of the train.

    ReplyDelete
  60. Fremont is not a good example.

    > While 8tph is generally not a problem at other transit agencies, BART is just unable to do it at the Fremont terminal. 3-5 minute delays are common for inbound trains terminating at Fremont. I say that based on very extensive experience.

    > There are various reasons for this, but one problem is that they split trains there.

    This is more of a scheduling problem. If the Daly City to Fremont trains left Daly City 2 minutes earlier, most of the congestion at Fremont goes away. Right now, the schedule conflict happens because both (Orange and Green) trains share a pair of arrival/departure times - 01 and 07. One arrives when the other departs, then the other departs when one arrives. So, lots of crossovers every hour.

    It is common to have police holds and also common to have minor delays at MacArthur, so by the time you get to Fremont, a few minutes are gone.

    The make and break pieces (the non-revenue part that goes between Fremont and the yard) just waits until there is a time slot and does not get in the way (in theory).

    Eventually (it's in the annual reports) BART will turn back Fremont to DC trains at 24th Street. That will allow some fine tuning of the congestion at Fremont. Not to mention freeing up ten badly needed cars.

    Speaking of freeing up cars, by speeding up all of the Richmond to Fremont tracks (except MacArthur to Lake Merritt) to 70 mph, would shorten the run time to 45 minutes, and free up another 12 cars.

    ReplyDelete
  61. This can be dealt with in the short term by having some trains skip Embarcadero and make their first stop at Montgomery.

    Okay, firstly, Embarcadero has the single highest ridership in the entire BART system. Sure, it needs better platform access (many BART stations do), but skipping it is not really an effective strategy as far as riders are concerned. Secondly, that only works if there is a slot open ahead of the train in question, which there usually isn't.

    ReplyDelete
  62. If you skip Embarcadero, you get double the load (and thus double the dwell time) at Montgomery, making that the new bottleneck. How does that help, unless the bottleneck is actually at the escalators at Embarcadero? My point is that in a passenger rail system, it's not unusual to have the limiting factor for capacity be the dwell time, specifically the maximum dwell time on the busiest part of the line. Dwell time tends to dominate other factors like terminal layout, at least assuming that the terminal is designed and scheduled properly (not like Fremont). On "commuter rail" systems, it's typical for the longest dwell time to be at the downtown terminal, because that's where a majority of riders are headed to or from, and so depending on the layout and the number of platforms, the dwell time can be more of a constraint than the conflicts in the station throat, and that it therefore doesn't make sense to have a very fancy and expensive layout there, if you're limited by the dwell time at the platforms anyway.

    ReplyDelete
  63. Arcady: You are referring to BART station dwells at their busiest station being their most significant bottleneck. Here is some data from TCRP Report # 13 Figure 4.2 BART Montgomery Station dwell time components p.m. peak February 9, 1995. The following averages were recorded for 33 trains: Operating on a 153 second headway the passenger flow time was 15.8 seconds with a 6.6 second standard deviation which is 39% of the 40.4 second dwell time with a 7.7 second standard deviation. Therefore with three doors instead we save about 6 seconds. Aside from the 5 seconds required to open and close the train doors nearly half the dwell time is consumed by last minute arrivals blocking door closure or operator uncertainty as to whether boarding has completed for all twenty doors. There is a $1.5 billion incentive to find another approach for reducing station dwells by 6 seconds rather than scrapping 669 cars in order to add a third door. The third door per side approach would also reduce available seating area and add to outside noise and drafts admitted through the doors for each car.
    One way to reduce dwells at the busiest stations would be to erect something similar to the fare gate barrier 5 feet from and parallel to the platform edge. The barrier’s turn-styles would restrain boarding passengers until alighting passengers have had a chance to proceed and cut-off late arrival entries. Station information panels above every train doorway could clearly describe where the currently stopped train is going. The station boarding system computer network could determine the loading state of an arriving train from pressure readings from each car’s air suspension system. If the passenger information system stated clearly (using the same destination indication station panels above each train doorway) that the platform edge barriers would block entry sooner opposite the most heavily loaded train cars than sections opposite nearly empty train cars passengers would have an incentive to move to sections where rapid boarding is most likely. Note: The blocking aspect of this system should be active only when the arriving train has an immediate follower. Passenger safety, convenience, and comfort should always remain the first priority in that order.

    ReplyDelete
  64. But there is a far more effective approach to increasing BART capacity. The same TCRP report and my own observations confirm that BART’s train separation control forces a 90 second minimum station close-up time. All other North American close-up times for similar operations range from 53 to 60 seconds as of 1996 except for Vancouver’s Sky-Train. Strictly adhering to BART’s current safety standards a moving block system would allow a 36 to 44 second close-up time with their current equipment. (The uncertainty stems from the closest an in service train controlled by a moving block system is allowed to approach its leader. This minimum distance could range from 50m to 200m.) If magnetic track brakes were installed enabling a firm minimum braking rate then BART’s safety braking rate could safely be raised to the maximum in-service braking rate of 4.4 feet/second squared and the minimum distance between train ends could fall to 50m. These modified in-service separation rules would make a 33 second close-up performance level possible for BART’s 700 foot trains. (All these close-up time estimates include 2 seconds of additional time in order to allow for a gradual rate of acceleration increase, a smooth transition from acceleration to braking and a gradual braking rate reduction as the following train approaches a complete stop.) A collateral benefit from installing track brakes is the possibility of instantly locking rolling stock wheels to the track during an earth-quake even while the train is in motion. (Of course pinning the sleepers to the ground is necessary in order to assure that the track structure remains flat and therefore train cars have a better chance of remaining up-right during a severe earth-quake.)
    The MTC 2002 Bay Crossing Study claims that in order for BART to meet their trans-bay demand by 2035 they would need a 45 train per hour each direction trans-bay capacity. In order for BART to meet this future demand their report suggests a second trans-bay pair of tubes plus another Downtown San Francisco Subway parallel to their current set at a cost of $7 billion to $10 billion in 2002 dollars. If the capacity enhancement approach outlined above allowed an 80 second headway then 45 trains per hour capacity would be produced without any additional track construction in Downtown San Francisco or under the Bay or the need to scrap 669 cars in order to add a third door per side for all BART cars.

    ReplyDelete
  65. Adirondacker1280024 January, 2011 22:11

    John, come up from inhaling your formulae for a moment. Trains have been running on track with more than one train on them for over 175 years. Some of them have been electrified for over 100. Operators all over the world struggle to get to 25 trains per hour on a track. In a few instances they manage 30 and in one or two cases 40. You aren't going to be able to maintain safe stopping distances and wedge 40 trains an hour into the Tranbay Tubes. Not if you want people to get between San Francisco and Oakland in a reasonable time.

    ReplyDelete
  66. Since the ultimate subject of discussion is CAHSR and not BART, which is just a comparison, let me point out that the presumed train control system, ETCS, is capable of 32 tph on a running line.

    ReplyDelete
  67. The other thing is that BART has already committed to scrapping their entire fleet in favor of new trains with more doors. I also suspect that data from 1995 is a bit out of date and BART ridership has increased somewhat since then. And for the uncertainty of when to close the doors, the solution is simply doors that close faster: that reduces the window during which a passenger can realize "oh no, the door is closing!", and then proceed to run to the door and hold it open. Ideally you'd close the train doors once there is nobody close enough to the train to get to a door before it finishes closing, and the faster the doors close, the smaller this distance is. I also wonder what effect the practice of lining up in front of the doors has on all this.

    Adirondacker: didn't the East Side IRT run 35 tph on the express back in the 1950s? 30 tph certainly seems to be no problem for the NYC Subway even today, and 39 tph is doable in Moscow. It doesn't even matter whether you use fixed-block or moving-block, just that your signal system is optimized for the way your trains actually run.

    ReplyDelete