08 June 2010

Bi-Level or Bust

A while back, Bob Doty (Director of the Peninsula Rail Program) was asked about the incompatibility of platform heights between HSR and Caltrain. His response:
The primary issue of non-compatible equipment between Caltrain and high-speed trains is that they will not be able to share the same platform edge if level-boarding access is to be provided.
Level-boarding access refers to the platform and train floor both being the same height--similar to boarding an elevator--and is mandated by the Americans with Disabilities Act. For an introduction to the subject, covering the various regulatory and engineering constraints, refer to an earlier discussion of platform height. In short, accessible boarding will have to be provided at every train door.

Doty's comment is predicated on two important but unstated assumptions: (1) Caltrain must operate bi-level train cars, and (2) HSR high platforms are not compatible with bi-level train cars. Let's examine those two assumptions.

Must Caltrain operate bi-level train cars?

Since the late 1950s, the peninsula commute has traditionally used railroad cars that have two levels, upstairs and downstairs. Bi-level cars are such a familiar fixture on the peninsula that one might forget to even question how or why they came to be that way. Even future Caltrain vehicles are systematically depicted as bi-levels. The reasons to use bi-level cars include:
  • Bi-levels maximize passenger count per train, to provide a given transportation capacity using the fewest trains and crews.
  • Bi-levels maximize linear density, or how many passengers can sit within a given train length. This can be useful when platform length can't be extended, such as in Menlo Park or Burlingame, where the platforms are boxed in by grade crossings.
  • Bi-levels provide a seat for every passenger, with no standees. The average Caltrain trip length is 23 miles or roughly half an hour, longer than most people would be willing to stand. Because of this, by the way, Caltrain's capacity is defined not in terms of passengers per hour, but seats per hour.
  • Bi-levels minimize the mass per seat of the train to reduce energy consumption under intensive start-and-stop usage. (Note, mass per seat can be a misleading metric if a train is designed to provide significant amounts of standing room.)
  • Bi-levels take advantage of the generous vertical clearances available on the peninsula corridor.
  • Bi-levels have been used on the peninsula as far back as most people can remember.
  • Bi-levels differentiate Caltrain's brand image from BART.
Now consider these counterpoints:
  • Providing a seat for every passenger results in Caltrain operating at an average capacity of just 39 percent. The numbers don't lie, Caltrain actually hauls around significantly more empty seats than passengers! A few peak-hour express trains do operate near 100% of seating capacity, but conversely, many off-peak trains operate nearly empty. Specifying a new vehicle fleet in terms of peak seating capacity (as opposed to passenger capacity) will perpetuate this extremely low and wasteful average load factor.
  • Shorter trip times, thanks to fast-accelerating, short-dwell EMUs, may reduce the need for seating if there's a comfortable place to stand. Most of Caltrain's current fleet doesn't have anywhere to stand at all; if you do, you will quickly be shooed to a seat by a conductor. The need for a seat for every passenger is a self-fulfilling truth: everyone needs a seat because there's nowhere to stand!
  • Speeding up service, and especially the turnaround times at each end of the line, allows more trains to be operated using the same number of vehicles and crews. More frequent trains can meet a given level of passenger demand using fewer seats per train.
  • 100% grade separation for HSR means that stations will no longer be boxed in by grade crossings and the length of platforms can be increased (within reason, of course.) Extending platforms across grade-separated streets has the additional benefit of providing direct platform access from both sidewalks of those streets, making stations more accessible to pedestrians.
  • Caltrain plans to increase peak rush-hour track capacity from today's 5 trains per hour per direction; a conceptual schedule shows as many as 10 trains per hour per direction. Once again, moving more trains per hour can meet a given level of passenger demand with smaller-capacity trains.
Perhaps the need for double-deckers isn't as pressing as it once was.

One could argue that growing ridership would eventually require bi-level cars anyway--similar to this gargantuan commuter train in New Jersey--an argument that isn't entirely without merit, since rail vehicles typically last for one-third of a century. But this argument would need to be based on a serious, quantitative ridership study.

So, must Caltrain use bi-levels? The answer is not as obvious as one might first expect.

It's even less obvious that Caltrain should provide an actual seat for every customer. Providing convenient and comfortable places to stand, and a more flexible floor plan for standees and bicycles, would increase the average load factor and make more efficient and profitable use of the new vehicle fleet. Rush-hour crush loads could still be met by increasing train length (something that is inherently easy to do with EMUs) and by increasing train frequencies.

Are high platforms compatible with bi-level train cars?

The most important implication of this question is how to pull off the transition from the existing low platforms to the high platforms typically used for high-speed rail, without interrupting Caltrain service during construction. One possible solution to this transition conundrum has already been described.

Assuming such a transition were feasible, would it preclude the use of bi-level train cars?

The quick answer is no. You don't have to go further than Chicago or the Northeast Corridor to see plenty of bi-level trains that board using high platforms--to say nothing about numerous examples abroad. The slightly longer answer is that high platforms don't jive with Caltrain's plans to acquire European bi-level EMUs. The European products that Caltrain is evaluating (bi-level EMUs from Bombardier, Siemens or Alstom... and by now, hopefully also Stadler, which just recently entered this market) all feature doors on the lower level. Low-floor boarding has undeniable advantages:
  • It's what the builders already provide off-the-shelf.
  • It allows doors to be optimally spaced out along the length of the train, providing good passenger circulation and reducing station dwell times.
  • It makes for a much easier transition from today's 8-inch platforms to the ADA-mandated level-boarding without interruption in Caltrain service.
  • It allows Caltrain to retain its relatively young Bombardier sub-fleet, if the platform height is established at 25 inches (the floor height of a Bombardier car). One can of course question the wisdom of making a fundamental corridor architecture decision, such as platform height, on the basis of a sub-fleet of 25 standard-issue commuter cars that would fetch excellent prices on the second-hand market.
All that being said, European bi-level 25 kV EMUs with high-level boarding do actually exist, although not in a form that would be directly applicable here on the peninsula. Without belittling the intrinsic packaging complexity of such vehicles, it is conceivable that one could be procured from within the broad product families of the major manufacturers, which typically feature a modular approach that provides some flexibility in car shell design.

High-level boarding would definitely introduce a big wrinkle in Caltrain's existing plans, but those plans ought to be carefully re-examined in the context of a shared corridor with HSR.

Incompatible Caltrain and HSR platforms would have several crippling drawbacks for the peninsula corridor. The operational benefits of "any train, any track, any platform" are too great to ignore--perhaps even great enough to smash the bi-level paradigm and the low-boarding paradigm. Beware of unvalidated assumptions!


  1. How do the ADA requirements affect the type of bi-level cars that Caltrain can use? I.e., high-boarding-level bi-level train cars seem to have both steps up and steps down from boarding-level to the seating-levels. I presume a train with just this type of car would not pass muster with the ADA?

    How about if a train had mixed single-level and bi-level cars? They run those all the time on commuter lines in Japan. Generally, the single-level cars have sideways facing seats and lots of standing room, plus some open spaces for someone with a wheelchair. The bi-level cars are called "green cars" and are considered a "premium" ride. You can upgrade your ticket and grab one of these seats for a more comfortable ride if you're going to be traveling a long distance.

    If you're interested, here's a cool video of how you can use the "Suica" pre-paid fare card in Japan to grab one of these premium seats.


    FWIW, I agree that the idea of "everyone needs a seat" is an anachronism and should be retired, along with Caltrain's ancient trains! And yeah, duh, same platform height as HSR should be a no-brainer.

  2. Note that European low-floor EMUs feature 550mm ATOR boarding with 4650mm high cars (EBO G2 or UIC-GC loading gauges), so doorstep-to-roof height is 4100 mm. AAR Plate F allows 5181 mm high cars, that in turn allow 1081 mm high platforms. With some quirks, like raising 2nd floor level few inches directly above door, it's feasible to build cars:

    - with high-platform level boarding in NEC/Shinkansen range of 1220-1250 mm ATOR

    - with both levels spanning whole length of the car, thus maximizing floor area and seat count

    - with enough space for traction and A/C equipment underfloor, so it doesn't take space for passengers (unlike Flirt, hello Richard ;))

    Of course, single level train with 3+2 seating provides the same seat count as traditional bilevels, so it's the obvious way to go. The wides cars used in US today are 3200 mm wide, enough for 3+2, but that's quite close to comfort limit width-wise. For the record, Shinkansens are 3400 mm wide, Reginas 3450 mm and DOD outline for tanks with side armor is 3650 mm. (It's no problem to use wider platforms than cars, most European level boarding designs have to use such bridges zillion times a day due to outdated regulations governing structure clearance and vehicle width.)

  3. Dan S:

    Many, if not most bilevel cars have single level seating areas between the door and the end of the car, which are at the level of the high platforms. Wheelchairs would use this space.

  4. Good write-up, Clem. However, you're missing three arguments against bilevels:

    1. They have higher axle load than single levels.

    2. There are somewhat fewer off-the-shelf products.

    3. While the average trip on Caltrain may be half an hour long, many trips are much shorter, allowing people to stand. On commuter trains, the people whose trips are the longest typically board early and get seats; it's the short-range commuters who have to stand. In addition, HSR is going to poach some SF-SJ ridership, shortening the average Caltrain trip.

  5. IMHO, counting on standees to increase ridership on a regional train isn't a winning strategy.

    That said, your point regarding empty off-peak seats is valid. The easiest way to deal with this is to run trains less frequently off-peak in the hope of aggregating demand over a still-acceptable time interval. For example, 7.5min headway during peak, 15min off-peak.

    Another strategy is to vary train length during the day, preferably using automatic Scharfenberg couplers. If the basic EMU unit is a 100m 4-car trainset, two or even three could be strung together during rush hour to boost line throughput and slash operating overheads in terms of passengers/hour. The downside is that in addition to long platforms, you need adequate pedestrian flow capacity through the station and sufficient connecting transportation capacity. A full 12-car train would swamp SF Muni at 4th & King, forcing passengers to incur additional hassle or wait times. At the TBT, different story.

    A third strategy would be to eliminate cars for full-size bicycles in favor of a discount program for pre-qualified folding models available to passengers who purchase an annual pass. In addition, install expanded bicycle parking (bi-level, please!) at the stations with the highest demand for cycling as the connecting mode of transportation. Once Caltrain gets to use the TBT, demand for space to take full-size cycles along will anyhow fall.

    None of these options depends on bi-level cars, which tend to have lower pedestrian flow capacity within the vehicle and therefore require longer dwell times. It would be possible to compensate for that by implement one-way traffic along the aisles: designate half the doors as entrances and the other half as exits and, marking platforms to encourage passengers waiting to board to queue up. Works fine in Japan, but passengers would have to get used to that in California.

    Final note: the Talgo 22 (sorry, brochure is in German) is a lightweight bi-level EMU design with short, wide cars featuring 3+2 seating and walk-through decks on both levels. I believe the platform height is the lower of the UIC standards, 550mm. Thanks to the generous loading US gauge, a modified design with even wider cars would be possible, enabling 3+3 seating or an extra-wide aisle. The self-steering axles would not squeal in the tight corners of the DTX tunnel.

  6. One more point: single-level trainsets can potentially offer higher acceleration than bi-level designs of the same length and power level.

    This could be very relevant if the corridor upgrade program is modified to feature a main line with a minimum speed of 90mph (where feasible) and a limited number of siding sections. This main line would be shared by HSR and a redefined Caltrain service I've dubbed Firebird. It would replace all existing service levels.

    Caltrains would exit and rejoin at the main line's minimum speed, cp. hopping off and back onto a fast-moving conveyor belt. For minimum siding length, useful line throughout and high line speeds, that requires hefty acceration.

    Maximum line throughput, i.e. 20tph, each excursion to a siding would need to take 2x3 = 6 minutes, allowing exactly one through train (HSR or other Caltrain) to overtake the one on the siding. For reasonable dwell times and service, that means stopping at 2 stations per excursion onto a siding.

    Fortunately, Caltrain's geographic ridership distribution would permit six distinct siding sections in the 3-4 mile range (incl. acc and dec sections):

    1. San Bruno(new)-Millbrae (main line + HSR platforms possibly elevated thanks to BART tracks north of station)

    2. San Mateo-Hillsdale

    3. San Carlos-RWC (optional tail track

    4. Menlo Park-Palo Alto-Calif. Ave. (any 2 for one overtake, all three for 2 overtakes; reconfigured Univ Ave/Alma intersection)

    5. Mountain View-Sunnyvale (main line in tunnel at 85 and Whisman, perhaps 237; curved road overpass between Moffett and either W Evelyn or Hope)

    6. Santa Clara-SJ Diridon (aerial over just CEMOF for 2 new south bound lanes; single level at SJ Diridon; W San Carlos remains overpass; close Auzerais + W Virginia; possibly gauntlet tracks through Gardner for 55mph top speed; mid-day stabling at new yard at 880/87 via Milpitas line turn-off or at southern end of PCJPB's ROW).

    In addition, the 22nd St and Tamien stations would receive limited service without bypass tracks. Service to other stations would be discontinued, but those only comprise 9% of current ridership.

    Basically, stopping at any station on the main line reduces line throughput by 1tph. Since there are 20 to play with, that's acceptable up to a point.

    HSR express trains running at 125mph would need 3-4 empty slots ahead of their own. Some of those could initially be occupied by Caltrains that then get overtaken in the first half of the SF-SJ route. Alternatively, HSR trains could initially catch up with any Caltrains in front of them and then lose the time gained at Millbrae and/or the mid-peninsula HSR stop.

    Total quad-tracked distance: ~20 miles instead of 50 as per the CHSRA proposal. However, those do include all of the most difficult = expensive sections, because building extra tracks is hard where people live, i.e. there is ridership to be had.

    I reckon a 13-stop train from SF 4th & King to SJ would take around 55 min, comparable to today's baby bullets but serving more stops.

  7. Sorry,

    "3. San Carlos-RWC (optional tail track"

    should have read

    "3. San Carlos-RWC (optional tail track in Dumbarton wye if trains serving only SF-RWC are desired. Those would be extended to Union City if and when Dumbarton rail ever gets implemented)

  8. @ Rafael

    The Talgo 22 looks like the most brilliant application Talgo technology I've ever seen. Does anyone know if you can fit a wheelchair from one car to the next on the bottom floor?

  9. @ Peter -

    yes, rolling a wheelchair to anywhere on the lower deck should be possible, provided the aisle is wide enough. It would also be easy to roll suitcases around.

    All Talgo trains feature the company's trademark wheelset technology, with two independent single wheels connected by a cradle that supports the cars well above their center of gravity - hence the 2% passive tilt feature. Due to damping, it kicks in a little later than the active systems, but it's much less complex.

    Each short, wide car is supported by its own wheelset on one end and by a hitch to the next car's on the other. Each wheelset is attached two bar linkages that passively steer in curves (up to a point). This ensures the wheel axles are always perpendicular to the tracks, so the flanges don't squeal in tight curves.

    Of course, since they're independent, the wheels can rotate at slightly different speeds, there's no top-of-rail slip, either.

    Of course, the power cars do have regular bogies so the above is not 100% true for the entire train. The downside of Talgo's wheelsets and the low center of gravity they make possible is that the individual wheels cannot be powered - at least, they haven't come up with motors compact enough to fit on the inside and the loading gauge usually doesn't permit mounting them on the outside. Of course, each wheel does have a brake.

    My guess is the biggest problems with the Talgo 22 are a high price tag for a ~100m configuration and insufficient acceleration performance for a ~200m version. The design has been around for a number of years, but I'm not sure if any railway has actually bought any.

    The company does also sell the bi-level cars without tractor/trailers for use in locomotive-drawn unpowered multiple units. They're rated for speeds up to 125mph, same as the powered trainset.

  10. Reality Check09 June, 2010 14:18

    The Talgo 22 brochure document properties shows a date of April 2003 ... so the 22 appears neither new or commercially successful; the Talgo Wikipedia page doesn't even mention the 22.

  11. So then the Talgo 22 isn't an EMU. It's just a regular train that happens to have seating in the power car...

  12. How much would it cost to upgrade Caltrain to 125 mph? This could make HSR operations considerably easier...

  13. @ Caelestor

    More like make Caltrain operations easier, enabling them to use the otherwise-likely-dedicated HSR tracks for overtakes of Caltrain locals without interfering with HSR due to speed mismatch.

  14. Or Caltrain could seek to purchase the Warratah or Millennium EMUs from Downer/EDI, which are built for CityRail in Sydney/New South Wales. The Millennium trains are quite nice since they're quiet, comfortable, have fast acceleration and have boarding heights roughly similar to Siemen's Velaro and other HSR trains.

  15. I think the Millennium EMUs have a platform height which just so happens to be exactly the same as the AGV's...

  16. @ Reality Check -

    as I said, IDK if Talgo has had any takers. Most operators need high acceleration performance from their regional rail rolling stock because that lets them improve service quality without having to lay a lot of new track.

    Now, if Talgo could figure out how to build a powered wheelset, they might have a winner on their hands. As it is, the only option is to deploy short trainsets and couple them together. That's probably why operators prefer to buy the cars, attach a locomotive and run intercity services at 100-125mph.

    @ Peter -

    no, the 22 is an EMU design. It's self-propelled.

  17. @ Caelstor, Peter -

    Caltrain's stations are quite close together, there's usually one every 1.5-3 miles. Accelerating to 125mph and then braking is too ambitious.

    However, there are clusters of consecutive stations that each deliver high ridership, e.g. Menlo Park + Palo Alto + Calif. Ave or Mountain View + Sunnyvale.

    If you set the main line speed at a moderately elevated level, e.g. 90mph (where feasible), you can use something like a Stadler FLIRT to run at that speed between two clusters, switch to a siding, brake, dwell for 60s, accelerate-cruise-brake, dwell at a second station for 60s, accelerate and rejoin the main line to get to the next cluster.

    If the acceleration performance is high enough, e.g. 1m/s^2 average from 0 to 90mph, you can keep up with a 3 minute headway pattern even if the stations are about 3 miles apart.

    Simplest case: a Caltrain is running at 90mph. There's another train running at the same speed 3min = 7200m behind it. The lead train cuts over to a siding and is overtaken by the second train. After spending exactly 6 minutes on the siding, the first train returns to the main line at 90mph.

    The second train could be an HSR train or else, another Caltrain leapfrogging the first. In practice, if you alternate between HSR and Caltrain and, have each Caltrain serve each siding, you really can pump through 10tph for each service during peak periods - as much or more than either optimistically claim they will need in the 2030 timeframe.

    Note that as long as the acceleration performance is there, it doesn't matter how long a given train is. That's the beauty of EMUs, though the OCS has to be designed to deliver sufficient juice. Do the math and you have massive passenger transportation capacity with just ~20miles of quad tracking instead of 50 as espoused by CHSRA.

    Admittedly, many of the sidings would be hard to construct because the ridership is highest where the people live in close proximity to the tracks. This isn't a free lunch, though you'd need ZERO new track between Bayshore and 4th & King as long as 16th St gets an underpass. 22nd St would remain a station on the main line, serving it reduces throughput capacity by 1tph (same for Tamien, possibly Burlingame), so you don't stop there often.

    Irksome Atherton and South SF would also stay at two tracks, but Menlo Park, downtown Palo Alto, San Mateo and Mountain View would all need four. Bayshore already has quad tracks but it just gets 137 boardings/day (out of almost 37000 total). Unless/until the Baylands development happens, service there would be suspended.

    Downside: this service concept only covers the subset of stations that together deliver 90% of current ridership. Those used to boarding at secondary or tertiary stations today would be SOL.

    On the other hand, a "local" would serve 13 stations - more than any baby bullet - and still achieve SJ-SF 4th & King line haul times in the 50-55 min range - better than the fastest baby bullet.

  18. (cont'd)

    Upside: you'd also get HSR service. Express HSR trains traveling at 125mph would require an initial headway of 9-12 minutes so they could gradually gain on the preceding Caltrain over the 48.5 mile distance between SF and SJ. However, they can give back some of those gains by stopping at Millbrae and/or the mid-peninsula station to reduce the impact on throughput.

    Constraint for HSR: during Caltrain's rush hour, HSR trains would either run at max. 90mph non-stop between SF and SJ or, they'd run at max. 125mph but make a stop or two. True SF-LA express trains that don't even stop in SJ could run off-peak, when Caltrain can make do with 5 tph (or less).

    Note that the superelevation levels required for 90 or 125mph operation in curves are not suitable for freight trains. Either those have to become lighter and faster, at night, or you abandon the freight corridor. The only location where gauntlet tracks for freight might possibly make sense is in the Gardner district in San Jose.

    Did I forget to mention that SJ Diridon would be a single-level station and that Caltrain and HSR would share track all the way out to the southern tip of the PCJPB ROW, even if that means all other operators have to install ETCS level 2 boxes in their locomotives and submit to traffic control according to a timetable? Oops, my bad.

    Of course, if Caltrain selects bi-level rolling stock, acceleration will not be as good. Either the headway would have to be 4-5min (reduced throughput and flexibility for both services) or the nominal line speed would have to closer to 75mph (longer line haul for same station pattern).

  19. Great post. It is becoming more and more clear that there has never been a fundamental shift in Caltrain's thinking about how to plan the corridor in regards to HSR. One would have hoped that the passage of Prop 1A would have finally pushed them to take HSR seriously. Calrain does say they want to use HSR tracks for passing locals (i.e. the shared use alternative), but it does not appear they value having cross platform transfers at HSR stations, evidenced by the lack of effort to deal with the antiquated CA rule that freight can't be encumbered by high platforms because some operator may be hanging off the side of the train. Rather it appears the path of least resistance is being taken, which will lead us to different platform heights, which will also lead to cumbersome transfers stations like the Millbrae fiasco. BTW, if BART ever manages to get to Santa Clara, the same lame set up is being envisioned as is the case in Millbrae. Rather than weaving the tracks so there are two cross platforms like there are at MacArthur BART station (which works very well) Caltrain will be off to the west and BART off to the east.

  20. Rafael, I'm not aware of any steel-on-steel train that can maintain 1 m/s^2 up to 200 km/h. However, some can maintain about 0.7 up to 160, in which case the train accelerates to its top speed in 2 km.

  21. @ Daniel -

    yes, the fundamental problem is that CHSRA and Caltrain are separate bureaucracies. Neither is willing to compromise on its service model to maximize combined bang for taxpayer buck. Neither is yet prepared - or feels empowered - to abandon the capital investment program that might (perhaps) make sense if the other didn't exist.

    And yet, Caltrain cannot survive unless it gets electrification plus full grade separation, because it will never actually be permitted to run 8-10tph during rush hour, because that would snarl up cross traffic at the remaining grade crossings. CBOSS wouldn't change that basic fact. Without a substantial increase in train frequency, Caltrain is unlikely to attract the additional ridership it needs to structurally improve its fare box returns, i.e. reduce the operating subsidy it needs (at least on a per-passenger basis).

    Conversely, CHSRA cannot succeed in getting the HSR project fully funded unless it shows it is doing all it can to drive down the price tag. Sharing not just the ROW but 60% of the actual track with Caltrain would be a good start. So would some scenario analysis of costs with and without UPRR service, so politicians can decide if they want to spend the delta.

    The knee-jerk assumption that all existing services must be preserved in their present state, forever, needs to be jettisoned. Neither CHSRA not Caltrain will do that of their own volition. Either funding shortfalls or political leadership have to force to do some scenario-based integrated operations planning and develop a plan for capital improvements on that basis.

  22. @ Alon Levy -


    I re-did my calculations and you're right, I did have a bug. The figure you give for the average acceleration from 0-90mph, 0.7m/s^2, is indeed more realistic.

    With a line speed of 90mph, a 3 min headway and the most critical segment (2.8mi between Caltrain stations worth stopping at), feasible dwell time goes down from my target 60s to just 44s. For another segment with two stations 2.2mi from one another, a 56s dwell is still possible.

    Shinkansen trains in Japan operate with dwells as low as 50 seconds, thanks to level boarding. However, those don't admit bicycles. Also, riders there are very aware of the need to board and disembark quickly because causing a delay means losing face and the fines are hefty, too. In California, not so much.

    As I pointed out elsewhere, dwell times can be improved by organizing passengers waiting to board on the platform: white painted lines to encourage queuing, distribution along the length of the platform, keeping the way clear for disembarking passengers, keeping those moving etc.

    In addition, enforcing one-way foot traffic along the aisle of the train speeds up throughput as long as there aren't any standees blocking the path to the exit or new passengers stowing baggage before the train gets moving.

    If a 60s dwell time is deemed necessary at all stations, the headway would have to be a little longer, about 200 seconds. That would still mean 18 tph throughput on the main line, which isn't exactly shabby.

  23. I don't know what you had for breakfast, Rafael... thank you Reality Check for dismissing this Talgo 22 nonsense, which did what exactly for platform compatibility?

    The service plan you suggest is interesting but you'd need to play with it a little bit to see if it would really work. Try the string diagram spreadsheet here. One major issue we have to contend with is off-nominal conditions: you can't plan every overtake to be nailed at exactly the right place and time. It's one thing to have a service plan that works when everything runs on time, and quite another to provide a robust and recoverable service plan. That's why timetable planners get paid the big bucks.

    Can't agree more that operational planning is the right place to invest, before deciding what capital improvements to make. Work smarter, not harder, as Richard likes to say.

  24. Btw, it turns out I underestimated the acceleration performance of available bi-level EMUs. The Stadler DOSTO comes in at 1.1m/s^2 starting acceleration vs. 1.2 for the FLIRT. The higher rated power of the bi-level train allows it to achieve 0.68m/s^2 average for 0-90mph in spite of the greater cross-section, i.e. greater drag.

    It's worth pointing out that trains are such long vehicles that lateral drag accounts for a significant fraction of the total. A very long single-level train need not be all that much better than a shorter bi-level design with the same number of seats.

    Still, the DOSTO guzzles more electric power than the the FLIRT to get to 90mph at full tilt, so seat capacity utilization had better be high. Bi-level designs make sense if a line is nearing capacity or, there's limited train throughput capacity at a terminal station (as opposed to the main line).

    Then again, station and throat design shouldn't constrain line capacity. That's why busy stations are almost always built with run-through tracks these days. TJPA, take note: force CHSRA to accept dwell times on the order of a <10 minutes rather than 30-40 just so the operator doesn't have to employ as many cleaning ladies.

  25. @ Clem -

    the Talgo 22 was just interesting because of the wheelchair question. It's probably not the right design for Caltrain, though Richard Mlynarik has suggested that HSR should use whatever platform height is optimal for Caltrain. IIRC, he said the TGV Duplex uses 550mm.

    My analysis so far is really just back-of-the-envelope to showcase the possibility. At the moment, it looks like I'm already sailing fairly close to the wind, but it looks like a solvable puzzle.

    Richard is right that it would be a lot easier to solve if HSR trains didn't need to run through Santa Clara county, home to 4 of my 6 siding sections. But then, Altamont is off the table at this point.

    Recovering from off-nominal conditions is of course the hard part, because it forces the operator to sacrifice system performance to gain enough buffer time.

  26. And I'll agree with Rafael that the Talgo 22 was only interesting as a novelty. It doesn't appear to have actual practical use (given that 7 years later it's still not used anywhere). It was a cool idea with no real practical application.

  27. Rafael, the dwell times for commuter trains are not the same as for intercity trains. For intermediate stations on a regional line, 50 seconds is very high, even by US standards. Even Metro-North and the LIRR don't dwell as long - my current recollection is that they dwell about 20-30 seconds.

    Bicycles have nothing to do with it. I've seen people take bicycles on the New York City Subway and BART; dwells were still about 20 seconds. Instead, it's an issue of level boarding.

  28. There's off the shelf train for train that can maintain 1 m/s^2 to 90 mph - it's two Viaggio (aka RailJet) cars pushed/pulled by ES64U4 (aka Taurus). For the record, such consist has 1.63 m/s^2 of initial acceleration - more than PCC streetcars, and whooping 35.0 kW/t.

    BTW, Viaggio is actually quite suitable train for Baby Bullet service - in consist of ES64U4 + four cars. Such train is cheap, it has high acceleration (over 1 m/s^2 in 0-100 km/h range), high top speed (230 km/h) and enough power to reach such speed (6000 kW continuous, 6400 kW hourly - yielding 21.7 kW/t and 23.19 kW/t respectively, comparable to 23.9 of N700). Such trains would even be capable of continuing on CHSRA's line south to Gilroy, and if the wire is extended, to Monterery.

  29. Yes, Caltrain's conceptual schedule shows 10 tph, BUT it shows most of those trains terminating at 4th and King. People want to go to TBT, not 4th and King. If anything, limited capacity into TBT would increase the need for bilevel cars.

  30. @ Alon Levy, dejv -

    Caltrain will have level boarding, whatever platform height is chosen.

    If dwell times can really be as short as 30 seconds, that means the nominally required acceleration/deceleration in-between stations on a given siding is much more modest. In the worst case I looked at, it came to 0.48m/s^2 for a 180s headway. The reserve allows the operator to regain a few seconds if he needs to get back on schedule.

    In the acc/dec sections connecting the siding to the main line, you'd typically want high values because they keep those sections short to minimize the amount of track you need to lay and maintain. However, choosing a more conservative value, e.g. 0.6m/s^2 even though your train can deliver 0.7, again makes the schedule more robust.

    Thanks for the Viaggio links, dejv. With shorter nominal dwell times, extreme acceleration performance may not even be needed in this case. That would be good news for Caltrain in terms of vendor choice, trainset cost and the required OCS performance.

  31. Sorry, the Viaggio links were broken, so once again:

    * Viaggio Comfort, used by ÖBB under Railjet trademark

    * Viaggio Light - low floor version limited to 160-200 km/h, probably derived from Desiro ML, AFAIK not yet built. The pdf also contains nice diagram showing platform height mess of Europe (with omitting lowest platforms at 210 mm ATOR, just like US 200 mm standard)

    * Complete Viaggio range, showing scaled-down ES64U4 designs

    (links taken from this danish site of Siemens.)

    The nice thing about push-pull sets is that at off-peak and night time, you can uncouple two locos and put them in front of freight train. Two Tauri have around the same tractive effort as single american standard mainline diesel, but they retain it to speeds in excess of 100 km/h (unlike the diesels with Vcrit around 20-25 km/h) - see this vid to get sense of it.

    Such operation of course necessitates:
    - compatible couplers (either use Scharfenberg at different level to AAR coupler, or just add automatic brake/air connections to AAR coupler)
    - compatible braking regulations (common air pipe pressure standards)
    - either reinforcing loco frame to FRA standards or limiting train weight

    This way, the only other locos needed for freight would be a few diesel (or battery-electric) shunters to serve unelectrifiable spurs, thus completely eliminating need for UPRR's locos access to Caltrain tracks.

  32. If the loading gauge is *really* tall enough, you can simply have 'high-level' boarding and 'gallery car' seating like on the Metra Electric line in Chicago.

    I really see no justification for a boarding level lower than 550mm.

  33. A few random thoughts:

    * New Jersey Transit only got bilevels because the approach to their main terminal at NY Penn is at capacity. It's two tracks, and they're putting 24 trains per hour through in the peak direction (4 of those trains being Amtrak). The trains had to be a custom design to fit in the constrained loading gauge. The only other NY-area railroad to use double deckers is the LIRR, which uses them on the diesel-powered outer branches. SEPTA does not use bilevels, but MARC and MBTA do, while VRE and Metra use gallery cars.

    * NJT's bilevels have entrances on the middle level, with one set of doors at the ends of the cars allowing for both low and high platform entry, and another set at the inner end of the middle level for high platforms only. The seating is 2+2 in the upper and lower level. MBTA bilevels only have the end vestibules, and 2+3 seating. And Metra has high-platform gallery MUs.http://oppugn.us/posts/1276122413.html

    * The "everyone gets a seat" metric is reasonable, but only under the assumption that everyone is going to SF. The real metric ought to be "everyone on the train for more than 15 minutes should get a seat", which, if you assume everyone is going to SF, means that every trip is like that and everyone gets a seat. But that's not true anymore, and the peak load point tends to be right before Millbrae. And while Caltrain doesn't encourage short distance riders, they still exist, and are basically a free source of revenue given that they don't need seats at all.

    * Bicycles, for better or worse, have become an integral part of Caltrain's service, but I think the bike car arrangements can be improved somewhat. I would suggest having two or more bike cars per train, and using an arrangement like ACE has, with racks for individual bikes rather than stacking bikes on top of each other. The queue of people waiting to put their bikes in the racks sometimes extends out the doors of the train, which adds dwell time.

  34. neroden: Metra's MU gallery cars use pretty much the same body shells as Caltrain's regular gallery cars. The door cutout is higher, and the floor of the vestibule is slightly higher, with ramps down to the seating area similar to the ramps at the ends of the cars (which are to provide a standard 51" ATOR inter-car passageway)

  35. Almost all of Cityrail stock is EMU bilevel high boarding. The exceptions are the Endeavours and Hunters, which are single level high boarding DMU's (built up from engine one car, driver-trailer the other, inter-train access when coupled engine-car to engine-car)

    The intercity V-sets have narrowish doors, a door to the mezzanine, and steps up and down, all seated. Dwell times in the busy Wynyard and Town Hall stations for V-Sets heading up to West Sydney and the Blue Mountains is high, and of course they leave in the evening peak with a large number of standing passengers crowded into the aisles.

    By contrast, the inner suburban Tangaras have wide doors, opening directly into the mezzanine with metro seating and standing room on the mezzanine and all seated in the upper and lower decks. They have much less delay for access and egress in the busy Sydney City stations.

    Of course the intercity V-sets have toilets and the inner suburban Tangaras (and the more recent Millenium Bug) do not.

    The new outer suburban OSCARS are similar, with a disabled access toilet on one side. Since there is inter-train access on the mezzanine level, that means that any wheelchair passenger entering at a mezzanine has access to a toilet.

  36. Thanks to everybody who responded. So I guess the better solution is single-level trains (which is what every other country uses anyway). Keeps frequency high and entices customers.

  37. @ Caelestor -

    There are actually plenty of regional rail operators all over the world who use bi-level cars. If platforms are short and/or your lines are congested, that's what you choose (loading gauge permitting).

    That said, those same operators usually have some single-level rolling stock as well, to avoid moving a boatload of empty seats off-peak and on weekends.

    Clem's point is that Caltrain has a choice and it has selected bi-levels (i.e. 550mm platform height) almost by default, which implies sacrificing compatibility and hence, operational flexibility at the TBT, Millbrae, the mid-peninsula station and SJ Diridon. Space is at a premium in all of these locations.

    Harmonization is a Good Thing(tm), the question is if Caltrain should adopt CHSRA's preference or vice versa. Meeting in the middle (~760mm) doesn't really work for either operator.

    The only reason single-level HSR trainset operators prefer high boarding is that it delivers a level deck. Designers like it because there's plenty of space to sling heavy technical components like transformers under the car bodies to enable distributed traction. Bi-level HSR designs can have high boarding (above the bogies) or low boarding (in the middle of the car), depending on what the customer wants.

    However, no-one has delivered a bi-level design capable of 220mph with distributed traction yet. The latest generation TGV Duplex is rated at 200mph but still features a tractor-trailer layout, with implications for starting acceleration and hill climbing ability.

    The latest trend appears to be increasing the number of seats per unit of platform length in single-level HSR designs to levels that were previously only feasible with bi-levels. The trick is using shorter, wider cars that fully exploit the available loading gauge to deliver more that four seats across and leveraging that to install more motor bogies. The real innovations are in reducing the size of those transformers (and battery packs), because they have to go somewhere close to the ground and operators want seats in the end cars.

    So for now, Caltrain and CHSRA have each concluded that the other's platform height preference doesn't work for them. Besides, bureaucrats like to stay off each others' turf.

  38. Reality Check10 June, 2010 14:25

    Here's a not-so-nice solution: could either Caltrain or HSRA (depending on which one won't budge on platform height) get equipment with 2 sets of doors at each platform height per car for level boarding at either platform type?

    Yeah, it's far from optimal in numerous ways, and it'd probably make for an ugly and expensive and interior-space-inefficient train and so on, but I just wanted to throw it out there as grist for the mill.

    Or would it be OK to have each trainset have some cars with high doors and others with low ones? Of course, then you'd have entire cars that would be unavailable for boarding/alighting at stations with the "wrong" platform height. And riders traveling between station pairs with differing platform heights would have to alight using a different car than they boarded.

  39. Reality Check: no. ADA will almost certainly not allow it. How would a passenger in a wheelchair get from a high platform station to a low platform one? That's the whole point of level boarding, and if not for that you can just use steps and traps to allow high-floor trains to use both high and low platforms. Actually, given HSR dwell time and crewing requirements, it might actually make sense to install a lift in HSR trains, but would definitely be a bad idea in commuter trains.

  40. Federal law (49CFR38.175, under Accessibility Specifications for Transportation Vehicles) actually requires all high-speed rail vehicles to use level-boarding high platforms.

    This is a dark unused corner of ADA law where HSR is lumped in with monorails, maglevs and other gadgetbahn miscellany not already covered by other regulations.

    If you attach any validity to the letter of the law, there it is-- we have no other choice but high platforms, unless the law is changed or a waiver is granted. The likelihood of that happening is fairly close to zero.

  41. Doesn't "high platform" mean higher than 8 inches in this context? I've seen the ADA boarding platforms for Bombardier cars referred to as "mini high platforms" before.

  42. The simple way to address this problem is to require any company that wants to sell caltrain rolling stock to make sure it is at the same boarding height as HSR. Similarly, CAHSRA needs to define, soon, the guidelines for rolling stock or to at least say boarding heights must either be x height or must be within three inches (or whatever is allowed for ADA) of the height of the platform.

    It would also be nice if the Authority or Caltrain put it's foot down and started to actually say what is going to happen instead of remaining very quiet on this subject. This would be helpful for potential vendors since they could begin to market their product(s) to the interested parties so better planning could take place.

    The important thing that is not explicitly said in many of the topics on this and other blogs is to not re-invent the wheel. Keep it simple. Keep it practical. Don't let outside parties influence the modernization of Caltrain and the building HSR so that the end product becomes lost in the process.

  43. The trick is using shorter, wider cars that fully exploit the available loading gauge
    The Japanese and Swedish approach is completely opposite - use full-length full-width cars and widen loading gauge in curves. Given that CAHSR is brand-new system and bulk of legacy lines supports DOD's 3650 mm envelope (I suppose also with full-length cars), the best way for CA, NV and AZ railway system is to keep height compatibility with NEC but use wider cars - in 3400-3650 mm range. That would require retrofitting of Amtrak's LD cars with some bridges like the one I linked in my first post under this blog (but not that wide - the gap between the Desiro Classic and TSI platform is 260 mm), but that would also allow high platforms along STRACNET lines, making feasible single platform standards for all passenger RRs of American southwest.

  44. @ dejv -

    you're quite right, any discussion regarding single vs. bi-level rolling stock for Caltrain needs to consider the loading gauge.

    In the context of the grade separation project, a lot of track will have to be moved or built from scratch, so Caltrain isn't necessarily stuck with whatever loading gauge it has now. Wide-body single-level cars could offer very high seat capacity per unit of platform length and make harmonizing platform height at HSR levels a no-brainer.

    The one caveat is the DTX tunnel. The current design features several very tight curves. SF being what it is, it would be difficult to secure local environmental approval to increase curve radius beyond the tweaks TJPA has already made.

    IFF CHSRA ever comes to its senses on dwell times at the TBT, TJPA could switch to a one-way single-track loop tunnel to deliver a run-through station. They already floated that option, Caltrain liked it, CHSRA killed it.

    If it is revisited, both operators would need to consider the implications of slightly curved platforms. Doors located in the middle of cars, rather than at the end, would then a really Good Idea(tm).

    Ironically, run-through tracks at the TBT would give Caltrain so much throughput capacity that harmonized platform heights would matter much less than Clem's post implies.

  45. "49CFR38.175 ... actually requires all high-speed rail vehicles to use level-boarding high platforms. ... unless the law is changed or a waiver is granted. The likelihood of that happening is fairly close to zero."

    A waiver for level boarding is certainly out of the question -- as it should be.

    As for "high platforms", the preamble to that section of the regulation strongly suggests it was written by the sort of people who have never seen anything except Amtrak North East Corridor, and the "high level" business (which appears to have no legal definition, anyway) came along out of sheer laziness and intellectual disability. My bet is that it would be easily and fairly readily severed: spirit rather than the letter of the law and all.

    "Since the systems envisioned will not be affected by station platform set-back requirements needed along freight train lines, this section requires level boarding from high platforms."

    I'd say the odds of obtaining a waiver to allow level boarding but not "high platform" boarding are very good. Certainly a far easier undertaking than Caltrain's CBOSS-is-now-the-law FRA-lovefest. I'd also say the odds of the think-way-way-WAY-inside-the-box technical geniuses at CHSRA applying for, understanding the need for, or even thinking about such a waiver are zero.

  46. Rafael -

    If it is revisited, both operators would need to consider the implications of slightly curved platforms. Doors located in the middle of cars, rather than at the end, would then a really Good Idea(tm).

    Don't forget that platform can also be in the outside of curve, where door in the middle of car won't help minimize platform gap. ;)

    ADA regulations mandate 76 mm gap or narrower. If you keep 40 for geometry tolerances, you're left with f=36 mm for curvature - with L=26 m cars, that means minimum radius of track in curved platform is:

    L^2 / 8f = 26^2 / 8*0,036 = 2347.2 = 2350 m

    Pretty straight track, ain't it?

    Richard -

    BTW, more than half of German platform edges is in 200-380 mm height range, despite 760mm standard is there for more than century (since 1904 permitted, since 1928 recommended and since 1991 mandated). That's the single reason, that drives low-floor vehicles in Europe - not technical, but economic one. The advent of low-floor vehicles was for compatibility with such platforms in 200-840 mm range. They do they job fine, but it's not perfect (weight, stairs and costs are non-negligible drawbacks of low-floor vehicles) and they effectively prevent a change to operationally cheapest solution possible, high platform in 1100-1300 mm range. There's no need for USA to intentionally bring such nonsense where it isn't necessary at all.

  47. dejv: As I've said elsewhere I think there are very good arguments -- good enough to have changed my decades-long prejudice -- to consider a standard sub-700mm platform, the argument primarily being compatibility with an optimal design of double-deck inter-city vehicle. (Boarding into lower level, doors spaced evenly at quarter points along vehicle length.)

    It's a happy technical fact that there are many existing regional/local single-deck designs that are compatible with that lower platform height (for European rail-historical reasons); and of course the majority of the world's regional/local double-deckers work at ~450-~700mm "for free".

    And yes, I am aware of the disadvantages of not being able to mount equipment under-floor. I suggest first that roof-mounting modular propulsion and AC units on single deck equipment is really the way forward on the rapid maintenance front ("Lirex", etc.); and secondly that the loss of interior space when equipment is internally mounted doesn't seem disastrous anyway for regional (not metro) trains.

    Another way to look at your "such nonsense" is that high platforms were based on the historical technology -- historically the only possibility, and today one of a number of possibilities -- of single deck continuous floor cars with a pair of under-floor 2 axle bogies, and that locking all future train designs into compatibility with that configuration on a built-from-scratch new regional+intercity system isn't necessarily the optimal solution.

    Anyway, I think these are arguments which in a better world would receive a great deal of attention and careful analysis, particularly with regard to long-term inter-city station platform throughput.

    And the larger point is that Doty is outright lying about an incompatability of "commuter" and "high speed" equipment, whether at 400mm or 1300mm or some point in between, and is deliberately causing billions of dollars of additional cost and crippling San Francisco Peninsula corridor operations permanently, egregiously, and unnecessarily.

  48. There is one class of railcar where there are absolutely no compatibility constraints on the train-platform interface, and these are pretty much without exception built single-deck with high floors: cars for new-build rapid transit lines. The lines where low floor vehicles are used are either tramways, where this allows for level boarding and makes for a much more convenient interface between platform and street, or mainlines, where compatibility with existing platforms is an issue. The few mainline style services that are new-build with no compatibility issues and use low-floor trains (Ottawa, Austin, San Diego) do so because they are very small systems and need to but their fleet of 3-12 vehicles off the shelf. So in a way, the low platforms in San Diego are a side effect of the numerous 8 inch platforms in Switzerland.

  49. Dear Arcady,

    Metros are inherently single-deck and the stations are inherently (or overwhelmingly) completely segregated from surrounding neighbourhoods (ie usually underground.) High floor is natural and simple and free of any downside in this context.

    The argument that persuades me to consider lower platforms for regional and intercity is, again, that if you ever plan to operate double-deck equipment, and if you have no historical interoperability constraints, then it seems to make sense to design to allow the best design of double-deck equipment.

    As for new-build systems (segregated metros and trams excepted), about the only ones I can think of are Shinkansen (1960s), Spain's AVE (1990s, and built around European interopability), and Taiwan's HSR. Everything else has worked around and with existing rail infrastructure. (To be sure I don't know whether Turkey's new high speed lines do so -- though TSI must figure there -- nor do I know what's up in South America.)

    California is something like Japan, Spain and Taiwan in having a new system completely segregated from the existing rail network. (Yes, I know this is a simplification, what with gauge changing trains, mini-Shinkansen, etc.) Unlike Japan it is being designed and built decades after both double-deck and low-floor single-deck train designs have been developed, and unlike Spain there are no interoperability reasons to go with the TSI. Perhaps like Taiwan there are reasons to just clone a Shinkansen system. Or perhaps there are reasons to take a different route, even bearing in mind the issues -- issues I've harped on about for 20 years -- with doing anything but exactly copying existing designs and standards. Perhaps.

  50. Argentina uses Iberian gauge and Brazil meter gauge, so their systems are going to be totally segregated. However, they're not good examples to follow, because they're like the US in that they're building Flight Level Zero airlines for prestige reasons. This is especially true for Argentina, which is doing some shady deals with Alstom and SNCF that would make PB proud. I'm less sure about Brazil - its proposal is strong in principle, but it's still just on paper at this stage. If it gets built, it will probably succeed on the strength of the cities' size (same reason I think CAHSR will succeed even with Kopp, Diridon, and Pringle in charge).

    Even Taiwan got embroiled in shady deals. It intended to get a European system, built by SNCF and DB. Then the corruption involved was exposed, and a few dead bodies later, the government decided to drop the bid and build Shinkansen instead.

  51. Richard -

    I question the very need for double-deck intercity vehicle in western USA context. The whole point of double deck trains is to increase net floor area where you can't increase neither length, nor width. This is pretty common in Europe, where even in major stations, platforms are limited to 200 m and width is limited by dysfunctional, dumb, outdated (etc.) 3150/250 rule.

    But even then, the net floor area gains are pretty minor, because:
    - you can add the second deck between trucks only
    - the staircases and aisle in second deck should be clear, so it is dead space
    - the limited space for traction equimpent on roof above trucks forces the TE to take up another piece of valuable floor area.

    All things combined, well-designed single-deck train often provides the same value as double deck one:

    - single deck DB 440.0: 240 seats (incl. folding) / 70.9 m = 3.39 seats/m
    - single deck DB 425: 236 seats (incl. folding) / 67.5 m = 3.50 seats/m
    - double deck Dosto: 535 seats / 150.0 m = 3.57 seats/m
    - double deck Desiro: 378 seats / 100.0 m = 3.78 seats/m
    - double deck CityElefant: 310 seats / 79.2 m = 3.92 seats/m

    not much difference, is it? All the train feature 2+2 seating. Now, let us take into account trains with 3+2 seating, all single deck with exception of E1:

    - spacium 3.06: 405 seats / 94.3 m = 4.29 seats/m
    - M-7: (110 + 101) / 2*25.908 = 4.07 seats/m
    - N700: 1523 / 400 = 3.8 seats/m
    - E1: 1235 / 300 = 4.12 seats/m

    So you can see it for yourself, european double-deck commuter trains with 2+2 seating, including seat-count-optimized CityElefant, have systematicaly lower seating capacity than single-deck trains with 2+3 seating - including Spacium with lots of space dedicated to standing, strollers, bikes or wheelchairs. In addition, low-floor EMU either needs roof-mounted traction, adding 10-20 % to carbody weight, or the TE must take space for passengers.

    All that stuff combined, advocating narrow low-floor trains for brand-new system running on tracks that support whooping 3650 mm loading gauge width (allowing comfortable 3+3 seating) seems incredibly short-sighted.

    Alon -

    the Brazil's line is to be standard-gauge and I wouldn't be surprised if Argentina follows, at least for the line north to Brazil.

  52. The middle seat sucks. I'd much rather climb a short flight of stairs than sit in the middle seat. Even on long-ish commuter runs people stand instead of sit in the middle seat.

  53. dejv, I know that the Brazilian and Argentinian HSR systems will be standard-gauge. This is why they do not need to consider historical loading gauge issues, unlike countries where HSR runs on legacy track for much of the way.

  54. @ Richard Mlynarik -

    your argument boils down to the notion that CHSRA should purchase bi-level HSR trains. There are no such trains that are capable of 220mph operation and it's far from certain that there ever will be.

    Putting heavy equipment like transformers on the roof instead of under the floor would require much heavier car superstructures to cope with the static and dynamic loads. It would also raise the center of gravity, never a good idea for a fast-moving vehicle.

    As dejv points out, the trend these days is to build new tracks to a wider loading gauge such that you can get (almost) the same seat count per unit length by seating 5 or even 6 abreast without significant loss of passenger comfort. And of course, such designs can have a continuous high floor with plenty of room to sling all the technical components underneath.

    If Adirondacker12800 doesn't like such seating arrangements, he's free to buy a first/business class ticket.

    Clem's argument is that Caltrain should also pick wide-body single level designs suitable for high platforms, though with some room reserved for standees.

  55. @ dejv -

    Regarding the door-in-the-middle concept, stations in curves would have to have side platforms for both directions on the inside of the curve. Neither conventional side platforms nor island platforms would work for both directions.

    However, the TBT is really the only station in the entire Caltrain corridor that could possibly present a problem in this regard and then only for CHSRA, which wants to give the future operator the chance to run full-length trains.

    The new TBT has been designed, but it hasn't been built yet. In principle, the design could still be changed to accommodate narrower curved platforms that would only be used for one track each. This might mean going from 6 to 5 tracks, but if accepting curved platforms enables a loop tunnel that makes all of 5 of those run-through tracks, you'd still end up with more throughput capacity for the station as a whole.

    Of course, all of this presumed CHSRA will drop its ridiculous 30-40min dwell time requirement just so cleaning can happen underground in SF rather than in SoCal.

    Btw: your calculation regarding gap width is too optimistic IMHO. You need to allow more than 40mm for geometry variations. The minimum curvature radius I read would be acceptable to CHSRA is 10,000m.

  56. The minimum curvature radius I read would be acceptable to CHSRA is 10,000m.

    Look it up, it's 20 km, and then only as a design exception requiring special approval.

  57. Rafael, nothing you write shows any understanding of anything Richard Mlynarik wrote.

  58. @ anon @ Jun 13 7:37am -

    Richard Mlynarik wrote:

    "[...] consider a standard sub-700mm platform, the argument primarily being compatibility with an optimal design of double-deck inter-city vehicle."

    How is that supposed to mean anything other than that CHSRA (i.e. the agency tasked with planning the inter-city service) should harmonize with the low platforms Caltrain prefers, because bi-level is supposedly inherently better than single level?

    For reference, the Alstom AGV is based on a platform height of 1155mm. Other designs trains rated for operation at 215-225mph have similar requirements (exception: Talgo 350).

  59. To throw yet another wrench into the calculations of single vs bi-level, you could fit nearly as many seats in the N700 if you went 2+2 and reduced the seat pitch to that of a TGV.

  60. If you're curious as to my numbers:

    1160mm seat pitch in coach class of N700,
    920-940mm seat pitch in coach class TGV:

    (1000 / 1160) x 5 = 4.31 seats/m in seating area
    (1000 / 920) x 4 = 4.35 seats/m in seating area

    Given that the TGVs seat pitch is more like what americans would expect from a business class airline seat, you could give them business-class width as well and fit just as many seats as the Japanese do.

    The issue with the N700 at least, would be that you'd have seats with blocked views.

  61. @ Andy Duncan -

    the N700 is definitely an interesting option, though the version JR uses is rated at just 300km/h, not the 350-360 that CHSRA needs.

    All or almost all of the axles are powered, but each of the motors is quite small, on the order of 300kW. To reach the higher speed, roughly (360/300)^3=1.72 times the rated power would be needed, since aerodynamic drag dominates total resistance at high speeds.

    Using longer reduction gear ratio would let the motors hit rater rotor speed at the desired top speed of the vehicle, at the expense of reduced acceleration at low speeds.

    Still, rated torque would have to increase by a factor of (360/300)^2=1.44 to achieve the required power level. This would compensate, potentially even overcompensate the traction force lost to the longer gear ratio.

    The axle load is low at 715/64=11.2 tons, so the real question would be if there's enough space to fit beefier motors, power converters, transformers and cooling systems.

    I believe the Japanese shinkansen use a platform height of 1250mm.

    Note that the Russian and Chinese versions of the Siemens Velaro also feature wide bodies and 3+2 seating in economy class. The Chinese version is already rated at 350km/h in commercial service.

    IIRC, the Chinese state railway originally settled on 1200mm platforms for the CRH2 but has since raised that 1250 for the CRH3. That's a difference that can be bridged with the pneumatic portion of the suspension system, so the two should be interoperable.

    Those numbers are much higher than for off-the-shelf regional trains, at least European designs. For example, the Stadler FLIRT is built to the UIC's 550mm standard.

    We shouldn't forget that it's perfectly possible to implement level transfer platforms between trains built to different platform heights by putting the respective rails at different elevations. That would still prevent each from using the other's platform tracks, though.

    In the SF peninsula context, the TBT will be a terminal station while Millbrae, RWC/PA/MV and San Jose will be transfer stations. Reading between the lines, I think Clem is most concerned about Caltrain slated to get just one island and two platform tracks at the TBT, even though it will have to support many more passengers during rush hour.

  62. JR Central is promoting an export model of the N700 that can reach 330 km/h and has even better acceleration than the current N700.

  63. @ Alon Levy -

    perhaps this version features newer motor designs with both higher rated speed and higher rated torque. I have no doubt the Japanese engineers can figure out the propulsion issue.

    I just wonder if CHSRA's new CEO Roelof van Ark (ex-Alstom) is open to purchasing a Japanese design.

  64. an export model of the N700 that can reach 330 km/h

    Maybe the damn train wouldn't need to reach 350 km/h (220 mph) to make the timing work, if they could stop dropping new obstacles in its path!

    Every time they add 30 seconds to the run time, the train needs to have a top speed about 1 mph faster.

    The solution isn't more torque or anything related to the cube of speed. The solution is to engineer an alignment entirely devoid of frivolous "iconic" civil engineering features. Transbay. San Bruno. San Jose. The list goes on and on.

  65. @ Rafael, curved platforms -

    such design requires more digging than straight platforms, because one island platform is narrower than two single-edge platforms. In addition, it locks up train design to single car lenght, because longer cars would strike the platform and shorter would leave too wide gap for ADA compliance (and this goes with square of length).

    IMO, much better is to drop the "no tunnel under buildings" requirement, keep Transbay's throat straight and use 3rd Street to connect to 4th & King station (this also allows much wider curve radii and simpler turnout geometry).

    @ Rafael, power for 350 km/h operation -

    AFAIK, China runs their CRH3's at 350 km/h with Siemens's off-the-shelf SF 500 family bogies, so the 8800 kW per 200m trainset works for them. N700's power is similar and JR often claim they use overgenerous safety margins, so I guess 350 km/h operation is feasible with unmodified Shinkansen designs.

    The real trouble is once you start demanding high performance on long steep grades.

  66. @ Clem

    I agree with the "iconic" structures. CHSRA shouldn't be responsible for curing local politicians' small penis syndrome.

    Could they keep Transbay's throat straight if they build the trainbox now? Or will those damn support columns have to be built at the same time?

  67. Um... HSRA pretty much is a bunch of local politicians with small penis syndrome. And perhaps this tradeoff between higher train speed and "iconic" structures is not one that would have to be made, if the travel time were not prescribed by law, which seems like a crazy way to engineer a railroad. Anyway, as a denier, I'm not too worried about it: it's not like the central section is ever going to get built anyway. Even if they ever finish the Peninsula part of the route, there won't be any money or political will left to build the rest, which will be fine because by that point, they'd need trains to go 300 mph to make the legally required travel time thanks to more "iconic" slowness.

  68. My point wasn't about the speed of the trains, or even a statement that the authority should acquire the N700s (though I do think there's good reason to go with widebody trains).

    The point was that all the talk about how the Shinkansen fit more seats onto the train by going with 3+2 instead of 2+2 is incorrect, or at least misleading, given the looser seat pitch.

    Per linear-meter of seating area, the shinkansen fit almost exactly the same number of seats as the narrower TGVs, they just have more linear meters of seating area in a given trainset.

    Where they get the additional area is important. Obviously compared to the TGV much of that additional area comes from the longer cars and EMU layout, which yields more usable space, but even versus the Velaros, the shinkansen has better seating numbers than the 2+2 velaro where seat pitch is around 970mm (4.1 seats/meter in seating areas).

    Perhaps they get the space from having fewer amenities? Smaller and/or fewer bathrooms? Better equipment placement yielding fewer equipment closets?

    The point is that if you could get TGV duplex-level of capacity from a 2+2 widebody EMU with nice big seats. Why would you bother with a duplex?

  69. Why would you bother with a duplex?

    Because all the kewl kids have them.

  70. Adirondacker -

    The middle seat sucks.

    Could you measure it when you'll travel with that vehicle again? In my experience, it depends on width, if it is at least 550-600 mm wide, there's nothing uncomfortable about it.

    Andy -

    If I read WP correctly, the 1160 mm value applies to green class of half-length train. The half length train has 546 seats, equalling 545 of TGV Duplex, but way below half of full length train - 1323/2 = 661, so the full length train must use 3+2 for reserved seating or have smaller seat pitch or both.

    Ground plan of Velaro E shows that it's single biggest space waster is Cafeteria car.

  71. Could you measure it when you'll travel with that vehicle again? In my experience, it depends on width, if it is at least 550-600 mm wide, there's nothing uncomfortable about it.

    I've never measured them and have no plans to. The middle seats are just as wide as the rest of the seats. That they suck isn't my personal opinion but one you can verify on rush hour trains. People will stand instead of sitting in the middle seat. Their arrival was widely bemoaned and the announcement that the bi/multilevels would be two by two was heralded. . . the middle seat sucks.

  72. @Dev:

    Yeah, my other source only listed 1160, if it's 1040 for the coach class, then you're at 4.81 seats/meter in seating areas.

    The 8 car sets are less space efficient than the 16s simply because the 16s aren't two 8's chained together, they have two full passenger cars in place of the middle two end cars in a 200x2 consist. The seat pitch is the same, though the mix of coach/green might be different.

    WP says the 2+2 Velaro can seat "up to" 536 seats, I'm guessing that's in an all-coach config.

    Bringing it down to 1040 means you can't fit the same number of seats in a 2+2 as the N700 or a TGV duplex, but you could still fit more than in the regular TGVs, and maybe still more than the AGV.

  73. The N700 and E5 Shinkansen offer 1040 mm in regular class. On both trainsets, the middle seat is 460 mm wide, to increase comfort, and the other seats are 430-440 mm wide. Some middle cars can be configured with 100 seats, but most have less; the N700 is overall more efficient than the E5, but that's only because it's longer.

  74. Discussions over seat capacities are all very interesting, but also quite irrelevant.

    Standard practice among rail agencies in California is to run way more seats than needed. That seems very likely outcome here, with San Jose being intermediate stop. If CHSRA had wanted to plan for efficient seat utilization, they would have begun with optimal network layout rather than fiddle with seat pitch.

  75. And yet, Caltrain cannot survive unless it gets electrification plus full grade separation, because it will never actually be permitted to run 8-10tph during rush hour, because that would snarl up cross traffic at the remaining grade crossings.

    Total BS. Just because you read it somewhere on the internet doesn't make it true. There are plenty of LRT lines running nearly double that amount of traffic over hundreds of grade crossings just here in the US. And no, the additional length of commuter rail trains doesn't catch up to the LRT's total time of blockage per hour.

    Middle seats are a complete non-starter. This isn't China, cattle car configurations aren't necessary or wise. Some of you techno fetishists need to take a break from mommy's basement computer and spend some time in the real world. 99% don't give a flying fig about Japanese train width, they don't want to sit in a damn middle seat, especially with a brand new train system. For choice riders, customer preferences trump nerd olympics.

  76. Adirondacker1280017 June, 2010 20:57

    99% don't give a flying fig about Japanese train width

    Depending on the source you consult Shinkansen are an outrageously commodious astounding 6 inches wider than a standard North American passenger railroad car. In nice round number 150 mm/15 cm.

  77. DE: Discussions over seat capacities are all very interesting, but also quite irrelevant.

    Standard practice among rail agencies in California is to run way more seats than needed.

    Well, that makes number of seats per train more imporant rather than less.

    Anonymous: Middle seats are a complete non-starter. (...) 99% don't give a flying fig about Japanese train width, they don't want to sit in a damn middle seat, especially with a brand new train system.

    The choice riders value speed in first place and comfort in second. In any city with good subway system, you'll see plenty of affluent white-collar guys straphanging. Back to the middle seats - compartment cars on IC/EC services always feature a middle seat and they're anything but uncomfortable. The little trick how to the railways achieve that is that those seats have over 600mm/2ft lateral space, so even quite wide persons can sit there comfortably without interfering to the other seats.

    Shinkansen seats are about 3/4 of that value, similar to seats of city buses, where a single person often occupies 1.5 of seat, so it's not good arrangement for European or North American railways IMO.

    adirondacker -

    compared to Bombardier's Bilevels, it's full 400 mm wider.

  78. Adirondacker1280018 June, 2010 07:15

    compared to Bombardier's Bilevels, it's full 400 mm wider.

    It depends on the source you consult. Most of them cluster around Shinkansen being 11 feet wide and standard North American passenger cars being 10'6". YMMV.

    The meme is that Shinkansen are extraordinarily wide. They are compared to conventional Japanese cars or the IRT. They aren't all that much wider than standard run of the mill North American cars.

  79. Oh, sorry, I meant MultiLevel and it's only 350 mm narrower. It's still more than half of full-width seat though.

  80. Adirondacker1280018 June, 2010 19:37

    The Multilevels feature 2 x 2 seating and wider aisles. And as the sales page from Bombardier mention, up to 30 % more seating than single levels.

  81. If you look at the Bomabardier bi-levels, they actually have three levels, because the part of teh bogies is mid-way between 'upstairs and 'downstairs'. Around Toronto, the entrances are on the lowest level (low platform), but the same train design would still work just as well with the doors on the middle level (high platform).

    In short, bi-levels do not dictate high or low level platforms - door placement does.