09 December 2008

Lemonade from a Lemon: Millbrae Take 2

The comments in response to Focus on: Millbrae were quite thoughtful and identified another couple of reasonable options for reconfiguring the station for the arrival of HSR. They are labeled Option C and Option D in the expanded diagram below.

(click to enlarge-- and yes, this figure is getting too big...)

After some consideration, Option D stood out to me as the least-terrible way of making lemonade out of the existing Millbrae lemon.

Option D pro:
  • 4 platform tracks for HSR / Caltrain
  • Cross-platform transfers between Caltrain Express / HSR and Caltrain local, in both directions
  • No demolition and reconstruction of the existing station mezzanine
  • Platforms stay pretty much where they are
  • "Site One" transit-oriented development to the west is minimally impacted
  • Keeps It Simple, Stupid (compared to nearly every other solution)
Option D con:
  • Loss of northbound cross-platform transfer between BART and Caltrain
  • Non-stop HSR passes platforms at ~125 mph
  • Need to establish physical separation of BART and Caltrain on adjacent tracks
While this is probably the best technical solution, it remains to be seen whether it is politically workable, especially because it destroys the BART cross-platform transfer, which was a huge selling point for the existing station. Then again, follow the money... the smaller station footprint and lesser impact to the neighboring development may favor Option D.


  1. @ Clem -

    that's a very nice picture. However, I think your approach focuses a little too much on the station layout and perhaps not enough on the services that may be provided.

    For example, you assume that

    - Caltrain will maintain "baby bullet" service after real bullet trains become operational.

    - Caltrain will decide not to operate regional HSR trains using equipment separate from both legacy rolling stock and the Siemens Desiro EMUs it intends to use for commuter rail service. Regional HSR service would require equipment capable of the speeds and accelerations defined by CHSRA for the Gilroy to SF segment, but no more.

    - station layout dictates how the approach tracks will run, rather than vice versa. HSR will run on the inside tracks throughout the corridor, crossing tracks should be avoided. That likely rules out your option C.

    - there is significant value in retaining existing BART service to Millbrae even after the new TTC and the DTX tunnel to it become available.

    - FRA will permit track sharing for legacy FRA-compliant and new non-compliant equipment not just for Caltrain but also for other operators incl. UPRR


    So here's an option E for you:

    - number the existing tracks from west to east as 1 through 5

    - current configuration is: 1&2 for Caltrain (FRA-compliant), 3-5 for BART

    - proposed configuration:

    --- make do with 5 tracks at the station

    --- tracks 2&3 for non-compliant standard gauge trains only, principally bullet trains (multiple operators possible). Frequently used as bypass tracks, but stopping should be possible. Install Japanese-style platform fences plus floor markings and use loudspeaker announcements to warn people when a train is about to zoom past at high speed. The platforms need to be raised for level boarding and 1350' long.

    --- tracks 1&5 for standard gauge mixed traffic (legacy FRA-compliant and new non-compliant Caltrain equipment plus the occasional UPRR freight train). Raise platforms and extend to 1350' if and when Caltrain legacy rolling stock is mustered out and, HSR traffic becomes heavy enough to justify having four platforms at which bullet trains may stop. Additional FRA approval may be required.

    --- track 4 for either a BART shuttle or preferably, for a new Caltrain service between Gilroy/SJ and SFO. The latter would require variable gauge technology and BART-specific in-cab signaling upgrades to selected non-compliant Caltrain EMU equipment, plus approval from both BART and FRA.

    Might require transition from overhead catenary to overhead conductor rails inside SFO station to cope with low ceiling.


    Like your option D, this option E requires ripping up some BART tracks and sorting out how standard and broad gauge tracks will run without crossing each other north of the station.

    If BART service to Millbrae is discontinued entirely, that has implications for overnight storage of trains, both BART's and those used for the new Caltrain service to SFO suggested above.

  2. Rafael,

    Some thoughts in response to your comments.

    - Don't be fooled by the "bullet" in baby bullet. It is a marketing term used to sell plain old skip-stop express service, not a real "bullet" train. It is very important that Caltrain maintain skip-stop express service on the peninsula even after HSR begins operating. Operating skip-stop express service works best when locals and expresses can perform cross-platform connections, hence my perceived need for 4 platform faces at Millbrae.

    - The EMUs can handle Caltrain express without a requirement for yet a third type of equipment.

    - HSR may not always run on the inside tracks throughout the corridor, even if the CHSRA has planned it that way in the preliminary work.

    - BART to Millbrae will be retained: it provides connectivity from the south bay to the entire western portion of San Francisco not accessible by Caltrain. While it was implemented very wastefully, the Caltrain-BART connection does have some non-zero value.

    - Speaking of service: Running some Caltrain services into SFO, with all the technological complications that implies, is a ridiculous proposition in my opinion-- a poor solution to the real problem.

    This is the real problem with the Millbrae SFO connection: it is impossible to get a single-seat ride from the peninsula rails to the airline of your choice. It was possible before the Millbrae lemon was built: a free shuttle van met every Caltrain and dropped you off at the airline of your choice. It's amazing to me that several hundred million dollars later, Millbrae today requires consecutive rides on TWO BART trains (for a fee) and then a third ride on Airtrain, to achieve the same end result. It's totally impractical and nobody in their right mind probably does this.

    What is needed (and what existed previously) is a SINGLE SEAT, FREE ride from rail services (be it Caltrain or HSR) to the airline of your choice (NOT just the international terminal served by BART-- ALL the terminals). The least-terrible way to achieve that, given the existing mess, is to run Airtrain right down to Millbrae.

    - your option E will be quite challenging for crossing track 5 over track 4, considering that you will have to run freight trains around and then over the BART tunnel box. Option D is far simpler and easier in that respect since BART stays physically segregated with no need for yet more concrete pouring on that side of the station.

  3. For the record, there is no problem with running express service at 125 mph adjacent to platforms. Amtrak operates Acela Express service adjacent to platforms at 120-150 mph without special measures (beyond a system announcing the approach of a high speed train) at the following stations:

    New Carrollton, MD (120 mph)
    Seabrook, MD (125 mph)
    Bowie State, MD (125 mph)
    Odenton, MD (125 mph)
    Aberdeen, MD (125 mph)
    Kingston, RI (150 mph)
    Attleboro, MA (135 mph)
    Mansfield, MA (150 mph)
    Sharon, MA (135 mph)
    Canton Junction, MA (135 mph)
    Readville, MA (120 mph)
    Hyde Park, MA (120 mph)
    Forest Hills, MA (120 mph)

    Some of these are fairly busy commuter rail stations (e.g. New Carrollton, Forest Hills).

  4. Option D was the option that I proposed in the other Millbrae blog and I agree with Clem in that it will be the least destructive and therefore least expensive solution. Sacrificing cross platform transfers between BART and Caltrain isn't the worth thing in the world, there could easily be coordination between BART and Caltrain to allow time for passengers to get up the stairs on the platforms and down the other stairs and get on the train. Like I said before, using this configuration maintains the Express trains down the center tracks without cross-overs.

    @ mike -

    Great info! It sounds like running a HSR train at 125 mph through a station is definitely possible utilizing the system used on the Japanese systems w/announcements at the stations.

  5. @ Clem -

    running the northbound Caltrain track wouldn't be a problem at all, it would just split off from the other three south of the BART tail tracks.

    Between Lions Gate Park and approx. I-380, the only way to grade separate the standard gauge tracks is a gantry-based aerial anyhow. There's no avoiding additional concrete there.

    Note that your option D also usurps a BART track for northbound service. If you look closely at the satellite pix of how the tracks run where the southern ramp veers off to SFO, you'll see the problem with that.

    Option E with a plain old BART shuttle to SFO would be easier than noodling with variable gauge technology etc. The main issue is BART's ticketing system, which - as usual - is incompatible with everyone else's.

    Another issue is that anyone coming up the peninsula would still have to change trains twice before reaching check-in. If special Caltrain gear could run directly into SFO as a guest of BART's, that would be both faster and cheaper than running the AirTrain across to Millbrae.

  6. I love Option D (note: typo in Option of Option D on image) - regain the unused BART track, good local vs express std gauge transfers (whether Caltrain -> HSR or local -> skip-stop), minimal intrusion/demolition. Add it AirTrain coming down from SFO to the mezzanine level (does anyone have a floor plan of that level?) and this would be a right reasonable solution:

    * Anything <-> SFO, no problem. Go upstairs and catch the AirTrain. If you happen to catch the Millbrae train rather than SFO train from the East Bay it's a non-issue.
    * HSR <-> Caltrain, no problem, right across the platform.
    * Caltrain/HSR <-> BART - a pain, you have to go up to mezzanine and back down. This seems ok to me - once HSR and Caltrain go to the Transbay Terminal, there will be less interest in switching to BART (as if there was now) cause the train will be significantly faster - 13 mins vs 31 mins. Now if they can provide a nice tunnel to Montgomery this would be killer.

    So the real X-factor in this to me is the AirTrain. That would make this station really click but I don't think it's on anyone's radar. It really should be included in this project - it would really be the icing on the cake.

  7. @ Rafael -

    Regarding freight, isn't that a problem down the whole length of the system? If/when Caltrain goes to non-compliant stock, isn't that and FRA problem here or anywhere else on the line? Wouldn't tracks 1 & 4 in Option D be the same thing as you are referring to as 1 & 5 in your option?

    I guess I don't understand the objection.

  8. @ timote -

    I recommended track 5 for northbound local Caltrain service because it runs east of the ramp tracks toward SFO. Similarly, track 3 runs west of them. That would allow BART to operate a Millbrae-SFO shuttle without any crossing of broad and standard gauge tracks.

    If BART decided it did not want to operate that shuttle for whatever reason, then the idea of a new Caltrain service from Gilroy/SJ to SFO using variable gauge etc. would kick in, using track 4 at Millbrae as part of a short single-track section. On the southbound leg, such a service would have to cross the HSR tracks, though.

    Wrt FRA mixed traffic rules, Caltrain is applying for a waiver to operate both its FRA-compliant legacy diesel and new non-compliant EMU rolling stock for a transition period. In addition, UPRR still runs the occasional heavy freight train up the peninsula.

    That can be dealt with safely via guaranteed time separation, but engineers will have to be mindful of gradients and load carrying capacity when they design grade separation solutions, e.g. between Millbrae station and I-380 and, in Mountain View. If might help if UPRR agreed to deploy only lighter, European-style freight trains in the SF peninsula or, it gave up trackage rights between Santa Clara and SF altogether (with some financial compensation).

    The only other section of the starter line in which mixed traffic will be an issue is between Fullerton and Anaheim. Metrolink uses FRA-compliant rolling stock and the ROW in this relatively short section is too narrow to accommodate four tracks side-by-side.

    If FRA has a serious problem with that, one option would be to lay some new tracks from Redondo Junction to Anaheim ARTIC via Paramount, Stanton and a tunnel under Disneyland. A spur between Paramount and Long Beach airport might be possible as well, though it would require the construction of an underground loop past the terminals there.

    Separately, there is also a parallel ROW right next to I-5 between Cudahy and Disneyland; it may still be in use by the Seal Beach Naval Weapons Station.

    In the SF peninsula, there really isn't an alternative to the Caltrain corridor, especially since the option to run trains up the East Bay and then across to SF was rejected. Running tracks across the Bay Bridge again has been rendered all but impossible by design decisions for the new east span and, a second transbay tube is unlikely to be built anytime soon.

  9. Instead of a tunnel under Disneyland, the 5 in that section has enough room for 12 lanes but only has 10, and probably wouldn't expand to 10 because the widening of the LA county expansion looks like it is going to be 10 lanes.

    It seems like that would be a lot cheaper than a tunnel

  10. A route along highway 5 sounds like a good solution to. It is a very straight route that could easily cut through Anaheim without expensive tunneling.

    Another question: It sounds like the LA area route is much less developed and more complicated than the SF-SJ route. I know we were talking in other threads about how the Merced-Bakerfield route will be built first but we also discussed the possibility of shorter routes in LA-Anaheim and SF-SJ opening up early. Does it make more sense to put all of the projects resources towards finishing the SF-SJ route first and get it running because the plan is more developed? It seems to me that trying to juggle the construction of both the LA and SF-SJ routes will be slower than just concentrating on getting the more developed route into an operating, paying route.

    Sorry for being completely off topic but the previous posts led me there...

  11. Looking at the corridor, I'm thinking HSR should NOT be running on the inside tracks, even though that's the conceptual layout. HSR and express tracks fit so much more neatly on the west side, for long long distances.

    The catch is that you want to be able to switch from local to express tracks, and at *those* stations you don't want to foul the tracks going in the other direction.

  12. Alex wrote: "It sounds like the LA area route is much less developed and more complicated than the SF-SJ route."

    Not my impression. Less developed, yes. But also a *lot* less complicated. Land use patterns in LA mean that there is width for a four-track line without encroaching on any expensive property, pretty much from Union Station all the way to Bakersfield.

  13. @ Ben -

    the whole point would be to have a station at Disneyland, one of the main reasons for people to take HSR to Anaheim in the first place.

    A Metrolink line from LA US via Huntington Park, Paramount, Artesia, Cypress, Stanton and Disneyland to Anaheim would permit an intermodal with the Metro Green Line and free up the tricky Fullerton-Anaheim section for HSR. It is too narrow for four tracks and FRA might yet have a cow about mixed traffic in that short section.

    Note that there is a freight track just west of I-5 between Norwalk and Disneyland. It may or may not still be in use for supplying the Seal Beach Naval Weapons Center.

    @ Alex -

    Merced-Bakersfield is needed to secure FRA approval for running non-compliant off-the-shelf bullet trains from Europe or Asia at 220mph. The agency doesn't even have rules for speeds that high yet. It is also needed for vendor pre-qualification.

    SF-SJ and LA-Anaheim will likely be the first sections to see commercial service. LA to Bakersfield involves negotiating narrow ROWs in the LA basin and then a lot of tunneling. Same for SJ to Chowchilla/Merced.

    @ anon @ 1:41 -

    track fouling is indeed why CHSRA would like to stick with inside tracks all the way to SF. If nothing else, FRA may require it because Caltrain will continue to operate some legacy FRA-compliant equipment alongside its new non-compliant EMUs for quite some time (assuming it gets its FRA waiver to do so).

    In other words, contributors to this blog may want to adjust their point of view. The HSR tracks are for fast trains, they need to be as straight and uncomplicated as possible. Caltrain tracks are for slower trains and have to work around that constraint. If that means moving stations or reconfiguring them, so be it.

  14. Rafael,

    > If BART decided it did not want to operate that shuttle for whatever reason, then the idea of a new Caltrain service from Gilroy/SJ to SFO using variable gauge etc. would kick in

    I'm not sure if I'm making myself clear. The idea of running BART or Caltrain (!?!??) from Millbrae to the SFO International Terminal is inferior to simply running Airtrain out to Millbrae. The key requirements aren't technical gadgets like variable gauge; they are transportation requirements like a one seat, free ride to any SFO terminal from any train serving Millbrae.

    Driving it home once more:
    (1) one seat ride
    (2) free
    (3) any airport terminal
    (4) any train serving Millbrae

    With respect to the Millbrae-SFO connection, your suggestions achieve none of the above.

    > In addition, UPRR still runs the occasional heavy freight train up the peninsula.

    The freight trains are more than occasional. There are typically two round trips every night.

    > track fouling is indeed why CHSRA would like to stick with inside tracks all the way to SF.

    There is no track fouling with HSR on the outside tracks, if Caltrain stations have island platforms in the middle. As Richard M pointed out in another thread, the greatest benefit to Caltrain is to be able to change tracks (during the all-too-frequent service disruptions) without fouling HSR traffic. I'll come back to this good idea in a future post.

    Thanks for your thoughts.

  15. @ Clem -

    maybe I'm the one not being clear: unfortunately, I just don't think the AirTrain will ever be extended to Millbrae. They'd have to build a new bridge/ramp across 101, run it along the western edge of the freeway and approach Millbrae station from the east. It's quite unclear who would pay for that.

    Messing with the ramp built for BART would be difficult (seismic safety) and eliminate BART access to its current overnight parking spaces south of Millbrae.

  16. @ anon @ 1:41, Rafael,

    Then which is it? Rafael, you are saying the LA route is difficult and will require a lot of tunneling because the ROWs aren't big enough. Anon, you said that the LA area is simple with wide ROWs that will easily accommodate 4 tracks. Just taking a cruise around LA in Google Maps, I don't see very much space in most of the areas that the HSR line is proposed to be going.

  17. SFO supports HSR and John Martin, the head of SFO, worked with his peers to support 1A. SFO will gain a lot of traffic from HSR, especially if an airline is involved in operations.

    SFO makes money. It can extend AirTrain. A ticket surcharge on HSR at Millbrae could help fund the extension. The airport could do it on its own. It is in SFO's interest and the interest of it's clients, the airlines, to make the connection between HSR and the terminals as easy as possible. That looks to be an extension of Airtrain to Millbrae.

  18. Absolutely agree with Clem - by far the best airport connection solution is to extend AirTrain to Millbrae.

    Presumably the cheapest way to do this is to convert the southern section of the SFO Wye to AirTrain and then build AirTrain at grade along the Caltrain ROW to Millbrae. I have mapped this as Alternative A. The good news is that there is currently ~100' between the eastern Caltrain track and Monterey St., so there might actually be enough space for 4 Caltrain/HSR tracks + AirTrain (though the BART tunnel machinery buildings will need to be rebuilt).

    A faster route for AirTrain would be to build a wye at the southwestern point of the AirTrain system and connect to the Caltrain ROW from that point. I have mapped this as Alternative B. I assume this is more expensive, but maybe not. It also has a key operational advantage: unlike in Alternative A, AirTrains never need to run in opposite directions on the same track. (In Alternative A, at the point marked Bottleneck, Millbrae-bound AirTrains must cross the southern track running westbound while AirTrains from the rental car station run eastbound on the same track).

    IF there is insufficient political or financial resources to extend AirTrain to Millbrae, then I think that one aspect of Rafael's plan does merit serious consideration. Specifically, the unused southern section of the SFO Wye could be converted to standard gauge track with 25kV AC catenary. There are 3 BART tracks from the SFO Wye to SFIA Station, of which BART currently uses one. The southern track from SFO Wye to SFIA Station could be converted to standard gauge and then Caltrain (or another operator) could run limited stop, high speed service from SJ Diridon to SFIA (e.g., SJ-Mtn View-PA-San Mateo-SFIA), perhaps every 30 minutes during the day. This would be similar to Heathrow Express and Gatwick Express in London.

    Note that there would be absolutely no mixing of BART and conventional rail equipment. There is just no way this will never happen. Even ignoring the incompatible track gauge and regulatory/crash compatibility issues, there is no way that BART would ever want to expose its entire system to the delays that would arise because of late Caltrain/HSR movements that would block arriving/departing BART trains at SFIA. It is like asking BART to build some grade crossings somewhere on its line. Cannot dispatch with them. Cannot run on-time with them. Cannot maintain reliability with them. CAN'T DO IT.

  19. > Messing with the ramp built for BART would be difficult (seismic safety) and eliminate BART access to its current overnight parking spaces south of Millbrae.

    Rafael, would it? SFO could partially rebuild the southern ramp (over 101) to connect ~5 m higher with the Airtrain. The rubber-tired equipment used by Airtrain can climb much steeper grades than BART, so only a limited portion of the ramp over 101 would need to be modified in order to connect to Millbrae. Then, Airtrain would run on a new elevated guideway along the future HSR tracks, terminating in a second Millbrae station mezzanine built to the north of the existing one, with a direct escalator connection to the HSR platforms.

    The BART tail tracks in Millbrae would still be accessible from SFO by reversing at San Bruno (as hapless Millbrae-to-SFO passengers currently must!). Besides, there are three tracks in the SFO BART station, one of which is never used and can serve as parking if needed.

    That sounds a lot easier to me than running variable-gauge who-knows-what only to the international terminal, to achieve exactly what suboptimal service that doesn't already exist? ;-)

    Whatever you propose, remember to maximize concrete pouring. That's how Millbrae got built the way it is.

  20. @ Alex -

    tunneling is required in Soledad Canyon and Tehachapi pass to achieve sufficiently high speeds, not because the ROW is too narrow. It is tight between LA US and Sylmar, but CHSRA figure they can squeeze in there. If necessary, they'd stack their tracks on top of the existing freight tracks rather than tunnel there.

    @ mike -

    thx for the map, I see now what you had in mind. Adding those wyes would be possible but non-trivial. The supposedly cheaper option of re-purposing the southern BART ramp could actually end up costing the same because constructing a ramp to bridge that elevation difference while BART continues to operate is a complex task.

    Btw, I did consider the impact of tricked-out Caltrain EMUs serving SFO. While on broad gauge tracks, it would be under BART dispatch control via the same Alcatel box installed in every BART trainset. It would be no different than to old BART shuttle service except that if a Caltrain misses its "slot", it will have to wait on the ramp for a while.

    Caltrain would have a huge incentive to ensure its operations into SFO did not negatively impact BART's clockwork scheduling, because BART could and would deny guest privileges if it happened too often. There a similar situation in Tokyo, where some private regional railroads have trackage rights on city-owned subway lines and vice versa.

    Crash compatibility I'm not too worried about, since the new Caltrain EMUs will be lightweight non-compliant gear and feature all the active safety equipment found on BART trains. FRA may well want to see some crash simulations, but Caltrain already has some experience with those in another context (see appendix C).

    @ Clem -

    "Whatever you propose, remember to maximize concrete pouring. That's how Millbrae got built the way it is."

    You do realize that my proposal calls for virtually no concrete pouring whatsoever, don't you? No new tracks, no new ramps, just some sleepers. Well, that and raising and lengthening the HSR platforms, but that's unavoidable in any case.

    Anyone on the BART network would still get directly to SFO, only now anyone approaching from the south on Caltrain would as well. Only those who use HSR would have an extra transfer. It would be easy enough to tack the cost of that onto the HSR ticket so it is perceived as a courtesy ride.

    Bringing the AirTrain to Millbrae would require quite a bit of concrete pouring (see Mike's maps) just to save HSR customers - and only them - one transfer. The AirTrain's fairly slow, too. Even if SFO decides to shell out for the extension and raise airport taxes to pay for it, I'm not sure it would be as big an improvement as you claim.

    In terms of bang for buck, I suspect none of these flights of fancy will beat re-instating the trusty old SamTrans courtesy shuttle bus.

  21. > In terms of bang for buck, I suspect none of these flights of fancy will beat re-instating the trusty old SamTrans courtesy shuttle bus.

    Amen to that, brother! I miss it.

  22. @ rafael

    Well the extension of airtrain would service caltrain (ore-electrification vehicles), Millbrae BART (a fairly minor point considering you really have to catch the wrong line, but it would allow for god-forbid BART extension), HSR (non-trivial use case considering the length HSR is going to to pick up airports), and bus (I don't know if Millbrae is a bus hub, but most BART stations are.) Plus you can make revenue off that big parking lot all of a sudden for super easy long term airport parking.

    The tech spec says top speed of airtrain is 30 mph, not bad for this purpose. I bet it seems slow cause of the frequent stops...

  23. Errr that was supposed to say pre-electrification. Dang phone spell correction :-)