13 August 2011

Corridor Capacity Study, Free Edition

You may have heard that Caltrain is working on a corridor capacity study, to see how much high-speed rail traffic could be accommodated on the corridor without adding too many passing tracks or destroying commuter service quality.

While we wait for the outcome of this study, below is a decomposition of the problem into a range of possible solutions.

Some solutions have very little downside for HSR or Caltrain service:
Other solutions require compromise between the needs of high-speed rail passengers, the needs of commuter passengers, and the needs of surrounding communities.
  • Slow down HSR, with lower top speed and all stops (Millbrae and Redwood City) made by all trains
  • Close low-traffic Caltrain stations such as Atherton and Hayward Park
  • Make peak-hour trains skip more stops-- which unfortunately denies frequent commuter service to communities precisely at the times when it is most needed
  • Build additional grade separations wherever four tracks are required
Before we make any of these painful compromises, however, the most effective measures that do not require compromising service quality should be prioritized and vigorously pursued.

If one had to bet a six-pack of one's favorite microbrew (worth several orders of magnitude less than LTK Engineering's consulting contract), the outcome of the capacity study is likely to be:
  • With 6 Caltrains per hour during the rush, spare capacity available for HSR is minimal on the existing tracks
  • Capacity can be increased most effectively by matching train average speeds, i.e. slowing down HSR and/or speeding up Caltrain
  • HSR trains will probably average no more than 60 mph between SF and SJ
  • A mid-line overtake facility between Whipple @ Redwood City and 9th Ave @ San Mateo (requiring minimal new grade separations) would improve peak corridor capacity from 6 to 8 tph
We shall soon find out if we got our money's worth.


  1. Ah yes, LTK Engineering: The same bozos that recommended FRA-compliant DMUs for 'SMART'. Who better to help determine Caltrain's "PTC/CBOSS requirements".

    Looking forward to yet another useless "study"...

  2. Some minor quibbles re your nice diagram:

    * If "Close low-traffic stations" means "Hayward Park, Atherton and College Park are toast" then this is a no-downsisde winner for both peninsula Caltrain riders and for any rail operators on the corridor. No compromise involved.

    If it means "run a brain-damaged service pattern (like today's Caltrain, or like Caltrain staff proposed skip-stop nightmares" then it is indeed a compromise -- one that isn't necessary, of course.

    * "Automatic train stop assist" (ie docking at fixed platform locations) is a bit if a stretch. It will be decades, if ever, before regional traffic on the corridor could benefit from this level of operating discipline, and it certainly has its downsides (being a technical pioneer, another system that can fail, etc.) Plenty of cost and some risk, near-zero benefit.

    There's a care to be made for 400m trains attempting to ooze and squeeze into 400+epsilon dead-end platforms at Transbay only, but that's a ways off. (And the real solution is 150m unit double-deck trains operated in 300m tandem anyway!)

    * "Lower speed limit from 125mph to 100-110mph" also doesn't seem like it has any downsides to me. The effect on HS traffic, especially given unmitigated speed restrictions along the corridor, is minimal to the point of being almost unmeasurable in practice. The upsides (capacity provided for a given level of infrastructure, reduced noise, possibly reduced environmental mitigation engineering, possibly reduced regulation, certainly reduced amounts of expensive passing tracks) are numerous and far larger.

    * LTK. Good God. America's Finest Transportation Planning Professionals. Drunk Engineer (I'll buy you a beer anytime) knows the score, and that's not half of it. Where do they even find these people?

  3. Adirondacker1280013 August, 2011 22:47

    Lower speed limit from 125 to 100-110

    Or let the commuter trains go 125.
    The locals will be poking along on the local tracks. The express trains can surf along at 125, the same speed as the HSR trains.

    Automatic stop assist to line up doors and eliminate platform overshoots

    Line up with what?
    Is overshooting the platform such a serious problem that it needs an expensive technological fix?

    would improve peak corridor capacity from 6 to 8 tph

    OOO! Two trains an hour! Makes all the deliciousness of service to Sacramento and commuter service from the East Bay rather difficult doesn't it?

  4. On the topic of cal train capacity, I noticed that the new Santa Clara station is being built to allow 3 tracks between the existing southbound and new northbound platform, and the tunnel is long enough to fit another track if the existing platform is moved. Quad tracking the section from Lawrence to College Park would allow more effective overtaking and require no new grade separations and only two new bridges.

  5. Or let the commuter trains go 125.
    The locals will be poking along on the local tracks. The express trains can surf along at 125, the same speed as the HSR trains.

    The issue is that express trains make at least 5 stops along the corridor, HSR makes 0-2. I'm not saying that 125 won't work (it can if you fool around with the schedules enough), but the regional expresses will have to spend some time on the local tracks.

    Is overshooting the platform such a serious problem that it needs an expensive technological fix?

    No, but a computer could probably approach the station faster and still be able to hit the platform accurately.

    That is, unless you use BART's control system apparently.

    In any case, it doesn't seem like it would provide much benefit unless the corridor is heavily congested.

  6. The locals will be poking along on the local tracks.

    Yes, but the point of this exercise is to minimize where you have local / express tracks. None of this would be a challenge with four tracks everywhere. Work harder, not smarter?

    Is overshooting the platform such a serious problem that it needs an expensive technological fix?

    Yes. As a daily rider I have witnessed many a platform overshoot, so this is statistically a not-insignificant source of delays. Not that it would show up in today's on-time statistics, which consider anything within 5 minutes of the generously padded terminus arrival time to be "on-time". But if you're doing timed overtakes without four tracks everywhere you can't afford this sort of sloppiness. This is not cutting edge technology.

  7. Adirondacker1280014 August, 2011 11:50

    You don't need a computer to spot a train at the platform.

    As a former daily rider of the NYC Subway, the one with the sophisticated signaling system originally installed in 1904, I can tell where the conductor will be. At the striped board. They do it thousands of times a day, Everyday. Since 1904. I'm gonna go out on a limb here and suggest that they acquired those skills on the El which had been doing for decades when the subway opened. In 1904.


    It's been my experience that those same skills are demonstrated on railroad systems all over the world. Maybe it would be cheaper to send Caltrain staff for some training.

  8. Adirondacker1280014 August, 2011 13:35


    With a bit of an explanation.

    ....and line up with what?

  9. I do not commute on Caltrain and in 20 years living on the peninsula I really cannot recall experiencing or witnessing over-runs; maybe one. The bike riders wait where the bike car will stop. The wheel chair bound wait next to the ramp. Then engine spots next to the sign. I understood Clem to mean that an automatic system would reduce an already small risk. Thanks Adirondacker for yet another poke in the eye. Can't wait til next time.

  10. Re platform overshoots: they're common at Caltrain. Crews do whatever they can to not have them broadcast and recorded as incidents, and the slack schedule adherence reporting standards aid this. I never kept track, but I'd make a wild guesstimate that this happens in on over 5% of Caltrain runs (based on very approximately one episode every couple weeks, 10 trips/week, on a trip covering ~60% of the stops.)

    Two causes: failing to apply brakes at the correct distance before an expected stop (cause: lack of attention; lack of training; lack of organizational discipline); failure to remember that an approaching station is a stop (causes: ditto.) In short, a Lack of Situational Awareness -- not encouraging, all things said.

    Fixes: less arbitrary timetables with more regular stop patterns (better for crews AND passengers); better training; better supervision; better in-cab data.

    Any post-stone-age train-driver interface will show aspects of the next several approaching signals; speed limits as dictated by civil and signals; braking to comply with those speeds; next station stop; next station stop time; and time variance (seconds ahead of or behind schedule) of the current train run from the system timetable. This is pretty much 1990s technology. Expect at CBOSS implementation by 2060.

    Going from this level of routine tech (driver assistance, machine-enforced positioning at ~20-30m accuracy) to platform docking precision (~1m positioning) is a fairly big and non-standardised step today.

    TMI on a minor point!

  11. The most cost effective incremental Peninsula Rail Corridor improvements will enable faster, more reliable and frequent service that generates less noise; all without increasing visual blight. The first major improvement should, as Clem suggested, start by adding express tracks past the already mostly grade separated San Carlos through Hayward Park station segment. Note that extending passing tracks beyond 6,000 feet for accelerating from a station or 3,000 feet for braking before a station will not reduce local-express train conflict even for 110 mph peak speed local trains. This passing segment is near the center of local-train-only served station locations on the line. The foregoing express track enhanced segment could accommodate, without mutual interference, two 25 Kw/metric ton peak traction power 1000 foot long all-stop trains, eight similar one stop trains, and at least 12 non-stop trains per hour using a four-aspect 2,100 foot block length signal system protecting a 2.1 ft/sec2 assured braking rate 110 mph track.

  12. Passing segments need to be longer than one station. Unless you want to hold trains for longer than they would normally stop (this may be necessary under some schedules but not desirable if it can be avoided).

  13. Another possible solution. One wonders if HSR trains could abosorb some baby bullet commuters, thereby reducing the number of peak hour Caltrains. This would require close corridation and careful planning of where to seat people (i.e. so commuters could find plenty of seats in say 2 cars of the HSR trains). It would also require allowing HSR trains to stop at mulitple Caltrain stops, rather than HSR-only stations.

    For NB trains, long distance riders exiting the HSR train in SJ or mid-Peninsula, would be replaced by commuters. For SB trains, commuters getting off at mid-Peninsula stations/SJ would be replaced by long-distance riders in those locations. HSR makes more money and capacity if freed up.

  14. Overshoots happened to a train I was riding twice within a year, and I'm not even a daily rider. More like on weekends every other weekend.

    Once was a Metrolink train that forgot to stop at Tustin and another was a Surfliner that forgot to stop at Norwalk, which only a minority of Surfliner trains stop at.

  15. @Daniel: that's a fine idea, if commuter fare media are made valid on high-speed trains. It also has the nice benefit of forcing platform interface compatibility between HSR and Caltrain, something I've been harping on for years.

  16. Adirondacker1280015 August, 2011 20:25

    It doesn't force any compatibility between HSR and Caltrain. It just means that the commuters waiting on the island platform that serves both kinds of trains can get on either. And the people traveling south of San Jose will love all the empty seats.

  17. It's not a bad idea, but in order to substitute a full express run, a HST would have to be at least half empty by the time it reached San Jose.

  18. Re "One wonders if HSR trains could abosorb some baby bullet commuters, ..."
    Alon got there (though he described it as "trolling", which barely seems the case) long ago.

    I don't believe it is terribly feasible for a number of reasons.

    Two alternatives.

    Either San José (the Capital of Silicon Valley!) is so important that exotic 400kmh-capable super-cost HSR trains via Los Banos run largely empty between provincial SF and teeming, bustling, TOD-tastic SJ, with plenty of seats available to soak up regional commuters. (This being the PBQD/MTC/SVLG/VTA/C4HSR model of the universe.)

    In this case I suggest that it is a mistake to not terminate and turn many or most of those exotic HS-capable trains in SJ, and instead use cheaper, more appropriate, higher capacity and less-overengineered Caltrain equipment north of SJ. Yeah, we all know the jive about "transfers kill ridership", but backing that up with cold numbers in the face of hundreds of millions of dollars of commercially-financed and rapidly-depreciating exotic rolling stock is another matter.

    The other possibility is that Los Banos-SJ-SF trains end up with few empty seats after pausing at Diridon Intergalactic to watch the tumbleweeds roll past.

    In which case, well, somebody is in fraudulent receipt of $15 billion or more of public cash.

    In short, not an infeasible operating plan in terms of getting trains from point A to point B, but it has considerable economic and/or practical passenger capacity problems.

  19. Well, Caltrain now has a place on their website for a very very early summary of preliminary results, almost completely devoid of detailed technical information.

    We want a timetable! That is the product, far more important than a PowerPoint.

  20. By overshoots, do you mean missing a station completely (I'm a sorta regular Caltrain rider whose never seen that) or the more common scenario where the train overshoots, then has to back up 20-30 feet ish?

  21. The latter. It can consume 1-2 minutes that can ruin a timed overtake opportunity by delaying other trains. This sort of unpredictability is why the Caltrain planning folks abhor overtakes. I'm thinking it makes more sense to fix the cause (long and unpredictable dwell times) than the syptom (botched overtake timing)

  22. Much, much worse cause for unpredictable dwell times is the complete lack of level-platform boarding. One single wheelchair is all it takes to screw up a schedule.

    Too bad they didn't fix that problem when they had the chance.

  23. The meeting of the San Mateo group yesterday (8-17-2011) had the slide show presented on the capacity study.

    It was pretty obviously more of a PR presentation, than mush else. Today the press is full of all the claims that this will solve all the problems on the peninsula.

    This is all predicated on CalTrain needing 6 trains, each way, during rush hours.

    Without additional passing tracks that leaves at most 2 trains per hour for HSR, each way. Apparently all based on CalTrain staying at 79 MPH and HSR up to 110 MPH.

    They then propose a 7 to 8 miles set of 4 tracks and studies from Heywood Park to San Carlos. (note Belmont is in that stretch and Belmost is not happy; they are furious)

    The present CalTrain timetable has 5 trains each hour leaving San Jose between 7 and 8 and 4 trains per hour leaving SF between 6 and 7 and 5 trains per hour between 7 and 9.

    So, I guess CalTrain figures they can get by with only 6 trains per hour for quite some time. (during rush hour.)

    UPRR and freight ignored, except when asked they had to admit the UPRR wants to hold onto their present agreement that says they can run freight after 5 PM.

    So it is early. Simitian is out saying this is the solution. vanArk saying its looks good and worth investigating.

  24. This comment has been removed by the author.

  25. Has the CHSR-Caltrain Consortium really paid LTK Engineering to analyze the resultant capacity when mixing 110 mph HSR trains with new electrified 79 mph Caltrain on a mostly two track corridor where capacity and robust scheduling is in question? The combination of more frequent stopping and 31mph lower top speed for all Caltrain runs than hsr will drastically reduce potential capacity and schedule reliability. My August 14th comment outlined where at least 20 trains per hour could shift their order with a 4 station length passing track if all trains maximum speeds were 110 mph. If Foucault Magnetic Brakes are active where deceleration rates are completely independent of surface friction are developed to the point where maximum assured braking rates were increased to the maximum normal service braking rate = 4.4 ft/sec2 the time between 110 mph trains could fall to 30 seconds. (Note: Quick recovery from delays is more likely if scheduled frequency is below track capacity. Combining queuing delayed trains or activating spare trains stored between the local center tracks are two other low cost techniques for enabling prompt recovery to scheduled frequency after an en-route delay.)
    Given the lightly used often skipped stations in the 12 mile gap between Millbrae and 22nd Street where trains are near their peak loading Caltrain passengers’ average speeds would increase significantly if Caltrain peak speeds approached those for hsr. Most of the development cost for safe operation for 110 mph passenger train operation is spent on improving track quality. Buying new Caltrain equipment that will not fully utilize future 110 mph track ignores a tremendous speed, capacity and reliability Peninsula Rail Corridor improvement for little additional cost.

  26. For example: Redwood City platform is 600' long. The consist (specifically including the engine) is 480'. The wheel chair loading platform is 16'. This discussion made it sound like the 480' train was overshooting the 600' platform a target of about +/- 60'. Just wondering approximately how many of the gross overshoots are over by due to trying to line up the door within a fraction of 16'? (All the more reason to get level boarding platforms I know.) Since I have zero experience with rail operations can anyone comment on what the crew has to do to hit the mark on the platform? It ist some combination of regenerative braking with applied air brakes? Is it primarily the skill of the crew? Minimum dwell time would require accurate operation every time. Maybe in 40 or 50 years room-temp superconductor mag-lev trains will be common and stop on a dime but does the Caltrain crew work with traditional diesel-electric braking?

  27. @James: for a trained train operator, stopping a train isn't any more complicated than stopping a car. The key is to remember to initiate the stopping process.

  28. I believe in versatility, i.e. common platform height/level boarding, common signal system, any train, any track, any station, compatibility with freight, standard trains, and steam excursions. I also believe in cost containment, something totally foreign to the HSRA, as profit seeking contractors appear to be operating/driving the train. HSR does not need to be so gold (BART) plated.

    Regarding “commuter service quality” / “metrics that matter,” convenience plays a major role in attracting riders, more so than the speed of the service. People do not like to wait and I have heard transit experts say that people perceive the waiting time as 2 or 3 times more than it actually is. People would rather be on a moving vehicle even a slow one. People also do not like to transfer which adds additional time and expense to their commute. Ridership to San Francisco would be significantly higher if Caltrain dropped customers off in the downtown CBD.

    The system also needs to be convenient throughout the day and in the evening in order to provide the commuter service quality customers seek; weekend service must also be convenient. Sure the service may be good during peak hours but if a customer has to leave early for some reason, and then they have to deal with one train per hour, if you stay late its even worse, not the best way to gain loyal customers. Granted the current peak hours has some gross holes at some station pairs which can be solved by better operating practices, electrification, and an upgraded signal system. Convenience and waiting time should carry more weight than time on the train itself; however we must not dismiss the value of express trains, since people really do like them… kind of an oxymoron LOL. Also we shouldn’t dismiss seats per hour, while the customer might no care so much about seats per hour, they do want to have a comfortable ride and that means providing enough seats for the expected ridership.

    There are important metrics missing from your formula such as costs. Cost of gasoline has been a driving factor, in increasing Caltrain ridership throughout the years. I have been analyzing Caltrain (Southern Pacific) ridership, schedules, and fares for nearly 30 years, and increased gas prices have a profound influence on increased ridership. As gas prices stabilize and or decrease, the ridership did not drop accordingly, indicating that once people try Caltrain (public transit), they find that it is not so bad, they can still save money and they stay with transit.

    Historically, large losses of ridership on Caltrain are driven by hefty fare increases/loss of convenience trains (causing a large ridership loss in 1980-1983), 911, dot com bust, poor economy, weekend shutdown, etc. were responsible for more recent ridership losses.

  29. In 5 years of riding the New York City Subway regularly - of which it was my primary method of commuting in 2 - I do not recall having ever seen an overshoot. And, on the legacy IRT, the system I took the most often, the platforms are barely any longer than the trains.

  30. Lobotomyoriffic.

    Combines all the following highly desirable US public transportation attributes:
    * Poor service.
    * Ridiculous service gaps.
    * Bizarre stopping patterns.
    * Insane capital costs.
    * Incompatible stations.
    * Tunnel under Millbrae.
    * Pissing away money in SJ.
    * Poor equipment utilization.
    * Hugely negative return on investment.

    Let's see if LTK Engineering Services with their PCJPB partners -- America's FInest Transportation Planning Professionals together combing their skills and insights -- can manage to do even worse.

    It's a race to the bottom!

  31. Richard you left out a very important part.

    All this being led by by those experts who sit in Sacramento or Washington and know how to solve these problems.

    Joe Simitian
    Rich Gordon
    Anna Eschoo

    Right behind them and pushing are Larry Patterson of San Mateo and Wunderman, and dozens of other sheep from the Democratic caucus.

  32. Adirondacker1280020 August, 2011 13:29

    In 5 years of riding the New York City Subway regularly....

    Union Square on the IRT and the loop at South Ferry - without computers....


  33. Hell, I rode the Berlin U-Bahn every day for thirteen years and never saw an overshoot.

  34. "for a trained train operator, stopping a train isn't any more complicated than stopping a car. The key is to remember to initiate the stopping process."

    Ah! Forgetful engineers-- so that's the problem. Should be easy to fix.

    Caltrain trains typically take around 85 seconds for the last mile to a stop, which is... maybe 10-15? seconds more than Amtrak trains betw Emeryville and Sacramento. Think Caltrain engineers are more forgetful? All of them?

  35. "Think Caltrain engineers are more forgetful? All of them?"

    No, but the ones who completely miss a station without trying to stop at least have the excuse that Caltrain has a ridiculous number of stopping patterns. That is not an issue for Amtrak, where they all stop at the same stations every time.

  36. Adirondacker1280003 September, 2011 18:55

    That is not an issue for Amtrak, where they all stop at the same stations every time.

    No they don't. Northeast Regional stopping patterns are highly variable.

  37. No doubt he's responding to my comparison with Oakland-Sacramento trains. In any case, saying "for a trained train operator, stopping a train isn't any more complicated than stopping a car" is silly, of course.