More Reader Mail: Morning Delays And A Possible Answer at SF Appeal

This morning, Reader Sam noticed some delays, and an odd appearence of an N train at Church St. Station! Read on, and post something in the comments if you saw something similar early this morning:

This morning, I headed to the East Portal stop to catch the inbound N to 4th & King. When I got there (around 7:10), the Next Muni screen showed something like 50 & 55 minutes for the next arrival. Based on that, I ran down to the Church St Station to try to catch the T (my only other hope for getting to Caltrain). Much to my surprise, the next train to show up at the station was the N-Judah!
I thought my problems were over, since I should have had plenty of time to catch my 7:59 train, but then the train I was on sat between the 2nd & 4th street stops for over 10 minutes, and as far as I could see, there were a lot more trains stacked up at the turnaround. Another train came up 4th and turned onto King, following the route for the KT, but it’s sign said N-Judah as well. Luckily, the driver was sympathetic to our plight, and he eventually lowered the stairs and opened the doors, giving us just enough time to run to the train.
I still don’t know what caused all this commotion, so I’d love to hear if you get any more info.

The SF Appeal has this article about some short term changes MUNI is making while the NTSB investigates the crash.
UPDATE: And a review of Twitterpals’s posts indicates it was another severed overhead wire at Church and Duboce.
Another one? Eesh.

This entry was posted in Reader Mail. Bookmark the permalink.

12 Responses to More Reader Mail: Morning Delays And A Possible Answer at SF Appeal

  1. Sam Etler says:

    I tend to take Muni to Caltrain from the Castro and have found that it’s not worth the hassle and risk of missed trains to try and get an N or a K/T to go all the way around the city. I’ve found that getting off at Powell and walking down 4th St. generally gets me there anywhere from 10-20 minutes faster. It’s about a 1 mile walk, so if you’re a slow walker or just can’t do the walk then perhaps it’s not as good of a solution.
    If you can though, it takes away the need to wait for a K/T (which will be crowded and a single car when it eventually arrives), the need to take something else and then switch to an N at Embarcadero which tends to get backed up in the morning, and the annoying waiting as Muni navigates the short but painfully slow route from the tunnel to 4th and King, all the time wondering if you’ll make your train or not. Oh, and exercise. Always a plus.

  2. anonymouse says:

    On the few occasions that I’ve actually taken Muni to Caltrain, I’ve found that transferring to the 30/45 at 4th/Market is actually not a bad option. Definitely faster than the N or K/T, because of the potential delays at Embarcadero and especially between 2nd and 4th. And yes, I’ve been on an N before where the driver lowered the stairs and opened the doors to let Caltrain-bound passengers make a run for it.

  3. anonymouse says:

    Also on the topic of automatic versus manual mode: why is the automatic mode so bad that it would take so much longer to get into the station? I’ll admit I don’t regularly go to West Portal, but if the automatic system is causing delays, clearly it needs to be fixed rather than compromising safety for a workaround. Of course that would require going into the vast pile of ancient and unmaintained code that Muni may not even own the rights to.
    It also baffles me that there’s no “supervised manual” mode between full auto and full manual, where the driver runs the train, but the ATCS still enforces max speed and train spacing, and that the transition from street running to subway and vice versa is so inelegant.

  4. SLKinSF says:

    Meanwhile, more MUNI weirdness.
    Was walking down Judah today and saw an outbound train, stopped at 10th Ave, where it was letting off passengers. At 10th. No other trains around, no clue why this train was doing its business just ONE BLOCK past the ‘regular’ stop at 9th.
    I know it doesn’t rate high on the fail scale—this week especially. Just another small bizarre data point, another signpost on MUNI’s road to Hell. But I’m really starting to get the feeling we’re headed for a rerun of MUNI’s Dark Ages (back in the mid-to-late ’80s).

  5. Brian says:

    In automatic mode, one train must clear the West Portal station completely in order for the next one to arrive. Sometimes clearing that station can take a long time because there’s a lot of cross traffic from cars, pedestrians, and other trains.
    Putting the train into manual mode allows the driver to get into the station faster and unload/load faster, because the lead train clears the platform a lot faster than it clears the station. It’s even faster if one of the trains is a single-car train because the station can unload/load both trains at once.
    After the wreck, I surmised that they would stop this practice, and indeed, they have. It has the potential to lead to backups, and at least two days this week for me, we have been backed up for a while waiting to get into West Portal station.
    My own solution would be to split off the L underground at Claremont Circle and keep it underground on Taraval until it’s past 19th Ave. It would improve safety, speed up the K, L, and M, ease equipment needs somewhat (faster trains mean that you need fewer of them), improve Muni Metro coverage slightly (put an L station over by Taraval and Madrone), improve traffic, and even reduce noise a little by eliminating three streetcar turns.
    But of course, I’m crazy, because that would just be too nice. Maybe we can hope for something a little less ambitious, like an ATCS that actually works correctly.

  6. Alex says:

    @Brian: or make West Portal a non-car intersection, or make it right turn only, or extend the K to St. Francis circle and keep the L above ground only, etc, etc, etc.
    @mouse: Supposedly the ATCS can be used such that it only enforces a speed limit… that speed limit in the West Portal station area is 30mph (the driver in this case was only going 23mph).

  7. anonymouse says:

    ATCS, also known by its brand name Seltrac, is a system designed for 100% segregated metros (think Vancouver SkyTrain) and makes assumptions accordingly. These assumptions include that the safe stopping location for each station is a single point, and that trains stay within the system. It is also not a very flexible system. They haven’t been able to add T trains into the system, and changing the stop points to stop zones is a much more complicated change. Even if the train is allowed to enter the platform area (which it generally is, it’s a moving block system), it won’t be able to open its doors, because it’s not at the station point as far as ATCS is concerned.
    As for why West Portal is different, it has to do with manual mode and the way ATCS works. It’s a communications-based system: this means that in the normal functioning of the system, each train’s onboard computer is constantly telling the central computer where it is, and the central computer tells it how far it can go without hitting another train. Both the communication and the train’s determination of its position uses the inductive-loop cable that you can see on the tracks in the subway. For backwards compatibility, emergencies, or if the train’s on-board computer breaks, there is also a backup system to ensure that the train doesn’t just disappear. This backup system is a fixed-block system, and thus can only locate a non-communicating train as being within a block, and trains under ATCS control can only go as far as the beginning of that block. How does this have to do with West Portal, and why is West Portal different? Well, my conjecture is that at West Portal, because outbound trains switch from ATCS to street mode, they become “non-communicating” trains. This means that they can now only be located to within a block, and this block covers the whole of the West Portal station platform, thus any following train still in Auto mode would only be able to get as far as the start of the block, or possibly the signal at the crossover. Drivers would bypass the ATCS to avoid this and keep the service moving.
    I personally consider this to be a failure of the ATCS system. If a safety system is so restrictive or dysfunctional that it is routinely bypassed, then it is as good as not having a safety system in the first place. Ultimately, I think Muni is going to have to either accept the responsibility for modifying ATCS until it actually does what Muni needs it to do, or else replace it with something else that does.

  8. Alex says:

    I’m not sure what else I can add here, you’ve summed it up quite well.
    At least for this week, trains do not appear to be taken out of auto mode until the operator is ready to depart. However, the ATCS setup won’t allow for double berthing. Thus even with the trains in auto mode at the platform, following trains won’t advance to the platform. As a result, you’ve got driver socialization, supervisors spending a few minutes trying to give instructions to drivers (apparently nobody’s heard of using a radio), supervisors scrambling to find drivers, automobile and pedestrian traffic delaying each and every train.
    Looking at the ATCS screenshots (, you can see various dots along the track as if there are potentially various stop points at the stations that could be used as station points.
    Along with SelTrac, the previous cab signaling system is still in place. Cab signaling is what was used for the Boeings without a huge number of problems. The speed limits are lower than with SelTrac, but much faster than if you’re in fully manual mode. What does the MTA want to do? Remove the cab signaling system. ::facepalm::
    Additionally, Alcatel (now Thales) never did bother to make the ATCS work properly. In theory Alcatel was contractually obligated to ensure at most one hour of ATCS downtime a year. We know how well that’s worked out…
    P.S. Docklands Light Rail in London suffered quite a few teething problems with SelTrac around the same time we did. The tube switched as well(!) on the Jubilee, Northern, and Piccadilly lines. Fat lot of good it did them (took me ~2 hours to get from Heathrow to Liverpool St because of various delays).

  9. anonymouse says:

    I thought the Bredas didn’t have the onboard equipment for cab signalling? It would have been handy to have and may have avoided the infamous Muni Meltdown. And I don’t see any reason they can’t run close headways at high speeds with just a cab signal system, Moscow Metro does 39 tph (scheduled capacity) at 50 mph with such a system with no problem. As for London, only the DLR has Seltrac. The main Tube lines don’t have it in revenue service, though it’s being installed and tested on the Jubilee with, unsurprisingly, lots of problems and delays. I think the problem comes down to the fact that Seltrac is a sort of all in one system, which works great if you do everything entirely the Seltrac way, but starts failing badly if you want to deviate even a bit from that, or try to make non-Seltrac parts and systems work with it.

  10. Alex says:

    After yesterday’s failure and chaos, I don’t have anything positive to say, so I’ll keep it short. I’m pretty sure that the Bredas have cab signaling equipment (as indicated by the switch). That equipment is almost certainly post-meltdown.
    SelTrac (and pretty much anything Thales/Alcatel has ever touched) is utter shit. Thales advertises SelTrac as being installed on DLR, Jubilee, Piccadilly, and Northern lines. Of course they also told SF that SelTrac was proven technology when we were their first customer.

  11. anonymouse says:

    I’m pretty sure SelTrac was installed on the DLR before Muni, and definitely on the Scarborough RT, Vancouver SkyTrain, and Detroit peoplemover. But I’m also pretty sure that nobody ever tried to implement Seltrac on a system as operationally complex as the Muni Metro.

  12. Alex says:

    Alcatel/Thales is pretty good at spinning things. You’ll see things like their big press release touting its success with the MUNI Metro. They tout their amazing ability to handle 60 trains per hour without once mentioning how much shorter the trains need to be (partially a problem with the Bredas, partially SelTrac).
    So it’s a bit difficult for me to read through some of the DLR stuff and infer success. That said, DLR has been using an Alcatel moving block system since 1990. They upgraded in 1999 (I don’t know if this was the completion date or the start date) to the same SelTrac system MUNI was sold on. Worth noting, however, is that DLR stations are only long enough for one car.
    Even more importantly only MUNI and the Scarborough RT actually use drivers. SelTrac had (has?) only ever been proven for light duty use. Even with drivers, the RT was designed to operate like a light duty people mover.
    Transportation Systems Design, Inc. out of Oakland were huge proponents of MUNI going to SelTrac. It would be interesting (to me at least) to see what they think of SelTrac now that we’ve been forced to live with it for ten years.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.