request for professional info: has anyone figured out how to estimate bus layover needs?

So you’re designing a new rail line, which will require bus connections at many points.  These bus lines are important.  They extend the reach of the rail line, and they are fundamental to how everything fits together as a complete network that anyone can use.

On some parts of the alignment, it appears logical that many bus lines will terminate at a rail station.  This means that they will need to lay over — sit for a while so that drivers can have breaks — and that takes space.  You also want your rail stations to be vibrant urban places, so redevelopment needs are likely to conflict with bus layover needs for urban space, and it’s obvious which one will excite politicians and citizens more.

Yet the layover is essential.  If it’s not there, buses have to drive further to get to other layover locations somewhere beyond the station, and all that time is money out of the operating budget, that comes at the expense of service that customers can actually use.

The critical problem is how to estimate long-term layover needs.  At many agencies, operational service planners who spend almost all their time working in a 1-5 year timeframe are suddenly asked how many layover bays they will need for a project that opens ten years from now and needs to function for 50 years after that.  Obviously, they have no clue.  They have a sense of how the logic of the network tends to cause lines to converge at a station or not, but they can’t guess how high demand will be in 40 years, or even what the size and shape of the buses running then will be.  (One reason to expect a renaissance of double deckers is simply the intense presure to conserve curb space while maximizing capacity — a big problem for the double-decker’s competitor, the articulated bus.)

So faced with all that uncertainty, all planning staff can do is guess very high about how much space they’ll need, which amplifies the conflict with other development — and also with project construction cost.  In the worst case, the estimated layover need conflicts so dramatically with everyone else’s needs that they get ignored.

The conflict is totally understandable from everyone’s point of view.

Has any agency solved this problem to everyone’s satisfaction?

I’m pretty sure the answer is no, but would love to see some inspiring stories about how you’ve gotten close.

31 Responses to request for professional info: has anyone figured out how to estimate bus layover needs?

  1. Erik Griswold September 20, 2012 at 1:26 pm #

    If you know anyone at what is now the Movia agency (formerly HUR, before that HT and once apon a time KS) in Greater Copenhagen, I can say that they have certainly been grappling with the problem for some time now:
    As for bays, do you need to construct a facility that uses them, or could you use a run through pick-up like that used in Harvard Square in Cambridge, Mass.?

  2. Alan Robinson September 20, 2012 at 2:53 pm #

    From an amateur — There seem to be quite a few design questions that can be answered even into the far future.
    1) Are the busses expected to pulse, in which case the layover bay should also be the bus stop, with one bay per route.
    2) Is there parking at the station that can be designed to provide additional layover space if the need arises, or other open space such as a public square?
    3) How important is layover space as compared to connectivity to the street network or time to transfer.
    4) If need be, can a potential bus route connect at an adjoining station where more may be available. This needs to be done in a way so not to impede bus to bus transfers.
    I can think of several examples of transfer station designs, but only one where there was any large conflict in the design: the proposed underground bus loop at UBC (which had many shortcomings.) The original plans sacrificed capacity and expense for both developable land, location. It seems that the location of a permanent UBC loop may finally be resolved by locating the station further from the center of campus. This is a solution that doesn’t work for rail stations.

  3. Beta Magellan September 20, 2012 at 4:29 pm #

    Bus terminal capacity is one of the main rationales for two of the CTA’s extension projects—southern extensions of the Red and Orange lines, (the latter of is essentially dormant). Both termini are currently over-capacity and the buses feeding into the lines either serve a large, impoverished area with long commutes (Red) or have to travel on a very congested arterial where they don’t pick up or drop off many passengers (Orange). Glancing over the LPA reports, though, I don’t see anything about projected increases in bus traffic in the future—both extensions are in areas that are either stagnant or losing population, so it doesn’t look like increased bus service is anticipated.

  4. Morgan Wick September 20, 2012 at 6:23 pm #

    Can a route continue past the station on the other side, serving two markets at once? This would be an especially good idea if you want a high-frequency grid… 🙂

  5. Red Yoshimaru September 20, 2012 at 8:07 pm #

    I was thinking the same as Morgan Wick. Would it be better to have the bus lines not terminate at the station but continue past and serve other communities? That way, layover space can be built elsewhere and most likely not have to meet a large demand by buses, assuming that buses would head in different directions after stopping at the station. In other words, smaller layover spaces constructed in multiple locations might be more advantageous than a single large layover area at a rail station.
    However, I do see the benefit of a layover at a station. It can create an opportunity for an “intelligent transportation system” (ITS), where bus schedules can be digitally synced with rail schedules to help ease transfers for commuters. In addition, if a bus line travels from the far reaches of a suburb to a rail station, it might be detrimental to extend the already long line even further past the station.

  6. Joel N. Weber II September 20, 2012 at 8:37 pm #

    Harvard Square in Cambridge, Massachusetts, USA does have 86 run through, but there are also a number of routes that terminate there: 1, 71, 72, 73, 77, and 96 come to mind. It should be noted that 1 and 77 are the Massachusetts Avenue routes; if you want to continue along Massachusetts Avenue through Harvard Square, you have to change buses, and even walk a bit to do so (77 runs through the tunnel, with an awkward exit southbound through the doors on the right and the platform on the left if you’re arriving on a diesel bus from Arlington; 1 does not use the tunnel).
    I think the other commenters have some good ideas in suggesting letting buses run through without long layovers if both the bus and rail services will be sufficiently frequent.
    If you want the buses to depart shortly after the train arrives (similar to the way the MBTA’s Mattapan Line departs Ashmont a few minutes after a Red Line rapid transit train arrives at Ashmont), one bay per route may be a good approach, but if you want to allow long layovers to recover from buses stuck in traffic, it may not be enough storage space in extreme cases.
    In a lot of cases, the morning peak direction for buses will be toward the rail station, and the afternoon peak direction will be away from the rail station. If you want to minimize deadhead miles, you need mid-day storage at the rail station for all of the buses which are used during the peak travel time minus the number used for mid-day service. You probably have some sense of how many passengers the rail system might be able to accommodate in a few decades, and what share of them are likely to arrive by bus, which might tell you something about how many buses that might be. You do also want to estimate how many peak direction trips a single vehicle is going to make during a single morning commute period of perhaps three or four hours, which is largely a function of the running time of the route; a route an hour long in each direction may see each bus only making one or two peak direction trips during that peak travel period, but a 25 minute long per direction route might be able to make three or four peak direction trips.

  7. ant6n September 20, 2012 at 8:48 pm #

    I had the same idea as previous commenters, about not having layovers at transit hubs. It reminds me a bit of the issue of terminal capacity for commuter rail stations in downtown — you don’t need any if you through-route the service.

  8. zefwagner September 21, 2012 at 12:57 am #

    Extending beyond the transit hub is often just a waste of money, because the adjacent area is low-density nothingness or the edge of the service area. It’s also often most challenging to find layover space in neighborhoods. You have to find a curb where people don’t mind having buses parked, and a bathroom that drivers can access. It is a lot easier to build that space at a rail station.

  9. Eric September 21, 2012 at 1:45 am #

    (One reason to expect a renaissance of double deckers is simply the intense presure to conserve curb space while maximizing capacity — a big problem for the double-decker’s competitor, the articulated bus.)
    I would assume double-deckers are much slower to load/unload – the buses will be shorter, but more of them will be at the curb at once, so what has been gained?
    Can a route continue past the station on the other side, serving two markets at once?
    If the station is the middle not the beginning of the route, then bus departures from the station (typically the main boarding site) do not take place on a set schedule, which is less convenient for passengers.

  10. Nicholas Barnard September 21, 2012 at 2:48 am #

    So I’m yet another amateur theorizing…
    What about placing a significant amount of the layover space underground? Have a garage space that is pull in and pull out (one way through the complex with two ramps) to maximize the number of vehicles that can be placed into the space.
    It won’t fit every possible configuration, but if you’re building a new rail station that already has to go underground somewhat, might as well take advantage of that.

  11. Nicholas Barnard September 21, 2012 at 2:50 am #

    @Eric Double-deckers from what I’ve experienced have a good deal of self-sorting going on: people who are going short distances stay downstairs, people who are going long distances go upstairs.
    Also from what I’ve experienced the line to go down the stairs doesn’t take much longer than walking a normal aisle. So the load/unload time is more a function of the number of passengers and doors.

  12. dejv September 21, 2012 at 3:44 am #

    Amateur observation from Brno, Czech Republic: if the space is constrained at tram connection end, there is a push for:
    * move layover (and slack time to compensate for possible delays) to the far termini
    * if the former is not possible, find the suitable spot nearby where the layovers can take place while not spending fortune on empty rides
    On the side of organization, no layovers are done in AM and PM peaks

  13. Barry Watkins September 21, 2012 at 7:18 am #

    I would assume double-deckers are much slower to load/unload – the buses will be shorter, but more of them will be at the curb at once, so what has been gained?
    This is a common concern, but it depends on the design of the bus. Some new double decker models have twin staircases, and up to three doors, so passenger flow is fine.
    As to dwell time, remember that stop capacity is a function of dwell time + clearance time (the time taken to get into and out of a stop). If you have a bus that delivers twice the capacity per length, then yes the dwell time per pax is the same, but you only have one clearance movement, not two. Depending on whether the bus is in mixed traffic or not, or has upstream traffic signals, the total clearance time of a bus can hit 30 seconds.
    But back to my first point, it’s really hugely dependent on vehicle design – the number and size of stairs, the number and width of doors, the location and volume of seating. These can all be optimised to ensure DD buses function efficiently.
    (There’s a prototype bus being developed in Australia…. double decker, 14.5m, twin front axles, twin rear axles (tag steer), two staircases, three doors, capacity up to 150. I’m waiting with anticipation to see if this actually becomes a reality).

  14. Barry Watkins September 21, 2012 at 7:32 am #

    The only rigorous way I know to determine the appropriate number of layover spaces is through scheduling software (HASTUS etc).
    Configured correctly, scheduling software will utilise available layover space to build the most efficient network. Using an existing network (because proposed/future networks are rarely entered into such software), schedule the network based on a small number of layover space (maybe even zero). Then continually add layover spaces, one by one, and see how network costs (driver hours, dead running, peak fleet) change. There will be a law of diminishing returns as the number of spaces increases. You can also re-run the scenarios to determine whether providing a driver’s meal room reduces trips back to the depot.
    The next step is to have an understanding of the costs of developing each space, and then do a basic economic assessment of costs vs benefits.
    For future scenarios, the best I can suggest is to ramp up frequencies on the network you have and see what happens. Even though the network may be very different in the future, operations in the station precinct won’t change that much. They may intensify, but the only radical change would be to use more or less through-routing.
    As others have suggested, the first places to look are unutilised bus bays and kerbside space. Both of these irk me though. The former because I don’t like seeing out-of-service buses sitting idle in a bus station. It’s confusing for the passengers, and many drivers elect to leave the engines running, creating noise and fumes. The latter, because eventually someone will decide that the kerbspace is too valuable, and force you out (leaving you in real trouble).
    I’m working on layover projects in two major Australian cities, and it’s a challenge. If you think it’s hard making buses sexy, then layovers are the hardest sell of all.

  15. Neil September 21, 2012 at 9:09 am #

    This is sidestepping the question, but I’d be thrilled if bus bay area wasn’t frequently overwhelmed by surface parking area. Then we could have a conversation about bus bay size vs. (re)development potential.

  16. Erik Griswold September 21, 2012 at 3:09 pm #

    Can’t you increase the frequency of use of the berth/bay and just double or triple up?

  17. Jo Walton September 21, 2012 at 4:56 pm #

    Something that seems to work well for the STM in Montreal is having actual bays etc in some stations where it’s convenient and in others just having a few spots and having rapid driver turnover. The bus comes in, the new driver is waiting, and the two swap places. The old driver goes on break (every metro station has a driver break room and bathrooms) and at the end of the break they’re ready to hop on and exchange with the next driver coming in. When you have every 10 minute buses and you don’t care which one you catch, and you have this system for driver breaks, and it doesn’t really matter which driver is on which actual vehicle either, and you can just mostly keep everything running.
    It also allows for shifting drivers around as needed, because hey, they’re having their break at the metro, you can get them to hop onto the metro and go where you want them tp pick up a bus. And it means you don’t have the thing you’ve talked about here where to finish a run means a driver gets into overtime — they just finish up at an earlier metro and a different driver finishes the trip. Lots of inbuilt flexibility.

  18. Jarrett September 22, 2012 at 9:50 am #

    @Jo Walton. What you describe is called “operator fallback” and most US agencies shudder at the complexity of it. Are you aware of other Canadian agencies doing this? thanks!
    I suspect that is actually the ONLY solution to the problem of layover space in high demand areas. It also reduces fleet requirements! All the other solutions mentioned here involve adding eternal operations cost.

  19. Henry September 22, 2012 at 10:24 am #

    Are bus bays for layovers really necessary? In New York, even though there’s a huge demand for parking, the MTA uses regular parking spaces in the surrounding neighborhood around the bus stop.

  20. Jo Walton September 23, 2012 at 5:29 am #

    Jarrett: I don’t know if other Canadian agencies do it.
    But why is it complex? Because it separates driver and vehicle and you have to keep track of both of them? I guess that’s more work, but more work for a spreadsheet has got to be better than acres of downtown used as bus bays, surely?

  21. david vartanoff September 23, 2012 at 1:16 pm #

    @ Jarrett & Jo Walton. “Shuddering” @ fallback is silly. In this era of GPS locators/radios/cell phones/RFID fare collection, what’s to fear? AC Transit in the East Bay, and Muni in SF regularly swap drivers at convenient corners as shifts change.

  22. Robert Madison September 23, 2012 at 3:56 pm #

    Back in my CTA scheduling days, we determined recovery needs based on recorded reliability of the route. Of course, that doesn’t help when you’re planning for the future on a service that doesn’t exist yet.
    But you’re always going to have to trade off between recovery time and reliability. One of the problems that those planning for theoretical operational models (such as a short bus layover and operators just taking fallback) is that if your bus isn’t on time getting in, it’s not going to be on time leaving.
    In that sense, you kind of have to guess what the variability of the traffic and running time is going to be, and schedule your service based on that (since I’m not one to be bothered with theoretical “models” of traffic and such, I’m not going to endorse any such method for predicting future traffic variances).
    This is really a case of having one’s cake and eating it, too. You want reliable service, but you also don’t want to take up too much space for layovers.
    As for the issue of double-deckers, I’d be curious to know if the claims that double deckers don’t take any longer for dwells than articulated buses are based on plans for seated loads, or full-crush-capacity-with-standees-from-the-back-row-to-the-windshield loads. The comment made above about passengers sorting themselves out with short-distance passengers staying downstairs and longer-distance passengers going upstairs seems to imply the luxury of said passengers to sort themselves (implying that there’s room to move around). Again, my experience being with CTA, it’s tough enough for folks to get out when they’re a few feet from the door. Not sure how having to navigate a staircase on a crush-loaded bus would work.
    Back to the original question of how to plan for layover capacity, ultimately, it really depends (as noted) on the type of system you’re planning, and what you expect the service needs to be. If it were just about drivers taking breaks, 4-5 minutes should be sufficient (operator mealbreaks can be done with reliefs, and CTA does do that sort of “fallback” with operators being relieved, going on their meal break, then picking up another bus when they’re done, perhaps 40-50 minutes later). But it’s not about that, it’s also about reliability of service. And it’s about how much service you plan to provide. If you’re going to need to run three-minute headways, you’ll obviously need extra space because just the layover alone would exceed that. If you’re talking about 15-minute headways, then you won’t have buses stacking up. With multiple routes, if they need to pulse (all show up and leave at the same time), you need one spot per route. If they can be staggered, then with proper scheduling, you should be able to develop some kind of pattern (including interlining, or possibly even through-routing if that’s beneficial for some reason) to minimize the number of buses in the terminal at once.
    Long story short, there’s no one-size-fits-all answer. There are too many variables that probably nobody can answer before a line is developed.

  23. Max Wyss September 26, 2012 at 5:06 am #

    From an amateur: I was exposed to two operators in my former hmetown(s).
    The first scenario is pretty much what is described in the article. A S-Bahn station serving about 10000 inhabitants, recently rebuilt, 4 trains per hour and direction on weekdays. The station is served by 6 bus lines, 5 of them terminating. The station has two “double” length platforms plus a “single” length platform (commonly used by two minibuses (MB Sprinter type)). Weekday operation is every 30 minutes for 4 lines, and hourly for the other two. The schedules are kind of coordinated to provide connections to the trains, which means that the platforms will be heavily used every 30 minutes.
    Operation is such that the drivers normally drive from the garage to begin service and then drive back to the garage at the end of their shift. In some cases, the driver for the next shift drives from the garage to the station with an automobile, and the driver ending the shift drives the automobile back.
    Now, to the original question: there is one space dedicated for parking a bus, not at the platforms, but at a border of the site. For short breaks (which are required by legisation, the layover is just sufficient. The layover time has been scheduled to provide the short break time.
    The other operator is urban with high density schedules (main bus lines every 7/8 minutes; secondary lines evry 15 minutes). Here, the drivers change at specified stops (if the bus passes near a garage, it would be at that stop, otherwise at the stop easiest to reach from the garage. Drivers do, however, not need to check in at the garage, which means that they can get to the changing stop directly from home.

  24. Pete Brown October 1, 2012 at 1:32 pm #

    Regarding double decker capacity, a typical UK bus outside London will have a single entrance/exit doorway and seat 45 upstairs and 29 plus up to 21 standees on the lower deck. There will be a space for wheelchairs and/or buggies on the nearside behind the wheel arch. The stairs will be located over the front offside wheel arch and will ascend towards the front of the bus with a quarter turn at the foot and the top. At bus stops alighting passengers should have priority, drivers should enforce this. Many urban systems use exact fare systems and loading can be fairly swift.
    (page 4 for seating layouts)
    Some typical double decker bus action in Brighton:

  25. Paul K McGregor October 4, 2012 at 6:34 am #

    Based on my professional experience, here’s a couple of approaches that I was actually a part of. While I was at VTA in San Jose back in the early 2000s, I had to develop a preliminary bus network that would be expected to serve the appropriate BART stations. Of course, this is based on current operating conditions and doesn’t really get into the long-term future growth potential. This network was also used to develop ridership estimates. You can get a preliminary idea of bus bay needs. Once the number of bays is established, it then becomes a scheduling exercise. But those questions aren’t usually asked until you actually get ready to implement the service.
    In Dallas, they had routes especially in the Oak Cliff area that operated in pairs through to downtown. It made sense to truncate one route at a rail station and operate the other route through to downtown to serve the area north of the rail line. Those downtown routes still operated through the station so it didn’t really save any bus bays. At the time, DART was operating articulated buses and the bays were being designed for 40 foot buses. I asked the question of designing for longer buses but I was told that DART would be phasing out the articulated buses and there would be no need for longer bays. While that may hold true for the 10-15 year timeframe, would that be the case for the long term?
    Ideally, you would want to have a way to provide enough for your immediate needs and still provide some way to expand for future expansion if needed. It always seems like you get a constrained space and you don’t get enough space or you get a lot of bus bays and not enough buses to use them. The same goes for figuring out the appropriate number of parking spaces for park & ride lots. It just really comes down to the amount of space that is available and the adequacy of your assumptions that are used.
    In regards to the discussion on fallbacks, it is a common practice on rail and can be used for bus operations but only in specific situations. You would probably not do it on a systemwide basis because of the complexity involved. I am sure with the use of HAUSTUS and Trapeze, it does simplify the process, but for the field supervisors, I am sure it creates headaches for them.

  26. Nathanael October 4, 2012 at 5:57 pm #

    “The only rigorous way I know to determine the appropriate number of layover spaces is through scheduling software (HASTUS etc).”
    Which runs right up against the unreliability of bus schedules. For an exclusive-ROW urban rail system, you can use scheduling software to figure out the layover needs with great accuracy.
    The conclusion I get is this:
    (1) If you expect all your bus routes to be quiet rural routes along uncrowded roads, you don’t really have a layover problem: just count the destinations and count the buses.
    (2) If you expect to have dense urban bus routes stuck in traffic, you need exclusive ROW, and if you need exclusive ROW, you want rail.
    So we’re getting down to the idea that maybe the bus interchange point needs to be TEMPORARY, until the rail lines are expanded further, 10 or 20 years down the road.
    This makes it easier to figure out the bus layover needs (since you don’t have to project as far in the future), but also implies that you should be very sure to allow for future rail expansion in multiple directions.

  27. Nathanael October 4, 2012 at 6:00 pm #

    I’m surprised to hear that there are complaints about “complexity” of “operator fallback”. “Operator fallback” has been best practice for high-efficiency operations since what, the 19th century I think? Train operators don’t even blink at the idea.

  28. Julian Wearne October 13, 2012 at 11:05 pm #

    Yep. I fail to see why you need bays for buses to park for the entire time that a driver is on break. What difference does it make if a driver gets off the 9.45 bus, takes his break then climbs aboard the 10.30 bus to continue his shift?
    I’m very curious as to why American transit agencies shudder at the complexity of such a situation?

  29. Scott October 17, 2012 at 4:54 pm #

    Julian, I think the problem is not having a driver get off the bus, take a break and drive another bus. The problem is sending the driver unsupervised to another point in the system. My experience in not just transit but working in general tells me that adding extra complexity to getting a driver where you want is just providing extra opportunities to be late, either unintentionally or intentionally.
    The busier a station gets the less it makes sense to have the bus connection at the walk-in entrance and another entrance should be developed that is designed for good bus connectivity or at least to split the load. Layover could be accomplished at the back of the site or by purchasing some spaces at the nearest parking lot. Someone mentioned underground bus storage and by that logic, elevated bus storage could also be an option but it seems that either of those options only makes sense if there are lots of buses that are laying-over and requiring a bathroom break.
    I currently live in a city with lots of low bridges and doubt that double-decker buses will be able to run most of the routes. I think articulated buses boarding time could be easily improved if the local transit agency adopted some BRT style elements on some of the busiest routes.

  30. Julian Wearne October 20, 2012 at 3:34 pm #

    @Scott – I’m not talking about the bus driver moving unsupervised to another part of the network. I’m talking about the bus driver getting off at a hub, a different driver climbing on board then 30-45 minutes later (however long breaks should be scheduled for) the same thing happening again. All at the same hub, all on the same route. That way drivers can still start and finish their shifts at the same place, and they can still drive the route they’re shifted for on the day. The only thing that changes is the vehicle they’re driving.

  31. üsküdar tercüme bürosu October 21, 2012 at 2:00 am #

    I don’t have the same idea as previous commenters, about not having layovers at transit hubs. It reminds me a bit of the issue of terminal capacity for commuter rail stations. Thats not positive. It would be better.