Applying Highway Concepts to Transit

One of this blog’s recurrent themes is that we need to notice when people are thinking about transit as though it worked just like roads and cars.  Our transportation bureaucracies are full of people who’ve been trained to understand traffic, and who sometimes struggle to extend that mental framework to transit.  One of the most important American “bibles” on public transit, the Transit Capacity and Quality of Service Manual, was explicitly designed to imitate the structure and style of the AASHTO Highway Capacity Manual, because it saw traffic experts as one of its key audiences.

In the general public, too, it’s common to encounter well-intentioned people who unconsciously think like motorists even as they try to talk about transit; one sign of this is when people talk about vehicle speed as though it were much more important than frequency.  For cars it is.  For most urban transit it isn’t, with the exception of the most rigidly scheduled commutes.

Along these lines, consulting engineer Steven Kolarz wrote me to ask about this famous diagram from the AASHTO Green Book, explaining the tradeoff between access and movement as highway planners understand these terms:

Access vs Movement Hwy

Steven writes:

An interstate highway has great mobility (manifested as high speeds), but poor accessibility (wide spacing in interchanges).  Meanwhile local streets have the opposite – poor mobility (low speeds), but high accessibility (numerous intersections, curb cuts, etc).  Perhaps it is time we adapt this idea more explicitly to transit.  You already briefly touched on this idea in the controversial, but well needed, streetcar posting.   Though true that the exact implementation is a transit technology can be very fluid, there are overarching features and implementations that can help dictate what is appropriate and inappropriate.  Specifically, I am thinking of inappropriate as an 89 mi. High Speed Rail line, and a 64 mi. Light Rail line.

Yes, that’s what I’m up to when I talk about the balance between distance (which requires speed) and stop spacing (which causes delay) in posts such as this one.  (I would clarify, though, that there’s not necessarily anything wrong with a 64 mile light rail line, as long as it’s fast in the middle where the most people are on it.  But most long light rail lines in North America are the opposite; they have downtown in the middle and go through downtown on-street, making that the slowest segment.)

The speed vs stop spacing issue in transit is reasonably analogous to this speed/access distincion in highways.  In transit, travel time is determined mostly by what can get in your way, and by how often, and for how long, you have to stop.  You could even make an analogy between a typical local bus stop, where the bus stops only if requested to, and a driveway access to an arterial, which causes delay to through traffic only if it’s being used.  In this analogy, a major traffic signal that affects most traffic would be comparable to a rapid transit stop where every trip stops because there’s always someone getting on or off.

This tradeoff between speed and stops in transit determines some of the most fundamental categories of transit service: local vs. rapid vs. express as I defined them here.  This is an area where transit/traffic analogies work reasonably well.

The real problem for leading highway thinkers across the conceptual gap into the transit world is frequency, a concept of paramount importance in transit that has no analogue in motoring, apart from the small waits required by signal cycles.  This is such a persistent conceptual barrier between motorists and transit people that I’ve started coining vulgar alliterative slogans such as “Frequency is freedom!”   Because it really is.

Finally, when interacting with highway engineering, it’s crucial to notice that the highway engineering meaning of the terms access (or accessibility) and mobility (or movement) is quite different to the way those words are used in sustainable transport planning and urbanist thinking.  The terms access and movement in the email and diagram above refer exclusively to the tradeoff as it affects vehicles.  If everything that matters is a vehicle, then access and mobility do come into conflict, and the different standard road types, from freeway to cul-de-sac, represent different points of balance between them, as the diagram indicates.

But when sustainable transport people talk about mobility and access, they’re extending the concept to apply to all humans rather than vehicles, which makes the issue vastly more complex and multidimensional.  They’re also transforming access into something more purposeful, and assigning it a moral weight that it doesn’t have in the highway world.  In the sustainable transport formulation, mobility is how far you can go, while access is how many desirable things you can do.  You can improve your access, but not your mobility, by moving closer to work or moving in with your romantic partner.  In this urbanist formulation, access can be improved through putting desirable things closer together — a process that we all consider when we decide where to live, and which urban designers do in the aggregate when they design or redesign communities in response to the demands that our individual decisions have generated.  For that reason, access, in urbanist thinking, has taken on a moral superiority over mobility which it doesn’t have in highway thinking.

So be careful with mobility or movement and access or accessibility.  Urbanists and sustainable transport people mean one thing.  Highway engineers mean something much simpler — so simple in fact that you can make a diagram of it.


20 Responses to Applying Highway Concepts to Transit

  1. Tom West April 16, 2010 at 6:22 am #

    “The speed vs stop spacing issue in transit is reasonably analogous to this speed/access distincion in highways”
    It’s not just analogous – stop spacing in transit is *exactly* the same as the amount access controls on roads. Stop spacing and access points both are places where you join/leave the service/highway in question.
    Frequency/headway doesn’t occur to road planners, because the effective headway of a car is zero minutes, and hence in-vehicle time is exactly the same as journey time. However, with transit, journey time is in-vehicle time plus some sort of allowance for headway.

  2. EngineerScotty April 16, 2010 at 8:12 am #

    Many transit debates can often be understood in these terms. The ideal line for a given user is one which stops at his residence, and which takes him directly to where he wishes to go without intervening delay. Stops, transfers, obstructions, and such are all contrary to that goal–as is any infrastructure near his residence which doesn’t address his need directly.
    To put it another way–we desire access and mobility for ourselves, and seek to deny it to others where it would inconvenience us.
    Obviously, such a goal cannot be satisfied for everyone simultaneously, so our transportation networks (transit, roads, whatever) are a compromise. But still, a common theme emerges:
    * People want high-access services close to them.
    * People want high-mobility services close to someone else.
    This explains the construction of freeways, and the freeway revolts, in the US–in places where the suburbs had greater political power; high-mobility roadways (freeways) are built to connect distant communities, sometimes without regard to the urban neighborhoods destroyed by this. In places where the urbanites have the political power to resist, they often do so–freeways to the burbs do far more harm than good to urban dwellers. The whole streetcar-vs-lrt-vs-metro debate is another instance–urban dwellers often don’t care about crosstown or suburban commuters (especially if the needed infrastructure is an eyesore or a physical impediment to their community), and want something to provide easy access to within a small urban area; users outside the urban core hate taking twenty minutes to go three miles through downtown Portland. Likewise, bus-vs-rail is also often an instance of this debate, especially in the context of local bus vs LRT (a common argument heard in Portland).

  3. CroMagnon April 16, 2010 at 8:49 am #

    Another good post.
    Curious that terms and meanings with regard to access and mobility take on moral connotations. I don’t think they should.
    In mature urban areas, accessibility is usually quite established, such that additions of higher-grade rail transit are really about mobility. Accessibility is relegated to how people existing transit riders can get to the new system and enjoy the improved mobility.

  4. Scott Mercer April 16, 2010 at 10:16 am #

    Don’t have the hard facts on this, but it does seem to me that the light rail trains of the pre-Transit removal era in the USA (they were called interurban trains back then) DID have the “fast part in the middle,” because they connected one or more downtowns To each other. The middle part was unpopulated suburban or even rural countryside, allowing fast running speeds. There were many of these interuban systems throughout North America. I believe reading that there were even a few that had routes so long that they offered sleeping facilities for overnight (or late night) trips.
    Service within downtowns, or connecting downtowns with nearby suburbs, was done with “streetcars,” which almost to a system were single cars, much smaller than the “light rail” systems used today. Most of our current systems in North America do use the “hub and spoke” pardigm with downtown at the middle.
    Los Angeles was perfectly suited to the interurban model, because it always had multiple activity centers, even prior to the coming of the railroads, with several small cities spread wide apart in a large basin surrounded by mountains.
    Maybe we can get back to the “interurban” model? The only new services in North America I can think of that fits that model is the River Line in New Jersey, and maybe the Sprinter in San Diego County. (Both diesel DMU systems, and not electric, unfortunately.)

  5. M1EK April 16, 2010 at 1:14 pm #

    Tom West puts it well – frequency (time to next vehicle) is like an extra penalty on top of trip time.
    This doesn’t mean, as you seem to claim, that speed is irrelevant, or that it’s even less important than frequency. Look at the popularity of Houston and Austin’s express buses among choice commuters compared to the lowly city bus (much more frequent, much slower; patronage mostly among the transit-dependent).
    In other words, speed will easily trump frequency in many applications. In only a few will frequency trump speed.

  6. EngineerScotty April 16, 2010 at 1:43 pm #

    Reliability is, of course, an important variable–which is why transit is popular with commuters, and many transit systems are geared towards commutes: Transit has more reliable performance during peak load times than do highways.
    Generally, speed becomes more important as distances grow longer. The flight from Portland to Tokyo only departs once per day, but that isn’t a problem. A local bus line that only ran once per day would be FTMP useless, unless that run were timed to serve some other once-daily external event, worth scheduling around.

  7. Cap'n Transit April 16, 2010 at 8:25 pm #

    Frequency/headway doesn’t occur to road planners, because the effective headway of a car is zero minutes, and hence in-vehicle time is exactly the same as journey time.

    Not exactly. The closest analogue is parking: in a lot, how far do you have to walk to get to your car? How long do you have to drive to get out of the lot? Do you have to go down ramps? If parking on the street, how long does it take you to find a spot?
    People like to think that the effective headway of a car is zero minutes, but it’s really not. I think it’s interesting how the negatives you encounter when driving are not considered inherent to driving or cars, but just things that other people are doing wrong. On the other hand, negatives about transit like inconvenient schedules, crowding, dirt and noisy cell phone talkers are considered inherent aspects of transit.

  8. ws April 17, 2010 at 2:43 pm #

    I’ve always wondered why there can’t be variable transit times for commuters (like an express train) where a train only stops at major nodes during peak hours.
    Would that be too inconvenient/confusing for some people to justify that? Would it make sense for a HC transit system to do this in order to make up for its inefficiencies?
    Specifically for Tri-Met: a Hillsboro – Sunset TC – Pioneer Courthouse – Lloyd Center (Cascade Station – Airport express line) or Cleveland Ave TC.
    I could see the benefits of speed, but I could also see the benefit of frequency if there were a good 7-8 sets of express trains that came in during the 7:30AM – 8:30AM rush hour time (with maybe one or two non-express trains intermixed and timed right as to not be “passed” by express trains that make every stop for people).
    Am I crazy or would this work very well for existing users, but also for getting new users onto the system who need the speed and frequency of a car?
    I’ll be the first to admit I don’t know what I’m talking about in regards to this topic, and maybe the everyday riders who benefit from the smaller stops would get kind of screwed.

  9. Alon Levy April 17, 2010 at 3:47 pm #

    The access/mobility tradeoff in transit is complicated by the fact that the freeway-oriented thinking in terms of capacity doesn’t apply.
    When you build road networks, the highest-capacity links are the freeways, which serve the highest level of traffic. Arterial streets have lower capacity, local streets even less.
    Transit works differently: high capacity on a train means lots of people standing, which limits trip length. So for the most part, the highest-capacity, highest-cost technology is the subway, which is at the middle of the access-mobility spectrum. Higher-mobility commuter services sometimes compromise and configure schedules and build rolling stock to be more like a subway close in and more like traditional commuter rail further out.

  10. Jonathan Parker April 17, 2010 at 5:27 pm #

    Interurbans in the modern sense are a major regulatory/legal challenge since most railroads often do not permit LRT-style (non-FRA compliant) vehicles on or even *near* their tracks as was often done in say the 1920s. That’s why you see the push towards the hybrid DMUs like the River Line, which are blurring the lines between LRT and FRA-compliant DMUs. Temporal separation is possible as on the River Line, but that option is often less than ideal for transit angencies, since it places strict time limits on service hours, which often restrict the type of frequency and extent of service that supports a car-free or urbanist lifestyle. It’s a shame that building an expansive transit system like the red car, is basically infeasible today.

  11. Alon Levy April 18, 2010 at 1:31 am #

    Did my comment get eaten?
    I tried to post a comment saying that highways and transit differ in how they treat capacity. With highways, the highest-capacity technology, the freeway, is located at the mobility end of the spectrum. With transit, the highest-capacity technology, the subway, is located in the middle; this leads to many hybrid systems, trying to match the capacity of a subway at longer scales. Traditional commuter rail doesn’t work: people can’t stand at subway crush loads for an hour. This is what leads to RER-style systems, where the train sacrifices some capacity to allow exurban commuters to sit.
    As for speed vs. frequency: sometimes, what actually matters is pricing, ticketing convenience, and access to multiple destinations. In New York, the subway absolutely killed parallel commuter lines for these reasons; the LIRR still has a few percent less traffic than it did before the 7 was extended to Flushing.

  12. Peter Parker April 18, 2010 at 6:14 am #

    I do think there’s some useful analogies between transit and road planning.
    Especially if you take the view that modal share stats represents the best available evaluation of the relative connectivity of the road and transit networks (ie the assumption that most people make sensible transport choices based on the quality of available services).
    In many cases roads set the standards to which transit should aspire. Especially when it comes to connectivity.
    Eg should you be penalised every time you turn a corner? Roads don’t, so neither should transit fares penalise transfers.
    What happens before an intersection? You get a sign telling you what direction, road and destination the turnoff is before you approach it. Does transit provide equivalent real-time information about connecting services, preferaby in the vehicle so you can decide there and then when to get off?
    As for planning, roads are conceived as a network. Justification for new roads is often in terms of improving connectivity.
    Both road planners and drivers regard ‘network thinking’ as logical, whereas transit authorities, operators and passengers sometimes struggle to plan or view the system as a connected network.
    I agree that frequency has no analogy with roads, so the parallel has limitations.

  13. dejv April 18, 2010 at 2:18 pm #

    High average speed is important to make service attractive and to keep operating costs down. If agency bumps speed on frequent service, it can offer the same headway with fewer vehicles and drivers.

  14. michelle April 19, 2010 at 9:07 am #

    Transit access probably compares more directly with parking than roadway access. The amount of time spent looking for parking or traveling from the parking space to the final destination is somewhat comparable to time spent traveling to and waiting at a transit stop or station.

  15. francis April 19, 2010 at 3:17 pm #

    “A local bus line that only ran once per day would be FTMP useless, unless that run were timed to serve some other once-daily external event, worth scheduling around.”
    classic example: the school bus. In Oakland, CA, there are a number of once-a-day bus routes that essentially serve as school buses. However, anybody can ride them.

  16. EngineerScotty April 19, 2010 at 7:35 pm #

    Running express trains generally requires doubling the number of tracks (or getting exceedingly clever) so express trains can pass locals.
    Perhaps with clever signalling, you can have a dual-tracked line for most of the route, with a third passing track at skipped stations (so express trains can pass a local stopped there). With only one bypass, non-peak-direction express trains may have to stop and wait for peak-direction trains.
    One other issue with a MAX-like system is that pedestrians can cross the tracks. The safety issue is mitigated when the trains are slowed down by the need to stop at the station, but if you add express trains to the mix, that introduces a significant safety issue, which would probably mean a lot of station redesign.

  17. ws April 19, 2010 at 8:34 pm #

    “Running express trains generally requires doubling the number of tracks (or getting exceedingly clever) so express trains can pass locals.”
    In my express train hypothetical where I did absolutely zero calculations or math, the local trains during peak hours would be so limited that they would never get passed up.
    I think one could get clever to pencil it out, but that’s just my guess.

  18. Alon Levy April 19, 2010 at 9:06 pm #

    EngineerScotty: if the trains run on a schedule, it’s fine to run local and express trains on two tracks with four-track passing sections. It’s standard on competent commuter rail networks, it’s standard on every high-speed rail network, and it’s even done on two subway lines in Tokyo.
    However, for light rail, running trains on a schedule instead of at fixed headways is too much of a hassle. That’s why the technology almost never works with passing sections.

  19. anonymouse April 19, 2010 at 10:22 pm #

    Nonetheless, the Hudson Bergen Light Rail ran express trains around a local schedule with, I believe 15-minute headways per branch. They did some amount of crossing over and using the opposite-direction track to let the expresses pass the locals. The main problem is that the bigger the difference in speeds, the more restriction this places on your line’s capacity and reliability.

  20. Ben April 20, 2010 at 6:24 pm #

    There are many different kinds of road transport to address different needs, and likewise there needs to be different kinds of public transport to address different needs. With roads you have local arterials with high accessibility and low speed, highway collector lanes with limited accessibility and high speed, and highway express lanes with very limited access designed for long distance travel. Likewise, we have local transit (bus, streetcar), rapid/express transit (subway, BRT, LRT) with limited access and increased speed, and commuter/regional transit (commuter train, highway express bus) for long distance. While you can find compromises, it is impossible to have a one-size-fits-all transit solution.
    When it comes to speed vs frequency, value becomes a defining factor as well. It may be faster to take the less frequent but faster commuter train; but if it isn’t that much faster, becomes too stressful to time it properly, or costs too much many will find themselves taking more local transit options instead.
    Here is a good example: I live very close to a Toronto commuter GO station. From here to Union station in downtown by TTC and GO, it would take 53 minutes (7:50-7:55 on TTC Leslie St bus, 8:09-8:43 on Richmond Hill GO train). Assuming I left at about the same time, it would take 55 minutes by TTC alone (7:47-8:07 on Steeles Ave express bus, 8:13-8:42 on TTC Yonge subway train). Now pretending that both routes cost the same, many would continue to use the TTC exclusively from this area, since the stress of timing it exactly is not worth 2 extra minutes in savings. However, if you compare travel times from points further out in the Toronto area, one may be more willing to time the GO over local transit, since time savings could be more significant.
    Finally, this is a good reason of why subway or elevated transit stop spacing should be focused towards speed outside core areas: Because access to the train is slower due to navigating the stations. One will be more willing to invest extra time getting into and out of a station if the transit ride is fast enough to make up for this loss. However, if you are using grade separated transit for local transit (like on Bloor-Danforth subway line), then the passenger is losing time on both the ride and navigating stations.