Stephen Rees's blog

Thoughts about the relationships between transport and the urban area it serves

Cost Comparisons of Transportation modes

with 21 comments

The following has been circulated to a number of email list serves. I am copying it here in its entirety and without further comment from me as this issue has been frequently raised here, and I am sure that my readers will find it useful.

Free download here:

This research has been improved by inclusion of a longer list of external costs associated with each mode. Pollution costs and the value of GHG reductions are not included in these externalities. They are things like cost of road construction and upkeep, and parking charges. The biggest change here is moving away from the average cost to park a car downtown to the average cost to park a car generally, which lets the cars look better cost wise. If our focus is narrowly on trips within and to center cities this would change back.

At any rate, the Prius does slightly better than the tram per mile in this work. Buses, skytrain, and light rail behind them.

Of significance, we also have a cost per trip. Given that cars make longer trips generally than do busses or trams, in this computation trams do far better than the prius. This to us is significant as its not the distance that matters to the traveler, its the destination. In areas where there is a synergy between landuses and mode, as is the case in “streetcar city” neighborhoods generally, trips tend to be, on average, and whatever the mode, much shorter.

Finally, as we state in the work, this is not a definitive set of answers to the question of what mode is “best”. There are too many variables, too many assumptions, and too much future uncertainty. At the same time we do believe that the work help clarify a set of issues that are quite muddy, and most often dealt with in a disintegrated way.

Professor Patrick M. Condon
University of British Columbia
James Taylor Chair in Landscape and Liveable Environments
2357 Main Mall
Vancouver, BC – V6T 1Z4
604 822 9291

Written by Stephen Rees

November 25, 2009 at 11:12 am

21 Responses

Subscribe to comments with RSS.

  1. Figure 10 got my attention. I cannot think of the last time I was on a Skytrain with less than 20 people aboard. No way that the average utilization of Skytrain is less than 20% of capacity. If you blow that up to 30%, the rest of the stats for Skytrain (per passenger-mile) become less negative. And that is one of the stronger arguments for Skytrain, people would rather use it than a bus.

    Pat Johnstone

    November 25, 2009 at 4:22 pm

  2. from the caption “typical vehicle occupancies for the transit modes were calculated from operating data reported from existing systems using these vehicles”. You seem to think that casual observation is a better source of data than the operators of the systems themselves supply. This seems to me to be unlikely. However, as I am not the author of the report, I have no insight into how this data was calculated. However, it is obvious that there is spare capacity on every transit system off peak weekdays, at peak periods in the “return” direction, and on weekends. On SkyTrain in recent years there has been a lot of spare capacity on the Millennium Line which has only recently seen full trains at peak periods

    Stephen Rees

    November 25, 2009 at 4:33 pm

  3. I suspect the reason why SkyTrain average utilization is low compared to other modes is the high frequency of trains in off-peak hours. For systems with drivers, the frequency is often way less in off-peak hours due to the high cost of the drivers.

    It is certainly great as a user having trains every 5 minutes instead of having to wait for 15 or 20 minutes.

    One of the many problems with such high-level cost analysis is that it does not take into account some of the other benefits of the various systems nor does it really give any idea what is the best solution for a particular corridor.


    November 25, 2009 at 5:27 pm

  4. Figure 10 is important as it sets ridership for the other analyses, but there is little detail on how they calculated vehicle occcupancy. Perhaps skytrain’s off-peak capacity is the reason why the numbers are skewed, but IMO off-peak frequency is not a bad thing for riders. Portland’s new Green line MAX runs every 30 minutes weekday evenings.

    A few small things skimming thru:
    -why did they use skoda trolleys for the analysis instead of new flyers? (Fig 10)
    -they use the combino supra for the tram comparison – but the photo looks like they used Budapest’s 72 meter long combino trainsets – it is not clear what they used for the analysis.


    November 25, 2009 at 10:18 pm

  5. Come on Stephen, Pat Johnston is right, the Skytrain numbers are utterly wrong and… why the author tooks some 2003 number for the Skytrain?

    let’s use the most recently available translink number:
    average trip length on skytrain: 10km
    ( 37.5Million
    boarding.year= 73.5 million

    mkI car capacity=80, number of MK1=150
    mkII car capacity=150, number of MKII=60

    work for a garde 7: compute the skytrain average car capacity
    work for a grade 8: compute the % of skytrain car occupancy…

    you can easily verify that the number relative to the skytrain in the study are utterly wrong on all accounts!

    Mezzanine asks right questions:
    we could also add why use a tramway with regenrating braking and not the equivalent for bus (last hybrid bus on 99B line for example)

    So Wwy he has use the 2003 year for Skytrain occupancy (year of Millenium line opening, hence lowest occupancy in decades!) and not not the latest available statistic, if it is not to deceive the reader?

    the answer is: the author has a very strong bias in favor of “streetcar” ala Portland and will choose fact supporting his conclusion:
    Streetacr ala Portland are always the best, and if you don’t want it, better to have a car!

    So I give little to no credibility to this type of study

    …but thanks for posting 😉

    I have already noticed problem with a previous study emanating of the same author in


    November 25, 2009 at 11:02 pm

  6. I get on SkyTrain at Nanaimo every morning. At that point the outbound trains run almost empty so there is certainly a lot of unused capacity. It’s not in the direction most people care about, but it has to be part of the calculation. Even in a peak period the average load on a train is (one full + one nearly empty)/2 or roughly 55% full. In off peak hours that utilization is significantly lower so an average value of 25% or less is to be expected.

    Contrast that with my experiences in the 1990s on the #4 trolley bus. Because there were major destinations at both ends of the route buses in both directions filled up. I haven’t used that bus in a long time so I have no more recent observations to contribute.

    I find it interesting that Professor Condon says it’s easy to get the capital costs for SkyTrain projects when all I see are cover-ups and half-truths. No two sources agree on how much Canada Line cost to build. We were told repeatedly that it was “on budget”. There was a nice chart on the project website showing how closely actual expenses lined up with the $1.335B budget.

    Then we’re told by Global TV the final cost was $1.98B and by TransLink that it came in “on budget” at $2.05B. Disclosures during the Susan Heyes court case suggest the real figure is another 25% higher again, fully double the original budget.

    Professor Condon and the rest of us don’t know for sure how much SkyTrain cost to build and we don’t know how much it costs to operate because TransLink and Victoria don’t release that data, at least not in an audited form.


    November 25, 2009 at 11:03 pm

  7. A note, re: SkyTrain car capacity.

    A SkyTrain Mk.1 capacity (all seats filled and standees @ 4 persons per m/2) was originally 75 persons.

    A Bombardier Mk.2 car capacity (all seats filled and standees at 4 persons per m/2) was about 30% higher or about 100 persons.

    TransLink loves to calculate capacity with SkyTrain (and RAV) cars at 6 persons per metre/2, which on paper looks good and may inflate ridership numbers for statistics.

    Toronto & Detroit rate their ICTS (almost identical to a Vancouver Mk.1) car at 70 persons per car. During Expo 86, BC Transit claimed that they could pack 100 people into a Mk.1 car, yet could not verify the claim, but it did make for some high ridership numbers during the last weeks of Expo 86. BC Transit claimed 300,000 persons a day used SkyTrain; not bad for, four car trains operating at 5 minute headways (10 minute headways from Stadium Station to Waterfront Station).

    Here is the problem with SkyTrain, BC Transit and TransLink have inflated car capacity to suit their need to show high ridership. Their numbers are suspect, thus any study using their numbers must be suspect as well.


    November 25, 2009 at 11:38 pm

  8. We currently have a SkyTrain system, and I’m sure we would benefit from Trams and LRT as well (serving a different geography). Each mode complements the others as well as serving different types of trips. The characteristics of the data presented in the report have more to do with govn’t policy and priorities than they do with technology/mode.

    Instead of berating the SkyTrain system for having so much extra capacity, why not make better use of it?

    Its not that there’s a lack of travelers out there to use it (look at our plugged roads). Perhaps we need much better transit to connect to it? More employment centres created along it? Completing the long-promised Evergreen line would do wonders for Millenium Line ridership, as would connecting the Millenium Line to the Canada Line. (stop treating it like a feeder service to downtown, and create a network of rapid transit so people can get to multiple destinations)

    Reducing the subsidies to the private automobile would also boost ridership. (why are we subsidizing 2 competing forms of transportation?).

    It would not take a committed Govn’t long to make the right investments and policy changes such that the numbers in the report would appear quite different than they do now.


    November 26, 2009 at 12:22 am

  9. The only way to use excess capacity on the SkyTrain light metro is to again densify housing along the line. Here lies the Achilles heel of SkyTrain and light metro, to maintain higher commercial speeds, stations are further apart, making the transit system much less user friendly.

    Car drivers will not leave their cars to take a bus to a metro; they may leave their cars to take ‘rail’ transit if it caters to their transit needs.

    80% of SkyTrain’s passengers first take a bus to the metro and this figure probably represents established transit users who would not normally use a car.

    In the late 70’s, light-metro was thought to have replaced most light rail systems by the Millennium, but the opposite happened, light metro could not compete against LRT and has mostly disappeared into obscurity or morphed into specialist roles like airport people movers.

    It is good to see UBC finally doing studies on the subject of urban transit and the next step must be offering degrees in public transportation and/or urban transit instead of transportation and transit being an adjunct of the ‘Planning Faculty’.

    D. M. Johnston

    November 26, 2009 at 7:52 am

  10. […] Experiment [The New York Times] Walking, biking good for you and the planet: Study [Vancouver Sun] Cost Comparisons of Transportation modes [Stephen Rees's Blog] Prize-winners rise from brownfields [The Globe and Mail] Wind-power project […]

    re:place Magazine

    November 26, 2009 at 9:38 am

  11. As Stephen rightly pointed out, my personal observations do not represent data, as they are rather non-systematic. But this raises another issue, which is peak usage vs. off-peak usage vs. mean usage vs. median usage. The differences make comparison between transportation types more difficult. The “average” usage of a typical car, be it Prius or Hummer, may be less than one passenger, as cars generally spend more time parked than driving…

    “TransLink loves to calculate capacity with SkyTrain (and RAV) cars at 6 persons per metre/2, which on paper looks good and may inflate ridership numbers for statistics.”
    Actually, doesn’t this deflate the ridership numbers? 100 passenges in a car would be a car operating at capacity in Montreal, but at only 66% utilization in Vancouver.

    Pat Johnstone

    November 26, 2009 at 9:44 am

  12. Thanks for the link, Stephen.

    I was disappointed that they made no mention of travel time. Shorter travel times provide real social and economic benefits, and absolutely need to be taken into consideration. Of course, better land use planning will lessen the need for longer trips, which in turn will lessen the impact of travel time. Nevertheless, longer trips will always be necessary to an extent, and it’s important that people will be able to move around the metro area efficiently. This is one area where SkyTrain clearly beats much of the competition studied in this report (although of course we could have built mass metro rapid transit for far less than what SkyTrain costs).

    In a semi-related aside, David made reference to riding the #4 bus in Vancouver in the 90s. I used to ride that bus all the time, and yes it was full because it was one of only 2 buses connecting UBC (and much of the west side) with south Granville and downtown (and onwards from there in many cases). Riding that bus was a terrible experience – it could take nearly an hour to get downtown from UBC. Use has decreased significantly since the 99, 44 and 84 (B-Line or B-Line-like services all) were added.

    Darren Thomson

    November 26, 2009 at 12:05 pm

  13. Another of the many problems with the study is that the costs for tram, LRT and SkyTrain include the capital costs for the new infrastructure required to run the vehicles on. This typically represents additional transportation capacity. To compare this against the Prius, costs for new highway and parking construction should have been included as well. If this was included, it would have shown more the true costs of adding additional automobiles to a city.

    The problem with this “study” is it does nothing to help the debate. Those who can find info that supports their views will I’m sure find some juicy bits but it is full of so many holes that anyone who doesn’t agree with the conclusions can easily dismiss the results.


    November 26, 2009 at 12:20 pm

  14. Yes, a higher stated capacity would, in essence, inflate the demoninator of the ratio making for a lower proportion of riders.

    It’s not like they conduct passenger counts by how many “carloads” they’ve got – this isn’t a drive-in movie theatre.

    Ron C.

    November 26, 2009 at 1:11 pm

  15. I agree that figure 10 is confusing and not well explained. Typical vehicle occupancies for transit modes were calculated from
    OPERATING DATA but for private automobiles they used an average of observed occupancy for TRIPS to or from work. I would think the ‘operating data’ for most private vehicles would show a typical occupancy of almost 0. Most cars have 0 occupants for most of the day. Most transit vehicles, on the other hand, would at least have a couple of customers throughout the day, even when most cars are sitting idles in car parks. Thus would not the occupancy of transit modes during these off-peak times, when the occupancy of most cars is 0, bring down the ‘typical occupancy’ figure for transit?

    Chris S.

    November 26, 2009 at 6:44 pm

  16. How they calculate the number of passengers is the major problem of the survey. In all the years I have used SkyTrain I have never seen anybody actually checking the number of passengers. There is no physical system counting the people entering platforms and I bet that few systems around the world do, except those with smart cards..and even then, do they?.
    As for capacity.. European systems routinely use the figure of 4 people per m2. as maximum possible capacity. Believe me, even with slim Europeans this is rush hour crush (try to stand with 3 friends within a space just a tiny bit more than 3 ft X 3 ft.)

    To talk about something else, the only way to take drivers out of their cars is the stick and carrot approach. Stick as in reducing car lanes and street parking and making parking in whatever few places are left very expensive (this is routinely done in other jurisdictions). Carrot or candy is a free secure park and ride lot with a monthly pass, or a small fee allowing a car driver to park all day long i na secure lot, and get a free 1 day transit pass.

    Yesterday my building hosted an info session about the Evergreen line. It was quite interesting. People in my building don’t want SkyTrain because the elevated structure will be right in front of their windows. They asked why the LRT was nixed. The Evergreen guy said it was as expensive as SkyTrain, much more noisy etc. less capacity etc. assertions that several of my neighbours refuted as –much to my delight— they have seen and used recently built LRT in other towns…they mentioned park and ride lots, incentive for drivers using tarnsit, building stations in a business/ housing complex owned by TransLink…the answer was that all these ideas were good but had not been seriously considered yet.

    Our whole transit system badly need to be reviewed by a panel of foreign experts (including people from Toronto??) who should recommend a solution. But this would only work is the federal government was willing to give as much money to Vancouver as they have promised Toronto, and if the BC Premier and minister of transportation were forced to study several rapid transit systems around the world and officially report their findings.
    This is my Christmas wish for Metro Vancouver.

    Red frog

    November 26, 2009 at 7:26 pm

  17. I probably shouldn’t stir the pot with this one, but LRT can be noisy

    As for the off-peaks, the Expo line is much busier in the off peak PM rush than the AM, as people head downtown for evening events. As for the M line, it’s been “full” for a couple of years now, and pass-ups are now common in the AM at stations west of Holdom when the train consists of only 2 MK-IIs, and also sees significant ‘off peak direction’ travel thanks to SFU and the U-Pass. Commercial Drive is a busy place with people coming and going in both directions.


    November 27, 2009 at 11:01 pm

  18. In the time honoured tradition of brain storming, let me piggy-back-and-front on Patrick’s excellent report.

    1. Eye on the Prize: Walkable Neighbourhoods or Quartiers

    Past modernism, with the era of the suburbanization of North America receding in the rear-view mirror, we face the challenge of building walkable places to live. Places of high density and high urban quality, served with transit, and presenting convenience and amenity.

    At the core of these new neighbourhoods or quartiers most of our needs, including a transit stop, will be within a 5 minute walking distance of the front door of 20,000 or more people.

    Walking distances to a regional transpiration stop can be 10-15 minutes. However, this will increase the catchment to 140,000 or more folks (living in 3.5 storey house rows).

    2. Principle: Considerations About the Resulting Urban Quality Trumps Cost-Based Transportation Analysis

    While it is important to keep track of the costs of the various transportation options, the overarching value we are seeking to implement is ‘urban quality’.

    The urban quality delivered by the Prius and the SUV is not the kind we want. Thus, regardless of their cost implications, automobile modes of transportation present a reduced value to their neighbourhoods.

    The suburb has reached its breaking point largely due to the amount of space that must be given over to store and operate cars in a population where everyone owns one.

    Whether it is in the congested highway, the gigantic parking lot, or the agglomeration of garages and underground parkades, accommodating the personal automobile has destroyed our streets and pushed our houses and shops so far apart that we can’t get anywhere by walking.

    Having to drive everywhere we go reinforces the vicious cycle of more congestion, unliveable streets, and accelerating deterioration in urban quality.

    3. Principle: Walking is the primary form of public transportation

    In “walkable” neighbourhoods and quartiers walking is the primary form of public transpiration (Patrick captures this in the concept of using transit to “extend the walking trip”).

    An initial target might be to make 50% of all household trips walking trips. Eventually, one can see having no more than one or two trips per day requiring transportation: work, and specialty shopping or entertainment. For the latter two, driving off-peak may still represent the best option.

    The Vancouver trolleys have failed to deliver a walkable urbanism. Outside of the West End, Vancouver as we have it today is still very much a “driving city” where walking is a trek, not a daily convenience. Trolleys, I would suggest, also lose value in the high density quartier.

    4. Principle: Redevelopment is triggered by the construction of Transportation

    We heard Mayor Adams of Portland in the summer of ’09 state as a matter of fact that developers buy up the land along the tram lines.

    We need to reach the same level of land absorption for the footprint of a pedestrian shed centred on the regional and commuter transportation stop (Transit Oriented Development or T.O.D. on every WestCoast Express, Expo, Millenium, and Canada line station).

    Ten years ago, building up the pedestrian shed centred on the Maple Ridge West Coast Express Station (WCE) was a non-starter with Council and staff in Maple Ridge.

    Yet, a 2-job family in Maple Ridge choosing to live within walking distance of WCE would have one member commuting to work in Vancouver by walking. The second job-holder could drive to work in, say, Port Coquitlam over the Pitt River Bridge.

    Further on into implementing a regional transportation strategy, the second bread winner would be riding a street car to Port Coquitlam.

    A more likely scenario might be implementing BRT first.

    We can duplicate this TOD pattern all along the feral BC Electric R.O.W. from Kitsilano and Arbutus all the way to Chilliwack.

    5. Principle: We need both long-trip and short-trip transportation

    It is not a question of choosing one transpiration mode over another. We need an array of transportation options, coordinated with a build out plan.

    Community Planning and Transportation Planning should become flip sides of the same coin: good urbanism.

    We also need what the NYC Commissioner called “The Willy Wonka” fare card that works across all modes. Topping up your fare card amounts to pre-buying trips, and flushing cash-flow into the system up front.

    6. Principle: We need a progression of intensifying transportation modes.

    The freedom of the automobile built the suburbs. The roads, servicing, communication and power grids are all in place.
    However, now that the sheer volume of traffic has taken that freedom away, we can reintroduce enhanced mobility by increasing densities with strategic intensification.

    One crucial element is to have the intensification footprint coordinated with a transportation system (T.O.D. or the transit-stop-in-the-quartier approach).

    7. Based on urban design analysis, we can consider the following transit implementation sequence:

    (i) private automobile
    (ii) Buses and Trolleys on dedicated lanes
    (iii) BRT on a dedicated lanes (with signal control),
    (iv) Surface rail replacing BRT
    (v) Skytrain (if we must use that technology) in the ground

    I am not a transportation expert. This approach is based solely on an urban design analysis of Vancouver arterials and strategies for increasing local populations with human-scale intensification. Obviously, it requires input from transportation planners.

    BRT has the advantage of having an initial implementation that can take place in months, not years (think of the growing Evergreen Line fiasco, it could have been BRT for the last 10 years).

    BRT brings a significant advantage for enhancing local urban quality. It can reducing vehicular trips, remove automobile lanes from the arterial R.O.W. By building medians with trees, it can introduce greening and reduce pedestrian crossing distances.

    Furthermore, BRT reserves space in the arterial for future surface rail when the numbers materialize (human and financial).

    8. Density-Plus

    We can increase density over suburban levels by a factor of 15-times and still achieve free-hold, low-rise, dual-aspect affordable housing (I know, a tall order).

    However, considerations about the quality of the fronting urban space—the arterial with transit, increased greening, and reduced numbers of vehicular lanes—are crucial. Missed steps here directly affect the livability of family homes, and the urban quality of the walkable, high-density neighbourhoods.

    Thus, as stated earlier, transit is not the trump card, even if it is a primary element of good urbanism. Yet, transit implementation can be pressed into doing double-duty as a key player for elevating the urban quality in the high density neighbourhood or quartier.

    9. Skytrain Losing Ground

    Once we get used to considering ‘resulting urban quality’ as part of the transportation decision matrix, and understanding that transit implementation plays a shaping as well as leading role in creating good urbanism, we can begin to see short comings in elevated systems.

    When Skytrain leaves areas of low density, the elevated guideway is a blight on the all important realm of the ground plane.

    As Port Moody may be about to find out (and we can see in New Westminster on the south side of Begbie Square) in high density quartiers the only thing that is worse than Skytrain in the air is Skytrain on the ground!

    Thus, Skytrain technology in the neighbourhood or quartier loses priority unless the system runs in subway tunnels. (Patrick has some interesting things to show regarding costs at this point).

    10. Helping to shape an answer to complex choices

    By considering issues of urban quality we have been able to show the transportation modes stacking up differently.

    In some cases (auto, trolley, Skytrain) we see the transportation mode losing ground when implementation is in a high density, human-scale, walkable neighbourhoods or quartiers.

    In other cases (BRT, surface rail, subway) thinking of applications in a neighbourhood or quartier sees the transportation option providing comparative advantages in sustaining or improving the urban quality, as well as delivering on the amenity and convenience of fast, frequent and reliable public transportation.

    Lewis N. Villegas

    November 28, 2009 at 12:40 pm

  19. WRT passenger counts, the Canada Line (not Expo Line or M-Line as far as I know) has automatic passenger counters (using lasers)mounted within the overhead “Fare Paid Zone” signs. It automatically counts the number of crossings into the Fare Paid Zone.

    Here’s a pic by Tafryn Palecloud:

    Ron C.

    November 30, 2009 at 1:13 pm

  20. FWIW, I was on the M Line last Sunday around noon, I counted exactly 20 passengers on our MK-1 car

    On Tuesday morning at 5:45 Am there were 30 on an MK-II


    December 4, 2009 at 10:38 pm

  21. I read this paper with interest, if only because “formal” presentations of this type often get more notice than they deserve, and their underlying methodology is rarely challeneged by amateur readers including politicians.

    Like many others who commented here, I have a big problem with the average occupancy data. First off, the source is in the USA, and it’s unclear whether the information is typical for a more densely used system like Vancouver’s or Toronto’s. Also, the type of fleet examined excludes two important modes: subway (such as in Toronto, not Skytrain) and commuter rail).

    At first glance, one might suspect that the writer wanted to show LRT in a very favourable light compared with Skytrain. However, going over the top with assumptions that can be challenged works contrary to that purpose, if indeed that’s what was intended. (For the record, I prefer LRT, but understand that Skytrain has a role in the Vancouver context.)

    The question of “used” capacity is important to service quality. Riders are very sensitive to wait times, and although it may look good to the bean counters to run only enough service to handle the observed demand, this may actually drive riders away especially during the off peak periods. In Toronto, over half of all riding is outside of the peak, although it is spread over many more hours and therefore generates a lower average load. We may reduce service in the name of “saving” energy and emissions “wasted” on half-empty buses, but in the process lose the very riders we hoped to get out of their cars.

    Another aspect of utilization depends on the nature of neighbourhoods served by a route and the degree to which travel is between many points, or all focussed on a single node. In the former, a vehicle can handle many more trips with a given capacity than if everyone wants to go to one location (a mall, a major transfer station), and there is better chance of bi-directional use of the vehicle. In the worst case where all demand is oriented in the peak direction, the usage of a bus cannot be better than 50% because it will be empty for the counter-peak run, and unless most riders board near the outer end of the line, there will be a lot of “empty” capacity inbound also. That’s how transit works, and people who complain that buses are wasting resources because they are not jammed all day everywhere are way off base (although it may suit a right-wing anti-transit agenda).

    Probably the biggest flaw in this paper (at least in the pdf version) is that it does not include tables of the underlying factors and calculations that would allow a reader to change the assumptions and see what happens.

    Steve Munro

    December 11, 2009 at 8:27 am

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

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

%d bloggers like this: