Adam Fitch wants to share this pdf with us:

Tunnel Vision vs Green Vision

See also, UBC line: The Adam Fitch proposal

[edited on May 30th 2010]

This post is written in the context of the decision to move from the current honor system to the barrier controlled access rail transit system in Vancouver. The capital cost involved by the move is estimated at $100 millions [3]. the cost of fare evasion on the Vancouver rail system is estimated at $3.5 millions per year [4].

Fare evasion

At the beginning the Paris subway had human fare control at its gates. In a move to save on fare control operating cost, the Parisian transit agency, had started to implement turnstile, pretty much in the style seen on the Toronto TTC or the Vancouver seabus…
But Quickly, it appeared that the lack of human control once in the subway system was a pretty good incentive to dodge the turnstiles.

The full display of fare evasion as well as turnstile dodging technique associated with perceived impunity of such behavior, has encouraged wide spreading of similar conducts in the Paris subway, then adopted by people of all conditions as illustrated below


A French countryman, Jacques Chirac, dodging a turnstile in the Paris subway on Dec 5, 1980, station Auber. He was then Mayor of Paris, he will become head of State (credit photo bnf)

That has lead the French transit agency, the ratp to adopt more elaborate faregate, now featuring full door.

  • either the door stay open long enough, and several people (following close enough) can pass the gate with a single fare.
  • or the door close very quickly, and you can’t cross the gate with any luggage, or stroller.

In despite of tremendous investment in fare gating, fare evasion is estimated at 10% in the Parisian subway what is in fact comparable to the access free subway in Berlin [7] and way much more than the 5% measured in Vancouver [4] or 6% on the access free subway of Los Angeles in 2007 [8]

in any case, the turnstiles are an impediment slowing down the flow of transit riders, and the good thing introduced by smart-cards, is that it allow to mitigate this point.

One will note that if fare evasion was the justification for a fare gating, subsidy could be then not necessary. In reality fare evasion on Vancouver transit system is as low as 2.5% system wide (5% on the Skytrain, what amount to $3.5 millions revenue lost a year, which could be only partially recovered by turnstiles) [4], that is nowhere near to able to justify an $100 million investment [3]. In comparison of other systems [7][9][10], there is a generally good level of compliance in Vancouver, observation that the casual observer could have confirmed during the Olympic games, by watching the sometime hour long line-up at the fare vending machine .


It has been a strange and unsubstantiated claim done by the British Columbia government that the fare gates could increase the security on the skytrain [2].
The opposite could be more true: In fact, the diversion of resource going to the maintenance and amortization of the fare gates, instead of human staffing are of nature to make the system less safe.
The creativity of the fare gate dodgers and other smugglers being boundless, it appears that nothing is able to replace human staffing as the picture below illustrates and could also apply to Canada [6]


Police, at Paris Gare du Nord, in April 2007, makes sure everyone is paying his fare!

Nevertheless, the unsubstantiated BC government claim will suffice to justify to allocate $70 millions of subsidy by senior government toward a fare gating system on the Vancouver rail network [3]

The smart card

Another strange association has been done in BC between the turnstile and the smart card. Both can be put in place separably, as it is done on numerous transit network.

smartcard access to the subway of Rennes, France, is done without turnstile. Nevertheless, notice how the smart card readers are placed in prominent position on the farepaid zone line (credit photo wikipedia commons)

The distance based pricing

That seems the only reason a network transit the size of the Vancouver one, could wish to adopt turnstiles control. Controlling entry and exit of the network effectively allows the transit operator to charge by the distance, as done in some rail network, more noticeably on the Hong Kong MTR [5]

Still, in this case, one will find curious the government interference in a matter which should be a priori leads by economic consideration.

Even so: the logic would like that the smart card come first, since it can work currently in the 3 zones model, which is common to numerous network, like the Paris one, and turnstiles in a second phase. Curiously, according to the VancouverSun [1], it is the reverse we gonna see, and suffer all the inconvenience of the faregate, including compromised ease of access to the transit system for people with special needs; from the traveler with luggage, to the wheelchairs, without the advantage of the smartcard.

[1] Olympic commuters sticking with public transit, Vancouver Sun, May 25, 2010

[2] Free rides approach end of the line on SkyTrain, CBC, November 09, 2007.

[3] Canada, BC and TransLink Invest in Transit Security Improvements, press release from Canda government, April 09, 2009

[4] Fare evasion Internal Audit, by PriceWaterHouseCoopers, Translink, September 2007

[5] Notice that such distance based pricing model can a priori apply only to the rail network, and not the bus one. Nevertheless, the smart card can allow implementation of a bus route based pricing like on the model of Hong Kong

[6] Montreal police quell subway brawl, Globe and Mail, December 27, 2009.

[7] TTC fare collection study, TTC, October 2000.

[8] Metro Rail Gating Study, Metro Los Angeles, November 15, 2007

[9] Toronto TTC has a fare evasion of as low as 0.7% on its subway [7], but one should note that all the turnstiles lines are constantly monitored by human staff, via strategic location of ticket office at the stations, as well as additional staffing at rush hours. Non staffed entrance, are equipped of full rotating door, unable to accommodate people with special need (wheelchair, stroller,…). In conclusion, it could be hard to conclude that the low fare evasion number is achieved by turnstiles alone. This observation can be confirmed by the number from the New York subway able to reduce from 3.5% to 0.5% the level of fare evasion through policy measures [7].

[10] The level of fare evasion is estimated at 5% in Montreal, where the subway is accessed throug fare gate, as estimated by the Montreal Transit Agency (stm communiqué, April 21th, 2008 )


The Zurich Model

April 26, 2010

Entry edited on April 27

The Zurich metropolitan area is home of around 1.6 million of people on 2103 sq km [1] and boost one of the highest transit ridership in Europe if not in the world. This has obviously drawn the attention of the transit observers and advocates, speaking then of a Zurich model, but do they draw the right conclusions?

Some authors, like Paul Mees, use the Zurich Model to support the assertion that transit efficiency is not correlated to density, and so can work well in low density suburbia [5], while others noting that Zurich has no subway per sei, could conclude that a dense enough all at grade bus/streetcar service could be a better option than a traditionally more hierarchical network like we know in subway rich cities like Toronto or Munich.

A bit of history

In 1960 and 1973, Zurich has rejected twice subway plan which was more or less aiming to replace its streetcar network, then considered guilty of creating congestion.
That being said, subway work has been initiated before the failed referendum, and ~2km of already built tunnel have later been re-used by 2 streetcar lines, but the city will then mainly bring some efficiency improvement to its network [7].
In 1983, a referendum will give the green light for a cross city rail tunnel with new underground station, enabling the introduction of the S-Bahn, a regional rapid rail network working in a way similar to the Parisian RER, which will be inaugurated in 1990.

This S-Bahn came with a fare integration between the different transport operators, and a more general re-organization of the rail service with what is called “Taktfarplan” or “regular timetable” (that means that you train start at regular interval, usually at least once an hour , e.g. 8:04 , 9:04, 10:04,…) which has been underway on the Swiss national network since 1982.

Though that the frequency of the S-Bahn is not high enough to forget the timetable, the later one are easier to remember, and so simple they can figure on the network map itself.

The settlement structure

Zurich regional area density is low and could be compared to the one of Houston, TX or Edmonton, AB, but eventually such comparison are not telling the real story as much as the topographic map below can say:

extract of Zurich area topographic map showing how the urbanization follow a linear model along valley separated by strong physical barrier

The Zurich area topography draws urbanization along natural valley corridors where sit the railway network. While it is not a pure model of urban clusters; the Zurich area can be compared to the typical north American urban sprawl model either. It has rather developed a linear urbanization model along transportation corridors.

More, one can see that the geography force indirect route to go from point A to point B as soon as those points are not in the same valley, whether you take your car or transit. The usual transit network disadvantage of providing indirect route for “non radial” become less problematic in the case of Zurich.

The Socio-economic structure

Zurich city with a population of 380,000, host 320,000 jobs, for an active population of only 200,000. 50% of its work force come from outside its boundaries.

One should also note that the Zurich city population has declined since the 60’s when it was host of 440,000 people, at the advantage of its suburbs.

The Zurich socio-economic pattern could be compared to Calgary or Seattle for the very high ratio job/population. both inner cities foster good transit ridership number by North American standard, but this can be mostly due to a very centralized job market favoring a good transit market share in the CBD rather than a good transit system per sei [6], as could illustrates the statistic below.

Zurich city [3]92632611

Conurbation Size (in km2) Transit Car Walk/bike
Zurich urban area < 2,100 41 40 19
Seattle urban area 2,100 7 88 4
Seattle city [1] 370 18 67 10
Seattle DT [1] 23 33 39
Calgary city [2] 726 17 75 7
Calgary DT [2] 27 22 51

The table above verifies the correlation between high transit ridership and a strong CBD stated by J. Michael Thomson [9], but there are some interesting facts to underline:

  • Zurich’s walk/bike mode share is pretty weak and in line with the one in Seattle
  • car share in the Down Town of the both North American cities, is below the one in Zurich

The Swiss city is significantly smaller that the considered North American ones, so it could have been interesting to compare thing on a similar size, but the conclusion we can probably draw is that Zurich tends to make a difference in its suburbs more than in its center, where it seems that the high transit mode share is achieved more at the expense of the bike/walk mode than the car on.

The Network

It is constituted of a 380 km S-bahn network. In comparison the Zurich’s streetcar network has 70km of track, and the streetcar route are around 7km long, not venturing much farther than 5km of the Zurich center. The map below overlay this streetcar network on the s-bahn to illustrates the coverage zone of both.

The Zurich s-bahn network and in thin red line, the tram network “roughly” mapped to scale

The Results…or does Zurich has been too far?

The Zurich S-bahn has been an undeniable success with ridership increasing from 159,000 ride/day in 1989 to 356,000 in 2007 [5]. A second cross city tunnel is currently under construction.
The Zurich Transit Priority program [7] has also produced positive effect. Though speed is not the only element of the program improvement, it is still an important one: as an example, the table below provides the evolution of the average speed on the Zurich’s trams network (number from [8])

1960 1970 1990
16km/h 14.5km/h 15,5km/h

On a financial note, The Zurich public transit had a recovery fare box of around 45% in 1997, requiring CHF360 millions of subsidy yearly: it could be due to a political choice of low fare, but one should keep in mind that the Zurich model is not necessarily a self sustained one. The table below represents the operating cost and fare-box recovery (number from [7])

years operating cost farebox revenue % fare-box recovery
1991 522.6 277.5 53%
1992 563.6 286.9 51%
1993 583.2 292.8 50%
1994 600.7 294.4 49%
1995 605.4 298.1 49%
1996 639.2 298.8 47%
1997 660.2 300.0 45%

For matter of comparison, the table below shows the evolution of the ridership in some selected Swiss cities since 1980.

Intra city commuting pattern on last 20 years in selected swiss cities

It is remarkable that the already very high transit market share in both the Zurich region and city has make some gain after the 90’s when it tends to stagnate in the other Swiss cities. But as already noticed when comparing the modal split with Calgary and Seattle, is that this gain has been done quasi exclusively at the expense of the bike/walk share mode which tends to be low in Zurich and not only by Swiss standard:

  • All things happen like if the improvement of the Zurich city surface transit, by better accessibility and frequency but not necessarily significantly improved speed, compete more with the walking or biking option than the driving option, but
  • One should also question if Zurich offer, like other Swiss cities, has not reach a limit touching the “hard core motorists group” which could not consider transit under any circumstance.

The graph below compare the inner city mode share with the region urban area one for some selected Swiss cities:

mode share in selected swiss cities in the inner (ref. 3) and the urban area (ref. 4)

As mentioned before, where the Swiss cities exhibit specificity is more at the regional level than at the local level: Even at the region level, the transit market share is greater than the car one: this is probably contributing to the overall high level of transit share including in the inner area, but there is more to it:

  • The walk or bike share is still non negligible even at the regional level: that seems to indicates that the suburban shape is suitable to those mode, and sufficiently transit friendly
  • The walk and bike mode in the urban area of Zurich seems more important that in the inner city itself: it confirms the fact that the city’s level of Transit service tend to cannibalize those mode more than the car one.
  • car mode share is smaller in Bale and Bern than in Zurich. Those city being smaller, one could have expect the reverse.

While numbers show an higher transit share mode in Zurich than in any other Swiss cities, it doesn’t translate with a smaller car mode share showing the limits of the “Zurich model”.

On a side note, one could note that the German speaking cities perform better than the French one (Lausanne and Geneva), reflecting difference we can also see between France and Germany in term of transportation market share: Eventually beyond an unified public policy of a nation, some cultural trend tied to language could be identified making this policy more or less effective.


As we have seen, Zurich can’t be really invoked to justify good transit in low density area. As well one shouldn’t ignore the specific socio-economic structure of Zurich providing a favorable ground for Public transit.
The defeating of the Subway proposal has not translated in lack of a hierarchical network in Zurich: Eventually Zurich could have been too small to support a 3 level hierarchy (bus-subway-S-Bahn) like in bigger city such as Paris, but the “rapid rail” component is here providing a rapid transit backbone in the Zurich area.
We should also ask the question if Zurich has been too fa in its quest for high Transit ridership: this one lately seems to have been done at the expense of the walk or bike mode rather than the car one, and we should ask the question of what should be an efficient transportation goal:

  • Increase the transit mode share or decrease the car mode share?

If the former is the goal, then Zurich is a model, but if it is the later, the point is more moot.
Overall the Swiss urban areas achieve a remarkable non automobile mode share, and may be we should talk more of a “Swiss model” than a “Zurich model”: The Swiss model is not only constituted of a good local transit, like we could find in other European cities, but rely on an excellent regional rail network. The Public transportation option is still of excellent quality at any level from local to the Intercity train sustaining a real culture of public transportation starting as soon as the kindergarten with some initiatives like the walking-bus …and eventually that could be the real lessons of the “Swiss Model”, Zurich is capitalizing on.

[1] numbers from department of planning and development, city of Seattle, Jan 2004

[2] numbers from Mobility Monitor April 2008, city of Calgary, Jan 2004

[3] numbers from LITRA, Switzerland, 2004 citing statistics from the Office fédéral de la statistique. Those statistic discriminate intra city commuters and inter city commuters: To give fair comparison we have mixed both using a 2/3 weight for intra commuting pattern and 1/3 for inter commuting pattern reflecting the fact that Zurich city has one third more jobs than active residents. The intra commuting number are kept as is for figure comparing Swiss cities between each others

[4] numbers from Projet d’agglomération
Lausanne-Morges (PALM)
, December 2007
, citing the census 2000 from the Office fédéral de la statistique

[5] “Transport for Suburbia: Beyond the Automobile Age”, Earthscan, Paul Mees, 2010

[6] One could see Zach Shaner’s comparison of Vancouver and Seattle illustrating the kind of Transit service offered in Seattle.

[7] Implementation of Zurich’s transit priority program, Andrew Nash, Mineta Transportation Institute, San Jose, October 2001

[8], Dr Felix Laube and Dr Rolf Bergmaier, Transport Engineering Australia, 2000

[9] Great Cities and their traffic, J. Michael Thomson, Littlehampton Book Services, 1977

the agency overseeing the “guided transportation systems” in France publishes some numbers worth to be repeated [1]. Though that the sample sizes prevent to draw definitive conclusions: we can still exhibit some trends: not surprisingly multi year studies tend to show that subways [2] are order of magnitude safer than LRTs [3]. This said, it is interesting to probe the source of tram accidents, what is provided by the graphs below.

partition and severity of tram accidents per year function of their location along the lines

Accidents partition per mode, and transportation modal share

It appears, that the bulk of accidents happen at intersections where they involve third parties. If car are responsible of most of the conflict, it is mainly, pedestrians and cyclists whose pay a disproportionate human toll considering the transportation modal sharing [11]. Furthermore, a study of the Belgium institute on road safety shows that while tram/pedestrian conflicts represent 2.1% of the overall pedestrian conflicts in Brussels, they result in more than 6.7% of vehicle/pedestrian conflicts with severe injuries [4] while that pedestrians represent more than 50% of the overall fatalities on the french trams network [1]. However, a non negligible number of accidents happen outside platforms and crossings: most of them involve emergency braking of the trams, which are responsible of most of the passenger casualties. The french agency has further detailed the pattern of crossing accident, and provides statistic per crossing:

yearly number of tram accidents per crossing, according to their severity type

Comparison with the US

It can be interesting to compare the french statistic to the American one, as reported by the Bureau of Transportation Statistics.

Accident rate comparison between USA and France on LRt and subway network

Though that the accident ratio between subway and LRT witnessed in France is also founded in USA, there is a striking difference in the number of casualties per million of trip.

One explanation to it could be the suicide ratio:

  • statistics are not including casualties due to suicide, but suicide characterization can be different according to the country. Thought that casualties due to suicide are not well documented, anecdotal evidences seem to show that the French authorities are more willing than the North American ones to classify an accident as a suicide: Some officious counting report around 70 suicides per year on the Parisian RATP metro alone [5], when this number is of around 30 in New York City [6], and 15 in Toronto [7]. For purpose of a study on the suicides in the Montreal subway, the researchers have requalified fatalities, considered as accident by the coroner, as suicide [8].

That said, the American LRTs still seem more prone to accident than their french counterpart. We can attempt some explanations to it. .

  • LRT accidents are significantly due to third parties, and eventually the measure of accident/trip is unfavorable to the less patronized US LRT vehicle. This explanation can be countered by the fact that busy LRT lines involve busy pedestrians traffic around their route, hence increasing also the chance of accident.
  • Average speed of french LRTs, usually in the 15 to 20km/h range is significantly slower than their american counter part
  • Design of European LRT could be more permissibe too
    • Front design of low floor european LRT seems less prone to drag pedestrian under the railcar
    • All low floor design reduce the chance of fall inside the car in case of emergency braking
  • More frequent LRT could increase the public awareness of their presence
  • Due to the above factor, French LRT seem also less attractive than their US counterpart to suicide candidate

Compared to even recent American design, the European tram design features all low floor train,with “housed” coupler into an all “soft angle” front design, and offers an unobstructed view fro the driver…all these eventually help to prevent or reduce accident consequences (credit photo, Northfolk LRT: LRTA, Brussel tram in Vancouver: Stephen Rees)

Nantes, a real life example

A tramway accident in Nantes (credit photo: Presse Ocean)

To provide some more reality to the statistic, we provide the example of the Nantes Trams network [9]:
it has opened in 85, has 3 lines, totalizing 42km, and carrying an average of 266000 riders /day.

  • One accident every 2 days
  • One accident in 4 involves injuries

Interestingly enough, according to the Nantes transit agency, their BRT records a rate of accident twice less than their trams, though their buses go faster [9]. It is eventually due to a better designed right of way for the bus than for the trams .

[1] see Accidentologie des tramways, Service Technique des Remontées Mécaniques et des Transports Guidés DES TRAMWAYS, 2006 and
Accidentologie des metros, Service Technique des Remontées Mécaniques et des Transports Guidés DES TRAMWAYS, 2006

[2] French subways include also the “VAL” family of subway

[3] French LRTs include also the guided bus systems

[4] Etude des accidents entre un tram et un pieton en region de Bruxelles-capitale

[5] Suicides dans le métro : deux morts par semaine à Paris, France Info, October 30, 2007

[6] Epidemiology of suicide in the New York City subway system, Sandro Galea and al. , APHA 134th Annual meeting and Session, November 4-8, 2006, Boston

[7] More than 150 people killed themselves in subways from 1998-2007, TTC says, National Post, Rob Roberts, November 26, 2009

[8] Qui se tue dans le métro de Montréal?, Brian L. Mishara, UQAM, Dec 1996

[9] Un accident de tramway en moyenne tous les deux jours, October 7, 2008, Presse Ocean, Nantes, France

[10] The rate per million of passengers is not necessarily the most relevant, but it is the only one readily available from the french statistics, which are averaged on the number of available years after 2003 to provide a more relevant sample size. For USA, to increase the sample size, the accident statistics are the average of year 1994 to 2006, as provided by the BTS 2009 report

[11] Transportation mode share of 14 metropolitan area with tram in France, from “Les deplacements a Nantes metropole Etude N 80, decembre 2009, Insee Pays de Loire, France citing “enquêtes nationales transports et communication 1993-1994, transports et déplacements 2007-2008”, Insee, SOeS and Inrets.