Measuring the Transit supply In Vancouver (*)

When it comes to service delivery, the TransLink narrative goes like this:

Delivered transit service hours have fallen behind the population growth since 2010 reaching levels last in 2008. That is leading to more crowding, more pass-ups and a worsening of the overall transit experience [1][18].

The graph presented to support this thesis is usually a truncated version of the below one:

A problem with this narrative using the total service hours delivered by the TransLink subsidiaries and contractors is that it magnifies the 2010 peak, by including service provided for the Olympic Games. A second issue is that it includes the technical services which could vary greatly without affecting the transit supply. Below is an example of such differences [2]:

route	Revenue hour service	Total hour service	difference in %
All	3,841,860	4,950,000	29%
555	13,500	21,400	60%
96B	42,900	62,400	44%

.

Revenue service or service supply means service dedicated to move transit passengers (passenger can use the provided service). Total service is the revenue service + technical service (deadhead run, layover…). That is matching the APTA definitions. Translink’s reports tend to easily interchange the both terms.

The relatively important difference between the total service and the effective revenue service had already been noticed as an optimization avenue by the 2012 TransLink commissioner’s review [17]. The more fundamental issue is that the service/hour provided is not representative of the Transit supply:

• The replacement of a 40 foot bus by a 60 foot bus wouldn’t increase the service hours per capita, but it could address overcrowding.
• Faster bus routes infer less hours of service but are improving the service offer.
• The replacement of a bus route by a rail one, offering much faster and higher capacity vehicles, can both address crowding while improving the offer, while resulting in a decrease in total service hours.

Seat.Kilometres Supply

The seat.km metric; which needs to be understood as (seat+standee).km in the transit world; is a much better way to evaluate the transit supply, and for this reason is widely used in the passenger transportation industry.

As an example: 1 hour of coach service on the express route 555 using the Hwy 1 HOV lane can provide ~3600 seat.km when one hour of C23 Shuttle bus in Vancouver’s Yaletown, provides only ~320 seat.km. Differences in average speed and vehicle capacity drastically affect the offered service which is reflected by the seat.km metric:

The effect of the introduction of the Canada line service in late 2009 is clear. Though service hours may have stayed stable since 2011, the seat.km supply has slightly increased thanks to a greater use of articulated buses. The advent of routes 96B and 555, having higher speed than average, also provides more seat.km at constant service hours. Is this enough to keep pace with the population growth?

The point is moot. If a downtrend can be observed since 2011, we are nowhere near the 2008 level. The introduction of rapid transit lines tends to exhibit a positive long term trend.

Canadian and International Comparisons

To provide a larger perspective, the Vancouver transit supply is compared to other Canadian metropolitan areas, using numbers as provided by the Transportation Association of Canada [4]. The Vancouver numbers have been normalized to correlate with those provided by the association [5] . Vancouver tends to exhibit favorable trends when compared to its Canadian peers:

Vancouver pales when compared to Megalopolises such as Paris, London or Hong Kong [6], but its Transit supply is much greater than in Portland and comparable to the ones of European metropolises of population size closer to Metro Vancouver, such as Lille or Lyon [7]. Nevertheless, this comes with one caveat: both Lille and Lyon are fed by an important suburban train network which has not been accounted for in the following figure:

The above international comparison is assuming 4 standees per m2 to estimate the vehicle capacity [9]:

system	bus	LRT	Metro	RER/MTR/Skytrain
Vancouver	76			386
Hong Kong	105	146 [10]		200 [10]
London [11]	79	252[12]	728	509
Paris [11]	83	230	586	1772
Portland	76	166 [13]

The Occupancy rate
Is the Transit supply good enough or not?

The occupancy rate [14] can be a good proxy to assess the relevance of the supply: the higher the occupancy rate is, the more likely crowding issues will arise. On the other hand, a low occupancy rate could suggest an excess of capacity.

Crowding experienced locally with a low occupancy rate could suggest that the transit supply deployment is not optimal, but some other issues could arise: A directional demand unbalance makes crowding difficult to address without deploying excess capacity on the underused direction.

Possibly a transit world specific: even the busiest systems don’t achieve an occupancy rate greater than 30%. In that light, the TransLink system appears to be a heavily used one.

It is worthwhile to note that TransLink estimates the average transit trip length at ~8km [15] when TfL estimates the average bus trip length at 3.5km and the Underground trip length at 8km [16]. Similarly the average bus or tram trip length is 3.3km and the subway trip length 5km in Paris. The reliability of trip length data could be an issue but a consequence of longer trips in Vancouver is that TransLink needs to provide more seat.km per trip than London or Paris.

(*) This article has been first published in the December 2014 newsletter from Transport Action BC.

---

[1] Mayors’ council on regional transportation Regional Transportation Investments: a Vision for Metro Vancouver – June 12,2014

[2] Difference between the GTFS data (revenue hr) and the Translink 2013 Annual report (Total service hr). see more in this post

[3] Supply is computed on the first Friday following Labour Day (usually one of the busiest Transit days of the year) of each year from GTFS schedule and fleet deployment observations. The vehicles’ capacity used are the maximum as displayed on the concerned vehicles. see more in this post

[4] Transportation Association of Canada. Urban Transportation Indicators, Fourth Survey. Ottawa :2010

[5] Numbers otherwise differ, possibly due to different assumptions, such as on the vehicles’ capacity. The urban areas, used by the association [4], don’t match either the area covered by the transport agencies, so numbers are subject to caution.

[6] Numbers for Paris come from the Observatoire de la mobilité en Ile-de-France, London numbers from TfL [16] and Hong-Kong numbers from the 2013 MTR Annual report.

[7] Number for Portland, including population, comes from the APTA, and includes the scheduled services provided by Trimet, C-Tran, SMART and Portland city.

[8] Numbers from the Certu (“Annuaire statistique Transports Collectifs Urbains”, 2014) with bus capacity normalized at 83.

[9] Agencies could have different standards (e.g. 6 persons per sqm in Hong Kong). The vehicle capacity is per bus or consist (train) unless otherwise specified. When different vehicle types are used, a vehicle revenue.km weighted average is used.

[10] The capacity is per car. Hong Kong Tram capacity is 125, and Hong Kong Airport train capacity is 120 per car.

[11] Vehicle Capacity number from Report on mobility an transport #1 – Institut D’aménagement et d’urbanisme- November 2014”.

[12] Weighted average of a DLR train capacity (280) and a Tramlink train (200).

[13] The capacity is per vehicle, the Portland streetcar capacity is 200.

[14] Also called Load factor.

[15] Translink: 2014 Business Plan, Operating and Capital – Budget. New Westminster 2014.

[16] Transport for London. Travel in London: report 7. London 2014.

[17] Shirocca consulting Translink Efficiency review. 2012,

[18] A narrative largely echoed by Lower Mainland translink advocates as illustrated here.

Advertisement

Registre du Rhone: some bike helmet datas from France

Below we provide mainly some number coming from [1], using a data base; the “registre du Rhone” (Lyon district, France); recording detailed information for 14 478 cyclist victims of road accident in the Rhone county (France). The data under studies have been collected from 1996 to 2007. The level of detail is very high compared to some other data base, and they can help to clarify the helmet debate:

see wikipedia for the AIS and IIS definition and meaning

Injured Body part among cyclists
reported numbers are only for cyclist where the helmeted/unhelmeted status is known

	helmeted 1461		unhelmeted 5620		helmeted 1461		unhelmeted 5620
Body region	%yes	%AIS3+	%yes	%AIS3+	yes	AIS3+	yes	AIS3+
head	16,6%	1,0%	19,5%	2,3%	242	15	1095	128
face	18,3%	0,1%	24,3%	0,2%	268	1	1367	9
neck	3,3%	–	2,7%	0,0%	48	0	152	1
Thorax	9,6%	1,8%	6,3%	1,0%	140	26	355	55
abdomen	3,9%	0,1%	3,6%	0,2%	57	2	200	14
spine	7,5%	0,3%	5,0%	0,1%	110	5	281	8
Upper Extremity	53,5%	3,2%	44,7%	3,4%	781	47	2513	189
Lower Extremity	33,7%	1,8%	31,2%	1,9%	493	27	1751	107
skin	17,8%	–	15,2%	–	260	0	857	0

This table shows that the helmeted cyclist are less often severely injured to the head, less often injured and seevrely injured to the face. But they are more often injured and severely injured to the Thorax and spine, and more often injured to the upper extremity. This suggesthelemeted cyclist get involved in more severe accident. 2 reasons can be advanced for it

• Helmeted cyclists have a different cyclist practice that non helmeted one (that relate to the infamous MAMIL)
• There is a risk compensation at play along helmeted cyclist

the effect of it is to give an over estimate of the helmet efficiency

The figure below illustrates the % of head injuries oberved for injuried cyclists with an helmet and without:

The bike helmet seems to have no significant effect at reducing superficial head injuries but it reduces the risk of severe head injury (AIS3+) by 50%. However such severe head injuries represent less than 2% of the total cyclists injuries

So we are looking at a global effect of 1%, but severe head injuries can have long term disabilities consequence, so a reduction by 50% of severe head injuries could be considered a worth endeavour:

Below a more detailled view of the head and face injuries specificities:

Head or face Injuried cyclist: type of lesion according to helmet wearing or not

	Helmeted 238		Unhelmeted 1078
lesions	N =	%	N	%
Unconsciousness without additional descripted lesion or head trauma without further indication	194	73,5%	668	49,3%
skin lesion	43	16,3%	459	33,9%
brain	21	8,0%	150	11,1%
skull	6	2,3%	65	4,8%
little brain	0	–	4	0,3%
nerves	0	–	4	0,3%
brainstem	0	–	2	0,1%
Destruction skull and brain	0	–	2	0,1%
Total	264	100%	1354	100%

In despite of a pretty huge database, significativity is not reached on many lesions types.

According to [1], Among the 222 cyclists injuried to the skull or its content

• 30 are dead
• 42 will have severe heavy brain disabilities (IIS3) – that represents 0.3% of the injuried cyclist
• 146 will have light brain disabilities – that represents 1% of the injuired cyclist

The effect of the helmet on long term disabilities seems to be measured on a very small cohort:

	unhelmeted 35	helmeted 3
fatalities	15	0
%fatalities among total injuried	0,3%	–
%fatalities among head injuries	1,4%	–
Severe head disabilities (IIS3+)	20	3
% Severe disabilities among total injuried	0,4%	0,2%
% Severe disabilities among head injuried	1,9%	1,3%
% Other severe disabilities in addition of severe head disabilities [10]	19,8%	21,4%

Again, we don’t reach significativity to be able to conclude that a “bike helmet saves life”, but it could well reduce significantly the risk of severe head disabilities. The number could suggests 50%, but again with very little confidence.

In anycase, the effect is measured in tenth of 1% of the total injuries: the absolute number of victim is very small.

How bike injuries compare to other transportation mode?

When involved in an accident, cyclists have the lowest fatailities rate compared to other individual transportation mode. They have also the best chance to survive an accident without long term disabilities. Urban cyclists are even much safer than rural or suburban cyclists:

	Cyclists
Injury severities	Uurban 6584	Rural 1726	0-10years 3364	Pedestrians 10131	car 53151	Mopette Motorbike 21831
MAIS 1	63,7%	53,7%	72,5%	60,4%	81,6%	61,7%
MAIS 2	28,7%	35,3%	22,4%	24,7%	13,4%	27,0%
MAIS 3	5,8%	7,9%	4,2%	10,0%	2,8%	8,6%
MAIS 4 and 5	0,9%	2,0%	0,3%	2,3%	1,0%	1,5%
fatalities	0,6%	0,8%	0,1%	2,6%	1,2%	1,1%
no disabilities	78,4%	73,8%	88,2%	66,1%	55,3%	73,7%
light diabilities (MIIS1-2)	20,2%	23,8%	11,1%	29,6%	42,8%	23,8%
severe disabilities (MIIS3-6)	0,6%	1,2%	0,2%	1,6%	0,7%	1,3%

When urban cycling is considered alone, helmeted cyclists are observed to have 20% less chance to be severly injured that non helmeted one, whenever involved in an accident, however significativity is not reached [1].

The figures below eventually help to compare the different pattern of injuries, according to transportation mode, and eventually relativize the impact of a bike helmet on the total number of both severe head injuries and severe disabilities.

Injuries type according to transportation mode (from [2])

Long term severe disabilities type according to transportation mode (from [2])- head injuries represents ~60% of the long term disabilities fro both pedestrians and cyclists

those figures provide absolute numbers for the Rhone district (1.7Millions) inhabitant for years 2007-2010…
Even if a bike helmet is able to reduce the severe head injuries and disabilities, by 50%, the saving in term of public health could be relatively insignificant but the more generalized economic cost can be real: It has became clear that an helmet legislation is counterproductive, but even promotion campaigns in favor of bike helmet could be counter productive:

• The money spend on helemet couldn’t justify the health care saving, and could be better invested somewhere else
• Helmet campaign promotions paint cycling as a dangerous transportation, discouraging people to cycle, conducting to a less healthy population

One could object that a low absolute bike injuries number is due to a relatively low cycle ridership. The dramatic increase in cycle ridership over the last decade, especially in urban area, is well documented, and still the number of total injuries has not increased but is decreasing [2]:

Bike accident evolution in France

The more people bike, the more a bike helmet becomes irrelevant…unless you believe pedestrians should wear an helmet too!

---

More citation to come

[1] Vélo et casque, Journée spécialisée, jeudi 28 mai 2009, Rapport UMRESTTE 0908, May 2009

[2] Bilan 2011 de la sécurité routière en France

[10] doesn’t express it like it, but I am assuming it is due to a typo.