PreprintPDF Available

Cost-benefit of bicycle infrastructure with e-bikes and cycle superhighways

Preprints and early-stage research may not have been peer reviewed yet.
Preprint

Cost-benefit of bicycle infrastructure with e-bikes and cycle superhighways

Abstract and Figures

The paper carries out a cost-benefit assessment of an ambitious cycle superhighway infrastructure for the Greater Copenhagen area. In the analysis, we differentiate between e-bikes and conventional bikes to account for differences in the estimated consumer surplus with respect to travel time savings, external health benefits and safety costs. The assessment shows that the investigated bike infrastructure is beneficial with a rate of return on investment between 8%-28% depending on the assumptions. It is revealed that increasing shares of e-bikes render lower benefits. This is because e-bikes provide lower health benefits, which cannot outweigh the increased surplus from travel time savings. The study also suggest that most benefits are non-local benefits. This further suggests that, while bike infrastructure investments historically have been undertaken by municipalities and local authorities, it could be relevant to revise the role of the investment strategy from the state perspective.
Content may be subject to copyright.
1
Cost-benefit of bicycle infrastructure with e-bikes and cycle superhighways
Jeppe Rich*
1
, Anders Fjendbo Jensen*, Ninette Pilegaard*, Martin Hallberg#
* Technical University of Denmark, Bygningstorvet Bygning 116, Kongens Lyngby 2800, Denmark
# MOE Artelia Group, Buddingevej 272, 2860 Søborg, Denmark
Abstract
In this paper a cost-benefit analysis is performed to evaluate an ambitious cycle superhighway infrastructure in
the Greater Copenhagen area. In the analysis, we separate the effects of electric and conventional bikes and the
estimation of user benefits thus allow differentiation with respect to travel time savings due to different travel
speed profiles and different external effects regarding health and safety for different bicycle technologies. The
cost-benefit analysis show that the proposed bicycle infrastructure has a positive net present value with an internal
rate between 6%-23% depending on different assumptions. The cost-benefit performance of the analysed bicycle
infrastructure thereby exceeds other types of network infrastructure that is often prioritised. At the specific level,
it is found that larger shares of e-bikes implies lower benefits as these bikes provide lower health benefits and
larger accident costs. These costs exceeds the higher surplus from travel time savings. The study also show that
most benefits are non-local benefits, suggesting that it could be relevant to revise the investment strategy to have
a national perspective rather than a local perspective at the municipality level, which is the common practise today.
Keywords: Bicycle infrastructure, e-bikes, cost-benefit, external costs, travel demand, cycle superhighways.
1
Corresponding author rich@dtu.dk
2
1 Introduction
Historically, bicycles have mainly been considered a transport mode for shorter distances and with
little potential for mode substitution with respect to trips and total mileage on longer trips. As a result,
from an overall infrastructure and societal welfare perspective, bicycles have received little attention
compared to infrastructure investments for cars and public transport. This is also reflected in the
recent Danish national infrastructure investment plan (Danish Government, 2019), where national
bicycle infrastructure investments represent 0.7% of the total investment between 2020 and 2030
while bike travel accounts for close to 14.5% of all journeys and 10% of the total daily travel time
(Christensen and Baescu, 2019).
The recent popularity of e-bikes (electric bicycles) combined with better bicycle infrastructure, e.g. in
the form of cycle superhighways, represents two fundamental changes that can potentially shift the
level of welfare benefits of bicycle infrastructure. Cycle superhighways is an initiative that provide
improved conditions for cyclists with particular focus on commuters. A specific focus area is to
ensure that the bicycle infrastructure is connected and goes beyond municipality borders to provide an
improved bicycle alternative for longer trips.
In this paper we present a method for assessing the welfare economic effects of bicycle infrastructure
when different bicycle technologies such as bikes, e-bikes, and speed-pedelecs exist at the same time.
While the estimation of user benefits effects can be made by adapting the rule-of-a-half
approximation and using the standard cost-benefit framework (e.g. Kidokoro, 2004), the existence of
different bicycle technologies in combination with new bike infrastructure introduces several
challenges when calculating the welfare economic effects. The main challenges are expressed in the
following statements:
1) Different technologies have different travel speed profiles. This means that the user benefits
are different across technologies even for similar origin-destination pairs. Because e-bikes
travel faster than conventional bicycles, travel time benefits will be higher if a common value-
of-time is used.
2) Bicycle infrastructure may be designed to benefit specific technologies. As an example, it is
possible that super cycle highways will benefit the faster e-bicyclists, because of fewer
obstacles and a better path design. This generally means that a dedicated route choice is
required for every technology where travel time is measured by accumulating travel time for
all link types and by measuring waiting time at stops and crossings.
3) Different bike technologies will have different health and safety effects. Conventional bikes
have higher positive health effects and lower safety risks compared to e-bikes.
4) The speed at which the different bike technologies enter the market, e.g. the share of e-bikes
at a given point in time, is generally uncertain. However, due to the points raised above, the
accumulated net present value of benefits could be affected by the speed of the market
penetration.
The contribution of the paper is to describe these challenges, propose a practical solution how welfare
benefits of implementing infrastructure can be measured when accounting for differences between
types of bicycles. This is presented in a specific case of a cycle superhighway infrastructure for
Copenhagen.
In summary, the challenges raised above generally mean that e-bikes could have higher expected user
benefits and at the same time lower external benefits compared to conventional bicycles. As a result,
it is not clear if the net benefit of e-bikes is higher than conventional bikes. However, aggregating
different bike technologies would in any case lead to inaccurate and uncertain measurement of
benefits. It is the aim of this paper to individually consider and implement all of these effects in a
welfare economic analysis of bike infrastructure.
3
1.1 Literature review of bicycle infrastructure cost-benefit assessment
There is relatively firm evidence that bicycle infrastructure affects bicycle demand (e.g. Krizek et al.,
2009; Amiret al., 2016 and Van Goeverden et al., 2015). Not only will bicycle infrastructure affect the
number of trips, but also the speed of travel (Eriksson et al. 2019, Schleinitz et al. 2017) and the
choice of route (Stinson and Bhat, 2003; Zimmermann et al. 2017). For cycle superhighways,
increased speed on such infrastructure is investigated in a recent paper by Houkes and Klingen
(2019). The results indicate that e-bikes travel at significantly higher speeds between cities when these
are connected by cycle superhighways. A specific Danish investigation was conducted for Denmark
(Supercykelstier, 2019). It was found that the number of cyclists along the routes increased by 2 -
68% within the first year and for these new cyclists, 9 - 26% were reported to come from car. Skov-
Petersen et al. (2017) used linear regression to assess the number of cyclists on a specific route which
was updated to a superhighway route. Nearby parallel bicycle routes were used as reference. The
reference case was examined before the improvement, as well as one and two years after the
improvement. By controlling for variables such as weather, seasonal effects, and time of day, the
results showed a significant increase in the number of bicycle trips of up to 61% for certain periods of
the year.
However, while there is evidence that infrastructure affect demand in many different ways, there are
very few studies that examine the combined benefit-cost performance of bike infrastructure. In
Denmark, the only recent relevant study is by Incentives (2018) who examined the same bicycle
infrastructure investments as in this paper. However, their study was limited in mainly three ways.
Firstly, demand effects were not based on a model but only on coarse assumptions about how bicycle
behaviour was expected to evolve. Secondly, there was no distinction between e-bikes and
conventional bikes. Thirdly, values for external effects (health and accidents) were largely outdated
and not based on the newest literature. They found benefits translated into an internal rate of return of
11%, which by all means is an impressive result compared to most other public transport and car
infrastructure investments.
2 Cost-benefit analyses of bicycle infrastructure
A cost-benefit analysis of an infrastructure investment is composed of different elements. First, direct
user benefits are calculated based on changes in the consumer surplus resulting from changes in travel
time or costs. Secondly, a calculation of the external costs from the project, i.e. possible costs related
to congestion, health, safety or emissions. Thirdly, construction costs and maintenance costs are taken
into account and finally the effects on tax revenue. The result of the cost-benefit analysis is provided
as the sum of all these elements for the starting year and future years represented as a net present
value (NPV) for the chosen base year.
Below we provide a description of the calculation of these terms.
2.1 Direct travel time benefits
For any type of transport infrastructure or policy, the calculation of travel time benefits can be based
on a representation of demand matrices and corresponding matrices that express travel time and costs.
The calculation is based on a baseline, e.g. a ‘do-nothing’ situation, and a scenario that represents a
‘do-something’ situation. The calculation of the user benefits can be based on the well-known ‘rule-
of-a-half’ approximation which requires that every relevant transport market is evaluated separately to
avoid aggregation bias (Kidokoro, 2004). In the situation with different bike technologies, the change
in consumer surplus  is calculated for trips between origin and destination by mode and
technology is given by
4

 
Here represents the cost matrix in the baseline scenario, while represents the cost matrix for the
project scenario. Moreover, and represent the corresponding demand matrices. Cost matrices
are expressed in monetary units while demand matrices are expressed in number of trips. The cost
matrices and are includes a monetised value of travel time. This calculation is based on a value-
of-time (VoT) multiplied by the travel time. In this case, it involves the direct travel time and the
added waiting time at intersections. The projection of VoT for future years is based on GDP growth in
order to account for productivity effects over time.
When analysing bike infrastructure, there is a potential substitution effect between transport modes.
Indeed, it is possible that more people will use bikes instead of cars if it becomes easier to travel by
bike. This will reduce the number of cars in the road network and thereby reduce congestion. Only if
the underlying demand model represents such effects, the aggregated consumer surplus for all
transport modes will adequately represent the combined effect.
In the present paper, we ignore effects within the other transport markets (e.g., car and public
transport) and consider only the direct bicycle demand benefits. That is, we consider only the user
benefits  for bikes
      

In this sense, we include the substitution effects that result from increased bicycle demand due to
improved infrastructure. However, we ignore the second-order effect it may have in other markets,
e.g. congestion effects in the transport network. The reason why these effect are ignored is because,
for long-range scenarios, it is generally uncertain how such effects evolve. Among other things, these
effects depend technology development in the car market and the level of congestion. The latter
depend on the development of the car road network, which we consider fixed throughout the different
scenario. Also, we expect these effects to be modest as it is a second-order effect with respect to
bicycle demand.
The calculation of user benefits for bikes typically depends on a corresponding demand model. In this
paper, the demand model is formulated as a large-scale discrete-choice model for mode of transport
and choice of destination. The model is linked to a bicycle assignment model that distinguishes
between different bike technologies. In this paper, we deliberately focus on the cost-benefit
assessment and refer to Hallberg et al. (2020) for more details with respect to the demand model.
When calculating the user benefits, we consider the effect for the base year 2030. Benefits are then
interpolated from this to the years 2020-2069. This is accomplished by applying the population
growth and assuming that the growth in demand is proportional to this growth rate. We believe this is
an uncontroversial assumption because demand, all other things being equal, should be approximately
proportional to the underlying population growth when congestion effects can be assumed minimal
(Paulsen et al., 2019).
To calculate the user benefits for the years in the construction phase, it is assumed that the
construction phase is from 2020-2030 and that there is a linear phasing-in of benefits during this
5
period. Hence, in 2020, only 10% of the benefits are included in the calculation, while in 2030, all
benefits are included. Because the project period is 50 years, this has relatively little impact on the
total project evaluation.
2.2 Health, safety and emission benefits
Most evidence indicates that active travel modes such as bike and walk have positive health effects
(Brown et al., 2015). However, as demonstrated in a recent large systematic review of reductions in
all-cause mortality from walking and bicycling (Kelly et al, 2014), there is substantial variation
between the different studies. One question is how much active modes contribute to the total physical
activity. If cyclists take health benefits into consideration when making their travel choice and thereby
substitute other physical activities with cycling, the health benefit are partly internalized as discussed
in Börjesson and Eliasson (2012). Börjesson and Eliasson (2012) present some evidence of
substitution based on a traveller survey. It was found that health benefits could be overestimated by as
much as 60% if this substitution is ignored. However, no studies focus exclusively on the magnitude
of this substitution and the substitution degree remains uncertain to this point in time. As a result,
many studies appears to assume zero substitution (Genter et al., 2008; Brown et al., 2016). This
assumption is supported by more recent evidence (Foley, 2018; Foley et al., 2019) that suggest that
bicycling does not substitute other physical activities.
Another question when measuring physical activity is to what extent the use of e-bikes represent
exercise. Physical activity is here often measured using the Metabolic Equivalent of Tasks (MET),
where a value of 1 represent the metabolic rate associated with being at rest. Haskell et al. (2007)
suggests that in order to promote and maintain health, the exercise intensity should be at least 3 MET.
Several studies have indicated that the average MET while riding an e-bike is well above 3 and most
likely between 5-7 MET as suggested by De Geus et al. (2007). Hence, usage of e-bikes is indeed a
physical activity, but compared to usage of conventional bikes, the level of physical activity is lower.
Due to this, it is expected that health benefits are lower for e-bikes compared to conventional bikes
(Berntsen et al., 2017).
A recent study (COWI, 2020) estimates the Danish external costs of biking and distinguished for the
first time between bikes and e-bikes. The estimated values are based on updated values from the
recent review in Foley (2018) and Foley et al. (2019). These studies link the effect of physical
exercise with specific diagnoses such as cancer, diabetes and heart disease and is monetised in a
Danish unit cost per kilometre by applying the value of a statistical life’, which is commonly used in
appraisals schemes. The value is also based on the assumption that cycling does not substitute other
physical activity. This is based on recent findings in Berntsen, et al. (2017) and Foley et al. (2019) as
discussed above. However, it remains unclear to what degree such assumptions can be transferred
between different countries with different bicycle cultures.
While bicycling gives rise to positive health benefits, it also gives rise to negative effects due to
increased accident rates. Accident costs are based on the estimated loss of future productivity when
involved in accidents (distinguishing between accident types: fatal, seriously injured and injured) and
the cost of medical treatment. In the study by COWI, estimates were corrected for systematic
underreporting of non-fatal accidents and the difference between accident risk for bikes and for e-
bikes was estimated.
The empirical basis for understanding safety aspects of e-bikes is relative limited and the uncertainty
is thereby significant. For Denmark, e-bike safety has been studied in HVU (2019). Based on
observed data for bicycle accidents, it was found that the risk factor of riding e-bike was
approximately 70% higher for e-bikes. While the data does not allow a detailed empirical assessment
of why this happens, a hypothesis is the effect is related to increased speed and the fact that many e-
bikers are elderly and thereby more vulnerable when involved in accidents.
6
The updated value of the external costs related to cycling are presented in Table 1 below and
compared with the previous values. It is decomposed into effects relating to health and to safety
effects. These unit costs are used by the Ministry of Transport and Housing for appraisals of transport
policy and infrastructure projects (TERESA 5.0, 2019). Note that there was previously no
differentiation between different bike types, conventional bikes and e-bikes.
The external costs are applied directly to the difference in gross mileage for the different scenarios,
and values are forecasted using GDP per capita growth.
Version
Type of bike
Accident costs
Health costs
Total
Previous external costs (Teresa
5.08)
Bike
1.17
-3.56
-2.39
New external costs (Teresa 5.09)
Bike
1.49
-7.84
-6.35
E-Bike
2.51
-6.27
-3.76
Table 1: External costs for bike in 2020 prices.
The unit costs can be compared to values derived from using the HEAT assessment model (Kahlmeier
et al., 2018). If assuming that for one person travelling 1 km every day in a year at a speed of 14
km/h, it is found that the total health benefits are in the order of 337 EUR per year. The main driver is
a reduction in premature deaths by 0.0000800 combined with a monetized value of a statistical life
(VSL) of 4,540,000 EUR/death. If assuming an average trip length of 4 km per day, the average km
health benefits is close to 7 DKK/km and comparable to the values in Table 1.
3 Case study for Copenhagen
In the following we consider a two bicycle infrastructure scenarios for Copenhagen. Firstly, the
baseline, which represents the infrastructure of today, and subsequently an ambitious cycle
superhighway network in three stages. The infrastructure scenarios are then combined with different
assumptions regarding the mix of bicycle types. The two scenarios are presented in Figure 1 below.
7
Figure 1: Illustration of geographical area and three bicycle extension scenarios.
Readers can refer to Hallberg et al. (2020) for a detailed analysis of consumer surplus effects for all
other intermediate development stages of the network.
The network covers the greater Copenhagen area for which bicycle trips are competitive. This
boundary excludes long-distance trips beyond 70 km, which are largely irrelevant for assessing
welfare effects of bicycle infrastructure. We will refer to the network scenario as ‘Scenario 3’ as it
complies with the notion in Hallberg et al. (2020) and further allow us to distinguish between network
scenarios and technology mix scenarios to be defined later.
3.1 Cost of bicycle infrastructure
The implemented project consists of a comprehensive expansion of the cycle superhighway network
in the Copenhagen Region. In the scenario, the total length of the network is increasing from 162 km
to 749 km. Construction and maintenance costs vary according to the location of the routes. We use
the cost estimates that are applied in Incentives (2018). For completeness, we present an overview of
the applied cost estimates for five different route categories.
Inner ‘finger’ routes represent connections between the city centre and surrounding areas while the
remaining ‘finger’ routes are located further out but still connect in the same direction towards the
city centre. Radial routes are those that cross the city, while open country routes are located in rural
areas outside the city.
8
Category
Construction cost per km
Maintenance cost per km/year
Inner ‘finger’ city routes
1.9
0.1
Inner ‘radial’ routes
3.0
0.3
Radial routes
1.9
0.1
‘Finger’ routes
3.5
0.3
Routes in open country
3.3
0.3
Average
2.6
0.2
Table 2: Average construction and maintenance cost per km cycle superhighway path, million DKK,
2020 prices.
The costs are estimated by considering the specific path design of 55 specific extensions (Incentives,
2018) and by considering the need for additional connections, underpasses or overpasses or changes
to the bicycle path profile as described in Supercykelsti (2019).
We assume that the infrastructure will be implemented in 2030 but developed over a 10-year period
from 2020 to 2030. Due to this, a linear phasing-in of costs is assumed for the period. This includes
both construction costs and maintenance costs.
3.2 Assumptions related to bike technology
Different combinations of bike technology give rise to different speed profiles and different consumer
surplus effects. It has not been possible to model the composition of the bicycle fleet directly. As an
alternative, we have superimposed different realistic mix combinations when evaluating the demand
model. An enumeration of all technology scenarios, as proposed in Hallberg et al. (2020), is shown in
Figure 2.
Figure 2: The different combinations of technology in the baseline (M0) and in the scenario (M1).
The technology mix in the scenario M1 is represent a realistic future scenario for 2030, while the
baseline M0 reflect the share of the different technologies in 2020.
3.3 Other assumptions
The project is evaluated over a period of 50 years, which is standard for transport infrastructure
appraisal in Denmark. During the project, it is assumed that the infrastructure is maintained on a
0% 20% 40% 60% 80% 100%
Scenario
(M1)
Baseline
(M0)
Market share (%)
speed pedelec
e-bike
Regular bike
9
continuous basis. As a consequence the infrastructure uphold its full value at the end of the project
period. In other words, the residual value of the project is assumed equal to the initial value of the new
infrastructure because of maintenance. The residual value of investment is equal to the discounted
value of this residual, which is equivalent to 537 million DKK in 2020.
While we ignore second-order congestion effects that result from fewer cars due to mode substitution,
we include other effects that relates to car drivers. The estimated costs for car drivers that result from
annoyance and additional congestion, e.g. due to priorities for bikes, crossings and the fact that the
bicycle infrastructure may take up space that could otherwise have been used for cars, is estimated to
a NPV of approximately 800 million DKK and are based on a previous valuation from Incentives
(2018).
Other project specific assumptions are shown in Table 3 below.
Type
Value
Calculation period
2020-2069
distortionary effects
10%
Value-of-time
91 DKK/hour (2020, forecasted based on GDP)
Substitution effects for cars and other modes
Excluded due to uncertain prediction of congestion effects
Business trips
Excluded due to uncertain prediction
Driving costs for bikes
0.39 DKK/Km
Value-of-time for non-adults
50%
Taxation consequences from health benefits and safety costs
100% for health and 37% for safety
Discounting rate
4% (between years 1-35), 3% (between years 35-70)
Future consumer surplus effect based on interpolation
Proportional to population growth
Table 3: Project specific assumptions, year 2020 in 2020 prices.
Distortionary effects represent effects that result from distortionary labour taxes and their implied
effect on government tax revenue. When a project is financed by increasing labour taxes this distorts
the labour supply and hence implies an additional welfare loss. The common practice for Danish
appraisals studies is to assume a distortionary effect of 10%. The effect is calculated with respect to
the net effect for the public budget. In addition, the labour supply decision is distorted by changes in
transport costs related to commuting and business travel and hence, distortion effects are also
calculated for these.
In the paper, we show the results of the CBA using both old and new external costs related to the
health effect of the use of bikes. This is to underline the sensitivity of this particular assumption. For
safety costs, we apply only the updated values as these are fairly comparable to the previous values.
For the old external costs, we assume the same ratio between the values for conventional bicycles and
e-bikes as in the new external costs, both for health and safety costs.
3.4 Assumptions of demand model
Demand is based on a model for bicycle demand (Hallberg et al., 2020). The model was estimated on
data from the Danish National Travel Survey (TU data) and includes a joint model of choice of mode
and choice of destination. In addition, the mode for bikes has been divided into types of bike
technologies, e.g. conventional bike, e-bike and speed-pedelec. These different types, have been
defined with different speed profiles in the network (based on literature) and is further combined with
different assumptions regarding their market shares (refer to Figure 2). This is used to calculate
10
expected demand and resulting surplus effects in a comprehensive network for Copenhagen based on
Open Street Map (OSM).
The demand model is estimated as a nested choice model for the combined choice of mode (walk,
bicycle, public transport, car as driver and car as passenger) and destination. The model includes a
range of parameters including travel-time and travel cost variables. Travel time for bicycles is
expressed as a weighted average over different bicycle technologies, where the market share for the
different technologies are imposed externally as discussed in Section 3.2. This enable an assessment
of travel time differences across different technology scenarios. The calculation of travel time is based
on a dedicated route choice model and assumptions with respect to travel speed for the different types
of bicycle links (e.g., ordinary bicycle paths and cycle superhighways).
The model is calibrated (based on pivoting) to a 2030 OD bicycle matrix from the Danish National
Model (Rich and Hansen, 2016). This means that in 2030, the model reproduces the number of trips in
the baseline matrix when applied to the base network. For all future years, it is assumed that benefits
develop proportionally to the underlying population growth as described previously. The growth in
population for the area of interest has been based on official projections from Statistics Denmark for
2020-2070.
Transport indicators derived from the underlying transport model show that bicycle speed has
increased by between 2-4% as a result of the upgraded network. The number of trips increases by 4%
and the average distance increases by as much as 7-8%. While the latter represents a relative large
increase, it is worth remembering that the cycle superhighway network is primarily aimed at
improving conditions for longer distances. As newly generated bicycle trips are often substituted (to a
relatively high degree) from cars and public transport, this leads to longer distances.
One of the strongest assumptions of the demand model is that mode and destination choice is generic
across bike technology, as it was not possible to estimate different parameters for different types of
bikes. This can be seen as a critical limitation in the sense that an increase in the number of e-bikes
will generally not affect the overall bicycle preferences. This is not likely to be correct as faster bikes
in the future is expected to travel longer and take advantage of the increased speed. Hence, while on
the one hand 7-8% seem as a high increase, in the examined scenario, it encompasses also the distance
effects that may result from changed bike technology.
3.5 Cost-benefit results
The cost-benefit results are presented below in Table 4. We have compared a baseline (without any
changes to the bicycle network) with the fully developed network in Scenario 3 shown in Figure 1.
These two scenarios have then been compared for the two different technology scenarios M0 and M1
in Figure 2.
Several results can be revealed from the analysis. At the overall level, all scenarios render highly
positive benefits that range between 6% and 23% when measured in terms of the internal rate of
return. The single most important factor is the external health benefits that account for most of the
benefits. External health benefits are societal benefits that measure the effect of healthier people who
will live and work longer. However, most importantly, the society saves expenses for the health
system. Internal health benefits are benefits that are directly linked to the users and represent
improved quality of life. The effect of using the updated valuation is by all means significant as health
effects benefits more than doubles.
The fact that the health benefits are so dominating implies that related components, such as budgetary
consequences from health benefits’ are relative important as well. If the percentages are reduced,
benefits will be lower. In order to test the sensitivity of this assumption, a value of 50% for health
11
benefits was tested. This change reduces the internal rate of return on investment to 7%, 5%, 21% and
19% respectively (refer to the last row in Table 4 below).
While the user benefit effects are generally less important, it can be noted that the technology scenario
M1 has a higher user benefit for e-bikes compared to the baseline M0 technology scenario. This is
because the share of e-bikes increase. However, if we consider the total user benefits across all
technologies, the change in consumer surplus (by upgrading the infrastructure from the baseline
network scenario to Scenario 3) is higher for the baseline technology scenario M0 when compared to
the technology scenario M1. While this can seem counterintuitive, it is a consequence of the
difference of speeds for the bike types in the network. For conventional bikes, the increase in speed is
bigger than for e-bikes, when travelling on cycle superhighways in Scenario 3 instead of the base
scenario. This is due to the fact that there are limits for how fast you can travel without compromising
safety. As an example, for speed-pedelecs, it would not be reasonable to increase speed up to the
maximum velocity of these bikes because it would simply not be possible while at the same time
manoeuvring safely in the transport system. Hence, speed is not simply adjusted linearly but has a
decreasing slope for higher speeds. Consequently, the isolated effect of travel speed is lower in the
scenario as compared to baseline.
Net Present Value
Year 2020 (million DKK)
Old valuation of health effects
Updated valuation of health effects
Type of bike
Baseline M0
Scenario M1
Baseline M0
Scenario M1
Construction costs
-2249.95
-2249.95
-2249.95
-2249.95
Construction costs
-2787.45
-2787.45
-2787.45
-2787.45
Residual value
537.50
537.50
537.50
537.50
Cost of maintenance
-4704.45
-4704.45
-4704.45
-4704.45
Consumer effects for bike
5204.55
4904.48
5204.55
4904.48
Travel time savings
Conventional
4885.99
3778.32
4885.99
3778.32
Electric
481.91
1306.51
481.91
1306.51
Travel costs
Conventional
5.44
3.77
5.44
3.77
Electric
0.92
3.25
0.92
3.25
Intern health effects
Conventional
-145.28
-100.70
-145.28
-100.70
Electric
-24.43
-86.67
-24.43
-86.67
External road safety effects
-6202.22
-6534.67
-6202.22
-6534.67
Accidents
Conventional
-5177.61
-3794.65
-5177.61
-3794.65
Electric
-1024.61
-2740.03
-1024.61
-2740.03
Other effects
12611.15
11214.18
29129.13
26078.95
Tax effects
Conventional
0
0
0
0
Electric
0
0
0
0
External health effects
Conventional
11348.18
8317.03
25072.61
18375.59
Electric
1068.23
2856.68
2360.15
6311.54
Labour supply distortion
Conventional
385.63
82.51
1758.07
1088.37
Electric
121.88
325.92
279.33
746.98
Labour supply benefits
Conventional
474.62
368.14
474.62
368.14
Electric
106.82
285.67
236.01
631.15
Disbenefits for car drivers
-818.16
-818.16
-818.16
-818.16
Total Net Present Value
4659.09
2629.59
21177.07
17494.36
Rate of return on investment
0.08
0.06
0.23
0.20
12
Table 4: Presentation of detailed cost-benefit calculation for base Scenarios and Scenario 3 for the
baseline technology M0 and scenario M1. All prices are for 2020.
When we compare the results, expressed as the rate of return, with other types of investments in road
and public infrastructure projects, the current benefits are generally higher. Traditional public
transport investments are often in the range of 3-5% (Annema et al., 2007) and in some cases even
negative. The performance of road infrastructure varies with the type of investment. Intersections and
local projects can have very high returns, while motorway projects and larger projects have lower
returns and in most cases is below 8-10% (Hyewon and Euijune, 2019).
While the modelling of bike demand is not the aim of this paper, it is, however, clear that results
depend heavily on the outcome of the demand model. In this case in particular, the net generated
bicycle kilometres are important, as they are the main driver of the health benefits.
The demand model is a first attempt to integrate bicycle technology in a cost-benefit analysis of
bicycle infrastructure investments. However, it is clear that more research is needed as how different
bike technologies should be integrated. It is definitely not an appropriate assumption to estimate
generic mode and destination choice across these bicycle types as we hereby implicitly assume that
faster e-bikes will travel approximately similar distances as conventional bikes. However, at present
there are few revealed observations for e-bikes, and it has not been possible to estimate separate
parameters for these.
The analysis is generally based on many assumptions, some of which are conservative and others
which are optimistic. In Table 5 below, all important assumptions are considered with respect to their
expected consequence for the assessment.
Assumptions
Comment
Impact on final benefits
Only benefits for bikes are
considered.
Underestimates congestion benefits
for cars and generally discards
benefits in other markets. It also
means that feed-back effects from
reduced congestion in the car network
are not included.
Underestimate
No business trips considered.
Underestimates the value of travel
time savings because the value-of-
time for these trips is higher.
Underestimate
The geography of Copenhagen may
differ from other cities in terms of
lack of hilly terrain and weather
conditions during winter.
Lack of hilly terrain may
underestimate the preferences towards
e-bikes. However, as shares of e-bikes
are super-imposed prior to the
assessment this can be adjusted
accordingly.
Uncertain
Choice model (destination and mode
choice) is generic for bike types and is
based on TU data (mostly) without e-
bikes.
Due to this limitation, it is implicitly
assumed that a fast bike in the future,
will travel largely the same distance as
a conventional bike of today. Hence,
we discard behavioral distance effects
and focus only on distance effects that
result from the improved network (if
people have similar preferences as
today).
Underestimate
Driving costs similar across different
types of bikes.
Underestimates costs for e-bikes as
these have higher costs.
Overestimate
We are not considering bike and
public transport combinations.
It could be that the infrastructure
could also benefits such combinations
and thereby lead to higher benefits.
Underestimate
Bicycle congestion not considered.
Could potentially increase the effect
of upgrading infrastructure although
the effect is likely to be small.
Overestimate
13
Uncertain if bicycle infrastructure can
actually accommodate the speed
increase in a safe way.
Could lead to lower speeds.
Overestimate
In the bicycle route model we use a
shortest-path only.
It could lead to an overestimate of
benefits because, in reality, people do
not travel by the shortest route.
Overestimate
Bike travel does not substitute other
physical activities.
This is likely not the case for all
people. If, however, substitution were
to be considered, it would be
necessary to also compensate the
‘saved time’ for these other activities.
Overestimate
Public revenue effects
Health effects and other externalities
are not assumed to have effect on the
public revenue. These are included as
‘0’ in Table 4 to highlight that this is
an element that may need
consideration in other studies.
Underestimate
Bike travel does not cause people to
be less physical active.
This obviously renders higher values
for the external costs and thereby
higher benefits.
Overestimate
Table 5: Assumptions and their expected consequences for the assessment.
The geography of Copenhagen may differ from other cities in terms of lack of hilly terrain and
weather conditions during winter. It is not strictly clear how this could affect the assessment. Lack of
hilly terrain may underestimate the preferences towards e-bikes. Weather conditions during winter
could have a similar effect. However, as shares of e-bikes are super-imposed prior to the assessment
this can be adjusted accordingly.
The external costs may also implies a second order effect that result from budgetary consequences of
health benefits and safety costs. This effect would have a positive effect on the evaluations. However,
the size of this effect is uncertain and it is therefore not included.
It is difficult to assess the net effect of the limitations in Table 5 as it would require a substantial
model development. However, with respect to the derived transport effects, a few observations can be
made. As the distances increase by 7-8% solely due to the changes in the bike infrastructure, this
suggests that the demand model is relatively sensitive to distance. However, considering the fact that
the distance effect from increasing shares of e-bikes and speed-pedelecs by construction is zero, the
effect is likely more realistic. After all, in the scenario with more e-bikes, 30% of the bikes have a
considerably higher speed and are known to travel longer distances (as commented in Section 2).
Furthermore, these bikes will operate in a highly improved network, which will benefit bikes that
travel at higher speeds.
With respect to the valuation of health effects, this effect turns out to be very important as
demonstrated in the calculation in Table 4. In particular, it can be questioned whether physical activity
is substituted or not and more research is indeed needed with respect to this question. This could be
particularly relevant if new bicycle technology attracts other types of users in the future, e.g. older
people which are less physical active in general. It is also a question to what extent such results can be
generalized to other countries, especially between countries where the bicycle culture is very
different.
When judging all effects combined and the potential limitations, it is concluded that the investigated
bicycle structure, with a high probability, is beneficial for society. It is demonstrated that the level of
the internal rate of return on investment is likely better than conventional projects for cars and public
transport. It is also found that among many different effects, health benefits are the single most
important factor. This suggest that most benefits are national benefits and that investments in bicycle
infrastructure should not only be a matter for local authorities but for the state as well.
14
4 Conclusion
In the paper, we have presented a cost-benefit analysis of a bicycle infrastructure that includes
different bike technologies, which, to our knowledge, is a new approach. The work is based on a
separate bicycle demand model, which has been documented in Hallberg et al. (2020). Demand
effects are combined with a newly updated valuation of health and accident costs for different bicycle
technologies.
As a case study, the paper consider an ambitious development of the bicycle network in the Greater
Copenhagen area. The upgraded network represent an added network of 855 km cycle superhighway,
which is expected to be fully implemented by 2045. The results suggest that the bicycle infrastructure
is strongly beneficial for all scenarios. The most beneficial scenario is a scenario where the e-bike
share is comparable to the share of today and where external costs for health are updated to comply
with the most recent literature. The internal rate of return on investment is 23% in that case.
The least beneficial scenario assumes an increasing number of e-bikes and a valuation of health
benefits that applies the previous valuation practice. The internal rate of return is 6% in that case.
Scenarios with higher shares of e-bikes will on the one hand render smaller health benefits per km and
on the other hand, higher safety costs. The reduction in these external benefits (when compared to
conventional bikes) is not counteracted by a corresponding increase in consumer surplus from travel
time savings. This is partly because the valuation of consumer benefits is minor compared with the
high valuation of the external effect linked to the increased mileage, but also because the model
framework cannot represent travel demand for e-bikes correctly. From a demand perspective, e-bikes
behave largely as ordinary bikes, and despite being faster, they will not travel longer on average.
In summary, the paper establishes a relatively positive case for a bike infrastructure investment plan
for Copenhagen. Due to the substantial and robust social benefits across the various assumptions, it is
worthwhile considering whether bike infrastructure should have a higher degree of government
funding in the future. After all, health benefits are non-local benefits that have a positive effect on
public health expenditure.
References
Annema, J.A., Koopmans, C. and Wee, B.V. (2007) Evaluating Transport Infrastructure Investments:
The Dutch Experience with a Standardized Approach, Transport Reviews, 27:2, 125-150, DOI:
10.1080/01441640600843237
Berntsen, Sveinung, Lena Malnes, Aleksander Langåker and Elling Bere (2017): Physical activity
when riding an e-assisted bicycle, International Journal of Behavioral Nutrition and Physical
Activity, 14:55
Brown, V., Diomedi, B.Z., Moodie, M., Veerman, J.L., Carter, R. (2015). A systematic review of
economic analyses of active transport interventions that include physical activity benefits, Transport
Policy, https://doi.org/10.1016/j.tranpol.2015.10.003.
Börjesson, M., Eliasson, J., 2011. The value of time and external benefits in bicycle appraisal.
Technical Report. Url: www.cts.kth.se.
Christensen, H., Oana, B. (2019). Annual TU Report For Denmark. Url:
https://www.cta.man.dtu.dk/transportvaneundersoegelsen/udgivelser
COWI (2020). Transportøkonomiske enhedspriser for cykling. Report for the Ministry of Transport.
Url: https://www.trm.dk/nyheder/2020/det-er-sundere-at-cykle-end-hidtil-antaget
15
Danish Government, 2019. Agreement for a coherent Denmark - Investment Plan 2030 (In Danish).
Technical Report. https://www.regeringen.dk/media/6531/aftale-om-et-sammenhaengende-danmark-
investeringsplan-2030.pdf
Eriksson, J., Forsman, Å., Niska, A., Gustafsson, S., Sörensen, G., 2019. An analysis of cyclists’
speed at combined pedestrian and cycle paths. Traffic Injury Prevention.
https://doi.org/10.1080/15389588.2019.1658083
Foley, L., Dumuid, D., Atkin, A. J., Olds, T., & Ogilvie, D. (2018). Patterns of health behaviour
associated with active travel: A compositional data analysis. The International Journal of Behavioral
Nutrition and Physical Activity, 15, Article 26.
Foley, L., Dumuid, D., Atkin, A. J., Wijndaele, K., Ogilvie, D., Olds, T. (2019). Cross-sectional and
longitudinal associations between active commuting and patterns of movement behaviour during
discretionary time: A compositional data analysis. PLoS One. doi: 10.1371/journal.pone.0216650
Genter, J.A., Donovan S., Petrenas B., 2008. Valuing the health benefits of active transport modes,
NZ Transport Agency Research Report 359. New Zealand Transport Agency, Wellington, NZ.
Geus, D. B., De Smet S, Nijs J, Meeusen R. Determining the intensity and energy expenditure during
commuter cycling. Br J Sports Med. 2007;41(1):8-12.
Goeverden, K.V., Nielsen, T.S., Harder, H., Nes, R.V. (2015). Interventions in bicycle infrastructure,
lessons from Dutch and Danish cases. Transportation Research Procedia 10 (2015) 403412.
Hallberg, M., Rasmussen, T.K., Rich, J. (2020). Modelling the impact of cycle superhighways and
electric bicycles. In submission for Transportation part A.
URL: https://www.dropbox.com/s/mdxujrb1r3cgm24/Bicycle_paper_revision1_TRA.pdf?dl=0
Haskell, W. L., Lee, I. M., Pate, R. R., et al. 2007. Physical Activity and Public Health. Medicine and
Science in Sports and Exercise, 39(8), 1423-34.
Houkes, R. Klingen, J. (2019). To what extent do cycling highways increase bicycle commuting
speed? Paper presented at CRB 2019, Amsterdam.
HVU (2019). E-bike accidents (In Danish). Tema rapport nr. 19. Accident Investigation Board
Denmark.
URL:
www.hvu.dk/SiteCollectionDocuments/Havarikommission%20Rapport16%20Elcykelulykker.pdf
Hyewon, S., Kim, E. (2019). Meta-analysis of rate of return on road projects, Transportation Letters,
11:4, 190-199, DOI: 10.1080/19427867.2017.1300984
Incentive, 2018. Samfundsøkonomisk analyse af supercykelstierne. Technical Report.
Url: https://supercykelstier.dk/wp-content/uploads/2018/06/Incentive_Samfunds%C3%B8konomisk-
analyse-af-supercykelstier.pdf
Kahlmeier, S., Gotschi., T., Cavill., N., et al. (2017). Health economic assessment tool (HEAT) for
walking and for cycling Methods and user guide on physical activity, air pollution, injuries and carbon
impact assessments. WHO Regional Office for Europe (2017)
URL: https://www.euro.who.int/__data/assets/pdf_file/0010/352963/Heat.pdf%3Fua%3D1
Kelly, P., Kahlmeier, S., Götschi, T., Orsini, N., Richards, J., Roberts, N., Scarborough, P., Foster,
C. (2014). Systematic review and meta-analysis of reduction in all-cause mortality from walking
16
and cycling and shape of dose response relationship. International Journal of Behavioral
Nutrition and Physical Activity volume, 2014, 11:132.
Kidokoro, Y. (2004). Cost-Benefit Analysis for Transport Networks: Theory and Application, Journal
of Transport Economics and Policy, University of Bath, vol. 38(2), pages 275-307.
Krizek, K. J., Barnes, G., Thompson, K., 2009. Analyzing the effect of bicycle facilities on commute
mode share over time. Journal of Urban Planning and Development 135, 6673.
doi:10.1061/(ASCE)0733-9488(2009)135:2(66).
Paulsen, M., Rasmussen, T.K., Nielsen, O.A. (2019). Fast or forced to follow: A speed heterogeneous
approach to congested multi-lane bicycle traffic simulation, Transportation Research Part B:
Methodological. https://doi.org/10.1016/j.trb.2019.07.002.
Rich, J., Hansen, C.O. 2016; The Danish National Passenger Model Model specification and results.
European Journal of Transport and Infrastructure Research. 16(4): 573-599.
Schleinitz, K., Petzoldt, T., Franke-Bartholdt, L., Krems, J.F., & Gehlert, T. (2017). The German
Naturalistic Cycling StudyComparing cycling speed of riders of different e-bikes and conventional
bicycles. Safety Science, 92, 290-297. dx.doi.org/10.1016/j.ssci.2015.07.027
Skov-Petersen, H., Jacobsen, J.B., Vedel, S.E., Nielsen, T.S., Rask, S. (2017). Effects of upgrading to
cycle highways - An analysis of demand induction, use patterns and satisfaction before and after,
Journal of Transport Geography, https://doi.org/10.1016/j.jtrangeo.2017.09.011.
Stinson, M., A., Bhat, C., R. (2003). Commuter Bicyclist Route Choice: Analysis Using a Stated
Preference Survey. Transportation Research Record. 2003;1828(1):107-115. doi:10.3141/1828-13
Supercykelstier, 2019. Supercykelsti cykelregnskab. Technical Report.
URL: https://supercykelstier.dk/wp-content/uploads/2019/11/Cycle-superhighway-bicycle-account-
2019-UK-final-2.pdf
TERESA 5.0 (2019). Model for Socio-economic Appraisal, Ministry of Transport Transport
URL: https://www.cta.man.dtu.dk/modelbibliotek/teresa
Zahabi, A. H. S., Chang, A., Miranda-Moreno, L.F., Patterson, Z., 2016. Exploring the link between
the neighborhood typologies, bicycle infrastructure and commuting cycling over time and the
potential impact on commuter GHG emissions. Transportation Research Part D 47, 89103. URL:
http://dx.doi.org/10.1016/j.trd.2016.05.008
Zimmermann, M., Mai, T., Frejinger, E., 2017. Bike route choice modeling using gps data without
choice sets of paths. Transportation Research Part C: Emerging Technologies 75, 183 196.
https://doi.org/10.1016/j.trc.2016.12.009
17
Term
Explanation
Consumer surplus effects
Represent a monetized expression of travel time savings.
Cycle superhighway
A cycle superhighway represent upgraded cycle infrastructure that facilitate improved
safety, higher speed and fewer intersections. Different variants exist depending on whether
the infrastructure is separated from the car road infrastructure.
HEAT assessment model
An international model developed to provide, among other things, external costs (and
benefits) associated with transport.
Internal health effects
Effects of improved health following the use of bicycle experienced by the users themselves.
Labour supply distortion
Distortionary effects of labour taxation. When government expenditures are to be financed
with increased taxation this implies a reduced labour supply and as a result the welfare cost
of an additional investment is higher than the direct cost.
Labour supply benefit
Labour supply is similarly affected by commuting, business and freight transport costs and
is to be corrected accordingly.
OSM
Open Street Map (OSM) is a detailed open-source digital map used to represent the bicycle
network.
MET
Is the Metabolic Equivalent of Tasks and represent a measure of physical activity, where 1
represent the state of being at rest.
Internal rate of return on
investment
The annual rate of return on investment expressed in fixed prices for a given base year.
Residual value
Represent the value of the invested infrastructure after the project period.
rule-of-a-half
Linear approximation of demand function to calculate changes in consumer surplus that is
commonly applied in cost-benefit assessment studies.
Speed-pedelec
Type of electric bicycle with a maximum travel speed up to 45 km/h. The market share of
these bicycles are low and the maximum assumed speed on the network is lower than the
maximum speed and depends on the infrastructure.
TERESA 5.0
A comprehensive Danish cost-benefit assessment model that applies to a wide range of
CBA assessments for transport projects. The model provides a system for calculating the
net-present-value for a project based on a range of inputs.
Value-of-time
Represent the monetary value of travel time savings.
VSL
Monetized value of a statistical life (VSL)
Table 6: Explanation of key terms in the context of the study.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Objective: In Sweden, cyclists, pedestrians, and moped riders share the space on combined pedestrian and cycle paths, and their speeds may differ greatly. Both actual speed and speed differences can potentially influence the number of accidents on the shared paths. As a starting point, this article studies the speed component and how cyclists’ speed varies at pedestrian and cycle paths depending on the day, week, and year; road user composition; and road design. Methods: Three data sources were used: Existing measurements of cycle speed and flow in 3 different Swedish municipalities, Eskilstuna (1 site, January–December 2015), Linköping (6 sites, 4 weeks in September–October 2015), and Stockholm (10 sites, 1–5 days in August–September 2015); complementary measurements of cycle speed and flow in Linköping (4 sites, 1–10 days in August–September 2016) and Stockholm (1 site, only part of 2 days in August 2016) were also conducted within the project, in addition to roadside observations of bicycle types at the 5 new sites. Results: The average speed of cyclists on the paths varied between 12.5 and 26.5 km/h. As expected, the lower average speeds were found in uphill directions, near intersections, and on paths with high pedestrian flows. The higher speeds were found in downhill directions and on commuter routes. In all, 70%–95% of road users observed on pedestrian and cycle paths were cyclists, and 5%–30% were pedestrians. The most common type of bicycle was a comfort bike, followed by a trekking bike. Electric-assisted bicycles and racer bikes occurred at all sites, with proportions of 1%–10% and 1%–15%, respectively. The 2 sites with the highest proportion of electric-assisted bicycles and racer bicycles also had the highest average speeds. The differences in average speed throughout the day, week, and year could only be assessed at one of the sites. Only small differences were found, with the most noticeable being that the average speed was lower in January and February (13.8 km/h) compared to the rest of the year (15.3–16.1 km/h). The average speed was also lower during daytime (14.7 km/h) than during other parts of the day (15.4–15.8 km/h). Conclusions: The relationship between bicycle type and measured speed was not entirely clear, but the results suggest that paths with higher proportions of electric and racer bicycles have higher average speeds. There also appears to be a connection between average speed and the width of the distribution; that is, the higher the average speed, the wider the speed distribution. More research is needed on how speed levels and speed variance affect accident risk.
Article
Full-text available
Background: Active living approaches seek to promote physical activity and reduce sedentary time across different domains, including through active travel. However, there is little information on how movement behaviours in different domains relate to each other. We used compositional data analysis to explore associations between active commuting and patterns of movement behaviour during discretionary time. Methods and findings: We analysed cross-sectional and longitudinal data from the UK Biobank study. At baseline (2006-2010) and follow up (2009-2013) participants reported their mode of travel to work, dichotomised as active (walking, cycling or public transport) or inactive (car). Participants also reported activities performed during discretionary time, categorised as (i) screen time; (ii) walking for pleasure; and (iii) sport and do-it-yourself (DIY) activities, summed to produce a total. We applied compositional data analysis to test for associations between active commuting and the composition and total amount of discretionary time, using linear regression models adjusted for covariates. Adverse events were not investigated in this observational analysis. The survey response rate was 5.5%. In the cross-sectional analysis (n = 182,406; mean age = 52 years; 51% female), active commuters engaged in relatively less screen time than those who used inactive modes (coefficient -0.12, 95% confidence interval [CI] -0.13 to -0.11), equating to approximately 60 minutes less screen time per week. Similarly, in the longitudinal analysis (n = 4,323; mean age = 51 years; 49% female) there were relative reductions in screen time in those who used active modes at both time points compared with those who used inactive modes at both time points (coefficient -0.15, 95% confidence interval [CI] -0.24 to -0.06), equating to a difference between these commute groups of approximately 30 minutes per week at follow up. However, as exposures and outcomes were measured concurrently, reverse causation is possible. Conclusions: Active commuting was associated with a more favourable pattern of movement behaviour during discretionary time. Active commuters accumulated 30-60 minutes less screen time per week than those using inactive modes. Though modest, this could have a cumulative effect on health over time.
Article
Full-text available
Background: Active travel (walking or cycling for transport) is associated with favourable health outcomes in adults. However, little is known about the concurrent patterns of health behaviour associated with active travel. We used compositional data analysis to explore differences in how people doing some active travel used their time compared to those doing no active travel, incorporating physical activity, sedentary behaviour and sleep. Methods: We analysed cross-sectional data from the 2014/15 United Kingdom Harmonised European Time Use Survey. Participants recorded two diary days of activity, and we randomly selected one day from participants aged 16 years or over. Activities were categorised into six mutually exclusive sets, accounting for the entire 24 h: (1) sleep; (2) leisure moderate to vigorous physical activity (MVPA); (3) leisure sedentary screen time; (4) non-discretionary time (work, study, chores and caring duties); (5) travel and (6) other. This mixture of activities was defined as a time-use composition. A binary variable was created indicating whether participants reported any active travel on their selected diary day. We used compositional multivariate analysis of variance (MANOVA) to test whether mean time-use composition differed between individuals reporting some active travel and those reporting no active travel, adjusted for covariates. We then used adjusted linear regression models and bootstrap confidence intervals to identify which of the six activity sets differed between groups. Results: 6143 participants (mean age 48 years; 53% female) provided a valid diary day. There was a statistically significant difference in time-use composition between those reporting some active travel and those reporting no active travel. Those undertaking active travel reported a relatively greater amount of time in leisure MVPA and travel, and a relatively lower amount of time in leisure sedentary screen time and sleep. Conclusions: Compared to those not undertaking active travel, those who did active travel reported 11 min more in leisure MVPA and 18 min less in screen time per day, and reported lower sleep. From a health perspective, higher MVPA and lower screen time is favourable, but the pattern of sleep is more complex. Overall, active travel was associated with a broadly health-promoting composition of time across multiple behavioural domains, which supports the public health case for active travel.
Article
Full-text available
Objective The objective of the present study is to investigate the effects of improvements made to two large, interconnected bicycle infrastructure in the western suburbs of Copenhagen, Denmark, on bicycle volumes and mode share, and cyclists' behaviour, perceptions, and experiences. Methods Effects are assessed by analysing data from automatic counting stations during 35 months to measure the changes in bicycle volumes on the investigated routes. Furthermore, a questionnaire survey repeated three times – before, and one and two years after opening the improved routes - is used. Findings are supported by a control survey at a nearby facility, which was not influenced by the infrastructure improvements. Results The investments related to the two investigated cases of infrastructure improvements resulted in a significant increase in the volume of bicyclist two years after the improvements. On one of the routes, the “Albertslund Route”, on weekdays during the rush hour in daylight, an increase from 126 to 203 bicyclists/h was recorded, whereas an increase from 24 to 32 bicyclists/h was recorded at “Vestvolden” for the same period. Most of the increase could be attributed to relocation of bicyclists from other routes. Induced cycling trips – trips that were not previously made by bicycle - were estimated to account for only 4–5% of the bicyclists two years after improvements. Bicyclists using the improved route express an increase in satisfaction with the quality of the facilities, which is significantly higher than at the control site. Conclusion Data from the counting stations provides useful information if measured over a long period. This is necessary to correct for factors such as climate effects and temporal variation. Investments in cycle infrastructure in the investigated case led to a higher number of bicyclists who were mainly relocated from other routes. A minor increase in the modal share of cyclists was observed two years after the infrastructure improvements. Furthermore, the investments resulted in a higher degree of satisfaction among active bicyclists. If measured over a longer period, this could lead to a higher modal share due to a potential social advertising effect.
Article
Full-text available
Background The objectives of the present study were to compare time spent cycling, exercise intensity, and time spent in moderate- (MPA) and vigorous intensity physical activity (VPA) when cycling on an E-bike and a conventional bicycle on two “cycling-to-work” routes with differences in topography, defined as a hilly and a flat route. Methods Eight adults (23–54 years, two women) cycled outdoors on a conventional bicycle and an E-bike, on a flat (8.2 km) and a hilly (7.1 km) route, resulting in 32 journeys. Duration, elevation, and oxygen consumption were recorded using a portable oxygen analyser with GPS. A maximal cardiorespiratory fitness test was performed on a cycle ergometer. Resting metabolic rate was obtained by indirect calorimetry with a canopy hood. Results The participants spent less time (median (IQR)) cycling on the E-bike compared with the conventional bicycle, on both the hilly (18.8 (4.9) vs. 26.3 (6.4) minutes) and the flat (20.0 (2.9) vs. 23.8 (1.8) minutes) routes. Lower exercise intensity was observed with the E-bike compared with the conventional bicycle, both on the hilly (50 (18) vs. 60 (22) % of maximal oxygen uptake) and the flat (52 (19) vs. 55 (12) % of maximal oxygen uptake) routes. In both cycling modes, most time was spent in MVPA (92–99%). However, fewer minutes were spent in MVPA with the E-bike than the conventional bicycle, for both the hilly (26% lower) and the flat (17% lower) routes. Cycling on the E-bike also resulted in 35 and 15% fewer minutes in vigorous intensity, respectively on the hilly and flat routes. Conclusion Cycling on the E-bike resulted in lower trip duration and exercise intensity, compared with the conventional bicycle. However, most of the time was spent in MVPA. This suggests that changing the commuting mode from car to E-bike will significantly increase levels of physical activity while commuting.
Article
Full-text available
The paper describes the structure of the new Danish National Passenger model and provides on this basis a general discussion of large-scale model design, cost-damping and model validation. The paper aims at providing three main contributions to the existing literature. Firstly, at the general level, the paper provides a description of a large-scale forecast model with a discussion of the linkage between population synthesis, demand and assignment. Secondly, the paper gives specific attention to model specification and in particular choice of functional form and cost-damping. Specifically we suggest a family of logarithmic spline functions and illustrate how it is applied in the model. Thirdly and finally, we evaluate model sensitivity and performance by evaluating the distance distribution and elasticities. In the paper we present results where the spline-function is compared with more traditional function types and it is indicated that the spline-function provides a better description of the data. Results are also provided in the form of a back-casting exercise where the model is tested in a back-casting scenario to 2002.
Article
Copenhagen is world-known for its large proportion of cyclists, forming a diverse group with a large variation of equipment and physical abilities. This leads to a considerable speed heterogeneity which needs to be taken into account when modelling the traffic on dedicated bicycle paths. Nevertheless, existing studies on bicycle traffic simulation have either neglected such speed heterogeneity altogether or modelled it by dividing cyclists into a few discrete classes ignoring the entirety of the speed distribution. This paper proposes an efficient bicycle traffic simulation model with continuously speed heterogeneous cyclists and corresponding congestion effects. Based on individual-specific desired speeds and headway distance preferences, the model shows realistic speed-flow relationships validated with on-site observations while being capable of delaying rapid cyclists more often than slower ones in moderate traffic flows. The scalability of the model allows it to be large-scale applicable for network loading purposes, and thus suitable for evaluating impacts of cycling related infrastructure investments.
Article
This study analyzes the impact of economic and spatial factors on the economic return on Korean road projects, using meta-regression models and Monte Carlo simulation. This paper finds that the internal rate of return (IRR) is negatively affected by the distance and land acquisition cost (LAC) in urban areas while it is positively related to the population density, ratio of manufacturing workforce to total workforce, and LAC in rural areas. In addition, net benefits can be generated if a road length is less than 5 km. Finally, the IRR tends to be highly sensitive to the population density.
Article
Concerned by the nuisances of motorized travel on urban life, policy makers are faced with the challenge of making cycling a more attractive alternative for everyday transportation. Route choice models can help achieve this objective by gaining insights into the trade-offs cyclists make when choosing their routes and by allowing the effect of infrastructure improvements to be analyzed. We estimate a link-based bike route choice model from a sample of GPS observations in the city of Eugene on a network comprising over 40,000 links. The so-called recursive logit (RL) model (Fosgerau et al., 2013) does not require to sample any choice set of paths. We show the advantages of this approach in the context of prediction by focusing on two applications of the model: link flows and accessibility measures. Compared to the path-based approach which requires to generate choice sets, the RL model proves to make significant gains in computational time and to avoid paradoxical accessibility measure results discussed in previous works, e.g. Nassir et al. (2014).