Exploring the Design of Urban Bike Sharing Systems intended for Commuters in the Netherlands

More Info
expand_more

Abstract

Ever since the rise of the industrial revolution, one of the greatest challenges for governments relates to the variety of traffic problems that have emerged in urbanized areas (Buchanan, 2015; Headicar, 2015). A vastly growing and promising worldwide spread transport policy is the development of urban bike sharing programmes. These programmes can be found in cities all over the world and have been proven to contribute to: a modal shift towards the use of more sustainable transportation modes, a reduction in externalities of non-sustainable modes, an increase in the accessibility of urban regions and a decrease in bicycle parking and traffic congestion (e.g. DeMaio, 2009; Fishman et al., 2014). These systems provide the public with shared use of a bicycle fleet for a small fee, making cycling a possible mode choice at any location throughout an urban region (Shaheen et al., 2010a). There is currently only one bike sharing system found in the Netherlands: the Public Transport bicycle (PT-bicycle). This system is found at medium to large railway stations where the PT-bicycle can be used for the last leg of the journey (NS, 2015). Although this system is very successful, it does not provide the variety of trip options and flexibility that urban bike sharing systems offer in other countries. Its success does indicate however that there is indeed a demand for bicycles at locations where the privately owned bicycle is not available. This raises the question whether there is a market for destination-based bike sharing systems in the Netherlands which are not limited to train stations, and what such a system should look like. Since commuting and business travel highly contribute to urban traffic congestion and emissions in the Netherlands, it is important to increase the use of sustainable modalities for these travel motives. The objective of this research is therefore to provide recommendations for bike sharing system design in order to introduce the service as an attractive mode option to commuters and business travellers in the Netherlands. This objective corresponds to the main research question: How should an urban bike sharing system be designed in order to attract commuters and business travellers in the Netherlands? The first step in this research is to determine the basic elements of a bike sharing concept aimed towards Dutch commuters and business travellers based on extensive background research and expert interviews. This concept is developed to serve as a basis for testing preferences with regards to bike sharing system design and its use as well as to determine the characteristics of Dutch commuters and business travellers that are interested in using a shared bicycle. Based on this research it was found that several conditions are to be met in order for this bike sharing concept to be a realistic representation of how bike sharing will exist in Dutch cities in the future, so that preferences and user characteristics can be determined. Firstly, the concept should be able to provide a mix of both traditional and electric bicycles so that preferences regarding the type of bicycle can be studied in more detail. Secondly, the bicycles should be able to be picked up at any location thereby creating a hypothetical situation in which all potential users are able to access the system, allowing their preferences to be tested. Thirdly, the system should be flexible with regards to two aspects: the system should allow the bicycles to be able to be locked anywhere and at any time and in addition the drop-off or return location of the bicycles should be flexible. This means that the shared bicycle should be allowed to be returned at any docking station. Lastly, the pricing structure presented in the concept has to allow the potential user to pay for use through a fixed fee per kilometre so that the trip costs reflect the cost structure that is expected to be used in Dutch bike sharing systems. Preferences and the perception with regards to making use of a shared bicycle as well as preferences regarding several system characteristics and the characteristics of potential users are studied using a web survey which consists of a questionnaire as well as a stated choice (SC) experiment. The questionnaire consists of questions that study the characteristics of potential users, the current trip characteristics of the respondents and questions that explore user preferences regarding the use of a shared bicycle, while the SC experiment is used to study preferences regarding several system characteristics. A SC experiment is used here as this method allows us to be able to determine the influence of design attributes upon the choices that are observed, thereby gaining understanding of how different characteristics or attributes are balanced against each other in the bike share mode choice. The SC experiment that is developed in order to study how several system characteristics influence the bike share mode choice consists of a series of choice sets, in which for every choice set three alternatives are presented to the respondent. The first two involve shared bicycle alternatives varying in attribute values, while the third alternative is a no-choice alternative, representing any other transport mode. The attributes that make up the shared bicycle alternatives are the bicycle type (traditional versus electric), access time, egress time and the trip costs. While making their choices, respondents are told to assume that they have to travel a certain distance with the shared bicycle, which varies across the choice sets. This context variable allows for examining how different trip distances influence the preference for certain attribute values and the general preference for a shared bicycle over other transport modes. The web survey is distributed amongst a sample of the population of Dutch commuters and business travellers to gather stated preference data on potential users of a bike sharing system that aims to facilitate commuting and business trips. In order to incorporate commuters and business travellers who vary regarding their preferences towards driving and cycling, the survey was distributed amongst employees from one large and three smaller employers whom all have multiple office locations throughout the country. The 293 respondents were found to be predominantly male, on average aged 47 years old, have different educational backgrounds and professions. A large portion of the respondents uses a (company) car for their commuting trip, while public transportation and the bicycle are chosen by smaller portions of the respondent sample. This sample is therefore believed to be a reasonable representation of the population. From the observed choices made in the SC experiment and a mixed logit (ML) model taking into account panel effects is estimated which includes taste heterogeneity for the bike sharing constant as well as the bicycle type parameter. This ML panel model shows that for this model setup there is a high degree of variation in unobserved preferences for bike sharing over other modalities. However, there is no preference for a shared bicycle compared to other modalities or vice versa when averaged over all respondents. Next to the bike sharing constant, the results show that the trip cost and interaction effects with trip cost are the most important attributes influencing the commuters’ bike share mode choice. The parameter estimates for the random parameters bicycle type and the bike sharing constant are by far the most accurately measured estimates. Other attributes such as the trip distance and education seem secondary attributes in determining the bike share mode choice. In addition there is quite a bit of unobserved variation in the importance of the bicycle type parameter. Furthermore, the interaction effect of trip distance on bicycle type shows that the traditional bicycle is preferred for shorter distances while the electric bicycle is preferred for trip distances of over 4.5 kilometres. The difference in utility between both types of bicycles however is relatively small. With regards to the utility of trip distance itself, it is found that commuters are open towards using a shared bicycle for trips up to 8 kilometres. In addition to determining the influence of design attributes upon the choices that are observed, an SC experiment can be used to predict the choice probabilities of people opting for a shared bicycle based on the estimated bike share mode choice model. Several system design scenarios are tested using this method to determine recommendations for bike sharing system design. The traditional bicycle is found to be preferred for short distances while the electric bicycle is preferred for longer distances. Due to the electric bicycles being much more costly than the traditional bicycles however, it is recommended to only provide electric bicycles in a bike sharing system if the system specifically focuses on facilitating medium to long distance trips and enough funds are available to implement these bicycles while still aiming to limit trip costs. Depending on the target groups and the types of trips the system has to facilitate, one can either opt for a fine-grained network of docking stations, which expands the reach of the system, or for a coarsely distributed network of docking station that minimizes costs. As it is recommended to design a flexible bike sharing system in the Netherlands, meaning that users can return the shared bicycle at any docking station, redistribution of bicycles will be necessary. A method to decrease the cost of redistribution is the implementation of user-based redistribution schemes. As the trip costs have a large effect on the attractiveness of the system, these should be minimized which can be achieved by making choices on the distribution of docking stations and the bicycle type that is provided. In addition, employers should be stimulated to include the use of shared bicycles into the travel allowances that are granted to their employees. Based on the results from the questionnaire and the SC experiment it can be concluded that a significant group of people is interested in using a shared bicycle for their commuting trips, and even more so for business trips if their current trip characteristics allow them to make use of a shared bicycle. From the perspective of the user there are three preferred system design scenarios with regards to commuting trips. The first two scenarios describe a bike sharing system implemented on a citywide scale, with either a coarsely distributed or a fine-grained network of docking stations, which only provides traditional bicycles limits the trip costs. The third option is a system that only focuses on a limited group of commuters and business travellers in a smaller region. Such a system would minimize access and egress time through providing electric bicycles exactly where needed. To implement such a system however, trip costs need to be minimized through providing users with a travel allowance aimed towards using a shared bicycle. It can be concluded that the decision on whether to implement a bike sharing system in Dutch cities is not straightforward. Due to the nature of the Dutch population, it remains unclear whether a shared vehicle system will be found to be attractive by a significant part of the population. In addition, as bicycle usage does not need to be stimulated as much as in foreign countries, and the focus lies more on decreasing traffic congestion and increasing the accessibility of urban regions, it is less likely that these problems can be solved through the implementation of a costly citywide bike sharing system. Systems introduced in smaller regions like industrial areas will however be able to tackle specific problems and thereby achieve the goals that were set with regards to traffic congestion and accessibility. The introduction of bike sharing systems at specific locations is therefore more likely to be effective as well as feasible. The decision on whether citywide bike sharing systems in Dutch cities are to be implemented should therefore not be taken lightly.