Demand and behaviour models for shared, electric and autonomous car mobility

Andrea Vanesa Papu Carrone

Research output: Book/ReportPh.D. thesis

193 Downloads (Pure)

Abstract

Transport decarbonisation is an important part of the solution towards meeting the Paris Agreement goals of limiting global warming to 1.5 degrees. However, reducing transport related emissions is a challenge that will require a variety of ambitious technological and policy solutions. New sustainable transport agendas must include actions to reduce unnecessary travel, shift to more sustainable modes, and rapidly scale-up the use of electric vehicles and low-carbon fuels.

Concurrent technological advances towards information and communication technology (ICT), battery technology and automation create opportunities for innovating traditional transport systems. The Three Revolutions in urban transport, namely shared, electric and automated vehicles, will fundamentally affect urban travel. Each of these revolutions addresses different societal needs, but also entails different implementation challenges and costs. Nevertheless, to reach the settled emission objectives, the approach to decarbonising transport must have multiple strands. Transitioning to cleaner vehicles and increasing shared mobility and ridership can provide significant gains; however, additional behavioural changes will likely also be required. Shifts to active transport modes and an overall reduction of travel demand (specially, for private vehicles) should be promoted.

This PhD thesis concerns several aspects within the modelling of behaviour and demand for car mobility. The work is split into three main parts: Part I, Modelling changes in car ownership, presents choice models that extend the current understanding of car ownership behaviour; Part II, Modelling of car sharing, covers two papers that investigate the potential of free-floating car sharing in Denmark; the third and final part of the thesis Part III, Modelling of electric and autonomous car mobility, deals with challenges that arise from adoption of new vehicles technologies such as electrification and automation.

The first part of the thesis, Part I, begins with two papers that present important empirical findings from choice models that are developed to extend the current understanding of car ownership behaviour. Throughout this part of the thesis, it is repeatedly highlighted the importance of understanding both the aggregate dynamics of car ownership behaviour over time, and the individual decision-making that triggers changes in car ownership behaviour, to ensure reliable forecasting of travel demand. The two studies presented within the first part of this thesis, take advantage of cross-sectional travel survey data collected consistently over a period of 12 years. Furthermore, they present two different approaches to reveal changes in car ownership. The first paper proposes a novel approach by estimating a joint discrete choice model on cross-sectional data from multiple points in time, while time-dependent parameters in the utility specification are included to explore the longitudinal effects. The study reveals that young age cohorts are less likely to own cars compared to their earlier generations, while senior age cohorts are more attached to car ownership compared to earlier generations. The second paper of the thesis, develops a methodology that investigates changes in household car ownership based on irregularly spaced panel data. The paper raises awareness regarding the possibility of using recurrent respondents from cross-sectional surveys conducted over long time periods to derive panel data. This is a special type of panel, given the irregular separation between observations; therefore, the study proposes a panel-specific weighted likelihood estimation. The results of a generalised ordered logit model formulated as a first-difference approach, provides empirical evidence concerning the importance of transport infrastructure in nudging individuals behaviour (e.g. improving public transport accessibility decreases the likelihood to acquire a new car), and of certain life course events towards increasing the likelihood of changes in car ownership (e.g. start of work-life and partnership or co-habitation increase the likelihood to acquiring a new car).

The second part of the thesis, Part II, investigates the potential of free-floating car sharing in Copenhagen throughout two studies that explore users preferences, demand drivers and impacts on the traditional transport modes. The first paper within this part of the thesis, reveals users preferences with respect to free-floating car sharing in a mode choice setup relying on a stated preference survey. The collected mode choices are modelled as a mixed logit that accounts for panel correlation among choices of the same individual. The paper first derives the ’transaction costs’ (the added cost that individuals perceive from having to share a car) for access time, parking search time and reservation of a vehicle; and then goes on to calculate the elasticities of the attributes. The latter exercise shows that free-floating car sharing in Copenhagen is a stronger substitute to public transport than to bicycle and private cars. The second paper, studies geographical demand drivers and the performance of an electric free-floating car sharing service in Copenhagen by analysing a large transaction dataset facilitated by a free-floating car sharing service operator. This paper investigates the determinants of demand by observing the idle time of vehicles, i.e. the amount of time free-floating car sharing vehicles remain parked between drop-off and the subsequent pick-up. Two hazard-based duration models are presented in the study. The first one identifies successful and unsuccessful areas and
explores geographical demand drivers; while the second predicts the idle time for the specific vehicles conditional on knowing their drop-off location, time, weather and battery state of charge. The two models could be used by free-floating car sharing operators in practise, to understand the strategic decision of what zones to include within the operation area, and to define the operation relocation strategy based on vehicle idle time predictions, respectively. The ultimate aim of this part of the thesis is to provide a better understanding of the opportunities and challenges concerning free-floating car sharing systems in order to help policy makers and car sharing service operators design systems that increase usability.

The final part, Part III, deals with new vehicle technologies, as electrification and automation. The first paper, explores users willingness-to-pay for electric vehicles and vehicle-to-grid applications and provides a comprehensive understanding of consumers’ preferences needed to accelerate the adoption of electric vehicles. The analysis relies on a mixed logit model that is jointly estimated on five stated choice panel datasets that correspond to five Nordic countries. It provides willingness-to-pay measures for different characteristics of electric vehicles (driving range, acceleration, recharging time, electricity source, vehicleto-grid capability) and compares these across countries. The paper provides a valuable discussion regarding how different background policies affect the willingness-to-pay for electric vehicles and vehicle-to-grid applications. The second paper, develops a microscopic simulation framework that is used to investigate the effects of different levels of market penetration of autonomous vehicles on motorway traffic performance. The study aims at quantifying travel time and vehicle throughput in a congested segment of a motorway in several scenarios. The simulation results indicate that in the idealised situation in which autonomous vehicles operate alone, a substantially improved capacity utilisation is attained. However, when drivers of regular vehicles are mixed with autonomous vehicles, the capacity quickly degrades. Finally, from a strategical perspective, the results underline that dedicated lanes for autonomous vehicles are preferable.

In summary, this PhD has contributed to research within behavioural modelling of car mobility, covering topics related to changes in car ownership behaviour, preferences and impacts of free-floating car sharing and challenges related to adoption of new vehicle technologies such as electric and autonomous vehicles. The thesis provides results relevant for policy and practise as findings suggest the importance of offering attractive mobility solutions to motivate the adoption of shared transport modes and cleaner fuels to reduce the transport related CO2 emissions. Specifically, the results from this thesis suggest that young age cohorts of today are less attached to car ownership than their earlier generations. However, it is also acknowledged that certain life events (such as co-habitation and start of work-life) trigger changes in car ownership. Therefore, it should be in the interest of policy makers to provide and inform about innovative and comprehensive mobility solutions to discourage young individuals of initiating car-oriented mobility. Within new shared mobility solutions, this thesis focuses on the case of free-floating car sharing in the city of Copenhagen and delivers findings regarding users perceived costs of sharing a car and geographical demand drivers. Furthermore, it contributes to understanding the interaction between free-floating car sharing services and the traditional transport alternatives through a stated preference survey and a transactions data analysis. Finally, the thesis investigates the challenges for electric vehicles adoption, finding a strong link between consumers willingness-to-pay and policy incentives as well as significant differences in willingness-to-pay across different countries of the same region. Lastly, it investigates the challenges for autonomous vehicles operation in the early stages of market penetration, showing that significant benefits in terms of travel time and vehicle throughput would not be gained and only implementing lane segregation would allow to take advantage of the improved operational features of autonomous vehicles.
Original languageEnglish
PublisherTechnical University of Denmark
Number of pages158
Publication statusPublished - 2021

Fingerprint

Dive into the research topics of 'Demand and behaviour models for shared, electric and autonomous car mobility'. Together they form a unique fingerprint.

Cite this