Exploring the future of mobility: unveiling the power of agent-based models
Peter Hogeveen defended his PhD thesis at the Department of Mechanical Engineering on January 30th.
The mobility transition plays a large role in the transition towards a sustainable society. However, it also results in increasingly complex interactions with the energy system. Policy makers are in need of tools and methods to assist them in that complex and dynamic landscape. In his PhD research Peter Hogeveen explored how so-called 'agent-based models' can provide that much needed understanding and guidance of the mobility transition. Agent-based models are simulation models where the dynamics result bottom-up from interactions of the system's components (instead of from mathematical functions), much like in real life. The general conclusion is that agent-based models developed with the proposed approach can provide valuable guidance for the mobility transition.
In his thesis, Hogeveen starts by discussing the challenges of modelling the mobility system in transition. The main challenge is that we often don't have a perfect understanding of the current behavior and even if we would have that, the behavior is going to change drastically by definition of the transition. So how can we make accurate models of the future mobility system? Or any system in transition for that matter. The discussion led to the proposal of a modelling approach meant to simplify understanding of the mobility system and quantify outcomes of scenarios, both of which are very welcome to policy makers. The defined approach distinguishes itself by focusing on modelling the current behavior first (based on real-world data and without attempting to model the decision-process that behavior), and then to adjust the model behavior to (prospected) scenarios. The result is that scenario definition, instead of modelling algorithms, is the most crucial part of the research. This opens up the study to the rest of society, as scenarios can be defined by any stakeholder without complex modelling expertise.
Key factors of success
Next, this approach was showcased in three research articles on shared autonomous electric vehicles, electric vehicle charging behavior, and the charging flexibility of electric vehicles. While performing the research, Hogeveen found three key factors of success for the approach: 1) adequate data, 2) modelling mobility behavior without underlying decision-making, and 3) plausible scenario definition. The thesis also presents two additional articles on the future of mobility that use the modelling results of the first three papers to discuss societal impacts.
Lack of explorative modelling studies
During his research Hogeveen discovered that there is a significant lack of explorative modelling studies in the field of mobility. The majority of studies focus on the near-future. However, with our long-term climate targets, the distant future is extremely relevant to explore. Another interesting finding is that the majority of models in scientific literature are significantly less realistic and accurate one might initially believe. He perceives both to be a direct result of the challenges related to modelling systems in transition. With his proposed modelling approach he provides a solution for future researchers.
Valuable guidance for mobility transition
The general conclusion of my thesis is that agent-based models developed with the proposed approach can provide valuable guidance for the mobility transition. It can significantly increase understanding of the modeled system and it can be a powerful tool for exploring scenarios. It does so without modelling complex decision-making mechanisms that are poorly understood. However, it is crucial to provide the correct context of the model results and to convey that the model approach holds explorative, but not predictive value. Scenarios are defined, not predicted. This leads to one of the core advantages of the approach: defined scenarios can be discussed and easily adjusted to explore other quantified narratives, both of which are key to policymaking processes.
Title of PhD thesis: Exploring the future of mobility - On the value of agent-based modeling of the transitioning mobility system. Supervisors: Prof. Maarten Steinbuch and Prof. Geert Verbong.