The Mobility Transition Model (MoTMo) is an agent-based model that simulates private mobility demand in Germany with a time horizon up to 2035. It provides projections for exploring a sustainable mobility transition, for example, to investigate potential shifts to new technologies, such as electric vehicles, and new behaviours, such as car sharing. It is an instrument for experimenting with different options – such as possible policy measures or alternative exogenous developments. The resulting different scenarios for the future can then be compared, for example, in terms of the dynamics and spatial distribution of the different means of transport and total greenhouse gas emissions from the private mobility sector. MoTMo is used in the Decision Theatre to support discussions about sustainable mobility.
Available Mobility Choices
In MoTMo, means of transport that persons can use to satisfy their mobility demand are classified into five mobility types: combustion engine cars, electric cars, public transport, car sharing, and non-motorized mobility, that is, walking or biking.
The model provides a temporal evolution for these mobility types, summarized in form of a mobility sector. For some features, the temporal evolution is exogenously given, for example the trend to an increasing number of heavy cars (SUVs) is based on empirical data. In other cases, the temporal evolution is (partly) endogenous, that is, it depends on the number of persons that use a mobility type; the effect is known, for example for production price development, as learning-by-doing.
Agents and Graph Structure
MoTMo has three types of agents, connected through a multilayered network. Persons represent the individuals that chose a mobility type to cover their everyday travels. Persons are connected to other persons that form their social network. Households represent collectives of agents that take coordinated mobility decisions. Every person belongs to exactly one household. Cells build a regular 5 by 5 km grid of locations representing the German map. They account for spatial information, like population density, and developments, like charging infrastructure built. Every household is located in a cell.
MoTMo uses a synthetic population of Germany for persons in households, and the households’ location in cells, i.e. agents statistically match distributions of the German population in terms of age, income, household type, and population density.
MoTMo persons aim to maximize the utility gained from their mobility choice. The utility function takes account of four types of consequences from the use of a mobility type: costs, ecology, innovation, and convenience. Different persons may weight these consequences differently.
However, persons do not know the true consequences of all mobility types; they base their choice on an information exchange with their peers in the social network about utility gained from their current mobility types. The actual decision for a mobility type is then made at the household level, to take into account the budget constraint based on the household’s income.
”Convenience” of a means of transport is difficult to capture as it may depend on many circumstances (access and availability, travel time needed, feeling of safety, etc.). To abstract from these many aspects in a useful way, MoTMo considers convenience as a function of population density to account for the fact that for example a car is more convenient in sparsely populated areas than in big cities and the other way around for public transport. The convenience curves are Gaussian functions, using different parameters for the different mobility types; examples are shown in the figure.
Options for Scenario Selection
To explore possible futures of private mobility demand in Germany, different scenarios can be composed by choosing among options of three kinds: policy measures, investment strategies, and alternative expectations on exogenous developments. In the business-as-usual (BAU) scenario, where no option is activated, current trends are projected into the future. Options are implemented through parameter changes in the model, for example price parameters or those of the convenience curves.
In the Decision Theatre, participants discuss a scenario choice in small groups, as seen in the picture. They can pick up to two options of each kind.
MoTMo simulations of the different scenarios composed from the list of options have to be run on high performance computing infrastructures, such as those at the Zuse Institute Berlin. Due to the amount of detail in the agent-based setting, one can then explore results at various levels of detail, for example by zooming in to different regions or household types. The figures show examples, such as the modal split, that is, the development of the distribution of the different mobility types over time, or the resulting mobility emissions. In particular, the scenarios chosen here show that there may be trade-offs: while the first scenario choice results in fewer combustion engine cars – a goal frequently voiced by DT participants – the second scenario produces lower emissions. Other example visualisations include maps and a so called Sankey plot that indicates the proportions of agents switching from each mobility type to each other one.