Promoting active and low-emission mobility is of prime importance towards 'net zero' in transportation. Hence, it is essential to assess and prioritize the effect of measures that can promote or inhibit active mobility before they are implemented. In this manner, measures can be prioritized and implemented according to their effectiveness. Transport demand models are proper tools for that. However, demand models are usually complex and users often need modelling expertise. Apart from that, active mobility modes are currently insufficiently considered in existing models. Cycling, for instance, is partially not considered at all. At the same time, there is a lot of open data that potentially enable precise modelling and thus allow assessing the effects of measures. Nevertheless, the current question is how to effectively leverage open data and information and how to develop a simple, robust, and plausible model which can be used to simulate measures to promote active mobility.
The contribution will present a transport demand model based on open data and freely available information and algorithms. The model is based on the classic sequential four-step model structure (trip generation, distribution, mode choice, assignment). It includes the modes ‘walking’, ‘cycling’, ‘car’ and ‘public transport’ (tram, bus, train). Various open data sources and information were used for model development and validation. For example, OpenStreetMap (OSM) data and freely available census data, as well as freely accessible information (e.g. mobility indicators) from publications of the national (MiD) and urban (SrV) household surveys on traffic behaviour in Germany are used for trip generation. For the development and validation of trip distribution and mode choice, OSM data, free timetable data (GTFS) and MiD data is used (among other). In addition, freely available transport networks and information on route choice models are used for traffic assignment. The used methods and data within the model steps will be explained in more detail in the article. The model was implemented in Python. As the focus is on active mobility, model application is illustrated for measures promoting cycling. Thus, model quality reveals around 87% accuracy in the last step (route choice). Further model validation (e.g. based on counts, backcasting) will be carried out before TRA2024.
The result of the study is a simple, but robust, plausible and accurate transport demand model that can take into account modes of climate-friendly mobility. It was initially developed for the city of Dresden (Germany), but claims to be transferrable to all other cities where corresponding data is available. Established methods of transport demand modelling (e.g. logit models) are used but other methods (e.g. machine learning approaches) could be an option to improve the model in the near future. In addition, the model could be applied to other (German or European) cities to validate its prognostic capabilities. The final aim of the project is to install a web-platform to make the model accessible to everyone, so that e.g. traffic planners without modelling expertise can model traffic demand and estimate the effects of measures towards a climate-friendly mobility system.