Longitudinal Microsimulation As a Tool to Merge Transport Planning and Traffic Engineering Models "The Mobitopp Model"
S Schnittger, D Zumkeller, University of Karlsruhe, DE
Transport planning and traffic engineering are unlikely bedfellows, having quite different properties and necessary skills. As a consequence, sometimes traffic engineers and transport planners show different personalities, the latter being more conceptual and the former being more operational.
Early applications of traffic flow models as well as four step planning models were built on aggregate data, since computer capacity was limited. Later developments brought microsimulation approaches based on elements such as cars or trains etc. (for the traffic flow models) and individuals (for the planning models). However the resolution of time and space was very high for traffic flow models (time resolution in seconds but only for a few number of detailed modelled links), whereas for transport planning models the resolution was quite low (15 to 60 minutes but for a city or a country). The reverse applies to the coverage since transport planning models require much more detailed spatial coverage than traffic engineering models.
The improvement of our transport system has to be seen as a optimization process covering both the short-term operational aspects as well as long-term infrastructure development. For that reason the unlikely bedfellows clearly have to be merged into one comprehensive model to exploit their sometimes diverging aims for optimization. And since the elementary basic unit of a microscopic transport model (one individual) represents more or less a basic unit of a traffic flow model (one car), the natural common basis for a comprehensive modelling approach is the microlevel. Furthermore, the improvement of our transport system as a whole aims for change of both the infrastructure and the behavioural context of its use. As a consequence, the new modelling concept should be longitudinal, being based on longitudinal behavioural data (such as the German Mobility Panel) and a time scale for the operational aspects. Such a concept provides the option to model decisions simultaneously, which gives room for the inclusion of so-called closed loops. Finally, the new concept has to balance the problem of coverage and resolution since the planning aspects are concerned with decades, whereas the operational aspects (i.e. transport telematics) with seconds or minutes. The same applies to the spatial coverage, which clearly is a major challenge for the presented mobiTopp model.
The database on the demand side is constituted by longitudinal mobility data being composed of the German Mobility Panel (1994 ongoing for the daily mobility) and a special extension for interurban travel (so called INVERMO, 2001 to 2003). This integrated dataset has a sample size of about 15000 person-weeks, which is continuously increasing. For applications in specific planning cases (selected cities, states etc.) the social-economic variables of a selected case are used to form a subsample being valid to describe the microscopic demand of that area. As a consequence all households and persons within this specific planning case will be synthesized, and thus provide the platform for the overall microsimulation. The second step matches the planned and performed activity patterns of all individuals with destination and mode choice modules to create full consistency and plausibility of the microscopic data set describing demand. Based on that information even existing information deriving from stated response methods to describe reactions to interventions can be implemented easily and with respect to potentially restrictive parameters of persons and their households.
The simulation itself performs on intervals of one minute covering a so-called ?average day?. Decisions necessary in order to manage changes as reactions to interventions or distortions of the traffic flow through the day are performed by a rules-based approach taking all potential options and restrictions of the affected individuals into account. At this stage links to the network have to be mentioned, which are based on a traffic flow model and provide potential interaction as a reaction to the actual load factors of certain links or routes. These so called closed loops are a necessary prerequisite for applications in the field of transport telematics.
A first application of this model is underway in the region of Ulm, a medium size city between the capital cities Stuttgart and Munich. In this case implementation of new technological mobility services are in progress in order to increase the reliability and regularity of public transport and the flexibility of the supply by additional taxi services. These new services include functions such as the interconnection of a computer aided operational centre (RBL), connection maintenance integrating different public transport companies, dynamic (real time) passenger information over several public transport companies, and traveller information services observing specific trips for spontaneous travellers and for commuters. Additionally, the new flexible taxi services will be offered and can be booked online. The associated demonstrators will be available in a test bed in the city of Ulm as earlier as next year. To estimate the effects applicable to the entire city of Ulm including customization effect results, the application of the new model approach will be used to form a platform for the evaluation during the test phase.
Association for European Transport