Aggregation and decomposition methods in traffic assignment: towards consistent and efficient planning models in a multi-scale environment

Abstract

Transport models adopt a simplified version of reality to model the movement of people within a transport system. This simplification limits the accuracy of any model. This research focuses on developing novel techniques that, depending on the application context, try to maximise the level of simplification given the minimum result accuracy that is required. To do so, we explore both aggregation and decomposition methods. Besides maximising simplification, we also investigate the requirements to ensure consistency between models that operate in the same spatial domain. In this, so called, multi-scale setting, it is paramount that differences in results between models can be attributed to a particular set of simplifying assumptions. To date, hardly any efforts have been made to formalise, or assess the conditions that need to be satisfied in order to achieve this much desired consistency. The focus of this work is therefore twofold; (i) exploit the combination of both model and application characteristics to achieve the best possible result with the least amount of computational burden, (ii) develop methodology to construct transport model representations in a multi-scale environment following the identified conditions that guarantee consistency between various model granularities.