Census-Based Travel Demand Generation for Transportation Simulations

During the last decades, a lot of progress has been made in understanding the dynamics of traffic flow models. Real world applications of these models require the ability to model traffic in networks, which has been an important research topic lately. We focus on the problem of travel demand generation in transportation networks. Our investigations are based on real world data for the Portland/Oregon area. We present a microscopic approach for iterative activity assignment exemplarily for home-to-work trips. It provides a method to generate real-world macroscopic data - in our case it is the travel time distribution resulting from census data - in a network traffic simulation under simulation feedback. The underlying assignment is based on a simple ansatz to split the probability of choosing a workplace in a particular distance into a term which describes the accessibility of workplaces, and the individuals’ function of travel time acceptance. In combination with the census data, this approach provides the macroscopic acceptance function, which turns out to be an exponentially decaying function plus a ‘repulsive’ behavior for small travel times. Furthermore, these investigations demonstrate that iterative activity assignment on a microscopic level is computationally feasible even for realistically sized transportation systems.