Architecture for Modular Microsimulation of Real Estate Markets and Transportation

TL;DR

This paper introduces a modular microsimulation architecture integrating land use, travel demand, and traffic models, utilizing open-source platforms and GPU acceleration for comprehensive urban planning analysis.

Contribution

It presents a novel modular framework combining UrbanSim and ActivitySim with GPU-accelerated traffic microsimulation, enabling detailed and scalable urban modeling.

Findings

Successful integration of land use, travel demand, and traffic models

GPU-accelerated traffic microsimulation for full-model analysis

Open-source platform facilitating regional planning studies

Abstract

Integrating land use, travel demand, and traffic models represents a gold standard for regional planning, but is rarely achieved in a meaningful way, especially at the scale of disaggregate data. In this paper, we present a new architecture for modular microsimulation of urban land use, travel demand, and traffic assignment. UrbanSim is an open-source microsimulation platform used by metropolitan planning organizations worldwide for modeling the growth and development of cities over long (~30 year) time horizons. ActivitySim is an agent-based modeling platform that produces synthetic origin-destination travel demand data, developed from the UrbanSim model and software framework. For traffic assignment, we have integrated two approaches. The first is a static user equilibrium approach that is used as a benchmark. The second is a traffic microsimulation approach that we have highly parallelized to run on a GPU in order to enable full-model microsimulation of agents through the entire modeling workflow. This paper introduces this research agenda, describes this project's achievements so far in developing this modular platform, and outlines further research.