Year: 2017
Type: conference-paper
Abstract: The simulation of large-scale scenarios requires high-performance simulations. MATSim, an agent-based transport simulation, is increasingly reaching limits. The traditional approach is to scale the scenarios, i.e. simulating only 10% of a population instead of 100%. This paper suggests MacroSim, a macroscopic mobility simulation module for MATSim, to overcome the current per- formance limits within MATSim. It uses volume-delay functions to estimate travel times for links based on experienced usages of these links. This allows to decouple agents and thus allows a parallelization of the mobility simulation within MATSim per design. A preliminary implementation of MacroSim showed promising results (7 to 50 times faster than the current mobility simulation depending on the scenario size). Given its limitations - most important no back propagation of traffic congestion - MacroSim is suggested as a complementary mobility simulation to the current implementation for cases where scenario size and simulation performance are more important than precise traffic dynamics.
Source: Procedia Computer Science
Authors: Patrick M. Bösch, Francesco Ciari
Institution: ETH Zurich
Language: Spanish
Subfield: Transportation
Field: Social Sciences
Domain: Social Sciences
Open Access status: diamond
OpenAlex ID: W2625726666