Real-Time Metering Control and Behavior Responses: An Agent-Based Travel and Traffic Microsimulation Approach
Corresponding Author: Chenfeng Xiong, University of Maryland
Presented By: Chenfeng Xiong, University of Maryland
Abstract
Being one of the effective strategies in Active Traffic Management (ATM) and Integrated Corridor Management (ICM), ramp metering control has received increasing research attention. Most existing studies analyze ramp metering at the corridor level and in the context of traffic conditions and impacts. How travelers may behaviorally respond to the ramp metering is often neglected. This is partially due to lack of behavioral data. More importantly, a comprehensive modeling framework that models travel behavior and traffic dynamics is not readily available for evaluating ramp metering control.
This paper develops an integrated agent-based microsimulation to model travel behavior and traffic dynamics and applies the modeling approach to dynamic ramp metering control. Minute-by-minute ramp metering operation is controlled by efficient ALINEA and ALINEA/Q algorithm. The integrated model employs a dynamic traffic simulation engine, DTALite, to simulate queuing and traffic dynamics. Time-varying ramp metering operation not only encourages dynamic route changes, but can also lead to day-to-day travel behavioral adjustments due to its significant impact on traffic conditions in the affected freeway mainline as well as the local area. The agent-based model captures how an agent learn about the ramp meters, gathering experience and traffic information, updating the knowledge, and eventually adapt her/his behavior.
The proposed model captures a wide spectrum of behavioral responses. In the meantime, bounded rational theory and satisficing behavior are conceptualized such that drivers will not search for and/or switch to their “shortest paths” just for infinitesimally small travel time savings under the influence of ramp metering. Instead, travelers will re-equilibrate by adapting themselves to new situations, such as the introduction of ramp meters. By replacing the conventional User Equilibrium with this behavioral equilibrium process, the computational efficiency of integrated simulation models has been significantly enhanced.
The abovementioned unique features allow the model to predict realistic behavior responses and be applied in large-scale practices. We demonstrate the model in a multi-corridor region between Washington D.C. and Baltimore in Maryland. The application showcases the behaviorally rich en-route diversion and departure time responses to ramp metering control. The large-scale simulation implementation can be used to assess corridor-level and regional traffic impacts.