Off-line calibration of a macroscopic Dynamic Traffic Assignment model: Iterative demand-supply parameters estimation

Dynamic Traffic Assignment (DTA) is an increasingly popular and effective tool for analyzing complex transportation networks which integrates both demand and supply variations in time to replicate prevailing traffic conditions and forecast future network states. Application of these models requires meticulous calibration of both demand and supply parameters to generate consistent and reliable results. The aim of this paper is to present a calibration approach for a macroscopic DTA model in an off-line environment. We estimate Time Dependent Origin Destination (TDOD) demand matrices by a bi-level optimization problem and manually adjust supply parameters afterward. The calibration procedure and relative effectiveness of the iterative demand-supply calibration compared with demand-only calibration is investigated via a real network coded in VISUM. Link traffic volumes are used as measures of performance to compare observed and estimated values. The output of this paper is a guideline to calibrate macroscopic DTA models which can be used (i) for deployment of intelligent transportation systems, (ii) as an initial solution in on-line systems to efficiently combine historical and real-time data and provide more accurate anticipatory information, and (iii) appraise subsequent network performance and predict consequences of various traffic management strategies.