详情

原文传递 Using Collected Data to Improve Dynamic Traffic Assignment Modeling: Volumes 1 and 2.

题名：	Using Collected Data to Improve Dynamic Traffic Assignment Modeling: Volumes 1 and 2.
作者：	Juri, N. R.; Boyles, S. D.; Zhu, T.; Perrine, K.; Chen, A.; Li, Y.
关键词：	Dynamic traffic assignment, User delay cost, Link transmission model (LTM), Network loading, Demand uncertainty, Sensitivity analysis, Queue spillback, Freeway land closure, Link transmission model, Queue length, Urban congestion, Performance metrics, Strategic planning
摘要：	Volume 1: In the field, queue spill-back contributes substantially to urban congestion. Therefore, most dynamic network loading models either explicitly include spillback, or frame a failure to model spill-back as an unfortunate consequence of mathematical or computational tractability that should be relaxed in future work. While models with spillback are undeniably more realistic, they are also less robust to errors in input demand. We show that when there is high uncertainty in input demand, excluding spillback can actually reduce error by reducing sensitivity to demand errors. Our demonstrations include a small network that can be solved analytically, and dynamic user equilibrium on two real-world networks (Austin and San Antonio). We conclude that model realism must be carefully balanced against accuracy of the input parameters, and the sensitivity of the model to any such errors in these inputs. Volume 2: Realistic estimations of queue propagation and traffic delays due to freeway lane closures are critical for planning and managing work zones. The current practical methods include deterministic queuing theory and microsimulation. However, these approaches are either too simple to capture the complex traffic entry/exit patterns or require significant modeling effort. This paper proposes a framework to use kinematic wave theory to estimate queue position and the user’s traffic delay. Link transmission model (LTM) is implemented in the analysis of work zone impacts due to its efficiency and is validated in different scenarios. A comparison of LTM method with microsimulation suggests that LTM method produces comparable queue propagation and dissipation patterns, with runtimes of only one second. LTM method estimations were also consistent with field data collected on IH35 through Austin, TX, suggesting that model results are realistic, and could support decision-making given adequate input data and parameter calibration. The method presented can be used by agencies to support planning and operational decisions, such as the placement of variable message signs, for events that happen frequently and cannot be modeled in extensive detail. It can also facilitate the computation of meaningful performance metrics to communicate with stakeholders for strategic planning, and to conduct cost-benefit analyses, among others.
报告类型：	科技报告