J. Barceló
Universitat Politècnica de Catalunya

J. Casas, D. García and J. Perarnau
TSS-Transport Simulation Systems

Sedona Modeling Workshop, 2005

INTRODUCTION

Traffic assignment models based on user equilibrium approaches are one of the most widely used tools in transportation planning analysis. The modeling hypotheses, soundly based on the well established concept of equilibrium, lead to nice mathematical models for which there are efficient algorithms that provide solutions in terms of the expected flows on network links, Florian and Hearn, 1995. Modeled flows offer a static average view of the expected use of the road infrastructure under the modeling hypothesis. This information has usually been sufficient for planning decisions, namely at regional and metropolitan levels, as far as the optimization algorithms, based on user equilibrium principle, have a quite efficient computing performance even for very large networks.

However, the evolution of advanced technologies and their application to modern traffic management systems require in most cases a dynamic view complementing the static estimates provided by the assignment tools. The planned infrastructure is probably sufficient for average demand, but time-varying traffic flows, i.e. at peak periods, combined with the influence of road geometry, can produce undesired congestion that cannot be forecasted or analyzed with the static tools, as far as models based on static network equilibrium cannot capture the time dependent traffic phenomena as queue spillbacks or the time evolution of congestion. This is a clear case for changing in the analysis methodology.

The lack of an algorithmic implementation of dynamic equilibrium models computationally efficient for medium or large networks, and the progress made by microscopic simulators in the late 90�s led practitioners to think of the convenience of interfacing a macroscopic approach for transportation modeling based on user equilibrium traffic assignment models, and a microscopic approach, as implemented in microscopic traffic simulation models, as a way of overcoming these limitations, and in recent years commercial software interfacing EMME/2 and AIMSUN, VISUM and VISSIM, or TRIPS and DYNASIM has put to work in practice these ideas.

However, as far as microscopic models are data intensive, the fact that the calibration of models for large networks is not usually an easy task, and the failure to achieve in some cases a consistent dynamic assignment led researchers to consider a different modeling paradigm, the so called mesoscopic models, to fill the gap between the upper level aggregated approach used in the macroscopic models and the lower level fully disaggregated approach of the microscopic models. Mesoscopic models, describing in a simplified way the flow dynamics and implementing heuristically a dynamic equilibrium, can computationally succeed in the analysis of large networks. DYNASMART, Jayakrisham et al. 1994, DYNAMIT, Ben-Akiva 2002, and more recently DTASQ, Florian et al. 2001, and Mahut et al 2004, are good examples of such approach.

Although the criticisms on the ability of microscopic modeling to achieve a proper dynamic assignment have been overcome, proving that a proper implementation of link costs and route choice can achieve the expected results, Barcel� and Casas 2004, and the improvement in hardware and software have speeded up dramatically the computational performance of microscopic simulators, the fact is that each approach plays, and very likely will pay for a long time, a well defined role in transportation analysis as far as each one has unique characteristics to answer appropriately certain questions that the other cannot answer that well. Consequently the discussion is not whether one approach is better or more appropriate than other, or if there is a unique approach that can replace satisfactorily all others, but which is the most appropriate use of each approach and how can then work together in a common framework, if possible, generalizing the former macro-micro interfaces to a combination of macro, meso and micro, in a new methodological framework that could be considered as an extension of the traditional four steps planning model into a six steps planning model, Jayakrisham 2005, in which the four steps of the usual static, strategic long term planning, would find a follow up in term of a dynamic medium term planning, and short term dynamic tactical analysis or, in other words, the corresponding macro, meso and micro levels.

A proper interfacing of two modeling approaches, macro and micro, based on so different principles raises a set of methodological questions, Montero et al. 2001. A microscopic route based simulation in which vehicles follow time changing paths from origins to destinations can be used to get a deeper insight of the performance of a planned infrastructure, complementing in that way the estimated of a classical planning exercise based only on traffic assignment. A proper interface between the two models requires that both network representations at macroscopic and microscopic levels respectively are consistent, and that the same Origin-Destination matrix is used in both approaches. A way of satisfying accurately this requirement is to use a model building tool that automatically translates the network representation at disaggregate level, that of the microscopic view, into the aggregate level of the macroscopic view. Assuming that network consistency is appropriately solved there still remains what is likely the most critical consistency problem: how to synchronize changes in the models, that is how a change at macro level is translated into the meso or micro level and reciprocally. Another key methodological questions are whether the time dependent paths used in the microscopic simulation can be interpreted in terms of a dynamic user equilibrium or not; or how these time dependents paths at micro level correspond to time dependent paths at meso level, Wilco Burhout et al. 2005, discusses these consistency problems of making a meso and a micro approach work together. This paper presents an integrated framework to support a transport analysis methodology that combines the three approaches, overcoming these problems. From a modeling point of view there are two main issues to achieve these objectives:

• Computer modeling issues: an efficient integration of transport analysis tools requires an specific computer modeling approach to overcome the inefficiencies and limitations of transport models based on different network representations communicating by means of file exchanges
• Traffic modeling issues concerning the modeling paradigms to be used at each level and their compatibility.

Download (2 Mb)