University of California Transportation Center
Spatial Models of Morning Commute Consistent with Realistic Traffic Behavior
- Author(s): Lago, Alejandro
- et al.
Urban planners are increasingly concerned about the sprawling suburban development in metropolitan areas around the world, which they often blame for growing traffic congestion and excessive highway investment needs. This dissertation seeks to shed light on this issue by studying the relationship between morning commute congestion and urban form.
The causes and consequences of traffic congestion have been extensively studied in the economics and engineering literatures. Unfortunately, most conclusions have been drawn from very idealized models, which either fail to consider adequately the spatial nature of congestion, by neglecting the effects of physical queues and merging interactions, or overlook dynamic aspects, such as commuters’ departure time adaptation during the rush-hour.
To better capture the spatial-dynamic nature of morning commute traffic, this dissertation proposes a new analytical framework that explicitly incorporates spatially distributed commuter origins, realistic traffic behavior and commuter timing decisions. The work combines the departure-time equilibrium theory (as first proposed by Vickrey ) with the spatial model of traffic dynamics of Newell  and the model of merge traffic interactions of Daganzo [1994, 1995a].
Focus is placed on idealized urban configurations, where traffic behavior can be studied analytically and general insights can be gained. We first study the equilibrium problem in a stylized two-origin network. This enables us to understand the fundamental role of merging bottlenecks and queue spillovers when commuters have different origins. The analysis is then extended to model congestion behavior in long freeway corridors and monocentric cities. We develop an exact procedure to solve the dynamic departure-time equilibrium for single-destination freeway tree networks. Solutions are characterized for cases with and without an alternative street network. The results show that the location-based congestion cost is very dependent on the spatial behavior of queues and that congestion can be reduced by altering the freeway access priorities given to different origins. At the same time, urban sprawl is shown to contribute not only to larger travelled distances but also to increased overall delays. Sprawl effects, however, are not as severe as often assumed.
We finally propose some closed-form continuous approximations for the location based congestion cost. These formulae provide an improved and simple representation of the dependence of congestion on the spatial distribution of population that can be easily incorporated to study policy issues. The design of more effective measures to reduce congestion and control urban development is an immediate example.