Network User Equilibrium with Elastic Demand: Formulation, Qualitative Analysis and Computation

Terry L. Friesz, Ke Han, Amir H. Meimand, Tao YaoarXiv 2012 · DOI · arXiv

TL;DR

This paper introduces a novel mathematical framework for modeling how users choose departure times and routes in transportation networks when demand isn't fixed but adjusts to travel conditions. They formulate this complex problem as a "differential variational inequality" (DVI), which offers a unified way to analyze and potentially solve these dynamic user equilibrium scenarios.

What Problem Does This Paper Address?

Understanding how people make travel decisions in a dynamic transportation network is crucial for efficient urban planning and traffic management. Traditionally, models often assume a fixed amount of travel demand. However, in reality, demand is "elastic" – it changes based on factors like travel time, congestion, and desired arrival times. This means that if travel becomes too inconvenient, some people might choose to travel at different times or not at all.

The challenge is to model this "dynamic user equilibrium" (DUE) where not only are travel times on chosen paths equilibrated for all users, but the total travel demand itself is determined endogenously by these travel conditions. Previous work has tackled DUE with fixed demand using differential variational inequalities, but extending this to elastic demand, where users simultaneously choose departure times and routes, is a significant hurdle. This paper aims to bridge that gap by developing a DVI formulation for this more realistic scenario.

Methods

The core contribution of this paper is the formulation of dynamic user equilibrium with elastic demand as a differential variational inequality (DVI). This approach combines elements of differential equations (to model the dynamic flow of traffic over time) with variational inequalities (to capture the equilibrium conditions where no user can unilaterally improve their situation).

Key aspects of their method include:

  • Differential Variational Inequality (DVI) Formulation: They extend existing DVI frameworks to incorporate elastic demand. This involves defining path delay operators that account for travel time and schedule delay (penalties for arriving too early or too late).
  • Endogenous Demand: Unlike fixed-demand models, the travel demand itself is determined by the equilibrium conditions. The inverse demand function, which relates travel cost to demand, plays a crucial role.
  • Simultaneous Departure Time and Route Choice: The model explicitly considers users making both decisions concurrently.
  • Optimal Control Problem Equivalence: They show that the DVI formulation is equivalent to an optimal control problem, which provides a pathway for developing computational algorithms.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Unified Framework: The DVI formulation provides a powerful and unified mathematical framework to analyze dynamic user equilibrium with elastic demand, encompassing both traffic flow dynamics and user behavior.
  • Endogenous Demand is Key: Recognizing that travel demand is not static but a function of travel conditions is critical for realistic modeling. The paper demonstrates how to integrate this elasticity into the DVI framework.
  • Complexity of Elastic Demand: The authors highlight that extending DUE to elastic demand is not straightforward, particularly due to the infinite-dimensional and finite-dimensional terms involved in the DVI and the dependence of demand on terminal state variables.
  • Algorithmic Potential: By linking the DVI to an optimal control problem, the paper lays the groundwork for developing efficient algorithms to solve these complex problems, which is essential for practical applications.

Key Results

The paper's primary result is the successful formulation of elastic demand dynamic user equilibrium as a differential variational inequality (DVI). They establish the equivalence between this DVI and the definition of dynamic user equilibrium, providing a theoretical foundation for further analysis and computation. The paper outlines the structure of the Hamiltonian and adjoint equations for this DVI, which are crucial for developing solution algorithms.

Limitations and Future Work

The authors explicitly mention that existence results for the formulated DVI are under development and will be presented in a future paper. They also state that computational algorithms and examples are also under development and will be ready for a future submission. Future work will involve analyzing various network flow models (Vickrey, LWR-Lax, link delay, kinematic wave) within this DVI framework and demonstrating the capability of analytically investigating models with vehicle spillback. The paper concludes by stating that the qualitative and numerical analyses will be expanded upon in future work.

Summary updated

Abstract

In this paper we present a differential variational inequality formulation of dynamic network user equilibrium with elastic travel demand. We discuss its qualitative properties and provide algorithms for and examples of its solution.
BibTeX
@misc{friesz2012network,
  author = {Terry L. Friesz and Ke Han and Amir H. Meimand and Tao Yao},
  title = {{Network User Equilibrium with Elastic Demand: Formulation, Qualitative Analysis and Computation}},
  year = {2012},
  eprint = {1211.4614},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1211.4614},
}
Figure