A New Traffic Assignment Method for Small and Medium Communities

Abstract

This paper describes a new equilibrium traffic assignment method, compares it with prior methods, and illustrates how it can provide improved results for the models that are applied in small and medium-sized communities. The new origin user equilibrium (OUE) method, based on an algorithm developed by Robert Dial, provides superior convergence and more reliable estimates of link flows than the conventional method provided in most travel demand forecasting software packages. Utilizing a regional model for Victoria, British Columbia, the authiors compare traffic assignment results for test cases at different levels of assignment convergence and using different algorithms for computing user equilibrium. The authors demonstrate that OUE can achieve levels of convergence that are virtually indistinguishable from the true equilibrium solution and do so rather quickly on commonly available computers. The test cases illustrate that low levels of convergence lead to gross errors and spurious effects in traffic assignment, but that these problems disappear with higher levels of convergence. Consequently, the ability of OUE to achieve high levels of convergence quickly leads to much more reasonable and reliable traffic assignment results. Small community planners often rely on select link and select zone analysis to understand who benefits from road improvement projects and to share that information with stakeholders. The authors examine the select link analysis that is derived from different traffic assignment algorithms and convergence levels. The authors show that the results are very sensitive to these factors and that, with a proportionality correction, the OUE method provides more robust and reliable select link analysis.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 4 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.