Incorporating environmental justice measures into equilibrium-based transportation network design models

Abstract

This research outlines three major challenges of incorporating Environmental Justice (EJ) into metropolitan transportation planning and proposes a new variation of the user equilibrium discrete network design problem (UEDNDP) for achieving EJ amongst population groups. Needed data is compared with what is currently available on spatial distribution of race and income, spatial distribution of trip ends, trip tables, network performance, and cost estimates of improvements. Several conflicting definitions of equity are offered, as well as applications for each within the context of EJ. The importance of choosing a correct unit of analysis is discussed, with particular emphasis on how the geographic unit of analysis is a poor proxy for the group unit – that is theoretically required as the analysis’ purpose is to compare performance measures between groups. Research into the UE-DNDP examines nine potential objective functions focused on maximizing equity of congestion and travel time. Assuming knowledge of the origin-destination travel matrices by population group, numerical analysis is conducted to assess the performance of each proposed formulation. The lower level UE problem is solved using the Frank-Wolfe method, and due to the hard combinatorial nature of EJ-UE-DNDP, a selectorecombinative genetic algorithm is implemented to efficiently search the solution space for feasible network improvement strategies. The results of numerical analysis suggest that both pareto-optimal and utility-based approaches can be successfully applied, and that the most effective formulations minimize the difference between the change in congestion or travel time across population groups due to the selected improvement projects.

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 — 2 citations exist upstream.

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