Time of Day and Location Modeling on Managed Lanes
Managed lanes (MLs) on the Katy Freeway were designed to provide a faster and more reliable travel option than the adjacent general-purpose lanes (GPLs), with variable toll pricing used to regulate demand and generate revenue. Although existing demand models perform reasonably well at the aggregate level, they often fail to adequately capture ML choice behavior at the individual traveler level. Data from Katy Freeway travelers indicated that travelers were more likely to use MLs when toll rates and ML travel time variability were higher. While the positive association of ML usage with toll rates and travel time variability may appear counterintuitive, it actually reflected typical peak-period conditions characterized by congestion, during which both tolls and travel time variability on MLs were higher. Because toll rates were strongly correlated with peak periods, including both variables in the same utility function raised concerns about multicollinearity. To address this limitation, the present study incorporated a trigonometric functional form within the logit model, along with trip characteristics and experience-based measures of slower ML performance, to explicitly capture 24-h patterns in lane-choice decisions. In addition, start and end location variables were introduced to further examine how lane-choice behavior varies across different segments of the Katy Freeway corridor. The findings demonstrate that incorporating time-of-day effects alongside start and end variables delivered substantial improvements in model performance, particularly for location attributes. Applying a trigonometric functional form not only highlights the importance of time-of-day effects in lane-choice modeling but also addresses potential multicollinearity between toll and peak-period variables. The likelihood of travelers choosing MLs increased with higher toll rates, larger speed differentials, and greater travel time variability on MLs. In contrast, travelers who have previously experienced or observed instances in which MLs performed worse than GPLs were less likely to select MLs on subsequent trips. Although the influence of such negative experiences attenuated over time as memories faded, repeated occurrences reinforced these perceptions and prolonged their impact on future lane-choice behavior. For temporal travel patterns, ML utility exhibited a bimodal daily pattern, with elevated utility during the morning and afternoon peak periods and lower utility during overnight and early morning hours.
Authors
- Mark Whitman Burris (ORCID: https://orcid.org/0000-0003-4216-708X)
- Suphanut Jamonnak (ORCID: https://orcid.org/0000-0003-4707-8529)
- Natchaphon Leungbootnak (ORCID: https://orcid.org/0000-0002-7123-8994)
- Evan Cheung (ORCID: https://orcid.org/0009-0007-3302-6564)
Institutions
- Texas Department of Transportation (US)
- Texas A&M University (US)
Publication Details
- Journal
- Journal of Transportation Engineering Part A Systems
- Published
- 2026-11-20
- DOI
- https://doi.org/10.1061/jtepbs.teeng-10016
- Primary Topic
- Transportation Planning and Optimization
- Type
- article
- Field-Weighted Citation Impact
- 0.00