Recently, the focus of transportation planning has evolved from accommodating long-term mobility needs to providing near-term and more efficient transportation systems management and operations (TSMO) solutions, the result of limited transportation funding and road capacity build-out. This planning-for-operation concept calls for modeling tools that are sensitive to dynamic interactions between travel behavior and network supply so that the impacts of emerging TSMO strategies (e.g., variable road pricing, ramp metering, etc.) can be accurately estimated. The integration of activity-based travel demand models (ABM) and dynamic traffic assignment (DTA) models offer a perfect solution. However, existing operational integrated ABM-DTA models suffer from several limitations, including excessively long runtime and poor convergence quality, which severely hinders large-scale implementations.

This dissertation proposes to integrate operational ABM and DTA models based on an innovative behavioral foundation: behavior user equilibrium (BUE). Different from the normative behavior theory (i.e., user equilibrium, or UE), BUE is based on a positive theory of travel behavior that avoids impractical assumptions, such as complete information and perfect rationality. BUE describes what travelers actually do in the system and thus emphasizes the role of information acquisition, knowledge updating, and learning in travel decision-making. The BUE-based model saves runtime because DTA models no longer need to run iteratively to reach UE internally and fewer agents undergo behavioral adjustments through iterations. In addition to runtime savings, the BUE principle proposes an alternative way to explain the behavior adjustment process and provides improved behavioral realism. This BUE-based integration framework is applied to the Washington-Baltimore Metropolitan Area as a case study. The integrated model includes InSITE, an ABM developed for the Baltimore Metropolitan Council (BMC), and DTALite, a mesoscopic DTA model. The BUE-based integrated model is then compared with a traditional, sequentially integrated benchmark regarding model convergence and performance. Lastly, to enhance the transferability of the BUE-based integration approach, this dissertation develops a calibration method that estimates parameters associated with the BUE principle using readily available local data so that this integration framework can be easily applied to operational ABM and DTA models elsewhere.