A Simulation-Based Assessment of Downtown Traffic and Parking Systems
Open Access DepositedCurb access in dense downtown street networks is a recurring source of operational friction because multiple users rely on the same limited space for fundamentally different activities. In Washington, DC’s Chinatown–Penn Quarter area, passenger parking demand, commercial activity, and frequent delivery activity intersect within a constrained street network where parking search and curb conflicts can propagate into measurable traffic delay. This thesis develops an agent-based microsimulation framework to evaluate curb-management interventions under these mixed-user conditions, with an emphasis on how behavioral responses and limited curb capacity jointly shape both parking outcomes and network performance.The model is implemented in the traffic microsimulation software SUMO (Simulation of Urban MObility) with the Python TraCI (Traffic Control Interface) package and represents a real-world street network and curb inventory for the study area, including on-street parking supply and off-street garages used as an alternative for general parkers. Two agent classes are simulated. General parkers choose between curb search, early parking, and garage diversion based on perceived curb availability that evolves over the course of a trip. Delivery vehicles execute multi-stop tours and face a distinct constraint set, including a limited willingness or ability to use off-street options, and a fallback strategy when legal curb parking is unavailable (continued cruising versus illegal parking). To establish a baseline behavioral and traffic regime prior to policy tests, the thesis evaluates a large parameter grid spanning traffic and parking demand levels, curb and garage pricing regimes, and delivery fallback strategies. Because these simulations initialize from an empty network, the analysis also implements an explicit equilibrium workflow, using a time-series warm-up detection approach with additional safeguards to distinguish genuine steady-state behavior from late stabilization and jam-state artifacts. Baseline results show that stable operating conditions occupy a relatively narrow region of the tested parameter space, with gridlock emerging as the dominant failure mode as curb demand rises and when the curb–garage price gap induces additional curb search. Within runs that converge to equilibrium, delivery fallback behavior plays a central role in network performance
allowing illegal parking increases delay and can push marginal scenarios into instability, while cruising shifts impacts toward additional VMT and longer travel times for delivery tours. Two curb-management interventions are then evaluated. The first intervention introduces delivery bays by converting selected on-street spaces on edges identified as frequent locations of delivery parking difficulty. Bay implementations at multiple scales (15, 30, and 45 bays) consistently reduce delivery vehicle illegal-parking time in the scenarios where illegal parking is otherwise prevalent, with benefits that increase as bay supply expands and evidence of diminishing marginal gains at higher demand levels. Network-delay effects are more context-dependent, reflecting the tradeoff between improved delivery outcomes and the loss of general curb capacity in a constrained network. The second intervention introduces a perfect parking information scenario within a defined radius, allowing vehicles to route directly toward available parking based on known occupancy. For delivery vehicles in cruising-fallback scenarios, parking information reduces tour lengths and improves operational efficiency by enabling more direct access to legal curb space. At the same time, the intervention increases route lengths for general curb-parkers and reduces the likelihood of convergence to equilibrium in several settings, consistent with a behavioral shift toward additional curb search travel and away from garage diversion. Combined scenarios suggest that information provision can dominate delay outcomes in some regimes, while targeted bays provide a more direct and reliable reduction in delivery-specific parking difficulty. Overall, the thesis demonstrates that curb interventions in mixed-user downtown systems produce coupled effects that are not well captured by single-user or purely static representations of parking. The proposed modeling framework provides a way to test pricing, space reallocation, and information strategies under heterogeneous behavior while explicitly accounting for network feedbacks and stability. The results support designing delivery accommodations, information provision, and pricing policy as an integrated package, with attention to behavioral response, capacity tradeoffs, and the conditions under which the network remains within a stable operating regime.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.