Project Number: PSR-25-SP07

Research Project Funding: $ 80,000

Project Start and End Date: 5/15/26 – 5/15/27

Project Description: 

Curb space in dense urban cores is under intense pressure from freight deliveries, service vehicles, and passenger car parking activities. Without a data-driven view of curb regulations and demand, cities face double-parking, spillback congestion, and safety conflicts. This project addresses the gap by creating an open-data-based digital twin that links curb regulations, observed curb activity proxies, and network performance to support actionable curb management decisions.

We are developing a strategic curb digital twin for a portion of downtown Los Angeles (DTLA), built using publicly available data to ensure transparency and replicability. In Phase 1, we integrated multiple open datasets (GIS networks from the LA GeoHub, land use data from DataLA, OpenStreetMap) and developed heterogeneous freight demand models distinguishing commercial and residential delivery behaviors. Phase 2 adds an analytical curb allocation optimization layer with targeted microsimulation validation. While the broader research agenda includes multimodal curb demand modeling (pursued in parallel work), Phase 2 addresses the policy question: given competing freight delivery and passenger parking demands, how should curb space be optimally allocated across blockfaces and time-of-day periods?

Phase 2 develops a hierarchical curb allocation framework with three tasks. 

Task 1: Blockface Demand Inference. Block-level freight demand from Phase 1 will be disaggregated to individual blockfaces using frontage land-use, business activity, and existing curb function as weights. Passenger curb demand will be estimated by fusing parking occupancy data where available with land-use, employment, and traffic-flow proxies. The output is a blockface-by-time-bin demand matrix covering both freight and passenger streams. 

Task 2: Analytical Curb Allocation Model. An analytical curb occupancy model will be developed that relates blockface-level demand, dwell-time distributions, and allocated curb capacity to overflow risk for freight and passenger vehicles. Although demand is estimated at the blockface level, curb allocation is solved jointly across linked blockfaces within a block or corridor, allowing for demand substitution as drivers circulate to nearby faces when the first-choice face is full. Unserved demand is represented as probabilistic overflow outcomes, with double-parking more prevalent for freight vehicles and cruising or diversion more prevalent for passenger vehicles. The model supports an optimization identifying curb allocations that minimize total system cost, defined as a weighted sum of freight double-parking events, passenger cruising time, and unserved demand, subject to total curb supply constraints and regulatory requirements. The baseline model treats loading zones and parking spaces as separate, non- interchangeable pools; time-of-day switching scenarios then explore flexible allocation under alternative weighting of freight versus passenger demand. 

Selected policy scenarios (e.g., current allocation, optimized allocation, and variants such as time-of-day  switching) will be tested in the Phase 1 SUMO environment to validate analytical recommendations against traffic performance indicators and test sensitivity to key assumptions. Additional policy options will be explored through scenario analysis to examine how alternative curb management strategies affect double-parking, cruising, and network performance. Results will be synthesized into policy-oriented curb allocation guidance and documented in the final report.