Methods for Quantifying Urban Freight Infrastructure Capacity and Utilization

Abstract

Two major components of urban freight infrastructure, curb space and off-street commercial vehicle parking, will be analyzed to provide novel methods for estimating their respective capacity in Seattle’s downtown commercial core. This report will also look to develop a framework for estimating parking events from sensor data collected from occupancy detectors place along a curb face using hierarchical clustering. This framework will be tested on existing video data and its performance will be assessed using metrics including but not limited to total occupancy over time, length of parking overstay, and true positive parking event identifications.First, Seattle’s Central Business District (CBD) was studied to compare off-street parking capacity in urban loading bays and loading docks with on-street parking at curb segments designated for commercial vehicle loading and unloading. The Federal Highway Administration’s vehicle classification criteria was used to distinguish parking capacities by vehicle class. Upper and lower bounds of vehicle dimensions were researched for each vehicle class to utilize a parallel parking formula to determine commercial vehicle occupancy for each CVLZ and Loading/Unloading segment in Seattle’s CBD. Utilization scenarios were created for off-street loading bays to calculate the total off-street parking capacity. The two results, for off-street and on-street commercial parking capacities, were compared to see the significance of off-street parking capacity in Seattle’s CBD. In 25 out of the 40 cases, the facilities in off-street operations had equal to or greater capacity than the space dedicated to commercial vehicle deliveries at the curb. In all scenarios tested, off-street parking consisted of at least one in every three potential commercial vehicle loading spaces that exist in Seattle’s CBD. The development and implementation of a sensor data clustering algorithm that turns sensor activity into estimated parking events will also be discussed. This process is done by determining the time and spatial dissimilarity between any two sensor events caused by parking activity at the curb. Optimal configurations of space and time factors demonstrate the potential for application in certain situations, specifically those with smaller study areas with uniform parking activity. The quantification of certain parking metrics from estimated sensor data is also discussed, specifically in overstay and total curb occupancy over time. These metrics are calculated for parking events estimated from both video and sensor data. The former is used to calibrate the latter for several days of video recordings at different block faces in Seattle’s Belltown district. Calibrating the hierarchical clustering algorithm with estimated parking events from video recordings means the algorithm is validated by visual estimations of parking events, which are treated as a baseline for performance in this research. It also serves as an assessment for the quality of the clustering algorithm and can determine whether the clustering algorithm provides useful information on parking activity for a given area.