Development of New Supply Models Using Big Data in Maryland
Corresponding Author: Jonathan Avner, Whitman, Requardt, and Associates
Presented By: Jonathan Avner, Whitman, Requardt & Associates, LLP
Observed speed data has been used for both validation of congested speeds in travel demand models and for the calibration of volume delay functions for traffic assignment for several years. Initially observed speed data was collected by employing floating car methods with stop watches and soon replaced with GPS. Most recently, MPOs and DOTs have begun to rely upon third party sources of observed speed data with the common example being INRIX or the NPMRDS. Maryland State Highway Administration (SHA) maintains its own system to collect and process observed speed data. From this system SHA is able to extract vehicle based speed information on 15 minute intervals on data collected on a cross section of roadways including interstates, expressways and major arterials.
This presentation will discuss the approach taken to calibrate the volume delay function parameters using the observed speed data from SHA. Challenges of the analysis includeddefining the input free flow speed and link level capacity as well as dealing with over-saturated conditions. Various approaches were tested and will be presented including the use of model speeds and capacities as well as maximum observed speeds and max flow to represent capacity. The approaches revealed an inconsistency between the model and observed speeds. This inconsistency resulted in a second goal of the analysis which was to define a new speed and capacity logic that was developed using link level attributes from the Maryland Centerline file and calibrated using the observed speed data.
The ability of the MSTM to more accurately represent the true congested speeds on the network is an important contribution to SHA as they are now able to predict several of the performance measures currently only available from existing speed data.