Travel Demand Model - OKI

OKI/MVRPC Travel Demand Model – Version 6.02.4.2 Multi-stage implementation of synthetic matrix estimationSynthetic matrix estimation procedures attempt to adjust the interchange values in a trip table throughan iterative process of assigning the table to the network, calculating deviations from coded trafficcounts, and using this information to re-factor the trip table. Since the traffic counts coded to thenetwork represent daily link flows that include both trips with both internal and external origins anddestinations, the seed matrix used to initiate the process should also include both internal and externalorigins and destinations.Initial development of the truck model using the SME procedure in TransCAD was based on a seed matrixin which the internal-internal trip tables generated through the Quick Response truck trip generation anddistribution processes were combined with the EE/EI truck trip tables that were formed from the ODOTexternal station survey. A thorough analysis of changes between these initial trip table estimates and theoutput of the SME procedure revealed that some significant changes were being made to the EE/EI flowpatterns. Despite excellent matches to the count data, it was decided that the EE/EI trip flow patternsformed from the expanded ODOT external station survey should be preserved to the extent possible. Toremedy the situation, a multi-stage approach to SME was developed whereby only the internal-internalseed matrix tables would be calibrated and the external station tables left intact. The steps in this multistageSME procedure are as follows:1. Perform a preliminary assignment of the internal-internal seed matrix and save the resultingflows in a new data field.2. Perform a preliminary assignment of the EE/IE truck trip tables and save the resulting flowsin a new data field.3. For each link, calculate the ratio of saved internal-internal truck volume to total saved truckvolumes (internal-internal plus EE/EI).4. For links with daily traffic counts, multiply these ratio times the traffic count totals toapproximate the proportion of count volume that represents internal O-D demand and savethis in a new field as a new calibration count target value.5. Add the saved EE/IE truck trips to the pre-loaded auto volumes for each link, creating a newfield for pre-loaded vehicle volumes.6. Apply the SME procedure to the internal-internal O-D seed matrix, using the new calibrationcount values and the new pre-loaded vehicle volumes.7. Combine the adjusted internal-internal truck trip O-D matrix with the EE/EI truck trip tablesfrom the ODOT external station survey to form the final trip table.Despite the loss of information from using partial count data for calibration, this multi-stage approachresulted in a good fit to the full count data, while producing reasonable zonal trip ends and interchangeflow values.Truck Model - Base-Year Truck Model Development 23