SANDAG Truck Model

PB Americas, Inc.
San Diego Association of Governments (SANDAG)
Goals and Objectives: 

To develop an urban freight analysis capability to support congestion and air quality assessments, using existing data resources.


Multi-class, mixed passenger and freight traffic assignment, treatment of special freight generators, including special attention to US/Mexico border POE data.

Overview and Description: 

The SANDAG Truck Model is an aggregate flow model developed to make use of existing data sources. It follows the traditional and sequential trip generation-distribution-traffic assignment modeling framework, and offers a well-documented example of the data and modeling issues associated with doing so at a detailed level of spatial resolution and on a limited data collection budget.   

•    Internal and External Truck Trip-Based Models. The truck model includes both trip –based generation and subsequent distribution models. Two different models are fitted, based on two different levels of geographic scale: one for local or internal to San Diego County truck trips, and one for longer distance, external to the region truck trips across the entire United States (see Figure 1). The latter model disaggregates FAF flows to US between all US counties that begin, end, or pass truck trips through San Diego County (the latter including truck trips that go into or originate in Mexico). Three truck types are modeled (Light-Heavy, Medium-Heavy and Heavy-Heavy Duty Trucks) based on Emfac2007 (emissions modeling) categories.
•    Trip Generation/Attraction: The trip-based internal truck generation and attraction model is based on employment (aggregated to 7 categories) and household data, supplemented by data on some large special freight genertors (military facility international airport, cruise-ship terminal), with trip rates taken from previous SCAG study.
•    Trip Distribution: The origin-to-destination distribution of internal rtuck trips uses a spatial interaction (entropy maximizing) model with balanced origin and destination totals. The external model disaggregtes the FAF2 141-x 141 commodity flows matrix to an inter-county matrix, and uses commodity-to-truck payload and empty truck factors to derive estimates of annual average weekday truck trips by the same three truck sizes classes used in the internal truck model. These O-D trips are temporalily assigned to an external highway network in order to catpure the number of trucks entering and leaving San Diego County by 12 cordon stations. A destination constrained spatial interaction model is used to disribute externally originating trips entering the region to internal TAZs, and a production constrained spatial interaction model is used to distribute internally origination trips leaving the region via one of the 12 external cordon stations. The model can handle truck size class specific friction factors.
•     Traffic Assignment: The truck flows from both the internal and external models are then combined, along with passenger vehicle flows, in a multi-class traffic assignment using truck size-specific passenger car equivalents (PCEs) to derive congestion level of service (LOS) impacts on the network. To match up with the passenger traffic assignment, this truck assignment is divided into average weekday morning peak, afternoon  peak and off-peak traffic time of day (TOD) assigments, based on within region WIM station data; and using a separate source of data for border POE flows when breaking up daily truck trips into TOD volumes.  

    Figure 1. Schema of the SANDAG Truck Model


Spatial Detail: 

4,670 internal TAZs plus 12 external stations.

Mode Detail: 


Commodity Detail: 

The focus is on truck trips. Commodity flows are aggregated into 3 truck size classes.


The model can support highway network traffic congestion and air quality assessments.

System Platform: 

The internal truck model is run in TransCAD Version 5, using a GDISK script to facilitate model re-runs. The external truck trips model is written in Java, and treated as an exogenous input to the internal truck trips model. A FORTRAN program prepares the output from this Java-based routine for input to the TransCAD multi-class assignment software.

Data Sources: 

•    FHWA FAF2  O-D-Commodity flow data
•    County employment and number of households data by SANDAG TAZ
•    County and Caltrans truck counts and WIM station data
•    County and Caltrans special generator (e.g. airport tonnage, military base employment) data
•    SCAG (Sothern California Association of Governments) model estimated trip generation rates, and also friction factors by truck size class.

Example Inputs & Outputs: 

•    Inputs: Employment and special freight generator data, WIM, FAF commodity flow data, US and local highway network, truck size class data, truck empty, payload friction factors.  
•    Outputs: O-D local and long distance truck O-D flows and associated network assigned volumes by vehicle class.

User Manual/User Interface: 

An Appendix in the above report (PB Americas, Inc., 2008) describes how to run the external truck model using Java, TransCAD Assignment and Fortran codes.

Website & Contact Information: