BTS produces several datasets that are central to freight planning and modeling in the U.S. The datasets include the Freight Analysis Framework (FAF), Commodity Flow Survey (CFS), Vehicle Inventory and Use Survey (VIUS), the National Transportation Atlas Database (NTAD), and others. Major updates to these programs include: (1) development of a new commodity flow disaggregation approach, with application to FAF and release of county-level FAF estimates; (2) ongoing development of a new FAF forecast process that aims to incorporate mode choice and local socioeconomic data from across the U.S.; (3) increase in the 2022 CFS sample size and a new "Customer Pickup" mode, with releases of CFS tables, public-use file (shipment microdata), and subarea estimates throughout 2025; (4) reinstatement of the VIUS in 2021, with planning underway for the next survey; (5) continuous improvement of and addition to NTAD geospatial freight (and passenger) products; (6) other BTS freight-related programs may be discussed, such as the Transborder data program or the Vulnerability and Resiliency program. This presentation will serve to educate transportation planners and modelers about these essential freight data sources and their ongoing updates, focusing mainly on the FAF. It will also serve as a listening session, allowing BTS to hear about use cases for the data and providing data users an opportunity to shape these data programs in future years.

Abstract Background
Trucks remain the dominant mode of freight transportation in the U.S. and forecasts indicate that it will continue to grow, making the accurate estimation of truck volume and types across major corridors and regions increasingly critical for infrastructure planning, air quality monitoring, and regulatory compliance. Agencies rely on regional travel demand models to estimate current and future truck movements, which are calibrated and validated using external data sources such as truck classification counts. However, these data sources often employ different vehicle classification systems, making direct comparisons challenging.

For example, weight-based classifications, such as the Gross Vehicle Weight Rating (GVWR) system, used in EPA emission calculation models, does not map one to one on axle-based classification systems like the FHWA-13 scheme used in traffic operation and roadway maintenance. Vehicle classification schemes provided in national datasets such as the Vehicle Inventory and Use Survey (VIUS) help with creating a cross walk between various classification schemes, but the process is complicated and may introduce discrepancies. These inconsistencies hinder the ability of transportation agencies to harmonize datasets for model validation, policy analysis, and performance monitoring.

Description of Application
This study reviews common truck classification schemes used in the U.S. and develops a crosswalk methodology to facilitate data integration across different systems. Specifically, it:

·       Compares Vehicle Classification Systems: Reviews and contrasts the GVWR-based system, FHWA-13 classification, and vehicle class definitions used in national datasets such as VIUS.

·       Develops Cross-Classification Approaches: Proposes a method for converting between different classification schemes, enabling agencies to integrate disparate datasets for freight/truck modeling applications.

·       Enhances Model Validation and Input Preparation: Improves the process of using traffic count data and sample truck GPS data for calibrating/validating freight/truck models by aligning classifications across data sources.

·       Expands on Prior Work: Builds on prior analysis conducted for the Southern California Association of Governments (SCAG) and extends the methodology for broader statewide and national applications.

A part of this study is the incorporation of the newly released 2023 National VIUS dataset. By including this data source, the analysis will provide fresh insights into truck fleet characteristics and operations, further enhancing the cross-classification data and methodology.

Statement on Why Application is Noteworthy
This work is notable for its contribution to consistent truck classification schemes, which is essential for ensuring consistency in freight/truck modeling, regulatory compliance, and policy development. By creating a systematic approach to integrating weight-based and axle-based classification systems, this study provides a valuable resource for transportation agencies at the regional, state, and national levels. The inclusion of the 2023 National VIUS dataset further strengthens the applicability of the findings, ensuring that agencies have access to the most relevant and up-to-date information for truck operational characteristics.

Project Status
Major milestones, including the development of crosswalk tables, and addition of the 2023 National VIUS will be completed by Fall 2025.

Authors: David Ory, Virginie Amerlynck, Rebekah Straub, John Gliebe,  Greg Giaimo, Zhujoun Jiang

The Ohio Department of Transportation (ODOT) maintains and is working to improve representations of both commercial and personal travel in both the 3C (regional model used for large MPOs) and statewide (OSWM) modeling systems. Many types of travel fall neatly into one of these two categories. For example, a person commuting from home to work in their personally owned automobile falls cleanly into the personal travel category. A person delivering potato chips to grocery stores and other food outlets in a company-owned vehicle falls cleanly into the commercial vehicle category. In these cases, we can develop algorithms and assumptions that align with the type of travel in which these categories of persons engage.

A minority of travel is more complicated, as it has both personal and commercial components. For example, a real estate agent may travel to an office where they spend most of the day but take a client out in the middle of their workday to show them two or three or four properties. In some ways, this is like an office worker traveling to another office for a business meeting. In other ways, this is like someone traveling from place to place delivering goods or providing a service (e.g., housecleaning). It is not obvious whether the types of behavioral representations included in the personal or commercial travel models best represent this type of travel. Nor is it obvious how to operationalize a definition that categorizes travel as either personal or commercial to ensure that each model has a clearly defined scope, such that the same type of travel is not represented in both models.

To address this challenge, the team began by creating a framework of travel with the following categories: transport a person, deliver goods to a business, deliver goods to a residence, and provide a service at a business or home. Using this framework as a guide, we looked for information about each type of travel in the Ohio Household Travel Survey data and, informed by this analysis, crafted model designs. The survey investigation led to numerous recommendations for improving household travel surveys, which can be shared at MoMo. The resulting model design included recommendations to modify both the 3C system and the commercial vehicle system (known as DCOM). The design strives to allow each model to do what it does best. For example, the demand for package delivery is done in 3C, where individual households are described and simulated, and the logistics of delivering the resulting packages is handled in DCOM. Additional details on the design will be presented and discussed at MoMo.

Abstract Background
California’s freight sector is undergoing significant changes due to regulatory requirements aimed at reducing emissions and improving the sustainability of goods’ movement at a very fast pace. The California Statewide Freight Forecasting and Travel Demand Model (CSF2TDM) plays a key role in evaluating the impacts of various scenarios by forecasting commercial vehicle travel flows and analyzing the interaction between truck movements, and changes in transportation infrastructure. The CSF2TDM integrates socioeconomic factors, land-use changes, and multimodal infrastructure investments to evaluate the effect of various policies on VMT and on-the-road emission. Given the rise of electric vehicle (EV) adoption, incorporating electric trucks into this model is essential for evaluating future transportation scenarios, energy use, and emissions impacts. There is not adequate historical data regarding EV trucks operation, therefore we need to develop a tool to easily evaluate various scenarios.

Description of Application
This presentation outlines the methodology for implementing electric trucks into the freight modeling component of CSF2TDM, focusing on both non-drayage and drayage truck trips.

Corridor Selection:
We identified six key freight corridors targeted for truck electrification infrastructure by 2035. Zones in proximity of these corridors are selected to capture truck trips that can access charging stations without significant detours.

Adoption Rate Application (Non-Drayage Trucks):
Using EMFAC (California’s on the road emissions model) data, we apply electric truck trip adoption rates based on truck type (light, medium, and heavy). These rates are applied to total truck trips in the model to estimate scenario electric truck trips.

Regional Variations:
The methodology accounts for potential regional differences in electric truck adoption, with data disaggregated by county to reflect varying levels of policy support and incentives for electrification.

Drayage Truck Implementation:
For drayage trucks, adoption rates from the California Energy Commission (CEC) Statewide Electric Truck Study are applied to all port zone trips. Scenario testing is based on variations in adoption rates defined by the CEC's work.

This framework provides a scalable and adaptable approach for integrating electric truck scenarios into the CSF2TDM, enabling policymakers and planners to analyze the effects of electrification on freight movements.

Statement on Why Application is Noteworthy
This application is important as it enhances the CSF2TDM’s capability to evaluate electric truck adoption scenarios across California's freight network. By incorporating both corridor-based and regional adoption variations, the model offers a simple and yet inclusive analysis of how electrification can influence transportation efficiency, emissions, and energy consumption. The streamlined approach for non-drayage and drayage truck scenarios serves as a practical and replicable solution for integrating emerging vehicle technologies into large-scale transportation models.

Project Status
Major milestones for this project will be completed by Fall 2025. These include finalizing data integration, scenario testing, and refining model outputs for key electrification corridors and regions. Results will also be shared in the presentation.

Travel demand modeling is a key aspect of planning for the transportation needs of the future. While regional models of passenger travel have long included sophisticated methods, commercial vehicle movement has generally been addressed in more simplistic ways. As models are driven by data, this is largely the result of the availability of information. However, in recent years new robust data sources have become available which have dramatically enhanced our understanding of commercial vehicle behavior.

Commodity flow data such as the freight analysis framework (FAF) has been used in models by converting commodity tonnage into trucks with assumptions for empty trucks. This captures generalized movements at a large scale, in and out of the region. With the advent of ubiquitous location tracking on vehicles – i.e. electronic logging devices (ELD) - data are available in much higher quantities and at extremely granular levels. These data can be harnessed to build models of truck travel patterns derived from recent, local data, which will more accurately represent contemporary regional truck demand patterns and short-distance movements.

For the New York Metropolitan Transportation Authority (NYMTC), we harnessed LOCUS Truck expanded GPS-based origin-destination data to estimate a model of light, medium, and heavy commercial vehicle movements within the region. It used a traditional 3 step approach (trip generation, trip distribution, and assignment) to create trips between transportation analysis zones (TAZs) and load them onto the highway network. A regression of trips by origin over the employment and local characteristics determines the number of trips. A destination choice model is used to better reflect the complex ways in which local characteristics and travel time affect trip destinations.

While many past truck models have depended on limited survey data from a few metropolitan areas around the country to model localized movements, recently available data has allowed us to derive models from the actual travel they will be synthesizing. Assumptions regarding average time or distance traveled have been replaced with robust trip length distributions. The generation of trips across the region is much better understood as well as its change over time. FAF and other data sources still provide important information for analyzing commercial vehicle movement, Big Data has made vast improvements possible.

Since the rise of e-commerce in the late 90s, there has been a steady shift from traditional shopping to online shopping. During the first year of COVID-19 pandemic the pace of growth in online shopping dramatically accelerated. Understanding the factors affecting shopping behavior and implications of online shopping trends for personal activity travel behavior, travel demand modeling, logistics management and physical distribution of goods is the first step toward better planning and tackling of associated externalities in urban areas. This paper explores the determinants of frequency of online shopping among households in the Greater Phoenix Metropolitan Area where Maricopa Association of Governments (MAG) conducts regional level planning. A decision tree model, a supervised learning algorithm, is developed to predict the frequency of online shopping for the region’s population. Aggregated estimates of weekly online transaction data at the traffic analysis zone (TAZ) level is used for the analysis of online shopping trends and model estimation. The results show a diverse range of factors including consumers’ socio-economic characteristics affect frequency of online shopping. The result of this study will provide the basis for forecasting last mile delivery truck trips. The results of the paper contribute to freight planning studies towards more sustainable goods distribution including congestion mitigation policies for urban freight, emission consideration, and curbside management.

Background

Over the past decade, the explosive growth of e-commerce has produced a greater presence of commercial pickup and delivery activities in more places. The demand for home deliveries of parcels and food has increased commercial traffic during all times of day in residential neighborhoods, while suppliers have built new distribution centers throughout metropolitan regions. This has effectively shifted supplier zones closer to consumer zones and enabled the use of different vehicle types. In addition, transportation network company (TNC) vehicle operators, who provide pickup and delivery services for multiple businesses in the same day, have become prominent. These trends pose challenges to commercial vehicle analytics and forecasting--integration of household demand, changing expectations for the spatial distribution of commercial vehicle traffic, and non-establishment based carriers for local pickups and deliveries.

Description

In this presentation, we will discuss how the San Diego Association of Governments (SANDAG) updated their modeling tool to better reflect the realities of the e-commerce era, all within the larger context of needing forecasts for long-range planning and emissions analysis which, in California, requires careful accounting of VMT/VHT by vehicle size. Laying the groundwork for model development, SANDAG fielded a new household travel survey and a new establishment survey in 2022, which enabled them to update both their resident activity-based model (ABM) and their tour-based commercial vehicle model (CVM) in parallel. The GPS-assisted establishment survey also included a separate survey of TNC drivers.

The CVM was redesigned to better meet the needs of the agency. One key component was integrating household demand for package and food deliveries and for commercial service visits, which was made possible by using the household survey results and by designing the CVM to read in synthetic households from the resident model as forecasting inputs.

The simulation of commercial vehicle movements featured two-level destination type and zone choice models. The upper level choice of types--residential, warehouse, ports, and other commercial land use--reflect customer types and route and stop purposes while providing levers for calibration and validation. The TNC survey was critical in developing a separate market segment for TNC pickup and delivery routes. The simulation also allowed for time of day and elapsed time to influence stop and location choices dynamically.

As a further step towards integrating residential and commercial demand, SANDAG developed the CVM within the ActivitySim open-source ABM modeling framework. Development of the CVM in ActivitySim required some modifications to the core software architecture and entirely new model specifications; however, the software engineering team was able to produce a fully operational model which reflected the desired commercial vehicle behavior and ensured compatibility with other SANDAG modeling tools.

Why Application is Noteworthy

This work is noteworthy because it represents a holistic approach by an MPO to update its commercial vehicle modeling tools from the standpoint of identifying the need to address changing analytical requirements, data development, design/specification, and software implementation. Direct integration of household demand, TNCs, and adapting ActivitySim are all novel features for a commercial vehicle model.

Project is Complete