Freight-Demand Modeling to Support Public-Sector Decision Making (2010)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

S U M M A R Y The private sector is largely responsible for development and management of the nation’s freight flow system, but public agencies at all levels face important investment and policy decisions that may affect those flows. Decisionmakers need to understand the large and shifting increases in traffic generated, for example, by ports, inland terminals, and mega- destination centers. In 2004, U.S.DOT launched the Freight Model Improvement Program (FMIP) as a joint effort with the U.S. Department of Agriculture, DOE, and the Army Corps of Engineers, and with support from Oak Ridge National Laboratory. Each of these agencies has developed models for national-level analysis in support of their own unique missions. Some state and regional agencies have undertaken their own modeling efforts. Given the growth in freight and its importance to national, state, and regional economies, public-sector agencies need improved capabilities to analyze freight demand. Study Objective The objective of this project was to • Investigate, identify, and report on high-priority, high-payoff improvements in freight-demand models and other analysis tools; • Conduct research on several of these improvements; and • Develop a guidebook to assist model developers in implementing freight transportation plan- ning, including these improvements. Current Needs and Practices A framework for categorizing existing models was developed. This framework included not only the model categories but the status of how that model is being utilized. That frame- work and the general findings of the literature review are shown in Table S.1. In-person interviews, telephone, and Web surveys were conducted to identify the freight transportation needs and concerns of public decisionmakers. The needs identified by those findings are shown in Figure S.1. The outreach showed general satisfaction with the methods available to support freight planning, but dissatisfaction with the data that are available to support freight planning. Research The review of the model framework and the survey responses suggested a number of research topics that could be improved through additional research. Those topics that were Freight-Demand Modeling to Support Public-Sector Decision Making 1

2selected for additional research as part of this project will be discussed later. The topics iden- tified but not selected include the following: • Freight data to support model specification, calibration, and validation; • Better methods to consider nonfreight trucks; • Better incorporation of labor and equipment productivity in freight models; • Improved methods for nonhighway freight assignment; • Simplified methods for considering the economic impact of freight improvements; and • Better consideration and forecasting of trip by empty and repositioning freight vehicles. The three topics that were selected for additional research were as follows: • Developing trip distribution and other chaining data through the use of GPS data; • Developing temporal and seasonal commodity flow factors; and • Developing mode-choice parameters using public datasets. Model Category Description Model Development Model Implementation Public Sector Applications Time Series Short-, medium-, and long-term forecasts of freight demand and freight activity Behavioral Models how companies perceive and select from the many available freight choices. Includes choice- based and survey-based demand models Commodity- Based and Input-Output Estimate current and forecasted freight traffic generation and distribution by linking industrial activity through input-output models of economic activity Multimodal Network Link-node network representations of freight supply useful for determining travel times, costs, reliability, and overall level of service Microsimulation and Agent-Based Microsimulation models the individual movement of large numbers of units and their attributes, while agent-based modeling defines potential actors in freight transporatation and an allowable set of actions and interactions Supply Chain/ Logistics Supply chains define the life cycle of products from raw materials to the final consumer, including production, inventory, and transportation Network Design Private-sector models for locating factories, distribution centers, warehouses, and other freight generating facilities Routing and Scheduling Private-sector models for locating factories, distribution centers, warehouses, and other freight generating facilities Other and Emerging Topics Hybrid models, real-time decision making Widely used, state of the practice Emerging model, limited use Lacking research or application Table S.1. Comparison of model development and implementation in the literature to public-sector applications.

3The research on trip distribution and other chaining data through the use of GPS data suggested that it is possible to use unobtrusive GPS subscription data to obtain a large num- ber of records containing information about truck trips. The locations and times of GPS readings can be used to determine truck stop locations, the land uses at those locations, the next land use served on a trip, and the travel time and distance to the next stop. The infor- mation was examined in four diverse metropolitan areas (Baltimore, Chicago, Los Angeles, and Phoenix). The data suggested similar patterns for all of the metropolitan areas. The land use at the origin was most often linked with the same land use at the destination. This ranged from a high of 65 percent of the truck trips from origins with residential land uses to desti- nations with residential land uses in Los Angeles to a low of 40 percent of the truck trips from origins with low-density land uses to destinations with low-density land uses in Chicago. Average trip characteristics ranged from a low of 38 min and 9.2 mi in Baltimore to 44 min and 11.3 mi in Los Angeles. The research on temporal and seasonal commodity flow factors suggested that it is possi- ble to assign truck commodity flows to the highway network and to compare the flows with observed truck counts. Although continuous count information was not available from every state, the information available suggests that commodity flow patterns are stable for all commodities throughout the year and that the annual traffic is approximately equivalent to 310 average weekdays (Monday through Friday) and 295 average mid-weekdays (Tues- day through Thursday). The peak hourly traffic is approximately 8 percent of daily traffic and that same percentage of daily traffic occurs during each of the hours beginning at 11 A.M. and ending at 5 P.M. The research on mode-choice parameters using public datasets, which used commodity flow surveys as a revealed-preference survey for developing mode-choice models, suggests that for all commodities, distance is the most important decision variable for mode choice. The only other decision variables that appear to explain mode choice are size of shipment (e.g., tons shipped) and value of the shipment (e.g., dollar per tons), and those variables were only significant as cross products with distance. For many commodities, dividing the distance traveled into short-haul and long-haul markets slightly improved the ability to explain mode Freight Costs and Benefits Performance Measures Estimate of Mode Shifting Estimate of Time-of-Day Shifting Route Diversion Estimates Freight Forecasts Existing Routings Facility Flow Information 3.5 3.3 3.4 2.9 3.2 3.4 3.9 3.5 1 3 2 4 5 Figure S.1. Public-sector freight analysis needs.

4choice. Overall, for many commodities, none of the variables tested at the scale of geography were significant, suggesting that mode-choice decisions are subject to business decisions for which data are not available and are made based on conditions in geographies smaller than the Freight Analysis Framework Version 2 (FAF2) regions that were used. Guidebook Based on the literature review and model framework, the survey of public decisionmakers, and the additional research in support of this project, the study team developed a proposed 10-step process. This process follows standard practices used to support transportation analy- ses for public decisionmakers, and typically is a vehicle or commodity-based process. The process proposed for freight is generally similar to the process used by models to support other transportation decisions. The process includes the following steps: Step 1—What freight policy alternatives need to be evaluated? Step 2—What performance measures support those policy measures? Step 3—What forecasting models can be used to support decisions? Step 4—How much freight? Trip generation: productions and attractions by commodity in tons. Step 5—Where does the freight go? Trip distribution: trip table Os and Ds. Step 6—What mode does freight use? Mode choice: trip table Os and Ds by mode. Step 6a—Direct acquisition of commodity OD tables: alternate ways to get freight OD tables. Step 6b—Economic/land use model: alternate ways to get freight OD table by mode. Step 7—How many freight trucks? Payload and temporal factors: trip table Os and Ds by mode by vehicle. Step 8—What service and other trucks must be considered with freight? Nonfreight vehicle OD tables. Step 9—What facilities do freight vehicles use? Assignment of modal vehicles to networks. Step 9a—What facilities do freight vehicles use? Direct estimation. Step 10—How do freight vehicles perform on the network? Estimation of benefits. Steps 4 through 10 follow the modeling steps shown in Figure S.2. This figure also shows the alternate paths that can be followed. These are indicated in the previous list of steps as alternatives (e.g., Step 6a) to the major steps.

5Findings and Conclusions Although models always can be improved, the freight planners who support public decision- makers expressed general satisfaction with available models but concern about the availabil- ity of freight data. Research concentrated on ways to use existing data to develop data inputs for the models. The research showed that existing and readily available data can be used to develop the inputs required by freight models. Economic Inputs Step 4 Trip Generation Step 5 Trip Distribution Step 6 Mode Choice Step 7 Payload and Temporal Factors Step 8 Service (nonfreight) trucks Step 9 Modal Assignment Step 10 Benefits Analysis Step 6a Acquired Commodity Flow Tables Multimodal By Mode Step 6b Economic Modeling Step 9a Trend Analysis Figure S.2. Model methods.