Transportation agencies differ widely in their analysis resources and capabilities. This article is designed to provide tools and guidance for agencies across the spectrum – from those that have no travel demand models to those that have sophisticated transportation, land use and economic models. In all cases, the methods and tools described here are intended to demonstrate how wider transportation and economic impacts can be calculated. While spreadsheet tools are presented, individual agencies are encouraged to adapt and modify these tools and methods as appropriate for their own uses.
A common feature of the tools and methods discussed in article is that they rely upon forms of transportation system performance and traffic zone information that are commonly available to transportation agencies. No further information is required on the types of businesses that stand to benefit from proposed transportation projects. However, it is noted that the addition of information on the types of business areas served and types of freight being carried can enable sharper distinctions between projects in terms of their relative impacts on productivity. Here we also discuss optional tools that enable a more complete analysis of productivity benefits and impacts.
- Use of Travel Demand Models
- Agencies with a Four-Step Travel Demand Model
- Agencies with No Travel Demand Model
- Agencies with an Activity-Based Model
- Reliability Analysis Tools
- SHRP2 – C11 Reliability Analysis Tool for Sketch Planning
- SHRP2 – L07 Reliability Analysis Tool for Design Treatments
- SHRP2 – L08 FREEVAL-RL
- Accessibility Analysis Tools
Use of Travel Demand Models
The analysis process requires data on travel volumes and the performance of highway or transit links. This represents a challenge, given the wide differences in travel demand modeling capabilities among state DOTs and MPOs. Some agencies have sophisticated traffic simulation models, others have standard four-step transportation modeling systems, and still others have no formal models but rely on engineering estimates to generate transportation impacts. This chapter discusses the implications of different agency modeling capabilities.
Agencies with a Four-Step Travel Demand Model
Agencies with travel demand and network modeling capabilities can derive the Standard Traveler Benefits. Measures of aggregate changes in vehicle or trip rates, vehicle miles of travel and vehicle hours of travel are typically generated by the models. Safety impacts can be calculated based on information about link volumes, link type categories and average collision and injury incidence rates by mode and link type.
With travel models, an agency can also potentially derive most or all of the core data required for analysis of wider transportation impacts such as reliability, accessibility and intermodal connectivity. To do so, the following issues should be addressed:
- It is important to distinguish traffic conditions on each relevant link for multiple peak and off-peak time periods, because the reliability metrics depends in part on volume/capacity ratios (which can vary tremendously over the course of a day).
- It is important to distinguish the truck percentage of traffic on each relevant link, as much of the reliability effect on productivity is derived from truck delays that trigger added costs of safety stocks and stranded inventory, as well as loading dock overtime.
- Market access measurement depends on forecasts of project impacts on trip volumes, travel times and costs between zonal pairs, which can be readily calculated with travel demand models. It is also important to separate out the effects of induced traffic volumes in a travel demand model, since market access (agglomeration) effects related to scale economies can generate additional traffic that should be distinguished from traffic growth due to travel time savings alone.
- To correct the induced traffic and market access interactions, it may also be desirable to incorporate reliability effects into the measure of generalized costs between zones in the travel demand modeling process.
- Intermodal connectivity measurement can be enabled by identifying the zonal location of intermodal terminals, and then using network skims to calculate the times, distances and costs of ground travel to those terminals from surrounding zones. Additional information on connecting air, marine and rail services is embedded within the intermodal connectivity tool.
Note: In the long run, travel models can also be enhanced to more directly value the unique challenges of intermodal connectivity. For example, it is challenging for a model of one urban area to take into account the extra value (and economic gains) that results from better access to an air or rail freight terminal that provides better access to more distant areas. One idea is to treat intermodal terminals as transportation network nodes with added trip attraction, since improved access to those nodes generates higher benefits. However, the implications for trip distribution are unclear.
Agencies with No Travel Demand Model
Many rural regions and some rural states do not have travel demand models because the road network and population settlement pattern is sparse. As a result, there is a limited likelihood of travelers switching routes, modes or destinations. In these cases, a travel demand model is simply not justified. Standard traveler benefits may simply boil down to the following calculation:
Benefits = Cost Saving per Vehicle x Traffic Volume
(where the cost saving may include the value of travel time, expense and safety changes).
Yet, that does not mean that rural transportation projects have no further effects on business productivity or economic competitiveness. There are several counter examples. One is a transportation investment intended to improve reliability by making the highway or railroad less prone to closure by snow or flooding. Another example is a package of improvements to highway geometrics that reduce or eliminate sporadic delays caused by large trucks blocking traffic making wide turns. An additional example is a project to redesign at-grade railroad crossings to eliminate sporadic traffic backups when large freight trains block the road.
In all of these examples, it is possible to apply sketch planning methods that involve spreadsheets to assess route, mode, and destination shifts when there are only a handful of origin and destination zones. That approach is likely to be sufficient for intermodal connectivity as well as standard traveler benefit measures. Engineering estimates may be utilized to estimate project impacts on average peak period delays and required buffer times for the reliability analysis. In addition, external GIS systems or planner estimates may also be used to enable simple spreadsheet calculations of market access benefits, as long as the number of zones is very small.
Agencies with an Activity-Based Model
Some larger MPOs have moved towards activity-based modeling (ABM), which typically involves Monte Carlo micro-simulation to represent the choices made by a sample of travelers, and the constraints on those choices. Such methods have great potential to take account of factors such as the information available to the transportation users and the inter-personal and inter-temporal connections that affect their choices in ways impossible or simply impractical with conventional matrix-based models. While these methods may aid planning by providing more realistic scenarios, they raise questions for assessing the benefit of proposed projects.
These questions concern the data they provide and the use of micro-simulation. Each run of the model uses random numbers in forecasting choices, so the results from each set of inputs (such as a proposed highway improvement) reflect just one draw from a complex distribution. Ideally, the model should be run repeatedly to find the average results in the base and project cases, but it appears that this is not always done in practice. In addition, changing the seed values of the micro-simulation can alter the results when the model is run only a few times. This suggests that the seed file should be made consistent for the base and project cases and that the model should be run multiple times. As a result, there may be some concern that ABM, while valuable for understanding transportation behavior, may be less suitable for prioritization processes that call for calculating the benefits productivity of proposed projects.
Reliability Analysis Tools
The research team identified three spreadsheet-based tools that can be used for assessing reliability impacts of highway projects as part of a productivity impact calculation. All three were funded by the Strategic Highway Research Program 2 (SHRP2), administered by the Transportation Research Board. Each has a different intended use, and hence requires different types of inputs. The three reliability analysis tools are: (1) the C11 simplified reliability analysis tool for sketch planning, (2) the L07 reliability analysis tool for highway project designs and (3) the L08 “FreeVal” reliability analysis tool for freeway modeling. (Their letter-number designations are references to the corresponding SHRP2 Projects.) Key differences among them are summarized in Table 1
Table 1. Reliability Analysis Tools
Note that all three tools focus on road traffic characteristics, speeds and effects of traffic incidents on queuing. As such, they are not applicable for other modes, though their basic designs may be of use in developing custom tools for other applications because the general concepts of delay incidents and queuing also apply for air, rail and marine travel.
SHRP2 – C11 Reliability Analysis Tool for Sketch Planning
Overview. The C11 Reliability Analysis Tool for Sketch Planning was developed by Cambridge Systematics and Weris (2014) for SHRP2 Project C11.
It is a spreadsheet designed to function as a sketch planning tool for highway capacity projects that have impact on both travel time and reliability. The tool estimates total delay costs and disaggregates it into recurring delay (i.e., travel time delay that is due to speed slowdown), and non-recurring delay (i.e., delay that follows random traffic incidents – vehicle collisions and breakdowns). Costs associated with the non-recurring delay are referred to as reliability related costs.
Operation. The foundation of the model is the use of travel time distribution functions estimated in SHRP2 Project L03. These travel time distribution functions are measured in terms of a travel time index (TTI), which is the ratio of average travel time under congested conditions, divided by average travel time under free flow conditions. The TTI distribution is truncated at a lower bound of 1, which represents vehicles travelling at free-flow speed. It is also truncated at a higher bound of 6, implying a vehicle speed of 1/6 of free-flow speed (e.g., a highway with free flow speed of 60 mph functioning at only 10 mph). A travel time index of 1.8 implies that vehicles take 80 percent longer to travel the route compared to if they travelled at the free-flow speed.
The initial calculations made by the model are intended to fit the TTI distribution for the local route. This is undertaken by estimating ‘recurring’ travel time delay through the use of a generic speed/flow relationship and incident delay which is a function of the v/c ratio, number of lanes, and the length and type of time period. This allows the calculation of the mean TTI, as well as the calculation of other measures of the travel time index distribution (e.g., 80th or 95th percentile). Together, that data enables estimation of a generalized time equivalent measure of reliability for the route in question. From this measure, total delay costs can be calculated and then disaggregated into recurring delay and reliability costs.
Modal and Regional Coverage. The tool was developed for analysis of individual roadway links. It is most appropriately applied in cases where reliability impacts occur in discrete identifiable locations. (Using it to calculate reliability impacts for an urban network is cumbersome and requires extrapolating from the core of the underlying method to the model.)
Inputs and Outputs. To use this tool, users must specify a road link and then input the following information:
- Traffic data – Average Annual Daily Traffic (AADT) and Annual traffic growth rate (%)
- Truck data – Percent trucks in the traffic stream (combinations + single units)
- Capacity data – peak capacity as determined with Highway Capacity Manual procedures
- Road/highway data – either the G/C ratio (effective green time divided by cycle length) for signalized highways, or type of terrain (flat, rolling, or mountainous) for freeways and rural two-lane highways
Results are displayed for the base condition and improvement scenarios. A variety of reliability metrics are produced to allow users flexibility in interpreting the results. They also permit users to make independent estimates of the value of reliability if they want to use alternative measures of the reliability space:
- Delay – recurring delay (hours), incident delay (hours), total delay (hours)
- Travel time index (TTI) – overall, 95th percentile and 80th percentile
- Percent of trips < 45 mph and < 30 mph
- Congestion Cost – due to recurring delay, unreliability and total
Interpretation of Results. The “recurring delay costs” estimated by the reliability analysis tool correspond to the congestion delays estimated using standard travel benefit analysis (which recognizes travel time delay impacts as well as vehicle operating cost and safety impacts). So, the “recurring delay costs” should be removed from the delay estimated by the reliability analysis tool to avoid double counting. This also removes an element of delay costs not captured within STB analysis – that of expected delay due to traffic incidents. As the reliability analysis tool does not separate out expected delay due to congestion from expected delay due to incidents, it is not possible to include the expected delays due to incidents specific to the facility in the estimate of the value of productivity impact.
Impact on Productivity. Travel time and reliability costs will have a clear impact on productivity for business and commercial traffic (i.e., passengers traveling on employers’ business and for trucks or freight). The tool separates commercial traffic from auto driver or passenger traffic, but does not separate auto driver or passenger traffic between business, commuting, and personal trips. To estimate the impact on productivity due to reliability benefits for business auto drivers and passengers, the analyst will need to adjust the output of the tool to identify the component of the total reliability benefits that will have an impact on business productivity. Typically, these costs will be estimated using a standard percentage of business travel.
SHRP2 – L07 Reliability Analysis Tool for Design Treatments
Overview. The Reliability Analysis Tool for Design Treatments was developed by HDR Engineering for SHRP2 Project L07.
The reliability analysis tool for design treatments is implemented as a spreadsheet with a customized (Visual Basic) user interface. The tool estimates travel time delay, reliability, and safety benefits for a specific design treatment. It also estimates the impacts on the travel time index (TTI) by hour and calculates other reliability measures of effectiveness, such as lateness index, standard deviation, buffer index, and semi-variance. This tool is more appropriate than the reliability analysis tool developed in SHRP2 Project C11 if a proposed improvement involves one of the 16 tailored design treatments programmed into the model.
Operation. Like the reliability analysis tool developed in SHRP2 Project C11, the design treatments tool estimates reliability impacts using relationships estimated in SHRP2 Project L03. Unlike the C11 analysis tool, the design treatments tool can estimate impacts for 16 tailored design treatments. The tool uses four variables to estimate reliability impacts: Critical Demand/Capacity Ratio (dccrit), Lane Hours Lost (LHL), Hours with Rainfall Exceeding 0.05 inches (R0.05”), and Hours with Snowfall Exceeding 0.01 inches (S0.01”). For each tailored design treatment, the tool includes rule-of-thumb variable impacts estimated based on a literature review. These impacts are available under an assumptions option. The tool also allows the user to enter custom treatments by making manual changes in the four key variables.
Modal and Regional Coverage. The tool was designed to analyze individual roadway links with generally homogenous conditions (e.g., segments between successive interchanges). However, the tool can be used for a slightly larger segment if the highway geometry does not change drastically. It is best applied for discrete locations.
Inputs and Outputs. A graphical user interface (GUI) prompts the user to enter site-specific inputs using pull-down menus. In many cases, default values are provided. The site inputs are divided among the following tabs:
- Geometry – basic information about the facility, its location, and geometry
- Demand – traffic demand date for each hour of the day including peak-hour factors and heavy vehicle percentages
- Incident – information about crash and non-crash incidents in terms of frequency, duration, and cost
- Weather – hourly rain and snowfall entered by the user or from 10-year average data for proxy sites included in tool
- Event – percent demand impacts of user-defined special events by hour of day
- Work Zones – capacity impacts of short-term and long-term work zones by hour of day.
SHRP2 – L08 FREEVAL-RL
Overview. The Institute for Transportation Research and Education (ITRE) at North Carolina State University developed FREEVAL-RL to estimate travel time reliability impacts using the freeway reliability analysis methodology developed in SHRP2 Project L08. The tool is based upon FREEVAL, a spreadsheet-based implementation of the 2010 Highway Capacity Manual (HCM) procedures for the operational analysis of under- and oversaturated freeway facilities.
FREEVAL-RL is a more sophisticated analysis tool than the reliability tools previously described. FREEVAL-RL is able to test the reliability impacts of projects by dynamically modeling multiple combinations of demand and operating conditions along a corridor using a Monte Carlo simulation (i.e., repeated random sampling). The tool allows users to model up to 70 highway segments along a single corridor. Each segment can vary from the next in terms of demand and highway geometry. The HCM methods in FREEVAL-RL allow the tool to estimate the traffic impact in each segment and traffic queuing to spread from one segment to the next.
Operation. Modeling travel time reliability in FREEVAL-RL starts with a seed file. This file contains information on the overall freeway corridor as well as the geometry and demand of individual segments along the corridor. This serves as the base run for the Monte Carlo simulation. The user defines scenarios for the Monte Carlo simulation in terms of demand multipliers, demand patterns, weather probabilities, and incident probabilities. These scenario inputs are used to generate and run multiple scenarios until the results converge.
The user is able to control the number of scenarios run and eliminate highly unusual (i.e., low probability) combinations of events. FREEVAL-RL summarizes the results of the scenario runs in terms of probability density functions, cumulative distribution functions, and other reliability performance measures for the facility. Since the tool uses Monte Carlo simulation, modeling freeway improvements can be very time consuming (i.e., require several hours for a model run) and require relatively fast computer processors.
Modal and Regional Coverage. Unlike the other reliability analysis tools, FREEVAL-RL is able to model an entire freeway corridor. The tool has constraints in terms of the number of segments and time period modeled, but only very long or highly congested corridors will exceed these constraints. Regional impacts can be modeled by stringing together several corridor analyses. However, the impacts of one corridor on another cannot be taken into account and FREEVAL-RL model runs can be very resource intensive.
Inputs and Outputs. To use this tool, users must develop a seed file with the following information:
- Seed File Management – study period, analysis year, seed file demand, terrain type, ramp metering , and other corridor-level information
- Individual Segment Data – length, number of lanes, segment demand, free-flow speed, capacity, percent trucks, adjustment factors, ramp demands, percent trucks on ramps, number of lanes on ramp, and ramp orientation.
The model uses scenario data to generate the Monte Carlo runs. Scenario data include:
- Demand Multipliers – demand adjustments by month and day of week
- Demand Pattern Configuration –demand patterns adjustments for specific months and days of week
- Weather Probability – monthly occurrence of rain, snow, and low visibility conditions
- Incidence Probability – detailed incidence data or estimates from “data poor” equations.
FREEVAL-RL generates analysis scenarios from the previous information. The user is given the ability to eliminate low probability scenarios from the analysis to save on model run time. Results are displayed in terms of several facility reliability performance measures:
- Travel Time Index (TTI) – mean, 50th, 80th, and 95th percentiles
- Misery Index
- Semi-Standard Deviation
- Reliability Rating
- Percent Vehicle-Miles Traveled (VMT) at TTI > 2
The tool also provides a summary of the percent contribution of recurring and nonrecurring delays. Additional analysis details are provided to test model calibration.
Interpretation of Results. Unlike the other two reliability tools, FREEVAL-RL does not calculate a monetized value of travel time reliability benefits for inclusion in benefit-cost analysis. The user must calculate these values outside of FREEVAL-RL using one of the freeway reliability performance measures reported by the tool. The user would need to multiply the change in reliability (80th percentile TTI – 50th percentile TTI or the change in the semi-standard deviation) by the value of reliability (value of time times the reliability ratio). The addition of travel time reliability benefits would not double count any benefits already included in standard benefit-cost analysis.
Accessibility Analysis Tools
The research team identified two spreadsheet-based tools that can be used for assessing the market access impacts of ground transportation projects as part of a productivity impact calculation. The Effective Density Tool and Fixed Threshold Tool were both funded by the Strategic Highway Research Program 2 (SHRP2), administered by the Transportation Research Board, and programmed by Texas A&M University’s Transportation Institute. These two spreadsheet-based tools illustrate how it is possible to tailor measurement of agglomeration impacts to capture either urbanization or localization effects. They are accompanied by a detailed literature review and report on their uses (Texas A&M Transportation Institute, 2014).
Each accessibility tool has a different intended use and ability to capture or reflect “localization effects” (support for clustering of businesses with similar or complementary activities) and “urbanization effects” (support for enhanced access to labor, supplier or customer markets). This comes from their different capabilities regarding zonal system detail, zonal activity measures, decay factors and threshold factors, as shown in Table 2
Table 2. Market Access Analysis Tools
These differences lend each tool to a different form of use:
- As a general approach, the Effective Density tool can be used with zonal employment data to capture the effect of transportation projects on broadening economic markets – reflecting access among firms and employees. This approach, taken in the UK, can represent a composite of business localization and urbanization effects.
- Alternatively, the Effective Density tool can be used with zonal population as well as employment data to better capture benefits of broader labor and shopper market access (urbanization benefits for industries gaining scale economies from population access).
- The Fixed Threshold Accessibility tool is set up to use zonal employment data for specified industry sectors that have specific clustering characteristics (regarding relevant types of businesses and a proximity or connectivity thresholds), and thus can best capture localization benefits– i.e., support for clustering and interaction among specific types of businesses (e.g., high tech clusters or same-day supply chains).
The options for zonal metrics, decay and threshold factors are of particular note. The choice of zonal attributes will affect relative rankings of projects. To understand why, consider that most major US cities have office employment more concentrated in central zones while population, housing and retailing are more dispersed and prevalent in outlying areas. As a result, measuring effective density in terms of zonal employment might show greater gain with improved radial access to the urban center, while measuring effective density in terms zonal population might show greater gain with a new circumferential (non-radial) suburban corridor project.
Decay and Threshold Functions can be particularly important for accuracy regarding impacts on business workforce access and supply chains interactions. For instance, same-day parts delivery (needed for just-in-time processes) is limited to legal and practical limits on driver hours/day, while labor market access is limited by acceptable commuting travel times (which tend to attenuate rapidly as travel times go beyond the 40-60 minutes threshold). Further work is needed to understand how decay and threshold functions vary by industry.