Transportation system efficiency reduces congestion on roads, resulting in less time in traffic and more time with family, recreation. It also means reduced cost of road improvements. Roads really take a beating from heavy and constant traffic. But fewer cars means fewer road projects like street widening, and less road maintenance. That saves money for everyone.
Our current goal is to calculate–based on the trip-impact information that we already produce–the ROI of the ACCS TDM programs in terms of user and public or societal benefits. Put another way, we wish to estimate the monetary value of time saved, pollutant-emissions reduced, fuel saved, and road-repair work avoided, for every dollar invested in TDM programs. Though noise reduction and travel-time reliability are benefits associated with improved system efficiency, we are not considering those benefits at this time.
We have sketched out a calculation method for reaching our goal in anticipation of future funding and research work that our team will undertake.
In order to measure the benefits of Arlington’s TDM, instead of analyzing how system efficiency would improve with TDM added, our method will require us to calculate how system efficiency would suffer if TDM were not present. This is because existing traffic in the region reflects vehicle and VMT reductions that Arlington’s TDM already generates. The change in congestion (and thus time loss) and other factors under this scenario will represent the value of TDM’s ROI.
Calculating Congestion and Time-Savings Benefits of TDM
In our view, calculating the congestion and time-savings benefits of TDM will be the most technically complex part of our work.
Though a variety of existing tools might be used to estimate changes in travel conditions in a hypothetical “no-TDM” scenario, none of the off-the-shelf sketch planning or network-based models appear to be reasonable options for our work. The sketch-planning models are typically used to analyze a single corridor without considering impacts beyond an immediate project area, while Arlington’s TDM programs are directed on a regional level.
The network-based models were overly complicated for our use; they would require outputs from a network travel-demand model, which is beyond the scope of our analysis. Many models are also outdated, with economic values and default parameters from very old data sources; in most cases it would be difficult to update the factors due to the complex relationships embedded in the models. Also, existing models typically cannot accommodate multiple modes simultaneously, so they could not easily be applied for automobile and other modes together.
Ultimately, we have focused on developing a spreadsheet method that utilizes the standard speed-flow curve calculations defined in the Transportation Research Board’s Highway Capacity Manual (HCM) (Transportation Research Board, 2000; Muench, 2006). We are still in the exploratory phase of this approach, but we realize that it will require data on trip origin, destination, and travel routes to assign TDM trip and VMT reductions to specific roadways. Our analysis will then use the spreadsheet tool to calculate the speed at the existing, “with-TDM” condition and the projected “no-TDM” scenario from which we will estimate the change in speed for the selected roads.
Of course, our approach does have methodological problems that we must resolve. First and foremost, while we can calculate the number of SOV trips that our TDM programs remove from the road network each day, the surveys that form the basis of our calculations have not collected information about the routes that TDM-program participants currently take for commuting or would take on SOV trips if they were not participating in TDM. In other words, we do not know which roads those trips would burden if they were added back on in our hypothetical scenario. Without this information, we cannot reasonably assume whether our programs are removing trips from highly congested roads (where congestion-reduction and time-savings benefits would be large) or from uncongested roads (where benefits would be small or non-existent).
We do have survey-based information about trip origins and destinations, however, and this will be the basis for assigning trips to the road network for input in the HCM model. In addition, we will collect regional data on the routes chosen by commuters between those origins and destinations.
With these data sources, we will first cross-tabulate the trip origins, destinations, and modes reported by participants in earlier surveys. We will then eliminate any trips that do not begin, end, or likely pass through Arlington County. The vast majority of trips impacted by the ACCS TDM programs fall into one of these three categories; therefore, we believe that we can limit our analysis to these trips for simplification purposes.
We will then assign these trips to routes on the road network by using the regional data on route distribution. For example, for the origin/destination pair of ZIP code 20009 (Washington, DC) and 22209 (Arlington, VA), we will define the share of trips that travel through the District and across the Arlington Memorial Bridge to Rosslyn as well as the share that travel through Georgetown and over the Key Bridge, and other routes. Then we will multiply the total number of trips between origin/destination pairs (as cross-tabulated from earlier surveys) by this route distribution, which will give us an approximation of how many trips would be travelling on each road between each origin and destination in the absence of TDM.
The goal will be to assign trips to specific freeway and arterial road segments; however, we realize that there may not be enough respondents from each origin/destination pair from which to create reasonable distributions of trips across the road network. If sufficient cases are not available, we will need to decrease the geographic resolution of our analysis and aggregate trips to corridors of parallel roadways.
With these TDM-influenced trips assigned to the road network, we will then gather data for existing peak-period trips, or the base to which we will add the TDM-shifted trips. We will aggregate our existing-trip data to the same geographic resolution as the trip assignments made for the vehicle trips removed by Arlington’s TDM programs. At present we believe that the best source of observed, existing trips is regional data based on analysis of satellite imagery to determine roadway traffic volumes during the AM and PM peak periods. For non-freeway road networks, we will consult with the Departments of Transportation for Virginia, Maryland, and the District of Columbia.
Using a geographic information system (GIS), we will then add the TDM-shifted trips, distributed across the road network, to the number of existing peak-period trips on the same roadways, and produce a sum that equals the new traffic volume that would be present on each roadway without TDM. From the change in observed vehicle volume to the vehicle volume in the absence of TDM, we will derive the change in travel speed, and thus delay.
Example curve that describes the relationship between traffic volume and speed. Source: Muench 2006
Finally, we will multiply the number of vehicles and travelers who experience this increased delay by the monetary value of time spent commuting in order to determine the monetary value of the time lost to congestion in a scenario without TDM.
In order to reach this goal, we will need to resolve a basic problem with our HCM approach. As shown in the graph below, the HCM approach is designed to measure speed changes in free-flow travel conditions, and does not adequately address changes in already-congested conditions; the speed flow curves are calibrated only through LOS “E”, which occurs before traffic flow breaks down. (Muench, 2006).
Example speed-flow curves and their relationship to traditional level of service (LOS) on freeways. Source: Muench, 2006.
Furthermore, we have yet to determine at what points on the road network we should measure the change in travel speed. The most logical points are at known bottleneck locations, where congested conditions already exist, but the method of adding back trips that are not currently on the network might also be appropriate at locations that are upstream of the bottlenecks in order to assess how much farther the travel delay would extend if the TDM programs were not in place.
Also, we must determine how we will assign VMT impacts when trips cross analysis area boundaries. A vehicle trip should be counted on every segment of road it uses. Thus, trips removed from a road that originates, ends, or passes through Arlington should be counted on each Arlington road it uses. But the VMT impacts will be more dispersed, with some travel outside of the County. Because some VMT-related benefits, such as emissions reduced, are not location specific, while others, such as road maintenance cost savings, are specific to where travel occurs, it might be reasonable to use different VMT counts for different benefits, with some VMT excluded from the calculation.
Finally, we must determine how our analysis should address induced demand and diversion from other roadways when TDM actions create capacity. Can they be ignored, because diverted traffic frees space elsewhere and induced traffic reflects economic opportunity?
Determining the Monetary Value of Other TDM Benefits
Calculating the monetary value of reduced vehicle-operation costs, accidents, roadway maintenance, and vehicle emissions will be relatively straightforward. We can multiply trips and VMT reduced (as we calculate with our impact-calculation method) by the generally accepted factors in order to reach monetized cost savings.
For roadway maintenance and vehicle-operation costs, we will utilize the standard monetary values from AAA and the National Highway Traffic Safety Administration as a function of trips and VMT. We will then multiply the trips and VMT reduced as found in our impact calculations by these values. At this time we have not developed a methodology for adding transit costs in to our estimates given the wide variation in fare structures around the region and the distance-based fare structure used by WMATA’s Metrorail. Also, we do not intend to calculate savings to individuals who sell a vehicle because of their mode shift.
Using the EPA value for the cost of accidents per unit of VMT, we will calculate the dollar value of traffic accidents avoided due to decreased automobile travel.
For air pollution, which does not have a direct monetary value, we will use monetary values from the EPA, such that we can multiply trip and VMT reductions with volumes of air pollutants generated per trip and per mile of travel and then multiply the total volume of air pollutants removed by the monetary value of those pollutants. We recognize that internal combustion engines operate less efficiently, and thus produce more pollution, at low speed. Therefore, we will also calculate the value of increased pollution from automobiles that will result from slower travel speeds that are the result of the increased congestion calculated in the “no-TDM” scenario of our congestion and time-savings benefit calculations.
Summing Benefits and Comparing to the Costs of TDM
To complete our calculations, we will sum the monetary values of time savings, pollution reductions, vehicle-operating-cost reductions, and road-maintenance savings to produce and divide that total value with the dollar cost to ACCS for providing TDM services. The quotient of this calculation will be the benefit of TDM per dollar invested in TDM.
Creating a Transferrable Calculation Tool for Estimating TDM Trip Impacts
As a final goal for our system-efficiency research, we are seeking to modify the Arlington TDM impact calculation method for easy transferability to other regions around the United States using their local service and travel inputs. Our goal will be to develop a tool that will provide a calculation system with the flexibility to accommodate different, commonly-offered TDM services, along with guidance for defining multiplier factors from local data or for selecting reasonable surrogate calculation inputs where local data are not available.
We will begin with a review of trip-reduction and air-quality analysis for TDM programs. We will accomplish this through a literature review and, as necessary, follow-up interviews with practitioners and researchers.
Based on this review, we will select the types of services (e.g., guaranteed ride home, carpool matching, transit subsidies, etc.) that are most commonly offered by TDM programs and compile results of existing TDM evaluations that have been conducted for these services. We will use these studies to document the range of trip- and emission-reduction factors generated by the services in various settings and under various implementation conditions.
In order to understand the local differences that impact TDM effectiveness, we will document the environmental factors (e.g., land use patterns, city size, transit availability) and demographic factors (e.g., labor participation rate, income, overall commute mode split) that existing research shows to have effects on TDM vehicle trip reduction, and collect indicators for these factors in the areas served by the TDM programs studied in earlier trip-reduction and air-quality research.
Using this information, we will then build a TDM-impact-calculation tool with options that allow users to select which types of TDM services are under study, and which allows for detailed inputs derived from local data sources, the use of default values based on our earlier research, or a mix of the two.
Written guidance will accompany the calculation tool to help analysts select specific calculation multiplier factors to estimate vehicle trip and VMT reductions for the TDM program under study, given the environmental and demographic conditions in the program’s service area.
Works Cited
Muench, S. (2006). Freeway & Highway Level of Service. Retrieved from http://courses.washington.edu/cee320w/lectures/Freeway%20&%20Highway%20LOS.ppt
Transportation Research Board. (2000). Highway Capacity Manual 2000. Washington, DC: National Research Council.