ITS - Intelligent Transportation Systems Report ITS Home Page
Table of Contents

3.0 Performance Measures and Analysis Approach

The AMS methodology includes the capability to convert all impact/performance measures to non-mode specific measures such as person trips.  These mode-independent performance measures will be produced by an interface tool that can translate AMS model components outputs into non-mode-specific performance measure output.  Since ICM is multimodal, the operational impacts need to be measures beyond the traditional network-based measures.  The AMS framework outputs will be converted to performance measures, such as person travel time or trip reliability, in order to evaluate and compare operations among the alternative paths and properly portray the collective corridor-wide performance.

The AMS methodology is flexible enough to accommodate the analysis needs of corridors in a range of urban areas across the nation.  For example, corridors in the nation’s largest metropolitan areas may have broad and complex corridors served by multiple layers of transit operations.  Corridors in smaller or developing markets may have simpler corridors with more limited transit operations.  Certain types of analysis may not be relevant in particular jurisdictions, just as complex feedback between classes of analytical models may not be required in some corridors.

The performance and benefit-cost framework outlined in this chapter establishes traffic analysis goals and objectives; and sets expectations, needs, and issues related to the corridor study analysis performance measures, expected output, and project prioritization/cost-benefit requirements.  The following objectives of this framework are to ensure that performance measures and analysis methods:

This chapter presents:

3.1 Corridor Analysis Approach

Figure 3.1 provides an overview of the ICM AMS corridor analysis approach.  It describes the process that should be followed in the AMS of Pioneer Site corridors.

Figure 3.1 Overview of the ICM AMS Analysis Approach

Figure 3.1 is a flow diagram overview of the ICM AMS Analysis Approach. The diagram consists of four approach steps: Initiation, Existing Conditions, Mitigation Measures, and Analysis. The initiation step involves the kickoff meeting, the development of performance measures, the selection of analysis tools and evaluation approaches, and the work scope. The existing conditions step involves two substeps: Summarizing the existing conditions and analyzing them. Summarizing the existing conditions involves collecting data from available reports, archived ITS data, census data, 511 data, etc., while analyzing the existing conditions involves identifying problem areas in terms of recurrent congestion or safety and identifying the causes of the problems. The mitigation step involves identifying capacity, operational and transit improvements and determining performance measures and cost estimates to arrive at near-term and long-term strategies. The analysis step entails evaluating the mitigation strategies under consistent performance measures to prioritize the strategies and projects. Throughout the entire four-step approach there should be stakeholder involvement.

The following steps are involved in this process:

The data will be analyzed to determine causes of existing recurrent traffic congestion problems in the corridor.  Locations of freeway bottlenecks will be identified, as well as other locations that may constitute mobility constraints in the corridor, such as freeway ramps or arterial intersections.  The data also will be analyzed to quantify the magnitude of non-recurrent congestion in the corridor.  The results of the existing conditions analyses will be summarized in an Existing Conditions Technical (ECT) memorandum.  At a minimum, the ECT memorandum will include a description of the roadway and transit network, including a map showing the corridor study network; and a detailed description of existing traffic performance on the corridor with specific explanations of the causes of congestion problems.

Develop ICM strategies and projects – In this step, the analysis team will refine the ICM strategies developed for the corridor.  The proposed strategies will be segregated into short- and long-term implementation timelines, consistent with the Concept of Operations documents developed at Pioneer Sites.  For the identified mitigation strategies, the analysis team will develop performance measures and prepare planning-level cost estimates.  The list of strategies and projects then will be summarized in a technical memorandum.

Evaluation of congestion mitigation strategies and projects – In this step, the analysis team will evaluate the ICM strategies and projects making use of AMS framework described in previous chapters.  The analysis is intended to identify locations of freeway bottlenecks, changes in aggregate congestion levels in the corridor, and changes in peak-period travel times and delays.  Based on the analysis, a prioritized list of recommended measures will be developed, including a narrative explaining the rationale for the prioritization.  A technical memorandum will be produced summarizing the results of the analysis; and a prioritized list of ICM strategies, including recommendations for any modifications to proposed projects and strategies.

3.2 Performance Measures

This section provides details on the proposed performance measures to be used in the evaluation of ICM strategies.  To be able to compare different investments within a corridor, a consistent set of performance measures should be used.

The following are primary objectives of the proposed performance measures:

As much as is feasible, the primary performance measures should be reported for existing and future conditions, and should be easily calculated to evaluate any proposed improvement scenarios.  To the extent possible, the measures should be reported by:

The proposed performance measures focus on the following four key areas:

  1. Mobility – Describes how well the corridor moves people and freight;
  2. Reliability – Captures the relative predictability of the public’s travel time; and,
  3. Safety – Captures the safety characteristics in the corridor, including crashes (fatality, injury, and property damage).

Mobility

Mobility describes how well the corridor moves people and freight.  The mobility performance measures are readily forecast, making them useful for future comparisons.  There are two primary types of measures proposed to quantify mobility:  1) travel time and 2) delay.  Other proposed measures that are commonly used to describe mobility are volume-based measures derived from distance and travel times, such as total vehicle-miles traveled (VMT), person-miles traveled (PMT), vehicle-hours traveled (VHT), and person-hours traveled (PHT).  Person-hours of delay (PHD) can also be used as a mobility measure.  Descriptions of these performance measures, including how they can be calculated and caveats surrounding their use, are provided below.

Travel Time

Travel time is defined as the average travel time for the entire length of the corridor or segment within a study corridor by facility type (e.g., mainline, HOV, local street) and by direction.  Travel times should be computed for peak periods and by hour, and used in calibrating traffic analysis tools for AMS.

When developing real travel time data, any gaps in the detection system will have to be accounted for.  In cases where there are very limited data, field observations, as well as input from local agencies, will have to be utilized to validate assumptions and analysis conclusions of the ICM study teams.  Some additional data collection will have to be performed when critical to the analysis.

Where gaps exist in ITS system coverage, probe vehicle travel times can be used as being representative of the travel time over the gap.  A field observation may also reveal that travel times over the section missing detection is at free-flow speeds – meaning that probe vehicle runs may not have to be done.

Travel times are inputs to the subsequent measures:  delay and reliability.

Delay

Delay is defined as the total observed travel time less the travel time under non-congested conditions, and is reported as either vehicle-hours or person-hours of delay.  Delays should be calculated for freeway mainline and HOV facilities, transit, and surface streets.

Many transportation agencies define the freeway congested speed threshold as 35 mph, because this is in the speed range at which traffic flow breaks down and becomes stop and go.  Vehicles traveling at freeway speeds above 35 mph are not considered to experience any delay.  Delay is calculated by using the following formula:

Equation is the formula for calculating delay. Delay equals the number of vehicles affected per hour multiplied by the distance they were affected multiplied by the duration they were affected multiplied by the subtraction of the inverse of 35 mph from the inverse of the congested speed.

Many agencies use a freeway lane capacity 2,200 vehicles per hour per lane (vphpl).  2,000 or 2,200 vphpl are commonly used by engineers as the design capacity, or the bottleneck capacity of an urban freeway lane.  Figure 3.2 shows an example summary of average daily delay on a hypothetical corridor.

For arterials and surface streets a similar threshold is needed to separate congested speed from free flow – the speed limit or the 85th percentile travel speed can be used to make the calculation applicable to arterials and surface streets.

Figure 3.2 Example Average Daily Delay by Day of Week and Time of Day

Figure 3.2 is a graph of the average daily delay by day of week and time of day. Along the x-axis are the hours of the day beginning with 0:00 and ending with 23:00. Along the y-axis is the average daily vehicle hours of delay based on 60 mph beginning at 0 and ranging to 1,500 hours. There are three lines on the graph: weekday, Saturday and Sunday/holiday. The weekday line experiences a spike during the morning peak period to about 950 hours and another at during the p.m. peak period to about 775 hours; it can be noted that the morning spike has s duration from about 6:00 a.m. to about 10:00 a.m. while the afternoon spike begins at 1:00 p.m. and ends at about 8:00 p.m. The Saturday line has the highest spike reaching about 1,550 hours and has a duration beginning at 2:00 p.m. and ending at 5:00 p.m. There is no real spike in the Sunday line, which reaches a maximum of about 175 hours of delay at 5:00 p.m.

VMT, PMT and RPMT, VHT and VMT, and PHD

Vehicle and person-miles or hours traveled (VMT and PMT) and person hours of delay (PHD) are relatively straightforward calculations once travel times and delays are established.  VMT is computed by segment by time period as the flow × the segment length.  Along a corridor, multiple segment VMT can be summed to arrive at the corridor-level VMT.  PMT is simply VMT × average vehicle occupancy.  Where transit ridership and vehicle occupancy data are available, PMT can be calculated for a segment by multiplying ridership by distance.  In this case, total PMT is:

(Total Autos) × (Segment Length) × (Average Auto Occupancy) + (Total Transit Ridership) × (Segment Length)

If specific transit ridership is not available, PMT can be computed as:

(Total VMT) × (Average Vehicle Occupancy)

As with VMT, PMT can be aggregated from the segment level to the corridor level.  It is advised to use household survey vehicle occupancy data with caution, since it is based on a sample size that may not be factored appropriately for corridor-level analyses.

PHD is computed by multiplying vehicle hours of delay times average auto occupancy.  Autos should be computed as total vehicles less transit vehicles, if possible.  For transit, estimate total transit ridership during the peak period and multiply it by delay per transit vehicle hours of delay, where applicable (i.e., distinguish between HOV and mainline speeds to use for transit travel times).  If transit-specific data is not available, multiply vehicle hours of delay by average vehicle occupancy gathered from other sources.

A multimodal performance measure that can useful in ICM AMS is “Reliable PMT” (RPMT) – this measure can help summarize transit, arterial, and freeway performance into one measure that describes corridor performance.  The “Reliable” part of RPMT can be derived by comparing the PMT for a certain scenario to a target maximum or optimal RPMT (RPMT*).  This can be calculated analytically by loading a simulation network incrementally to some maximum throughput level before systemwide decline.  An acceptable RPMT could then be defined as an RPMT that does not deviate more than x percent from RPMT*.  RPMT is not an observable field measure; however, high RPMT is likely associated with values of observable field measures like travel time, travel time reliability, and bottleneck throughput.

Reliability

Reliability captures the relative predictability of the public’s travel time.  Unlike mobility, which measures how many people are moving at what rate, the reliability measure focuses on how much mobility varies from day to day.  Analysis techniques that can be used to forecast travel time reliability include the use of simulation models (e.g., performing multiple runs of the same forecast scenario, while adjusting flows and/or other input variables); or the ITS Deployment Analysis System (IDAS) methodology; or the Highway Economic Requirements System (HERS) methodology; or the TTI “buffer index” method.

To illustrate the importance of the reliability measures, Figure 3.3 shows two hypothetical corridors of the same length having the same average weekday travel time of around 22 minutes (i.e., they have the same level of mobility).  However, Corridor “B” on the right side of the figure has a much wider day-to-day variation in travel time, and is less reliable than corridor “A.”  Even though they both experience the same average level of mobility, it is very likely that the travelers on Corridor “B” will remember those days where the travel time exceeded 35 minutes much more than the travelers on Corridor “A” will remember those few days where their travel time barely exceeded 25 minutes.

The “buffer index” method can be used to estimate reliability in conjunction with  other measures, such as the standard deviation of travel time or using other percentile measures (e.g., the 85th percentile travel time).  The buffer index is defined as the extra time (or time cushion) that travelers must add to their average travel time when planning trips to ensure on-time arrival.  On-time arrival assumes the 95th percentile of travel time distribution.

Figure 3.3 Illustrative Difference Between Mobility and Reliability

Figure 3.3 is composed of two graphs illustrating the difference between mobility and reliability. Each graph has the date along the x-axis and the daily travel time along the y-axis. The first graph is for corridor A and the second is for corridor B. Both corridors are the same length and have the same average weekday travel time of around 22 minutes (i.e., they have the same level of mobility). However, Corridor B has a much wider day-to-day variation in travel time, and is less reliable than corridor A. Data points for corridor A range between 17 minutes and 27 minutes, while data points for corridor B range from 17 minutes to 37 minutes.

The buffer index is fairly easy to communicate to the general public.  It also is presented as a percentage, which makes it comparable among the different corridors and modes.  Figure 3.4 illustrates two ways to present the buffer index.  The first chart in the figure shows the average travel time and the buffer index by hour of the day.  The second chart on the right averages the travel time over the four periods of the day (a.m., p.m., midday, and evening/early morning).

Figure 3.4 Different Ways to Present the Buffer Index

Figure 3.4 has two graphs depicting different ways to present the Buffer Index. The first graph presents the Buffer Index by hour and the second by period. Both graphs have average travel time along the y1-axis and the buffer index along the y2-axis but the first graph has hours along the x-axis ranging from 0 to 23, while the second graph’s x-axis is divided into four periods: a.m., midday, p.m. and night. The first graph has two lines, one for the average travel time and the second for the buffer index, while the second graph is made up of vertical bars for the average travel time and points and lines for the buffer index.

For example, a buffer index of 30 percent for a corridor of 10 miles has the same relative reliability as a 30 percent buffer index for a corridor of 20 miles.  The FHWA has a web site with more detailed information on how to apply the buffer index for planning purposes and provides links to additional resources at http://ops.fhwa.dot.gov/publications/tt_reliability/.

To illustrate, a buffer index of 40 percent means that, for a trip that usually takes 20 minutes, a traveler should budget an additional 8 minutes to ensure on-time arrival most of the time:
            Average travel time = 20 minutes
            Buffer index = 40 percent
            Buffer time = 20 minutes × 0.40 = 8 minutes

The average travel time can be estimated as described above in the travel time calculation discussion from above.  The 95th percentile travel time can be obtained by sorting each day’s travel time for the given hour.  The buffer index is the difference between the 95th percentile travel time and the average travel time for the year divided by the average travel time:

Equation: The Buffer Index is equal to the an expression with the subtraction of the Average Travel Time from the 95th Percentile Travel Time in the numerator and the Average Travel Time in the denominator.

Safety

For the safety performance measure, it is suggested to use the number of accidents and accident rates from accident databases linked to highway databases.  The highway database contains description elements of highway segments, intersections and ramps, access control, traffic volumes and other data.  Accident databases contain specific data for accidents on state highways.  Each accident record contains a ramp, intersection, or highway post mile; and includes other data, such as the following:

Safety is a fairly difficult measure to forecast.  There are some tools available that estimate the potential reduction in accidents, given certain types of improvements.  For the purpose of ICM AMS, safety analysis can be conducted qualitatively using expected levels of improvement in safety as a result of deploying mitigation strategies (e.g., major improvement, minor improvement, none, slightly worse, etc).

3.3 Analysis of Non-Recurrent Congestion

Collectively, all the tools in the ICM AMS framework are capable of supporting the analysis of both recurrent and non-recurrent corridor scenarios.  The non-recurrent scenarios that will be supported include major and minor incidents (unplanned), special events, weather, and work zones.  These non-recurring scenarios entail various combinations of increases of demand and decreases of capacity.  The relative frequency of non-recurrent conditions is also important to estimate in this process – based on archived traffic conditions.  Otherwise, resource allocation may be drawn towards highly unlikely events.

Figure 3.5 depicts this approach:  key ICM impacts may be lost if only “normal” travel conditions are considered; the proposed scenarios take into account high and low travel demand, incidents, work zones, and weather conditions.  These are sources of variation in the performance of the transportation system.  Possible analysis scenarios are depicted in Figure 3.6.

Figure 3.5 Key ITS Impacts May Be Lost If Only “Normal” Conditions Considered

Figure 3.5 is a diagram depicting incidents that may occur that will affect the impacts of Intelligent Transportation System performance. The diagram shows four incidents located on a graph with travel demand along the x-axis and incident-severity along the positive y-axis and weather severity along the negative y-axis. The examples of incidents affecting ITS impacts are weekend construction, tractor trailer fire during rush hour, a snowstorm, and a special event.
Source: Wunderlich, K., et al., Seattle 2020 Case Study, PRUEVIIN Methodology, Mitretek Systems.  This document is available at the FHWA Electronic Data Library (http://www.itsdocs.fhwa.dot.gov/).

Figure 3.6 Sources of System Variation: Classifying Frequency and Intensity

Figure 3.6 is a three-dimensional graph illustrating sources of system variation. The three axes are defined by increasing traffic demand, increasing incident severity or frequency, and increasing weather severity. Examples of events at the extremes are shown: Snow/sleet conditions is the example given for an incident with high weather and incident severity/frequency but a low traffic demand; a thunderstorm during p.m. rush hour is an incident with sever weather and a high traffic demand but a low incident severity/frequency; and, a weekend freeway closure is given as a sever incident with low traffic demand and low weather severity.
Source: Wunderlich, K., et al., Seattle 2020 Case Study, PRUEVIIN Methodology, Mitretek Systems.  This document is available at the FHWA Electronic Data Library (http://www.itsdocs.fhwa.dot.gov/).

3.4 Analysis Approach – Existing Conditions

This section outlines the general approach for evaluating existing conditions.  There are “rules of thumb” that should be applied where possible:

The timeframe for analysis should focus on weekday peak-period conditions.  Weekends should be assessed where such an analysis would influence projects or strategies to be used, or where weekend conditions vary considerably from the weekday.

Peak-period analyses should be performed at a minimum, and hourly estimates should also be used where appropriate data are available.  Mid-day and off-peak periods would be of interest, if data are readily available.  If after an assessment of the availability of existing traffic data the study teams determine that data gaps exist, the analysis teams should make a recommendation for additional data collection.

In addition to corridor-wide performance for existing conditions, individual bottlenecks should be evaluated to determine their overall contribution to corridor congestion.  Once bottlenecks are identified field observations need to be performed to determine and document the cause of the bottlenecks.

Existing and future corridor conditions should also be assessed for arterials in addition to freeways.  Mean speeds and average traffic volumes can be used to describe arterial traffic performance, for both the baseline and mitigation strategies.

3.5 Analysis Approach – Future Conditions

Two analysis levels are recommended for the analysis:

  1. First, in comparing alternatives, a low-level/screening analysis should be used to screen out non-viable alternatives; and
  2. Second, viable alternatives that emerge from the screening analysis should be assessed using higher-level, more detailed analysis.

Analysis Timeframe

In conducting the analysis, the study area and analysis timeframe should be defined so as to contain congestion both spatially and temporally.  The primary focus in ICM AMS traffic analyses is on peak weekday periods – not only peak-hour conditions.  Weekends should be assessed where the analysis would potentially influence the selection of projects or strategies.

If assumptions need to be made for peak spreading in future traffic conditions, the analyst can check the reasonableness of future queues and travel demand, and make peak-spreading assumptions that would result in queues and delays that would be acceptable by the traveling public.  The peak-spreading approach should be thoroughly documented and applied to both future baseline and alternatives so that benefits of the improvements can be demonstrated in a consistent way across alternatives.

Analysis of ICM Strategies

The identified ICM strategies will be segregated into short- and long-term implementation timelines.  The identified strategies should then be grouped into analysis scenarios.  Figure 3.7 shows the desired organization of analysis scenarios and results.

Figure 3.7 Example Analysis Scenarios

Figure 3.7: graph representing example analysis scenarios

Cost Estimation

For the identified mitigation strategies, the analysis team should prepare planning-level cost estimates, including life-cycle costs (capital, operating, and maintenance costs).  Costs should be expressed in terms of the net present value of various components.  The analysis team can use consistent percentages for soft costs, such as design and contingency costs.  Also, the FHWA Cost Database can be used to assist in producing capital and operating and maintenance (O&M) costs for ICM strategies.

3.6 Output

The output of each ICM corridor analysis should be a well written narrative of the identified problems and recommended solutions, and a clear and concise description of an implementation sequence and schedule for project and strategies for any given corridor.  In addition to the narrative, output and reports should be graphical to the extent possible, and then tabular.  Output performance measures must be consistent across corridor analyses, existing and future conditions, and mitigation strategies and scenarios.

Each corridor report should include the following chapters:

next | previous