U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

 
REPORT
This report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-13-026    Date:  March 2014
Publication Number: FHWA-HRT-13-026
Date: March 2014

 

Guidance on The Level of Effort Required to Conduct Traffic Analysis Using Microsimulation

CHAPTER 7.  ALTERNATIVES ANALYSIS

OVERVIEW

The primary purpose of building models is to evaluate alternatives in support of the decisionmaking process. The alternatives analysis process has two fundamental parts: the simulation modeling of alternatives and the sequencing of the alternatives to analyze. From an overall workflow perspective, the alternatives analysis step is similar for all projects in terms of how the modeling steps are conducted. However, the number of alternatives and workflow for what is being analyzed and reported can be very different and therefore needs to be customized based on the issues and the nature of the proposed project and improvements.

There are three main areas of consideration for an alternatives analysis, as shown in figure 21. There is the simulation modeling work itself, which includes carrying forward relevant calibration parameters, satisfying the required number of model runs, forecasting traffic, and ensuring that error checking has been conducted. The second area is the sequencing of alternatives and timeframes that are to be simulated. Time should be invested up front to investigate and lay out an optimal approach for the sequencing of the simulation analysis to reduce the overall level of effort. The third area is the presentation of modeled results, where the report summaries must be designed to be illustrative and effective at communicating differences.

This figure shows an alternatives analysis workflow. Alternatives analysis is at the top and has three separate arrows pointing down to boxes labeled simulation modeling steps, alternatives analyzed/sequencing, and reporting results comparative analysis.
 
Figure 21. Flowchart. Alternatives analysis workflow.

Key considerations for the alternative analysis step include the following:

In some occasions, the simulation analysis may not be sensitive enough to distinguish between subtle nuances between similar alternatives. For example, simulation models cannot really distinguish between operational characteristics of two alternatives featuring a difference of 20 ft in the length of a storage lane. Similarly, simulation tools may not adequately analyze the differences between other design features such as shoulder widths, lane widths, and taper rates of acceleration and deceleration lanes. It is important that parameters are identified upfront about the realistic capabilities of what the simulation can and cannot be used to evaluate.

ALTERNATIVE ANALYSIS REQUIREMENTS

The requirements for an alternatives analysis depend on the type of project. The focus of this report is interstate freeway improvement projects. The level-of-effort case studies presented in chapter 1 include two projects requiring FHWA approval for interstate access requests. One project was a national highway system freeway (TH 100) that does not require the same FHWA approval as an interstate project, and the other was a comprehensive system plan (I‑5 in San Diego, CA). The national highway system project mostly followed the interstate access request process.

The interstate access request process is fairly strict and requires a number of analyses, including the existing conditions, year opening, and design year. Depending on the sequence of construction and project phasing, additional interim year analyses may be required. This workflow is illustrated in figure 22. The interstate access request process is described in FHWA’s Interstate System Access Informational Guide.(10)

This illustration highlights the interstate access analysis requirements. There are four boxes from left to right labeled existing (calibration), year opening analysis (baseline and alternatives), interim year analysis (baseline and alternatives), and design year analysis (baseline and alternatives).
 
Figure 22. Illustration. Interstate access analysis requirements—analysis years.

Baseline Condition

In all of types of projects, it is important to have a baseline for comparison. The baseline condition generally reflects some of the current planned geometric and operational conditions with future year traffic. When future planned/approved projects are coded into the model, the baseline condition is representative of the no-build condition.

Depending on the complexity of the network and of the alternatives analyzed, it is beneficial to use the ODME process (described in chapter 5) to analyze future alternatives that have interchange or ramp reconfigurations. Constructing O‑D matrices by hand (to reflect the new ramp termini or interchange reconfiguration) introduces the potential for errors, and, depending on the size of the system, it can be labor intensive.

SIMULATION OF ALTERNATIVES

The actual simulation of alternatives must follow the same principles and guidelines for base development and error checking. However, there are some specific considerations for the alternatives analysis that are captured in this section as follows:

SEQUENCING OF ANALYSIS

Managing the number of alternatives to be simulated in an alternatives analysis must be given careful consideration. If there are too few analyses, the process will have little credibility. Conversely, if there are too many analyses, the process can become bogged down with information overload, and it will become costly.

Simulation analysis is a powerful tool that allows for an effective and comprehensive analysis of an alternative. However, depending on the size and complexity of the model, it can also be a costly resource to deploy. It is essential to strike a balance between using the modeling process correctly and using the modeling process too much.

There are three different sequencing approaches, which are discussed in the following subsections. This is an illustrative exercise only and is provided to demonstrate how planning the alternatives analysis can be accomplished, potentially in a streamlined fashion. The disclaimer on this exercise is that there is no “one size fits all” solution; each project must consider the requirements and the best approach to satisfy the project. These examples demonstrate alternative methods that can reduce the amount of effort expended and improve the effectiveness of the analysis.

Incomplete Analysis

At times, particularly when resources are limited, there is a strong desire to skip the entire seven-step simulation process and go straight to the analysis of the alternative that is thought to be the best. This approach is incomplete and unacceptable for many reasons. For example, bypassing the model calibration step works against the credibility of the analysis. Also, the comparison between the baseline results and the alternative is most useful when focused on the differences in performance measures values rather than the absolute values.

An example analysis workflow of incomplete analysis is illustrated in figure 23. In this extreme case, the model calibration is not conducted, there is no baseline (no-build) for comparison, and there is only one alternative analyzed.

This illustration shows an example analysis workflow of incomplete analysis. This illustration shows two areas labeled “Interim Year Analyses” and “Design Year Analyses” separated by a vertical dashed line. The figure shows that only alternative 1 was being considered in both areas.
 
Figure 23. Illustration. Incomplete alternatives analysis approach.

 

Excessive Alternatives Analysis

The opposite of an incomplete alternatives analysis approach is one that is too excessive. In dealing with multiple stakeholders, there is a temptation to satisfy every request to analyze solutions in the microsimulation tool. At the beginning of the process, this may seem reasonable; however, as controversy occurs on a project and more ideas begin to be voiced, there is pressure to simulate more and more ideas.

One advantage of simulation is that when an alternative is modeled, the results provide useful information as to how to refine a design concept. If the initial/core set of alternatives is not managed to a reasonable number, then it is possible that the proliferation of subalternatives can add another layer of excessive alternatives analysis.

The last consideration is how the analysis years are sequenced. If the design year is the decisionmaking timeframe on the final solution, then it is advantageous to start with that timeframe to make a decision, reduce the number of alternatives, and then focus the analysis on the interim years on fewer alternatives. If not and the analysis proceeds in a linear fashion from existing to year opening to interim and then design year, then the modeler becomes obligated to carry forward all the alternatives (and sub-alternatives) forward without making a decision until all the analyses are completed.

A diagram of the number of alternatives and sub-alternatives that could easily occur as a result of the aforementioned issues is illustrated in figure 24.

This illustration shows the number of alternatives and sub-alternatives that could be investigated. The layout is the same as in figure 23. There are two areas labeled “Interim Year Analyses” and “Design Year Analyses” separated by a vertical dashed line. However, unlike figure 23, this figure shows excessive alternative analysis starting with a baseline model and then modeling alternatives 1 through N with three versions of each alternative for both the interim year analyses area and the design year analyses area.
 
Figure 24. Illustration. Excessive alternative analysis.

 

Streamlined Sequencing

There are many ways to approach an alternatives analysis that can satisfy the requirements and be time effective. Some simple streamlining techniques include the following:

An example is shown in figure 25. In addition to the simulation time, it is important to consider the stakeholders that must absorb the model information and make a decision. These techniques can be effective in managing stakeholders’ needs, requests, and expectations.

This illustration shows an example of the streamlined alternatives analysis method. The layout is the same as in figure 23 and figure 24. There are two areas labeled “Interim Year Analyses” and “Design Year Analyses” separated by a vertical dashed line. Unlike figure 23 and figure 24, in this figure, the design year analyses comes before interim year analyses. In this analysis method, the number of analyses needed to be performed is first reduced by pre-screeing the alternatives with less intensive analytical techniques. Unsuitable alternatives are eliminated before proceeding to full blown anlaysis. The illustration then shows only three alternatives that need to be analysied in the interim year analyses (baseline, alternative 2b, and alternative 4b).
 
Figure 25. Illustration. Streamlined alternatives analysis method.

Sample Relative Level-of-Effort Approaches

The previous sample alternative analysis workflows are generic cases to illustrate the issues with too little analysis, too much analysis, and a modified approach. Assigning a value of 100 h for each analysis box in the preceding figures, the results depicted in figure 26 make a strong argument for planning a streamlined analysis before the modeling work is to begin.

This bar graph shows an effort comparison for the different alternative analysis approaches. Hours of effort are on the y-axis, and type of analysis is on the x-axis, which include incomplete analysis, excessive analysis, and streamlined analysis. The bar graph shows approximately 200 h for incomplete analysis, 3,200 h for excessive analysis, and 1,400 h for streamlined analysis.
 
Figure 26. Graph. Comparison of the effort for different alternative analysis approaches.

EXAMPLES ON REPORTING RESULTS OF THE COMPARATIVE ANALYSIS

Tabular Methods

Simulation models produce excellent MOEs that help provide a quantitative assessment of alternatives. The models have the capacity to produce a lot of data. Reducing these data down to a few core tables of essential information is needed for an effective decisionmaking process. Table 19, table 20, and figure 27 are examples of data summaries. The color code in figure 27 indicates the following speeds:

Table 19. Sample MOE summary table.(7)

Analysis Segment Design Year
2005 2015
(Alternative A)
2015
(Alternative B)
2025
Speed
(mi/h)
Density
(vehicles/
lane/mi)
Speed
(mi/h)
Density
(vehicles/
lane/mi)
Speed
(mi/h)
Density
(vehicles/
lane/mi)
Speed
(mi/h)
Density
(vehicles/
lane/mi)

I-94 merge to High Ridge exit

64 (64) 11 (6) 32 (64) 34 (9) 64 (64) 15 (9) 64 (64) 16 (11)

High Ridge exit to High Ridge entrance

64 (64) 10 (4) 7 (64) 111 (9) 63 (64) 14 (6) 63 (64) 13 (7)

High Ridge entrance to
I-94 diverge

58 (63) 12 (4) 8 (62) 108 (7) 61 (64) 15 (6) 61 (63) 17 (7)
Note: Table values are listed as XX (YY), where XX represents the morning peak average, and YY represents the afternoon average. Bolded cells represent where the average speed dropped below 30 mi/h for any peak period for that alternative.

 

Table 20. Delay comparison between two scenarios in vehicle-hours—I 210 VISSIM simulation.

Direction Scenario 13 Scenario 14 Difference Percent Difference
Westbound 14,109 10,757 -3,351 -23.8 percent
Eastbound 971 920 -51 -5.3 percent
Total 15,080 11,677 -3,403 -22.6 percent

 

This graph shows a speed diagram for the I-210 VISSIM simulation. The simulation time period is on the y-axis starting at 15:00 and ending at 18:55 in 5-min time steps. The x-axis lists 21 different points in the simulation network (Fair Oaks 1, Marengo, Lake 2, Hill 1, Allen San Gabriel, Michillinda, Baldwin, Santa Anita 2, Huntington 1, Myrtle Ave, Mountain, Buena Vista, Irwindale, Vernon, Azusa 2, Citrus 2, Sunflower Ave, Lone Hill Ave, and Foothill BL). Numerical speed values at each location are given at the specific time, and each box is color coded from green to yellow to red as the speeds drop below 65 mi/h. The color coded graph is a quick way to see where the bottle necks and congestion are occurring in the network and at what time the onset is and when they dissipate.
 
Figure 27. Graph. Speed diagram for I-210 VISSIM simulation.

 

Graphical Techniques

Graphical techniques are another way of representing the simulation analysis results in an effective manner. When developing a graphical representation of data, it is important to clearly explain the criteria for any shading or coloring used to highlight where there may be issues with the depicted operations. Furthermore, it is always good practice to accompany graphics with a corresponding narrative explaining the context, findings, and recommendations. Figure 28 through figure 30 include examples of graphics that can be used.

This illustration shows a sample comparison of project alternatives. It shows the various design alternatives at a particular freeway section for 2005, 2015(a), 2015(b), and 2025. It shows the design for 2005 fails in the year 2015 (shown in the 2015 (a) design option). An alternative design, shown as 2015 (b), adds a 3,550-ft weaving lane between the high ridge on-ramp and the I-94 off-ramp adding capacity, and now this design passes. The same design is then compared to 2025 predicted traffic, and the design still passes.
 
Figure 28. Illustration. Sample comparison of project alternatives using schematic drawing.(7)

 

This bar graph shows the delay by segment for two scenarios (scenarios 13 and 14). Delay in vehicle-hours is on the y-axis, and eight segments are on the x-axis. The x-axis segments and delays for scenarios 13 and 14, respectively, are SR-57 to Azusa on (2,350 and 2,400), Azusa on to I-605 off (2,000 and 1,500), I-605 off to Santa Anita on (4,500 and  4,100), Santa Anita on to Baldwin on (1,200 and 1,300), Baldwin on to Rosemead on (1,000 and 450), Rosemead on to Altadena on (1,900 and 700), Altadena on to Lake on (900 and 150), and Lake on to SR-134 (200 and 50).
 
Figure 29. Graph. Delay by segment for two scenarios in vehicle-hours—I-210 VISSIM simulation.

 

This bar graph shows the delay by segment for all scenarios (base and scenarios 1, 3, 5, 7, and 9). Delay in vehicle-hours is on the y-axis, and eight segments are on the x-axis. The x-axis segments and delays for base and scenarios 1, 3, 5, 7, and 9, respectively, are SR-57 to Azusa on (25,000; 2,250; 2,000; 1,800; 1,850; and 1,750), Azusa on to I-605 off (1,250; 1,200; 1,000; 750; 700; and 450), I-605 off to Santa Anita on (5,100; 5,075; 5,050; 4,000; 3,800; and 1,750), Santa Anita on to Baldwin on (1,000; 970; 965; 1,400; 1,250; and 1,100), Baldwin on to Rosemead on (1,025; 1,000; 1,050; 1,300; 1,200; and 800), Rosemead on to Altadena on (1,025; 1,025; 1,000; 1,050; 680; and 2,050), Altadena on to Lake on (650; 980; 985; 850; 100; and 1,250), and Lake on to SR-134 (300; 325; 400; 450; 150; and 400).
 
Figure 30. Graph. Delay by segment for all scenarios in vehicle-hours—I-210 VISSIM simulation.

 

Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101