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

CHAPTER 2.  TRANSPORTATION AGENCY AND PRACTITIONER EXPERIENCES

This chapter presents the major ideas and trends that were extracted from interviews conducted with transportation agency staff and consultants. A questionnaire was developed and sent out to targeted transportation agency representatives and consultants. Each questionnaire consisted of 30 questions spread across four categories: experience, building of the project’s framework, model calibration, and beyond calibration. The terms “interviews,” “surveys,” and “questionnaires” are used interchangeably throughout the section.

The recipients of the interview were given the option to either fill out the questionnaire in writing or provide input during a phone interview. In total, 12 completed surveys were received—10 by agency representatives and 2 by consultants. Table 6 contains a list of the interviewees.

Table 6. Interviewee list.

PROJECT EXPERIENCE

Analyses Conducted

The survey participants were involved in various traffic analysis studies in different roles. They conducted, managed, supervised, and provided support for analysis of construction projects, interchange operations studies, corridor studies, and planning studies. The surveyed individuals have ample experience in conducting/‌leading traffic and transportation studies involving simulation and/‌or traffic analysis tools.

Network Sizes Involved

The survey participants worked on a variety of network sizes ranging from a single intersection or interchange to several miles of interstate freeway, as well as a variety of analysis types including complex regional- and corridor-level analyses. While a typical project would include around 15 mi of freeway with crossing arterials, some projects involved freeway systems of more than 50 mi in addition to arterial network studies with several hundred intersections.

Project Type and Stage

Numerous projects, including local agency roadway operation improvement projects, safety projects, freeway modernization projects, and planning and corridor studies, were undertaken by the different surveyed agency representatives and consultants. Analysis included the entire process from early planning to final design stages.

Analysis Tools Used

Various analysis tools were used by the survey subjects. Those included deterministic tools as well as macroscopic and microscopic simulation analysis tools.

BUILDING OF THE ANALYSIS FRAMEWORK

Conducting the Analyses

Transportation agency representatives provided answers on who conducted the analyses, which is highlighted in figure 8. While some regularly resorted to consultants, others conducted in-house analyses (mix 1), and some used both in-house plus consultants (mix 2).

 
Figure 8. Graph. Agency responses of who conducted the analyses.

Developing RFPs

Some transportation agencies have in-house guidelines that are used to develop RFPs (such as FDOT’s Project Development and Environmental Studies).⁽³⁾ Others rely on RFPs for similar projects. Refinements to RFPs are made on a case-by-case basis depending on the unique nature, scope, type, size, and complexity of the project.

Identifying the Project Needs and Questions to Address

Respondents indicated that the needs and questions to be addressed by the analysis are based on field investigation and performance assessment. Studies are expected to address both existing and future needs. Any issues or challenges known in the study are recognized in the scope so they can be properly addressed in the analysis.

Scoping Projects

The interviewees were consistent in their indication that a planning and development team is formed within agencies for each project. The team consists of planning, traffic, and preliminary design staff. The team determines the initial scope and then works with a consultant to develop the refined scope of services.

The scope of work may be influenced by many factors such as funding availability, geographic homogeneity, influence of adjacent interchanges and intersections, or amount of time elapsed since the last study of any particular section. Study areas are typically extended beyond bottleneck extents so that the causes of congestion are reasonably replicated. Traffic patterns and the duration of congestion help determine the temporal extent of the analysis. Sufficient lead time prior to the beginning of congestion is included in the models.

Prior to specifying the scope of work, some agencies ask project managers to complete a scoping worksheet that describes the geographical limits, level of detail, any existing data or resources available for the project, and any previous models. Some agencies follow guides to develop their scoping documents.^(4,5)

Determining the Level of Effort Required for the Analyses

At most transportation agencies surveyed, a labor-hour estimate is prepared by the consultant and reviewed by the project manager and the agency’s program leader. A project schedule is developed, including anticipated documentation delivery dates and agency review times. The labor-hour estimate and project schedule are then submitted to the agency for review.

Analysis schedules are dependent on the availability of traffic counts, seasonal factors determining when to conduct the counts, availability of travel demand models (TDMs), funding constraints, and project phasing.

Consultants responding to the survey generally use a bottom-up approach to develop budget and schedule estimates by task. Half of the surveyed agencies and consultants (6 out of 12) do not use a guide to determine their budget and schedule.

Determining Data Requirements

Data requirements are based on the project’s goals and scope and depend on the analysis tool used. The data requirements also are influenced by the amount of time and resources available to complete the work. Most surveyed agencies and consultants indicated that meeting data requirements is relatively challenging due to resource availability in terms of both time and budget. It was acknowledged that there is a conflict between the need for comprehensive model calibration data and limited data collection budgets.

Special methods have been developed by agencies such as WisDOT to resolve data discrepancies and maximize the use of existing data. Some State agencies mentioned O-D data as being the most challenging because they does not exist in a form that is ideal for model calibration. Furthermore, survey participants reported that there are no guidelines on the temporal consistency of collected data or the geographic coverage of freeway and arterial data collection. Most interviewees agreed that issues with data comprehensiveness, temporal consistency, geographic coverage, and overall resources are among the key reasons for schedule and budget overruns.

Determining Performance Measures

Performance measures are usually identified by agencies with input from consultants. They depend on the project type and the study objectives. Measures are reported by corridor subsections, direction, and facility type (i.e., mainline, on-ramps, off-ramps, and local streets). Typical performance measures used in the analysis include the following:

• Density.
  
  
• Volume.
  
  
• Speed.
  
  
• Level of service (LOS).
  
  
• Delay.
  
  
• Travel time.
  
  
• Queue length.

Performance measures such as travel time reliability, emissions, fuel consumption, and safety were not mentioned by the survey participants.

Selecting Analysis Tools

Most often, the tool that is used in the analysis is identified by the agency usually based on its staff/project experience. At times, consultants are asked to suggest the appropriate tool(s), and the assessment is then based on the type of analysis (scope and goals of the project) and the resources available (available data, schedule, and budget). Macrosimulation techniques are used for high-level screening of alternatives; however, most of the interviewees mentioned that they use microsimulation, especially for the analysis of traffic operations in congested environments.

MODEL CALIBRATION

Model Calibration Guidelines

A total of 7 out of the 12 interviewees (2 consultants and 5 agency representatives) indicated that they follow the FHWA model calibration guidelines, including volumes II, III, and IV of the Traffic Analysis Toolbox.^(1,6,7) Some survey participants mentioned having encountered ambiguities and challenges while using the FHWA guidelines.

Ideas offered by the interviewees for guideline improvement include the following:

• Standards to limit the number of in-route removal of vehicles.
  
  
• Unreleased vehicles from zones.
  
  
• Logic check of demands between key O-D pairs.
  
  
• Path checks for key O-D pairs.
  
  
• A method to help select the typical day to simulate.

Several of the respondents developed their own set of policies and procedures. WisDOT has expanded its model calibration criteria and guidelines over time. ODOT is in the process of updating their Protocol for Vissim Simulation.⁽⁸⁾ WSDOT has its own document that was developed specifically for each project involving Federal funds or Federal facilities in order to guide the requirements and expectations regarding model calibration. Florida and Minnesota also use their own model calibration guidelines.

Model Output Versus Real-World Data

A comparison between observed data and modeled output was performed via model calibration and validation. The data used in the comparison included volume, speed, travel time, queue length, and location of bottlenecks. Sources of observed data comprised field traffic volume and speed counts, applicable information from existing studies, and historical traffic counts. Survey participants suggested that models must also provide a qualitative look and feel that matches the way traffic operates in the real world.

Determination of Appropriate Number of Model Runs

Some interviewees acknowledged the usage of statistical methods to determine the appropriate number of model runs. (See references 1 and 6–8). The use of statistical methods to determine the appropriate number of model runs has emerged as an area where additional guidance is needed. Chapter 6 of this report contains an example on how to determine the appropriate number of model runs using statistical methods.

Documentation of Calibration Results

Interviewees consistently responded that the model calibration effort is usually documented via a report/‌memorandum. Calibration documentation contains the calibration method, including criteria, measures, and existing conditions as well as a summary of the calibration results documenting how performance measures from the baseline model relate to measured real-world data. Also, any parameters that are modified based on calibration results for the existing condition are documented in the report and carried forward to future alternatives. The model calibration report is considered a major milestone in the model development process.

The following assumptions were made concerning the consistency in analysis:

• Microsimulation and externally developed forecasts:Traffic volumes and/or O-D flows from externally developed forecasts are often used as inputs to microsimulation models. These inputs usually come from TDMs, which have different underlying assumptions and procedures than simulation models. Linking refined microsimulation zones to TDM traffic analysis zones is used as a means to maintain consistency between the two types of models.
  
  
• Base models and future year models: After baseline model calibration is complete for existing conditions, the calibrated model parameters are carried forward to future year models. This helps maintain consistency between baseline and future year models.

Documentation on the Level of Effort Required for Calibration

While some agencies document the level of effort (number of person-hours required) for model calibration in the scope of work, other agencies do not track the level of effort associated with model calibration. Consultants track their time through timesheets. Model calibration activities are usually tracked separately by task.

ANALYSIS AND RESULTS

Development of Analysis Scenarios

Respondents indicated that analysis scenarios are developed in close coordination with project stakeholders. Scenarios are usually driven by the planning needs of the project and by the availability of resources. The types and numbers of scenarios to be modeled are usually dependent on the geometric design and operational solutions being considered.

Analysis of Different Operational Conditions

Most interviewees’ agencies conduct their analyses for the typical day. Some agencies use the 30th highest hour of the year to help establish morning and afternoon average day peak period analyses. It is not standard practice to model nonrecurring congestion (e.g., incidents, weather events, fluctuations in demand, and construction activities), and there are no model calibration criteria associated with these operational conditions.

Presentation of Analysis Results

Survey responses indicated that analysis results are presented through memoranda/‌reports, screen shots from the simulation, and video clips. Reports usually contain summary tables and charts for network measures, graphs, and maps for location-specific LOS. Reports also contain time-space diagrams for freeway corridor queues/‌congestion.

Decision Development from Analysis Results

Results of the analysis (including the comparison between performance measures for different analysis scenarios) provide useful information for decisionmaking on various projects and for prioritization and staging of programmed and planned projects. Interviewees stated that decisions are made based on a collaborative process involving the project development team, the management team, local jurisdictions, and stakeholders. In general, the charge of the analysts is to provide unbiased technical information for the clients/‌decisionmakers to review and take appropriate action.

Funding Level Specification

Specifying funding levels for the analysis is done on a project-by-project basis, and no general guidance is available. Microsimulation is recommended for facilities with significant congestion and/or operational problems, whereas simpler tools, such as deterministic methods and macrosimulation, are recommended for less complex projects. Factors influencing the project costs include the length of the corridor, presence/‌inclusion of parallel streets, existing level of congestion, time period of the simulation, routing assignment used, number and types of analysis scenarios desired, availability of existing data, and comprehensiveness and internal consistency of calibration data.

CONCLUSIONS

Through reviewing and summarizing agency and practitioner experience, this chapter highlighted current procedures and challenges in traffic simulation analyses. A number of key issues emerged from the survey findings where guidance is needed to assist in improving traffic analysis processes.