CHAPTER 5

Summary of Findings

Synthesis Objectives, Scope, and Methodology

The objective of the synthesis was to review and document DOT practices regarding the use of operational traffic simulation models. The synthesis scope included various topics related to operational traffic simulation models, such as extent of use, typical applications, software, modeling processes, data sources, and skill set development (e.g., training).

Methods used to achieve the synthesis objectives included a literature review, survey, and follow-up interviews. Various sources, such as guidance documents (general and DOT-specific), research reports, journal articles, and other resources were reviewed and synthesized. An online survey questionnaire was distributed to all 50 state DOTs and the District of Columbia DOT. Survey responses were received from 49 DOTs, a response rate of 96%. Case examples for six DOTs (those of Colorado, Indiana, South Carolina, Texas, Virginia, and Washington) were developed through follow-up interviews.

Summary of Key Findings by Topic

Key findings from the synthesis based on the literature review, survey results, and follow-up interviews are described in the following sections, which are organized by topic.

Key Synthesis Findings for Extent of Use and Applications for Operational Traffic Simulation Models

• All 49 state DOTs that responded to the survey use operational traffic simulation models.
• Twenty-seven responding DOTs apply operational traffic simulation models for 25 or fewer projects annually.
• The need for the use of operational traffic simulation models is often determined on a project-specific basis. DOTs typically reserve the use of operational traffic simulation models for more complex situations that cannot be adequately analyzed using other tools. Eighteen responding DOTs require approval for the use of operational traffic simulation models on each project. In some cases, scoping meetings at the beginning of the project provide an opportunity to discuss the use of simulation as the appropriate tool.
• At least 35 responding DOTs use operational traffic simulation models to some extent for each of these applications: signal retiming analyses, freeway design alternatives analyses, arterial design alternatives analyses, TIAs, and design visualization and communication. Of these applications, operational traffic simulation models are most often used for freeway design alternative analyses, mixed design alternative analyses, and signal retiming analyses. Seven responding DOTs use operational traffic simulation models for evacuation route analyses. Twenty-three responding DOTs use operational traffic simulation models most frequently for freeways.

• Examples of specific applications of operational traffic simulation models include, but are not limited to CAVs; environmental and planning stages (e.g., development of environmental impact statements and modeling for noise, emissions, and fuel consumption); complex interchanges and interchanges with high arterial flow; evaluation of alternatives for intersection layout, lane configuration, storage lengths, and signal operations; performance of traffic analysis for IARs and Interchange Modification Reports; maintenance of traffic analysis for work zone traffic control scenarios; truck climbing lanes; roundabouts; oversaturated conditions; multimodal contexts (e.g., pedestrian signal timing, light rail); and other situations that simpler tools do not handle appropriately.
• The specialized application used by the highest number of DOTs is signal optimization (38 responding DOTs); the specialized application used by the lowest number of DOTs is CAV analysis (three responding DOTs). Thirteen responding DOTs indicated the use of hybrid/MRM.

Key Synthesis Findings for Guidance and Other Resources for Operational Traffic Simulation Models

• Resources available from FHWA cover topics such as tool selection, modeling guidance, MRM, and use of simulation for CAV applications. Modeling guidance from FHWA is available through the 2004 TAT Volume III (Dowling et al. 2004), which was updated in 2019. The 2019 update includes more detailed guidance on data collection analysis, model calibration, and alternatives analysis and suggests the use of cluster analysis for the identification of travel conditions.
• Another general resource (under development by the TRB Standing Committee on Traffic Simulation) is the TSSM, which provides guidance on topics such as modeling resolutions, scenario development, and modeling processes and includes case studies with commentary on how to apply the guidance (List 2021).
• The literature review identified 21 state DOTs as having one or more published guidelines or other documents detailing traffic simulation. These resources cover topics such as tool selection, project scoping and management, model calibration, and review checklists. Regarding types of resources, responding DOTs have most often developed guidance documents, followed by suggested calibration parameters and procedures for model development and review.
• The availability of documentation regarding development of procedures for maintenance and archiving of data or performance studies on the benefits of—or return on investment for—the use of operational traffic simulation models is infrequent.
• Responding DOTs most often use state-specific guidance (19 responding DOTs), followed by ad hoc project-based decisions (15 responding DOTs) for the calibration of operational traffic simulation models. Six responding DOTs primarily use the 2004 version of the TAT; four responding DOTs primarily employ the 2019 version of the TAT. Five responding DOTs (those of Missouri, Nevada, New Hampshire, New Mexico, and Oklahoma) use resources from other states (those of Florida, Oregon, Utah, Virginia, Washington, and Wisconsin).
• There is general movement toward the use of the 2019 TAT guidelines; however, data availability for the cluster analysis is viewed as a significant challenge to broader implementation.
• Twenty responding DOTs have a process for documenting deviations from modeling guidance on specific projects.

Key Synthesis Findings for Modeling Practices for Operational Traffic Simulation

• Initial scoping meetings are often conducted at the beginning of a project to discuss the project scope and modeling approach.

• Although nearly all (46) responding DOTs reuse or adapt previously developed operational traffic simulation models to some extent, this practice is only used with a frequency of “usually” or “always” by 13 responding DOTs, and simulation models are typically project-specific.
• The factors believed to be most important for operational traffic simulation models are (1) justifying the need for simulation analysis, (2) scheduling constraints, (3) level of modeling effort, and (4) data availability.
• The most frequently used simulation modeling resolution is microscopic, followed by macroscopic. Approximately one-fourth of the responding DOTs use hybrid/MRM, and DTA is sometimes used on larger projects. Although microscopic resolution is the type most frequently discussed in the literature, example applications for other modeling resolutions found in the literature include signal optimization and large-scale evacuation planning (macroscopic); evaluating land development impacts (mesoscopic); and dynamic reversible lane systems and work zones (MRM).
• The survey results indicate that the following calibration metrics are used, on average, at least sometimes by responding DOTs: volumes, visual inspection, travel times, queue length, intersection LOS, and freeway density. The calibration metrics used most often by responding DOTs are volumes and visual inspection.
• Most state DOTs with published guidance use variations of the 2004 TAT calibration targets, with slight modifications that frequently reference peer states.
• Example aspects of simulation modeling that are studied in the literature include calibration tools and processes and car-following models.

Key Synthesis Findings for Software and Data for Operational Traffic Simulation Models

• DOTs use various software packages for operational traffic simulation modeling, and software updates are typically installed by DOTs on a regular basis. The software packages used by the highest number of responding DOTs are SimTraffic and Vissim.
• Consultants often play an active role in the selection of software tools, and the software deemed most suitable for the specific application is chosen. In some cases, RFPs with traffic simulation requirements will specify the software that will be used.
• Comparisons in the literature of software packages have found that each package has different strengths.
• Twenty-one responding DOTs adopt new versions of operational traffic simulation software at least every two years; the frequency of adopting new software versions can vary depending on the software, schedule, and costs.
• The survey results indicate that the most frequently used data sources for operational traffic simulation models are traffic counts, field observations, aerial imagery, and online map data; the least frequently used data sources are drone footage and transit data. Other data sources include, but are not limited to, travel times, speeds, historical data for comparison and validation, projection data for forecasting from the travel demand model count data, subscriptions to vendor data, and data from local agencies and consultants.
• Twenty-two responding DOTs fuse data from different sources.

Key Synthesis Findings for Review and Documentation of Operational Traffic Simulation Models

• Eleven DOTs include traffic simulation model review checklists in their guidance. These checklists vary in length, specificity, and content and cover topics such as network coding; driving behavior; performance measure outputs; calibration thresholds and parameters; and documentation.

• For review processes, responding DOTs most often use reviews of performance measures that are reported from the model, model input data, and animation.
• A wide range of MOEs is used for both uninterrupted flow and interrupted flow, with speed, travel time, and density/LOS the most frequently used for uninterrupted flow, and delay/LOS and queue length the most frequently used for interrupted flow. Examples of other methods and measures used to assess model performance include bottlenecks, throughput visual cues, and engineering judgment.
• The deliverables required by the highest number of responding DOTs are (1) a summary of MOEs, (2) simulation model files, (3) a technical memorandum of results, (4) volume diagrams, and (5) a methods and assumptions document (typically submitted in the early project stages).
• Animation is sometimes employed for presenting project results and conveying the design and performance of alternatives to various audiences, including in public information meetings, workshops, and presentations.
• Twenty-two responding DOTs require all simulation results to be reported with a minimum of 10 simulation seeds.
• Peer reviews are sometimes performed for larger consultant projects, but their use is infrequent.
• Post-construction verification of operational traffic simulation models is infrequent (indicated by only two responding DOTs).

Key Synthesis Findings for DOT Training, Staffing, and Management for Operational Traffic Simulation Models

• Training is typically provided both online and in person by software vendors; in some cases, in-house training is given. Eleven responding DOTs have developed training materials for operational traffic simulation models.
• Most simulation models are developed by consultants, with review and oversight from DOT staff. Twenty-seven responding DOTs use consultants for more than 75% of their operational traffic simulation models, and no responding DOTs perform all simulation modeling in-house.
• Some local agencies conduct operational traffic simulation modeling, often using a consultant.
• Internal DOT staff members who implement the development and review of operational traffic simulation models are placed in various divisions, mostly in the areas of operations and design.

Key Synthesis Findings for Challenges and Opportunities for Operational Traffic Simulation Models

• Challenges faced by DOTs in the use of operational traffic simulation models include, but are not limited to, (1) data availability; (2) cost; (3) staffing limitations; (4) guidance and training needs; (5) difficulty in obtaining and validating data from different sources—especially during the COVID-19 pandemic; (6) diversity of software tools and versions that are used for different types of projects and analyses; (7) archiving and reusing models due to software version changes and project duration; and (8) demonstrating the return on investment for new approaches.
• Examples of ways that DOTs are working toward enhancing their use of operational traffic simulation modeling include (1) encouraging consistency in modeling practices; (2) providing more direction on modeling implementation; (3) expanding the use of simulation for multimodal applications (e.g., bicycle, pedestrian, transit, rideshare); (4) applying simulation for safety analysis; (5) continuing the transition to the 2019 TAT guidelines; (6) using mesoscopic modeling for TSMO alternatives and regional multimodal mobility programs; (7) adding other software platforms; and (8) increasing the use of MRM.

• State DOTs are interested in learning about various aspects of other state DOTs’ experiences with operational traffic simulation models, such as how they approach operational traffic simulation modeling programmatically, their successes, and their struggles.

Suggestions for Future Research

This synthesis has identified some gaps in existing knowledge. Additional research is needed to enhance practices for the use of operational traffic simulation models in the United States. Suggestions for future research include the following:

• Creation of guidance (possibly including case examples) demonstrating the impact of new calibration and data collection recommendations (e.g., comparing the 2004 and 2019 TAT microsimulation guidance regarding factors such as modeling effort and data requirements) to help demonstrate how the additional effort can lead to greater accuracy and better-informed decisions
• Further development and publication of the TSSM
• Development of national guidance on operational traffic simulation models that could be considered for adoption and use
• Development of additional guidance regarding procedures for the maintenance and archiving of data
• Research on the benefits of—or return on investment for—the use of operational traffic simulation models
• Development of guidance and case studies for the use of operational traffic simulation models for safety analyses
• Development of case examples on the use of operational traffic simulation models by MPOs and local agencies
• Development of case examples showing post-construction validation of operational traffic simulation models
• Development and broad dissemination of reviewer training materials for operational traffic simulation models
• Creation of guidance for using artificial intelligence and machine learning to automate the calibration of operational traffic simulation models
• Guidance on how to fuse data sources and leverage existing data for operational traffic simulation models, especially in cases where data availability is limited
• Peer exchange to help state DOTs to share information regarding practices for operational traffic simulation models
• Development of a guide for best practices and reliable data for incorporating nonmotorized road users into operational traffic simulation models