Traffic Analysis Toolbox Volume III: Guidelines for Applying Traffic Microsimulation Modeling Software 2019 Update to the 2004 Version

Foreword Regarding the 2019 Update

The purpose of the Guidelines for Applying Traffic Microsimulation Modeling Software is to provide a recommended process for using traffic microsimulation software in transportation operations analyses. The guidelines provide the reader with a seven-step process that begins with planning a microsimulation analysis for the project and ends with the final project report. The process is generic, in that it is independent of the specific software tool used in the analysis. In fact, the first step in the process involves picking the appropriate tool for the job at hand.

The 2004 version of this guide created a set of standard practices for effective application of microsimulation tools. The guide addressed the most critical need of the day, namely to provide a systematic process that began with project scoping, tool selection, data collection and input preparation, model testing and calibration, alternatives analysis, and project documentation.

When the 2019 update effort was initiated, the Federal Highway Administration (FHWA) engaged more than two dozen experts from around the world to provide their views on the current state of the microsimulation practice and how this document might be best improved. These experts responded to a structured set of questions to identify the most pressing current needs and to provide prioritization among competing needs.

Taken as a whole, the panel responses reinforced the concept of an increasingly complex and demanding landscape for microsimulation analyses. The experts noted a trend towards more complex applications of microsimulation tools on larger networks, the evaluation of increasingly complex alternatives, and the increasing importance of microsimulation studies in influencing both operational and investment decisionmaking.

Another crucial consensus observation among the experts was the rise of more comprehensive and detailed data sources available to the microsimulation analyst. Pre-2004, data were relatively scarce, and simulation analysts were often forced to rely on a set of vehicle count data limited to some larger facilities within the system. In this era, notions of microsimulation calibration were adapted from travel demand forecasting methods and targeted the accurate reflection of average vehicle counts at detector locations within the network. The panel pointed out that new forms of time-dynamic data were now available but often ignored or ineffectively incorporated within simulation studies.

Identified High-Priority Areas

The expert panel prioritization exercise identified four high-priority areas for the 2019 update:

• Fully Integrate Time-Dynamic Representation of Congestion. The panel indicated that current practice in microsimulation was too focused on fixed or average demand patterns and that congestion development and dissipation over time were often poorly reflected in tool application. This issue resulted in unrealistic analyses that failed to reflect accurately the performance of alternatives under congestion conditions. An identified need for an updated version of this document was to provide more emphasis on time-dynamic data in simulation development, calibration, and alternatives analysis.
• Require Better Representation of Recurrent and Non-Recurrent Conditions. Simply put, the panel indicated that many simulation analyses used to justify investments failed to represent adequately incidents, weather, and variations in travel demand. In some cases, the impact of particular alternatives (traveler information, incident management, road-weather applications) were not accounted for within the simulation study, and suffered relative to other alternatives (e.g., geometric improvements). The identified need was to provide guidance on how to integrate a variety of data sources to create meaningful and realistic models for these travel conditions. This guidance should also include the tailoring of driver behavioral model parameters (e.g., gap acceptance and car-following distance) to the specific conditions being modeled.
• Remove Subjective Calibration Criteria. Some elements of guidance presented as best-practice examples within the guidebook included subjective criteria (e.g., the phrase "to the analyst's satisfaction"). This resulted in two issues. First, these criteria often led to contention among stakeholders since "satisfaction" is highly subjective. Second, in the absence of clear FHWA guidance on calibration, examples used as illustrations within the document were misinterpreted as FHWA guidance. The identified need was to create and document FHWA guidelines on a data-driven calibration process based on statistically-derived and objective criteria.
• Emphasize Accurate Bottleneck Modeling. The accurate representation of the bottleneck location, onset time, and duration were identified as a critical aspect of modeling congested systems. The panel recommended more emphasis on modeling and calibrating bottleneck dynamics.

Goals

From these recommendations, FHWA focused available resources for the 2019 update on the following goals:

• Encourage comprehensive experimental design based a range of varying travel conditions, rather than a normative "average" day.
• Focus calibration and alternatives analysis on the representation of time-dynamic system performance measures including bottleneck formation and dissipation.
• Eliminate all subjective criteria, to be replaced with criteria that are statistically valid and derived from observed data.
• Develop a calibration process that is data-driven, repeatable, and potentially automatable.

It is hoped that these guidelines will assist the transportation community in moving away from outdated practices that were developed in the relatively data-poor past dependent on the analysis of averages or the assumption that unrealistic "normal" conditions always prevail, and towards conducting data-driven, statistically-valid objective analyses.

Format of the Update

In order to meet the goals of the update, a complete rewrite of the document was neither feasible nor desirable. Instead, the 2019 update retains the general structure and intent of the main sections of the 2004 guidebook [1], with updated materials throughout the document. Significantly enhanced and detailed technical guidance has been included in chapters dealing with data collection and analysis (Chapter 2), model calibration (Chapter 5), and alternatives analysis (Chapter 6). A more complex corridor-based example problem is used to illustrate application of the updated guidance. A series of technical appendices from the 2004 document on ancillary topics have been either incorporated into the body of the document or removed. References have been updated, particularly with respect to the larger body of current and projected Traffic Analysis Tools volumes. Finally, a complete end-to-end case study using a large microsimulation model for a hypothetical work zone alternatives analysis is included in an appendix. This case study illustrates the application of the detailed guidance on data collection and analysis, calibration and alternatives analysis.

Aside from this section, for the purposes of clarity, the 2019 update is presented as a complete document without reference to original source (2004 or 2019). That is, the specific passages and updates introduced into this version are not identified separately from materials which appeared in the original version.