Traffic Analysis Toolbox Volume III: Guidelines for Applying Traffic Microsimulation Modeling Software 2019 Update to the 2004 Version
Microsimulation is the modeling of individual vehicle movements on a second or sub-second basis for the purpose of assessing the traffic performance of highway and street systems, transit, and pedestrians. Microsimulation analyses are increasingly visible and important — fostered both by the continued evolution of microsimulation software capability and increasing application within transportation engineering and planning practices. These guidelines provide practitioners with guidance on the appropriate application of microsimulation models to traffic analysis problems, with an overarching focus on existing and future alternatives analysis.
The use of these guidelines will aid in the consistent and reproducible application of microsimulation models. They are also intended to support further the accuracy and credibility of analyses using these tools. As a result, practitioners and decision makers will be equipped to make informed decisions that will account for current and evolving technology. It is hoped that these guidelines will assist the transportation community in moving away from outdated practices that were developed in the relatively data-poor past dependent on the analysis of averages or the assumption that unrealistic "normal" conditions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. Depending on the project-specific purpose, need, and scope, elements of the process described in these guidelines may be enhanced or adapted to support the analyst and the project team. It is strongly recommended that the respective stakeholders and partners be engaged prior to and throughout the application of any microsimulation model. This further supports the credibility of the results, recommendations, and conclusions, and minimizes the potential for unnecessary or unanticipated tasks.
Guiding Principles of Microsimulation
Microsimulation can provide the analyst with valuable information on the performance of the existing transportation system and potential improvements. However, microsimulation can also be a time-consuming and resource-intensive activity. The key to planning and conducting an effective (and cost-effective) microsimulation analysis revolves around a set of guiding principles:
Overview of the Microsimulation Analytical Process
The overall process for developing and applying a microsimulation model to a specific transportation analysis problem consists of seven major tasks. Each task is summarized below and described in more detail in subsequent chapters. A flow chart, complementing the overall process, is presented in Figure 1. This figure is intended to be a quick reference that will be traceable throughout the document.
Chapter 1 addresses the management, scope, and organization of microsimulation analyses. In this section, the guidebook describes key issues for the management of a microsimulation study. This includes defining the project purpose, identifying the project influence area and analytic time period(s), characterizing the alternatives to be evaluated, selecting performance measures, developing a general technical approach (including tool selection) and estimating staff time and other costs.
Chapter 2 discusses the steps necessary to collect and prepare input data for use in microsimulation models. This task involves the collection and preparation of all of the data necessary for the microsimulation analysis. Microsimulation models require extensive input data, including: roadway geometry, controls, travel demand, and calibration data, among others. Chapter 2 also includes guidance on analyzing contemporaneous data on traffic counts (demand), weather, incidents, and key performance measures (e.g., travel time) to characterize a set of operational and travel conditions.
Chapter 3 discusses the coding of input data to create a base model. The goal of base model development is a model that is verifiable, reproducible, and accurate. It is a complex and time-consuming task with steps that are specific to the software used to perform the microsimulation analysis. The details of model development are best covered in software-specific user's guides. For this reason, the development process may vary. This report provides a general outline of the model development task.
Chapter 4 presents error-checking methods. The error-checking task is necessary to identify and correct model coding errors so that they do not interfere with the model calibration task. Coding errors can distort the model calibration process and cause the analyst to adopt incorrect values for the calibration parameters. Error checking involves various tests of the coded network and the demand data to identify input coding errors.
Chapter 5 provides guidance on a systematic calibration of the error-free base model developed in Step 4 to reproduce observed throughput and other performance measures for distinct travel conditions characterized in Step 2. First, individual bottleneck capacities are obtained from observed data. The analyst then performs systematic adjustment of relevant tool parameters to reproduce this throughput rate in model outputs. Second, aggregate demand crossing key strategic internal boundaries (screenlines) are characterized from observed data and are reproduced through alteration of travel demand flow rates. Third, a statistical analysis of observed data creates a series of time-dynamic ranges for key performance measures (e.g., bottleneck throughput and travel times) in each of the travel conditions identified in Chapter 2. One model variant for each travel condition is then calibrated to fall within these time dynamic statistical envelopes, passing four distinct calibration criteria. When these criteria are met for all operational condition model variants, calibration is complete and alternatives analysis may begin.
Chapter 6 explains how to use microsimulation models in a well-designed experiment to support an analysis of alternative improvements proposed for the modeled transportation system. The set of calibrated microsimulation model variants corresponding to the set of travel conditions are run several times to test various project alternatives. The first step is to develop a baseline demand pattern, potentially for a future year. Then the various improvement alternatives are coded into the simulation model representing each travel condition. Randomness in simulation outputs is analyzed to determine the required number of runs to assess statistical validity satisfactorily when comparing the impacts of competing alternatives. The analyst then runs each model variant for the required number of replications for each alternative to generate the necessary output to generate the key performance measures. A statistical test is then performed to determine whether differences between two alternatives are statistically significant, i.e., to determine if the test meets the minimum criteria bounding the risk that the differences are only related to randomness in model outputs.
Chapter 7 provides guidance on the documentation of microsimulation model analysis. This task involves summarizing the analytical results in a final report and documenting the analytical approach in a technical document. This task may also include presentation of study results to technical supervisors, elected officials, and the general public. The final report presents the analytical results in a form that is readily understandable by the decision makers for the project. The effort involved in summarizing the results for the final report should not be underestimated, since microsimulation models produce a wealth of numerical output that should be tabulated and summarized.
Organization of This Document
United States Department of Transportation - Federal Highway Administration