Vol3 Guidelines

Traffic Analysis Toolbox Volume III:
Guidelines for Applying Traffic

Microsimulation Modeling Software
PUBLICATION NO. FHWA-HRT-04-040 JULY 2004
Research, Development, and Technology

Turner-Fairbank Highway Research Center
6300 Georgetown Pike
McLean, VA 22101-2296
Foreword
Traffic simulation software has become increasingly more popular as a traffic analysis tool
used in transportation analyses. One reason for this increase in the use of simulation is the
need to model and analyze the operation of complex transportation systems under
congested conditions. Where some analytical techniques break down under these types of
conditions, simulation shines. However, despite the widespread use of traffic simulation
software, there are a variety of conflicting thoughts and practices on how simulation
should be used.
The purpose of the Guidelines for Applying Traffic Microsimulation Modeling Software is to
provide a recommended process for using traffic simulation software in transportation
analyses. The guidelines provide the reader with a seven-step process that begins with
project scope 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. It is hoped that these
guidelines will assist the transportation community in creating a more consistent process
in the use of traffic simulation software.
This document serves as Volume III in the Traffic Analysis Toolbox. Other volumes
currently in the toolbox include: Volume I: Traffic Analysis Tools Primer and Volume II:
Decision Support Methodology for Selecting Traffic Analysis Tools.
The intended audience for this report includes the simulation analyst, the reviewer of
simulation analyses, and the procurer of simulation services.
Jeffery A. Lindley, P.E.

Director
Office of Transportation Management
Notice
This document is disseminated under the sponsorship of the U.S. Department of Transportation
in the interest of information exchange. The U.S. Government assumes no liability for the use of
the information contained in this document.
The U.S. Government does not endorse products or manufacturers. Trademarks or

manufacturers’ names appear in this report only because they are considered essential to the
object of the document.
Quality Assurance Statement

The Federal Highway Administration (FHWA) provides high-quality information to serve
Government, industry, and the public in a manner that promotes public understanding.
Standards and policies are used to ensure and maximize the quality, objectivity, utility, and
integrity of its information. FHWA periodically reviews quality issues and adjusts its programs
and processes to ensure continuous quality improvement.
Technical Report Documentation Page
1. Report No. 2. Government Accession No. 3. Recipient’s Catalog No.
FHWA-HRT-04-040
4. Title and Subtitle 5. Report Date
Traffic Analysis Toolbox June 2004
Volume III: Guidelines for Applying Traffic Microsimulation 6. Performing Organization Code
Software
7. Author(s) 8. Performing Organization Report No.
Richard Dowling, Alexander Skabardonis, Vassili Alexiadis
9. Performing Organization Name and Address 10. Work Unit No.
Dowling Associates, Inc.
180 Grand Avenue, Suite 250 11. Contract or Grant No.
Oakland, CA 94612 DTFH61-01-C-00181
12. Sponsoring Agency Name and Address 13. Type of Report and Period Covered
Office of Operations Final Report, May 2002 – August 2003
Federal Highway Administration 14. Sponsoring Agency Code
400 7th Street, S.W.
Washington, D.C. 20590
15. Supplementary Notes
FHWA COTR: John Halkias, Office of Transportation Management
16. Abstract
This report describes a process and acts as guidelines for the recommended use of traffic microsimulation
software in transportation analyses. The seven-step process presented in these guidelines highlights the aspects
of microsimulation analysis from project start to project completion. The seven steps in the process include:
1) scope project, 2) data collection, 3) base model development, 4) error checking, 5) compare model MOEs to
field data (and adjust model parameters), 6) alternatives analysis, and 7) final report. Each step is described in
detail and an example problem applying the process is carried through the entire document.
The appendices to report contain detailed in information covering areas such as: a) traffic microsimulation
fundamentals, b) confidence intervals, c) estimation of simulation initialization period, d) simple search
algorithms for calibration, e) hypothesis testing of alternatives, and f) demand constraints.
This is the third volume in a series of volumes in the Traffic Analysis Toolbox. The other volumes currently in
the Traffic Analysis Toolbox are:
Volume I: Traffic Analysis Tools Primer (FHWA-HRT-04-038)
Volume II: Decision Support Methodology for Selecting Traffic Analysis Tools (FHWA-HRT-04-039)
17. Key Words 18. Distribution Statement
Traffic simulation, traffic analysis tools, highway No restrictions. This document is available to the public
capacity, simulation guidelines through the National Technical Information Service,
Springfield, VA 22161.
19. Security Classif. (of this report) 20. Security Classif. (of this page) 21. No of Pages 22. Price
Unclassified Unclassified 146
Form DOT F 1700.7 (8-72) Reproduction of completed pages authorized
SI* (MODERN METRIC) CONVERSION FACTORS
APPROXIMATE CONVERSIONS TO SI UNITS
Symbol When You Know Multiply By To Find Symbol
LENGTH
in inches 25.4 millimeters mm
ft feet 0.305 meters m
yd yards 0.914 meters m
mi miles 1.61 kilometers km
AREA
2 2
in square inches 645.2 square millimeters mm
2 2
ft square feet 0.093 square meters m
2 2
yd square yard 0.836 square meters m
ac acres 0.405 hectares ha
2 2
mi square miles 2.59 square kilometers km
VOLUME
fl oz fluid ounces 29.57 milliliters mL
gal gallons 3.785 liters L
3 3
ft cubic feet 0.028 cubic meters m
3 3
yd cubic yards 0.765 cubic meters m
3
NOTE: volumes greater than 1000 L shall be shown in m
MASS
oz ounces 28.35 grams g
lb pounds 0.454 kilograms kg
T short tons (2000 lb) 0.907 megagrams (or "metric ton") Mg (or "t")
TEMPERATURE (exact degrees)
o o
F Fahrenheit 5 (F-32)/9 Celsius C
or (F-32)/1.8
ILLUMINATION
fc foot-candles 10.76 lux lx
2 2
fl foot-Lamberts 3.426 candela/m cd/m
FORCE and PRESSURE or STRESS
lbf poundforce 4.45 newtons N
lbf/in2 poundforce per square inch 6.89 kilopascals kPa
APPROXIMATE CONVERSIONS FROM SI UNITS

Symbol When You Know Multiply By To Find Symbol
LENGTH
mm millimeters 0.039 inches in
m meters 3.28 feet ft
m meters 1.09 yards yd
km kilometers 0.621 miles mi
AREA
2 2
mm square millimeters 0.0016 square inches in
2 2
m square meters 10.764 square feet ft
2 2
m square meters 1.195 square yards yd
ha hectares 2.47 acres ac
2 2
km square kilometers 0.386 square miles mi
VOLUME
mL milliliters 0.034 fluid ounces fl oz
L liters 0.264 gallons gal
m3 cubic meters 35.314 cubic feet ft3
3 3
m cubic meters 1.307 cubic yards yd
MASS
g grams 0.035 ounces oz
kg kilograms 2.202 pounds lb
Mg (or "t") megagrams (or "metric ton") 1.103 short tons (2000 lb) T
TEMPERATURE (exact degrees)
o o
C Celsius 1.8C+32 Fahrenheit F
ILLUMINATION
lx lux 0.0929 foot-candles fc
2 2
cd/m candela/m 0.2919 foot-Lamberts fl
FORCE and PRESSURE or STRESS
N newtons 0.225 poundforce lbf
2
kPa kilopascals 0.145 poundforce per square inch lbf/in
*SI is the symbol for the International System of Units. Appropriate rounding should be made to comply with Section 4 of ASTM E380.
(Revised March 2003)
ii
Table of Contents
Introduction ............................................................................................................................ 1
1.0 Microsimulation Study Organization/Scope ........................................................... 11

1.1 Study Objectives ................................................................................................... 11
1.2 Study Breadth ....................................................................................................... 11
1.3 Analytical Approach Selection ........................................................................... 14
1.4 Analytical Tool Selection (Software) ................................................................. 15
1.4.1 Technical Capabilities .............................................................................. 15
1.4.2 Input/Output/Interfaces......................................................................... 16
1.4.3 User Training/Support ............................................................................ 16
1.4.4 Ongoing Software Enhancements .......................................................... 16
1.5 Resource Requirements ....................................................................................... 16
1.6 Management of a Microsimulation Study ........................................................ 18
1.7 Example Problem: Study Scope and Purpose .................................................. 19
2.0 Data Collection/Preparation ........................................................................................ 23

2.1 Required Data ....................................................................................................... 23
2.1.1 Geometric Data.......................................................................................... 24
2.1.2 Control Data .............................................................................................. 24
2.1.3 Demand Data............................................................................................. 24
2.2 Calibration Data ................................................................................................... 27
2.2.1 Field Inspection ......................................................................................... 28
2.2.2 Travel Time Data....................................................................................... 28
2.2.3 Point Speed Data....................................................................................... 29
2.2.4 Capacity and Saturation Flow Data ....................................................... 29
2.2.5 Delay and Queue Data ............................................................................. 30
2.3 Data Preparation/Quality Assurance ............................................................... 30
2.4 Reconciliation of Traffic Counts......................................................................... 31
2.5 Example Problem: Data Collection and Preparation ...................................... 32
3.0 Base Model Development............................................................................................ 35

3.1 Link-Node Diagram: Model Blueprint.............................................................. 35
3.2 Link Geometry Data............................................................................................. 38
3.3 Traffic Control Data at Intersections and Junctions ........................................ 39
3.4 Traffic Operations and Management Data for Links ...................................... 39
3.5 Traffic Demand Data ........................................................................................... 39
3.6 Driver Behavior Data ........................................................................................... 40
3.7 Events/Scenarios Data ........................................................................................ 40
3.8 Simulation Run Control Data ............................................................................. 40
3.9 Coding Techniques for Complex Situations..................................................... 41
3.10 Quality Assurance/Quality Control (QA/QC) Plan ...................................... 42
3.11 Example Problem: Base Model Development .................................................. 42
iii
Table of Contents (continued)
4.0 Error Checking ............................................................................................................... 45

4.1 Review Software Errors....................................................................................... 45
4.2 Review Input......................................................................................................... 45
4.3 Review Animation ............................................................................................... 46
4.4 Residual Errors ..................................................................................................... 48
4.5 Key Decision Point ............................................................................................... 49
4.6 Example Problem: Error Checking .................................................................... 49
5.0 Calibration of Microsimulation Models ................................................................... 53

5.1 Objectives of Calibration ..................................................................................... 53
5.2 Calibration Approach .......................................................................................... 54
5.3 Step 1: Calibration for Capacity ......................................................................... 55
5.3.1 Collect Field Measurements of Capacity ............................................... 56
5.3.2 Obtain Model Estimates of Capacity...................................................... 57
5.3.3 Select Calibration Parameter(s)............................................................... 58
5.3.4 Set Calibration Objective Function ......................................................... 59
5.3.5 Perform Search for Optimal Parameter Value ...................................... 60
5.3.6 Fine-Tune the Calibration ........................................................................ 61
5.4 Step 2: Route Choice Calibration........................................................................ 62
5.4.1 Global Calibration..................................................................................... 62
5.4.2 Fine-Tuning................................................................................................ 62
5.5 Step 3: System Performance Calibration ........................................................... 63
5.6 Calibration Targets............................................................................................... 63
5.7 Example Problem: Model Calibration ............................................................... 65
6.0 Alternatives Analysis.................................................................................................... 73

6.1 Baseline Demand Forecast .................................................................................. 73
6.1.1 Demand Forecasting................................................................................. 73
6.1.2 Constraining Demand to Capacity......................................................... 74
6.1.3 Allowance for Uncertainty in Demand Forecasts ................................ 74
6.2 Generation of Alternatives.................................................................................. 74
6.3 Selection of Measures of Effectiveness (MOEs) ............................................... 75
6.3.1 Candidate MOEs for Overall System Performance ............................. 75
6.3.2 Candidate MOEs for Localized Problems ............................................. 76
6.3.3 Choice of Average or Worst Case MOEs............................................... 76
6.4 Model Application ............................................................................................... 77
6.4.1 Requirement for Multiple Repetitions ................................................... 77
6.4.2 Exclusion of Initialization Period ........................................................... 78
6.4.3 Avoiding Bias in the Results ................................................................... 78
6.4.4 Impact of Alternatives on Demand ........................................................ 78
6.4.5 Signal/Meter Control Optimization ...................................................... 78
iv
Table of Contents (continued)
6.5 Tabulation of Results ........................................................................................... 79
6.5.1 Reviewing Animation Output ................................................................ 79
6.5.2 Numerical Output..................................................................................... 80
6.5.3 Correcting Biases in the Results.............................................................. 81
6.6 Evaluation of Alternatives .................................................................................. 83
6.6.1 Interpretation of System Performance Results ..................................... 83
6.6.2 Hypothesis Testing ................................................................................... 85
6.6.3 Confidence Intervals and Sensitivity Analysis ..................................... 85
6.6.4 Comparison of Results to the HCM ....................................................... 87
6.7 Example Problem: Alternatives Analysis ......................................................... 88
7.0 Final Report .................................................................................................................... 93

7.1 Final Report........................................................................................................... 93
7.2 Technical Report/Appendix............................................................................... 93
Appendix A: Traffic Microsimulation Fundamentals .................................................... 95
Appendix B: Confidence Intervals ..................................................................................... 105
Appendix C: Estimation of the Simulation Initialization Period................................. 111
Appendix D: Simple Search Algorithms for Calibration............................................... 113
Appendix E: Hypothesis Testing of Alternatives ............................................................ 119
Appendix F: Demand Constraints ...................................................................................... 129
Appendix G: Bibliography................................................................................................... 133
v
List of Figures
1. Microsimulation model development and application process. ............................. 5
2. Prototypical microsimulation analysis task sequence............................................... 17
3. Example problem study network................................................................................. 21
4. Example of link-node diagram. .................................................................................... 36
5. Link-node diagram for example problem................................................................... 43
6. Checking intersection geometry, phasing, and detectors. ........................................ 50
7. Minimization of MSE. .................................................................................................... 61
8. Impact of mean queue discharge headway on MSE of modeled saturation

flow. .................................................................................................................................. 67
9. Average speed by minute of simulation (three links): Before calibration. ............. 68
10. Average speed by minute (three links): After calibration. ....................................... 68
11. Computation of uncaptured residual delay at the end of the simulation

period. .............................................................................................................................. 83
12. Ramp meter geometry. .................................................................................................. 89
13. Generation of driver characteristics............................................................................. 99
14. Variation in the results between repetitions............................................................... 106
15. Illustration of warmup period. ..................................................................................... 112
16. Golden Section Method. ................................................................................................ 115
17. Example contour plot of the squared error................................................................. 116
18. Dual-objective, dual-parameter search........................................................................ 117
19. Bottleneck, gateway, and study area. .......................................................................... 130
20. Example proportional reduction of demand for capacity constraint...................... 131
vi
List of Tables
1. Milestones and deliverables for a microsimulation study........................................ 18
2. Vehicle characteristic defaults that can be used in the absence of better data....... 27
3. Example node-numbering convention. ....................................................................... 38
4. Wisconsin DOT freeway model calibration criteria. ................................................. 64
5. Impact of queue discharge headway on predicted saturation flow........................ 66
6. System performance results for example problem. ................................................... 71
7. Summary of analytical results. ..................................................................................... 91
8. Minimum number of repetitions needed to obtain the desired confidence

interval. ............................................................................................................................ 109
9. Null hypothesis............................................................................................................... 120
10. Minimum repetitions for distinguishing alternatives. .............................................. 122
vii
Introduction
Microsimulation is the modeling of individual vehicle movements on a second or subsecond

basis for the purpose of assessing the traffic performance of highway and street systems,
transit, and pedestrians. The last few years have seen a rapid evolution in the
sophistication of microsimulation models and a major expansion of their use in
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 and will further support the credibility of the tools of today and
tomorrow. As a result, practitioners and decisionmakers will be equipped to make
informed decisions that will account for current and evolving technology. 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 consult 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.
Organization of Guidelines
These guidelines are organized into the following chapters and appendixes:
Introduction (this chapter) highlights the key guiding principles of microsimulation and
provides an overview of these guidelines.
Chapter 1.0 addresses the management, scope, and organization of microsimulation

analyses.
Chapter 2.0 discusses the steps necessary to collect and prepare input data for use in
microsimulation models.
Chapter 3.0 discusses the coding of input data into the microsimulation models.
Chapter 4.0 presents error-checking methods.
Chapter 5.0 provides guidance on the calibration of microsimulation models to local

traffic conditions.
1
Chapter 6.0 explains how to use microsimulation models for alternatives analysis.
Chapter 7.0 provides guidance on the documentation of microsimulation model analysis.
Appendix A provides an introduction to the fundamentals of microsimulation model

theory.
Appendix B provides guidance on the estimation of the minimum number of

microsimulation model run repetitions necessary for achieving a target confidence level
and confidence interval.
Appendix C provides guidance on the estimation of the duration of the initialization

(warmup) period, after which the simulation has stabilized and it is then appropriate to
begin gathering performance statistics.
Appendix D provides examples of some simple manual search algorithms for optimizing
parameters during calibration.
Appendix E summarizes the standard statistical tests that can be used to determine
whether two alternatives result in significantly different performance. The purpose of
these tests is to demonstrate that the improved performance in a particular alternative is
not a result of random variation in the simulation model results.
Appendix F describes a manual method for constraining external demands to available

capacity. This method is useful for adapting travel demand model forecasts for use in
microsimulation models.
Appendix G provides useful references on microsimulation.
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
obtaining a cost-effective microsimulation analysis is to observe certain guiding principles
for this type of analysis:
• Use of the appropriate tool is essential. Do not use microsimulation analysis when it is
not appropriate. Understand the limitations of the tool and ensure that it accurately
represents the traffic operations theory. Confirm that it can be applied to support the
purpose, needs, and scope of work, and can address the question that is being asked.
2
• Traffic Analysis Toolbox, Volume II: Decision Support Methodology for Selecting Traffic
Analysis Tools (Federal Highway Administration (FHWA), presents a methodology for
selecting the appropriate type of traffic analysis tool for the task.
• Do not use microsimulation if sufficient time and resources are not available.
Misapplication can degrade credibility and can be a focus of controversy or
disagreement.
• Good data are critical for good microsimulation model results.
• It is critical that the analyst calibrate any microsimulation model to local conditions.
• Output of a microsimulation model is different from that of the Highway Capacity

Manual (HCM) (Transportation Research Board (TRB)). Definitions of key terms, such
as “delay” and “queues,” are different at the microscopic level of microsimulation
models than at the macroscopic level typical of the HCM.
• Prior to embarking on the development of a microsimulation model, establish its

scope among the partners, taking into consideration expectations, tasks, and an
understanding of how the tool will support the engineering decision. Identify known
limitations.
• To minimize disagreements between partners, embed interim periodic reviews at

prudent milestones in the model development and calibration processes.
Additional information on traffic microsimulation fundamentals is provided in appendix A.
Terminology Used in These Guidelines
Calibration: Process where the analyst selects the model parameters that cause the model
to best reproduce field-measured local traffic operations conditions.
Microsimulation: Modeling of individual vehicle movements on a second or subsecond

basis for the purpose of assessing the traffic performance of highway and street systems.
Model: Specific combination of modeling software and analyst-developed input/

parameters for a specific application. A single model may be applied to the same study
area for several time periods and several existing and future improvement alternatives.
Project: To reduce the chances of confusing the analysis of a project with the project itself,
this report limits the use of the term “project” to the physical road improvement being
studied. The evaluation of the impact of a project will be called an “analysis.”
3
Software: Set of computer instructions for assisting the analyst in the development and
application of a specific microsimulation model. Several models can be developed using a
single software program. These models will share the same basic computational
algorithms embedded in the software; however, they will employ different input and
parameter values.
Validation: Process where the analyst checks the overall model-predicted traffic
performance for a street/road system against field measurements of traffic performance,
such as traffic volumes, travel times, average speeds, and average delays. Model
validation is performed based on field data not used in the calibration process. This report
presumes that the software developer has already completed this validation of the
software and its underlying algorithms in a number of research and practical applications.
Verification: Process where the software developer and other researchers check the
accuracy of the software implementation of traffic operations theory. This report provides
no information on software verification procedures.
Microsimulation Model Development and Application Process
The overall process for developing and applying a microsimulation model to a specific
traffic analysis problem consists of seven major tasks:
1. Identification of Study Purpose, Scope, and Approach.

2. Data Collection and Preparation.
3. Base Model Development.
4. Error Checking.
5. Calibration.
6. Alternatives Analysis.
7. Final Report and Technical Documentation.
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. It is intended to be
a quick reference that will be traceable throughout the document. This report’s chapters
correspond to the numbering scheme in figure 1.
4
Project Scope
1 - Define project purpose
- Identify influence areas
See Chapter 1
- Select approach
- Select model
- Estimate staff time Work prior to
actual modeling
Data Collection
2 - Traffic volumes
- Base maps/inventory See Chapter 2
- Field observations
Base Model Development

3 - Input data See Chapter 3
- Develop quality assurance
Error Checking Initial

4 - Review Inputs See Chapter 4 modeling
- Review Animation
Working Model
Before Calibration
Compare Model
5 MOEs to Field Data
- Volumes & speeds match? Adjust Model Parameters
- Congestion in right places? - Modify Global Parameters
- Modify Link Parameters
- Modify Route Choice Parameters
No Calibration
Acceptable
Match
See Chapter 5
Yes
Calibrated Model
Alternatives Analysis
6 - Forecast Demand
- Base Case See Chapter 6
- Project Alternatives Model
Application
Final Report
7 - Key Results See Chapter 7
- Technical Documentation
Developed by the FHWA Traffic Analysis Tools Team and later adapted from Advanced Corsim Training
Manual, Short, Elliott, Hendrickson, Inc., Minnesota Department of Transportation, September 2003.
Figure 1. Microsimulation model development and application process.
5
To demonstrate the process, an example problem is also provided. The example problem
involves the analysis of a corridor consisting of a freeway section and a parallel arterial.
For simplicity, this example assumes a proactive traffic management strategy that
includes ramp metering. Realizing that each project is unique, the analyst and project
manager may see a need to revisit previous tasks in the process to fully address the issues
that arise.
Organization and management of a microsimulation analysis require the development of

a “scope” for the analysis. This scope includes identification of project objectives, available
resources, assessment of verified and available tools, quality assurance plan, and
identification of the appropriate tasks to complete the analysis.
Task 1: Microsimulation Analysis Organization/Scope
The key issues for the management of a microsimulation study are:
• Securing sufficient expertise to develop and/or evaluate the model.
• Providing sufficient time and resources to develop and calibrate the microsimulation
model.
• Developing adequate documentation for the model development process and

calibration results.
Task 2: Data Collection and Preparation
This task involves the collection and preparation of all of the data necessary for the
microsimulation analysis. Microsimulation models require extensive input data, including:
• Geometry (lengths, lanes, curvature).

• Controls (signal timing, signs).
• Existing demands (turning volumes, origin-destination (O-D) table).
• Calibration data (capacities, travel times, queues).
• Transit, bicycle, and pedestrian data.
In support of tasks 4 and 5, the current and accurate data required for error checking and
calibration should also be collected. While capacities can be measured at any time, it is
crucial that the other calibration data (travel times, delays, and queues) be gathered
simultaneously with the traffic counts.
6
Task 3: Base Model Development
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, and, for this reason, the development
process may vary. This report provides a general outline of the model development task.
The method for developing a microsimulation model can best be thought of as the building
up of several layers of the model until the model has been completed. The first layer (the
link/node diagram) sets the foundation for the model. Additional data on traffic controls
and link operations are then added on top of this foundation. Travel demand and traveler
behavior data are then added to the basic network. Finally, the simulation run control data
are input to complete the model development task. The model development process does
not have to follow this order exactly; however, each of these layers is required in some
form in any simulation model. The model development task should also include the
development and implementation of a quality assurance/quality control (QA/QC) plan to
reduce the introduction of input coding errors into the model.
Task 4: Error Checking

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.
Task 5: Microsimulation Model Calibration

Each microsimulation software program has a set of user-adjustable parameters that
enable the practitioner to calibrate the software to better match specific local conditions.
These parameter adjustments are necessary because no microsimulation model can
include all of the possible factors (both onstreet and offstreet) that might affect capacity
and traffic operations. The calibration process accounts for the impact of these “unmodeled”
site-specific factors through the adjustment of the calibration parameters included in the
software for this specific purpose.
Therefore, model calibration involves the selection of a few parameters for calibration and
the repeated operation of the model to identify the best values for those parameters. This
can be a time-consuming process. It should be well documented so that later reviewers of
the model can understand the rationale for the various parameter changes made during
calibration. For example, the car-following sensitivity factor for a specific freeway segment
(link 20 to 21) has been modified to a value of 95 to match the observed average speed in
this freeway segment.
7
The key issues in calibration are:
• Identification of necessary model calibration targets.
• Allocation of sufficient time and resources to achieve calibration targets.
• Selection of the appropriate calibration parameter values to best match locally

measured street, highway, freeway, and intersection capacities.
• Selection of the calibration parameter values that best reproduce current route choice
patterns.
• Calibration of the overall model against overall system performance measures, such
as travel time, delay, and queues.
Task 6: Alternatives Analysis With Microsimulation Models
This is the first model application task. The calibrated microsimulation model is run
several times to test various project alternatives. The first step in this task is to develop a
baseline demand scenario. Then the various improvement alternatives are coded into the
simulation model. The analyst then determines which performance statistics will be
gathered and runs the model for each alternative to generate the necessary output. If the
analyst wishes to produce HCM level-of-service (LOS) results, then sufficient time should
be allowed for post-processing the model output to convert microsimulation results into
HCM-compatible LOS results.
The key issues in an alternatives analysis are:
• Forecasting realistic future demands.
• Selecting the appropriate performance measures for evaluation of the alternatives.
• Accurate accounting of the full congestion-reduction benefits of each alternative.
• Properly converting the microsimulation results to HCM LOS (reporting LOS is

optional).
Task 7: Final Report/Technical Documentation
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 decisionmakers for the project. The effort involved in summarizing the results for the
8
final report should not be underestimated, since microsimulation models produce a
wealth of numerical output that must be tabulated and summarized.
Technical documentation is important for ensuring that the decisionmakers understand

the assumptions behind the results and for enabling other analysts to reproduce the
results. The documentation should be sufficient so that given the same input files, another
analyst can understand the calibration process and repeat the alternatives analysis.
9
1.0 Microsimulation Study Organization/Scope
Microsimulation can provide a wealth of information; Project Scope

- Define project purpose
1
however, it can also be a very time-consuming and - Identify influence areas
- Select approach
- Select model
See Chapter 1
Work prior to
resource-intensive effort. It is critical that the manager

- Estimate staff time
actual modeling
Data Collection
2 - Traffic volumes
See Chapter 2
effectively coordinate the microsimulation effort to

- Base maps/inventory
ensure a cost-effective outcome to the study. The 3 - Input data

See Chapter 3
primary component of an effective management plan

- Review Animation
is the study scope, which defines the objectives, Working Model

Before Calibration
breadth, approach, tools, resources, and time schedule 5

Compare Model
MOEs to Field Data
- Congestion in right places?
for the study. This chapter presents the key

- Modify Global Parameters
No Calibration
Acceptable
components of an overall management plan for Match

See Chapter 5
Yes
achieving a cost-effective microsimulation analysis. Calibrated Model
6 - Forecast Demand
Application
1.1 Study Objectives 7

Final Report
- Key Results
See Chapter 7
Before embarking on any major analytical effort, it is wise to assess exactly what it is the
analyst, manager, and decisionmakers hope to accomplish. One should identify the study
objectives, the breadth of the study, and the appropriate tools and analytical approach to
accomplish those objectives.
Study objectives should answer the following Precise study objectives will ensure a
questions: cost-effective analysis.
It can be just as useful to identify
• Why is the analysis needed? what WILL NOT be analyzed, as it is
to identify what WILL be analyzed.
• What questions should the analysis answer?
(What alternatives will be analyzed?)
• Who are the intended recipients/decisionmakers for the results?
Try to avoid broad, all-encompassing study objectives. They are difficult to achieve with
the limited resources normally available and they do not help focus the analysis on the
top-priority needs. A great deal of study resources can be saved if the manager and the
analyst can identify upfront what WILL NOT be achieved by the analysis. The objectives
for the analysis should be realistic, recognizing the resources and time that may be
available for their achievement.
1.2 Study Breadth

Once the study objectives have been identified, the next step is to identify the scope or
breadth of the analysis—the geographic and temporal boundaries. Several questions
related to the required breadth of the analysis should be answered at this time:
11
• What are the characteristics of the project being analyzed? How large and complex is
it?
• What are the alternatives to be analyzed? How many of them are there? How large
and complex are they?
• What measures of effectiveness (MOEs) will be required to evaluate the alternatives

and how can they be measured in the field?
• What resources are available to the analyst?
• What are the probable geographic and temporal

scopes of the impact of the project and its The geographic and temporal scopes
of a microsimulation model should be
alternatives (now and in the future)? How far sufficient to completely encompass all
and for how many hours does the congestion of the traffic congestion present in the
extend? The geographic and temporal primary influence area of the project
boundaries selected for the analysis should during the target analysis period.
encompass all of the expected congestion to
provide a valid basis for comparing
alternatives.1
• What degree of precision do the decisionmakers require? Is a 10-percent error

tolerable? Are hourly averages satisfactory? Will the impact of the alternatives be very
similar or very different from those of the proposed project? How disaggregate an
analysis is required? Is the analysis likely to produce a set of alternatives where the
decisionmakers must choose between varying levels of congestion (as opposed to a
situation where one or more alternatives eliminate congestion, while others do not)?
Development of a logical terminus of an improvement project versus a model has been

debated since the early days of microsimulation; in the end, it is a matter of balancing
study objectives and study resources. Therefore, the modeler needs to understand the
operation of the improvement project to develop logical termini.
The study termini will be dependent on the “zone of influence,” and the project manager
will probably make that determination in consultation with the project stakeholders. Once
1
The analyst should try to design the model to geographically and temporally encompass all
significant congestion to ensure that the model is evaluating demands rather than capacity;
however, the extent of the congestion in many urban areas and resource limitations may preclude
100 percent achievement of this goal. If this goal cannot be achieved 100 percent, then the analyst
should attempt to encompass as much of the congestion as is feasible within the resource
constraints and be prepared to post-process the model’s results to compensate for the portion of
congestion not included in the model.
12
that has been completed, the modeler then needs to look at the operation of the proposed
facility. When determining the zone of influence, the modeler needs to understand the
operational characteristics of the facility in the proposed project. This could be one
intersection beyond the project terminus at one end of the project or a major generator 3.2
kilometers (km) (2 miles (mi)) away from the other end of the project. Therefore, there is
no geographical guidance that can be given. However, some general guidelines can be
summarized as follows:
Interstate Projects: The model network should extend up to 2.4 km (1.5 mi) from both
termini of the improvement being evaluated and up to 1.6 km (1 mi) on either side of the
interstate route. This will allow sufficient time and distance for the model to better
develop the traffic-stream characteristics reflected in the real world.
Arterial Projects: The model network should extend at least one intersection beyond those
within the boundaries of the improvement and should consider the potential impact on
arterial coordination as warranted. This will capture influences such as the upstream
metering of traffic and the downstream queuing of traffic.
The model study area should include areas that might be impacted by the proposed
improvement strategies. For example, if an analysis is to be conducted of incident
management strategies, the model study area should include the area impacted by the
diverted traffic. All potential problem areas should be included in the model network. For
example, if queues are identified in the network boundary areas, the analyst might need to
extend the network further upstream.
A study scope that has a tight geographic focus,

Microsimulation is like a microscope.
little tolerance for error, and little difference in the It is a very effective tool when you
performance of the alternatives will tend to point to have it pointed at the correct subject.
microsimulation as the most cost-effective Tightly focused scopes of work
approach.2 A scope with a moderately greater ensure cost-effective microsimulation
geographic focus and with a timeframe 20 years in analyses.
the future will tend to require a blended travel
demand model and microsimulation approach. A scope that covers large geographic areas
and long timeframes in the future will tend to rule out microsimulation and will instead
require a combination of travel demand models and HCM analytical techniques.
2
Continuing improvements in data collection, computer technology, and software will eventually
enable microsimulation models to be applied to larger problems.
13
1.3 Analytical Approach Selection
Traffic Analysis Toolbox, Volume II: Decision Support Methodology for Selecting Traffic Analysis
Tools3 (a separate document) provides detailed guidance on the selection of an appropriate
analytical approach. This section provides a brief
summary of the key points. Microsimulation models are data-
Microsimulation takes more effort than macroscopic intensive. They should only be used
simulation, and macroscopic simulation takes more when sufficient resources can be
effort than HCM-type analyses. The analyst should made available and less data-
employ only the level of effort required by the intensive approaches cannot yield
satisfactory results.
problem being studied.
The following are several situations where microsimulation is the best technical approach
for performing a traffic analysis:
• Conditions violate the basic assumptions of the other available analytical tools.
• Conditions are not covered by the other available analytical tools.
• There is testing of vehicle performance, guidance, and driver behavior modification
options.
For example, most of the HCM procedures assume that the operation of one intersection
or road segment is not adversely affected by conditions on the adjacent roadway.4 Long
queues from one location interfering with another location would violate this assumption.
Microsimulation would be the superior analytical tool for this situation.
Because they are sensitive to different vehicle performance characteristics and differing
driver behavior characteristics, microsimulation models are useful for testing intelligent
transportation system (ITS) strategies designed to modify these characteristics. Traveler
information systems, automated vehicle guidance, triple- or quadruple-trailer options,
new weight limits, new propulsion technologies, etc., are all excellent candidates for
testing with microsimulation models. The HCM procedures, for example, are not
designed to be sensitive to new technology options, while microsimulation allows
prediction of what the effect of new technology might be on capacity before the new
technology is actually in place.
3
Available at http://ops.fhwa.dot.gov/Travel/Traffic_Analysis_Tools/traffic_analysis_tools.htm.
4
The one exception to this statement is the recently developed freeway systems analysis
methodology presented in Highway Capacity Manual 2000 (HCM 2000), which does explicitly treat
oversaturated flow situations.
14
1.4 Analytical Tool Selection (Software)
The selection of the appropriate analytical tool is a Software selection is like picking out a
key part of the study scope and is tied into the suit of clothes. There is no single suit
selection of the analytical approach. Some of the that fits all people and all uses. You
key criteria for software selection are technical must know what your software needs
capabilities, input/output/interfaces, user are and the capabilities and training of
the people that you expect to use the
training/support, and ongoing software software. Then you can select the
enhancements.5 software that best meets your needs
and best fits the capabilities of the
Generally, it is a good idea to separate the selection people who will use it.
of the appropriate analytical tool from the actual
implementation of the tool. This can be accomplished through a selection process that is
independent from any project-level analytical activities. Traffic Analysis Toolbox, Volume II:
Decision Support Methodology for Selecting Traffic Analysis Tools, identifies several criteria
that should be considered in the selection of an appropriate traffic analysis tool and helps
identify the circumstances when a particular type of tool should be used. A methodology
also is presented to guide the users in the selection of the appropriate tool category. This
report includes worksheets that transportation professionals can use to select the
appropriate tool category and assistance in identifying the most appropriate tool within
the selected category. An automated tool that implements this methodology can be found
at the FHWA Traffic Analysis Tools Web site at:
http://ops.fhwa.dot.gov/Travel/Traffic_Analysis_Tools/traffic_analysis_tools.htm
1.4.1 Technical Capabilities
The technical capabilities of the software are related to its ability to accurately forecast the
traffic performance of the alternatives being considered in the analysis. The manager must
decide if the software is capable of handling the size of problems being evaluated in the
study. Are the technical analytical procedures that are incorporated into the software
sensitive to the variables of concern in the study? The following is a general list of
technical capabilities to be considered in the selection of software:
5
Additional discussions on software selection criteria can be found in Traffic Analysis Toolbox,
Volume II: Decision Support Methodology for Selecting Traffic Analysis Tools, and in Shaw, J.W., and
D.H. Nam, “Microsimulation, Freeway System Operational Assessment, and Project Selection in
Southeastern Wisconsin: Expanding the Vision” (paper presented at the TRB Annual Meeting,
Washington, D.C., 2002).
15
• Maximum problem size (the software may be limited by the maximum number of
signals that can be in a single network or the maximum number of vehicles that may
be present on the network at any one time).
• Vehicle movement logic (lane changing, car following, etc.) that reflects the state of the
art.
• Sensitivity to specific features of the alternatives being analyzed (such as trucks on

grades, effects of horizontal curvature on speeds, or advanced traffic management
techniques).
• Model parameters available for model calibration.
• Variety and extent of prior successful applications of the software program (should be
considered by the manager).
1.4.2 Input/Output/Interfaces
Input, output, and the ability of the software to interface with other software that will be
used in the study (such as traffic forecasting models) are other key considerations. The
manager should review the ability of the software to produce reports on the MOEs needed
for the study. The ability to customize output reports can also be very useful to the analyst.
It is essential that the manager or analyst understand the definitions of the MOEs as
defined by the software. This is because a given MOE may be calculated or defined
differently by the software in comparison to how it is defined or calculated by the HCM.
1.4.3 User Training/Support
User training and support requirements are another key consideration. What kind of
training and support is available? Are there other users in the area that can provide
informal advice?
1.4.4 Ongoing Software Enhancements
Finally, the commitment of the software developer to ongoing enhancements ensures that
the agency’s investment in staff training and model development for a particular software
tool will continue to pay off over the long term. Unsupported software can become
unusable if improvements are made to operating systems and hardware.
1.5 Resource Requirements

The resource requirements for the development, calibration, and application of
microsimulation models will vary according to the complexity of the project, its
16
geographic scope, temporal scope, number of alternatives, and the availability and quality
of the data.6
In terms of training, the person responsible for the initial round of coding can be a
beginner or on an intermediate level in terms of knowledge of the software. They should
have supervision from an individual with more experience with the software. Error
checking and calibration are best done by a person with advanced knowledge of
microsimulation software and the underlying algorithms. Model documentation and
public presentations can be done by a person with an intermediate level of knowledge of
microsimulation software.
A prototype time schedule for the various model development, calibration, and
application tasks is presented in figure 2, which shows the sequential nature of the tasks
and their relative durations. Data collection, coding, error checking, and calibration are the
critical tasks for completing a calibrated model. The alternatives analysis cannot be started
until the calibrated model has been reviewed and accepted.
Project Schedule
Task
1. Project Scope
2. Data Collection
3. Develop Base Model
4. Error Checking
5. Calibration
6. Alternatives Analysis
7. Final Report
Figure 2. Prototypical microsimulation analysis task sequence.
6
Some managers might devote about 50 percent of the budget to the tasks that lead up to and
include coding of the simulation model, including data collection. Another 25 percent of the
budget might go toward calibration. The remaining 25 percent might then go toward alternatives
analysis and documentation. Others might divide the resources into one-third each for data
collection and model coding, calibration, and alternatives analysis and documentation.
17
1.6 Management of a Microsimulation Study
Much of the management of a microsimulation study is the same as managing any other
highway design project: establish clear objectives, define a solid scope of work and
schedule, monitor milestones, and review deliverables. The key milestones and
deliverables for a microsimulation study are shown in table 1:
Table 1. Milestones and deliverables for a microsimulation study.
Milestone Deliverable Contents

1. Study scope 1. Study scope and schedule Study objectives, geographic and
2. Proposed data collection plan temporal scope, alternatives,
data collection plan, coding
3. Proposed calibration plan
error-checking procedures,
4. Coding quality assurance plan calibration plan and targets
2. Data collection 5. Data collection results report Data collection procedures,
quality assurance, summary of
results
3. Model 6. 50% coded model Software input files
development
4. Error checking 7. 100% coded model Software input files
5. Calibration 8. Calibration test results report Calibration procedures, adjusted
parameters and rationale,
achievement of calibration
targets
6. Alternatives 9. Alternatives analysis report Description of alternatives,
analysis analytical procedures, results
7. Final report 10. Final report Summary tables and graphics
highlighting key results
11. Technical documentation Compilation of prior reports
documenting model
development and calibration,
software input files
Two problems are often encountered when managing microsimulation models developed
by others:
• Insufficient managerial expertise for verifying the technical application of the model.
• Insufficient data/documentation for calibrating the model.
18
The study manager may choose to bring more expertise to the review of the model by
forming a technical advisory panel. Furthermore, use of a panel may support a project of
regional importance and detail, or address stakeholder interests regarding the acceptance
of new technology. The panel may be drawn from experts at other public agencies,
consultants, or from a nearby university. The experts should have had prior experience
developing simulation models with the specific software being used for the particular
model.
The manager (and the technical advisory panel) must have access to the input files and the
software for the microsimulation model. There are several hundred parameters involved
in the development and calibration of a simulation model. Consequently, it is impossible
to assess the technical validity of a model based solely on its printed output and visual
animation of the results. The manager must have access to the model input files so that he
or she can assess the veracity of the model by reviewing the parameter values that go into
the model and looking at its output.
Finally, good documentation of the model calibration process and the rationale for
parameter adjustments is required so that the technical validity of the calibrated model
can be assessed. A standardized format for the calibration report can expedite the review
process.
1.7 Example Problem: Study Scope and Purpose

The example problem is a study of the impact of freeway ramp metering on freeway and
surface-street operations. Ramp metering will be operational during the afternoon peak
period on the eastbound on-ramps at two interchanges.7
Study Objectives: To quantify the traffic operation benefits and the impact of the
proposed afternoon peak-period ramp metering project on both the freeway and nearby
surface streets. The information will be provided to technical people at both the State
department of transportation (DOT) and the city public works department.
Study Breadth (Geographic): The ramp metering project is expected to impact freeway
and surface-street operations several miles upstream and downstream of the two metered
on-ramps. However, the most significant impact is expected in the immediate vicinity of
the meters (the two interchanges, the closest parallel arterial streets, and the cross-
connector streets between the interchanges and the parallel arterial). If the negative impact
7
This example problem is part of a larger project involving metering of several miles of freeway.
The study area in the example problem was selected to illustrate the concepts and procedures in
the microsimulation guide.
19
of the ramp meters is acceptable in the immediate vicinity of the project, then the impact
farther away should be of a lower magnitude and, therefore, also acceptable.8
There is no parallel arterial street north of the freeway, so only the parallel arterial on the
south side needs to be studied. Figure 3 illustrates the selected study area. There are six
signalized intersections along the study section of the Green Bay Avenue parallel arterial.
The signals are operating on a common 90-second (s) cycle length. The signals at the ramp
junctions on Wisconsin and Milwaukee Avenues also operate on a 90-s cycle length.
The freeway and the adjacent arterial are currently not congested during the afternoon
peak period and, consequently, the study area only needs to be enlarged to encompass
projected future congestion. The freeway and the arterial are modeled 1.6 km (1 mi) east
and west of the two interchanges to include possible future congestion under the “no meter”
and “meter” alternatives.
Study Breadth (Temporal): Since there is no existing congestion on the freeway and
surface streets, and the ramp metering will occur only in the afternoon peak hour, the
afternoon peak hour is selected as the temporal boundaries for the analysis.9
Analytical Approach: There are concerns that: (1) freeway traffic may be diverted to city
streets causing unacceptable congestion, and (2) traffic at metered on-ramps may backup
and adversely impact surface-street operations. A regional travel demand model would
probably not be adequate to estimate traffic diversion caused by a regional ramp metering
system because it could not model site- and time-specific traffic queues accurately and
would not be able to predict the impact of the meters on surface-street operations. HCM
methods could estimate the capacity impact of ramp metering; however, because these
methods are not well adapted to estimating the impact of downstream congestion on
upstream traffic operations, they are not sufficient by themselves for the analysis of the
impact of ramp meters on surface-street operations. Microsimulation would probably
assist the analyst in predicting traffic diversions between the freeway and surface streets,
and would be the appropriate analytical tool for this problem.
Analytical Tool: The analyst selects a microsimulation tool that either incorporates a
procedure for the rerouting of freeway traffic in response to ramp metering, or the analyst
8
Of course, if the analyst does not believe this to be true, then the study area should be expanded
accordingly.
9
If existing or future congestion on either the freeway or arterial was expected to last more than
1 h, then the analysis period would be extended to encompass all of the existing and future
congestion in the study area.
20
supplements the microsimulation tool with a separate travel demand model analysis to
predict the rerouting of traffic.10
Resource Requirements and Schedule: The resource requirements and schedule are
estimated at this time to ensure that the project can be completed with the selected
approach to the satisfaction of the decisionmakers. The details of the resources and
scheduling are not discussed here because they are specific to the resources available to
the analyst and the time available for completion.
Milwaukee
Wisconsin
Mainline USA
1st Street
3rd Street
4th Street
6th Street
Green Bay
Figure 3. Example problem study network.
10
For this example problem, a microsimulation tool without rerouting capabilities was selected. It
was supplemented with estimates of traffic diversions from a regional travel demand model.
21
2.0 Data Collection/Preparation
This chapter provides guidance on the identification,

Project Scope
1 - Define project purpose
See Chapter 1
- Select approach
collection, and preparation of the data sets needed to - Select model

actual modeling
develop a microsimulation model for a specific

Data Collection
2 - Traffic volumes
project analysis, and the data needed to evaluate the 3

- Input data See Chapter 3
calibration and fidelity of the model to real-world


4 - Review Inputs See Chapter 4
conditions present in the project analysis study area.

modeling
- Review Animation
Working Model
There are agency-specific techniques and guidance Before Calibration
Compare Model
documents that focus on data collection, which 5 MOEs to Field Data

- Volumes & speeds match?
Adjust Model Parameters
should be used to support project-specific needs.

No Calibration
Acceptable
Match
See Chapter 5
Yes
A selection of general guides on the collection of Calibrated Model
traffic data includes:

6 - Forecast Demand
Application
Final Report
Introduction to Traffic Engineering: A Manual for Data

Collection and Analysis, T.R. Currin, Brooks/Cole, 2001, 140 pp., ISBN No. 0-534378-67-6.
Manual of Transportation Engineering Studies, H. Douglas Robertson, Joseph E. Hummer,

and Donna C. Nelson, Institute of Transportation Engineers, Washington, DC, 1994, ISBN
No. 0-13-097569-9.
Highway Capacity Manual 2000, TRB, 2000, 1200 pp., ISBN No. 0-309067-46-4.
These sources should be consulted regarding appropriate data collection methods (they
are not all-inclusive on the subject of data collection). The discussion in this chapter
focuses on data requirements, potential data sources, and the proper preparation of data
for use in microsimulation analysis.
If the amount of available data does not adequately support the project objectives and
scope identified in task 1, then the project team should return to task 1 and redefine the
objectives and scope so that they will be sufficiently supported by the available data.
2.1 Required Data
The precise input data required by a microsimulation model will vary by software and the
specific modeling application as defined by the study objectives and scope. Most
microsimulation analytical studies will require the following types of input data:11
11
The forecasting of future demands (turning volumes, O-D table) is discussed later under the
Alternatives Analysis task.
23
• Road geometry (lengths, lanes, curvature).
• Traffic controls (signal timing, signs).
• Demands (entry volumes, turning volumes, O-D table).
• Calibration data (traffic counts and performance data such as speed, queues).
In addition to the above basic input data, microsimulation models also require data on
vehicle and driver characteristics (vehicle length, maximum acceleration rate, driver
aggressiveness, etc.). Because these data can be difficult to measure in the field, it is often
supplied with the software in the form of various default values.
Each microsimulation model will also require various control parameters that specify how
the model conducts the simulation. The user’s guide for the specific simulation software
should be consulted for a complete list of input requirements. The discussion below
describes only the most basic data requirements shared by the majority of
microsimulation model software.
2.1.1 Geometric Data
The basic geometric data required by most models consist of the number of lanes, length,
and free-flow speed.12 For intersections, the necessary geometric data may also include the
designated turn lanes and their vehicle storage lengths. These data can usually be
obtained from construction drawings, field surveys, geographical information system
(GIS) files, or aerial photographs.
2.1.2 Control Data
Control data consist of the locations of traffic control devices and signal-timing settings.13
These data can best be obtained from the files of the agencies operating the traffic controls
or from field inspection.
2.1.3 Demand Data
The basic travel demand data required for most models consist of entry volumes (traffic
entering the study area) and turning movements at intersections within the study area.
12
Some microsimulation models may allow (or require) the analyst to input additional geometric
data related to the grades, horizontal curvature, load limits, height limits, shoulders, onstreet
parking, pavement condition, etc.
13
Some models may allow the inclusion of advanced traffic control features. Some models require
the equivalent fixed-time input for traffic-actuated signals. Others can work with both fixed-time
and actuated-controller settings.
24
Some models require one or more vehicular O-D tables, which enable the modeling of
route diversions. Procedures exist in many demand modeling software and some
microsimulation software for estimating O-D tables from traffic counts.
Count Locations and Duration
Traffic counts should be conducted at key locations within the microsimulation model
study area for the duration of the proposed simulation analytical period. The counts
should ideally be aggregated to no longer than 15-minute (min) time periods; however,
alternative aggregations can be used if dictated by circumstances.14
If congestion is present at a count location (or upstream of it), care should be taken to
ensure that the count measures demand and not capacity. The count period should ideally
start before the onset of congestion and end after the dissipation of all congestion to
ensure that all queued demand is eventually included in the count.
The counts should be conducted simultaneously if resources permit so that all count
information is consistent with a single simulation period. Often, resources do not permit
this for the larger simulation areas, so the analyst must establish one or more control
stations where a continuous count is maintained over the length of the data collection
period. The analyst then uses the control station counts to adjust the counts collected over
several days into a single consistent set of counts representative of a single typical day
within the study area.
Estimating Origin-Destination (O-D) Trip Tables
For some simulation software, the counts must be converted into an estimate of existing
O-D trip patterns. Other software programs can work with either turning-movement
counts or an O-D table. An O-D table is required if it is desirable to model route choice
shifts within the microsimulation model.
Local metropolitan planning organization (MPO) travel demand models can provide O-D
data; however, these data sets are generally limited to the nearest decennial census year
and the zone system is usually too macroscopic for microsimulation. The analyst must
usually estimate the existing O-D table from the MPO O-D data in combination with other
data sources, such as traffic counts. This process will probably require consideration of
O-D pattern changes resulting from the time of day, especially for simulations that cover
an extended period of time throughout the day.
A license plate matching survey is the most accurate method for measuring existing O-D
data. The analyst establishes checkpoints within and on the periphery of the study area
14
Project constraints, traffic counter limitations, or other considerations (such as a long simulation
period) may require that counts be aggregated to longer or shorter periods.
25
and notes the license plate numbers of all vehicles passing by each checkpoint. A matching
program is then used to determine how many vehicles traveled between each pair of
checkpoints. However, license plate surveys can be quite expensive. For this reason, the
estimation of the O-D table from traffic counts is often selected.15
Vehicle Characteristics
The vehicle characteristics typically include vehicle mix, vehicle dimensions, and vehicle
performance characteristics (maximum acceleration, etc.).16
Vehicle Mix: The vehicle mix is defined by the analyst, often in terms of the percentage of
total vehicles generated in the O-D process. Typical vehicle types in the vehicle mix might
be passenger cars, single-unit trucks, semi-trailer trucks, and buses.
Default percentages are usually included in most software programs; however, the vehicle
mix is highly localized and national default values will rarely be valid for specific
locations. For example, the percentage of trucks in the vehicle mix can vary from a low of
2 percent on urban streets during rush hour to a high of 40 percent of daily weekday
traffic on an intercity interstate freeway.
It is recommended that the analyst obtain one or more vehicle classification studies for the
study area for the time period being analyzed. Vehicle classification studies can often be
obtained from nearby truck weigh station locations.
Vehicle Dimensions and Performance: The analyst should attempt to obtain the vehicle
fleet data (vehicle mix, dimensions, and performance) from the local State DOT or air
quality management agency. National data can be obtained from the Motor and
Equipment Manufacturers Association (MEMA), various car manufacturers, FHWA, and
the U.S. Environmental Protection Agency (EPA). In the absence of data from these
sources, the analyst may use the defaults shown in table 2.
15
There are many references in the literature on the estimation of O-D volumes from traffic counts.
Appendix A, chapter 29, HCM 2000 provides a couple of simple O-D estimation algorithms. Most
travel demand software and some microsimulation software have O-D estimation modules
available.
16
The software-supplied default vehicle mix, dimensions, and performance characteristics should
be reviewed to ensure that they are representative of local vehicle fleet data, especially for
simulation software developed outside the United States.
26
Table 2. Vehicle characteristic defaults that can be used in the absence of better data.
Maximum Maximum Maximum

Speed Accel. Decel. Jerk
Vehicle Type Length (ft) (mi/h) (ft/s2) (ft/s2) (ft/s3)
Passenger car 14 75 10 15 7
Single-unit truck 35 75 5 15 7
Semi-trailer truck 53 67 3 15 7
Double-bottom 64 61 2 15 7
trailer truck
Bus 40 65 5 15 7
1 ft = 0.305 m, 1 mi/h = 1.61 km/h, 1 ft/s2 = 0.305 m/s2, 1 ft/s3 = 0.305 m/s3
Sources: CORSIM and SimTraffic™ technical documentation

Notes:
Maximum Speed: Maximum sustained speed on level grade in miles per hour (mi/h)
Maximum Accel.: Maximum acceleration rate in feet per second squared (ft/s2) when starting from
zero speed
Maximum Decel.: Maximum braking rate in ft/s2 (vehicles can actually stop faster than this;
however, this is a mean comfort-based maximum)
Jerk: Maximum rate of change in acceleration rates in feet per second cubed (ft/s3)
2.2 Calibration Data
Calibration data consist of measures of capacity, System performance data (travel

traffic counts, and measures of system performance times, delays, queues, speeds) must
such as travel times, speeds, delays, and queues. be gathered simultaneously with the
Capacities can be gathered independently of the traffic counts.
traffic counts (except during adverse weather or
lighting conditions); however, travel times, speeds,
delays, and queue lengths must be gathered simultaneously with the traffic counts to be
useful in calibrating the model.17 If there are one or more continuous counting stations in
the study area, it may be possible to adjust the count data to match the conditions present
when the calibration data were collected; however, this introduces the potential for
17
It is not reasonable to expect the simulation model to reproduce observed speeds, delays, and
queues if the model is using traffic counts of demand for a different day or time period than when
the system performance data were gathered.
27
additional error in the calibration data and weakens the strength of the conclusions that
can be drawn from the model calibration task.
Finally, the analyst should verify that the documented signal-timing plans coincide with
those operating in the field. This will confirm any modifications resulting from a signal
retiming program.
2.2.1 Field Inspection
It is extremely valuable to observe existing operations in the field during the time period
being simulated. Simple visual inspection can identify behavior not apparent in counts
and floating car runs. Video images may be useful; however, they may not focus on the
upstream conditions causing the observed behavior, which is why a field visit during
peak conditions is always important. A field inspection is also valuable for aiding the
modeler in identifying potential errors in data collection.
2.2.2 Travel Time Data
The best source of point-to-point travel time data is “floating car runs.” In this method,
one or more vehicles are driven the length of the facility several times during the
analytical period and the mean travel time is computed. The number of vehicle runs
required to establish a mean travel time within a 95-percent confidence level depends on
the variability of the travel times measured in the field. Free-flow conditions may require
as few as three runs to establish a reliable mean travel time. Congested conditions may
require 10 or more runs.
The minimum number of floating car runs needed to determine the mean travel time
within a desired 95-percent confidence interval depends on the width of the interval that
is acceptable to the analyst. If the analyst wishes to calibrate the model to a very tight
tolerance, then a very small interval will be desirable and a large number of floating car
runs will be required. The analyst might aim for a confidence interval of ± 10 percent of
the mean travel time. Thus, if the mean travel time were 10 min, the target 95-percent
confidence interval would be 2 min. The number of required floating car runs is obtained
from equation 1:
2
 s
N =  2 ∗ t 0.025, N −1  (1)
 R
where:
R = 95-percent confidence interval for the true mean

t0.025,N-1 = Student’s t-statistic for two-sided error of 2.5 percent (totals 5 percent) with N-1
degrees of freedom (for four runs, t = 3.2; for six runs, t = 2.6; for 10 runs, t = 2.3)
28
(Note: There is one less degree of freedom than car runs when looking up the appropriate
value of t in the statistical tables.)
s = standard deviation of the floating car runs
N = number of required floating car runs
For example, if the floating car runs showed a standard deviation of 1.0 min, a minimum
of seven floating car runs would be required to achieve a target 95-percent confidence
interval of 2.0 min (± 1.0 min) for the mean travel time.
The analyst is advised that the standard deviation is unknown prior to runs being
conducted. In addition, the standard deviation is typically higher under more congested
conditions.
2.2.3 Point Speed Data
Traffic management centers (TMCs) are a good source of simultaneous speed and flow
data for urban freeways. Loop detectors, though, may be subject to failure, so the data
must be reviewed carefully to avoid extraneous data points. Loop detectors are typically
spaced 0.5 to 0.8 km (0.3 to 0.5 mi) apart and their detection range is limited to 3.7 meters
(m) (12 feet (ft)). Under congested conditions, much can happen between detectors, so the
mean speeds produced by the loop detectors cannot be relied on to give system travel
times under congested conditions.
The loop-measured free-flow speeds may be reliable for computing facility travel times
under uncongested conditions; however, care should be taken when using these data.
Many locations have only single-loop detectors in each lane, so the free-flow speed must
be estimated from an assumed mean vehicle length. The assumed mean vehicle length
may be automatically calibrated by the TMC; however, this calibration requires some
method of identifying which data points represent free-flow speed, which data points do
not, and which ones are aberrations. The decision process involves some uncertainty. In
addition, the mix of trucks and cars in the traffic stream varies by time of day, thus the
same mean vehicle length cannot be used throughout the day. Overall, loop-
estimated/-measured free-flow speeds should be treated with a certain amount of caution.
They are precise enough for identifying the onset of congestion; however, they may not be
an accurate measure of speed.
2.2.4 Capacity and Saturation Flow Data
Capacity and saturation flow data are particularly valuable calibration data since they
determine when the system goes from uncongested to congested conditions:
• Capacity can be measured in the field on any street segment immediately downstream
of a queue of vehicles. The queue should ideally last for a full hour; however,
29
reasonable estimates of capacity can be obtained if the queue lasts only 0.5 hour (h).
The analyst would simply count the vehicles passing a point on the downstream
segment for 1 h (or for a lesser time period if the queue does not persist for a full hour)
to obtain the segment capacity.
• Saturation flow rate is defined as “the equivalent hourly rate at which previously
queued vehicles can traverse an intersection approach under prevailing conditions,
assuming that the green signal is available at all times and no lost times are
experienced, in vehicles per hour or vehicles per hour per lane.”18 The saturation flow
rate should be measured (using procedures specified in the HCM) at all signalized
intersections that are operating at or more than 90 percent of their existing capacity.
At these locations, the estimation of saturation flow and, therefore, capacity will
critically affect the predicted operation of the signal. Thus, it is cost-effective to
accurately measure the saturation flow and, therefore, capacity at these intersections.
2.2.5 Delay and Queue Data
Delay can be computed from floating car runs or from delay studies at individual
intersections. Floating car runs can provide satisfactory estimates of delay along the
freeway mainline; however, they are usually too expensive to make all of the necessary
additional runs to measure all of the ramp delays. Floating cars are somewhat biased
estimators of intersection delay on surface streets since they reflect only those vehicles
traveling a particular path through the network. For an arterial street with coordinated
signal timing, the floating cars running the length of the arterial will measure delay only
for the through movement with favorable progression. Other vehicles on the arterial will
experience much greater delays. This problem can be overcome by running the floating
cars on different paths; however, the cost may be prohibitive.
Comprehensive measures of intersection delay can be obtained from surveys of stopped

delay on the approaches to an intersection (see the HCM for the procedure). The number
of stopped cars on an approach is counted at regular intervals, such as every 30 s. The
number of stopped cars multiplied by the counting interval (30 s) gives the total stopped
delay. Dividing the total stopped delay by the total number of vehicles that crossed the
stop line (a separate count) during the survey period gives the mean stopped delay per
vehicle. The stopped delay can be converted to the control delay using the procedure in
appendix A of chapter 16 in the HCM.
2.3 Data Preparation/Quality Assurance

Data preparation consists of review, error checking, and the reduction of the data
collected in the field. Data reduction is already well covered in other manuals on data
18
HCM 2000.
30
collection. This section consequently focuses on review and error checking of the data. The
following checks of the data should be made during the data preparation step:
• Geometric and control data should be reviewed for apparent violations of design
standards and/or traffic engineering practices. Sudden breaks in geometric continuity
(such as a short block of a two-lane street sandwiched in between long stretches of a
four-lane street) may also be worth checking with people who are knowledgeable
about local conditions. Breaks in continuity and violations of design standards may be
indicative of data collection errors.
• Internal consistency of counts should be reviewed. Upstream counts should be

compared to downstream counts. Unexplained large jumps or drops in the counts
should be reconciled.19
• Floating car run results should be reviewed for realistic segment speeds.
• Counts of capacity and saturation flow should be compared to the HCM estimates for
these values. Large differences between field measurements and the HCM warrant
double-checking the field measurements and the HCM computations.20
2.4 Reconciliation of Traffic Counts

Inevitably, there will be traffic counts at two or more nearby adjacent locations that do not
match. This may be a result of counting errors, counting on different days (counts
typically vary by 10 percent or more on a daily basis), major traffic sources (or sinks)
between the two locations, or queuing between the two locations. In the case of a freeway,
a discrepancy between the total traffic entering the freeway and the total exiting it may be
caused by storage or discharge of some of the vehicles in growing or shrinking queues on
the freeway.
The analyst must review the counts and determine (based on local knowledge and field
observations) the probable causes of the discrepancies. Counting errors and counts made
19
There is no guidance on precisely what constitutes a “large” difference in counts. The analyst
might consider investigating any differences of greater than 10 percent between upstream and
downstream counts for locations where no known traffic sources or sinks (such as driveways)
exist between the count locations. Larger differences are acceptable where driveways and parking
lots could potentially explain the count differences.
20
There is no guidance on precisely how large differences can be between the HCM and field
measurements before the field measurements may be suspect. The analyst might consider 25-
percent differences to be within the normal range of accuracy of the HCM and take a second look
at the calculations and field measurements if the differences are greater than that.
31
on different days are treated differently than counting differences caused by midblock
sources/sinks or midblock queuing.
Discrepancies in the counts resulting from counting errors or counts made on different
days must be reconciled before proceeding to the model development task. Inconsistent
counts make error checking and model calibration much more difficult. Differing counts
for the same location should be normalized or averaged assuming that they are
reasonable. This is especially true for entry volumes into the model network. Intersection
turning volumes should be expressed as percentages based on an average of the counts
observed for that location. This will greatly assist with calibrating the model later.
Differences in counts caused by midblock sources (such as a parking lot) need not be
reconciled; however, they must be dealt with by coding midblock sources and sinks in the
simulation model during the model development task.
Differences in entering and exiting counts that are caused by queuing in between the two
count locations suggest that the analyst should extend the count period to ensure that all
demand is included in both counts.
Accurate vehicle classification counts and accurate travel speeds can also affect the traffic
volumes. Trucks and other large vehicles and inaccurate speeds can skew the volume
counts.
2.5 Example Problem: Data Collection and Preparation

The same ramp metering example problem discussed in chapter 1.0 is continued here.
Now the task is to identify, gather, and prepare the data for the study area and the
afternoon peak-hour analytical period.
Road Geometry: Aerial photographs, construction drawings, and field inspections are
used to obtain the lengths and the number of lanes for each section of the freeway, ramps,
and surface streets. Turn lanes and pocket lengths are determined for each intersection.
Transition lengths for lane drops and additions are determined. Lane widths are
measured if they are not standard widths. Horizontal curvature and curb return radii are
determined if the selected software tool is sensitive to these features of the road and
freeway design. Free-flow speeds are estimated based on the posted speed limits for the
freeway and the surface streets.
Traffic Controls: Existing signal settings were obtained from the agencies’ records and
verified in the field. The controllers at the interchange ramp terminals are all fixed-time,
having a cycle length of 90 s. The signals along Green Bay Avenue are traffic-actuated.
Demands: Field measurements of traffic volumes on the freeway mainline and ramps, and
turning-movement counts at each intersection were conducted for a 2-h period during the
32
afternoon peak period (4:00 to 6:00 p.m.). The peak hour was determined to be between
5:00 and 6:00 p.m., and the highest 15-min volumes occurred between 5:30 and 5:45 p.m.
Vehicle Characteristics: The default vehicle mix and characteristics provided with the
microsimulation software were used in this analysis.21
Calibration Data: The model will be calibrated against estimates of capacity and traffic
counts, and the system performance will be assessed against travel time data.
Saturation flows for protected movements at traffic signals were estimated using the HCM
2000 methodology and verified through field observations at a few approaches with long
queues.22
Capacity values for basic freeway segments were estimated using the HCM 2000
procedures.23
The traffic count data have already been discussed.
The system performance calibration data were obtained at the same time as the traffic
counts, which consisted of travel times obtained from floating car runs, delays at traffic
signals, and speeds on the freeway.
Data Preparation: The input data were reviewed by the analyst for consistency and
accuracy prior to data entry. The turning, ramp, and freeway mainline counts were
reconciled by the analyst to produce a set of consistent counts for the entire study area.
After completion of the reconciliation, all volumes discharged from upstream locations are
equal to the volumes arriving at the downstream location. Based on local knowledge of
the area, the analyst determined that midblock sources were not required for the surface
streets in the study area.
21
The selected software had been developed with defaults appropriate to the United States, and this
was considered to be sufficiently accurate for this planning study of ramp metering.
22
The HCM was used to estimate surface-street saturation flow rates because none of the
intersections were currently operating nor were they expected to operate within 90 percent of
their capacity. Thus, an estimate is sufficiently accurate. Field measurements were not required.
23
The HCM was used to estimate freeway capacity because there is no freeway congestion under
existing conditions that would enable field measurement of capacities. Another method would be
to observe capacities on a similar type of facility in the area that operated under congested
conditions. This method may be preferred if a comparable facility can be found.
33
3.0 Base Model Development
This chapter provides general guidance on the Project Scope

1
procedures for developing a microsimulation model. - Identify influence areas
- Select approach
- Select model
See Chapter 1
Work prior to
- Estimate staff time
There are many software programs for performing 2 Data Collection

- Traffic volumes
actual modeling
See Chapter 2
this task and each has its own unique coding

methodology. The guidance provided here is 3 - Input data

See Chapter 3
intended to give general advice on model

- Review Animation
development; however, the analyst should consult Working Model

Before Calibration
the specific microsimulation software documentation 5

Compare Model
MOEs to Field Data
for information on available data input tools and Acceptable No

Calibration
techniques. Match
See Chapter 5
Yes
Calibrated Model
Building a model is analogous to building a house. 6 - Forecast Demand

- Base Case
- Project Alternatives
See Chapter 6
Model
You begin with a blueprint and then you build each

Application
Final Report
element in sequence—the foundation, the frame, the

roof, the utilities and drywall, and finally the interior details. The development of a
successful simulation model is similar in that you must begin with a blueprint (the link-
node diagram) and then you proceed to build the model in sequence—coding links and
nodes, filling in the link geometries, adding traffic control data at appropriate nodes,
coding travel demand data, adding traveler behavior data, and finally selecting the model
run control parameters.
3.1 Link-Node Diagram: Model Blueprint

The link-node diagram is the blueprint for constructing the microsimulation model. The
diagram identifies which streets and highways will be included in the model and how
they will be represented. An example link-node diagram is shown in figure 4. This step is
critical and the modeler should always prepare a link-node diagram for a project of major
complexity.
The link-node diagram can be created directly in the microsimulation software or offline
using various types of computer-aided design (CAD) software. If the diagram is created in
the microsimulation software, then it is helpful to import a map or aerial photograph into
the software over which the link-node diagram can be overlaid. If the diagram is created
offline using CAD software, then it is helpful to import a map or photograph into the
CAD software.
35
Freeway
Interchange
Link-Node
Diagram
Figure 4. Example of link-node diagram.
Nodes are the intersection of two or more links. Nodes are usually placed in the model
using x-y coordinates and they can be at a place that represents an intersection or a
location where there is a change in the link geometry. Some simulation software may also
warrant consideration of a z coordinate. The node locations can be obtained from aerial
photographs, maps, or physical measurements.
36
Links are one-directional segments of surface streets or freeways.24 Links represent the
length of the segment and usually contain data on the geometric characteristics of the road
or highway between the nodes. Ideally, a link represents a roadway segment with
uniform geometry and traffic operation conditions.25
The analyst should consider establishing a node-numbering scheme to facilitate error

checking and the aggregation of performance statistics for groups of links related to a
specific facility or facility type (see table 3 for an example of a node-numbering scheme
designed to enable the rapid determination of the facility type in text output). Much of the
output produced by microsimulation software is text, with the results identified by the
node numbers at the beginning and end points of each link. A node-numbering
convention can greatly facilitate the search for results in the massive text files commonly
produced by simulation software.26
If the link-node diagram was created offline (using some other software besides the
microsimulation software), then the information in the diagram needs to be entered (or
imported) into the microsimulation software. The x-y coordinates and identification
numbers of the nodes (plus any feature points needed to represent link curvature) are
entered. The nodes are then connected together to develop the link-node framework for
the model.
24
Some software programs do not always use a link-node scheme, while others allow the analyst to
code both directions of travel with a single link. The two-way links coded by the user are then
represented internally (inside the software) as two one-way links.
25
The analyst may find it desirable to split links for performance reporting purposes. For example,
it may be desirable to report density separately for the 460 m (1500-ft) influence area on the
freeway near a ramp merge or diverge. However, the analyst should be cautious about the
potential impact of split links on the ability of the software to accurately simulate vehicle
behavior.
26
Some software programs provide analytical modules that assist the analyst in displaying and
aggregating the results for specific groups of links. This feature reduces the necessity of adopting
a node-numbering convention; however, a numbering convention can still result in a significant
labor savings when reviewing text output or when importing text results into other software
programs for analytical purposes.
37
Table 3. Example node-numbering convention.
Range
Segments From To Description
0’s 1 99 Miscellaneous
100’s 100 199 Northbound Freeway Mainline
200’s 200 299 Northbound Freeway Ramps
300’s 300 399 Southbound Freeway Mainline
400’s 400 499 Southbound Freeway Ramps
500’s 500 599 Eastbound Freeway Mainline
600’s 600 699 Eastbound Freeway Ramps
700’s 700 799 Westbound Freeway Mainline
800’s 800 899 Westbound Freeway Ramps
900’s 900 999 Arterials
3.2 Link Geometry Data

Once the modeler has completed coding the link-node diagram, then the modeler must
input the physical and operational characteristics of the links into the model. This phase is
when the modeler must code the following:
• Number of lanes.
• Lane width.
• Link length.
• Grade.
• Curvature.
• Pavement conditions (dry, wet, etc.).
• Sight distance.
• Bus stop locations.
• Crosswalks and other pedestrian facilities.
38
• Bicycle lanes/paths.
• Others.
The specific data to be coded for the links will vary according to the microsimulation
software.
3.3 Traffic Control Data at Intersections and Junctions

Most microsimulation uses a time-step simulation to describe traffic operations (which is
usually 1 s or less). Vehicles are moved according to car-following logic in response to
traffic control devices and in response to other demands. Traffic control devices for
microsimulation models will vary. Listed below are some examples:
• No control.
• Yield signs.
• Stop signs.
• Signals (pretimed, actuated, real-time traffic adaptive).
• Ramp metering.
3.4 Traffic Operations and Management Data for Links

Traffic operations and management data for links consist of the following:
• Warning data (incidents, lane drops, exits, etc.).
• Regulatory data (speed limits, variable speed limits, high-occupancy vehicles (HOVs),
high-occupancy toll (HOT), detours, lane channelizations, lane use, etc.).
• Information (guidance) data (dynamic message signs and roadside beacons).
• Surveillance detectors (type and location).
3.5 Traffic Demand Data

Traffic demands are defined as the number of vehicles and the percentage of vehicles of
each type that wish to traverse the study area during the simulation time period.
Furthermore, it may be necessary to reflect the variation in demand throughout the
simulation time period. In most software programs, the traffic entering into the network is
usually defined by some parameter, and traffic leaving the network is usually computed
based on parameters internal to the network (turning movements, etc.). The modeler must
code the traffic volume by first starting from the external nodes (this is where the traffic is
put into the model). Once all of the entering traffic volumes at the external nodes are
39
coded, the modeler will then go into the model and define the turning movements and
any other parameter related to route choice. The key traffic demand data are:
• Entry volumes by vehicle type and turning fractions at all intersections or junctions
(random walk simulators).
• O-D/path-specific and vehicle data (path-specific simulators).
• Bus operations (routes and headways/schedules).
• Bicycle and pedestrian demand data.
3.6 Driver Behavior Data

Default values for driver behavior data are usually provided with the microsimulation
software. Users may override the default values of any driver behavior parameters if valid
observed data are available (e.g., desired free-flow speed, discharge headway, startup lost
time at intersections, etc.). Driver behavior data include:
• Driver’s aggressiveness (for minimum headway in car-following, gap acceptance for

lane changing, response to yellow interval).
• Availability of (real-time) information for the driver.
• Driver’s response to information (for pretrip planning and/or en route switching).
3.7 Events/Scenarios Data

Events data include:
• Blockages and incidents.

• Work zones.
• Parking activity in curb lane.
This data is optional and will vary according to the specific application being developed
by the analyst.
3.8 Simulation Run Control Data

All simulation software contain run control parameters to enable the modeler to
customize the software operation for their specific modeling needs. These parameters will
vary between software programs. They generally include:
40
• Length of simulation time.
• Selected MOEs or output (e.g., reports, animation files, or both).
• Resolution of simulation results (e.g., temporal and special resolution).
• Other system parameters to run the model.
3.9 Coding Techniques for Complex Situations

Microsimulation software allows the modeler to develop a model to represent the real-
world situation. However, not all possible real-world situations were necessarily
contemplated when the software was originally written. This is when the modeler, with a
good understanding of the operation of the software, can “extend” the software to
simulate conditions not originally incorporated into the microsimulation software.
However, this should only be attempted after the tools, resources, and skills available are
fully appreciated. This is because new tools continue to become available that account for
selected real-world situations.
Provided below are some examples of how basic microsimulation software can be applied
to less standard situations, noting that calibration (discussed in chapter 5.0) must also
consider the following situations:
• A curb lane is heavily used for parking, loading, and unloading activities. As a result,
this lane may be blocked virtually all of the time. If the simulation software cannot
correctly replicate the real situation, the modeler may consider removing this lane
from the link.
• There is a drawbridge in the real-world network; however, the software does not have
explicit “drawbridge” modeling capability. If boat traffic is very heavy and is
equivalent to vehicular traffic, the modeler may consider coding the drawbridge as an
intersection with fixed-time signal control. If boat traffic is irregular, the analyst might
consider coding the drawbridge as a semi-actuated traffic signal-controlled
intersection. If boat traffic is rare, then the drawbridge might be coded as a link with
events that result in a 100-percent capacity loss with a certain probability of
occurrence during the analytical period.
• Traffic regularly backs up on a freeway off-ramp, backing up onto the freeway.

However, instead of stopping on the freeway and blocking a lane, the queue forms on
the shoulder of the freeway, keeping the right-hand lane open. If this is the case in the
study area, the modeler may artificially extend the off-ramp length to realistically
model the traffic in the field.
41
3.10 Quality Assurance/Quality Control (QA/QC) Plan
Once the modeler is satisfied that the model is coded, the modeler must then vet the
model for completeness. This could be a QA/QC process. Many of the models input are
developed in a format that can be exported to a spreadsheet format. Therefore, the
modeler can use this type of tool to aid in vetting the model for completeness and
accuracy. The next chapter on error checking describes quality control tests in more detail.
3.11 Example Problem: Base Model Development

Continuing with the same example problem from the previous chapters, the task is now to
code the base model.
The first step is to code the link-node diagram. How this is best done can be determined
from the software user’s guide. Figure 5 shows the link-node diagram for the example
problem.
The next steps are to input:
1. Link geometry data.

2. Intersection traffic control data.
3. Traffic operations and management data.
4. Traffic count data.
5. Run control parameters.27
27
The software-provided default values for driver behavior (aggressiveness, awareness, etc.) were
used in this example problem. They were considered to be sufficiently accurate for this analysis.
42
43
Figure 5. Link-node diagram for example problem.
4.0 Error Checking
The error correction step is essential in developing a 1

Project Scope
working model so that the calibration process does not

See Chapter 1
- Select approach
- Select model
result in parameters that are distorted to compensate

actual modeling
Data Collection
2 - Traffic volumes
for overlooked coding errors. The succeeding steps of - Field observations
the calibration process rely on the elimination of all

Error Checking
major errors in demand and network coding before

Initial
- Review Animation
calibration.
Working Model
Before Calibration
Compare Model
Error checking involves various reviews of the coded

No Calibration
Acceptable
network, coded demands, and default parameters. Match

See Chapter 5
Yes
Error checking proceeds in three basic stages: Calibrated Model
(1) software error checking, (2) input coding error

6 - Forecast Demand
Application
checking, and (3) animation review to spot less obvious 7

Final Report
- Key Results
See Chapter 7
input errors.
4.1 Review Software Errors

The analyst should review the software and user group Web sites to ensure that he or she
is aware of the latest known “bugs” and user workarounds for the software. The analyst
should ensure that he or she is using the latest version and “patch” of the software.
4.2 Review Input

A checklist for verifying the accuracy of the coded input data is provided below:
1. Link and node network:
a. Check basic network connectivity (are all connections present?).

b. Check link geometry (lengths, number of lanes, free-flow speed, facility type, etc.).
c. Check intersection controls (control type, control data).
d. Check for prohibited turns, lane closures, and lane restrictions at the intersections
and on the links.
2. Demand:
a. Check vehicle mix proportions at each entry node/gate/zone.
b. Check identified sources and sinks (zones) for traffic.
c. Verify zone volumes against traffic counts.
d. Check vehicle occupancy distribution (if modeling HOVs).
45
e. Check turn percentages (if appropriate).
f. Check O-Ds of trips on the network.
3. Traveler behavior and vehicle characteristics:

a. Check and revise, as necessary, the default vehicle types and dimensions.
b. Check and revise the default vehicle performance specifications.
The following techniques may be useful to increase the efficiency and effectiveness of the
error-checking process:1
• Overlay the coded network over aerial photographs of the study area to quickly verify
the accuracy of the coded network geometry.
• If working with software that supports three-dimensional modeling, turn on the node
numbers and look for superimposed numbers. They are an indication of
unintentionally superimposed links and nodes. Two or more nodes placed in the same
location will look like a single node in a three-dimensional model when viewed in two
dimensions. The links may connect to one of the nodes, but not to the other.
• For a large network, a report summarizing the link attributes should be created so that
their values can be easily reviewed.
• Use color codes to identify links by the specific attribute being checked (e.g., links
might be color-coded by free-flow speed range). Out-of-range attributes can be
identified quickly if given a particular color. Breaks in continuity can also be spotted
quickly (e.g., a series of 56-km/h (35-mi/h) links with one link coded as 40 km/h
(25 mi/h)).
4.3 Review Animation

Animation output enables the analyst to see the vehicle behavior that is being modeled
and assess the reasonableness of the microsimulation model itself. Running the simulation
model and reviewing the animation, even with artificial demands, can be useful to
identify input coding errors. The analyst inputs a very low level of demand and then
follows individual vehicles through the network. Aberrant vehicle behavior (such as
unexpected braking or stops) is a quick indicator of possible coding errors.
A two-stage process can be followed in reviewing the animation output:
1. Run the animation at an extremely low demand level (so low that there is no
congestion). The analyst should then trace single vehicles through the network and
see where they unexpectedly slow down. These will usually be locations of minor
1
Some of these techniques may not be available or necessary for some software programs.
46
network coding errors that disturb the movement of vehicles over the link or through
the node. This test should be repeated for several different O-D zone pairs.
2. Once the extremely low demand level tests have been completed, then run the
simulation at 50 percent of the existing demand level. At this level, demand is usually
not yet high enough to cause congestion. If congestion appears, it may be the result of
some more subtle coding errors that affect the distribution of vehicles across lanes or
their headways. Check entry and exit link flows to verify that all demand is being
correctly loaded and moved through the network.
The animation should be observed in close detail at key congestion points to determine if
the animated vehicle behavior is realistic. If the observed vehicle behavior appears to be
unrealistic, the analyst should explore the following potential causes of the unrealistic
animation in the order shown below:
Error in Analyst Expectations: The analyst should first verify in the field the correct
vehicle behavior for the location and time period being simulated before deciding that the
animation is showing unrealistic vehicle behavior. Many times, analyst expectations of
realistic vehicle behavior are not matched by actual behavior in the field.2 Field inspection
may also reveal the causes of vehicle behavior that are not apparent when coding the
network from plans and aerial photographs. These causes need to be coded into the
model if the model is expected to produce realistic behavior.3
Analyst Data Coding Errors: The analyst should check for data coding errors that may be
causing the simulation model to represent travel behavior incorrectly. Subtle data coding
errors are the most frequent cause of unrealistic vehicle behavior in commercial
microsimulation models that have already been subjected to extensive validation. Subtle
coding errors include apparently correctly coded input that is incorrect because of how it
is used in the model to determine vehicle behavior.4
2
Analysts should not expect classic macroscopic traffic-flow concepts to apply at the microscopic
individual-vehicle level. Macroscopic flow concepts (e.g., no variance in mean speed at low flow
rates) do not apply to the behavior of an individual vehicle over the length of the highway. An
individual vehicle’s speed may vary over the length of the highway and between vehicles, even at
low flow rates. Macroscopic flow theory refers to the average speed of all vehicles being relatively
constant at low flow rates, not individual vehicles.
3
A TMC with high-density camera spacing will be very helpful in reviewing the working model.
Many TMCs are now providing workstations for traffic analysis/simulation staff.
4
For example, it could be that the warning sign for an upcoming off-ramp is posted in the real
world 0.40 km (0.25 mi) before the off-ramp; however, because the model uses warning signs to
identify where people start positioning themselves for the exit ramps, the analyst may have to
code the warning sign at a different location (the location where field observations indicate that
the majority of the drivers start positioning themselves for the off-ramp).
47
A comparison of model animation to field design and operations cannot be
overemphasized. Some of the things to look for include:
• Overlooked data values that need refinement.
• Aberrant vehicle operations (e.g., drivers using shoulders as turning or travel lanes,
etc.).
• Previously unidentified points of major ingress or egress (these might be modeled as

an intersecting street).
• Operations that the model cannot explicitly replicate (certain operations in certain
tools/models), such as a two-way center turn lane (this might be modeled as an
alternating short turn bay).
• Unusual parking configurations, such as median parking (this might be modeled

operationally by reducing the free-flow speed to account for this friction).
• Average travel speeds that exceed posted or legal speeds (use the observed average
speed in the calibration process).
• Turn bays that cannot be fully utilized because of being blocked by through traffic.
• In general, localized problems that can result in a systemwide impact.
4.4 Residual Errors

If the analyst has field-verified his or her expectations of traffic performance and has
exhausted all possible input errors, and the simulation still does not perform to the
analyst’s satisfaction, there are still a few possibilities. The desired performance may be
beyond the capabilities of the software, or there may be a software error.
Software limitations can be identified through careful review of the software

documentation. If software limitations are a problem, then the analyst will have to work
around the limitations by “tricking” or “forcing” the model to produce the desired
performance.5 If the limits are too great, the analyst might seek an alternate software
program without the limitations. Advanced analysts can also write their own software
interface with the microsimulation software (called an “application program interface”
(API)) to overcome the limitations and produce the desired performance.
Software errors can be tested by coding simple test problems (such as a single link or
intersection) where the result (such as capacity or mean speed) can be computed manually
and compared to the model. Software errors can only be resolved by working with the
software developer.
5
For example, a drawbridge that opens regularly might be coded as a traffic signal.
48
4.5 Key Decision Point
The completion of error checking is a key decision point. The next task—model
calibration—can be very time-consuming. Before embarking upon this task, the analyst
should confirm that error checking has been completed, specifically:
• All input data are correct.
• Values of all initial parameters and default parameters are reasonable.
• Animated results look fine based on judgment or field inspection.
Once the error checking has been completed, the analyst has a working model (though it
is still not calibrated).
4.6 Example Problem: Error Checking
Continuing with the same example problem from the previous chapters, the task is now to
error check the coded base model.
Software: The latest version of the software was used. Review of the model
documentation and other material in the software and user groups’ Web sites indicated
that that there were no known problems or bugs related to the network under study and
the scenarios to be simulated.
Review of Input Data and Parameters: The coded input data were verified using the
input files, the input data portion of the output files, static displays, and animation.
First, the basic network connectivity was checked, including its consistency with coded
geometry and turning restrictions. All identified errors were corrected. For example, one
link with exclusive left- and right-turn lanes (no through traffic) was coded as feeding a
downstream through link.
Static network displays were used extensively to verify the number of lanes, lane use, lane
alignment (i.e., lane number that feeds the downstream through link), and the location of
lane drops. At this step, the consistency of link attributes was checked. For example, is the
free-flow speed of 87 km/h (55 mi/h) coded for all freeway links?
Next, the traffic demand data were checked. Input volumes at the network entrances were
specified in four time slices. The input values were checked against the collected data.
49
Special attention was given to the traffic patterns at the interchange ramp terminals to
avoid unrealistic movement. The software provisions (and options) were exercised to
force the model not to assign movements to travel paths that were not feasible.6
The vehicle characteristics and performance data were reviewed.
The model displays and animation were used to verify the input data and operation of the
traffic signals coded at each intersection. For fixed-time signals, the phasing and signal
settings were checked (see figure 6). For actuated signals, initial simulations were
performed with reduced, but balanced, volumes to ensure that all phases were activated
by the traffic demand. This was done because often the inappropriate settings of the phase
flags cause signals to malfunction within the simulation and produce unreasonable
results. This step also involves checking the location of the detectors and their association
with the signal phases.
Figure 6. Checking intersection geometry, phasing, and detectors.
Review Animation: Following the checking of the input data, the model was run using
very low demand volumes to verify that all of the vehicles travel the network without
6
An example of this would be coding to address freeway drivers who do or do not make U-turns
at an interchange (i.e., get off and then get back on the freeway in the opposite direction).
50
slowdowns. This step uncovered minor errors in the links alignments that needed to be
adjusted.
Next, the traffic demands were specified to about 50 percent of the actual volumes and the
simulation model was rerun. Animation was used to verify that all demands were
properly loaded in the network links and the traffic signals were properly operating. The
link and system performance measures (travel time, delay) were also checked for
reasonableness (i.e., they should reflect free-flow conditions).
Careful checking of the animation revealed subtle coding problems. For example, the
coded distance of a warning sign (for exiting vehicles) or the distance from the start of the
link to the lane drop affects the proper simulation of driver behavior. These problems
were corrected.
Key Decision Point: The model, as revised throughout the error-checking process, was
run with all the input data (actual demands) and the default model parameters. The
output and animation were also reviewed and discussed with other agency staff who
were familiar with the study area. The conclusion is that the model is working properly.
51
5.0 Calibration of Microsimulation Models
Upon completion of the error-checking task, the analyst 1

Project Scope
has a working model. However, without calibration,

See Chapter 1
- Select approach
- Select model
actual modeling
the analyst has no assurance that the model will 2 Data Collection
- Traffic volumes
correctly predict traffic performance for the project. 3


- Input data See Chapter 3
Error Checking
Calibration is the adjustment of model parameters to

Initial
- Review Animation
improve the model’s ability to reproduce local driver

Working Model
Before Calibration
behavior and traffic performance characteristics.

Compare Model
Calibration is performed on various components of the Acceptable No

Calibration
overall model.
Match
See Chapter 5
Yes
Calibrated Model
The importance of calibration cannot be

6 - Forecast Demand
Application
overemphasized. Recent tests of six different software 7

Final Report
- Key Results
See Chapter 7
programs found that calibration differences of

13 percent in the predicted freeway speeds for existing conditions increased to differences
of 69 percent in the forecasted freeway speeds for future conditions.7
5.1 Objectives of Calibration

Calibration is necessary because no single model The objective of calibration is to
can be expected to be equally accurate for all improve the ability of the model to
possible traffic conditions. Even the most detailed accurately reproduce local traffic
microsimulation model still contains only a portion conditions.
of all of the variables that affect real-world traffic
conditions. Since no single model can include the whole universe of variables, every
model must be adapted to local conditions.
Every microsimulation software program comes with a set of user-adjustable parameters

for the purpose of calibrating the model to local conditions. Therefore, the objective of
calibration is to find the set of parameter values for the model that best reproduces local
traffic conditions.
For the convenience of the analyst, the software developers provide suggested default
values for the model parameters. However, only under very rare circumstances will the
model be able to produce accurate results for a specific area using only the default
parameter values. The analyst should always perform some calibration tests to ensure that
the coded model accurately reproduces local traffic conditions and behavior.
7
Bloomberg, L., M. Swenson, and B. Haldors, Comparison of Simulation Models and the Highway
Capacity Manual, Preprint, Annual Meeting, TRB, Washington, DC, 2003.
53
The fundamental assumption of calibration is that the travel behavior models in the
simulation model are essentially sound.8 There is no need to verify that they produce the
correct delay, travel time, and density when they are given the correct input parameters
for a link. Therefore, the only remaining task for the analyst is to adjust the parameters so
that the models correctly predict local traffic conditions.
5.2 Calibration Approach

Calibration involves the review and adjustment of potentially hundreds of model
parameters, each of which impacts the simulation results in a manner that is often highly
correlated with that of the others. The analyst can easily get trapped in a never-ending
circular process, fixing one problem only to find that a new one occurs somewhere else.
Therefore, it is essential to break the calibration process into a series of logical, sequential
steps—a strategy for calibration.
To make calibration practical, the parameters must be divided into categories and each
category must be dealt with separately. The analyst should divide the available calibration
parameters into the following two basic categories:
• Parameters that the analyst is certain about and does not wish to adjust.
• Parameters that the analyst is less certain about and willing to adjust.
The analyst should attempt to keep the set of adjustable parameters as small as possible to
minimize the effort required to calibrate them. Whenever practical, the analyst should use
observed field data to reflect local conditions. This observed data will serve as the
nonadjustable values for certain calibration parameters, thus leaving the set of adjustable
parameters to a minimum. However, the tradeoff is that more parameters allow the
analyst more degrees of freedom to better fit the calibrated model to the specific location.
The set of adjustable parameters is then further subdivided into those that directly impact
capacity (such as mean headway) and those that directly impact route choice. The capacity
adjustment parameters are calibrated first, then the route choice adjustment parameters
are calibrated.
Each set of adjustable parameters can be further subdivided into those that affect the
simulation on a global basis and those that affect the simulation on a more localized
basis. The global parameters are calibrated first. The local link-specific parameters are
used after the global calibration to fine-tune the results.
8
The analyst determines the soundness of the simulation software when selecting it for use in the
study.
54
The following three-step strategy is recommended for calibration:
1. Capacity Calibration: An initial calibration is

Calibration Strategy:
performed to identify the values for the
capacity adjustment parameters that cause the 1. Calibrate capacity parameters.
model to best reproduce observed traffic 2. Calibrate route choice
capacities in the field. A global calibration is parameters.
performed first, followed by link-specific fine- 3. Calibrate overall model
tuning. The HCM can be used as an alternative performance.
source of capacity target values if field
measurements are not feasible.
2. Route Choice Calibration: If the microsimulation model includes parallel streets, then
route choice will be important. In this case, a second calibration process is performed,
but this time with the route choice parameters. A global calibration is performed first,
followed by link-specific fine-tuning.
3. System Performance Calibration: Finally, the overall model estimates of system

performance (travel times and queues) are compared to the field measurements for
travel times and queues. Fine-tuning adjustments are made to enable the model to
better match the field measurements.
5.3 Step 1: Calibration for Capacity

The capacity calibration step adjusts the global and link-specific capacity-related
parameters in the simulation model to best replicate local field measurements for capacity.
This is an important step because capacity has a significant effect on predicted system
performance (delay and queues).9 The objective of this calibration step is to find a set of
model parameters that causes the model to come as close as possible to matching the field
measurements for traffic capacity.
The capacity calibration step consists of two phases:

Capacity calibration has two phases:
(1) a global calibration phase and (2) a fine-tuning
phase. Global calibration is first performed to 1. Global calibration
identify the appropriate network-wide value of the 2. Link fine-tuning
capacity parameter(s) that best reproduces
9
Historically, it has been the practice to calibrate microsimulation models to all the traffic counts in
the field. The majority of these counts will be at noncritical locations. The recommended strategy
is to focus (at this point in the calibration process) only on the critical counts at the bottlenecks
and to get the model to reproduce these counts correctly. Once this has been done, the rest of the
counts are used later to check the route choice aspects of the model. All of this presupposes that
the demands have been entered correctly and have already been checked against the counts at the
entry gates as part of the first step (error checking).
55
conditions in the field. Link-specific capacity parameters are then adjusted to fine-tune the
model so that it more precisely matches the field-measured capacities at each bottleneck.10
The capacity calibration procedure is as follows:
1. Collect field measurements of capacity.

2. Obtain model estimates of capacity.
3. Select calibration parameters.
4. Set calibration objective function.
5. Perform search for optimal parameter value.
6. Fine-tune the calibration.
The following subsections explain this procedure in more detail.
5.3.1 Collect Field Measurements of Capacity
The identification of locations for field measurements of capacity will depend on the
existing traffic conditions within the study area.
For nonsignalized facilities (freeways, rural highways, and rural roads), the analyst should
identify locations where queues persist for at least 15 min and measure the flow rate at the
point where the queue discharges. This observed flow rate is measured only while an
upstream queue is present. It is totaled across all lanes and converted to an equivalent
hourly flow rate. This is the field-measured capacity of the facility at this point.
For signalized intersections, the analyst should identify the approach legs that frequently
have queues of at least 10 vehicles per lane and measure the saturation flow rate per hour
per lane using the procedures outlined in appendix H to the signalized intersection
chapter of HCM. The capacity of the signalized intersection approach is then the
saturation flow multiplied by the portion of green time in the signal cycle:
g
c=s (2)
C
10
It is certainly possible to use link-specific parameters exclusively during the capacity calibration
step; however, this eliminates the benefits of the global parameter adjustment. The global
adjustment ensures the accuracy of the model-predicted capacities on all links (even those not
currently congested). Adjustment of link-specific parameters ensures the model accuracy only for
the specific link.
56
where:
c = capacity (vehicles per hour (veh/h))

s = saturation flow (veh/h per green phase)
g = effective green time
C = cycle length
Higher saturation flows and lower startup lost times result in higher capacities.
Several measurements of maximum flow rates should be made in the field and averaged.
Procedures are provided in the appendix for estimating how many measurements are
required to estimate capacity within a desired confidence interval (see Appendix B:
Confidence Intervals). If capacity cannot be measured in the field, then the HCM
methodology can be used to estimate capacity. The HCM methods should not be
considered a default technique since the estimates are not as accurate as direct field
measurements.
5.3.2 Obtain Model Estimates of Capacity
Microsimulation models do not output a number called “capacity.” Instead, they output
the number of vehicles that pass a given point. Thus, the analyst must manipulate the
input demand as necessary to create a queue upstream of the target section to be
calibrated so that the model will report the maximum possible flow rate through the
bottleneck.
If the model does not initially show congestion at

If the model initially shows no
the same bottleneck locations as exist in the field, congestion at the bottlenecks, the
then the demands coded in the model are demand must be temporarily
temporarily increased to force the creation of increased before calibration can
congestion at those bottlenecks. These temporary proceed.
increases must be removed after the capacity
calibration has been completed, but before the route choice calibration step.
If the model initially shows congested bottlenecks at

All “false” bottlenecks in the model
locations that DO NOT exist in the field, it will be must be cleared before calibration
necessary to temporarily increase the capacity at can proceed.
those false bottlenecks (using temporary link-specific
headway adjustments). These temporary adjustments are then removed during the fine-
tuning phase.
For nonsignalized facilities (freeways, rural highways, and rural roads), the simulated
queue should persist for at least 15 min of simulated time, across all lanes and links
feeding the target section. The simulated capacity is then the mean flow rate at the target
section (measured at a detector placed in the target section and summed across all lanes)
averaged over the 15-min or longer period that the queue is present. The result is then
divided by the number of lanes and is converted to an hourly flow rate.
57
For signalized intersections, the coded demand should be increased, as necessary, to
ensure the necessary queues of at least 10 vehicles at the start of the green phase. A
detector is placed in the model at the stop line to measure the discharge headways (on a
per lane basis) of the first 10 vehicles crossing the detector in each lane.11 The per lane
headways are averaged for each lane and then averaged across lanes. The result is then
converted to an hourly flow rate per lane.
Just as the field measurements of capacity were repeated several times and the results
were averaged, the model runs should be repeated several times and the maximum flow
rate at each location should be averaged across the runs. The minimum required number
of runs to obtain a value of capacity within a desired confidence interval can be calculated
using the procedures provided in appendix B (Confidence Intervals).
5.3.3 Select Calibration Parameter(s)
Only the model parameters that directly affect capacity are calibrated at this time. Each
microsimulation software program has its own set of parameters that affect capacity,
depending on the specific car-following and lane-changing logic implemented in the
software. The analyst must review the software documentation and select one or two of
these parameters for calibration.
This chapter does not intend to describe all of the parameters for all of the simulation
models that are available. An illustrative list of capacity-related parameters for freeways
and signalized arterials is given below:
Freeway Facilities:
• Mean following headway.

• Driver reaction time.
• Critical gap for lane changing.
• Minimum separation under stop-and-go conditions.
Signalized Intersections:
• Startup lost time.

• Queue discharge headway.
• Gap acceptance for unprotected left turns.
For example, in CORSIM, the two parameters that

Mean headway is a good global
most globally affect capacity are “headway factor by parameter for calibrating capacity. It
vehicle type” (record type 58) and “car following may have different names in
different software.
11
The headways for the first three vehicles are discarded.
58
sensitivity factor” (record type 68). There are numerous other parameters in CORSIM that
affect capacity; however, they mostly apply only to specific conditions (e.g., “acceptable
gap in near-side cross traffic for vehicles at a stop sign” (record type 142)).
For the fine-tuning phase in CORSIM, the link-specific capacity calibration parameter is
“mean queue discharge headway,” which can be found in record type 11 for city streets,
and in record type 20 for freeways (“link-specific car-following sensitivity multiplier”).
5.3.4 Set Calibration Objective Function
It is recommended that the analyst seek to minimize the mean square error (MSE)
between the model estimates of maximum achievable flow rates and the field
measurements of capacity. The MSE is the sum of the squared errors averaged over
several model run repetitions. Each set of repetitions has a single set of model parameter
values p with different random number seeds12 for each repetition within the set.13
Select a set of model parameters p to minimize:
1
MSE = ∑
R r
( M ltpr − Fl ) 2 (3)
Subject to:
pmmin ≤ pm ≤ pmmax for all user-adjustable model parameters pm
where:
MSE = mean square error

Mltpr = model estimate of queue discharge flow rate (capacity) at location l and time t,
using parameter set p for repetition r
Fl = field measurement of queue discharge flow rate (capacity) at location l
12
Microsimulation models assign driver-vehicle characteristics from statistical distributions using
random numbers. The sequence of random numbers generated depends on the initial value of the
random number (random number seed). Changing the random number seed produces a different
sequence of random numbers, which, in turn, produces different values of driver-vehicle
characteristics.
13
Some researchers have calibrated models using the percent MSE to avoid the unintended
weighting effect when combining different measures of performance (such as volumes and travel
time) into one measure of error. The percent MSE divides each squared error by the field-
measured value. The effect of using percent MSE is to place greater weight on large percentage
errors rather than on large numerical errors. The simple MSE is recommended for calibration
because it is most sensitive to large volume errors.
59
R = number of repetitive model runs with fixed parameter values pm and different
random number seeds14
pm = value of model parameter number m
pmmin, pmmax = user-specified limits to the allowable range of parameter values pm (limits
are necessary to avoid solutions that violate the laws of physics, vehicle performance
limits, and driver behavior extremes)
5.3.5 Perform Search for Optimal Parameter Value
The analyst must now find the capacity adjustment factor(s) p that minimizes the MSE
between the model and the field measurements of capacity. The calibration problem is a
nonlinear least-squares optimization problem.
Since simulation models are complex, it is not usually possible to formulate the models as
a closed-form equation for which traditional calculus techniques can be applied to find a
minimum value solution. It is necessary to use some sort of search algorithm that relies on
multiple operations of the simulation model, plotting of the output results as points, and
searching between these points for the optimal solution. Search algorithms are required to
find the optimal solution to the calibration problem.
There are many software programs available for identifying the optimal combination of
calibration parameters for minimizing the squared error between the field observations
and the simulation model. The Optimization Technology Center Web site of the Argonne
National Laboratory and Northwestern University lists several software programs for
nonlinear least-squares parameter estimation. See the following:
www-fp.mcs.anl.gov/otc/Guide/SoftwareGuide/Categories/nonlinleastsq.html
Appendix D describes a few simple search algorithms that can be applied manually or
with the aid of spreadsheets. Figure 7 illustrates how a single parameter search (m = 1)
might look using one of these simple search algorithms for the number of model runs (R)
set to 4.
14
Since the objective is to minimize error, dividing by a constant R (the number of repetitions)
would have no effect on the results. However, R is included in the objective function to
emphasize to the analyst the necessity of running the model several times with each parameter
set.
60
Run #1
Run #2
Run #3
e
ap
Run #4
Sh
on
Mean Square Error
ti
nc
Fu
ly
ke
Li
0 0.1 0.4 0.7 1.0
Mean Headway (Parameter)
Figure 7. Minimization of MSE.
5.3.6 Fine-Tune the Calibration
Once the optimal global capacity calibration parameter values have been identified, there
will still be some locations where model performance deviates a great deal from the field
conditions. Therefore, the next phase is to fine-tune the predicted capacity to match the
location-specific measurements of capacity as closely as possible.
Link-specific capacity adjustments account for

Link-specific adjustment factors are
roadside factors that affect capacity, but are not often still necessary after global
typically coded in the microsimulation network calibration; however, they should be
input data set (such as presence of onstreet parking, used sparingly.
numerous driveways, or narrow shoulders). Most
simulation software programs have link-specific capacity (or headway) adjustment factors
that apply only to the subject link. These capacity adjustment factors are used to fine-tune
the model calibration. Link-specific adjustment factors should be used sparingly since
they are not behavior-based. They are fixed adjustments that will be carried through all
future runs of the simulation model.
61
5.4 Step 2: Route Choice Calibration
Once the analyst is satisfied that the model reproduces as closely as possible the field-
measured capacities, the next step is to then calibrate the route choice parameters in the
model to better match the observed flows. The temporary demand adjustments used in
the previous capacity calibration step are reversed. The model-predicted volumes are then
compared to the field counts and the analyst adjusts the route choice algorithm
parameters until the best volume fit is achieved.15
If the model network consists of only a single facility, then no route choice calibration is
possible or needed. This step is skipped.16 This step is also skipped for microsimulation
software that does not have route choice capabilities.
Route choice calibration proceeds in two phases: (1) global calibration and (2) link-specific
fine-tuning.
5.4.1 Global Calibration
Global calibration of route choice consists of the application of a route choice algorithm
and associated parameters.17 The specific parameters vary by algorithm and software
program, but usually involve weightings placed on the actual cost and travel time for each
route. Additional parameters may be related to the familiarity of the driver with each
route and the amount of error in the driver’s perception of the cost and time for each
route. The analyst must review the software documentation and select one or two of these
parameters for calibration. Global calibration then proceeds through the same process as
was used to calibrate capacity. The MSE between the field counts and the model volume
estimates is minimized using one of the available nonlinear optimization techniques.
5.4.2 Fine-Tuning
Once the global calibration has been completed, link-specific adjustments to cost or speed
are made during the fine-tuning phase. The fine-tuning has been completed when the
calibration targets for volumes have been met (see section 5.6 on calibration targets).
15
The specific route choice algorithm parameters will vary by software program. They generally
relate to the driver’s awareness of, perception of, and sensitivity to travel time, delay, and the cost
of alternate routes.
16
For a single-facility network, if there are still some remaining volume errors after the capacity
calibration step, then the input demands should be checked for errors.
17
Some software programs allow selection of the algorithm and its associated parameters.
62
5.5 Step 3: System Performance Calibration
In this last step of the calibration, the overall traffic performance predicted by the fully
functioning model is compared to the field measurements of travel time, queue lengths,
and the duration of queuing. The analyst refines link free-flow speeds and link capacities
to better match the field conditions. Since changes made at this step may compromise the
prior two steps of calibration, these changes should be made sparingly. The next section
suggests calibration targets for this last step of the review.
5.6 Calibration Targets

The objective of model calibration is to obtain the best match possible between model
performance estimates and the field measurements of performance. However, there is a
limit to the amount of time and effort anyone can put into eliminating error in the model.
There comes a point of diminishing returns where large investments in effort yield small
improvements in accuracy. The analyst needs to know when to stop. This is the purpose
of adopting calibration targets for the model.
Calibration targets are developed based on the minimum performance requirements for
the microsimulation model, taking into consideration the available resources. The targets
will vary according to the purpose for which the microsimulation model is being
developed and the resources available to the analyst.
Table 4 provides an example of calibration targets that were developed by Wisconsin DOT
for their Milwaukee freeway system simulation model. They are based on guidelines
developed in the United Kingdom.18
18
“Traffic Appraisal in Urban Areas, Highways Agency,” Design Manual for Roads and Bridges:
Volume 12, Section 2, Department for Transport (formerly Department of Environment, Transport,
and the Regions), London, England, May 1996 (www.official-
documents.co.uk/document/deps/ha/dmrb/index.htm)
63
Table 4. Wisconsin DOT freeway model calibration criteria.
Criteria and Measures Calibration Acceptance Targets

Hourly Flows, Model Versus Observed
Individual Link Flows
Within 15%, for 700 veh/h < Flow < 2700 > 85% of cases
veh/h
Within 100 veh/h, for Flow < 700 veh/h > 85% of cases
Within 400 veh/h, for Flow > 2700 veh/h > 85% of cases
Sum of All Link Flows Within 5% of sum of all link
counts
GEH Statistic < 5 for Individual Link Flows* > 85% of cases
GEH Statistic for Sum of All Link Flows GEH < 4 for sum of all link
counts
Travel Times, Model Versus Observed

Journey Times, Network
Within 15% (or 1 min, if higher) > 85% of cases
Visual Audits
Individual Link Speeds
Visually Acceptable Speed-Flow To analyst’s satisfaction
Relationship
Bottlenecks
Visually Acceptable Queuing To analyst’s satisfaction
*The GEH statistic is computed as follows:
GEH =
(E − V ) 2 (4)
(E + V ) / 2
where:
E = model estimated volume

V = field count
Source: “Freeway System Operational Assessment,” Paramics Calibration and Validation
Guidelines (Draft), Technical Report I-33, Wisconsin DOT, District 2, June 2002.
64
Another example of suggested calibration targets is “Theil’s Inequality Coefficient,” which
is broken down into three parts, each of which provides information on the differences
between the model measures and the target measures.19
5.7 Example Problem: Model Calibration
The same example problem from the previous chapters is continued. The task is now to
calibrate the model.
A review of the results produced by the working model at the end of the model
development task indicates that there are discrepancies between the observed and
simulated traffic performance. The purpose of the calibration process is to adjust the
model parameters to better match field conditions.
Calibration for Capacity—Step 1: Global Calibration
Field Data on Capacity: The network under study is not congested. Therefore, field
measurements of the capacity values on the freeway links and saturation flows at the
traffic signals on the arterials cannot be obtained.
Two potential future bottleneck locations were selected on the study area network. The
capacities for these potential bottleneck locations were estimated using the HCM 2000
procedures. The estimated value for the saturation flow for protected movements at
signalized intersections was 1800 vehicles per hour of green per lane (vphgpl) and the
capacity of the freeway links was 2100 vphgpl.
Model Estimates of Capacity: The model estimates of capacity can be obtained from
detector measurements or from the throughput values reported at the end of the
simulation run. However, it is necessary to have upstream queues for the throughput
values to represent capacity.
Because of the lack of congestion, the existing volumes on the network links had to be
artificially increased to trigger congestion upstream of the bottleneck locations (the model
throughput volumes at these bottleneck locations would then be the model-estimated
capacities).
Select Calibration Parameters: The global parameter calibration was performed using the
following parameters:
• Mean queue discharge headway at traffic signals: The default value for this parameter
in the model is 1.8 s/veh (which is equivalent to a saturation flow of 2000 vphgpl
19
Further discussions of Theil’s Inequality Coefficient can be found in the Advanced CORSIM
Training Manual, Minnesota DOT, 2002, and in Hourdakis, J., P. Michalopoulos, and J.
Kottommannil, “Practical Procedure for Calibrating Microscopic Traffic Simulation Models,” TRB,
TRR 1852, Washington, D.C., 2003.
65
under ideal conditions). The queue discharge headway per vehicle type is randomly
assigned by the model, depending on the driver’s aggressiveness factor.
• Mean headway (or car-following sensitivity factor) for the freeway links: This
represents the minimum distance that a driver is willing to follow vehicles. The values
of the headway (or sensitivity factors) depend again on driver aggressiveness. Lower
values would cause higher throughput.
Set Calibration Objective Function: The criterion chosen to determine the optimal
parameter value was the minimization of the MSE between the model estimates of
saturation flow/capacity and the field estimates.
Table 5 shows the values of the mean queue discharge headway tested and the resulting
simulated saturation flows. Ten replications for each assumed value of the parameter
were performed to estimate the model-generated saturation flow within 5 percent of its
true value. The mean values of the saturation flows were used in the calculation of MSE.
Figure 8 shows the values of MSE for each parameter value. The results indicate that a
value of 1.9 s produces the minimum MSE for this data set.
Table 5. Impact of queue discharge headway on predicted saturation flow.
Parameter Sample Sample

Value 1 2 3 4 5 6 7 8 9 10 Mean St Dev
1.8 1925 1821 1812 1821 1861 1848 1861 1875 1763 1801 1839 45
1.9 1816 1857 1777 1809 1855 1815 1807 1786 1752 1784 1806 33
2.0 1772 1743 1720 1728 1799 1711 1693 1764 1673 1690 1730 40
Note: Each column labeled 1 through 10 represents a model run repetition using a different
random number seed.
66
7000
6423
Modeled Saturation Flow 6000
5000
4000
3338
3000
2000
1022
1000
0
1.8 1.9 2.0
Mean Queue Discharge Headway (in seconds)
Figure 8. Impact of mean queue discharge headway on MSE of modeled

saturation flow.
A similar procedure was followed for the calibration of the mean headway values for
freeway links. Model runs were performed assuming higher through volumes on the
network links.
The results indicated that the default minimum headway (car-following sensitivity factor
in CORSIM) must be reduced by 5 percent (i.e., 95 percent of its default value for the
model throughput to match the observed value of 2100 vphgpl). This value minimizes the
MSE.
Figures 9 and 10 illustrate the effect of this parameter on freeway performance. Before
calibration (figure 9), the average speed on a series of freeway links is much lower than
the speeds after calibration of the headways in the HCM 2000 (figure 10).
Calibration for Capacity—Step 2: Fine-Tuning
For this example, field capacity measurements were only available for two links on the
network—one for the freeway and one for the surface street. These measures were used in
the previous step to perform the global calibration. Given that no other field capacity data
were available, no further fine-tuning (or link-level calibration) is needed.
67
Figure 9. Average speed by minute of simulation (three links): Before calibration.
Figure 10. Average speed by minute (three links): After calibration.
68
Calibration for Route Choice
The prior calibration step has ensured that the model is predicting the correct capacities.
With the correct capacities, the model should no longer have bottlenecks at the incorrect
locations with too low or too high discharge rates. The upstream and downstream traffic
volumes on each of the links should now better match the observed counts. Any remaining
differences (assuming that error checking caught errors in the source node volumes)
should be the result of route choice errors in the model. Therefore, the purpose of route
choice calibration is to better match the counts at the non-bottleneck locations on the
network.
The particular software selected for this example problem does not predict changes in
route choice (no O-D table was provided for the model). Therefore, there is no route
choice to be calibrated.20 Assuming that the analyst is confident in the quality of the turn
counts entered into the model (and converted inside the model into equivalent turn
probabilities for routing traffic), any remaining errors in the link volumes can be corrected
by manual adjustments to midblock source and sink node volumes.
In this example problem, the freeway, ramp, and intersection counts were all adjusted at
the data preparation stage to produce an internally consistent set of volumes for every link
and node. There are no unaccounted for increases or decreases in volumes between nodes.
Thus, no midblock source/sink nodes were required. The model-predicted link volumes
based on the source node volumes and the intersection turn percentages are consistent
with the link counts (within the range of variation expected because of the random
components of the simulation model).
System Performance Calibration
Once calibration has been completed to the analyst’s satisfaction, the next step is to
calibrate the overall performance of the model.
The model predictions were compared with the field data on speed, travel time, and delay
on the freeways and arterials.
Based on this comparison, the free-flow speed distribution for the freeway was adjusted
so that the model-generated free-flow speeds were within a more narrow range than the
default distribution. This reflects commuter traffic behavior.
No other adjustments were made to the model parameters.
20
Actually, the analyst could (in theory) calibrate the percentage turns that were input at each
intersection and ramp junction to better match the observed link flows downstream from the
source nodes; however, this is currently impractical (there are too many variables to be adjusted
without the assistance of a good computerized optimization algorithm).
69
Ten repetitions of the calibrated data set were performed and the link output was
processed to produce performance measures to be compared with the field data.
The comparison is shown in table 6. The simulated values represent the mean value of the
MOEs based on the 10 model runs for the third time slice of the simulated peak hour.
Comparisons with field measurement of delay are not shown here because it was unclear
how the field data on delay were collected. In general, users must be cautious when they
compare field versus simulated delays because the delay is typically defined differently in
most software programs than in the approaches commonly used for field measurements.
The results indicate that the model satisfies the criteria for calibration (shown in table 4).
70
Table 6. System performance results for example problem.
Travel Time (minutes) Speed (mi/h) Density (vehicle/mi/lane)
Abs. % Abs. % Abs. %

Freeway Segment Field Sim. Diff. Diff. Field Sim. Diff. Diff. Field Sim. Diff. Diff.
EB west of Wisconsin off 1.44 1.47 0.03 2.3% 53.5 52.42 -1.04 1.95% 38.1 36.4 -1.64 4.3%
EB Wisconsin off to on 0.22 0.22 -0.01 2.4% 53.4 51.84 -1.56 2.93% 38.9 33.0 -5.94 15.3%
EB Wisconsin onto 1.00 0.95 -0.05 5.0% 48.1 51.29 3.18 6.61% 38.6 34.4 -4.23 11.0%
Milwaukee off
EB Milwaukee off to on 0.22 0.22 0.00 0.1% 55.0 52.24 -2.76 5.01% 30.3 31.4 1.13 3.7%
EB east of Milwaukee on 1.12 1.15 0.04 3.4% 53.8 51.96 -1.83 3.40% 34.0 33.9 -0.14 0.4%
WB east of Milwaukee off 1.11 1.13 0.02 2.1% 54.3 53.44 -0.85 1.57% 29.1 28.8 -0.25 0.9%
WB Milwaukee off to on 0.22 0.22 0.00 1.0% 55.0 52.65 -2.35 4.26% 26.5 27.1 0.59 2.2%
WB Milwaukee onto 0.96 0.94 -0.02 2.4% 50.1 51.94 1.87 3.73% 30.4 31.0 0.61 2.0%
Wisconsin off
WB Wisconsin off to on 0.22 0.22 0.00 0.8% 55.0 52.46 -2.54 4.61% 28.5 29.6 1.09 3.8%
WB west of Wisconsin on 1.42 1.48 0.06 4.0% 54.2 52.10 -2.09 3.85% 32.1 32.9 0.78 2.4%
71
Average 2.4% 3.79% 4.6%
Travel Time (minutes)

Abs. %
Arterial Segment Field Sim. Diff. Diff.
SB Milwaukee between 0.44 0.43 -0.01 3.0%
ramps
NB Milwaukee between 0.47 0.48 0.01 3.1%
ramps
SB Wisconsin between ramps 0.57 0.58 0.01 2.4%
NB Wisconsin between 0.74 0.66 -0.08 11.0%
ramps
Average 4.9%
6.0 Alternatives Analysis
Project alternatives analysis is the sixth task in the 1

Project Scope
microsimulation analysis process. It is the reason for

See Chapter 1
- Select approach
- Select model
actual modeling
developing and calibrating the microsimulation model. 2 Data Collection

- Traffic volumes
The lengthy model development process has been - Field observations

3 - Input data
completed and now it is time to put the model to work.

See Chapter 3

- Review Animation
The analysis of project alternatives involves the Working Model

Before Calibration
forecasting of the future demand for the base case and

Compare Model
the testing of various project alternatives against this Acceptable No

Calibration
baseline future demand. The analyst must run the

Match
See Chapter 5
Yes
model several times, review the output, extract relevant Calibrated Model
6 - Forecast Demand
statistics, correct for biases in the reported results, and - Base Case
- Project Alternatives
See Chapter 6
Model
Application
perform various analyses of the results. These analyses 7

Final Report
- Key Results
See Chapter 7
may include hypothesis testing, computation of

confidence intervals, and sensitivity analyses to further support the conclusions of the
analysis.
The alternatives analysis task consists of several steps:
1. Development of Baseline Demand Forecasts.

2. Generation of Project Alternatives for Analysis.
3. Selection of Measures of Effectiveness.
4. Model Application (Runs).
5. Tabulation of Results.
6. Evaluation of Alternatives.
6.1 Baseline Demand Forecast

This step consists of establishing the future level of demand to be used as a basis for
evaluating project alternatives.
6.1.1 Demand Forecasting
Forecasts of future travel demand are best obtained from a travel demand model. These
models require a great deal of effort and time to develop and calibrate. If one does not
already exist, then the analyst may seek to develop demand forecasts based on historic
growth rates. A trend-line forecast might be made, assuming that the recent percentage of
growth in traffic will continue in the future. These trend-line forecasts are most reliable for
relatively short periods of time (5 years or less). They do not take into account the
potential of future capacity constraints to restrict the growth of future demand. Additional
information and background regarding the development of traffic data for use in highway
73
planning and design may be found in National Cooperative Highway Research Program
(NCHRP) Report 255, Highway Traffic Data for Urbanized Area Project Planning and Design.
6.1.2 Constraining Demand to Capacity
Regardless of which method is used to estimate future demand (regional model or trend
line), care must be taken to ensure that the forecasts are a reasonable estimate of the actual
amount of traffic that can arrive within the analytical period at the study area. Regional
model forecasts are usually not well constrained to system capacity and trend-line
forecasts are totally unconstrained. Appendix F provides a method for constraining future
demands to the physical ability of the transportation system to deliver the traffic to the
microsimulation model study area.
6.1.3 Allowance for Uncertainty in Demand Forecasts
All forecasts are subject to uncertainty. It is risky to design a road facility to a precise
future condition given the uncertainties in the forecasts. There are uncertainties in both
the probable growth in demand and the available capacity that might be present in the
future. Slight changes in the timing or design of planned or proposed capacity
improvements outside of the study area can significantly change the amount of traffic
delivered to the study area during the analytical period. Changes in future vehicle mix
and peaking can easily affect capacity by 10 percent. Similarly, changes in economic
development and public agency approvals of new development can significantly change
the amount of future demand. Thus, it is good practice to explicitly plan for a certain
amount of uncertainty in the analysis. This level of uncertainty is the purpose of
sensitivity testing (explained in a separate section below).
6.2 Generation of Alternatives

In this step, the analyst generates improvement alternatives based on direction from the
decisionmakers and through project meetings. They will probably reflect operational
strategies and/or geometric improvements to address the problems identified based on
the baseline demand forecasts. The specifics of alternatives generation are beyond the
scope of this report. Briefly, they consist of:
• Performing a microsimulation analysis1 of the baseline demand forecast to identify the

deficiencies.2
• Identifying alternative improvements that solve one or more of the identified problems.3
1
A sketch-planning analysis using HCM techniques can be performed, if desired, to identify
conceptual project alternatives prior to the development of a calibrated simulation model. The
calibrated simulation model can then be used to refine the conceptual plans.
2
Deficiencies are identified by comparing the measured performance against the project objectives
and agency performance standards identified in task 1 (Project Scope).
74
6.3 Selection of Measures of Effectiveness (MOEs)
MOEs are the system performance statistics that
When reviewing model output, the
best characterize the degree to which a particular analyst should look at a few key
alternative meets the project objectives (which were systemwide performance measures
determined in the Project Scope task). Thus, the plus “hot spots” (indicators of
appropriate MOEs are determined by the project localized breakdowns in the system).
objectives and agency performance standards rather
than what is produced by the model. This section, however, discusses what output is
typically produced by microsimulation software so that the analyst can appreciate what
output might be available for constructing the desired MOEs.
Microsimulation, by its very nature, can bury the analyst in detailed microscopic output.
The key is to focus on a few key indicators of system performance and localized
breakdowns in the system (locations where queues are interfering with systems
operation).
6.3.1 Candidate MOEs for Overall System Performance
As explained above, the selection of MOEs should be driven by the project objectives and
the agency performance standards; however, many MOEs of overall system performance
can be computed directly or indirectly from the following three basic system performance
measures:
• Vehicle-miles traveled (VMT) or vehicle-kilometers traveled.
• Vehicle-hours traveled (VHT).
• Mean system speed.
These three basic performance measures can also be supplemented with other model
output, depending on the objectives of the analysis. For example, total system delay is a
useful overall system performance measure for comparing the congestion-relieving
3
In support of the purpose and scope of the project, this identification of alternatives must take
into consideration the range of demand conditions, weather conditions, incidents, and
operational management strategies. Additional information and background may be found in the
following documents:
Bunch, J., S. Hatcher, J. Larkin, G. Nelson, A. Proper, D. Roberts, V. Shah, and K. Wunderlich,
Incorporating ITS Into Corridor Planning: Seattle Case Study, ITS Electronic Document Library
(EDL)#11303, August 1999.
Wunderlich, K., Incorporating ITS Into Planning Analyses: Summary of Key Findings From a Seattle
2020 Case Study, Report No. FHWA-OP-02-031, May 2002.
75
effectiveness of various alternatives. The number of stops is a useful indicator for signal
coordination studies.
6.3.2 Candidate MOEs for Localized Problems
In addition to evaluating overall system performance, the analyst should also be evaluating
if and where there are localized system breakdowns (“hot spots”). A hot spot may be
indicated by a persistent short queue that lasts too long, a signal phase failure (green time
that fails to clear all waiting vehicles), or a blocked link (queue that backs up onto an
upstream intersection).
A blocked link is the most significant indicator of localized breakdowns. A queue of

vehicles that fills a link and blocks an upstream intersection can have a significant impact
on system performance. A link queue overflow report can be developed to identify the
links and times during the simulation period when the computed queue of vehicles
equaled (and, therefore, probably actually exceeded) the storage capacity of the link.4
Signal phase failures, where the provided green time is insufficient to clear the queue,
indicate potential operational problems if the queues continue to build over several cycles.
The analyst should develop a signal phase failure report to indicate when and where
signal green times are not sufficient to clear out all of the waiting queues of vehicles during
each cycle.
At the finest level of detail, the analyst may wish to develop a report of the presence of
persistent queues of a minimum length. This “hot spot” report would identify persistent
queues on a lane-by-lane basis of a minimum number of vehicles that persist for a minimum
amount of time. This report points the analyst to locations of persistent long queues (even
those that do not overflow beyond the end of a link) during the simulation period.
6.3.3 Choice of Average or Worst Case MOEs
Microsimulation models employ random numbers to represent the uncertainty in driver

behavior in any given population of drivers. They will produce slightly different results
each time they are run, with a different random number seed giving a different mix of
driver behaviors. The analyst needs to determine if the alternatives should be evaluated
based on their average predicted performance or their worst case predicted performance.
The average or mean performance is easy to compute and interpret statistically. The
analyst runs the model several times for a given alternative, using different random
number seeds each time. The results of each run are summed and averaged. The standard
deviation of the results can be computed and used to determine the confidence interval for
the results.
4
This report will be less useful if the analyst has split long sections of roadway into arbitrary short
links for other reporting purposes. The result may be many “false alarms” of blocked links that do
not actually obstruct an upstream intersection.
76
The worst case result for each alternative is slightly more difficult to compute. It might be
tempting to select the worst case result from the simulation model runs; however, the
difficulty is that the analyst has no assurance that if the model were to be run a few more
times, the model might not get an even worse result. Thus, the analyst never knows if he
or she has truly obtained the worst case result.
The solution is to compute the 95th percentile probable worst outcome based on the mean
outcome and an assumed normal distribution for the results.5,6 The equation below can be
used to make this estimate:
95% Worst Re sult = m + 1.64 • s (5)
where:
m = mean observed result in the model runs

s = standard deviation of the result in the model runs
6.4 Model Application

The calibrated microsimulation model is applied in this step to compute the MOEs for
each alternative. Although model operation is adequately described in the software user’s
guide, there are a few key considerations to be taken into account when using a
microsimulation model for the analysis of alternatives.
6.4.1 Requirement for Multiple Repetitions
Microsimulation models rely on random numbers to generate vehicles, select their

destination and route, and determine their behavior as they move through the network.
No single simulation run can be expected to reflect any specific field condition. The results
from individual runs can vary by 25 percent and higher standard deviations may be
expected for facilities operating at or near capacity.7 It is necessary to run the model
several times with different random number seeds8 to get the necessary output to determine
5
Many statistical phenomena approximate a normal distribution at large sample sizes. Even
though most microsimulation analysts usually work with relatively few model repetitions, the
assumption of normal distribution is usually good enough for most analyses.
6
Note that when computing the 95th percentile queue on the macroscopic level, it is typically
assumed that the arrival of the vehicles are Poisson distributed. Microsimulation models predict
the arrival patterns of vehicles, so the Poisson distribution assumption is not necessary when
estimating 95th percentile queues using microsimulation data.
7
See example in appendix B.
8
Those with a more sophisticated statistical aptitude may elect to use variance reduction
techniques that employ a single common random number seed to reduce the number of required
repetitions. These techniques are described in Joshi, S.S., and A.K. Rathi, “Statistical Analysis and
(Footnote continued on next page...)
77
mean, minimum, and maximum values. The analyst must then post-process the runs to
obtain the necessary output statistics (see appendix B for guidance on the computation of
confidence intervals and the determination of the minimum number of repetitions of
model runs).
6.4.2 Exclusion of Initialization Period
The initialization (warmup) period before the system reaches equilibrium for the
simulation period should be excluded from the tabulated statistics (see appendix C for
guidance on identifying the initialization period).
6.4.3 Avoiding Bias in the Results
The simulation geographic and temporal limits should be sufficient to include all
congestion related to the base case and all of the alternatives. Otherwise, the model will
not measure all of the congestion associated with an alternative, thus causing the analyst
to underreport the benefits of an alternative. See the subsection below on the methods for
correcting congestion bias in the results.
6.4.4 Impact of Alternatives on Demand
The analyst should consider the potential impact of alternative improvements on the base
case forecast demand. This should take into consideration the effects of a geometric
alternative, an operational strategy, and combinations of both. The analyst should then
make a reasonable effort to incorporate any significant demand effects within the
microsimulation analysis.9
6.4.5 Signal/Meter Control Optimization
Most simulation models do not currently optimize signal timing or ramp meter controls.
Thus, if the analyst is testing various demand patterns or alternatives that significantly
change the traffic flows on specific signalized streets or metered ramps, he or she may
need to include a signal and meter control optimization substep within the analysis of
each alternative. This optimization might be performed offline using a macroscopic signal
timing or ramp metering optimization model. Or, the analyst may run the simulation
model multiple times with different signal settings and manually seek the signal setting
that gives the best performance.
Validation of Multi-Population Traffic Simulation Experiments,” Transportation Research

Record 1510, TRB, Washington, DC, 1995.
9
The analyst might consult Predicting Short-Term and Long-Term Air Quality Effects of Traffic-Flow
Improvement Projects, Final Report, NCHRP Project 25-21, TRB, Washington, DC, 2003.
78
6.5 Tabulation of Results
Microsimulation models typically produce two types of output: (1) animation displays
and (2) numerical output in text files.10 The animation display shows the movement of
individual vehicles through the network over the simulation period. Text files report
accumulated statistics on the performance of the network. It is crucial that the analyst
reviews both numerical and animation output (not just one or the other) to gain a
complete picture of the results.11 This information can then be formatted for inclusion in
the final report.
6.5.1 Reviewing Animation Output
Animation output is powerful in that it enables the analyst to quickly see and qualitatively
assess the overall performance of the alternative. However, the assessment can only be
qualitative. In addition, reviewing animation results can be time-consuming and tedious
for numerous model repetitions, large networks, and long simulation periods. The analyst
should select one or more model run repetitions for review and then focus his or her
attention on the key aspects of each animation result.
Selection of Representative Repetition
The analyst has to decide whether he or she will review the typical case output, or the
worst case output, or both. The typical case might give an indication of the average
conditions for the simulation period. The worst case is useful for determining if the
transportation system will experience a failure and for viewing the consequences of that
failure.
The next question that the analyst must decide is how to identify which model repetition
represents typical conditions and which repetition reflects worst case conditions. The total
VHT may be a useful indicator of typical and worst case conditions. The analyst might
also select other measures, such as the number of occurrences of blocked links (links with
queue overflows) or delay.
If VHT is selected as the measure and the analyst wishes to review the typical case, then
he or she would pick the model run repetition that had the total VHT that came closest to
falling within the median of the repetitions (50 percent of the repetitions had a VHT less
10
Some software programs also produce static graphs that can be very useful for gaining insight
into the input or the results.
11
Animation output shows the results from just one run of the simulation model. Drawing
conclusions about traffic system performance from reviewing just one animation result is like
trying to decide if the dice are fair from just one roll. One needs to roll the dice several times,
tabulate the results, and compute the mean and standard deviation of the results to have the
information needed to determine if the dice are fair.
79
than that, and 50 percent has more VHT). If the analyst wished to review the worst case,
then he or she would select the repetition that had the highest VHT.
The pitfall of using a global summary statistic (such as VHT) to select a model run
repetition for review is that overall average system conditions does not mean that each
link and intersection in the system is experiencing average conditions. The median VHT
repetition may actually have the worst performance for a specific link. If the analyst is
focused on a specific link or intersection, then he or she should select some statistic related
to vehicle performance on that specific link or intersection for selecting the model run
repetition for review.
Review of Key Events in Animation
The key event to look for in reviewing animation is the formation of persistent queues.
Cyclical queues at signals that clear each cycle are not usually as critical unless they block
some other traffic movement. The analyst should not confuse the secondary impact of
queues (one queue blocking upstream movement and creating a secondary queue) with
the root cause of the queuing problem. Eliminating the cause of the first or primary queue
may eliminate all secondary queuing. Thus, the analyst should focus on the few minutes
just prior to formation of a persistent queue to identify the causes of the queuing.
6.5.2 Numerical Output
Microsimulation software reports the numerical

To avoid pitfalls, it is critical that the
results of the model run in text output files called analyst understand how the results
“reports.” Unless the analyst is reviewing actual have been accumulated by the
vehicle trajectory output, the output reports are software.
almost always a summary of the vehicle activity
simulated by the model. The results may be summarized over time and/or space. It is
critical that the analyst understands how the software has accumulated and summarized
the results to avoid pitfalls in interpreting the numerical output.
Microsimulation software may report instantaneous rates (such as speed) observed at

specific instances of time, or may accumulate the data over a longer time interval and
either report the sum, the maximum, or the average. Depending on the software program,
vehicle activity that occurs between time steps (such as passing over a detector) may not
be tallied, accumulated, or reported.
Microsimulation software may report the results for

It is good practice to cross-check
specific points on a link in the network or output to ensure that the analyst
aggregated for the entire link. The point-specific understands how it is computed by
output is similar to what would be reported by the software.
detectors in the field. Link-specific values of road
performance are accumulated over the length of the link and, therefore, will vary from the
point data.
80
The key to correctly interpreting the numerical output of a microsimulation model is to
understand how the data were accumulated by the model and summarized in the report.
The report headings may give the analyst a clue as to the method of accumulation used;
however, these short headings cannot usually be relied on. The method of data
accumulation and averaging can be determined through a detailed review of the model
documentation of the reports that it produces, and, if the documentation is lacking, by
querying the software developers themselves.
An initial healthy skepticism is valuable when reviewing reports until the analyst has
more experience with the software. It helps to cross-check output to ensure that the
analyst understands how the data is accumulated and reported by the software.
6.5.3 Correcting Biases in the Results
To make a reliable comparison of the alternatives, it is important that vehicle congestion

for each alternative be accurately tabulated by the model. This means that congestion
(vehicle queues) should not extend physically or temporally beyond the geographic or
temporal boundaries of the simulation model. Congestion that overflows the time or
geographic limits of the model will not normally be reported by the model, which can bias
the comparison of alternatives.
The tabulated results should also exclude the initial and unrealistic initialization period
when vehicles are first loaded on the network.
Ideally, the simulation results for each alternative would have the following characteristics:
• All of the congestion begins and ends within the simulation study area.
• No congestion begins or ends outside of the simulation period.
• No vehicles are unable to enter the network from any zone (source node) during any
time step of the simulation.
It may not always be feasible to achieve all three of these conditions, so it may be
necessary to adjust for congestion that is missing from the model tabulations of the results.
Correction of Output for Blocked Vehicles
If simulation alternatives are severely congested, then the simulation may be unable to
load vehicles onto the network. Some may be blocked from entering the network on the
periphery. Some may be blocked from being generated on internal links. These blocked
vehicles will not typically be included in the travel time (VHT) or delay statistics for the
model run.12 The best solution is to extend the network back to include the maximum back
12
The analyst should verify with the software documentation or developer how statistics on
blocked vehicles are accumulated in the travel time and delay summaries.
81
of the queue. If this is not feasible, then the analyst should correct the reported VHT to
account for the unreported delay for the blocked vehicles.
Microsimulation software will usually tally the excess queue that backs up outside the
network as “blocked” vehicles (vehicles unable to enter the network) for each time step.
The analyst totals the number of software-reported blocked vehicles for each time step of
the simulation and multiplies this figure by the length of each time step (in hours) to
obtain the vehicle-hours of delay. The delay resulting from blocked vehicles is added to
the model-reported VHT for each model run.
Correction of Output for Congestion Extending Beyond the End of the

Simulation Period
Vehicles queues that are present at the end of the simulation period may affect the
accumulation of total delay and distort the comparison of alternatives (cyclical queues at
signals can be neglected). The “build” project alternative may not look significantly better
than the “no-build” option if the simulation period is not long enough to capture all of the
benefits. The best solution is to extend the simulation period until all of the congestion
that built up over the simulation period is served. If this is not feasible, the analyst can
make a rough estimate of the uncaptured residual delay by computing how many vehicle-
hours it would take to clear the queue using the equation given below:
Q2
VHT (Q ) = (6)
2•C
where:
VHT(Q) = extra VHT of delay attributable to a queue present at the end of the simulation
period
Q = number of vehicles remaining in the queue at the end of the simulation period
C = discharge capacity of the bottleneck in veh/h
The equation computes the area of the triangle created by the queue and the discharge
capacity after the end of the simulation period (see figure 11 below):
82
Vehicles
ity
ac
C ap
Queue
Time
Figure 11. Computation of uncaptured residual delay at the end of the

simulation period.
Note that this is not a complete estimate of the residual delay since it ignores the
interaction of vehicles left over from the simulation period that interfere with traffic
arriving during later time periods.
6.6 Evaluation of Alternatives
This step involves the evaluation of alternatives using the microsimulation model results.
First, the interpretation of system performance results is discussed. Then, various analyses
are discussed for assessing the robustness of the results. The ranking of alternatives and
cost-effectiveness analyses are well documented in other reports and are not discussed
here.
6.6.1 Interpretation of System Performance Results
This subsection explains how to interpret the differences between alternatives for the three
basic system performance measures (VMT, VHT, and system speed).
VMT provides an indication of total travel demand (in terms of both the number of trips
and the length of the trips) for the system.13 Increases in VMT generally indicate increased
demand (car, bus, and truck). VMT is computed as the product of the number of vehicles
13
Person-miles traveled (PMT), if available, is the preferred measure of travel demand since it takes
into account the number of people in each vehicle or mode.
83
traversing a link and the length of the link, summed over all links. Since VMT is computed
as a combination of the number of vehicles on the system and their length of travel, it can
be influenced both by changes in the number of vehicles and changes in the trip lengths
during the simulation period. The following can cause changes in VMT between one
alternative and the next:
• Random variations between one alternative and the next (use of the same random
number seed in both runs can reduce or eliminate the variation).
• Changed demand.
• Changed congestion causing vehicles to take different paths (increased congestion

may also reduce the number of vehicles that can complete their trip during the
simulation period, also decreasing VMT).
• Inability of the model to load the coded demand onto the network within the
simulation period (increased congestion may force the model to store some of the
vehicle demand off-network because of bottlenecks at loading points; in this situation,
increased congestion may actually lower VMT since stored vehicles cannot travel any
distance during the simulation period).
VHT provides an estimate of the amount of time expended traveling on the system.14
Decreases in VHT generally indicate improved system performance and reduced traveling
costs for the public. VHT is computed as the product of the link volume and the link travel
time, summed over all links. Since VHT is computed as a combination of the number of
vehicles and the time spent traveling, it can be influenced both by changes in demand (the
number of vehicles) and changes in congestion (travel time). Changes in VHT between
one alternative and the next can be caused by the following:
• Random variations between one alternative and the next.
• Changed demand.
• Changed congestion.
• Demand stored off-network because of excessive congestion at load points (increased

congestion that causes demand to be stored off-network may reduce VHT if the
software does not accumulate delay for vehicles stored off-network).
Mean system speed is an indicator of overall system performance. Higher speeds

generally indicate reduced travel costs for the public. The mean system speed is computed
from the VMT and VHT as follows:
14
Person-hours traveled (PHT), if available, provides a preferred measure of travel delay since it
takes into account the number of people delayed in each vehicle. This is especially important for
comparing the performance of HOV alternatives.
84
Mean System Speed = VMT/VHT (7)
Changes in the mean system speed between one alternative and the next can be caused by
the following:
• Random variations between one alternative and the next.
• Changed link speeds and delays caused by congestion.
• Changes in vehicle paths caused by congestion.
• Changes in vehicle demand.
• Changes in the number of vehicles stored off-network caused by excessive congestion

at loading points.
Total system delay, if available, is useful because it reports the portion of total travel time
that is most irritating to the traveling public. However, defining “total system delay” can
be difficult. It depends on what the analyst or the software developer considers to be ideal
(no delay) travel time. Some sources consider delay to include only the delay caused by
increases in demand above some base uncongested (free-flow) condition. Others add in
the base delay occurring at traffic control devices, even at low-flow conditions. Some
include acceleration and deceleration delay. Others include only stopped delay. The
analyst should consult the software documentation to ensure the appropriate use and
interpretation of this measurement of system performance.
6.6.2 Hypothesis Testing
When the microsimulation model is run several times for each alternative, the analyst may
find that the variance in the results for each alternative is close to the difference in the
mean results for each alternative. How is the analyst to determine if the alternatives are
significantly different? To what degree of confidence can the analyst claim that the
observed differences in the simulation results are caused by the differences in the
alternatives and not just the result of using different random number seeds? This is the
purpose of statistical hypothesis testing. Hypothesis testing determines if the analyst has
performed an adequate number of repetitions for each alternative to truly tell the
alternatives apart at the analyst’s desired level of confidence. Hypothesis testing is
discussed in more detail in appendix E.
6.6.3 Confidence Intervals and Sensitivity Analysis
Confidence intervals are a means of recognizing the inherent variation in microsimulation

model results and conveying them to the decisionmaker in a manner that clearly indicates
the reliability of the results. For example, a confidence interval would state that the mean
delay for alternative x lies between 35.6 s and 43.2 s, with a 95-percent level of confidence.
If the 95-percent confidence interval for alternative y overlaps that of x, the decisionmaker
85
would consider the results to be less favorable for either alternative. They could be
identical. Computation of the confidence interval is explained in appendix B.
A sensitivity analysis is a targeted assessment of the reliability of the microsimulation

results, given the uncertainty in the input or assumptions. The analyst identifies certain
input or assumptions about which there is some uncertainty and varies them to see what
their impact might be on the microsimulation results.
Additional model runs are made with changes in demand levels and key parameters to
determine the robustness of the conclusions from the alternatives analysis. The analyst
may vary the following:
• Demand.
• Street improvements assumed to be in place outside of the study area.
• Parameters for which the analyst has little information.
A sensitivity analysis of different demand levels is particularly valuable when evaluating

future conditions. Demand forecasts are generally less precise than the ability of the
microsimulation model to predict their impact on traffic operations. A 10-percent change
in demand can cause a facility to go from 95 percent of capacity to 105 percent of capacity,
with a concomitant massive change in the predicted delay and queuing for the facility.
The analyst should estimate the confidence interval for the demand forecasts and test the
microsimulation at the high end of the confidence interval to determine if the alternative
still operates satisfactorily at the potentially higher demand levels.
The analyst should plan for some selected percentage above and below the forecasted
demand to allow for these uncertainties in future conditions. The analyst might consider
at least a 10-percent margin of safety for the future demand forecasts. A larger range
might be considered if the analyst has evidence to support the likelihood of greater
variances in the forecasts.
To protect against the possibility of both underestimates and overestimates in the

forecasts, the analyst might perform two sensitivity tests—one with 110 percent of the
initial demand forecasts and the other with 90 percent of the initial demand forecasts—for
establishing a confidence interval for probable future conditions.15
Street improvements assumed to be in place outside the simulation study area can also
have a major impact on the simulation results by changing the amount of traffic that can
15
Note that the percentage confidence interval (such as a 95 percent confidence interval) has not
been stated here, so it cannot be claimed that there is a certain probability of the true value falling
within this 10 percent range. This is merely a sensitivity test of the impact of the demands being
10 percent lower or 10 percent higher than that forecast, without knowing the likelihood of it
actually happening.
86
enter or exit the facilities in the study area. Sensitivity testing would change the assumed
future level of demand entering the study area and the assumed capacity of facilities
leaving the study area to determine the impact of changes in the assumed street
improvements.
The analyst may also run sensitivity tests to determine the effects of various assumptions
about the parameter values used in the simulation. If the vehicle mix was estimated,
variations in the percentage of trucks might be tested. The analyst might also test the
effects of different percentages of familiar drivers in the network.
6.6.4 Comparison of Results to the HCM
It is often valuable when explaining microsimulation model results to the general public
to report the results in terms of HCM levels of service. However, the analyst should be
well aware of the differences between the HCM and the microsimulation analysis when
making these comparisons.
Delay and Intersection Level of Service (LOS)
Delay is used in the HCM to estimate the LOS for signalized and unsignalized
intersections. There are distinctions in the ways microsimulation software and the HCM
define delay and accumulate it for the purpose of assessing LOS.
The HCM bases its LOS grades for intersections on estimates of mean control delay for the
highest consecutive 15-min period within the hour. If microsimulation output is to be
used to estimate LOS, then the results for each run must be accumulated over a similar 15-
consecutive-minute time period and averaged over several runs with different random
number seeds to achieve a comparable result.
This still may not yield a fully comparable result, because all microsimulation models
assign delay to the segment in which it occurs. For example, the delay associated with a
single approach to a traffic signal may be parceled out over several upstream links if the
queues extend beyond one link upstream from the intersection. Thus, when analysts seek
to accumulate the delay at a signal, they should investigate whether the delay/queues
extend beyond the single approach links to the signal.
Finally, the HCM does not use total delay to measure signal LOS. It uses “control delay.”
This is the component of total delay that results when a control signal causes a lane group
to reduce speed or to stop. It is measured by comparison with the uncontrolled condition.
The analyst needs to review the software documentation and seek additional
documentation from the software vendor to understand how delay is computed by the
software.
87
Density and Freeway/Highway LOS
If microsimulation model reports of vehicle density are to be reported in terms of their

LOS implications, it is important to first translate the densities reported by the software
into the densities used by the HCM to report LOS for uninterrupted flow facilities.16
HCM 2000 defines freeway and highway LOS based on the average density of passenger
car equivalent vehicles in a section of highway for the peak 15-min period within an hour.
For ramp merge and diverge areas, only the density in the rightmost two lanes is
considered for LOS. For all other situations, the density across all lanes is considered.
Trucks and other heavy vehicles must be converted to passenger car equivalents using the
values contained in the HCM according to vehicle type, facility type, section type, and
grade.
Queues
HCM 2000 defines a queue as: “A line of vehicles, bicycles, or persons waiting to be served
by the system in which the flow rate from the front of the queue determines the average
speed within the queue. Slowly moving vehicles or people joining the rear of the queue
are usually considered part of the queue.” These definitions are not implementable within
a microsimulation environment since “waiting to be served” and “slowly” are not easily
defined. Consequently, alternative definitions based on maximum speed, acceleration,
and proximity to other vehicles have been developed for use in microsimulation.
Note also that for most microsimulation programs, the number of queued vehicles
counted as being in a particular turn-pocket lane or through lane cannot exceed the storage
capacity of that lane. Any overflow is reported for the upstream lane and link where it
occurs, not the downstream cause of the queue. Unlike macroscopic approaches that
assign the entire queue to the bottleneck that causes it, microsimulation models can only
observe the presence of a queue; they currently do not assign a cause to it. So, to obtain the
95-percent queue length, it may be necessary to temporarily increase the length of the
storage area so that all queues are appropriately tallied in the printed output.
6.7 Example Problem: Alternatives Analysis

The same example problem from the previous chapters is continued here. The task now is
to apply the calibrated model to the analysis of the ramp metering project and its
alternatives.
Step 1: Baseline Demand Forecast
A 5-year forecast was estimated using a straight-line growth approach assuming 2 percent
growth per year uncompounded. The result was a forecasted 10-percent increase in traffic
16
Note that density is NOT used as an LOS measurement for interrupted flow facilities, such as city
streets with signals and intersections with stop signs.
88
demand for the corridor. The forecasted growth for individual links and ramps varied
from this average value.
Since the existing conditions were uncongested and the growth forecast is a modest
10 percent growth, the forecasted demand was not constrained because of anticipated
capacity constraints on the entry links to the corridor.
Since a 5-year forecast is being performed, it was considered to be fairly reliable for such a
short time period. No extra allowance was added to the forecasts or subtracted from them
to account for uncertainty in the demand forecasts.
Step 2: Generation of Alternatives
Two alternatives will be tested with the calibrated model—no-build and build. The build
alternative consists of ramp metering on the two eastbound freeway on-ramps. The no-
build alternative has no ramp metering. Figure 12 below illustrates the coding of one of
the ramp meters.
Figure 12. Ramp meter geometry.
Step 3: Selection of MOEs
The following system MOEs were selected for evaluation of the alternatives: VMT, VHT,
and delay (vehicle-hours). The selected indicator of localized problems was a “blocked
link,” indicating that the queue filled up and overflowed the available storage in the link.
89
It was opted to report the mean results rather than the 95-percent worst case results.
Step 4: Model Application
The model was run 10 times for each alternative. The results were output into a
spreadsheet and averaged for each alternative.
The impact of ramp meters on route choice was estimated outside of the model using a
regional travel demand model to predict the amount of diversion. The regional model
predicted that diversion was implemented in the simulation model by manually adjusting
the turn percentages at the appropriate upstream intersections and ramp junctions.
The initialization period was automatically excluded from the tabulated results by the
selected software program.
The results were reviewed to determine if building queues17 were extending beyond the
physical boundaries of the model or the temporal boundaries of the analytical period.
None was found, so it was not necessary to correct the model results for untabulated
congestion.
Because of the modest differences in congestion between the alternatives, induced

demand was not considered to be a significant factor in this analysis. No adjustments
were made to the baseline demand forecasts.
Signal/meter control optimization was performed outside of the model using macroscopic
signal optimization software and ramp meter optimization software. The recommended
optimal settings were input into the simulation model. Separate optimizations were
performed for the no-build and build alternatives.
Step 5: Tabulation of Results
The model results for 10 repetitions of each alternative were output into a spreadsheet and
averaged for each alternative (see table 7 below). A review of the animation output
indicated that post-model corrections of untallied congestion were not necessary.18
17
Recurrent queues at signals that occur during each cycle are not considered to be building queues
indicative of unserved demand that might bias the results for an alternative.
18
No increasing queues indicating underserved demand were found.
90
Table 7. Summary of analytical results.
Measure of Future
Effectiveness Existing No-Build Build
VMT
Freeway 35,530 39,980 40,036
Arterial 8,610 9,569 9,634
Total 44,140 49,549 49,670
VHT
Freeway 681.6 822.3 834.2
Arterial 456.5 538.5 519.5
Total 1138.1 1360.8 1353.7
Delay (VHT)
Freeway 33.1 90.4 101.0
Arterial 214.3 269.7 248.9
Total 247.4 360.1 349.9
Step 6: Evaluation of Alternatives
Under the no-build scenario, the total delay on the corridor increased by 46 percent over
existing conditions. The VMT increased by 12 percent and the total travel time increased
by 20 percent. Most of the delay increases were on the freeway mainline links.
Under the improved scenario (ramp metering plus signal optimization), systemwide delay
was reduced by about 3 percent (from the no-build scenario) with a slight increase in
VMT.19 Freeway mainline traffic conditions improved at the expense of the on-ramp
traffic.
The improvements are operationally acceptable (no spillbacks from the ramp meters to the
arterial network).
19
The improvement is modest enough that it might be worth establishing confidence intervals for
the results and performing some sensitivity analysis and hypothesis tests to confirm the
robustness of the conclusion.
91
7.0 Final Report
This chapter discusses the documentation of the Project Scope

1
microsimulation analysis results in a final report and
See Chapter 1
- Select approach
- Select model
the technical report or appendix supporting the final 2 Data Collection

- Traffic volumes
actual modeling
See Chapter 2
report.

7.1 Final Report

- Review Animation
Working Model
Before Calibration
Compare Model
The final report presents the assumptions, analytical

steps, and results of the analysis in sufficient detail for Acceptable

Match
No
Calibration
decisionmakers to understand the basis for and Yes

See Chapter 5
Calibrated Model
implications of choosing among the project 6

- Forecast Demand
See Chapter 6
alternatives. The final report, however, will not usually

- Base Case
Application
Final Report
contain sufficiently detailed information to enable other 7 - Key Results

See Chapter 7
analysts to reproduce the results. That is the purpose of

the technical report/appendix.
The effort involved in summarization of the results should not be underestimated, since
microsimulation models produce a wealth of numerical output that must be tabulated and
summarized. The final report should include the following:
1. Study Objectives and Scope.
2. Overview of Study Approach (tools used, methodology, rationale).
3. Data Collection (sources and methods).
4. Calibration Tests and Results (which parameters were modified and why).
5. Forecast Assumptions (assumed growth inside and outside of the study area, street
improvements, etc.).
6. Description of Alternatives.
7. Results.
7.2 Technical Report/Appendix

The technical report/appendix documents the microsimulation analysis in sufficient detail
to enable an analyst to reproduce the results (the version or release of the software used is
included). It may be an appendix to the final report or a separate document.
The technical report/appendix is a vital step in preserving the rationale for the various
decisions that were made in the process of developing, calibrating, and operating a
microsimulation model. The documentation should be sufficient so that given the same
93
input files, another analyst can understand the calibration process and repeat the
alternatives analysis.
The technical report/appendix should include the model input and output files (in
electronic format) for the final model calibration run and alternatives analysis model runs.
In addition to a diskette with the calibration and alternatives analysis files (input and
output), the technical report/appendix should include a printed listing of the files on the
diskette with a text description of the contents and purpose of each file.
94
Appendix A: Traffic Microsimulation
Fundamentals
This section provides an overview of the technical aspects of microsimulation. It identifies

and defines the terminology typically used, and what model types and procedures are
generally employed in microsimulation. This section also provides a general introduction
to traffic simulation models. Readers interested in additional details on the technical
aspects of microsimulation may consult the Revised Monograph on Traffic Flow Theory
(Chapter 10: Traffic Simulation) or other documents.
A.1 Overview
Simulation models are designed to emulate the behavior of traffic in a transportation
network over time and space to predict system performance. Simulation models include
the mathematical and logical abstractions of real-world systems implemented in computer
software. Simulation model runs can be viewed as experiments performed in the
laboratory rather than in the field.
A.2 System Update Mechanism

Traffic simulation models describe the changes in the system state through discrete
intervals in time. There are generally two types of models, depending on whether the
update time intervals are fixed or variable:
Discrete Time (Time-Scan) Models: The system is being updated at fixed-time intervals.
For example, the model calculates the vehicle position, speed, and acceleration at 1-s
intervals. The choice of the update time interval depends on how accurately the system
needs to be simulated at the expense of computer-processing time. Most microsimulation
models employ a 0.1-s resolution for updating the system. Traffic systems are typically
modeled using time-scan models because they experience a continuous change in state.
Discrete Event (Event-Scan) Models: In these models, the time intervals vary in length
and correspond to the intervals between events. For example, a pre-timed traffic signal
indication (e.g., green) remains constant for 30 s until its state changes instantaneously to
yellow. The operation of the signal is described by recording its changes in state when
events occur, rather than monitoring the state of the signal each second. Event-scanning
models typically achieve significant reductions in computer run time. However, they are
suitable for simulating systems whose states change infrequently.
95
A.3 Traffic Stream Representation
Simulation models are typically classified according to the level of detail at which they
represent the traffic stream. These include:
Microscopic Models: These models simulate the characteristics and interactions of

individual vehicles. They essentially produce trajectories of vehicles as they move
through the network. The processing logic includes algorithms and rules describing how
vehicles move and interact, including acceleration, deceleration, lane changing, and
passing maneuvers.
Mesoscopic Models: These models simulate individual vehicles, but describe their
activities and interactions based on aggregate (macroscopic) relationships. Typical
applications of mesoscopic models are evaluations of traveler information systems. For
example, they can simulate the routing of individual vehicles equipped with in-vehicle,
real-time travel information systems. The travel times are determined from the simulated
average speeds on the network links. The average speeds are, in turn, calculated from a
speed-flow relationship.
Macroscopic Models: These models simulate traffic flow, taking into consideration
aggregate traffic stream characteristics (speed, flow, and density) and their relationships.
Typically, macroscopic models employ equations on the conservation of flow and on how
traffic disturbances (shockwaves) propagate in the system. They can be used to predict
the spatial and temporal extent of congestion caused by traffic demand or incidents in a
network; however, they cannot model the interactions of vehicles on alternative design
configurations.
Microscopic models are potentially more accurate than macroscopic simulation models.
However, they employ many more parameters that require calibration. Also, the
parameters of the macroscopic models (e.g., capacity) are observable in the field. Most of
the parameters of the microscopic models cannot be observed directly in the field (e.g.,
minimum distances between vehicles in car-following situations).
A.4 Randomness in Traffic Flow

Simulation models are also classified by how they represent the randomness in the traffic
flow, including:
Deterministic Models: These models assume that there is no variability in the driver-
vehicle characteristics. For example, it is assumed that all drivers have a critical gap of 5 s
in which to merge into a traffic stream, or all passenger cars have a vehicle length of 4.9 m
(16 ft).
Stochastic Models: These models assign driver-vehicle characteristics from statistical

distributions using random numbers. For example, the desired speed of a vehicle is
randomly generated from an assumed normal distribution of desired speeds, with a mean
of 105 km/h (65 mi/h) and a standard deviation of 8 km/h (5 mi/h). Stochastic
96
simulation models have routines that generate random numbers. The sequence of random
numbers generated depends on the particular method and the initial value of the random
number (random number seed). Changing the random number seed produces a different
sequence of random numbers, which, in turn, produces different values of driver-vehicle
characteristics.
Stochastic models require additional parameters to be specified (e.g., the form and
parameters of the statistical distributions that represent the particular vehicle
characteristic). More importantly, the analysis of the simulation output should consider
that the results from each model run vary with the input random number seed for
otherwise identical input data. Deterministic models, in contrast, will always produce the
same results with identical input data.
A.5 Microsimulation Process
Microsimulation models employ several submodels, analytical relationships, and logic to

model traffic flow. Detailed descriptions of each submodel are beyond the scope of this
section. Instead, this document focuses on some key aspects of the simulation process that
will probably affect the choice of the particular tool to be used and the accuracy of the
results.
Simulation models include algorithms and logic to:
• Generate vehicles into the system to be simulated.

• Move vehicles into the system.
• Model vehicle interactions.
A.5.1 Vehicle Generation
Generating Vehicles Into the Traffic Stream
At the beginning of the simulation run, the system is empty. Vehicles are generated at the
entry nodes of the analytical network, based on the input volumes and an assumed
headway distribution. Suppose that the specified volume is V = 600 veh/h for a 15-min
analytical period and that the model uses a uniform distribution of vehicle headways.
Then, a vehicle will be generated at time intervals:
H = mean headway = 3600/V = 3600/600 = 6 s (8)
If the model uses the shifted negative exponential distribution to simulate the arrival of
vehicles at the network entry node instead of the uniform distribution, then vehicles will
be generated as time intervals:
h = (H - h min ) [- ln(1 - R)] + H - hmin (9)
97
where:
h = headway (in seconds) separating each generated vehicle

hmin = specified minimum headway (e.g., 1.0 s)
R = random number (0 to 1.0)
Generating Driver-Vehicle Attributes
When a vehicle is generated at the entry of the network, the simulation model assigns
driver-vehicle characteristics. The following characteristics or attributes are commonly
generated for each driver-vehicle unit:
Vehicle: Type (car, bus, truck), length, width, maximum acceleration and deceleration,
maximum speed, maximum turn radius, etc.
Driver: Driver aggressiveness, reaction time, desired speed, critical gaps (for lane
changing, merging, crossing), destination (route), etc.
Note that the different models may employ additional attributes for each driver-vehicle
unit to ensure that the model replicates real-world conditions. Each attribute may be
represented in the model by constants (e.g., all passenger cars have a vehicle length of 4.9
m (16 ft)), functional relationships (e.g., maximum vehicle acceleration is a linear function
of its current speed), or probability distributions (e.g., driver’s desired speed is obtained
from a normal distribution). Most microsimulation models employ statistical distributions
to represent the driver-vehicle attributes. The statistical distributions employed to
represent the variability of the driver-vehicle attributes and their parameters must be
calibrated to reflect local conditions.
Figure 13 illustrates a generic process for generating driver-vehicle attributes for a

stochastic microscopic simulation model. Drivers are randomly assigned an
“aggressiveness index” ranging from 1 (very aggressive) to 10 (very cautious), drawn from
a uniform distribution to represent the range of human behavior. Once the value of the
aggressiveness has been assigned (X = 7 in this example), it is used to assign the value of
the free-flow speed (97 km/h) and the acceptable deceleration (-3.4 m/s2 (-11 ft/s2)) from
the corresponding distribution of speeds and accelerations.
98
Source: Traffic Flow Theory Monograph (Chapter 10: Simulation)
Figure 13. Generation of driver characteristics.
99
A.5.2 Vehicle Movement
This section briefly describes the process for simulating vehicle movement and how it is
impacted by the physical environment. The physical environment—the transportation
network under study—is typically represented as a network of links and nodes. Links are
one-way roadways with fixed design characteristics and nodes that represent intersections
or locations where the design characteristics of the links change. Simulation models have
different limits on the network size (maximum number of links and nodes).
Vehicles, in the absence of impedance from other vehicles, travel at their desired speed on
the network links. However, their speed may be affected by the link-specified geometry
(horizontal and vertical alignment), pavement conditions, and other factors. For example,
the simulation model computes the actual vehicle speed as the minimum value from the
desired speed and the speed computed for the specified vertical and horizontal alignment.
However, not all microsimulation tools model the sensitivity of vehicle speeds to link
design characteristics.
Vehicles proceed through the network until they exit the system at their destination.
Typically, there are two types of simulation models: turning-fraction based or O-D based.
In O-D-based models, the O-D matrix is input, and when a vehicle is generated at an
origin, it is assigned its destination. The vehicle then exits the network at the specified
destination. In turning-fraction-based models, the vehicle destination is randomly
assigned at the entry of the link, based on specified turning volumes (or fractions) at the
downstream end of the link. For example, for a vehicle entering a signalized intersection
approach, the vehicle destination (going through or turning) is randomly determined
based on the input turning-volume fractions for the particular link. This also implies that
turning-fraction models are not well suited for tracing the performance of individual
vehicles throughout the network and evaluating the effectiveness of certain ITS options
(e.g., a fraction of vehicles being equipped with real-time information systems).
Simulation models employ a number of approaches to guide vehicles within the network.
They typically employ warning signs to advise the simulated vehicle to change lanes
because it needs to exit at the downstream off-ramp, its lane is ending, or there is a
blockage downstream. The location of the warning signs may significantly affect the
accuracy of the simulation. For example, vehicles traveling on the freeway at a speed of 97
km/h (60 mi/h) (27 m/s (88 ft/s)) may miss their exit without this advance warning if
they are traveling in the median lane of a multilane freeway and are required to make
multiple lane changes to exit at the downstream end of a short link.
Vehicle Interactions
Microscopic models simulate the interactions of individual vehicles as they travel in the
analytical network using car-following, lane-changing, and queue discharge algorithms.
100
Car Following
The interaction between a leader and follower pair of vehicles traveling in the same lane,
is generally assumed to be a form of stimulus-response mechanism:
Response follower (t+T) = (sensitivity) • (stimulus) t (10)
where:
T = reaction time (time lag) for the response of the following vehicle
Car-following models for highway traffic were proposed since the inception of the traffic-
flow theory in the early 1950s. Extensive experimentation and theoretical work were
performed at the General Motors (GM) laboratories and elsewhere. Most of the existing
simulation models employ fail-safe car-following algorithms. Such algorithms simulate
car-following behavior by maintaining a minimum safe distance between vehicles subject
to stochastically generated vehicle characteristics (e.g., maximum acceleration and
deceleration rates). The fail-safe car-following algorithms currently implemented in most
simulation models consist of the following components:
1. An equation that calculates the acceleration (deceleration) of the following vehicle in

response to the motion of its leader to maintain a target headway (spacing), depending
on the driver-vehicle characteristics.
a f = F ( vl , v f , s , T , X i ) (11)
where:
af = acceleration of the following vehicle after a reaction time T

vl and vf = speeds of the leading and following vehicles, respectively
s = distance between vehicles
Xi = parameters specific to the particular car-following model
2. The computed acceleration rate above must not exceed the maximum acceleration rate
for the specific vehicle type and must not result in a higher speed than the vehicle’s
desired speed.
3. Furthermore, the computed acceleration of the follower above must satisfy the “safe-
following” rule (i.e., the follower should always maintain a minimum separation from
the leader). If the value of the safe-following acceleration is smaller then the car-
following acceleration computed above, the former is implemented.
101
Lane Changing
The modeling of lane changing is based on the gap-acceptance process. A vehicle may
change lanes if the available gap in the target lane is greater than its critical gap. Typically,
three types of lane changes are modeled:
1. A mandatory lane change occurs when a vehicle is required to exit the current lane.
Merging from an on-ramp onto the freeway is an example of a mandatory lane
change. A vehicle must execute a lane change under several circumstances:
a. Current lane is ending (lane drop)

b. Must move to the outer lane(s) to exit the freeway via an off-ramp
c. Current lane will not accommodate its movement or vehicle type (e.g., HOV lane)
d. Lane is blocked because of incidents or other events
2. A discretionary lane change occurs when a vehicle changes lanes to improve its
position (i.e., travel at the desired speed). If a driver is impeded in the current lane
because of a slower moving vehicle in front, he or she may consider changing lanes to
improve his or her position (i.e., resume his or her desired speed). The lane-changing
logic determines which of the adjacent lanes (if more than one is available) is the best
candidate. To be a candidate, the lane’s channelization must accommodate the subject
vehicle, be free of blockage, and not end close by. Lane changing is performed subject
to the availability of gaps in the candidate lane and the acceptability of the gaps by the
subject vehicle.
On a surface-street network, discretionary lane changes may occur because of various

impedance factors perceived by the driver, including:
a. Queue length at the stop line.

b. Heavy truck traffic.
c. Bus transit operations.
d. Turning movements from shared lanes.
e. Pedestrian interference with turning vehicles.
f. Parking activities.
3. An anticipatory lane change occurs when a vehicle may change lanes in anticipation
of slowdowns in its current lane further downstream as a result of merging or
weaving. The decision to change lanes is based on the difference in speed at the
location of the merge or weave, between the vehicles in the current lane and the
adjacent lane not directly involved in the merge or weave. The lane-changing logic
recognizes that a driver may accelerate or decelerate to create acceptable gaps to change
lanes. Consider a vehicle A that desires to merge into a gap between vehicles B and C (B
currently is the leader of C). Vehicle A will accept the gap if the time headway between
it and vehicle B is greater than some critical value g(A,B) and the time headway
102
between it and vehicle C is greater than some critical value g(A,C). However, the critical
time headway values are not constant, but are dependent on the following
considerations:
• Critical time headways depend on vehicle speeds. Vehicle A will accept a smaller
critical headway if it is going slower than vehicle B.
• Lane changing takes place over a finite period of time. During this time period, the
driver who is changing lanes can adjust his or her position with respect to the new
leader by decelerating.
• The new follower may cooperate with the driver who is changing lanes by
decelerating to increase the size of the gap.
These considerations are typically combined in a risk measure. More aggressive drivers
would accept a higher risk value (i.e., shorter gaps and higher acceleration/
deceleration rates) to change lanes. Moreover, the risk value may be further increased,
depending on the type of lane change and the situation. For example, a merging
vehicle reaching the end of the acceleration lane may accept much higher risk values
(forced lane changes).
The model’s lane-changing logic and parameters have important implications for traffic
performance, especially for freeway merging and weaving areas. In addition, the time
assumed for completion of the lane-change maneuver affects traffic performance because
the vehicle occupies both lanes (original and target lanes) during this time interval.
Furthermore, one lane change is allowed per simulation time interval.
103
Appendix B: Confidence Intervals
This section explains how to compute the confidence intervals for the microsimulation
model results. It explains how to compute the minimum number of repeated field
measurements or microsimulation model runs needed to estimate the mean with a certain
level of confidence that the true mean actually falls within a target interval.
Three pieces of information are required: (1) sample standard deviation, (2) desired length
of the confidence interval, and (3) desired level of confidence.
The required number of repetitions must be estimated by an iterative process. A

preliminary set of repetitions is usually required to make the first estimate of the standard
deviation for the results. The first estimate of the standard deviation is then used to
estimate the number of repetitions required to make statistical conclusions about
alternative highway improvements.
When all of the repetitions have been completed, the analyst then recomputes the
standard deviation and the required number of repetitions based on all of the completed
repetitions. If the required number of repetitions is less than or equal to the completed
number of repetitions, then the analysis has been completed. If not, the analyst either
relaxes the desired degree of confidence in the results or performs additional repetitions.
B.1 Why Are Multiple Model Runs Required?

Multiple repetitions of the same model are required because microsimulation results will
vary depending on the random number seed used in each run. The random number seed
is used to select a sequence of random numbers that are used to make numerous decisions
throughout the simulation run (Should a vehicle be loaded on the network now or later?
Should the driver be aggressive or timid? Which vehicle type should it be? Will the driver
choose the shortest path or a slightly longer one instead?). The outcomes of all of these
decisions will affect the simulation results. The results of each run will usually be close to
the average of all of the runs; however, each run will be different from the other.
Figure 14 shows the range of results that can occur

Microsimulation model results can
simply by varying the random number seed used in vary by 25 percent between runs.
each run. The figure shows the mean system vehicle
speed for a freeway-to-freeway interchange during
each minute of the simulation run. A total of six simulation runs (called “processes” in the
figure) were made. As can be seen, after the initial warmup period, results diverge. The
range of mean speeds output by each process is about 8.1 to 9.7 km/h (5 to 6 mi/h) for
most of the length of the simulation period. Running a longer simulation period does not
reduce this variation. At the end of the simulation, the range of results is about 9.7 km/h
(6 mi/h), or 25 percent of the mean system speed at the end of the simulation period.
105
Vehicle Mean Speed
70
60
process-001
process-002
50 process-003
process-004
process-005
40 process-006
mph
mi/h
30
20
10
0
16 1
16 4
16 7
16 0
16 3
16 6
16 9
16 2
16 5
16 8
16 1
16 4
17 7
17 0
17 3
17 6
17 9
17 2
17 5
17 8
17 1
17 4
17 7
0
:2
:2
:2
:3
:3
:3
:3
:4
:4
:4
:5
:5
:5
:0
:0
:0
:0
:1
:1
:1
:2
:2
:2
:3
16
Simulation Time
Source: I-580/I-680 Freeway Interchange Model, Dowling Associates, 2002
1 mi/h = 1.61 km/h
Figure 14. Variation in the results between repetitions.
B.2 Estimation of Sample Standard Deviation

The standard deviation (an estimate of the variance) is required to estimate the number of
repetitions. However, it takes a minimum number of repetitions to estimate the standard
deviation in the first place. So either the analyst estimates the standard deviation directly
(based on past experience) or executes a few model run repetitions (each run using a
different random number seed) and uses the equation below to compute an initial
estimate of the sample standard deviation.
∑ (x − x )
2
s 2
= (12)
N −1
where:
s = standard deviation
x = variable (such as delay) for which the sample variance is desired
106
x = average value of the variable produced by the model runs
N = number of model runs
Unless the analyst already knows the standard deviation from experience, it is
recommended that four repetitions be performed for the initial estimation of the standard
deviation. This initial estimate is then revisited and revised later if and when additional
repetitions are performed for the purposes of obtaining more precise estimates of mean
values or for alternatives analysis.
B.3 Selection of Desired Confidence Level

The confidence level is the probability that the true mean lies within the target confidence
interval. The analyst must decide to what degree he or she wishes to know the interval in
which the true mean value lies. The usual approach is to pick a 95-percent level of
confidence; however, analysts may choose higher or lower levels of confidence. Higher
levels of confidence require more repetitions.
B.4 Selection of Desired Confidence Interval

The confidence interval is the range of values within which the true mean value may lie.
The length of the interval is at the discretion of the analyst and may vary according to the
purposes for which the results will be used. For example, if the analyst is testing
alternatives that are very similar, then a very small confidence interval will be desirable to
distinguish between the alternatives. If the analyst is testing alternatives with greater
differences, then a larger confidence interval can be tolerated. Smaller confidence intervals
require more repetitions to achieve a given level of confidence. Confidence intervals that
are less than half the value of the standard deviation will require a large number of
repetitions to achieve reasonable confidence levels.20
B.4.1 Computation of Minimum Repetitions
It is impossible to know in advance exactly how many model runs will be needed to
determine a mean (or any other statistical value) to the analyst’s satisfaction. However,
after a few model runs, the analyst can make an estimate of how many more runs may be
required to obtain a statistically valid result.
The required minimum number of model repetitions is computed using the following
equation:
s
CI1−α % = 2 ∗ t(1−α / 2 ), N −1 (13)
N
20
With such a tight confidence interval, the analyst may be striving for a degree of precision not
reflected under real-world conditions.
107
where:
CI(1-alpha)% = (1-alpha)% confidence interval for the true mean, where alpha equals the
probability of the true mean not lying within the confidence interval
t(1-alpha/2),N-1 = Student’s t-statistic for the probability of a two-sided error summing to alpha
with N-1 degrees of freedom, where N equals the number of repetitions
s = standard deviation of the model results
Note that when solving this equation for N, it will be necessary to iterate until the
estimated number of repetitions matches the number of repetitions assumed when
looking up the t statistic. Table 8 shows the solutions to the above equation in terms of the
minimum number of repetitions for various desired confidence intervals and desired
degrees of confidence.
108
Table 8. Minimum number of repetitions needed to obtain the
desired confidence interval.
Desired Range Desired Minimum

(CI/S) Confidence Repetitions
0.5 99% 130
0.5 95% 83
0.5 90% 64
1.0 99% 36
1.0 95% 23
1.0 90% 18
1.5 99% 18
1.5 95% 12
1.5 90% 9
2.0 99% 12
2.0 95% 8
2.0 90% 6
Notes:
1. Desired Range = desired confidence interval (CI)
divided by standard deviation (S)
2. For example, if the standard deviation in the delay is
1.5 s and the desired confidence interval is 3.0 s at a 95-
percent confidence level, then it will take eight
repetitions to estimate the mean delay to within ± 1.5 s.
109
Appendix C: Estimation of the Simulation
Initialization Period
Simulation model runs usually start with zero vehicles on the network. If the simulation
output is being compared to field measurements (as in calibration), then the artificial
period where the simulation model starts out with zero vehicles (the warmup period)
must be excluded from the reported statistics for system performance. Some software
programs will do this automatically. For others, the warmup period must be computed
offline by the analyst. This section explains how to identify the warmup period.
The number of vehicles present at any one time on the network is used to determine
whether the model has reached equilibrium and, therefore, can start tallying performance
statistics for the network. Once the number of vehicles present on the network ceases to
increase by a minimum specified amount, then the warmup period is deemed to have
been concluded (see figure 15).
If the number of vehicles and the mean speed do not level off within the first 15 min, it
could be that the demand coded by the analyst for the system is greater than the system
capacity. In this case, congestion will never level off. This will result in less accurate
congestion statistics since the system never clears the congestion. The analyst should
consider extending the starting and end times of the simulation to incorporate lower
demand periods before and after the peak period.
If it is not feasible to extend the simulation period to uncongested time periods, the
analyst should choose a warmup period that is equal to at least twice the estimated travel
time at free-flow conditions to traverse the length of the network. For example, if the
freeway being modeled is 8.1 km (5 mi) long, it takes roughly 5 min to traverse its length
at the free-flow speed, so the warmup period is set at 10 min.
Note that in this example, the number of vehicles in the system is generally steadily
increasing. The system never reaches equilibrium. Initialization is achieved when the
number of vehicles entering the system is approximately equal to the number leaving the
system (even though this slight decline is later superceded by greater increases). The
generally increasing trend in the number of vehicles present in the system suggests that
the simulation period should be extended earlier and later to incorporate lower demand
periods at the beginning and the end of the peak period.
111
Number of Veh icles i n System
16
:2
1
5,00
1,000
1,500
2,000
2,500
0
16 : 00
:2
4
16 : 00
:2
7
16 : 00
:3
0
16 : 00
:3
16 3: 0
:3 0
6
16 : 00
:3
9
16 : 00
:4
2
16 : 00
:4
5
Warm-up Perio d = 20 minutes
16 : 00
:4
16 8: 0
:5 0
1
16 : 00
:5
112
4
16 : 00
:5
7
17 : 00
:0
17 0: 0
:0 0
Simulati on Time 3
17 : 00
:0
6
17 : 00
:0
9
17 : 00
:1
2
Figure 15. Illustration of warmup period. 17 : 00
:1
17 5: 0
:1 0
8
17 : 00
:2
Source: I-680/I-580 Interchange Microsimulation Model, Dowling Associates, 2002
1
17 : 00
:2
4
17 : 00
:2
7
17 : 00
:3
0:
00
Appendix D: Simple Search Algorithms for
Calibration
Since simulation models are complex, it is not typically possible to formulate the models
as a closed-form equation for which traditional calculus techniques can be applied to find
a minimum value solution. It is necessary to use some sort of search algorithm that relies
on multiple operations of the simulation model, plotting of the output results as points,
and searching between these points for the optimal solution. Search algorithms are
required to find the optimal solution to the calibration problem. The calibration problem is
a nonlinear least-squares optimization problem.
There are many software programs available for identifying the optimal combination of
calibration parameters for minimizing the squared error between the field observations
and the simulation model. The Argonne National Laboratory (www-fp.mcs.anl.gov/
otc/Guide/SoftwareGuide/Categories/nonlinleastsq.html) lists several software
programs for nonlinear least-squares parameter estimation.
The sections below illustrate some simple approaches available for single-parameter
estimation and dual-parameter estimation when working with a stand-alone simulation
model. Estimation of three or more parameters, however, would require the use of a
software program.
D.1 Single-Parameter Search Algorithm (Golden Section)1

Several methods are available for finding the value of a single parameter that minimizes
the squared error between the model and the observations. These methods include Newton’s
Method, Secant Method, Quadratic Approximation Methods, and the Golden Section
Method. The Golden Section Method is illustrated in the example below.
D.1.1 Example of Golden Section Method
Objective: Find the global value of the mean headway between vehicles that minimizes
the squared error between the field counts of traffic volumes and the model estimates.
Approach: Use the Golden Section Method to identify the optimal mean headway.
1
References: Hillier, F.S., and G.J. Lieberman, Introduction to Operations Research, Sixth Edition,
McGraw-Hill, New York, 1995; Taha, H.A., Operations Research, An Introduction, Seventh Edition,
Prentice-Hall, New York, 2003 (if the seventh edition is not available, look for the sixth edition
published in 1996).
113
Step 1: Identify the maximum and minimum acceptable values for the parameter
to be optimized.
This step brackets the possible range of the parameter in which the optimal solution is
presumed to lie. The user can select the appropriate range. For this example, we will set
the minimum acceptable value for the mean headway at 0.5 s and the maximum at 2.0 s.
The larger the range, the more robust the search. However, it will take longer to find the
optimum value. The smaller the range of acceptable values, the greater the likelihood that
the best solution lies outside of the range.
Step 2: Compute the squared error for maximum and minimum values.
The simulation model is run to determine the volumes predicted when the maximum
acceptable mean headway is input and again for the minimum acceptable headway. Either
randomization should be turned off or the model should be run several times with each
mean headway and the results averaged. The squared error produced by the model for
each mean headway is computed.
Step 3: Identify two interior parameter values for testing.
The two interior points (x1 and x2) are selected according to specific ratios of the total
range that preserve these ratios as the search range is narrowed in subsequent iterations.
The formulas for selecting the two interior mean delays for testing are the following:
x1 = min+ 0.382 • (max− min) (14)
x2 = min+ 0.618 • (max− min) (15)
where:
x1 = lower interior point value for the mean delay to be tested

x2 = upper interior point value for the mean delay to be tested
min = lower end of the search range for the mean delay
max = upper end of the search range for the mean delay
The minimum and maximum ends of the search range are initially set by the user (based
on the acceptable range set in step 1); however, the search range is then gradually
narrowed as each iteration is completed.
Step 4: Compute squared error for two interior parameter values.
The simulation model is run for the new values of mean delay (x1 and x2) and the squared
errors are computed.
114
Step 5: Identify the three parameter values that appear to bracket the optimum.
This step narrows the search range. The parameter value (x1 or x2) that produces the
lowest squared error is identified. (If either the minimum or the maximum parameter
values produce the least-squared error, the search range should be reconsidered.) The
parameter values to the left (lower) and right (higher) of that point become the new
minimum and maximum values for the search range. For instance, in figure 16, parameter
value x2 produces the lowest squared error.
Squared Error
Search Range
0 Min x1 x2 Max
Parameter
Figure 16. Golden Section Method.
Step 6: Return to step 3 and repeat until the uncertainty in the location of the
optimal parameter value is satisfactory.
The Golden Section search is repeated until the range of values in which the optimum
parameter value lies is small enough to satisfy the user’s requirements. After about
10 iterations, the uncertainty in the optimal value of the parameter is reduced by a factor
of 100. Therefore, if an initial range of 0.5 to 2.5 s is specified for the mean headway (a
range of 2.0 s), this range will be reduced to 0.2 s after 10 iterations of the Golden Section
Method. The user should obtain the optimal value of the mean headway to within ± 0.1 s.
115
D.2 Simple Two-Parameter Search Algorithm
In the case where two model parameters are to be optimized, a contour plot approach to
identifying the optimal values of the parameters can be used. One first identifies the
acceptable ranges of the two parameters and then exercises the model for pairs of values
of the parameters. The squared error is computed for each pair of parameter values and is
plotted in a contour plot to identify the value pairs that result in the lowest squared error.
An example of this approach is shown in figure 17 for optimizing the mean headway and
mean reaction time parameters. The search starts with the default values, blankets the
region with a series of tests of different values, and then focuses in more detail on the
solution area.
3.8 3.8
3.2
3.5
2.1
3.1
Mean Headway
Default 1.3 0.7

Optimal Headway 0.5
1.6 0.6
1
2.8 0.8
Optimal Reaction Time
2 3.2
3.2 3
3.4
Default
Contours = Squared Error x 105
0
Mean Reaction Time
Figure 17. Example contour plot of the squared error.
D.3 Dual-Objective, Two-Parameter Search Algorithm2

This section addresses the case where the analyst wishes to consider two model
calibration objectives separately, rather than as the weighted sum of the squared errors
2
Adapted from Gardes, Y., A.D. May, J. Dahlgren, and A. Skabardonis, Bay Area Simulation and
Ramp Metering Study, California PATH Research Report UCB-ITS-PRR-2002-6, University of
California, Berkeley, February 2002.
116
(single objective) discussed above. The analyst in this case can use a variation of the two-
dimensional search algorithm identified above with up to two parameters. Instead of
plotting a single set of contour lines for the squared error, the analyst plots two sets of
contour lines (one set for each objective) for each pair of parameters. The analyst then
visually identifies the acceptable solution range where both objectives are satisfied to the
extent feasible. The following example illustrates this algorithm.
The analyst has identified the following two objectives for the model calibration:
• Model should predict a mean speed for the freeway of 56 km/h (35 mi/h) (± 3 km/h
(2 mi/h)).
• Model should predict a maximum observed flow rate for the bottleneck section of
2200 veh/h per lane (± 100 veh/h).
Figure 18 shows how changing the mean headway and the mean reaction time changes
the values of these two objectives (speed and capacity). The solution region shows the
pairs of values of mean headway and mean reaction time that meet both objectives.
Capacity = 2200 vphl Capacity = 2000 vphl
Capacity = 2400 vphl

Speed
Speed ==35
35mph
mi/h
Solution Region
Mean Headway
Default
Speed
Speed ==30
30mph
mi/h
Speed = 25=mi/h
Speed 25 mph
Default
0
Mean Reaction Time
vphl = vehicles per hour per lane 1 mi/h = 1.61 km/h
Figure 18. Dual-objective, dual-parameter search.
117
Appendix E: Hypothesis Testing of Alternatives
When the microsimulation model is run several times for each alternative, the analyst may
find that the variance in the results for each alternative is close to the difference in the
mean results for each alternative. How is the analyst to determine if the alternatives are
significantly different? To what degree of confidence can the analyst claim that the
observed differences in the simulation results are caused by the differences in the
alternatives and not just the result of using different random number seeds? This is the
purpose of statistical hypothesis testing. Hypothesis testing determines if the analyst has
performed an adequate number of repetitions for each alternative to tell the alternatives
apart at the analyst’s desired level of confidence.
E.1 Estimation of the Required Number of Model Repetitions
This section identifies how to estimate the minimum number of model run repetitions that
would be required to determine if two alternatives with results a given distance apart are
significantly different. This estimate, however, requires a preliminary estimate of the
standard deviation of the model run results for the alternatives, which, in turn, require
some preliminary runs to estimate the standard deviation.
The procedure involves the following steps:
1. Perform a preliminary set of model run repetitions for each alternative.
2. Estimate the standard deviation and the mean difference between the alternatives
from the preliminary runs and then compute the required number of runs using the
equations in this subsection.
3. If the required number of runs is greater than the preliminary number of runs, the
analyst should perform the additional repetitions for each alternative and recompute
the mean difference and standard deviation using the augmented set of model run
repetitions.
E.1.1 Estimation of Pooled Standard Deviation
The analyst should perform about six to eight model repetitions of each alternative to
estimate the pooled standard deviation for all alternatives according to the following
equation:
s x2 + s 2y
s 2p = 2 (16)
where:
sx = standard deviation of model run results for alternative x

sy = standard deviation of model run results for alternative y
119
The preliminary model repetitions used to estimate the pooled estimate of the standard
deviation of the model run results can also be used to estimate the likely difference of the
means for each alternative.
E.1.2 Selection of Desired Confidence Level
A 95-percent confidence level is often selected for hypothesis testing. This means that
there is a 5-percent chance (often called “alpha” in the textbooks) that the analyst will
mistakenly reject the null hypothesis when it is really true (type I error). If a higher
confidence level is desirable, it comes at the cost of increasing the likelihood of making a
type II error (accepting the null hypothesis when it is really false) (table 9).
Table 9. Null hypothesis.
True False
Accept Hypothesis OK Type II Error
Reject Hypothesis Type I Error OK
The study objective may determine the desired confidence level. For example, if the
objective is to design an alternative with a 95-percent probability that it will provide
significant improvements over the current facility, then this is the appropriate confidence
level for the determination of the number of model repetitions required.
E.1.3 Selection of Minimal Difference in the Means
The likely minimal difference in the means between the alternatives should be identified
by the analyst. This is the target sensitivity of the simulation tests of the alternatives.
Alternatives with mean results farther apart than this minimal difference will obviously
be different. Alternatives with mean results closer together than this minimal difference
will be considered to be indistinguishable.
The study objectives have some bearing on the selection of the minimal difference to be
detected by the simulation tests. If the study objective is to design a highway
improvement that reduces mean delay by at least 10 percent, then the tests should be
designed to detect if the alternatives are at least 10 percent apart.
The preliminary model repetitions used to estimate the pooled estimate of the standard
deviation of the model run results can also be used to estimate the likely difference of the
means for each alternative. The smallest observed difference in these preliminary runs
would be the selected minimal difference of the means to be used in determining the
required number of repetitions.
120
E.1.4 Computation of Minimum Repetitions
Assuming that the analyst wishes to reject the null hypothesis that the means of the two
most similar alternatives are equal with only an alpha percent chance of error against the
counter hypothesis that the mean of alternative x is different than y, then the number of
repetitions required can be computed according to the following equation:3
x − y > t(1−α / 2 ); 2 n − 2 • s p 2
n
(17)
where:
x − y = absolute value of the estimated difference between the mean values for the two
most similar alternatives x and y
sp = pooled estimate of the standard deviation of the model run results for each alternative
n = number of model repetitions required for each alternative
t = t statistic for a confidence level of 1-alpha and 2n-2 degrees of freedom4
Note that when solving this equation for N, it is necessary to iterate until the estimated
number of repetitions matches the number of repetitions assumed when looking up the t
statistic. Table 10 provides solutions for this equation for various target mean difference
ranges and levels of confidence.
3
If the analyst intends to perform hypothesis tests on only a few pairs of alternatives, then the
equation provided should be sufficiently accurate. However, if the analyst plans to perform
hypothesis testing of all possible pairs of alternatives, then this equation will underestimate the
required number of repetitions needed to achieve the desired confidence level. The analyst
should consult Lane, D.M., Hyperstat OnLine, An Introductory Statistics Book and Online Tutorial for
Help in Statistics, Chapter 12: Introduction to Between-Subjects ANOVA (“All pairwise
comparisons among means…”), Rice University (www.davidmlane.com/hyperstat).
4
Note that this is a two-sided t-test for the null hypothesis that the means are equal versus the
hypothesis that they are different.
121
Table 10. Minimum repetitions for distinguishing alternatives.
Minimum Minimum
Difference of Desired Repetitions per
Means Confidence Alternative
0.5 99% 65
0.5 95% 42
0.5 90% 32
1.0 99% 18
1.0 95% 12
1.0 90% 9
1.5 99% 9
1.5 95% 6
1.5 90% 5
2.0 99% 6
2.0 95% 4
2.0 90% 4
Notes:
1. The minimum difference in the means is expressed in
units of the pooled standard deviation: x − y .
Sp
2. For example, if the pooled standard deviation in the
delay for two alternatives is 1.5 s, and the desired
minimum detectable difference in the means is a 3.0-s
delay at a 95-percent confidence level, then it will take
four repetitions of each alternative to reject the
hypothesis that the observed differences in the
simulation results for the two alternatives could be the
result of random chance.
E.2 Hypothesis Testing for Two Alternatives

To determine whether simulation output provides sufficient evidence that one alternative
is better than the other (e.g., build project versus no-build), it is necessary to perform a
statistical hypothesis test of the difference of the mean results for each alternative. A null
hypothesis is specified: “The model-predicted difference in VHT for the two alternatives
occurred by random chance. There really is no significant difference in the mean travel
time between the alternatives.” A statistic is computed for a selected level of confidence,
and if the difference between the two means is less than that statistic, then the null
hypothesis is accepted and it is concluded that there is insufficient evidence to prove that
the one alternative is better than the other. The analyst either makes more model
repetitions for each alternative (to improve the sensitivity of the test) or relaxes his or her
standards (confidence level) for rejecting the null hypothesis.
122
The specification of the problem is:
Null Hypothesis:
H0 : µx − µy = 0 (18)
against
H1 : µ x − µ y ≠ 0 (19)
where:
µx = mean VHT (or some other measure) for alternative x

µy = mean for alternative y
This is a two-sided t-test with the following optimal rejection region for a given alpha
(acceptable type I error):
x − y > t(1−α / 2 ); ( n + m − 2 ) • s p 1
n
+ 1
m
(20)
where:
x − y = absolute value of the difference in the mean results for alternative x and
alternative y
sp = pooled standard deviation
t = Student’s t-distribution for a level of confidence of 1-alpha and n+m-2 degrees of

freedom
n = sample size for alternative x
m = sample size for alternative y
( n −1)s x2 + ( m −1)s 2y
s 2p = (m + n − 2 ) (21)
where:

sx = standard deviation of the results for alternative x
sy = standard deviation of the results for alternative y
123
The probability of mistakenly accepting the null hypothesis is alpha (usually set to 5
percent to get a 95-percent confidence level test). This is a type I error.
There is also a chance of mistakenly rejecting the null hypothesis. This is called a type II
error and it varies with the difference between the sample means, their standard
deviation, and the sample size.5
E.3 Hypothesis Testing for Multiple Alternatives

When performing hypothesis testing on more than one pair of alternatives, it is most
efficient to first determine if any of the alternatives are significantly different from the
others. An analysis of variance (ANOVA) test is performed to determine if the mean
results for any of the alternatives are significantly different from the others:
• If the answer is yes, the analyst goes on to test specific pairs of alternatives.
• If the answer is no, the analysis is complete, or the analyst runs more model
repetitions for each alternative to improve the ability of the ANOVA test to
discriminate among the alternatives.
E.3.1 Analysis of Variance (ANOVA) Test
The ANOVA test has three basic requirements:
• Independence of samples (random samples).
• Normal sampling distribution of means.
• Equal variances of groups.
Levine’s test of heteroscedasticity can be used for testing whether or not the variances of
the model run results for each alternative are similar. Less powerful nonparametric tests,
such as the Kruskal-Wallis test (K-W statistic) can be performed if the requirements of the
ANOVA test cannot be met.
However, the ANOVA test is tolerant of modest violations of these requirements and may
still be a useful test under these conditions. The ANOVA test will tend to be conservative
if its requirements are not completely met (less likely to have a type I error with a lower
power of the test to correctly reject the null hypothesis).
5
Analysts should consult standard statistical textbooks for tables on the type II errors associated
with different confidence intervals and sample sizes.
124
To perform the ANOVA test, first compute the test statistic:
MSB
F= (22)
MSW
where:
F = test statistic
MSB = mean square error between the alternatives (formula provided below)
MSW = mean square error among the model results for the same alternative (within
alternatives)
The formulas below show how to compute MSB and MSW:
∑n
i =1
i • ( xi − x ) 2
MSB = (23)
g −1
where:
MSB = mean square error between the alternatives (i = 1 to g)

ni = number of model runs with different random number seeds for alternative i
x i = mean value for alternative i
x = mean value averaged across all alternatives and runs
g = number of alternatives
and
∑ (n
i =1
i − 1) • s i2
MSW = (24)
N−g
where:
MSB = mean square error between the alternatives (i = 1 to g)

ni = number of model runs with different random number seeds for alternative i
s i2 = variance of the model run results for alternative i

N = total number of model runs summed over all alternatives
g = number of alternatives
125
The null hypothesis of equal means is rejected if:
F > F1-alpha,g-1,N-g (25)
where:
F1-alpha,g-1,N-g = F statistic for a type I error of alpha (alpha is usually set at 5 percent for a 95-
percent confidence level) and g-1 and N-g degrees of freedom. N is the total number of
model runs summed over all alternatives; g is the number of alternatives
If the null hypothesis cannot be rejected, then the analysis is either complete (there is no
statistically significant difference between any of the alternatives at the 95-percent
confidence level) or the analyst should consider reducing the level of confidence to below
95 percent or implementing more model runs per alternative to improve the sensitivity of
the ANOVA test.
E.3.2 Pairwise Tests of Some Pairs of Alternatives
If performing hypothesis tests for only a few of the potential pairs of alternatives, the
standard t-test can be used for comparing a single pair of alternatives:
x − y > t(1−α / 2); ( n + m − 2) • s p 1

n
+ 1
m
(26)
x − y = absolute value of the difference in the mean results for alternative x and
alternative y
t = t distribution for a level of confidence of 1-alpha and n+m-2 degrees of freedom
( n −1)s x2 + ( m −1)s 2y
s 2p = (m + n − 2 ) (27)
where:

sx = standard deviation of the results for alternative x
sy = standard deviation of the results for alternative y
126
If one merely wishes to test that the best alternative is truly superior to the next best
alternative, then the test needs to be performed only once.
If one wishes to test other possible pairs of alternatives (such as second best versus third
best), it is possible to still use the same t-test; however, the analyst should be cautioned
that the level of confidence diminishes each time the test is actually performed (even if the
analyst retains the same nominal 95-percent confidence level in the computation, the mere
fact of repeating the computation reduces its confidence level). For example, a 95-percent
confidence level test repeated twice would have a net confidence level for both tests of
(0.95)2, or 90 percent.
Some experts, however, have argued that the standard t-test is still appropriate for
multiple paired tests, even at its reduced confidence level.
E.3.3 Pairwise Tests of All Pairs of Alternatives
To preserve a high confidence level for all possible paired tests of alternatives, the more
conservative John Tukey “Honestly Significantly Different” (HSD) test should be used to
determine if the null hypothesis (that the two means are equal) can be rejected.6
The critical statistic is:
xi − x k
ts = (28)
MSE
n
where:
ts = Studentized t-statistic
xi = mean value for alternative i
MSE = mean square error = MSB + MSW
N = number of model runs with different random number seeds for each alternative (if the
number of runs for each alternative is different, then use the harmonic mean of the
number of runs for each alternative)
Reject the null hypothesis that the mean result for alternative i is equal to that for
alternative k if:
ts > t1-alpha,g-1 (29)
6
Adapted from Lane, D.M., Hyperstat OnLine, An Introductory Statistics Book and Online Tutorial for
Help in Statistics, Rice University (www.davidmlane.com/hyperstat).
127
where:
t1-alpha,g-1 = t statistic for a desired type I error of alpha (alpha is usually set at 5 percent to
obtain a 95-percent confidence level) and g-1 degrees of freedom, with g equal to the total
number of alternatives tested, not just the two being compared in each test
Some experts consider the HSD test to be too conservative, failing to reject the null
hypothesis of equal means when it should be rejected. The price of retaining a high
confidence level (the same as retaining a low probability of a type I error) is a significantly
increased probability of making a type II error (accepting the null hypothesis when it is
really false).
E.4 What To Do If the Null Hypothesis Cannot Be Rejected

If the null hypothesis of no significant difference in the mean results for the alternatives
cannot be rejected, then the analyst has the following options:
• Increase the number of model run repetitions per alternative until the simulation
performance of one alternative can be distinguished from the other.
• Reduce the confidence level from 95 percent to a lower level where the two
alternatives are significantly different and report the lower confidence level in the
results.
• Accept the result that the two alternatives are not significantly different.
128
Appendix F: Demand Constraints
Microsimulation results are highly sensitive to the amount by which the demand exceeds
the capacity of the facility, so it is vital that realistic demand forecasts be used in the
analysis. The following steps outline a procedure for manually reducing the forecasted
demands in the study area to better match the capacity of the facilities feeding the study
area.
Step 1: Identify Gateway Bottlenecks
The analyst should first identify the critical bottlenecks on the facilities feeding the traffic
to the boundaries of the microsimulation study area. Bottlenecks are sections of the
facilities feeding the model study area that either have capacities less than other sections
of the freeway or demands greater than the other sections. These are the locations that will
probably be the first ones to experience congested conditions as traffic grows.
These bottlenecks may be located on the boundary of the microsimulation study area, in
which case they are identical to the gateway zones on the boundary of the
microsimulation model study area. Bottlenecks within the microsimulation model study
area can be disregarded since they will be taken into account by the microsimulation
model.
Inbound bottlenecks are congested sections feeding traffic to the microsimulation model
area. Outbound bottlenecks are congested sections affecting traffic leaving the
microsimulation model area. If an outbound bottleneck will probably create future queues
that will back up into the microsimulation model study area, then the model study area
should be extended outward to include the outbound bottleneck. If the future outbound
queues will not back up into the model study area, then these bottlenecks can be safely
disregarded.
129
Microsimulation
Model Study Area
Gateway
Bottleneck
Figure 19. Bottleneck, gateway, and study area.
Step 2: Estimate Excess Demand at Inbound Bottlenecks
If the forecasted hourly demand at a bottleneck (in the inbound direction toward the
model) exceeds its capacity, the proportion of the demand that is in excess of the available
hourly capacity should be computed:
D−C
X= (30)
C
where:
X = proportion of excess demand

D = forecasted demand (veh/h)
C = estimated capacity (veh/h)
Step 3: Reduce Forecasted Demand Inbound at Gateways
The forecasted hourly demands for the off-ramps between the bottleneck and the gateway
entering the microsimulation study area should be reduced in proportion to the amount
by which the forecasted bottleneck demand exceeds its capacity:
Dconst = Dunconst * (1 − X ) (31)
where:
Dconst = constrained demand (veh/h) for a downstream off-ramp or exit point
Dunconst = unconstrained demand forecast (veh/h)
130
X = proportion of excess demand
It is suggested that the off-ramp demand be reduced in proportion to the reduction in

demand that can get through the bottleneck, assuming that the amount of reduction in the
downstream flows is proportional to the reduction in demand at the bottleneck. If the
analyst has superior information (such as an O-D table), then the assumption of
proportionality can be overridden by the superior information. The constrained
downstream gateway demand is then obtained by summing the constrained bottleneck,
off-ramp, and on-ramp volumes between the bottleneck and the gateway to the study
area.
Figure 20 illustrates how the proportional reduction procedure would be applied for a
single inbound bottleneck that reduces the peak-hour demand that can get through from
5000 veh/h to 4000 veh/h. Since there is an interchange between the bottleneck and the
entry gate to the microsimulation study area, the actual reduction is somewhat less
(800 veh/h) at the gate.
Microsimulation
Model Area
Duc = 4500
Dc = 3700
On-Ramp
Duc = 500 Off-Ramp
Dc = 500 Duc = 1000
Dc = 800
Duc = 5000
Dc = 4000
Bottleneck
Duc = 5000
Dc = 5000
Figure 20. Example proportional reduction of demand for capacity constraint.
Starting upstream of the bottleneck, there is an unconstrained demand for 5000 veh/h.
Since the bottleneck has a capacity of 4000 veh/h, the downstream capacity constrained
demand is reduced from the unconstrained level of 5000 veh/h to 4000 veh/h. Thus, 1000
vehicles are stored at the bottleneck during the peak hour. Since it is assumed that the
stored vehicles are intended for downstream destinations in proportion to the exiting
volumes at each off-ramp and freeway mainline, the downstream volumes are reduced
the same percentage as the percentage reduction at the bottleneck (20 percent). A 20-
percent reduction of the off-ramp volume results in a constrained demand of 800 veh/h.
The on-ramp volume is unaffected by the upstream bottleneck, so its unconstrained
131
demand is unchanged at 500 veh/h. The demand that enters the microsimulation study
area is equal to the constrained demand of 4000 veh/h leaving the bottleneck, minus the
800 veh/h leaving the freeway on the off-ramp, plus 500 veh/h entering the freeway at
the on-ramp, which results in a constrained demand of 3700 veh/h.
132
Appendix G: Bibliography
G.1 Traffic Microsimulation Fundamentals

Revised Monograph on Traffic Flow Theory, N.H. Gartner, C.J. Messer, and A. Ruthi
(editors), FHWA, 1994 (www.tfhrc.gov/its/tft).
May, A.D., Traffic Flow Fundamentals, Prentice Hall, Englewood Cliffs, NJ, 1990.
Ahmed, K.I., Modeling Drivers’ Acceleration and Lane Changing Behavior, Ph.D. Dissertation,
Massachusetts Institute of Technology, Cambridge, MA, February 1999.
Elefteriadou, L., G. List, J. Leonard, H. Lieu, M. Thomas, R. Giguere, R. Brewish, and

G. Johnson, “Beyond the Highway Capacity Manual: A Framework for Selecting Simulation
Models in Traffic Operational Analyses,” Transportation Research Record 1678, National
Academy Press, 1999, pp. 96-106.
G.2 Calibration and Validation of Microsimulation Models

Abdullai, B., J. Sheu, and W. Recker, Simulation of ITS on the Irvine FOT Area Using
“Paramics 1.5” Scalable Microscopic Traffic Simulator, Phase I: Model Calibration and Validation,
California PATH Program Research Report UCB-ITS-PRR-99-12, University of California,
Berkeley, 1999.
Ben-Akiva, M., A. Davol, T. Toledo, H.N. Koutsopoulos, W. Burghout, I. Andréasson,

T. Johansson, and C. Lundin, “Calibration and Evaluation of MITSIMLab in Stockholm,”
Paper presented at the Annual Meeting, TRB, Washington, DC, 2002.
Fellendorf, M., and P. Vortisch, “Validation of the Microscopic Traffic Flow Model VISSIM in
Different Real-World Situations,” Paper presented at the Annual Meeting, TRB, Washington,
DC, 2001 (www.itc-world.com/library).
“Freeway System Operational Assessment,” Paramics Calibration and Validation Guidelines

(Draft), Technical Report I-33, Wisconsin DOT, District 2, Milwaukee, WI, June 2002.
Gardes, Y., A.D. May, J. Dahlgren, and A. Skabardonis, Bay Area Simulation and Ramp
Metering, California PATH Program Research Report.
Hellinga, B.R., Requirements for the Calibration of Traffic Simulation Models, Canadian Society of
Civil Engineering (CSCE), University of Waterloo, Ontario, Canada, 1998.
Jayakrishnan, R., J. Oh, and A.B. Sahraoui, Calibration and Path Dynamics Issues in Microscopic
Simulation for Advanced Traffic Management and Information Systems, UCI-ITS-WP-00-22,
Institute of Transportation Studies, University of California, Irvine, December 2000.
133
Lee, D-H., X. Yang, and P. Chandrasekar, “Parameter Calibration for Paramics Using Genetic
Algorithm,” Paper No. 01-2339, Annual Meeting, TRB, Washington, DC, 2001.
Ma, T., and B. Abdulhai, “A Genetic Algorithm-Based Optimization Approach and Generic
Tool for the Calibration of Traffic Microscopic Simulation Parameters,” Paper presented at the
Annual Meeting, TRB, Washington, DC, 2002.
Owen, L.E., G. Brock, S. Mastbrook, R. Pavlik, L. Rao, C. Stallard, S. Sunkari, and Y. Zhang, A
Compendium of Traffic Model Validation Documentation and Recommendations, FHWA Contract
DTFH61-95-C-00125, FHWA, 1996.
Project Suite Calibration and Validation Procedures, Quadstone, Ltd., Edinburgh, United
Kingdom, February 2000 (www.paramics-online.com/tech_support/ customer/cust_doc.htm
(password required)).
Rakha, H., B. Hellinga, M. Van Aerde, and W. Perez, “Systematic Verification, Validation and
Calibration of Traffic Simulation Models,” Paper presented at the Annual Meeting, TRB,
Washington, DC, 1996.
Rao, L., L. Owen, and D. Goldsman, “Development and Application of a Validation

Framework for Traffic Simulation Models,” Proceedings of the 1998 Winter Simulation
Conference, D.J. Medeiros, E.F. Watson, J.S. Carson, and M.S. Manivannan (editors).
Skabardonis, A., Simulation of Freeway Weaving Areas, Preprint No. 02-3732, Annual Meeting,
TRB, Washington, DC, 2002.
Traffic Appraisal in Urban Areas: Highways Agency, Manual for Roads and Bridges, Volume 12,
Department for Transport, London, England, United Kingdom, May 1996
(www.archive.official-documents.co.uk/document/ha/dmrb/index.htm).
Wall, Z., R. Sanchez, and D.J. Dailey, A General Automata Calibrated With Roadway Data for
Traffic Prediction, Paper No. 01-3034, Annual Meeting, TRB, Washington, DC, 2001.
G.3 Analysis of Results

Benekohal, R.F., Y.M. Elzohairy, and J.E. Saak, A Comparison of Delay From HCS, Synchro,
Passer II, Passer IV, and CORSIM for an Urban Arterial, Preprint, Annual Meeting, TRB,
Joshi, S.S., and A.K. Rathi, “Statistical Analysis and Validation of Multi-Population Traffic
Simulation Experiments,” Transportation Research Record 1510, TRB, Washington, DC, 1995.
Prassas, E.S., A Comparison of Delay on Arterials by Simulation and HCM Computations, Preprint
No. 02-2652, Annual Meeting, TRB, Washington, DC, 2002.
Tian, Z.Z., T. Urbanik II, R. Engelbrecht, and K. Balke, Variations in Capacity and Delay
Estimates From Microscopic Traffic Simulation Models, Preprint, Annual Meeting, TRB,
134
Trueblood, M., “CORSIM…SimTraffic: What’s the Difference?,” PC-TRANS, Winter/Spring
2001.
Gibra, I.N., Probability and Statistical Inference for Scientists and Engineers, Prentice-Hall,
Englewood Cliffs, NJ, 1973.
Lane, D.M., Hyperstat OnLine, An Introductory Statistics Book and Online Tutorial for Help in
Statistics, Rice University, Houston, TX (www.davidmlane.com/hyperstat).
Gerstman, B., and M. Innovera, StatPrimer, Version 5.1, California State University, San Jose
(www.sjsu.edu/faculty/gerstman/StatPrimer).
135
Recycled
Recyclable HRDO-03/07-04(WEB)E

Vol3 Guidelines

Uploaded by

Copyright:

Available Formats

Vol3 Guidelines

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Vol3 Guidelines

Uploaded by

Copyright:

Available Formats

Traffic Analysis Toolbox Volume III:

Guidelines for Applying Traffic

Research, Development, and Technology

Jeffery A. Lindley, P.E.

The U.S. Government does not endorse products or manufacturers. Trademarks or

Quality Assurance Statement

APPROXIMATE CONVERSIONS FROM SI UNITS

1.0 Microsimulation Study Organization/Scope ........................................................... 11

2.0 Data Collection/Preparation ........................................................................................ 23

3.0 Base Model Development............................................................................................ 35

4.0 Error Checking ............................................................................................................... 45

5.0 Calibration of Microsimulation Models ................................................................... 53

6.0 Alternatives Analysis.................................................................................................... 73

7.0 Final Report .................................................................................................................... 93

Appendix A: Traffic Microsimulation Fundamentals .................................................... 95

Appendix B: Confidence Intervals ..................................................................................... 105

Appendix C: Estimation of the Simulation Initialization Period................................. 111

Appendix D: Simple Search Algorithms for Calibration............................................... 113

Appendix E: Hypothesis Testing of Alternatives ............................................................ 119

Appendix F: Demand Constraints ...................................................................................... 129

Appendix G: Bibliography................................................................................................... 133

1. Microsimulation model development and application process. ............................. 5

2. Prototypical microsimulation analysis task sequence............................................... 17

3. Example problem study network................................................................................. 21

4. Example of link-node diagram. .................................................................................... 36

5. Link-node diagram for example problem................................................................... 43

6. Checking intersection geometry, phasing, and detectors. ........................................ 50

7. Minimization of MSE. .................................................................................................... 61

8. Impact of mean queue discharge headway on MSE of modeled saturation

9. Average speed by minute of simulation (three links): Before calibration. ............. 68

10. Average speed by minute (three links): After calibration. ....................................... 68

11. Computation of uncaptured residual delay at the end of the simulation

12. Ramp meter geometry. .................................................................................................. 89

13. Generation of driver characteristics............................................................................. 99

14. Variation in the results between repetitions............................................................... 106

15. Illustration of warmup period. ..................................................................................... 112

16. Golden Section Method. ................................................................................................ 115

17. Example contour plot of the squared error................................................................. 116

18. Dual-objective, dual-parameter search........................................................................ 117

19. Bottleneck, gateway, and study area. .......................................................................... 130

20. Example proportional reduction of demand for capacity constraint...................... 131

1. Milestones and deliverables for a microsimulation study........................................ 18

3. Example node-numbering convention. ....................................................................... 38

4. Wisconsin DOT freeway model calibration criteria. ................................................. 64

5. Impact of queue discharge headway on predicted saturation flow........................ 66

6. System performance results for example problem. ................................................... 71

7. Summary of analytical results. ..................................................................................... 91

8. Minimum number of repetitions needed to obtain the desired confidence

9. Null hypothesis............................................................................................................... 120

10. Minimum repetitions for distinguishing alternatives. .............................................. 122

Microsimulation is the modeling of individual vehicle movements on a second or subsecond

Chapter 1.0 addresses the management, scope, and organization of microsimulation

Chapter 4.0 presents error-checking methods.

Chapter 5.0 provides guidance on the calibration of microsimulation models to local

Chapter 7.0 provides guidance on the documentation of microsimulation model analysis.

Appendix A provides an introduction to the fundamentals of microsimulation model

Appendix B provides guidance on the estimation of the minimum number of

Appendix C provides guidance on the estimation of the duration of the initialization

Appendix F describes a manual method for constraining external demands to available

Appendix G provides useful references on microsimulation.