An Introduction to Urban Travel Demand Forecasting - A Self Instructional Text
Click HERE for graphic.
USER-ORIENTED MATERIALS FOR
UTPS
AN INTRODUCTION TO URBAN TRAVEL
DEMAND FORECASTING
-- a self-instructional text --
U.S. DEPARTMENT OF TRANSPORTATION
FEDERAL HIGHWAY ADMINISTRATION
URBAN MASS TRANSPORTATION ADMINISTRATION
1977
TABLE OF CONTENTS
Page
Preface x
Purpose of Text x
General Content x
Who the Text is For xi
Acknowledgments xii
Chapter One: TRAVEL DEMAND FORECASTING IN THE URBAN
TRANSPORTATION PLANNING PROCESS
THE URBAN TRANSPORTATION PLANNING PROCESS 1-1
Organization 1-3
Planning Work Program 1-4
The Transportation Plan 1-6
The Long-Range Element 1-10
Plan Refinement 1-11
The Transportation Improvement Program 1-12
The Continuing Process 1-14
A SUMMARY OF THE TRADITIONAL TRAVEL DEMAND
FORECASTING PROCESS 1-21
Urban Activity Forecasts 1-24
Trip Generation 1-25
Table of Contents (continued)
Page
Trip Distribution 1-27
Mode Usage 1-28
Trip Assignment 1-30
SPECIFICATING Calibration, VALIDATION AND
FORECASTING 1-34
Chapter Two:INFORMATION NEEDS FOR TRAVEL DEMAND FORECASTING
INTRODUCTION 2-1
THE STUDY AREA 2-2
Defining the Boundaries 2-2
Subdividing the Area for Forecasting 2-3
URBAN ACTIVITIES 2-6
TRANSPORTATION SYSTEM 2-8
Network Geometry 2-8
Level of Service 2-11
TRAVEL INFORMATION 2-14
Origin-Destination Data 2-14
Initial Decisions 2-16
Processing Travel Data 2-21
iv
Table of Contents (continued)
Page
Chapter Three: TRIP GENERATION ANALYSIS
INTRODUCTION 3-1
BASIC CONSIDERATIONS IN TRIP GENERATION 3-2
Amount of Urban Activity 3-2
Character of Urban Activity 3-3
Other Considerations 3-3
Special Generators 3-4
SPECIFYING THE TRIP GENERATION MODEL 3-6
Trip Production Model Structure 3-7
Trip Attraction Model Structure 3-8
Internal-External Trip Generation 3-9
TRIP GENERATION MODEL CALIBRATION 3-12
Introduction 3-12
Developing Trip Production Rates 3-12
Developing Trip Attraction Rates 3-17
TRIP GENERATION MODEL APPLICATION 3-20
Trip Production Model Application 3-21
Trip Attraction Model Application 3-27
v
Table of Contents (continued) Page
Chapter Four: TRIP DISTRIBUTION ANALYSIS
INTRODUCTION 4-1
BASIS FOR THE GRAVITY MODEL FORMULA 4-2
GRAVITY MODEL PARAMETERS 4-6
Trip Productions and Attractions 4-6
Friction Factors 4-7
Socioeconomic Adjustment Factors 4-8
SAMPLE PROBLEM 4-11
CALIBRATING THE GRAVITY MODEL 4-17
Developing Friction Factors 4-17
Establishing Socioeconomic Adjustment Factors 4-24
APPLICATION OF THE CALIBRATED GRAVITY MODEL 4-26
Chapter Five: MODE USAGE ANALYSIS
INTRODUCTION 5-1
FACTORS THAT AFFECT MODE USAGE 5-3
Characteristics of the Trip maker 5-3
Characteristics of the Trip 5-4
Characteristics of the Transportation System 5-5
vi
Table of Contents (continued) page
MODELS FOR ESTIMATING MODE USAGE 5-9
Direct Generation Mode Usage Models 5-10
Trip Interchange Mode Usage Models 5-17
Consideration of Auto Occupancy 5-20
Chapter Six: TRIP ASSIGNMENT ANALYSIS
INTRODUCTION 6-1
HIGHWAY TRIP ASSIGNMENT 6-4
Highway Network File Development 6-4
Defining the Network 6-4
Coding the Network 6-7
Evaluating the Network File 6-9
HIGHWAY TRIP ASSIGNMENT PROCEDURES 6-13
Minimum Path Techniques 6-14
Minimum Path with Capacity Restraint 6-16
Multiroute Probabilistic Assignment 6-17
TRANSIT TRIP ASSIGNMENT PROCEDURES 6-20
Defining the Network 6-20
Coding the Network 6-23
Evaluating the Network File 6-25
vii
Table of Contents (continued)
Chapter Seven: REVIEW QUIZ
viii
PREFACE
This self-instructional text is part of a coordinated series
of user-oriented materials designed to aid planners in
understanding and applying the computer package UTPS -- the Urban
Transportation Planning System. The series of materials will also
include:
- audio-visual presentations,
- case study workbooks,
- computer-interactive programmed learning, and
- a UTPS general information manual.
Purpose of Text
This text is intended to provide a general introduction to the
travel demand forecasting process, and as such it does not deal
with specific UTPS programs. It is designed to give you some basic
background information so that you can begin to understand the
basis of computer programs.
ix
General Content
The presentation is based on what is known as the traditional
four-step travel demand forecasting process: trip generation, trip
distribution, mode usage, and network assignment. The text is
structured to teach you the "language" of travel demand forecasting
and to give you an understanding of the traditional process. This
information provides the foundation for continued study in travel
forecasting, covering subjects that are not dealt with here -- such
as sketch planning, behavioral modeling, and direct demand models.
The chapter on trip generation deals only with cross-
classification analysis. There are two reasons for this:
limitations on the size of this book, and the feeling that cross-
classification will overshadow regression analysis as the most
commonly used technique. The gravity model is the only trip
distribution technique discussed, for similar reasons. A list of
references is given at the end of each chapter in this text. You
will find these lists helpful in your continued study of topics for
which you require additional information.
Who the Text is For
This text was prepared primarily for technical people who are
not familiar with the urban transportation planning process. The
self-instructional format was chosen so that you can learn at your
own pace: you can take as much time as you need to get a thorough
understanding of the subject matter. The text is in a modular form
to encourage study of sections you find to be most useful. You can
stop as often as you like to review or think about the section that
you have just read.
x
This is not an ordinary textbook. There is a short quiz at the
end of each section. These quizzes are not tests; they are
designed so that you can gauge how well you are learning. Take the
time to answer the quiz questions. They reinforce the reading and
make learning much easier. The answers to each quiz are in the
lower-right-hand corner of the quiz page. Keep the answers covered
until you've made a reasonable attempt at answering the questions.
Since travel demand forecasting has a language all its own, an
extensive glossary of terms is provided in the back of the book.
To get abetter understanding of the material in this text, you
should first see two audio-visual presentations included in the
series of user-oriented materials: "An Overview of Urban Trans-
portation Planning" and the audio-visual complement to this text,
"An Introduction to Urban Travel Demand Forecasting. While
seeing those presentations is not essential, you will find them
very helpful.
xi
Acknowledgments
This text was prepared by Dr. G. Scott Rutherford of De Leuw,
Cather and Company, with assistance from Mark G. Bergstralh of Roy
Jorgensen Associates, Inc. , under contract DOTFH-11-8806. The
project was jointly funded by the Urban Planning Division of the
Federal Highway Administration (FHWA) and the Office of Planning
Methodology and Technical Support of the Urban Mass Transportation
Administration (UMTA). Dr. Ricardo dePaul Dobson and Dr. Leonard
Goldstein of FHWA served as contract managers. Reviews,
suggestions, and technical assistance were provided by Messrs.
David Gendell, James McDonnell and George Schoener, all of FHWA;
and Mr. Samuel Zimmerman of UMTA. The authors are indebted to
these individuals for their contribution to this text.
A substantial portion of the material for this text comes
directly or indirectly from existing FHWA and UMTA publications;
however, the authors take full responsibility for any errors or
omissions.
xii
1-1
Chapter One
TRAVEL DEMAND FORECASTING IN THE
URBAN TRANSPORTATION PLANNING PROCESS
Urban transportation planning is the process that leads to
decisions on transportation policies and programs. In this process,
planners develop information about the impacts of implementing
alternative courses of action involving transportation services,
such as new highways, bus route changes, or parking restrictions.
This information is used to help decision-makers (elected officials
or their representatives) in their selection of transportation
policies and programs.
The transportation planning process relies on travel demand
forecasting, which involves predicting the impacts that various
policies and programs will have on travel in the urban area. The
forecasting process also provides detailed information, such as
traffic volumes, bus patronage, and turning movements, to be used
by engineers and planners in their designs. A travel demand
forecast might include the number of cars on a future freeway or
the number of passengers on a new express bus service. It might
also predict the amount of reduction in auto use that would occur
in response to a new policy imposing taxes on central-area parking.
To help you understand the role of travel demand forecasting
in urban transportation planning, this first chapter is divided
into two sections: first, a brief look at the planning process, and
then forecasting as it is applied in planning. The urban
transportation planning process is discussed in the FHWA/UMTA
audio-visual presentation titled "An Overview of Urban Transporta-
tion Planning. Therefore, this discussion of planning is limited.
1-2
Click HERE for graphic.
THE URBAN TRANSPORTATION PLANNING PROCESS
1-3
Click HERE for graphic.
Organization
Transportation planning is an extensive undertaking. The need
for an effective organization is obvious; this phase of the
planning process ensures efficient direction from public officials
and guidance from the citizens of the study area.
The planning process must operate within the frame work of the
goals and objectives of the study area. Early in the process, ways
to promote interaction with public officials, public agencies ar.d
the citizens of the area must be defined to make sure that the
goals and objectives reflect current community values.
The governor of each state designates a Metropolitan Planning
Organization (MPO) to be responsible, with the state, for urban
transportation planning. Usually,. one MPO is designated for each
urbanized area or group of adjacent areas.
The MPO Is planning activities are carried out in cooperation
with state and local agencies. The staffs of other local agencies
can be contracted by the MPO for work on some of the elements of
the planning process.
The organization phase, then, is designed to ensure that all
available inputs are used to the fullest extent possible to
organize planning to meet the needs, goals, and objectives of the
community.
1-4
Click HERE for graphic.
Planning Work Program
Planning work programs are developed to make sure that the
planning is conducted efficiently and comprehensively, and that it
addresses all pertinent issues. As you can see in the diagram,
there are two parts to this phase: the prospectus and the unified
planning work program.
The prospectus establishes a multi-year framework for the
planning process. It summarizes the planning procedures, discusses
the important issues that will be addressed during planning,
describes the responsibilities of each agency that is participating
in planning, and describes the status of all elements in the
planning process.
The unified planning work program has two functions. It
describes all urban transportation and transportation-related
planning activities that are anticipated over the next year or two,
and it documents work to be performed with federal planning
assistance.
1-5
QUIZ
Urban transportation planning leads to decisions on
transportation __________ and programs. Travel demand
forecasting predicts the impacts on __________ that these
policies and programs will have.
MPO stands for_______________________ ________________________.
It is designated by the _____________ to carry out planning
activities within the state.
policies; travel
Planning work programs consist of two parts:
Metropolitan Planning
- developing a _____________,and Organization;
governor
- developing a __________ __________ prospectus; unified
_________ ___________________. planning work
program
1-6
Click HERE for graphic.
The Transportation Plan
There are two elements in the preparation of a comprehensive
transportation plan to guide improvements to the transportation
system. The Transportation Systems Management (TSM) element and
the Long-Range element work together to formulate a transportation
improvement program. Let's look more closely at these two
elements.
1-7
Click HERE for graphic.
The Transportation Systems Management Element
The key word here is management. The TSM Element is concerned
with making existing systems as efficient as possible and with
making provisions for an area's short-range transportation needs.
Automobiles, taxis, trucks, terminals, public transit, pedestrians,
and bicycles are all parts of the urban transportation system.
There are four basic categories of actions to increase the
efficiency of the different parts of the system.
1. Actions to ensure the efficient use of existing road space.
These actions include measures to manage and control the
flow of motor vehicles. Installing reversible lanes to
accommodate rush-hour traffic and improving intersection
capacities are examples.
2. Actions to reduce vehicle use in congested areas.
Encouraging carpooling and other forms of ride-sharing are
examples.
1-8
3. Actions to improve transit service. People can be
encouraged to use transit by such actions as providing
park-and-ride services from fringe areas to the central
business district.
4. Actions to improve internal transit management efficiency -
- for example, developing management tools, such as
information systems, or marketing campaigns.
To decide which actions to implement, you need a clear
understanding of how each would affect the transportation system
and the region as a whole. That is, what are the results of
selecting a particular course of action? "Planning tools" are
developed so that you can predict impacts and provide the
information necessary for decision-makers to evaluate alternatives
and select the best courses of action.
Travel demand forecasting tools provide input to this process
by predicting travel impacts on transportation systems and their
users. These.predictions are one part of the planning process
where travel demand forecasting plays an important role.
1-9
QUIZ
TSM means _____________________ ____________________
_____________________, with the emphasis being on________.
The major purpose of the TSM element is to increase the _________
of our existing systems.
Installing exclusive bus lanes is an example of Transportation
Transportation System Management. True or false ? Transportation
systems
Management,
management
efficiency
Reducing vehicle use in congested areas is an True
example of a TSM action. True or false? True
1-10
Click HERE for graphic.
The Long-Range Element
To make provisions for the long-range transportation needs of the
urban area, the long-range element identifies facilities to be
constructed, major changes to be made to existing facilities, and
long-range policy actions. This element of the transportation plan
might, for example, consider future land development policies by
adding a highway link, or installing a busway system.
As in the TSM element, many long-range alternatives must be
evaluated before decisions can be made. Thus, planning tools for
analysis of long-range alternatives must be developed. These tools
also provide information to help decision-makers select the most
promising alternatives.
Once again, travel demand forecasting plays an important role as a
major contributor to the planning tools used in evaluating
alternatives.
1-11
Click HERE for graphic.
Plan Refinement
After the long-range plan elements have been selected, the
plan is refined by further detailed studies. In the plan
refinement stage, the corridor in which improvements are planned
might be studied in detail, along with studies of various types of
technology --for example., buses versus rapid transit -- and
studies to determine the proper staging (scheduling) of the planned
projects.
Travel demand forecasting plays a role here in refining
estimates of such things as patronage, market areas, congestion and
turning movements.
1-12
Click HERE for graphic.
The Transportation Improvement Program
After selections have been made. in the Transportation Systems
Management and Long-Range elements, and after plan refinement, a
Transportation Improvement Program is developed. This program
ensures that the transportation plan will be implemented in an
orderly, efficient manner and represents a statement as to how the
transportation system will improve in the next few years. The
program has two major elements: the staged multi-year element and
the annual element.
The staged multi-year element describes the general aspects of the
program over the next three to five years. This element indicates
priorities among the projects identified for implementation, groups
projects into appropriate staging periods, and makes estimates of
costs and revenues, for the program period.
The annual element identifies the details of projects that will be
implemented within the next year. For each project, the annual
element contains:
- a basic description,
- costs involved,
- revenue sources, and
- the local agency that is responsible for
implementation.
1-13
QUIZ
The long-range element of the transportation plan is primarily
concerned with capital-intensive improvements. True or
false_______?
Adopting development policies that complement efficient
transportation systems is an example of the long-range element.
True or false?______________________________
In plan refinement, we are concerned with specific
___________in which improvements are
planned. True
True
The__________________ _______________ corridors
is a statement of improvements
to the transportation system in the next few years. Transportation
Improvement
Program
1-14
Click HERE for graphic.
The Continuing Process
The continuing process (the stage that most urban areas are in now)
consists of monitoring changes that could make it necessary to mod-
ify the transportation plan, updating the data that serve as a base
for planning, updating the methods used in transportation
planning -- including those for travel forecasting -and reporting
on activities and findings.
1-15
Transportation studies have been made in all areas with populations
over 50,000. Planners have made inventories of the characteristics
that affect travel in urban areas. These inventories, which make up
the data base, include:
- population, - travel,
- land use, - laws and
ordinances,
- economic activity, - financial
resources, and
- transportation systems, - community
values.
In the continuing process, one task is to update these
inventories as necessary to make sure that the data base for
planning is complete and accurate. This update is done using
secondary sources, such as the census or small sample surveys.
Complete large-scale surveys are no longer feasible due to their
high cost.
Monitoring changes is extremely important. You can imagine
how a decision to install a major shopping center or sports arena
near your city would affect transportation and the transportation
plan. In the continuing process, we monitor both the
transportation system and its performance.
In addition to monitoring changes and updating the data base,
the continuing process provides information to interested local
agencies in the form of reports and technical assistance as
required.
1-16
Through careful studies of data on people's travel behavior,
relationships have been developed to predict how many trips people
will make, where they will go, by which mode of transportation, and
by which specific route. These relationships are the basis for
travel demand forecasting and must be reviewed and reevaluated, if
necessary, in the continuing process.
The continuing process ensures that the transportation plan will
respond to the area's transportation needs -- needs that are
constantly changing. The level of effort involved in the
continuing process will depend on the size of the urban area and
the complexity of its problems.
1-17
Click HERE for graphic.
1-18
Click HERE for graphic.
There are several different techniques for travel demand
forecasting from which to choose, depending on the requirements of
the analysis. These techniques differ in complexity, cost, level
of effort, sophistication and accuracy, but each has its place in
travel forecasting. The graph at right shows the general
relationship of these planning tools to their most likely
applications. Each tool is explained briefly below.
SKETCH TOOLS: Sketch planning is the preliminary screening of
possible configurations or concepts. It is used to compare a large
number of proposed policies in enough analytical detail to support
broad policy decisions. Useful in both long- and short-range
regional planning and in preliminary corridor analysis, sketch
planning, at minimum data costs, yields aggregate estimates of
capital and operating costs, patronage, corridor traffic flows,
service levels, energy consumption, and air pollution.
The planner usually remains in the sketch planning mode until
he completes his comparisons of possibilities or finds a strategic
plan worthy of consideration at a finer level of detail.
TRADITIONAL TOOLS: Traditional tools treat the kind of detail
appropriate to tactical planning; they deal with many fewer
alternatives than sketch tools, but in much greater detail. Inputs
include the location of principal highway facilities and delineated
transit routes.
1-19
At this level of analysis the outputs are detailed estimates
of transit fleet size and operating requirements for specific
service areas, refined cost and patronage forecasts, and level-of
service measures for specific geographical areas. Household
displacements, noise, and aesthetic factors can also be evaluated
The cost of examining an alternative at the traditional level
is 10-20 times its cost in sketch planning, although default
models, which dispense with many data requirements, can be used for
a less expensive first look. Apparently promising plans can be
analyzed in detail, and problems uncovered at this stage will
suggest a return to sketch planning to accommodate new constraints.
MICRO-ANALYSIS TOOLS: Micro-analysis tools are applicable as
the time to implement a project grows near. They are the most
detailed of all planning tools. At this level of analysis, one may
wish, for example, to make a detailed evaluation of the extension,
rescheduling., or repricing of existing bus service; to analyze
passenger and vehicle flows through a transportation terminal or
activity center; or to compare possible routing and shuttling
strategies for a demand-activated system. Final analysis at this
level is prohibitively expensive except for subsystems whose
implementation is very likely, and whose design refinements would
bring substantial increases in service or significant reductions of
cost. it is most effective in near-term planning when a great many
outside variables can be accurately observed or estimated. It is
sometimes necessary, however, to use micro-analysis tools to
supplement the output of traditional longer-range planning.
Although traditional tools are the focus of this text, we
should keep in mind their relationship to sketch and micro-analysis
tools.
1-20
QUIZ
The continuing planning process is a process of
monitoring,_____________, and _________________.
Travel demand forecasting provides important inputs to the TSM
element, to the ________ element, in plan refinement, and in the
__________ process.
To quickly investigate a large number of alternatives, the
most appropriate method for travel forecasting is ____________.
updating,
reporting
sketch planning
long-range,
continuing
That was a brief look at urban transportation planning. Now, go on
to a summary of the Traditional Travel Demand Forecasting
Process, a short discussion on one method planners use to forecast
the amount and character of travel. This material will introduce
you to the entire process, after which its components are discussed
in some detail.
1-21
A SUMMARY OF THE TRADITIONAL TRAVEL DEMAND FORECASTING PROCESS
Click HERE for graphic.
In general, travel demand forecasting attempts to quantify the
amount of travel on the transportation system. Demand for
transportation is created by the separation of urban activities.
The supply of transportation is represented by the service
characteristics of highway and transit networks. These basic
relationships are shown in the diagram at right.
There are many methods available to forecast travel demand. A
discussion of all forecasting is beyond the scope of this text;
our major objective is to provide a basic foundation for the
process. Therefore, what has become known as the "traditional
four-step process" will be discussed.
The process considered here has been developing over the past
25 years for forecasts of urban travel. In terms of the planning
process, this discussion will focus on planning tools for the long-
range element of the transportation plan, as modified and updated
in the continuing process. Understanding the traditional approach
will greatly help your understanding of other aspects of travel
demand forecasting. However, this presentation is not intended to
endorse any particular method for doing travel forecasting -- it
only represents an example of how several agencies have done travel
forecasting in the past.
1-22
There are four basic phases in the traditional travel demand
forecasting process.
- Trip Generation forecasts the number of trips that will
be made.
- Trip Distribution determines where the trips will go.
- Mode Usage predicts how the trips will be divided among
the available modes of travel.
- Trip Assignment predicts the routes that the trips will
take, resulting in traffic forecasts for the highway
system and ridership forecasts for the transit system.
The following page shows how these phases fit together into the
forecasting process.
1-23
Click HERE for graphic.
Urban activity forecasts provide information on the location
and intensity of future activity in an urban area and provide
primary input to trip generation.
Descriptions of the highway and transit networks provide the
information necessary to define the "supply" of transportation in
the area; the four phases predict the travel "demand.
The feedback arrows shown represent checks of earlier
assumptions made on travel times and determine if adjustments are
necessary. If not, the process is complete.
1-24
Click HERE for graphic.
Urban Activity Forecasts
Urban activity forecasts provide estimates of where people
will live and where businesses will locate in the future. These
forecasts also include the intensity of activity, such as the
number of households and number of employees of businesses. The
figure on the right shows the number of households and employment
for two zones of the mythical 5- zone town known as UPTOWN. An
actual forecast might include several additional factors and
considerably more detail, such as employment by type and households
by socioeconomic group.
These forecasts are done for small parcels of land called
zones. Zones vary in size, with the smallest about the size of a
block in the downtown area and the largest on the urban fringe
being several square miles. More about establishing zones will be
described in the next chapter.
Zonal urban activity forecasts are based on the following:
- total urban area population and employment estimates;
- location behavior of people and businesses; and
- local policies regarding land development,
transportation, zoning, sewers, etc.
These activity forecasts are direct inputs to the next stage of the
process, trip generation analysis.
1-25
Click HERE for graphic.
Trip Generation
Trip generation is the process by which measures of urban
activity are translated into numbers of trips. For example, the
number of trips that are generated by a shopping center is quite
different from the number of trips generated by an industrial
complex that takes up about the same amount of space. In trip
generation, the planner attempts to quantify the relationship
between urban activity and travel.
The inventory data discussed earlier is.the analyst's input
for trip generation analysis. Surveys of travelers in the study
area show the numbers and types of trips made;by relating these
trips to land use patterns, the analyst is able to forecast the
number of trips that will be made in the future, given forecasts of
population and other urban activity.
1-26
Here's a simplified example. The UPTOWN survey data show that
Zone I has employment of 900 people and, from the figure above,
attracts 4, 511 trips. By dividing the trips by employees, we find
about 5 trips attracted per employee. This rate can then be used
to predict attractions for future employment levels.
The output of trip generation analysis is a table of trip ends
-- the number of trips that are produced and the number that are
attracted.
As mentioned earlier, the study area is divided into zones for
analysis purposes. After trip generation analysis, the planner
knows how many trips are produced by each zone, and how many are
attracted by each zone. In addition, the planner knows the
purposes for the trips the trips are put into several categories,
like trips from home to work, or home to shop.
There are basically two tools for trip generation analysis:
multiple linear regression and cross-classification. In this text,
you'll see the use of cross-classification because this procedure
is easy to understand and update, and because it produces reliable
results.
1-27
Click HERE for graphic.
Trip Distribution
After trip generation, the analyst knows the numbers of trip
productions and trip attractions each zone shown below will have.
But, where do the attractions in Zone I come from and where do the
productions go? What are the zone-to-zone travel volumes?
Trip distribution procedures determine where the trips
produced in each zone will go -- how they will be divided among all
other zones in the study area. . The output is a set of tables that
show the travel flow between each pair of zones. The figure on the
right shows where Zone l's trip productions are distributed.
The decision on where to go is represented by comparing the
relative attractiveness and accessibility of all zones in the area
-- a person is more likely to travel to a nearby zone with a high
level of activity than to a distant zone with a low level of
activity.
There are several types of trip distribution analyses: the
Fratar method, the intervening ,, opportunity model, and the
gravity model. In this text, we will discuss only the gravity
model, because it is the most widely used method.
1-28
Click HERE for graphic.
Mode Usage
In this phase of travel demand forecasting, we analyze
people's decisions regarding mode of travel -- auto, bus, train.,
etc.
In our flow chart of the travel demand forecasting process,
mode usage comes after trip distribution. However, mode usage
analyses can be done at various points in the forecasting process.
Mode usage analyses are also commonly done within trip generation
analyses. The most common point is after trip distribution, since
the information on where trips are going allows the mode usage
relationship to compare the alternative transportation services
competing for users.
1-29
Before we can predict how travel will be split among the modes
available to the traveler, we must analyze the factors that affect
the choices that people make. Three broad categories of factors
are considered in mode usage:
- the characteristics of the trip maker,
- the characteristics of the trip, and
- the characteristics of the transportation system.
The planner looks at how these characteristics interact to
affect the trip maker's choice of mode. When the relationships
have been discovered, the planner can predict how the population of
the future will choose from among the modes that will be available.
Generally, at this point in the forecasting process some
consideration is given to predicting the number of occupants in
autos for those choosing that mode. This consideration of auto
occupancy can either be included in the mode usage relationship
with each level of occupancy being considered a separate mode, or a
separate relationship might be developed.
1-30
Trip Assignment
Trip assignment is the procedure by which the planner predicts
the paths the trips will take. For example, if a trip goes from a
suburb to downtown, the model predicts which specific roads or
transit routes are used. The trip assignment process begins by
constructing a map representing the vehicle and transit network in.
the study area. The network maps show the possible paths that
trips can take.
The intersections (called nodes) on the network map are
identified, so that the sections between them (called links) can be
identified. After the links are identified by nodes, the length,
type of facility, location in the area, number of lanes, speed, and
travel time are identified for each link. If transit is available,
additional information, which identifies fares, headways (time
between vehicles), and route descriptions, is included on a
separate network. This information allows the computer to
determine the paths that the traveler might take between any two
points on the network and to assign trips between zones to these
paths.
The output of trip assignment analysis shows the paths that
all trips will take, and therefore the number of cars on each
roadway and the number of passengers on each transit route.
1-31
Using these analyses of trip generation, trip distribution,
mode usage, and trip assignment, the planner can obtain realistic
estimates of the effects of policies and programs on travel demand.
Once travel demand is known, the planner can assess the performance
of alternative transportation systems and identify various impacts
that the system will have on the urban area, such as energy use,
pollution, and accidents. With information on how transportation
systems perform, and the magnitude of their impacts, planners can
provide decision-makers with some of the information they need to
evaluate alternative methods of supplying a community with
transportation services.
1-32
QUIZ
Urban activity forecasts are a major input to the ________
phase.
Mode usage analyzes the characteristics of the traveler, the
trip, and the _________ ________.
Trip generation is the process by which the analyst predict the
_______ that will be made in the future.
trip generation
Trip distribution forecasts the ___________ of
the trip productions. transportation
system
trip ends or
productions and
Mode usage is always analyzed after the trip
generation analysis. True or false? attractions
destinations
false
1-33
One of the first steps in trip assignment is the preparation of
a___________ ___________.
The trip assignment step predicts the ________trips will take
through the network and hence the _________ of travel on various
segments of the network.
network map
Checking initial assumptions about travel times between
zones is a task known as _________.
paths, volumes
feedback
1-34
SPECIFICATION, CALIBRATION, VALIDATION AND FORECASTING
Before forecasting travel, a considerable effort must be made
to analyze inventory data and establish relationships among travel
choices and several other variables. Discovering the reasons for
making travel decisions, such as where or how to travel, is done in
two steps. First, the types of models to be used and their
variables are specified; and second, those models are calibrated to
reproduce observed travel. Follow the figure on page 1-36 to see
how these steps relate to one another.
In model specification, a choice must be made among several
mathematical formulations and many possible variables. Research has
shown which formulations and variables will probably yield the best
results; therefore, the task is somewhat simplified to testing a few
options. During this step, the level of analysis for the models
must be specified; that is, a decision must be made whether to model
individual travel behavior or that of a larger group, such as a
zone.
The calibration process is basically an attempt to duplicate
travel for the year in which calibration data is available. The
year for which data is available is termed the "base year." Surveys
are taken to see how people travel in the study area; base year data
is used to calibrate the trip generation, trip distribution, mode
usage, and trip assignment relationships separately. The
calibration process also includes intuitive tests of models to see
if the variables and their coefficients are reasonable. The
important point to remember in the calibration process is that the
output of each model is compared individually to base-year data.
1-35
After the models are calibrated they should go through a
validation process by applying these calibrated models sequentially
in the base year. These models are applied in exactly the same way
as they would be applied in the forecast year. This is necessary to
see if the procedure produces reasonable comparisons to the base
year observed data. For example, if models are calibrated using
household information (income, family size, access time to bus,
parking costs, etc.), they should be validated at the same level of
aggregation to be used in forecasting (average zone income, average
zone access time to bus,, etc.). This validation will show how well
the entire chain of calibrated models can forecast observed travel
in the same way they will be used to forecast future travel. If the
series of models cannot produce traffic volumes and transit
riderships similar to what is observed on roadways and bus lines,
then the models must be reevaluated and appropriate adjustments
made.
The validation process can continue in future years by
comparing measured traffic volumes with model estimates. This
continued reappraisal of the travel forecasting models assures that
forecasts will remain as accurate as possible.
1-36
Click HERE for graphic.
In the forecasting process, estimates of new levels of activity
in the zones and/or alternative transportation system descriptions
are used. Then, the calibrated models produce estimates of travel
on the alternative systems to be tested.
This diagram shows the specification, calibration, validation
and forecasting process.
1-37
Click HERE for graphic.
Before proceeding, you should be aware that the travel demand
forecasting process is not without problems.
You will notice after studying the diagram at right that the
feedback to the Highway and Transit Networks implies that these
networks are updated and the process begins again with new forecasts
of Urban Activity and Trip Generation based on the results of
Highway and Transit Assignments. This seems logical. The pattern
of urban activities should be influenced by the transportation
system. In fact, this update seldom takes place, because of time
and budget constraints. The process, as practiced, therefore
implies that the location and intensity of activities and trip gen-
eration are independent of the transportation system. Clearly, this
is not the case.
1-38
The forecasting process has also come under criticism because
of its sequential nature; first, a decision to make a trip is
modeled, then a destination, then a mode, and finally a path. Is
this the way people make travel decisions? Some analysts have
thought not, and have attempted with varying degrees of success to
place the entire process within a single model.
The forecasting techniques are being improved continually in
terms of their theoretical basis and ability to respond to changing
requirements. Recent research is aimed at overcoming the problems
mentioned and enhancing the capabilities to represent travel
behavior accurately.
In spite of some criticisms, the current forecasting process
has been shown to work quite well over extensive periods of time
with major changes in the amount of activity in the area, rising
affluence of households, and changes to the transportation system.
1-39
QUIZ
Specification decides which model formulation will be used in the
forecasting process. True or false ?
The process of working with the models to reproduce travel be-
havior is known as ____________.
To use the calibrated models for forecasting, new levels of
activity and/or alternative _______ _______ are input.
true
calibration
transportation
system
descriptions
This section completes Chapter One. If you feel you have a
general knowledge of urban transportation planning and how travel
forecasting relates to planning, go on to the next chapter. If you
are unsure, go back and review this chapter.
1-40
For further study on introductory transportation planning or
forecasting, review the following documents:
Introduction to Urban Travel Demand Forecasting, UMTA, March
1974. "Summary" - NTIS - PB 23 847/AS $3. 75. "Volume I -
Demand Modelling. " - NTIS - PB #23 6- 848/ AS $9.25. A
comprehensive instructional text on modern demand modeling
approaches.
Computer Programs for Transportation Planning -
PLANPAC/BACKPAC General Information, FHWA, April 1977. Reprint
available from FHWA, HHP-22, Washington, D. C. 20590. This
reference provides very useful technical information on several
phases of the analytical planning process.
UTPS Users Guide (on UTPS tape), Chapter II "Planning
Technical Review." This tape provides a terse outline form review
of planning steps useful as a refresher but not as a basic
instructional text.
A Review of Operational Urban Transportation Models, DOT,
April 1973 NTIS - PB #222-109 $6.75. This provides a good review
of models of land use, travel demand, network analysis and some
evaluation aspects.
2-1
Chapter Two
INFORMATION NEEDS FOR
TRAVEL DEMAND FORECASTING
INTRODUCTION
Prior to embarking on the forecasting process as outlined in
Chapter One, a significant amount of work is necessary to accumulate
the information that drives the forecasting process.
These information needs include broad items like defining the
area for which forecasting will be done and specific items such as
identifying detailed information about streets and bus routes. In
order to understand the travel forecasting process and its
terminology, a brief exposure to its information requirements is
essential. For the purpose of this discussion, information needs
have been divided into four broad categories, as follows:
- the study area, - transportation system, and
- urban activates, - travel.
With knowledge of these four categories of information, the
transportation planner has the data necessary to begin the travel
demand forecasting process.
2-2
THE STUDY AREA
Defining the Boundaries
Obviously, before forecasting travel for an urban area, the
planner must clearly define the exact area to be considered. This
planning area generally includes all the developed land plus the
undeveloped land that the urban area will encompass in the next 20
to 30 years.
The boundary of the planning area is known as the cordon line.
In addition to considering future growth, the establishment of the
cordon line might take into account political jurisdictions, census
area boundaries, and natural boundaries. The cordon should
intersect a minimum number of roads to save on subsequent interview
requirements.
2-3
Click HERE for graphic.
Subdividing the Area for Forecasting
The study area must be divided into analysis units to enable the
planner to link information about activities, travel, and
transportation to physical locations in the study area.
The transportation analysis units are known as zones. These
zones vary in size depending on the density or nature of urban
development. In the central business district (CBD), zones maybe as
small as a single block and in the undeveloped area they may be as
large as 10 or more square miles. An area with a million people
might have 600 to 800 zones, and an area of 200, 000 people might
have 150 to 200 zones. The zones attempt to bound homogeneous urban
activities; that is, a zone may be all residential, all commercial,
all industrial, etc. Zones also should consider natural boundaries
and census designations.
An important consideration in establishing zones is their
comparability with the transportation network to be used. As a
general rule, the network should form the boundaries of the zones.
Zones are usually grouped into larger units known as districts.
Districts might contain 5 to 10 zones, and a city of a million might
have about 100 districts. Districts often follow travel corridors,
political jurisdictions, and natural boundaries such as rivers.
2-4
Click HERE for graphic.
Dividing the area into zones, districts, sectors, and rings is a
great help to the transportation planner in organizing information
and interpreting forecasting results.
2-5
QUIZ
The study area boundary is known as the ___________.
The study area is divided into small analysis units
known as __________.
cordon line
Zones can be grouped to form ____________.
zones
districts
2-6
URBAN ACTIVITIES
Once the study area has been divided into appropriate analysis
units (zones and districts), information about activities in these
areas can be gathered. Knowledge about the forecasting procedure is
essential at this point, since only data relevant to the calibration
and forecasting, process need be gathered. Collecting data that is
not eventually used is a wasteful practice.
In Chapter One, urban activity forecasting was briefly
mentioned as the source of information on activities that might
influence travel in the urban area. These activity forecasts are
done on a zonal basis., providing the intensity and characteristics
of activities in each zone in the study area. The results of a
typical activity analysis provide the planner with present levels of
activities in zones to help in predicting future levels that provide
a basis for forecasting.
2-7
Zone activity information might appear as follows for a few sample
zones.
Zone 39: Heavy industry
Employment -- 400 blue collar
-- 123 white collar
Zone 3: Central Business District
Employment -- 623 retail
-- 1200 non- retail
Zone 136: Suburban shopping center
Parking spaces -- 700
Employment -- 120 retail
-- 43 non-retail
Zone 89: Residential
Population -- 1200
Households -- 400
Average Income -- $12,000
More detailed information might be required depending on the
modeling requirements. However, the trend is to base models on the
simplest information possible from which reasonable forecasts of the
information can be made and the models applied to these future
forecasts.
2-8
TRANSPORTATION SYSTEM
The transportation system.allows the urban activities to
communicate with one another; that is, people travel to work, to
shop, and to visit friends. This communication takes place on
streets, highways, subways, and bus routes.
If an urban area were a flat plane with no obstacles to
movement, the transportation system could be described quite easily.
In cities, this is clearly not the case; some areas are not directly
connected, some roads are faster than others, and some areas have no
transit service. This variation in accessibility requires the
planner to describe the transportation system in terms of its
geometry (what's connected with what) and its level of service (how
well points are connected).
Network Geometry
The transportation system consists of networks that represent
the available modes (auto, bus, etc.). The network description is an
abstraction of what is actually on the ground, and as such does not
include every local street or collector street in the area. A
network description is developed to describe auto and truck travel,
with a separate description for transit, if transit is a
consideration. These descriptions could include the geometry of the
transportation system.
2-9
Click HERE for graphic.
Network geometry includes numbering the intersections (called
nodes for assignment purposes). Numbering the nodes allows us to
identify the segments between them (called links). In transit
networks we also identify groups of links over which specific routes
pass (called lines). This geometric description of the
transportation network shows all possible ways that travel can take
place between points in the area.
In the illustration at right, link 34-37 is defined by its
nodes; if link 34-37 is a two-way street, link 37-34 would also be
defined.
In the network description, zone centroids (centers of
activity) are identified; they are connected to nodes by imaginary
links called centroid connectors. Centroids are used as the points
at which trips are "loaded" onto the network.
Sample vehicle and transit network maps are shown on the next
page for the mythical UTOWN.
2-10
Click HERE for graphic.
2-11
Level of Service
Once the transportation network has been described in terms of
how points can be connected, it is necessary to quantify the ease
with which these connections are made. Whether two zones are
connected by an arterial street or freeway is an important
distinction in travel forecasting. Travel speeds and ]link capacity
on the freeway would probably be greater than on the arterial
street; this level of service difference must be quantified and
included as part of the transportation system description.
For the vehicle network description, specific items must be
collected to determine the level of service on each link. These
requirements include physical items such as the link length and
number of lanes, whether the type of facility under consideration is
a freeway, arterial, etc., and where in the urban area the link is
located.
2-12
For the transit network, the description needs some additional
information about the service characteristics of the system. The
transit network is different from the vehicle network in that both
the links and the sequence of links to establish routes (lines) must
be identified. In that both links and lines must be represented,
the network description is two-tiered. The first level is a system
of links that define the segments of travel facilities between
nodes. Travel times, speeds, and distances are required for links.
The second level -- the network of lines -- overlays the links and
defines the fixed routes. This requires that each line be
identified individually along with its service headways and the
series of links over which it travels.
With these items of information about the highway and transit
networks, the planner can determine how each zone in the area is
connected in terms of time and cost to all other zones and hence the
level of service the transportation system provides.
The measures of time and cost can usually be expressed in terms
of a general measure known as impedance, which represents a
"generalized cost" of traversing a particular link.
2-13
QUIZ
Once zones have been established) information about the character
and intensity of ______ is gathered.
The transportation system is usually described in terms of its
______ and level of service.
Numbering the _______ on the network provides definitions of
between the nodes.
Time and/or cost of travel between zones is an indicator
of the _____ provided by the transportation system.
activities
geometry
A transit ______ describes a specific route in the
transit network. nodes, links
_______ is a combination of time and cost used level of service
to describe the difficulty in using any link.
line
impedance
2-14
TRAVEL INFORMATION
Information on how, when, and where people are currently
traveling is of obvious importance in the forecasting process. This
information is studied to determine the underlying factors causing
people to make certain travel decisions so that models can be
calibrated and used to forecast how people will travel in the future
or in response to changing conditions now.
Origin-Destination Data
To be able to analyze trip-making, the planner needs
information on where trips come from, where they go, by what mode,
for what purpose, and characteristics about the tripmaker and
activities at the origin and destination of the trip. This
information is termed origin destination data.
Origin-destination survey data are generally available in
sufficient detail and of proper statistical stability to allow
accurate estimates of the model parameters. The home interview
survey provides the most complete and accurate information for
computing the parameters. However, since home interviews have been
done in all large cities and are extremely expensive,
they are no longer done for large samples. Small-sample
surveys are now done to update past surveys.
2-15
Although the best data are collected through person-to-person
interviews, other techniques can provide satisfactory results.
Telephone interviews, on-board transit surveys, mailed
questionnaires, and pick-up postal cards are sometimes used.
No matter which technique is used, the data must be reliable:
they must be complete and unbiased.
The following information is typical of that collected about
each household that is contacted in home interviews:
- address of household,
- number of persons who live there,
- number of cars available to the residents,
- occupation of the head of the household, and
- income of the household.
For each trip made by a member of the household, the following
information is usually collected:
- identification of the person who made the trip, address
of the origin of the trip,
- address of the destination of the trip,
- time at which the trip started,
- time at which the trip ended,
- purpose at the origin,
2-16
- purpose at the destination,
- mode of transportation,
- type of activity at the origin, and
- type of activity at the destination.
In addition to home interview surveys, truck-and-taxi surveys
and external cordon surveys are generally made to provide a complete
picture of travel in the study area.
Counts of vehicle traffic and transit ridership are also made
at various locations throughout the urban area. They can be used to
check on the accuracy of the surveys and later, to check on the
functioning of the models.
Initial Decisions
Before proceeding with the processing of travel data, some
decisions must be.made. Here are two:
- whether to use vehicle trips or person trips in the
analysis, and
- how to stratify trip purpose classifications.
2-17
Vehicle Trips or Person Trips
Deciding whether to use vehicle trips or person trips is one of
the first decisions the planner must make. The decision is directly
related to the needs and objectives of the study. Areas with
significant public transit issues, regardless of size, require mode
usage models to be developed making person trip models for trip
generation and trip distribution necessary.
Stratification of Trip Purpose Classifications
The planner must decide how many and which trip purpose
classifications to use.
The use of classifications ranges from only one to nine or more,
depending on the size of the study area and the scope and objectives
of the study.
The number of trips in each classification should be considered.
The home interview survey data can be studied to make sure that
enough trips are recorded in each classification used to allow
accurate forecasting.
The amount of data preparation time, computer time, and analysis
time should also be considered, in deciding on the stratification.
2-18
The trip purpose breakdown in most widespread use is as follows.
1. Home-based work -- trips between a person's home and
place of employment for the purpose of working.
2. Home-based other -- trips between a person's home and any
other destination for any other purpose.
3. Non-home-based -- trips that have neither end at home,
regardless of purpose. These may include truck and taxi
trips.
4. Internal-external trips -- trips with one end inside the
study area and one outside the study area.
5. Through trips - - trips that have neither end in the study
area but pass through it.
6. Truck and taxi trips -- if required, but often included as
part of non-home-based trips.
2-19
Defining Productions and Attractions
It is generally felt that household characteristics related to
travel are easier to identify and forecast. For this reason,
planners use the home as a base to predict travel. This convention
requires a distinction in the terms origin, destination, production,
and attraction, which often cause confusion in transportation
planning. A trip origin is always the beginning point of the trip,
and a trip destination is always the ending point. This doesn't
hold true for productions and attractions, however. Here are a few
rules to follow.
Rule 1: Trips that either begin or end at the
traveler's home are produced at the home end.
Rule 2: Trips that either begin or end at the
traveler's home are attracted to the non-home
end.
Rule 3: Trips that begin at a non-home location and
end at another non-home location are produced
at the origin and attracted to the
destination.
The example on the next page should help.
2-20
A traveler goes from home, to the office, to a store, then back
home.
Click HERE for graphic.
Trip number 1 was produced at home and attracted to the office --
Rules 1 and 2 -- therefore a home-based work trip.
Trip number 2 was produced at the office and attracted to the store
-- Rule 3 -- therefore a non-home-based other trip.
Trip number 3 was produced at home and attracted to the store --
Rules 1 and 2 --therefore a home-based other trip.
2-21
Processing Travel Data
Developing Calibration Files
From the edited trip information, a file that includes an entry
for each trip surveyed can be developed. Each entry represents a
complete description of one trip; this description includes
information about the trip maker along with where the trip was
produced, where it was attracted, its purpose, mode. etc.
The calibration file can then be used in various phases of the
forecasting process to provide the observed travel information
against which models are calibrated.
Click HERE for graphic.
Developing Trip Tables
The calibration file just mentioned provides the basic data to
generate what is called a trip table. A trip table is simply a
matrix showing how many trips are interchanged between the various
zones in the study area. A simple three-zone trip table is shown at
right. This table indicates that 10 trip productions from Zone I
stay in Zone 1, 30 go to Zone 2, 20 to Zone 3 , etc. The sum of the
rows of the trip table is equivalent to zone trip productions, and
the sum of the columns is zone trip attraction. For example, the
table shows Zone I produces 60 trips and attracts 70 trips.
A trip table is produced for each trip purpose category
defined.
2-22
Developing Initial Travel Impedances
You will recall from Chapter One that the travel forecasting
process relies on feedback to check initial assumptions on travel
times. Initial travel times are best derived by loading the trip
tables produced by the calibration file onto the network that
provides estimates of zone-to-zone travel times.
The network assignment procedure will be discussed in detail in
Chapter Six. This brief description is provided because impedance
estimates are required in trip distribution and mode usage analysis,
which are discussed prior to network assignment.
2-23
QUIZ
Travel information allows the planner to ________ the necessary
travel models.
Extensive home interviews are now required to obtain travel data.
True or false? ________.
A trip from a person's office to home is a non-home-based trip.
True or false? _________.
calibrate
Internal-external trips have one end inside the area and False
the other end _______ the area.
False
A _____ file is used to establish travel outside
models.
calibration
For a production/attraction trip table, the row
sums are the zone _______ ______. trip productions
2-24
For further reading about surveys and information needs, refer to
the following:
Urban Origin-Destination Surveys, 1973. Available from FHWA, HHP-
20, Washington, D. C. 20590.
Urban Mass Transportation Travel Surveys, FHWA, August 1972.
Available from Government Printing Office, Stock No. 5001- 0037,
$2. 10.
3-1
Chapter Three
TRIP GENERATION ANALYSIS
INTRODUCTION
Click HERE for graphic.
To predict the amount of travel that will take place in the
forecast year, or to analyze current travel, the planner must
understand -and quantify -- the relationship between present urban
activity and present trip-making. If the urban activity forecasts
are accurate and the relationship between urban activity and trip
making does not change,then predictions based on these relationships
will also be accurate.
In the trip generation phase of transportation planning, the
planner is concerned only with the number of trip ends. A trip end
is defined as the beginning or ending of a trip; therefore, a trip
from home to work has two trip ends. The other characteristics of
trips -- destinations, modes and paths -- are considered in other
phases.
Trip generation, then, is the process by which the
transportation planner predicts the number of trip ends.
3-2
Urban activity, as it affects trip generation, is usually
described in terms of amount of activities and character of
activities. Measures of amount and character are provided by the
urban activity forecasts and are input to the trip- generation
phase. Transportation system variables, from the descriptions of
the highway and transit networks, can also be used in trip
generation; however, they have not generally been included in
practice, since strong relationships have not yet been discovered.
BASIC CONSIDERATIONS IN TRIP GENERATION
Amount of Urban Activity
Relationships obviously exist between the amount of activities
and travel. All else being equal, a zone with a larger number of
households or employees will generate more trips than a zone with a
smaller number. Establishing the amount of activity is, therefore,
a key element in analyzing trip generation.
The amount of activity is usually stated in terms of a measure
like the number of employees, households, or retail sales in a zone.
3-3
Character of Urban Activity
Measures of the amount of activity usually are not enough to
develop a good relationship between activities and travel. The
character of the activities is important, too.
For residential land uses, character is described in terms of
socioeconomic variables like family size, family income, and car
availability. Generally, high-income or large families make more
trips than low-income or small families. And obviously, three-car
families generally make more trips than one-car families.
For nonresidential activities, character reflects the type of
activity; for example, industrial, retail, and commercial. As you
can imagine, the number of trips that are generated by a major
shopping center is usually higher than the number of trips that are
generated by a warehouse of the same size.
Other Considerations
It certainly seems logical that the level of service provided
by - the transportation systems would affect the trip generation
rates. You might expect areas with excellent freeways and high
quality bus service to generate more trips than areas with poor
facilities. The fact is that strong relationships between trip
generation and the quality of transportation have not been shown;
and therefore variables to describe the transportation system are
seldom included in trip generation analysis.
3-4
Trip generation relationships are often developed with location
of activities as a consideration, especially when considering the
generation rates for retail activities in the downtown area as
opposed to other shopping areas.
Special Generators
Area wide trip generation analysis must be somewhat general in
the treatment of a wide diversity of activities in an urbanized
area. There are some concentrations of activities of such size or
unusual nature to warrant special consideration in trip generation
analysis. Such generators might include airports, sports stadiums,
hospitals, military bases, large regional shopping centers, and
suburban office complexes.
Special generators are usually handled separately from the area
wide analysis using special surveys and generation rates specific to
each activity or group of similar activities, such as large shopping
centers.
Special generators are relatively few in number in any urban
area, but may represent a significant portion of trips and therefore
justify separate treatment. Their influence on nearby arterials and
highways could not be adequately captured without separate analysis.
3-5
QUIZ
Trip generation is the process by which the planner attempts to
understand and quantify the relationship between ________ and
_________.
Measures of _______ and ________ of urban activities are input to
trip generation.
urban activity,
trip-making
Special generators represent large or
unusual concentrations of trip ends. amount, character
True or false? _____________________
True
3-6
SPECIFYING THE TRIP GENERATION MODEL
While there are several general alternative structures for
specifying trip generation models, the procedure called Cross-
Classification Analysis has become most widely accepted and will be
the only one discussed in this text. This procedure is based on
household trip making as its level of analysis.
A great deal of research has established a recommended
procedure for using cross-classification for trip generation
analysis. The procedure provides the planner a basic model
structure. This structure can be altered for local situations by
substituting or adding variables. There are separate recommended
model structures for each type of trip as follows:
- trip productions,
- trip attractions, and
- internal-external trip generations.
3- 7
Trip Production Model Structure
The forecasting of trip productions is based on relating trip making
to various household characteristics such as income, auto
availability, or household size. The matrix below shows a possible
trip production model structure with household trip rates cross-
classified by income and auto availability.
Click HERE for graphic.
In actual. use, the above table might have additional income
categories or additional variables, depending on data availability
and results obtained when using the relationship.
3-8
Trip Attraction Model Structure
In contrast to trip production modeling, which focuses on household
characteristics, the trip attraction model structure is directed
toward the activities that might attract the trip productions.
These activities might be households, stores, offices, factories,
etc.
In order to analyze trip attractions, the number of trips attracted
to certain activities is related to a measure of the amount of that
activity. For example, the number of trips attracted might be
related to the number of employees in a factory or the number of
employees in a store.
The structure of the trip attraction model relates trip ends by
purpose to the amount, character, and in some cases location of the
activities as shown below:
Click HERE for graphic.
In the attraction model structure, the amount of activity is
reflected in the rate per unit measure, the character of activity by
the type of activity, and the location by the downtown versus other
retail employment classification.
3-9
Internal-External Trip Generation
So far in this chapter, only internal trips -- trips that begin and
end within the boundary of the study area, the "cordon line" -- have
been discussed. Internal trips generally account for 80 to 90
percent of all travel within a study area, so 10 to 20 percent of
the travel in the area remains to be analyzed. As a matter of fact,
trips without both origins and destinations within the study area
run as high as 50 percent of travel in smaller urban areas. Trips
that have one end outside the cordon line are called internal-
external trips, and trips that go through the study area but neither
begin or end within the cordon line are called through trips.
Through trips are generally handled by a factoring process. Growth
factors are developed, based upon forecasts of transportation
facility development that may add to or subtract from the
attractiveness of travel through the study area. Population and
economic activity in the regions from which the through trips began
are also analyzed and forecast. In this section, the concern is
mainly with internal-external trips -- those that begin or end
inside the study area.
There are several ways to handle internal-external travel. The
recommended way is to group internal- external trips so that they
are "produced" at the cordon line and "attracted" to internal zones.
The number of attractions is a function of the character of the
internal zone. As a rule, more trips are attracted to the central
business district than to the outlying zones of the study area.
3-10
Usually there are not enough internal-external trips for them to be
analyzed alone. Therefore it is suggested that the planner treat
the internal ends of internal- external trips as a proportion of all
other trips.
The following ratio, calculated by zone, forms the basis for this
approach.
Number of internal ends of internal-external trips/zone
Number of all other internal trip ends/zone
After these ratios have been determined. the averages of the zonal
ratios can be calculated and analyzed, by rings, from the central
business district. If no pattern is apparent, an alternative is to
average the zonal ratios by districts and then examine those ratios
for a pattern. If a large traffic generator attracts a significant
amount of external traffic, it should be analyzed separately.
In forecasting, the ratios are applied by area type (central
business district, fringe business, central city, suburbs, etc.) to
the forecasted internal trip ends to yield forecasts of internal--
external trip attractions.
At the cordon line station -- the production end of internal-
external trips -- the forecast of future trip ends should be based
upon a growth factor that reflects the expected growth within the
travel corridor of the external station, including the area beyond
the cordon.
3-11
QUIZ
The recommended approach to trip generation analysis is called
___________-__________________.
Trip production forecasting is based on ________ characteristics.
The trip attraction model relates trip ends to units of
_____________.
cross-classification
household
Trip ends that have one end outside the cordon line are
known as ___________ trips.
activity
Internal-external
cordon line
Trip productions for internal- external trips are estimated at the
__________.
3-12
TRIP GENERATION MODEL CALIBRATION
Introduction
Model calibration is the process in which the relationships
between travel and variables influencing travel are quantified. In
trip generation analysis, trip productions and attractions are
related to urban activities and their characteristics.
Developing Trip Production Rates
In this section, you'll see some examples of how a cross-
classification matrix is developed for trip production rates. The
purpose here is to show you the process for developing rates.
Consider the development of trips per household, stratified by auto
availability and income. In this example, there are 20 households in
the sample for cross-classification analysis. Naturally, in practice
there would be many more households included in the sample.
The table on the next page shows the data that might have been
collected in home interviews.
3-13
Household Trips Income Autos
1 2 4,000 0
2 4 6,000 0
3 10 17,000 2
4 5 11,000 0
5 5 4,500 1
6 15 17,000 3
7 7 9,500 1
8 4 9,000 0
9 6 7,000 1
10 13 19,000 3
11 8 18,000 1
12 9 21,000 1
13 9 7,000 2
14 11 11,000 2
15 10 11,000 2
16 11 13,000 2
17 12 15,000 2
18 8 11,000 1
19 8 13,000 1
20 9 15,000 1
In this table, you see the numbers of trips, the family incomes and
numbers of autos available to the twenty families in the sample.
3-14
A matrix is established, based upon family income and auto
availability. The numbers inside the matrix below represent the
household sample numbers.
Click HERE for graphic.
Now, the average of the number of trips for the households in
each cell of the matrix above is obtained. For example, the average
trip rate for households with 2 or more autos and an income between
$12,000 and $15,000 is 11.5 -- households 16 and 17 made a total of
23 trips, so 23 is divided by 2 to arrive at the average rate of
11.5 trips. These averages are then substituted for the household
numbers in the matrix, as shown below.
Click HERE for graphic.
3-15
Click HERE for graphic.
The data from the second matrix can now be plotted on a graph,
as shown at right.
The plotted points are fit with smooth curves, and the lines
are extended -- extrapolated -based on the shape of the curves and
logic. The curve values are then used to develop a completed
matrix, which can be used to make forecasts of trips. For example,
on the basis of information from the graph, we can predict that
families that have an income of $10,000 and one auto will make about
7 trips per day. When used in forecasting, incomes are corrected
for inflation and represented in constant dollars.
Click HERE for graphic.
3-16
Click HERE for graphic.
An additional relationship is necessary for the trip production
process. Generally in transportation planning, we are interested in
trip making associated with some specific purpose and therefore the
total trip productions must be split by trip purpose.
Most urban areas can use three categories of trip purpose:
home-based work (HBW),
home-based other (HBO),
and non-home-based (NHB).
Additional categories may be justified, depending on local needs.
Using survey data, the relationship at right can be developed from
cross-classification tables. These curves relate the expected split
of trips into various purpose categories to household income. Once
again, these relationships can be borrowed from similar areas if no
data are available, or if the available data are found inadequate.
3-17
Developing Trip Attraction Rates
In the previous section, you saw the development of trip production
rates based on household characteristics. We now turn our attention
to the attraction end of the trip. Attraction rates are developed
by analyzing the urban activities that attract trips.
Trips are attracted to various locations, depending on the
character, location, and amount of activities taking place. For
example, a high-rise office building that employs thousands of
people will attract more trips than a small dress shop.
Trip attraction rates might be borrowed from other urban areas or
developed from survey data relating the number of trip attractions
to activity character, location, and amount. For example, we might
discover that retail stores in the central business district employ
1,000 people and attract 9,700 trips. Using this information, we
can make the following calculation:
CBD Retail Trips per Employee = 9,700 trips/1,000 employees = 9.7
trips/employee.
3-18
If we refine the calculations further and establish a separate rate
for each trip purpose and for various types of urban activities, we
might end up with the following table:
Click HERE for graphic.
Using these trip rates and estimates of future households and
employment in each zone, the planner can forecast trip attractions
for the area. An example application of this table will follow
later in the chapter.
This concludes our discussion of trip generation model calibration.
After the quiz, which follows, go on to the application example.
* Note: These trips would be made to households other than the
residence of the tripmaker.
3-19
QUIZ
The purpose of model calibration is to ___________ the
relationship between travel and appropriate variables.
Trip production models use cross-classification to develop
relationships between trip making and ______ characteristics.
quantify
Trip production and attraction rates can be
developed from travel surveys or _________________. household
borrowed
from
similar
areas
3-20
TRIP GENERATION MODEL APPLICATION
In order to understand how the production and attraction models are
used, the remainder of this chapter is devoted to a single example.
To simplify the example, three household income groups are used --
low, medium and high. In actual applications, more income groups
might be used or other variables included, such as household size or
location in the urban area.
The trip production example is shown first, followed by trip
attractions.
3-21
Trip Production Model Application
The diagram below outlines the trip production forecasting
procedure, showing how the three
components are interrelated. Study this diagram carefully and then
go onto a UTOWN example.
Click HERE for graphic.
3-22
As can be seen on the preceding page, the process begins with inputs
of the number of households cross-classified by income and auto
availability (or other variables, as provided for in the calibration
process). These inputs can be obtained in several ways.
First, the urban activity forecasting procedure might produce
estimates of households by income and auto availability for each
zone. If this is the case, the required data is at hand and can be
used directly.
The FHW publication "Trip Generation Analysis" describes a procedure
for deriving the number of households by income and auto
availability using functions relating the average zone income to
distribution of household income and auto availability.
Alternatively, the planner might assume that zones already
developed will maintain their current breakdown of households by
income and auto availability or, if the zone population is changing,
the percent of households in each income/auto category would remain
constant. Zones being developed could be related to existing zones
of similar socioeconomic status and their households divided into
income/auto categories using the similar existing zones as a basis.
For example, if a zone is being developed in a similar manner to an
existing one, and the existing zone has 500/o of its households with
high incomes and two or more autos, then the new zone will likely
have 50% of its households in the same income/auto category. For
long-range forecasts corrections to account for real increases in
income might be necessary. Clearly the planner must exercise a good
deal of judgment to obtain the desired information.
3-23
Any of the three methods just described could yield estimates of the
number of households by income and auto availability for each zone
in the area. The table below shows an example for zone #1 of UTOWN.
Click HERE for graphic.
3-24
Using the information about households from the previous page,
the trip production relationship shown at right can be applied to
obtain the number of trip productions.
Click HERE for graphic.
The calculations for Zone I are as follows:
Click HERE for graphic.
3-25
Next, with estimates of trip productions by zone and household
income, the curves at right can be entered to split the trips by
trip purpose.
Click HERE for graphic.
The calculations for Zone I are as follows:
Zone 1:
Trip Household Percent Trip Productions
Productions Income Group by Purpose by Purpose
15 Low 15% HBW 2
57% HBO 9
28% NHB 4
17% HBW 39
231 Medium 52% HBO 120
31% NHB 72
18% HBW 72
339 High 50% HBO 199
32% NHB 128
Summary Zone 1: 113 Home-Based Work Productions
328 Home-Based Other Productions
204 Non-Home-Based Productions
3-26
Similar calculations for other zones yield the following:
Trip Production
Zone # Home-Based Work Home-Based Other Non-Home Based
1 113 328 204
2 404 1395 747
3 602 1726 1079
4 263 821 478
5 428 1213 766
Total 1810 5483 3274
This completes the forecast of trip productions for UTOWN. The next
task is to estimate trip
attractions.
3-27
With estimates of the number of households and employment levels in
each zone, and with the calibrated trip attractions can be made.
Click HERE for graphic.
Zone 1 (downtown): Home Based Other Attractions
60 households x 1 trip end/household = 60 attractions
220 downtown retail employees x 5 trip ends/emplo yee=1100 attractions
650 non-retail employees x 2 trip ends/employee = 300 attractions
Zone 1 home-based other attractions = 2460 attractions
Similarly, calculations for other.purposes for Zone 1 yield:
Home-based work attractions = 1479
None-home-based attractions = 1370
3-28
The attractions for the remainder of UTOWN zones calculated in a
similar manner show the following:
Zone # Home-Based Work Home-Based Other Non-Home Based
1 1479 2460 1370
2 144 1750 1075
3 104 650 495
4 64 500 370
5 112 680 450
Total 1903 6040 3760
The total number of trip productions and trip attractions
should be in balance for any given area. Given the approximate
nature of these rates, the likelihood of a close balance is not
great. The comparison for the UTOWN example is as follows:
Total Trip Total Trip Productions/
Productions Attractions Attractions
Home-Based Work 1810 1903 1810/1903 = 0.95
Home-Based Other 5483 6040 5483/6040 = 0.91
Non-Home-Based 3274 3760 3274/3760 = 0.87
3-29
As you can see, the productions and attractions do not balance;
and, therefore, adjustments must be made. If these totals were off
more than 20%. The entire procedure would have to be investigated.
Since the trip productions are based on household data, they are
felt to be more reliable and therefore act as control totals for
trip generation. The trip attractions by zone are multiplied by the
ratio of total productions over total attractions so that the totals
will be equal. If work trip productions and attractions do not
balance, a special analysis should be undertaken since the
attractions are based on employment and therefore should be quite
accurate.
You will recall from Chapter Two that non-home-based trips
represent true beginnings and endings and therefore must be in
balance by zone. The non-home-based trip productions act only as
control totals, since the production equations have no way of
knowing where these trip ends are produced. Therefore, the non-
home-based trip ends from the attraction forecasting procedure
represent productions and attractions for each zone. These trip
ends are factored to equal the total number of non-home-based trips
estimated in the trip production procedure.
3-30
To adjust the attractions for Zone #1 of UTOWN, the following
procedure is followed:
Home-Based Work UTOWN Zone *1 Attractions = 1479
Total UTOWN HBW Attractions = 1903
Total UTOWN HBW Productions = 1810
Total Productions/Total Attractions = 0.95
Therefore, to adjust Zone #1 attractions: 1479 x 0. 95 = 1407
Similarly:
Home-Based Other: 2460 x 0. 91 = 2233
None-Home-Based: 1370 x 0. 87 = 1192
3-31
The summary for Zone #1 now reads:
Zone #1:
Click HERE for graphic.
Note that the adjusted non-home-based attractions are set equal to
productions.
3-32
By continuing this process for all zones, the following final
estimate of trip generation is produced:
Click HERE for graphic.
The total productions and attractions are now in close balance
(round-off errors cause some minor discrepancies), and the trip
generation process is complete. Internal- external trip ends and
special trip generators will not be developed into examples in this
text.
The next phase of travel demand forecasting now begins -- the
distribution of these trip ends throughout the area.
3-33
For additional information about trip generation analysis, consult
the following documents:
Guidelines for Trip Generation Analysis, FHWA, June 1967.
(Reprinted April 1973) Available from FHWA, HHP-22, Washington,
D.C. 20590.
Trip Generation by Land Use, Maricopa Association of Governments,
Maricopa County, Arizona, April 1974. A synthesis of trip
generation rates for differing land uses. Useful in developing
synthetic models, checking models or site planning. Available
from FHWA, HHP- 22, Washington D.C. 20590.
Trip Generation Analysis, FHWA, August 1975. Available from FHWA,
HHP-20, Washington, D.C. 20590.
Trip Generation, Institute of Transportation Engineers, 1976.
Available from ITE, members $15/nonmembers $18.
4-1
Chapter Four
TRIP DISTRIBUTION ANALYSIS
INTRODUCTION
Click HERE for graphic.
As you saw earlier in this text, our forecasts begin with urban
activity forecasts - - how much activity will there be, and how will
it be allocated to the various zones in the study area?
The trip generation phase of planning uses the data from urban
activity forecasts, and other related studies,, to forecast the
number of trips that will be "generated" in each zone.
Trip distribution is the process by which the planner determines
where the generated trips will be attracted. That is, each zone's
trip productions are connected to all the zones to which they are
attracted.
There are several methods of distributing trips. Although other
techniques supply accurate results, in this text we will discuss
only the development of the most widely used technique, the Gravity
Model.
4-2
BASIS FOR THE GRAVITY MODEL FORMULA
The gravity model derives its name and basic premise from Isaac
Newton's law of gravity. Newton's law states that the attractive
force between any two bodies is directly related to the masses of
the bodies and inversely related to the distance between them.
Similarly, in the gravity model, the number of trips between two
areas is directly related to activities in the area represented by
trip generation and inversely related to the separation between the
areas represented as a function of travel time. Therefore, areas
with large amounts of activity tend to exchange more trips and areas
farther from each other tend to exchange fewer trips.
4-3
The gravity model formula appears as follows:
Pi Aj F(t)ij
Tij = ---------------
n Aj F(t)ij
ä
j=1
where:
Tij = the number of trips produced in zone i and
attracted to zone j
Pj = the trips produced to zone i
Aj = the trips attracted to zone j
F(t)ij = the friction factor for interchange i j (based on
travel time between i and j
i = origin zone
j = destination zone
n = number of zones in the study area.
At first, this formula might look elaborate and confusing, but it's
really quite simple. Let's look at each part of the formula.
4-4
The gravity model states that the trips produced in zone i --
Pi
will be distributed to each other zone j --
Tij
according to the relative attractiveness of each zone j --
Aj
---
äAj
and the relative accessibility of each zone j -
F(t)ij
-------
äF(t)ij
and, that's it.
4-5
Here's what the formula means.
attractiveness and
accessibility
characteristics of j
--------------------
Trips between i and j = trips produced at i X attractiveness and
accessibility
characteristics of
all zones in the
area.
Thus, zone j gets a portion of zone i's trip productions according
to its characteristics as compared to the characteristics of all
other zones in the study area. This leads to the term "share model"
often applied to the gravity model and other models having these
characteristics.
In practice, a separate gravity model is developed for each trip
purpose, since different trip purposes have different distribution
characteristics -- a person who travels from one side of the city to
the other for work might not go so far to see a movie or to go
shopping.
In this text, we will not discuss the derivation or the proof of
this model -- our purpose is to show you how to use the model as a
tool in distributing trips. For more information about the formula
itself, refer to the references at the end of the chapter.
Before the number of trip interchanges can be computed, several
parameters must be defined.
4-6
GRAVITY MODEL PARAMETERS
The first two parameters that must be considered for the
gravity model are trip productions and attractions for each zone.
Trip Productions and Attractions
Estimating trip productions and attractions, you recall, is the
purpose of trip generation. Let's review the definitions of
production and attraction. The gravity model distributes trips from
production zone to attraction zone. To define productions and
attractions, it is necessary to classify all trips as either home-
based or non-home-based.
Home-based trips are always considered to be produced at the
home end and attracted at the non-home end, whether the trips begin
at the home end or the non-home end. A trip from work to home is
produced at home and attracted at work. Does this sound
unreasonable ? It's not -- the traveler probably began at home, went
to work, and returned home. The attraction for both trips was work.
By the same token, a trip from home to a shopping center and the
return trip home would be classified as home-based -- both trips are
produced at home and attracted at the shopping center.
Non-home based trips are always produced by the origin zone and
attracted by the destination zone.
4-7
In actual gravity model applications, the trip attraction parameter
gives way to a term called the "attraction factor. " This factor is
a product of a' balancing procedure to ensure that the number of
trips sent to a zone by the gravity model is equivalent to its trip
attractions estimated by the trip generation procedure. More about
this procedure can be found in the references at the end of the
chapter.
Friction Factors
Newton's gravity theory states that the force of gravity is
inversely proportional to the distance between two bodies. The
effect of distance between zones on the amount of travel between
zones isn't so neatly defined -- if you were going to go shopping at
one of the two identical shopping centers described below, which
would you choose?
Shopping Center A--7 miles away/8 minutes by freeway
Shopping Center B--5 miles away/20 minutes by city street
You'd probably choose shopping center A, because the trip takes less
than half as long as a trip to shopping center B even though B is
closer. Of course, this example is for travel by automobile; the
gravity model can distribute transit trips as well.
Friction factors represent the effect that various levels. of travel
time have on travel between zones. These factors are determined in
the calibration process.
4-8
Travel time between each pair of zones in the study area is
determined by the trip assignment process.
A more general term than travel time for the separation of zones is
"impedance." Impedance can represent traveltime, cost, distance, or
a combination of factors. Generally, impedance is a weighted sum of
various types of times (walking, waiting, riding) and types of cost
(fares, operating cost, tolls, parking cost). In the past,
traveltime was used in the gravity model to measure separation and
to develop friction factors. We now have the capability to include
other factors, such as tolls and operating cost, in the impedance
function, which more accurately represent the separation between
zones.
Socioeconomic Adjustment Factors
Another parameter that can be used reflects the unique socioeconomic
characteristics of the various zones - - characteristics that are
not otherwise accounted for - - and how these characteristics affect
the travel patterns in the study area.
These adjustment factors affect the number of trip interchanges
determined by the model.
4-9
Although the gravity model provides for these adjustments, few
cities have found it necessary to use them -- but, where they are
used, they show the effect of certain social and economic conditions
that otherwise are not accounted for. Like the friction factors,
socioeconomic factors are determined in the calibration process.
They should be used with extreme caution and only when justified to
account for an area's unique characteristics.
If socioeconomic factors are found necessary, they appear as
the Kji parameters in the gravity model, as follows:
Pi Aj F(t)ij Kij
Tij = --------------------
n Aj F(t)ij Kij
ä
j=1
4-10
QUIZ
The gravity model distributes trips from the ______ zone to the
__________ zone.
Home-based trips are always produced at the ____________ end and
attracted to the ____________________ end.
Non-home based trips are always produced by the zone of
____________ and attracted by the zone of________________________.
The gravity model is the only acceptable method of dis-
tributing trips. True or false? ______________ production,
attraction
The most often used measure of the separation of home,
is ____________. non-home
origin,
destination
False
A more general term, which also represents the separ-
ation of zones, is __________. traveltime
impedance
Traveltime is determined in conjunction with the trip
_________________ process. assignment
4-11
SAMPLE PROBLEM
Click HERE for graphic.
The simplified sample problem using UTOWN that follows is
designed to show you how the gravity model distributes trips.
We will distribute the 602 work trip productions from Zone 3 to
Zones 1, 2, 4 and 5 at right. The numbers of work trip productions
and attractions are determined in the trip generation phase.
Socioeconomic(Kij )factors are not used in this problem.
4-12
The number of attractions in each zone is used as the estimate
of the zone's attractiveness. You would expect more of zone 3's 602
trips to go to zone 1 than to zone 4. But, the travel impedance to
zone 4 is 10 minutes, compared with zone 1's 20 minutes. The
shorter traveltime affects the drivers in zone 3. The gravity model
is designed to take both attractiveness and traveltime into account
in distributing trips from zone 3.
The first thing to do is to list the zones, with their
attractions, to show the relative attractiveness of each zone:
Zone Attractions Aj
1 1080
2 531
3 76
4 47
5 82
On the basis of this tabulation, you'd expect zone I to get most of
zone 3's 602 trips, followed by zone 2, then zone 5, then zone 3
(intrazonal trips), and zone 4.
4-13
Now we look at traveltime, which represents the separation between
zones. Let's add these times to the table:
Impedance
From Zone To Zone Attractions Aj (minutes)
3 1 1080 20
3 2 531 7
3 3 76 5*
3 4 47 10
3 5 82 25
*Naturally, even trips that begin and end in the same zone take
time, so they also must be assigned an impedance.
4-14
The next step is to enter the friction factors for each zone. These
factors are determined from origin-destination data, and reflect the
area wide effect of traveltime on drivers' willingness to drive to
various destinations. Obtaining these factors is the principal
operation of gravity model calibration, which is discussed in later
sections of this chapter. The chart at right shows some friction
factors for our sample study area. The friction factors represent
accessibility -- with higher values being greater accessibility.
Friction Factors fro Sample Problem
Traveltime Friction factor
(minutes) F(t)ij
3.0 87
5.0 45
7.0 29
10.0 18
15.0 10
20.0 6
25.0 4
30.0 3
40.0 2
The product of the attractions and the friction factor
represents each zone's relative attractiveness and accessibility --
as shown in the far-right column. This product is usually referred
to as the "accessibility index" of a zone.
Click HERE for graphic.
We now have all of the information that is necessary to distribute
zone 3's 602 trips.
4-15
Let's take another look at the gravity model formula:
Pi Aj F(t)ij
Tij = ----------------
n Aj F(t)ij
ä
j=1
The socioeconomic adjustment factor (Kij) was not used in this
sample problem. Let's "plug in" the values that we already have,
for trips going from zone 3 to 2.
Click HERE for graphic.
4-16
The numbers of trips from zone 3 to the other zones are calculated
in the same manner. Here's the calculation for intrazonal trips --
trips that remain in zone 3:
Click HERE for graphic.
Zone 3's 602 trips will be distributed as follows: 78
intrazonal trips; 147 trips to zone 1; 350 trips to zone 2; 19 trips
to zone 4; and 8 to zone 5.
As you can see, the computations in even a very simple
situation can be complex and cumbersome. In an actual situation,
several more steps are necessary, such as balancing attractions to
assure the gravity model does not distribute more trips to a zone
than it attracts according to trip generation analysis. These
complex computations led to the use of computers in using the
gravity model. The computer will do these computations for you --
the purpose of this example was merely to show you how the model
works.
4-17
CALIBRATING THE GRAVITY MODEL
Calibration of the gravity model -- making sure that it
accurately describes the travel patterns of the study area -- is
accomplished by developing friction factors and, if found necessary,
developing socioeconomic adjustment factors.
Developing Friction Factors
Friction factors, you will recall, attempt to show the effect
that traveltime or impedance has on trip making. It is necessary to
go through a trial-and-adjustment process to fit the model to the
specific area that is being studied. Friction factors determined by
this process are empirical measures of the effect of time or
impedance on travel, and are "tailored" to each urban area.
One way to begin the trial-and-adjustment process for
determining the friction factors is to use the factors from a past
study in a similar urban area.
Friction factors from other areas can be used as a substitute
for calibrating the gravity model if not enough information is
available for a traditional calibration. An FHWA handbook of
typical friction factors is referenced at the end of this chapter.
4-18
Three items are used as input to the gravity model for calibration:
1. Production-attraction trip tables for each trip purpose;
2. traveltimes for all zone pairs, including intrazonal
traveltimes (impedance would be substituted if combinations
of costs and times are used); and
3. initial friction factors for each increment of traveltime
(optional -- if none are input, initial estimates are
provided analytically).
4-19
QUIZ
Friction factors attempt to show the effect of ___________ on
trip interchange.
One way to estimate the initial friction factors is to use the
factors from a previous study in a similar area. True or false?
In order to calibrate the gravity model, two items are traveltime
(or
required inputs: impedance)
__________, and
___________. True
trip tables,
traveltimes or
impedances
4-20
In order to calculate the friction factors, let's take another look
at the gravity model formula. We'll leave out the socioeconomic
adjustment factors again.
Pi Aj F(t)ij
Tij = ----------------
n Aj F(t)ij
ä
j=1
where: Tij = the number of trips produced at i and attracted to
J;
Pi = the trip production at i;
Aj = the attractions at J;
F(t)ij = the friction factor for interchange i j (based on
traveltime
between zones);
i = an origin zone;
j = a destination zone; and
n = the number of zones in the study area.
The trip production values Pi, and trip attraction values Aj are
obtained from the input trip tables (from the survey data). The
initial friction factors can be based on judgment, obtained from
similar areas, or calculated by the program.
4-21
The calibration process involves adjusting the friction factor
parameter until the planner is satisfied that the model adequately
reproduces the trip distribution as represented by the input trip
table (until the model's trip table substantially agrees with the
table from the survey data using indications such as trip time
frequency distribution and average trip time).
The process begins by using the gravity model to distribute
trips based on initial inputs. Then, the total trip attractions at
all zones J, as calculated by the model, are compared to those
obtained from the input "observed" trip table. If this comparison
shows significant differences, the attraction Aj is adjusted for
each zone where a difference is observed. The model is rerun until
the calculated and observed attractions are reasonably balanced.
While these adjustments are being made, the friction factors F (t)
are not changed.
4-22
Click HERE for graphic.
After the trip attractions are reasonably balanced, the model's
trip table and the input traveltime table can be used for two
calculations: the trip time frequency distribution and the average
trip time. These values can be compared to the values from the
input trip table; if there are significant differences, the friction
factors are adjusted, and the process begins again. The figure at
right shows the plot of the observed traveltime distribution along
with plots for an initial calibration and after four calibration
runs.
Each time the friction factors are adjusted, the attractions
must be compared again -and, if necessary., the attraction factors
are adjusted. The basic process for adjusting the friction and
attraction factors is shown on the next page; study it carefully.
4-23
Click HERE for graphic.
As you can see from the diagram above, the model adjusts the
attractions' and the friction factors until the output trip table
balances with the input table -- the factors that are produced can
be used to estimate future trip distributions.
4-24
Establishing Socioeconomic Adjustment Factors
There may be factors other than travel impedance that affect
the travel patterns of an area. The effect of social and economic
conditions can be accounted for in the gravity model using the K
factors that we mentioned earlier.
In smaller urban areas, K factors usually aren't needed, but in
larger areas, these factors are sometimes necessary in order to get
accurate results from the gravity model program. For example,
friction factors are developed for each trip purpose; since trips
between all zones in the area are used to develop these factors,
they represent the average area-wide effect of traveltime on trip
making. However, there is some evidence that indicates that
friction factors vary from zone to zone, depending on the
characteristics of the people in each zone. Socioeconomic
adjustment factors are intended to account for any factors, other
than traveltime and trip purpose, that affect travel. patterns. For
example, if low-income people live near the central business
district, the model may incorrectly distribute many of their trips
because they are close to the CBD. K factors would adjust the
interchanges so that this won't happen. However, this problem might
be solved better by having separate trip distributions for various
types of employment, income, or other factors.
These K factors are determined during the calibration of the
gravity model. It's done after the attractions are adjusted, and
after the final friction factors are obtained. Of course, adding K
factors to the calibration process results in revised interchange
estimates -- we add adjustment factors for interchanges that would
not otherwise be adequately estimated. It must be strongly
emphasized that K factors should be used only when their use can be
adequately explained by some specific circumstances. They should
not be used simply because the gravity model might show unreasonable
values for some trip interchanges.
4-25
QUIZ
The output of the gravity model calibration is a set of
_____________ _______________________.
If the output attractions don't balance with the input
attractions, the attractions are adjusted and the model is
___________.
If the trip time frequency distribution and average trip time
inputs and outputs don't compare satisfactorily, the ________
factors are adjusted.
The inputs to calibration for friction factors consist of:
friction factors
______________;
______________; and
______________.
rerun
friction
initial friction
factors;
K factors balance the effect that unique ________ travel
conditions have on trip interchange. time tables;
