1. Travel Demand case.

Station-to-station movements of
Basic purpose of transportation planning passengers may have to be estimated in the
and management case of a rapid transit service along the
 match transportation supply with travel corridor. At the macro level the concern may
demand, which represents the need for be total energy consumption by the
transportation infrastructure. transportation sector or the total emission of an
Existing travel pattern air pollutant; for these cases, information on
 necessary for identifying and analyzing total vehicle-miles traveled (VMT) on each
existing traffic-related problems. functional class of roads will be needed.
Prediction of future travel demand It is important to recognize that the nature of
 essential task of the long-range problems to be examined dictates the level of
transportation planning process for planning to be used as well as the technique
determining strategies for for travel demand analysis. The discussion of
accommodating future needs. this chapter will be oriented mostly to meso
Strategies scale or area-wide travel demand analysis that
 land use policies is commonly performed in urban transportation
 pricing programs planning studies.
 expansion of transportation supply- Even for this type of analysis for an urban area
highways and transit service. at the meso scale, the approach and details of
"Micro," "meso," and "macro" techniques and models to be used would
 used to describe the level of detail or the depend on the size of the area as well as the
size of an area used for an analysis. resources available for carrying out the work.
"Site-specific," "corridor," and "area-wide" For example, a small urban area may not have
or "metropolitan" the manpower or funding needed for carrying
 describe variations in the scope of a out large-scale surveys and developing
problem. advanced
Example of a micro-level or site-specific mathematical models. The need for
analysis customizing the planning and modeling
 case of a congested road intersection. approaches based on specific situations was
 In this case traffic engineers would be discussed in detail by Grecco et al.
interested in detailed traffic flow (1976).
characteristics, including turning
movements of vehicles along each
approach and pedestrian volumes 2. CHARACTERISTICS OF TRAVEL
across each approach. 2.1. Spatial and Temporal
 Management strategies in this case Variations
would involve traffic operation and The total magnitude of travel demand alone is
roadway design-oriented techniques. not sufficient for detailed planning and
Corridor-level analysis management purposes.
 cover a larger area, say 10 miles long The spatial and temporal distributions of travel
and 2 miles wide. are also important items of information to be
 A major highway with severe congestion considered in determining supply strategies.
problem may require [Link] The peaking of travel at certain time periods
The origin and destination of trips and modal requires a level of transportation supply that is
choice of travelers would be of interest in this not needed at other times. However, due to the
nature of supply, which cannot be adjusted
easily, large investments have to be made to
provide roadway or transit service capacities to Choice of domicile and workplace, lifestyles
accommodate peak period travel, and this and different travel needs of individuals and
capacity is not utilized efficiently at other times. families make the comprehension of trip-
An imbalance in the directional distribution of making characteristics of a large metro area
travel also creates similar inefficiencies. very complex. These complexities may be
The spatial orientation of trips has important illustrated through trips made by a typical
influence on supply requirements and costs. A household on a given weekday.
few typical spatial distribution patterns of trips Assume that this household has four members,
in urban areas are listed below: including two children who go to grade school,
• Travel along dense corridors, which are and two cars. It can be seen that there are at
usually radial connecting suburbs to central least 11 trips made by this household at
business district (CBD) different times of day. Most of the trips are auto
• Diffused travel pattern caused by urban trips, and two trips are taken in the walk mode.
sprawl Travel demand modeling attempts to capture
• Suburb to suburb or circumferential travel such spatial and temporal variations in travel at
• Travel within large activity centers in an aggregate level, such as a zone, in which a
CBD and suburbs number of households, businesses, and offices
Different modes of transportation may be exist.
needed to serve these different travel patterns.
For example, fixed-route public transit service
usually is efficient for concentrated travel along 2.2. Classification of Travel by Trip Purpose
a dense corridor, but it is not ideally suited to and Market Segments of travel demand must
serve a diffused travel pattern in a cost- be recognized.
effective manner Trip purposes such as work, shopping, and
social recreation, and trip-maker's
characteristics such as income and car
ownership, are important factors influencing
the elasticity of demand, reflecting its
sensitivity with respect to travel time and cost.
For example, work trips may be more likely to
use public transit for a given level of service
than trips of other trip purposes.
For a metropolitan study, it is useful to classify
travel according to spatial orientation and trip
purpose as shown in Figure. The concept of
market segmentation is applicable to the
classification of travel based on trip purpose,
trip-makers' characteristics, and spatial-
temporal concentration. This concept is used in
the field of marketing for developing different
types of consumer products targeted to match
different tastes and preferences of potential
users /buyers of these products.
The concept of market segmentation is Travel demand is measured and expressed in
applicable to public transportation planning. A different ways for different types of analysis.
single type of transit service is not suitable for Examples of different units of measurement
all transit market segments. For example, are:
express buses may be needed for a commuter 1. Trip (between two areas)
market segment. Taxicabs serve a different 2. Trip end (in a given area)
market segment. 2. Traffic volume (on a road segment)
1. Person trip and vehicle trip
2. Passenger vehicle and freight vehicle
3. Person-miles traveled and vehicle-miles
traveled
The definition of each of these units should be
understood clearly and an appropriate unit of
measurement should be used to match the
case being analyzed. For example, for a
parking study, trip end is the appropriate unit
for expressing parking demand. For estimating
the number of lanes to be provided in a road
segment, the demand should be expressed in
terms of traffic volume. As pointed out earlier,
the appropriate unit of travel for estimating fuel
consumption and/or air pollution attributable to
transportation is vehicle-miles traveled (VMT).

4. MEASURING EXISTING TRAVEL

Detailed information on existing travel is
needed for two purposes: analyzing existing
problems and developing mathematical models
for forecasting travel. A variety of surveys can
be performed for gathering information related
to existing travel demand.
However, travel surveys are expensive, and
therefore care must be taken to identify the
types of information that really would be useful
for specific purposes, and then the most
suitable procedures should be selected for
gathering the information. Sampling techniques
are useful, and adequate time and care must
be devoted to developing sampling
procedures. There are several different types
of survey techniques, some of which are
suitable for automobile travel, some for transit
travel, and some for general passenger
3. UNITS FOR MEASURING TRAVEL movement. Survey procedures for freight
DEMAND vehicles and commodity movements may be
very different in certain respects from those of trip purpose and mode of travel can be
passenger travel. Good references for travel determined in different ways:
demand-related survey techniques are Dresser 1. Home interviews (for internal travel)
and Pearson (1994), Stopher and Metcalf 2. Roadside interviews at cordon stations
(1996), and Travel Survey Manual (Cambridge (for external-internal and through trips)
Systematics, Inc. and Barton-Aschman 3. On-board survey on transit vehicles
Associates, Inc. 1996). All three of these techniques involve sampling
and require careful planning before their
4.1. Time Frame for Travel Surveys execution. The Federal Highway Administration
Since travel demand varies during a given year (FHWA 1975b) and Urban Transportation
according to the season (or month of year) and Systems Associates (1972) developed detailed
day of week, a decision must be made guidelines for O-D survey procedures.
carefully to select a specific time frame or The reliability of the results of an O-D survey
reference for surveys. For urban transportation depends on its sampling scheme and sample
studies it is a common practice to develop size, and this issue was examined by
travel demand information for an average Makowski, Chatterjee, and Sinha (1974).
weekday in the fall. However, the time can be Full-scale origin-destination surveys were
different based on the nature of the problem to widely used during the 1960s and 1970s to
be analyzed. For example, in the case of a develop a variety of information, including
tourist-oriented urban area the major concern "desire lines" of travel. Their use has
may be traffic problems during weekend days decreased because of the cost and also due to
or holidays, and surveys may be done to the use of synthetic or borrowed disaggregate
capture information for one of those days. A travel models, which require less survey data.
few major types of surveys are discussed in 4.3. Traffic Volume and Passenger
the following sections. Counts
For determining the use of various roadway
4.2. Origin-Destination Surveys facilities and assessing their level of service,
The classification of trips into the three classes vehicle counts are taken at selected locations
of internal, external-internal (and vice versa), along roadways.
and through trips is useful for meso-scale and Short-count techniques are useful provided
metropolitan-level as well as small-area appropriate expansion factors are developed
studies. This classification scheme is useful for based on previous or ongoing research on
developing forecasting fluctuations of traffic by hour, weekday, and
procedures/models as well as policies and month. All state Departments of Transportation
strategies for accommodating travel because (DOTs) have extensive programs for gathering
strategies for each of traffic volume data on an annual basis.
these classes of travel would be different. These vehicle counts usually are taken with
For example, through trips may require a machines.
bypass facility. External-internal trips may have For urban transportation studies screen lines
to be intercepted before they reach a heavily and cut-lines are established in the study area
congested area such as the central business to select traffic count locations and take counts
district (CBD). in an organized manner so that the major travel
The origins and destinations (O-D) of trips movements can be measured and analyzed.
along with some other characteristics such as These counts are also used for checking the
results of travel forecasting models. Similarly,
traffic counts are taken at special traffic specific factor such as travel time or fare based
generators such as airports and large colleges on actual observation.
and universities to capture their unique travel- The coefficient should exclude the effect of
generating characteristics. other factors that also may be influencing
For analyzing the use of a transit service, demand at the same time.
passenger counts are taken on-board transit Mayworm, Lago, and McEnroe (1980) give
vehicles and / or at selected stops or stations. information on demand elasticity models.
These passenger counts usually are taken by More information on the elasticity of transit use
observers who are assigned to specific transit with respect fare may be found in Parody and
vehicles and/or transit stops / stations Brand (1979) and Hamberger and Chatterjee
according to a survey plan. (1987).

5. FORECASTING FUTURE 5.2. Stated Preference Surveys and Concept

DEMAND Tests for Forecasting For transit planning it is
The need for travel demand forecasts arises in sometimes necessary to ask people about their
various contexts of planning, short-range as preferences and their likes and dislikes for
well as long-range. Travel forecasting is one of various service characteristics. These surveys
the most important and difficult tasks of are used for determining how to improve an
transportation planning. There are different existing service and/or designing a new
types of travel prediction techniques, and the service, and also for forecasting ridership on a
one to be used in a particular case must be new service.
compatible with the nature of the problem and These attitudinal and stated-preference
scope of planning. surveys need sound statistical design for
Constraints of available time and resource also selecting the sample and analyzing the results.
influence the selection of a technique. A discussion on stated preference survey may
be found in the Travel Survey Manual. In the
5.1. Predicting Response to Service Changes field of marketing, concept tests are performed
Using Elasticity for estimating the potential demand for a new
Coefficients consumer product, and this approach can be
For short-range planning or a transportation extended to ridership forecasts for new
systems management innovative transit services.
(TSM) study, it is often necessary to predict the Hartgen and Keck (1976) describe a survey-
effect of a proposed change in transportation based method of forecasting ridership on a
service that can be implemented in the near new dial-a-bus service. The interpretation of
future. For example, a planner may be asked results of opinion-based surveys must be done
to evaluate the impact on transit ridership of carefully in order to account for any bias
improving transit service in a travel corridor by reflected in apparent results. Chatterjee,
providing increased frequency and/or limited McAdams, and Wegmann (1983) present a
stop service. The impact of changing the fare case study involving noncommitment bias in
structure on public opinion on the anticipated usage of a
transit ridership may also be of interest. In new transit service.
these cases demand elasticity
coefficients, if available from past studies, 5.3. Forecasting Future Travel on Road
would be useful. Typically an elasticity Segments and/or Transit
coefficient is developed with respect to a Lines
A variety of forecasting procedures are Another advantage of the stepwise, or
available, ranging from the extrapolation of sequential, procedure is that it generates
past trends to complex mathematical models several types of useful information at the end
involving several steps. A transportation of various steps. The disadvantage of the
planner must recognize the advantages and procedure is that it needs a large amount of
disadvantages of each procedure. Two data for model development. It also requires a
procedures are examined for illustration. sound knowledge of one of the available
Direct Estimation of Traffic Volume by Trend computer software models that is specially
Analysis. If traffic volume data are available for designed for developing and applying these
a road segment or a transit line of interest for models. A great deal of research has been
several years in the past, the historical trend performed and is still being continued to
can be identified and extrapolated to estimate improve this stepwise modeling procedure. It
future volumes. This approach, of course, is should be acknowledged that the staff of the
appropriate if the trend is expected to continue, former Bureau of Public Works and later the
which commonly is true for short-range staff of the Federal Highway Administration
forecasts. Trend-based forecasts are and
appropriate also for aggregate values such as Urban Mass Transportation
total VMT or transit rides in an urban area. Administration made tremendous contributions
However, major changes in the land to the development of various procedures and
development pattern and/or transportation computer software. An historical overview of
network can cause substantial changes in the the development of planning and modeling
travel pattern, and if such changes are likely procedures in the United States is presented
then trend extrapolation will not be appropriate. by Weiner (1992).
Therefore, for long-range forecasts of traffic The stepwise procedure is popularly known as
volumes on individual segments of a road the four-step modeling process because it
network or the number of passenger trips on includes four major steps: trip generation, trip
individual transit routes, trend analysis is not distribution, mode choice, and traffic
used. assignment. Additionally, network analysis
must be done to develop a few types of
Stepwise/Sequential Procedure. A widely used information that are needed for the other steps.
travel estimation procedure for long-range These steps and procedures involved with
forecasts of traffic volumes on a highway each are discussed in detail in the following
network uses several steps in a sequence, as sections.
shown in the flowchart of the Figure below.
Each step requires a particular type of model
or procedure, and there are different choices of
models at each step. One of the major
advantages of this procedure is its ability to
reflect several types of changes that may occur
in the future:
1. Changes in trip-making rates
2. Changes in development pattern,
resulting in altered travel pattern
3. Changes in transportation mode usage
4. Changes in transportation network
  task can be accomplished with either
aggregate or disaggregate models (both
used in planning studies)
For aggregate models
 total number of trips (trip ends)
1. TRIP GENERATION generated in a zone is used as the
very important step because it sets up dependent variable
not only the framework for the following
tasks but also some of the controlling For disaggregate models
values, such as the total number of trips  trips made by a household (or a
generated in the study area by location business establishment) are used as the
and trip purpose. dependent variable
Commonly used units for trip generation  trip ends generated by households
analysis and/or any other trip generating units
 household such as business establishments in a
 dwelling unit (DU) zone are combined to produce the zonal
 business establishment. (total) value.
Traffic Zones – larger areas that are used to 3. TRIP GENERATION BY HOUSEHOLDS
aggregate the results of a trip generation Household-generated trips - comprise more
analysis for a study area. than 80 percent of all trips in an urban area.
Comprise the remaining portion of total
A thorough analysis of all these types of trips travel
requires a large amount of data.  Trips by nonresidents
Origin-destination (O-D) surveys - using this  commercial vehicles such as taxis and
the data are collected. trucks
 public utility
 public service vehicles
Trips generated by households
Home-based trips - one end, either
origin or destination, located at the
home zone of the trip maker.
Non-home-based trip - if both ends of a
trip are located in zones where the trip
maker does not live.
MODEL DEVELOPMENT – made simplified if
it is assumed that the home end of a trip is a
2. Models for Internal Passenger Trips: production (P) in the zone of the trip maker's
Aggregate and Disaggregate Models residence irrespective of whether it represents
GOAL of trip generation models the origin or destination of the trip, given the
 estimate the total number of trip ends for predominance of home-based trips in an urban
each purpose generated in each traffic area
zone based on socioeconomic and / or  According to this approach, the non-
land use data for the respective zones. home end of a home-based trip is
considered to be attraction (A).
 For a non-home-based trip
 o neither its origin nor its combinations of household size and
destination at the trip maker's auto ownership.
residence, production and  used widely for estimating trip
attraction are synonymous with production by households.
origin and destination, Chatterjee and Khasnabis (1973) - discussed
respectively. alternative techniques for a statistical analysis
of these models
(P) = Summation of (No. of households of a
NOTE: For ... category) x (Trip rate for households of that
 home-based trips the activity at the non- category and for that specific purpose) -
home end determines the trip purpose When the households in a traffic zone are
 non-home-based trips usually are not cross-classified by size and auto ownership,
further stratified by purpose. total trips made by households in a zone for a
4. Cross-Classification or Category Models specific purpose.
for Household
Household trip rates - vary significantly
according to certain socioeconomic
characteristics and the size of a household.
Household characteristics that have been Choice of household characteristics
found to be significant for trip generation and  used for developing the various
are commonly used in trip generation models categories for trip production rates may
are: vary from one study to another.
1. Household size One advantage of disaggregate models
2. Auto ownership developing these models a full-scale O-
3. Income D survey is not needed.
5. MODELS FOR TRIP ATTRACTIONS
Aggregate Models - common practice to use
in the form of regression equations for trip
attractions.
Dependent variable = total number of
trip attractions for a specific trip purpose
in a traffic zone
Independent variables - employment-
related, and they represent zonal total
values.
Hypothetical examples of trip attraction (A)
models are presented below:

Cross-classification (Category Models)

 includes trip production rates for Full-scale O-D survey - development of
different types or categories of aggregate models usually requires this.
households defined in terms of different
Coefficients of the regression equations - urban area, and the planners may have to
vary from area to area. analyze these trips thoroughly.
Choice of independent variables and trip- Comprehensive study - should pay attention
purpose categories also may vary from one to travel demand of all categories, although the
study to another. level of detail may vary.
Home-based productions in a zone may not be Analysis of commercial vehicle travel -
equal to the corresponding attractions in the neglected in most urban transportation studies.
same zone These vehicles are garaged in nonresidential

Non-home-based productions in a zone should locations and include trucks of all sizes, taxicabs,

be equal to corresponding attractions in the rental cars, service vehicles of plumbers and
electricians, etc.
same zone.
Area-wide (total) productions (p's) - should
8. Forecasting Variables Used in Trip
be equal area-wide (total) attractions
Generation Models
Independent variables - availability of data for
this is an important issue that can influence the
selection of a variable in developing trip
generation models
availability of data for the base year is less
Estimated area-wide (total) productions -
problematic than that for future years.
when synthetic or borrowed models are used it
if data for an independent variable are not available
would not be equal to the estimated area-wide for the base year, they cannot be used in model
(total) attractions. development.
Zona attractions - to achieve a balance this is Before building and adopting a model
adjusted proportionately such that the adjusted  model developer must recognize
area-wide attractions equal area-wide whether the independent variables used
productions in the model can be forecast by the
responsible planning agency
Adjustment or scaling factors for  if such forecasts would be very difficult
attractions: then it may be desirable to avoid using
those variables in the model.
Aggregate values of socioeconomic parameters
used in trip generation models are not very difficult
to forecast with the existing state of the art.
Difficulty usually involves the task of disaggregating
7. Commercial Vehicle Traffic in an Urban socioeconomic data at the zonal level.
Area Example: it may not be very difficult to predict the
Proportion of each category of trips varies total number of households in each zone along with
according to the size and other characteristics their average size and auto ownership.

of an urban area. However, it would be difficult to cross-classify the

Example: the proportion of through trips is usually predicted number of households in a zone

larger in smaller size areas. In some cases, trips of according to specific categories based on

one or more of these other categories may be the household size and auto ownership.

cause of major problems and thus will require

special attention.
Example: through traffic may be the major
issue in the case of a small or medium-sized
 - thought of as a factor which
distributes the total trips produced
by a zone among all the possible
destination zones.
In this sense, the sum of the expression over all destinations
should be equal to unity. Thus:

TRIP DISTRIBUTION MODELS

 Predict the no. of trips that will be made
between a pair of zones for a particular trip
purpose
 mathematically describe the
the number of trips attracted to a zone (when such
DESTINATION-CHOICE phase of the
data is independently available), Tj (Trips at
sequential demand analysis procedure
destination j) is used as a surrogate for f(aj) (in situ
Factors (for a given trip purpose) which affect
attributes).
the number of trips between two zones:
Example (GRAVITY MODEL)
 Number of trips produced by the origin zone
Consider the following six-zone model of a town.
 The degree to which the in-situ attributes of
Zones 1, 2, and 3 are fully residential areas and
the destination zone attract trip-makers. The
Zones 4, 5, and 6 are purely shopping areas. The
attributes which gain importance vary with
shopping areas, the shopping trips attracted (per
the trip purpose.
day), the shopping trips produced (per day), and
 The factors that inhibit travel, between a
the travel distances. are as shown in Table 9.1. The
pair of zones. These factors could be: travel
cells which have a "_" imply that those data are
time, travel distance, travel cost and so on
irrelevant to the problem.
Example:
Determine the trip distribution between the zones
If one is modelling the number of shopping trips for the following different scenarios:
attracted to a zone then the type of attributes of
the zone which assume importance will be the (a) Use the origin constrained gravity model,
total shopping floor area, the number of retail assuming f(aj) to be a linear function of the
outlets, and the likes.
shopping area (in square meters) with a slope of
On the other hand, if one is modelling the 0.01 and constant term of 10. Also assume h(d ij) to
number of work trips attracted to a zone then
the type of attributes of the zone which assume be dij2 where dij is the distance in km.
importance will be the number of offices, type of (b) Use the origin-destination constrained gravity
offices, and so forth.
model with the same relevant.
GRAVITY MODEL uses the following basic
form to determine the trips between an origin
zone, I and a destination zone j:

ORIGIN-CONSTRAINED GRAVITY MODEL

 This relation states that a small addition in
opportunities [dVi(j)] after the jth destination causes a
correspondingly small increase in the probability
[dPi(j)]. This increase is proportional to the probability
that no destination is chosen by the jth destination
and the number of additional opportunities.

Using Pi (0) = 0 (since it is assumed that Vi (0) =) &

solving the differential eq., we get
 Pi (j) = 1 – exp [ - LVi (j)]
And since, by definition, Πij = Pi (j) - Pi (j-1), we
have

INTERVAL OPPORTUNITIES MODEL  Πij = exp [ - LVi (j-1)] - [ - LVi (j)]

POSTULATE OF TRAVEL BEHAVIOR on which this

model is based:
Probability of choice of a particular destination
(from a given origin for a particular trip-purpose)
 proportional to the opportunities for trip- Example (INTERVAL OPPORTUNITIES MODEL)

purpose satisfaction at the destination and For the 1200 shopping trips from Zone A, three
inversely proportional to all such destinations exist. The destinations are Zones X, Y,
opportunities that are closer to the origin. and Z. The shopping areas available in each of the
Inverse proportionality zones and their distances from Zone A are given in
 interpreted as proportionality to the Table 9.2. Assuming a proportionality constant of
probability that none of the closer 0.35 and assuming a thousand square metres of
destinations (opportunities) are chosen. shopping area as one opportunity, determine the
In-situ attractive properties of the destination trip distribution from Zone A.
 modelled as opportunities and the
impedances are measured in terms of the
number of opportunities which are closer.’
In order to formulate the postulate as a
mathematical model, the following notations are
used:
L: Constant of proportionality (or calibration
DESTINATION CHOICE MODEL
constant) POSTULATE OF TRAVEL BEHAVIOR on which this
model is based:
J: Total number of destinations
1. The probability that Destination j - will be
j: Index indicating the position of a destination (from
chosen from origin i for a particular trip-
the given origin), j = 1 for the farthest purpose if the perceived utility derived from
choosing j is greater than the perceived
Vi(j): Cumulative function of opportunities up to and
utility derived from choosing any other
including the jth destination from origin i.
destination
Pi(j): Probability that a destination is chosen from Deterministic component - approximate utility that
can be obtained from the destination, given the
the ith origin by the time the jth destination is
destination’s in-situ attributes and the impedances.
reached
Stochastic component - approximation of the
Πij: Probability that the jth destination is chosen random variability assumed to be present in the
utility because of the fact that this is a quantity
from the ith origin.
which is perceived by humans
 dPi(j) = L {1-Pi(j) }dVi(j)  ui (j) = vi (j) + ei (j)
Where
ui (j) – utility derived from destination j
vi (j) – deterministic component
ei (j) – stochastic component
2. The probability that destination j will be
chosen from origin i for a particular trip-
purpose is equal to the probability that the
perceived utility derived from j is greater
than the perceived utility derived from
choosing any other destinations.

Πij = known as the multinomial logit model

Example (DESTINATION CHOICE MODEL)
For the data given in Example 9.2 & assuming v A (j)
= (0.5 x shopping area of Zone j in ‘000 sq.m) –
(0.23 x distance to j in km), determine the trip
distribution using the multinomial logit model.

ENTROPY MODEL - trip distribution

 is not a behavioral model.
 does not strive to predict the trip distribution
by modelling the human behavior aspects
related to choosing a destination.
 to determine a distribution of trips which is
most likely to occur assuming that each trip
occurs independently of another.