Introduction To Travel Demand Forecasting

Introduction to Travel
Demand Forecasting
THE FOUR-STEP FORECASTING MODEL
The classic forecasting model has the following sub models:
a. Trip generation and attraction
b. Trip distribution
c. Modal split
d. Traffic assignment
THE FOUR-STEP FORECASTING MODEL
ORIGIN-DESTINATION TABLE (OD
MATRIX)
• Once the study area is defined, it is divided into a number of zones
normally defined by political or administrative boundaries. These are
called internal zones. Depending on the coverage of the study, a zone
may correspond to one barangay, town, or city.
• The study may also consider some zones outside the study area,
which may greatly affect the trip patterns, These are called external
zones.
ORIGIN-DESTINATION TABLE (OD
MATRIX)
• For example, If Metro Manila is taken as
a study area, there is a need to consider
the outlying areas of Cavite, Laguna,
Rizal, and Bulacan, among others
METHODS FOR ESTIMATING TRIP
GENERATION AND ATTRACTION
A number of methods are available for estimating trip generation and
attraction. The more commonly used ones may fall under any of the
following:
a. Growth rate method

b. Category analysis
c. Regression model
Growth Rate Method
The number of trips is assumed to be influenced by several variables
like population, car ownership, or income. If future values of these
variables can be estimated statistically, then the future number of trips
may be based on the growth rates of these variables and may be
estimated.
Category Analysis
Let us say that the number of cars is considered the main variable in
determining trip making in a certain area. Based on the present
number of households and trip making activities, the future trips can
be estimated as follows:
Category Analysis
Suppose the following table shows the current condition based on

interviews:
Category Analysis
From these data, the average trip generation rate per category can be
estimated by dividing the number of trips by the number of households
in each category.
Category Analysis
These trip generation rates are then applied to the forecasted number
of households per category to obtain the future trip generation.
Category Analysis
The city of Vigan had the following land use activity statistics during
morning peak hour in 2000. Estimate the total trip generation during
that period.
Category Analysis
Estimated person trips can be computed by multiplying column a with

column b.
Category Analysis
The table summarized the total trip generation. Residential units

produce about 20,000 person trips in the morning. Trip attraction to
various establishments is also obtained by considering all person trips
coming in during the morning peak hour. It must be noted that total
trip attraction does not equal total trip production because there are
trips coming from outside the city of Vigan.
Category Analysis
Regression Analysis
The different indices or dependent variables normally considered that

may have influence on trip making.
Regression Analysis
In the Metro Manila Urban

Transportation Integration Study
(ALMEC Corp 1999), the following
trip generation and attraction
models are utilized:
TRIP DISTRIBUTION
The objective of this model is to determine from which zones these trip
generations (Oi) are generated and to which zones these trip
attractions (Dj) are bound.
A number of models have been developed in the past to distribute trips
and these are classified as follows:
a. Present pattern method
b. Model method
TRIP DISTRIBUTION
Present pattern method
a. Uniform factor method
b. Average factor method
c. Detroit method
d. Fratar method
TRIP DISTRIBUTION
Present pattern method
Some of the disadvantages of the present pattern method are the
following:
A. Complete OD matrix is required.
B. Change of land use is not considered; where the present trip
distribution is zero, future is also zero.
C. If there is error in current OD matrix, the error is expanded with the
same rate.
TRIP DISTRIBUTION
Model Method
A number of model methods are patterned after Newton’s law of
gravitation. Hence, these are termed gravity models.
a. Basic type of gravity model
b. Bureau of Public Roads type gravity model
c. Voohees type gravity model
MODAL SPLIT
Calculation of choice ration may be done using either binary or multiple
choice. Under the method of multiple choice, all possible modes are
identified, and the share of each mode is calculated. The use of binary
choice is relatively easy compared to multiple choice.
MODAL SPLIT
The Generalized Cost
The generalized cost attempts to put costing on quantifiable and
sometimes unquantifiable costs that may be incurred during a trip
using a specified mode. Two cost items may be quantified easily: cost
(in ordinary sense, such as fare when taking public transport) and travel
time.
MODAL SPLIT
Considering several competing modes, the mode that will have the
least generalized cost will be chosen. Supposing that there are two
modes, with corresponding generalized costs:
MODAL SPLIT
The shares of trip makers likely to use modes 1 and 2 are shown by the
areas of the time value distribution in figure
MODAL SPLIT
Disaggregate Choice Model
This method originated from microeconomics. It assumes that a person
will use a particular mode with maximum utility for him.
MODAL SPLIT
Example: A calibrated utility function for travel in a medium-sized city
by car, bus, and light rail is
Where X1 is the cost of travel in pesos and X2 is the travel time (min).
Calculate the modal split for the given values:
MODAL SPLIT
Example: A calibrated utility function for travel in a medium-sized city by
car, bus, and light rail is
Where X1 is the cost of travel in pesos and X2 is the travel time (min).
Calculate the modal split for the given values:
If a parking fee of P30 per trip is imposed, what would be the split to the
other two modes?
MODAL SPLIT
MODAL SPLIT
Similarly, Pbus and Plight rail can be obtained using the same formula.
The calculations are summarized in the table below. The current modal
share shows that car and light rail have almost the same share.
Now, when a parking fee of P30 is imposed to each traveller, the utility
of car will be reduced as follows
(30+30)
MODAL SPLIT
Probabilities are recomputed and the results are shown below.
The share of car is greatly reduced from 36 percent to 24 percent, an

overall reduction of 33 percent. This results in car users shifting to bus
and rail, which now have an increased share of 32 percent and 44
percent, respectively.
ROUTE ASSIGNMENT
Given a road network, car drivers will normally use the route or
routes through which they will reach their destination in the least
time or distance. The same assumption is used to assign trips or
vehicles in the four-step model.
ROUTE ASSIGNMENT
The Shortest Path
A prerequisite to route assignment is the determination of the shortest
path from one node of the network to another. Given a road network
with known characteristics such as distance or travel time, the shortest
path may be found using Dijkstra’s algorithm.
ROUTE ASSIGNMENT
The Shortest Path
The algorithm or procedure follows these basic steps:
• First, label the staring node with a value equal to 0. This label is
permanent as this will not be changed.
• Consider each node adjacent to this node and give them temporary
labels. (Only those without permanent labels are considered.)
• If a node is unlabelled, it is given a label equal to the weight of the link plus the
value of the previously labelled node.
• If a node is labelled, calculate the value of the label and, if this is less than the
current value, then exchange its value with the smaller one; otherwise, leave
the label unchanged.
ROUTE ASSIGNMENT
The Shortest Path
The algorithm or procedure follows these basic steps:
• Choose the node with the smallest temporary label and make the
label permanent.
• Repeat steps b and c until the final node has been given a permanent
label. The shortest path has a length given by its permanent value.
ROUTE ASSIGNMENT
The Shortest Path
Example: Consider the road
network shown in figure. The
nodes are labelled a to k. The
links have values in travel time
in minutes. Determine the
shortest path (least travel time)
from a to k using Dijsktra’s
algoritm.
ROUTE ASSIGNMENT
Network Assignment
Given choices or routes, it is but natural to assume that the trip makers
will consider the route that would require the least time or least
distance. Based on this assumption, the methods used to assign the
trips within the road network are the following:
a. All-or-nothing assignment
b. Constant assignment ratio
c. Incremental assignment
ROUTE ASSIGNMENT
Network Assignment
• All-or-nothing assignment - Considering one OD pair, all trips are
assigned to the shortest path from point of origin to point of
destination. After these trips are loaded into the network, the level of
service of the roads in the network may change.
a. Find the minimum path between zones i and j.
b. Assign all trips Tiy the minimum path.
c. Repeat steps 1 and 2 for all OD pairs.
ROUTE ASSIGNMENT
Network Assignment
• Constant assignment ratio - In this method, it is assumed that the
number of trips assigned to a route is inversely proportional to the
travel time or cost of that route, i.e., more trips will be assigned to a
route providing shorter travel time or lesser cost. The basic steps for
the constant assignment ratio method are as follows:
a. Find out several possible routes between zones i and j.
b. Calculate the level of service (time or cost) on each route.
c. Divide the distributed trip to each route inversely proportional to LOS.
d. Repeat for all OD pairs.
ROUTE ASSIGNMENT
Network Assignment
• Constant assignment ratio.
Suppose that 100 trips are to be distributed among 4 routes with the
following travel times. Assign these trips using constant assignment
ratio.
ROUTE ASSIGNMENT
Network Assignment
• Constant assignment ratio.
The inverse of time for each route is computed. This serves as the level
of service of the route. The assigned trips to each route are obtained in
proportion to this LOS. Higher values would have larger share of
assigned trips. The result of assignment is shown in the table below.
ROUTE ASSIGNMENT
Network Assignment
• Incremental assignment - This method considers the influence of
previously assigned trips. It is based on the Wardrop’s Princicple,
which states that out of several routes available between zones i and
j, the routes that are used have equal level of service. The routes that
are not used have lower LOS.
ROUTE ASSIGNMENT
Network Assignment
• Incremental assignment
Consider the trips originating from one zone to be assigned. The
incremental assignment method takes the following steps:
a. Assign 1/n of the trips to route (by using all-or-nothing assignment or
constant assignment ratio).
b. Calculate the new level of service for the state that 1/n trips are
already assigned.
c. Assign next 1/n trips.
d. Repeat until all trips are assigned.

Introduction To Travel Demand Forecasting

Uploaded by

Copyright:

Available Formats

Introduction To Travel Demand Forecasting

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Introduction To Travel Demand Forecasting

Uploaded by

Copyright:

Available Formats

Introduction to Travel

a. Growth rate method

Suppose the following table shows the current condition based on

Estimated person trips can be computed by multiplying column a with

The table summarized the total trip generation. Residential units

The different indices or dependent variables normally considered that

In the Metro Manila Urban

The share of car is greatly reduced from 36 percent to 24 percent, an

You might also like