Ch t 13:

Chapter 13
Transportation Demand
A l i
Analysis

Satish Ukkusuri , Ph.D.

Ph D
Assistant Professor
Rensselaer Polytechnic Institute,
JEC 4032
Troy, NY 12180-3590
Previous Lecture Recap
2

™Transportation as a comprehensive discipline

™Frequently unforeseen issues complicate plans (e.g. NYC
congestion pricing, I-35 bridge collapse)
™Introduction to Intelligent Transportation Systems
lli
™Intelligent
™I Transportation
T i Systems
S use technology
h l to
help solve transportation problems without simply
buildingg more roads

™Question: What are common applications

of ITS in modern motor vehicles?
Lecture Outline
3

™More on Intelligent Transportation Systems (ITS)

™Different components of the ITS system

™Transportation Demand Analysis

™Four-step process
™Introduction to mathematical modelling theories for each
of the four steps; next lecture will focus on more
practical applications (i.e. examples)
Advanced Vehicle Control System (AVCS)
4

™Two major components: Collision Avoidance

Systems and Navigation
™I l d
™Include:
™Intelligent/Adaptive Cruise Control
™Collision Avoidance Systems
™Automated Highway System (AHS)
Intelligent/Adaptive Cruise Control
5

The design ideology behind a Intelligent/Adaptive Cruise Control system.

Advanced Public Transportation System (APTS)
6

™ ITS technology applied to public transportation

operations.
™ Real-time:
R l ti
™ Information to users.
™ Scheduling displays at
bus and subway stations.
™ Fleet management.
™ Technologies used: AVL, DGPS and EDI.
ITS in Transit
7

Adaptive Signal Timing and Route Destination

Communication Control Display
Automated Vehicle
Location and
Automated Fare Identification
Collection and
Passenger
Counting RTA

$OR

Smart Card Reader Vehicle

Diagnostics
Driver Information Display
p y
Silent Alarm
Terminal operations
8

hl complex
™Highly l operations.
™Many agents interact.
™Large terminals can contain greater than 10,000
containers.
™Container location and identification are a major
issue. ™ Crane operators can
earn large salaries and
work short schedules
because they must be
extremely efficient
™ Lots of research money
available for developing
computer algorithms for
operating
ti cranes
Advanced Rural Transportation System
9

(ARTS)
™Designed to enhance safety in rural areas.
™Includes:
™Weather advisory systems
™Emergency signaling
™Route guidance
™Tourism information

A highway advisory sign for an ARTS.

ITS Research @ RPI
10

™ATIS (Advanced Traveler Information System)

™Drivers input a destination and based upon the travel
time of other drivers already on the network, an
optimal route can be determined to bring travelers to
ttheir
e dest
destination
at o in a ttimely
e y fashion.
as o

The electronic devices that make up the

ATIS system.

Software used to determine optimal routes.

ITS Research @ RPI
11

™ETTM (Electronic Toll and Traffic Management)

™Through the use of two Electronic Toll Tag Readers
placed a known distance from each other, speed, travel
times and volumes can be calculated and used in the
management
a age e t oof spec
special
a ttraffic
a ce events.
e ts
™ Question: What do
you think is the main
reason for 5mph or
15mph speed limits
on EZ-Pass at NY toll
booths?
™ Answer: Safety
Electronic Toll Tag

Electronic Toll Tag Reader

ITS Issues and Challenges
12

Traffic Transit
Freewayy
Si
Signall M
Managementt
Management
Control
ITS is not yet
a system,
system the Multi-Modal
Travel
pieces are Electronic Inf. Centers
Electronic
Fare
being put Toll
Payment
together.
Emergency
RR Grade Response Incident
Crossing Safety Management Management
The Future…
13

™Increased Multi-technology Integration

™Multifunctional electronic cards (Smart cards)
™Electronic Toll Collection + Fare + Shopping
™Increased Multi-institutional Integration
™Non-traditional players will jump to the
transportation wagon.
™Credit card companies + Banks
™Communication Companies
™Increased Functional Integration
™Management + Traffic Information + Demand
Estimation
The Future…
14

™Increased consolidation of the transportation

industry.

™Increased consolidation of the high-tech

i d t
industry.

™Increased consolidation of information. The

privacy “issue” will be overcome.
Summary
15

™The issues of congestion, safety, access,

environment, funding and security present unique
and interesting challenges to transportation
engineers.
™Intelligent Transportation Systems are present in
vehicles, public transit, freight operations and
enable us to make Intelligent transportation
decisions.
™ITS is still in development and the future of ITS will
allow for the integration and consolidation of
technologygy and information.
Chapter 13: Components of Demand Analysis
16

™There are four parts of demand analysis:

™Tripp Generation
™Trip Distribution
p
™Mode Split
™Traffic Assignment
™There are other steps that can be considered part
of planning but these are the main ones
Demand Modeling Sequence
17
Trip Generation
18

™Some basic terms:

™Home Based Trip p (HB)-
( ) a tripp where the home of
the trip maker is either the origin or the
destination
™Non Home Based Trip (NHB)- a trip where the
home is neither end of the trip
™Utility Functions - what value does each user put
on being able to travel (eg. Utility of getting
groceries is higher than going to the movies)
Trip Generation
19

™Pi = f(z1,z2,z3…), trip ™Ai = f(y1,y2,y3…), trip

production for zone I attraction for zone I

™Personal Trip ™Personal Trip

Productions factors z’s
z s… Attractions factors y’s
y s…
™Income ™Industrial,
™Household structure commercial and
™Family size other services
™Car Ownership ™Zonal Employment
™Value of Land ™Recreational
Resources
sd a density
™Residential d s y
™Accessibility
Trip Distribution
20

O 1 2 3 4
™Goal is to figure out how D
many people l go from
f 1 T11 T12 T13 T14
zone i to zone j
™Often represented with
2 T21 T22 T23 T24
Origin-Destination Matrix
g g to
™VERY challenging
figure this out from only 3 T31 T32 T33 T34
knowing X1 trips from
zone 11, X2 trips from 4 T41 T42 T43 T44
zone 2, etc
Factors Influencing the Choice of Mode
21

™Characteristics of the journey

™Trip Purpose (Business, Leisure, etc.)
™Time of day of the journey (Morning, Afternoon, Night)
™Characteristics of the transport facility
™Quantitative factors:
™Relative travel time: in-vehicle, waiting and walking times
™Relative money costs (fares
(fares, fuel
fuel, and direct costs)
™Availability and cost of parking
™Qualitative factors
™Comfort and Convenience
™Reliability and regularity
otect o a
™Protection and
d secu
security
ty
Mode Split
22

™Using the O-D matrix, break up each trip count Tij

™Of the trips,
p , how manyy are byy car,, bus,, train,, etc
™More O-D matrices can be generated representing
just the passenger car trips, or the transit trips
O 1 2 3 4 O 1 2 3 4
D D
1 C11 C12 C13 C14 1 T11 T12 T13 T14

2 C21 C22 C23 C24 2 T21 T22 T23 T24

3 C31 C32 C33 C34 3 T31 T32 T33 T34

4 C41 C42 C43 C44 4 T41 T42 T43 T44

A simplified OD Matrix
23
Mode Choice
24

™ Utility – what the individual seeks to maximize.

™ Example: we have this utility function:
Vcar = 0.25 − 1.25 * IVT − 0.3 * (C / I ) + 1.1* NCAR

Alternative specific
constant
In-Vehicle
In Vehicle Time Cost/Income
Number of Cars

™ The IVT, ACC, and NCAR variables take on the values of 0 or 1.

These are Discrete Choice variables.
Mode Choice
25

™ For Utility functions, mathematical transformations exist to

understand the probability distribution of a specific utilities
of choices.
choices
™ Logit Function:
exp( V 1 )
P1 =
exp(( V 1 ) + exp(( V 2 )

™ Probit Function: 1 x1 2 2 ρx1 x2 x2 2

∞ V −V2 + x1 exp( [( ) − + ( ) ])
1− ρ 2 σ1 σ 1σ 2 σ2
P1 = ∫ ∫
−∞ −∞ 2πσ 2σ 1 1 − ρ 2
dx1dx 2

™The Probit function has a covariance matrix of the Normal

distribution in the form of: σ 12 σ 1σ 2 ρ
Σ = ( )
σ2
1 σ 2 ρ σ 2
Mode Choice
26

™ Useful Properties of Disaggregate Demand (DM) models:

™ DM are based on theories of individual behavior
™ DM models are estimated using individual data
™ They may be more efficient than conventional models
™ Variability in the information can be utilized
™ They can be applied at any aggregation level
™ They are less likely to suffer from bias
™ DM models are probabilistic
™ The expected number of people using a certain travel option is the sum
of the probabilities of choosing that alternative:
N i = ∑n
P in

™ An independent set of decisions may be modeled separately considering

each one as a conditional choice.
™ DM models have explanatory variables with coefficients that explain utility
utility.
™ Flexible representation of policy variables.
™ The coefficients have a direct marginal utility interpretation.
Mode Choice: Discrete Choice Models
27

™ The framework for generating discrete choice

models is the random utility theory, which says
has the following postulates:
1. Individuals belong to a given homogeneous
postulate Q, and they act rationally and possess
perfect information.
2. There is a certain set of alternatives A ={A1,…,AN},
and a set of measured attributes X of the individuals
and their alternatives.
ƒ A given individual q is endowed with a set of attributes
(q) ε A
x ε X and A(q)
ƒ We can determine the choice set of alternatives in later
lectures
Traffic Assignment
28

™Apply the O-D matrices to the existing networks

p , we know that 4000 vpd
™For example, p travel from
zone 1 to zone 4. What roads will they use?

1
3

5 6
2
Methods for Traffic Assignment
29

™Shortest Path/Lowest Cost Algorithms

g
™Static Network Assignment
™Dynamic Network Assignment
y
™System Optimal
p Network Loading
g
™Equilibrium models
Equilibrium models
™Non-Equilibrium
™Non
Traffic Assignment Methods
30

™Introduction
™The main objectives of these methods are:
™P i
™Primary
ƒ To obtain good aggregate network measures (e.g. total
revenue by a bus service)
ƒ T estimate
To i zone to zone travell costs/times
/i for
f a given
i level
l l
of demand
ƒ To obtain reasonable link flows and to identify heavily
congested areas
areas.
™Secondary
ƒ To estimate the routes used between each O-D pair
ƒ T estimate
To i which
hi h O-D
O D pairs
i use a particular
i l link
li k or route
ƒ To obtain turning movements for the design of future
junctions.
Traffic Assignment Methods
31

™Introduction
™The basic inputs required for assignment models are:
™Trip Matrix that expresses estimated demand
™A network of links and their properties including
speed flow curves
speed-flow
™Principles or route selection rules thought to be
relevant to the p
problem in question
q
Traffic Assignment Methods
32

™Route Choice
™When selecting a route between two points, we often
look at journey time,
time distance,
distance monetary costs,
costs
congestion, queues, scenery, comfort, etc.
™Different travelers choose different routes for two
diff
different
t reasons:
™Individual perceptions
™Congestion effects
™Each assignment method has several steps which must
be treated in turn. Their basic functions are:
™T identify
™To id tif suitable
it bl proportions
ti off the
th trip
t i matrix
t i to
t
these routes or trees
™To search for convergence
g
Traffic Assignment Methods
33

™Route Choice
™Example: Which route would you choose?

Bypass: 3000 v/h capacity

Congestion area

A B
T
Town center:
t 1000 v/h
/h capacity
it
All or Nothing Assignment
34

™ The main assumptions are:

™ No congestion effects
™ Drivers consider the same attributes for route choice, and weigh
them the same way.
™ Pair by Pair Method
™ This is the simplest but not necessarily the most efficient method
™ We start from an origin and take each destination in turn
™ We initialize all VA,B = 0. Then for each pair (i,j):
™1 set B to the destination j
™1.
™2. if (A, B) is the back link of B then increment VA,B by Tij (i.e.
make VA,B = VA,B + Tij)
™3 set B to A
™3.
™4. if A = i terminate (i.e. process the next (i,j) pair), otherwise
return to step 2
All or Nothing Assignment
35

™ Once Through Method

™ Sometimes called the cascade method, as it accumulates flow from
nodes to links following the minimum cost trees from an origin i.
i Let
VA be the cumulative flow at node A:
™1. set all VA = 0 except for the destinations j for which Vj =Tij
™2 set B equal to the most distant node from i
™2.
™3. increment VA by VB where A is the back node of B (i.e. make
VA = VA + VB)
™4 increment
™4. i t VA,B by
b VB (i.e.
(i makek VA,B = VA,B + VB)
™5. set B equal to the next most distant node; if B = i then the
origin has been reached, begin processing the next origin,
otherwise
th i proceed d with
ith step
t 3.3
Congested Assignment
36

™Wardrop’s 1st principle for Equilibrium

™ “Under equilibrium conditions traffic arranges itself in
congested networks in such a way that no individual trip
maker can reduce his/her path costs by switching
routes.”
outes
™If all trip makers perceive costs in the same way (no
stochastic effects), then: “Under equilibrium conditions
ttraffic
ffi arranges it
itself
lf in
i congested
t d networks
t k such h that
th t allll
used routes between O-D pairs have equal and minimum
costs while all unused routes have ggreater or equal
q
costs.”
Congested Assignment
37

™Wardrop’s Equilibrium
™Several techniques have been proposed as reasonable
approximations to Wardrop’s equilibrium: some of them
are simple heuristic approaches and the most interesting
ones
o es follow
o o a more o e rigorous
go ous mathematical
at e at ca pprogramming
og a g
framework.
™We must ask the following questions about these
t h i
techniques:
™Is the solution stable?
™Does it converge to the correct solution (Wardrop
Equilibrium)?
™Is it efficient in terms of computational requirements?
Congested Assignment
38

™ Wardrop’s Equilibrium
™ We can use the following indicate to measure how close a solution
is to Wardrop
Wardrop’ss Equilibrium:

∑ T ijr ( C ijr − C * ij )
δ = ijr

∑ ij
T ij C *
ij

The excess cost of travel over a particular route relative to

the minimum cost of travel for that (i
(i, j) pair.
pair
Congested Assignment
39

™Wardrop’s 2nd Principle

™“Under social equilibrium conditions traffic should be
arranged in congested networks in such a way that the
average (or total) travel costs are minimized.
Congested Assignment
40

™ Hard and Soft Speed-Change Methods

™ In algorithmic terms the two variants can be described as follows:
™1 Select an initial set of current link costs,
™1. costs usually free-flow
free flow
travel times, make n = 0.
™2. Build the set of minimum cost trees with the current costs and
assign (all or nothing) the matrix to them obtaining a new set of
flows; increment n by (make n = n+1).
™3. (a) Hard speed change: recalculated the link costs
corresponding to the new flows (b) Soft speed change: calculate
the current link costs as the arithmetic average of the costs in
the previous iteration and those calculated in (a).
™4 If the flows or current link costs have not changed significantly
™4.
in two consecutive iterations, stop: otherwise, proceed to step 2.
Congested Assignment
41

™ Incremental Assignment
™ We need a trip matrix T and applying a set of proportional factors pn
such that ∑ n p n = 1
™ The algorithm is the following:
™1. Select an initial set of current link costs, usually free-flow travel
times Initialize all flows Va =0;
times. 0; select a set of fractions pn of the
trip matrix T such that ∑ n p n = 1 ; make n =0.
™2. Build the set of the minimum cost trees (one for each origin)
using the current costs; make n = n+1. n+1
™3. Load Tn = pn T all or nothing to these trees, obtaining a set of
n −1
auxiliary flows Fa: accumulate flows on each link V a = V a + Fa
n

™4 Calculate
™4. C l l t a new sett off currentt lilinkk costs
t based
b d on the
th flows;
fl V an
if not all fractions of T have been assigned proceed to step 2.,
otherwise stop.
Economic Evaluation
42

™Once individual projects have been modeled:

™Compute travel times with and without project
™Estimate economic savings attributable to project
™Forecast economic savings
™C
™Compute economici indicators
i di off performance
f
Differences between Planning and Design
43

Planning Design

How many people

H l Whats
Wh t the
th best
b t route
t
want to get from A to build from A to B?
to B?
How much will How much will it
people pay to use cost to build?
this route?
What modes are How can they be
best to use (bus, integrated together?
train, car)?