CHAPTER – 8

TRAFFIC ASSIGNMENT

8.1 DESCRIPTION OF TRANSPORT NETWORK

Route assignment, route choice, or traffic assignment concerns the selection of routes
(alternative called paths) between origins and destinations in transportation networks. It is the
fourth step in the conventional transportation forecasting model, following trip generation, trip
distribution, and mode choice. The zonal interchange analysis of trip distribution provides origin-
destination trip tables. Mode choice analysis tells which travellers will use which mode. To
determine facility needs, costs and benefits, the number of travellers on each route and link of the
network should be known (a route is simply a chain of links between an origin and destination),
and therefore to undertake traffic (or trip) assignment. Suppose there is a network of highways
and transit systems and a proposed addition. We first want to know the present pattern of traffic
delay and then what would happen if the addition were made.
Each number of trips from a particular origin to a particular destination must be assigned to the
alternative routes connecting the pair of nodes. While the path flows are of particular interest
from the traveller’s point of view, the network designer is usually more interested in the
corresponding link flows to detect bottlenecks.

8.1.1 Objectives of Traffic Assignment

During the classic traffic assignment stage a set of rules or principles is used to load a fixed trip
matrix onto the network and thus produce a set of links flows. This is not, however, the only
relevant output from the assignment stage; this has several objectives which are useful to
consider in detail. Not all of them receive the same emphasis in all situations nor can be achieved
with the same level of accuracy. The primary and secondary objectives of traffic assignment are
:
1. Primary:
 To obtain good aggregate network measures, e.g. total motorway flows, total
revenue by bus service;
 To estimate zone-to-zone travel costs (times) for a given level of demand;
 To obtain reasonable link flows and to identify heavily congested links.
 2. Secondary :
 To esttimate the ro
outes used beetween eachh O-D pair;
 To analyse which O-D pairs use
u a particullar link or rooute;
 To ob
btain turning movements for the desiign of future junctions.
In
n general teerms we sh
hall attain the primaryy objectivess more accuurately thann the
secondarry ones. Eveen within obj
bjectives we are likely too be more aaccurate withh those earliier in
the list. This
T is essen
ntially becau
use our mod
dels are morre likely to estimate coorrectly aggrregate
than disaaggregate vallues.
Itt is a criticaal step in th
he sequentiaal procedure, which dettermines linkk and OD ttravel
times, th
hereby influeencing OD choice and mode choicce through a “feedbackk” mechanism
m, or
through the solution
n of a route choice model integrateed with a m
model of vaariable OD fflows
(demand)). A schematic diagram for traffic asssignment is given in Figgure 8.1.

Figuree 8.1: Schem

matic Diagra
am for Depiicting Trafffic Assignmeent

8.1.2 Inp
puts and Ou
utputs in Assigning a Network
To carry out a traffic assignment, the followiing data are rrequired:
1. The
T number of trips thatt will be maade from onne zone to annother (this informationn was
determined frrom trip distrribution phaase.
2. Available
A hig
ghway or tran
nsit routes between zonees.
3. How
H long it will
w take to trravel on eacch route.
4. A decision ru
ule (or algoriithm) that ex
xplains how m
motorists orr transit userss select a rouute.
5. External
E tripss that were not consideered in the pprevious tripp generationn and distribbution
stteps.
 All
A assignment techniquees are based
d on route seelection. Thee choice of rroute is baseed on
several criteria
c such
h as travel tiime, length of travel, c ost of traveel, comfort, conveniencee and
safety.
The
T highway network is described
d by
y a system off links and nnodes (Figuree 8.2).
Link: A link is a section of a hig
ghway netwo
ork between two intersecctions.
Node: A node is eith
her the centro
oid of a zonee or the two or more inteersections.

Figure 8.2 Alternate Routes

R for T
Traffic Assiggnment
The
T number of
o available paths betweeen any pair of zones deepends on thhe mode of trravel.
In the case of privatee transportatiion modes a driver has ggood deal off freedom too select a posssible
path betw
ween different sets of paath variation
ns. Private annd public traavel demandd mainly deppends
upon trav
vel time and
d travel costt. These aspects insist thhe vehicularr traffic to sselect a partiicular
path betw
ween any paiirs of zones.
Traffic
T assignment is th
he stage off transportatiion networkk problem wherein thee trip
nges are allocated to the network. Asssigning of a network givves the folloowing outputts:
interchan
1. To
T estimate in
nter-zonal deemand by mode.
m
2. To
T determinee trip-maker’s choice of paths
p betweeen all zones along the neetwork.
3. To
T predict resulting flow
ws on the ind
dividual linkss of the entiire network oof the considdered
mode.
m
Trip
T assignm
ment processes involves earlier estim
mated demaand and netw
work descriiption
(links an
nd nodes) as inputs and individual
i fllows w.r.t inndividual linnks as outpuuts. The estim
mates
of link utilization
u caan be used to
t assess the likely level of service and to annticipate poteential
capacity problems.
8.1.3 Asssumptions in Trip Assignments
 A time perio
od of substaantial length, compared with the duuration of triips, in whicch the
ngestion in the networrk is relativvely constannt (either hiigh or low)); for
level of con
 example, a period of one or two hours. Such models are static, in contrast to dynamic
models.
 An input trip table giving the flow per hour from each origin zone to each destination
zone.
 A road network description consisting of nodes, links and link travel a time-flow
function, which increases indefinitely as flow increases without limit.
 In practice, the travel time pattern for each link is typically defined on its own flow,
ignoring the flows of opposing or conflicting links. However the links nominal capacity
may reflect the effect of intersecting links. The link capacity is not a strict upper limit on
flow.
 Drivers have perfect information about travel times (deterministic), or perfect
information plus a perception error (limited stochastic case), models with truly stochastic
times are much more difficult, and not considered.
8.1.4 Applications of Trip Assignment
Some of the applications of traffic assignment analysis to the network are:
1. To determine the deficiencies in the existing transport system by assigning the future
trips to the existing transport system.
2. To evaluate the effects of limited improvements and additions to the existing transport
system by assigning estimated future trips to the improved networks.
3. To develop construction priorities by assigning estimated future trips for the intermediate
years to the transport system proposed for those years.
4. To test alternative transportation system proposals by systematic and readily repeatable
procedures.
5. To provide design hour volumes on highway and turning movements at junctions.
Thus the assignment processes is applicable for both transport planners and highway facility
design engineers. The main need of traffic assignment processes is to evaluate:
 How the proposed transport system will work presently, and to the latter date.
 For the geometric design of highways and intersections.
Developments in computer technology made it possible to facilitate traffic assignment
techniques computations so laborious. A computer network assignment procedure requires:
 A way of coding the modal network for computer processing.
  An understanding of the factors affecting the trip-maker’s path preferences.
 A computer algorithm that is capable of producing the trip-maker’s preferred paths.
For computer analysis, the network is coded, key punched and stored in the computer memory.
The computer is then made to select the minimum path between the zones and assign predicted
trips to these paths. Traffic volumes are thus accumulated for each section of the network.
8.2 ROUTE CHOICE BEHAVIOUR:
The most fundamental element of any traffic assignment is to select a criterion which
explains the choice by driver of one route between an origin-destination pair from among the
number of potential paths available.

8.2.1 Wardrop criterion:

Wardrop had proposed two criteria that might be used to predict the paths between an origin
and destination pair. They are:
1. System optimal equilibrium criterion.
2. User optimal equilibrium criterion.

System equilibrium criterion:

“the trip times on all the routes actually used are equal and less than those which would
be experienced by a single vehicle on any unused route”
The first criterion is quite likely in practice. It is assumed that traffic will tend to settle
down into an equilibrium situation in which no driver can reduce his journey time by choosing a
new route.
The relationship labelled “average travel time” shows the variation in the average travel
time faced by individual vehicles as the traffic volume increases.
For example, when the traffic volume is 2000vehicles per hour, each vehicle will
experience an average travel time of 5.0 minutes per mile. When the volume is 3000 vehicles per
hour, then each vehicle will experience an average travel time of 15.0 minutes per vehicle.
The second relationship labelled “marginal travel time,” shows the increase in the
aggregate travel time experienced by all vehicles as traffic volume increases.

At low traffic volumes, the marginal or additional total travel time contributed to all
vehicles by an additional vehicle entering the traffic stream is small. At a volume of 3000
vehicles per hour the average travel time per vehicle is 15 minutes. The marginal travel time
imposed on the set of vehicles by one additional vehicle is 105 minutes. The extra vehicle would
actually experience a travel term increased by 90 minutes.
 Fig
gure 8.3 Va
ariation of Travel
T Timee with Trafffic Volume

Wardrop’s
W firrst criterion is equivalen
nt to the nottion of averaage cost priccing of econnomic
theory. Drivers
D mig
ght be regarrded as actiing selfishlyy in that thhey considerr only their own
individuaal travel timees in making
g route choicce decisions,, and not thee manner in which their route
choice in
nfluences thee aggregate travel
t time experienced
e by all motoorists. That iis, they base their
route cho
oice decision
ns on the aveerage travel time
t relationnship.
User equ
uilibrium critterion:
“The
“ averag
ge journey times
t mum which implies thaat the
of all motorists aare a minim
aggregatee vehicle hou
urs spent in travelling iss a minimum
m”.
A user optimal pattern reefers to the in
ndividual pooint of view of each mootorists who tends
to be unaaware of, or at least to be unresponsive to, certaain costs he iimposes on oothers. Hencce the
driver wiill rationally
y not take intto account th
hese costs inn his travel rrelated decissions. Driverrs are
only con
ncerned with
h the privatee costs that they
t must bbear themselves such as petrol costss, the
time cosst of making
g the trip and
a maintaining the veehicle. Conssequently, m
motorists tennd to
underestiimate the soccial cost of trip
t that shou
uld include aall impacts oof their activities on otheers.
A path flow is
i called Waardrop user equilibrium when no drriver has lesss costly alteernate
route.
 Wardrop restated “each path in use operates at minimum cost and each unused path
shows at least minimum cost”.
Wardrop’s second criterion implies that motorists select their routes according to the
marginal cost criterion of economic theory. Drivers are thought to as acting as though they are
aware of the way in which their route choices influence travel times of all drivers using the road
network. If motorists react to the marginal costs they create when they choose a route, then the
total travel time of all vehicles using the system will be minimized.
A number of studies of the route selection behaviour of motorists have shown that
motorists behave according to some criterion intermediate between the two criteria. Blunden and
Taylor both argue that wardrop’s second criterion describes the route choices of motorists better
than Wardrop’s first criterion. However, this evidence is far from conclusive and most of the
traffic assignment techniques used on a routine basis in urban transport studies assumes that
Wardrop’s second criterion governs trip-distribution behaviour. The other trip-distribution
techniques assume that the minimum individual average travel times govern trip-distribution
behaviour.
8.2.2 Diversion Curves
Diversion curve models was developed in early 1950s to know how many drivers would be
diverted from arterial streets to a proposed freeway in order to make decisions related to the
geometric design and capacity of proposed urban freeways.
This model employs empirically derived curves to compute the percentage of trips that would
use the freeway in route between two points on some measure of relative impedance between the
freeway route and the fastest arterial route between the two points.

California diversion curves

These curves used travel-time and travel-distance differences between two alternative paths to
estimate the percentage of trips that would use the freeway.
The formula for determining percentage usage of the freeway in route between two points is
given by
Where
p = percentag
ge usage,
d = distance saved
s in milees,
t = time saveed in minutess,
The diversion curve developed
d by Californiaa diversion cuurve methodd is shown inn figure 8.4.

Fig
gure 8.4 Caliifornia diverrsion curves
8.2.3 Dettroit diversiion curves
The Detrroit Area Traansportation study estim
mated diversioon from a soomewhat diffferent viewppoint,
still using
g a two-paraameter appro
oach. In thiss case, the paarameters foound to be reelated to freeeway
usage weere the ratio of expressw
way speed to
o arterial speeed, and the ratio of exppressway disstance
to arterial distance.. In each case
c the minimum
m appplicable paath was useed for the ratio
computattions.
The diversion curve developed
d by Detroit Arrea Transporrtation Studyy is given in fig 8.5.
 Fiigure 8.5 Deetroit diversiion curves
8.2.4 Burreau of Pub
blic Roads Diversion
D Cu
urve
Undoubtedly, the mo
ost widely used method of diversionn is that whiich is availabble in the Buureau
of Public Road’s series of traafic planning computerr programs. This form
m of diversioon is
dependen
nat on one parameter
p on
nly, the ratio
o of travel ttimes by thee quickest coombined artterial-
freeway route to the quickest arrterial-only route.
r With a one-param
meter relationnship, one ssingle
diversion
n curve defin
nes the relatiionship. Thee form of S-sshaped diverrsion curve iis similar to those
used in th
he Detroit sttudy for high
her speed rattios. Total frreeway usagee occurs wheen the travell time
ratios falll below 48% while no
o freeway ussage can bee anticipatedd when the travel time ratio
exceeds 150% of th
he quickest surface routte.Diversionn curve deveeloped by B
Bureau of P
Public
Road’s iss given in Fiigure 8.6.
 100

EQUAL TIME
90

Zone‐to‐Zone usage of freeway in

80
70
PASSENGER CARS
60
percent

50 50% USAGE
40
30
20
10
0
0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5
TRAVEL TIME RATIO

Figure 8.6 Diversion curve developed by Bureau of Public Road’s

The following formula can be fitted to these curves:

Where p = percent of traffic diverted to new system

tr = travel time ratio = .

8.3 THE MINIMUM PATH
8.3.1 The minimum path algorithm
Tree building is an important stage in any assignment method for two related reasons.
Firstly, it is performed many times in most algorithms, at least once per iteration. Secondly, a
good tree building algorithm can save a great deal of computer time and costs. An algorithm
searches out routes from a street network according to the driver route selection criterion. There
are two basic algorithms in general use for finding the shortest paths in road networks, one due to
Moore (1957) and one due to Dijkstra (1959). The two will be discussed using a more
convenient node-oriented notation: the length (cost) between links A and B in the network is
denoted by dA,B. the path or route is defined by a series of connected nodes, A-C-D-H, etc.,
whilst the length of the path is the arithmetic sum of the corresponding link lengths in the path.
Let dA denote the minimum distance from the origin of the tree S to the node or centroid A: pA is
the predecessor or backnode of A so that the link (pA, A) is part of the shortest path from S to A.
8.3 ROUTE ASSIGNMENT TECHNIQUES:
8.4.1 All-or-nothing assignment:
In this method the trips from any origin zone to any destination zone are loaded onto a single,
minimum cost, path between them. This model is unrealistic as only one path between every O-
D pair is utilised even if there is another path with the same or nearly same travel cost. Also,
traffic on links is assigned without consideration of whether or not there is adequate capacity or
heavy congestion; travel time is a fixed input and does not vary depending on the congestion on
a link. However, this model may be reasonable in sparse and uncongested networks where there
are few alternative routes and they have a large difference in travel cost. This model may also be
used to identify the desired path: the path which the drivers would like to travel in the absence of
congestion. In fact, this model's most important practical application is that it acts as a building
block for other types of assignment techniques. It has a limitation that it ignores the fact that link
travel time is a function of link volume and when there is congestion or that multiple paths are
used to carry traffic.
8.4.2 Multipath traffic assignment:
Mclaughlin developed one of the first multipath traffic assignment techniques. A driver route
selection criterion is used by Mclaughlin which is a function of travel time, travel cost, and
accident potential. The minimum resistance paths between each origin and destination pair are
calculated with all the link resistances set to values which correspond to a zero traffic volume.
The minimum resistance value between an origin and destination pair with resistance values less
than this maximum value are identified.
Mclaughlin used certain principles of linear graph theory to accomplish the multipath
assignment. Using an electrical analogy it is possible to identify through variable y that
corresponds to current, or traffic flow. An across variable x may be identified that corresponds to
potential difference, or traffic pressure.
Two postulates from linear graph theory may be introduced that are known as the vertex and
circuit postulates. At any vertex

Where e = the number of oriented terminal graphs, or elements

Yi = the through variable of the ith element
 ai = 0 if the ith element is not connected to V
= 1 if the ith element is oriented away from V
= -1 if the ith element is oriented toward V
For any circuit,

Where Xi = the through variable of the ith element

bi = 0 if the ith element is not in the jth circuit
= 1 if the ith element orientation is the same as the jth circuit
= -1 if the ith element orientation is opposite to the jth circuit
A subgraph is then established for each origin and destination pair with these representing two
vehicles. The connecting elements are the acceptable paths between the verticles plus one flow
driver element that corresponds to the car travel demand between the origin and destination pair.
The travel demand is assigned among the potential paths in accordance with the path resistance
values calculated during the path building phase. The traffic assigned to each path must be such
that the alternative paths have an equal across variable value.
The across variable X, the resistance value R(y) and the through variable y for each path are
assumed to be related as follows:
X = R(y)y
Equation is analogous to Ohm’s law in that potential is equal to the resistance times the flow. In
this case the resistance along a path is assumed to be a function of the flow along that path.
8.4.3 Cap
pacity restrraint assignm
ment
Capacityy restraint asssignment atttempts to approximate
a an equilibriium solution
n by iteratinng
between all-or-nothiing traffic loadings annd recalculaating link trravel times based on a
congestioon function that
t reflects link capacitty. Unfortunnately, this m
method does not convergge
and can flip-flop
f back and forth in
i loadings on
o some linkks.
This tripp assignmen
nt technique uses voluume-densityy relationshiips. As traaffic volumees
increase, travel speedds decrease due to increeased congesstion. Capaccity restraintt uses the linnk
freeflow speeds andd capacities to control the
t final assigned link speeds. Recall from thhe
session on
o network developmennt that link speeds and capacities are determinned by usinng
look-up ttables that reelate these variables
v to the
t facility tyype or functtional classiffication of thhe
link, the area type su
urrounding the
t link, andd the number of lanes. T
The capacity
y per lane is a
function of the LOS used for cooding the loook-up table. A LOS D ccapacity is normally
n useed
with the Bureau
B of Puublic Roads (BPR) form
mula.
The BPR
R function is the most coommonly useed function for
f relating changes
c in trravel speed tto
increasess in travel voolume. The B
BPR functionn is specifiedd as follows:

Figure 8.8
8 Variatioon of Travell Time Ratioo

88
Where:
T = congested link
l travel time;
To = original (fr
free flow) linnk travel tim
me;
v = assigned trraffic volumee; and
c = the link caapacity
EXA
AMPLE
Examplee 1
To demoonstrate how the most coommon assiggnment workks, an exampple network is consideredd.
This netw
work has twoo nodes haviing two pathhs as links.
Let us suuppose a casee where travvel time is noot a functionn of flow as shown
s in oth
her words it is
constant as shown inn the figure below.
b

Figgure 8.9
Two Link Problem
P withh constant traavel time funnction
All or noothing:
The traveel time functtions for bothh the links iss given by:
= 10;; t2 =15
And totall flows from
m 1 to 2.
q12 = 12
Since thee shortest patth is Link 1 all flows aree assigned too it making =12 and = 0.
User Equuilibrium:
Substitutting the traveel time in eqquations 1 - 5 yield to

Min : Z (x) =
= 10x1 + 15x2
Substitutting: x1 + x2 = 12

89
Substitutting , inn the abovve formulattion will yield
y the unconstraine
u ed
formulatiion as below
w:
Min Z(x)) = 10x1 + 155(12 – x1)
Differenttiate the above equatioon w.r.t and equatte to zero, and solving
g for annd
then lleads to the solution
s = 12, = 0.
0
System O
Optimizationn:
Substitutting the traveel time in eqquation: (6-8)), we get thee following:

Min: Z (x) =
= 10x1 + 15x2
Substitutting: x1 + x2 = 12
Substitutting the above
a formuulations takess the followiing form:
Min Z(x) = 10x1 + 15(112 – x1)
Differenttiate the above equatioon w.r.t and equatte to zero, and solving
g for annd
then lleads to the solution
s = 12, = 0.
0
Comparison of resultts:
After sollving each of
o the formuulations the results are ttabulated in Table 8.5. One
O can infe
fer
that if thhe travel tim
me is indepeendent of thhe flow, theen essentiallly there in no
n differencce
between the various assignment
a t
types.

T
Table 8.5: Comparison
C of results foor example 1

Type T
TSTT

A
AON 10 15 12 0 1
120 1220

U
UE 10 15 12 0 1
120 1220

SO 10 15 12 0 1
120 1220

Examplee 2

90
 Figu
ure 8.10
Two Linkk Problem with
w variable travel time function
f
Let’s now
w take a casee where travvel time funcctions for botth the links aare given by:
= 10 + 3x1; t2 =155 + 2x2
And totall flows from
m 1 to 2.
q12 = 12
All or Noothing Assig
gnment:
Assume which makees t1 = 10 and
a t2 = 15. Since the shhortest path is Link 1 aall
flows aree assigned too it making x1 =12 and x2 = 0.
User Equuilibrium:
Substitutting the traveel time in eqquations 1 - 5 yield to

Min : Z (x) =

= 10x1 + + 15x2 +
Substitutting: x1 + x2 = 12
Substitutting , inn the abovve formulattion will yield
y the unconstraine
u ed
formulatiion as below
w:

Min Z = 10x1 +
Z(x) + 155(12-x1) +
Differenttiate the abbove equatiion w.r.t x1 and equatee to zero, and solvinng for x2 annd
then x2 lleads to the solution
s x1 = 5.8, x2 = 6.2.
System O
Optimizationn:
Substitutting the traveel time in eqquation: (6-8)), we get thee following:

91
Min: Z (x) =
= 10xx1+ 3x12+ 15xx2+2x22
Substitutting: x1 + x2 = 12
Substitutting the above
a formuulations takess the followiing form:
Min Z(x) = 10x1+ 3x12+ 15(12-x1) +2(12-x1)2
Differenttiate the abbove equatioon w.r.t zerro, and solvving for x1 and then x2 leads to thhe
solution x1* = 5.3, x2* = 6.7, and z(x*) = 327.55.
Comparison of resultts:
After sollving each of
o the formuulations the results are ttabulated in Table 8.6. One
O can infe
fer
that unliike earlier, the variouss assignmennt types shoows consideerable differrences in thhe
performaance. AON has
h obviouslyy the worst solution
s andd SO has the best.

ble 8.6: Com

Tab mparison off results for example 2

Tyype TSTT

AO
ON 10 1
15 12 0 4677.44 552

UE
E 27.4 27.4
2 5.8 6.2 2399.0 328.8

SO
O 30.1 25.6
2 5.3 6.7 3277.5 327.5

Limitatioon of conveentional assiignment moodels:

The speccific limitatioons of the assignment moodels are higghlighted below.
Most of tthe cost funcctions, such as the BPR function,
f do not take intoo considerattion emissionn-
related faactors.
Interactioons betweenn links are nnot considereed; the traveel time on onne link is in
ndependent oof
the volum
mes on otherr links. This is an obviouus oversimpllification. A
At intersectionns, link travel
times aree affected by volumes oon other approaches annd opposing left turns. On freewayys,
merging and weaviing conditioons can greeatly affect travel tim
mes. Queuing caused by
b

92
bottlenecks on other links can also be a factor. Queues build as volumes approach the
bottleneck
Although some software packages allow node-based capacities, delays, or performance
functions which allows for better modelling of intersection dynamics. However, many of the
problems described above cannot be eliminated through network solutions. Some of these
issues can be addressed by considering the effects of flows on other links and the delays at a
junction, on the link under investigation.
ERRORS AND VALIDATION:
In contrast with the other three steps- trip generation, trip distribution, and model split- there
is no standard calibration procedure to ensure that the assignment stage reproduces closely
matched observations. Since the assignment of trips to the network is the final output of the
modelling process, assignment provides an ideal opportunity to conduct a number of error
checks. Many of these checks focus on the highway and transit assignment results compared
to observed values. Some of the more important aspects to consider as part of validation and
error checking are:
 The plotted trees will show if network building follows logical routes.
 VMT is an important indicator of adequate model calibration since it addresses all
four steps in the travel demand modeling process. The observed VMT is obtained
from a traffic counting program and compared with the modelled values. The
allowable difference between the two values should be 5% or less for regional
models.
 For the base year, compare the external station volumes with the actual counts. For
the forecast year, compare the volumes to the base year counts, and evaluate the
reasonableness of the growth.
 Compared counted and modelled screen line volumes these differences should be in
the range of plus or minus 15%.
 Compare assigned volumes to ground counts for groups of links. Assigned volumes
should reasonably compare to ground counts, recognizing that both assigned volumes
and ground counts are estimates.

93