Professional Documents
Culture Documents
UTSP-Part IV (Compatibility Mode)
UTSP-Part IV (Compatibility Mode)
Dr. Bhargab
g Maitra
Department of Civil Engineering
Indian Institute of Technology Kharagpur
India
Email: bhargab@civil.iitkgp.ernet.in
Indian Institute of Technology, Kharagpur
General
• Traffic Assignment
g is the process
p of allocating
g
given set of trip interchanges to the specified
transportation system or routes
• Traffic assignment requires a complete
description of proposed or existing transportation
system and a matrix of inter zonal trip movements
• The major aims of traffic assignment procedures
are to:
9Estimate the volume of traffic on the links of
the network and possibly the turning
movements at intersections
Indian Institute of Technology, Kharagpur
9Furnish estimates of travel costs between trip
origins and destinations for use in trip
di t ib ti
distribution
9Obtain aggregate network measures, e.g. total
vehicular flows, total distance covered by the
vehicle, total system travel time
9Estimate zone-to-zone travel costs (times) for a
given level of demand
9Obtain reasonable link flows and to identify
heavily congested links
9Estimate the routes used between each origin-
destination (O-D) pair
Indian Institute of Technology, Kharagpur
9Estimate zone-to-zone travel costs (times) for a
given level of demand
9Obtain reasonable link flows and to identify
heavily congested links
9Estimate the routes used between each origin-
destination (O-D) pair
9Analyze which O-D pairs that uses a particular
link or path
9Obtain turning movements for the design of
future junctions
Indian Institute of Technology, Kharagpur
• The
Th minimum
i i path
th computed
t d
prior to the trip assignment
will not be the minimum after
the trips are assigned
Indian Institute of Technology, Kharagpur
• A number of iterative procedures are available to
converge this difference
• The relation between the link flow and link
impedance is called the link cost function
• An example of link cost function
a 3 5 7 f
4 1
c e
6
• So many ways, 5 links can be selected to ensure
connectivity of all the 6 nodes
• But, for the minimum spanning tree (shown by
yellow lines)
lines), the total cost is minimum
Indian Institute of Technology, Kharagpur
Step 1 4
b d
1 2
a 3 5 7 f
4 1
c e
6
Indian Institute of Technology, Kharagpur
Step 2
4
b d
1 2
a 3 5 7 f
4 1
c e
6
Indian Institute of Technology, Kharagpur
Step 3
4
b d
1 2
a 3 5 7 f
4 1
c e
6
Indian Institute of Technology, Kharagpur
Step 4
4
b d
1 2
a 3 5 7 f
4 1
c e
6
Indian Institute of Technology, Kharagpur
Step 5
4
b d
1 2
a 3 5 7 f
4 1
c e
6
Indian Institute of Technology, Kharagpur
Prim’s Algorithm
• Start at any vertex in a graph (vertex A, for
example), and find the least cost vertex (vertex B,
for example) connected to the start vertex.
• Now, from either 'A' or 'B', find the next least
costly vertex connection, without creating a cycle
(
(vertex
t C,C ffor example).
l )
• Now, from either 'A', 'B', or 'C', find the next least
costly vertex connection, without creating a cycle,
and so on
• Eventually, all the vertices will be connected,
without any cycles, and an MST will be the result
Indian Institute of Technology, Kharagpur
Example
Step 1 4
b d
1 2
a 3 5 7 f
4 1
c e
6
Indian Institute of Technology, Kharagpur
Step 2
4
b d
1 2
a 3 5 7 f
4 1
c e
6
Indian Institute of Technology, Kharagpur
Step 3
4
b d
1 2
a 3 5 7 f
4 1
c e
6
Indian Institute of Technology, Kharagpur
Step 4
4
b d
1 2
a 3 5 7 f
4 1
c e
6
Indian Institute of Technology, Kharagpur
Step 5
4
b d
1 2
a 3 5 7 f
4 1
c e
6
Indian Institute of Technology, Kharagpur
Shortest path
• Findingg a path
p between two vertices (or
( nodes))
such that the sum of the weights of its constituent
edges is minimized
• Formally, given a weighted graph (that is, a set V
of vertices, a set E of edges, and a real-valued
weight
i ht function
f ti f : E → R),
R) andd one element
l t v off V,
V
find a path P from v to a v' of V so that
– Σf(p),
Σf(p) p P, P is minimal among all paths
connecting v to v'
Indian Institute of Technology, Kharagpur
Dijkstra's algorithm
• It solves the single-pair,
g p , single-source,
g , and
single-destination shortest path problems
Indian Institute of Technology, Kharagpur
Algorithm
• Let's call the node we are starting
g as initial node
and from which Y be the distance to an other
node
• Assign some initial distance values and try to
improve them step-by-step and assign to every
node a distance value; Steps
S are
9Set it to zero for initial node and to infinity for
all other nodes and mark all nodes as unvisited
9Set initial node as current
9For current node, consider all its unvisited
neighbours and calculate their distance
Indian Institute of Technology, Kharagpur
• For example, if current node (A) has distance of 6,
and an edge connecting it with another node (B) is
2 the distance to B through A will be 6+2=8
2,
• If this distance is less than the previously
recorded distance (infinity in the beginning
beginning, zero
for the initial node), overwrite the distance
• When we are done considering all neighbours of
the current node, mark it as visited
• A visited node will not be checked ever again; its
distance recorded now is final and minimal
• Set the unvisited node with the smallest distance
(from the initial node) as the next "current node"
and continue from step 3
Indian Institute of Technology, Kharagpur
Example
9 5
6 6
2 11
3 4
14
9
10 15
1
7 2
14 9 5 6
6 22
2 (7+10)=17>9
11
3 4
14
9
10 15
0
1 7
7 2
Indian Institute of Technology, Kharagpur
• Now node 3 become the current node from which
the distance of node 4 is (9+11)= 20<22 and node
6 is (9+2)= 11<14
• The distance of node 6 from node 1 is more than
the distance from node 33, so
(9+2)=11<14
9 5 6
6
2 9 11
(9+11)=20<22
3 4
14
9
10 15
0
1 7
7 2
Indian Institute of Technology, Kharagpur
• Now node 6 become the current node from which
the distance of node 5 is (11+9)= 20
• And distance from
f node 5 to node 4 is ((20+6)=
) 26
> 20, so
(11+9)=20
11
9 5 6
6
2 9 11
20
3 4
14
9
10 15
0
1 7
7 2
Indian Institute of Technology, Kharagpur
Floyd’s Algorithm
• In Floyd’s
oyd s method, et od, the
t e nodes
odes are
a e numbered
u be ed as 1,,
2, . . . . . . . . , n
• The distances found at any y stage
g of the algorithm,
g ,
say when the nodes 1 to k may be used as
intermediate nodes, are stored in an n by n matrix,
Dk, with elements dk[i, j]
• The arc lengths of the network will form the matrix
D0 and d the
th final
fi l matrix
t i will
ill be
b Dn
• The algorithm is formulated as follows
9Step 0: Create the n by n matrix D0 with
elements
Indian Institute of Technology, Kharagpur
d0[i, j] = d[i, j] (the length of arc (i, j), if this exists)
= 0 (if i = j)
= ∞ (if arc (i, j) does not exist)
P[i j] = i
P[i,
Set k = 0
9Step 1: Define the matrix Dk-1[i,j] with elements
dk-1[i, j] = min [dk[i, j], dk[i, k+1] + dk[k+1, j]]
If the later value is chosen, then change P[i, j] to
P[k+1, j]
9Step 2: Increase k by one
If k=n stop, otherwise return to step 1
Indian Institute of Technology, Kharagpur
9Step 2: Increase k by one
If k=n stop, otherwise return to step 1
• When the algorithm stops it will have found the
shortest path length between every pair of nodes
in the network
Indian Institute of Technology, Kharagpur
Example
• Determine
ete e tthe
esshortest
o test pat
paths
s bet
between
ee aall pa
pairs
soof
nodes on transportation network shown in Figure
g
• Branch lengths are shown on the figure
g
Indian Institute of Technology, Kharagpur
• Starting matrix Do is as
follows:
• All elements along the main
diagonal of matrix Do equal
zero since by definition dij0 = 0
for i = j
• We
W note element
l d1,02 off matrix
i Do
• This element equals 8 since the length of the
branch connecting nodes 1 and 2 is 8
• Element d3,01 equals infinity since the network has
no
o branch
b a c whichc issooriented
e ted from
o node
ode 3 to node
ode 1
• Element d5,01 of matrix Do equals infinity as well
since there is direct branch linking nodes 5 and 1
Indian Institute of Technology, Kharagpur
Figure 8: Areas under travel-time versus flow plots under various flow distributions
Indian Institute of Technology, Kharagpur
• Another interesting feature is that at equilibrium
flow, the sum of the areas under the travel time
versus flow
fl plots
l t isi the
th least
l t
• This feature can be easily seen by visually
inspecting the sum of the areas shown in Figure 8
• Figure 8 (a) corresponds to the equilibrium flow,
and the sum of the areas is equal to area ACDE +
area ABFG
• Figure 8 (b),
(b) corresponds to an instance of non-
equilibrium flow conditions with Route 2 carrying
all the flow ((i.e.,, 110);
); in this case the relevant sum
of the areas is simply equal to area HIJK
Indian Institute of Technology, Kharagpur
• Figure 8(c) represents another instance of non-
equilibrium flow where Route 2 carries a flow of 90
and
d the
th restt 20 are carried
i d by
b Route
R t 1;1 ini this
thi case
the relevant sum of areas is area LMNO + area
LPQR
• Finally, Figure 8 (d) represents another instance of
non-equilibrium
non equilibrium flow conditions with Routes 1, 2,
and 3 carrying flows of 30, 70, and 10,
respectively; the relevant sum of the areas in this
case is area TUVW + area TXYZ + area Tαβγ. Note
that the sum of the areas is the least in
Indian Institute of Technology, Kharagpur
• The solution to the above equilibrium conditions
given by the solution of an equivalent nonlinear
mathematical
th ti l optimization
ti i ti program
• Where k is the path, xa
equilibrium flows in link a, a
ta travel time on link a, frsk
flow on path k connecting
O D pair
O-D i r-s, qrs trip
t i rate
t
between r and s, and δrsa,k
is a constraint and is
given by
Indian Institute of Technology, Kharagpur
Example
Subject to
Link 2
O D
Li k 3
Link
• t1 = 10 [1+0
[1+0.15(x
15(x1/2)4 ] time units
• t2 = 20 [1+0.15(x2/4)4 ] time units
• t3 = 25 [[1+0.15(x
( 3/3))4 ] time units
• x1 + x2 + x3 = 10 flow units
Indian Institute of Technology, Kharagpur
Iteration Algorithmic step link1 Link 2 link3
(increment)
1 Update t11=10
10 t21=20
20 t31=25
25
Increment loading w11=2.5 w21=0 w31=0
Summation x11=2.5 x21=0 x31=0
2 Update
U d t t12=14
14 t22=20
20 t32=25
25
Increment loading w12=2.5 w22=0 w32=0
Summation x12=50 x22=0 x32=0
3 Update t13=69 t23=20 t33=25
Increment loading w13=0 w23=2.5 w33=0
Summation
Su at o x13=5.0
=5 0 x23=2.5
=2 5 x33=0
4 Update t14=69 t24=20.5 t34=25
Increment loading w14=0 w24=2.5 w34=0
S
Summation
ti x14=5.0 x24=5.0 x34=0
Travel time at convergence t1* =69 t2* =27.3 t3* =25
Indian Institute of Technology, Kharagpur
Capacity restraint assignment
• Capacity restraint assignment attempts to
approximate an equilibrium solution by iterating
between all-or nothing traffic loadings and
recalculating
l l ti link
li k travel
t l times
ti based
b d on a
congestion function that reflects link capacity
• Unfortunately,
Unfortunately some times
times, this method does not
converge and the flow flip-flops between some
links whereas the other links does not get loaded
at all
• So to remedy this situation, we need to use
modified capacity restraint algorithm
Indian Institute of Technology, Kharagpur
• The algorithm can be summarised as follows
9 Step 0: Initialization: Perform all-or-nothing
assignment based on ta0 = ta(0), a. Obtain a set
of link flows {xa0}. Set iteration counter n = 1
9 Step
St 1: 1 U
Update:
d t S Sett tan = ta(x
( an-11),
) a.
9 Step 2: Network loading: Assign all trips to the
network
t k using
i all-or-nothing
ll thi based
b d on travel
t l time
ti
{tan}. This yield a set of link flows {xan}
9 Step 3: Convergence test: If maxa { an an‐1 } ≤ k k,
stop.
• Otherwise,
Otherwise set n := n+1 and go to step 1
Indian Institute of Technology, Kharagpur
Example
Link 1
Link 2
O D
Li k 3
Link
• t1 = 10 [1+0
[1+0.15(x
15(x1/2)4 ] time units
• t2 = 20 [1+0.15(x2/4)4 ] time units
• t3 = 25 [[1+0.15(x
( 3/3))4 ] time units
• x1 + x2 + x3 = 10 flow units
Indian Institute of Technology, Kharagpur