De Transportation,

liv
er
ed
to
Assignment, &
yo
Transshipment ub
y
Problems
tbs
re
so
urce
s.w
ix.
Soniaco Rebai
m/
tbs
Tunisian Business School
re
so
ur
ce
s
Introduction
D eli
ve
re
In this chapter we discuss three special types of LP
 dt
oy
problems ou
by
 Transportation problems tbs
re
 Assignment problems ou
s
rce
 Transshipment problems s.w
ix.
These problems are special cases co of an LP problem.
 m/
tbs
While each of these problems can beresolved by the
so
ur
simplex algorithm, there are specialized algorithms ce for
s
each type of problem that are much more efficient than
the simplex algorithm. 2
Transportation
D Problems
eli
ve
re
dt
oy
Formulating Transportation
ou Problems
by
Example 1 tbs
re
so
A company has three electricurpower
ce plants that supply
s.w
the power needs of four cities. The ix. supply of each
co
m/
plant, the demand of each city, andtb the cost of
sre
transportation of a unit from each plant to each so city are
ur
ce
given in the following table. Formulate an LP s to

minimize the cost of meeting each city’s demand. 3

Transportation
D Problems
eli
ve
re
dt
oy Supply
City 1 ou City 2 City 3 City 4
by (Million kwh)
tbs
Plant 1 8 6 re 10 9 35
so
urce
Plant 2 9 12 s13
.w 7 50
ix.
co
Plant 3 14 9 16 m/ 5 40
tbs
re
Demand so
45 20 30 30 ur
(Million kwh) ce
s

4
Transportation
D Problems
eli
ve
re
dt
Xij : number of million
o y kwh produced at plant i and sent to city j.
ou
Min Total cost of shipping by power
tbs
re
Min Z = 8X11 + 6X12 + 10X13 +so9X14 + 9X21 + 12X22 + 13X23 +
ur
ce
7X24 + 14X31 + 9X32 + 16X33 s.w+ 5X34
ix.
co
Supply constraints m/
tbs
X11 + X12 + X13 + X14 ≤ 35 re
so
ur
X21 + X22 + X23 + X24 ≤ 50 ce
s
X31 + X32 + X33 + X34 ≤ 40
5
Transportation
D Problems
eli
ve
re
dt
Demand constraints
oy
ou
X11 + X21 + bXy31 ≥ 45
tbs
re
X12 + X22 + X32 ≥ 20 so
u
rce
X13 + X23 + X33 ≥ 30 s.w
ix.
co
X14 + X24 + X34 ≥ 30 m/
tbs
re
All the Xij’s must be nonnegative. so
ur
ce
Xij ≥ 0 (i = 1, 2, 3; j = 1, 2, 3, 4) s

6
Transportation
D Problems
eli
ve
re
dt
General Description
o yo
ub
In general a transportation problem is concerned with
yt
bs
re a single product from several
specifying how to distribute
so
ur
sources to several destinations ceat minimum cost.
s.w
1. A set of m supply points from which ix. a good is shipped.
co
m/
Each can supply at most si units. i = 1,tb2, …, m.
sre
2. A set of n demand points to which the good so is shipped.
ur
ce
Each must receive at least dj units. j = 1, 2, …,sn.
3. Each unit produced at i and shipped to j incurs a cost cij.
7
Transportation
D Problems
eli
ve
re
dt
oy
ou
m Sources n Destinations
by C11: X11
a1 tbs b1
re
so Demand
Supply
a2
u rce
b2
required by
Capacity s.w destination n
from source m ix.
am co bn
Cmn: Xmn m/
tbs
re Number of
Cost to send so
a unit from m ur units to send
to n cfrom n to m
es

8
Transportation
D Problems
eli
ve
re
dt
oy
ou
by
tbs
re
so
urce
s.w
ix.
co
m/
tbs
re
so
ur
ce
s

9
Transportation
D Problems
eli
ve
re
dt
Note that the previous transportation problem is balanced.
oy
ou
by
City 1 tbs 2
City City 3 City 4
Supply
re (Million kwh)
so
ur
Plant 1 8 6 ce 10 9 35
s.w
ix.
Plant 2 9 12 13 com 7 50
/tb
sre
Plant 3 14 9 16 5 so 40
ur
ce
Demand s
45 20 30 30 125
(Million kwh)
10
Transportation
D Problems
eli
ve
re
dt
oy
ou
by
tbs
re
so
urce
s.w
ix.
co
m/
tbs
re
so
ur
ce
s

11
Transportation
D Problems
eli
ve
re
dt
Note that the previous transportation problem is balanced.
oy
ou
by
City 1 tbs 2
City City 3 City 4
Supply
re (Million kwh)
so
ur
Plant 1 8 6 ce 10 9 35
s.w
ix.
Plant 2 9 12 13 com 7 50
/tb
sre
Plant 3 14 9 16 5 so 40
ur
ce
Demand s
45 20 30 30 125
(Million kwh)
12
Transportation
D Problems
eli
ve
re
d
 Note that it tois relatively easier to solve a balanced
yo
u b than a unbalanced one.
transportation problemyt
bs
re
 For this reason, it is desirable to formulate a transportation
so
ur
problem as a balanced one. ces
.w
ix.
Balancing a Transportation Problem co
m/
tbs
 If total supply exceeds total demand, we re can balance a
so
transportation problem by creating a dummy udemand rce point
s
having a demand equal to the amount of excess supply.
13
Transportation
D Problems
eli
ve
re
dt
 Since shipments o y to the dummy demand point are not real,
ou
by
they are assigned a cost tbs of zero. Shipments to the dummy
re
so
ur
point indicate unused supply capacity.ce
s.w
ix.
co
Example 2 m/
tbs
re
so
Suppose that in the previous example, the demand ur
ce for city 1
s
were reduced to 40 million kwh.
14
Transportation
D Problems
eli
ve
re
dt
oy
ou
City 1 bCity
Supply
y 2 tbs
City 3 City 4
(Million kwh)
re
Plant 1 8 6 so 10 9 35
ur
ce
Plant 2 9 12 13 s.w 7 50 125
ix.
Plant 3 14 9 16 5co 40
m/
tbs
Demand re
(Million kwh)
40 20 30 30 so
ur
ce
s
120
15
Transportation
D Problems
eli
ve
re
To balance thisdproblem
to we would add a dummy demand point
y
(Dummy City 5) withoua demand of 125-120 = 5 million kwh.
by
tbs
re
so 3 City 4 Dummy Supply
City 1 City 2 City ur
ce City 5 (Million kwh)
Plant 1 8 6 10
s.w9 0 35
ix.
Plant 2 9 12 13 7 com 0 50
/tb
Plant 3 14 9 16 5 s0re 40
so
Demand ur
(Million kwh)
40 20 30 30 5 ce 125
s

16
Transportation
D Problems
eli
ve
re
 If total supplyd tis
o yless than total demand, the problem has no
ou
feasible solution. For by example, if plant 1 had only 30 million
tbs
kwh of capacity, then a rtotal es of only 120 million kwh would
ou
be available, which would be rce insufficient to meet the total
s.w
ix.
demand of 125 million kwh. co
m/
 It is sometimes desirable to allow the tpossibility bs
re of leaving
so
some demand unmet. In such situation a penalty ur is often
ce
s
associated with unmet demand.
17
Transportation
D Problems
eli
ve
re
Example 3 d to
yo
ub
Two reservoirs are available to supply the water needs of three
y tbs supply up to 50 million gallons of
cities. Each reservoir can
re
water per day. Each city wouldso like to receive 40 million gallons
ur
ce
per day. The penalties are $20 at scity
.w 1, $22 at city 2, and $23
ix.
at city 3, the transport costs are given co below. Formulate a
m/
balanced transportation problem that can tbe bs used to minimize
re
the sum of shortage and transport costs. so
ur
ce
City 1 City 2 City 3 s
Reservoir 1 7 8 10
Reservoir 2 9 7 8 18
 Transportation
D Problems
eli
ve
re
Transportation
d Tableau
to
It is not necessary yo to explicitly build the LP model of a
ub
transportation problem.y Instead, we usually represent the
tbs
transportation problem using re a transportation tableau
so
ur
Destinations
Dest. 1 Dest. 2
c e
Dest. s 3 Dest. 4 Supply
Number of available
Decision
Variable C11 C12
. w
C13 ix C14
units to ship from
source 1 (i)
Point 1 X11 X12 X13 X.14
co s 1
mC/
Sources

Point 2 X21
C21
X22
C22
X23
C23
X24 tbs s2
24
Cost for sending
re
s o
a unit from 2 (i)
s3 ur
C31 C32 C33 C34 to 4 (j)
Point 3 X31 X32 X33 X34
ce
Demand d1 d2 d3 d4 s
Number of units
required in
destination 1 (j) 19
Transportation
D Problems
eli
ve
re
Example 4 d to
yo
ub
A company must determine for a given product the quantity to
y
manufacture during eachtbsof the next four quarters. Demand is
re
as follows: 40, 60, 75 and so 25 units respectively. At the
ur
ce
s.w
beginning of the first quarter, the company has an inventory of
ix.
10 units. During each quarter, the company co can produce up to
m/
40 units at a cost $400 per unit. By having tbs employees work
re
overtime during a quarter, the company can produce so additional
ur
ce quarter, a
units at a cost of $450 per unit. At the end of each s
holding cost of $20 per unit is incurred. Formulate a balanced
transportation problem to minimize the total cost. 20
Transportation
D Problems
eli
ve
re
Solution dt
oy
Supply points: ou
by
Point 1: Initial inventorybst s = 10
re 1
so (RT) production s = 40
Point 2: Quarter 1 regular-time
ur 2
ce
Point 3: Quarter 1 overtime (OT) sproduction
.w s3 = 150
ix.
Point 4: Quarter 2 RT production co s4 = 40
m/
Point 5: Quarter 2 OT production tbs s5 = 150
re
Point 6: Quarter 3 RT production so s6 = 40
ur
Point 7: Quarter 3 OT production s7ce=s150
Point 8: Quarter 4 RT production s8 = 40
Point 9: Quarter 4 OT production s9 = 150 21
Transportation
D Problems
eli
ve
re
dt
Since total demand o y = 200 units at most 200–10 = 190 units
ou
by
will be produced duringtbanysre quarter.
so
ur
ce
Since 40 units must be produced s.w on RT before any units
ix.
co
are produced on OT, OT production during m/ any quarter will
tbs
re
so
never exceed 190-40 = 150 units. ur
ce
s

22
Transportation
D Problems
eli
ve
re
Demand points:dt
oy
Point 1: Quartero1u demand d1 = 40
by
tbs
Point 2: Quarter 2 demand d2 = 60
re
Point 3: Quarter 3 demand so
ur d3 = 75
ce
Point 4: Quarter 4 demand s.dw4 = 25
ix.
Since total supply = 770 and total demand co
m/ = 200 units, we
tbs
must add a dummy demand point of 770 – r200 es = 570 units
ou
to balance the problem. The cost of shipping raceunit s from
any supply point to the dummy demand point is 0.
23
Point 5: Dummy demand d5 = 770-200 = 570
Transportation
D Problems
eli
ve
re
C13 correspond d t to a shipment from quarter 1 RT to quarter
oy
ou
3 demand. A such shipmentby means producing 1 unit on RT
tbs
during quarter 1 that is used re to meet 1 unit of quarter 3’s
so
ur
demand. Thus C13 = 400 + 2*20ce=s 440
.w
To ensure that no units are used toixmeet .co demand during a
m/
quarter prior to their production, a cost tbs of M (a large
re
so
positive number) is assigned to any cell that corresponds ur to
ce
s
an earlier quarter. For instance, C41 = M.
The transportation tableau is as follows: 24
Transportation
D Problems
eli
ve
re
dt
Quarter 1 Quarter 2 Quarter 3 Quarter 4 Quarter 4 Supply
o0 y 20 40 60 0
Point 1 ou 10
by420
Point 2
400
tbs 440 460 0
40
re 490
450 470 so 510 0
Point 3
ur 150

420 e
c
Point 4
M 400
s.w440 0
40
ix.
Point 5
M 450 470 490
co 0
150
m/
Point 6
M M 400 420 tb0 s 40
r
0 es
Point 7
M M 450 470
ou
150
rce
Point 8
M M M 400 0
40 s
M M M 450 0
Point 9 150

Demand 40 60 75 25 570 770 25

Transportation
D Problems
eli
ve
re
dt
Resolution of oay Transportation Problem
ou
by
tbs is solved in two phases:
The transportation problem
re
so
ur
1. Determination of an initial basicce feasible solution
s.w
ix.
2. Finding an optimal solution through co the sequential
m/
tbs
re
improvement of the initial feasible solution sou
rce
s

26
Transportation
D Problems
eli
ve
re
dt
o yFeasible Solutions
Finding Basic o ub
There are many waysy tbto find a feasible solution. We will
sre
discuss three methods thatscan
ou be used to find a BFS for a
rce
balanced transportation problem.s.wi
x.c
om
 Northwest Corner Rule (NWCM) /tb
sre
 Minimum Cost Method (MCM)
so
ur
ce
s
 Vogel’s Method (VM)

27
Transportation
D Problems
eli
ve
re
dt
Northwest Corner Rule (NWC)
o yo
 We begin in theu bNorthwest
yt (upper-left) corner of the
bs
matrix and assign as remuch as we can (considering
so
supply and demand) and updateur remaining supply and
ce
s.w
demand. ix.
co
 We move either down or to the right m/ (depending on
tbs
whether supply or demand has been redepleted). so We
ur
again assign as much as possible and continue ce to the
s
Southeast (lower-right).
28
Transportation
D Problems
eli
ve
re
We illustrate the d t use of the NWCM by finding a BFS for the
oy
ou
first balanced transportation by problem example:
tbs
r
City 2 es City 3
City 1
ou City 4 Supply

Plant 1
8 6 rce10 9
35 X
35 s.w
9 12 13 ix
Plant 2 10 20 20 .co 7 50 40 20 X
m/
14 9 16 5tb
Plant 3 10 30 sre40 30
so
Demand 45 20 30 30 ur
ce
10 X 10 s
X X

29
Transportation
D Problems
eli
ve
re
We have nowdobtainedto the BFS
yo
x11=35, x21=10, x22=20, u b x23=20, x33=10 et x34=30.
yt
TC = 35*8 +10*9 + 20*12bs+r 20*13 + 10*16 +30*5 = 1180
es
ou
rce
City 1 City 2 City 3s
.w City 4 Supply
8 6 10 ix 9
Plant 1 35 . co 35
m/7
Plant 2 10
9
20
12
20
13 tbs 50
re
so
40 ur
14 9 16 30 5
Plant 3 10
ce
Demand 45 20 30 30 s

30
Transportation
D Problems
eli
ve
re
d t Method (MCM)
Minimum Costo yo
 Note that the NWCM u b is simple but ineffective. Indeed,
yt
this method do not look bs at the costs! Thus it can yield an
re
so high shipping cost. Then
initial BFS that has a very ur
ce
determining an optimal solution s.w may require many
ix.
iterations. co
m/
 The MCM examines the costs to build tan bs initial solution.
re
so
The cell with the lowest cost is chosen, and we ur assign as
ce
many units as possible to the cell (considerings supply
and demand).
31
Transportation
D Problems
eli
ve
re
 We then dreduce
to supplies and demands by the
yo
u b out ineligible cells (those in rows
assignment and mark yt
bs
or columns where theressupply or demand has been
ou
rce
depleted). We repeat this processs.w until all supplies and
ix.
demands are depleted. co
m/
t
 We move either down or to the rightbsr(depending
es on
ou
whether supply or demand has been depleted). rce We
s
again assign as much as possible and continue to the
Southeast (lower-right). 32
Transportation
D Problems
eli
ve
re
Vogel’s Method
dt
o
(VM)
yo
 This method recognizes
ub that it may be wise to make a
small sacrifice for ay tbigger
bs gain. It computes a penalty
for each row and column re if the lowest cost cell is not
so
selected. That is, it figures uout
rce what it would cost to take
the second best cost. s.w
ix.
 The penalty is the cost differencecobetween m/ the lowest
cost cell and next lowest cost value tin bs each row and
re
column. so
ur
 We select the cell associated with the largest cpenalty es to
assign units to, and proceed essentially like the MCM.
We will have to recalculate some of the penalties on
33
each iteration.
Transportation
D Problems
eli
ve
re
 dt
Step 1: Determine the difference between the lowest
oy
two cells in all rowsou and columns, including dummies.
by
tbs
 Step 2: Identify the row re or column with the largest
so
ur
difference. ce
s.w
ix.
 Step 3: Allocate as much as possible co to the lowest-
m/
tbs
cost cell in the row or column with re the highest
so
ur
difference. ce
s
 Step 4: Stop the process if all row and column
34
requirements are met. If not, go to the next step.
Transportation
D Problems
eli
ve
re
dt
oy
 Step 5: Recalculate the differences between the two
ou
lowest cells remaining by in all rows and columns. Any
tbs
re
row and column with zero so supply or demand should
ur
ce
not be used in calculating further s.w differences. Then go
ix.
to Step 2. co
m/
tbs
re
so
ur
ce
s

35
Transportation
D Problems
eli
ve
re
Tutorial dt
oy
Find a starting BFSouusing
by NWCM, MCM and VM for each of
the following models: tbs
re
(a) so(b) (c)
ur
ce
6s 7
0 2 1 6 1 2
.w 5 1 8 12
ix.
2 1 5 7 0 4 2 12
co 2 4 0 14
2 4 3 7 3 1 5 11 m/
tbs 3 6 7 4
5 5 10 10 10 10 re9 10 11
so
ur
ce
s

36
Transportation
D Problems
eli
ve
re
dt
Before discussing o y further resolution steps, let us introduce
ou
the following definition:by
tbs
Definition re
so
ur
An ordered sequence of at least ce four different cells is called
s.w
a loop if ix.
co
m/
1. Any two consecutive cells lie in eithertbsthe same row or
re
same column so
ur
ce
2. No three consecutive cells lie in the sames row or
column
37
Transportation
D Problems
eli
ve
re
d
3. The last cellto in the sequence has a row or column in
yo
common with the u bfirst cell in the sequence.
yt
bs
Illustration re
so
ur
ce
s.w
ix.
co
m/
tbs
re
so
ur
ce
s

38