LINEAR PROGRAMMING TECHNIQUES - I

M.L. KANSAL
NEEPCO Chair Professor,
Department of Water Resources Development and Management,
Indian Institute of Technology Roorkee, ROORKEE 247 667

2.1 INTRODUCTION

Conventional engineering approaches following sectoral problem definitions and solutions

appeared to be inadequate to tackle complex problems inherent in water resources development
and management. Since the early sixties, serious attempts have been made to redefine the water
resources development and management problems within the framework of system analytical
concepts.
Water resource problems deal with both design and analysis. Analysis is concerned with
determining the behavior of an existing system or a trial system that is being designed. In many
cases determination of the behavior of the system is determining operation of the system or the
response of a system under specific inputs. The design problem is to determine the sizes of
components of the system. Usually design is estimated and is then analyzed to see if it performs
according to the specifications.
Irrespective of the choice of a conventional or system analytical approach, towards a given water
resources problem, the solution can be visualized as a sequence of decision problems.
Obviously, the decision environment should be simplified to distinguish between the important
and the secondary features. One can look at the situation as a decision making problem whose
solution requires identifying the following three major components:
1. What are the decision alternatives?
2. What are the constraints/ restrictions under which the decision is to made?
3. What is an appropriate objective criterion for evaluating the alternative?
Consequently, a conceptual model will represent the reality and shall require answer to these
three components in decision-making problems. In general, various steps for solving the
decision making problems in system engineering are:
 Identify and formulate the problem
 Construct a mathematical model to represent the system under study
 Derive a solution from the model
 Test the model and the solution derived from it.
 Establish control over the solution
 Put the solution to work (or implement the solution)
Thus, once the problem is identified, the next crucial step is to formulate the problem
mathematically. For mathematical modeling, the definition of the alternatives or the decision
variables of the problem are identified. Here, decision variables represent some physical or
economic quantities that are of interest to the decision maker. These are used to construct the
objective function and the constraints of the model. The decision variables usually have lower
and upper bounds. Only rarely are the variables in an optimization problem permitted to take on
any value from minus infinity to plus infinity. The model should have a well-defined objective
of maximization of certain benefits or minimization of cost involved. A general form of an
optimization problem is
Optimize f(x)
Subject to constraints
g(x) = 0
where, g(x) is a vector of m equations.
Every optimization problem has two essential parts : the objective function and the set of
constraints. The objective function describes the performance criteria of the system. Constraints
describe the system or process that is being designed or analysed and can be of two forms :
equality and inequality.
A solution of the model is feasible if it satisfies all the constraints. It is optimal if, in addition to
being feasible, it yields the best (maximum or minimum) value of the objective function. One
should notice that the system models are designed to optimize a specific objective criterion
subject to a set of constraints, the quality of the resulting solution depends on the accuracy of the
model in representing the real system.
Depending upon the nature of the objective function and the constraints, an optimization problem
can be classified as : (i) linear vs. nonlinear; (ii) deterministic vs. probabilistic; (iii) static vs.
dynamic; (iv) continuous vs. discrete; and (v) lumped parameters vs. distributed parameter.
In linear programming (LP) all of the mathematical expressions for the objective function and
the constraints are linear. Linear programming is by far the most widely used method of
constrained optimization. The largest optimization problems in the world are LPs having million
of variables and hundreds of thousands of constraints. With recent advances in both solution
algorithms and computer power, these large problems can be solved in practical amounts of time.
Other techniques include integer programming (in which variables assume integer values),
dynamic programming (in which original model can be decomposed into smaller sub problems),
network programming (in which the problem can be modeled as a network), and nonlinear
programming (in which the functions of the model are non-linear). A peculiarity of the most of
these programming techniques is that solutions are not generally obtained in a (formula-like)
closed form. Instead, algorithms determine them. An algorithm provides fixed computational
rules that are applied repetitively to the problem, with each repetition (called iteration) moving
the solution closer to the optimum. Some mathematical models may be so complex that it is
impossible to solve them by any optimization algorithms. In such cases, it may be necessary to
abandon the search for optimal solution and simply seek a good solution using heuristics or
thumb rules’ approach. Queuing or simulation models fall under such category. The queuing
models are purely mathematical and hence subject to certain assumptions, whereas, simulation is
flexible and can be used to analyze practically any queuing situation.
This chapter deals with the systems of linear objective function that is subject to linear
constraints. It is time now to introduce some LP problems examples that can be modeled
mathematically.
Problem 2.1
There are 10 possible projects that are being considered as components of a river basin
development plan like reservoirs, pumping facilities, flood control, improvement & hydro
electric power plants, etc. Investment cost and net annual benefits for each project are as follows

Project Investment cost Net annual

(million Rs) benefit
(million Rs)
1 120 4
2 150 3
3 90 5
4 210 9
5 160 7
6 180 6
7 90 4
8 100 5
9 80 2
10 140 7

Due to budget constraints the total investment cost for selected projects cannot exceed Rs. 750
million. Projects 2 & 3 can be implemented if project 6 is implemented. Project 10 can be
implemented if projects 8 & 9 are implemented. Project 4 cannot be selected if project 3 is
selected. At least one but not more than 3 of the following projects must be selected i.e, 5,6,7
and/or 8. Formulate this problem as Linear Programming model.

Solution
In this problem, there are ten decision variables, i.e., X1, X2,……X10 that represent the
various projects. The variables are of binary type. i.e., these can have values either 0 or 1. Once
the decision variables are identified, next step is to identify the objective function. Here, the
objective function is
Maximize Z = 4X1+3X2+5X3+9X4+7X5+6X6+4X7+5X8+2X9+7X10
The objective function is subject to various constraints. These are :
120X1+150X2+90X3+210X4+160X5+80X6+90X7+100X8+80X9+140X10<=750
(Cost constraint)
X2 + X3-2X6<= 0 or X2 - X6<= 0 and X3- X6<= 0
2X10 - (X8 + X9)<= 0 or X10 – X8<= 0 and X10- X9<= 0
X3 + X4<= 1
X5 + X6+ X7 + X8>= 1
X5 + X6+ X7 + X8<= 3

X1 , X2,…X10 >= 0 (non negative constraints)

Problem 2.2
City 1 produces 500 tons of waste per day, and city 2 produces 450 tons of waste per day. Waste
must be incinerated at incinerator 1 or 2, and each incinerator can process up to 500 tons of
waste per day. The cost to incinerate waste is Rs. 2000/- per ton at incinerator 1 and Rs. 1500/-
per ton at 2. Incineration reduces each ton of waste to 0.2 tons of debris, which must be dumped
at one of two landfills. Each landfill can receive at most 200 tons of debris per day. It costs
Rs. 150/- per km to transport a ton of material (either debris or waste). Distances (in km)
between locations are shown below. Formulate an LP that can be used to minimize the total cost
of disposing of the waste of both cities.

Incinerator Landfill
City City
1 2 1 2
1 3 1 1 5 8
2 4 5 2 9 6

Solution
In this problem, there are two cities, two incinerators, and two land fill sites. Let the
amount of waste/ debris transported from various cities to incinerators, and incinerators to
landfill sites are represented by eight decision variables (X11, X12, X21, X22, Y11, Y12, Y21, and
Y22 ) as shown below :

X11 Y11
C1 I1 L1

X12 X21 Y12 Y21

C2 X22 I2 Y22 L2

The objective function is to minimize the total cost of transportation from the cities to
incinerators, cost of incineration, and the cost of transportation of the debris from incinerators to
the landfill sites. Mathematically, this can be represented as

Minimize Z =
450X11+150X12+600X21+750X22+2000(X11+X21)+2500(X12+X22)+750Y11+1200Y12+1350Y21+900Y22
Subject to: -
120X1+150X2+90X3+210X4+160X5+80X6+90X7+100X8+80X9+140X10<=750
(Cost constraint)
X11+X12 = 500 (Waste production constraint)
X21+X22 = 450 (Waste production constraint)
Y11+Y21 <= 200 (Landfill Capacity constraint)
Y12+Y22 <= 200 (Landfill Capacity constraint)
X11+X21 <= 500 (Incinerator capacity constraint)
 X12+X22 <= 500 (Incinerator capacity constraint)
Y11+Y12 = 0.2 (X11+X21) (Debris production constraint)
Y21+Y22 = 0.2 (X12+X22) (Debris production constraint)

X11, X12, X21, X22, Y11, Y12, Y21, Y22>= 0 (non negative constraints)

Problem 2.3
Consider a system composed of manufacturing factory and a Waste Treatment Plant of a owner.
The manufacturing plant produces finished goods that sell for unit price of Rs 500/-. However
the finished goods production cost is of Rs 300/- per unit. In the manufacturing process 3 units
of waste are generated for each unit of finished goods. In deciding about the number of goods to
be produced, the manager must also decide how much waste will be discharged without
treatment so that net benefit to the company can be maximized and the water quality
requirements of wastewater is also met. The treatment plant has a capacity of 300 units of waste
with 70 % waste removal efficiency at the treatment cost of Rs 20/- per unit of waste. There is
also an affluent tax imposed on the waste discharge at the rate of Rs 25/- per each unit of waste
discharge. The water pollution control authority has set up on upper limit of 200 units on the
amount of waste that can be discharged by any manufacturer. Formulate a linear programming
model for this problem.
Solution
X1
FACTORY FINISHED GOODS

3X1 (waste product)

3X1 – X2
WASTE TREATMENT PLANT

X2 (discharged quantity without

treatment) (0.3) (3X1 – X2)

Maximize Z = 500 X1 - [300 X1 + 20 (3 X1 - X2) + 25 {X2 + 0.3 (3 X1 - X2)}]

= 117.5 X1 - 2.5X2
Subject to 3 X1 - X2  300 ------------------------------------------------------------A
X2 + 0.3 (3 X1 - X2)  200 Or 0.7X2+0.9 X1 ≤ 200----------------B
3X1 – X2  0 ----------------------------------------------------------------C
X1  0 ------------------------------------------------------------------------D
X2  0 ------------------------------------------------------------------------E

2.2 LINEAR PROGRAMMING – GRAPHICAL METHOD

As the name suggest, the most important feature of LP problems is the presence of linearity in
the problem. The problem, thus reduces to maximizing or minimizing a linear function subject
to a number of linear inequalities/equalities (constraints) as discussed above. The technique is
used in a wide range of applications. The simplest form of LP problem is that which involves
two decision variables, as it can be solved by simple graphical technique, as interpretation of a 2-
variable graph is easy. However, for a large number of decision variable, graphical method does
not work well and there is a need for systematic solution of such problems. Simplex method can
be used in case of large number of variables and constraints. Following sections deals with such
problems.

Problem 2.4
Suppose that a farmer wants to sow two crops viz wheat and jowar for his remote land where
only ground water is available to him for irrigation. He can’t pump more than 60 units of water
in a year to avoid the problem of mining in that region. Each unit production of wheat requires 6
units of water and that of jowar requires 5 units of water, also one unit of wheat production
requires 10 units of land and one unit of jowar production requires 20 units of land, where as
total land available to him is 150 units of land. Since he is already sowing wheat in his other
land, he doesn’t want to sow more than 8 units of wheat. The profit per unit production of wheat
is Rs 500 and that for jowar is Rs 450. What should be the policy of the farmer about sowing of
the two crops so that he maximizes the profit?
Solution
Identify the problem and write in mathematical form:
In this problem, there are two decision variables, i.e.,
X1 = no of units of Wheat production
X2 = no of units of Jowar production
Objective function: - Maximize Z = 500 X1 + 450 X2
Subject to: - 6 X1 + 5 X2 <= 60 (water consumption constraints)
10 X1 + 20 X2 <= 150 (land constraints)
X1 <= 8 (upper limit of wheat production)
X1 >= 0, X2 >= 0 (non negative constraints)
General form of Linear Programming
Maximize or Minimize Z = C1 X1 + C2 X2 + ………. + Cn Xn
Subject to a11 X1 + a12 X2 + ……….. + a1n Xn = b1
a21 X1 + a22 X2 + ……….. + a2n Xn = b2
..
..
..
am1 X1 + am2 X2 + ……….. + amn Xn = bm

 a11 a12 a1n   x1   b1 

a a 22 a 2 n   x  b 
In matrix form   2 2
21

     
     
a m1 am2 a mn   x m  bm 
To derive the solution, plot the graph between X1 and X2

All the equations in the problem can be written in the form of: -
Y = mx + C (straight line)
Z = 500 X1 + 450 X2 =====> X2 = (Z / 450) – (500 X1 / 450)
6 X1 + 5 X2 = 60 =====> X2 = (60 / 5) – (6 X1 / 5)
10 X1 + 20 X2 = 150 =====> X2 = (150 / 20) – (10 X1 / 20)

Where ever the solution is feasible in nature, policy space must be convex in nature.

Effective constraints: All the constraints that affect the objective function , i.e.
6 X1 + 5 X2 = 60 and 10 X1 + 20 X2 = 150
Redundant constraint  X2  8

Coordinate of points in the policy space where the slope of boundary changes are:
1. (0,0)
2. (0, 7.5)
3. (45/7,30/7)
4. (8,12/5)
5. (8,0)
The objective function will be maximum at point (45/7, 30/7) & shall be
Z max = 500 x 45/7 + 450 x 30/7
= 5142.86
So policy of the farmer should be
X1 = 45/7 i.e. wheat production 6.429 units
X2 = 30/7 i.e. jawar production 4.286 units
Sensitivity Analysis :
Shadow Price or Reduction Cost
There is always a shadow price connected with effective constraint.
Maximize Z1 = 500 X1 + 450 X2
If 6 X1 + 5 X2  61 ------- A If 10 X1 + 20 X2 = 151
10 X1 + 20 X2  150 ------- B 6 X1 + 5 X2 = 60
X1  8 ------- C from these equation
Considering inequality A, B, C 14 X1 = 89 ====> X1 = 6.351
X1 = 6 75 X2 = 4.371
X2 = 4 17 Z151 = 500 x 6.36 + 450 x 4.37
= 5145.714
Z1 = 5221 73 Difference (Z151) - (Z150) = 2.86
Difference (Z1 61) - (Z1 60) = 78.57 Shadow price
Extra benefit

1. What will be Z1 when 6 X1 + 5 X2  59 ?

2. Find out the shadow price if, 10 X1 + 20 X2  149 and
10 X1 + 20 X2  151 ?
Limits of RHS Values of Constraints

L  b1  U
6 x 8 + 5 x 3.5 = 65.5
6 x 0 + 5 x 7.5 = 37.5
Water constraint varies between 65.5 & 37.5
L  b2  U
10 x 0 + 20 x 12 = 240
10 x 8 + 20 x 2.4 = 128
Land constraint varies between 240 & 128
L  b3  infinity
6.43 x 1 = 6.43
6.43  b3  
This constraint can vary from 6.43 to infinity.
Limits of Objective Coefficients
Limits of C1 Limits of C2
-6/5  - C1 / 450  -1/2 -6/5  - 500 / C2  -1/2
540  C1  225 416(2/3)  C2  1000

Note :
1. If the slope of the objective function matches, with the slope of effective constraint then
there will be multiple solutions to the problem.
2. If the feasible region is bounded, the solution will be bounded, and if the feasible region
is unbounded the solution will be unbounded.
2.3 LINEAR PROGRAMMING (SIMPLEX METHOD)

The Graphical method for solving the LP problems is useful when the number of variables is
two. However, if the number of variables is more than two, it will be difficult to follow the
graphical method as the graphs of multiple variables will be very difficult to construct and
interpret. Under such situations, a systematic approach is required to find the optimal solution.
Simplex method is such approach which identifies the optimal solution of LP problem. The
steps of Simplex method will be explained with the help of an illustrative example.

Problem 2.5
Suppose an organization is manufacturing two products. The profit per tonne of the two
products is Rs. 50 and Rs. 60 respectively. Both the products require processing in three types of
machine (m/c). The following table indicates the available m/c hours per week and the time
required on each m/c for one tonne of the two products. Formulate and solve this LP problem
using Simplex method.

Profit/tonne Product 1 (Rs. 50) Product 2 Total available m/c

(Rs. 60) hr./week
M/c 1 2 1 300
M/c 2 3 4 509
m/c 3 4 7 812

Solution
Objective Function Maximize Z = 50 X1 + 60 X2
Subject to
2 X1 + X2  300
3 X1 + 4 X2  509
4 X1 + 7 X2  812
X1  0; X2  0
We have to change the inequality constraints to equality constraints, i.e.
2 X1 + X2 + X3 = 300
3 X1 + 4X2 + X4 = 509
4 X1 + 7 X2 + X5 = 812
X1  0, X2  0, X3  0, X4  0, X5  0
Maximize Z = 50 X1 + 60 X2 + 0 X3 + 0 X4 + 0 X5
where, X3, X4 and X5 are slack variables.
For basic feasible solution;
X1 = 0, X2 = 0 -------------------------- Non basic variable
X3 = 300, X4 = 509, X5 = 812 ------- Basic variable
Z = 0 ------------------------------------- Basic feasible solution
Total no. of variables = 5; Basic variable = 3
Combinations = 5C3 = 5*4*3 / 1*2*3 = 10
 Basic Cj 50 60 0 0 0
CB Ratio test
Variable XB X1 X2 X3 X4 X5
0 X3 300 2 1 1 0 0 300/1=300
0 X4 509 3 4 0 1 0 509/4=127.25
0 X5 812 4 7 0 0 1 812/7=116
Zj – Cj -50 -60 0 0 0

Z = 50 x 0 + 60 x 0 = 0 Pivotal element

CB ----- Cost coefficient corresponding to basic variable

Cj ----- Cost coefficient corresponding to all variable

Summary of steps to be followed in Simplex Method:

Step-1 In the Zj – Cj row, we locate the quantities, which are negative. If all the quantities are
positive, the inclusion of non-basic variable will not increase the value of the objective function.
If there is more than one negative value, we choose that basic variable corresponding to which
the (Zj-Cj) value is least as this is likely to increase the profit most.
Step-2 Let Xj be the incoming basic variable and the corresponding elements of the J th column
be denoted by Y1j, Y2j, Y3j, ….. and so on. If the present value of the basic variables are XB1,
X B1 X B 2 X B 3
XB2, XB3, ….. then, we compute minimum of , , for Y1j > 0 , Y2j > 0 , Y3j > 0
Y1 j Y2 j Y3 j
X
If the minimum occurs corresponding to Br then the rth basic variable will become non-basic
Y1r
variable.
Step-3
1. Revise the basic variable as per step 2 and modify the corresponding’s value.
2. Then make the basic variable coefficient as 1 by dividing each element of the
corresponding row by the value of the pivotal element.
3. Carry out the operations on the remaining rows so that the elements corresponding to new
basic variable in remaining constraints become zero.
4. Compute Zj – Cj value and repeat the steps given above till all the values of Zj – Cj are
either zero or positive.
Conditions :
 If the Zj – Cj value corresponding to non basic variables is also zero then it indicates basic
variables is also zero then it indicates that the problem has multiple solutions.
 If the Zj – Cj value corresponding to non basic variables are same then either of the two is
considered & it has multiple solutions.
 If in the course of simplex computation, the Zj – Cj is less than zero but Yij  0 for all
values of I then problem has no finite solution.
The most negative value of (Zj – Cj) should enter into the basis. Min. of [300/1 , 509/4 , 812/7 ]
should leave the basis. It may be noted that only the positive values of ratios are considered.
Next, make the pivotal element as unity and then CB is replaced by Cj
CB Basic Cj 50 60 0 0 0
Remark
Variable XB X1 X2 X3 X4 X5
0 X3 184 10/7 0 1 0 -1/7 184/(10/7)= 128.8
0 X4 45 5/7 0 0 1 -4/7 45/(5/7)= 63
60 X2 116 4/7 1 0 0 1/7 (812/7)/(4/7)= 203
Zj – Cj -110/7 0 0 0 60/7

Z = 60 x 812/7 = 6960 improving

Basic Cj 50 60 0 0 0
CB Remark
Variable XB X1 X2 X3 X4 X5
0 X3 94 0 0 1 -2 1 94
50 X1 63 1 0 0 7/5 -4/5 -
60 X2 80 0 1 0 -4/5 3/5 400/3 = 133.33
Zj – Cj 0 0 0 22 -4
Z = 50 X1 + 60 X2
= 50 x 63 + 60 x 80
= 3150 + 4800 = 7950 improving

Basic Cj 50 60 0 0 0
CB Remark
Variable XB X1 X2 X3 X4 X5
0 X5 94 0 0 1 -2 1
50 X1 691/5 1 0 4/5 -1/5 0
60 X2 118/5 0 1 -3/5 2/5 0
Zj - Cj 0 0 4 14 0

Now all the values of Zj – Cj are non-negative, hence

Z = 50 x (691/5) + 60 x (118/5)
= 8326 ------This will be the maximum value because (Zj – Cj) is  0 in the above table

Problem 2.6 : Solve Farmer’s problem (2.4) using Simplex Method.

Mathematical form of the problem is
Maximise Z= 500X1+450X2
Subject to
6X1+5X2 ≤ 60
10X1+20X2 ≤ 150
X1 ≤ 8
X1 ≥0, X2 ≥ 0.
Solution :
Standard Form
6X1+5X2 + S1 = 60
10X1+20X2 + S2 =150
X1 +S3 = 8
X1 ≥ 0, X2 ≥ 0.
Tableau 1

Basic Cj 500 450 0 0 0 Ratio

CB
Variable XB X1 X2 S1 S2 S3 Test
0 S1 60 6 5 1 0 0 10
0 S2 150 10 20 0 1 0 15
0 S3 8 1 0 0 0 1 8→
Z= 0 Zj - Cj -500 ↑ -450 0 0 0

Tableau 2
Basic Cj 500 450 0 0 0
CB Ratio Test
Variable XB X1 X2 S1 S2 S3
0 S1 12 0 5 1 0 -6 12/5 = 2.4→
0 S2 70 0 20 0 1 -10 70/20 = 3.5
500 X1 8 1 0 0 0 1 -
Z = 4000 Zj - Cj 0 -450↑ 0 0 500

Tableau 3
Basic Cj 500 450 0 0 0
CB Ratio Test
Variable XB X1 X2 S1 S2 S3
450 X2 12/5 0 1 1/5 0 -6/5 -
0 S2 22 0 0 -4 1 14 22/14= 1.57→
500 X1 8 1 0 0 0 1 8
Z = 5080 Zj - Cj 0 0 90 0 -40↑

Tableau 4
Basic Cj 500 450 0 0 0
CB Ratio Test
Variable XB X1 X2 S1 S2 S3
450 X2 30/7 0 1 -1/7 3/35 0
0 S3 11/7 0 0 -2/7 1/14 1
500 X1 45/7 1 0 2/7 -1/14 0
Z = 5142.86 Zj - Cj 0 0 550/7 20/7 0
The objective function will be maximum at point (45/7, 30/7) and
Z max= 500 x 45/7 + 450 x 30/7 = 5142.86
So policy of the farmer should be
X1 = 45/7 i.e. wheat production 6.429 units
X2 = 30/7 i.e. jawar production 4.286 units

Problem 2.7
Solve the following linear programming problem
Maximize Z = 3 X1 + 2 X2 + 5 X3
Subject to
X1 + 2 X2 + X3 430
3 X1 + 2 X3  460
X1 + 4 X3  420
X1  0 , X2  0 , X3  0

Solution
We have to change the inequality constraints to equality constraints, i.e.
X1 + 2 X2 + X3 + X4 = 430
3 X1 + 2X3 + X5 = 460
X1 + 4 X3 + X6 = 420
X1  0, X2  0, X3  0, X4  0, X5  0 X6  0
Maximize Z = 3 X1 + 2 X2 + 5 X3 + 0 X4 + 0 X5 + 0 X6
where, X4, X5 and X6 are slack variables.
For basic feasible solution;
X1 = 0, X2 = 0, X3 = 0 -------------------------- Non basic variable
X4 = 430, X5 = 460, X6 = 420 ------- Basic variable
Z = 0 ------------------------------------- Basic feasible solution

Total no. of variables = 6; Basic variable = 3

Combinations = 6C3 = 6*5*4 / 1*2*3 = 20

Basic Cj 3 2 5 0 0 0
CB Ratio test
Variable XB X1 X2 X3 X4 X5 X6
0 X4 430 1 2 1 1 0 0 430/1=430
0 X5 460 3 0 2 0 1 0 460/2=230
0 X6 420 1 0 4 0 0 1 420/4=105
Zj – Cj -3 -2 -5 0 0 0
Z=3x0+2x0 + 5 x 0= 0
Pivotal element
 Basic Cj 3 2 5 0 0 0
CB Ratio test
Variable XB X1 X2 X3 X4 X5 X6
0 X4 325 3/4 2 0 1 0 -1/4 325/2=162.5
0 X5 250 5/2 0 0 0 1 -1/2
5 X3 105 1/4 0 1 0 0 1/4
Zj – Cj -7/4 -2 0 0 0 5/4

Z = 5 x 105 = 525 improving Pivotal element

Basic Cj 3 2 5 0 0 0
CB Ratio test
Variable XB X1 X2 X3 X4 X5 X6
2 X2 325/2 3/8 1 0 1/2 0 -1/8 (325/2)/(3/8)=433.33
0 X5 250 5/2 0 0 0 1 -1/2 250/(5/2)=100
5 X3 105 1/4 0 1 0 0 1/4 105/(1/4)=420
Zj – Cj -1 0 0 1 0 1

Z = 3 X1 + 2 X2 + 5 X3 Pivotal element
= 2 x 162.5 + 5 x 105
= 850 improving

Basic Cj 3 2 5 0 0 0 Ratio
CB
Variable XB X1 X2 X3 X4 X5 X6 test
2 X2 125 0 1 0 1/2 -3/20 -1/20
3 X1 100 1 0 0 0 2/5 -1/5
5 X3 80 0 0 1 0 -1/10 3/10
Zj – Cj 0 0 0 1 2/5 4/5

Now all the values of Zj – Cj are non-negative, hence

Z = 3 X1 + 2 X2 + 5 X3
= 3 x 100 + 2 x 125 + 5 x 80
= 950 ------This will be the maximum value because (Zj – Cj) is  0 in the above table

Problem 2.8 (Multiple Solution)

Objective Function Maximize Z = 2000 X1 + 3000 X2

Subject to
6 X1 + 9 X2  100
2 X1 + X2  20
X1  0; X2  0
We have to change the inequality constraints to equality constraints, i.e.
6 X1 + 9 X2 + X3 = 100
2 X1 + X2 + X4 = 20
X1  0, X2  0, X3  0, X4  0
Maximize Z = 2000 X1 + 3000 X2 + 0 X3 + 0 X4
where, X3 and X4 are slack variables.
For basic feasible solution;
X1 = 0, X2 = 0 -------------------------- Non basic variable
X3 = 100, X4 = 20------------------------ Basic variable
Z = 0 ------------------------------------- Basic feasible solution
Total no. of variables = 4; Basic variable = 2
Combinations = 4C2 = 4*3 / 1*2 = 6

Basic Cj 2000 3000 0 0

CB Ratio test
Variable XB X1 X2 X3 X4
0 X3 100 6 9 1 0 100/9=11.11
0 X4 20 2 1 0 1 20/1=20
Zj – Cj -2000 -3000 0 0

Z = 2000 x 0 + 3000 x 0 = 0 Pivotal element

CB ----- Cost coefficient corresponding to basic variable
Cj ----- Cost coefficient corresponding to all variable

Basic Cj 50 60 0 0
CB Remark
Variable XB X1 X2 X3 X4
3000 X2 100/9 6/9 1 1/9 0
0 X4 80/11 4/3 0 -1/9 1
Zj – Cj 0 0 1000/3 0

Now all the values of Zj – Cj are non-negative, hence

Z = 3000 x (100/9)
= 33333.33 ------This will be the maximum value because (Zj – Cj) is  0 in the above table
Still if we go further

Basic Cj 50 60 0 0
CB Remark
Variable XB X1 X2 X3 X4
3000 X2 20/3 0 1 1/6 1/2
2000 X1 20/3 1 0 -1/12 9/12
Zj – Cj 0 0 1000/3 3000

Now also all the values of Zj – Cj are non-negative, hence

Z = 2000 x (20/3) + 3000 x (20/3)
= 33333.33
Both the values are same. So it is found that the problem has multiple solution.
2.4 PHASE I AND II METHOD OF SOLVING THE LP PROBLEM

When problem has  &  both type of constraints then the problem can be solved in two phases
which is explained through an example here:

Problem 2.9
Minimize Z  12.5 x1  14.5 x 2
Subject to
x1  x2  2000 or  x1  x2  2,000
0.4 x1  0.75 x2  1000 or  40 x1  75 x2  100,000
0.075 x1  0.1x2  200 or  75 x1  100 x2  200,000
x1  0 & x2  0

Solution:
Maximize Z  12.5 x1  14.5 x 2
Subject to
x1  x 2  x3  2000
0.4 x1  0.75 x 2  x 4  1000
0.075 x1  0.1x 2  x5  200
where, x3 & x4 are surplus variable and x5 is slack variable. x3 , x4 , x5  0
Now,
x1  x 2  x3  x 6  2000
0.4 x1  0.75 x 2  x 4  x 7  1000
0.075 x1  0.1x 2  x5  200
where, x6 & x7 are artificial variable and x6 , x7  0
Phase-I
Minimize Z  x6  x7 Or, Maximize Z  x6  x7

Subject to
x1  x 2  x3  x 6  2,000
40 x1  75 x 2  x 4  x 7  100,000
75 x1  100 x 2  x5  200,000
Coeff. Basic Cj 0 0 0 0 0 -1 -1
Ratio
Basic variable XB X1 X2 X3 X4 X5 X6 X7
-1 X6 2000 1 1 -1 0 0 1 0 2000
-1 X7 100000 40 75 0 -1 0 0 1 100000/75 =1333.33
0 X5 200000 75 100 0 0 1 0 0 2000
Zj – Cj -41 -76 1 1 0 0 0

Coeff. Basic Cj 0 0 0 0 0 -1 -1
Ratio
Basic variable XB X1 X2 X3 X4 X5 X6 X7
X6 (2000/3)/(7/15)
-1 2000/3 7/15 0 -1 1/75 0 1 -1/75
=1428.57
(4000/3)/(8/15)
0 X2 4000/3 8/15 1 0 -1/75 0 0 1/75
=2500
0 X5 200000/3 65/3 0 0 4/3 1 0 -4/3 40000/13=3076.92
Zj – Cj -7/15 0 1 -1/75 0 0 76/75

Coeff. Basic Cj 0 0 0 0 0 -1 -1
Ratio
Basic variable XB X1 X2 X3 X4 X5 X6 X7
0 X1 10000/7 1 0 -15/7 1/35 0 15/7 -1/35
0 X2 4000/7 0 1 8/7 -1/35 0 -8/7 8/525
0 X5 250000/7 0 0 325/7 5/7 1 -325/7 -5/7
Zj – Cj 0 0 0 0 0 1 1
Now all the values of Zj-Cj are non-negative

Phase-II
Now we introduce the original problem
Cj -12.5 -14.5 0 0 0
Coeff. Basic
Ratio
Basic variable XB X1 X2 X3 X4 X5
-12.5 X1 10000 / 7 1 0 -15/7 1/35 0
-14.5 X2 4000 / 7 0 1 8/7 -1/35 0
0.0 X5 250000/7 0 0 325/7 5/7 1
Zj - Cj 0 0 143/14 2/35 0
Since all the values of Zj-Cj are non negative, hence, the solution is optimum. Therefore,
X1 = 10000 / 7 and X2 = 4000 / 7 Z = 12.5 x 10000 / 7+ 14.5 x 4000 / 7=183000 / 7
Note : If in the course of computation by simplex method one or more artificial variable remain
basic variable at the end of phase-I computation then the problem has no feasible solution,
i.e. at the end of phase-I artificial variable must go out.

The graphical solution of the problem is as follows :

Coordinate of points in the policy space where the slope of boundary changes are:
A. (0,2000) ;

B. ;

C. (2500, 0) ;

D.

The objective function will be minimum at point B & shall be

Z min = 12.5× (10000/7) + 14.5× (4000/7)
= 183000/7 = 26142.86
2.5 BIG-M METHOD

Problem 2.10 Solve Problem 2.9 by Big-M method.

Solution :
Maximize Z  12.5 x1  14.5 x 2  M ( x6  x7 )
Subject to
x1  x 2  x3  x 6  2,000
40 x1  75 x 2  x 4  x 7  100,000
75 x1  100 x 2  x5  200,000
x1 , x 2 , x3 , x 4 , x5 , x 6 , x 7  0
Coeff. Basic Cj -12.5 -14.5 0 0 0 -M -M
Ratio
Basic variable XB X1 X2 X3 X4 X5 X6 X7
-M X6 2000 1 1 -1 0 0 1 0 2000
-M X7 100000 40 75 0 -1 0 0 1 1333.33
0 X5 200000 75 100 0 0 1 0 0 2000
-41M -76M
Zj - Cj M M 0 0 0
+12.5 +14.5

Coeff. Basic Cj -12.5 -14.5 0 0 0 -M

Ratio
Basic variable XB X1 X2 X3 X4 X5 X6
-M X6 2000/3 7/15 0 -1 1/75 0 1 10000/7=1428.57
-14.5 X2 4000/3 8/15 1 0 -1/75 0 0 2500
0 X5 200000/3 65/3 0 0 4/3 1 0 40000/13=3076.92
(-7/15)M (-M/75) =142
Zj - Cj +(143/30) 0 M +(29/150) 0 0

Cj
Coeff. Basic -12.5 -14.5 0 0 0
Ratio
Basic variable XB X1 X2 X3 X4 X5
-12.5 X1 10000/7 1 0 -15/7 1/35 0
-14.5 X2 4000/7 0 1 8/7 -1/35 0
0 X5 250000/7 0 0 325/7 5/7 1
Zj - Cj 0 0 143/14 2/35 0

Since all the values of Zj-Cj are non-negative, the result is optimum. Therefore, X1 =10000/7
and X2 = 4000/7 Z = 12.5 x10000/7 + 14.5 x 4000/7= 26,142.86
Problem 2.11 .
Maximise Z = 12X1 + 15 X2 + 9X3
Subject to
8 x1 + 16 x2 + 12 x3  250
4 x1 + 8 x2 +10 x3  80
7 x1 + 9 x2 + 8 x3 = 105
x1 , x2 , x3  0
Solution:
Maximize Z = 12 x1 + 15 x2 + 9 x3

Subject to
8 x1 + 16 x2 + 12 x3 + x4 = 250
4 x1 + 8 x2 +10 x3 - x5 = 80
7 x1 + 9 x2 + 8 x3 = 105
where, x5 is surplus variable and x4 is slack variable. x4 , x5  0
Now,
8 x1 + 16 x2 + 12 x3 + x4 = 250
4 x1 + 8 x2 +10 x3 - x5 + x6 = 80
7 x1 + 9 x2 + 8 x3 + x7 = 105
where, x6 & x7 are artificial variable and x6 , x7  0
Revised objective function is
Maximize Z = 12 x1 + 15 x2 + 9 x3 – M (x6+ x7 )
Subject to
8 x1 + 16 x2 + 12 x3 + x4 = 250
4 x1 + 8 x2 +10 x3 - x5 + x6 = 80
7 x1 + 9 x2 + 8 x3 + x7 = 105
x1 , x2 , x3 , x4 , x5 , x6 , x7  0
Coeff. Basic Cj 12 15 9 0 0 -M -M
Ratio
Basic variable XB X1 X2 X3 X4 X5 X6 X7
0 X4 250 8 16 12 1 0 0 0 250/12=20.833
-M X6 80 4 8 10 0 -1 1 0 80/10=8
-M X7 105 7 9 8 0 0 0 1 105/8=13.12
-11M -17M -18M
Zj - Cj 0 +M 0 0
-12 -15 -9
Coeff. Basic Cj 12 15 9 0 0 -M
Ratio
Basic variable XB X1 X2 X3 X4 X5 X7
-0 X4 154 16/5 32/5 0 1 6/5 0 385/8=48.12
9 X3 8 2/5 4/5 1 0 -1/10 0 20
-M X7 41 19/5 13/5 0 0 4/5 1 205/19=10.789
-19/5M - -13/5M - -4/5M -
Zj - Cj 42/5 39/5
0 0
9/10
0

Cj 15
Coeff. Basic 12 9 0 0
Ratio
Basic variable XB X1 X2 X3 X4 X5
0 X4 2270/19 0 80/19 0 1 10/19 227/8=28.37
9 X3 70/19 0 10/19 1 0 -7/38 7
12 X1 205/19 1 13/19 0 0 4/19 205/13=15.769
Zj - Cj 0 -39/19 0 0 33/38

Coeff. Basic Cj 15
12 9 0 0
Ratio
Basic variable XB X1 X2 X3 X4 X5
0 X4 90 0 0 -8 1 2
15 X2 7 0 1 19/10 0 -7/20
12 X1 6 1 0 -13/10 0 9/20
Zj - Cj 0 0 39/10 0 3/20

Since all the values of Zj-Cj are non-negative), the optimum values are achieved, so X1 = 6 and
X2 = 7  Z = 12 x 6 + 15 x 7 + 9 x 0 = 177

2.6 DUAL OF A LP PROBLEM

Every LP problem is associated with another LP problem known as Dual Problem. The original
problem is known as Primal Problem. The formulation of Dual LP problem is substantially
helpful in understanding the LP. The variable of Dual LPP are also known as Dual Variables
have important economic interpretations which can be used by a decision maker for planning his
resources. Under certain circumstances the dual problem is easier to solve than Primal Problem.
The solution of Dual Problem lead to the solution of the Primal Problem and thus efficient
computational technique can be developed through the concept of duality. Finally in the
problems of competitive strategy, solution of both the Primal and the Dual Problem is necessary
to understand the problem fully.
Formulation of a Dual Problem
If the Primal problem is in Standard form then dual can be formulated using the following rules:
Primal Dual
1. Objective is to Maximize 1. Objective is to Minimize
2. Variables are j, i.e., Xj 2. Constraints are j
3. Constraints are I 3. Variables are I (Wi)
4. If variable Xj is unrestricted in sign 4. Jth Constraint is = type
5. If constraint is = type 5. Corresponding variable (Wi) is unrestricted
6.  type constraint 6.  type variable
7. > inequality constraint 7. < 0 variable
8.  0 variable 8.  inequality constraint

1. If the Primal problem is a maximization problem, then, the dual problem will be a
minimization problem
2. Numbers of constraints in the Primal problem are equal to the number of Dual variables.
Further, the numbers of constraints in the Dual problem will be equal to the number of
variables in the Primal Problem.
3. The profit coefficients of the Primal Problem appear on the right hand side of the constraints
of dual problem.
4. The primal problem has  type constraints while the dual problem has  type constraints.
5. The coefficient of the primal problem, which appears, left to right be placed top to bottom in
the constraints of the dual problem & vice-versa.

If the primal problem is non-standard form, the structure of the dual-problem remains
unchanged, however, if a constraint is  type the corresponding dual variable is negative or zero
and if a constraint in the primal is equal to type then the corresponding dual variable is
unrestricted in sign.
Properties of Dual Problem
1. If the primal problem is in standard form, the solution of dual problem can be obtained from
the Zj-Cj values of the slack variable in the final simplex table
2. The maximum value of the objective function of the primal problem is the minimize value of
the objective function of the dual problem.
3. If the number of constraints and variables in a primal problem are m & n respectively. The
number of constraints and variable in the dual problem will be n & m respectively.
4. Dual of the dual LP problem is again a primal problem.
5. A linear program can also be unbounded or infeasible. Duality theory tells that if the primal
is unbounded then the dual is infeasible by the weak duality theorem. Likewise, if the dual is
unbounded, then the primal must be infeasible. However, it is possible for both the dual and
the primal to be infeasible.
6. If either the primal or dual problem has a finite optimal solution, the other one also possesses
the same, and the optimal value of the objective functions of the two are equal.
7. Complementary slackness property of primal-dual relationship states that for a positive basic
variable in the primal, the corresponding dual variable will be equal to zero. Alternatively,
 for a non-basic variable in the primal (which is zero), the corresponding dual variable will be
basic and positive. For example, if the slack variable in the primal problem are Y1, Y2, Y3,
….. Ym, and the surplus variable in the dual problem are Z1, Z2, Z3, ….. Zn, then in the
optimal solution

a. If Xi >0  Zi = 0
b. If Zi >0  Xi = 0
c. If Wi >0  Yi = 0
d. If Yi >0  Wi = 0

Physically, the shadow price or the marginal cost of a resource, which is the unique price that is
equal to the increase in profit to be realized by one additional unit of the resource, is referred as
the dual variable that means this is the highest price the manufacturer would be willing to pay for
a resource. If the shadow price of a resource is zero that means there is already unutilized amount
and the profit is not increased until the current supply is totally exhausted.

Write the dual of the following primal problems.

Problem : 2.12
The primal problem is
Maximise Z = X1 - X2 + 3X3
Subject to
x1 + x2 + x3  10
2 x1 - x3  2
2 x1 - 2 x2 - 3 x3  6
x1 , x2 , x3  0
Solution
Dual of the above primal is
Minimize Z = 10 w1+ 2 w2+ 6 w3
Subject to
w1+ 2 w2+ 2 w3  1
w1- 2 w3  - 1
w1- w2- 3 w3  3
w1 , w2 , w3  0
Problem :2.13
The primal problem is
Minimize Z = X1 - 3X2 - 2X3
Subject to
3x1 - x2 + 2x3  7 or -3x1 + x2 - 2x3  - 7
2 x1 - 4x2  12
-4 x1 + 3 x2 + 8 x3 = 10
x1 , x2  0, x3 is unrestricted in sign
Solution
Dual of the above primal is
Maximize Z = -7 w1+ 12 w2+ 10 w3
Subject to
-3w1+ 2 w2- 4 w3  1
w1- 4 w2 + 3 w3  - 3
-2w1+8 w3 = - 2
w1 , w2 ,  0 , w3 is unrestricted in sign

Problem : 2.14
The primal problem is
Minimize Z = X1 + 2X2
Subject to
2x1 + 4x2  160 or -2x1 - 4x2  -160
x1 - x2 = 30
x1  10
x1 , x2  0
Solution
Dual of the above primal is
Maximize Z = -160 w1+ 30 w2+ 10 w3
Subject to
- 2w1+ w2+ w3  1
- 4w1 - w2  2
w1 , w3 ,  0 , w2 is unrestricted in sign
Problem :2.15
The primal problem is
Maximize Z = 2X1 + 3X2 + 4X3
Subject to
2x1 + 3x2 + 5x3  2 or -2x1 - 3x2 - 5x3  -2
3x1 + x2 + 7x3= 3
x1 + 4x2 + 6x3  5
x1 , x2  0 , x3 is unrestricted in sign
Solution
Dual of the above primal is
Maximize Z = -2 w1+ 3 w2+ 5w3
Subject to
-2w1+ 3w2+ w3  2
-3w1+ w2+ 4w3  3
-5w1+ 7w2+ 6w3 = 4
w1 , w3 ,  0 , w2 is unrestricted in sign

Problem :2.16
The primal problem is
Minimise Z = 3X1 + X2 + X3 - X4
Subject to
2x1 - x2 + 3x3 - x4 = 1
x1 + x2 - x3 + x4 = 3
2 x2 – x3  4
x1 , x2 , x3 , x4  0,
Solution
Dual of the above primal is
Maximize Z = w1+ 3 w2+ 4 w3
Subject to
2w1+ w2  3
-w1+ w2 + 2 w3  13
3w1- w2 - w3  1
 -w1+ w2  -1
w1 , w2 is unrestricted in sign , w3  0

Problem 2.17 Write the dual of the following problems :

(i) Maximize 5 x1  12 x 2  4 x 3
Subject to
x1  2 x 2  x 3  10
2 x1  x 2  3 x 3  8
x1 , x 2 , x 3  0

(ii) Maximize 12 x1  15 x 2  9 x 3
8 x1  16 x 2  12 x 3  250
4 x1  8 x 2  10 x 3  80
Subject to 7 x1  9 x 2  8 x 3  105
x1 , x 2 , x 3  0

2.7 DUAL SIMPLEX METHOD

In simplex algorithm presented earlier the procedure starts at a basic feasible solution. The
successive iterations remain basic and feasible but move towards optimality until the optimal is
reached at the last iteration. This was known as Primal Simplex method.
In the dual simplex method the linear programming starts at a basic solution that is better than
optimal but infeasible and successive iterations remain basic and better than optimal as they
move towards feasibility. At the last iteration the feasible optimal solution is found.
The generalized simplex algorithm combines both the primal and dual simplex methods in one
algorithm. It deals with the problem that starts both non optimal and infeasible. In this algorithm
successive iterations are associated with basic (feasible and infeasible) solutions. At the final
iteration the solution is both optimal and feasible.

Problem 2.18
Objective Function Minimize Z = X1 + 2 X2 + 3 X3

Subject to 2 X1 - X2 + X3  4 or - 2 X1 + X2 - X3 < - 4
X1 + X2 + 2 X3 < 8 or X1 + X2 + 2 X3 < 8 (same)
X2 - X3 > 2 X2 + X3  - 2
X1  0; X2  0 X3  0
We have to change the inequality constraints to equality constraints, i.e.
- 2 X1 + X2 - X3 + X4 = - 4
X1 + X2 + 2 X3 + X5 = 8
X1 + X2 + X6 = - 2

X1  0, X2  0, X3  0, X4  0
For basic feasible solution;
X1 = 0, X2 = 0 -------------------------- Non basic variable
X4 = -4, X5 = 8 , X6 = -2------------------------ Basic variable
Z = 0 ------------------------------------- Basic feasible solution
Prepare the first simplex table and compute the Zj – Cj values. The slack variable which gives
the most negative value will be the leaving variable. For example, in this problem, slack variable
corresponding to – 4, is the leaving variable.
CB Basic Cj 1 2 3 0 0 0
Variable XB X1 X2 X3 X4 X5 X6
0 X4 -4 -2 1 -1 1 0 0
0 X5 8 1 1 2 0 1 0
0 X6 -2 0 -1 1 0 0 1
Zj 0 0 0 0 0 0
Zj-Cj -1 -2 -3 0 0 0
X4 -2 1 -1 1 0 0
Ratio (-1)/(-2) (-3)/(-1)
= 1/2 =3

For identifying the entering variable, we do the ratio test. In ratio test, repeat the row
corresponding to the leaving variable just below the (Zj – Cj) row. Compute the ratio for each
variable except the positive and zero values in the denominator. Identify the minimum ratio
value. The variable corresponding to minimum ratio will be the entering variable.
Then make the pivotal row and revise the table as is done in simplex table as shown below. -2 is
the pivotal element.

CB Basic Cj 1 2 3 0 0 0
Variable XB X1 X2 X3 X4 X5 X6
1 X1 2 1 - 1/2 1/2 - 1/2 0 0
0 X5 6 0 3/2 3/2 1/2 1 0
0 X6 -2 0 -1 1 0 0 1
Zj 1 - 1/2 1/2 - 1/2 0 0
Zj-Cj 0 -5/2 -5/2 - 1/2 0 0
X6 0 -1 1 0 0 1
Ratio (-5/2)/(-1)
= 5/2
Here, -1 is the pivotal element.
CB Basic Cj 1 2 3 0 0 0
Variable XB X1 X2 X3 X4 X5 X6
1 X1 3 1 0 0 - 1/2 0 -½
0 X5 3 0 0 3 1/2 1 3/2
2 X2 2 0 1 -1 0 0 -1
Zj 1 2 -2 - 1/2 0 -5/2
Zj-Cj 0 0/1 -5 - 1/2 0 -5/2

In above table, all values of XB are positive.

X1 = 3, X2 = 2, X3=0
Z=7

Problem 2.19
Objective Function Minimize Z = 3 X1 + 2 X2
Subject to 3 X1 + X2  3 or -3 X1 - X2  -3
4 X1 + 3 X2  6 -4 X1 - 3 X2  -6
X1 + X2  3 X1 + X2  3
X1  0; X2  0
We have to change the inequality constraints to equality constraints, i.e.
-3 X1 - X2 + X3 = -3
-4 X1 - 3 X2 + X4 = -6
X1 + X2 + X5 = 3
X1  0, X2  0, X3  0, X4  0

For basic feasible solution;

X1 = 0, X2 = 0 -------------------------- Non basic variable
X3 = -3, X4 = -6 , X5 = 3------------------------ Basic variable
Z = 0 ------------------------------------- Basic feasible solution
Total no. of variables = 5; Basic variable = 2

Combinations = 5C3 = 5*4*3 / 1*2*3 = 10

Basic Cj 3 2 0 0 0 Max. Ratio

CB
Variable XB X1 X2 X3 X4 X5 (Zj – Cj)/ yrj (for - ve yrj only)
0 X3 -3 -3 -1 1 0 0
0 X4 (-)6 -4 -3 0 1 0 (-) 3/-4, - 2/(-)3
0 X5 3 1 1 0 0 1
Zj – Cj -3 (-) 2 0 0 0
Z=3x0+2x0 = 0 Pivotal element
CB ----- Cost coeff. corresponding to basic var.
Cj ----- Cost coeff. corresponding to all variable

Most negative B.V. leaves the basis and for entering basic, do the max. ratio test.

Basic Cj -3 -2 0 0 0
CB Max. Ratio test
Variable XB X1 X2 X3 X4 X5
0 X3 -1 -5/3 0 1 -1/3 0 X1 enters the basis and (- 1/3)/(-5/3)
2 X2 2 4/3 1 0 -1/3 0 and (- 2/3)/(-1/3), i.e., 0.2, 2, i.e., X3
0 X5 1 -1/3 0 0 1/3 1 leaves the basis
Zj – Cj -1/3 0 0 -2/3 0
Z = 2x2
Pivotal element
= 4
Basic Cj 3 2 0 0 0 Max. Ratio
CB
Variable XB X1 X2 X3 X4 X5 test
3 X1 3/5 1 0 -3/5 -1/5 0
2 X2 6/5 0 1 4/5 -3/5 0
0 X5 6/5 0 0 -1/5 2/5 1
Zj – Cj 0 0 -1/5 -3/5 0

Now all the values of XB are non-negative, hence this is the feasible solution.
Z = 3 x (3/5) + 2 x (6/5)
= 21/5

Exercise 2.20
A farmer has 100-acre farm. He is growing vegetables and fruits in his farm and sell all the
tomatoes, lettuce and radish produced in the farm. The price that he can obtain is Rs 10 per kg
for tomato, Rs 7.50 for price of the lettuce and Rs 20 per kg for radish. The average yield per
acre is 2000 kg of tomato, 3000 pieces of lettuce and 1000 kg of radish. The fertilizer is available
at Rs 5 per kg and the amount required per acre is 100 kg for tomato and lettuce and 50 kg for
radishes. The laborer required for sowing cultivating and harvesting per acre is 5 man-days for
tomato and radish and 6 man-days for lettuce. A total of 400 man-days are available at the rate of
Rs 100 per man-day. Formulate the LPP for the maximizing the farmer’s benefit.
Exercise 2.21
Maximize 12 x1  3 x2  x3
Subject to
10 x1  2 x 2  x 3  100
7 x1  3 x 2  2 x3  77
2 x1  4 x 2  x3  80
x1  0 , x 2  0 , x 3  0
REFERENCES

1. Dantzig,G.B., and Thapa, M.N. (1997). Linear Programming 1: Introduction.

Springer-Verlag, New York.
2. Haimes,Y.Y.,1977, ‘Hierarchical Analyses of Water Resources Systems- modeling and
Optimization of Large Scale Systems’, McGraw Hill, New Delhi.
3. Hall,W.A., and J.A. Dracup, Water Resources Systems Engineering, Tata McGraw-Hill
Publishing Company, New Delhi, 1979.
4. IGNOU, 1992, 'Operations Research-MS-51’, IGNOU, New Delhi.
5. Loucks, D.P., J.R. Stedinger, and D.A. Haith, Water Resources Systems Planning and
Analysis, Prentice Hall Inc., New Jersey, 1981.
6. Mays, L.W. and Tung, Y.K., 1992, ‘Hydrosystems Engineering and Management’, McGraw
Hill, New Delhi.
7. Rao, S.S., 1992,’Optimization Theory and Applications’. Second Edition, Wiley Eastern
Limited, New Delhi.
8. Sharma, J.K., 1997, 'Operations Research’ Macmillan India Limited, New Delhi..
9. Taha, H.A., 2002, 'Operations Research An Introduction', Seventh Edition, PHI, New Delhi.
10. Wurbs, R.A. and James, W.P., 2002,’Water Resources Engineering’, PHI, New Delhi.