9 Chapter 4 Simplex Method
9 Chapter 4 Simplex Method
9 Chapter 4 Simplex Method
net/publication/311694393
SIMPLEX METHOD
CITATIONS READS
0 1,685
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Dalgobind Mahto on 17 December 2016.
ESSENTIALS OF
OPERATIONS RESEARCH
Chapter 4: Linear Programming-Iii (Simplex Method)
4.1 INTRODUCTION
Simplex method is an algebraic procedure in which a series of repetitive operations are used to
reach at the optimal solution. Therefore, this procedure has a number of steps to find out a
solution of the problem. There may be any number of variables and constraint in the problem ,
however, it is very tough to solve manually, when there are more than 4 variables. It requires a
computer for large number of variables
Tableau. The current matrix, with auxiliary row and/or columns, as it appears at an
iterative stage in the standard simplex method.
Surplus Variable. A slack variable when an inequality is of the “greater than or equal
to” form
The simplex method can be understood in a better way with the help of an example
Solution
The first simplex table is shown in Table 4.1. In order to understand the rule, the table is
thoroughly explained as follows.
Row 1 contains C or the contribution to total profit with the production of one unit of
each product X1 and X2. This row gives the coefficients of the variables in the objective
function which will remain the same. Under column 1(CJ) is provided profit unit of 4
variables S1 S2 S3 S4 which is zero.
All the variables S1, S2, S3, S4 are listed under solution Mix. Their profit is zero and
written under column 1 (CJ) as explained above.
The constraints variables are written to the right of solution mix. These are X1 X2 S1, S2,
S3 and S4 Under these are written coefficient of variable and under each are written the
coefficients X1, X2, S1, S2, S3 and S4 in first constraint equation are 3,2,1,0,0 and 0,
respectively which are written under these variables in the first level. Similarly,, the
remaining 3 rows represents the coefficient of the variables as they appear in the other 3
constraint equation The entries in the quantity column represent the right hand side of
each constraint equation. These values are 1,200, 1,500, 350 and 200 receptivity, for the
given problem.
The Zj values in the second row form the bottom refer to the amount of gross profit that
is given up by the introducing one unit the solution. The subscript j refer to the specific
variable being considered. The Zj values under the quantity column is the total profit for
their solution. In the initial column all the Zj -values will be zero because no real product
is being manufactured and hence there is on gross profit to be lost if they are replaced. 5.
The bottom row of the table contains net profit per unit obtained by introducing one unit
of a given variable into the solution. This row is designated as the CJ – Zj row. The
procedure for calculating Zj and Cj – Zj values is given below
Calculation of Zj
Table 4.2
Cj x X1 Cj x X2 Cj x S1
0x3=0 0 x 2 =0 0x1=0
+ + +
0 x 2= 0 0 x 6= 0 0 x 0= 0
+ + +
0 x 1= 0 0 x 0= 0 0 x 0= 0
+ + +
0x0=0 0 x 1= 0 0 x 0= 0
In the same way, Zs2, Zs2 and Zs4 can be calculated as 0 each
Calculation of Cj - Zj
Cx1 – Zx1 = 10 – 0 = 10
Cx2 – Zx2 = 20 – 0 = 20
Cs1 – Zs1 = 0 – 0 = 0
Cs2 – Zs2 = 0 – 0 = 0
Cs3 – Zs3 = 0 – 0 = 0
Cs4 – Zs4 = 0 – 0 = 0
The total profit for this solution is Rs zero.
Step 3.
Determine the variable to be brought into the solution. An improved solution is possible if there
is a positive value in Cj – Zj row. The variable with the largest positive value in the Cj – Zj, row is
subjected as the objective is to maximize the profit, The column associated with this variable is
referred to as ‘Key column‘ and is designated by a small arrow beneath this column. In the given
example , 20 is the largest possible value corresponding to X2 which is selected as the key
column.
Step 4.
Determine which variable is to be replaced. To make this determination, divine each amount in
the contribution quantity column by the amount in the comparable row of Key column, X2 and
choose the variable associated with smallest quotient as the one to be replace. In the given
example, these value are calculated as
for the S1 row – 1200/2 = 600
for the S2 row – 1500/6 = 250
for the S3 row – 350/2 =
for the S4 row – 200/1 = 200
Since the smallest quotient is 200 corresponding to S4 , S4 will be replaced, and its row is
identified by the small arrow to the right of the table as shown. The quotient represents the
maximum value of X which could be brought into the solution.
Step 5.
Calculate the new row values for entering the variable. The introduction of X2 into the solution
requires that the entire S4 row be replaced. The values of X2, the replacing row, are obtained by
dividing each value presently in the S4 row by the value in column X2 in the same row. This
value is termed as the key or the pivotal element since it occurs at the intersection of key row and
key column.
The row values entering variable X2 can be calculated as follows:
0/1 = 0 ; 1/1 = 1 ; 0/1 = 0 ; 0/1 = 0 ; 0/1 = 0; - 1/1 = 1; 200/1 = 200
Step 6.
Update the remaining rows. The new S2 row values are 0, 1, 0, 0, 1 and 200 which are same as
the previous table as the key element happens to be 1. The introduction of a new variable into the
problem will affect the values of remaining variables and a second set of calculations need to be
performed to update the initial table. These calculations are performed as given here:
Updated S1 row = old S1 row – intersectional element of old S1 row x corresponding element of
new X2 row.
= 3 – [2 x 0] = 3
= 2 – [2 x 1] = 0
= 1 – [2 x 0] = 1
= 0 – [2 x 0] = 0
= 0 – [2 x 0] = 0
= 0 – [2 x 1] = -2
= 1200 – [2 x 200] = 800
Now, again rewriting the second simplex table, we have updated elements as presented in Table
4.3 as follows
Table 4.3
The new variable entering the solution would be X1. It will replace the S2 row which can be
shown as follows:
for the S1 row – 800/2 = 266.7
for the S2 row – 300/6 = 150
for the S3 row – 350/2 = 350
for the S4 row – 200/1 =
As he quotient 150 corresponding of S2 row is the minimum, it will be replaced by X1
in the new solution. The corresponding elements of S2 row can be calculated as
follows:
Revised simplex table can now be written as shown in Table 4.4 below:
Table 4.4
Now the new entering variable will be S4 and it will replace S1 as shown below:
350/7 = 50
150/-3 = -50
200/3 = 66.7
200/1 = 200
In these figures, 50 represent the minimum quotient which corresponds to row S1. The
negative sign is not considered.
The new elements of S1 row to be replaced by S4 can be calculated as follows
The new elements of S4 row would be
0/7 = 0; 0/7 = 0; 1/7 = 1/7; (-3/2) x (1/7) = - 3/14; 0/7 = 0; 1; 7/7 = 1; 350/7 = 50
The updated elements of the other rows can be calculated as follows:
Elements of updated X1 row
1 – [-3 x 0] = 1
0 – [-3 x 0] = 0
0 – [-3 x 1/7] = 3/7
1 / 2 – [-3 x 3/14] = 1/7
0 – [-3 x 0] = 0
-3 – [-3 x 1] = 0
150 – [-3 x 50] = 300
Elements of updated S3 row
0 – [3 x 0] = 0
0 – [3 x 0] = 0
0 – [3 x 1/7] = 3/7
1 / 2 – [3 x 3/14] = -1/7
Revised simplex table can now be written as shown in Table 4.5 below:
Table 4.5
As there is no positive value in Cj – Zj row it represents the optimal solution, which is given as:
X1 = 300 units : X2 = 150 units
And the maximum profit Z = Rs 6,000
Subject to 2 x1 + x2 - 6 x3 = 20,
6 x1 +5 x2 + 10 x3 ≤ 76,
8 x1 - 3 x2 + 6 x3 ≤ 50,
x1 , x2 , x3 ≥ 0.
Solution: PHASE 1: It consist of the following steps:
Step 1. Set up the problem in the Standard Form
The given constraints , after the introduction of slack and artificial variable take the form
2 x1 + x2 - 6 x3 + A = 20 ,
6 x1 +5 x2 + 10 x3 + s2 = 76 ,
8 x1 - 3 x2 + 6 x3 + s3 = 50,
x1 , x2 , x3 ,s2 , s3 , A1 ≥ 0
The new (artificial) objective function is
Minimize w = A1
The new objective function is called the infeasibility form or auxiliary (dummy) objective
function. Thus the problem for phase 1 in the standard form can be written as
Cj 0 0 0 0 0 1
Cb Basic x1 x2 x3 s2 s3 A1 b θ
1 A1 2 1 -6 0 0 1 20 10
0 s2 6 5 10 1 0 0 76 38/3
0 s3 (8) -3 6 0 1 0 50 25/4 ←
Zj 2 1 -6 0 0 1 20
Cj- Zj -2 -1 6 0 0 0
↑ Initial b.f.s for phase 1 problem
Table 1.1
Step 3 . Perform optimality Test
Since Cj – Zj is negative under some variable, table 1.1 is not optimal.
Step 4 . Iterate Toward an Optimal Solution
In the table 1.1 , x1 is the incoming variable , s3 is the outgoing variable and (8) is the key
element. This element is made unity in table 1.2. Performing iteration to get an optimal solution
result in the following table:
Cj 0 0 0 0 0 1
CB Basic x1 x2 x3 s2 s3 A1 b θ
1 A1 0 (7/4) -15/2 0 -1/4 1 15/2 30/7←
0 s2 0 29/4 11/2 1 -3/4 0 77/2 154/29
0 x1 1 -3/8 3/4 0 1/8 0 25/4 -50/3
Zj 0 7/4 -15/2 0 -1/4 1 15/2
Cj- Zj 0 -7/4 15/2 0 1/4 0
↑ Second b.f.s for phase 1 problem
Table 1.2
Cj 0 0 0 0 0 1
CB Basic x1 x2 x3 s2 s3 A1 b
0 x2 0 1 -30/7 0 -1/7 4/7 30/7
0 s2 0 0 256/7 1 2/7 -29/7 52/7
0 x1 1 0 -6/7 0 1/14 3/14 55/7
Zj 0 0 0 0 0 0 0
Cj- Zj 0 0 0 0 0 1
Optimal b.f.s for phase 1 problem
Table 1.3
Since cj – Zj is non-negative under all columns, table 1.3 is optimal. Also since Wmin = 0 and
no artificial variable appear in the basic , this tableau gives a b.f.s to the original problem.
PHASE II: PHASE II of the simplex method finds optimal solution to the original problem .
Objective function for the initial table of phase II is the objective function of the original (given
problem) . The remaining part of initial table for phase II is the last table for phase I, with the
only difference thet Zj and cj-Zj row in the last table for phase I are changer to account for
changes in the cost coefficients. Table 1.3 represent the initial table for phase II computatations.
Note that in table 1.3 contains no artificial variable, it is considered in table 1.4.
Cj 5 -4 3 0 0
CB Basic x1 x2 x3 s2 s3 b
-4 x2 0 1 -30/7 0 -1/7 30/7
0 s2 0 0 256/7 1 2/7 52/7
5 x1 1 0 -6/7 0 1/14 55/7
Zj 5 -4 90/7 0 13/14 155/7
Cj- Zj 0 0 -69/7 0 -13/14
Optimal b.f.s for phase 1 problem
Table 1.4
Since cj- Zj is either negative or zero under all variable columns. Table 1.4 gives optimal basic
feasible solution for the original problem . The optimal solution is
x1 = 55/7 , x2 = 30/7 , x3 = 0 : Zmax = 155/7.
Subject to 2 x1 + x2 ≤ 10,
2 x1 +5 x2 ≤ 20,
2 x1 + 3 x2 ≤ 18,
x1 , x2 ≥ 0.
(i). Indicate that this problem has an alternate optimal basic feasible solution.
(ii). Find the optimal solution.
(iii). Hence show that this problem has multiple optimal solutions.
Solution:
Step 1. Set up the problem in the Standard Form
Introducing slack variables s1 , s2 , s3 the problem can be expressed in the standard form as
Minimize Z = 4 x1 + 10 x2 + 0 s1 + 0 s2+ 0s3,
Subject to 2 x1 + x2 + s1 + 0 s2+ 0 s3 = 10,
2 x1 +5 x2 + 0 s1 + s2 + 0 s3 = 20,
2 x1 + 3 x2 + 0 s1 +0 s2 + s3 = 18 ,
x1 , x2 , s1 ,s2 , s3 ≥ 0.
Cj 4 10 0 0 0
CB Basic x1 x2 s1 s2 s3 b θ
0 s1 2 1 1 0 0 10 10
0 s2 2 (5) 0 1 0 20 4←
0 s3 2 3 0 0 1 18 6
Zj 0 0 0 0 0 0
Cj- Zj 4 10 0 0 0
↑ Initial basic feasible solution
Table 1.5
In the table 1.5 , x2 is the incoming variable , s2 is the outgoing variable and (5) is the key
element. In table 1.6 , s2 is replaced by x2
Cj 4 10 0 0 0
CB Basic x1 x2 s1 s2 s3 b θ
0 s1 (8/5) 0 1 -1/5 0 6 15/4←
10 x2 2/5 1 0 1/5 0 4 10
0 s3 4/5 0 0 -3/5 1 6 15/2
Zj 4 10 0 2 0 40
Cj- Zj 0 0 0 -2 0
↑ Optimal basic feasible solution
Table 1.6
Since cj- Zj is either negative or zero under all variable columns , table 1.6 is optimal. The
optimal solution is
x1=0(non-basic),
x2=4 (basic),
s1=6 (basic),
s2=0 (non-basic),
s3=6 (basic),
Zmax = 40,
(i). In table 1.6 , element of net evaluation row (cj –Zj row) under x1 column ( Corresponding to
the non-basic variable x1) is zero. It indicates that by introducing x1 into solution , there will be
no improvement in the value of objective function Z. In other word , it indicates the existence of
an alternate optimal basic feasible solution.
(ii). Choosing x1 as the incoming variable and slack s1 as the outgoing variable in table 1.6 and
carrying out the iteration yield table 1.7.
Cj 4 10 0 0 0
CB Basic x1 x2 s1 s2 s3 b
4 x1 1 0 5/8 -1/8 0 15/4
10 x2 0 1 -1/4 1/4 0 5/2
0 s3 0 0 -1/2 -1/2 1 3
Zj 4 10 0 2 0 40
Cj- Zj 0 0 0 -2 0
Alternate Optimal basic feasible solution
Table 1.7
Table 1.7 gives the following alternate optimal basic feasible solution :
x1= 15/4 (Basic),
x2 = 5/2 (Basic),
s1= 0 (non- basic),
s2= 0 (non- basic),
s3= 3 (non- basic),
Zmax = 40
Note that the maximum value of the objective function remains the same . The above two are the
basic feasible optimal solutions. Now if two basic feasible optimal solution are known an infinite
number of non-basic feasible optimal solution can be derived by taking any weighted average of
these two solution.
For example, if
Example (202):
Maximize Z = 107 x1 + x2 + 2 x3,
Subject to 14 x1 + x2 - 6 x3 +3 x2 = 7,
16 x1 +1/2 x2 -6 x3 ≤ 5,
3 x1 - x2 - x3 ≤ 0 ,
x1 , x2 , x3 , x4 ≥ 0.
Solution : Step 1 . Set up the problem in the standard form
The first constraint is of equality form . It can be divided throughout by 3 to give unit
coefficient to x4
i.e 14/3 x1 + 1/3 x2 - 2 x3 + x4 = 7/3
Variable x4 , now , occur in constraint 1 with unit coefficient and it occurs in no other constraint
and hence can be treated as a slack variable. Introducing slack variables , say x5 and x6 in second
and third constraints , the problem can be expressed in standard form as
Maximize Z = 107 x1 + x2 + 2 x3 +0 x4+ 0 x5 + 0 x6
Subject to 14/3 x1 + 1/3 x2 - 2 x3 + x4+ 0 x5 + 0 x6 = 7/3 ,
16 x1 + 1/2 x2 - 6 x3 + 0 x4+ x5 + 0 x6 = 5,
3 x1 - x2 - x3 + 0 x4+ 0 x5 + x6 = 0,
x1 , x2 , x3 , x4 , x5 , x6 ≥ 0,
Cj 107 1 2 0 0 0
CB Basic x1 x2 x3 x4 x5 x6 b θ
0 x4 0 17/9 -4/9 1 0 -14/9 7/3 -21/4
0 x5 0 35/6 -2/3 0 1 -16/3 5 -15/2
107 X1 1 -1/3 -1/3 0 0 1/3 0 -0
Zj 107 -107/3 -107/3 0 0 107/3 0
Cj-Zj 0 110/3 113/3 0 0 -107/3
↑
Table 1.9 Second basic feasible solution