1

CHAPTER 3 Artificial Variables Techniques

3.1 INTRODUTION LPP in which constraints may also have and = signs after ensuring that all b i 0 are considered in this section. In such cases basis of matrix cannot be obtained as an identity matrix in the starting simplex table, therefore we introduce a new type of variable called the artificial variable. These variables are fictitious and cannot have any physical meaning. The artificial variable technique is a device to get the starting basic feasible solution, so that simplex procedure may be adopted as usual until the optimal solution is obtained. To solve such LPP there are two methods. (i) The Big M Method or Method of Penalties. (ii) The Two-phase Simplex Method. 3.2 THE BIG M METHOD The following steps are involved in solving an LPP using the Big M method. Step 1 Express the problem in the standard form. Step 2 Add non-negative artificial variables to the left side of each of the equations corresponding to constraints of the type or = However, addition of these artificial variable causes violation of the corresponding constraints. Therefore, we would like to get rid of these variables and would not allow them to appear in the final solution. This is achieved by assigning a very large penalty (-M for maximization and + M for minimization) in the objective function. Step 3 Solve the modified LPP by simplex method, until anyone of the three cases may arise.

BIG M METHOD CRIERION: OPTIMALITY The optimal solution to the augmented problem to the original problem if there are no artificial variables with nonzero value in the optimal solution. BIG M METHOD CRIERION: NO FEASIBLE SOLUTION If any artificial variable is in the basis with non zero value at the optimal solution of the augmented problem, then the original problem has no feasible solution. The solution satisfies the constraints but does not optimize the objective function, since it contains a very large penalty M and is called pseudo optimal solution.

BIG M METHOD CRIERION: DEGENERATE SOLUTION If at least one artificial variable in the basis at zero level and the optimality condition is satisfied then the current solution is an optimal basic feasible solution (though degenerated solution).

Note: While applying simplex method, whenever an artificial variable happens to leave the basis, we drop that artificial variable and omit all the entries corresponding to its column from the simplex table. Example 3.1 Use Big M method to solve the example 2.2 Turkey Feed Problem as given in Table 3.1. TABLE 3.1 Turkey meals Data Ingredient A B C Cost per pound Compostitoin of each Pound Feed(Oz.) Brand Feed 1 Brand Feed 2 5 4 1/2 Rs.2 10 3 0 Rs.3 Minimum Monthly Requirement Per Turkey(Oz.) 90 48 3/2

Minimize z = 2x 1 + 3x 2
subject to these constraints:

5x 1 + 10x 2 90 ounces 4x 1 + 3x 2 48 ounces

1 3 x 1 ounces 2 2

(ingredient A constraint) (ingredient B constraint) (ingredient C constraint)

x1 0 , x 2 0

SOLUTION Step 1. Express the problem in standard form Slack variables s1, s2 and s3 are subtracted from the left-hand sides of the constraints to convert them to

equations. These variables s1, s2 and s3 are called negative slack variable or surplus variable. Variable s1 represents excess of requirement of 90 units on ingredient A, s2 represents excess of requirement of 48 on on ingredient B and s2 represents excess of requirement of 1.5 on ingredient C . Since they represent 'free' feeds, the cost associated with them in the objective function are zeros. The problem, therefore, can be written as follows Minimize z = 2x 1 + 3x 2 + 0s1 + 0s 2 + 0s 3 subject to these constraints:

5x 1 + 10x 2 s1 = 90 4x 1 + 3x 2 s 2 = 48
x1 s3 = 3 x 1 , x 2 , s1 , s 2 , s 3 0 Step 2. Find initial basic feasible solution Putting x1 = x2 = 0, we get s1= -90, s2 =-48 and s3 =-3 as the first basic solution but it is not feasible as s1, s2 and s3 have negative values that do not satisfy the non-negativity restrictions. Therefore, we introduce artificial variables A1 , A2, A3 in the constraints, which take the form Minimize z = 2x 1 + 3x 2 + 0s1 + 0s 2 + 0s 3 subject to these constraints:

5x 1 + 10x 2 s1 + A1 = 90 4x 1 + 3x 2 s 2 + A 2 = 48
x1 s 3 + A 3 = 3 x 1 , x 2 , s1 , s 2 , s 3 , A 1 , A 2 , A 3 0 Now artificial variables with values greater than zero violate the equality in constraints established in step 1. Therefore, A1,A2,and A3 should not appear in the final solution. To achieve this, they are assigned a large unit penalty (a large postive value, +M) in the objective function, which can be written as Minimize z = 2 x 1 + 3x 2 + 0s1 + 0s 2 + 0s 3 + MA 1 + MA 2 + MA 3 subject to these constraints:

5x 1 + 10x 2 s1 + A1 = 90 4x 1 + 3x 2 s 2 + A 2 = 48
x1 s 3 + A 3 = 3 x 1 , x 2 , s1 , s 2 , s 3 , A 1 , A 2 , A 3 0

Eliminating A 1 , A 2 , A 3 from the first, second and third equations modified objective function can be written as Minimize z = (10M 2) x 1 (13M 3) x 2 + Ms1 + Ms 2 + Ms 3 + 141M Or z + (10M 2) x 1 + (13M 3) x 2 Ms1 Ms 2 Ms 3 = 141M

Problem, now, has eight variables and three constraints. five of the variables have to be zeroised to get initial basic feasible solution to the 'artificial system'. Putting, x 1 = x 2 = s1 = s 2 = s 3 = 0, and A 1 = 90, A 2 = 48, A 3 = 3 we get The starting feasible solution is A 1 = 90, A 2 = 48, A 3 = 3 and z = 141M. Note that we are starting with a very heavy cost (compare it with zero profit in maximization problem) which we shall minimze during the solution procedure. Table 3.2 represents the problem and its solution. . TABLE 3.2 Initial simplex tableau To improve this solution, we determine that x2 is the entering variable, because 10M-2 is the largest positive entry in the bottom row.

x1
5 4 1 10M-2

x2
10 3 0 13M-3

s1 -1 0 0 -M

s2 0 -1 0 -M

s3 0 0 -1 -M

A1 1 0 0 0

A2 0 1 0 0

A3 0 0 1 0

b 90 48 3 141M

B.V. s1 s2 s3 z

Ratio 90/5=19 48/4=12 3/1=3

Let M =50 as default value, then we have

x1
5 4 1 498

x2
10 3 0 647 Entering

s1 -1 0 0 -50

TABLE 3.3 First iteration simplex tableau s2 s3 A1 A2 A3 b B.V. 0 -1 0 -50 0 0 -1 -50 1 0 0 0 0 1 0 0 0 0 1 0 90 48 3 7050 s1 s2 s3 z

Ratio 90/10=9 --

Leaving variable

48/3=16

Current z-value

x1
1/2 5/2

x2
1 0 0 0

s1 -1/10 3/10 0 14.7

s2 0 -1 0 -50

TABLE 3.4 Second iteration simplex tableau s3 A1 A2 A3 b B.V. 0 0 -1 -50 1/10 -3/10 0 -64.7 0 1 0 0 0 0 1 0 9 21 3 1227 s1 s2 s3 z

Ratio 9/0.5=18 42/5=8.4

Leaving variable

1
174.5

3/1=3

Entering

Current z-value TABLE 3.5 Third iteration simplex tableau s3 A1 A2 A3 b B.V. 1/2 5/2 -1 124 Entering 1/10 -3/10 0 -64.7 0 1 0 0 -1/2 -5/2 1 -174.5 7.5 13.5 3 703.5 x2 A1 x1 z

x1
0 0 1 0

x2
1 0 0 0

s1 -1/10 3/10 0 14.7

s2 0 -1 0 -50

Ratio 7.5/0.5=15
Leaving variable

27/5=5.4

Current z-value

x1
0 0 1 0

x2
1 0 0 0

s1 -0.16 0.12 0.12 -0.24

s2 0.20 -0.40 -0.40 -0.20

TABLE 3.6 Optimal simplex tableau s3 A1 A2 A3 b 0 1 0 0 0.16 -0.12 -0.12 -49.76 -0.20 0.40 0.40 -49.8 0 -1 0 -50 4.8 5.4 8.4 31.20

B.V. x2 s3 x1 z

Since all z coefficients are negative it becomes optimal solution with minimum cost Rs. 31.20. Hence, the minimum cost solution is to purchase 8.4 pounds of brand 1 feed and 4.8 pounds of brand 2 feed per turkey per month. EXAMPLE 3.2 Use Big M method to Maximize z = 3x1 + 2x2 Subject to the constraints 2x 1 + x 2 2 3x 1 + 4 x 2 12

x1 , x 2 0

SOLUTION Step 1. Express the problem in standard form Slack variables s1 and s2 are add and subtracted from the left-hand sides of the constraint 1 and 2 respectively to convert them to equations. These variable s2 is called negative slack variable or surplus variable. Variable s1 represents excess of availability of 2 units on constraint 1, s2 represents excess of requirement of 12 on constraint 2 . Since they represent 'free', the cost/profit coefficients associated with them in the objective function are zeros. The problem, therefore, can be written as follows Maximize Z = 3x1 + 2x2+0s1+0s2 2 x 1 + x 2 + s1 = 2 3x 1 + 4x 2 s 2 = 12

x 1 , x 2 , s1 , s 2 , 0
Step 2. Find initial basic feasible solution Putting x1 = x2 = 0, we get s1=2, s2 =-12 as the first basic solution but it is not feasible as s2 have negative values that do not satisfy the non-negativity restrictions. Therefore, we introduce artificial variables A1 in the second constraint, which take the form 2 x 1 + x 2 + s1 = 2 3x 1 + 4 x 2 s 2 + A 1 = 12

x 1 , x 2 , s1 , s 2 , A1 0
Now artificial variables with values greater than zero violate the equality in constraints established in step 1. Therefore, A1 should not appear in the final solution. To achieve this, they are assigned a large unit penalty (a large negative value, - M) in the objective function, which can be written as

Maximize z = 3x 1 + 2x 2 + 0s1 + 0s 2 MA1

Subject to 2 x 1 + x 2 + s1 = 2 3x 1 + 4 x 2 s 2 + A 1 = 12

x 1 , x 2 , s1 , s 2 , A1 0
Eliminating A1 from the second equation, modified objective function can be written as

Maximize z = (3 + 3M) x 1 + (2 + 4M) x 2 + 0s1 Ms 2 12M

Let M =50 as default value, then, we have

Maximize z = 153x 1 + 202x 2 + 0s1 5s 2 600

Problem, now, has five variables and two constraints. Three of the variables have to be zeroised to get initial basic feasible solution to the 'artificial system'. Putting x 1 = x 2 = s 2 = 0 , we get The starting feasible solution is s1 = 2, A1 = 12 and z = 600 . Note that we are starting with a very heavy negative profit which we shall maximize during the solution procedure. Table 3.7 represents the problem and its solution. TABLE 3.7 First iteration simplex tableau To improve this solution, we determine that x2 is the entering variable, because -202 is the smallest entry in the bottom row.

x1
2 3 -153 .

x2
1 4 -202 Entering

s2 0 -1 50

s1 1 0 0

A1 0 1 0

b 2 12 -600

B.V. s1 A1 z

Ratio 2/1=2

Leaving variable

12/4=3

Current z-value TABLE 3.8 Second iteration simplex tableau

x1
2 -5 251 .

x2
1 0 0

s2 0 -1 50

s1 1 -4 202

A1 0 1 0

b 2 12 -196

B.V. x2 A1 z

Ratio 2/1=2 12/4=3

Since all Z coefficients and an artificial variable appears in the objective row at positive level, the solution of given LPP does not possess any feasible solution. . EXAMPLE 3.3 Solve the LPP Minimize z = 4x1 +x2 Subject to 3x 1 + x 2 = 3 4x 1 + 3x 2 6 x 1 + 2x 2 3

x1 , x 2 0

SOLUTION Since the objective function is minimization, we covert it into maximization using Min z = Max (-z) Subject to 3x 1 + x 2 = 3 4x 1 + 3x 2 6 x 1 + 2x 2 3

x1 , x 2 0
Convert the given LPP into standard form by adding artificial variables A1, A2, surplus variable s1 and slack variable S2 to get the initial basic feasible solution.

Maximize z = 4x 1 x 2 + 0s1 + 0s 2 MA1 MA 2

Subject to 3x 1 + x 2 + A 1 x 1 + 2x 2 + S 2 =3 =3 4 x 1 + 3x 2 S1 + A 2 = 6

x1 , x 2 0
Modified objective function:

Maximize z = (7M 4x 1 ) + (4M x 2 ) Ms1 9M

The starting feasible solution is A1 = 3, A2 = 6, S2 = 3. Let M =50 TABLE 3.9 Initial iteration simplex tableau

x1
3 4 1 -346

x2
1 3 2 -199

s1 0 -1 0 50

A2 1 0 0 0

A1 0 1 0 0

s2 0 0 1 0

b 3 6 3 -450

Basic Varible A2 A1 s2 z

To improve this solution, we determine that x2 is the entering variable, because -346 is the smallest entry in the objective row. TABLE 3.10 First iteration simplex tableau

x1
3

x2
1

s1 0

A2 1

A1 0

s2 0

b 3

B.V. A2

Ratio 3/1=2

Leaving variable

4 1 -346 Entering

3 2 -199

-1 0 50

0 0 0

1 0 0

0 1 0

6 3 -450

A1 s2 z

6/4=1.5 3/1=3

Current z-value TABLE 3.11 Second iteration simplex tableau

x1
1 0 0 0

x2
.33 1.67 1.67 -83.67 Entering

s1 0 -1 0 50

A2 .33 -1.33 -0.33 115.3

A1 0 1 0 0

s2 0 0 1 0

b 1 2 2 -104

B.V. x1 A1 s2 z

Ratio 1/.33 2/1.67 2/1.67

Leaving variable

Current z-value TABLE 3.12 Third iteration simplex tableau s1 0.2 -0.6 1 -0.2 A2 .33 -1.33 -0.33 48.4 A1 0 1 0 50.2 s2 0 0 1 0 b 0.6 1.2 0 -3.6 B.V. x1 A1 s1 z
Leaving variable

x1
1 0 0 0

x2
0 1 0 0

Ratio 1/.2 2/1

Entering

Current z-value TABLE 3.13 Optimal simplex tableau

x1
1 0 0 0

x2
0 1 0 0

s1 0 0 1 0

A2 .40 -0.2 1 48.6

A1 0 0 -1 50

s2 -0.2 0.6 1 0.2

b 0.6 1.2 0 -3.6

B.V. x1 x2 s1 z

since all coefficients of z are positive solution is optimum and is given by x1=0.6=3/5, x2 =1.2=6/5, and Max z = -3.6 = -18/5 Min z= Max (-z)= 18/5.

10

EXAMPLE 3.4 Consider the following linear programming model and solve it using Big M method. Minimize z = 12 x 1 + 18x 2 + 15x 3 Subject to 4 x 1 + 8x 2 + 6 x 3 64 3x 1 + 6x 2 + 12 x 3 96 x1 , x 2 , x 3 0 SOLUTION The canonical form of the given problem is shown below: Minimize z = 12x 1 + 18x 2 + 15x 3 + 0s1 + 0s 2 + MA 1 + MA 2 Subject to 4 x 1 + 8x 2 + 6x 3 s1 + A 1 = 64 3x 1 + 6x 2 + 12 x 3 s 2 + A 2 = 96 x 1 , x 2 , x 3 , s1 , s 2 , A 1 , A 2 0 The starting feasible solution is A1 = 64, A2 =96. Let M =50, Then we have modified objective function: Minimize z = (12 7M ) x 1 + (18 14M ) x 2 + (15 18M ) x 3 + 50s1 + 50s 2 + 160M = 338x 1 682 x 2 885x 3 + 50s1 + 50s 2 + 8000 Subject to 4 x 1 + 8x 2 + 6x 3 s1 + A 1 = 64 3x 1 + 6x 2 + 12 x 3 s 2 + A 2 = 96 x 1 , x 2 , x 3 , s1 , s 2 , A 1 , A 2 0 TABLE 3.14 Initial simplex tableau x3 6 12 885 s1 -1 0 -50 s2 0 -1 -50 A1 1 0 0 A2 0 1 0 b 64 96 8000 B.V. A1 A2 z

x1
4 3 338

x2
8 6 682

To improve this solution, we determine that x3 is the entering variable, because 885 is the largest entry in the objective row.

11

TABLE 3.15 First iteration simplex tableau

x1
4 3 338

x2
8 6 682

x3 6 12 885 Entering

s1 -1 0 -50

s2 0 -1 -50

A1 1 0 0

A2 0 1 0

b 64 96 8000

B.V. A1 A2 z

Leaving variable

Current z-value s2 0.5 -0.08 23.75 A1 1 0 0 A2 -0.5 0.08 -73.75 b 16 8 B.V. A1 z

TABLE 3.16 Second Iteration Simplex tableau

x1
2.5 0.25 116.75

x2
5 0.5 239.5 Entering

x3 0 1 0

s1 -1 0 -50

Leaving variable

x3

920

Current z-value TABLE 3.17 Optimal Simplex tableau s1 -1 0 -2.1 s2 0.5 -0.08 -0.20 A1 1 0 -47.9 A2 -0.5 0.08 -49.8 b 3.2 6.4 153.6 B.V. x2 x3 z

x1
2.5 0.25 -3

x2
1 0 0

x3 0 1 0

optimum z-value since all coefficients of Z are negative solution is optimum and is given by x1=0, x2 =3.2=6/5, x3 = 6.4 and Min z = 153.6 EXERCISES 3.2 Use Big M method to solve the following LPP. 1. Minimize z = 12x1 +20x2 Subject to 6 x 1 + 8x 2 100 7 x 1 + 12 x 2 120 x1 , x 2 0 [Ans: x1 =15, x2= 4 and Min z =205] 2. Maximize z = 2 x 1 + x 2 + 3x 3 , Subject to

12

x 1 + x 2 + 2x 3 5 2x 1 + 3x 2 + 4 x 3 = 12 x1 , x 2 , x 3 0 [Ans: x1 =3, x2=2, x3,=0 and Max z = 8] 3. Minimize z = 4 x 1 3x 2 + 1x 3 , Subject to x 1 + 2x 2 + 4x 3 12 3x 1 + 2x 2 + x 3 8 x1 , x 2 , x 3 0 [Ans. x1 =0, x2=10/3,x3 =4/3 and Min z=34/3] 4. Maximize z = 2x1 + 4x 2 + x 3 Subject to x 1 2x 2 x 3 5 2x 1 x 2 + 2x 3 = 2 x 1 + 2x 2 + 2x 3 1 x1 , x 2 , x 3 0 [Ans: Unbounded solution] 5. Maximize z = x 1 + 3x 2 + 5x 3 , Subject to 3x 1 + 10x 2 + 5x 3 15 33x 1 10x 2 + 9 x 3 33 x 1 + 2x 2 + x 3 4 x1 , x 2 , x 3 0 [Ans: No solution]

3.3 THE TWO PHASE METHOD In the preceeding section we observed that it was frequently necessary to add artificial variables to the constraints to obtain an initial basic feasible solution to an L.P. problem. If the problem is to be solved, the artificial variables must be driven to zero. The two-phase method is another method to handle these artificial variables. Here the L.P. problem is solved in two phases. PHASE I In this phase we find an initial basic feasible solution to the original problem. For this all artificial variables are to be driven to zero. To do this an artificial (Auxilary) objective function (r) is created which

13

is the sum of all the artificial variables. This new objective function is then minimized, subject to the constraints of the given (original) problem, using the simplex method. At the end of phase I, two cases arise: TWO PHASE METHOD CRIERION: NO FEASIBLE SOLUTION If the minimum value of r > 0, and at least one artificial variable appears in the basis at a positive level, then the given problem has no feasible solution and the procedure terminates.

TWO PHASE METHOD CRIERION: OPTIMALITY If the minimum value of r = 0, and no artificial variable appears in the basis, then a basic feasible solution to the given problem is obtained. The artificial column (s) are deleted for phase II computations. If the minimum value of r = 0 and one or more artificial variables appear in the basis at zero level, then a feasible solution to the original problem is obtained. However, we must take care of this artificial variable and see that it never becomes positive during phase II computations. Zero cost coefficient is assigned to this artificial variable and it is retained in the initial table of phase II. If this variable remains in the basis at zero level in all phase II computations, there is no problem. However, the problem arises if it becomes positive in some iteration. In such a case, a slightly different approach is adopted in selecting the outgoing variable. The lowest non-negative replacement ratio criterion is not adopted to find the outgoing variable. Artificial variable (or one of the artificial variables if there are more than one) is selected as the outgoing variable. The simplex method can then be applied as usual to obtain the optimal basic feasible solution to the given L.P. problem. PHASE II Use the optimum basic feasible solution of phase I as a starting solution for the original LPP. Assign the actual costs to the variable in the objective function and a zero cost to every artificial variable in the basis at zero level. Delete the artificial variable column from the table which is eliminated from the basis in phase I. Apply simplex method to the modified simplex table obtained at the end of phase I till an optimum basic feasible is obtained or till there is an indication of unbounded solution. REMARKS 1. In phase I, the iterations are stopped as soon as the value of the new (artificial) objective function becomes zero because this is its minimum value. There is no need to continue till the optimality is

14

reached if this value becomes zero earlier than that. 2. Note that the new objective function is always of minimization type regardless of whether the original problem is of maximization or minimization type. EXAMPLE 3.5 Consider the example 3.4, the following linear programming model and solve it using the two-phase method. Minimize z = 12x1 + 18x 2 + 15x 3 Subject to 4 x 1 + 8x 2 + 6x 3 64 3x 1 + 6x 2 + 12 x 3 96 x1 , x 2 , x 3 0 SOLUTION The canonical form of the given problem is shown below: Minimize z = 12 x 1 + 18x 2 + 15x 3 Subject to 4 x 1 + 8x 2 + 6x 3 s1 + A 1 = 64 3x 1 + 6x 2 + 12 x 3 s 2 + A 2 = 96 x 1 , x 2 , x 3 , s1 , s 2 , A 1 , A 2 0 Phase 1 The model for phase 1 with its revised objective function is shown below. The corresponding initial table is presented in Table 3.18. Further iterations towards optimality are shown in Tables 3.19 and 3.21. Auxilary objective function: Minimize r = A 1 + A 2 = 160 7 x 1 14x 2 18x 3 + s1 + s 2 subject to 4 x 1 + 8x 2 + 6x 3 s1 + A 1 = 64 3x 1 + 6x 2 + 12 x 3 s 2 + A 2 = 96 x 1 , x 2 , x 3 , s1 , s 2 , A 1 , A 2 0 TABLE 3.18 Initial Table of Phase 1 (Example 3.5) s2 0 -1 -1 A1 1 0 0 A2 0 1 0 b 64 96 160 B.V. A1 A2 r x3 6 12 18 s1 -1 0 -1

x1
4 3 7

x2
8 6 14

15

TABLE 3.19 First iteration simplex tableau

x1
4 3 7

x2
8 6 14

x3 6 12 18 Entering

s1 -1 0 -1

s2 0 -1 -1

A1 1 0 0

A2 0 1 0

b 64 96 160

B.V. A1 x3 r

Leaving variable

Current r-value

TABLE 3.20 Second iteration simplex tableau

x1
2.5 0.25 2.5

x2
5 0.5 5

x3 0 1 0

s1 -1 0 -1

s2 0.5 -0.093 0.5

A1 1 0 0

A2 0.5 -0.093 1.5

B.V.

64 Leaving Avariable 1 96 16 Current r-value A2 r

Entering

TABLE 3.21 Optimal simplex tableau of pahase 1

x1
0.5 0 0

x2
1 0 0

x3 0 1 0

s1 -0.2 0.1 0

s2 0.1 -0.13 0

A1 -0.1 -0.093 0

A2 -0.1 0.13 0

b 3.2 6.4 0

B.V. x2 x3 r

The set of basic variables in the optimal table of phase 1 does not contain artificial variables. So, the given problem has a feasible solution. Phase 2 The only one iteration of phase 2 is shown in Table 3.22. One can verify that Table 3.23 gives the optimal solution. The solution in Table 3.23 is optimal and feasible. The optimal results are presented below. by x1=0, x2 =3.2=6/5, x3 = 6.4 and min z=153.6 TABLE 3.22 Initial Table of Phase 2

x1
0.5 0 -12

x2
1 0 -15

x3 0 1 -18

s1 -0.2 0.1 0

s2 0.1 -0.13 2.1

b 3.2 6.4 0

B.V. x2 x3 z

TABLE 3.23 Optimal Table of Phase 2

16

x1
0.5 0 3

x2
1 0 0

x3 0 1 0

s1 -0.2 0.1 0

s2 0.1 -0.13 2.1

b 3.2 6.4 153.6

B.V. x2 x3 z

The optimal results are presented by x1=0, x2 =3.2=6/5, x3 = 6.4 and min z=153.6 EXAMPLE 3.6 Use the two-phase simplex method to maximize Maximize z = 5x 1 4 x 2 + 3x 3 Subject to 2x 1 + x 2 6x 3 = 20

6x 1 + 5x 2 + 10 x 3 76 8x 1 3x 2 + 6 x 3 50 x1 , x 2 , x 3 0 SOLUTION The canonical form of the given problem is shown below: Maximize z = 5x 1 4 x 2 + 3x 3 Subject to 2x 1 + x 2 6x 3 + A1 = 20

6x 1 + 5x 2 + 10 x 3 + s1 = 76 8x 1 3x 2 + 6x 3 + s 2 = 50 x 1 , x 2 , x 3 , s1 , s 2 , A 1 0 Phase 1 The model for phase 1 with its revised objective function is shown below. The corresponding initial table is presented in Table 3.24. Further iterations towards optimality are shown in Tables 3.50 and 3.21. Auxilary objective function: Minimize r = A 1 = 20 2 x 1 x 2 + 6 x 3 Subject to

17

2x 1 + x 2 6x 3 + A1

= 20

6x 1 + 5x 2 + 10x 3 + s1 = 76 8x 1 3x 2 + 6 x 3 + s 2 = 50 x 1 , x 2 , x 3 , s1 , s 2 , A 1 0 TABLE 3.24 Initial Table (Example 3.6) x3 -6 10 6 -6 A1 1 0 0 0 s1 0 1 0 0 s2 0 0 1 0 b 20 76 50 20 B.V. A1 s1 s2 r

x1
2 6 8 2

x2
1 5 -3 1

TABLE 3.25 First Iterationl simplex tableu

x1
2 6 8 2 Entering

x2
1 5 -3 1

x3 -6 10 6 -6

A1 1 0 0 0

s1 0 1 0 0

s2 0 0 1 0

b 20 76 50 20

B.V. A1 Leaving variable s1 s2 r

Current r-value TABLE 3.26 Second iteration simplex tableau

x1
0 0 1 0

x2
1.75 1.75 7.25 -0.38 1.75 Entering

x3 -7.5 5.5 0.75 -7.5

A1 1 0 0 0

s1 0 1 0 0

s2 -0.25 -0.75 0.13 -0.25

b 7.5

B.V. A

1 Leaving variable

38.5 6.25 7.5

s1 x1 r

Current r-value TABLE 3.27 Optimal Table of Phase 1 x3 -4.29 35.57 -0.86 0 A1 0.57 -4.14 0.21 -1 s1 0 1 0 0 s2 -0.14 0.29 0.07 0 b 4.29 7.43 7.86 0 B.V. x2 s1 x1 r

x1
0 0 1 0

x2
1 0 0 0

18

Optimum r-value Phase 2 The only one iteration of phase 2 is shown in Table 3.28. One can verify that Table 3.29 gives the optimal solution. The solution in Table 3.29 is optimal and feasible. TABLE 3.28 Initial Table of Phase 2 x3 -4.29 35.57 -0.86 -3 x3 -4.29 35.57 -0.86 9.86 s1 0 1 0 0 s1 0 1 0 0 s2 -0.14 0.29 0.07 0 s2 -0.14 0.29 0.07 0.93 b 4.29 7.43 7.86 0 b 4.29 7.43 7.86 22.14 B.V. x2 s1 x1 z B.V. x2 s1 x1 z

x1
0 0 1 -5

x2
1 0 0 4

TABLE 3.29 Optimal Table of Phase 2

x1
0 0 1 0

x2
1 0 0 0

Optimum z-value The optimal results are presented by x1=7.86, x2 =-4.29, x3 = 0 and max z=22.14. EXAMPLE 3.7 Use two-phase simplex method to solve the following LPP. Minimize z = 8x 1 + 4 x 2 + 2 x 3 Subject to 4x 1 + 2x 2 + x 3 3x 1 + 2x 2 x1 + x 2 + x 3 8 10 4

x1 , x 2 , x 3 0 SOLUTION The canonical form of the given problem is shown below: Minimize z = 8x 1 + 4 x 2 + 2 x 3 subject to

19

4x 1 + 2x 2 + x 3 + s1 3x 1 + 2x 2 + s2

=8 = 10

x 1 + x 2 + x 3 s 3 + A1 = 4 x 1 , x 2 , x 3 , s1 , s 2 , s 3 , A 1 0 Phase 1 The model for phase 1 with its revised objective function is shown below. The corresponding initial table is presented in Table 3.30. Further iterations towards optimality are shown in Tables 3.31 and 3.28. Auxilary objective function: Minimize r = A 1 = 4 x1 x 2 x 3 subject to 4x 1 + 2x 2 + x 3 + s1 3x 1 + 2x 2 + s2 =8 = 10

x 1 + x 2 + x 3 s 3 + A1 = 4 x 1 , x 2 , x 3 , s1 , s 2 , s 3 , A 1 0

TABLE 3.30 Initial Table (Example 3.7)

x1
4 3 1 1

x2
2 2 1 1

x3 1 0 1 1

s3 0 0 -1 -1

s1 1 0 0 0

s2 0 1 0 0

A1 0 0 1 0

b 8 10 4 4

B.V. s1 s1 A1 r

TABLE 3.31 First iteration simplex tableu

x1
4 3 1 0 Entering

x2
2 2 1 1

x3 1 0 1 1

s3 0 0 -1 -1

s1 1 0 0 0

s2 0 1 0 0

A1 0 0 1 0

b 8 4 4

B.V. s1

10

Leaving variable

s2 r

A1

Current r-value

20

TABLE 3.32 Second iteration simplex tableu

x1
1 0 0 0

x2
0.5 0.5 0.5 0.5

x3 0.25 0.75 0.75 0.75 Entering

s3 0 0 -1 -1

s1 0.25 -0.75 -0.25 -0.25

s2 0 1 0 0

A1 0 0 1 0

b 2 4 2 2

B.V. x1 s2 A1 r

Leaving variable

Current r-value

TABLE 3.33

Optimal Table of Phase 1 s2 0 1 0 0 A1 -0.33 1 1.33 -1 b 1.33 6 2.67 0 B.V. x1 s2 x3 r

x1
1 0 0 0

x2
0.33 1 0.67 0

x3 0 0 1 0

s3 0.33 -1 -1.33 0

s1 0.33 -1 -0.25 -0.33

Optimum r-value

Phase 2 The only one iteration of phase 2 is shown in Table 3.34. One can verify that Table 3.35, itself gives the optimal solution. The solution in Table 3.35 is optimal and feasible. TABLE 3.34 Initial Table of Phase 2

x1
1 0 0 -8

x2
0.33 1 0.67 -4

x3 0 0 1 -2

s3 0.33 -1 -1.33 0

s1 0.33 -1 -0.25 -0.33

s2 0 1 0 0

b 1.33 6 2.67 0

B.V. x1 s2 x3 z

TABLE 3.35 Optmal Table of Phase 2

x1

x2

x3

s3

s1

s2

B.V.

21

1 0 0 0

0.33 1 0.67 0

0 0 1 0

0.33 -1 -1.33 0

0.33 -1 -0.25 2

0 1 0 0

1.33 6 2.67 16

x1 s2 x3 z

The optimal results are presented by x1=1.33, x2 = 0, x3 = 2.67 and max z = 16. EXAMPLE 3.8 Use two-phase simplex method to solve Maximize Z = 5x1 + 3x2 Subject to 2x 1 + x 2 1 x 1 + 4x 2 6 x1 , x 2 0 SOLUTION The canonical form of the given problem is shown below: Maximize Z = 5x1 + 3x2 subject to 2 x 1 + x 2 + S1 =1 x1 , x 2 0 Phase 1 The model for phase 1 with its revised objective function is shown below. The corresponding initial table is presented in Table 3.36. Further iterations towards optimality are shown in Tables 3.38 and 3.28. Auxilary objective function: Minimize r = A 1 = 6 x 1 4x 2 + s 2 Subject to 2 x 1 + x 2 + S1 =1 x1 , x 2 0 Initial basic feasible solution is given by S1 =1, A1 = 6. x 1 + 4x 2 S 2 + A1 = 6 x 1 + 4x 2 S 2 + A1 = 6

22

TABLE 3.36

Initial Table (Example 3.8) A1 0 1 0 b 1 6 6 B.V. s1 A1 r

x1
2 1 1

x2
1 4 4

s2 0 -1 -1

s1 1 0 0

TABLE 3.37 First iteration simplex tableau

x1
2 1 1

x2
1 4 4

s2 0 -1 -1

s1 1 0 0

A1 0 1 0

b 1 6 6

B.V. s1 A1 r

TABLE 3.38

Optimal Table of Phase 1 b 2 1 2 B.V. x2 A1 r

x1
2 -7 -7

x2
1 0 0

s2 0 -1 -1

s1 1 -4 -4

A1 0 1 0

Since all coefficients of objective row are negative as shown in Table 3.38, z is an optimum feasible solution to the auxiliary LPP is obtained. But as Min r > 0, and an artificial variable A, is in the basis at a positive level, the original LPP does not possess any feasible solution. EXERCISES Use two-phase method to solve the following LPP. 1. Maximize z = 2 x 1 + x 2 + x 3 subject to 4 x 1 + 6x 2 + 3x 3 8 3x 1 6x 2 4x 3 1 2x 1 + 3x 2 5x 3 4 x1 , x 2 , x 3 0 [Ans: x1=9/7, x2 =10/21, x3=0 and max z= 64/21]

23

2. Maximize z = 2 x 1 + 3x 2 + 5x 3 subject to 3x 1 + 10x 2 + 5x 3 x 1 + 2x 2 + 15 x3 4

33x 1 10x 2 + 9x 3 33 x1 , x 2 , x 3 0 [Ans: Infeasible Solution.] 3. A company possesses two manufacturing plants, each of which can produce three products X, Y, Z from common raw material. However, the proportions in which the products are produced are different in each plant and so are the plants operating cost per hour. Data on production per hour and costs are given in Table 3.38 together with current orders in hand for each product. TABLE 3.39 Product X 2 4 50 Y 4 3 24 Z 3 2 60 Operating cost per hour (Rs) 9 10

Plant I Plant II Order on hand

Use simplex method to find the number of production hours needed to fulfil the orders on hand on a minimum cost. [Ans: Min z =195, x1 = 35/2, x2 =15/4] 4. Reduce the following LPP to the canonical form, so as to minimize z: Minimize z = 2x 1 + x 2 + 3x 3 subject to 5x 1 + 2 x 2 3x 1 + 2x 2 + 4 x 3 2 x 1 + 5x 2 x1 , x 2 , x 3 0 [Ans: Min z =4.88, x1 = 1.5, x2 = 0, x3 = 0.63] 5 7 3

24

5. Maximize z = 3x 1 + 5x 2 + 10x 3 subject to 3x 1 + 10 x 2 + 5x 3 x 1 + 2x 2 + 15 x3 4

33x 1 10x 2 + 9x 3 33 x1 , x 2 , x 3 0 [Ans: No feasible solution] 6. Maximize z = 3x 1 + 5x 2 + 4 x 3 subject to x 1 + x 2 + 5x 3 = 10 2 x 1 + x 2 + 2 x 3 16 x 1 + 4 x 2 + 3x 3 24 x1 , x 2 , x 3 0 [Ans: Min z = 39, x1 = 5, x2 = 4, x3 =1] 7. Use the two-phase method to

Minimize z = x 1 + x 2
subject to 2x 1 + x 2 4 x1 + 7x 2 7 x1 , x 2 0 [Ans: Min z = 31/13, x1 = 21/13, x2 = 10/13] 8. Solve the following linear programming problem, using the two phases of the simplex method

Minimize z = 2x 1 + x 2
subject to 5x 1 + 10 x 2 3x 3 = 8 x1 + x 2 + x 4 = 1 x1 , x 2 , x 3 , x 4 0 [Ans: Min z = 4/5, x1 = 0, x2 = 4/5, x3 =0, x4 = 1/5]

25

9. Use the two-phase method to solve the problem : . Maximize z = x 1 2 x 2 + 3x 3 Subject to 3x 1 + 10 x 2 + 5x 3 15 2x 1 + 1x 2 + 3x 3 = 2 2x 1 + 3x 2 + 4x 3 = 1 x1 , x 2 , x 3 0 (Ans: No feasible solution) 10. Minimize z = 12x 1 + 20x 2 subject to 6 x 1 + 8x 2 100 7 x 1 + 12x 2 120 x1 , x 2 0 Maximize z = 2x 1 + x 2 + 3x 3 11. subject to x 1 + x 2 + 2x 3

2 x 1 + 3x 2 + 4 x 3 = 12 x1 , x 2 , x 3 0 12. Minimize Z = 4 x1 + 3x 2 + x3 subject to x 1 + 2 x 2 + 4 x 3 12 3x 1 + 2 x 2 + x 3 8 x1 , x 2 , x 3 0 13. Consider the following problem. Solve it using any appropriate method. Max z = 3x 1 + 2 x 2 + 3x 3 subject to 2x 1 + x 2 + x 3 4 x 1 + 4x 2 + 2 x 3 12 2x 1 + 2x 2 + x 3 8 x1, x 2 , x 3 0

26

14. Using any appropriate method, solve the following problems. If it is not possible to arrive at an optimal solution, state clearly, the reasons for that and the characteristics that you see in your final solution. (i) (ii) Max z = 3x 1 + 2 x 2 + x 3 Max z = 2x 1 + x 2 subject to x 1 x 2 10 2 x 1 40 x1 , x 2 0 (iii) Max z = 2 x 1 x 2 + 3x 3 subject to x 1 x 2 + 5x 3 10 2 x 1 x 2 + 3x 3 40 x 1 , x 2 ,x 3 0 15. A company manufactures three types of leather belts, namely A, B and C. The unit profits from then three varieties are Rs. 10, Rs. 5 and Rs. 7 respectively. Leather is sufficient for only 800 belts per day (all types together). Belt A requires thrice the time of belt B and belt C requires twice the tim of B. If all the belts of type B are produced, a maximum of 1,000 belts per day can be produced. Belt A requires a fancy buckle and 150 such buckles per day are available. There are sufficient number buckles for the other varieties. Determine how many belts of each type be produced to maximize the total profit. 16. A factory works 8 hours a day, producing three products viz. A, B and C. Each of these products is processed in three different operations viz. 1, 2 and 3. The processing times in minutes for each of these products in each of the operations are given in Table 3.40 along with utilisation of the processes and the cost and price in rupees for each of these three products which have unlimited demand. TABLE 3.40 Processing time (rains.) in operation Product A B C Utilisation 1 4 2 3 80% 2 3 1 4 70% 3 1 4 5 90% Cost/unit (Rs.) 10 8 5 Price/unit (Rs.) 16 12 10 subject to 2x 1 + x 2 2

3x 1 + 4 x 2 + x 3 12 x1 , x 2 , x 3 0

(i) Determine the optimal product mix using the simplex method. (ii) Give the interpretations for the values obtained in the final simplex table.

27

17. Three types of cutting tools are produced in a factory using a lathe, a grinder and a polisher. The duration in hours required to produce one batch of tools on each of these machines are given in the following table 3.41 along with costs, selling prices of each batch of tools and the minimum number of hours available on each machine per week. TABLE 3.41 Tool type A B C per week Processing time (hrs.) per batch on a Lathe 7 3 4 50 Grinder 2 3 4 46 Polisher 5 8 2 80 Cost the (Rs.) per batch 100 65 80 Selling prices (Rs.) per batch 145 100 120

(i) Determine the optimum production schedule and maximum profit per we7k. (ii) The factory wants to double the capacity of grinder at an additional c6st of Rs. 50 per week. Should it go for it ? Substantiate your result. 18. A manufacturer produces three products A, B and C. Each product requires processing on two machines I and Il. The time required to produce one unit of each product on a machine is TABLE 3.42 Product A B C Time to produce one unit (hrs.) Machine I 0.5 0.7 0.9 Machine II 0.6 0.8 1.05

There are 850 hours available on each machine. The operating cost is Rs. 5/hr. for machine I and Rs. 4/hr. for machine II. The market requirements are at least 90 units of A, at least 80 units of B and at least 60 units of C. The manufacturer wishes to meet the requirements at minimum cost. Solve the problem by the simplex method. Minimize z = (0.5x5 + 0.6 x 4) x 1 + (0.7 x5 + 0.7 x 4) x 2 + (0.9 x 5 + 1.05x 4) x 3 = 4.9 x 1 + 6.7 x 2 + 8.7) x 3

28

Subject to 0.5x 1 + 0.7 x 2 + 0.9x 3 850 0.6x 1 + 0.8x 2 + 1.05x 3 850 x1 x2 90 80 x 3 60 x1 , x 2 , x 3 0 SELF TEST 1.A basic feasible solution is a solution to an LP problem that corresponds to a corner point of the feasible region. a. True b. False 2. In preparing a ~! constraint for an initial simplex tableau, we would a. add a slack variable. b. add a surplus variable. c. subtract an artificial variable. d. subtract a surplus variable and add an artificial variable. 3. In the initial simplex tableau, the solution mix variables can be a. only slack variables. b. slack and surplus variables. c. artificial and surplus variables. d. slack and artificial variables. 4. Even if an LP problem involves many variables, an optimal solution will always be found at a corner point of the n-dimensional polyhedron forming the feasible region. a. True b. False 5. Which of the following in a simplex tableau indicates that an optimal solution for a maximization problem has been found? a. all the Cj Zj values are negative or zero b. all the C. Z. values are positive or zero c. all the substitution rates in the pivot column are negative or zero

29

d. there are no more slack variables in the solution mix 6. To formulate a problem for solution by the simplex method, we must add slack variables to a. all inequality constraints. b. only equality constraints. c. only "greater than" constraints. d. only "less than" constraints. 7. If in the optimal tableau of an LP problem an artificial variable is present in the solution mix, this implies a. infeasibility. b. unboundedness. c. degeneracy. d. alternate optimal solutions. 8. If in the final optimal simplex tableau the C. Z. value for a nonbasic variable is zero, this implies a. feasibility. b. unboundedness. c. degeneracy. d. alternate optimal solutions. 9. In a simplex tableau, all of the substitution rates in the pivot column are negative. This indicates a. there is no feasible solution to this problem. b. the solution is unbounded.