Management Science Module 7

Z00_REND1011_11_SE_MOD7 PP2.
QXD 2/21/11 12:39 PM Page 1
MODULE 7
Linear Programming:
The Simplex Method
LEARNING OBJECTIVES
After completing this chapter, students will be able to:
1. Convert LP constraints to equalities with slack, 4. Recognize special cases such as infeasibility,
surplus, and artificial variables. unboundedness and degeneracy.
2. Set up and solve LP problems with simplex tableaus. 5. Use the simplex tables to conduct sensitivity
3. Interpret the meaning of every number in a simplex analysis.
tableau. 6. Construct the dual problem from the primal problem.
CHAPTER OUTLINE
M7.1 Introduction M7.8 Solving Minimization Problems
M7.2 How to Set Up the Initial Simplex Solution M7.9 Review of Procedures for Solving LP
M7.3 Simplex Solution Procedures Minimization Problems
M7.4 The Second Simplex Tableau M7.10 Special Cases
M7.5 Developing the Third Tableau M7.11 Sensitivity Analysis with the Simplex Tableau
M7.6 Review of Procedures for Solving LP Maximization M7.12 The Dual
Problems M7.13 Karmarkar’s Algorithm
M7.7 Surplus and Artificial Variables
Summary • Glossary • Key Equation • Solved Problems • Self-Test •

Discussion Questions and Problems • Bibliography
M7-1
Z00_REND1011_11_SE_MOD7 PP2.QXD 2/21/11 12:39 PM Page 2
M7-2 MODULE 7 • LINEAR PROGRAMMING: THE SIMPLEX METHOD
M7.1 Introduction
In Chapter 7 we looked at examples of linear programming (LP) problems that contained two
decision variables. With only two variables it is possible to use a graphical approach. We plotted
the feasible region and then searched for the optimal corner point and corresponding profit or
cost. This approach provides a good way to understand the basic concepts of LP. Most real-life
LP problems, however, have more than two variables and are thus too large for the simple graph-
ical solution procedure. Problems faced in business and government can have dozens, hundreds,
or even thousands of variables. We need a more powerful method than graphing, so in this chap-
ter we turn to a procedure called the simplex method.
Recall that the theory of LP How does the simplex method work? The concept is simple, and it is similar to graphical
states the optimal solution will LP in one important respect. In graphical LP we examine each of the corner points; LP theory
lie at a corner point of the tells us that the optimal solution lies at one of them. In LP problems containing several variables,
feasible region. In large LP we may not be able to graph the feasible region, but the optimal solution will still lie at a corner
problems, the feasible region point of the many-sided, many-dimensional figure (called an n-dimensional polyhedron) that
cannot be graphed because it has
represents the area of feasible solutions. The simplex method examines the corner points in a
many dimensions, but the
concept is the same.
systematic fashion, using basic algebraic concepts. It does so in an iterative manner, that is,
repeating the same set of procedures time after time until an optimal solution is reached. Each
iteration brings a higher value for the objective function so that we are always moving closer to
the optimal solution.
The simplex method Why should we study the simplex method? It is important to understand the ideas used to
systematically examines corner produce solutions. The simplex approach yields not only the optimal solution to the decision
points, using algebraic steps, variables and the maximum profit (or minimum cost), but valuable economic information as
until an optimal solution is well. To be able to use computers successfully and to interpret LP computer printouts, we need
found. to know what the simplex method is doing and why.
We begin by solving a maximization problem using the simplex method. We then tackle a
minimization problem and look at a few technical issues that are faced when employing the sim-
plex procedure. From there we examine how to conduct sensitivity analysis using the simplex
tables. The chapter concludes with a discussion of the dual, which is an alternative way of look-
ing at any LP problem.
M7.2 How to Set Up the Initial Simplex Solution

Let us consider the case of the Flair Furniture Company from Chapter 7. Instead of the graphi-
cal solution we used in that chapter, we now demonstrate the simplex method. You may recall
that we let
T = number of tables produced
C = number of chairs produced
and that the problem was formulated as
Maximize profit = $70T + $50C (objective function)
subject to 2T + 1C … 100 (painting hours constraint)
4T + 3C … 240 (carpentry hours constraint)
T, C Ú 0 (nonnegativity constraints)
M7.2 HOW TO SET UP THE INITIAL SIMPLEX SOLUTION M7-3
Converting the Constraints to Equations

Slack variables are added to The first step of the simplex method requires that we convert each inequality constraint (except
each less-than-or-equal-to nonnegativity constraints) in an LP formulation into an equation.1 Less-than-or-equal-to con-
constraint. Each slack variable straints 1…2 such as in the Flair problem are converted to equations by adding a slack variable
represents an unused resource. to each constraint. Slack variables represent unused resources; these may be in the form of time
on a machine, labor hours, money, warehouse space, or any number of such resources in various
business problems.
In our case at hand, we can let
S1 = slack variable representing unused hours in the painting department
S2 = slack variable representing unused hours in the carpentry department
The constraints to the problem may now be written as
2T + 1C + S1 = 100
and
4T + 3C + S2 = 240
Thus, if the production of tables (T) and chairs (C) uses less than 100 hours of painting time
available, the unused time is the value of the slack variable, S1. For example, if T = 0 and
C = 0 (in other words, if nothing is produced), we have S1 = 100 hours of slack time in the
painting department. If Flair produces T = 40 tables and C = 10 chairs, then
2T + 1C + S1 = 100
21402 + 11102 + S1 = 100
S1 = 10
and there will be 10 hours of slack, or unused, painting time available.
To include all variables in each equation, which is a requirement of the next simplex step,
slack variables not appearing in an equation are added with a coefficient of 0. This means, in ef-
fect, that they have no influence on the equations in which they are inserted; but it does allow us
to keep tabs on all variables at all times. The equations now appear as follows:
2T + 1C + 1S1 + 0S2 = 100
4T + 3C + 0S1 + 1S2 = 240
T, C, S1, S2 Ú 0
Because slack variables yield no profit, they are added to the original objective function with 0
profit coefficients. The objective function becomes
Maximize profit = $70T + $50C + $0S1 + $0S2
Finding an Initial Solution Algebraically

Let’s take another look at the new constraint equations. We see that there are two equations and
four variables. Think back to your last algebra course. When you have the same number of un-
known variables as you have equations, it is possible to solve for unique values of the variables.
But when there are four unknowns (T, C, S1, and S2, in this case) and only two equations, you can
1This is because the simplex is a matrix algebra method that requires all mathematical relationships to be equations,
with each equation containing all of the variables.
FIGURE M7.1
Corner Points of the Flair C
Furniture Company
Problem
100
B = (0, 80)
80
Number of Chairs
2T + 1C 100
60
40 C = (30, 40)
4T + 3C 240
20
D = (50, 0)
(0,0) A
0 20 40 60 80 T
Number of Tables
let two of the variables equal 0 and then solve for the other two. For example, if T = C = 0, then
A basic feasible solution to a S1 = 100 and S2 = 240. A solution found in this manner is called a basic feasible solution.
system of n equations is found by The simplex method begins with an initial feasible solution in which all real variables (such
setting all but n variables equal as T and C) are set equal to 0. This trivial solution always produces a profit of $0, as well as
to 0 and solving for the other slack variables equal to the constant (right-hand-side) terms in the constraint equations. It’s not
variables. a very exciting solution in terms of economic returns, but it is one of the original corner point
solutions (see Figure M7.1). As mentioned, the simplex method will start at this corner point (A)
and then move up or over to the corner point that yields the most improved profit (B or D). Fi-
nally, the technique will move to a new corner point (C), which happens to be the optimal solu-
Simplex considers only corner tion to the Flair Furniture problem. The simplex method considers only feasible solutions and
points as it seeks the best hence will touch no possible combinations other than the corner points of the shaded region in
solution. Figure M7.1.
The First Simplex Tableau

To simplify handling the equations and objective function in an LP problem, we place all of the
coefficients into tabular form. The first simplex tableau is shown in Table M7.1. An explanation
of its parts and how the tableau is derived follows.
CONSTRAINT EQUATIONS We see that Flair Furniture’s two constraint equations can be ex-
pressed as follows:
SOLUTION QUANTITY
MIX T C S1 S2 (RIGHT-HAND SIDE)
Here are the constraints in
tabular form. S1 2 1 1 0 100
S2 4 3 0 1 240
TABLE M7.1 Flair Furniture’s Initial Simplex Tableau
ns
n
ns
um
n
um
lum
um
col
col
col
t co
n
um
ix
les
es
nm
uni
col
iab
abl
tio
per
var
ant
ari
duc
al v
nst
fit
ck
Pro
Pro
Sla
Co
Re
Cj SOLUTION $70 $50 $0 $0 Profit per unit row
MIX T C S1 S2 QUANTITY
$0 S1 2 1 1 0 100 Constraint equation rows
$0 S2 4 3 0 1 240
Zj $0 $0 $0 $0 $0 Gross profit row
Cj - Zj $70 $50 $0 $0 $0 Net profit row
The numbers 12, 1, 1, 02 in the first row represent the coefficients of the first equation,
namely, 2T + 1C + 1S1 + 0S2. The numbers 14, 3, 0, 12 in the second row are the algebraic
equivalent of the constraint 4T + 3C + 0S1 + 1S2.
The initial solution mix begins As suggested earlier, we begin the initial solution procedure at the origin, where T = 0 and
with real, or decision, variables C = 0. The values of the other two variables must then be nonzero, so S1 = 100 and S2 = 240.
set equal to zero. These two slack variables constitute the initial solution mix; their values are found in the
quantity (or right-hand-side [RHS]) column. Because T and C are not in the solution mix, their
initial values are automatically equal to 0.
This initial solution is a basic feasible solution and is described in vector, or column,
form as
T 0
Here is the basic feasible solution C 0
in column form. D T = D T
S1 100
S2 240
Variables in the solution mix are Variables in the solution mix, which is called the basis in LP terminology, are referred to as ba-
called basic. Those not in the sic variables. In this example, the basic variables are S1 and S2. Variables not in the solution mix
solution are called nonbasic. or basis and that have been set equal to zero (T and C in this case), are called nonbasic variables.
Of course, if the optimal solution to this LP problem turned out to be T = 30, C = 40, S1 = 0,
and S2 = 0, or
T 30
D T = D T 1in vector form2

C 40
S1 0
S2 0
then T and C would be the final basic variables, and S1 and S2 would be the nonbasic variables.
Notice that for any corner point, exactly two of the four variables will equal zero.
SUBSTITUTION RATES Many students are unsure as to the actual meaning of the numbers in the
columns under each variable. We know that the entries are the coefficients for that variable.
Under T are the coefficients a b, under C are a b, under S1 are a b , and under S2 are a b.
2 1 1 0
4 3 0 1
Substitution rates are numbers But what is their interpretation? The numbers in the body of the simplex tableau (see Table M7.1)
in the body of the table. This can be thought of as substitution rates. For example, suppose we now wish to make T larger than
paragraph explains how to 0, that is, produce some tables. For every unit of the T product introduced into the current solu-
interpret their meaning. tion, 2 units of S1 and 4 units of S2 must be removed from the solution. This is so because each
table requires 2 hours of the currently unused painting department slack time, S1. It also takes 4
hours of carpentry time; hence 4 units of variable S2 must be removed from the solution for
every unit of T that enters. Similarly, the substitution rates for each unit of C that enters the
current solution are 1 unit of S1 and 3 units of S2.
Another point that you are reminded of throughout this chapter is that for any variable ever
to appear in the solution mix column, it must have the number 1 someplace in its column and 0s
in every other place in that column. We see that column S1 contains a b, so variable S1 is in the
1
0
solution. Similarly, the S2 column is a b , so S2 is also in the solution.2
0
1
ADDING THE OBJECTIVE FUNCTION We now continue to the next step in establishing the first
simplex tableau. We add a row to reflect the objective function values for each variable. These
contribution rates, called Cj, appear just above each respective variable, as shown in the follow-
ing table:
Cj $70 $50 $0 $0
SOLUTION
$0 S1 2 1 1 0 100
$0 S2 4 3 0 1 240
The unit profit rates are not just found in the top Cj row: in the leftmost column, Cj indicates the
unit profit for each variable currently in the solution mix. If S1 were removed from the solution
and replaced, for example, by C, $50 would appear in the Cj column just to the left of the term C.
THE Zj AND Cj Zj ROWS We can complete the initial Flair Furniture simplex tableau by adding
two final rows. These last two rows provide us with important economic information, including
the total profit and the answer as to whether the current solution is optimal.
We compute the Zj value for each column of the initial solution in Table M7.1 by multiply-
ing the 0 contribution value of each number in the Cj column by each number in that row and
2Ifthere had been three less-than-or-equal-to constraints in the Flair Furniture problem, there would be three slack vari-
ables, S1, S2, and S3. The 1s and 0s would appear like this:
SOLUTION MIX S1 S2 S3
S1 1 0 0
S2 0 1 0
S3 0 0 1
IN ACTION Resource Allocation at Pantex
C ompanies often use optimization techniques such as LP to

allocate limited resources to maximize profits or minimize costs.
tex must allocate scarce resources among competing demands,
all of which are important.
The team charged with solving the resource allocation prob-
One of the most important resource allocation problems faced lem at Pantex developed the Pantex Process Model (PPM). PPM is
by the United States is dismantling old nuclear weapons and a sophisticated optimization system capable of analyzing nuclear
maintaining the safety, security, and reliability of the remaining needs over different time horizons. Since its development, PPM
systems. This is the problem faced by Pantex, a $300 million has become the primary tool for analyzing, planning, and sched-
corporation. uling issues at Pantex. PPM also helps to determine future re-
Pantex is responsible for disarming, evaluating, and maintain- sources. For example, it was used to gain government support for
ing the U.S. nuclear stockpile. The company is also responsible for $17 million to modify an existing plant with new buildings and
storing critical weapon components that relate to U.S.–Russian $70 million to construct a new plant.
nonproliferation agreements. Pantex constantly makes trade-offs
in meeting the requirements of disarming some nuclear weapons Source: Based on Edwin Kjeldgaard, et al. “Swords into Plowshares: Nuclear
versus maintaining existing nuclear weapon systems, while effec- Weapon Dismantlement, Evaluation, and Maintenance at Pantex,” Interfaces
tively allocating limited resources. Like many manufacturers, Pan- 30, 1 (January–February 2000): 57–82.
The Z-row entry in the quantity the jth column, and summing. The Zj value for the quantity column provides the total contribu-
column provides the gross profit. tion (gross profit in this case) of the given solution:
Zj 1for gross profit2 = 1Profit per unit of S12 * 1Number of units of S12
+ 1Profit per unit of S22 * 1Number of units of S22
= $0 * 100 units + $0 * 240 units
= $0 profit
The Zj values for the other columns (under the variables T, C, S1, and S2) represent the gross
profit given up by adding one unit of this variable into the current solution. Their calculations
are as follows:
Zj = 1Profit per unit of S12 * 1Substitution rate in row 12
+ 1Profit per unit of S22 * 1Substitution rate in row 22
Thus,
Zj 1for column T2 = 1$02122 + 1$02142 = $0
Zj 1for column C2 = 1$02112 + 1$02132 = $0
Zj 1for column S12 = 1$02112 + 1$02102 = $0
Zj 1for column S22 = 1$02102 + 1$02112 = $0
We see that there is no profit lost by adding one unit of either T (tables), C (chairs), S1, or S2.
The Cj - Zj row gives the net The Cj - Zj number in each column represents the net profit, that is, the profit gained
profit from introducing one unit minus the profit given up, that will result from introducing 1 unit of each product or variable
of each variable into the solution. into the solution. It is not calculated for the quantity column. To compute these numbers, simply
subtract the Zj total for each column from the Cj value at the very top of that variable’s column.
The calculations for the net profit per unit (the Cj - Zj row) in this example follow:
COLUMN
T C S1 S2
Cj for column $70 $50 $0 $0
Zj for column 0 0 0 0
Cj Zj for column $70 $50 $0 $0
It is obvious to us when we compute a profit of $0 that the initial solution is not optimal. By
examining the numbers in the Cj - Zj row of Table M7.1, we see that the total profit can be
We reach an optimal solution increased by $70 for each unit of T (tables) and by $50 for each unit of C (chairs) added to the
when the Cj - Zj row has no solution mix. A negative number in the Cj - Zj row would tell us that profits would decrease if
positive numbers in it. the corresponding variable were added to the solution mix. An optimal solution is reached in the
simplex method when the Cj - Zj row contains no positive numbers. Such is not the case in our
initial tableau.
M7.3 Simplex Solution Procedures

Here are the five simplex steps. After an initial tableau has been completed, we proceed through a series of five steps to com-
pute all the numbers needed in the next tableau. The calculations are not difficult, but they are
complex enough that even the smallest arithmetic error can produce a wrong answer.
We first list the five steps and then carefully explain and apply them in completing the sec-
ond and third tableaus for the Flair Furniture Company data.
Five Steps of the Simplex Method for Maximization Problems

1. Variable entering the solution 1. Determine which variable to enter into the solution mix next. One way of doing this is by
has the largest positive Cj - Zj. identifying the column, and hence the variable, with the largest positive number in the
Cj - Zj row of the preceding tableau. This means that we will now be producing some of
the product contributing the greatest additional profit per unit. The column identified in this
step is called the pivot column.
2. Variable leaving the solution is 2. Determine which variable to replace. Because we have just chosen a new variable to enter
determined by a ratio we must the solution mix, we must decide which basic variable currently in the solution will have to
compute. leave to make room for it. Step 2 is accomplished by dividing each amount in the quantity
column by the corresponding number in the column selected in step 1. The row with the
smallest nonnegative number calculated in this fashion will be replaced in the next tableau.
(This smallest number, by the way, gives the maximum number of units of the variable that
may be placed in the solution.) This row is often referred to as the pivot row. The number
at the intersection of the pivot row and pivot column is referred to as the pivot number.
3. New pivot-row calculations are 3. Compute new values for the pivot row. To do this, we simply divide every number in the
done next. row by the pivot number.
4. Other new rows are calculated 4. Compute the new values for each remaining row. (In our Flair Furniture problem there are
with Equation M7-1. only two rows in the LP tableau, but most larger problems have many more rows.) All
remaining row(s) are calculated as follows:
1New row numbers2 = 1Numbers in old row2
Number above Corresponding number in
- C £ or below ≥ * £ the new row, that is, the ≥ S (M7-1)
pivot number row replaced in step 3
5. Finally Zj and Cj - Zj rows 5. Compute the Zj and Cj - Zj rows, as demonstrated in the initial tableau. If all numbers in
are recomputed. the Cj - Zj row are 0 or negative, an optimal solution has been reached. If this is not the
case, return to step 1.
M7.4 THE SECOND SIMPLEX TABLEAU M7-9
M7.4 The Second Simplex Tableau

Here we apply the five steps to Now that we have listed the five steps needed to move from an initial solution to an improved
Flair Furniture. solution, we apply them to the Flair Furniture problem. Our goal is to add a new variable to the
solution mix, or basis, to raise the profit from its current tableau value of $0.
First, T (tables) enters the Step 1. To decide which of the variables will enter the solution next (it must be either T or C,
solution mix because its Cj - Zj since they are the only two nonbasic variables at this point), we select the one with the largest
value of $70 is largest. positive Cj - Zj value. Variable T, tables, has a Cj - Zj value of $70, implying that each unit
of T added into the solution mix will contribute $70 to the overall profit. Variable C, chairs, has
a Cj - Zj value of only $50. The other two variables, S1 and S2, have 0 values and can add noth-
ing more to profit. Hence, we select T as the variable to enter the solution mix and identify its
column (with an arrow) as the pivot column. This is shown in Table M7.2.
Step 2. Since T is about to enter the solution mix, we must decide which variable is to be re-
placed. There can only be as many basic variables as there are constraints in any LP problem, so
either S1 or S2 will have to leave to make room for the introduction of T, tables, into the basis.
To identify the pivot row, each number in the quantity column is divided by the corresponding
number in the T column.
For the S1 row:
100 1hours of painting time available2
2 1hours required per table2
= 50 tables
For the S2 row:

240 1hours of carpentry time available2
4 1hours required per table2
= 60 tables
The smaller of these two ratios, 50, indicates the maximum number of units of T that can be
produced without violating either of the original constraints. This corresponds to point D in
S1 leaves the solution mix Figure M7.2. The other ratio (60) corresponds to point E on this graph. Thus, the smallest ratio
because the smaller of the two is chosen so that the next solution is feasible. Also, when T = 50, there is no slack in constraint
ratios indicates that the next 1, so S1 = 0. This means that S1 will be the next variable to be replaced at this iteration of the
pivot row will be the first row. simplex method. The row with the smallest ratio (row 1) is the pivot row. The pivot row and the
pivot number (the number at the intersection of the pivot row and pivot column) are identified in
Table M7.3.
Step 3. Now that we have decided which variable is to enter the solution mix (T) and which
is to leave 1S12, we begin to develop the second, improved simplex tableau. Step 3 involves
TABLE M7.2
Cj $70 $50 $0 $0
Pivot Column Identified
in the Initial Simplex SOLUTION QUANTITY
MIX T C S1 S2 (RHS)
Tableau
$0 S1 2 1 1 0 100
$0 S2 4 3 0 1 240
Zj $0 $0 $0 $0 $0
Cj Zj $70 $50 $0 $0 total profit
Pivot column
FIGURE M7.2
Graph of the Flair C
Furniture Company
Problem F = (0, 100)
100
B = (0, 80)
80
Number of Chairs
60
40 C = (30, 40)
20 D = (50, 0)
E = (60, 0)
(0,0) A
0 20 40 60 80 T
Number of Tables
computing a replacement for the pivot row. This is done by dividing every number in the pivot
The new pivot row is computed row by the pivot number:
by dividing every number in the
pivot row by the pivot number. 2 1 1 0 100
= 1 = 0.5 = 0.5 = 0 = 50
2 2 2 2 2
The new version of the entire pivot row appears in the accompanying table. Note that T is now
in the solution mix and that 50 units of T are being produced. The Cj value is listed as a $70 con-
tribution per unit of T in the solution. This will definitely provide Flair Furniture with a more
profitable solution than the $0 generated in the initial tableau.
Cj SOLUTION MIX T C S1 S2 QUANTITY

$70 T 1 0.5 0.5 0 50
TABLE M7.3
Cj $70 $50 $0 $0
Pivot Row and Pivot
Number Identified in SOLUTION
the Initial Simplex
Tableau $0 S1 2 1 1 0 100 Pivot row
$0 S2 4 3 0 1 240
Pivot number
Zj $0 $0 $0 $0 $0
Cj Zj $70 $50 $0 $0
Pivot column
M7.4 THE SECOND SIMPLEX TABLEAU M7-11
Step 4. This step is intended to help us compute new values for the other row in the body of the
tableau, that is, the S2 row. It is slightly more complex than replacing the pivot row and uses the
formula (Equation M7.1) shown earlier. The expression on the right side of the following equa-
tion is used to calculate the left side.
We can now recompute the S2

a b ⴝ a b ⴚ ca b : a bd
Number in Number in Number Below Corresponding Number
row.
New S2 Row Old S2 Row Pivot Number in the New T Row
0 = 4 - (4) * (1)
1 = 3 - (4) * (0.5)
2 = 0 - (4) * (0.5)
1 = 1 - (4) * (0)
40 = 240 - (4) * (50)
This new S2 row will appear in the second tableau in the following format:

$70 T 1 0.5 0.5 0 50
$0 S2 0 1 2 1 40
Now that T and S2 are in the solution mix, take a look at the values of the coefficients in their re-
spective columns. The T column contains a b, a condition necessary for that variable to be in
1
We note that the T column
0
contains a b and the S2 column
1
the solution. Similarly, the S2 column has a b, that is, it contains a 1 and a 0. Basically, the
0
0
1
contains a b . These 0s and 1s
0
1 algebraic manipulations we just went through in steps 3 and 4 were simply directed at producing
indicate that T and S2 are in the 0s and 1s in the appropriate positions. In step 3 we divided every number in the pivot row by the
pivot number; this guaranteed that there would be a 1 in the T column’s top row. To derive the
basis (the solution mix).
new second row, we multiplied the first row (each row is really an equation) by a constant
(the number 4 here) and subtracted it from the second equation. The result was the new S2 row
with a 0 in the T column.
Step 5. The final step of the second iteration is to introduce the effect of the objective function.
This involves computing the Zj and Cj - Zj rows. Recall that the Zj entry for the quantity col-
umn gives us the gross profit for the current solution. The other Zj values represent the gross
We find the new profit in the
profit given up by adding one unit of each variable into this new solution. The Zj values are cal-
Z row.
culated as follows:
Zj 1for T column2 = 1$702112 + 1$02102 = $70
Zj 1for C column2 = 1$70210.52 + 1$02112 = $35
Zj 1for S1 column2 = 1$70210.52 + 1$021-22 = $35
Zj 1for S2 column2 = 1$702102 + 1$02112 = $0
Zj 1for total profit2 = 1$7021502 + 1$021402 = $3,500
Note that the current profit is $3,500.
TABLE M7.4
Cj $70 $50 $0 $0
Completed Second
Simplex Tableau for SOLUTION
Flair Furniture
$70 T 1 0.5 0.5 0 50
$0 S2 0 1 2 1 40
Zj $70 $35 $35 $0 $3,500
Cj Zj $0 $15 $35 $0
The Cj - Zj numbers represent the net profit that will result, given our present production
The Cj - Zj row indicates the mix, if we add one unit of each variable into the solution:
net profit, given the current
solution, of one more unit of
each variable. For example, C COLUMN
has a profit of $15 per unit. T C S1 S2
Zj for column $70 $35 $35 $0
The Zj and Cj - Zj rows are inserted into the complete second tableau as shown in Table M7.4.
Interpreting the Second Tableau

Table M7.4 summarizes all of the information for the Flair Furniture Company’s production mix
decision as of the second iteration of the simplex method. Let’s briefly look over a few impor-
tant items.
We can look at the current CURRENT SOLUTION At this point, the solution point of 50 tables and 0 chairs 1T = 50, C = 02
solution as a corner point in the generates a profit of $3,500. T is a basic variable; C is a nonbasic variable. Using a graphical LP
graphical method. approach, this corresponds to corner point D, as shown earlier in Figure M7.2.
RESOURCE INFORMATION We also see in Table M7.4 that slack variable S2, representing the
amount of unused time in the carpentry department, is in the basis. It has a value of 40, implying
that 40 hours of carpentry time remain available. Slack variable S1 is nonbasic and has a value
of 0 hours. There is no slack time in the painting department.
SUBSTITUTION RATES We mentioned earlier that the substitution rates are the coefficients in the
heart of the tableau. Look at the C column. If 1 unit of C (1 chair) is added to the current solu-
Here is an explanation of the tion, 0.5 units of T and 1 unit of S2 must be given up. This is because the solution T = 50 tables
meaning of substitution rates. uses up all 100 hours of time in the painting department. (The original constraint, you may re-
call, was 2T + 1C + S1 = 100.) To capture the 1 painting hour needed to make 1 chair, 0.5 of
a table less must be produced. This frees up 1 hour to be used in making 1 chair.
But why must 1 unit of S2 (i.e., 1 hour of carpentry time) be given up to produce 1 chair?
The original constraint was 4T + 3C + S2 = 240 hours of carpentry time. Doesn’t this indi-
cate that 3 hours of carpentry time are required to produce 1 unit of C? The answer is that we
are looking at marginal rates of substitution. Adding 1 chair replaced 0.5 table. Because 0.5
table required 10.5 * 4 hours per table2 = 2 hours of carpentry time, 2 units of S2 are freed.
Thus, only 1 more unit of S2 is needed to produce 1 chair.
M7.5 DEVELOPING THE THIRD TABLEAU M7-13
Just to be sure you have this concept down pat, let’s look at one more column, S1, as well.
The coefficients are a b . These substitution rate values mean that if 1 hour of slack painting
0.5
-2
time is added to the current solution, 0.5 of a table (T) less will be produced. However, note that
if 1 unit of S1 is added into the solution, 2 hours of carpentry time 1S22 will no longer be used.
These will be added to the current 40 slack hours of carpentry time. Hence, a negative substitu-
tion rate means that if 1 unit of a column variable is added to the solution, the value of the corre-
sponding solution (or row) variable will be increased. A positive substitution rate tells us that if
1 unit of the column variable is added to the solution, the row variable will decrease by the rate.
Can you interpret the rates in the T and S2 columns now?
NET PROFIT ROW The Cj - Zj row is important to us for two reasons. First, it indicates whether
the current solution is optimal. When there are no positive numbers in the bottom row, an opti-
mum solution to an LP maximization problem has been reached. In the case of Table M7.4, we
The Cj - Zj row tells us (1) see that Cj - Zj values for columns T, S1, and S2 are 0 or negative. The value for column C (15)
whether the current solution is means that the net profit can be increased by $15 for each chair added into the current solution.
optimal and (2) if it is not, which Because the Cj - Zj value for T is 0, for every unit of T added the total profit will remain
variable should enter the solution unchanged, because we are already producing as many tables as possible. A negative number,
mix next. such as the -35 in the S1 column, implies that total profit will decrease by $35 if 1 unit of S1 is
added to the solution. In other words, making one slack hour available in the painting depart-
ment (S1 = 0 currently) means that we would have to produce one-half table less. Since each
table results in a $70 contribution, we would be losing 0.5 * $70 = $35, for a net loss of $35.
Later in this chapter we discuss in detail the subject of shadow prices. These relate to
Cj - Zj values in the slack variable columns. Shadow prices are simply another way of inter-
preting negative Cj - Zj values; they may be viewed as the potential increase in profit if one
more hour of the scarce resource (such as painting or carpentry time) could be made available.
We mentioned previously that there are two reasons to consider the Cj - Zj row carefully.
The second reason, of course, is that we use the row to determine which variable will enter the
solution next. Since an optimal solution has not been reached yet, let’s proceed to the third sim-
plex tableau.
M7.5 Developing the Third Tableau

Since not all numbers in the Cj - Zj row of the latest tableau are 0 or negative, the previous so-
lution is not optimal, and we must repeat the five simplex steps.
C (chairs) will be the next Step 1. Variable C will enter the solution next by virtue of the fact that its Cj - Zj value of 15
solution mix variable because it is the largest (and only) positive number in the row. This means that for every unit of C (chairs)
has the only positive value in the we start to produce, the objective function will increase in value by $15. The C column is the
Cj - Zj row. new pivot column.
Step 2. The next step involves identifying the pivot row. The question is, which variable cur-
rently in the solution (T or S2) will have to leave to make room for C to enter? Again, each num-
ber in the quantity column is divided by its corresponding substitution rate in the C column:
50
For the T row: = 100 chairs
0.5
40
For the S2 row: = 40 chairs
1
TABLE M7.5
Cj $70 $50 $0 $0
Pivot Row, Pivot
Column, and Pivot SOLUTION
Number Identified in
the Second Simplex $70 T 1 0.5 0.5 0 50
Tableau $0 S2 0 1 2 1 40 Pivot row
Pivot number
Zj $70 $35 $35 $0 $3,500
Cj Zj $0 $15 $35 $0 (total profit)
Pivot column
We replace the variable S2 These ratios correspond to the values of C (the variable entering the solution mix) at points
because it is in the pivot row. F and C seen earlier in Figure M7.2. The S2 row has the smallest ratio, so variable S2 will leave
the basis (and will become a nonbasic variable equal to zero) and will be replaced by C (which
will have a value of 40). The new pivot row, pivot column, and pivot number are all shown in
Table M7.5.
The pivot row for the third Step 3. The pivot row is replaced by dividing every number in it by the (circled) pivot number.
tableau is replaced here. Since every number is divided by 1, there is no change:
0 1 -2 1 40
= 0 = 1 = -2 = 1 = 40
1 1 1 1 1
The entire new C row looks like this:

$50 C 0 1 2 1 40
It will be placed in the new simplex tableau in the same row position that S2 was in before (see
Table M7.5).
Step 4. The new values for the T row may now be computed:
Number Number Number Corresponding

£ in new ≥ = £ in old ≥ - C £ above pivot ≥ * £ number in new ≥ S
T row T row number C row
The new T row is computed here. 1 = 1 - 10.52 * 102
0 = 0.5 - 10.52 * 112
1.5 = 0.5 - 10.52 * 1-22
-0.5 = 0 - 10.52 * 112
30 = 50 - 10.52 * 1402
M7.5 DEVELOPING THE THIRD TABLEAU M7-15
Hence, the new T row will appear in the third tableau in the following position:

$70 T 1 0 1.5 0.5 30
$50 C 0 1 2 1 40
The final step is again computing Step 5. Finally, the Zj and Cj - Zj rows are calculated for the third tableau:
Zj 1for T column2 1$702112 + 1$502102
the Zj and Cj - Zj values.
= = $70
Zj 1for C column2 = 1$702102 + 1$502112 = $50
Zj 1for S1 column2 = 1$70211.52 + 1$5021-22 = $5
Zj 1for S2 column2 = 1$7021-0.52 + 1$502112 = $15
Zj 1for total profit2 = 1$7021302 + 1$5021402 = $4,100
The net profit per unit row appears as follows:
COLUMN
T C S1 S2
Zj for column $70 $50 $5 $15
An optimal solution is reached All results for the third iteration of the simplex method are summarized in Table M7.6. Note that
because all Cj - Zj values are since every number in the tableau’s Cj - Zj row is 0 or negative, an optimal solution has been
zero or negative. reached.
That solution is
The final solution is to make 30 T = 30 tables
tables and 40 chairs at a profit C = 40 chairs
of $4,100. This is the same as the
graphical solution presented S1 = 0 slack hours in the painting department
earlier. S2 = 0 slack hours in the carpentry department
profit = $4,100 for the optimal solution
TABLE M7.6
Cj $70 $50 $0 $0
Final Simplex Tableau
for the Flair Furniture SOLUTION
Problem
$70 T 1 0 1.5 0.5 30
$50 C 0 1 2 1 40
Zj $70 $50 $5 $15 $4,100
Cj Zj $0 $0 $5 $15
T and C are the final basic variables, and S1 and S2 are nonbasic (and thus automatically equal
to 0). This solution corresponds to corner point C in Figure M7.2.
It’s always possible to make an arithmetic error when you are going through the numerous
Verifying that the solution does simplex steps and iterations, so it is a good idea to verify your final solution. This can be done in
not violate any of the original part by looking at the original Flair Furniture Company constraints and objective function:
constraints is a good way to
check that no mathematical First constraint: 2T + 1C … 100 painting department hours
errors were made. 21302 + 11402 … 100
100 … 100✓
Second constraint: 4T + 3C … 240 carpentry department hours
41302 + 31402 … 240
240 … 240✓
Objective function: profit = $70T + $50C
= $701302 + $501402
= $4,100
M7.6 Review of Procedures for Solving LP Maximization Problems

Before moving on to other issues concerning the simplex method, let’s review briefly what
we’ve learned so far for LP maximization problems.
I. Formulate the LP problem’s objective function and constraints.
II. Add slack variables to each less-than-or-equal-to constraint and to the problem’s objective
function.
III. Develop an initial simplex tableau with slack variables in the basis and the decision
variables set equal to 0. Compute the Zj and Cj - Zj values for this tableau.
IV. Follow these five steps until an optimal solution has been reached:
Here is a review of the five 1. Choose the variable with the greatest positive Cj - Zj to enter the solution. This is
simplex steps. the pivot column.
2. Determine the solution mix variable to be replaced and the pivot row by selecting
the row with the smallest (nonnegative) ratio of the quantity-to-pivot column
substitution rate. This row is the pivot row.
3. Calculate the new values for the pivot row.
4. Calculate the new values for the other row(s).
5. Calculate the Zj and Cj - Zj values for this tableau. If there are any Cj - Zj num-
bers greater than 0, return to step 1. If there are no Cj - Zj numbers that are greater
than 0, an optimal solution has been reached.
M7.7 Surplus and Artificial Variables

To handle » and ⴝ constraints, Up to this point in the chapter, all of the LP constraints you have seen were of the less-than-or-
the simplex method makes a equal-to 1…2 variety. Just as common in real-life problems—especially in LP minimization
conversion like it made to ◊ problems—are greater-than-or-equal-to 1Ú2 constraints and equalities. To use the simplex
constraints. method, each of these must be converted to a special form also. If they are not, the simplex tech-
nique is unable to set up an initial solution in the first tableau.
M7.7 SURPLUS AND ARTIFICIAL VARIABLES M7-17
Before moving on to the next section of this chapter, which deals with solving LP mini-
mization problems with the simplex method, we take a look at how to convert a few typical
constraints:
Constraint 1: 5X1 + 10X2 + 8X3 Ú 210
Constraint 2: 25X1 + 30X2 = 900
Surplus Variables
We subtract a surplus variable to Greater-than-or-equal-to 1Ú2 constraints, such as constraint 1 as just described, require a differ-
form an equality when dealing ent approach than do the less-than-or-equal-to 1…2 constraints we saw in the Flair Furniture
with a » constraint. problem. They involve the subtraction of a surplus variable rather than the addition of a slack
variable. The surplus variable tells us how much the solution exceeds the constraint amount. Be-
cause of its analogy to a slack variable, surplus is sometimes simply called negative slack. To
convert the first constraint, we begin by subtracting a surplus variable, S1, to create an equality:
Constraint 1 rewritten: 5X1 + 10X2 + 8X3 - S1 = 210
If, for example, a solution to an LP problem involving this constraint is X1 = 20, X2 = 8,
X3 = 5, the amount of surplus could be computed as follows:
5X1 + 10X2 + 8X3 - S1 = 210
51202 + 10182 + 8152 - S1 = 210
100 + 80 + 40 - S1 = 210
-S1 = 210 - 220
S1 = 10 surplus units
There is one more step, however, in preparing a Ú constraint for the simplex method.
Artificial Variables
There is one small problem in trying to use the first constraint (as it has just been rewritten) in
setting up an initial simplex solution. Since all “real” variables such as X1, X2, and X3 are set to 0
in the initial tableau, S1 takes on a negative value:
5102 + 10102 + 8102 - S1 = 210
0 - S1 = 210
S1 = -210
All variables in LP problems, be they real, slack, or surplus, must be nonnegative at all times. If
S1 = -210, this important condition is violated.
Artificial variables are needed To resolve the situation, we introduce one last kind of variable, called an artificial variable.
in » and ⴝ constraints. We simply add the artificial variable, A 1, to the constraint as follows:
Constraint 1 completed: 5X1 + 10X2 + 8X3 - S1 + A 1 = 210
Now, not only the X1, X2, and X3 variables may be set to 0 in the initial simplex solution, but the
S1 surplus variable as well. This leaves us with A 1 = 210.
Let’s turn our attention to constraint 2 for a moment. This constraint is already an equality,
so why worry about it? To be included in the initial simplex solution, it turns out, even an equal-
ity must have an artificial variable added to it:
Constraint 2 rewritten: 25X1 + 30X2 + A 2 = 900
The reason for inserting an artificial variable into an equality constraint deals with the usual
problem of finding an initial LP solution. In a simple constraint such as number 2, it’s easy to
guess that X1 = 0, X2 = 30 would yield an initial feasible solution. But what if our problem had
10 equality constraints, each containing seven variables? It would be extremely difficult to sit
down and “eyeball” a set of initial solutions. By adding artificial variables, such as A 2, we can
provide an automatic initial solution. In this case, when X1 and X2 are set equal to 0, A 2 = 900.
Artificial variables have no Artificial variables have no meaning in a physical sense and are nothing more than compu-
physical meaning and drop out tational tools for generating initial LP solutions. If an artificial variable has a positive (nonzero)
of the solution mix before the value, then the original constraint where this artificial variable was added has not been satisfied.
final tableau. A feasible solution has been found when all artificial variables are equal to zero, indicating all
constraints have been met. Before the final simplex solution has been reached, all artificial vari-
ables must be gone from the solution mix. This matter is handled through the problem’s objec-
tive function.
Surplus and Artificial Variables in the Objective Function

Whenever an artificial or surplus variable is added to one of the constraints, it must also be in-
cluded in the other equations and in the problem’s objective function, just as was done for slack
variables. Since artificial variables must be forced out of the solution, we can assign a very high
Cj cost to each. In minimization problems, variables with low costs are the most desirable ones
and the first to enter the solution. Variables with high costs leave the solution quickly, or never
To make sure that an artificial enter it at all. Rather than set an actual dollar figure of $10,000 or $1 million for each artificial
variable is forced out before the variable, however, we simply use the letter $M to represent a very large number.3 Surplus vari-
final solution is reached, it is ables, like slack variables, carry a zero cost. In maximization problems, we use negative M.
assigned a very high cost (M). If a problem had an objective function that read
Minimize cost = $5X1 + $9X2 + $7X3
and constraints such as the two mentioned previously, the completed objective function and con-
straints would appear as follows:
Minimize cost = $5X1 + $9X2 + $7X3 + $0S1 + $MA 1 + $MA 2
subject to 5X1 + 10X2 + 8X3 - 1S1 + 1A 1 + 0A 2 = 210
25X1 + 30X2 + 0X3 + 0S1 + 0A 1 + 1A 2 = 900
M7.8 Solving Minimization Problems

Now that we have discussed how to deal with objective functions and constraints associated with
minimization problems, let’s see how to use the simplex method to solve a typical problem.
The Muddy River Chemical Company Example

The Muddy River Chemical Corporation must produce exactly 1,000 pounds of a special
mixture of phosphate and potassium for a customer. Phosphate costs $5 per pound and potas-
sium costs $6 per pound. No more than 300 pounds of phosphate can be used, and at least 150
pounds of potassium must be used. The problem is to determine the least-cost blend of the two
ingredients.
3A technical point: If an artificial variable is ever used in a maximization problem (an occasional event), it is assigned
an objective function value of - $M to force it from the basis.
M7.8 SOLVING MINIMIZATION PROBLEMS M7-19
Linear Programming Modeling

IN ACTION in the Forests of Chile
problem,” says Professor Andres Weintraub. “The model and its

F aced with a series of challenges in making short-term harvest-
ing decisions, Forestal Arauco, a Chilean forestry firm, turned to
concepts became the natural language to discuss the operations.
They had to negotiate the parameters, and the model would do
professors at the University of Chile for LP modeling help. One of the dirty work. The system had to run in a few minutes to allow
the problems in short-term harvesting of trees is to match de- discussion and negotiation; that was a critical feature for the suc-
mand of products—defined by length and diameter—with the cess of this tool,” he adds.
supply of standing timber. The LP program took about two years to develop, and the re-
The manual system used at the time by foresters led to a sig- searchers were careful to observe two cardinal rules: (1) The solu-
nificant amount of waste of timber, where higher diameter logs, tion approach had to be comfortable and clear to the user, and
suited for export or sawmills, ended up being used for pulp, with (2) the system had to provide answers to the user in a fast devel-
a considerable loss in value. An LP model, labeled OPTICORT by opment, so the user could see quick improvements.
the professors, was the logical way to get better schedules.
“The system not only optimized the operational decisions in Source: Based on J. Summerour. “Chilean Forestry Firm a ‘Model’ of
harvesting, but also changed the way managers looked at the Success,” OR/MS Today (April 1999): 22–23.
Here is the mathematical This problem may be restated mathematically as

formulation of the minimization
problem for Muddy River Minimize cost = $5X1 + $6X2
Chemical Corp. subject to X1 + X2 = 1,000 lb
X1 … 300 lb
X2 Ú 150 lb
X1, X2 Ú 0
where
X1 = number of pounds of phosphate
X2 = number of pounds of potassium
Note that there are three constraints, not counting the nonnegativity constraints; the first is an
equality, the second a less-than-or-equal-to, and the third a greater-than-or-equal-to constraint.
Graphical Analysis
Looking at a graphical solution To have a better understanding of the problem, a brief graphical analysis may prove useful.
first will help us understand the There are only two decision variables, X1 and X2, so we are able to plot the constraints and fea-
steps in the simplex method. sible region. Because the first constraint, X1 + X2 = 1,000, is an equality, the solution must lie
somewhere on the line ABC (see Figure M7.3). It must also lie between points A and B because
of the constraint X1 … 300. The third constraint, X2 Ú 150, is actually redundant and nonbind-
ing since X2 will automatically be greater than 150 pounds if the first two constraints are ob-
served. Hence, the feasible region consists of all points on the line segment AB. As you recall
from Chapter 7, however, an optimal solution will always lie at a corner point of the feasible re-
gion (even if the region is only a straight line). The solution must therefore be either at point A
or point B. A quick analysis reveals that the least-cost solution lies at corner B, namely
X1 = 300 pounds of phosphate, X2 = 700 pounds of potassium. The total cost is $5,700.
FIGURE M7.3
X2
Muddy River Chemical
Corporation’s Feasible X1 300
Region Graph 1,000 A
800
B
600
X1 + X2 = 1,000
400
X2 150
200 F G H
100
E D C
0 200 400 600 800 1,000 X1
You don’t need the simplex method to solve the Muddy River Chemical problem, of course.
But we can guarantee you that few problems will be this simple. In general, you can expect to
see several variables and many constraints. The purpose of this section is to illustrate the
straightforward application of the simplex method to minimization problems. When the simplex
procedure is used to solve this, it will methodically move from corner point to corner point until
the optimal solution is reached. In Figure M7.3, the simplex method will begin at point E, then
move to point F, then to point G, and finally to point B, which is the optimal solution.
Converting the Constraints and Objective Function

First, insert slack, surplus, and The first step is to apply what we learned in the preceding section to convert the constraints and
artificial variables. This makes objective function into the proper form for the simplex method. The equality constraint,
it easier to set up the initial X1 + X2 = 1,000, just involves adding an artificial variable, A 1:
simplex tableau in Table M7.7.
X1 + X2 + A 1 = 1,000
The second constraint, X1 … 300, requires the insertion of a slack variable—let’s call it S1:
X1 + S1 = 300
The last constraint is X2 Ú 150, which is converted to an equality by subtracting a surplus vari-
able, S2, and adding an artificial variable, A 2:
X2 - S2 + A 2 = 150
Finally, the objective function, cost = $5X1 + $6X2, is rewritten as
Minimize cost = $5X1 + $6X2 + $0S1 + $0S2 + $MA 1 + $MA 2
The complete set of constraints can now be expressed as follows:

1X1 + 1X2 + 0S1 + 0S2 + 1A 1 + 0A 2 = 1,000
1X1 + 0X2 + 1S1 + 0S2 + 0A 1 + 0A 2 = 300
0X1 + 1X2 + 0S1 - 1S2 + 0A 1 + 1A 2 = 150
X1, X2, S1, S2, A 1, A 2 Ú 0
Rules of the Simplex Method for Minimization Problems

The minimization simplex rules Minimization problems are quite similar to the maximization problems tackled earlier in this
are slightly different. Now, the chapter. The significant difference involves the Cj - Zj row. Our objective is to minimize cost,
new variable to enter the solution and a negative Cj - Zj value indicates that the total cost will decrease if that variable is selected
mix will be in the column with to enter the solution. Thus, the new variable to enter the solution in each tableau (the pivot col-
the negative Cj - Zj value umn variable) will be the one with a negative Cj - Zj that gives the largest improvement. We
indicating the greatest
choose the variable that decreases costs the most. In minimization problems, an optimal solution
improvement.
is reached when all the numbers in the Cj - Zj row are 0 or positive—just the opposite from
the maximization case.4 All other simplex steps, as seen in the following, remain the same.
Steps for Simplex Minimization Problems

1. Choose the variable with a negative Cj - Zj that indicates the largest decrease in cost to
enter the solution. The corresponding column is the pivot column.
2. Determine the row to be replaced by selecting the one with the smallest (nonnegative)
quantity-to-pivot column substitution rate ratio. This is the pivot row.
3. Calculate new values for the pivot row.
4. Calculate new values for the other rows.
5. Calculate the Zj and Cj - Zj values for this tableau. If there are any Cj - Zj numbers less
than 0, return to step 1.
First Simplex Tableau for the Muddy River Chemical Corporation Problem
Now we solve Muddy River Chemical Corporation’s LP formulation using the simplex method.
The initial tableau is set up just as in the earlier maximization example. Its first three rows are
shown in the accompanying table. We note the presence of the $M costs associated with artifi-
cial variables A 1 and A 2, but we treat them as if they were any large number. As noted earlier,
they have the effect of forcing the artificial variables out of the solution quickly because of their
large costs.
4We should note that there is a second way to solve minimization problems with the simplex method: It involves a sim-
ple mathematical trick. It happens that minimizing the cost objective is the same as maximizing the negative of the cost
objective function. This means that instead of writing the Muddy River objective function as
Minimize cost = 5X1 + 6X2
we can instead write
Maximize 1-Cost2 = -5X1 - 6X2
The solution that maximizes 1-Cost2 also minimizes cost. It also means that the same simplex procedure shown earlier
for maximization problems can be used if this trick is employed. The only change is that the objective function must be
multiplied by 1-12.
Cj SOLUTION MIX X1 X2 S1 S2 A1 A2 QUANTITY

$M A1 1 1 0 0 1 0 1,000
$0 S1 1 0 1 0 0 0 300
$M A2 0 1 0 –1 0 1 150
The numbers in the Zj row are computed by multiplying the Cj column on the far left of the tableau
times the corresponding numbers in each other column. They are then entered in Table M7.7:
Zj 1for X1 column2 = $M112 + $0112 + M102 = $M
Zj 1for X2 column2 = $M112 + $0102 + $M112 = $2M
Zj 1for S1 column2 = $M102 + $0112 + $M102 = $0
Zj 1for S2 column2 = $M102 + $0102 + $M1-12 = - $M
Zj 1for A 1 column2 = $M112 + $0102 + $M102 = $M
Zj 1for A 2 column2 = $M102 + $0102 + $M112 = $M
Zj 1for total cost2 = $M11,0002 + $013002 + $M11502 = $1,150M
The Cj - Zj entries are determined as follows:
COLUMN
X1 X2 S1 S2 A1 A2
Cj for column $5 $6 $0 $0 $M $M
Zj for column $M $2M $0 $M $M $M
Cj Zj for column $M $5 $2M $6 $0 $M $0 $0
TABLE M7.7
Cj $5 $6 $0 $0 $M $M
Initial Simplex
Tableau for the SOLUTION
MIX X1 X2 S1 S2 A1 A2 QUANTITY
Muddy River
Chemical Corporation $M A1 1 1 0 0 1 0 1,000
Problem $0 S1 1 0 1 0 0 0 300
$M A2 0 1 0 1 0 1 150 Pivot row
Pivot number
Zj $M $2M 0 $M $M $M $1,150M
Cj – Zj $M 5 $2M 6 $0 $M $0 0 (total cost)
Pivot column
Here is the initial simplex This initial solution was obtained by letting each of the variables X1, X2, and S2 assume a
solution. value of 0. The current basic variables are A 1 = 1,000, S1 = 300, and A 2 = 150. This complete
solution could be expressed in vector, or column, form as
X1 0
X2 0
S1 300
F V = F V
S2 0
A1 1,000
A2 150
An extremely high cost, $1,150M from Table M7.7. is associated with this answer. We know that
this can be reduced significantly and now move on to the solution procedures.
Developing a Second Tableau

We examine whether the current In the Cj - Zj row of Table M7.7, we see that there are two entries with negative values, X1 and
solution is optimal by looking at X2. In the simplex rules for minimization problems, this means that an optimal solution does not
the Cj - Zj row. yet exist. The pivot column is the one with the negative entry in the Cj - Zj row that indicates
the largest improvement—shown in Table M7.7 as the X2 column, which means that X2 will en-
ter the solution next.
Which variable will leave the solution to make room for the new variable, X2? To find out,
we divide the elements of the quantity column by the respective pivot column substitution rates:
1,000
For the A 1 row = = 1,000
1
300
For the S1 row = (this is an undefined ratio, so we ignore it)
1. largest negative —> 2. smallest 0
150
A2 is the pivot row because 150 is For the A 2 row = = 150 (smallest quotient, indicating pivot row)
the smallest quotient. 1
Hence, the pivot row is the A 2 row, and the pivot number (circled) is at the intersection of the X2
column and the A 2 row.
The entering row for the next simplex tableau is found by dividing each element in the pivot
row by the pivot number, 1. This leaves the old pivot row unchanged, except that it now repre-
sents the solution variable X2. The other two rows are altered one at a time by again applying
the formula shown earlier in step 4:
1New row numbers2 = 1Numbers in old row2
-ca b * a bd
Number above or below Corresponding number
pivot number in newly replaced row
A 1 Row S1 Row
1 = 1 - 112102 1 = 1 - 102102
0 = 1 - 112112 0 = 0 - 102112
0 = 0 - 112102 1 = 1 - 102102
1 = 0 - 1121-12 0 = 0 - 1021-12
1 = 1 - 112102 0 = 0 - 102102
-1 = 0 - 112112 0 = 0 - 102112
850 = 1,000 - 11211502 300 = 300 - 10211502
TABLE M7.8
Cj $5 $6 $0 $0 $M $M
Second Simplex
Muddy River
Chemical Corporation $M A1 1 0 0 1 1 1 850
Problem $0 S1 1 0 1 0 0 0 300 Pivot row
Pivot number
$6 X2 0 1 0 1 0 1 150
Zj $M $6 $0 $M – 6 $M –$M 6 $850M $900
Cj –Zj $M 5 $0 $0 $M 6 $0 $2M 6
Pivot column
The Zj and Cj - Zj rows are computed next:

Zj 1for X12 = $M112 + $0112 + $6102 = $M
Zj 1for X22 = $M102 + $0102 + $6112 = $6
Zj 1for S12 = $M102 + $0112 + $6102 = $0
Zj 1for S22 = $M112 + $0102 + $61-12 = $M - 6
Zj 1for A 12 = $M112 + $0102 + $6102 = $M
Zj 1for A 22 = $M1-12 + $0102 + $6112 = - $M + 6
Zj 1for total cost2 = $M18502 + $013002 + $611502 = - $850M + 900
COLUMN
X1 X2 S1 S2 A1 A2
Zj for column $M $6 $0 $M – 6 $M –$M 6
Cj – Zj for column –$M 5 $0 $0 –$M 6 $0 $2M – 6
All of these computational results are presented in Table M7.8.

The solution after the second The solution at the end of the second tableau (point F in Figure M7.3) is A 1 = 850,
tableau is still not optimal. S1 = 300, X2 = 150. X1, S2, and A 2 are currently the nonbasic variables and have zero value.
The cost at this point is still quite high, $850M + $900. This answer is not optimal because not
every number in the Cj - Zj row is zero or positive.
Developing a Third Tableau

The new pivot column is the X1 column. To determine which variable will leave the basis to
make room for X1, we check the quantity column–to–pivot column ratios again:
850
The third tableau is developed For the A 1 row = = 850
1
in this section.
1smallest ratio2
300
For the S1 row = = 300
1
150
For the X2 row = = undefined
0
TABLE M7.9
Cj $5 $6 $0 $0 $M $M
Third Simplex
Muddy River
Chemical Corporation $M A1 0 0 –1 1 1 –1 550 Pivot row
Problem $5 X1 1 0 1 0 0 0 300 Pivot number
$6 X2 0 1 0 –1 0 1 150
Zj $5 $6 –$M 5 $M 6 $M $M 6 $550M 2,400
Cj – Zj $0 $0 $M 5 $M 6 $0 $2M 6
Pivot column
Hence, variable S1 will be replaced by X1.5 The pivot number, row, and column are labeled
in Table M7.8.
To replace the pivot row, we divide each number in the S1 row by 1 (the circled pivot num-
ber), leaving the row unchanged. The new X1 row is shown in Table M7.9. The other computa-
tions for this third simplex tableau follow:
Here are the computations for A 1 Row S1 Row
1 - 112112 0 - 102112
the fourth solution.
0 = 0 =
0 = 0 - 112102 1 = 1 - 102102
-1 = 0 - 112112 0 = 0 - 102112
1 = 1 - 112102 -1 = -1 - 102102
1 = 1 - 112102 0 = 0 - 102102
-1 = -1 - 112102 1 = 1 - 102102
550 = 850 - 11213002 150 = 150 - 10213002
The Zj and Cj - Zj rows are computed next:

Zj 1for X12 = $M102 + $5112 + $6102 = $5
Zj 1for X22 = $M102 + $5102 + $6112 = $6
Zj 1for S12 = $M1-12 + $5112 + $6102 = - $M + 5
Zj 1for S22 = $M112 + $5102 + $61-12 = $M - 6
Zj 1for A 12 = $M112 + $5102 + $6102 = $M
Zj 1for A 22 = $M1-12 + $5102 + $6112 = - $M + 6
Zj 1for total cost2 = $M15502 + $513002 + $611502 = $550M + 2,400
COLUMN
X1 X2 S1 S2 A1 A2
Zj for column $5 $6 $M 5 $M 6 $M $M 6
Cj – Zj for column $0 $0 $M 5 $M 6 $0 $2M 6
5 At this point, it might appear to be more cost-effective to replace the A 1 row instead of the S1 row. This would remove
the last artificial variable, and its large $M cost, from the basis. The simplex method, however, does not always pick the
most direct route to reaching the final solution. You may be assured, though, that it will lead us to the correct answer. In
Figure M7.3, this would involve moving to point H instead of point G.
The third solution is still not The solution at the end of the three iterations (point G in Figure M7.3) is still not optimal
optimal. because the S2 column contains a Cj - Zj value that is negative. Note that the current total cost
is nonetheless lower than at the end of the second tableau, which in turn is lower than the initial
solution cost. We are headed in the right direction but have one more tableau to go!
Fourth Tableau for the Muddy River Chemical Corporation Problem

The pivot column is now the S2 column. The ratios that determine the row and variable to be
replaced are computed as follows:
1row to be replaced2
550
For the A 1 row: = 550
1
1undefined2
300
For the X1 row:
0
1not considered because it is negative2

150
For the X2 row:
-1
Here are the computations for Each number in the pivot row is divided by the pivot number (again 1, by coincidence). The
the fourth solution. other two rows are computed as follows and are shown in Table M7.10:
X1 Row X1 Row
1 = 1 - 102102 0 = 0 - 1-12102
0 = 0 - 102102 1 = 1 - 1-12102
1 = 1 - 1021-12 -1 = 0 - 1-121-12
0 = 0 - 102112 0 = -1 - 1-12112
0 = 0 - 102112 1 = 0 - 1-12112
0 = 0 - 1021-12 0 = 1 - 1-121-12
300 = 300 - 10215502 700 = 150 - 1-1215502
Zj 1for X12 = $0102 + $5112 + $6102 = $5

Zj 1for X22 = $0102 + $5102 + $6112 = $6
Zj 1for S12 = $01-12 + $5112 + $61-12 = - $1
Zj 1for S22 = $0112 + $5102 + $6102 = $0
Zj 1for A 12 = $0112 + $5102 + $6112 = $6
Zj 1for A 22 = $01-12 + $5102 + $6102 = $0
Zj 1for total cost2 = $015502 + $513002 + $617002 = $5,700
COLUMN
X1 X2 S1 S2 A1 A2
Zj for column $5 $6 $1 $0 $6 $0
Cj Zj for column $0 $0 $1 $0 $M 6 $M
M7.9 REVIEW OF PROCEDURES FOR SOLVING LP MINIMIZATION PROBLEMS M7-27
TABLE M7.10
Cj $5 $6 $0 $0 $M $M
Fourth and Optimal
Solution to the SOLUTION
Muddy River
Chemical Corporation $0 S2 0 0 1 1 1 1 550
Problem $5 X1 1 0 1 0 0 0 300
$6 X2 0 1 1 0 1 0 700
Zj $5 $6 $1 $0 $6 $0 $5,700
Cj – Zj $0 $0 $1 $0 $M 6 $M
The optimal solution has been On examining the Cj - Zj row in Table M7.10, only positive or 0 values are found. The
reached because only positive or fourth tableau therefore contains the optimum solution. That solution is X1 = 300, X2 = 700,
zero values appear in the S2 = 550. The artificial variables are both equal to 0, as is S1 . Translated into management
Cj - Zj row. terms, the chemical company’s decision should be to blend 300 pounds of phosphate 1X12
with 700 pounds of potassium 1X22. This provides a surplus 1S22 of 550 pounds of potassium
more than required by the constraint X2 Ú 150. The cost of this solution is $5,700. If you look
back to Figure M7.3, you can see that this is identical to the answer found by the graphical
approach.
Although small problems such as this can be solved graphically, more realistic product
blending problems demand use of the simplex method, usually in computerized form.
M7.9 Review of Procedures for Solving LP Minimization Problems

Just as we summarized the steps for solving LP maximization problems with the simplex
method in Section M7.6, let us do so for minimization problems here:
I. Formulate the LP problem’s objective function and constraints.
II. Include slack variables in each less-than-or-equal-to constraint, artificial variables in each
equality constraint, and both surplus and artificial variables in each greater-than-or-equal-to
constraint. Then add all of these variables to the problem’s objective function.
III. Develop an initial simplex tableau with artificial and slack variables in the basis and the
other variables set equal to 0. Compute the Zj and Cj - Zj values for this tableau.
IV. Follow these five steps until an optimal solution has been reached:
1. Choose the variable with the negative Cj - Zj indicating the greatest improvement
to enter the solution. This is the pivot column.
2. Determine the row to be replaced by selecting the one with the smallest (nonnega-
tive) quantity-to-pivot column substitution rate ratio. This is the pivot row.
3. Calculate the new values for the pivot row.
4. Calculate the new values for the other row(s).
5. Calculate the Zj and Cj - Zj values for the tableau. If there are any Cj - Zj numbers
less than 0, return to step 1. If there are no Cj - Zj numbers that are less than 0, an
optimal solution has been reached.
M7.10 Special Cases

In Chapter 7 we addressed some special cases that may arise when solving LP problems graphi-
cally (see Section 7.7). Here we describe these cases again, this time as they refer to the simplex
method.
Infeasibility
A situation with no feasible Infeasibility, you may recall, comes about when there is no solution that satisfies all of the prob-
solution may exist if the problem lem’s constraints. In the simplex method, an infeasible solution is indicated by looking at the
was formulated improperly. final tableau. In it, all Cj - Zj row entries will be of the proper sign to imply optimality, but an
artificial variable 1A 12 will still be in the solution mix.
Table M7.11 illustrates the final simplex tableau for a hypothetical minimization type of LP
problem. The table provides an example of an improperly formulated problem, probably con-
taining conflicting constraints. No feasible solution is possible because an artificial variable, A 2,
remains in the solution mix, even though all Cj - Zj are positive or 0 (the criterion for an opti-
mal solution in a minimization case).
Unbounded Solutions
Unboundedness describes linear programs that do not have finite solutions. It occurs in maxi-
mization problems, for example, when a solution variable can be made infinitely large without
No finite solution may exist in violating a constraint. In the simplex method, the condition of unboundedness will be discov-
problems that are not bounded. ered prior to reaching the final tableau. We will note the problem when trying to decide which
This means that a variable can be variable to remove from the solution mix. As seen earlier in this chapter, the procedure is to di-
infinitely large without violating vide each quantity column number by the corresponding pivot column number. The row with
a constraint. the smallest positive ratio is replaced. But if all the ratios turn out to be negative or undefined, it
indicates that the problem is unbounded.
Table M7.12 illustrates the second tableau calculated for a particular LP maximization
problem by the simplex method. It also points to the condition of unboundedness. The solution
is not optimal because not all Cj - Zj entries are 0 or negative, as required in a maximization
problem. The next variable to enter the solution should be X1. To determine which variable will
leave the solution, we examine the ratios of the quantity column numbers to their corresponding
numbers in the X1, or pivot, column:
30
Ratio for the X2 row:
-1
Negative ratios unacceptable
10
Ratio for the S2 row:
-2
Since both pivot column numbers are negative, an unbounded solution is indicated.
TABLE M7.11
Cj $5 $8 $0 $0 $M $M
Illustration of
Infeasibility SOLUTION
$5 X1 1 0 2 3 1 0 200
$8 X2 0 1 1 2 2 0 100
$M A2 0 0 0 1 1 1 20
Zj $5 $8 $2 $31 M $21 M $M $1,800 20M
Cj Zj $0 $0 $2 $M 31 $2M 21 $0
M7.10 SPECIAL CASES M7-29
TABLE M7.12
Cj $6 $9 $0 $0
Problem with an
Unbounded Solution SOLUTION
MIX X1 X2 S1 S2 QUANTITY
$9 X2 1 1 2 0 30
$0 S2 2 0 1 1 10
Zj $9 $9 $18 $0 $270
Cj Zj $15 $0 $18 $0
Pivot column
Degeneracy
Degeneracy is another situation that can occur when solving an LP problem using the simplex
method. It develops when three constraints pass through a single point. For example, suppose a
problem has only these three constraints X1 … 10, X2 … 10, and X1 + X2 6 20. All three con-
straint lines will pass through the point 110, 102. Degeneracy is first recognized when the ratio
Tied ratios in the simplex calculations are made. If there is a tie for the smallest ratio, this is a signal that degeneracy
calculations signal degeneracy. exists. As a result of this, when the next tableau is developed, one of the variables in the solution
mix will have a value of zero.
Table M7.13 provides an example of a degenerate problem. At this iteration of the given
maximization LP problem, the next variable to enter the solution will be X1, since it has the only
positive Cj - Zj number. The ratios are computed as follows:
10
For the X2 row: = 40
0.25
20
For the S2 row: = 5 Tie for the smallest ratio indicates degeneracy
4
10
For the S3 row: = 5
2
Cycling may result from Theoretically, degeneracy could lead to a situation known as cycling, in which the simplex algo-
degeneracy. rithm alternates back and forth between the same nonoptimal solutions; that is, it puts a new
variable in, then takes it out in the next tableau, puts it back in, and so on. One simple way of
dealing with the issue is to select either row (S2 or S3 in this case) arbitrarily. If we are unlucky
and cycling does occur, we simply go back and select the other row.
TABLE M7.13
Cj $5 $8 $2 $0 $0 $0
Problem Illustrating
Degeneracy SOLUTION
MIX X1 X2 X3 S1 S2 S3 QUANTITY
$8 X2 0.25 1 1 2 0 0 10
$0 S2 4 0 0.33 1 1 0 20
$0 S3 2 0 2 0.4 0 1 10
Zj $2 $8 $8 $16 $0 $0 $80
Cj Zj $3 $0 $6 $16 $0 $0
Pivot column
TABLE M7.14
Cj $3 $2 $0 $0
Problem with
Alternate Optimal SOLUTION
Solutions
$2 X2 1.5 1 1 0 6
$0 S2 1 0 0.5 1 3
Zj $3 $2 $2 $0 $12
Cj – Zj $0 $0 –$2 $0
More Than One Optimal Solution

Alternate optimal solutions Multiple, or alternate, optimal solutions can be spotted when the simplex method is being used
may exist if the Cj ⴚ Zj by looking at the final tableau. If the Cj - Zj value is equal to 0 for a variable that is not in the
value ⴝ 0 for a variable not solution mix, more than one optimal solution exists.
in the solution mix. Let’s take Table M7.14 as an example. Here is the last tableau of a maximization problem;
each entry in the Cj - Zj row is 0 or negative, indicating that an optimal solution has been
reached. That solution is read as X2 = 6, S2 = 3, profit = $12. Note, however, that variable X1
can be brought into the solution mix without increasing or decreasing profit. The new solution,
with X1 in the basis, would become X1 = 3, X2 = 1.5, with profit still at $12. Can you modify
Table M7.14 to prove this?
M7.11 Sensitivity Analysis with the Simplex Tableau

In Chapter 7 we introduce the topic of sensitivity analysis as it applies to LP problems that we
have solved graphically. This valuable concept shows how the optimal solution and the value of
its objective function change, given changes in various inputs to the problem. Graphical analysis
is useful in understanding intuitively and visually how feasible regions and the slopes of objec-
tive functions can change as model coefficients change. Computer programs handling LP prob-
lems of all sizes provide sensitivity analysis as an important output feature. Those programs use
the information provided in the final simplex tableau to compute ranges for the objective func-
tion coefficients and ranges for the RHS values. They also provide “shadow prices,” a concept
that we introduce in this section.
High Note Sound Company Revisited

In Chapter 7 we use the High Note Sound Company to illustrate sensitivity analysis graphically.
High Note is a firm that makes compact disk (CD) players (called X1) and stereo receivers
(called X2). Its LP formulation is repeated here:
Maximize profit = $50X1 + $120X2
subject to 2X1 + 4X2 … 80 1hours of electricians’ time available2
3X1 + 1X2 … 60 1hours of audio technicians’time available2
High Note’s graphical solution is repeated in Figure M7.4.
M7.11 SENSITIVITY ANALYSIS WITH THE SIMPLEX TABLEAU M7-31
FIGURE M7.4
X2
High Note Sound
(receivers)
Company Graphical
Solution
60
40 Optimal Solution at Point a
X 1 = 0 CD Players
a = (0, 20) X 2 = 20 Receivers
Profits = $2,400
20 b = (16, 12)
Isoprofit Line: $2,400 = 50X1 + 120X2
10
0 10 20 30 40 50 60 X1
c = (20, 0) (CD players)
Changes in the Objective Function Coefficients

In Chapter 7 we saw how to use graphical LP to examine the objective function coefficients.
A second way of illustrating the sensitivity analysis of objective function coefficients is to
consider the problem’s final simplex tableau. For the High Note Sound Company, this tableau is
shown in Table M7.15. The optimal solution is seen to be as follows:
f Basic variables
X2 = 20 stereo receivers
S2 = 40 hours of slack time of audio technicians
f Nonbasic variables
X1 = 0 CD players
S1 = 0 hours of slack time of electricians
Basic variables (those in the solution mix) and nonbasic variables (those set equal to 0) must be
handled differently using sensitivity analysis. Let us first consider the case of a nonbasic
variable.
TABLE M7.15
Cj $50 $120 $0 $0
Optimal Solution by
the Simplex Method SOLUTION
$120 X2 0.5 1 0.25 0 20
$0 S2 2.5 0 - 0.25 1 40
Zj $60 $120 $30 $0 $2,400
Cj - Zj - $10 $0 - $30 $0
Nonbasic variables are variables NONBASIC OBJECTIVE FUNCTION COEFFICIENT Our goal here is to find out how sensitive the
that have a value of zero. problem’s optimal solution is to changes in the contribution rates of variables not currently in
the basis (X1 and S1). Just how much would the objective function coefficients have to change
before X1 or S1 would enter the solution mix and replace one of the basic variables?
The answer lies in the Cj - Zj row of the final simplex tableau (as in Table M7.15). Since
this is a maximization problem, the basis will not change unless the Cj - Zj value of one of the
nonbasic variables becomes positive. That is, the current solution will be optimal as long as all
numbers in the bottom row are less than or equal to 0. It will not be optimal if X1’s Cj - Zj
The solution is optimal as long as value is positive, or if S1’s Cj - Zj value is greater than 0. Therefore, the values of Cj for X1
all Cj ⴚ Zj ◊ 0. and S1 that do not bring about any change in the optimal solution are given by
Cj - Zj … 0
This is the same as writing
Cj … Zj
Since X1’s Cj value is $50 and its Zj value is $60, the current solution is optimal as long as the
profit per CD player does not exceed $60, or correspondingly, does not increase by more than
$10. Similarly, the contribution rate per unit of S1 (or per hour of electrician’s time) may
increase from $0 up to $30 without changing the current solution mix.
The range over which Cj rates In both cases, when you are maximizing an objective function, you may increase the value
for nonbasic variables can vary of Cj up to the value of Zj. You may also decrease the value of Cj for a nonbasic variable to neg-
without causing a change in the ative infinity 1- q 2 without affecting the solution. This range of Cj values is called the range of
optimal solution mix is called the insignificance for nonbasic variables.
range of insignificance.
- q … Cj 1for X12 … $60
- q … Cj 1for S12 … $30
BASIC OBJECTIVE FUNCTION COEFFICIENT Sensitivity analysis on objective function coefficients
of variables that are in the basis or solution mix is slightly more complex. We saw that a change
in the objective function coefficient for a nonbasic variable affects only the Cj - Zj value for
Testing basic variables involves that variable. But a change in the profit or cost of a basic variable can affect the Cj - Zj values
reworking the final simplex of all nonbasic variables because this Cj is not only in the Cj row but also in the Cj column. This
tableau. then impacts the Zj row.
Let us consider changing the profit contribution of stereo receivers in the High Note Sound
Company problem. Currently, the objective function coefficient is $120. The change in this
value can be denoted by the Greek capital letter delta 1¢2. We rework the final simplex tableau
(first shown in Table M7.15) and see our results in Table M7.16.
Notice the new Cj - Zj values for nonbasic variables X1 and S1. These were determined in
exactly the same way as we did earlier in this chapter. But wherever the Cj value for X2 of $120
was seen in Table M7.15, a new value of $120 + ¢ is used in Table M7.16.
Once again, we recognize that the current optimal solution will change only if one or more
of the Cj - Zj row values becomes greater than 0. The question is, how may the value of ¢ vary
so that all Cj - Zj entries remain negative? To find out, we solve for ¢ in each column.
From the X1 column:
-10 - 0.5¢ … 0
-10 … 0.5¢
-20 … ¢ or ¢ Ú -20
M7.11 SENSITIVITY ANALYSIS WITH THE SIMPLEX TABLEAU M7-33
TABLE M7.16
Cj $50 $120 + $0 $0
Change in the Profit
Contribution of Stereo SOLUTION
Receivers
$120 + X2 0.5 1 0.25 0 20
$0 S2 2.5 0 – 0.25 1 40
Zj $60 + 0.5Δ $120 + Δ $30 + 0.25Δ $0 $2,400 + 20Δ
Cj Zj $10 0.5Δ $0 $30 – 0.25Δ $0
This inequality means that the optimal solution will not change unless X2’s profit coeffi-
cient decreases by at least $20, which is a change of ¢ = -$20. Hence, variable X1 will not
enter the basis unless the profit per stereo receiver drops from $120 to $100 or less. This, inter-
estingly, is exactly what we noticed graphically in Chapter 7. When the profit per stereo receiver
dropped to $80, the optimal solution changed from corner point a to corner point b.
Now we examine the S1 column:
-30 - 0.25¢ … 0
-30 … 0.25¢
-120 … ¢ or ¢ Ú -120
This inequality implies that S1 is less sensitive to change than X1. S1 will not enter the basis un-
less the profit per unit of X2 drops from $120 all the way down to $0.
The range of optimality is the Since the first inequality is more binding, we can say that the range of optimality for X2’s
range of values over which a profit coefficient is
$100 … Cj1for X22 … q
basic variable’s coefficient can
change without causing a change
in the optimal solution mix. As long as the profit per stereo receiver is greater than or equal to $100, the current production
mix of X2 = 20 receivers and X1 = 0 CD players will be optimal.
In analyzing larger problems, we would use this procedure to test for the range of optimal-
ity of every real decision variable in the final solution mix. The procedure helps us avoid the
time-consuming process of reformulating and resolving the entire LP problem each time a small
change occurs. Within the bounds set, changes in profit coefficients would not force a firm to al-
ter its product mix decision or change the number of units produced. Overall profits, of course,
will change if a profit coefficient increases or decreases, but such computations are quick and
easy to perform.
Changes in Resources or RHS Values

Making changes in the RHS values (the resources of electricians’ and audio technicians’ time)
result in changes in the feasible region and often the optimal solution.
The shadow price is the value SHADOW PRICES This leads us to the important subject of shadow prices. Exactly how much
of one additional unit of a should a firm be willing to pay to make additional resources available? Is one more hour of ma-
scarce resource. Shadow pricing chine time worth $1 or $5 or $20? Is it worthwhile to pay workers an overtime rate to stay one
provides an important piece of extra hour each night to increase production output? Valuable management information could
economic information. be provided if the worth of additional resources were known.
Fortunately, this information is available to us by looking at the final simplex tableau of an
LP problem. An important property of the Cj - Zj row is that the negatives of the numbers in
its slack variable 1Si2 columns provide us with what we call shadow prices. A shadow price is
The negatives of the numbers in the change in value of the objective function from an increase of one unit of a scarce resource
the Cj ⴚ Zj row’s slack variable (e.g., by making one more hour of machine time or labor time or other resource available).
columns are the shadow prices. The final simplex tableau for the High Note Sound Company problem is repeated as Table
M7.17 (it was first shown as Table M7.15). The tableau indicates that the optimal solution is
X1 = 0, X2 = 20, S1 = 0, and S2 = 40 and that profit = $2,400. Recall that S1 represents
slack availability of the electricians’ resource and S2 the unused time in the audio technicians’
department.
The firm is considering hiring an extra electrician on a part-time basis. Let’s say that it will
cost $22 per hour in wages and benefits to bring the part-timer on board. Should the firm do
this? The answer is yes; the shadow price of the electrician time resource is $30. Thus, the firm
will net $8 1= $30 - $222 for every hour the new worker helps in the production process.
Should High Note also hire a part-time audio technician at a rate of $14 per hour? The
answer is no: The shadow price is $0, implying no increase in the objective function by making
more of this second resource available. Why? Because not all of the resource is currently being
used—40 hours are still available. It would hardly pay to buy more of the resource.
RIGHT-HAND-SIDE RANGING Obviously, we can’t add an unlimited number of units of resource
without eventually violating one of the problem’s constraints. When we understand and com-
pute the shadow price for an additional hour of electricians’ time ($30), we will want to
The range over which shadow determine how many hours we can actually use to increase profits. Should the new resource be
prices remain valid is called added 1 hour per week, 2 hours, or 200 hours? In LP terms, this process involves finding the
right-hand-side ranging. range over which shadow prices will stay valid. Right-hand-side ranging tells us the number of
hours High Note can add or remove from the electrician department and still have a shadow
price of $30.
Ranging is simple in that it resembles the simplex process we used earlier in this chap-
ter to find the minimum ratio for a new variable. The S1 column and quantity column from
Table M7.17 are repeated in the following table; the ratios, both positive and negative, are
also shown:
QUANTITY S1 RATIO
20 0.25 20>0.25 = 80
40 - 0.25 40> -0.25 = -160
TABLE M7.17
Cj $50 $120 $0 $50
Final Tableau for the
High Note Sound SOLUTION
Company
$120 X2 0.5 1 0.25 0 20
$0 S2 2.5 0 –0.25 1 40
Zj $60 $120 $30 $0 $2,400
Cj – Zj –$10 $0 –$30 $0
⎫⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪⎪
⎬
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎭
Objective function increases by $30

if 1 additional hour of electricians’
time is made available
M7.12 THE DUAL M7-35
The smallest positive ratio (80 in this example) tells us by how many hours the electricians’
time resource can be reduced without altering the current solution mix. Hence, we can decrease
the RHS resource by as much as 80 hours—basically from the current 80 hours all the way down
to 0 hours—without causing a basic variable to be pivoted out of the solution.
The smallest negative ratio 1-1602 tells us the number of hours that can be added to the re-
source before the solution mix changes. In this case, we can increase electricians’ time by 160
hours, up to 240 1=80 currently + 160 may be added2 hours. We have now established the
range of electricians’ time over which the shadow price of $30 is valid. That range is from 0 to
240 hours.
The audio technician resource is slightly different in that all 60 hours of time originally
available have not been used. (Note that S2 = 40 hours in Table M7.17.) If we apply the ratio
test, we see that we can reduce the number of audio technicians’ hours by only 40 (the smallest
positive ratio = 40>1) before a shortage occurs. But since we are not using all the hours cur-
rently available, we can increase them indefinitely without altering the problem’s solution. Note
that there are no negative substitution rates in the S2 column, so there are no negative ratios.
Hence, the valid range for this shadow price would be from 201= 60 - 402 hours to an un-
bounded upper limit.
The substitution rates in the slack variable column can also be used to determine the actual
values of the solution mix variables if the right-hand side of a constraint is changed. The follow-
Changes in the RHS values of ing relationship is used to find these values:
New quantity = Original quantity + 1Substution rate21Change in right-hand side2
constraints may change the
optimal quantity values of the
solution mix variables. For example, if 12 more electrician hours were made available, the new values in the quan-
tity column of the simplex tableau are found as follows:
ORIGINAL QUANTITY S1 NEW QUANTITY

20 0.25 20 + 10.2521122 = 23
40 - 0.25 40 + 1-0.2521122 = 37
Thus, if 12 hours are added, X2 = 23 and S2 = 37. All other variables are nonbasic and remain
zero. This yields a total profit of 50102 + 1201232 = $2,760, which is an increase of $360 (or
the shadow price of $30 per hour for 12 hours of electrician time). A similar analysis with the
other constraint and the S2 column would show that if any additional audio technician hours
were added, only the slack for that constraint would increase.
M7.12 The Dual

Every LP primal has a dual. Every LP problem has another LP problem associated with it, which is called its dual. The first
The dual provides useful way of stating a linear problem is called the primal of the problem; we can view all of the prob-
economic information. lems formulated thus far as primals. The second way of stating the same problem is called the
dual. The optimal solutions for the primal and the dual are equivalent, but they are derived
through alternative procedures.
The dual contains economic information useful to management, and it may also be easier to
solve, in terms of less computation, than the primal problem. Generally, if the LP primal in-
volves maximizing a profit function subject to less-than-or-equal-to resource constraints, the
dual will involve minimizing total opportunity costs subject to greater-than-or-equal-to product
profit constraints. Formulating the dual problem from a given primal is not terribly complex,
and once it is formulated, the solution procedure is exactly the same as for any LP problem.
Let’s illustrate the primal–dual relationship with the High Note Sound Company data. As
you recall, the primal problem is to determine the best production mix of CD players 1X12 and
stereo receivers 1X22 to maximize profit:
subject to 2X1 + 4X2 … 80 1hours of available electrician time2
3X1 + 1X2 … 60 1hours of audio technician time available2
The dual variables represent the The dual of this problem has the objective of minimizing the opportunity cost of not using the
potential value of resources. resources in an optimal manner. Let’s call the variables that it will attempt to solve for U1 and
U2. U1 represents the potential hourly contribution or worth of electrician time; in other words,
the dual value of 1 hour of the electricians’ resource. U2 stands for the imputed worth of the au-
dio technicians’ time, or the dual technician resource. Thus, each constraint in the primal prob-
lem will have a corresponding variable in the dual problem. Also, each decision variable in the
primal problem will have a corresponding constraint in the dual problem.
The RHS quantities of the primal constraints become the dual’s objective function coeffi-
cients. The total opportunity cost that is to be minimized will be represented by the function
80U1 + 60U2, namely,
Minimize opportunity cost = 80U1 + 60U2
The corresponding dual constraints are formed from the transpose6 of the primal constraints
coefficients. Note that if the primal constraints are …, the dual constraints are Ú :
2 U1 + 3 U2 Ú 50 : Primal profit coefficients
4 U1 + 1 U2 Ú 120 Coefficients from the second primal constraint
:
: Coefficients from the first primal constraint

Let’s look at the meaning of these dual constraints. In the first inequality, the RHS constant
($50) is the income from one CD player. The coefficients of U1 and U2 are the amounts of
each scarce resource (electrician time and audio technician time) that are required to produce
a CD player. That is, 2 hours of electricians’ time and 3 hours of audio technicians’ time are
used up in making one CD player. Each CD player produced yields $50 of revenue to High
Note Sound Company. This inequality states that the total imputed value or potential worth
of the scarce resources needed to produce a CD player must be at least equal to the profit de-
rived from the product. The second constraint makes an analogous statement for the stereo
receiver product.
example, the transpose of the set of numbers a b is a b . In the case of the transpose of the primal
6For
a b a c
c d b d
coefficients a b , the result is a b . Refer to Module 5, which deals with matrices and determinants, for a review
2 4 2 3
3 1 4 1
of the transpose concept.
M7.12 THE DUAL M7-37
Dual Formulation Procedures

The mechanics of formulating a dual from the primal problem are summarized in the follow-
ing list.
These are the five steps for Steps to Form a Dual

formulating a dual.
1. If the primal is a maximization, the dual is a minimization, and vice versa.
2. The RHS values of the primal constraints become the dual’s objective function coefficients.
3. The primal objective function coefficients become the RHS values of the dual constraints.
4. The transpose of the primal constraint coefficients become the dual constraint coefficients.
5. Constraint inequality signs are reversed.7
Solving the Dual of the High Note Sound Company Problem

The simplex algorithm is applied to solve the preceding dual problem. With appropriate surplus
and artificial variables, it can be restated as follows:
Minimize opportunity cost = 80U1 + 60U2 + 0S1 + 0S2 + MA 1 + MA 2
subject to 2U1 + 3U2 - 1S1 + 1A 1 = 50
4U1 + 1U2 - 1S2 + 1A 2 = 120
The first and second tableaus are shown in Table M7.18. The third tableau, containing the opti-
mal solution of U1 = 30, U2 = 0, S1 = 10, S2 = 0, opportunity cost = $2,400, appears in
Figure M7.5 along with the final tableau of the primal problem.
TABLE 9.18 First and Second Tableaus of the High Note Dual Problem
Cj 80 60 0 0 M M
SOLUTION
MIX U1 U2 S1 S2 A1 A2 QUANTITY
First $M A1 2 3 -1 0 1 0 50
tableau
$M A2 4 1 0 -1 0 1 120
Zj $6M $4M - $M - $M $M $M $170M
Cj - Zj 80 - 6M 60 - 4M M M 0 0
Second $80 U1 1 1.5 –0.5 0 0.5 0 25

tableau
$M A2 0 –5 2 -1 -2 1 20
Zj $80 $120 - 5M - $40 + 2M - $M $40 - 2M $M $2,000 + 20M
Cj - Zj 0 5M - 60 - 2M + 40 M 3M - 40 0
7 If the jth primal constraint should be an equality, the ith dual variable is unrestricted in sign. This technical
issue is discussed in L. Cooper and D. Steinberg. Methods and Applications of Linear Programming. Philadelphia:
W. B. Saunders, 1974, p. 170.
FIGURE M7.5
Comparison of the Primal’s Optimal Solution
Primal and Dual Cj $50 $120 $0 $0
Optimal Tableaus Solution Quantity
Mix X1 X2 S1 S2
$120 X2 0.5 1 0.25 0 20
$0 S2 2.5 0 –0.25 1 40
Zj 60 120 30 0 $2,400
Cj – Zj –10 0 –30 0
Dual’s Optimal Solution
Cj 80 60 0 0 M M
Solution Quantity
Mix U1 U2 S1 S2 A1 A2
80 U1 1 0.25 0 –0.25 0 0.25 30
0 S1 0 –2.5 1 –0.5 –1 0.5 10
Zj 80 20 0 –20 0 20 $2,400
Cj – Zj
0 40 0 20 M M – 20
We mentioned earlier that the primal and dual lead to the same solution even though they
are formulated differently. How can this be?
The solution to the dual yields It turns out that in the final simplex tableau of a primal problem, the absolute values of the
shadow prices. numbers in the Cj - Zj row under the slack variables represent the solutions to the dual prob-
lem, that is, the optimal Uis (see Figure M7.5). In the earlier section on sensitivity analysis we
termed these numbers in the columns of the slack variables shadow prices. Thus, the solution to
the dual problem presents the marginal profits of each additional unit of resource.
It also happens that the absolute value of the Cj - Zj values of the slack variables in the
optimal dual solution represent the optimal values of the primal X1 and X2 variables. The mini-
mum opportunity cost derived in the dual must always equal the maximum profit derived in the
primal.
Also note the other relationships between the primal and the dual that are indicated in
Figure M7.5 by arrows. Columns A 1 and A 2 in the optimal dual tableau may be ignored because,
as you recall, artificial variables have no physical meaning.
GLOSSARY M7-39
M7.13 Karmarkar’s Algorithm

The biggest change to take place in the field of LP solution techniques in four decades was the
1984 arrival of an alternative to the simplex algorithm. Developed by Narendra Karmarkar, the
new method, called Karmarkar’s algorithm, often takes significantly less computer time to solve
very large-scale LP problems.8
As we saw, the simplex algorithm finds a solution by moving from one adjacent corner
point to the next, following the outside edges of the feasible region. In contrast, Karmarkar’s
method follows a path of points on the inside of the feasible region. Karmarkar’s method is also
unique in its ability to handle an extremely large number of constraints and variables, thereby
giving LP users the capacity to solve previously unsolvable problems.
Although it is likely that the simplex method will continue to be used for many LP prob-
lems, a new generation of LP software built around Karmarkar’s algorithm is already becoming
popular. Delta Air Lines became the first commercial airline to use the Karmarkar program,
called KORBX, which was developed and is sold by AT&T. Delta found that the program
streamlined the monthly scheduling of 7,000 pilots who fly more than 400 airplanes to 166 cities
worldwide. With increased efficiency in allocating limited resources, Delta saves millions of
dollars in crew time and related costs.
Summary
In Chapter 7 we examined the use of graphical methods to (3) replacing the pivot row, (4) computing new values for each
solve LP problems that contained only two decision variables. remaining row, and (5) computing the Zj and Cj - Zj rows
This chapter moves us one giant step further by introducing the and examining for optimality. Each tableau of this iterative
simplex method. The simplex method is an iterative procedure procedure is displayed and explained for a sample maximiza-
for reaching the optimal solution to LP problems of any dimen- tion and minimization problem.
sion. It consists of a series of rules that, in effect, algebraically A few special issues in LP that arise in using the simplex
examine corner points in a systematic way. Each step moves us method are also discussed in this chapter. Examples of infeasi-
closer to the optimal solution by increasing profit or decreas- bility, unbounded solutions, degeneracy, and multiple optimal
ing cost, while maintaining feasibility. solutions are presented.
This chapter explains the procedure for converting less- Although large LP problems are seldom, if ever, solved by
than-or-equal-to, greater-than-or-equal-to, and equality con- hand, the purpose of this chapter is to help you gain an under-
straints into the simplex format. These conversions employed standing of how the simplex method works. Understanding the
the inclusion of slack, surplus, and artificial variables. An ini- underlying principles help you to interpret and analyze com-
tial simplex tableau is developed that portrays the problem’s puterized LP solutions.
original data formulations. It also contains a row providing This module also provides a foundation for another issue: an-
profit or cost information and a net evaluation row. The latter, swering questions about the problem after an optimal solution has
identified as the Cj - Zj row, is examined in determining been found, which is called postoptimality analysis, or sensitivity
whether an optimal solution had yet been reached. It also analysis. Included in this discussion is the analysis of the value of
points out which variable would next enter the solution mix, or additional resources, called shadow pricing. Finally, the relation-
basis, if the current solution was nonoptimal. ship between a primal LP problem and its dual is explored. We
The simplex method consists of five steps: (1) identifying illustrate how to derive the dual from a primal and how the
the pivot column, (2) identifying the pivot row and number, solutions to the dual variables are actually the shadow prices.
Glossary
Artificial Variable A variable that has no meaning in a Basic Feasible Solution A solution to an LP problem that
physical sense but acts as a tool to help generate an initial corresponds to a corner point of the feasible region.
LP solution.
8For details, see Narendra Karmarkar. “A New Polynomial Time Algorithm for Linear Programming,” Combinatorica
4, 4 (1984): 373–395, or J. N. Hooker. “Karmarkar’s Linear Programming Algorithm,” Interfaces 16, 4 (July–August
1986): 75–90.
Basis The set of variables that are in the solution, have posi- Range of Insignificance The range of values over which a
tive, nonzero values, and are listed in the solution mix col- nonbasic variable’s coefficient can vary without causing a
umn. They are also called basic variables. change in the optimal solution mix.
Cj - Zj Row The row containing the net profit or loss that Range of Optimality The range of values over which a ba-
will result from introducing one unit of the variable sic variable’s coefficient can change without causing a
indicated in that column into the solution. change in the optimal solution mix.
Current Solution The basic feasible solution that is the set Right-Hand-Side Ranging A method used to find the range
of variables presently in the solution. It corresponds to a over which shadow prices remain valid.
corner point of the feasible region. Shadow Prices The coefficients of slack variables in the
Degeneracy A condition that arises when there is a tie in the Cj - Zj row. They represent the value of one additional
values used to determine which variable will enter the solu- unit of a resource.
tion next. It can lead to cycling back and forth between two Simplex Method A matrix algebra method for solving LP
nonoptimal solutions. problems.
Infeasibility The situation in which there is no solution that Simplex Tableau A table for keeping track of calculations at
satisfies all of a problem’s constraints. each iteration of the simplex method.
Iterative Procedure A process (algorithm) that repeats the Slack Variable A variable added to less-than-or-equal-to
same steps over and over. constraints in order to create an equality for a simplex
Nonbasic Variables Variables not in the solution mix or ba- method. It represents a quantity of unused resource.
sis. Nonbasic variables are equal to zero. Solution Mix A column in the simplex tableau that contains
Pivot Column The column with the largest positive number all the basic variables in the solution.
in the Cj - Zj row of a maximization problem, or the Substitution Rates The coefficients in the central body of
largest negative Cj - Zj improvement value in a minimiza- each simplex table. They indicate the number of units of
tion problem. It indicates which variable will enter the solu- each basic variable that must be removed from the solution
tion next. if a new variable (as represented at any column head) is en-
Pivot Number The number at the intersection of the pivot tered.
row and pivot column. Surplus Variable A variable inserted in a greater-than-or-
Pivot Row The row corresponding to the variable that will equal-to constraint to create an equality. It represents
leave the basis in order to make room for the variable enter- the amount of resource usage above the minimum required
ing (as indicated by the new pivot column). This is the usage.
smallest positive ratio found by dividing the quantity Unboundedness A condition describing LP maximization
column values by the pivot column values for each row. problems having solutions that can become infinitely large
Primal–Dual Relationship Alternative ways of stating an without violating any stated constraints.
LP problem. Zj Row The row containing the figures for gross profit or
Quantity Column A column in the simplex tableau that gives loss given up by adding one unit of a variable into the
the numeric value of each variable in the solution mix column. solution.
Key Equation
(M7-1) 1New row numbers2 = 1Numbers in old row2 Formula for computing new values for nonpivot rows in the
simplex tableau (step 4 of the simplex procedure).
-ca b * a b
Number above or below Corresponding number
pivot number in newly replaced row
Solved Problems
Solved Problem M7-1

Convert the following constraints and objective function into the proper form for use in the simplex
method:
Minimize cost = 4X1 + 1X2
subject to 3X1 + X2 = 3
4X1 + 3X2 Ú 6
X1 + 2X2 … 3
SOLVED PROBLEMS M7-41
Solution
Minimize cost = 4X1 + 1X2 + 0S1 + 0S2 + MA 1 + MA 2
subject to 3X1 + 1X2 + 1A 1 = 3
4X1 + 3X2 - 1S1 + 1A 2 = 6
1X1 + 2X2 + 1S2 = 3
Solved Problem M7-2

Solve the following LP problem:
subject to 2X1 + 1X2 … 40
X1 + 3X2 … 30
Solution
We begin by adding slack variables and converting inequalities into equalities.
Maximize profit = 9X1 + 7X2 + 0S1 + 0S2
subject to 2X1 + 1X2 + 1S1 + 0S2 = 40
1X1 + 3X2 + 0S1 + 1S2 = 30
The initial tableau is then as follows:
Cj $9 $7 $0 $0
SOLUTION
$0 S1 2 1 1 0 40
$0 S2 1 3 0 1 30
Zj $0 $0 $0 $0 $0
Cj – Zj 9 7 0 0
The correct second tableau and third tableau and some of their calculations follow. The optimal solu-
tions, given in the third tableau, are X1 = 18, X2 = 4, S1 = 0, S2 = 0, and profit = $190.
Step 1 and 2 To go from the first to the second tableau, we note that the pivot column (in the first
tableau) is X1, which has the highest Cj - Zj value, $9. The pivot row is S1 since 40/2 is less than 30/1,
and the pivot number is 2.
Step 3 The new X1 row is found by dividing each number in the old S1 row by the pivot number,
namely, 2>2 = 1, 1>2 = 0.5, 1>2 = 0.5, 0>2 = 0, and 40>2 = 20.
Step 4 The new values for the S2 row are computed as follows:
Corresponding
a b = a b = Ca b * £
Number in Number in Number below
number in ≥ S
new S2 row old S2 row pivot number
new X1 row
0 = 1 - 3112 * 1124
2.5 = 3 - 3112 * 10.524
- 0.5 = 0 - 3112 * 10.524
1 = 1 - 3112 * 1024
10 = 30 - 3112 * 12024
Step 5 The following new Zj and Cj - Zj rows are formed:

Zj1for X12 = $9112 + 0102 = $9 Cj - Zj = $9 - $9 = 0
Zj1for X22 = $910.52 + 012.52 = $4.5 Cj - Zj = $7 - 4.5 = $2.5
Zj1for S12 = $910.52 + 01-0.52 = $4.5 Cj - Zj = 0 - 4.5 = $4.5
Zj1for S22 = $9102 + 0112 = $0 Cj - Zj = 0 - 0 = 0
Zj1profit2 = $91202 + 01102 = $180
Cj $9 $7 $0 $0
SOLUTION
$9 X1 1 0.5 0.5 0 20
0 S2 0 2.5 - 0.5 1 10 Pivot row
Zj $9 $4.5 $4.5 $0 $180
Cj - Zj 0 2.5 - 4.5 0
Pivot column
This solution is not optimal, and you must perform steps 1 to 5 again. The new pivot column is X2, the
new pivot row is S2, and 2.5 (circled in the second tableau) is the new pivot number.
Cj $9 $7 $0 $0
SOLUTION
$9 X1 1 0 0.6 - 0.2 18
7 X2 0 1 - 0.2 0.4 4
Zj $9 $7 $4 $1 $190
Cj - Zj 0 0 -4 -1
The final solution is X1 = 18, X2 = 4, profit = $190.
Solved Problem M7-3

Use the final simplex tableau in Solved Problem M7-2 to answer the following questions.
a. What are the shadow prices for the two constraints?
b. Perform RHS ranging for constraint 1.
c. If the right-hand side of constraint 1 were increased by 10, what would the maximum possible
profit be? Give the values for all the variables.
d. Find the range of optimality for the profit on X1.
Solution
a. Shadow price = -1Cj - Zj2
For constraint 1, shadow price = -1-42 = 4.
For constraint 2, shadow price = -1-12 = 1.
b. For constraint 1, we use the S1 column.
QUANTITY S1 RATIO
18 0.6 18> (0.6) = 30
4 - 0.2 4> ( - 0.2) = - 20
SOLVED PROBLEMS M7-43
The smallest positive ratio is 30, so we may reduce the right-hand side of constraint 1 by 30 units
(for a lower bound of 40 - 30 = 10). Similarly, the negative ratio of -20 tells us that we may in-
crease the right-hand side of constraint 1 by 20 units (for an upper bound of 40 + 20 = 60).
c. The maximum possible profit = Original profit + 101shadow price2
= 190 + 10142 = 230
The values for the basic variables are found using the original quantities and the substitution rates:
ORIGINAL QUANTITY S1 NEW QUANTITY

18 0.6 18 (0.6)(10) 24
4 –0.2 4 (0.2)(10) 2
X1 = 24, X2 = 2, S1 = 0, S2 = 0 1both slack varialbles remain nonbasic variables2

profit = 91242 + 7122 = 230 1which was also found using the shadow price2
d. Let ¢ = change in profit for X1.
Cj 9Δ 7 0 0
SOLUTION
9Δ X1 1 0 0.6 0.2 18
7 X2 0 1 0.2 0.4 4
Zj 9Δ 7 4 (0.6)Δ 1 (0.2)Δ 190 18Δ
Cj Zj 0 0 4 (0.6)Δ 1 (0.2)Δ
For this solution to remain optimal, the Cj - Zj values must remain negative or zero.
-4 - 10.62¢ … 0
-4 … 10.62¢
-20>3 … ¢
and
-1 + 10.22¢ … 0
10.22¢ … 1
¢ … 5
So the change in profit 1¢2 must be between -20>3 and 5. The original profit was 9, so this solu-
tion remains optimal as long as the profit on X1 is between 2.33 = 9 - 20>3 and 14 = 9 + 5.
Self-Test
䊉 Before taking the self-test, refer to the learning objectives at the beginning of the chapter, the notes in the margins, and the
glossary at the end of the chapter.
䊉 Use the key at the back of the book to correct your answers.
䊉 Restudy pages that correspond to any questions that you answered incorrectly or material you feel uncertain about.
1. A basic feasible solution is a solution to an LP problem c. there is more than one optimal solution.
that corresponds to a corner point of the feasible region. d. the solution is degenerate.
a. True 10. The pivot column in a maximization problem is the
b. False column with
2. In preparing a Ú constraint for an initial simplex a. the greatest positive Cj - Zj.
tableau, you would b. the greatest negative Cj - Zj.
a. add a slack variable. c. the greatest positive Zj.
b. add a surplus variable. d. the greatest negative Zj.
c. subtract an artificial variable. 11. A change in the objective function coefficient 1Cj2 for a
d. subtract a surplus variable and add an artificial variable. basic variable can affect
3. In the initial simplex tableau, the solution mix variables a. the Cj - Zj values of all the nonbasic variables.
can be b. the Cj - Zj values of all the basic variables.
a. only slack variables. c. only the Cj - Zj value of that variable.
b. slack and surplus variables. d. the Cj values of other basic variables.
c. artificial and surplus variables. 12. Linear programming has few applications in the real
d. slack and artificial variables. world due to the assumption of certainty in the data and
4. Even if an LP problem involves many variables, an opti- relationships of a problem.
mal solution will always be found at a corner point of the a. True
n-dimensional polyhedron forming the feasible region. b. False
a. True 13. In a simplex tableau, one variable will leave the basis and
b. False be replaced by another variable. The leaving variable is
5. Which of the following in a simplex tableau indicates a. the basic variable with the largest Cj.
that an optimal solution for a maximization problem has b. the basic variable with the smallest Cj.
been found? c. the basic variable in the pivot row.
a. All the Cj - Zj values are negative or zero. d. the basic variable in the pivot column.
b. All the Cj - Zj values are positive or zero. 14. Which of the following must equal 0?
c. All the substitution rates in the pivot column are nega- a. basic variables
tive or zero. b. solution mix variables
d. There are no more slack variables in the solution mix. c. nonbasic variables
6. To formulate a problem for solution by the simplex d. objective function coefficients for artificial variables
method, we must add slack variables to 15. The shadow price for a constraint
a. all inequality constraints. a. is the value of an additional unit of that resource.
b. only equality constraints. b. is always equal to zero if there is positive slack for that
c. only “greater than” constraints. constraint.
d. only “less than” constraints. c. is found from the Cj - Zj value in the slack variable
7. If in the optimal tableau of an LP problem an artificial column.
variable is present in the solution mix, this implies d. all of the above.
a. infeasibility. 16. The solution to the dual LP problem
b. unboundedness. a. presents the marginal profits of each additional unit of
c. degeneracy. resource.
d. alternate optimal solutions. b. can always be derived by examining the Zj row of the
8. If in the final optimal simplex tableau the Cj - Zj value primal’s optimal simplex tableau.
for a nonbasic variable is zero, this implies c. is better than the solution to the primal.
a. feasibility. d. all of the above.
b. unboundedness. 17. The number of constraints in a dual problem will equal
c. degeneracy. the number of
d. alternate optimal solutions. a. constraints in the primal problem.
9. In a simplex tableau, all of the substitution rates in the b. variables in the primal problem.
pivot column are negative. This indicates that c. variables plus the number of constraints in the primal
a. there is no feasible solution to this problem. problem.
b. the solution is unbounded. d. variables in the dual problem.
DISCUSSION QUESTIONS AND PROBLEMS M7-45
Discussion Questions and Problems

Discussion Questions M7-14 If a primal problem has 12 constraints and eight
variables, how many constraints and variables will
M7-1 Explain the purpose and procedures of the simplex
its corresponding dual have?
method.
M7-15 Explain the relationship between each number in a
M7-2 How do the graphical and simplex methods of solv-
primal and corresponding numbers in the dual.
ing LP problems differ? In what ways are they the
same? Under what circumstances would you prefer M7-16 Create your own original LP maximization problem
to use the graphical approach? with two variables and three less-than-or-equal-to
constraints. Now form the dual for this primal
M7-3 What are slack, surplus, and artificial variables?
problem.
When is each used, and why? What value does each
carry in the objective function?
Problems*
M7-4 You have just formulated an LP problem with 12 de-
cision variables and eight constraints. How many M7-17 The first constraint in the High Note example in this
basic variables will there always be? What is the dif- chapter is
ference between a basic and a nonbasic variable? 2X1 + 4X2 … 80 1hours of electrician time available2
M7-5 What are the simplex rules for selecting the pivot
column? The pivot row? The pivot number? Table M7.17 gives the final simplex tableau for this
example on page M7-34. From the tableau, it was
M7-6 How do maximization and minimization problems
determined that the maximum increase in electrician
differ when applying the simplex method?
hours was 160 (for a total of 240 hours).
M7-7 Explain what the Zj value indicates in the simplex (a) Change the right-hand side of that constraint to
tableau. 240 and graph the new feasible region.
M7-8 Explain what the Cj - Zj value indicates in the sim- (b) Find the new optimal corner point. How much
plex tableau. did the profit increase as a result of this?
M7-9 What is the reason behind the use of the minimum (c) What is the shadow price?
ratio test in selecting the pivot row? What might (d) Increase the electrician hours available by one unit
happen without it? more (to 241) and find the optimal solution. How
M7-10 A particular LP problem has the following objective much did the profit increase as a result of this one
function: extra hour? Explain why the shadow price from
the simplex tableau is no longer relevant.
Maximize profit = $8X1 + $6X2 + $12X3 - $2X4
M7-18 The Dreskin Development Company is building two
Which variable should enter at the second simplex apartment complexes. It must decide how many
tableau? If the objective function were units to construct in each complex subject to labor
and material constraints. The profit generated for
Minimize cost = $2.5X1 + $2.9X2 + $4.0X3 + $7.9X4 each apartment in the first complex is estimated at
which variable would be the best candidate to enter $900, for each apartment in the second complex,
the second tableau? $1,500. A partial initial simplex tableau for Dreskin
M7-11 What happens if an artificial variable is in the final is given in the following table:
optimal solution? What should the manager who
formulated the LP problem do? CJ $900 $1,500 $0 $0
M7-12 The great Romanian operations researcher Dr. Ima SOLUTION
Student proposes that instead of selecting the vari- MIX X1 X2 S1 S2 QUANTITY
able with the largest positive Cj - Zj value (in a 14 4 1 0 3,360
maximization LP problem) to enter the solution mix
10 12 0 1 9,600
next, a different approach be used. She suggests that
any variable with a positive Cj - Zj can be chosen, Zj
even if it isn’t the largest. What will happen if we Cj – Zj
adopt this new rule for the simplex procedure? Will
an optimal solution still be reached?
M7-13 What is a shadow price? How does the concept re- (a) Complete the initial tableau.
late to the dual of an LP problem? How does it relate (b) Reconstruct the problem’s original constraints
to the primal? (excluding slack variables).
*Note: means the problem may be solved with QM for Windows; means the problem may be
solved with Excel; and means the problem may be solved with QM for Windows and/or Excel.
(c) Write the problem’s original objective function. (b) Set up the initial simplex tableau. On the graph,
(d) What is the basis for the initial solution? identify the corner point represented by this
(e) Which variable should enter the solution at the tableau.
next iteration? (c) Select the pivot column. Which variable is the
(f) Which variable will leave the solution at the next entering variable?
iteration? (d) Compute the ratio of the quantity-to-pivot col-
(g) How many units of the variable entering the solu- umn substitution rate for each row. Identify the
tion next will be in the basis in the second tableau? points on the graph related to these ratios.
(h) How much will profit increase in the next solution? (e) How many units of the entering variable will be
M7-19 Consider the following LP problem: brought into the solution in the second tableau?
What would happen if the largest ratio rather
Maximize earnings = $0.80X1 + $0.40X2 + $1.20X3 than the smallest ratio were selected to deter-
- $0.10X4 mine this (see the graph)?
subject to X1 + 2X2 + X3 + 5X4 … 150 (f) Which variable is the leaving variable? What
X2 - 4X3 + 8X4 = 70 will the value of this variable be in the next
6X1 + 7X2 + 2X3 - X4 Ú 120 tableau?
(g) Finish solving this problem using the simplex
X1, X2, X3, X4 Ú 0
algorithm.
(a) Convert these constraints to equalities by adding (h) The solution in each simplex tableau is a corner
the appropriate slack, surplus, or artificial vari- point on the graph. Identify the corner point
ables. Also, add the new variables into the prob- associated with each tableau.
lem’s objective function. M7-22 Solve the following LP problem first graphically and
(b) Set up the complete initial simplex tableau for then by the simplex algorithm:
this problem. Do not attempt to solve.
(c) Give the values for all variables in this initial Minimize cost = 4X1 + 5X2
solution. subject to X1 + 2X2 Ú 80
M7-20 Solve the following LP problem graphically. Then 3X1 + X2 Ú 75
set up a simplex tableau and solve the problem using X1,X2 Ú 0
the simplex method. Indicate the corner points gen-
What are the values of the basic variables at each
erated at each iteration by the simplex method on
iteration? Which are the nonbasic variables at each
your graph.
iteration?
Maximize profit = $3X1 + $5X2 M7-23 The final simplex tableau for an LP maximization
subject to X2 … 6 problem is shown in the table at the bottom of this
3X1 + 2X2 … 18 page. Describe the situation encountered here.
X1, X2 Ú 0 M7-24 Solve the following problem by the simplex method.
What condition exists that prevents you from reach-
M7-21 Consider the following LP problem: ing an optimal solution?
Maximize profit = 10X1 + 8X2 Maximize profit = 6X1 + 3X2
subject to 4X1 + 2X2 … 80 subject to 2X1 - 2X2 … 2
X1 + 2X2 … 50 -X1 + X2 … 1
X1, X2 Ú 0 X1, X2 Ú 0
(a) Solve this problem graphically.
Tableau for Problem M7-23
CJ 3 5 0 0 –M
SOLUTION
MIX X1 X2 S1 S2 A1 QUANTITY
$5 X2 1 1 2 0 0 6
M A1 1 0 2 1 1 2
Zj $5 M $5 $10 2M $M $M $30 2M
Cj Zj 2 M 0 10 2M M 0
M7-25 Consider the following financial problem: kilogram of brand B supplement is $15. A kilogram
of brand A added to each production batch of Yum-
Maximize return on investment = $2X1 + $3X2
Mix provides a supplement of 1 unit of protein and
subject to 6X1 + 9X2 … 18 1 unit of riboflavin to each can. A kilogram of brand
9X1 + 3X2 Ú 9 B provides 2 units of protein and 4 units of ri-
X1, X2 Ú 0 boflavin in each can. Bitz-Karan must satisfy these
minimum nutrient standards while keeping costs of
(a) Find the optimal solution using the simplex
supplements to a minimum.
method.
(a) Formulate this problem to find the best combina-
(b) What evidence indicates that an alternate opti-
tion of the two supplements to meet the mini-
mal solution exists?
mum requirements at the least cost.
(c) Find the alternate optimal solution.
(b) Solve for the optimal solution by the simplex
(d) Solve this problem graphically as well, and illus-
method.
trate the alternate optimal corner points.
M7-29 The Roniger Company produces two products: bed
M7-26 At the third iteration of a particular LP maximization
mattresses and box springs. A prior contract requires
problem, the tableau at the bottom of this page is es-
that the firm produce at least 30 mattresses or box
tablished. What special condition exists as you im-
springs, in any combination. In addition, union labor
prove the profit and move to the next iteration?
agreements demand that stitching machines be kept
Proceed to solve the problem for the optimal solution.
running at least 40 hours per week, which is one pro-
M7-27 A pharmaceutical firm is about to begin production duction period. Each box spring takes 2 hours of
of three new drugs. An objective function designed stitching time, and each mattress takes 1 hour on the
to minimize ingredient costs and three production machine. Each mattress produced costs $20; each
constraints are as follows: box spring costs $24.
Minimize cost = 50X1 + 10X2 + 75X3 (a) Formulate this problem so as to minimize total
Subject to X1 - X2 = 1,000 production costs.
(b) Solve using the simplex method.
2X2 + 2X3 = 2,000
M7-30 Each coffee table produced by Meising Designers
X1 … 1,500
nets the firm a profit of $9. Each bookcase yields a
X1,X2,X3 Ú 0 $12 profit. Meising’s firm is small, and its resources
(a) Convert these constraints and objective function are limited. During any given production period of
to the proper form for use in the simplex tableau. one week, 10 gallons of varnish and 12 lengths of
(b) Solve the problem by the simplex method. What high-quality redwood are available. Each coffee
is the optimal solution and cost? table requires approximately 1 gallon of varnish and
M7-28 The Bitz-Karan Corporation faces a blending deci- 1 length of redwood. Each bookcase takes 1 gallon
sion in developing a new cat food called Yum-Mix. of varnish and 2 lengths of wood. Formulate Meis-
Two basic ingredients have been combined and ing’s production mix decision as an LP problem, and
tested, and the firm has determined that to each can solve using the simplex method. How many tables
of Yum-Mix at least 30 units of protein and at least and bookcases should be produced each week?
80 units of riboflavin must be added. These two What will the maximum profit be?
nutrients are available in two competing brands of M7-31 Bagwell Distributors packages and distributes indus-
animal food supplements. The cost per kilogram of trial supplies. A standard shipment can be packaged
the brand A supplement is $9, and the cost per in a class A container, a class K container, or a class
CJ $6 $3 $5 0 0 0
SOLUTION
$5 X3 0 1 1 1 0 3 5
$6 X1 1 3 0 0 0 1 12
$0 S2 0 2 0 1 1 1 10
Zj $6 $13 $5 $5 $0 $21 $97
Cj Zj $0 $16 $0 $5 $0 $21
T container. A single class A container yields a Thus, we see that the cost of carpeting a deluxe one-
profit of $8; a class K container, a profit of $6; and a bedroom unit is $1,100, the cost of carpeting a
class T container, a profit of $14. Each shipment regular one-bedroom unit is $1,000, and so on. A
prepared requires a certain amount of packing mate- total of $35,000 is budgeted for all new carpet in the
rial and a certain amount of time, as seen in the fol- building.
lowing table: Zoning regulations dictate that the building
contain no more than 50 condominiums when the
CLASS OF PACKING MATERIAL PACKING TIME conversion is completed—and no fewer than 25 units.
CONTAINER (LB) (HOURS) The development company also decides that to have a
A 2 2 good blend of owners, at least 40% but no more than
70% of the units should be one-bedroom apartments.
K 1 6
Not all money budgeted in each category need be
T 3 4 spent, although profit is not affected by cost savings.
Total amount of resource 120 lb 240 hours But since the money represents a bank loan, under no
available each week circumstances may it be exceeded or even shifted
from one area, such as carpeting, to another, such as
Bill Bagwell, head of the firm, must decide the opti- painting.
mal number of each class of container to pack each (a) Formulate Foggy Bottom Development Corpo-r
week. He is bound by the previously mentioned re- ation’s decision as a linear program to maximize
source restrictions, but he also decides that he must profits.
keep his six full-time packers employed all 240 (b) Convert your objective function and constraints
hours (6 workers, 40 hours) each week. Formulate to a form containing the appropriate slack, sur-
and solve this problem using the simplex method. plus, and artificial variables.
M7-32 The Foggy Bottom Development Corporation has M7-33 The initial simplex tableau on page M7-49 was
just purchased a small hotel for conversion to condo- developed by Tommy Gibbs, vice president of a
minium apartments. The building, in a popular area large cotton spinning mill. Unfortunately, Gibbs quit
of Washington, D.C., near the U.S. State Depart- before completing this important LP application.
ment, will be highly marketable, and each condo- Stephanie Robbins, the newly hired replacement,
minium sale is expected to yield a good profit. The was immediately given the task of using LP to deter-
conversion process, however, includes several op- mine what different kinds of yarn the mill should use
tions. Basically, four types of condominiums can be to minimize costs. Her first need was to be certain
designed out of the former hotel rooms. They are that Gibbs correctly formulated the objective func-
deluxe one-bedroom apartments, regular one-bed- tion and constraints. She could find no statement of
room apartments, deluxe studios, and efficiency the problem in the files, so she decided to recon-
apartments. Each will yield a different profit, but struct the problem from the initial tableau.
each type also requires a different level of invest- (a) What is the correct formulation, using real deci-
ment in carpeting, painting, appliances, and carpen- sion variables (that is, Xi’s) only?
try work. Bank loans dictate a limited budget that (b) Which variable will enter this current solution
may be allocated to each of these needs. Profit and mix in the second tableau? Which basic variable
cost data, and cost of conversion requirements, for will leave?
each apartment are shown in the accompanying
table.
TYPE OF APARTMENT
DELUXE REGULAR DELUXE TOTAL
RENOVATION ONE-BEDROOM ONE-BEDROOM STUDIO EFFICIENCY BUDGETED
REQUIREMENT ($) ($) ($) ($) ($)
New carpet 1,100 1,000 600 500 35,000
Painting 700 600 400 300 28,000
New appliances 2,000 1,600 1,200 900 45,000
Carpentry work 1,000 400 900 200 19,000
Profit per unit 8,000 6,000 5,000 3,500
Simplex Tableau for Problem M7-33
CJ $12 $18 $10 $20 $7 $8 $0 $0 $0 $0 $0 M M M M
SOLUTION
MIX X1 X2 X3 X4 X5 X6 S1 S2 S3 S4 S5 A1 A2 A3 A4 QUANTITY
$M A1 1 0 3 0 0 0 0 0 0 0 0 1 0 0 0 100
0 S1 0 25 1 2 8 0 1 0 0 0 0 0 0 0 0 900
M A2 2 1 0 4 0 1 0 1 0 0 0 0 1 0 0 250
M A3 18 15 2 1 15 0 0 0 1 0 0 0 0 1 0 150
0 S4 0 0 0 0 0 25 0 0 0 1 0 0 0 0 0 300
M A4 0 0 0 2 6 0 0 0 0 0 1 0 0 0 1 70
Zj $21M $14M $5M $5M $21M $M $0 $0 $M $0 $M $M $M $M $M $570M
Cj Zj 12 21M 18 + 14M 10 + 5M 20 5M 7 21M 8M 0 0 M 0 M 0 0 0 0
M7-34 Consider the following optimal tableau, where S1 and (g) How much could the right-hand side in con-
S2 are slack variables added to the original problem: straint number 2 be decreased before profit
would be affected?
CJ $10 $30 $0 $0 M7-35 A linear program has been formulated and solved.
SOLUTION The optimal simplex tableau for this is given at the
MIX X1 X2 S1 S2 QUANTITY bottom of this page.
(a) What are the shadow prices for the three con-
$10 X1 1 4 2 0 160
straints? What does a zero shadow price mean?
$ 0 S2 0 6 7 1 200 How can this occur?
Zj $10 $40 $20 $0 $1,600 (b) How much could the right-hand side of the first
Cj Zj 0 10 20 0 constraint be changed without changing the so-
lution mix (i.e., perform RHS ranging for this
constraint)?
(a) What is the range of optimality for the contribu- (c) How much could the right-hand side of the third
tion rate of the variable X1? constraint be changed without changing the
(b) What is the range of insignificance of the contri- solution mix?
bution rate of the variable X2? M7-36 Clapper Electronics produces two models of tele-
(c) How much would you be willing to pay for one phone-answering devices, model 102 1X12 and
more unit of the first resource, which is repre- model H23 1X22. Jim Clapper, vice president for
sented by slack variable S1? production, formulates their constraints as follows:
(d) What is the value of one more unit of the second
resource? Why? 2X1 + 1X2 … 40 (hours of time available on
(e) What would the optimal solution be if the profit soldering machine)
on X2 were changed to $35 instead of $30? 1X1 + 3X2 … 30 (hours of time available in inspec-
(f) What would the optimal solution be if the profit tion department)
on X1 were changed to $12 instead of $10? How Clapper’s objective function is
much would the maximum profit change? Maximize profit = $9X1 + $7X2
CJ 80 120 90 0 0 0
SOLUTION
120 X2 1.5 1 0 0.125 0.75 0 37.5
90 X3 3.5 0 1 0.125 1.25 0 12.5
0 S3 1.0 0 0 0 0.5 1 10.0
Zj 135 120 90 3.75 22.5 0 5,625
Cj Zj 55 0 0 3.75 22.5 0
Solving the problem using the simplex method, he were changed to X1 … 400, would the basis
produces the following final tableau: change? Would the values of X1, X2, and S2
change?
CJ $9 $7 $0 $0 M7-39 Formulate the dual of this LP problem:
SOLUTION Maximize profit = 80X1 + 75X2
1X1 + 3X2 … 4
$9 X1 1 0 0.6 –0.2 18
2X1 + 5X2 … 8
7 X2 0 1 –0.2 0.4 4
Find the dual of the problem’s dual.
Zj $9 $7 $4 $1 $190
M7-40 What is the dual of the following LP problem?
Cj – Zj 0 0 –4 –1
Primal: Minimize cost = 120X1 + 250X2
(a) What is the optimal mix of models 102 and H23 subject to 12X1 + 20X2 Ú 50
to produce? X1 + 3X2 Ú 4
(b) What do variables S1 and S2 represent?
M7-41 The third, and final, simplex tableau for the LP prob-
(c) Clapper is considering renting a second soldering
lem stated here follows:
machine at a cost to the firm of $2.50 per hour.
Should he do so? Maximize profit = 200X1 + 200X2
(d) Clapper computes that he can hire a part-time in- subject to 2X1 + X2 … 8
spector for only $1.75 per hour. Should he do so? X1 + 3X2 … 9
M7-37 Refer to Table M7.6, which is the optimal tableau for
the Flair Furniture Company problem. What are the solutions to the dual variables, U1 and
(a) What are the values of the shadow prices? U2? What is the optimal dual cost?
(b) Interpret the physical meaning of each shadow
price in the context of the furniture problem. CJ $200 $200 $0 $0
(c) What is the range over which the profit per table SOLUTION
can vary without changing the optimal basis MIX X1 X2 S1 S2 QUANTITY
(solution mix)? $200 X1 1 0 0.6 –0.2 3
(d) What is the range of optimality for C (number of
200 X2 0 1 –0.2 0.4 2
chairs produced)?
(e) How many hours can Flair Furniture add to or re- Zj $200 $200 $80 $40 $1,000
move from the first resource (painting depart- Cj – Zj 0 0 –80 –40
ment time) without changing the basis?
(f) Conduct RHS ranging on the carpentry depart- M7-42 The accompanying tableau provides the optimal so-
ment resource to determine the range over which lution to this dual:
the shadow price remains valid.
Minimize cost = 120U1 + 240U2
M7-38 Consider the optimal solution to the Muddy River
Chemical Corporation problem in Table M7.10. subject to 2U1 + 2U2 Ú 0.5
(a) For each of the two chemical ingredients, phosphate U1 + 3U2 Ú 0.4
and potassium, determine the range over which What does the corresponding primal problem look
their cost may vary without affecting the basis. like, and what is its optimal solution?
(b) If the original constraint that “no more than 300
pounds of phosphate can be used” 1X1 … 3002
CJ 120 240 0 0 M M
SOLUTION
MIX U1 U2 S1 S2 A1 A2 QUANTITY
$120 U1 1 0 –0.75 0.5 0.75 –0.5 0.175
240 U2 0 1 0.25 –0.5 –0.25 0.5 0.075
Zj $120 $240 –$30 –$60 $30 $60 $39
Cj – Zj 0 0 30 60 M – 30 M – 60
M7-43 Given the following dual formulation, reconstruct This means that a biophysicist can complete 8, 4,
the original primal problem: and 9 of tests 1, 2, and 3 per hour. Similarly, a bio-
chemist can perform 4 of test 1, 6 of test 2, and 4 of
Minimize cost = 28U1 + 53U2 + 70U3 + 18U4
test 3 per hour. The optimal solution to the lab’s pri-
subject to U1 + U4 Ú 10 mal problem is
U1 + 2U2 + U3 Ú 5
X1 = 8.12 hours and X2 = 13.75 hours
- 2U2 + 5U4 Ú 31
Total cost = $434.37 per day
5U3 Ú 28
12U1 + 2U3 - U4 Ú 17 The optimal solution to the dual problem is
U1,U2,U3,U4 Ú 0 U1 = 2.07, U2 = 1.63, U3 = 0
M7-44 A firm that makes three products and has three ma- (a) What is the dual of the primal LP problem?
chines available as resources constructs the follow- (b) Interpret the meaning of the dual and its solution.
ing LP problem: M7-46 Refer to Problem M7-45.
Maximize profit = 4X1 + 4X2 + 7X3 (a) If this is solved with the simplex algorithm, how
subject to 1X1 + 7X2 + 4X3 … 100 (hours on machine 1) many constraints and how many variables (in-
cluding slack, surplus, and artificial variables)
2X1 + 1X2 + 7X3 … 110 (hours on machine 2)
would be used?
8X1 + 4X2 + 1X3 … 110 (hours on machine 3) (b) If the dual of this problem were formulated and
Solve this problem by computer and answer these solved with the simplex algorithm, how many con-
questions: straints and how many variables (including slack,
(a) Before the third iteration of the simplex method, surplus, and artificial variables) would be used?
which machine still has unused time available? (c) If the simplex algorithm were used, would it be
(b) When the final solution is reached, is there any easier to solve the primal problem or the dual
unused time available on any of the three problem?
machines? M7-47 The Flair Furniture Company first described in
(c) What would it be worth to the firm to make an Chapter 7, and again in this chapter, manufactures
additional hour of time available on the third inexpensive tables (T) and chairs (C). The firm’s
machine? daily LP formulation is given as
(d) How much would the firm’s profit increase if an
Maximize profits = 70T + 50C
extra 10 hours of time were made available on
the second machine at not extra cost? subject to 4T + 3C … 240 hours of carpentry time available
M7-45 Management analysts at a Fresno laboratory have 2T + 1C … 100 hours of painting time available
developed the following LP primal problem: In addition, Flair finds that three more constraints
Minimize cost = 23X1 + 18X2 are in order. First, each table and chair must be in-
spected and may need reworking. The following
subject cost 8X1 + 4X2 Ú 120
constraint describes the time required on the average
4X1 + 6X2 Ú 115 for each:
9X1 + 4X2 Ú 116
0.5T + 0.6C … 36 hours of inspection/rework time available
This model represents a decision concerning number
of hours spent by biochemists on certain laboratory Second, Flair faces a resource constraint relating to
experiments 1X12 and number of hours spent by bio- the lumber needed for each table or chair and the
physicists on the same series of experiments 1X22. amount available each day:
A biochemist costs $23 per hour, while a biophysi- 32T + 10C … 1,248 linear feet of lumber available for
cist’s salary averages $18 per hour. Both types of sci- production
entists can be used on three needed laboratory
operations: test 1, test 2, and test 3. The experiments Finally, the demand for tables is found to be a maxi-
and their times are shown in the accompanying table: mum of 40 daily. There are no similar constraints re-
garding chairs.
T … 40 maximize table production daily
SCIENTIST TYPE MINIMUM TEST
LAB TIME NEEDED
These data have been entered in the QM for Win-
EXPERIMENT BIOPHYSICIST BIOCHEMIST PER DAY dows software that is available with this book. The
inputs and results are shown in the accompanying
Test 1 8 4 120
printout. Refer to the computer output in Programs
Test 2 4 6 115 M7.1A, M7.1B, and M7.1C in answering these
Test 3 9 4 116 questions.
(a) How many tables and chairs should Flair Furni- in the painting facility every day. Should Flair
ture produce daily? What is the profit generated sell time to the other firm? If so, how much?
by this solution? Explain.
(b) Will Flair use all of its resources to their limits (f) What is the range within which the carpentry
each day? Be specific in explaining your answer. hours, painting hours, and inspection/rework
(c) Explain the physical meaning of each shadow hours can fluctuate before the optimal solution
price. changes?
(d) Should Flair purchase more lumber if it is avail- (g) Within what range for the current solution can
able at $0.07 per linear foot? Should it hire more the profit contribution of tables and chairs
carpenters at $12.75 per hour? change?
(e) Flair’s owner has been approached by a friend
whose company would like to use several hours
PROGRAM M7.1A
QM for Windows Input
Data for Revised Flair
Furniture Problem
(Problem M7-47)
PROGRAM M7.1B
Solution Results (Final
Tableau) for Flair
Furniture Problem
(Problem M7-47)
PROGRAM M7.1C
Sensitivity Analysis
for Problem M7-47
BIBLIOGRAPHY M7-53
M7-48 A Chicago manufacturer of office equipment is des- Should the steel be purchased? All or part of the
perately attempting to control its profit and loss supply?
statement. The company currently manufactures (e) The accountants have just discovered that an er-
15 different products, each coded with a one-letter ror was made in the contribution to profit for
and three-digit designation. product N150. The correct value is actually
(a) How many of each of the 15 products should be $8.88. What are the implications of this error?
produced each month? (f) Management is considering the abandonment of
(b) Clearly explain the meaning of each shadow price. five product lines (those beginning with the let-
(c) A number of workers interested in saving money ters A through E). If no minimum monthly de-
for the holidays have offered to work overtime mand is established, what are the implications?
next month at a rate of $12.50 per hour. What Note that there already is no minimum for two of
should the response of management be? these products. Use the corrected value for N150.
(d) Two tons of steel alloy are available from an
overstocked supplier at a total cost of $8,000.
STEEL MINIMUM
ALLOY PLASTIC WOOD ALUMINUM FORMICA LABOR MONTHLY
REQUIRED REQUIRED REQUIRED REQUIRED REQUIRED REQUIRED DEMAND CONTRIBUTION
PRODUCT (LB) (SQ FT) (BD FT) (LB) (BD FT) (HOURS) (UNITS) TO PROFIT
A158 — 0.4 0.7 5.8 10.9 3.1 — $18.79

B179 4 0.5 1.8 10.3 2.0 1.0 20 6.31
C023 6 — 1.5 1.1 2.3 1.2 10 8.19
D045 10 0.4 2.0 — — 4.8 10 45.88
E388 12 1.2 1.2 8.1 4.9 5.5 — 63.00
F422 — 1.4 1.5 7.1 10.0 0.8 20 4.10
G366 10 1.4 7.0 6.2 11.1 9.1 10 81.15
H600 5 1.0 5.0 7.3 12.4 4.8 20 50.06
I701 1 0.4 — 10.0 5.2 1.9 50 12.79
J802 1 0.3 — 11.0 6.1 1.4 20 15.88
K900 — 0.2 — 12.5 7.7 1.0 20 17.91
L901 2 1.8 1.5 13.1 5.0 5.1 10 49.99
M050 — 2.7 5.0 — 2.1 3.1 20 24.00
N150 10 1.1 5.8 — — 7.7 10 88.88
P259 10 — 6.2 15.0 1.0 6.6 10 77.01
Availability
per month 980 400 600 2,500 1,800 1,000
Bibliography
See the Bibliography at the end of Chapter 7.

Management Science Module 7

Uploaded by

Copyright:

Available Formats

Management Science Module 7

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Management Science Module 7

Uploaded by

Copyright:

Available Formats

Z00_REND1011_11_SE_MOD7 PP2.

QXD 2/21/11 12:39 PM Page 1

Summary • Glossary • Key Equation • Solved Problems • Self-Test •

M7-2 MODULE 7 • LINEAR PROGRAMMING: THE SIMPLEX METHOD

M7.2 How to Set Up the Initial Simplex Solution

M7.2 HOW TO SET UP THE INITIAL SIMPLEX SOLUTION M7-3

Converting the Constraints to Equations

Finding an Initial Solution Algebraically

M7-4 MODULE 7 • LINEAR PROGRAMMING: THE SIMPLEX METHOD

The First Simplex Tableau

M7.2 HOW TO SET UP THE INITIAL SIMPLEX SOLUTION M7-5

TABLE M7.1 Flair Furniture’s Initial Simplex Tableau

Cj - Zj $70 $50 $0 $0 $0 Net profit row

D T = D T 1in vector form2

M7-6 MODULE 7 • LINEAR PROGRAMMING: THE SIMPLEX METHOD

M7.2 HOW TO SET UP THE INITIAL SIMPLEX SOLUTION M7-7

IN ACTION Resource Allocation at Pantex

C ompanies often use optimization techniques such as LP to

M7-8 MODULE 7 • LINEAR PROGRAMMING: THE SIMPLEX METHOD

M7.3 Simplex Solution Procedures

Five Steps of the Simplex Method for Maximization Problems

M7.4 THE SECOND SIMPLEX TABLEAU M7-9

M7.4 The Second Simplex Tableau

For the S2 row:

M7-10 MODULE 7 • LINEAR PROGRAMMING: THE SIMPLEX METHOD

Cj SOLUTION MIX T C S1 S2 QUANTITY

M7.4 THE SECOND SIMPLEX TABLEAU M7-11

We can now recompute the S2

Cj SOLUTION MIX T C S1 S2 QUANTITY

M7-12 MODULE 7 • LINEAR PROGRAMMING: THE SIMPLEX METHOD

Interpreting the Second Tableau

M7.5 DEVELOPING THE THIRD TABLEAU M7-13

M7.5 Developing the Third Tableau

M7-14 MODULE 7 • LINEAR PROGRAMMING: THE SIMPLEX METHOD

Cj SOLUTION MIX T C S1 S2 QUANTITY

Number Number Number Corresponding

M7.5 DEVELOPING THE THIRD TABLEAU M7-15

Cj SOLUTION MIX T C S1 S2 QUANTITY

M7-16 MODULE 7 • LINEAR PROGRAMMING: THE SIMPLEX METHOD

M7.6 Review of Procedures for Solving LP Maximization Problems

M7.7 Surplus and Artificial Variables

M7.7 SURPLUS AND ARTIFICIAL VARIABLES M7-17

M7-18 MODULE 7 • LINEAR PROGRAMMING: THE SIMPLEX METHOD

Surplus and Artificial Variables in the Objective Function

M7.8 Solving Minimization Problems

The Muddy River Chemical Company Example

M7.8 SOLVING MINIMIZATION PROBLEMS M7-19

Linear Programming Modeling

problem,” says Professor Andres Weintraub. “The model and its

Here is the mathematical This problem may be restated mathematically as

M7-20 MODULE 7 • LINEAR PROGRAMMING: THE SIMPLEX METHOD

Converting the Constraints and Objective Function

M7.8 SOLVING MINIMIZATION PROBLEMS M7-21

The complete set of constraints can now be expressed as follows:

Rules of the Simplex Method for Minimization Problems

Steps for Simplex Minimization Problems

M7-22 MODULE 7 • LINEAR PROGRAMMING: THE SIMPLEX METHOD

Cj SOLUTION MIX X1 X2 S1 S2 A1 A2 QUANTITY

M7.8 SOLVING MINIMIZATION PROBLEMS M7-23