Optimization Methods in Management Science

MIT 15.053, Spring 2013

Problem Set 2 (Second Group of Students)

Students with rst letter of surnames

IZ

Due: February 21 , 2013

Problem Set Rules:
1. Each student should hand in an individual problem set.
2. Discussing problem sets with other students is permitted. Copying from another person
or solution set is not permitted.
3. Late assignments will not be accepted. No exceptions.
4. The non-Excel solution should be handed in at the beginning of class on the day the
problem set is due. The Excel solutions, if required, should be posted on the website by
the beginning of class on the day the problem set is due. Questions that require an Excel
submission are marked with Excel submission . For Excel submission questions, only
the Excel spreadsheet will be graded.

Problem 1
(33 points total) Consider the following linear program:
max x1 + 2x2
s.t.:
Const 1 x1 x2
Const 2 x1 + x2
Const 3
x1
Const 4
x2
x1 , x2

2.5

0.

(LP)

(a) (5 points) Graph the feasible region of the LP. Is the feasible region unbounded?
Solution.
The feasible region and objective function are sketched in Figure 1. No.
The feasible region is bounded.
(b) (4 points) Are any of the above constraints redundant? If so, indicate which one(s). (For
large linear programs, eliminating redundant constraints can speed up the solution of the
linear program.)
Solution.
Yes, the constraint x2 3 is redundant since deleting this constraint does
not change the feasible region.
(c) (4 points) Solve the LP using the graphical method. Explain your approach.
Solution.

The optimal solution is clearly the point with coordinates (1, 3).

x2

x1

Figure 1: Feasible region for Part 1.A.

(d) (4 points) Is there more then one optimal solution? If so give two dierent solutions. If
not, explain using the graphical method why not?
(e) (9 points, 3 points each) Suppose we add the constraint 2x1 + x2 to (LP). For which
values of :
is the constraint redundant?

the optimal solution found above is no longer optimal?

the problem becomes infeasible?

To answer these questions, you should use the graph of the feasible region drawn in Part
(a).
Solution.
The constraint 2x1 + x2 has slope -2, and the feasible halfspace is the
one lying to the right of the line with equation 2x1 + x2 = . Therefore it is easy to see
that the constraint is redundant if 2x1 + x2 = passes through the point (0, 0) or if the
intercept on the x1 axis is negative. This happens for [, 0] (for = 0, the line
goes through the origin). A sketch of the situation is given in Figure 2. Note that the
value = 0 is acceptable. The solution (1, 3) is no longer optimal if it is cut o from the
feasible region by the constraint. The line 2x1 + x2 = passes through (1, 3) for = 5.
Therefore the solution is no longer optimal for (5, ] (the value = 5 is excluded).
Finally, the problem is infeasible if all feasible points are cut o. The last point to be cut
o is (2.5, 1.5), for > 6.5. Hence the problem is infeasible for (6.5, ].
(f) (4 points) Replace the objective function x1 + 2x2 with the objective function x1 + x2 , and
compute the values of for which the point (2.5, 1.5) is optimal.
Solution. Starting from the original objective function x1 +2x2 , which has slope 1/2,
we easily see that we have to decrease the slope of the objective function in order to make
2

x2

2x1 + x2 5

2x1 + x2 6.5

x1
2x1 + x2 0

Figure 2: Initial feasible region plus constraint 2x1 + x2 for Part 1.B.
(2.5, 1.5) optimal. Notice that the slope is equal to 1/. When = 1, the objective
function is parallel to the constraint x1 + x2 4 with slope 1, and the whole segment
between (1, 3) and (2.5, 1.5) is optimal. In order to decrease the slope from 1, we have
to decrease . As approaches 0, the objective function becomes a vertical line and the
whole segment between (2.5, 0) and (2.5, 1.5) is optimal.
It follows that (2.5, 1.5) is optimal for [0, 1].

Problem 2

(38 points total) A company makes three lines of tires. Its four-ply biased tires produce $6 in
prot per tire; its berglass belted line $4 a tire; and its radials $8 a tire. Each type of tire
passes through three manufacturing stages as a part of the entire production process.
Each of the three process centers has the following hours of available production time per
day:
1
2
3

Process
Modeling
Curing
Assembly

Hours
12
14
16

The time required in each process to produce one hundred tires of each line is as follows:

Tire
Four-ply
Fiberglass
Radial
1

Hours per 100 units

Modeling Curing Assembly
2
3
2
2
2
1
4
2
2

This problem is based on Problem 6 of Applied Mathematical Programming, Chapter 2.

(a) (5 points) Write a linear program to determine the optimum product mix for each days
production, assuming all tires are sold.
Solution.
We dene 3 decision variables that stand for the number of tires of each
type to be produced each day: xF P for the Fours-ply, xF G for Fiberglass, and xR for
radial. The problem can be formulated as follows:

max
6xF P + 4xF G + 8xV

s.t.:

Modeling : 0.02xF P + 0.02xF G + 0.04xR 12

Curing : 0.03xF P + 0.02xF G + 0.02xR 14

Assembly : 0.02xF P + 0.01xF G + 0.02xR 16

xF P , xF G , xR 0.
(b) (10 points) Excel submission Solve the problem using the simplex algorithm, employing
the Excel spreadsheet given with this problem set. Make sure you understand the formu
lation and the meaning of the variables, and ll in the missing coecients of the tableau.
The spreadsheet that you submit should contain the sequence of tableaus that leads to the
optimal tableau only. You will then use the same spreadsheet to help through the rest of
Problem 2, but make sure you submit a spreadsheet containing the sequence of tableaus
leading to the optimal tableau for the original data!
(c) (8 points) Using the Excel spreadsheet to carry out the calculations, answer the following
question:
(1) What is the initial feasible solution? Give the value for all the decision variables and
all the slack variables.
(2) What is the optimal solution? Give its objective function value and the value of all the
decision variables.
Solution.
The initial feasible solution is xF G = xF G = xR = 0, s1 = 12, s2 = 14, s3 =
16. The optimal solution is xF P = 400, xF G = 0, xR = 100, s1 = 0, s2 = 0, s3 = 6, with
objective function value 3200.
(d) (15 points, 5 points each) Using the Excel spreadsheet from the previous point to recompute
the optimal solution of the LP, answer the following questions:
(i) Suppose we increase the number of modeling hours per day from 12 to 13. How much
does the prot increase? (Comment: This is the shadow price of the constraint on the
modeling hours. Read the tutorial on Sensitivity Analysis in 2 Dimensions to learn
more about shadow prices.)
(ii) What would be the increase in the prot if we increase the number of assembly hours
per day from 16 to 17 (Assume that the number of modeling hours per day is 12. )
(iii) (Extra credit) Consider the answer to the previous two questions. What relation
ship do you observe between the shadow price of a constraint and the value of the
corresponding slack variable in the optimal solution?
Solution.

(i) The shadow price for the Modeling constraint is $150 per hour of production. There
fore we could pay as much as $150 in order to increase our production capabilities in
terms of Modeling time by one hour. Any larger amount would not be protable.
(ii) The shadow price for the Assembly constraint is 0. It is not protable to pay to
increase the number of Assembly hours.
(iii) Constraints such that the corresponding slack variable has zero value in the optimal
solution (i.e. tight constraints) have nonzero shadow price. Constraints such that the
corresponding slack variable has nonzero value in the optimal solution (i.e. non-tight
constraints) have zero shadow price.

Problem 3 (Second group of students)

(34 points total) A corporation that produces gasoline and oil specialty additives purchases four
grades of petroleum distillates, A, B, C, and D. The company then combines the four distillates
to make three mixtures (Deluxe, Standard, and Economy) according to specications of the
maximum and/or minimum percentages of grades A, C, or D in each blend, given in Table 1.

Mixture
Deluxe
Standard
Economy

Max %
allowed for
Additive A
60%
15%

Min %
allowed for
Additive C
20%
60%
50%

Max %
allowed for
Additive D
10%
25%
45%

Selling
price
$/gallon
7.9
6.9
5.0

Table 1: Specications of the three mixtures.

Supplies of the three basic additives and their costs are given in Table 2.

Distillate
A
B
C
D

Max quantity
available
per day (gals)
4000
5000
3500
5500

Cost
$/gallon
0.60
0.52
0.48
0.35

Table 2: Supplies and costs of petroleum grades.

We write a linear program to determine the production policy that maximizes prots. To do
that, we dene twelve decision variables to indicate the gallons of each of the basic petroleum
grades (additives) that contribute to the production of the three mixtures. We use the following
labels: xAD indicates the quantity of additive A that is used for production of mixture Deluxe,
xAS indicates the quantity of additive A that is used for production of mixture Standard, xAE
indicates the quantity of additive A that is used for production of mixture Economy. Similarly,
we use xBD , xBS , xBE to label the quantity of additive B used in each of the three mixtures,
xCD , xCS , xCE for additive C, and and xDD , xDS , xDE for additive D. The gallons of mixture
Deluxe produced can be computed as xAD + xBD + xCD + xDD ; similarly for mixture Standard
and Economy. The objective function is the total prot, computed as revenues minus costs.
2

This problem is based on Problem 11 of Applied Mathematical Programming, Chapter 1.

The problem can be formulated as

max

s.t.:
Availability A :
Availability B :
Availability C :
Availability D :
Mixture D Add A :
Mixture D Add C :
Mixture D Add D :
Mixture S Add A :
Mixture S Add C :
Mixture S Add D :
Mixture E Add C :
Mixture E Add D :

7.3xAD + 7.38xBD + 7.42xCD + 7.55xDD

6.3xAS + 6.38xBS + 6.42xCS + 6.55xDS
4.4xAE + 4.48xBE + 4.52xCE + 4.65xDE
xAD + xAS + xAE
xBD + xBS + xBE
xCD + xCS + xCE
xDD + xDS + xDE
0.4xAD 0.6xBD 0.6xCD 0.6xDD
0.2xAD 0.2xBD + 0.8xCD 0.2xDD
0.1xAD 0.1xBD 0.1xCD + 0.9xDD
0.85xAS 0.15xBS 0.15xCS 0.15xDS
0.6xAS 0.6xBS + 0.4xCS 0.6xDS
0.25xAS 0.25xBS 0.25xCS + 0.75xDS
0.5xAE 0.5xBE + 0.5xCE 0.5xDE
0.45xAE 0.45xBE 0.45xCE + 0.55xDE
xAD , xAS , xAE , xBD , xBS , xBE , xCD , xCS , xCE , xDD , xDS , xDE

4000
5000
3500
5500
0

0.

You are given an Excel spreadsheet that solves this problem. Suppose that the selling prices
are changed to $7.7 per gallon for Deluxe, $6.8 per gallon for Standard, and $4.9 per gallon
for Economy. In addition, the marketing department estimates that the demand for Deluxe
does not exceed 12000 gallons, therefore the corporation does not want to produce more than
this quantity for Deluxe. Assume that all other data remains unchanged. Update the Excel
spreadsheet and then answer the following questions:
(a) (10 points) What is the optimal prot and the optimal production-mix?
Solution.
The optimal prot is $97801.25 with optimal solution xAD = 3675, xBD =
4725, xBS = 275, xCD = 2400, xCS = 1100, xDD = 1200, xDS = 458.333, and all remaining
variables are zero.
(b) (5 points) No mixture Economy is produced in the optimal solution. What would the
minimum selling price need to be in order for Economy to be worth producing? (Be accurate
to within 5 cents).
Solution.
$5.75.

It is worth producing mixture Economy when it selling price increases to

(c) (6 points) Suppose that you can increase the selling price of Standard. What would be
the increase in the prot if the selling price of mixture Standard per gallon increases to
6.8 + s for s = $ 0.05, $0.1, and $0.15 . (Assume that the selling price of mixtures Deluxe
and Economy remain $7.7 and $4.9 per gallon, respectively.) The increase is the dierence
between the new prot and the prot from Part (a).
Solution.
The increase in the prot for s = $ 0.05, $0.1, and $0.15 is $ 91.6667,
$183.333, $275, respectively.
(d) (8 points) Based on your answer in Part (c), estimate the prot if the selling price of mixture
Standard increases to 7.2. What is the formula for the optimum prot if the selling price
6

of Standard per gallon increased to $6.8 + s? (You may assume that s is between 0.05 and
0.25).
Solution.

The increase in the prot at price s (for small enough s) is 20 s 91.6667.

(e) (5 points) Based on your formula in part (d), estimate the contribution (the increase in
optimal prot) if the selling price of Standard per gallon increases to 7.5. Use Excel solver
to see if your estimation is correct.
Solution.
The increase by the formula is $12.8334, and the increase in the prot by
solving the problem is $12.8334. The formulae is still correct.

MIT OpenCourseWare
http://ocw.mit.edu

15.053 Optimization Methods in Management Science

Spring 2013

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.