You are on page 1of 33

OPTIMIZATION TECHNIQUES

CASE STUDY
Presented To: Dr. Salman Habib
Presented By: 2020-IM-10
2020-IM-19
2020-IM-20
OUTLINE
■ Introduction to the Case
■ Defining Decision Variables
■ Objective Function
■ Constraints
■ Results and Validation
■ Sensitivity Analysis
2
CASE
■ Atlantic Management Systems is a consulting firm
that specializes in developing computerized decision
support systems for computer manufacturing
companies.
■ The firm currently has offices in Chicago, Charlotte,
Pittsburgh, and Houston.
■ It is considering opening new offices in one or more
cities including Atlanta, Boston, Washington DC,
St. Louis, Miami, Denver, and Detroit
3
■ The firm has $14 million available for this
purpose.
■ Because of the highly specialized nature of its
high-tech consulting work, the firm must
necessarily staff any new offices with a minimum
number of its employees from its existing offices.
■ However, a limited number of employees can
transfer to any new offices.
■ The cost of transferring employees depends on the
city they are leaving and the city to which they
might move.
4
Following are the costs for opening a new office in each of the prospective cities and the
start-up staffing needs at each office:
PROSPECTIVE SETUP COST STAFFING
OFFICE ($1,000,000S) NEEDS
EMPLOYEES
1. ATLANTA $1.7 9
2. BOSTON 3.6 14
3. DENVER 2.1 8
4. DETROIT 2.5 12
5. MIAMI 3.1 11
6. ST. LOUIS 2.7 7
7. WASHINGTON 4.1 18
DC 5
The numbers of employees available for transfer from each of the current offices are
as follows:
EXISTING OFFICE AVAILABLE EMPLOYEES

1.CHICAGO 24

2. CHARLOTTE 19

3. PITTSBURGH 16

4. HOUSTON 21

6
The cost of transferring an employee from the existing office to the new office differs
according to housing cost plus moving expenses plus the cost of living adjustments:
PERSPECTIVE NEW OFFICES(costs, $1,000s)

CURRENT ATLANT BOSTO DENVER DETROI MIAM ST. WASHINGT


OFFICE A N T I LOUIS ON DC

1. CHICAGO $19 $32 $27 $14 $23 $14 $41

2. CHARLOTTE 14 47 31 28 35 18 53

3. 16 39 26 23 31 19 48
PITTSBURGH

4. HOUSTON 22 26 21 18 28 24 43
7
The firm has ranked the possible new offices according to their profit potential, with
Washington, DC, being the best (i.e., greatest potential), as follows:

NEW OFFICE RANK


WASHINGTON DC 1
MIAMI 2
ATLANTA 3
BOSTON 4
DENVER 5
ST. LOUIS 6
DETROIT 7

8
■ The firm wants at least one new office:
• In the Midwest (i.e., Detroit and/or St. Louis).
• One new office in the Southeast (i.e., Atlanta or Miami).
■ Formulate and solve an integer programming model to help
Atlantic Systems determine:
• How many new offices it should open?
• Where they should be located?
• How to transfer employees?

9
DECISION VARIABLES
■ Xij = The number of employees to transfer from the
previous cities’ offices to those in new cities, where
(i=1,2,3,4) and (j=A,B,C,D,E,F,G).
■ Yj= new office at a different location.
■ Yj is the Binary decision variable to select whether or
not the city should be selected. It will be equal to 1 if
selected and 0 if not.

10
(a) ATLANTA
1.Chicago (b)
BOSTON
2.Charlotte (c) DENVER

(d) DETROIT
3.Pittsburgh
(e) MIAMI

4.Houston
(f) St. LOUIS

(g)
WASHINGTON 11
GENERIC FORM
■ Indices:
o i Existing cities (i=1, 2, 3, 4)
o j New cities (j= a , b, c, d, e, f, g)
■ Notations:
o Available employees
o Weights applied= Rj
at the old city to
transfer= Si
o New offices at different locations= Yj o Cost of
o Employee demand at new offices= Dj transportation= Cij
o No. of employees transfer from o Setup cost= Cj

existing to new(i-j)= Xij


12
New office Rank Weight
s
Washington, 1
DC(g)
7 OBJECTIVE
Miami (e) 2 6 FUNCTION
■ Maximize Z=
Atlanta (a) 3 5 5ya+4yb+3yc+1yd+6ye+2yf+
Boston (b) 4 4
7yg
■ Generic Form
Denver (c) 5 3 g
Max= ∑ Rj*Yj
St. Louis (f) 6 2 J=a
(j=a,b,c,d,e,f,g)
Detroit (d) 7 1
13
EMPLOYEE CONSTRAINTS PROSPECTIVE STAFFING
OFFICE NEEDS
FOR NEW OFFICES (Demand) EMPLOYEES
■ x1a + x2a + x3a + x4a = 9ya a. ATLANTA 9
■ x1b + x2b + x3b+ x4b = 14yb
b.BOSTON 14
■ x1c + x2c + x3c + x4c = 8yc
c.DENVER 8
■ x1d + x2d + x3d + x4d = 12yd
■ x1e + x2e + x3e + x4e = 11ye d.DETROIT 12

■ x1f + x2f + x3f + x4f = 7yf e.MIAMI 11

■ x1g + x2g + x3g + x4g = 18yg f.ST. LOUIS 7


4
18
 Generic Form: ∑ Xij = Dj * Yj ∀j
i=1 g.WASHINGT
ON DC
14
AVAILABLE EMPLOYEES CONSTRAINTS
(Supply)
EXISTING AVAILABLE
■ x1a + x1b + x1c + x1d + x1e + x1f + x1g OFFICE EMPLOYE
≤ 24 ES
1.CHICAGO 24
■ x2a + x2b + x2c + x2d + x2e + x2f + x2g
≤ 19
2. 19
■ x3a + x3b+ x3c + x3d + x3e+x3f + x3g CHARLOT
TE
≤ 16 3. 16
■ x4a + x4b + x4c + x4d + x4e + x4f + x4g PITTSBUR
GH
≤ 21 g 4. 21
 Generic Form: ∑ Xij ≤ Si ∀ i (i=1,2,3,4) HOUSTAN
J=a
15
BUDGET CONSTRAINT (Total $14 million)
■ 19x1a + 32x1b + 27x1c + 14x1d + 23x1e + 14x1f + 41x1g + 14x2a + 47x2b + 31x2c
+ 28x2d + 35x2e + 18x2f + 53x2g + 16x3a + 39x3b + 26x3c + 23x3d + 31x3e +
19x3f + 48x3g + 22x4a + 26x4b + 21x4c + 18x4d + 28x4e + 24x4f + 43x4g +
1,700ya + 3,600yb + 2,100yc + 2,500yd + 3,100ye + 2,700yf + 4,100yg ≤ 14000
 Generic Form: ∑∑ Cij * Xij + ∑ Cj * Yj ≤
i j j
B

16
■ yd + y f ≥ 1 At least one new office in the Midwest (i.e., Detroit and/or St.
Louis).
■ Generic form :
■ ya + ye = 1 One new office in the Southeast (i.e., Atlanta or Miami).
■ Generic form :

Non-negative Constraints:
■ xij ≥ 0
■ yj = 0 or 1

17
Non negative constraints:
X1a=0 X2a X3a X4a
X1b=0 X2b=0 X3b=0 X4b=0
X1c=0 X2c=0 X3c=0 X4c=0
X1d=0 X2d=0 X3d=0 X4d=0
X1e=0 X2e=0 X3e=0 X4e=0
X1f=0 X2f=0 X3f=0 X4f=0
X1g=0 X2g=0 X3g=0 X4g=0

18
Ya=0 or 1
Yb=0 or 1
Yc=0 or 1
Yd=0 or 1
Ye=0 or 1
Yf=0 or 1
Yg=0 or 1

19
■ Maximize Z=5ya+4yb+3yc+1yd+6ye+2yf+7yg
Subjected To:
■ x1a + x2a + x3a + x4a - 9ya = 0  19x1a + 32x1b + 27x1c + 14x1d + 23x1e + 14x1f +
■ x1b + x2b + x3b+ x4b - 14yb = 0 41x1g + 14x2a + 47x2b + 31x2c + 28x2d + 35x2e +
18x2f + 53x2g + 16x3a + 39x3b + 26x3c + 23x3d +
■ x1c + x2c + x3c + x4c - 8yc = 0
31x3e + 19x3f + 48x3g + 22x4a + 26x4b + 21x4c +
■ x1d + x2d + x3d + x4d - 12yd = 0 18x4d + 28x4e + 24x4f + 43x4g + 1,700ya +
■ x1e + x2e + x3e + x4e - 11ye = 0 3,600yb + 2,100yc + 2,500yd + 3,100ye + 2,700yf +
■ x1f + x2f + x3f + x4f - 7yf = 0 4,100yg ≤ 14000,
■ x1g + x2g + x3g + x4g - 18yg = 0
■ x1a + x1b + x1c + x1d + x1e + x1f + x1g ≤ 24
■ x2a + x2b + x2c + x2d + x2e x2f + x2g ≤ 19
■ x3a + x3b+ x3c + x3d + x3e+x3f + x3g ≤ 16
■ x4a + x4b + x4c + x4d + x4e + x4f + x4g ≤ 21
■ yd + yf ≥ 1
■ ya + ye = 1
■ xij ≥ 0
■ yj = 0 or 1
20
RESULTS:
■ It is obtained that the following 4 cities are selected for the
new offices: Variable Value
1)Atlanta ya Ya 1
2)Boston yb Yb 1
3)Miami ye Yc 0
4)Washington DC yg Yd 0
Ye 1
Yf 0
Yg 1
18
RESULTS:
Objective Value: 22
Variable Value Variable Value
X1a 0 X2c 0
X1b 0 X2d 0
X1c 0 X2e 0
X1d 0 X2f 0
X1e 11 X2g 0
X1f 0 X3a 0
X1g 13 X3b 0
X2a 9 X3c 0
x2b 0 X3d 0
22
RESULTS:
Variable Values Variable Value
X3e 0 X4g 5
X3f 0 Ya 1
X3g 0 Yb 1
X4a 0 Yc 0
X4b 14 Yd 0
X4c 0 Ye 1
X4d 0 Yf 0
X4e 0 Yg 1
x4f 0

23
RESULTS:
yd + yf ≥ 1 At least one new office in the Midwest (i.e., Detroit
and/or St. Louis).
ya + ye = 1 One new office in the Southeast (i.e., Atlanta or
Miami).

Variable Value Variable Value


Yd 1 Ya 1
yf 0 ye 0

24
Data Validation
AVAILABLE EMPLOYEES CONSTRAINTS Variabl Value
(Supply Constraints): e
■ x1a + x1b + x1c + x1d + x1e + x1f + x1g ≤ 24 X1e 11
0+0+0+0+11+0+13 24 =24
X1g 13
■ x2a + x2b + x2c + x2d + x2e +x2f + x2g ≤ 19
9 ≤ 19 X2a 9
9+0+0+0+0+0+0
■ x3a + x3b+ x3c + x3d + x3e+x3f + x3g ≤ 16 x4b 14
0+0+0+0+0+0+0 0 ≤ 16 x4g 5
■ x4a + x4b + x4c + x4d + x4e + x4f + x4g ≤ 21
0+14++0+0+0+5 19 ≤ 21

25
Data Validation
EMPLOYEE CONSTRAINTS FOR NEW OFFICES (Demand Constraints)

■x1a + x2a + x3a + x4a - 9ya = 0


0+ 9+0+0-9(1)=0 0=0
■x1b + x2b + x3b+ x4b - 14yb = 0
Variabl Value
Variable Value e
0+14+0+0-14(1)=0 0=0
■x1c + x2c + x3c + x4c - 8yc = 0 Ya 1 X1e 11
0+0+0+0-8(0)=0 0=0
■x1d + x2d + x3d + x4d - 12yd = 0
Yb 1 X1g 13
0+0+0+0-12(0)=0 0=0 Yc 0
X2a 9
Yd 0
Ye 1 x4b 14
Yf 0 x4g 5
Yg 1
26
Data Validation:
Variabl Value
■ x1e + x2e + x3e + x4e - 11ye = 0 e
X1e 11
11+0+0+0-11(1)=0 0=0
X1g 13
■ x1f + x2f + x3f + x4f - 7yf = 0
X2a 9
0+0+0+0-7(0)=0 0=0
x4b 14
■ x1g + x2g + x3g + x4g - 18yg = 0
x4g 5
13+0+0+0+5-18(1)=0 0=0

27
Validation:
■ BUDGET CONSTRAINT:
19x1a + 32x1b + 27x1c + 14x1d + 23x1e + 14x1f + 41x1g +
14x2a + 47x2b + 31x2c + 28x2d + 35x2e + 18x2f + 53x2g +
16x3a + 39x3b + 26x3c + 23x3d + 31x3e + 19x3f + 48x3g +
22x4a + 26x4b + 21x4c + 18x4d + 28x4e + 24x4f + 43x4g +
1,700ya + 3,600yb + 2,100yc + 2,500yd + 3,100ye +
2,700yf + 4,100yg ≤ 14000
23(11)+14(9)+26(14)+43(5)+1700+3600+3100+
13458 ≤ 14000

28
Data Validation
■ Objective Function:
Variable Value
Maximize Z= 5ya+4yb+3yc+1yd+6ye+2yf+7yg Ya 1
Z=5(1)+4(1)+3(0)+1(0)+6(1)+2(0)+7(1) Yb 1
Yc 0
Z= 22
Yd 0
Ye 1
Yf 0
Yg 1

29
Sensitivity Analysis
New office Rank Weight OBJECTIVE
Detroit(d) 1
■ Change the ranking s FUNCTION
St. Louis(f) 7
2
Atlanta (a) 3
6 ■Maximize Z=
Boston (b) 4
5 5ya+4yb+3yc+7yd+2
Denver (c) 5
4 ye+6yf+1yg
3
Miami(e) 6
2
Washington DC(g) 7
1 30
Sensitivity Analysis:
Results Variable Value
Ya 1
■ Objective Value is 25
Yb 1
1)Atlanta ya
2)Boston yb Yc 1
3)Denver yc Yd 1
4)Detroit yd Ye 0
5)St. Louis ye
Yf 1
Yg 0

31
Results: Variable Value
Variabl Valu Variable Value X3e 0
e e
X3f 0
X1a 0 X2c 0
X3g 0
X1b 0 X2d 0 X4a 0
X1c 0 X2e 0 X4b 14
X1d 12 X2f 0 X4c 7
X1e 7 X2g 0 X4d 0
X1f 0 X3a 0 X4e 0
X1g 0 X3b 0 x4f 0
X2a 9 X3c 1 x4g 0
x2b 0 X3d 0 32
THANK YOU

You might also like