Vous êtes sur la page 1sur 11

Discreet optimization of business processes

Assignment 1 Depot location problem



1.1 (A)- No depot capacity restrictions-first model
Data:
City number City Annual demand Annual depot costs
1 Enschede 20 75
2 Hengelo 25 90
3 Almelo 12 40
4 Oldenzaal 10 60
5 Delden 16 45


The problem can be formulated as an ILP with the variables X
d
and Y
sd
. The formulation for the
program is the following:
The qualitative model:
Objective: minimize the annual costs;
Restrictions:
- Each cities demand for products; the number of products needed to be delivered to each city.
- The number of products delivered; the number of products that can be delivered from each
depot to the cities.
- Transport costs; the cost of transporting a product from a certain depot to a certain city.
- The annual costs; the annual costs of keeping a depot operational;
- The annual demand of products;
The LP formulation of the problem:
Parameters: - TransportCost(s, d); transport costs from depot d to city s;
-Demand(s); city s demand of products;
-DepotCost(d); depot d annual costs;
-s {1, 2, 3, 4, 5}
-d {1, 2, 3, 4, 5}
Variables: - X(d); indicates if depot d is opened or closed;
-Y(s, d); indicates the number of products that depot d has to deliver to city s;
Constraints: - ad(s): SUM[d,Y(s,d)]>=demand(s); annual demand should be equal to the number of
products delivered
H(s,d): Y(s,d)<= X(d)*bigM;
Objective function
() () ( ) ( )


LP relaxation:
X(d) [0,1];
Y(s,d) [0,];
Results for the optimization of the LP-model
LP_SOLVE MESSAGES:
Feasible solution 295 after 15 iter, 4 nodes (gap 40.3%)
========================================
Value of objective function: 295.00000000
========================================
VARIABLE VALUES:
X_Almelo 1
X_Delden 1
X_Enschede 1
X_Hengelo 0
X_Oldenzaal 1
Y_Almelo_Almelo 12
Y_Almelo_Delden 0
Y_Almelo_Enschede 0
Y_Almelo_Hengelo 0
Y_Almelo_Oldenzaal 0
Y_Delden_Almelo 0
Y_Delden_Delden 16
Y_Delden_Enschede 0
Y_Delden_Hengelo 0
Y_Delden_Oldenzaal 0
Y_Enschede_Almelo 0
Y_Enschede_Delden 0
Y_Enschede_Enschede 25
Y_Enschede_Hengelo 0
Y_Enschede_Oldenzaal 0
Y_Hengelo_Almelo 0
Y_Hengelo_Delden 0
Y_Hengelo_Enschede 25
Y_Hengelo_Hengelo 0
Y_Hengelo_Oldenzaal 0
Y_Oldenzaal_Almelo 0
Y_Oldenzaal_Delden 0
Y_Oldenzaal_Enschede 0
Y_Oldenzaal_Hengelo 0
Y_Oldenzaal_Oldenzaal 10
========================================
The depots that are open are Almelo, Delden and Enschede.
Almelo is supplied by the depot in Almelo, Delden is supplied by the depot in Delden and Enschede,
Hengelo and Oldenzaal are supplied by the depot in Enschede.
The total costs for the optimal solution are 295. The integrality gap for this solution is 40.3%

Results for the LP-relaxation for the optimization of the model
LP_SOLVE MESSAGES:
========================================
Value of objective function: 210.00000000
========================================
VARIABLE VALUES:
X_Almelo 0.48
X_Delden 0.64
X_Enschede 0.8
X_Hengelo 0.2
X_Oldenzaal 0.4
Y_Almelo_Almelo 12
Y_Almelo_Delden 0
Y_Almelo_Enschede 0
Y_Almelo_Hengelo 0
Y_Almelo_Oldenzaal 0
Y_Delden_Almelo 0
Y_Delden_Delden 16
Y_Delden_Enschede 0
Y_Delden_Hengelo 0
Y_Delden_Oldenzaal 0
Y_Enschede_Almelo 0
Y_Enschede_Delden 0
Y_Enschede_Enschede 20
Y_Enschede_Hengelo 0
Y_Enschede_Oldenzaal 0
Y_Hengelo_Almelo 0
Y_Hengelo_Delden 0
Y_Hengelo_Enschede 20
Y_Hengelo_Hengelo 5
Y_Hengelo_Oldenzaal 0
Y_Oldenzaal_Almelo 0
Y_Oldenzaal_Delden 0
Y_Oldenzaal_Enschede 0
Y_Oldenzaal_Hengelo 0
Y_Oldenzaal_Oldenzaal 10
========================================
The solution for the LP-relaxation is not feasible because a depot can be only open or closed. This
solution shows that a depot is open only a certain percentage which is not an option in this model.
This solution can be used if we can model the capacity and the costs of each depot according to the
demand of products.
All the depots are open, but only for a certain percentage. This means that there are now transportation
costs because every city is supplied by the depot situated in it.






1.1(B)- No depot capacity restrictions- second model
The problem can be formulated as an ILP with the variables X
d
and Y
sd
. The formulation for the
program is the following:
The qualitative model:
Objective: minimize the annual costs;

Restrictions:
- The number of products delivered;
- Transport costs;
- The annual costs;
- The annual demand of products;
- Each city can be supplied by only one depot
The LP formulation of the problem:
Parameters: - TransportCost(s, d); transport costs from depot d to city s;
-Demand(s); city s demand of products;
-DepotCost(d); depot d annual costs;
-s {1, 2, 3, 4, 5}
-d {1, 2, 3, 4, 5}
VARIABLES: -X(d):{0,1}; indicates if a depot is open or closed
-Y(s,d): {0,1}; indicates whether city s is delivered to by depot d
CONSTRAINTS: -ad(s): SUM[d,Y(s,d)]>=demand(s); annual demand
-b(d): sum[(s),Y(s,d)]=x(d);
OBJECTIVE FUNCTION:

() () ( ) ( ) ()



LP relaxation:
X(d) [0, 1];




Optimal solution
LP_SOLVE MESSAGES:
Feasible solution 295 after 11 iter, 0 nodes (gap 0.0%)
========================================
Value of objective function: 295.00000000
========================================
VARIABLE VALUES:
X_Almelo 1
X_Delden 1
X_Enschede 1
X_Hengelo 0
X_Oldenzaal 0
Y_Almelo_Almelo 1
Y_Almelo_Delden 0
Y_Almelo_Enschede 0
Y_Almelo_Hengelo 0
Y_Almelo_Oldenzaal 0
Y_Delden_Almelo 0
Y_Delden_Delden 1
Y_Delden_Enschede 0
Y_Delden_Hengelo 0
Y_Delden_Oldenzaal 0
Y_Enschede_Almelo 0
Y_Enschede_Delden 0
Y_Enschede_Enschede 1
Y_Enschede_Hengelo 0
Y_Enschede_Oldenzaal 0
Y_Hengelo_Almelo 0
Y_Hengelo_Delden 0
Y_Hengelo_Enschede 1
Y_Hengelo_Hengelo 0
Y_Hengelo_Oldenzaal 0
Y_Oldenzaal_Almelo 0
Y_Oldenzaal_Delden 0
Y_Oldenzaal_Enschede 1
Y_Oldenzaal_Hengelo 0
Y_Oldenzaal_Oldenzaal 0
The optimal solution shows that the open depots are Almelo, Delden and Enschede.
The cities that have an open depot receive from that depot. The cities that dont have open
depots(Hengelo and Oldenzaal) receive from the depot in Enschede
The solution is different from the solution in 1.1a but the results for the costs are the same







LP-relaxation optimal solution
LP_SOLVE MESSAGES:
========================================
Value of objective function: 295.00000000
========================================
VARIABLE VALUES:
X_Almelo 1
X_Delden 1
X_Enschede 1
X_Hengelo 0
X_Oldenzaal 0
Y_Almelo_Almelo 1
Y_Almelo_Delden 0
Y_Almelo_Enschede 0
Y_Almelo_Hengelo 0
Y_Almelo_Oldenzaal 0
Y_Delden_Almelo 0
Y_Delden_Delden 1
Y_Delden_Enschede 0
Y_Delden_Hengelo 0
Y_Delden_Oldenzaal 0
Y_Enschede_Almelo 0
Y_Enschede_Delden 0
Y_Enschede_Enschede 1
Y_Enschede_Hengelo 0
Y_Enschede_Oldenzaal 0
Y_Hengelo_Almelo 0
Y_Hengelo_Delden 0
Y_Hengelo_Enschede 1
Y_Hengelo_Hengelo 0
Y_Hengelo_Oldenzaal 0
Y_Oldenzaal_Almelo 0
Y_Oldenzaal_Delden 0
Y_Oldenzaal_Enschede 1
Y_Oldenzaal_Hengelo 0
Y_Oldenzaal_Oldenzaal 0

The LP-relaxation solution is a feasible solution because it is the same as the solution for ILP formulation

From these results we can see that the dual value for Hengelo is the biggest with a value of 90. For
this reason if the demand of Hengelo will be ignored the objective will decline by 3 * 25 = 75, which
will lead to a new objective of 295 75 = 220.
c. As we can see the integrality gap for 1.1a is 85 and for 1.1b is 0. This means that 1.1b is the stronger
formulated model of the two situations






1.2 Depot capacity restrictions
Additional data:
City number City Maximum depot size
1 Enschede 80
2 Hengelo 25
3 Almelo 25
4 Oldenzaal 10
5 Delden 15
a) The problem can be formulated as an ILP with the variables X
d
and Y
sd
. The formulation for the
program is the following:
The qualitative model:
Objective: minimize the annual costs;

Restrictions:
- The number of products delivered;
- Transport costs;
- The annual costs;
- The annual demand of products;
- Each city can be supplied by only one depot;
- Depot capacity;
PARAMETERS: - TransportCost(s, d); transport costs from depot d to city s;
-Demand(s); city s demand of products;
-D
c
(d); depot d annual costs;
-D
cap
(d); depot d capacity;
-s {1, 2, 3, 4, 5}
-d {1, 2, 3, 4, 5}
VARIABLES: -X(d):{0,1}; this indicates if a depot d is open or closed;
-Y(s,d):{0,80}; this indicates the maximum capacity of a depot;
CONSTRAINTS
ad(s):SUM[(d),Y(s,d)]>=Demand(s); the demand cannot be greater than the capacity of the depot;
cr(d):SUM[(s),Y(s,d)]<=DepotCap(d); the number of products delivered from depot s cannot be greater
than the capacity of the depot;
a(s,d):Y(s,d)<= X(d)*bigM;
OBJECTIVE FUNCTION:

() () ( ) ( )



LP relaxation:
X(d) [0,1];
Y(s,d) [0,80];
LP_SOLVE MESSAGES:
Feasible solution 319 after 19 iter, 5 nodes (gap 48.1%)
Improved solution 305 after 28 iter, 14 nodes (gap 41.7%)
========================================
Value of objective function: 305.00000000
========================================
VARIABLE VALUES:
X_Almelo 1
X_Delden 1
X_Enschede 1
X_Hengelo 0
X_Oldenzaal 1
Y_Almelo_Almelo 12
Y_Almelo_Delden 0
Y_Almelo_Enschede 0
Y_Almelo_Hengelo 0
Y_Almelo_Oldenzaal 0
Y_Delden_Almelo 0
Y_Delden_Delden 15
Y_Delden_Enschede 1
Y_Delden_Hengelo 0
Y_Delden_Oldenzaal 0
Y_Enschede_Almelo 0
Y_Enschede_Delden 0
Y_Enschede_Enschede 20
Y_Enschede_Hengelo 0
Y_Enschede_Oldenzaal 0
Y_Hengelo_Almelo 0
Y_Hengelo_Delden 0
Y_Hengelo_Enschede 25
Y_Hengelo_Hengelo 0
Y_Hengelo_Oldenzaal 0
Y_Oldenzaal_Almelo 0
Y_Oldenzaal_Delden 0
Y_Oldenzaal_Enschede 0
Y_Oldenzaal_Hengelo 0
Y_Oldenzaal_Oldenzaal 10

The open depots are Almelo, Delden, Enschede and Oldenzaal.
Each city receives products from the depot that is situated in the city, except Hengelo which receives
from the depot in Enschede and there is also one product sent from Enschede to Delden.
Compared to the 1.1 model the optimal solution for 1.2 has a greater value, although the same number
of depots are open. This is because some cities have products delivered from two depots, because they
cannot supply the demanded number of product.
========================================
LP_SOLVE MESSAGES:
========================================
Value of objective function: 214.20000000
========================================
VARIABLE VALUES:
X_Almelo 0.48
X_Delden 0.6
X_Enschede 0.8
X_Hengelo 0.2
X_Oldenzaal 0.4
Y_Almelo_Almelo 12
Y_Almelo_Delden 0
Y_Almelo_Enschede 0
Y_Almelo_Hengelo 0
Y_Almelo_Oldenzaal 0
Y_Delden_Almelo 0
Y_Delden_Delden 15
Y_Delden_Enschede 0
Y_Delden_Hengelo 1
Y_Delden_Oldenzaal 0
Y_Enschede_Almelo 0
Y_Enschede_Delden 0
Y_Enschede_Enschede 20
Y_Enschede_Hengelo 0
Y_Enschede_Oldenzaal 0
Y_Hengelo_Almelo 0
Y_Hengelo_Delden 0
Y_Hengelo_Enschede 20
Y_Hengelo_Hengelo 5
Y_Hengelo_Oldenzaal 0
Y_Oldenzaal_Almelo 0
Y_Oldenzaal_Delden 0
Y_Oldenzaal_Enschede 0
Y_Oldenzaal_Hengelo 0
Y_Oldenzaal_Oldenzaal 10


The LP-relaxation solution shows that all the depots are open but they are functioning only at a certain
capacity. This is not possible with the current formulation of the problem. The integrality gap is 90.8, a
value which is pretty high.
1.3 Theory
a. Solve the problem with a simple heuristic method
Data: Transportation costs between cities
1 2 3 4 5
1 3 9 6 10
2 3 11 8 6
3 9 11 18 14
4 6 8 18 16
5 10 6 14 16
The method that has been applied for solving this problem is very simple. Consider only the depot that
has the lowest annual costs open and make the calculations to determine the total costs. After this open
one depot at a time in the order of the annual costs of the depot until all the depots are open. After
calculating the annual costs for all the options determine which the optimal solution is.
Each city that doesnt have an open depot in it will be supplied from the city that can supply with the
lowest transport costs.
The annual costs are the sum of the transport costs (Transport cost*annual demand) and the depot
costs.
First case: Only the depot in Almelo open
City Depot open Depot costs Supplied from Annual demand Transport costs Annual costs
Enschede no
0
Almelo
20 9
180
Hengelo
no
0 Almelo 25 11
275
Almelo
yes
40 Almelo 12 0
40
Oldenzaal
no
0 Almelo 10 18
180
Delden
no
0 Almelo 16 14
224

Total costs 899

Second case: The depots are open in Almelo and Delden
City Depot open Depot costs Supplied from Annual demand Transport costs Annual costs
Enschede no
0
Almelo
20
9 180
Hengelo
no
0 Delden 25
6 150
Almelo
yes
40 Almelo 12
0 40
Oldenzaal
no
0 Delden 10
16 160
Delden
yes
45 Delden 16
0 45

Total costs 575

Third case: The depots are open in Almelo, Delden and Oldenzaal
City Depot open Depot costs Supplied from Annual demand Transport costs Annual costs
Enschede no
0
Oldenzaal
20
6 120
Hengelo
no
0 Delden 25
3 75
Almelo
yes
40 Almelo 12
0 40
Oldenzaal
yes
60 Oldenzaal 10
6 60
Delden
yes
45 Delden 16
0 45

Total costs 340






Fourth case: The depots are open in Almelo, Delden, Oldenzaal and Enschede
City Depot open Depot costs Supplied from Annual demand Transport costs Annual costs
Enschede yes
75
Enschede
20
0 75
Hengelo
no
0 Enschede 25
3 75
Almelo
yes
40 Almelo 12
0 40
Oldenzaal
yes
60 Oldenzaal 10
0 60
Delden
yes
45 Delden 16
0 45

Total costs 295

Fifth case: All the depots are open
City Depot open Depot costs Supplied from Annual demand Transport costs Annual costs
Enschede yes
75
Enschede
20
0 75
Hengelo
yes
90 Hengelo 25
0 90
Almelo
yes
40 Almelo 12
0 40
Oldenzaal
yes
60 Oldenzaal 10
0 60
Delden
yes
45 Delden 16
0 45

Total costs 310

Looking at the total costs for all the five cases we can clearly see that the best solution is fourth one,
having th lowest costs of all the solutions.

b. The solution presented in 1.3a is the same solution as in 1.1a, meaning that this is the optimal
solution for this problem, given this particular situation
c. From the definition bigM is a large number associated with the artificial variables, represented by
the letter M. This is used for solving problems that contain constraints such as greater than or less
than.
In our case the constraint is less than or equal to and it refers to the annual demand that it has to
be less than the maximum value of the annual demand. This means that bigM must have a value
that is at least equal or greater than the biggest annual demand of products for one city.
By looking at the first table we can see that the biggest annual demand is from Hengelo, and it is 25.
This means that bigM must be at least 25.

Vous aimerez peut-être aussi