International University Deterministic Models in OR

INTERNATIONAL UNIVERSITY - VIETNAM NATIONAL UNIVERSITY

School of Industrial and Engineering Management

GROUP PROJECT

REPORT

COURSE: Deterministic Models in OPERATIONS RESEARCH

Tittle: A Location-Inventory Supply Chain Problem: Reformulation and

Piecewise Linearization - Industry: Supply chain

Lecturer: Dr. Ha Thi Xuan Chi

GROUP 8

No. Student ID Full name

1 IELSIU19145 Lê Thị Như Hà
2 IELSIU19215 Phạm Thị Yến Ngọc
3 IELSIU19051 Trần Thị Bích Ngọc
4 IELSIU19252 Nguyễn Thị Như Quỳnh
5 IELSIU20456 Huỳnh Thị Thu Uyên
6 IELSIU20328 Nguyễn Hữu Khoa

1
International University Deterministic Models in OR

Abstract

2
International University Deterministic Models in OR

Table of content

I. Introduction

1. Background of study (Show why it is important to study this

problem and background of the case study)
2. Problem statement (Show the problem question that needs to be
answered in the report)

II. Mathematical Model/ Algorithms

1. Parameters
2. Decision variables
3. Objective function
4. Constraints

III. Result analysis and Discussion

Show your result in interpretable form with analysis (sensitivity

analysis and solution discussion, future research direction)
IV. References

Show the sources (research paper, website,..) that you refer to

define problems, analyze data, mathematical models, and solve
the problem.
V. Appendix

CPLEX code
Input data (Excel/Table) used to run models
Output data obtained from CPLEX

3
International University Deterministic Models in OR

Table ..
Table ..
Table ..
Table ..
Table .

4
International University Deterministic Models in OR

Chapter I: Introduction
1. Background of study
In this project , we study a two-echelon inventory management problem with multiple
warehouses and retailers. In the background section, our team will give a few reasons why
this research is important and outline the roots of the problem.
In today's market, companies often pay attention to optimizing supply chains to ensure
efficient operation for their company or business. An efficient supply chain means they have
to reduce costs, and improve service levels. So supply chain management will involve
making decisions in a variety of situations: strategic decisions, tactical decisions, and
operational decisions. As follows: strategic-level decision involves long-term planning
activities for the company (such as locating a warehouse); tactical-level includes activities
with short-term influence such as transport and storage policy; operational decisions are for
weekly and daily plans. From the above level decisions, the decision of facility location and
inventory policy is very important, and they are dependent on each other. That is, the change
of one problem will affect the cost of the other problem and vice versa. In practice, however,
they are often dealt with sporadically, which leads to suboptimal results in supply chain
management. 
Therefore, the combination of two decision levels leading to joint location-inventory needs to
be considered to improve supply chain performance. Because such an integration will help
businesses get the most out of the important data from the interaction of the two issues. The
basic model addresses the delivery of a single product from a manufacturer to warehouses,
also named distribution centers (DCs), and from there to multiple retailers. The DCs can be
opened in multiple locations, which is something that is decided through the model. DCs and
retailers hold working inventory, based on product that has not yet been requested by retailers
or end-customers, respectively. The purpose of the model is to determine the optimal number
of open DCs, the optimal allocation of retailers to DCs, and the optimal inventory strategies
for both DCs and retailers. The objective is to minimize the fixed ordering, holding,
transportation and facility-related costs.
Many previous research papers have offered a solution to this model, but it has not really
worked due to the complexity of combining both the location and inventory problem.
Outstanding studies for this issue include the works of Chan, Chung, and Wadhwa (2005),
Jayaraman and Ross (2003), Diabat, Aouam, and Ozsen (2009), Shiguemoto and Armentano
(2010), and Diabat (2014). Furthermore, Lagrangian relaxation has been implemented by
Chen and Chu (2003), Diabat and Richard (2015), and Eskigunet al. (2005) and so on. The
limitation of the above papers is the need to use heuristic techniques due to the complexity of
the model, also, this technique cannot be used to solve extended problems. However, in the
current study, we need to solve this model using piecewise linearization to transform the
resulting MINLP to a mixed integer program and we solve it using CPLEX.
To sum up, one reason why our study is more important than ever is that in potential
extensions of the model, techniques such as Lagrangian relaxation almost do not give optimal
results for the model, whereas with the current approach in this paper, it is easy and
straightforward to incorporate the extension and again solve the problem with the help of
commercial software.

2. Problem statement
Let the set of retailers be I and the set of warehouses that can be located at certain
predetermined sites be J. The purpose of the problem is to distribute a single product from a

5
International University Deterministic Models in OR

manufacturing plant to warehouse J and from warehouse J to retailers I. When the product
reaches the retailers but has not yet been ordered by the customer, then the retailer will
determine the demand and keep the product as working inventory. For warehouses, working
inventory is products ordered from the manufacturer but not yet delivered to the retailers. We
also note that the supplier of the warehouse is only manufacturer, and shortages are not
permitted. In this model, the production unit will help determine the transportation costs
between the production  and the retailer, thereby making a better decision about which
warehouse needs to be opened. The model is illustrated more clearly through the diagram
below:

Figure 1: Supply chain network

In each echelon, we have to determine the cycle time by applying the power-of-two inventory
policy. Applying common production cycle policies often results in the same production
cycle times of end items. But with a power- of- two policy assume that the ratio of cycle is
power of two, that helps us to localize better results.
We can determine the associated costs as follows:
- Fixed order cost, the cost of placing an order
- Unit-inventory cost, the cost of holding one unit of product for one unit of time
- The cost of shipping one unit of product between facilities
- Fixed location cost, the cost of establishing and operating a warehouse.
- The objective of the model is to decide:
- The number of warehouses to open
- The location of these warehouses
- The set of retailers assigned to each warehouse
- The size and timing of orders for each facility.
- Repeatedly, the goal of the model is to:
- Minimizing the sum of inventory
- Minimizing shipping, ordering, and location costs and with the condition of satisfying
end-customer demand.

6
International University Deterministic Models in OR

Chapter II: Mathematical model

1. Single warehouse multi-retailer model
The distribution system includes a single warehouse, which responsible for providing
retailers with a single product.
Index sets:
I Set of retailers, indexed by i

Input parameters:
di Demand rate of retailer i
si Unit-shipping cost from the warehouse to retailer i
hi Unit-inventory cost per unit of time at retailer i
ki Fixed-order cost at retailer i
h^0
Unit-inventory cost per unit of time at the warehouse
❑trn Weight factor associated with transportation costs
❑inv Weight factor associated with inventory costs

Decision variables:
Ti Cycle-time of retailer i
T0 Cycle-time of warehouse

The echelon holding cost rates:

h^' 0=h^ 0 ∧h ' i=hi−h^ 0

The average cost of holding inventory for retailer i at the warehouse and at retailer i:
1 1
H i ( T 0 , T i ) = ( h i−h^ 0 ) . d i .T i + h^ 0 . di . max ⁡( T 0 , T i )
2 2

There are two cases of average holding cost:

 T i ≥ T 0: The warehouse places an order every time the warehouse does. Thus, the
warehouse does not hold inventory for the retailer
1 1
hi . d i . T i = ( h ' i −h^ 0 ) . d i . T i
2 2

 T i< T 0 : There is a specific portion of the warehouse inventory that is intended for
delivery to retailer i.
H i ( T 0 , T i ) =2h ' i . d i . T i + 2h^' 0 . d i .T i
1 1

Therefore, the average cost of a power-of-two policy is:

k
∑ T i +∑ H i ( T 0 , T i )
i i i
Thus, the formulation of the single warehouse multi-retailer problem is as follows:

7
International University Deterministic Models in OR

ki 1 1
min ∑ +∑ ( h + h^ ) . d . T + h^ d . max ( T 0 , T i )
i Ti i 2 i 0 i i 2 0 i

2. Multiple warehouse mutil-retailer problem

There are two decisions that need to be clearly considered in this regard: The first
one is the location of the warehouse and the second is choosing the warehouse to supply to
retailers.
Index sets:
I Set of retailers, indexed by i
J Set of possible warehouse locations, indexed by j

Input parameters:
di Demand rate of retailer i
fi Fixed cost of establishing and operating a warehouse at
location j
sij Unit-shipping cost from warehouse j to retailer i
^s j Unit-shipping cost from manufacturer to warehouse j
hi Unit-inventory cost per unit of time at retailer i
ki Fixed-order cost at retailer i
h^ j Unit-inventory cost per unit of time at warehouse j
k^ j
Fixed-order cost at warehouse j
❑trn Weight factor associated with transportation costs
❑inv Weight factor associated with inventory costs

Decision variables:
T ij Cycle-time of retailer i when served by warehouse j
T^ j Cycle-time of warehouse j

Binary decision variables:

Xj 1, if a warehouse is opened at location j
0, otherwise
Y ij 1, if a retailer i is served by warehouse at location j
0, otherwise

The average cost of the power-of-two inventory policy for the multiple warehouse multi-
retailer problem are included the fixed-order cost at warehouse j, the binary variables and the
inventory weight factor:
k^ k
∑ ❑inv T^ j X j + ∑ ∑ ❑inv T i Y ij +∑ ∑ ❑inv 12 [( ¿¿hi− h^ j )T ij + h^ j max ⁡(T^ j ,T ij )]d i Y ij ¿ ¿
j j i j ij i j

The fixed cost of establishing a facility:

∑ f jX j
j

Finally, there is the shipping cost between the assigned warehouses and retailers:

8
International University Deterministic Models in OR

∑ ∑ ❑trn (¿ ¿ sij + ^s j)d i Y ij ¿ ¿

i j
The formulation of the joint inventory-location problem:
k^ j ki
min X ,Y , T^ ,T ∑ (¿ f j+❑inv ) X j+ ∑ ∑ (¿inv ¿ ¿ +❑trn bij +❑inv c ij T ij +❑inv e ij max ⁡{T^ j ,T ij })Y ij ¿¿ ¿ ¿
j∈ J T^ j j ∈J i ∈ I T ij

We defined that:
1 1
b ij= ( s ij + s^ j ) d i ; cij = ( hi + h^ j) d i ; e ij = h^ j d i
2 2

The goal function is to minimize total inventory, shipping, ordering, and location costs, while
meeting end customer needs.

Constraints
Formulation Condition No.
ki 1 1 1
min ∑ +∑ ∙ d i ∙ T i+ h^ 0 ∙ d i ∙ max ( T 0 ,T i )
i T i i 2 ( hi−h^ 0 ) 2
T 0 /T i=2N i
∀ i∈ I 2
Ni ∈ Z ∀ i∈ I 3
T 0 >0 4
T i> 0 ∀ i∈ I 5

( )
k^ j
( )
ki 6
min X ∙Y ∙ T^ . T ∑ f j+ β inv X j + ∑ ∑ β inv + β trn bij + βinv c ij T ij + βinv e ij max {T^ j ,T ij } Y ij
j ∈J T^ j j∈ J i ∈I T ij
∑ Y ij =1 ∀ i∈ I 7
j∈ J
Y ij ≤ X j ∀ i∈ I , ∀ j∈ J 8
Y ij ∈ { 0,1 } ∀ j∈J 9
X j ∈ { 0,1 } ∀ j∈J 10
^ / T =2N
T ij
∀ i∈ I , ∀ j∈ J 11
I ij
N ij ∈ Z ∀ i∈ I , ∀ j∈ J 12
T^ j >0 ∀ j∈J 13
T ij >0 ∀ i∈ I , ∀ j∈ J 14

( ) ( )
1 2 15
⏞
4
⏞
3
k^ j ki
min ∑ f j X j + β inv X j +∑ ∑ β Y +β β inv c ij T ij Y ij +⏞
bij Y ij +⏞ β inv e ij max {T^ J , T ij } Y ij
T^ j
inv ij trn
j∈ J j∈J j ∈J T ij
k^ J k^ J n k^ J k^ J k^ J k^ J n
( )
16
≅ n p j + n +1 ( 1− p j ) = n +1 + n − n+1 p j
n

T^ J t^ J t^ J t^ J t^ J t^ J
k^j k^j k^j
(
k^j
)
n
17
X j ≃ n +1 X j+ n − n+1 P j X j
^
Tj ^
Tj ^
Tj Tj ^

9
International University Deterministic Models in OR

k^j k^j
( k^j k^j
)
n
18
β inv X j ≃ β inv n+1 + β inv − Aj
T^j T^ j
^ n
T T
j
^n+ 1
j

Anj ≤ P nj 19
n n
A j ≥ P j + M ( X j−1 ) 20
n
Aj≤ M X j 21
∑ P nj =1 22
n
n
P j ≥0 23

( )
ki ki ki ki n 24
β inv Y ij ≃ β inv n+1 + βinv − n+1 B ij
T ij T ij T ij T ij
n n
Bij ≤ Q ij 25
Bij ≥ Qnij + M (Y ij −1)
n
26
n
Bij ≤ M Y ij 27
∑ Qijn=1 28
n

Qnij ≥ 0 29
c ij C ij 30
C ij ≤T ij 31
C ij ≥T ij + M (Y ij −1) 32
C ij ≤ M Y ij 33
Y ij ≥ 0 34
Y ij ={0; 1 } 35
w ij Y ij 36
w ij ≥ T^j ∀ i, j 37
w ij ≥ T ij ∀ i, j 38
e ij Eij 39
Eij ≤W ij 40
Eij ≥W ij + M ( Y ij −1 ) 41
Eij ≤ M Y ij 42
W ij ≥ T^j 43
W ij ≥T j 44
^
T j ,T ij ≥ 0 45
Y ij ∈ { 0,1 } 46

( ( ) )
47
( )
k^j k^j k^j ki ki ki
min X ,Y , T^ ,T ∑ f j X j + β inv + β − A n
+ ∑ ∑ ( β + βinv n
− n+1 Bij + βtrn bb ij Y ij + βinv c ij C ij + β
t^ t^nj t^
n+1 inv n+1 j inv n+1 n
j∈ J
j j
j ∈J j ∈J t ij t ij t ij

∑ Y ij =1 ∀ i∈ I 48
j∈ J
Y ij ≤ X j ∀ i∈ I , ∀ j∈ J 49
Anj ≤ P nj 50
Anj ≥ P nj + M ( X j−1 ) 51
Anj ≤ M X j 52

10
International University Deterministic Models in OR

∑ P nj =1 53
n
n n
Bij ≤ Q ij 54
n n
Bij ≥ Qij + M ( Y ij −1 ) 55
n
Bij ≤ M Y ij 56
∑ Qijn=1 57
n
n
Qij ≥ 0 58
Cij ≤T ij 59
Cij ≥T ij + M (Y ij −1) 60
Cij ≤ M Y ij 61
Eij ≤W ij 62
Eij ≥W ij + M (Y ¿¿ ij−1)¿ 63
Eij ≤ M Y ij 64
W ij ≥ T^j 65
W ij ≥T ij 66
T^j ,T ij ≥ 0 67
N ij ∈ Z ∀ i∈ I , ∀ j∈ J 68
Y ij = {0,1 } ∀ i∈ I , ∀ j∈ J 69
X j ∈ { 0,1 } ∀ j∈J 70

11
International University Deterministic Models in OR

12
International University Deterministic Models in OR

13