VIETNAM NATIONAL UNIVERSITY – HOCHIMINH CITY

INTERNATIONAL UNIVERSITY
DEPARTMENT OF INDUSTRIAL ENGINEERING & MANAGEMENT

DETERMINISTIC MODELS IN OR

NEW MIX INTEGER APPROACH TO SOLVE A MULTI-LEVEL

CAPACIATED DISASSEMBLY LOT-SIZING PROBLEM WITH
DEFECTIVE ITEMS AND BACKLOGGING

Lecturer: Assoc.Prof. Ha Thi Xuan Chi

Group Number: 04
Class: IEIE22IU21 – Thursday (Period 1-3)

Student ID Member name % Contribution

IEIEIU21089 Nguyễn Hoàng Thái Công 100%

IEIEIU20125 Trần Ngô Anh Huy 60%

IELSIU21124 Huỳnh Nhật Nam 80%

IEIEIU21170 Trần Nhật Quang 100%

IELSIU20187 Trần CaoTrí 60%

6/2024

i
 ABSTRACT

Disassembly plays a vital role in the supply chain's functioning. Remanufacturing End-of-Life (EoL)
products is one of the finest options for producing large quantities of goods. A novel mix integer
programming model was presented in the study report for the deterministic capacitated multi-level
defective item and backlog lot-sizing problem. The setup cost, along with other typical and essential
characteristics in the industrial reality, such as inventory holding, external procurement, disposal,
backlog, and overload charges, was taken into consideration by the writers each time. But as the number
of variables expanded, so did the difficulty of the research. The authors' limits served as the foundation
for our research into maximizing profit while minimizing the aforementioned. Furthermore, determining
the ideal time and amount on the planning horizon through the disassembly of parent items was another
goal of our study. We tackle this problem using Mixed Integer Programming (MIP), a useful technique
for mathematical solution optimization. CPLEX is also used to facilitate the study's computational
results, which demonstrate the proven efficiency of the proposed MIP model in handling lot sizing
problems with many choice variables.

Keywords:

 Disassembly lot-sizing
 Mixed integer programming
 Defective items
 Backlogging
 EoL

ii
 TABLE OF CONTENTS

Chapter
Page

ABSTRACT..................................................................................................................ii
TABLE OF CONTENTS.................................................................................................iii
LIST OF TABLES.........................................................................................................iv
LISTS OF FIGURES.......................................................................................................v
CHAPTER 1 INTRODUCTION......................................................................................1
1.1 Background of Study...........................................................................................1
1.2 Problem Statement..............................................................................................2
1.3 Objective..........................................................................................................3
1.3.1 Scope:........................................................................................................3
1.3.2 Limitations:.................................................................................................4
CHAPTER 2 MATHEMATICAL MODEL FORMULATION................................................5
2.1 Index for model:.................................................................................................5
2.2 Parameters for model:..........................................................................................5
2.3 Decision variables:..............................................................................................6
2.4 Variables:.........................................................................................................6
2.5 Objective function:..............................................................................................6
2.6 Constraints:.......................................................................................................7
CHAPTER 3 RESULTS................................................................................................9
3.1 Data Collection and Processing...............................................................................9
3.1.1 Research structure.........................................................................................9
3.1.2 Some solution approaches..............................................................................10
3.1.3 Data.........................................................................................................11
3.2 Results Discussion............................................................................................14
3.2.1 Result presentation......................................................................................14
3.2.2 Result analysis.......................................................................................17
3.2.3 Sensitivity analysis......................................................................................18
CHAPTER 4 REFERENCES........................................................................................19
CONCLUSION...........................................................................................................20
CHAPTER 5 APPENDIX............................................................................................21
5.1 CPLEX CODE................................................................................................21
5.2 Output of changing disassembly capacity of the factory.............................24

3
 LIST OF TABLES

Table Page
Table 2-1: Index............................................................................................................5
Table 2-2: Parameters...............................................................................................6
Table 2-3: Decision variables............................................................................................6
Table 2-4: Variables.......................................................................................................6
Table 3-1: The generation of data.....................................................................................12
Table 3-2: Input data (1)................................................................................................12
Table 3-3: Input data (2)................................................................................................13
Table 3-4: Number of item j disassembled by item i..............................................................13
Table 3-5: Set of parents of item i....................................................................................14
Table 3-6: External demand of item I in period t..................................................................14
Table 3-7: Binary indicators...........................................................................................14
Table 3-8: Backlogging level at the end of each period..........................................................15
Table 3-9: Inventory level at the end of period t...................................................................15
Table 3-10: Sale quantity of item i in time-period t, ∀ i∈ E ¿1 }¿..............................................16

Table 3-12: Quantity of parent item i to disassemble in time-period t, ∀ 𝑖 ∈𝒜𝐶..........................16

Table 3-11: Defective quantity of item i in time-period t, ∀ i∈ E ¿1 }¿.......................................16

Table 3-13: Disassembly ‘Overtime’ in time-period t, ∀ 𝑡∈𝒯................................................17

Table 3-14: Quantity of item i to procure in time-period t, ∀ t ∈T ............................................17
Table 3-15: Result analysis............................................................................................17
Table 3-16: Changing disassembly capacity of the factory......................................................18
Table 3-17: Changing Setup Cost....................................................................................18

4
 LISTS OF FIGURES

Figure Page

Figure 3-1: Illustrative example of disassembly structure........................................9

Figure 5-1: Changing disassembly capacity of the factory Ut=140............................................24
Figure 5-2: Changing disassembly capacity of the factory Ut=200............................................25
Figure 5-3: Changing disassembly capacity of the factory Ut=240............................................26
Figure 5-4: Changing disassembly capacity of the factory Ut=480............................................27
Figure 5-5: Maximum profit when changing setup cost of item 5..............................................28

5
 CHAPTER 1 INTRODUCTION

1.1 Background of Study

The issue of concern in our study is the problem of massive over-production of goods. Therefore, the

disassembly system which is a methodical extraction of valuable components plays a vital role in the

supply chain. It is a divergent process where a recovered product is broken down into several parts and

or subassemblies, which can be termed as reverse logistics.

This process will collect resources from the EoL (End of life) product back to the original materials. The

goal is to optimize the supply chain, minimize costs and reduce environmental impacts and energy

consumption. The optimization is obtained by comparing the difference between the revenue achieved

by resale of the items recovered after disassembly and the costs tied to operating the disassembly

problem.

Lot sizing is among the most important and most difficult problems in production planning. Lot sizing

decisions give rise to the problem of identifying when and how much of a product to produce in order to

minimize the setup, production and holding costs. Disassembly lot-sizing (DLS) plays an important role

in the disassembly systems and is defined as the problem of determining the disassembly quantity and

timing of the End-of-life (EoL) products, with the purpose of fulfilling the demand of their parts over a

finite planning horizon.

Multi-level capacited disassembly lot-sizing problem is strongly NP-hard when there is a parent–

component relationship among the items. Raw materials after processing by several operations change to

end products. The output of a former operation (level) is the sequential input for a ladder operation.
1
Therefore, the demand of one level depends on the demand for its parents’ level.

1.2 Problem Statement

The disassembly process in re-manufacturing has attracted increasing headlight recently, due to its great

potential in energy saving and environment protection. The study focused on the disassembly lot sizing

(DLS) problem where researchers found the ideal quantity over the planning horizon while reducing the

disassembly system's expenses.

There are key summaries relating to DLS problem that researcher constructed to ensure the logic of the

study:

 Dynamic demands for the non-root items are deterministic.

 Demand can be satisfied by both disassembling the EoL products and purchasing of new non-root

items in the marketplace.

 All demands must be satisfied at the end of the planning horizon.

 EoL goods are consistently offered in the necessary quantity.

 Backlogging is allowed.

 Disassembled items may be defective.

 Disassembly operation time and setup time for all parent items are deterministic.

 Capacity and extra capacity are limited.

 Cost parameters are assumed to be constant.

 When non-root commodities are kept in stock to meet future demand, inventory holding costs occur.

 When backorders are delayed, the objective function may be adversely affected.

2
The authors' goal is to maximize the total profit, but meanwhile minimize the sum of holding,

backlogging, external procurement, setup, and overload costs. As a result, the best disassembly timing

and quantities to dismantle a parent item can be calculated, thus, researchers can determine the numbers

of every single non-root item sold, backordered, stored, and bought outside. Moreover, the amount of

additional capacity that can be added over a specific planning horizon for various time periods can be

found. The authors added setup time when preparing the disassembly operation since it is important in

manual disassembly operations.

In our case, compact and effective mixed-integer programming (MIP) is proposed to calculate the

optimal disassembly dates and quantities to meet all customer demand in the most efficient way. In our

report, CPLEX is going to be used to solve the MIP problem.

1.3 Objective

1.3.1 Scope:

First, our focus will be on checking whether the author’s formulation is correct or not via a small-scale

problem extracted from the big picture. If it is accurate, the large-scale, much more time-consuming

problem is also right.

The scope of the study is limited to the multi-level capacitated disassembly lot-sizing problem with

defective items and backlogging, which is a niche and specific problem. Therefore, we aim to solve it by

proposing a new mixed integer approach. Moreover, our study also seeks the involvement of developing

a mathematical model and formulating it into a mixed integer program (MIP). Then, with our

mathematical model and MIP completed, we can use computational experiments and numerical

examples to implement and test the previously proposed approach.

3
The study will explore novel algorithms, heuristics, or optimization methods to improve the efficiency

and effectiveness of finding solutions. Assuming everything proceeds according to plan, our method will

be able to take into account both the risk of backlog in the disassembly lot-sizing problem and the

existence of defective products. Consequently, it can be used to explore the mathematical model and

solution algorithm's handling of backlogged demand and defective items.

Finally, we hope to develop a model that accurately depicts the characteristics of the problem.

1.3.2 Limitations:

Finding the best solution is complex since it involves several stages of disassembly, capacitated

constraints, defective goods, and backlog. It is therefore expected that large-scale examples of the

problem will not be solved in a reasonable amount of processing time.

Moreover, many aspects of the research, such as fixed disassembly yields, independent needs, or

specific cost structures, are assumed or simplified, to aids develop the mathematical model by lessening

its complexity. These assumptions may restrict how the approach is applied in real-world scenarios.

Furthermore, software constraints, combined with our device's low memory and processing capacity,

could compromise the models' scalability. Owing to these constraints, the research may not be able to

resolve the issue in large-scale scenarios. This implies that it could be challenging to thoroughly and

accurately gather some types of data, such as disassembly yields, resource capacities, or demand trends,

among others.

In summary, even though the research attempts to suggest a novel mixed integer strategy, it could not be

as complete as current approaches. The study considers our attempt as a stand-alone entity and

4
concentrates more on proving the viability of the suggested strategy than on comparing it to other

methods.

CHAPTER 2 MATHEMATICAL MODEL FORMULATION

2.1 Index for model:

Index
t Index for time-period t, t ∈ 𝓣
i Index for item 𝑖, 𝑖 ∈ ℰ

Table 2-1: Index

2.2 Parameters for model:

Parameters
𝒯 Set of time periods in the current planning horizon
ℰ Set of items 𝑖, 𝑖 ∈ ℰ
𝒜 Set of leaf items 𝑖

𝒜 Set of sub-assembly’s items 𝑖

C

Ri,j Quantity of item j obtained from disassembling one unit of item 𝑖

Di,t External demand for item 𝑖, in time-period t
Ii,0 Starting inventory of item 𝑖
Li Lead time when disassembling parent item 𝑖
hi Inventory holding cost of one unit of item 𝑖
Si Setup cost of parent item 𝑖
𝛟i Set of parents of item 𝑖
bi Backlogging cost of one unit of item 𝑖
Ci Purchasing cost of one unit of item 𝑖
5
 𝘢i Defective rate of item 𝑖
Gi Operation time to disassemble parent item 𝑖
Ut Available capacity in time-period t
Ot Cost of increasing capacity by one unit in time-period t
Ft Disassembly capacity on overtime in time-period t
SCi Sale price of item 𝑖
STi Setup time of parent item 𝑖
M One enough large number

Table 2-2: Parameters

2.3 Decision variables:

Decision variables
Xi,t Quantity of parent item 𝑖 disassembled in time-period t, ∀ 𝑖 ∈ 𝒜𝐶
Zi,t Quantity of item 𝑖 procured in time-period t, ∀ 𝑖 ∈ ℰ\ {1}

Table 2-3: Decision variables

2.4 Variables:

Variables
δi,t Binary indicator of disassembly for item 𝑖 in time-period t, ∀ 𝑖 ∈ 𝒜𝐶
Ii,t+ Inventory level of item 𝑖 at the end of time-period t, ∀ 𝑖 ∈ ℰ\{1}
Quantity of item 𝑖 backordered in time-period t, ∀ 𝑖 ∈ ℰ\{1}
−
I i,t

Wi,t Quantity of defective item of item 𝑖 in time-period t, ∀ 𝑖 ∈ ℰ\{1}

Yt Disassembly ‘Overtime’ in time-period t, ∀ 𝑡 ∈ 𝒯
𝑄𝑖,𝑡 Sale quantity of item 𝑖 in time-period t, ∀ 𝑖 ∈ ℰ\{1}

Table 2-4: Variables

2.5 Objective function:

6
The objective of this problem is to optimize the overall profit by simultaneously reducing the

aggregate costs associated with inventory holding, backlogging, external procurement, setup,

and overload throughout the planning horizon.

2.6 Constraints:

 The inventory balance for each non-root item i at the end in each time-period t:

Ii,t + - Ii,t - = Ii,t ∀ iϵ ε \{1} , ∀ t ϵ T (2)

 Ii,t is the macro defined as follows:

Ii,t = Ii,t-1 +∑ R k,i . X k,t-Lk - D - Xi,t - Wi,t + Zi,t i,t

k ∈ ϕi

 All demands must be satisfied at the end of the planning horizon:

Iti,t + ∑ Di, 𝒯 ≥ 0, ∀i∈ e \{1} , ∀t∈ 𝒯 (3)

 A setup cost is generated in a period if any disassembly operation needs to be

performed in that period:

δi,t − Xi,t/M > 0, ∀i∈ 𝒜C , ∀t∈ 𝒯 (4)

 Defective quantity for each non-root item i in each time-period t:

 Disassembling capacity constraint in each time-period t:

∑ (𝑆𝑇𝑖. 𝛿𝑖,𝑡 + 𝐺𝑖. 𝑋𝑖,𝑡) ≤ 𝑈𝑡 + 𝑌𝑡, ∀t∈ 𝒯 (6)

7
 i∈ 𝒜
c

 The lower and upper bounds available for extra capacity on overtime in each time-period t:

0 ≤ 𝑌𝑡 ≤ 𝐹𝑡, ∀𝑡 ∈ 𝒯 (7)

 The sale quantity for each non-root item i and time-period t:

𝑄𝑖, 𝑡 = (𝐷𝑖,𝑡 − 𝑍𝑖,𝑡 − 𝐼𝑖,𝑡), ∀i∈ ℰ\{1}, ∀𝑡 ∈ 𝒯 (8)

 Equation (9) - (17) provide the conditions framing the decision variables:

Wi,t ≥ 0, ∀i∈ ℰ\{1}, ∀𝑡 ∈ 𝒯 (9)

Yt ≥ 0, ∀𝑡 ∈ 𝒯 (10)

Ii,t−≥ 0, ∀i∈ ℰ\{1}, ∀𝑡 ∈ 𝒯 (11)

Ii,t+ ≥ 0, ∀i∈ ℰ\{1}, ∀𝑡 ∈ 𝒯 (12)

Zi,t ≥ 0, ∀i∈ ℰ\{1}, ∀𝑡 ∈ 𝒯 (13)

Qi,t ≥ 0, ∀i∈ ℰ\{1}, ∀𝑡 ∈ 𝒯 (14)

Xi,t ≥ 0, ∀i∈ 𝒜C, ∀𝑡 ∈ 𝒯 (15)

Xi,t ≤ 0, ∀i∈ 𝒜, ∀𝑡 ∈ 𝒯 (16)

δi,t∈ {0,1}, ∀i∈ 𝒜C, ∀𝑡 ∈ 𝒯 (17)

8
 CHAPTER 3 RESULTS

3.1 Data Collection and Processing

3.1.1 Research structure

6 8
7
(1) (3) (3) (1)

DTL = 1 4 5 DTL = 0

(2) (1) (1)

DTL = 0 2 3

(1) (2)

1 DTL = 1

Figure 3-1: Illustrative example of

disassembly structure

Root item: (1)

Sub-assembly items: (2) (3) (4) (5)

Leaf items: (6) (7) (8)

- As illustrated in Fig. 1 above:

+ The root item is the first component in a disassembly process; components/leaf items and subassembly
items come next.

+ Across all EoL products, the usage of common parts lowers costs and material use.

+ The quantity obtained after deconstructing one unit of its parent is indicated by the number in

brackets.

+ The time interval between the order's release and the delivery of the disassembled products is known

as the disassembly lead time (L).

- After analyzing, we decided to use CPLEX Studio IDE 12.10. software to solve this problem.

Specifically, the structure of this model consists of one model file (*.mod), one data file (*.dat), and one

excel file (*.xlsx).

3.1.2 Some solution approaches

(Ullerich and Buscher 2013) formulated an integer linear programming model for multiple items with

capacity constraint in each period. (Ji et al. 2016) consider the importance of set up cost that was ignored

by (Ullerich and Buscher 2013).

This paper extended the above idea by adding start-up and the set-up costs and using a Lagrangian

relaxation heuristic to address a capacitated disassembly planning with parts commonality. Lost sales

and penalty cost was introduced in the research of capacitated lot-sizing disassembly problem for

overloading disassembly capacity by (Godichaud et al. 2016). An MIP model and a GA were proposed

to give a solution to this problem in a reasonable completion time for a large instance.

10
The combination between GA and Fix-and-Optimize heuristics by (Hrouga et al. 2016a) to solve the

multiple type products disassembly lot sizing problems with lost sales and capacity constraints. Then,

the authors extended the lot sizing problem by including lost sales, multi product types, two levels and

capacity constraint in (Hrouga et al. 2016b). Genetic algorithm and Fix-and-Optimize heuristics were

implemented to find the minimum sum of setup, inventory holding, lost sales and overload costs to solve

this problem. (Tian et Zhang 2018) formulated a Nonconvex Mixed Integer ¨Program. The

determination of accurate prices of returned products with suitable time and quality by using the Particle

Swarm Optimization (PSO) algorithm and dynamic programming.

3.1.3 Data

The sale price and purchasing cost of the model are calculated by the formula including two parameters:

γ represents the profit margin, and β represents the margin in relation to sale price. When: γ = 1.5 and β

= 1.8.

Demand is uniform distribution D U ( 0,100 ). The defective rate ai for each item i is calculated using

D U ( 0 ,10 ). For each item i , ∀ i∈ A C , we assume that the disassembly operation time Gi and setup time

S T i are D U (1 , 4 ) and D U (10 ,50 ), respectively.

The disassembly capacity, U t , and the available additional capacity, F t , are calculated using

D U ( 240,480 ) and D U ( 60,120 ). The lead time for disassembly is set to 0, 1, and 2 with the

probability of 0.3, 0.4, and 0.3, respectively. All the experiments are carried out at the parameter prices

shown in the table below.

Parameters Value
hi D U (12 , 20 )
11
 Si D U ( 0 ,1000 )
bi 2 ×hi
Ot D U ( 20 ,25 )
SCi γ × ( hi + Si ) ; γ=1.5
Ci β × S Ci ; β=1.8
Di ,t D U ( 0,100 )
ai D U ( 0 ,10 )
Gi D U (1 , 4 ), ∀𝑖 ∈ 𝒜𝐶
S Ti D U (10 ,50 ), ∀𝑖 ∈ 𝒜𝐶
Ut D U ( 240,480 )
Ft D U ( 60,120 )
Table 3-5: The generation of data

Index bi Ci ai Gi Ut Ft SCi S Ti
1 24 842.4 0 1 140 60 468 20
2 28 118.8 1 3 140 100 66 20
3 30 2470.5 1 4 140 120 1372.5 33
4 24 164.7 2 1 140 90 91.5 50
5 36 2489.4 2 2 1383 14
6 40 54 1 0 30 0
7 34 45.9 2 0 25.5 0
8 32 43.2 0 0 24 0

Table 3-6: Input data (1)

hi Si M Ot I0 L(k)
16 583 10000 20 100 1
20 425 23 100 0
17 810 21 100 1

12
 19 820 25 100 1
14 58 100 0
14 0 100 0
19 0 100 0
12 0 100 0

Table 3-7: Input data (2)

𝑹i,k 1 2 3 4 5 6 7 8
1 0 1 2 0 0 0 0 0
2 0 0 0 2 1 0 0 0
3 0 0 0 0 1 0 0 0
4 0 0 0 0 0 1 3 0
5 0 0 0 0 0 0 3 1
6 0 0 0 0 0 0 0 0
7 0 0 0 0 0 0 0 0
8 0 0 0 0 0 0 0 0

Table 3-8: Number of item j disassembled by item i

𝝓𝒊,𝒌 1 2 3 4 5 6 7 8
1 0 0 0 0 0 0 0 0
2 1 0 0 0 0 0 0 0
3 1 0 0 0 0 0 0 0
4 0 1 0 0 0 0 0 0
5 0 1 1 0 0 0 0 0
6 0 0 0 1 0 0 0 0
7 0 0 0 1 1 0 0 0
13
 8 0 0 0 0 1 0 0 0

Table 3-9: Set of parents of item i

D 1 2 3 4

1 0 0 0 0

2 100 100 100 80

3 47 100 70 40

4 20 100 38 38

5 100 44 34 74

6 37 98 37 36

7 57 65 57 47

8 100 100 100 100

Table 3-10: External demand of item I in period t

External demand is for non-root items, therefore Di,1=0 and demand for

non-root items is generated from the discrete uniform distribution D ∼ U

(0, 100).

3.2 Results Discussion

3.2.1 Result presentation

Maximum profit = 731,518

Item/period 1 2 3 4
1 0 0 0 0
2 0 1 1 1
3 0 0 1 0
4 1 1 1 0
5 0 0 0 0
6 0 0 0 0
14
 7 0 0 0 0
8 0 0 0 0

Table 3-11: Binary indicators

Item/period 1 2 3 4
1 0 0 0 0
2 0 0 0 0
3 0 0 0 0
4 0 0 0 0
5 0 0 0 0
6 8 0 0 0
7 0 10 0 0
8 0 0 0 0

Table 3-12: Backlogging level at the end of each period

Item/period 1 2 3 4
1 0 0 0 0
2 0 0 0 0
3 0 0 0 0
4 0 0 0 0
5 0 0 0 0
6 26 0 0 0
7 0 43 0 0
8 9 0 0 0

Table 3-13: Inventory level at the end of period t

Item/period 1 2 3 4
1 0 0 0 0

15
 2 100 100 100 80
3 47 100 70 40
4 20 100 38 38
5 100 44 34 74
6 19 98 8 33
7 57 32 57 47
8 100 0 0 0

Table 3-14: Sale quantity of item i in time-period t, ∀ i∈ E¿1 }¿

Item/period 1 2 3 4
1 0 0 0 0
2 0 0 1 0
3 53 208 198 260
4 0 0 1 102
5 0 10 2 0
6 45 0 0 0
7 43 142 0 52
8 0 0 0 0

Table 3-15: Defective quantity of item i in time-period t, ∀ i∈ E¿1 }¿

Item/period 1 2 3 4
1 154 136 150 0
2 0 54 36 70
3 0 0 4 0
4 80 8 33 0
5 0 0 0 0
6 0 0 0 0

16
 7 0 0 0 0
8 0 0 0 0

Table 3-16: Quantity of parent item i to disassemble in time-period t, ∀ 𝑖 ∈𝒜𝐶

1 0
2 100
3 120
4 90

Table 3-17: Disassembly ‘Overtime’ in time-period t, ∀ 𝑡∈𝒯

Item/period 1 2 3 4
1 0 0 0 0
2 0 0 0 0
3 0 0 0 0
4 0 0 0 0
5 0 0 0 0
6 0 0 29 3
7 0 0 0 0
8 0 100 100 100

Table 3-18: Quantity of item i to procure in time-period t, ∀ t ∈T

3.2.2 Result analysis

Revenue Total cost
Maximum Total Total Total Total Total
profit Total sale
purchase holding backlogging setup overload
price
cost cost cost cost cost
731,518 756,324 14,688 1,328 660 4,545 7,070
% Total Cost - 52% 5% 2% 16% 25%

17
 Table 3-19: Result analysis

From the table, purchasing cost takes the largest percentage of total cost, followed by overload cost and

setup cost.

3.2.3 Sensitivity analysis

 Changing disassembly capacity of the factory

Ut Profit
140 731,518
200 739,451.5
240 741,826.5
480 751,567.5

Table 3-20: Changing disassembly capacity of the factory

The above table shows that the higher disassembly capacity leads to the larger profit. This tendency

exhibits that capacity plays an important role in production. Therefore, finding ways to enhance the

manufacturing capacity is a considerable way to increase profitability.

 Changing setup cost

Setup cost is uniformly distributed with min =0 and max =1000. Changing setup cost of parent item 5

from 904 to 804:

Setup cost Selling price Purchasing cost Profit

904 1383 2489.4 731,518
804 1233 2219.4 693,718

Table 3-21: Changing Setup Cost

Setup cost decreases, total profit falls. Therefore, reducing setup cost may not increase total profit

18
because the selling price of item 5 is also reduced.

19
 CHAPTER 4 REFERENCES

[1] Slama, Ilhem, et al. "New mixed integer approach to solve a multi-level capacitated disassembly lot-sizing

problem with defective items and backlogging." Journal of Manufacturing Systems 56 (2020): 50-57.

[2] Taleb, K.N., Gupta, S.M., and Brennan, L. (1997). Disassembly of complex product structures with parts

and materials commonality. Production Planning & Control, 8(3), 255–269.

[3] Prakash P, Ceglarek D, Tiwari M. Constraint-based simulated annealing (cbsa) approach to solve the

disassembly scheduling problem. Int J Adv Manuf Technol 2012;60(9– 12):1125–37.

[4] Zhou, Fuli, et al. "Capacitated disassembly scheduling with random demand and operation time." Journal of

the Operational Research Society 73.6 (2022): 1362-1378.

[5] Ullerich, C., & Buscher, U. (2013). Flexible disassembly planning considering product conditions.

International Journal of Production Research, 51(20), 6209–6228. doi:10.1080/00207543.2013.825406

[6] Ji, X., Zhang, Z., Huang, S., & Li, L. (2016). Capacitated disassembly scheduling with parts commonality

and start-up cost and its industrial application. International Journal of Production Research, 54(4), 1225–1243.

[7] Godichaud, M., Amodeo, L., & Hrouga, M. (2016). Metaheuristic based optimization for capacitated

disassembly lot sizing problem with lost sales. Proceedings of 2015 International

[8] Hrouga, M., Godichaud, M., & Amodeo, L. (2016a). Heuristics for multi-product capacitated disassembly

lot sizing with lost sales. IFAC-PapersOnLine, 49(12), 628–633.

[9] Hrouga, M., Godichaud, M., & Amodeo, L. (2016b). Efficient metaheuristic for multi- product disassembly

lot sizing problem with lost sales. IEEE International Conference on Industrial Engineering and Engineering

Management, 2016–Decem, 740–744.

[10] Tian, X., & Zhang, Z. H. (2018). Capacitated disassembly scheduling and pricing of returned products

with price-dependent yield. Omega (United Kingdom).

20
 CONCLUSION

To maximize profit, the study focused on solving problems relating to

deterministic, capitalised, multi-level, and disassembly lot-sizing problems. This is

achieved by subtracting the sale price for the total amount of disassembly from

the total cost, which includes holding, backlog, external procurement, setup, and

overload costs.

This research proposes a Mixed-Integer Programming (MIP) model to

appropriately represent and solve the problem. The outcomes of randomly created

problems show that the MIP model is efficient in producing an ideal disassembly

plan within reasonable computation times. To be more precise, our model makes it

easier by offering such timetables for four items in eight intervals.

The results presented in Chapter 4 are adequate to satisfy the model's

assumptions. The sensitivity analysis on the disassembly capacity and its impact

on profit are also covered in the study. Larger disassembly capacities, hence,

translate into higher profits. When solving all restrictions and the objective function

in a short amount of time, the computational experiment on CPLEX IBM ILOG aids

in the mathematical problem's reduction in complexity.

In summary, the study that the authors have presented is correct and useful

when taking into account not only holding and setup costs as in other studies, but

also backlog items, outside procurement, disassembly "over time," and defective

products. They did not include the cost of disposal in the objective function,

despite taking into account products that are defective. Therefore, in order to
21
improve the reality in reverse logistics, our group suggests that disposal costs be

taken into account when solving the lot sizing problem.

22
 CHAPTER 5 APPENDIX

5.1 CPLEX CODE

//index
int indexofperiod = ...;
int indexofitem=...;
int firstleafitem=...;
int lastsubitem=...;
range period = 1..indexofperiod;
range itemj =2..indexofitem;
range itemi=1..indexofitem;
range leaf= firstleafitem..indexofitem;
range sub=2..lastsubitem;

//parameter
float T[period]=...; //set of time periods of the planning horizon
float E[itemi]=...;//set of items
float A[leaf]=...;///set of leaf item i
float Ac[sub] =...;//Set of sub-assemblies items
int R[itemi][itemi]=...;//Number of units of item j obtained from
disassembling one unit of item i
float D[itemi][period]=...;//external demfloat
I[itemi][period]=...;//inventory
float h[itemi]=...;//Per period inventory holding cost of one unit of item i
float S[itemi]=...;//Per period setup cost of parent item i
float b[itemi]=... ;//backlogging cost of one unit of item i
float C[itemi]=...; //purchasing cost of one unit of item i
float a[itemi]=...;//Defective rate of item i
float G[itemi]=...;//Disassembly operation time for parent item i
float phi[itemi][itemi] = ...; //Set of parents of item i
float U[period]=...;//Available capacity in time-period t
float F[period]=...;//Disassembly capacity on overtime in time-period t
float SC[itemi]=...;//Per-period sale price of item i
float ST[itemi]=...;//Setup time of parent item i
float M=...;//a large number
float O[period]=...;//Cost of adding a unit of extra capacity in time period
t
float I0[itemi]=...;//Starting inventory of item i
int L[itemi]= ...; //Disassembly lead time of parent item i

//desicion variables

dvar int+ X[itemi][period];//quantity of parent item to disassemble in time-

period t
dvar int+ Z[itemi][period];//quantity of procured item i to procure in time-
period t
23
dvar boolean binary[itemi][period];//Binary indicator of disassembly for
item i in time period t
dvar int+ Y[period];//Disassembly Overtime in time-period t
dvar int+ Iplus[itemi][period];//Inventory level of item i at the end of
time-period t
dvar int+ Iminus[itemi][period];//Backordered quantity of item i in time-
period t
dvar int+ W[itemi][period];//defective quantity of item i in time-period t
dvar int+ Q[itemi][period];//sale quatity of item i in time-period t
dvar int+ I[itemi][period]; //macro I

//objective function: Eq(1)

maximize sum (t in period , i in itemj) (SC[i] * Q[i][t] - h[i] * Iplus[i]
[t] - b[i] * Iminus[i][t] - C[i] * Z[i][t]) -sum (t in period) O[t] * Y[t]
- sum(t in period ,i in sub) S[i] * binary[i][t];

//Constraint
subject to
{
constraints_2: //The inventory balance for each non-root item i at the end
in each time-period t
forall (t in period,i in itemj)
Iplus[i][t]-Iminus[i][t]==I[i][t];

//defines the inventory balance for each non-root item i at the end in each
time-period t
forall (t in period,i in itemj: t==1)
I[i][t] == I0[i] + sum(k in itemi: phi[i][k] == 1 && t - L[k] >= 1)(R[k]
[i]*X[k][t - L[k]]) - D[i][t] - X[i][t] - W[i][t] + Z[i][t];

forall (t in period,i in itemj:t==2)

I[i][t] == I[i][t-1] + sum(k in itemi: phi[i][k] == 1 && t - L[k] >= 1)
(R[k][i]*X[k][t - L[k]]) - D[i][t] - X[i][t] - W[i][t] + Z[i][t];

forall (t in period,i in itemj:t==3)

I[i][t] == I[i][t-1] + sum(k in itemi: phi[i][k] == 1 && t - L[k] >= 1)
(R[k][i]*X[k][t - L[k]]) - D[i][t] - X[i][t] - W[i][t] + Z[i][t];

forall (t in period,i in itemj:t==4)

I[i][t] == I[i][t-1] + sum(k in itemi: phi[i][k] == 1 && t - L[k] >= 1)
(R[k][i]*X[k][t - L[k]]) - D[i][t] - X[i][t] - W[i][t] + Z[i][t];

constraint_3: //All demands must be satisfied at the end of the planning

horizon
forall (i in itemj, t in period)
I[i][t] + sum (l in period: l<=t &&l>=1) D[i][l]>=0;

24
constraint_4: //guarantees that a setup cost is generated in a period if any
disassembly operation needs to be performed in that period:
forall (i in sub, t in period)
X[i][t] <= M * binary[i][t] ;

contraint_6: //the disassembling capacity constraint in each time-period t

forall (t in period)
sum(i in sub) (ST[i] * binary[i][t] + G[i] * X[i][t]) <= U[t] + Y[t];

constraint_7: //the lower and upper bounds available for extra capacity on
overtime
forall ( t in period )
Y[t]<=F[t] && Y[t]>=0;

constraint_8: //sale quantity for each non-root item

forall (t in period, i in itemj)
Q[i][t] == D[i][t] - Z[i][t] - I[i][t];

//conditions:
forall (i in itemj, t in period)
{
Iminus[6][1]==8;
Iminus[7][2]==10;
W[i][t]>=0;
Iminus[i][t]>=0;
Iplus[i][t]>=0;
Z[i][t]>=0;
Q[i][t]>=0;
}
forall (i in sub, t in period)
X[i][t]>=0;
forall (i in leaf, t in period)
X[i][t]<=0;}

25
5.2 Output of changing disassembly capacity of the factory

Figure 5-2: Changing disassembly capacity of the factory Ut=140

26
Figure 5-3: Changing disassembly capacity of the factory Ut=200

27
Figure 5-4: Changing disassembly capacity of the factory Ut=240

28
Figure 5-5: Changing disassembly capacity of the factory Ut=480

29
Figure 5-6: Maximum profit when changing setup cost of item 5

30