### Abstract

Reverse logistics (RL) involves a sequence of operations that initiate at the consumer level and terminate at the manufacturer, opposite to the traditional forward approach of the supply chain. Recycling, reuse, and re-processing of products are activities of RL networks, all of which are becoming increasingly prevalent due to growing environmental and socio-economic concerns. Research has begun to study such networks in an effort to maximize efficiency and to improve operations. Previous work focused on developing a Mixed Integer Linear Programming (MILP) with an aim of determining the optimal location and capacity of important nodes of the RL network, such as inspection centers and remanufacturing facilities. Transportation decisions, such as whether to use in-house or outsourced vehicles, are often based on cost effectiveness. The problem is formulated for the case of a household appliance in the Gulf Cooperation Council (GCC) region. Sixty-eight cities are considered, leading to a very large number of variables and constraints; thus, a heuristic approach, namely a Genetic Algorithm (GA), is chosen to solve the problem. The main contribution of this paper is to develop a very efficient GA capable of solving a large scale problem in short time. The developed GA was capable of solving a very large problem (with 656,885 continuous variables, 2040 binary variables, 10 integer variables, and 100,340 constraints) with a gap of 0.3% and about 38.5 times faster than GAMS using a personal computer. The same GA succeeded to solve both large and small problems to optimality or with a gap that didn't exceed 1.5% and faster than GAMS. The technique that we used to code the GA reduced the number of variables and constraints to 92% and 86%, respectively. Furthermore, the reported results provide important insights on practical aspects of the problem, as well as useful points for the evaluation of the heuristic's performance.

Original language | English (US) |
---|---|

Pages (from-to) | 652-669 |

Number of pages | 18 |

Journal | Journal of Cleaner Production |

Volume | 151 |

DOIs | |

State | Published - May 10 2017 |

### Fingerprint

### Keywords

- Genetic Algorithms
- Metaheuristics
- Mixed integer programming
- Remanufacturing
- Reverse Logistics

### ASJC Scopus subject areas

- Renewable Energy, Sustainability and the Environment
- Environmental Science(all)
- Strategy and Management
- Industrial and Manufacturing Engineering

### Cite this

**A Genetic Algorithm for Reverse Logistics network design : A case study from the GCC.** / Alshamsi, Ahmed; Diabat, Ali.

Research output: Contribution to journal › Article

*Journal of Cleaner Production*, vol. 151, pp. 652-669. https://doi.org/10.1016/j.jclepro.2017.02.096

}

TY - JOUR

T1 - A Genetic Algorithm for Reverse Logistics network design

T2 - A case study from the GCC

AU - Alshamsi, Ahmed

AU - Diabat, Ali

PY - 2017/5/10

Y1 - 2017/5/10

N2 - Reverse logistics (RL) involves a sequence of operations that initiate at the consumer level and terminate at the manufacturer, opposite to the traditional forward approach of the supply chain. Recycling, reuse, and re-processing of products are activities of RL networks, all of which are becoming increasingly prevalent due to growing environmental and socio-economic concerns. Research has begun to study such networks in an effort to maximize efficiency and to improve operations. Previous work focused on developing a Mixed Integer Linear Programming (MILP) with an aim of determining the optimal location and capacity of important nodes of the RL network, such as inspection centers and remanufacturing facilities. Transportation decisions, such as whether to use in-house or outsourced vehicles, are often based on cost effectiveness. The problem is formulated for the case of a household appliance in the Gulf Cooperation Council (GCC) region. Sixty-eight cities are considered, leading to a very large number of variables and constraints; thus, a heuristic approach, namely a Genetic Algorithm (GA), is chosen to solve the problem. The main contribution of this paper is to develop a very efficient GA capable of solving a large scale problem in short time. The developed GA was capable of solving a very large problem (with 656,885 continuous variables, 2040 binary variables, 10 integer variables, and 100,340 constraints) with a gap of 0.3% and about 38.5 times faster than GAMS using a personal computer. The same GA succeeded to solve both large and small problems to optimality or with a gap that didn't exceed 1.5% and faster than GAMS. The technique that we used to code the GA reduced the number of variables and constraints to 92% and 86%, respectively. Furthermore, the reported results provide important insights on practical aspects of the problem, as well as useful points for the evaluation of the heuristic's performance.

AB - Reverse logistics (RL) involves a sequence of operations that initiate at the consumer level and terminate at the manufacturer, opposite to the traditional forward approach of the supply chain. Recycling, reuse, and re-processing of products are activities of RL networks, all of which are becoming increasingly prevalent due to growing environmental and socio-economic concerns. Research has begun to study such networks in an effort to maximize efficiency and to improve operations. Previous work focused on developing a Mixed Integer Linear Programming (MILP) with an aim of determining the optimal location and capacity of important nodes of the RL network, such as inspection centers and remanufacturing facilities. Transportation decisions, such as whether to use in-house or outsourced vehicles, are often based on cost effectiveness. The problem is formulated for the case of a household appliance in the Gulf Cooperation Council (GCC) region. Sixty-eight cities are considered, leading to a very large number of variables and constraints; thus, a heuristic approach, namely a Genetic Algorithm (GA), is chosen to solve the problem. The main contribution of this paper is to develop a very efficient GA capable of solving a large scale problem in short time. The developed GA was capable of solving a very large problem (with 656,885 continuous variables, 2040 binary variables, 10 integer variables, and 100,340 constraints) with a gap of 0.3% and about 38.5 times faster than GAMS using a personal computer. The same GA succeeded to solve both large and small problems to optimality or with a gap that didn't exceed 1.5% and faster than GAMS. The technique that we used to code the GA reduced the number of variables and constraints to 92% and 86%, respectively. Furthermore, the reported results provide important insights on practical aspects of the problem, as well as useful points for the evaluation of the heuristic's performance.

KW - Genetic Algorithms

KW - Metaheuristics

KW - Mixed integer programming

KW - Remanufacturing

KW - Reverse Logistics

UR - http://www.scopus.com/inward/record.url?scp=85016587109&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85016587109&partnerID=8YFLogxK

U2 - 10.1016/j.jclepro.2017.02.096

DO - 10.1016/j.jclepro.2017.02.096

M3 - Article

VL - 151

SP - 652

EP - 669

JO - Journal of Cleaner Production

JF - Journal of Cleaner Production

SN - 0959-6526

ER -