Browsing by Author "Erdogan, M. Serdar"

Now showing 1 - 2 of 2

Citation - WoS: 1
Citation - Scopus: 2
Design of multi-product multi-period two-echelon supply chain network to minimize bullwhip effect through differential evolution
(IEEE, 2017) Ozgur Kabadurmus; M. Serdar Erdogan; M. Fatih Tasgetiren; Erdogan, M. Serdar; Tasgetiren, M. Fatih; Kabadurmus, Ozgur
A supply chain network consists of facilities located in dispersed geographical locations. This network structure can be optimized to minimize total cost or total inventory by deciding the order quantities and distribution of links connecting the facilities. However bullwhip effect (i.e. amplification of order fluctuations) is an important performance metric for supply chains because as the order variance increases in the downstream of the supply chain (e.g. distributors) the demand variance in the upstream (e.g. manufacturer) amplifies and causes inefficiencies in the supply chain. In this study we optimize supply chain network structure for multi-product multi-period two-echelon supply chain networks to minimize bullwhip. Due to nonlinear structure of the objective function i.e. bullwhip effect this paper proposes a differential evolution (DE) algorithms employing variable neighborhood search (VNS) and constraint handling methods to optimize supply chain network structure. The proposed algorithm is tested over randomly generated test instances and its effectiveness is demonstrated.
Citation - Scopus: 5
Solving 0-1 Bi-Objective Multi-dimensional Knapsack Problems Using Binary Genetic Algorithm
(Springer Science and Business Media Deutschland GmbH, 2021) Ozgur Kabadurmus; M. Fatih Tasgetiren; Hande Oztop; Mehmet Serdar Erdoğan; Tasgetiren, M. Fatih; Erdogan, M. Serdar; Oztop, Hande; Kabadurmus, Ozgur
The multi-dimensional knapsack problem (MDKP) is a well-known NP-hard problem in combinatorial optimization. As it has various real-life applications the MDKP has been intensively studied in the literature. On the other hand far too little attention has been paid to the multi-objective version of the MDKP. In this chapter we consider the bi-objective multi-dimensional knapsack problem (BOMDKP). We propose a Binary Genetic Algorithm (BGA) with an external archive for the problem. Our proposed BGA algorithm also employs a binary local search. The non-dominated solution sets are obtained for various bi-objective benchmark instances with 100 250 500 and 750 items by employing the proposed BGA. Then the performance of the BGA is compared with other multi-objective algorithms from the literature i.e. MOEA/D and MOFPA. Furthermore it is observed that the Pareto-optimal solution set provided by Zitzler and Laumans for 500 items and 2 knapsacks includes 30 dominated solutions. Also the Pareto-optimal solutions for the scenario with 750 items are not reported in Zitzler and Thiele [43]. Hence the true Pareto-optimal solution sets are found for all benchmark problem instances using Improved Augmented Epsilon Constraint (AUGMECON2) method. The non-dominated solution sets of the BGA MOEA/D and MOFPA are compared with the Pareto-optimal solution sets for all test instances. The computational results indicate that the proposed BGA is more effective to solve the BOMDKP than the best-performing algorithms from the literature. © 2020 Elsevier B.V. All rights reserved.