Tesis yeri seçim problemleri için akış tabanlı modellerin ve çözüm metodolojilerinin geliştirilmesi
No Thumbnail Available
Date
2017
Journal Title
Journal ISSN
Volume Title
Publisher
TUBİTAK
Open Access Color
OpenAIRE Downloads
OpenAIRE Views
Abstract
Tesis yeri seçim problemleri, yoğun olarak akademik çalışmaların yürütüldüğü alanlardan_x000D_
biridir. Ancak, bazı araştırmacılar tarafından, tesis yeri seçim modellerinin gerçek hayat_x000D_
uygulamalarını temsil etme ve çözmedeki yeterliliği uzun süredir sorgulanmakta ve yeni_x000D_
modellerin geliştirilmesine ihtiyaç olduğu ifade edilmektedir. Literatürdeki modellerin büyük_x000D_
bir çoğunluğu, modellerin gerçek hayattaki uygulama alanlarını sınırlandıran belirli_x000D_
varsayımlara dayanmaktadır. Bu varsayımların en önemlilerinden biri, modellerde girdi olarak_x000D_
kullanılan serim ve veri yapısıyla ilgilidir. Literatürdeki modeller, düğümler arası mesafe_x000D_
matrisinde en kısa yol uzunluklarının kullanıldığı tam serim (complete network) yapısı üzerine_x000D_
kuruludur. Modellerde tam serim yapısının kullanılması, gerçek hayattaki serimlerin (örneğin,_x000D_
demiryolları ya da karayolları) tam serim yapısında olmasından ziyade, araştırmacıların_x000D_
bazen doğrudan bazen de dolaylı olarak kabul ettiği bir varsayıma dayanmaktadır._x000D_
Araştırmacılar, gerçek hayat serimlerine en kısa yol algoritmalarının uygulanması suretiyle,_x000D_
düğümler arasında en kısa yolların kullanıldığı bir tam serim yapısının oluşturulduğunu_x000D_
varsaymaktadır. Diğer bir ifadeyle, modellerde girdi olarak kullanılan serim yapısı, düğümler_x000D_
arası mesafelerin üçgen eşitsizliğini sağladığı tam serimdir. Bu yaklaşım genel olarak kabul_x000D_
görmekle beraber, gerçek serim ve veri yapısının modellerde doğrudan girdi olarak_x000D_
kullanılmaması, modelleme ve çözüm açısından bazı dezavantajlara sebep olmaktadır. Daha_x000D_
da önemlisi, gerçek hayatta en kısa yolların tercih edilmediği veya üçgen eşitsizliğinin_x000D_
sağlanmadığı birçok durum vardır. Söz konusu tespitlerden hareketle, literatürdeki_x000D_
yaklaşımlardan tamamen farklı olarak, tam olmayan gerçek serim yapısının modellerde_x000D_
doğrudan girdi olarak kullanıldığı tesis yeri seçim problemleri tanımlanmıştır. Projede, tesis_x000D_
yeri seçiminde klasikler arasında kabul edilmeleri ve diğer tesis yeri seçim modellerinin_x000D_
temelini oluşturmaları nedeniyle, p-ortanca ve p-hub ortanca problemleri ele alınmıştır. Bu_x000D_
problemlerin, ayrıt/düğüm kapasiteli, kapasitesiz, tek ve çoklu atama ile farklı topolojilere izin_x000D_
veren versiyonları için modeller ve çözüm yöntemleri geliştirilmiştir. Geliştirilen modeller, hem_x000D_
gerçek serim yapısı, hem de (üçgen eşitsizliğini sağlamayan dahil) tam serim yapısı ile doğru_x000D_
sonuçlar vermektedir. Geliştirilen formülasyonlarda, daha çok tesis-talep noktası atama_x000D_
kararlarına dayanan literatürdeki modellerin aksine, ayrıt tabanlı akışlar esas alınmıştır._x000D_
Modellerin çözümü için, Benders Ayrıştırma ve Lagrange gevşetme algoritmaları_x000D_
geliştirilmiştir. Modellerin ve geliştirilen algoritmaların performansları, çeşitli problemler_x000D_
kullanılarak test edilmiştir.
Facility location problems are one of the mostly-studied areas. However, some researchers_x000D_ have been questioning the applicability of the facility location models to solve real-world_x000D_ problems and stating that there is a need to develop new models to better model real-world_x000D_ problems for quite long time. Most models in the literature depend on several assumptions_x000D_ that limit their application areas in in real life. One of the most important assumptions is about_x000D_ the network and data structures used as an input in the models. The models in the literature_x000D_ are based on the complete network structure where the distance matrix represents the_x000D_ shortest-path distances between node pairs. This starting point is not necessarily from_x000D_ assuming that the underlying real-world network (e.g., physical network such as road and rail_x000D_ networks) on which the hub system will operate is complete. It is implicitly or explicitly_x000D_ assumed that a complete-network structure is constructed from the shortest-path lengths_x000D_ between origin-destination pairs on the underlying real-world network through a shortest-path_x000D_ algorithm. Thus, the network structure used as an input in most models is a complete_x000D_ network with the distances satisfying the triangle inequality. Even though this approach has_x000D_ gained acceptance, not using the real-world network and its data structure directly in the_x000D_ models may result in several computational and modeling disadvantages. More importantly,_x000D_ there are cases in which the shortest path is not preferred or the triangle inequality is not_x000D_ satisfied. In this regard, we take a new direction completely different from the literature and_x000D_ define the facility location problems directly on non-complete networks that are_x000D_ representative of many real- world networks. p-median and p-hub median problems have_x000D_ been addressed in the project as they are accepted among the classical facility location_x000D_ models and form the building blocks of many other facility location models. Arc/node_x000D_ capacitated, uncapacitated, single- and multi-assignment versions with general topologies_x000D_ (e.g., allowing tree structure between hubs) of these problems have been investigated. The_x000D_ models can be used for both real networks and complete networks (including the ones not_x000D_ satisfying the triangle inequality). Unlike most models in the literature that are based on_x000D_ facility-demand point assignments, the new formulations are based on arc-based flows. To_x000D_ solve the models, Benders decomposition and Lagrangean based algorithms have been_x000D_ developed. The performances of the proposed models and algorithms have been assessed_x000D_ using several test problems.
Facility location problems are one of the mostly-studied areas. However, some researchers_x000D_ have been questioning the applicability of the facility location models to solve real-world_x000D_ problems and stating that there is a need to develop new models to better model real-world_x000D_ problems for quite long time. Most models in the literature depend on several assumptions_x000D_ that limit their application areas in in real life. One of the most important assumptions is about_x000D_ the network and data structures used as an input in the models. The models in the literature_x000D_ are based on the complete network structure where the distance matrix represents the_x000D_ shortest-path distances between node pairs. This starting point is not necessarily from_x000D_ assuming that the underlying real-world network (e.g., physical network such as road and rail_x000D_ networks) on which the hub system will operate is complete. It is implicitly or explicitly_x000D_ assumed that a complete-network structure is constructed from the shortest-path lengths_x000D_ between origin-destination pairs on the underlying real-world network through a shortest-path_x000D_ algorithm. Thus, the network structure used as an input in most models is a complete_x000D_ network with the distances satisfying the triangle inequality. Even though this approach has_x000D_ gained acceptance, not using the real-world network and its data structure directly in the_x000D_ models may result in several computational and modeling disadvantages. More importantly,_x000D_ there are cases in which the shortest path is not preferred or the triangle inequality is not_x000D_ satisfied. In this regard, we take a new direction completely different from the literature and_x000D_ define the facility location problems directly on non-complete networks that are_x000D_ representative of many real- world networks. p-median and p-hub median problems have_x000D_ been addressed in the project as they are accepted among the classical facility location_x000D_ models and form the building blocks of many other facility location models. Arc/node_x000D_ capacitated, uncapacitated, single- and multi-assignment versions with general topologies_x000D_ (e.g., allowing tree structure between hubs) of these problems have been investigated. The_x000D_ models can be used for both real networks and complete networks (including the ones not_x000D_ satisfying the triangle inequality). Unlike most models in the literature that are based on_x000D_ facility-demand point assignments, the new formulations are based on arc-based flows. To_x000D_ solve the models, Benders decomposition and Lagrangean based algorithms have been_x000D_ developed. The performances of the proposed models and algorithms have been assessed_x000D_ using several test problems.
Description
Keywords
p-Ortanca Problemi, p-Hub Ortanca Problemi, Serim Akış-Tabanlı Modeller, Tesis Yeri Seçim Modelleri, Karışık Tamsayılı Programlama, Üçgen Eşitsizliği, Tam Olmayan Serimler, p-Median Problem, p-Hub Median Problem, Non-complete Networks, Triangle Inequality, Network Flow-Based Models, Facility Location Models, Mixed Integer Programming,
Turkish CoHE Thesis Center URL
Fields of Science
Citation
WoS Q
Scopus Q
Source
Volume
Issue
Start Page
1
End Page
92
Page Views
4
checked on Dec 05, 2025