Loading...
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 41493 (01)
- University: Sharif University of Technology
- Department: Industrial Engineering
- Advisor(s): Eshghi, Koorosh
- Abstract:
- Hub location problem arises where the flow has to be routed between a set of origins and destinations, but direct links between all nodes are impossible or too costly. In these situations, e.g. transportation and communication systems, hubs serve as transshipment points to collect, transfer, and distribute the flow. In hub location allocation problem, hubs are located and non-hub nodes are allocated to them. In this thesis, two different models of capacitated multiple allocation hub location problem are discussed. In the first model, flow collection of hubs from non-hub nodes and in the second model, sum of flow collection and distribution of hubs from/to non-hub nodes is addressed. The underlying network is assumed complete with unstructured distances, in contrast to networks in which distances satisfy the triangle inequality. To route the flow on each link between non-hub nodes and hubs and to prevent constructing underutilized networks, it is assumed that there is a flow threshold on non-hub and hub links. The flow threshold constraint is dealt in two cases, to be dependent or independent to distribution. Additionally, an unconventional approach to multiple allocation hub location problem is presented by simultaneously considering service level and cost. The service level is defined by limiting the cover radius of hubs. Therefore, the problems are categorized as multiple allocation hub location- covering problems. In all problems, two mixed integer programming formulations are developed. Since a good formulation can help the solving process, some properties of optimal solutions are shown and preprocessing techniques are presented. On the other hand, improving techniques are applied to both formulations. In the first formulation, facet-defining inequalities are constructed for the uncapacitated version and in the second formulation, flow cover and knapsack cover inequalities are generated for the capacitated version Computational experiences on medium sized instances of “Australia Post Dataset” show that using preprocessing techniques in both formulations and generating cover inequalities in the second formulation can significantly improve the lower bound of the usual linear programming relaxation.
- Keywords:
- Location ; Flow Threshold ; Flow Cover Inequality ; Knapsack Cover Inquality ; Preprocessing
Digital Object List
-
محتواي پايان نامه
- view
Bookmark