แบบจำลองทางคณิตศาสตร์และขั้นตอนสำหรับแก้ปัญหาการจัดเส้นทางโดยประยุกต์การจัดวางโครงข่าย

Vehicle routing is a major problem in commercial delivery because routing decision can affect the cost of the operation significantly. Effective routing decision can lead to significant cost savings but the decision is highly complex and difficult to analyze manually. Decisions based on experience o...

Full description

Saved in:
Bibliographic Details
Main Author: พรเทพ ตรีช่อวิทยา
Other Authors: มาโนช โลหเตปานนท์
Format: Theses and Dissertations
Language:Thai
Published: จุฬาลงกรณ์มหาวิทยาลัย 2008
Subjects:
Online Access:https://digiverse.chula.ac.th/Info/item/dc:44353
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Chulalongkorn University
Language: Thai
Description
Summary:Vehicle routing is a major problem in commercial delivery because routing decision can affect the cost of the operation significantly. Effective routing decision can lead to significant cost savings but the decision is highly complex and difficult to analyze manually. Decisions based on experience often seek feasibility of the solution rather than optimality. Mathematical models and algorithms are applied to the problem and significant savings have been shown. This thesis outlines a new approach in using mathematical programming model to tackle the vehicle routing problem. The model developed is an arc-based model with self-organizing network. The author presents mathematical formulation and solution approach and tests the model with a number of test data sets. The model shows applicability comparable to existing models in the literature but with two significant advantages. Compared to traditional arc-based model with sub-tour elimination, the proposed model is smaller in the number of rows. Compared to traditional path-based model with routings as variables, the proposed model is smaller in term of the number of variables. Thus, the proposed model is significantly smaller than traditional models and hence does not require advanced solution techniques such as row- or column-generation like the traditional models. In summary, the proposed model is easier to implement and require smaller memory unit. The computational test shows satisfactory results compared to the traditional models.