Incorporating Uncertainty in Truckload Relay Network Design Public Deposited

http://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/4f16c6224

Descriptions

Attribute NameValues
Creator
Abstract or Summary
  • In a relay network for full truckload (TL) transportation, facilities known as relay points (RPs) serve as exchange points where truck drivers can exchange trailers. This would help carriers to assign more regular tours to drivers when compared to the excessively long tours that exist in the traditional Point-to-Point (PtP) method. More regular driver tours would help to alleviate the driver turnover problem that significantly affects the industry. However, modifying the current system and completely replacing it with relay networks would not be practical. Instead a hybrid configuration known as truckload relay network design with mixed fleet dispatching (TLRND-MD) would allow certain loads to be delivered via the relay network while others are still served via the traditional PtP method. The strategic design of these hybrid networks entails locating RPs, determining the appropriate dispatching method for truckloads, and the selection of the appropriate route for those truckloads that are dispatched over the relay network. Most of the existing literature on the strategic design of truckload relay networks assumes deterministic parameters in the formulation of mathematical programs used to find optimal solutions. However, the TL transportation environment can be affected by uncertainty in terms of demands, travel times, transportation costs, disruptions, etc. Understanding the impacts of uncertainty on the design of relay networks for TL transportation is essential for making effective decisions. In this dissertation, we aim to explicitly incorporate demand uncertainty in the formulation of the TLRND-MD and the capacitated TLRND-MD problems. First, a robust optimization approach with a controllable level of conservatism is used to develop the robust counterpart for an existing mathematical model for the TLRND-MD problem. Solutions that perform well under any possible realization of the demand satisfying an uncertainty set are obtained for different network instances to show how incorporating uncertainty affects the total facility installation and transportation costs as well as other characteristics of the resulting truckload relay networks such as the required number of RPs. Then, we develop a two-stage stochastic programming formulation to capture demand uncertainty when demand is considered as a random variable governed by a posited probability distribution. Therefore, this approach allows us to optimize the expected transportation costs over scenarios of demand realizations. A Monte-Carlo simulation-based sampling algorithm known as Sample Average Approximation (SAA) is used to approximate the objective function value. Computational results are analyzed and compared with solutions obtained for the deterministic scenario. Finally, we propose to integrate the robust optimization and stochastic optimization approaches to incorporate demand uncertainty when the variability parameter which controls the level of conservatism in the robust formulation is also uncertain. We assume that the variability parameter follows a probability distribution, and develop a corresponding two-stage stochastic program. Based on the insights provided by the computational tests in this dissertation about the effect of uncertainty in the TLRND-MD and capacitated TLRND-MD problems, we conclude that robust optimization provides more conservative solutions compared to stochastic optimization, but it also provides a more tractable mathematical formulation. Considering the resulting network designs, robust optimization solutions show an increase in the transportation costs to move additional loads and more RPs needed when compared to deterministic solutions. On the other hand, as stochastic optimization does not deal with worst case values of demand, solutions obtained with this method tend to require fewer RPs compared to the deterministic solutions. However, stochastic programs seem to be computationally intractable for large size network instances and need to be coupled with efficient solution approaches to alleviate the computational burden.
Resource Type
Date Available
Date Copyright
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Committee Member
Academic Affiliation
Non-Academic Affiliation
Keyword
Subject
Rights Statement
Peer Reviewed
Language
Replaces
Additional Information
  • description.provenance : Approved for entry into archive by Julie Kurtz(julie.kurtz@oregonstate.edu) on 2017-03-24T18:54:28Z (GMT) No. of bitstreams: 2 license_rdf: 1223 bytes, checksum: d127a3413712d6c6e962d5d436c463fc (MD5) MokhtariZahra2017.pdf: 1585941 bytes, checksum: c03050c36ebfb999c1f947381d73113c (MD5)
  • description.provenance : Approved for entry into archive by Laura Wilson(laura.wilson@oregonstate.edu) on 2017-04-04T20:44:40Z (GMT) No. of bitstreams: 2 license_rdf: 1223 bytes, checksum: d127a3413712d6c6e962d5d436c463fc (MD5) MokhtariZahra2017.pdf: 1585941 bytes, checksum: c03050c36ebfb999c1f947381d73113c (MD5)
  • description.provenance : Made available in DSpace on 2017-04-04T20:44:40Z (GMT). No. of bitstreams: 2 license_rdf: 1223 bytes, checksum: d127a3413712d6c6e962d5d436c463fc (MD5) MokhtariZahra2017.pdf: 1585941 bytes, checksum: c03050c36ebfb999c1f947381d73113c (MD5) Previous issue date: 2017-03-13
  • description.provenance : Submitted by Zahra Mokhtari (mokhtarz@oregonstate.edu) on 2017-03-24T01:50:20Z No. of bitstreams: 2 license_rdf: 1223 bytes, checksum: d127a3413712d6c6e962d5d436c463fc (MD5) MokhtariZahra2017.pdf: 1585941 bytes, checksum: c03050c36ebfb999c1f947381d73113c (MD5)

Relationships

In Administrative Set:
Last modified: 08/16/2017

Downloadable Content

Download PDF
Citations:

EndNote | Zotero | Mendeley

Items