Skip to main content
Kent Academic Repository

An investigation into Supplier Selections and Contingency Freight Consolidation for Less-Than-Truckload Logisitics

Kaya, Rukiye (2022) An investigation into Supplier Selections and Contingency Freight Consolidation for Less-Than-Truckload Logisitics. Doctor of Philosophy (PhD) thesis, University of Kent,. (doi:10.22024/UniKent/01.02.97529) (KAR id:97529)

Abstract

This study deals with the supplier selection problem in which truck com panies are considered as supplier for the transportation service and freight consolidation scheduling problems for Third Party Logistic (3PL) companies. We present two novel investigations for the supplier selection problem. In the first one, we make some analyses on the commonly used methods for supplier selection problems which are the Multi-Criteria Decision Making (MCDM) methods, namely Analytic Hierarchy Process (AHP), Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) and VIekriter ijumsko KOmpromisno Rangiranje (VIKOR). Then we evaluate the results of each method based on the other two methods and conduct some tests for relying on a single method by the regret based measure approaches that we developed. In this way, we offer two effective approaches for combining the results of the individual MCDM approaches. Note that we do not propose an integration of the approaches, but a combination of the results of different MCDM methods in a systematic way instead of relying on a result of an indi vidual method. In the second study of supplier selection, we handle the issue of missing expert knowledge. When data is not available, researchers rely on expert knowledge. Therefore, there is a tendency to use MCDM methods for supplier selection problem due to working ability of MCDM methods with expert knowledge. However, experts do not always have full knowledge of all evaluation criteria. We offer a reliable solution for this problem. We integrate MCDM methods and Bayesian Network (BN) in a novel way that they can compensate each others' limitations with their strengths. We mainly rank the alternative suppliers with TOPSIS which has two inputs: weights of the deci sion criteria and initial decision matrix. We obtain the weights of the criteria from AHP and elicit the initial decision matrix from BN. Causal graphical structure and parameterization of BN is done by Decision Making Trial and Evaluation Laboratory (DEMATEL). Here, experts submit their knowledge about the decision criteria linguistically. Ranked Nodes tool of BN provides the experts to submit their knowledge with verbal expressions in an ordinal scale as low, medium, high. If the experts do not have full knowledge about some of the criteria BN estimates the missing value of criteria based on the available knowledge of the experts and causal relationship between the cri teria probabilistically. According to the obtained new knowledge(evidence) BN updates the values of the network and provides updated information to decision makers dynamically. Finally, we conducted sensitivity analyses for the value of knowledge followed by a case study. In the second part of this research, we investigate the freight consolidation scheduling problem. We address the problem in a particular way due to the preference of a 3PL company that operates in the UK. We consolidate the orders up to 3. First we investigate the possible consolidation configurations of orders as singleton(one), pair and triplets. We compute all the savings obtained by consolidation among non-consolidation case. Then we use these configurations and their saving values as input in our exact approaches like the 0-1 Integer Linear Programming (ILP) and the set partitioning formula tion which we developed. We also presented some tightening constraints into the set partitioning formulation and tested them for different size of the data sets. On the other hand we also tackled the problem using metaheuristics to overcome the computational time for larger instances. We offered Variable Neighbourhood Search (VNS) algorithm using six neighbourhood structures and two local searches: one performs within the route and the other one performs between the routes. The proposed neighbourhood structures are compatible with the purpose of the improvement of the consolidated ship ment configurations up to three requests. On the other hand to perturb the solution and improve it with the repair mechanism we offered Large Neigh bourhood Search (LNS) algorithm. In LNS algorithm, one of the removal operators performs effectively in a guided way by destroying the consolida tion configurations which have negative effect on savings. We also propose to hybridize the VNS/LNS algorithms. Lastly we discuss about the com putational results in terms of deviation from the optimal results and com putational time effectiveness. We finalize the study with a summary of the research, limitations and suggestions for further work. The thesis is made up of eight chapters. In the first chapter, the problem definition, a brief of the study and contributions are presented. In chapter 2, the literature review for supplier selection and order consolidation scheduling problems are given. Chapter 3 propose a deterministic rule for the combina tion of the results of different methods for supplier selection problem while Chapter 4 deals with the case of lack of complete expert knowledge for the supplier selection problem and proposes a novel MCDM-BN integration for this purpose. Chapter 5 discusses the order consolidation scheduling problem and defines all the possible configurations with their respective savings. In chapter 6, exact approaches for the order consolidation scheduling problem are provided, namely, a 0-1 ILP and a set partitioning formulation enhanced by valid inequalities. Chapter 7 treats the same scheduling problem by de signing and implementing two metaheuristics approaches, namely, variable neighbourhood search (VNS) and large neighbourhood search (LNS) as well as their hybridisation. Chapter 8 is the final chapter, it covers a summary of our findings and present limitations of the study and outlines suggestions as future work.

Item Type: Thesis (Doctor of Philosophy (PhD))
Thesis advisor: Spiegler, Virginia
Thesis advisor: Salhi, Said
DOI/Identification number: 10.22024/UniKent/01.02.97529
Uncontrolled keywords: Supplier Selection, Freight Consolidation, Less-Than-Truckload Logistics, MCDM, Bayesian Networks, VNS
Subjects: H Social Sciences
Divisions: Divisions > Kent Business School - Division > Department of Analytics, Operations and Systems
SWORD Depositor: System Moodle
Depositing User: System Moodle
Date Deposited: 21 Oct 2022 08:10 UTC
Last Modified: 24 Oct 2022 09:52 UTC
Resource URI: https://kar.kent.ac.uk/id/eprint/97529 (The current URI for this page, for reference purposes)

University of Kent Author Information

Kaya, Rukiye.

Creator's ORCID:
CReDIT Contributor Roles:
  • Depositors only (login required):

Total unique views for this document in KAR since July 2020. For more details click on the image.