A multi-objective solution approach for the design and planning of real-life retail distribution networks

1. Auflage 2017
ISBN: 978-3-9818126-3-3
Verlag: Zadek Management & Strategy

Buch, Englisch, Band 3, 231 Seiten, GEKL, Format (B × H): 148 mm x 210 mm, Gewicht: 340 g

Reihe: Zadek-Publikationen zur Logistik

Kurzfassung:
Ziel dieser Arbeit war die Entwicklung eines zweistufigen Verfahrens für die realitätsnahe Gestaltung und Planung von sekundären Distributionsnetzwerken im filialisierten Handel. In der ersten Stufe wird ein Optimierungsproblem gelöst, welches folgende Planungsentscheidungen berücksichtigt: Lagerauswahl, Zuordnung von Filialen zu Lagern, Auswahl von
Belieferungsrhythmen für die Filiale sowie Gestaltung von Rundfahrten für die Belieferung der Filiale unter Berücksichtigung von Zeitfenstern. Dabei erfolgt nicht nur die Optimierung der Logistikkosten, sondern auch die Glättung der Lagerlast über den Verlauf der Woche. Hybride Metaheuristiken werden sowohl für die Kostenoptimierung als auch für die Kosten-Last-Optimierung entwickelt. Als Ergebnis der ersten Stufe steht die Pareto-optimale Front, welche eine Menge von effizienten Lösungen beinhaltet, die sich in der Netzwerkstruktur bzw. Planungsqualität unterscheiden. Die Pareto-optimale Front dient als Eingangsgröße für die zweite Stufe des Verfahrens, welche den Entscheidungsträger in der Auswahl der geeignetsten Lösung unterstützt. Dafür wird der Analytische Hierarchieprozess (AHP) angewendet. Die Praxiseignung des zweistufigen Planungsverfahrens und die Vorteile eines gleichmäßigen Lastverlaufs werden im Rahmen einer Fallstudie basierend auf Realdaten untersucht.

Zielgruppe

Managers in retail sites, who are responsible for the design and planning of secondary distribution networks. The solution approach might be applied to plan new distribution networks or to optimize already existing networks.

Autoren/Hrsg.

Hayden Bofill, Paula Cristina

Zadek, Hartmut

Inhaltsverzeichnis

Table of Contents
Table of Contents ........................................................................................................................ i
List of Figures ............................................................................................................................. v
List of Tables ........................................................................................................................... viii
Abbreviations ............................................................................................................................ ix
1 Introduction ........................................................................................................................ 1
1.1 Research scope ............................................................................................................. 4
1.2 Research goals ............................................................................................................. 5
1.3 Thesis structure ............................................................................................................ 8
2 Supply Chain Management in the German retail sector ..................................................... 9
2.1 The importance of logistics for the trade sector in Germany....................................... 9
2.1.1 The trade sector in Germany ................................................................................ 9
2.1.2 Logistics as a success driver for retailers ........................................................... 13
2.2 Features of logistic networks in the chain store retail sector ..................................... 15
2.2.1 Forms of delivery and arising logistic networks ................................................ 16
2.2.2 Supply chain segmentation ................................................................................. 23
2.2.3 Interdependencies among warehousing, transportation, and store logistics ....... 25
2.3 Planning processes in the retail sector ....................................................................... 32
2.3.1 Planning processes within a general supply chain management framework ...... 32
2.3.2 A planning framework for chain store retail ...................................................... 35
2.4 Summary .................................................................................................................... 42
3 Single-objective and multi-objective combinatorial optimization ................................... 45
3.1 Linear combinatorial optimization ............................................................................. 46
3.1.1 Linear optimization problems ............................................................................. 47
3.1.2 Linear combinatorial optimization problems ..................................................... 48
3.2 Multi-objective linear combinatorial optimization .................................................... 51
3.2.1 Multi-objective combinatorial optimization problems ....................................... 51
3.2.2 Pareto-optimality ................................................................................................ 53
ii Table of Contents
3.3 Computational complexity ........................................................................................ 54
3.3.1 Decision problems ............................................................................................. 55
3.3.2 Time complexity function .................................................................................. 55
3.3.3 Polynomial time reducibility .............................................................................. 57
3.3.4 Complexity classes ............................................................................................. 57
3.3.5 Combinatorial optimization problems and complexity classes .......................... 60
3.4 Solution methods for single-objective combinatorial optimization .......................... 63
3.4.1 Exact methods for linear optimization problems ............................................... 64
3.4.2 Approximation algorithms ................................................................................. 66
3.4.3 Heuristical search methods ................................................................................ 67
3.5 Solution methods for multi-objective combinatorial optimization ........................... 75
3.5.1 Overview of solution methods ........................................................................... 75
3.5.2 Population-based modern heuristics .................................................................. 79
3.6 Summary ................................................................................................................... 81
4 Applications of relevant combinatorial optimization problems ....................................... 83
4.1 From single-objective to multi-objective design of retail distribution networks ...... 83
4.1.1 Periodic Location-Routing Problem with Time Windows (PLRPTW) ............. 84
4.1.2 Periodic Location-Routing Problem with Time Windows and Depot Balance . 90
4.2 Single-objective sequential and integrated optimization problems ........................... 92
4.2.1 Facility Location Problems ................................................................................ 92
4.2.2 Vehicle Routing Problems ................................................................................. 97
4.2.3 Location-Routing Problems ............................................................................. 102
4.3 Multi-objective sequential and integrated optimization problems .......................... 106
4.3.1 Multi-objective Facility Location Problems .................................................... 107
4.3.2 Multi-objective Vehicle Routing ..................................................................... 111
4.3.3 Multi-objective Location-Routing ................................................................... 114
4.4 Workload balance as optimization objective ........................................................... 116
4.5 Summary ................................................................................................................. 121
Table of Contents iii
5 Solution procedure for the Periodic Location-Routing Problem with Time Windows
and Depot Balance .............................................................................................................. 123
5.1 Solution method selected for multi-objective optimization (MOO) ........................ 124
5.1.1 NSGA-II ........................................................................................................... 124
5.1.2 ELSxPR ............................................................................................................ 133
5.1.3 Hybrid algorithm for the Periodic Location-Routing Problem with Time
Windows and Depot Balance ........................................................................... 140
5.2 Solution method selected for multi-criteria decision making (MCDM) .................. 144
5.3 Summary .................................................................................................................. 149
6 Application of the two-step solution procedure to a real-life case study ....................... 151
6.1 Input data ................................................................................................................. 152
6.2 First step: Network optimization ............................................................................. 156
6.2.1 Optimization results for North Germany .......................................................... 157
6.2.2 Optimization results for South Germany .......................................................... 166
6.3 Second step: Selection of the most suitable network ............................................... 172
6.3.1 Results for North Germany .............................................................................. 174
6.3.2 Results for South Germany .............................................................................. 178
6.4 Summary .................................................................................................................. 182
7 Conclusions and outlook ................................................................................................ 185
References .............................................................................................................................. 189
Appendix A: ELSxPR algorithm ............................................................................................ 227
Algorithm of the hybrid ELSxPR for the Location-Routing Problem ............................... 227
Algorithm of the evaluation of an individual from the ELS ............................................... 228
Appendix B: NSGA-II algorithm ........................................................................................... 229
Algorithm for the main loop ............................................................................................... 229
Algorithm for the fast-nondominated sorting procedure .................................................... 229
Algorithm for the crowding-distance-assignment (I) ......................................................... 230
Appendix C: Decision matrices for the Analytic Hierarchic Process (AHP) ......................... 231

Leseproben

Prologue:
Retail logistic costs amount to around 16% of the overall costs on average, whereas transport costs amount to around 30% of the logistic costs. Therefore, efficient cost management plays a crucial role for retailers. In order to reduce logistics costs, retailers have identified the importance of planning the operations of their own supply chains and have started to take it over from their suppliers. By developing planning processes for logistic networks, retailers can strengthen their own ability to optimize their networks.
Planning of logistic networks requires powerful tools to handle the size and complexity of the networks. In spite of superior models it can happen that the solutions, which are based solely on the optimization of costs, are not suitable for practical application. This is why tools that can handle more than one goal at a time are necessary. A great amount of work on general multi-objective optimization is available on the standing scientific literature. Yet, problems considering the particular requirements of retail distribution networks, particularly the balance of the workload of the warehouses throughout the week have not been studied yet. There are many reasons to consider the workload balance as a critical goal: economic reasons such as the use of resources, opportunity costs and economies of scale, organizational reasons such as personal planning and social fairness, as well as performance reasons e.g., planning reliability and operational performance.
Paula Cristina Hayden Bofill`s dissertation focuses on this problem as well as on other research gaps in this very current topic and proposes the Multiobjective Periodic Location-Routing Problem with Time Windows and Depot Balance. She received an excellent grade on her dissertation. Paula Cristina Hayden Bofill developed a two-step solution method for the realistic configuration and planning of secondary distribution networks in the retail chain store market. In the first step an optimization problem is taken into consideration, in which the following
decisions are made: selection of warehouses, allocation of retail stores to warehouses, definition of delivery patterns for retail stores as well as the design of delivery routes taking time slots into consideration. Thereby, not only the logistic costs are optimized, but also the workload within the warehouses during the week is equalized. Hybrid metaheuristic methods
were developed to optimize the costs and find a Pareto front for the cost-workload scenario. The Pareto front contains several efficient solutions which differ on the structure of the network and the quality of planning. The Pareto front is also the main input for the second step in the process that aids decision-maker in finding the most suitable solution. The analytic hierarchic process is used to make these decisions. The suitability for the practical application of the two-step planning process and the advantages of equalized load planning are supported by a case study based on real data.
Magdeburg, November 2017
Prof. Dr.-Ing. Hartmut Zadek