Ever-increasing customer expectations and fierce competition in global markets force manufacturing companies to continuously enhance competitiveness to stay profitable. In recent years, they have realized that manufacturing logistics, i.e. the management of material and information flows in manufacturing companies, has a considerable potential to reduce costs, improve customer service and provide them with a competitive advantage. Manufacturing logistics decision-making is, however, a complex and difficult task, and logistics professionals therefore continuously seek approaches and tools that help them take better decisions. One such approach is operations research (OR), which develops quantitative models and analyzes them to draw some conclusions about the model and, consequently, about the real world. OR has successfully supported manufacturing logistics for decades. It covers a wide variety of techniques, each with its strengths, limitations, success stories and group of advocates. They work well in certain situations, but none of them is a panacea that solves every problem. For logistics decision-makers, it is therefore crucial to understand the problem situations in which the different OR technique can provide added value, i.e. how applicable OR techniques are in different contexts. Research has shown that a mismatch between problem and model/technique is a frequent reason for failure of OR initiatives. There is also a considerable gap between the models and techniques described in the literature and those actually used for decision support in practice, which further emphasizes the need to understand their applicability, i.e. how useful they are in different situations.
A review of the literature revealed, however, that relatively little research has aimed to increase the understanding of the applicability of OR techniques. There is a paucity of literature providing details about the situations in which the techniques work well and there is relatively little guidance on selecting techniques. In practice, there seems to be considerable confusion and disagreement. Technique selection is in danger of being affected by personal preferences, and logistics professionals without OR background have little means to judge technique appropriateness.
The present thesis addresses these weaknesses in research and practice from the perspective of the operations management and logistics fields, which are concerned with effective decisionmaking in operations/logistics. The thesis’ overall objective is to increase knowledge on the applicability of OR techniques to support decision-making in manufacturing logistics, and to provide an overview of such knowledge for logistics professionals without OR background. This overall objective is achieved by means of three specific objectives: 1. To identify, classify and characterize the typical OR techniques used to support manufacturing logistics in practice, and to identify and classify the typical manufacturing logistics decisions supported by OR techniques. 2. To provide empirical evidence of how the applicability of OR techniques depends on different problem situation characteristics. 3. To develop guidelines that help logistics professionals understand if and how OR techniques can support a given real-world problem situation.
The overall methodological idea to achieve these objectives was to study a large number of successful OR applications, identify the areas in which the different OR techniques were useful, investigate how they were used, and develop guidelines based on findings and existing knowledge on OR applicability. Since literature contains hundreds of descriptions of OR applications, with details about the situations in which they took place, it was deemed appropriate to rely heavily on secondary literature. Two extensive surveys of successful applications described in the literature were carried out, one of the journal Interfaces, the other of Winter Simulation Conference proceedings. For a greater in-depth understanding, three case studies were performed as well. This provided a sample of close to 200 OR applications, which constituted the thesis’ empirical foundation. Thesis results were obtained by synthesizing this empirical data with existing literature and the researcher’s background and experience.
The main results of this thesis are as follows. (1) A classification of the main OR techniques used to support manufacturing logistics, namely deterministic optimization, discrete-event simulation, queuing theory and inventory theory. At such a high level of technique distinction, different techniques have different world views, provide decision support in different ways, are often practised by different people and are implemented in different types of software systems. At this level, technique selection is therefore of interest and importance not only to OR professionals, but also to logistics professionals responsible of taking sound decisions and seeking decision support. The thesis also includes a characterization of these techniques, based on the idea of paradigms, providing a general understanding of each technique’s key assumptions and properties.
(2) A classification of manufacturing logistics decisions supported by OR, including shortterm production planning/scheduling; plant location and distribution system design; production plant design; aggregate production and capacity planning; inventory management; the determination of production rules/policies; and transportation management. Integrated into a seven-by-four matrix, the two classifications provide a framework for systematic investigations of the applicability of OR techniques in manufacturing logistics.
(3) Substantial empirical evidence of the link between problem situation characteristics and OR usage. Focus is on five characteristics that seem to affect OR technique applicability, namely decision type, planning horizon, system scope, company size and industry. Empirical evidence was obtained from the two surveys performed as a part of this doctorate study, as well as from relevant surveys carried out by other researchers. The evidence is used to test claims made in the literature about the applicability of OR, as well as to put forward several new propositions. Additional empirical evidence of how problem situation characteristics affect technique applicability was obtained from the three case studies. In the first, Felleskjøpet Trondheim used deterministic optimization to support plant location and distribution system design; in the second, Gilde Norsk Kjøtt used discrete-event simulation to support production plant design and to determine production rules/policies; in the third, Mustad assessed the potential of multi-echelon inventory theory to reduce safety stocks in its global logistics network.
(4) Extensive guidelines on the applicability of OR techniques in manufacturing logistics. For the seven decision types typically supported by OR, these guidelines discuss OR technique applicability and provide links from detailed problem situation characteristics to suitable OR techniques. Furthermore, they contain descriptions of how OR techniques support the different decision types, with focus on practice-relevant issues such as input data requirements, the way the models are used in decision-making, relevant types of software systems, time/resource requirements etc. This provides an understanding of how OR works. Given a real-world problem situation in manufacturing logistics, the guidelines thus help assess if OR techniques can provide added value. They target people who need to be aware of the opportunities of OR without being OR professionals, such as logistics and operations managers. They are presented in a form and language that is relevant for this audience, without mathematics or computer jargon. Still, they can also be of interest to OR professionals, especially those new to the field; they highlighting promising application areas and contain structured references to hundreds of real-world OR applications described in the literature. The use and usefulness of the guidelines is illustrated by means of a real-world situation where they could have made OR technique selection more effective.
This thesis contributes to a theory of the practice of OR. The main benefits expected are less confusion about the areas in which OR techniques work well, more effective technique selection in practice, and increased exploitation of the opportunities of OR to support manufacturing logistics. Hopefully, it counteracts frequently returning discussions and even argument about the appropriateness of discrete-event simulation as opposed to optimization in logistics and supply chain management. Ultimately, such benefits will lead to more effective decision-making in manufacturing companies. For the research community, the thesis highlights practically relevant topics for future model development; it pinpoints areas in which further research is required to close the gap between theory and practice; and it can serve as a solid foundation for future research on OR applicability.