SAMINT-MILI-21019
Master’s Thesis 30 credits June 2021
Applicability of simulation analysis for planning agri-food supply chains
A case study at a Swedish farmer-owned cooperative.
Marcus Ahlqvist
Master’s Programme in Industrial Management and Innovation
Abstract
Applicability of simulation analysis for planning agri-food supply chains
Marcus Ahlqvist
Agri-food supply chains (ASCs) are becoming increasingly complex, and its actors are in need of sophisticated planning tools to remain competitive in an industry that have been moving away from small individual actors towards large multi-national, highly vertically integrated cooperatives. This case study views reality from an objectivist point of view and utilises a positivist approach to study this reality. It combines qualitative and quantitative methods to study an ASC operating in the Kingdom of Sweden. This ASC’s planning processes are investigated in order to identify processes that are applicable to simulation by considering model validation, verification, and credibility.
The simulation model allowed for system analyses from a strategic perspective and, hence, simplified the planning process of evaluating different scenarios. The model was intrinsically verified and validated in consultation with the supervisor and subject reader and was thus able to accurately imitate the real-world system.
The simulated scenarios comprised changes to the ASC’s infrastructure or design. The changes, in turn, comprised decommissions of one or more port- site storage facilities (HPs). Questions that were asked during the evaluation of the experiment results included what happens to the inventory levels of the non- decommissioned HPs when certain ones close? will the demand still be met? and if, then where, will queues arise in the system? It is shown that the non- decommissioned HPs will manage the closed HP’s volumes, but only to a certain extent. One closed HP does not cause severe problems, while two closed ones can create queues, which, in turn, will result in lower than desired inventory levels at the end of the harvesting period. Queues will arise from the closing of just one HP, although this queue is practically negligible, but as two are closed, the queues will create problems. The demand was able to be met even though an HP was closed, but to meet it while two HPs are closed, one of the non- decommissioned ones’ capacity must be increased. This, ultimately, generated or achieved for the host organisation a so called proof of concept (this is argued to generate credibility in the model). Some of the identified characteristics of their ASC are considered generic, while others can only be claimed to be specific the studied ASC. The study thus claims to have initiated a framework for the differentiation of strategic, tactical, and operational planning levels in an ASC.
Keywords: agri-food supply chain, simulation, applicability, planning, system analysis, strategic, tactical, operational.
Supervisor: Daniel Maidell
Subject reader: Matías Urenda Morris Examiner: David Sköld
SAMINT-MILI-21019
Printed by: Uppsala Universitet
Faculty of Technology
Visiting address:
Ångströmlaboratoriet Lägerhyddsvägen 1
Postal address:
Box 536 751 21 Uppsala
Telephone:
+46 (0)18 – 471 30 03 Telefax:
+46 (0)18 – 471 30 00 Web page:
http://www.teknik.uu.se/education/
Popular science summary
Industrialisation shifted firms into industries and self-service manufacturing into production lines. At a similar point in time came Taylorism or Scientific Management and, in turn, centralisation and specialisation of work. Nowadays, this is safe to consider as a building block – although the degree to which is arguable – to more modern operations research, where the topic of mismatching supply and demand, e.g. overstocking, stockout, and delivery delays, is popular to research.
Supply chains are becoming larger and more complex, costly, uncertain, and vulnerable.
Today’s globalisation and increasing food consumption, stress the need for efficient agri-food logistics. The industry is undergoing drastic transformation, whereby small independent actors are being overrun by tightly aligned, highly vertically-integrated global firms and cooperatives.
Agri-food supply chains (ASCs) are progressively resembling supply chains in the manufacturing sector, and competing producers need sophisticated planning tools to remain competitive. The automotive industry has had tremendous success with simulation tools, which lets the user visualise and analyse the turnout of, for example, a system reconfiguration before putting effort into making it a reality. However, very little has been researched in the domain of ASC simulation and, in particular, simulating ASC-specific planning processes.
This study’s purpose is to investigate the hierarchical differentiation of planning processes in an ASC, and to determine which of them that are applicable to discrete-event simulation.
This study views reality from an objectivist point of view and utilises a positivist approach to study it. A mixed qualitative and quantitative methodology is utilised to study an ASC operating in the Kingdom of Sweden. Its planning processes are investigated through interviews in order to identify processes that are applicable to a quantitative simulation modelling approach by considering model validation, verification, and credibility.
The simulation model is intended to allow system analyses from a strategic perspective and, hence, simplify the strategic planning process of evaluating different scenarios. These scenarios comprised changes to the ASC’s infrastructure or design. The changes, in turn, comprised decommissions of one or more port-site storage facilities (HPs). Questions that were going to be asked included what happens to the inventory levels of the non-decommissioned port storages when certain ones close? will the demand still be met? and if, then where, will queues arise in the system? Essentially, the other, non-decommissioned HPs will manage the closed HP’s volumes, but only to a certain extent. The experiments showed that one closed HP does not cause severe problems, while two closed ones can create queues, which, in turn, will result in lower than desired inventory levels at the end of the harvesting period. Queues will arise from the closing of just one HP, although this queue is practically negligible, but as two are closed, the queues will, as mentioned, create problems, not the least in terms of transportation expenses. The demand was able to be met even though an HP was closed, but to meet it while two HPs are closed, one of the non-decommissioned ones’ capacity must be increased.
Ultimately, these results showed the host organisation the potential of simulation analysis for
decision making – it resulted in a proof of concept. The study also initiated a framework for the
differentiation of strategic, tactical, and operational planning levels in the ASC. Some identified
characteristics of this ASC are considered generic, while others can only be claimed to be
specific the studied ASC. The strategic level is shown to be perfectly applicable to simulation
as a decision-support tool for planning ASCs, system analysis, and system design evaluation.
Forewords and acknowledgements
This research project constitutes a master’s degree thesis project within the programme Industrial Management and Innovation at Uppsala University in Sweden. It was conducted during the spring semester 2021, written by the student Marcus Ahlqvist in collaboration with a Swedish farmer-owned cooperative with operations in agriculture, machinery, bioenergy and production of food.
The author wants to extend particular gratitude towards the subject reader Matías Urenda
Morries for his encouragement throughout the tiresome validation process. The author is also
grateful for the generous encouragement and suggestions received throughout the project by
the supervisor Daniel Maidell, the time taken by the interviewees to provide the author with
great insight into their supply chain as well as the planning processes of it, and for the overall
warming welcome by the rest of the host organisation.
Table of Contents
1 Introduction ... 1
1.1 Background ... 1
1.2 Problem definition ... 3
1.3 Purpose and research questions ... 4
1.4 Project delimitations and study environment ... 4
2 Literature review ... 6
2.1 Supply chain management ... 6
2.2 The strategic, tactical, and operational supply chain hierarchy ... 9
2.2.1 Strategic planning and decision-making ... 10
2.2.2 Tactical planning and decision-making ... 10
2.2.3 Operational planning and decision-making ... 11
2.3 Simulation ... 11
2.3.1 Simulation in the planning of agri-food supply chains ... 13
2.3.2 Uniqueness of ASCs in terms of simulation ... 14
2.4 Innovation theory and diffusion of technology ... 14
2.5 Theoretical framework ... 16
3 Methodology ... 19
3.1 Research approach ... 19
3.2 Research design ... 20
3.3 Data collection ... 21
3.4 Sampling ... 22
3.5 Trustworthiness ... 22
3.6 Methodological limitations and critique ... 23
3.7 Ethical considerations ... 23
4 Empirical data and experiments ... 25
4.1 ASC and market dynamics ... 25
4.2 The supply chain at the host organisation ... 26
4.3 Controlling and planning the ASC ... 29
4.4 The ASC in numbers ... 32
4.5 Differentiating planning levels in the ASC ... 34
4.6 Quantitative modelling of the ASC ... 36
4.6.1 The generation of the base model ... 37
4.6.2 Exit strategies ... 41
4.7 Simulation experiments ... 43
4.7.1 Validation, verification, and credibility of the base model ... 43
4.3.2 Experiment 1 ... 45
4.3.3 Experiment 2 ... 47
4.3.4 Experiment 3 ... 48
5 Analysis ... 50
5.1 What characterises the planning process(es) of the ASC? ... 50
5.2 How are ASC-planning levels differentiated?... 51
5.3 Which ASC-planning process(es) and associated activities are applicable to discrete event simulation? ... 53
6 Discussion ... 55
6.1 Results ... 55
6.2 Sustainability ... 57
6.3 Methods ... 57
7 Conclusion ... 59
7.1 Main conclusions ... 59
7.2 Contributions ... 60
7.3 Future research ... 60
Reference list ... 61
Appendices ... 64
Appendix A ... 64
List of Figures
Figure 1 – An example of a supply chain network (Stadler, Kilger and Meyr, 2015, p. 4). ... 7
Figure 2 – The house of supply chain management (Stadler, Kilger and Meyr, 2015, p. 6) ... 8
Figure 3 – S-curves of the adoption of new payment systems (Wonglimpiyarat, p. 8). ... 15
Figure 4 – The traditional planning-level differentiation (Ahumada and Villalobos, 2009; Fleischman, Meyr, and Wagner, 2005; Misni and Lee, 2017; Schmidt and Wilhelm, 2000; Stadler, 2008; Tordecilla et al., 2020). ... 16
Figure 5 – Conceptual model of the system with a geographical demarcation. ... 27
Figure 6 – Conceptual model of the material flow from in-land storage to port-site storage. . 28
Figure 7 – Conceptual model of the material flow from in-land storage and counties to port- sites onto customers. ... 29
Figure 8 – Outcome of thematic analysis. A list of the ASC’s characteristics. ... 31
Figure 9 – Weekly distribution of purchased grain in tonnes. ... 32
Figure 10 – Summary of aspects that differentiate planning levels in the ASC. ... 36
Figure 11 – The base model. ... 39
Figure 12 – Weighting of flows from counties (Ls) to HPs. ... 40
List of Tables
Table 1 – The qualitative and quantitative methods used for data collection. ... 22
Table 2 – The qualitative and quantitative methods used for data analysis. ... 22
Table 3 – Held interviews. ... 22
Table 4 – Weekly distribution of purchased grain. ... 25
Table 5 – The distribution of material sourcing from counties to port-sites. ... 33
Table 6 – Distribution of received grain at each HP ... 33
Table 7 – Weighting of flows in between HPs. ... 41
Table 8 – An example of replayed rows from the loaded Excel-file. ... 43
Table 9 – Outcome of base model simulation in relation to actual real-world outcome. ... 44
Table 10 – Outcome of scenario 1 in relation to actual real-world outcome. ... 46
Table 11 – Outcome of scenario 1 in relation to base model outcome. ... 47
Table 12 – Outcome of scenario 2 in relation to actual real-world outcome. ... 47
Table 13 – Outcome of scenario 2 in relation to base model outcome. ... 48
Table 14 – Outcome of scenario 3 in relation to actual real-world outcome. ... 49
Table 15 – Outcome of scenario 3 in relation to base model outcome. ... 49
1 Introduction
This chapter will introduce the underlying research problem. The disposition is as follows:
First comes a background, followed by a problem definition and purpose motivation. Then, two research questions are posed, which intrinsically guide the project. Similarly, the following delimitations passage prevents it to surpass the research scope. Finally, the host organisation to the case study is introduced.
1.1 Background
Industrialisation brought more efficient and greatly streamlined means of production that shifted firms into industries and self-service manufacturing into production lines. Frederick W.
Taylor’s Taylorism or Scientific Management then theorised this and with it came centralisation and specialisation of work, which further improved economic efficiency and labour productivity. Taylor was a pioneer in the area of management research and has played an important role in modern operations or management research. It is safe to say that Taylorism can be considered a building block – although the degree to which is arguable – to more modern production improvement philosophies such as Toyota Production System (TPS) or Lean, Total Quality Management (TQM), and Six Sigma.
The topic of mismatching supply and demand, e.g. overstocking, stockout, and delivery delays, is popular to research in operations management (Wu et al., 2016). Problems of mis- matching stem from inherent factors of supply chains such as complexity, uncertainty, and vulnerability. Supply chains are becoming larger and integrated with more actors that are serving more customers across more borders, making them more complex, costly, uncertain, and vulnerable (Ibid). This challenges an efficient supply chain management (SCM). To deal with this, Wu et al (2016, p. 395) states “supply chains must become a lot smarter”. Stadler and Kilger (2015) emphasise that “correct reactions” to exceptions and variability will make the biggest impact on business performance, and state that planning itself reduces the number of exceptional situations.
There are often several production sites and other organisational units that make up a supply
chain, wherein thousands of decisions are being made every minute (Fleischmann, Meyr and
Wagner, 2015). Planning and controlling the like – the network of flows of goods and the vast
amount of information that is being generated by the activity that is SCM (Slack, Chambers
and Johnston, 2010) – far exceeds any manager’s “manual capacity” (Stadler, Kilger and Meyr,
2015).
Enterprise resource planning (ERP) systems are “a complete enterprise wide business solution…[comprising] software support modules such as: marketing and sales, field service, product design and development, production and inventory control, procurement, distribution, industrial facilities management, process design and development, manufacturing, quality, human resources, finance and accounting, and information services…” (1984, cited in Slack, Chambers and Johnston, 2010, p. 409). During the 1980s and 1990s, these systems became a standard IT-solution for almost all organisations (Stadler, Kilger and Meyr, 2015). However, according to Koch and Wailgum (2007, cited in Slack, Chambers and Johnston, 2010, p. 409) the planning part of the acronym is a “throwaway term”. What ERP systems do is both allow integration of all decisions and databases in “all departments and functions across a company”
(Ibid), and reflect consequences of decisions inter-departmentally and across functions in real time. Moreover, ERP systems can help in the improvement of internal processes, but the big efficiency-wins are found outside the company’s system limits, i.e. in the supply chain, through the integration of functions and processes between directly and indirectly linked firms that together make up the supply chain (Beamon, 1998; Croxton, Garcia-Dastugue, Lambert, and Roger, 2001; Fleischmann, Meyr and Wagner, 2015; Stadler, 2008; Stadler and Kilger, 2015;
Stadler, Kilger, and Meyr, 2015).
Today, with growing populations, globalisation, and environmental concerns, the efficiency of supply chains including agri-food supply chains (ASCs) are very much of relevance (Ahumada and Villalobos, 2009; Amer, Galal and El-Kilany, 2018; Borodin, Hnaien, Labadie and Bourtembourg 2013 and 2015; Galal and El-Kilany, 2016; Khan et al., 2020; Lucas and Chhajed, 2004). The growing population is shown to be the main reason for the increased global demand for perishable food, e.g. vegetables (Ahumada and Villalobos, 2009). Another reason is increased nutritional awareness (Ibid). Furthermore, ASCs have been and are still undergoing drastic transformation. The industry as a whole is globalising and small independent actors are being overrun by tightly aligned, highly vertically-integrated global firms and cooperatives.
This is resulting in increased need for planning, and an increased number of decisions concerning the ASC being made by the producers (Ibid).
With the consolidation and globalisation of ASCs, the agricultural logistics industry as a
whole is becoming progressively complex and competing producers need sophisticated
planning tools to remain competitive (Ibid). Furthermore, (Ibid) carries a discussion around the
above-mentioned consolidation of ASCs and states that ASCs are gradually resembling supply
chains in the manufacturing sector. Consequently, (Ibid) suggests, improvement programs and
concepts of supply chain planning and coordination should be applicable for ASCs. Such a
discussion is echoed in the thesis of Nyström and Hellberg (2020) who specifically emphasise other industries’ “tremendous success” with simulation-tools in their investigation of regulatory compliance versus continuous improvements in the medical device industry. Their analysis shows obvious results for the suitability of simulation in that industry’s complex system structures.
The realm of pure simulation science advocates a plethora of advantages that simulation undoubtedly will bring around. Essentially, one is able to visualise and analyse the turnout of a system (re)configuration (e.g. the performance of a supply chain redesign) or “something” that can be simulated, before putting effort into making that something a reality. That is as long as the modelled system sufficiently represents or imitates accurately the base system (Banks, 2000). To do that, the system needs imitate-able operating characteristics, i.e. information about what is happening inside the system or so-called system state variables (Ibid).
1.2 Problem definition
The different levels of decision-making involved in the planning of supply chains can be hierarchically classified as strategic, tactical, or operational. Strategic planning decisions are long-term, usually with a several year horizon. Tactical ones are medium to short-term, usually with a monthly to weekly planning horizon. Operational ones are short-term and can concern day-to-day operations. Ahumada and Villalobos (2009) cites the work of Simchi-Levi (2003) and Chopra and Meindl’s (2003) and say that the different classifications also depend on the effects a decision has on the overall supply chain.
Planning supply chains, the activity, includes the following functional areas: production planning, inventory control, and physical distribution. Ahumada and Villalobos (2009) identifies production, harvest, storage, and distribution as specific functional areas to ASC planning. Further on, (Ibid) lists several decisions that are associated to each function. For instance, production includes decisions concerning land-allocation and the timing of sowing;
harvest includes timing for collecting crops and determining what machinery to do it with;
storage includes decisions concerning what volumes to be stored and sold, respectively, and how to position inventory levels along the supply chain; and, finally, distribution includes deciding the mode of transport and distribution route.
The problem is that ASC-specific functional area-associated decisions (e.g. redesign of
supply chain) have not been classified according to the planning-level differentiation, that is,
whether certain decisions for an ASC are rather strategic, tactical, or operational. It is unclear
in the case of an ASC what effect(s) a decision shall have (e.g. total) to make it classify as
either strategic, tactical, or operational. It is also unclear which level(s) of planning and which ASC-planning activities are applicable to simulation. Furthermore, it is unclear whether ASCs are missing variables or, alternatively, encompasses certain variables that prohibits them from being simulated, or what other reasons that exists for keeping ASCs un-simulated.
1.3 Purpose and research questions
The purpose of the study is to investigate the hierarchical differentiation of planning processes in an ASC, and to determine which of them that are applicable to discrete-event simulation.
Three research questions are posed in order to guide the research project:
1. What characterises the planning process(es) of the ASC?
This question (Q1) aims at characterising the essence of the host organisation’s ASC. The factors may be unique for the studied ASC, or common for a supply chain.
2. How are ASC-planning levels differentiated?
This question (Q2) aims at understanding the hierarchical differentiation of ASC-specific levels of planning at the host organisation. It, obviously, presumes that there are hierarchical levels that differentiate the planning process within it. The basis on which the differentiation originates may include factors such as planning horizons and/or effect(s) of the execution of a decision.
3. Which ASC-planning process(es) and associated activities are applicable to discrete- event simulation?
This question (Q3) intends to identify ASC-processes that are applicable to discrete-event simulation by considering model verification, validation, and credibility (Sargent, 2010).
“Model verification is often defined as ensuring that the computer program of the computerized model and its implementation are correct” (Ibid, p. 166). “Model validation is usually defined to mean substantiation that a computerized model within its domain of applicability possesses a satisfactory range of accuracy consistent with the intended application of the model” (Ibid).
“Model credibility is concerned with developing in (potential) users the confidence they require in order to use a model and in the information derived from that model” (Ibid).
1.4 Project delimitations and study environment
The investigated supply chain only facilitates the distribution of Swedish grain in and out of in-
land and port-site storage facilities, including flows from farmers to the host organisation, intra-
organisational (re)distribution, and from the host organisation to its customer(s)). To add to
that, there are several types of grain, but these are aggregated into a singular type of material
flow. The study is limited to a time period of a little more than four months which limits the number of possible case studies to one. Furthermore, the investigation mainly deduce theories of operations management, including supply chain management (SCM) and advanced planning, simulation, and diffusion of innovation.
The study was conducted in collaboration with a Swedish agriculture organisation. The
organisation will not be named in this report due to reasons of confidentiality but instead,
henceforth, referred to as the host organisation. Besides agriculture, the organisation has
operations in machinery, bioenergy, and production of food. It facilitates a complex supply
chain with great amounts of detailed data. The investigated supply chain is delimited from the
overall supply chain, in the sense that it focuses only on the supply chain between farmers, in-
land storages, port-site storages, and the end consumer (see more in chapter 4.1). To add to that,
this supply chain’s performance indexes and general numerical data, is delimited to the period
of harvest (see more in chapter 4.1.1). The author and the host organisation have established
non-disclosure agreements (NDAs) regarding certain numerical and non-numerical data. Data
presented in this report will of course not be covered by NDAs.
2 Literature review
This chapter will introduce extant literature that covers the research area wherein the problem is embedded, including the concept of simulation in itself as well as the application of it in ASCs. Furthermore, the chapter will cover concepts like planning, supply chain management (SCM), simulation, and innovation including diffusion of technology. Finally, in the chapter’s last passage comes a summarized version of the theory that relates to the background, purpose, and research questions, ultimately comprising the project’s theoretical framework.
2.1 Supply chain management
According to Christopher (2005, p.17, cited in Stadler, Kilger and Meyr, 2015, p. 3) a supply chain is a “network of organizations that are involved, through upstream and downstream linkages, in the different processes and activities that produce value in the form of products and services in the hands of the ultimate consumer”. The up- and down-stream linkages comprise material, information, and financial flows. The organisations between which these flows are linked are usually legally separated (Stadler, 2008). Supply chain management (SCM), then, is the “task of integrating organizational units along a supply chain and coordinating material, information and financial flows in order to fulfil (ultimate) customer demands with the aim of improving the competitiveness of a supply chain as a whole” (Stadler, Kilger and Meyr, 2015, p.5).
Essentially, a supply chain is either intra-organisational meaning it takes place within a single firm, or it is inter-organisational meaning it comprises several firms usually on a global scale. It is, however, possible to treat a big, single global firm as a complex international supply chain (Ibid). Take IKEA for example, which consists of several different functional units: owns forests that produce the raw material; has its own factories producing parts, components and end products; and then its transport and logistics services organisation that secures transport to the end user. (Ibid, p. 4) states that the coordination and management of functional units and flows in the intra-organisational supply chain is less cumbersome but still a “formidable task”
and “mandatory – a prerequisite being no matter of course in today’s firms”. It is nevertheless
important in inter-organisational supply chains. It is because of the single top management level
in intra-organisational supply chains that makes decision-making and, in turn, managing easier
than in the inter-organisational supply chain, which consists of at least two firms, i.e. at least
two top management levels (Ibid). Figure 1 shows a supply chain with several divergent and
convergent flows, rather than singular flows, which is a typical setup of any supply chain (Ibid).
(Ibid) states that this is because many different customer orders are handled in parallel in today’s supply chains.
Figure 1 – An example of a supply chain network (Stadler, Kilger and Meyr, 2015, p. 4).
The main objective in SCM governing all endeavours and aspirations is increased competitiveness, which is realised through competing on customer service in a timely manner and as efficiently and profitably as possible (Ivanov, 2010; Maina and Perera, 2020; Misni and Lee, 2017; Stadler, 2008). Competing on customer service, a firm either satisfies a market standard service level at minimum cost, or provides a superior service level (Stadler, 2008).
This is also making SCM transition from an operational to a strategic role in the overall management of a firm (Maina and Perera, 2020).
The roof – competitiveness and customer service – of the house of SCM (see figure 2) rests on the two pillars integration and coordination (Stadler, Kilger and Meyr, 2015). Integration is meant to be of other organisational units along the supply chain and coordination refers to the flows along it. The first building block of integration is choosing suitable collaboration partners.
This is an obvious prerequisite to forming a supply chain and it relates greatly to strategy.
Collaboration partners should be aligned but even better is if they reinforce and “fit” each other in the supply chain, not only in terms of strategic intent but also in core competency (Porter, 2008 cited in Stadler, 2008). Secondly, to effectively contribute to the network of organisations, constituting the supply chain, firms must actually practice inter-organisational collaboration.
The third building block calls for a leadership that aligns involved partners’ strategies.
Figure 2 – The house of supply chain management (Stadler, Kilger and Meyr, 2015, p. 6)
The coordination pillar consists of the three building blocks: use of information and communication technology (ICT); process orientation; and advanced planning (Ibid). The utilisation of ICT will significantly simplify and improve processes and information flows.
Processes orientation aims at incorporating a process’ improvement or redesign and making it a standard. It is not only about a single process improvement but rather a permeated way of working, a philosophy for improvement. Finally, the challenge of coordinating plans, for instance of product or machine availability and distribution routes among involved partners and production sites, calls for advanced planning.
Customer service is at the helm of SCM. It is a notion made up of three stages of transaction:
pre-, during, and post-transaction (Stadler, 2008). A firm is able to perform different activities to influence the elements of these stages for the customer to experience good value from their purchase. Pre-purchase elements concern access to information and links to involved organisations. Routine orders, for instance, are not necessarily dependent on the suppliers’
ability to meet special requirements from the customer to secure the order, but it can be an
important element in bigger investments. The “during transaction” elements are those which
contribute to the very fulfilment of a transaction. Such activities can include delivery status and an instructional walkthrough of a machine. Finally, post transaction elements, that add value to an order overall in the eyes of the customer in terms of customer service, concern services provided after order fulfilment. This can include dealing with complaints and product exchange or maintenance.
Planning comprise preparatory activities or measures for taking decisions that will lead to the execution or carrying out of an activity. Along supply chains there are thousands of decisions being made and coordinated every minute, ranging from strategic to operational ones, being more or less important depending on the effect on the overall supply chain (Ahumada and Villalobos, 2009; Fleischmann, Meyr, and Wagner, 2005). The greater the effect, the more important is the decision, and the more and better planning should precede that decision (Fleischmann, Meyr, and Wagner, 2005). Increased complexity of a system will reduce information accuracy and, in turn, increase uncertainty (Ibid). Serious, rigorous planning will have to precede the decisions to be made, in order to ascertain good performance. Maina and Perera (2020, p. 33) state: “due to the rapid changes in consumer demand, technological swift and globalization, Supply Chains (SCs) have become complex entities to manage and in order to gain resilience in the SCs powerful technologies like simulation have been used which also act as key decision-making tool, as they mimic real life situations”.
2.2 The strategic, tactical, and operational supply chain hierarchy
Schmidt and Wilhelm (2000) discuss hierarchical decision-making levels in international
supply chains. They state that the top, strategic level set the environment in which the other
levels perform, i.e. the tactical and operational levels. They state that the strategic level
constraints the tactical level, which, in turn, constraints the operational level. For example, the
strategic level decides on a storage capacity, tactical positions that storage level, and operational
then gets to put that into use (Ibid). Misni and Lee (2017) reviews literature covering this
hierarchical division of decisions in reverse logistics networks. They point out that decisions
regarding product design (made at the strategic level) are generally under-appreciated, i.e. have
greater impact (or prescribe stronger constraints (Schmidt and Wilhelm (2000)) than is
generally believed. The levels interact in many ways and they must be integrated properly in
order to provide seamless customer service. Further, Misni and Lee (2017) state that strategic
decisions can (and should (Schmidt and Wilhelm, 2000)) be integrated with tactical decisions
since the operational decisions will come as consequences of these.
2.2.1 Strategic planning and decision-making
The strategic level assumes a long-term planning horizon, usually two up to five years (Fleischman, Meyr, and Wagner, 2005; Misni and Lee, 2017; Schmidt and Wilhelm, 2000;
Stadler, 2008). Strategic decisions creates the prerequisites for a supply chain and typically concern its design or structure (Fleischmann, Meyr, and Wagner, 2005). Due to the long planning horizon there is usually more uncertainty involved on this level, for example with political environments and exchange rates (Schmidt and Wilhelm, 2000). The decisions are noticeable over several years and include decisions regarding the location of facilities as well as the production technique, technology, and capacity that should be employed at each facility (Schmidt and Wilhelm, 2000). Tordecilla et al (2020, p. 1) states that strategic decisions include the determination of “the number and location of production facilities, the amount of capacity at each facility, the assignment of each market region to one or more locations, and supplier selection for sub-assemblies, components and materials”, which is consistent with Schmidt and Wilhelm (2000), Fleishmann, Meyr, and Wagner (2005), and Stadler (2008). So, strategic decisions are overarching and set the course for the overall strategy of the firm as well as for the supply chain as a whole. Their effects are total and act as the basis for lower-level planning and, in turn, lower-level decision-making.
2.2.2 Tactical planning and decision-making
The tactical level or mid-term planning (Fleischman, Meyr and Wagner, 2005) prescribes a 6–
24-month horizon (Misni and Lee, 2017; Schmidt and Wilhelm, 2000; Stadler, 2008).
Fleischman, Meyr and Wagner (2005) states that the notion of the tactical-level is contradictory in meaning in “the literature” and thus term both this and the bottom planning level operational.
They point out that some aspects of mid-term planning can fall within the scope of the strategic level, such as the setting of an “outline of the regular operations, in particular rough quantities and times for the flows and resources in the given supply chain” (Ibid, p. 82). The length of the planning horizon and the importance or effect (Ahumada and Villalobos, 2009) of a decision is what determines the classification (Ibid).
Schmidt and Wilhelm (2000) still stick to calling this level tactical and state that it, in a
broad sense, considers both production and transportation (see also Ahumada and Villalobos,
2009). It includes policies (such as going for either a finished-product inventory or assemble-
to-order system), inventory or storage levelling, and product variety. This is consistent with
Misni and Lee (2017, p. 92) who state that this level deals with “production planning and
inventory management as well as procurement”. Furthermore, tactical level planning has the
opportunity to refine decisions made under uncertainty on the strategic level (e.g. concerning
production plans), due to the shorter planning horizon. Still, the planning horizon on this level is relatively long, which induce some uncertainty in the decision-making. So, the tactical level concretises the strategic level’s abstraction of the firm’s overall strategy in a contemporary manner and, if needed, adjust for new developments that can impact it.
2.2.3 Operational planning and decision-making
Operational decisions are made on a day-to-day basis and, as mentioned, are constrained by or come as consequence of higher-level decisions (Fleischman, Meyr and Wagner ,2005; Misni and Lee, 2017; Schmidt and Wilhelm, 2000; Stadler, 2008). However, Fleischman, Meyr and Wagner (2005) choose to term this level short-term and state that its planning horizon is between a couple of days and up to three months. They state that the planning on this level require the highest level of detail and accuracy so as to allow all activities to be executed and controlled at any given moment. They further state that this level has direct impact on process and supply chain performance with effects including lead-times, delays, and customer service.
Similar claims are made by Schmidt and Wilhelm (2000) and Misni and Lee (2017). The former work points out that this level can have significant impact on customer service and the ability to achieve a unified logistics system since it schedules operations. They state that without proper scheduling, the final products will not be delivered to customers in time. The latter states that this level deals with, but is not limited to, vehicle planning, scheduling, and routing, as well as dynamic pricing of products, and order batching and picking.
The operational level operationalises decisions made on higher levels. It coordinates the strategic and tactical levels to allow for the execution of detailed and accurate activities of the concretised abstraction of the overall strategy “…to assure in-time delivery of final products to customers” (Schmidt and Wilhelm, 2000, p. 27).
2.3 Simulation
“Simulation is the imitation of the operation of a real-world process or system over time”
(Banks, 2000, p. 9), and there are several different types of simulation, for example spreadsheet
simulation, system dynamics, discrete-event simulation (DES), business games, and many,
many others (Kleijnen, 2005). Simulation is an indispensable problem-solving methodology
when it comes to describing and analysing the behaviour of any imitate-able system (Banks,
2000) and for conducting bottleneck analyses (Velumani and Tang, 2017). It allows the analyst
to ask What If questions (Maina and Perera, 2020) about a real-world system and through the
simulation be presented accurate real-world turn outs. Banks (2000) states that DES models
contrasts mathematical, descriptive, statistical, and input-output models by representing the interactions of a system’s components.
To the concept of simulation there are several other, underlying concepts including “system and model, events, system state variables, entities and attributes, list processing, and activities and delays” (Banks, 2000, p. 10). These will be explained below in a similar fashion as found in (Banks, 2000).
A system to be simulated is represented by a model. Depending on how the boundaries and limits of the model are set, the model’s accuracy of imitating the real-world system is affected.
Events are occurrences that changes the state of a system, e.g. the beginning of an operation or the arrival of raw material for an operation. If a given operation is simulated, the beginning of that operation is an internal or endogenous event while the arrival of raw material to that operation is an external or exogenous event since that occurrence is outside of the simulation (Ibid). External events must nevertheless be considered in the simulation for they do have an impact on the system.
System state variables represent all information required for the model to be able to explain what happens within a system. What is required depends on the purpose of the analysis, so, “the system state variables in one case may not be the same in another case even though the physical system is the same” (Banks, 2000, p. 11). System state variables remain constant over intervals of time in a DES model, and change only at certain events, while they are defined by certain equations in continuous models.
Entities may be dynamic or static and represent an object that needs explicit definition with attributes (Ibid). Attributes are local values and can represent, for example, an entity’s time of arrival or colour. Static entities serves other entities, which is what makes it static. Dynamic entities flows through the system. When a static entity serves a dynamic entity, it is called a resource from which the dynamic entity can request units. A resource can serve more than one dynamic entity simultaneously. However, if the resource denies the dynamic entity’s request for units, the dynamic entity is either put in queue, goes on to another resource, or is ejected from the system. Conversely, if the resource permits the dynamic entity’s request it remains for a while, putting the resource into a busy state, and then moves on. Other resource states include idle, failed, blocked, or starved (Ibid).
Queues are represented by lists which, in turn, follows a multitude of different processing
or scheduling policies, e.g. FIFO (fist in first out), LIFO (last in first out), or SPT (shortest
processing time). The processing order of a list can also be completely random or according to
certain values of attributes (e.g. red first then blue).
Banks (2000, p. 12) states that “an activity is a duration of time”. This duration is known in advance and, therefore, when the activity begins its end can be scheduled. Durations can be constant, random, results from an equation, inputs from a file to name a few. Similarly, a delay is a duration of time, but its length is unknown. It is an “indefinite” duration of time caused by a certain combination of system conditions (Ibid). When a resource denies the request for units of an entity that entity is put in queue, for which the duration is initially unknown. Beginnings and ends of both activities and delays are occurrences which change the state of the system, and are therefore considered events.
2.3.1 Simulation in the planning of agri-food supply chains
At the time of writing, combining “simulation” and “supply chain” in search on Google Scholar generates some 459 000 results. Changing “supply chain” to “agri-food supply chain” combined with simulation generates instead a mere 1 460, of which 1 160 are from 2014 and forwards.
This indicates that research concerned with simulating ASCs is relatively scarce, and that the majority of research concerned with this domain has been conducted during the last decade.
This is consistent in the exhaustive literature review of Ahumada and Villalobos (2009) and the work of Lucas and Chhajed (2004), wherein it is stated that the industry lacks simulation and “sophisticated” planning tools, respectively. In later reviews (Handayati, Simatupang and Perdana, 2015; Lou et al., 2018), which indeed specifically focuses on coordination issues in ASCs and on ASC management, respectively, only three works are listed within which simulation has been used as basis for analysis.
In one of the newer reviews on simulation in ASCs (Utomo et al., 2018, p. 803), it is stated that the characteristics on which the research in the area of simulation in ASCs is mainly concerned are “single echelon supply chains; cases from high- and middle-income countries;
unprocessed food products, empirical (as opposed to hypothetical) data; decision-making related to production planning and investment; and the use of black box validation.” Examples of applications of simulation in ASCs include, but are not limited to, Borodin et al (2015) wherein (discrete-event) simulation was used for redesign of a crop production supply chain.
They conclude that by “pooling” resources and “federating” efforts between cooperative and
individual growers, both parties will benefit. Galal and El-Kilany (2016), followed up in the
work of Amer, Galal and El-Kilany (2018), investigates the sustainability in and of ASCs using
simulation as tool for analysis. These works show that “reduced” order quantities can
significantly reduce costs and emission in the ASC, and that “adequate” order quantity increases
several ASC performance measures, respectively. Cruz, Pires-Ribeiro, and Barbosa-Póvoa
(2019) develops a mixed-integer linear programming model (which focuses on the strategic-
tactical decisions of capacity definition, selection of processing technologies, and the establishment of product flows to achieve expected net present value maximisation), as a tool for analysis for planning and decision-support to an ASC.
2.3.2 Uniqueness of ASCs in terms of simulation
There are several aspects that make the ASC stand out among supply chains of other industries.
The fundamental difference is the unreliable quality of the food throughout the entire chain and the fact that it is perishable (Ahumada and Villalobos, 2009; Paam et al., 2016; Routroy and Behera, 2017). Routroy and Behera (2017) state that the quality of food can be negatively impacted in terms of respiration, weight loss, texture, decay, storage temperature, external and internal browning, colouration, physical damage, taste, and aroma and nutritional deficiencies.
The perishing of the food that constitutes the material flowing through the supply chain occurs over time during handling and processing. This indicates that a more direct concern is time and suggestively the actual fundamental concern in an ASC, for it is over time that food will become perished.
This increases the importance of proper integration and coordination between firms and or functional units along the supply chain. Other aspects that differentiate an ASC from supply chains in other industries include, but are not limited to, price and demand variability and weather dependency (Ahumada and Villalobos, 2009; Paam et al., 2016; Routroy and Behera, 2017). In addition, the timing and length of a harvesting period is also uncertain, and so is the harvestable tonnage, i.e. the supply of material. This further adds to the model complexity and, in turn, its verification and validation.
2.4 Innovation theory and diffusion of technology
Innovation begins with an idea, or rather during the generation of a new idea (von Hippel,
1988). There are several sources of innovation, including users, firms’ research and
development (R&D) departments, and universities. Users are interested in creating something
useful for themselves, rather than making a profit out of it. Firms with R&D departments
account for the majority of research expenses in a developed country, and their intensity
positively correlate to its sales growth, new product sales, and profitability. Universities also
invest a lot in R&D and usually get to keep the rights to the results. Moreover, it is common for
R&D intensive firms to locate themselves in relatively close proximity to universities and other
firms which are R&D intensive, and which focus on similar technology (e.g. Silicon Valley)
(Schilling, 2017). Furthermore, there are different types of innovation, including product and
process innovations, radical and incremental ones, and competence enhancing or destroying
innovation. However, these are almost always relative, i.e. time, observer, and context dependent (Ibid). For example, small firms who manually plan their distribution routes every morning would most likely regard the logistics system of UPS or FedEx as a radical product innovation if it somehow “became” theirs. UPS would, however, most likely regard an upgrade or update to their own system as incremental, perhaps not even an innovation.
Innovation theory further tells us that there are five categories to help explain the diffusion of a technology in groups of adopters. These are in order of adoption termed: (1) innovators, who eventually constitute 2.5% of the total adopters of a technology; followed by (2) early adopters, who constitute 13.5%; then there is the (3) early and (4) late majority, constituting 34% each; and finally the (5) laggards with the remaining 16% (Shilling, 2017). Diffusion research centres on conditions of a culture or in a group that increases or decreases the likelihood of that culture or group to adopt a technology. However, it is not necessarily the adoption of a technology that is in focus. It can also be an idea, product, or practice (Ibid) such as simulation. Laggards are conservative and isolated compared to the other categories, and mainly interact with other laggards.
At the time a laggard has adopted a technology or practice, it can well be rendered obsolete by innovators, perhaps even by the rest of the world (Ibid). “Both the rate of a technology’s performance improvement and the rate at which the technology is adopted in the market place repeatedly have been shown to conform to an s-shaped curve” (Ibid, p. 50). Innovators adopts a new technology and sets the s-curve in motion and when the laggards adopts the technology the s-curve is “flattening”. At this time, that is, when the laggard finally adopts the technology, the innovators may well be on their way adopting yet another, new technology which sets in motion another “S” (see figure 3 which demonstrates s-curves in a technological change process of payment systems, suggesting a trend towards a cashless society (Wonglimpiyarat, p. 8)).
Figure 3 – S-curves of the adoption of new payment systems (Wonglimpiyarat, p. 8).
2.5 Theoretical framework
The industries that traditionally have been more researched, and where the application of simulation is dominated, are the semiconductor and automotive industries (Semini, Fauske and Strandhagen, 2006). These industries fall under the domain of manufacturing industries and their supply chains are typically categorizable as inter-organisational ones. Moreover, these supply chains utilise the traditional planning level differentiation and have a clear distinction between the strategic, tactical, and operational level (Ibid) (see figure 4 below).
The strategic level deals with the firm’s future market position and establishes the prerequisites for the supply chain. It designs the network of nodes in the chain and integrates and coordinates all actors across it. Decisions on this level are noticeable over several years and they come with the most uncertainty of all levels. The tactical level has the opportunity to refine decisions and plans made on the strategic level. This is because the tactical level deals with shorter planning horizons.
Schmidt and Wilhelm (2000, p. 16) states that what mainly can be compensated for is societal change – “political and financial processes” – the associated uncertainty of which is reduced as the horizon is shortened. The operational level has a narrower focus and takes decisions based on much more detailed and accurate data. The decisions are constrained by or come as consequence of higher-level decisions and are made on a day-to-day basis. For example, in a supply chain, the strategic level decides on a system or network infrastructure for the coming ten years – the location of and capacity employed at each facility. The tactical level forecasts demand and prescribes
Figure 4 – The traditional planning-level differentiation (Ahumada and Villalobos, 2009; Fleischman, Meyr, and Wagner, 2005; Misni and Lee, 2017; Schmidt and Wilhelm, 2000; Stadler, 2008; Tordecilla et al., 2020).
production and distribution policies for the coming year. Finally, the operational level assures in- time delivery of finished products to consumers during the coming weeks by basing their decisions – for example, from where, to where, how much, and by what mode of transport – on accurate forecasts that have the highest level of detail.
The manufacturing industry domain is also aware of where, or to what, DES is applicable.
Specific strategic aspects of supply chain networks, including those of distribution, facility location, and capacity expansion has received a lot of focus in prior research (Schmidt and Wilhelm, 2000).
According to Semini, Fauske and Strandhagen (2006), the majority of applications comprise (re)designs of production systems, followed by (re)evaluations of production rules and policies including lot sizes and WIP levels. Ibid shows that there is a multitude of possible applications but that these are more or less applicable depending on the problem context. Ibid (p. 1949) states that multi-echelon supply chains reduce the applicability of “quantitative modelling based on operations research and system analysis” because the multi-echelon context entails “new requirements, including alignment of network strategies and interests, mutual trust and openness among actors, high intensity information sharing, collaborative planning decisions and shared IT tools”. What is typically emphasised regarding the applicability of quantitative modelling approaches (such as DES) is that the problem context “must be of a technical nature and characterized by high consensus between stakeholders” (Ibid). Conversely, when a problem entails considerable amounts of human and organisational elements, quantitative modelling appear far less applicable, which is visible in the lack of applications in such contexts (Ibid). We also know from Schmidt and Wilhelm (2000, pp. 4-5) that “[a]ny model applicable to the strategic level must provide manufacturing capacity to satisfy forecast demand for all products and observe precedence relationships among assembly tasks.” Ibid does not explicitly state what a model on the tactical level shall include. However, they do state (pp. 17-18) that four basic issues must be modelled at the tactical level: “(1) what is the optimal assignment of operations to plants, taking into account plant capacities that result from strategic decisions and actual product demand that develops, (2) what is the optimal assembly policy for each product, (3) what is the optimal inventory level for each subassembly, taking into account the coordination of shared inventories and interdependent demand, and (4) what is the optimal lot size for assemble-to-order components.” Finally, from the same authors (pp. 5-22) we can see that
“[a] model for the operational level should be invoked daily to schedule operations relative to current information about jobs in process and the status of each”.
An ASC is usually an inter-organisational supply chain, wherein several actors or organisations,
both domestic and foreign ones, have been integrated (Ahumada and Villalobos, 2009), which
implies that multiple top management levels have to be coordinated. The planning of inter-
organisational supply chains is complex but have been studied thoroughly in multiple industries.
The planning in the agri-food logistics industry and that of its supply chains is complex, too, and
infers some unique properties or variables in terms of simulation. Still so far nothing in the literature
review has evidenced variables that should prohibit ASCs from being simulated. Furthermore, the
investigation of an ASC’s planning level differentiation should be able to draw inspiration from
other, more researched industries. Similarly, the investigation of an ASC’s simulation-ability should
be able to draw inspiration from other industries and extant similar works in the domain.
3 Methodology
This chapter explains the conduction of the project and applied methods. It will account for the research approach and design, sampling frame and sampling methods, as well as data collection and analysis for both qualitative and quantitative data. Then, a discussion is held regarding the trustworthiness. The last parts concern limitations and ethical considerations.
3.1 Research approach
Ontology and epistemology are different branches of philosophy concerned with the nature of being/reality, and the with theory of knowledge, respectively. Ontology answers to the central question “What is reality” (Brantnell, 2020b, p. 3), e.g. is there a supply and/or demand?
Epistemology, then, concerns how knowledge of reality can be generated (Bell, Bryman, and Harley, 2019), e.g. how can supply and demand be studied? Furthermore, ontology comprises two different viewpoints (Brantnell, 2020b) or extremes (Hellberg and Nyström, 2020), namely objectivism and constructionism. An objectivist reality is external and independent of the observer (Bell, Bryman, and Harley, 2019). This means that there is one unified reality in which all live and act. Conversely, a constructionist reality is not external but dependent on the observer (Ibid), “it is made up of our perceptions, actions and so on” (Brantnell, 2020b, p. 7).
Epistemology comprises several approaches to the study of reality, but the most common are positivism and interpretivism (Bell, Bryman, and Harley, 2019). Moreover, epistemological approaches are informed by ontological viewpoints. Positivism is connected to objectivism while interpretivism is connected to constructivism (Ibid). Human behaviour is interesting for both approaches, but positivism tries to explain while interpretivism tries to understand (Ibid).
This research project views reality from an objectivist point of view and utilises a positivist approach to study this reality.
According to Bell, Bryman, and Harley (2019) three different types of research strategies can provide a general orientation for the research project. Which strategy that is embraced also depends on ontological and epistemological viewpoints (Ibid). These research strategies are deduction, induction, and abduction. Moreover, the strategies explain the relationship between the researcher and their interaction between observations and established theory (Ibid).
Deductive strategies start with theory, from which hypotheses are generated and tested empirically as part of the research project. Such interaction between testing or observation and theory is more common in quantitative research (Ibid). Conversely, induction starts “with a clean slate” (Hellberg and Nyström, 2020, p. 22), i.e. starts with observations instead of theory.
Inductive research tries to, through observations, generate or develop theory out of that being
observed (Bell, Bryman, and Harley, 2019). Abduction, then, combines these strategies. The abductive researcher works iteratively, going back and forth in the interaction between observations and theory (Ibid). Brantnell (2020a, p. 10), in his presentation of scientific theory, states that the typical abductive process starts with a problem and “you have an idea [of how]
to solve it, you talk to people or do surveys, test the idea, revise the idea, talk to more people, revise the idea and so on with the aim of solving the problem.” Abduction is a common research strategy today and can comprise both qualitative and quantitative methods although qualitative methods are more commonly applied (Ibid).
This research project utilises an abductive strategy, approached with both qualitative and quantitative methods.
3.2 Research design
There are many different designs that a research project can take, including experimental, cross- sectional, longitudinal, comparative, and case study design (Bell, Bryman and Harley, 2019).
The case study design is focused, in order to obtain specific and detailed information. Further, it is said that there are three types of case study designs: intrinsic, instrumental, and multiple (Ibid). The latter, intuitively, comprise multiple cases in the same research project while the former two only uses one case. The cases here constitutes some phenomenon, which is studied in a certain setting. It can be an organisation, location, person, or event (Ibid). It is widely discussed whether findings from a case study can be generalisable, however, Flyvbjerg (2006, p. 8) suggests they can be, but “[i]t depends upon the case one is speaking of, and how it is chosen”.
This research project is an instrumental case study and utilises a mixed methodology of qualitative and quantitative methods. It is instrumental because it focuses on the broader phenomena of planning processes and applicability of simulation in ASCs (Crowe et al, 2011).
The case will provide insight to the planning of a Swedish ASC and allow for testing simulating
an ASC. This particular case is chosen because the organisation is sufficiently large in terms of
numbers of integrated organisations in the supply chain to include all relevant actors from
farmer to end consumer (European Commission, 2015), i.e. the studied ASC echoes real-world
ASC-complexity. This, together with the problem description (chapter 1.2), and the study
environment (chapter 1.5) makes good as a thick description of an ASC and strengthens the
transferability of the thesis.
3.3 Data collection
Qualitative research is typically concerned with words and meaning or understanding while quantitative research instead focus on numerical data and analysis (Bell, Bryman and Harley, 2019). The first two research questions in this thesis focuses on understanding and, ultimately, characterising ASCs so as to classify planning processes, activities, and decisions in relation to the traditional planning level differentiation. Since literature in this domain is scarce, an inductive, exploratory, and theory building approach that utilises qualitative methods was deemed necessary. It utilises text and document collection and semi-structured interviews.
More specifically, literature covering the concepts of characterising supply chains and planning hierarchy in different industries was collected for comparison during the theory building of the concepts for ASCs. Moreover, the semi-structured interviews comprised open-ended questions regarding sub-topics to these concepts that were found relevant during the literature review (see appendix A for interview topics).
One of the most common analysis methodologies in qualitative research is thematic analysis or thematization (Ibid). The goal with thematic analysis is to identify themes that explain data and findings. There are several ways to do this, but normally one starts off with transcribing interviews, from which the search can begin. One can look for repetitions, typologies, metaphors, transition of topics, similarities, or differences in how phenomenon are described, what is missing, or theory-related topics (Ibid). Some researchers do not differentiate between codes and themes, but this thesis assumes that code(s) come before/builds into a theme, which is synonymous with a category, and a theme connects to existing concepts. This analysis method is used for all interviews and qualitative data analysis in this thesis.
For research question three, the methodology is rather deductive – testing and developing (Dubois and Gadde, 2002) the theory built by the former two – and quantitative in nature, although it involves the same qualitative data collection methods as for Q1 and Q2. However, the data type differs in this text and document collection. Here it is numerical, representing a list of storage capacity at different sites, volumes of material flow, dates, and addresses (from and to). The material flow is an aggregated volume of all grain types into one product and is measured in tonnes, i.e. there is only one variant flowing between nodes in the model. The dates total at a time-span of more than 100 days and represent the hour at which a delivery was made.
Addresses are on a county level and total at 17 counties, which represent the “from-locations”.
The receiving nodes total at eleven (port) storage facilities. This data is saved in MS Excel files
and loaded into the DES software FACTS (Evoma.se) wherein the quantitative modelling is
conducted (see more in chapter 4).
Table 1 and 2 depicts the qualitative and quantitative methods used for data collection and for data analysis, respectively.
Table 1 – The qualitative and quantitative methods used for data collection.
Collection of data Qualitative Quantitative Methods Text and document
collection; Semi-structured interviews.
Text and document
collection; Semi-structured interviews.
Table 2 – The qualitative and quantitative methods used for data analysis.
Analysis of data Qualitative Quantitative Methods Transcription; Coding;
Thematic analysis.
Descriptive statistical analysis; Quantitative modelling and simulation using FACTS-software.
3.4 Sampling
The sample frame is host-company employees, texts, documents, and other (numerical) data files, from which units are sampled purposively (Bell, Bryman, and Harley, 2019). Primarily, the snowball sampling method is used, for, what appears relevant and interesting may change over time and the interviewees can recommend another unit accordingly. The units shall, however, be connected to the research questions and current theme to be theoretically saturated (Ibid). (See table 3 for a list of held interviews)
Table 3 – Held interviews.
