Improving supply chain visibility within logistics by implementing a Digital Twin

(1)

IN

DEGREE PROJECT INDUSTRIAL ENGINEERING AND MANAGEMENT,

SECOND CYCLE, 30 CREDITS STOCKHOLM SWEDEN 2020,

Improving supply chain visibility within logistics by implementing a Digital Twin

A case study at Scania Logistics

YLVA BLOMKVIST

LEO ULLEMAR LOENBOM

KTH ROYAL INSTITUTE OF TECHNOLOGY

SCHOOL OF INDUSTRIAL ENGINEERING AND MANAGEMENT

(2)

(3)

Improving supply chain visibility within logistics by implementing a Digital Twin

A case study at Scania Logistics

by

Ylva Blomkvist Leo Ullemar Loenbom

Master of Science Thesis TRITA-ITM-EX 2020:344 KTH Industrial Engineering and Management

Industrial Management SE-100 44 STOCKHOLM

(4)

Att förbättra synlighet inom logistikkedjor genom att implementera en Digital Tvilling

En fallstudie på Scania Logistics

av

Ylva Blomkvist Leo Ullemar Loenbom

Examensarbete TRITA-ITM-EX 2020:344 KTH Industriell teknik och management

Industriell ekonomi och organisation SE-100 44 STOCKHOLM

(5)

Master of Science Thesis TRITA-ITM-EX 2020:344

Improving supply chain visibility within logistics by implementing a Digital Twin

A case study at Scania Logistics

Ylva Blomkvist Leo Ullemar Loenbom

Approved

2020-06-08

Examiner

Hans Lööf

Supervisor

Bo Karlson

Commissioner

Scania AB

Contact person

Gerard Rosés Terrón

Abstract

As organisations adapt to the rigorous demands set by global markets, the supply chains that constitute their logistics networks become increasingly complex. This often has a detrimental effect on the supply chain visibility within the organisation, which may in turn have a negative impact on the core business of the organisation. This paper aims to determine how organisations can benefit in terms of improving their logistical supply chain visibility by implementing a Digital Twin — an all-encompassing virtual representation of the physical assets that constitute the logistics system. Furthermore, challenges related to implementation and the necessary steps to overcome these challenges were examined.

The results of the study are that Digital Twins may prove beneficial to organisations in terms of improving metrics of analytics, diagnostics, predictions and descriptions of physical assets.

However, these benefits come with notable challenges — managing implementation and

maintenance costs, ensuring proper information modelling, adopting new technology and leading the organisation through the changes that an implementation would entail.

In conclusion, a Digital Twin is a powerful tool suitable for organisations where the benefits outweigh the challenges of the initial implementation. Therefore, careful consideration must be taken to ensure that the investment is worthwhile. Further research is required to determine the most efficient way of introducing a Digital Twin to a logistical supply chain.

Keywords: Digital Twin, Digital Twins, Logistics, Supply chain visibility, Internet of Things, IoT, Cloud Computing, Machine Learning, Application Programming Interface, API, Cyber- physical systems, CPS, Manufacturing, Industry 4.0

(6)

Examensarbete TRITA-ITM-EX 2020:344

Att förbättra synlighet inom logistikkedjor genom att implementera en Digital Tvilling

En fallstudie på Scania Logistics

Ylva Blomkvist Leo Ullemar Loenbom

Godkänt

2020-06-08

Examinator

Hans Lööf

Handledare

Bo Karlson

Uppdragsgivare

Scania AB

Kontaktperson

Gerard Rosés Terrón

Sammanfattning

I takt med att organisationer anpassar sig till de hårda krav som ställs av den globala marknaden ökar också komplexiteten i deras logistiknätverk. Detta har ofta en negativ effekt på synligheten inom logistikkedjan i organisationen, vilken i sin tur kan ha en negativ påverkan på

organisationens kärnverksamhet. Målet med denna studie är att utröna de fördelar som

organisationer kan uppnå vad gäller att förbättra synligheten inom deras logistikkedjor genom att implementera en Digital Tvilling — en allomfattande virtuell representation av de fysiska

tillgångar som utgör logistikkedjan.

Resultaten av studien är att Digitala Tvillingar kan vara gynnsamma för organisationer när det gäller att förbättra analys, diagnostik, prognoser och beskrivningar av fysiska tillgångar.

Implementationen medför dock utmaningar — hantering av implementations- och

driftskostnader, utformning av informationsmodellering, anammandet av ny teknik och ledarskap genom förändringsarbetet som en implementering skulle innebära.

Sammanfattningsvis är en Digital Tvilling ett verktyg som lämpar sig för organisationer där fördelarna överväger de utmaningar som tillkommer med implementationen. Därmed bör beslutet om en eventuell implementation endast ske efter noggrant övervägande. Vidare forskning behöver genomföras för att utröna den mest effektiva metoden för att introducera en Digital Tvilling till en logistikkedja.

Nyckelord: Digital tvilling, Digitala tvillingar, Logistik, Supply chain visibility, Internet of Things, IoT, Molntjänster, Maskininlärning, Programmeringsgränssnitt, API, Cyberfysiska system, CPS, Tillverkning, Industri 4.0

(7)

VI

Acknowledgements

We would like to give a special thanks to Gerard Rosés Terrón, our supervisor at Scania, who has provided us with valuable guidance and contacts within Scania Logistics. We also want to direct special thanks to Bo Karlson, our supervisor at KTH, for his ideas and guidance

throughout the research process and thank our examiner Hans Lööf for his support.

We would also like to thank Peter Norrblom at Siemens and Jacob Edström at ABB Robotics for sharing their insights and expertise in Digital Twins. Lastly, we would like to thank the employees at Scania Logistics who helped us understand the logistics processes at Scania.

Ylva Blomkvist Leo Ullemar Loenbom Stockholm, June 2020

(12)

1

1 Introduction

1.1 Background

A functioning logistics supply chain is the lifeblood of an organisation. In organisations whose core businesses are centered around manufacturing both internal and external logistics play a vital part in ensuring that the core business of the organisation can proceed without stoppage (Tracey, 1998). While the basic objective of logistics can be summed up very briefly

— transporting something from point A to B — the means of which to do so can vary greatly according to the specific task at hand.

For many companies these tasks are becoming increasingly complex — as global markets put higher demands on customisation from customers, and both the goods produced and the production lines themselves involve more complex technology it is becoming more and more difficult to find solutions for the logistical challenges that surround the core businesses (Mangan & Lalwani, 2016).

As logistics supply chains become more complex, the process of attaining knowledge about the inner workings of the supply chain becomes increasingly difficult. This can quickly

become a problem for any organisation in which the supply chain has a vital role, as lacklustre transparency and visibility of your supply chain may cast shadows over inefficient processes that could otherwise be rectified (Creazza et al, 2010). Ensuring complete supply chain visibility not only grants access to a wealth of information that can serve as a basis for strategic decisions — it also gives greater ability to quickly react to changes and make rapid real-time operational decisions based on changes in demand or output (Caridi et al, 2014).

While supply chain visibility is important, obtaining and maintaining complete visibility throughout a supply chain network that spans all over the globe is a task bordering on the impossible without making use of new advances in digital technology (Branch, 2008). The manufacturing industry — and society as a whole — stands on the brink of what is commonly referred to as the fourth industrial revolution — or “Industry 4.0” (Lasi et al, 2014). The basic characteristics of Industry 4.0 are swift innovation and adaptation of new digital technologies and mindsets, with many new technologies gradually seeing more use and new applications within our modern industrial context (Tjahjono et al, 2017). While Industry 4.0 is still a rather abstract term, many of the technologies that make up the fourth industrial revolution are

(13)

2 seeing increasingly widespread usage — smart manufacturing, big data analytics algorithms and advanced human-machine interfaces to only name a few (Pfohl et al, 2015). These various technologies can be broadly categorised into four major groups (Erboz, 2017):

• Cyber-physical systems

• Internet of Things (IoT) platforms

• Cognitive computing

• Cloud computing

Within an industrial context of Industry 4.0 an area of technological research that is showing great promise is that of the Digital Twin (Uhlemann et al, 2017). A Digital Twin is a form of cyber-physical system that creates a high-fidelity virtual model of a physical asset by use of various IoT sensors. By use of machine learning algorithms, the wealth of data collected by the Digital Twin is then aggregated and analysed — facilitating strategic and operational decision making (Negri et al, 2017).

In this paper we will examine potential applications of Digital Twins within a logistical context, particularly when it comes to improving supply chain visibility of a complex supply chain. Our study is conducted at the behest of the Supply Chain Networks Intelligence department within Scania Logistics, and the focus of our study will be to understand which benefits and challenges an implementation of a Digital Twin within Scania Logistics would lead to.

1.2 Purpose

Scania, like many actors within the automotive industry, is now undergoing a transformation from being a supplier of trucks, buses and engines to gradually becoming a supplier of complete and sustainable transport solutions.

Scania aims to become a leader in efficient, connected and sustainable logistics within a global logistics network. The mission of Scania Logistics is to develop, manage and optimise inbound- and outbound logistics globally by creating flows with minimal environmental impact, high quality and minimal waste.

Supply Chain Networks Intelligence (or OISI) is tasked to develop initiatives that provide benefits for Scania Logistics. OISI is responsible for research concerning Web Applications, Advanced Analytics, Business Intelligence and New Technologies.

(14)

3 Currently Scania Logistics is facing several challenges. The push towards becoming suppliers of sustainable transport solutions is increasing demands for effective logistics solutions, as the numbers of actors involved and the complexity of issues both rapidly increase.

As such, Scania OISI has become interested in the concept of Digital Twin and the potential benefits and value that can be created by adopting and implementing a Digital Twin within Scania Logistics. The aim of this master thesis is to describe the concept itself and what it entails, which potential benefits that can be gained and which challenges that are likely to be encountered on the road to implementing the Digital Twin for a company like Scania

Logistics. Questions regarding the technical and digital implementation and questions regarding the process of change management within the organisation will be raised in this paper, to ensure a holistic view of the situation.

1.3 Research Questions

The aim of this thesis is to understand the benefits that an organisation like Scania Logistics stand to gain by implementing a Digital Twin, and to pinpoint any challenges that are likely to be encountered. Furthermore, as Digital Twins is a relatively new concept, there is a clear need to conceptualise Digital Twins and to describe the technologies involved in its creation.

The following research questions have thus been formulated:

1. How are Digital Twins conceptualised, and which technologies are necessary in their development?

2. Which potential benefits would the implementation of a Digital Twin lead to for an organisation like Scania Logistics?

3. Which potential challenges must be considered to achieve a successful implementation of a Digital Twin within an organisation like Scania Logistics?

4. Which steps are necessary for an organisation like Scania Logistics to overcome the aforementioned challenges in implementing a Digital Twin?

1.4 Delimitations

While Digital Twins are today primarily used within manufacturing processes, this research is conducted at the behest of Scania Logistics with the specific aim of improving the logistics supply chain. As such, Digital Twin from the standpoint of manufacturing will not be part of the scope of this thesis.

(15)

4 Furthermore, the implementation roadmap will not involve all potential applications of a Digital Twin within Scania Logistics. The research process involved periods of practical work experience together with experienced personnel at Scania Logistics, during which all of the basic processes within the Logistics department were mapped. While implementing a Digital Twin would doubtlessly be beneficial for all the sub-processes within Scania Logistics, creating an in-depth roadmap of implementation of an all-encompassing Digital Twin would be beyond the scope of this thesis. As such, the decision was made to focus on a sub-set of processes where the benefits of a Digital Twin were deemed greatest For Scania Logistics.

A summary of the benefits of a Digital Twin for the excluded processes will still be given within the chapter detailing the processes within Scania Logistics.

Furthermore, the outbreak of covid-19 in Sweden during the spring of 2020 led to clear and tangible limitations for the research process. For a significant part of the allotted research time company grounds were closed to external personnel — obstructing the process of obtaining empirical research data during the time of the outbreak.

The temporary suspension of all production units within Europe by Scania on the 25th of March led to further difficulties obtaining practical data from Scania production units during the scope of the project.

1.5 Expected Contributions

As Digital Twins are a recent addition to the research community the amount of existing research — particularly within the area of logistics — is somewhat limited. Furthermore, there is as of yet no clear scientific consensus regarding the definition of a Digital Twin within the research community. Therefore, a primary goal of this thesis is to make a scientific contribution to the field of Digital Twins and aid in constructing a consensus regarding the scientific definition of a Digital Twin.

While this research has been centered around the setting and specifications laid out by Scania Logistics, it is the aim of this thesis to further contribute to future research, so that it along with other articles may serve as a base for more general conclusions regarding the area of Digital Twins within logistics.

(16)

5

1.6 Outline

Chapter Content

1. Introduction

In this chapter, the background and the purpose of the study is described, followed by posing the research questions that form the basis of the thesis. This chapter will also introduce the delimitations of the report, concluding with sections devoted to expected contributions and an outline of the thesis.

2. Theoretical Framework

This chapter regards the theoretical framework of the thesis, centered around the concept of supply chain visibility and the impact it has on the core business of organisations.

3. Methodology

In this chapter the choice of research methods are explained. Concluding the chapter is a section on research ethics.

4. Results

The result chapter describes the overall results gathered through the methodology. The results of the literature review is presented and followed by information from interviews and work practice at Scania and other external sources.

5. Discussion

In this chapter the results from the literature review, interviews and work experience are discussed and analysed. The findings are contextualised within existing theoretical literature and the theoretical framework. Concluding the chapter is a road map towards the successful implementation of a Digital Twin within logistics.

6. Conclusion

The conclusion chapter answers the research questions and summarises findings from the previous chapters regarding benefits and challenges surrounding the implementation of a Digital Twin. This final chapter also proposes future research areas that may serve as a basis for strategic decisions and further research contributions.

(17)

6

2 Theoretical Framework

2.1 Supply Chain Visibility

Supply chain visibility (or SCV) is a part of the broader realm of supply chain management (or SCM) and is a concept centered around creating visibility throughout supply chains in order to improve internal decision making and operating performance. The study of supply chain visibility has been firmly brought into the scientific limelight over the recent years, as global megatrends towards globalisation has posed many challenges with regards to managing the existing supply chains of companies as they expand in accordance to the demands of the market (Caridi et al, 2014). At its core, supply chain visibility is centered around ensuring that the company has access to accurate and current information regarding their supply chains, in regards to both internal and external processes (Francis, 2008). This is achieved by identifying which supply chain processes are most critically affected by lack of visibility and establishing means by which to gather and share relevant information between all affected parties (Caridi et al, 2014).

Ensuring proper supply chain visibility within a company can have a myriad of positive effects — improving forecasting, planning, scheduling and execution of orders to only name a few (Wei & Wang, 2010). While visibility is closely linked to information sharing, it is important to distinguish the two as the sharing of information is an activity while visibility is a potential result of the activity of information sharing (Barratt & Oke, 2007).

The primary benefits of supply chain visibility are an overall enhancement of company performance, providing a basis for improved decision-making — both on a strategic and operational level (Wang & Wei, 2007).

On a strategic level, organisations benefit from proper supply chain visibility by obtaining the ability to quickly and efficiently reconfigure the supply chain — a skill that is becoming increasingly important when it comes to generating competitive advantage in rapidly evolving business environments. Wei and Wang propose in their paper from 2010 that there are four core visibility processes which, when enabled, allow an organisation to reconfigure their supply chain in accordance with both their own needs and outside demands. These are visibility for sensing, learning, coordinating and integrating.

(18)

7

2.1.1 Visibility for sensing

This metric indicates the extent of which the organisation can quickly get real-time information regarding internal and external processes and react to a changing business environment (Gosain et al, 2014). One of the most crucial items of information regarding supply chains is current business intelligence regarding customer demands, as the importance of information regarding market and customer information cannot be understated in terms of creating opportunities from changes in the business environment (Wei & Wang, 2010).

Organisations with strong systems of information sharing in place with business partners and actors within their supply chain may reap a multitude of benefits — being able to swiftly react to changes in customer preferences and demands and quickly acting upon new business opportunities that competitors without systems of information sharing in place may have been left unaware of (Madhavan et al, 1998).

2.1.2 Visibility for learning

This indicates the extent of which the organisation can gather and learn from new information and knowledge from both internal and external processes. Knowledge about the external processes that affect the company are crucial for maintaining business advantage, and

organisations may acquire new knowledge and capabilities via their partners within the supply chain (Teece et al, 1997).

As the process of evaluating and adapting to external processes is a key factor when it comes to discovering and implementing new business opportunities the value of active learning processes within the organisation targeted towards suppliers and customers cannot be understated (Teece, 2007). Bringing external knowledge from multiple external sources into one’s organisation may also lead to the ideas interlinking into new ideas unto themselves, further increasing the performance of the targeted processes (Decarolis & Deeds, 1999).

According to Zollo and Winter (2002), the process of dynamic learning is underbuilt by three primary mechanisms: experience accumulation (the very process of gathering knowledge) knowledge articulation (explaining the knowledge so that it can be shared with others) and knowledge codification (adopting the experiences garnered and adapting the knowledge to one’s organisation). Adopting dynamic learning mechanics within an organisation may have the further benefit of fostering a culture of reflection and continuous improvement, enabling processes within the company where employees raise ideas of their own that could be beneficial for their areas of work (Nonaka et al, 2000).

(19)

8

2.1.3 Visibility for coordinating

This represents how adept the organisation is at coordinating different areas of their supply chain — making decisions with overarching consequences for many different actors within the system. Having complete information regarding the supply chains is a necessity when it comes to maintaining a higher level of decision making regarding business decisions that impact the supply chain, as global product flows are the sum of a multitude of different actors (Simatupang et al, 2004).

According to Malone and Crowston (1994), coordination is the art of managing dependencies

— which processes are dependent on one another, what systems need to be in place in order to transfer goods and which tools are necessary to enable full usability of the supply chain.

When it comes to ensuring visibility for coordination, the focus should be on providing information for managing the different kinds of dependencies between the different actors within the supply chain (Wei & Wang, 2009).

The sharing of information between actors within the supply chain is key for ensuring proper visibility for coordination — sharing real-time data when and where items are to be delivered, establishing clear guidelines regarding necessary inventory levels and buffer stocks and order- and production forecasts to only name a few (Malone & Crowston, 1994). Furthermore, it is beneficial to share knowledge regarding the desired characteristics of the final product, as suppliers can then obtain and act upon information regarding product requirements and customer desires (Wei & Wang, 2009).

2.1.4 Visibility for integrating

This represents how adaptable the organisation is when it comes to adopting and integrating new methods and technologies in order to develop a strategic business advantage (Teece et al, 1997). When it comes to supply chain management, the development of a collective identity regarding the supply chain is a crucial step, as it enables a mindset that facilitates the

integration of processes between different actors within the supply chain (Dyer, 2000). In order to ensure this, information regarding all the key processes within the supply chain must be shared between the various actors, as understanding the core processes of others may lead to breakthroughs in improving the supply chain as a whole (Gosain et al, 2004).

It has been observed that long-term collaborations with high levels of information sharing between business partners tend to show higher levels of visibility with regards to integration

(20)

9 (Elgarah et al, 2005). Understanding the abilities, advantages and challenges of the other actors within the supply chain is beneficial when it comes to achieving harmonious synergy between the different actors, and having a clear common goal to strive towards is

advantageous for the organisation as a whole (Jap, 1999).

Maintaining proper supply chain visibility within an organisation by incorporating processes that strengthen these four aforementioned metrics can lead to a substantial increase in real time strategic- and tactical information, which in turn may lead to a dramatic effect in reducing demand distortion — also known as the bullwhip effect — which has the added benefits of lowered uncertainty within the organisation and increased customer satisfaction (Lee, 1997).

A core driver behind improved supply chain visibility is the adoption and implementation of new systems to gather, manage and analyse information (Mora-Monge et al, 2010). As such, information technologies such as Digital Twins are likely to have a tremendous beneficial impact with regards to all four core aspects of supply chain visibility as described above (Caridi et al, 2014).

(21)

10

3 Methodology

3.1 Research process

For this thesis, a qualitative research approach was chosen, given the scope of the research as an investigative thesis as opposed to one of a statistical nature. In accordance with the ideas laid forward by Blomkvist and Hallin in 2015, the appropriate initial method of investigation, considering the limited amount of previous investigated research about the topic, was to conduct an investigation from an exploratory perspective. Blomkvist and Hallin also propose that an abductive research approach is suitable when a phenomenon is heavily impacted by empirical data, an approach in line with the research that has been conducted.

The research process started with two parallel research trajectories (see figure 3.1), where one trajectory focused on internal research at Scania and the other trajectory focused on external research conducted outside of Scania. The internal research process began with a first round of interviews, conducted at the various departments at Scania Logistics in Södertälje, in order to get a brief overview of their functions. This was followed by a process of gathering further empirical data through work practice at relevant areas of Scania Logistics. To conclude the internal research process, additional semi-structured interviews with a number of employees at Scania Logistics were conducted in order to answer in-depth questions about the different areas upon which the research would be based.

In parallel to the internal research, the external research process started with a preliminary round of research of the concept of Digital Twins themselves. The information that was garnered in this stage was used to formulate some preliminary research questions that would guide the future research that was to be conducted. This was followed by a thorough literature review, which was conducted in order to define the concept of a Digital Twin and how it differs from other concepts in digitalisation. Furthermore, empirical data was collected through semi-structured interviews with experts in Digital Twin at external companies.

When the collecting of empirical data was concluded the data was analysed and discussed and conclusions were drawn. Lastly, new research questions and areas for future research were proposed for the benefit of future researchers.

(22)

11 Figure 3.1. Overview of the research process.

3.2 Data collection

3.2.1 Internal research at Scania

Getting to know Scania Logistics

In order to understand the logistics department at Scania, the first weeks of the research process were dedicated to conducting interviews with personnel from the departments within Scania Logistics. These interviews were approached without a formal interview structure, giving the interview subjects room to explain on their own terms. The interviews were

however similar in their general structure, starting off with a brief and simplified introduction about the topic of the master thesis followed by representatives from the different

subdivisions explaining the role of their division within the logistics supply chain. Clarifying questions were asked in order to give the interview subjects room to develop their

explanations throughout the interviews, concluding every session with questions about existing challenges and internal projects. The information gathered during these interviews served as a foundation for the initial scope of the project. During this initial round of interviews, any challenges and projects were described in too general terms to assess the applicability of a Digital Twin as a possible solution.

Our initial rounds of interviews at Scania were conducted with the following personnel:

• Head of Global Material Control.

(23)

12

• Head of Logistics Supplier Management.

• Head of Nordic Transport Control.

• Head Logistics Sustainability Manager.

• Head of Supply Chain Networks Intelligence.

• Head Sustainability Manager.

• Team leader within Packaging Planning.

• Head of Packaging Handling and Assortment.

• Head of Network and Supply Development.

• Head of Material Supply Engineering.

• Head of EDI Development and Support.

• Gerard Rosés Terrón, Logistics Developer and master thesis supervisor.

Work practice at Scania

While useful as a source of general knowledge about Scania Logistics, the initial round of interviews did not go into the level of detail that was deemed necessary for our research.

Thus, five work practice sessions were arranged with key personnel within Scania Logistics, with the main method of data gathering for these sessions being observation coupled with informal interviewing. This was deemed an appropriate method as the aim was to learn about both the systems and behaviours of personnel in the context of their daily work (Cooper, Reimann, Cronin & Noessel, 2014, p. 43). These work practice sessions were all roughly four hours in length per session, during which practical observation of work methods used by the personnel was gathered, asking clarifying questions when necessary. This provided valuable insight about the data systems used in the daily work and how the departments, clients and suppliers communicated with one another. This deeper, detailed knowledge was vital in order to assess whether the implementation of a Digital Twin would be beneficial to this area of the logistics process.

Additional in-depth interviews at Scania

Management at Scania Logistics are aware of the issues that exist within the organisation, and there are several internal projects to come up with solutions to these issues. The head of the department of Strategic Long-Term Development within Scania Logistics, has profound insights into these ongoing projects. The role of Strategic Long-Term Development is to examine possible solutions and new ways of working within the logistics department with a timeframe of ten or more years in the future. To establish which long-term projects and strategies were already in motion at Scania Logistics two semi-structured interviews were

(24)

13 conducted with the head of the department of Strategic Long-Term Development, in order to assess whether these projects could potentially be beneficial to or incorporated in a Digital Twin. Questions were posed regarding the processes of material, packaging and transport planning, with the intent of obtaining an overview of the current and future state of Scania Logistics and the applicability of a Digital Twin solution.

Further semi-structured interviews were also conducted with the project leader of the internal improvement project Scania Track. A full list of prepared interview questions was prepared for these interviews and can be found in appendix 1.1 and 1.2, but in accordance with semi- structured interview processes questions were also raised during the interview. Two virtual semi-structured interviews at a production line were also conducted in order to test and discuss our conclusions and to obtain insights regarding the inner workings and procedures of the supply chains at the Scania Production Units (or PRUs). These interviews were conducted with personnel affiliated with the Logistics Centers at the Scania PRU in Södertälje. Further questions about Scania Logistics were answered by Gerard Rosés Terrón, logistics developer and master thesis supervisor at Scania Logistics.

3.2.2 External research

Initial research

This phase of the research process served to create an overview of the concept of Digital Twins and to get an idea what to consider when conducting the literature review. During this phase, questions about Digital Twins were discovered and answered and organisations with prior knowledge about Digital Twin technology were pinpointed for future interviews.

Literature review

To obtain further in-depth knowledge about the concept of a Digital Twin, a literature review was conducted on articles available on various research platforms such as Google Scholar and KTH Library. This approach was chosen as it provides a broad selection of scientific articles from all over the world and provides ample opportunity to focus the research by way of filters.

The literature review was conducted in accordance with the methods laid out by Alvesson and Sandberg in 2011:

• Defining and refining the research question in order to form a clear understanding regarding the subject matter at hand.

(25)

14

• Determining the required characteristics of primary studies - establishing clear criteria for the basis with which to include or exclude literature and articles.

• Conducting a thorough information collection, gathering samples of potentially relevant literature.

• Conducting a selection of articles pertinent to the study according to our previously outlined metrics.

• Synthesising the gathered literature so that conclusions can be drawn.

• Presenting the results of our conclusions.

• Formulating questions for future research.

These search strings containing “Digital Twin” were varied with the search terms

“manufacturing”, “production” or “logistics”. It quickly became clear that there is a substantial research gap in Digital Twins as applied to logistics, with a majority of articles regarding Digital Twins focused on their application on manufacturing physical assets or products. This spurred a decision to exclude the search term “production design” from searches, as this area is not included in the research scope of the master thesis.

For the benefit of replicability, search terms used in Google scholar and KTH Library were:

digital twin, logistical flows, siemens, dhl logistics, machine learning, cloud computing, IoT, Internet of Things, API, application programming interface, VR, AR, IBM Digital Twin.

When there was a need to explain a basic concept, the research basis was extended to other sources and the information was cross-checked with other independent sources.

External expert interviews

In order to broaden the research about Digital Twins and not solely rely on information gathered from scientific articles, external interviews with companies involved with

implementation of Digital Twins were scheduled. Invitations for such interviews were sent to DHL, Siemens, IBM, Midroc and ABB, and interviews were then set up with representatives from Siemens and ABB Robotics. At Siemens, an interview was conducted with Peter Norrblom, Business Development Manager and responsible for the pre-sale and business development of Digital Twin solutions for prospective clients. At ABB Robotics, an interview was conducted with Jacob Edström, R&D Scrum Master with a focus on implementation of Digital Twin solutions. These interviews were conducted in a semi-structured manner and the questions asked can be found in appendix 1.3 and 1.4.

(26)

15

3.3 Source criticism

The report is based on a broad variety of sources — external experts, scientific articles and employees from different areas at Scania to name a few — all of which enhances the reliability of the result. However, there are many employees with unique experiences and insights about Scania Logistics, and considering the time constraints of the allotted time for the thesis research there was not enough time to interview every employee with knowledge in the matter. Furthermore, because of the necessary delimitations of the thesis project, the questions asked during the interviews were often targeted questions that did not always include generalisability in all aspects.

In addition, the experts in Digital Twin that were interviewed might have had further

knowledge that would have been valuable to this report, but were not always allowed to share all knowledge due to regulations from their employers. Lastly, one must recall that as Digital Twin is a relatively recent research area, the research gap — particularly within logistics — is substantial, and much research simply does not yet exist.

3.4 Validity, Reliability & Generalisability

An important criterion of rigorous quality is construct validity — the conceptualisation or operationalisation of the relevant concept (Gibbert, Ruigrok & Wicki, 2008). Rigorous quality means that it is important to create validity to the methods by which data is gathered, such as triangulating the data from different angles. Therefore, using different kinds of methods to gather data and construct the methodology of the report lends strength to the research that has been conducted. Gibbert, Ruigrok and Wicki (2008) raised four criteria used to assess the rigor of field research, one of them being external validity. External validity means that the applied theories must be applicable not only in the setting in which they are studied, but also in other parallel settings.

Reliability refers to replicability of the research — stating whether the research is transparent and precise enough to give the same results if the research was to be repeated (Gibbert,

Ruigrok & Wicki, 2008). While the aim is to rigorously document all methods and results in a transparent manner, given that the nature of the research process is built on human interaction the given results may not yield the exact same information twice. Furthermore, given the rapid pace of technological improvement the yielded results may differ. The interviewed staff

(27)

16 might also be exchanged, and ways of working changed, causing further issues with

replicability.

The combined body of research on Digital Twins will most likely develop even further after the publication of this report — meaning that the replicability of the report will further diminish over time.

The final factor is generalisability, in which Gibbert et al (2008) stated that only analytical and statistical generalisation can be applicable to qualitative research. Analytical

generalisation implies that there could be generalisation drawn between different case studies of a similar nature. As practical cases of research regarding the application of a Digital Twin in a logistical context is limited, this type of generalisation is not applicable to this research process. However, the results of this master thesis could serve as a basis for future research in tandem with other case studies in the field of Digital Twins.

3.5 Ethics

This research study was conducted at the request of and in collaboration with the automotive manufacturer Scania. A confidentiality agreement with Scania was signed, which dictated that no sensitive information was to be shared with any unauthorised personnel and that all

material and equipment given by Scania during was to be returned to the company at the end of term. A one-time monetary reward was given at the beginning and end of the project.

The research has been conducted in strict adherence to the basic principles of research ethics as dictated by the Swedish Research Council — clarity of information, consent,

confidentiality and good use (Vetenskapsrådet, 2002). All subjects of interviews were clearly informed of the purpose of the study before participating, and no interviewee were in any way forced to participate in the research. The terms of confidentiality were honoured by ensuring that all information dictated as confidential or sensitive (as per the priorly mentioned

confidentiality agreement) was treated as such in the final thesis report. The terms of good use will be ensured as the information and data collected in the study will not be used for any other purpose than that of the research elaborated upon within this thesis.

Any ethical implications of the results of the research itself will be discussed at the conclusion of the discussion chapter.

(28)

17

4 Results

The results of the conducted research will be divided into separate parts. The initial chapter regards the concept of the Digital Twin itself — presenting a scientific definition of the concept, going into depth about the technologies that make up a Digital Twin and which benefits and challenges they incur. The subsequent chapter regards the case study at Scania Logistics and lays out the premises of the current state of affairs based on interviews and personal work experience.

4.1 Defining the Digital Twin 4.1.1 The definition of a Digital Twin

The concept of Digital Twin was initially presented by Michael Grieves at a presentation about Product Lifecycle Management in 2003 at the University of Michigan (Grieves, 2014).

The first actionable Digital Twin was then developed by NASA in 2011 as a method to predict the structural behaviours of aircrafts by analysing and simulating them as digital models. Scientists at NASA later defined Digital Twin as “an integrated multi-physics, multi- scale, probabilistic simulation of a vehicle or system that uses the best available physical models, sensor update fleet history and so forth, to mirror the life of its flying twin.” (Lu et al, 2020). See figure 4.1 for a visual representation.

Figure 4.1. A visual representation of a Digital Twin in an aircraft (GE Digital, n. d.).

As the potential areas of applications of Digital Twin are growing more numerous by the day, it becomes apparent that there exists no established consensus in the research community of

(29)

18 what constitutes a modern Digital Twin. It may seem tempting for a company to

enthusiastically label their existing 3D-modelling solutions containing asset-tracking technologies as a Digital Twin — a line of action that risks diluting and short-selling the complexity and potential benefits of a fully-fledged Digital Twin (Negri et al, 2017).

In the interest of furthering the consensus and serving as a basis for further academic endeavours, the following list of attributes is a proposed list of what constitutes a Digital Twin. These attributes are based on individual research combined with observations from academic works published by Qi et al (2018), Lu et al (2020), Uhlemann et al (2017) and Negri et al (2017). The attributes are as follows:

• A Digital Twin is a virtual representation (or model) of a physical object or process.

• The Digital Twin is continuously updated with real-time data to reflect the current state and behaviour of the physical object or process.

• The Digital Twin can aid in visualising and analysing the physical object or process, and by use of machine learning further optimisations and predictions can be made.

By use of these attributes, one can broadly encompass most potential applications of Digital Twin, from something as contained as a part of a production line or something so large in scope as the entirety of a city.

Further deconstructing the concept, Qi et al proposed in their paper from 2018 five separate components that are necessary for a Digital Twin in a manufacturing setting. Building upon the components proposed by the aforementioned authors, using the following five general components one can create a Digital Twin for any conceivable set of applications.

• Crucially, there needs to exist some form of physical entities that can form the basis of the Twin. These physical entities are part of a system that has a clear objective or task

— a task that can then be recreated and monitored by the Digital Twin.

• Virtual models can then be created of the physical entities, which can then reflect all aspects of interest of the aforementioned physical entities.

• Digital Twin services integrate various functions such as management, control and optimisation to provide services according to the requirements of the Twin.

• The collected data is the lifeblood of the Digital Twin. This includes data from physical entities, virtual models and services. It also includes the data created by the twin itself in the cases of twins containing sophisticated machine learning algorithms.

(30)

19

• Without the methods of connecting the physical and the virtual worlds, there can be no Digital Twin. It is only through these connections that the aforementioned

optimisations and predictions can be made, and ensuring proper connectivity is thus of the utmost importance.

Any system that adheres to the aforementioned three attributes along with the five applications described above can be said to be a Digital Twin.

A brief example of a Digital Twin

To provide an idea of what the final product of Digital Twin, links to a demo version of a Digital Twin representation of a jet engine developed by Autodesk Forge is provided below.

https://forge-digital-twin.autodesk.io/ (Autodesk Forge, 2019). Snapshots of the demo can be found below in figure 4.2 and figure 4.3, showing live measurements of temperature as the engine operates.

Figure 4.2. Temperature shown in a Digital Twin (Autodesk Forge, 2019).

(31)

20 Figure 4.3. Temperature shown in a Digital Twin (Autodesk Forge, 2019).

4.1.2 The technologies enabling Digital Twin

Internet of Things (IoT)

Internet of Things (or IoT) refers to any objects that are interconnected to a network, often equipped with ubiquitous intelligence. By integrating several objects for interaction via embedded systems, a large distributed network of communicative devices is created — facilitating human-device and device-device communication (Xia et al., 2012). The introduction of IoT could provide new levels of visibility and adaptability and improve performance in various areas, from smart homes to supply chains.

In a supply chain, the data collected from several devices in different areas could be analysed and subsequently alert human operators of any potential problems by providing early

warnings (Ben-Daya, Hassini & Bahroun, 2019).

In order for IoT-devices to have any data of interest to communicate with one another, they must be equipped with sensors. The IoT-sensors come in all shapes and sizes, and are utterly indispensable for the concept of Digital Twins as they input the data that a Digital Twin could analyse and determine the current state of its physical twin (Haße et al, 2019).

Another technology used to communicate between IoT-devices in a network is RFID (Radio Frequency Identification Devices), which allows automated identification of devices by providing them with physical IoT tags. These can, through the common radio interface, communicate with an external RFID reader. IoT-sensors play an important role in translating

(32)

21 indications from the physical world into usable data that can be communicated over the IoT network. There are a broad range of complexity and costs between different kinds of sensors, all depending on which information has been deemed relevant to collect. Provided below are some examples of IoT-sensors:

• Proximity sensors, which emits a signal (which can be infrared, ultrasound or electromagnetic in nature) and registers variations in the return signals. This form of sensor has the benefit of removing the need for physical contact. Proximity sensors could be inductive, capacitive, ultrasonic, photoelectric or magnetic, and proximity sensors are often used for security and efficiency applications in many kinds of

industries. Many modern cars are also equipped with proximity sensors to help drivers with parking and seeing obstacles around the car. Proximity sensors can be used to detect, count and determine the position of objects and measure the movement of objects within the system.

• Position sensors, which are useful when sensing motions of an object in a particular area.

• Occupancy (or presence) sensors, which senses objects in a particular area.

Occupancy sensors can for example monitor temperatures, light and humidity.

• Other examples of sensors are motion sensors, velocity sensors, temperature sensors, pressure sensors, chemical sensors, water quality sensors, optical sensors and gyroscope sensors (Sehrawat & Gill, 2019).

The sensors themselves can affect the performance of the physical product. This means that the mass or volume of the sensors cannot be so that it runs a risk of affecting the measured results, something which is especially true if you insert a sensor in a process with high requirements in regards to precision. Physical sensors must thus always be included in product simulations (Edström, personal communication, April 16th, 2020).

Cyber-physical systems (CPS)

A cyber-physical system is broadly defined as a system which integrates the physical and the digital world. Through sensors, the system can communicate conditions of the physical world to the digital world and vice versa, allowing the corresponding asset to adapt and improve its efficiency. This creates a loop of information flowing through the system autonomously, without any human intervention (Zanero, 2017). Figure 4.4 describes how Digital Twins, IoT technologies and physical assets fit into the cyber-physical system.

(33)

22 Figure 4.4. A simplified structure of a Digital Twin’s connection to the physical world (Lu et al, 2019).

Machine Learning

Machine learning is one of the fastest growing technical fields today. It combines elements from both computer science and statistics and can be defined as a computer system whose performance improves automatically through experience. In layman's terms, machine learning focuses on two questions: “How can one construct computer systems that automatically improve through experience? and What are the fundamental statistical-computational- information-theoretic laws that govern all learning systems, including computers, humans, and organisations?” (Jordan & Mitchell, 2015).

The terms machine learning and artificial intelligence (AI) are often used interchangeably.

However, machine learning is considered to be a subset of AI, which is a broader term encompassing the entire field of artificial intelligence. Therefore, while machine learning is the same as AI, AI is not necessarily the same as machine learning (Camerer, 2018).

There are a huge variety of applications in which machine learning is used — speech

recognition, computer vision, language processing and robot control to only name a few. For some machine learning systems, it could be more effective to train the system by giving it example inputs and outputs instead of coding it manually. Today, machine learning is gradually being implemented in several different areas such as health care, education, manufacturing, marketing, consumer service and logistics (Jordan & Mitchell, 2015).

(34)

23 Machine learning structures could provide a Digital Twin with the ability of analysing and making decisions based on real-time data. Based on this data, the Digital Twin could also be able to make optimisations and predictions about possible future outcomes (Haße et al, 2019).

Cloud Computing

Cloud computing is a model for enabling shared network access to computing resources.

Using cloud computing, the user is able to access a pool of resources that are owned and maintained by a third party via the internet. The access to these resources is independent of their physical location. Examples of computing resources utilised in cloud computing are data storage, databases, computing power and services like data analytics processing (Arora et al, 2013).

The benefit of using a cloud is that the user does not have to buy and own expensive hardware or own any storage space. The user only pays for the cloud services that they use, always having the option to scale up or scale down on their demanded services. There are three primary different models of cloud services — Information as a Service (IaaS), Platform as a Service (PaaS) and Software as a Service (SaaS). While cloud computing already is a popular service, it has in some cases been questioned in terms of security as the cloud often has a various number of clients using the same services (Arora et al, 2013).

From a Digital Twin perspective, development and maintenance of a Digital Twin are both immensely computing power and storage intensive as it continually produces huge amounts of data that needs to be processed. A cloud service could therefore be a good option for many organisations that wish to keep costs low while maintaining flexibility (Heutger &

Kueckelhaus, 2019).

The cloud can also be beneficial when it comes to building up the infrastructure for storing data. Clouds contain substantial security measures, including the fact that all data is encrypted at every data center. With public cloud providers, data is also encrypted at rest. This means that when data is stored on a data disk, the data remains encrypted even if the data disk should be lost or stolen. Furthermore, cloud providers have stringent physical security measures at their data centers, so there cannot be said to be any more significant increased risk with storing data on the cloud rather than on an internal network solution. Many larger companies use private contractors for their cloud services — ABBs Digital Twin makes use of the services provided by Azure (Edström, personal communication, April 16th, 2020).

(35)

24 API (Application Programming Interface)

APIs (or Application Programming Interfaces) allows for communication between

applications such as databases, networks and IoT sensors. APIs are building blocks built by developers to be reused, so one does not have to redo the programming from scratch. A clear example of this is the Google Maps API that can be seamlessly integrated with contents from a third party in order to view points of interest nearby on the Google Map. In this case, the third party does not have to program an entirely new map — something which saves both time and resources (Conrad et al, 2016). As many modern organisations work with cloud computing, APIs help with transferring data efficiently between clouds, devices and other systems (Wodehouse, 2016). Figure 4.5 shows an overview of the role of APIs in application development.

Figure 4.5. Overview of the function of APIs for data gathering of assets, developers and end users (Wodehouse, 2016).

There is a distinction between private and public (or open) APIs. The aforementioned example with Google Maps API refers to a public API that is open source. Private APIs are only used internally inside of an organisation — at most extending to their partners. The benefit of private APIs is that they can be tailored to fit the exact needs of the organisation and can thus be simplified to suit that purpose. They can also provide a pool of data that co- workers can access so that they may work more efficiently (Wodehouse, 2016).

The benefit of using public APIs is that they are open-source and oftentimes free to use, which helps lower development costs. Public APIs may also be suitable if one wants to promote their brand or collect analytics about traffic and users using the API. It can also give developers the means to focus on developing the core functions of an application, as the basic functions could be provided by public APIs (Wodehouse, 2016). However, there are notable security risks connected to public APIs. In 2015, the Facebook API was compromised due to a security flaw in the algorithm, which led to thousands of Facebook accounts being

compromised by linking their accounts to their personal phone numbers (Conrad et al, 2016).

(36)

25 When it comes to security of the data contained within the Digital Twin, the single most important item is that there are APIs and clear access rights in place. Managing open APIs makes user management even more important. If a company gives rights to certain users to gain access to data, the greatest security risk lies in users having their data accounts

compromised. As such, solid security and password management by API users is mandatory (Edström, personal communication, April 16th, 2020). While there can definitely exist combinations of open and private APIs, most companies are likely to start with private for managing information that is to remain within the organisation, only allowing clear subsets of information to go out to third parties (Edström, personal communication, April 16th, 2020).

Augmented and virtual reality

Augmented and virtual reality are both becoming increasingly popular tools in terms of enhancing user experience. Virtual reality is a technology that imitates the real world and how the user experiences it virtually by creating a virtual world. This virtual world could be

anything from creating a high-fidelity simulacrum of the real world to simulating a specific part of the user experience. In contrast, augmented reality adds a layer of information to the real world rather than creating a whole new virtual world. A tangible example of everyday augmented reality is used by certain smart phone applications, where elements are added clearly visible through the phone camera that cannot be seen in the physical world (Ge et al, 2017). Figure 4.6 shows an example of how augmented and virtual reality differs from one another.

Figure 4.6. The difference between Virtual Reality and Augmented Reality (Nobel, 2019).

In the context of a Digital Twin, both augmented and virtual reality can be useful tools to view and inspect the Digital Twin either on a screen (2D) or in a physical space (3D). The aforementioned technologies such as IoT, cloud computing, APIs and machine learning all

(37)

26 provide and process the necessary data and infrastructure in order to create and visualise a Digital Twin in either augmented and virtual reality (Heutger & Kueckelhaus, 2019).

4.1.3 The benefits of a Digital Twin

As the scope and breadth of applications of Digital Twins can vary greatly, the potential benefits are equally numerous. A Digital Twin allows the user to conduct the daily operations of their twinned assets remotely, which aids in optimisation and lowering of service costs.

Digital Twins can also be used to automate monotonous activities that may often be prone to human error — something which subsequently allows resources to be centered around more value-adding activities (Heutger & Kueckelhaus, 2019).

Early adopters of Digital Twins often see value being added in three distinct categories:

improved decision making thanks to higher quality data, optimisation of day-to-day processes and freeing up resources which enables integration of new business models (Grieves, 2014).

For instance, a Digital Twin can enable a shift towards servitisation and product-as-a-service within companies that previously had no option whatsoever to do so, as managing an asset through its lifecycle becomes increasingly streamlined by use of a Digital Twin (Heutger &

Kueckelhaus, 2019).

The list of benefits can be further categorised into four distinct categories as outlined below (Heutger & Kueckelhaus, 2019).

Descriptive Value

By use of a Digital Twin, information of the twinned physical asset can immediately be observed from a distance. While this type of remote observation has many beneficial applications, it can be particularly desirable if the physical assets are located in hazardous locales (Heutger & Kueckelhaus, 2019). Remote visualisation by use of Digital Twins can also be immensely helpful in obtaining information for assets that operate outside of normal day-to-day operations — long distance hauling, data analysis from distant sensors and

remotely observing assets owned by an organisation operating in off-site production plants to only name a few potential applications (Grieves, 2014).

Analytical Value

Another benefit to using Digital Twins is the ability to gather data that is normally difficult to gauge — such as information generated from the interior of an asset (Heutger & Kueckelhaus,

(38)

27 2019). Using this information one can then make decisions and propositions in order to

improve subsequent generations of the asset (Grieves, 2014).

Diagnostics Value

Digital Twins can include systems that, based on the gathered data, can suggest likely root causes of the current state of the physical asset (Heutger & Kueckelhaus, 2019). These diagnostics systems can take the form of use of advanced analytics- and machine learning algorithms that are based on prior quantifiable data (Uhlemann et al, 2017).

Predictive Value

An area where Digital Twins can prove to be extremely beneficial is in the potential to create calculated predictions. By using the immense amount of data that is assembled by the Digital Twin, the likely future state of the physical asset can be predicted (Heutger & Kueckelhaus, 2019). While interesting in and of itself, the most sophisticated types of Digital Twins are not only able to merely predict the issues that may occur — they can also suggest a solution (Uhlemann et al, 2017). Digital Twins will likely be immensely important when it comes to developing systems of the future that will be capable of making autonomous decisions about which assets to create (and when to create them) in order to maximise value created — for both owners and customers (Heutger & Kueckelhaus, 2019).

The predictive capability of the Digital Twin is linked to the increasing availability of data and connectivity for devices, as well as advancements within big data analytics. Both short- and long-term predictions may be of value, allowing the business to react to rapid changes and business opportunities as well as creating forecasts and life expectancies of assets given their usage. The sample rate of data has to be set in relation to the stability and predictability of the asset in order to optimise costs, possibly reducing them for long-term predictions (Edström, personal communication, April 16th, 2020).

4.1.4 Challenges when implementing a Digital Twin

Despite the field of research still being relatively young, it is apparent that there are numerous challenges associated with the successful implementation of a Digital Twin. Building upon the issues raised by Modoni et al (2018), Kritzinger et al (2018) and Uhlemann et al (2017) the following list of challenges that are likely to be encountered when attempting to

implement a Digital Twin have been identified.

Improving supply chain visibility within logistics by implementing a Digital Twin

Improving supply chain visibility within logistics by implementing a Digital Twin

A case study at Scania Logistics

YLVA BLOMKVIST

LEO ULLEMAR LOENBOM

Improving supply chain visibility within logistics by implementing a Digital Twin

A case study at Scania Logistics

by

Ylva Blomkvist Leo Ullemar Loenbom

Att förbättra synlighet inom logistikkedjor genom att implementera en Digital Tvilling

En fallstudie på Scania Logistics

av

Ylva Blomkvist Leo Ullemar Loenbom

Table of Contents

Acknowledgements

1 Introduction

1.1 Background

1.2 Purpose

1.3 Research Questions

1.4 Delimitations

1.5 Expected Contributions

1.6 Outline

2 Theoretical Framework

2.1 Supply Chain Visibility

2.1.1 Visibility for sensing

2.1.2 Visibility for learning

2.1.3 Visibility for coordinating

2.1.4 Visibility for integrating

3 Methodology

3.1 Research process

3.2 Data collection

3.2.1 Internal research at Scania

3.2.2 External research

3.3 Source criticism

3.4 Validity, Reliability & Generalisability

3.5 Ethics

4 Results

4.1 Defining the Digital Twin 4.1.1 The definition of a Digital Twin

4.1.2 The technologies enabling Digital Twin

4.1.3 The benefits of a Digital Twin

4.1.4 Challenges when implementing a Digital Twin