Hydrogen System for Heavy Vehicles in Sweden Roles, Actors and Pathways for a Hydrogen Refueling Infrastructure Master’s thesis in Management and Economics of Innovation & Supply Chain Management SAMUEL ANDERSSON AXEL SÖRMAN DEPARTMENT OF TECHNOLOGY MANAGEMENT AND ECONOMICS DIVISION OF ENVIRONMENTAL SYSTEM ANALYSIS CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2021 www.chalmers.se Report No. E2021:029 www.chalmers.se Report NO. E 2021:029 Hydrogen System for Heavy Vehicles in Sweden Roles, Actors and Pathways for a Hydrogen Refueling Infrastructure SAMUEL ANDERSSON AXEL SÖRMAN Department of Technology Management and Economics Division of Environmental Systems Analysis Chalmers University of Technology Gothenburg, Sweden 2021 Hydrogen System for Heavy Vehicles in Sweden Roles, Actors and Pathways for a Hydrogen Refueling Infrastructure SAMUEL ANDERSSON AXEL SÖRMAN © SAMUEL ANDERSSON, 2021. © AXEL SÖRMAN, 2021. Supervisor: Tomas Kåberger, Department of Technology Management and Economics Examiner: Björn Sandén, Department of Technology Management and Economics Report no. E2021:029 Department of Technology Management and Economics Division of Environmental Systems Analysis Chalmers University of Technology SE-412 96 Gothenburg Telephone +46 31 772 1000 Gothenburg, Sweden 2021 iii Hydrogen System for Heavy Vehicles in Sweden Roles, Actors and Pathways for a Hydrogen Refueling Infrastructure SAMUEL ANDERSSON AXEL SÖRMAN Department of Technology Management and Economics Chalmers University of Technology Abstract The cost for renewable electricity has decreased in recent years, making it economically feasible to produce renewable fuels, such as hydrogen, to substitute fossil fuels. Hydrogen can with onboard fuel cells be used to generate electricity to power vehicles, thus constituting a renewable alternative for the transportation sector. Vehicles powered by hydrogen are dependant on a refueling infrastructure in order to be adopted. This study, conducted in collaboration with Volvo Group, has aimed at understanding how a hydrogen refueling infrastructure could develop by conducting interviews with multiple actors expected to be involved in it. It is concluded that the hydrogen infrastructure can consist of a system with both centralized and decentralized production units and a hydrogen production that is neither fully nor non-dedicated for the transportation sector. To manage the high costs associated with the establishment of a hydrogen infrastructure, actors are recommended to collaborate and engage in partnerships to mitigate the costs and risks of such projects. It is therefore not viable with a production unit that is dedicated for any sole purposes. Further, more hydrogen projects should be initiated in Sweden to increase the knowledge level and accelerate the development. Many actors believe that a framework laying out a pathway for the development is needed, something that the public sector could help establish. By undertaking a role of initiating and supporting hydrogen projects, Volvo Group could increase the knowledge about the technology and promote a hydrogen infrastructure development. Keywords: Hydrogen, green hydrogen, electrolysis, hydrogen refueling infrastructure, partnerships, FCEV, fuel cell trucks iv Acknowledgements This thesis constitutes the last part of our Master’s degree in Management & Economics of Innovation and Supply Chain Management. It was conducted during the spring 2021 within the Department of Technology Management and Economics at the division of Environmental System Analysis. The thesis was performed on behalf of Volvo Group to support their development of fuel cell trucks. First and foremost, we would like to thank all the interviewees that allowed us to take some of their valuable time to participate in the interviews and shared their knowledge and perspectives. Your insights have been invaluable for the study and we are very thankful for letting us take part of these. Secondly, we would like to thank all employees at Volvo Group, who reserved time to guide us through the processes and supported us with knowledge and data. A special thanks to all the employees at CampX for having us. We have had the opportunity to meet amazing people and you made us feel at home from the very first second. Finally, we would like to express our deepest gratitude to our main contributors. Tomas Kåberger, supervisor at Chalmers, who has encouraged us to think as engineers and helped us understand not only hydrogen solutions but the entire energy sector. Björn Sandén, examiner at Chalmers, who has provided great academic insights along the way and made us question the work to make it better. Niklas Gustafsson, supervisor from Volvo Group, who initiated the project and through his curiosity and optimism has been a great support during the entire work. This thesis would not have been possible without the knowledge, engagement and continuous support and guidance from all of you and for this we are truly grateful. Samuel Andersson & Axel Sörman Gothenburg, June 2021 vi Contents List of Figures x List of Tables xi 1 Introduction 1 1.1 Aim and Research Questions . . . . . . . . . . . . . . . . . . . . . . . 3 1.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.2.1 Hydrogen Projects in Sweden . . . . . . . . . . . . . . . . . . 4 1.2.2 Renewable Hydrogen Production . . . . . . . . . . . . . . . . 6 1.2.3 Electrolyzer Costs and Development . . . . . . . . . . . . . . 7 1.2.4 Hydrogen Distribution . . . . . . . . . . . . . . . . . . . . . . 9 1.2.5 Hydrogen Storage . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.2.6 Volvo and FCEVs . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.2.7 Swedish Electricity Production and Market . . . . . . . . . . . 12 1.2.8 BEV Infrastructure Development . . . . . . . . . . . . . . . . 14 2 Theory 15 2.1 Innovation and Technological Change . . . . . . . . . . . . . . . . . . 15 2.2 Testing and Development of new Technologies . . . . . . . . . . . . . 16 2.3 Networks and Partnerships . . . . . . . . . . . . . . . . . . . . . . . . 16 2.4 Waiting Game . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.5 Public-Private Partnerships . . . . . . . . . . . . . . . . . . . . . . . 19 2.6 Levelized Cost of Hydrogen . . . . . . . . . . . . . . . . . . . . . . . 19 2.7 Value Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.8 Design Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3 Methodology 22 3.1 Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 3.2 Research Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.2.1 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.2.2 Interviews . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.3 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.4 Research Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 4 Analysis and Findings 30 viii Contents 4.1 Findings from the Interview Study . . . . . . . . . . . . . . . . . . . 30 4.2 The Design Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 4.2.1 The Non-Dedicated Centralized Scenario . . . . . . . . . . . . 33 4.2.1.1 Possibilities and Challenges for the Scenario . . . . . 35 4.2.1.2 Value Chain . . . . . . . . . . . . . . . . . . . . . . . 36 4.2.1.3 Volvo’s Role . . . . . . . . . . . . . . . . . . . . . . . 38 4.2.1.4 The Scenario Over Time . . . . . . . . . . . . . . . . 38 4.2.2 The Dedicated Centralized Scenario . . . . . . . . . . . . . . . 39 4.2.2.1 Possibilities and Challenges for the Scenario . . . . . 39 4.2.2.2 Value Chain . . . . . . . . . . . . . . . . . . . . . . . 40 4.2.2.3 Volvo’s Role . . . . . . . . . . . . . . . . . . . . . . . 41 4.2.2.4 The Scenario Over Time . . . . . . . . . . . . . . . . 41 4.2.3 The Dedicated Decentralized Scenario . . . . . . . . . . . . . 42 4.2.3.1 Possibilities and Challenges for the Scenario . . . . . 43 4.2.3.2 Value Chain . . . . . . . . . . . . . . . . . . . . . . . 44 4.2.3.3 Volvo’s Role . . . . . . . . . . . . . . . . . . . . . . . 44 4.2.3.4 The Scenario Over Time . . . . . . . . . . . . . . . . 45 4.2.4 The Non-Dedicated Decentralized Scenario . . . . . . . . . . . 45 4.2.4.1 Possibilities and Challenges for the Scenario . . . . . 46 4.2.4.2 Value Chain . . . . . . . . . . . . . . . . . . . . . . . 46 4.2.4.3 Volvo’s Role . . . . . . . . . . . . . . . . . . . . . . . 47 4.2.4.4 The Scenario Over Time . . . . . . . . . . . . . . . . 47 4.3 LCOH calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.3.1 Small . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.3.2 Medium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.3.3 Large . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 4.3.4 Sensitivity Analysis and LCOH Discussion . . . . . . . . . . . 49 5 Discussion 52 5.1 The Four Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 5.2 Empirical Impressions . . . . . . . . . . . . . . . . . . . . . . . . . . 54 5.3 How to Develop the Hydrogen Infrastructure in Sweden? . . . . . . . 60 5.4 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 6 Conclusion 64 A Appendix A I B Appendix B III C Appendix C V D Appendix D VII ix List of Figures 1.1 Volvo’s time plan on how to reach a fossil free fleet by 2050. Reprinted with permission from [Volvo Group]. . . . . . . . . . . . . . . . . . . 12 4.1 The design space with the four different scenarios positioned, adapted from Hojčková et al. (2018). . . . . . . . . . . . . . . . . . . . . . . . 33 4.2 The non-dedicated centralized scenario marked in the design space. . . 34 4.3 A potential value chain for hydrogen infrastructure of the non-dedicated centralized scenario. . . . . . . . . . . . . . . . . . . . . 37 4.4 The dedicated centralized scenario marked in the design space. . . . . 39 4.5 A potential value chain for hydrogen infrastructure of the dedicated centralized scenario. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 4.6 The dedicated decentralized scenario marked in the design space. . . . 42 4.7 A potential value chain for hydrogen infrastructure of the dedicated decentralized scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 4.8 The non-dedicated decentralized scenario marked in the design space. . 46 4.9 A potential value chain for hydrogen infrastructure of the non-dedicated decentralized scenario. . . . . . . . . . . . . . . . . . . 47 5.1 The design space marked with the area where it is recommended for actors to position themselves in for future development of a hydrogen refueling infrastructure. . . . . . . . . . . . . . . . . . . . . . . . . . . 61 5.2 The design space showing potential development over time. . . . . . . 62 x List of Tables 1.1 The cost of hydrogen distribution in $/kg, depending on volume and distance (BloombergNEF, 2020a). . . . . . . . . . . . . . . . . . . . . 10 3.1 Summary of the sectors identified as important for the study to cover. 26 3.2 Summary and information about the interviews conducted. . . . . . . 27 xi 1 Introduction Climate change is one of the biggest and most defining challenges in the world today and poses a threat to all humankind (United Nations, 2015). Ever since the industrial revolution, energy systems of the western world have been based on fossil fuels which has increased greenhouse gas emissions and led to climate change (Apostolou & Xydis, 2019; Pivovar et al., 2018). In order to reverse this development, all industries around the world must adapt into using more sustainable energy sources (European Commission, 2019). The transportation sector contributes to 16% of the world’s total carbon dioxide emissions (Roser & Ritchie, 2020) and is thereby no exception to this situation. Especially since both the worldwide share of emissions from heavy-duty freight transport and the number of kilometers driven by road-freight transportation are increasing (Calvo Ambel, 2015; Hydrogen Council, 2017). There will not be one single solution that alone can decarbonize the industry, but rather a system of many new solutions and innovations. One such solution is to power vehicles with hydrogen instead of fossil fuels such as diesel (Çabukoglu et al., 2019; Farrell et al., 2003; Hydrogen Council, 2017). With a fuel cell, hydrogen can be converted into electricity which can be used to power an electric engine. These types of vehicles are called fuel cell electric vehicles (FCEV) and are renewable alternatives to the vehicles that mainly uses fossil fuels today. If the hydrogen used to drive these vehicles is renewable, the only emission from the vehicles is water (Dincer, 2012). Hydrogen has been used in industrial settings since the 18th century and is still regularly used today (International Energy Agency, 2020). Almost all of the pure hydrogen used today, 95% according to Philibert (2017), is produced from natural gas and coal by the process of steam methane reforming, also referred to as SMR. This is according to Santhanam et al. (2017) the cheapest and thereby the most popular method to produce hydrogen, despite it resulting in large emissions of carbon dioxide. However, it is possible to produce hydrogen from renewable energy sources. The main method to do this is through the electrochemical process of electrolysis, where an electrolyzer is used to split water molecules into hydrogen and oxygen (Ursua et al., 2011). Electrolyzers are powered by electricity, and if this electricity comes from renewable sources such as solar or wind power, the production of hydrogen is free from emissions. Therefore, renewable, or green hydrogen as it is often referred to (Dincer, 2012), has a great potential to decarbonize many industries and decrease the emissions of fossil fuels (IEA, 2020a). Despite the fact 1 1. Introduction that academics such as Ogden (1999) and Jensen and Ross (2000) discussed the potential of hydrogen and argued for the feasibility of FCEVs already 20 years ago, they are yet to be commercially successful on a large scale (IEA, 2020a). This is largely due to the lack of infrastructure and high costs associated with both the production of hydrogen and development of such an infrastructure (Hydrogen Council, 2020). However, as the cost of renewable energy and thereby the cost of producing renewable hydrogen has decreased in recent years (Roser, 2020), the interest to use hydrogen to decarbonize processes has increased. Many different industries are eager to develop hydrogen systems, and the transportation sector is no exception (IEA, 2019). Volvo Group, hereafter referred to as Volvo, is one of the world’s largest trucks manufacturers (Statista, 2020) and will play a big role in the transportation industry’s transformation to become more sustainable. They believe that the way to reduce the use of fossil fuels will be to shift towards using a mix of different sustainable alternatives, such as bio-fuels, battery electric vehicles (BEVs) and FCEVs (Volvo Group, n.d.). The different solutions have different characteristics and are best suited for different applications. BEVs take long time to charge and have a limited travel range due to the weight and size of batteries. For example, the electric truck Volvo FE Electric has according to Volvo Trucks (2019) a regular charging time of 6,5 hours and an eventual fast charging of 1,5 hours. Battery electric trucks are therefore suitable for driving short and predictable routes that are more or less the same every day. However, for heavy-duty truck applications that drive long range, have a flexible and variable usage and have a need for fast refueling, FCEVs are believed by some to be the solution to transition to more sustainable vehicles (Manoharan et al., 2019; Ogden, 1999). Kast et al. (2018) argue that fuel cell electric trucks technically already are a feasible solution that do not compromise the performance compared to conventional fossil fuel powered trucks. Combined with the absence of carbon dioxide and green house gases emissions during usage, FCEVs are a suitable option to conventional heavy-duty trucks in order to decarbonize the transportation sector (Manoharan et al., 2019; Ogden, 1999). The introduction and adoption of FCEVs is not only a matter of if and when fuel cell trucks will be developed. Many researchers, such as Çabukoglu et al. (2019), Jensen and Ross (2000), and Langford and Cherry (2012) argue that the introduction and adoption of hydrogen use in the transportation sector is also dependent on the construction of a hydrogen refueling infrastructure. Researchers have for a long time argued that it is the costs of constructing the infrastructure for hydrogen refueling which has hindered the development of FCEVs (Jensen & Ross, 2000; Ogden, 1999). Moreover, Forsberg and Karlström (2007) argue that accessibility of hydrogen refueling infrastructure is crucial in order for FCEVs to be accepted and Minutillo et al. (2021) even state that: “The most important factor in the development of FCEVs in the marketplace is the hydrogen infrastructure, from its production to its distribution" (p. 13668). Çabukoglu et al. (2019), Gim and Yoon (2012), and Langford and Cherry (2012) all describe that the absence of a refueling infrastructure for hydrogen discourages the development of fuel cell vehicles, and the absence of fuel cell vehicles discourages the development of a hydrogen refueling 2 1. Introduction infrastructure. This is sometimes referred to as the chicken-or-the-egg problem (Farrell et al., 2003; Gim & Yoon, 2012). In order for a successful roll-out of fuel cell vehicles, it is necessary that there is a hydrogen refueling infrastructure in place when this happens. Farrell et al. (2003) also argue that it usually takes relatively long time for these developments, which stresses the need to start this work in parallel with the development of fuel cell trucks. 1.1 Aim and Research Questions Given the critical role of infrastructure for a successful commercialization of FCEVs as well as the improved conditions for renewable hydrogen production, the aim of this study is to examine how a possible hydrogen refueling infrastructure can be established in Sweden and to identify the necessary roles and actors that will influence this development. There are many uncertainties on how the new hydrogen value chain will develop as well as different levels of knowledge for different actors. It will therefore be important to understand what roles that will be needed and which actors that could undertake each role to support the development of a hydrogen refueling infrastructure. The study therefore aims to answer the following questions: • How can a hydrogen refueling infrastructure develop in Sweden and what are the factors influencing this development? • Which role can Volvo Group take in the development of a hydrogen refueling infrastructure? 1.2 Background Hydrogen is one of the most used industry gases today and according to the International Energy Agency (2020), the worldwide demand for hydrogen has increased more than threefold since 1975. Hydrogen is primarily used in the industries of oil refining as well as ammonia, methanol and steel production (International Energy Agency, 2020). However, the increased urgency to achieve the goals set up by the Paris Agreement in combination with the rapidly falling prices of renewable energy (Roser, 2020) has made the interest to use renewable hydrogen solutions also for other processes increase. This has made some authors, such as Dou et al. (2017) to speculate about a future hydrogen economy and that we will see a rapid hydrogen development. This increased interest is also shown by authorities and legislators. In 2020, the European Union published their strategy on how to transform the hydrogen potential in the union into reality (European Commission, 2020) and many of its member states have followed and out forth their own strategies. In early 2021, Fossilfritt Sverige (2021) published a strategy suggesting how Sweden should act when introducing renewable hydrogen on a large scale. After the strategy was released, the Swedish Energy Agency was assigned to develop an official national hydrogen strategy by the Swedish Government (2021), which is expected to be presented later in 2021. In the European hydrogen strategy, the European 3 1. Introduction Union announces plans on investing € 470 billion in renewable hydrogen projects to increase the share of renewable hydrogen in EU’s energy mix from 2% today to 14% in 2050 (European Commission, 2020). In the budget for 2021, the Swedish government announced investments of SEK 550 million during 2021 and SEK 600 million in 2022 for infrastructure of electrified transports, in which hydrogen is included (Government of Sweden, 2020). This is an increase from earlier announced investments in infrastructure by SEK 1 billion. Furthermore, there is and has for a long time existed a strong academic interest and belief in the potential solutions achievable from renewable hydrogen (Apostolou & Xydis, 2019; Jensen & Ross, 2000; Manoharan et al., 2019; Ohta, 1979). In addition, Molin (2005) has showed that there is a high public willingness to use hydrogen vehicles. The main barrier that has hindered, and is hindering, the development and adoption of renewable hydrogen solutions is according to Lee et al. (2009) the cost, as green hydrogen production is more expensive than production via SMR. This perspective is shared by the Hydrogen Council (2020) who state that the cost of producing renewable hydrogen is what puts it of reach for everyday use. However, the Hydrogen Council (2020) further argues that the cost of hydrogen production equipment, such as electrolyzers, are expected to halve by 2030. This means that the investment costs of making a transition towards producing and using green hydrogen will decrease. Moreover, the cost of renewable energy is decreasing (Roser, 2020) which leads to that the cost of producing hydrogen from electrolysis will decrease. Fossilfritt Sverige (2021) suggests that locations in the world where the cost of renewable electricity is low will be able to produce renewable hydrogen to a competitive price already by 2022. In the strategy, it is exemplified that one such place is Scandinavia. As Sweden is a country with suitable conditions for renewable hydrogen production and the size of the country’s development provides a reasonable scope in relation to the project limitations, the study will primarily consider the hydrogen transition in Sweden. However, as Volvo is a large and international company operating worldwide, considerations will be made to ensure that findings and recommendations provide long term value for Volvo looking beyond the Swedish hydrogen infrastructure development. 1.2.1 Hydrogen Projects in Sweden As mentioned earlier, the interest for hydrogen solutions have increased greatly in the last years. This is driven by the European Green Deal, an initiative to make Europe climate neutral by 2050 from the European Commission (2019), as well as the hydrogen strategy from the European Commission (2020) where it is stated that € 470 billion will be invested in hydrogen projects until 2050. Despite the interest, the number of real projects have up until recently been low. For example, there only exists a handful of hydrogen refueling stations (HRS) in Sweden (Vätgas Sverige, n.d.), one of which is located in Mariestad. Today, this HRS produces up to 4 000 kg of hydrogen per year to supply 14 fuel cell cars owned by the municipality, but it has a capacity of producing 46 000 kg. The station has a storage capacity of 345 kg of compressed hydrogen at 200 bar. To get the electricity needed, solar panels with a capacity of 250 kW are connected next to the HRS (Alpman, 2021). 4 1. Introduction Another project is Hybrit, which is a joint-venture between the steel producer SSAB, iron miner LKAB and power company Vattenfall, dedicated to develop and implement technology of producing fossil free steel with the use of hydrogen (Hybrit, n.d.). In February 2021, the newly started company H2 Green Steel announced plans of constructing a factory to produce renewable steel with green hydrogen, that should be up and running already by 2024 (H2 Green Steel, n.d.). H2 Green Steel will construct an electrolyzer with a size of around 800 MW, in which it is possible to produce 180 000 kg hydrogen per day. In comparison, the average heavy duty truck in Sweden drove 40 410 km in 2020 (Trafikanalys, 2021). One of the few hydrogen trucks available today comes from Hyundai and has a range of 400 km with a hydrogen on-board storage of 33 kg (Hyundai, n.d.), meaning that if all heavy duty trucks were to be powered with hydrogen, the average Swedish heavy duty truck would have a hydrogen demand of around 3 334 kg/year. This indicates that the hydrogen demand, at least in an initial phase, will be far greater in industrial settings than for the transportation sector. The Swedish Energy Agency did during the spring of 2021 open up the application period to apply for financial support as part of IPCEI from the EU (Energimyndigheten, 2021a). IPCEI stands for Important Projects of Common European Interest and is an initiative from the EU to support cross-border projects that are of certain importance for European development. The Swedish Energy Agency received 21 applications for the IPCEI-call, with the majority coming from power and energy providers as well as industries, showing that at least 21 actors have concrete plans of making investments in hydrogen projects. In 2021, the Swedish Government included hydrogen projects as part of the type of projects that can get support from public funding via Industriklivet (Regeringskansliet, 2020). In April of this year, the Danish company Everfuel, which constructs and operates hydrogen refueling stations, announced plans to construct 15 HRS in Sweden by the end of 2023 (Everfuel, 2021). In a European outlook, Shell plans to build a 200 MW electrolyzer in the port of Rotterdam and in the project NortH2, the company wants to build a 10 GW offshore wind farm that will generate electricity for hydrogen production (Shell, n.d.). Another Swedish example is the steel manufacturer Ovako that has begun to transform production into using renewable hydrogen instead of propane in the heating process. The technique was tested and successfully demonstrated in full scale in 2020 and is to be implemented in 2022 given that Ovako can obtain financial support (Fossilfritt Sverige, 2021). Instead of going through and waiting for a slow application process for public funding, Ovako created a consortium to ensure the funding needed, which Volvo is a part of. Both the strategy from the European Commission (2020) and Fossilfritt Sverige (2021) state that the costs of constructing a hydrogen refueling infrastructure could be reduced by creating local clusters of hydrogen production. These clusters, or valleys as they are also referred to, would be built around areas which consist of industries that consume large amount of hydrogen and thereby could conjointly invest in production facilities of hydrogen, that is used within the cluster. 5 1. Introduction 1.2.2 Renewable Hydrogen Production Hydrogen is the most abundant element in the universe (Field, 1995). However, as only small quantities of hydrogen in its elemental form exist on earth (Lubitz & Tumas, 2007), it has to be produced by means of an energy input (Carmo et al., 2013). As mentioned, almost all of the hydrogen produced today is produced via steam methane reforming, leading to emissions of carbon dioxide (Philibert, 2017). With the use of an electrolyzer, hydrogen can be produced via water electrolysis which is a process which does not emit any carbon dioxide. Troostwijk and Diemann discovered the process of electrolysis in 1789 and it was the main method for hydrogen production up until the 1950s, when hydrogen produced via SMR became cheaper due to the low prices of natural gas (Carmo et al., 2013). In its most simple form, an electrolyzer consist of a cathode, an anode and a membrane. Water reacts on the anode side and the oxygen and hydrogen is then separated by the membrane, leading to that hydrogen gas is formed at the cathode side (Smolinka et al., 2015). A complete electrolyzer system consists of hundreds, up to thousands of such stacks. According to Carmo et al. (2013), there are three different types of water electrolysis technologies that in turn make way for three different types of electrolyzers. These are alkaline water electrolysis (AEM), solid oxide electrolysis (SOEC) and polymer electrolyte membrane electrolysis (PEM). Each technology comes with with its own strengths, weaknesses and characteristics. Carmo et al. (2013) and Schmidt et al. (2017) state that PEM electrolyzers have a short response time which enables dynamic operation and thereby is well suited to be powered with intermittent energy sources such as wind and solar. Furthermore, Schmidt et al. (2017) present that PEM electrolyzers have higher power density and cell efficiency than the other technologies. PEM water electrolysis can also produce hydrogen at higher rates compared to traditional alkaline water electrolysis (Zeng & Zhang, 2010) and is more mature than the solid oxide electrolysis technology which is yet to be commercialized (Schmidt et al., 2017). Moreover, the study by Schmidt et al. (2017) expects that PEM will be the dominant electrolysis technology in 2030. Therefore, the main focus and data used about electrolyzers in this study will be on PEM technology. In an electrolyzer, water and energy in the form of electricity are converted into hydrogen and oxygen (Carmo et al., 2013). The chemical reaction of electrolysis can be seen in Equation 1.1 (Carmo et al., 2013). H2O → H2 + 1 2O2 (1.1) If the electricity is produced from renewable energy sources, such as wind or solar, the hydrogen produced will consequently be renewable. According to Carmo et al. (2013) the production capacity of renewable electricity is increasing, which will reduce its costs. This has also been the case in the last years, as shown by Roser (2020). Since many industries plan to, or already have begun a transition towards renewable hydrogen solutions, the demand for renewable energy will increase greatly. Fossilfritt Sverige (2021) believes 48 TWh/year of renewable electricity will be 6 1. Introduction needed in order to produce the hydrogen needed to supply the planned hydrogen projects in Sweden. In comparison, the electricity generated from wind power in Sweden in 2020 was 27,6 TWh (Energimyndigheten, 2021c), meaning that the capacity of renewable electricity production will have to increase to support the increasing number of hydrogen projects. After hydrogen is produced, it can be fueled into a fuel cell vehicle. Fuel cells convert hydrogen into electricity in a chemical process on board the vehicle. This process is practically the same as the one that takes place in an electrolyzer, but reversed (O’hayre et al., 2016). The electricity generated from the hydrogen is fed into a small battery which can power electric motors from which a truck can be driven. As fuel cells generate electricity through a controlled chemical reaction rather than through combustion, the only emissions from a fuel cell vehicle driven by hydrogen is water (Chan, 2007). As with electrolyzers, there are multiple variants of fuel cells and the main difference between them is what type of electrolyte membrane that is used, with the PEM-technology being the most popular for vehicle applications (Manoharan et al., 2019). 1.2.3 Electrolyzer Costs and Development One of the main components in a hydrogen infrastructure is the HRS. Apostolou and Xydis (2019) present that there are mainly two types of HRS where the difference is whether the hydrogen is produced off-site or on-site. An off-site station gets hydrogen delivered to it and an on-site station produces hydrogen close or in direct proximity to to the site. In financial terms, the on-site stations will require larger investments, mainly due to the additional components needed for producing hydrogen, while the off-site station will include a cost for distributing the hydrogen to the station (Apostolou & Xydis, 2019). A HRS that produces hydrogen on-site includes an electrolyzer which supplies the refueling components with hydrogen. The hydrogen is compressed to a high pressure before being stored at the station. From the storage, the hydrogen is then guided through a cooling unit to lower the temperature and then delivered to the vehicle through a dispenser (Reddi et al., 2017). In order for the hydrogen transition to become feasible, both for the transportation sector and other industries, the cost of electrolyzers need to become sufficiently low so that hydrogen can be produced at a cost that is competitive with SMR production (Proost, 2019). The cost of producing hydrogen through water electrolysis mainly depends on two factors. The operational expenditures (OPEX) that mainly consists of the electricity price (Proost, 2019), and the capital expenditures (CAPEX) which are mostly related to the investment cost of the electrolyzer (Schmidt et al., 2017). According to Schmidt et al. (2017), it is the high capital costs of electrolyzers along with uncertainty regarding future cost and performance development that have hindered investments in electrolyzers. A PEM electrolyzer system consists of an electrolyzer with its stacks, power supply, a deionized water circulation system, components for processing hydrogen, components for cooling and some additional components, such as a compressed air valve (Mayyas et al., 2019). The electrolyzer 7 1. Introduction production industry is still very small and for example, Saba et al. (2018) state that only one or two large electrolyzer systems per year were manufactured as late as in the 1990s and that there still does not exist any mass production of electrolyzers today. Mayyas et al. (2019) thereby state that the two most influential aspects in order to reduce the costs for electrolyzers is manufacturing engineering and economies of scale. According to Gim and Yoon (2012), Minutillo et al. (2021), Morgan et al. (2013), Nguyen et al. (2019), Olateju et al. (2014), Ulleberg and Hancke (2020), and Viesi et al. (2017), electrolytic production of hydrogen can utilize economies of scale. The larger the electrolyzer is, the smaller will the unit price of hydrogen be (Olateju et al., 2014; Viesi et al., 2017). This is because the cost of an electrolyzer does not increase proportionally with the size (Gim & Yoon, 2012; Minutillo et al., 2021; Olateju et al., 2014). For example, Minutillo et al. (2021) show that an increase of the electrolyzer size with 300% leads to a 20% reduction of the hydrogen cost and Morgan et al. (2013) state that 60% of the capital costs could be reduced if all parts of a hydrogen production plant were to be scaled. IRENA (2020) summarizes investment costs of electrolyzers depending on scale from several studies and shows that an increase of the scale by 10 times, from electrolyzers with power in the magnitude of 10 MW to 100 MW can reduce costs by a third, from $ 750/kW to $ 500/kW. The cost of an electrolyzer in the magnitude of 1 GW would then accordingly be $ 400/kW (IRENA, 2020). In addition to the investment costs of the electrolyzer, the cost of producing hydrogen is also determined by the operational expenditures of the production facility. Minutillo et al. (2021) state that the operating costs of an electrolyzer mainly consist of the cost of the electricity to power the facility. The authors further show that the electricity cost impact of the total cost becomes larger as the electrolyzer size increases due to the economies of scale. Minutillo et al. (2021) suggest that large electrolyzers thereby could be particularly suitable in countries with low electricity prices. There are several other factors influencing the cost of producing hydrogen from electrolysis, one of which is the utilization rate of the electrolyzer (Philibert, 2017; Proost, 2019; Ulleberg & Hancke, 2020). According to Proost (2019), the cost of hydrogen increases greatly if the utilization is less than 2 000 hours per year, leading to a cost that makes renewable hydrogen an unviable option. Philibert (2017) discusses that the requirement for high utilization could make investments in electrolyzers particularly risky in an early phase, as demand of hydrogen might not be high enough to motivate a high capacity factor. Related to this, Ulleberg and Hancke (2020) propose that small-scale electrolyzers could prove to be cost-efficient since they have a higher potential to reach a higher number of operating hours and system utilization rate when the demand is low and unestablished. The cost of electrolyzers is also expected to decrease in the future. Since no mass production of electrolyzers is established yet, there exists a great potential for cost decrease by industrializing the manufacturing (Saba et al., 2018). Schmidt et al. (2017) state that production scaling-up alone could lead to a cost decrease of 17-30% and that this is larger than the impact of increased R&D funding. The authors argue 8 1. Introduction that the main drivers for this potential lies in the ability to standardize components and achieving economies of scale in the production plant when the manufacturing of electrolyzers ramps up. Some studies discuss the potential cost reduction of electrolyzers in terms of learning curves (Reddi et al., 2017; Saba et al., 2018). The learning rate indicates how much the cost of production is reduced by every doubling of cumulative capacity produced (Saba et al., 2018). Saba et al. (2018) state that electrolyzer production has a learning rate of 18%, showing a large potential for future cost decrease. Given the increasing demand for renewable hydrogen and thereby electrolyzers, production of electrolyzers is expected to increase which will lead to decreasing costs (Saba et al., 2018). IRENA (2020) states that the stacks of a PEM electrolyzer can achieve a lifetime of more than 50 000 hours, where the factors influencing the lifetime are the operating conditions, variable loads, gas permeation, the anode might dissolute, and the water could be contaminated. However, stacks and systems have different life times since stacks can be replaced when they become worn out. Carmo et al. (2013) state that the system of a PEM electrolyzer is expected to have a lifetime of 20 years (Carmo et al., 2013). 1.2.4 Hydrogen Distribution Distribution of hydrogen is an important subject that will influence the final cost of hydrogen (Demir & Dincer, 2018). Pivovar et al. (2018) argue that a lot of the value in hydrogen solutions lies in the high energy density of the element, meaning that hydrogen has a high amount of energy per unit volume. However, hydrogen has a very low density and is the lightest element in the periodic table, making it difficult to store and distribute (Demir & Dincer, 2018). Demir and Dincer (2018) state that "The viable development of an H2 transmission and distribution infrastructure referred to as ’hydrogen delivery infrastructure,’ is one of the most vital issues for an effective penetration of H2 into the energy system and the commercialization of the hydrogen energy driven automobiles" (p. 10421). There are mainly four alternatives to solve the matter of distributing hydrogen. Firstly, it is possible to transport hydrogen in gaseous form on for example trucks, railroad or ships. The same transportation modes can secondly be used to transport hydrogen in liquefied form. Thirdly, hydrogen can be distributed via a pipeline network. The final alternative is to produce hydrogen where it will be consumed, thus eliminating the need for distribution. In Sweden today, the main method for distribution of hydrogen is in compressed form on trucks (Fossilfritt Sverige, 2021). Liquid hydrogen is suitable when the transport distance is increased, because it has has higher density and allows for more hydrogen to be stored and distributed. To liquefy hydrogen, it does however need to be cooled down to -253°C, something that requires a lot of energy which makes this alternative more expensive (Steen, 2016). From a cost perspective, hydrogen distribution through pipelines is most suitable when the distances are long and the hydrogen flow is large (Fossilfritt Sverige, 2021). Wang et al. (2020) present plans on a hydrogen pipeline network in Europe, since hydrogen might be needed to be distributed to some locations where local production 9 1. Introduction of hydrogen is not suitable. Further, the authors suggest that by 2030, a pipeline network of 6800 km should be up and running and by 2040 the distance should be almost 23 000 km. This establishment of pipelines can be done by partly utilize already existing pipelines that are adjusted to distribute hydrogen. Of this network, 25% should be newly constructed pipelines and 75% should be adjusted from existing ones, which is estimated to cost between € 27-64 billions depending on multiple factors (Wang et al., 2020). Although, pipelines are not established today in Sweden with exception for some local ones for industrial use (Fossilfritt Sverige, 2021). In a report from BloombergNEF (2020a), the costs of distributing hydrogen is presented. It is clear that pipelines are a suitable option for large flows of hydrogen, which means more than 10 tons/day. For smaller amounts distributed, transportation by truck is reasonable. If the distance by truck exceeds 100 km, liquid hydrogen could prove to be economically viable, and over a distances of 1 000 km, liquid forms hydrogen are the best option. The cost for truck delivery up to 10 tons/day with distances up to 100 km with compressed gas ranges from $ 0.65/kg to $ 1.73/kg. These costs are visualized in Table 1.1 below, adapted from BloombergNEF (2020a). It is described whether the hydrogen is distributed as compressed gas, if it is partly compressed and liquid, or completely liquid, depending on distance. It is also shown what distribution mode to use, where pipeline is assumed for volumes greater than 10 tons/day and trucks for less distance. When the distances and volumes are very high, this combination leads to a combination of pipelines and distribution by ship. Table 1.1: The cost of hydrogen distribution in $/kg, depending on volume and distance (BloombergNEF, 2020a). 1 000 t/d 0,05 0,05-0.1 0,1-0.58 0,58-3.00 Pipeline 100 t/d 0,05-0.06 0,06-0.22 0,22-1.82 < 3.00 Pipeline 10 t/d 0,65-0,76 0,68-1,73 0,96-3,87 3,87-6,70 Truck 1 t/d 0,65-0,76 0,68-1,73 0,96-3,87 3,87-6,70 Truck - 10 km 100 km 1 000 km 10 000 km - - Compressed Compressed Part Liquid Liquid - The first method for distributing hydrogen is in gas form on truck. This is, according to BloombergNEF (2020a) suitable for less than 10 tons/day and distances below 100 km. An option when the distance increases is to transport hydrogen in liquid form. Due to higher density, this provides for more energy to be transported with the same volume. However, this also increases the energy demand and makes it more expensive. Liquefying hydrogen thereby requires larger demands, to justify the higher costs that comes from liquefying the hydrogen. Distributing hydrogen as compressed gas is the most common method used in Sweden. When the amount 10 1. Introduction of hydrogen that needs to be distributed is increased in combination with increased distance, pipelines are the best option considering costs (BloombergNEF, 2020a; Fossilfritt Sverige, 2021). The drawback with this alternative is the high investment costs needed to establish such a network, something that might not be viable given the initial small quantities of hydrogen demanded in the early phase. 1.2.5 Hydrogen Storage Hydrogen has the possibility to be stored and used when needed to even out the supply and demand. As hydrogen both can be converted from electricity and be used to produce electricity, it can be used to store electricity and balance the intermittent renewable electricity production, thus creating a stable power grid (Smolinka et al., 2015). This is possible because of the high energy density, and thus can make sure to face the challenges of the high electricity needs in combination with the potential of long-term storage (Steen, 2016). It is possible to store hydrogen in salt caverns and other natural formations, or by constructed caverns. The constructed caverns are relatively expensive. Steen (2016) further mentions the possibility to store the hydrogen in a pipeline. By producing hydrogen from electrolyzers, this can be added into the pipeline network of natural gas which in itself will act as a storage and reduce the need for dedicated hydrogen storage. Fossilfritt Sverige (2021) mentions some examples of hydrogen storage in Sweden. One example presented is the demonstration facility for Hybrit, a project with the ambition to create fossil-free steel, where an underground storage will be built. With a capacity of 100 m3 for the demonstration storage, it is believed that a storage 1 000 times larger can be used to balance the power grid as a complement to the hydro power. In the future, Hybrit will have a capacity of 100 GWh for increasing flexibility as well as matching supply and demand (Fossilfritt Sverige, 2021). 1.2.6 Volvo and FCEVs Volvo strives to do their part to make sure that the goals of the United Nations (2015) will be fulfilled and that their products and services will be emission free by 2050. However, since Volvo’s products have a long lifetime, the transition to only offer fossil-free vehicles needs to be made by 2040 in order to phase out all fossil vehicles until 2050 and meet the goals set up by the Paris Agreement. As mentioned, this will be achieved by offering a mix of BEVs, FCEVs and vehicles with internal combustion engines that for instance are fueled with bio-fuels (Volvo Group, n.d.). Volvo started production of battery electric trucks in 2019 and has set a goal of offering fuel cell trucks in the later half of this decade. An illustration over this plan can be seen in Figure 1.1 below. As seen in the figure, combustion engine vehicles are dominating the share of new trucks being produced today, hence a rapid transition will have to take place. 11 1. Introduction Figure 1.1: Volvo’s time plan on how to reach a fossil free fleet by 2050. Reprinted with permission from [Volvo Group]. To support the development of sustainable alternatives and fuel cell trucks, Volvo has engaged in several projects within the transportation sector. One of the largest projects is the joint-venture with Daimler Truck AG, called cellcentric. The joint-venture has the aim to speed up the development of fuel cells and was launched by the end of April 2021 (Volvo Group, 2021). Another recent project within this field is called H2Accelerate. In this project, Volvo collaborates with truck manufacturers Daimler Truck AG and IVECO as well as oil and energy providers OMV and Shell in order to create the right conditions for a large transition towards heavy FCEVs (Shell, 2020). A third initiative showing the increased activity regarding fossil-free vehicles is Volvo’s recent addition of a new business area called Volvo Energy, which was presented in the beginning of 2021. Volvo Energy will manage electromobility, which is something that includes the work with fuel cells (Volvo Group, 2020). 1.2.7 Swedish Electricity Production and Market A prerequisite as well as the major reason for the increasing interest and demand in hydrogen solutions is access to cheap renewable electricity. The electricity production in Sweden is partially based on renewable energy sources, with the non renewable share mainly coming from nuclear power (Energimyndigheten, 2020). In 2020, the total electricity production in Sweden was 159 TWh, with the majority coming from from renewable energy sources. In that year, the three largest energy sources were hydro power, nuclear power and wind power, producing 45%, 30% and 17% of the total production (Energimyndigheten, 2021c). According to Energimarknadsinspektionen (2021b), wind power is the energy resource with the lowest variable cost. Furthermore, BloombergNEF (2020b) states that wind power has become the cheapest source of electricity generation and the development of new 12 1. Introduction wind power farms is increasing. In Sweden, wind powered electricity production increased by 39% in 2020 compared to the year before, amounting to 27,6 TWh. This was the largest change for any of the energy sources (Energimyndigheten, 2021c). This increase occurred despite many investments and installations of new wind power production units were postponed to 2021 due to the Covid-19 pandemic. Energimyndigheten (2021b) believes that the installed capacity of wind power will continue to increase and that an additional power of 3 000 MW will be installed in 2021. Energimyndigheten (2021b) further estimates that the total electricity production in Sweden will increase to 179 TWh in 2023 and that the production of wind powered electricity will increase to 42 TWh in the same year. Globally, it is expected that renewable energy from wind and solar will grow to catch up and pass coal and oil in terms om production capacity (IEA, 2020b). Thereby, renewable energy sources, mainly hydro and wind power, already have, and will continue to have, a big role in the electricity production in Sweden. The expansion of renewable and thereby intermittent energy sources in Sweden will also likely lead to a more volatile electricity price in Sweden (Svenska Kraftnät, 2015). This increases the need for energy storage, which can be achieved with the use of hydrogen, but it also affects hydrogen production as the production cost will become more variable. Electricity is traded on the marketplace Nord Pool Spot, which is an energy-only marketplace that includes the Nordic and Baltic countries, and is owned by the countries’ transmission system operators (Energimarknadsinspektionen, 2021b). According to Energimarknadsinspektionen (2021a), the actors involved in trading electricity are electricity producers, large electricity consumers and electricity traders. Most of the electricity is traded on Nord Pool’s day-ahead market, also called the spot market, where the prices for the collective system as well for the individual areas are set through an auction process (Energimarknadsinspektionen, 2021b). The final price for the consumer is determined by the electricity price, margin of the electricity trader, taxes and an electricity grid cost where the electricity grid cost consists of one fixed part, and a variable part (Energimarknadsbyrån, 2020). The fixed part represents the cost to get access to the electricity grid and is dependent on how large the outtake is, whereas the variable part represents the cost to transport the electricity and is dependent on how much electricity that is consumed (Energimarknadsbyrån, 2020). In Sweden, electricity used for electrolytic processes are exempted from taxes (Skatteverket, 2020). The electricity grid cost differs depending where in Sweden the electricity is consumed. Furthermore, the average electricity cost has been very stable and only contributed for a small part of the total price for industries with a large electricity outtake in the last years (SCB, 2020). Because of this and the tax exemption, this study will only use data from Nord Pool’s spot price when making assumptions about the electricity price. There are three different types of networks for electricity distribution in Sweden (Energiföretagen, 2021). The first level is the high voltage transmission grid, where long-distance bulk quantities of electricity is transported. The second level is the regional grid which transports electricity from the transmission grid to the third level, the local grid. The local grid is the grid the distributes electricity out to the end-consumers (Energiföretagen, 2021). In Sweden, the majority of the electricity is 13 1. Introduction produced in the northern parts whereas most of it is consumed in the southern part of the country (Energimarknadsinspektionen, 2021b). To cope with this situation, Sweden is divided into four bidding markets. The two northern areas SE1 and SE2 have a surplus of electricity, while the to southern areas SE3 and SE4 have a shortage of electricity. This means that the electricity needs to be transported from the north to the south. Because of limitations to the electricity distribution grid, it is sometimes not possible to transport the electricity demanded, creating bottlenecks that lead to situations where even though it is possible to produce enough electricity to satisfy the demand, it is not possible to transport it to the consumers (Energimarknadsinspektionen, 2021a). This has made both industries and consumers in the southern parts of Sweden worried about the future electricity price and supply, and for example, Göteborg Energi (2021) has highlighted the problem saying that industries and the transportation sector will not be able to make a transition towards being renewable if the electricity grid is not expanded in the region. 1.2.8 BEV Infrastructure Development The introduction of BEVs is also an undergoing transition for the transportation sector which requires the development and construction of a new infrastructure. The adoption and market penetration of BEVs have come farther than the FCEVs and despite the different characteristics of BEVs and FCEVs, the development of a BEV-infrastructure and value chain could indicate the direction for a development of a hydrogen refueling infrastructure. Despite the sales of electric cars accelerating and that almost a third of all new registered cars in Sweden in 2020 where rechargeable (BIL Sweden, 2021), BCG (2021) state that the market still is in a very early stage and describes it as an anthill where participants are scurrying to test opportunities. Patterns are however, according to BCG (2021), emerging and both Arthur D Little (2021) and BCG (2021) present the roles needed for an infrastructure. These roles consist of an electricity provider, a hardware provider, site and asset owners, an operator and a service provider. These roles could be fulfilled by vertical specialists focusing on one role, or integrated actors that fulfill several roles in the value chain. Arthur D Little (2021) believes that the role of hardware and asset ownership are the roles with the highest revenue potential, despite the large investment requirements. Since the goal for the OEMs is to offer as many charging opportunities as possible, the role of an OEM should be to aggregate a system of charging possibilities and offer a convenient and easy-to-use experience (BCG, 2021). Arthur D Little (2021) stresses the importance of handling and participating in the development now, stating that actors who lose time now will have to pay for it many times over in a few years. Furthermore, both Arthur D Little (2021) and BCG (2021) believe that the market will consolidate and that large actors will try to capture larger parts of the value chain. 14 2 Theory This chapter presents theory that supports the analysis and give explanations to the factors that influence the development of a hydrogen refueling infrastructure. The theories included discuss innovation and technological change as well as the phases of technologies, stating that the hydrogen infrastructure development is in an early phase, which is often called the era of ferment. Organizational theories show that it is preferable to organize in networks when working with the development of new technologies, but that the collective action problem could hinder the actors of the network from undertaking the activities needed. Furthermore, the waiting game theory explains why some actors could be hindered from making investments in hydrogen infrastructure. Possible solutions to this are consortia and public-private partnerships, which are presented along with their benefits and drawbacks. Lastly, the equation for calculating levelized cost of hydrogen is presented, along with a description of what a value chain and a design space are. 2.1 Innovation and Technological Change The term creative destruction was coined by Schumpeter (2003) and describes the process of when the emergence of a new technology disrupts an industry, resulting in a new industry that develops and replaces the old one. This is what drives innovation and the society forward (Schumpeter, 2003). The creative destruction creates a time between the introduction and successful diffusion of an innovation, of which a company will be the only one serving the market and temporarily can raise their revenues over their costs. This profit called Schumpeterian rents and the behaviour of seeking Schumpeterian rents is what leads to innovation and technological development (Sautet, 2014). Development and diffusion of new technologies, products and industries tend to follow a logistic curve where their development starts off slow, then grows quickly to lastly shrink over time, corresponding to an S-curve (Anderson & Tushman, 1990; Griliches, 1957). According to Dosi (1982), discontinuous changes and innovations are associated with the emergence of a new paradigm and after the paradigm has been set, continuous changes and innovations then follow this trajectory. Dosi (1982) further claims that this creates cyclical movement where a new technology creates a new paradigm that incrementally evolves until a new and better technology emerges and creates a new paradigm, thus sparking a new life cycle. Abernathy and Utterback (1978) showed these 15 2. Theory dynamics by arguing and proving that the character of innovation of a technology changes depending of for how long it has existed and the maturity of the technology. This work, together with the notion of Schumpeterian rents explain how products and industries evolve as what is often referred to as the product or industry life cycle (Klepper, 1997). This theory can also be conceptualized to a technology life cycle, as technology tends to follow the same dynamics. The first stage of the life cycle is often referred to as an era of ferment (Anderson & Tushman, 1990). The era of ferment is characterized by extensive technology and market uncertainty. The high technical uncertainty is characterized by large variety and change in the technology design, rapid product innovation and an increasing general interest in the technology, something that leads to a lot of new actors entering the market (Anderson & Tushman, 1990). There is also a lack of a powerful prime mover which is able to drive the change (Jacobsson & Bergek, 2004). The phase ends when a dominant design emerges, either through the competitive environment or determined in a collaborative environment (Anderson & Tushman, 1990). 2.2 Testing and Development of new Technologies Developing and bringing new technologies to market is hard work that includes a lot of uncertainties. For example, Karlström and Sandén (2004) argue that it is difficult for new technologies to enter a market and that different actions are suitable in different phases. Eisenmann et al. (2012) state that development of new technologies is related to both resource constraints and uncertainty about the viability of the technology and its adoption. To mitigate this when the technology is in an era of ferment, the authors propose a method they call hypothesis-driven approach which helps to maximize learning and thereby establishment of the technology and guidance to how a dominant design might evolve. The purpose of this approach is to develop hypotheses about the technology and how it can be used, and then test these hypotheses with falsifiable tests (Eisenmann et al., 2012). The authors highlight that the tests should be as close to reality as possible, but at smaller scale to allocate the smallest amount of resources possible. By making such easy but practical tests, real feedback and knowledge about how users view the technology will be generated and this helps reduce the risk of offering a technology that no one wants (Eisenmann et al., 2012). According to Griliches (1957), successful new technologies typically enter a smaller niche market before diffusing and reaching more widespread use. Such niche markets are suitable places to test new technologies in practice and learn about the technology and customer demands before proceeding with the development (Eisenmann et al., 2012). 2.3 Networks and Partnerships According to Powell (1990) companies can organize in three distinct ways, namely using a market, hierarchy or network approach. A network builds on collaboration between actors that are characterized by mutual trust and blurred boundaries, in contrast to the arms-length-relationships in market and vertical integration of 16 2. Theory hierarchies (Powell, 1990). The author state that the network approach thereby allows companies to combine the flexibility and scalability of a market approach with the knowledge transfer of a hierarchy approach. Powell (1990) further argues that networks thereby are especially suitable when working with technological innovation, since this often requires cumulative knowledge. Furthermore, a network approach makes it possible to reduce risk by having several actors sharing it (Powell, 1990). Moreover, Powell et al. (1996) state that innovation takes place in networks of learning, rather than in individual firms. The authors further argue that this is especially prominent in fields with rapid technological development and change. There are some risks associated with organizing in networks. According to Glasmeier (1991), networks are suitable for promoting innovation within an existing technological framework, but it is difficult to achieve the same results in dynamic and fragmented periods of a systemic change. This is largely due to what is referred to as the collective action problem (Glasmeier, 1991). According to Glasmeier (1991), the concept means that all actors in a network must benefit from an action for them to act upon it. In a situation where many members of a network would benefit from an action, but each actor has an associated cost to the action which hinders them from undertaking it, the collective action problem hinders the action from being performed. Glasmeier (1991) further uses the decline of the Swiss watch industry as an example to illustrate the collective action problem, by arguing that the network organization was slow to react and form a single voice to respond and adapt to the new technology of electronic watches. This highlights the difficulties of undertaking a large systemic change in a network. When working in a network it is also important to understand incentives, roles and risks for all the actors involved, and work and set expectations accordingly to this. Adner (2006) states that the success of an innovation not only depends on the performance of the company, but also of the performance, integration and dependencies within the innovation system. The success of future fuel cell vehicles will not only depend on the performance of the trucks, but also of the performance of surrounding factors and complements in the network, such as add-on services, maintenance, the price of hydrogen and the refueling infrastructure. Adner (2006) argues that new networks and ecosystems present risks, one of which are integration risks. According to the author, integration risks are the risks associated with having a solution adopted along the value chain and in order for a network to take form, the entire value chain must be aligned for the product to be fully adopted. This speaks for that there must exist a network of hydrogen refueling tank stations in order for customers to adopt FCEVs, but also that there must exist alignment, agreement and incentives along the entire value chain for the network to develop. It is becoming more important for companies to compete with their network or ecosystem against others (Rese, 2006). Rese (2006) discusses the relationship between an OEM and supplier, and that one advantage of a close relationship is the cost reduction that can be achieved through coordination. Second, it could also lead to a productivity improvement within the network. Majava et al. (2013) further discusses some benefits from partnerships, some being asset flexibility, achieving 17 2. Theory economies of scale, better utilization of resources and reduced risks. But, to get there it is important to show trust and commitment to the partnership, have complementary skills between the firms and to have a proportional risk sharing. In other words, what can be achieved when companies partner up are that costs are reduced, both through coordination but also through higher utilization of assets, and that the risk is shared. Furthermore, Goedkoop and Devine-Wright (2016) state that establishing the trust required for a partnership, a lot of time and resources have to be spent. 2.4 Waiting Game Sometimes, as with the case of Schumpeterian rents, there are large first mover advantages to be reaped by the player that is first to market with a new technology. However, sometimes it could be advantageous to wait and see how the market reacts to a competitor’s launch of a new technology before moving forward with the development. If several companies in the same sector do this, it leads to a passive situation where the actors wait for someone else to make the first move, also referred to as a waiting game (Robinson et al., 2012). Bakker and Budde (2012) suggests that hype of a technology can help to overcome these waiting games. The authors state that hype and increased public interest in a technology can light a spark for actors to increase activity and to be the first one succeeding with the developed technology. However, Bakker and Budde (2012) also emphasize the risk of setback that can follow a hype, leading to a decreased level of innovation. The authors furthermore use the hydrogen development and fuel cell cars as an example to illustrate the risks of hype and according to the authors, hydrogen technology has historically undergone both hype and disappointment because it is a system that depends upon many actors to succeed. Another dimension determining the success of emerging technologies is whether they are environmentally friendly. Bakker and Budde (2012) state that renewable technologies are more likely to get the acceptance of the general public. due to their positive societal and environmental impact. However, these technologies usually perform worse than the conventional alternative, thus leading to a need for public funding to increase performance while lowering the cost. Another example mentioned by Bakker and Budde (2012) related to hydrogen refueling infrastructure, is that the fuel cell cars and hydrogen usage relies on an available infrastructure. Since no actor is willing to establish infrastructure elements unless there are no cars ready to use them and no car manufacturer will develop fuel cell vehicles before the infrastructure is ready, this creates a waiting game and a chicken-or-the-egg situation, as presented earlier. Technological expectations are according to Bakker and Budde (2012) strong when multiple actors share them and they can help to coordinate and decide upon future processes. The more resources and beliefs that are put into the work, the more does the chance of it to be successful increase, thus leading to high expectations. Bakker and Budde (2012) conclude that the actors that put most effort into the 18 2. Theory development tend to gain the most out of it when there is a hype. To manage this, the authors claim that it is not the company itself but rather the actors supporting the company that will do it. An example of this could be to establish long-term agreements or contracts that try to work against the potential disappointment phase and instead continues the hype. Bakker and Budde (2012) lastly comment that the waiting actors can lose competitiveness if the technology is delayed due to the waiting game. 2.5 Public-Private Partnerships When a technology is in an era of ferment, market power alone is not enough to support its adoption. At this stage, government investing in the new technology is especially important (Jacobsson & Bergek, 2004). One solution to this is public-private partnerships, which is a term that refers to the coordinated projects that take place between companies and the state (Linder, 1999). According to Klijn and Teisman (2003), the main idea is that both actors would achieve added value and the important part is the synergy effects that only would be achievable through a public-private partnership. Klijn and Teisman (2003) made a case study on three Dutch cases regarding public-private partnerships. One takeaway is the fact that it is difficult to make a reality of the plans for the public institutions. Usually, these plans are made on national level, but it is the local institutions and actors that are responsible for the actions and to make sure the ambitions are fulfilled. Klijn and Teisman (2003) argue that private actors have a focus on retrieving a market share and profits. Whereas private actors are devoted to consumer preferences and operate on the behalf of their shareholders, public actors are devoted to a public cause. Klijn and Teisman (2003) thereby argue that public actors are more risk averse than private actors, leading to tensions and public institutions trying to reduce risks associated with a project, resulting in a less innovative outcome. However, Un and Montoro-Sanchez (2010) argue that public funding is likely to have a positive outcome on innovations in service industries, since these can complement private funding and help with the complementary resources needed for innovation. 2.6 Levelized Cost of Hydrogen Levelized cost of energy (LCOE) is a measurement used to assess the energy generation cost of different technologies. The LCOE is calculated as the unit of the average total cost of constructing and operating an electricity generating asset, over the electricity generated during the production asset’s lifetime (Hansen, 2019). When the output of a process is hydrogen instead of electricity, the formula of LCOE can be derived into levelized cost of hydrogen (LCOH), which shows the cost of producing hydrogen for different technologies (Corporate Finance Institute, n.d.). There are many different methods and measurements to evaluate the cost of hydrogen, but according to Minutillo et al. (2021), LCOH is considered the most important indicator among such evaluation indexes. The equation to calculate the 19 2. Theory LCOH is shown in Equation 2.1: LCOH = ∑t n=1 CAPEXt + OPEXt (1 + r)t∑t n=1 mt (1 + r)t (2.1) CAPEX and OPEX are the capital and operating expenditures, r is the discount rate, m represents amount of hydrogen produced in kg, n is the lifetime of the system and t is the year in operation. 2.7 Value Chain A value chain is according to Kaplinsky (2000), all the activities that are needed for a service or product to reach a final consumer and beyond, from the beginning to the end. Kaplinsky (2000) presents a very simple example of a value chain consisting of three activities that are design, production and marketing, and within each activity there could be multiple activities. For example, production could include inward logistics and packaging. The author also mentions some important aspects of value chains, where governance is one of them. There are two types of value chains in relation to governance. A buyer-driven chain is when the buyer has an important role regarding governance. It is also possible to have a producer-driven chain, which means the producers have the most important roles, to coordinate and assisting both customers and suppliers in the chain (Kaplinsky, 2000). One important aspect presented by Walters et al. (2008) regarding managing the value chain is to focus on the entire chain instead of some parts. Since multiple organizations could be included in the chain, and these companies in turn develop their work and strategies continuously, the value chain will become more and more complex and the detailed perspective is not suitable (Walters et al., 2008). Although, Peppard and Rylander (2006) argue that the traditional value chain was more applicable to industries some time ago. Instead, since offerings are becoming dematerialized, it is more reasonable to view a value chain as a value network. In the network inter-firm relationships play an important role and value is co-created between actors. For instance, this opens up for alliances or take competitors into consideration, something that is missed in a value chain (Peppard & Rylander, 2006). 2.8 Design Space A design space is, according to Stankiewicz (2000), a technique that can be used to illustrate technological development. It consists of a number of operants that, when they are extended to their limit, create a design space which shows what is theoretically possible in the development (Stankiewicz, 2000). One example is the design space by Hojčková et al. (2018) that uses two dimensions to examine the development of possible renewable electricity systems, P - number of production 20 2. Theory units and G - number of independent grids. The authors can with the help of the design space then develop three extreme scenarios that are used for analyze the situation. Stankiewicz (2000) argues that design spaces can be seen as socially constructed cognitive spaces and that an existing design can be changed and modified by either adding a new dimension or by restructuring an existing dimension. By using a design space, it is possible to create clear representations depending on the different operants, where different positions in the design space creates different possibilities and situations (MacLean et al., 2020). 21 3 Methodology This chapter presents the methodology of the study. First, the research design is presented, showing that this study primarily used a qualitative approach and can be labeled as abductive. Then, a description of the research methods used, literature studies and interviews are presented. This is followed by an explanation of how the data was analyzed through a three step coding process. Lastly, the research quality of the study is discussed. 3.1 Research Design There are according to Williams (2007) three common research approaches that can be used in a study: quantitative, qualitative and mixed approaches. Which of the approaches to use depends on what type of data that is required to answer the research question (Williams, 2007). A qualitative research approach involves discovery (Williams, 2007) and is according to Yin (1994) suitable to answer how and why questions. Furthermore, Creswell and Creswell (2017) and Holme et al. (1997) state that a qualitative research approach is preferred when the study investigates an unknown reality and when the data is unstructured and needs to be subjectively interpreted by the researchers. The purpose of this study was to explore and understand how a hydrogen refueling infrastructure could develop in Sweden as well as what factors that would affect such a development. Because it was an explorative study of a new and unexplored development, a qualitative research approach was deemed as suitable and chosen to answer the research questions. There are several possible qualitative research methods, one of which are interview studies. Turner (2010) states that interviews provide in-depth information about interviewees’ experiences and viewpoints of a certain matter. Therefore, an interview study was an appropriate method to gain perspectives from different actors about their views on what a future hydrogen infrastructure network could look like and develop. In this study, a total of 19 interviews were conducted. The participants mainly consisted of representatives from companies that are believed to be a part of the hydrogen value chain in the future, either directly by producing and consuming hydrogen, or indirectly by offering services required for hydrogen production. To gain a broader understanding of the hydrogen infrastructure development, industry experts as well as representatives from the public sector and authorities 22 3. Methodology on a local, regional and national level were also interviewed. Furthermore, 8 meetings with representatives from Volvo were conducted to discuss their views on hydrogen development and to better understand what factors that will affect the transportation industry. According to Börjeson et al. (2006), it is suitable to develop scenarios when conducting research about future states or descriptions of development. To illustrate the different alternatives for how a hydrogen infrastructure can develop in Sweden, four different explorative scenarios were therefore developed. These were derived from and illustrated in a design space, as Stankiewicz (2000) argues that a design space is an appropriate tool to illustrate and analyze technological evolution. Furthermore, Hedin and Martin (1996) describe method triangulation as the use of multiple methods and states that it is recommended to use it in qualitative studies to enhance the credibility of the findings. To test and evaluate the feasibility of the scenarios that were developed, calculations were made and compared. The measurement LCOH was used to do this, as Minutillo et al. (2021) argue that LCOH is considered the most important indicator of hydrogen production costs. The calculations were conducted to better understand what factors that affect the costs of producing hydrogen and thereby understand what can or will be done, as well as how future external developments will affect the three scenarios. Data for the calculations were compiled in a literature study as well as from some of the interviews. This means that the study uses a mixed method approach, which is primarily qualitative but with some quantitative elements. Bell et al. (2018) make a distinction between inductive and deductive research approaches. A deductive approach derives hypotheses from existing theories or knowledge, which then are tested on the reality and either confirmed or discarded. An inductive approach on the other hand, aims at generating theories or hypotheses from empirical observations or findings. The two different approaches can be combined and complement each other, which is called an abductive research approach (Bell et al., 2018). This study used interviews as the primary data collection method, where the standpoints and perspectives from the interviewees were gathered and then summarized and generalized into recommendations for the development of a possible hydrogen refueling infrastructure. Rather than starting off with an initial hypothesis that was later tested, the subject of hydrogen refueling infrastructure was explored, which are clear indications of an inductive research approach. However, the study also consisted of deductive elements. The scenarios that were developed were not only analyzed based on the answers from the interviewees, as the feasibility of the scenarios also were evaluated by calculating the potential costs for hydrogen production in the different scenarios. This can be seen as a form of deductive hypothesis testing. Given that the study consists of elements that are both inductive and deductive, it should thereby be classified as an abductive study. The general research approach of the study is similar to the definition of grounded theory by Easterby-Smith et al. (2018). The authors describe grounded theory as a qualitative research approach where theory is generated based on the empirical 23 3. Methodology findings. This approach is usually based on interview studies where the researcher iteratively collects and analyzes interview data to enhance the understanding of an unexplored area. Bitsch (2005) thereby states that this approach can be used when the study originates from a broad research area rather than specific research questions. Taylor et al. (2015) argue that a grounded theory approach allows the researcher to choose new and adjust data collections during the data collection period. This study did not originate from a clear hypothesis or research question, and the process of collecting data and conducting literature studies was worked with iteratively, where the scenarios and the design space were adjusted as the sample size increased. Therefore, it could also be argued that elements of a grounded theory approach was included in the study. 3.2 Research Method There were mainly two qualitative methods used to collect data for the study. Firstly, a literature review was conducted to obtain knowledge about hydrogen technology and the current state of its development. The main method to collect data was however to conduct 19 semi structured interviews. 3.2.1 Literature Review In order to compile the background and theory section, a literature review was conducted. The purpose of the literature review was twofold and was thereby conducted in two phases. Firstly, it helped to set the context of the study and to gain knowledge about the current state of hydrogen applications, as well as understanding what research that had been conducted on the topic. The knowledge obtained was used both to develop questions for, and to analyze the interviews. The second phase of the literature review was done in parallel with the data collection and was conducted with the purpose to construct the scenarios to illustrate the empirical findings and to find theories to base the analysis of these scenarios on. The literature review included scientific articles, reports and press releases related to hydrogen production and use in both industrial applications and transportation. National hydrogen strategies, law acts, publications and websites from public sector organizations were also included. Webinars and online presentations about hydrogen were also attended. Articles were found by using scientific databases such as the one of the Chalmers Library, Google Scholar and Scopus. Moreover, articles were recommended by the supervisors. Literature was also found through the use of snowball sampling, i.e. looking at the reference lists of already selected articles (Wohlin, 2014). Since the technology of producing hydrogen with electrolysis has existed for close to 300 years but has experienced an increased interest in the last years, newer articles that were written in the past 10 years were premiered and preferred in the study if possible. However, older articles that were widely cited were also included. Furthermore, since hydrogen usage is a very current research area where new things and news are published every week, the literature review was worked with iteratively and updated during the entire writing process. 24 3. Methodology 3.2.2 Interviews The main data collection method used in this study was interviews with representatives from companies and organizations that will be involved in the hydrogen transition. The purpose of conducting interviews was to get qualitative data and subjective perspectives to be able to understand what different actors believe and how they plan to move forward working with hydrogen solutions. There are mainly three types of interview studies that can be used: structured, semi structured and unstructured (Bell et al., 2018). Given the exploratory nature of the study and that the purpose was to obtain the respondents’ opinions and perspectives, a semi structured approach was used. When using a semi structured approach, a predefined questionnaire is used but it is possible to mix the order of questions based on the answers from the interviewee, and ask follow-up questions that are not predefined by the questionnaire (Denscombe, 2009). Thereby, a semi structured approach allows for more flexibility than a structured, which was deemed necessary for this study. According to Patel and Davidson (2003), semi structured interviews also allow the respondent to declare thoughts and opinions without the influence from the interviewer, which gives answers that are not contaminated by the thoughts and hypotheses of the interviewer. The use of a semi structured approach was also deemed as an appropriate mix of efficiency and exploration. When using a semi structured approach, it is possible for the interviewee to bring up solutions or perspectives that could have been missed if a structured approach with a predetermined path would have been used (Gill et al., 2008). On the other hand, Gill et al. (2008) further elaborate that unstructured interviews risk spending a lot of time talking about areas that are not of interest to the study, since they are conducted completely without a questionnaire or predetermined subjects to discuss (Patel & Davidson, 2003). Marshall (1996) describes three different sampling techniques for a qualitative study: convenience, judgement and theoretical sampling. Convenience sampling is a technique were convenience is premiered and the most accessible subjects are selected, whereas judgement sampling involves a little bit more thought as the researcher actively selects the sample to answer the research questions. Theoretical sampling is an iterative process where theory that emerges from the data is used to select a sample to examine that theory. Marshall (1996) further argues that there is an overlap between the three broad categories. As several industries are looking to make a hydrogen transition, it was of particular interest for this study to include views and perspectives from as many different industries as possible. Therefore, a judgement sampling approach was chosen to select the interview sample. The first step of the sampling was to identify what industries and actors that will play a role in the hydrogen transition, which was done with the help of the literature study. After a breakdown of sectors had been done, the sampling began by contacting organizations in each sector, with the goal to cover at least one interview from each sector. Contact information to the respondents was either found online or with the help from the supervisors. A summary of the different sectors identified as important for the development of a hydrogen refueling infrastructure can be seen in Table 3.1. The study also utilized what Marshall (1996) refers to as snowball sampling, where 25 3. Methodology the interviewees were asked if they knew or had contact information to someone that could be interviewed for the study. By using this approach, it became easier to get in contact with the correct person at each organization. Based on some interviews, new actors and industries that will have a use of hydrogen were discovered. It could thereby also be argued that elements of theoretical sampling also was used in the study. Table 3.1: Summary of the sectors identified as important for the study to cover. Industry Role & Use of Hydrogen Steel Hydrogen Production & Consumption Chemistry Hydrogen Production & Consumption Refinery Hydrogen Production & Consumption Energy Provider Energy Provision & Hydrogen Production System Integrator Construct & Operate Electrolyzer and/or HRS Transportation Hydrogen Production & Consumption Forestry Hydrogen Production & Consumption Public Sector (local) Hydrogen Producer & Coordinating Activities Public Sector (regional) Investments & Coordinating Activities Public Sector (national) Investments, Set up Regulations & Coordinating Activities Industry Experts Knowledge In total, 19 interviews were conducted which accounted into more than 14 hours of interview time. The interviews were based on the template that can be found in Appendix A, albeit the order and follow up questions differed between the interviews. All of the respondents had a role which meant that they, in one way or another, worked with hydrogen and the development of hydrogen infrastructure. Information about the interviews is summarized in Table 3.2 The main focus of the interviews was to get information about how the organization that the interviewee represented works with hydrogen, their approach towards cross-sectoral collaboration and what they thought about the future development in terms of value chains, roles and hinders. Both the authors of this report participated in the interviews and all of the interviews except one were held online via video-conference software. The single interview that was not held online was instead conducted over phone. All but one interview were held in Swedish, which instead was held in English. The interviews were recorded with the permission of the respondents to ensure that the responses 26 3. Methodology were captured correctly. The parts of the recordings that were believed to be most relevant and of interest for the study were then transcribed to enable efficient data analysis. Table 3.2: Summary and information about the interviews conducted. Organization Industry Date Duration Vätgas Sverige Industry Expert 2021-02-24 53:36 REH2 Transportation 2021-03-09 53:31 Ojnveden Industry Expert 2021-03-15 29:30 Euromekanik System Integrator 2021-03-15 31:38 Rabbalshede Kraft Energy Provider 2021-03-16 53:47 Vattenfall Energy Provider 2021-03-17 44:48 Nilsson Energy System Integrator 2021-03-19 30:51 Mariestads kommun Public Sector (local) 2021-03-22 54:33 Ovako Steel 2021-03-22 36:40 Fossilfritt Sverige Public Sector (national) 2021-03-22 54:33 Västra Götalandsregionen Public Sector (regional) 2021-03-23 01:08:49 Lindholmen Science Park Industry Expert 2021-03-25 52:13 Preem Refinery 2021-03-25 29:33 SCA Forestry 2021-03-29 01:01:30 Höganäs Steel 2021-03-29 27:26 Port of Gothenburg Transportation 2021-03-30 53:19 Stena Teknik Transportation 2021-03-30 30:48 Shell Energy provider/Refinery 2021-04-07 45:41 Nouryon Chemistry 2021-04-20 54:35 Total interviews: 19 Total time 14:06:09 27 3. Methodology 3.3 Data Analysis Data was gathered from 19 different interviews. According to Easterby-Smith et al. (2018) it is possible to identify multiple codes when analyzing qualitative data. These codes could be summarized and divided into categories, which in turn could be developed into concepts or themes as they are interpreted. Bitsch (2005) has a similar view to this, called open, axial and selective coding, which refers to three levels of analysis. Open coding is mainly about identifying different units, similar to the first step as suggested by Easterby-Smith et al. (2018). Axial coding and selective coding can also be related to the other steps presented by Easterby-Smith et al. (2018). These stages were used to analyze the data gathered from the interviews. All interviews were recorded and thereafter summarized and transcribed. The interviews were divided between the authors and summarized individually. When summarizing, the major aspects were noted and if some ideas or thoughts of the interviewee were especially interesting, this was marked with a comment. For instance, small talk between the participants was not included in the transcription, but otherwise the majority of the data from the interview was noted. The authors then went through the transcriptions together and picked out interesting parts and answers, each of which were given a code word. After this phase, which can be seen as the first step of coding, Easterby-Smith et al. (2018) suggest that the codes should be developed into categories. This is when code words that relate to a similar topic are grouped together. In this study an example of a category is costs, which includes all type of costs, both investment costs and operating costs, related to the renewable hydrogen transition. The final coding step from Easterby-Smith et al. (2018) concerns the development of concepts or themes, based on the categories. Continuing with the example of the cost category, it was at this stage that it was linked with the hinder category, representing that the interviewees view high costs as a potential hinder to the development of an infrastructure. For this study, the codes and categories were analyzed further in Microsoft Excel. This was done by entering all the interviews horizontally and all codes and categories vertically. This creates a grid, in which all aspects were marked depending on what each interviewee believed. The marks covered for instance if the actor believed in the concept, if it was important or if it was seen as a challenge. All codes and categories thus received a score which shows the most important aspects of the hydrogen infrastructure development. 3.4 Research Quality Bell et al. (2018) believe that trustworthiness is the main criteria when evaluating the quality of a qualitative study. According to Bell et al. (2018) and Shenton (2004), trustworthiness has four aspects: credibility, transferability, dependability and confirmability. Credibility relates to internal validity of the study and that it is ensured that the study tests or measures what is intended (Shenton, 2004). Credibility is ensured for instance by providing the findings from the study to the people that have been participating (Bell et al., 2018). Further, the main topic of 28 3. Methodology research has been presented to all interviewees in order to increase credibility and the interviews were recorded to decrease the risk of missing or misunderstanding information. Shenton (2004) states that using well established research methods, method triangulation and debriefing results with superiors, all of which has been used in this study, are ways to increase the credibility of a study. Transferability refers to the external validity and to what extent the study can be used in the future for other researchers to compare their studies against (Bell et al., 2018; Shenton, 2004). This was done by providing contextual information about subjects such as the number of and information about the participants involved in the study, data collection methods and the amount of data collected, as is suggested by Shenton (2004). Ideally, this would make it easier for other researchers to understand the context and thus, making it possible to use this study for comparison against their own studies. Dependability of a study can be determined in terms of the documentation that is provided about how the study was designed and implemented (Bell et al., 2018). Being able to audit the study and conclude whether it has been conducted in a proper way influences the dependability. Bell et al. (2018) suggest that complete documentation from the study and its processes should be presented. In this study this has been done by providing a list over the interviews and what type of actors that were interviewed, along with a template of the interview guideline as well as a description of how the data collection and sampling was conducted. The fourth and final aspect of trustworthiness is confirmability which can be described as the influence of the authors’ own thoughts to the study (Bell et al., 2018). Shenton (2004) discusses that it is difficult to ensure real objectivity in qualitative studies, as the intrusion of biases are inevitable. To make sure this was not the case, the interviews were recorded, summarized and then discussed amongst the authors so the understanding was equal from each interview. This has been the same for all processes that have been done in parallel between the authors. For instance, if coding has been done by one of the authors, this has been discussed when it was done to make sure the other author believe the result is reasonable. Furthermore, in line with the recommendations from Gioia et al. (2013), the report clearly describes the trail of how data was collected to how it was analyzed and the conclusions were drawn. In studies like this, it is possible that a lot of the work is divided between the authors in order to increase efficiency. Although, this has not been the case for this study, as both authors have been eager to be involved in all parts of the work. This means for instance, that both authors have been present during all interviews and conducted them together. For smaller parts that have been done individually, for instance transcribing and summarizing interviews, these have later been discussed to make sure both authors were involved and up to date. Therefore, the work has always been made in parallel, and no author is more or less responsible for different parts of this study. The report has been written jointly and the the results are fully backed by both authors. 29 4 Analysis and Findings In this chapter, the findings from the interviews are presented. First, the themes that emerged from the coding of the interviews in the data analysis are introduced. This is followed by the development of a design space in which four scenarios are presented. Based on the interview data and theory from the literature study, strengths, weaknesses, roles in the value chain and feasibility of each scenario are presented. In relation to this, it is further analyzed how Volvo could be involved in the development. Lastly, calculations of the LCOH for hydrogen production and distribution are presented and discussed. 4.1 Findings from the Interview Study The most important aspect regarding the hydrogen infrastrcture development is the need for collaborations between actors. Almost half of the interviewees that discussed the need for collaboration also emphasized that this was crucial. Related to the collaboration aspect, it is also important to have a system perspective. This means that actors should not focus on internal processes, but try to find ways in how the entire system could function. Having a system perspective was also one of the other main takeaways from the interviews, and by doing this a majority of the interviewees argued that synergies would be easier to achieve. Although, the question regarding which actor that should coordinate this was brought up with no clear answer. Some argued for the actor initiating the project to coordinate it, while some argued for a public actor to take that role. One aspect brought up in one of the interviews is the potential power imbalance that might appear, and argued for a third party to be the coordinator. Further, similar on the topic of collaborations, the value chains that could develop can take many shapes and this was agreed by a majority of the interviewees. A few of them also mentioned this as a challenge, especially since hydrogen is applicable in many different settings. The second most discussed aspect was that the public sector should establish a framework for how hydrogen solutions could develop in Sweden. Within this area, it was both believed and expected that requirements regarding emissions will become tougher, something that would make hydrogen solutions more attractiv