Outdoor air pollution is linked to an estimated 4.2 million deaths each year worldwide (1). Tailpipe emissions from conventional internal combustion engine (ICE) vehicles are a major contributor to urban air pollution, and as such have been subject to ever tighter legislation for decades, requiring increasingly innovative improvements and catalytic emissions controls. We have now reached the point where a move away from the ICE is required to continue air quality improvements, with several countries going so far as banning new purely ICE vehicles in the coming years. This is where EVs will play their part – both pure EV and hybrid systems powered by LIB technologies, as well as fuel cell technologies, are set to see increased uptake and demand as we strive for cleaner air. In this article, we will add to the automotive-focused literature (2–4) and review what technologies are required to drive the uptake of pure EVs, and what the industry is doing now to respond to consumer requirements as this market rapidly grows.

There are several characteristic battery parameters that it is important to consider and contrast with consumer behaviours and expectations for automotive applications: perhaps most significant, the energy or capacity of the cell equates to the ‘miles in your tank’, and is an area where EVs have lagged behind the ICE in previous years. This is evolving, with the most successful EVs on the market now having an average range of 350 km (5). Range anxiety, equating to energy density, is a major theme for the battery materials industry, with contributions from and innovations required in three areas: the cathode, anode and electrolyte. Cost is also an important factor; as well as the material costs for the active components, analysis has shown that the electrode thickness within the cell is a major contributor to automotive cell costs (6) – materials with increased volumetric energy density are therefore additionally attractive from this perspective. There is also the practical cost benefit afforded by developing systems that can operate at higher voltage cut-offs (7), owing to the usable advantages, towards which multiple cell components can be developed and optimised. Herein, we review one topic of significant industry focus from each area: high-Ni cathode materials, with lithium nickel manganese cobalt oxide (NMC) 811 and beyond being commercialised within the next three years; high energy silicon anode technologies, expected to be at commercial scale in the next three to five years; and solid-state electrolytes, with significant progress expected from the next five years and beyond.

Whilst the cathode active material technology landscape remains diverse, with no one material that will meet all EV requirements, the general trend for passenger EVs is using high-Ni NMC, and lithium nickel cobalt aluminium oxide (NCA) materials. The layered Li Ni oxide (LNO), has been studied for the past 25 years, ever since the commercial application of the isostructural Li Co oxide (LCO) by Sony, Japan, in 1991; the relative low cost of Ni compared to Co was an initial driver for this work – and continues to be a factor today (8–12). Until relatively recently, automotive industry uptake was focused on lower Ni NMC variants, such as LiNi_1/3Mn_1/3Co_1/3O₂ (NMC 111), and lower energy chemistries such as Li Mn oxide (LMO), and Li iron phosphate (LFP). Tesla, USA, bucked the trend; as an early adopter of higher‐Ni NCA materials, it was ahead in the EV mileage stakes. Now, driven by consumer demand for more range, high-Ni is in vogue – the key for research and industry alike is to innovate-out the technical problems associated with LNO regarding its stability.

LNO tends towards non-stoichiometry, owing to the relative instability of Ni³⁺ compared to Ni²⁺, and the similar ionic radii of Ni²⁺ (0.69 Å) and Li⁺ (0.73 Å) (12, 13). It has been shown that synthesis conditions are key to prevent the formation of Ni²⁺ anti-site defects, with near-stoichiometric LNO requiring control of calcination temperature, atmosphere and Li content (12, 14). LNO is also known to undergo several phase transformations on electrochemical cycling; whilst a capacity of over 200 mAh g^–1 can be achieved, these transformations lead to significant capacity fade over the first cycles (15). Early research showed the benefits of incorporating relatively small amounts of other metals, most notably Co, Al and Mn, into the structure to impart stability and significantly improve capacity retention. Owing to the isostructural nature of its end members, all compositions in the series LiNi_1–x Co_x O₂ (x = 0–1) can be formed; Co³⁺ imparts stability by hindering the formation of Ni²⁺ anti-site defects (16). Conversely, doping Mn into the LNO structure has been shown to detrimentally effect the reversible capacity but to impart thermal stability benefits – a key property for battery safety (17, 18). The beneficial effect of Al substitution at low levels is two-fold: an improvement in capacity retention by minimising detrimental phase transformations and an increase in thermal stability (18, 19). There is, however, a limitation to the amount of Al that can be usefully incorporated into the structure; the addition of high-levels of an electrochemically inactive dopant will result in a reduction in capacity, and Al³⁺ has been shown to segregate and create localised defects within the lattice, due to the different ionicity of Al–O and Ni–O bonds (20).

This combined work has ultimately led to continued focus on the multiple metal dopant strategies found in NCA and NMC, where greater benefits are observed than in single dopant systems. Whilst not as catastrophic as those in LNO, NCA and high-Ni NMC materials (such as NMC 811) undergo significant structural changes on cycling, which their lower Ni counterparts (for example NMC 622, NMC 111) do not (Figure 1): at high states of charge, a transformation from the second hexagonal phase (H2) to the third hexagonal phase (H3) occurs in high-Ni materials that is associated with c lattice contraction and capacity fade (21–23). The addition of dopants to the bulk structure of LNO such as cobalt, manganese, aluminium, magnesium, titanium and combinations thereof has been shown to influence stability by affecting the volume change on cycling associated with the H2/H3 phase transformation (24–26).

Fig. 1.

Coating strategies have been employed to high‐Ni NCA and NMC systems, providing benefits in two key areas: handling and performance. The handling and processability of high-Ni materials is a well-known challenge, with surface reactivity towards the ambient resulting in the formation of Li hydroxide and Li carbonate impurities, and the resultant propensity of electrode slurries to gel: this creates obvious challenges before materials have even reached the cell (27–29). Once in the cell, these surface impurities contribute to resistance growth and side reactions resulting in gassing (30, 31). Moreover, the high-Ni surface itself is known to undergo phase changes upon cycling, with the formation of the rock salt phase Ni oxide also contributing to instability and capacity fade (32, 33). In its simplest sense, the application of an inactive coating such as Al oxide passivates the surface with respect to these undesirable side reactions, creating more benign materials that are easier to handle; but only so much of this type of coating can be applied before either significant capacity loss or resistance gains are observed (34). As such, the move toward active coatings, where the removal of an inherent risk of capacity loss does not limit the amount or depth of coating that can be applied, is very attractive. A notable example in this area is the extensive work by the Sun group, who have developed several generations of active coatings and complex morphologies for high-Ni materials (Figure 2): starting with a core@shell strategy, a low-Ni NMC was applied to the surface of a high-Ni NMC, creating a system that combined a high-energy core with a high-stability surface and building a system that was electrochemically active throughout (35). The drawback of this system was the observation that the shell layer broke away from the core on cycling, due to the mismatched volume changes within the core and shell NMC layers. To counteract this, the group developed a gradient coating strategy, whereby a lattice expansion or contraction mismatch was avoided by creating a continuous region of gradual compositional change, thus removing a core@shell interface (38). The Sun group further extended this work to look at deeper and multi-component gradients and their potential benefits (36, 37, 39). Such gradient systems can be viewed as a sophisticated hybrid between bulk doping and surface coating strategies, helping to mitigate the trade-offs associated with each strategy alone.

Fig. 2.

These gradient systems demonstrate the importance of considering morphology and process alongside composition in materials engineering. Another area of interest is the mitigation of microcrack formation through the control of primary particle shape, size and interfaces; fewer cracks means a more stable cathode electrolyte interface (CEI) layer, alleviating resistance growth and gas-generating side reactions (33, 40, 41). Most recently, this has led to particular interest in single crystalline morphologies, which promise greater long-term cycling stability compared to their polycrystalline counterparts by minimising the number of interfaces where microcracks can occur. The majority of published research in this area has focused on lower-Ni NMCs (i.e. NMC 622 or less), where reduction in gassing has been observed compared to polycrystalline counterparts, albeit at the cost of rate capability (42, 43). This lower Ni focus is in part due to the challenging nature of high Ni synthesis at the typically elevated temperatures required to form single crystalline materials compared to those used to generate polycrystalline materials. There are examples demonstrating similar advantages for a single crystalline morphology with up to 80% Ni content and efforts are clearly growing in this area: single crystalline NMC 811 has been shown to exhibit less gassing than its polycrystalline counterpart during high temperature storage (30). Zhu et al. undertook a broad study looking at NMCs from NMC 111 to NMC 811 prepared by multiple approaches and demonstrated the need to tune synthesis conditions to Ni content (44).

The engineering opportunities to overcome the challenges presented by high Ni materials continue to grow. As the automotive industry strives for higher energy, the drive to increase the Ni content of NCA and NMC type materials is clear – the common theme across the industry is to move from NMC 622 to NMC 811 and toward 90% Ni content to meet energy requirements, but also to reduce the Co content required, due to sourcing and cost challenges. Ultimately, a combination of the strategies reviewed above are required to develop and commercialise materials with a Ni content of 80% and above to meet the energy and stability requirements of the automotive industry.

Aligned with the drive toward higher energy cathode materials, there is a requirement to enhance and optimise LIB anode materials toward greater energy density, improved cycle life, lower cost per kilowatt hour and improved gravimetric and volumetric densities (3, 46). In particular, the use of higher energy cathode materials allows increased ampere hour per geometric area and volume of active cathode which is important to retain realistic active material loadings and thicknesses and achieve battery EV (BEV) cell and pack targets. A commensurate improvement in storable energy per area and volume of anode electrode is therefore also required. Cell manufacturers and original equipment manufacturers (OEMs) are increasingly moving beyond todays natural and synthetic graphite materials (or combinations of these) toward blending graphite with a higher energy density Si or Si oxide component to enhance cell level energy gravimetric and volumetric density (47). Table I illustrates examples of such Si containing materials (48).

The high natural abundance of Si and low operating voltage (0.2 V discharging potential compared to Li/Li⁺) single out Si as a highly promising anode material for LIBs (49). However, Si containing materials as battery anodes exhibit a number of challenges, with the greatest of these being significant volume expansion during the lithiation process (see Table I). Particle cracking or fragmentation, loss of electrical contact, ongoing parasitic reactions between electrolyte and ‘fresh’ surfaces, cell swelling and gassing all contribute to cycle life issues (see Figure 3 and Figure 4) (46). Various approaches can be deployed to address the volume change issue for pure Si anodes, including nano-engineering of the Si electrode structure (nanowires and nanoparticles, formation of secondary agglomerates) along with advanced binder combinations to create a flexible electrode structure (46, 50, 51). The addition of carbon dioxide into pouch cells has also been trialled to limit parasitic reactions (52). Formation of nanocomposites of Si–C via mechanical or chemical deposition processes, addition of other alloying components or the choice of a SiO_x material (where first cycle lithiation allows an irreversible reaction creating stabilising LiO_x and Li silicate components within the structure) can all bring improvements (50, 53). Incorporation of conductive carbon also addresses the challenge posed by the intrinsic low conductivity of Si containing materials (54).

Fig. 3.

Fig. 4.

A strategy of blending Si containing materials with existing graphite types is already in progress to achieve moderate capacity increase and lessen volume change, as illustrated by cell level calculations for this approach (for example Si:C 1:3 with capacity of 1100 mAh g^–1 by Andre et al.) (3, 47). Table I illustrates an additional challenge present in Si containing anodes in the form of lower first cycle efficiency (FCE) vs. graphite, related to reactions consuming Li between the electrolyte and anode, the formation of the SEI and associated reduction in useful Li inventory in the working cell, reducing effective watt hour per kilogram. Pre-lithiation approaches, where sacrificial Li containing materials are added to the Si anode during electrode fabrication or strategies such as electrochemical pre-lithiation of formed electrodes ahead of cell assembly are possible (55, 56) along with chemical pretreatments ‘artificial SEI formation’ (57, 58). However, these all represent additional steps and cost in a cell manufacturing process, also pre-lithiated materials and electrodes and Si nanoparticles require careful handling due to the reactivity of the materials with moisture and air (48).

Careful optimisation of the liquid electrolyte additives is also crucial to achieve prolonged cycle life and good FCE, with fluorinated additives, especially fluoroethylene carbonate (FEC), showing benefit (59). The discharge and charge voltage profile of Si containing anodes is slightly different to graphite-only examples, leading to reduced chance of Li plating during charging in Si anodes, but typically slightly lower discharge voltage with graphite, thus adjustments to cell balancing and understanding of the operational state of charge window in the usable voltage range are important for full cell (60).

Assessment of the sustainability of changing to Si containing anode components and advanced higher energy cell chemistries is also vital as electrification of the power train advances worldwide (61).

A further driver to increase the energy density of cells is to replace existing anode materials with metallic Li. Li metal was used as the first anode material in rechargeable Li-ion cells due to its very high energy density (3860 mAh g^–1) and low electrochemical potential (–3.040 V vs. the standard hydrogen electrode). However, numerous challenges prevented its widespread adoption, including low cycle life predominating from issues such as the formation of dendrites and unstable solid-electrolyte interfaces. Recently, there has been increasing investigations into using solid-state electrolytes to mitigate the challenges of using metal anodes, whilst maintaining their advantages.

In addition to potentially enabling the use of Li metal anodes, the evolution to solid state batteries has other advantages to conventional Li-ion cells (62). The primary reason is the displacement of the highly flammable cocktail of organic electrolytes that is used currently. This both reduces the risk of unwanted thermal events in the instance of cell misuse or damage, but it also results in a simpler packaging, further increasing the energy density (63) (Figure 5). In addition, solid state materials could offer increased electrochemical stability windows in comparison to existing organic electrolytes; potentially enabling alternative materials, such as higher voltage cathode materials, to be deployed.

Fig. 5.

The use of polymers as electrolytes in batteries was first pioneered in the 1970s (64, 66). This enables cells with high degrees of safety to be manufactured in various form factors. Polymer-based systems such as polyethylene oxide (PEO), polyvinylidene fluoride (PVDF), polyacrylonitrile and polymethyl methacrylate (PMMA) based electrodes have all been widely studied as polymer electrolytes (67). PEO-based polymer electrolytes have been studied the most due to their advantageous properties including lower cost, ability to solvate a wide variety of ions, relatively high chemical stability and the use of their moderate mechanical strength (~10⁶ Pa) to supress the growth of dendrites (68, 69). However, the low conductivity (~10^–7 S cm^–1) of the electrolyte systems, due to the crystallinity of the polymer chains, has been a limitation (70). Overall, the general uptake of polymer gel cells has been restricted by their lower energy densities and poor electrochemical stability compared to liquid electrolytes.

More recently, researchers have explored a range of solid inorganic materials, which allow ionic mobility through the solid. Numerous classes of these are currently being explored, all possessing different advantages and disadvantages (63, 71, 72). A summary of these are highlighted in Table II.

Researchers have looked to examine inorganic electrolyte materials with high ionic conductivities, such as Li₁₀GeP₂S₁₂, which exhibits high conductivity at RT (73). However, sulfide-based solid electrolytes are generally expensive, more challenging to synthesise and are sensitive to moisture, potentially releasing toxic gases. This brings challenges in their handling and subsequent fabrication.

Although most solid electrolytes have been shown to react with Li metal, garnet materials (such as Li₇La₃Zr₂O₁₂ (LLZO)), have shown the greatest stability (74, 75). In addition, they have relatively low costs and a wide electrochemical window (~6 V vs. Li metal) potentially enabling the use of higher voltage cathode materials; and are therefore attracting increasing investigations (74). The cubic phase of LLZO is found to offer greater ionic conductivity than the tetragonal phase. A typical strategy to promote this is to dope elements such as Al, tantalum and gallium into the structure thus stabilising the highly conductive cubic phase at RT (76).

Despite these advantages, a challenge in using LLZO remains its instability in the ambient atmosphere, due to CO₂ and moisture (77). This results in increased complexity upon subsequent material handling and processing. Further challenges include poor interfacial compatibility of LLZO with electrodes. To overcome this, methods to increase the wettability of the electrolyte have been explored, such as the atomic layer deposition of Al₂O₃ to reduce interfacial resistance by the formation of a desirable Li-Al-O layer (73); or alloying Li with other elements (such as Si, Al, Ge) to increase compatibility (72).

In addition to the preparation of materials capable of high levels of Li-ion conductivity, it is vital that these materials can be manufactured at an industrial scale at a reasonable cost. While there has been considerable interest in the use of oxides for an all solid electrolyte, their brittleness and fragility impose new challenges for mass production (78, 80). As a result the scale up of such activities is being explored using a variety of different processing technologies (Figure 6). Mature slurry-based technologies have been shown to provide dense layers using high throughput techniques. However, subsequent high temperature sintering inhibits the co-firing of solid electrolytes and cathode particles.

Fig. 6.

When using Li metal as an anode material it is vitally important to prepare dense electrolyte layers in the absences of holes. It has been suggested that a critical relative density of >93% are required to eliminate the formation of dendrites in LLZO electrolytes (79); with short circuits believed to propagate through voids and grain boundaries (81). To obtain highly sintered garnet-based solid electrolytes by conventional sintering techniques, generally high temperatures (>1200°C) and long sintering times (>30 h) are required. Such conditions can result in the decomposition of the solid electrolytes and loss of Li from the structure.

To overcome these challenges, alternative processes such as hot pressing, field-assisted sintering and spark plasma synthesis have been investigated to fabricate the optimal dense ceramic layer (82–85). To that end further evaluation of deposition and sintering technologies will be required to provide an economically viable solution.

There are also multiple technologies (such as Li‐sulfur and Li-air chemistry) that have the potential to deliver significant advances in performance, such as increased energy density (86). For example, Li-S chemistry benefits from the low cost and high abundance of S and an energy density significantly higher than current Li-ion cells (~2500 Wh kg^–1) (87, 88). However, these technologies currently suffer from technical challenges that limit their uptake. To fully maximise the benefit of these technologies, it is necessary to overcome the challenges of working with a Li metal anode. The use of solid-state electrolytes is a recent area where people have been exploring with the aim of enabling the technology via anode protection.

The demand for cleaner air is accelerating and this is giving rise to increased electrification in the automotive drivetrain. There is also a growing acceptance of vehicles with varying degrees of electrification, and this trend looks set to continue. Current concerns for increased energy density to counter consumer’s ‘range anxiety’ are leading to material developments to meet this. In particular, the careful design and manufacturing of cathode materials with high amounts of Ni and anode materials with increasing Si content are steadily improving these key parameters. Furthermore, significant exploration into next generation technologies, such as solid-state electrolytes, opens the possibility of redesigning the cell. While options to the type of material used and their processing remain; the replacement of conventional liquid electrolytes promises to deliver further improvements in energy density as well as other benefits, such as safety performance. These three examples highlight the major trends being investigated and introduced into automotive cells to meet the demands of society.

The tremendous growth of the global economy is directly related to increased energy demand and (currently) high greenhouse gas emissions. Substantial reduction in global emissions is required to minimise environmental hazard and ongoing climate change. Legislations are pushing for energy transition, replacing fossil fuels with alternatives for reduced emissions. Wind, solar and biomass are key-players for the energy future, as depicted in the latest statistics and forecast (1, 2). According to one of the possible energy transition scenarios, to accommodate the increasing energy demands with the least environmental impact, renewable sources will rapidly grow their share in the energy mix, while natural gas is foreseen to maintain a key role during the transition phase (1, 2). However, natural gas contributes to CO₂ emissions, with approximately 7 billion tonnes of CO₂ being produced on a yearly basis, with approximately 5% of this amount attributed to flaring (Figure 1). This percentage adds to both the environmental problem and to the waste of an important resource, methane (3–5).

Fig. 1.

ADREM (EU project Horizon 2020 No. 636820), focused on the development of novel reactor concepts that are capable of converting methane to higher chemicals with a compact, modular and flexible process design. The University of Zaragoza (UniZar), Spain; Delft University of Technology (TU Delft), The Netherlands; and SAIREM, Décines-Charpieu, France, investigated microwave reactor technology for methane non-oxidative coupling (MNOC). Katholieke Universiteit Leuven (KU Leuven), Belgium and Kemijski inštitut in Ljubljana, Slovenia, worked with plasma technology for methane non-oxidative coupling and dry reforming respectively. Ghent University, Belgium, investigated the gas solid vortex reactor (GSVR) for oxidative methane coupling (OCM). In the present paper, we give an overview of the technologies that were developed, the status, the main bottlenecks and the path forward.

2.1. Microwave Non-Oxidative Methane Coupling with Both a Multistage Monomodal Reactor and with a Travelling Wave Reactor

2.2. Plasma Non-Oxidative Coupling of Methane

2.3. Oxidative Coupling of Methane with a Gas-Solid Vortex Reactor

2.4. Plasma Dry Reforming

To assess the potential of the reactors that were developed in ADREM, a case study of valorising associated (flared) gas has been simulated. The feed is rich in methane (>95% vol) with a flowrate of 1000 Nm³ h^–1. All the cases include pretreatment for sulfur and CO₂ removal, while for comparison purposes, the downstream processing follows the conventional approach, with either cryogenic separation (for C2+ hydrocarbons) or methanol loop (for syngas to methanol conversion). The end product consists either of mixtures of products (i.e. ethane/ethylene) or product at low purity (for example, raw methanol). Further purification in centralised units is necessary to reach the required quality.

The specific energy (Table I) of each technology consists of the reactors’ energy demands and the downstream processing (DSP) intensity (the latter being directly related to methane conversion and productivity). The microwave and GSVR technologies have the lowest specific energy consumption, as a result of the upscaled microwave reactor design of SAIREM and the exothermic OCM reaction respectively. The plasma technology is more energy intensive predominantly due to numbering up of the modules in order to accommodate the required flow. The technologies that produce BTX and ethylene would obviously benefit from replacement of the cryogenic separation by energy-efficient and modular alternatives (for example, ethane/ethylene membranes (14) or adsorption based technology) to decrease the energy demand. For the plasma dry reforming, the product syngas enables alternative downstream processing (for example, a methanol reactor), but the high operational pressure of such a design still adds to the overall energy efficiency and complicates the modularity of the plant. However, the modular methanol reactor is already available in commercial scale (3).

The capital intensity (Table I) is a function of the conversion and selectivity and the ease of upscale. On one hand, low conversion results in a large recycle flow (due to unconverted methane), and more energy-demanding units. On the other hand the numbering up strategy to accommodate the required throughput implies high capital requirements for all the technologies. The MW reactor with the realised upscaled concept and the GSVR that can accommodate high flowrate, appear to be the most cost-competitive at the present development stage. Collectively, the first step of further development for the ADREM reactors is to improve the reactor performance in terms of conversion and selectivity.

Compared to flaring, for all the technologies the CO₂ emissions are low (25–80% decrease, depending on the technology), with the highest CO₂ emissions coming from the GSVR reactor (where CO₂ is a product) and the lowest emissions coming from plasma dry reforming (where CO₂ is the reactant). Applying the ADREM technologies in situations associated with gas flaring in remote locations will have a huge environmental benefit when renewable electricity is available in abundance.

During the project, partners have been developing new small scale gas-to-liquids (GTL) technology, where methane is valorised to chemicals. Two of the reactor technologies have been successfully demonstrated in TRL 5 (microwave and plasma). With tighter regulation on greenhouse gas emissions and flaring, there are clear opportunities for the ADREM technologies to find applications. The UniZar reactor has efficiently been upscaled (32x) and the GSVR reactor is designed in such a way that it can accommodate relatively high flowrate. The plasma reactors (both NPD and dry reforming) showed the highest conversions and selectivities, but they still need to improve the upscale strategy.

For further upscaling and demonstration of the technologies, it is required to improve productivity, conversion and mitigation of carbon formation. Different operating conditions (in terms of pressure, temperature, catalysis or reactor geometry) or in situ product separation could potentially enable higher conversions and selectivity and are planned for the next steps of development. Improving the reactor performance will decrease the unit size for each technology and simplify the downstream processing. Downstream processing is an essential point that should be developed and optimised once the selectivity and conversion are improved.

In July 2018 the UK government’s Road to Zero Strategy was announced, including the ambition to see at least half of new cars to be ultra low emission vehicles (ULEV) by 2030 (3). ULEVs are vehicles that emit less than 75 g of carbon dioxide from the tailpipe for every kilometre travelled; in practice, the term typically refers to battery EV (BEV), plug-in hybrid EV (PHEV) and fuel cell EVs. This built on the government’s commitment to “end the sale of new conventional petrol and diesel cars and vans by 2040”.

There are over 200,000 ULEVs in the UK as of the second quarter of 2019 (4) and while total ULEV registrations are still low, this is growing rapidly for several reasons, including government tax incentives and consumer appetite for decarbonisation. 2019 saw an 87% year on year increase in BEV registrations and a corresponding decrease in PHEV registrations due to subsidy changes (5). In this article the term EV is used to refer to both BEVs and PHEVs; currently EV stock is split between these two types, however in 2050 we expect most cars to be BEVs.

To model the uptake of various road transport types and fuels in our 2019 FES we utilise a total cost of ownership model. Assumptions on the increase and decrease of various factors including battery costs, fuel costs, vehicle efficiency and subsidies available for different scenarios feed into this model. The uptake rates for the different scenarios, in relation to the expected sales projections for all vehicles (determined by the total cost of ownership and the rate at which older vehicles are scrapped) gives the expected number of low carbon vehicles on the road (Table I).

The slowest growth scenario in FES projects only 2.3 million EVs to be owned in 2030 compared to a maximum of 11.5 million EVs in 2030 in the highest growth scenario. This represents 6.8% and 35% of cars being electric respectively in each scenario. By 2050 we expect almost all cars to be electric in all scenarios, although some petrol and diesel fuelled vans and heavy goods vehicles (HGVs) still exist in the slower decarbonisation scenarios. Although this shift towards EVs will cause an increase in overall electrical energy demand, the greater challenge lies in charging; i.e. where, when and how these vehicles are charged.

Traditionally the grid has been supplied by a relatively small number of large generators, primarily coal, gas and nuclear power stations. The energy system is transitioning from this centralised system where there were under one hundred generators primarily connected to the transmission network with flexible fossil fuel plant to help meet demand peaks, to the current state where there are thousands of smaller decentralised generators such as wind and solar farms mainly connected to the distribution network. Over the past 10 years this growth in renewables has led to new challenges in system operation, with wind and solar generation presenting issues due to generation variability.

Significant progress has been made decarbonising the electricity system since 2010 thanks to this growth in renewable generation. The carbon intensity of electricity is a measure of the level of CO₂ emissions that are produced per kilowatt hour of electricity consumed. The average carbon intensity of electricity has fallen 53% from 529 g CO₂ kWh^–1 in 2013 to 214 g CO₂ kWh^–1 in 2019 (6). The trend in emissions reduction is shown in Figure 2.

Fig. 2.

2.2.1 Phase Out of Coal

2.2.2 Increase in Renewable Generation

The Office of Gas and Electricity Markets (Ofgem), UK, defines flexibility as “modifying generation and/or consumption patterns in reaction to an external signal (such as a change in price) to provide a service within the energy system” (12). Demand on the electricity network varies throughout the day and across seasons. Peak demands are seen on winter weekday evenings, between 5 pm and 7 pm, with minimum demands seen historically overnight during the summer. The country needs electricity capacity to meet peak demand, which is variable, and hence the ability to increase this capacity through flexibility or to decrease the peak is pivotal.

Renewable generation always generates where it can as it has zero marginal cost. This is currently backed up by fossil fuel generation that can be turned up and down as required to help meet demand peaks. Between April and September solar generation meets a larger portion of demand during the daytime; generation is at its peak in the middle of the day when the sun is brightest. Solar generation provides relatively little contribution towards meeting evening peaks in demand, however. Wind generation output depends on the weather systems over the UK but is typically higher in winter. It is highly variable however, and the system needs to be able to manage multi-week spells with low levels of wind generation which can occur when a high-pressure system settles over the UK.

Output from large-scale transmission-connected generation is visible to the ESO and instantaneous changes in generation can be clearly seen and managed. Small-scale distribution-connected generation however, particularly embedded solar, may show up only as reduced demand on the transmission system which can make it difficult to forecast and manage.

To understand the impact of EVs on the electricity system it is necessary to understand how they charge today and how this may change in future. We commissioned a Network Innovation Allowance (NIA) project to develop a comprehensive picture of current charging profiles (14). The study successfully gathered together a database of over eight million real world charge events and generated a representative full year charging demand profile at hourly resolution across a range of different location types and charger sizes. This evaluation has delivered an improved understanding of charging behaviour and enabled us to generate a more nuanced and informed view of the future impact of EV growth on electricity demand.

Existing electricity system peak demand typically occurs between 5–6 pm on weekdays, which is earlier than the peak demand for EV charging (Figure 3). This evening peak in EV demand is dominated by residential charging and is likely the result of commuters plugging into charge when they arrive home from work (it tails off as those vehicles plugged in earlier finish charging). Workplace and public charging contribute to another smaller peak mid-morning on weekdays between 9–10 am. The reduction in workplace charging rates after 10 am suggests that generally commuter vehicles plugged in to workplace chargers when they arrive are fully charged by mid-morning and remain plugged in and no longer charging subsequently until they leave.

Fig. 3.

Other learnings from this study include the effect of temperature on demand, where average kilowatt hour of energy per EV per day increases by 1.6% for each one degree decrease in temperature. During public holidays demand also drops, particularly over Christmas and Easter where, despite an increase in demand at (primarily motorway based) rapid chargers, this is offset by a significant decrease in other types of charging. Weekend demand is also on average 25% lower than weekdays and shows a broader demand profile shape that peaks an hour earlier.

It is clear from the data that current charging patterns will contribute to increased peak loads on the electricity network at both distribution and transmission levels. This may present more of a problem for the distribution network where the existing peak demand is often later than on the transmission network. If charging patterns can be shifted to increase levels of overnight and daytime charging at the expense of evening charging this could have a beneficial network effect and help reduce carbon emissions, as peak demands are more likely to be met by dirtier fossil fuel generation peaking plants.

This study has captured the charging demand of plug-in cars, but as other vehicle segments electrify demand will change. This, for example, includes depot-based vans, taxis and buses that may show different demand profile characteristics and present different opportunities.

As part of FES 2019 we developed four scenarios setting out a credible range for how energy demand and generation could develop out to 2050 (Figure 4). This includes projections of the levels of renewable generation, EV take-up and flexibility.

Fig. 4.

Two of our scenarios met the national decarbonisation target at the time of an 80% reduction in 1990 emissions by 2050. These are Two Degrees, which relies primarily on centralised generation and Community Renewables which has a greater proportion of decentralised generation. The UK government has since tightened the 2050 target to net zero CO₂ emissions. It is likely that new policy and support will be put in place to achieve this aim, therefore we would expect that by 2030 the electricity system would be closer to Two Degrees and Community Renewables than the other two scenarios which did not meet the 80% reduction target. Net zero in 2050 was modelled as a sensitivity in FES 2019 and will be included in core scenarios in FES 2020.

Figure 5 shows the installed electricity generation capacity of different technologies in 2018 and the projected changes to this under the different scenarios in 2030 and 2050. In all scenarios overall capacity grows, but this is particularly noticeable in the faster decarbonising scenarios, Two Degrees and Community Renewables. These two scenarios have a higher proportion of renewable generation and much of this capacity is variable, with a low load factor, meaning more generation capacity is required to meet overall energy requirements at times of high demand, particularly in winter. The total installed capacity significantly exceeds forecast peak demands to account for this. Due to their lower load factor and variability, renewables are de-rated when calculating the capacity required to keep the lights on as they will not always be available to contribute at peak times (15).

Fig. 5.

Figure 5 also shows potential future avenues to add flexibility, with significant increases in interconnector capacity and storage capacity, particularly across the more decarbonised scenarios. Interconnectors will allow the UK to trade more electricity with mainland Europe at times of high demand or excess generation. Shorter duration storage projects could meet small periods of increased demand or provide flexibility services such as frequency response. Longer duration storage is well suited to covering longer periods of, for example, high or low wind, potentially co‐located with generation. Some of the other key outputs from FES 2019 are set out in Table II for 2030.

In the faster decarbonising scenarios of Two Degrees and Community Renewables, the growth of low-carbon capacity will contribute to periods of oversupply of electricity, particularly in the summer months beyond 2030. Inflexible renewable generation capacity will at times produce more electricity than total demand. The annual amount of excess electricity rises to 20–25 TWh (around 6% of total annual output) after 2040 in Community Renewables. Our modelling shows that at times of likely oversupply, excess electricity cannot be exported, as other countries that have decarbonised are likely to be facing similar issues. Nor can it be stored, as available storage is full.

Future markets will determine how this electricity could be used, stored or curtailed in the most efficient way; this could include use of electricity to produce hydrogen or charge EVs. This is likely to be attractive to consumers as power prices will be very low or negative at times of oversupply meaning consumers could be paid to use the electricity when carbon emissions are also likely to be low.

National Grid ESO has developed a Carbon Intensity forecasting tool (Figure 6) (6) in partnership with Environmental Defense Fund Europe, UK, University of Oxford Department of Computer Science, UK, and the World Wide Fund for Nature (WWF), Switzerland. It uses machine learning and power system modelling along with Met Office, UK, data to forecast the carbon intensity and generation mix 48 h ahead for each region in Great Britain. The forecast carbon intensity figures are accessible via a website, the National Grid ESO app and an application programming interface (API) to allow developers to produce applications that will enable consumers and smart devices to optimise their behaviour to minimise carbon emissions. WWF have implemented the API into a widget that can help people plan their energy use, switching devices on when energy is green and off when it is not.

Fig. 6.

The data from our EV innovation project suggests that EVs typically spend long periods of time plugged into residential or workplace charge points and current charging patterns result in vehicles starting to charge as soon as they are connected to the charger with little to no smart management of charging. Smart charging enables consumers to manage the time when their vehicle is charged. This could be to take advantage of lower prices or lower carbon electricity or to respond to external signals from third parties such as aggregators or network companies.

The government’s Automated and Electric Vehicles Act 2018 (16) sets out requirements for all new charge points sold or installed to be ‘smart’. This means they must be able to receive, process and react to information or signals, such as by adjusting the rate of charge or discharge; transmit, monitor and record information such as energy consumption data; comply with requirements around security; and be accessed remotely. This legislation aims to avoid infrastructure being a blocker to future smart charging developments.

EV batteries can be considered as a form of storage within the wider energy system, though the impact of EVs is fundamentally different to other forms of storage. This is because not all vehicles are connected to the system at any point in time, meaning that the available storage capacity from EVs is constantly varying. This creates natural diversity in availability and charging behaviour for EV batteries and means that the potential for EVs to increase, shift or decrease demand varies and is a fraction of the total capacity of EV batteries in Great Britain at any one time. BEV batteries are typically five to 10 times larger than PHEV batteries, so the relative mix of PHEVs to BEVs will also affect the total energy capacity available.

Consumers can be incentivised to take part in smart charging and delay the start of their charging period through time-of-use (ToU) tariffs and be guided by tools such as National Grid ESO’s Carbon Intensity app; these are already available to consumers to allow them to schedule their EV charging for times of lower prices or carbon emissions. A more dynamic form of smart charging involves in-home automation and smart management and optimisation of charging while the vehicle is plugged in without active involvement from the consumer. This would remove barriers for consumers to get involved and have a significant impact on the electricity system and resulting carbon emissions. This will become more important as the number of EVs on the system grows. These ToU tariffs are already available for consumers from some innovative energy suppliers such as Octopus Energy, UK, and are expected to become more widely available over time.

An additional avenue for EV to have a positive impact on the electricity network is through the use of V2G technology. This is where electricity stored in the battery of an EV can be supplied back into the network through a two-way V2G enabled charger. This process is likely to be managed by an aggregator triggering response from a large portfolio of vehicles contracted to deliver this capability, they would likely offer financial incentives to consumers to facilitate this. Individuals and businesses could also use this to take advantage of variable rate tariffs without the third-party involvement. There are a range of pilot projects developing this technology; in 2017 the UK’s innovation agency, Innovate UK, committed £25 million in support to eight real world V2G demonstrator projects undertaken by a range of organisations including energy suppliers, network operators and small and medium-sized enterprises (SMEs) (17).

Battery lifetimes are typically measured in the number of discharge cycles they can undergo without battery capacity falling below a certain threshold. The measurable impact of V2G on battery health is still at the research stage, with recent papers providing seemingly contradictory conclusions. Dubarry et al., 2017 (18) showed that additional battery cycling due to V2G would shorten battery life; while Uddin et al., 2017 (19) indicated that battery degradation could be avoided. These authors have since published a joint study in which they “jointly reconcile their previous conclusions by providing clarity on how methodologies to manage battery degradation can reliably extend battery life” (20). It is clear, however, that further research in this area is necessary to determine the effects of V2G and ensure it is an attractive proposition for both electricity networks and consumers.

Our FES 2019 scenarios consider how engaged vehicle owners are likely to be with smart technology and V2G and build these assumptions into our modelling of peak demand. We classify a consumer as participating in smart charging if they actively choose not to charge their EVs at peak times, wherever possible. We assume that only 2% of vehicle owners engage in V2G through to 2030 as the technology is still at an early stage, however that number then steadily increases to 2050, with the highest levels in the Community Renewables scenario. These participation rates are shown in Table III.

3.4.1 Impact on Peak Demand

3.4.2 Impact on Oversupply of Renewable Generation

Early research into alternative fuels for aviation was conducted following the fuel price increases in the USA in the 1970s (1) driven by concern around costs and security of supply. Today, with the aviation industry responsible for around 2% of all human-induced carbon dioxide emissions (2), its estimated contribution to manmade climate change more than double this when non-CO₂ impacts are taken into account (3), and rapid growth expected over the next decades, the development of alternative aviation fuel is driven largely by concerns around climate change. Global aviation activity grew by 140% between 2000 and 2019 (4) and passenger numbers have been anticipated to continue to grow at a compound annual growth rate of 3.5% over the next two decades (5).

Policies are beginning to be put in place which aim to reduce GHG emissions from the aviation sector. In 2016 the International Civil Aviation Organization (ICAO) adopted the Carbon Offsetting and Reduction Scheme for International Aviation (CORSIA) which aims to stabilise net CO₂ emissions from international civil aviation at 2020 levels (6). Whilst the remit of ICAO only covers international aviation, an increasing number of measures are being put in place by national governments which cover domestic flights and international flights. Flights within the European Union (EU) are included in the EU Emissions Trading System (EU ETS) since 2012. Domestic aviation is included in New Zealand’s Emissions Trading Scheme and other states such as Canada and China have indicated that domestic aviation will be brought within a national carbon pricing scheme (7).

Even taking into account fuel efficiency improvements that can be achieved by more modern aircraft design and improved operational measures, low-carbon fuels will be essential in order to meet targets for the decarbonisation of the sector (8). Several countries or regions including California (9), the UK (10) and The Netherlands (11) have included aviation fuel within national support schemes for low carbon fuels on an opt-in basis. Norway has introduced blending mandates for alternative fuels in aviation, and a number of other countries including Sweden and The Netherlands are considering similar policies (12).

In the past 20 years substantial progress has been made in the production and use of alternative aviation fuels. In February 2008 a Virgin Atlantic, UK, Boeing 747 (The Boeing Company, USA) became the first aeroplane flown by a commercial airline on a blend of kerosene and bio-jet fuel, and the first scheduled commercial flights on bio-jet fuel began in 2011 (13). Today more than 100,000 commercial flights have been carried out using alternative liquid aviation fuel (13), and at the time of writing there were six alternative fuel pathways certified by ASTM International, USA (14). Two additional ones have been approved in 2020.

One of the main challenges for low-carbon fuels replacing fossil kerosene is matching the same fuel energy density. The energy consumption of an aircraft is proportional to its mass and that is why the fuel energy density and the weight of aircraft components are key factors. Bio-jet has almost identical energy density to fossil kerosene, while hydrogen’s volumetric energy density is an order of magnitude lower, and electrochemical batteries’ volumetric and mass energy densities are also an order of magnitude lower (Figure 1).

Fig. 1.

This paper reviews the status of alternative fuel options for the aviation sector, covering liquid fuels, hydrogen and electricity. A schematic overview of all alternative fuel routes for aviation is provided in Figure 2. The paper then explores the prospects for future demand and supply of alternative drop-in liquid aviation fuels to 2030.

Fig. 2.

Renewable drop-in kerosene alternatives are synthetic liquid fuels produced from biogenic feedstocks or using renewable hydrogen and CO₂ (from waste streams or from the atmosphere) which are functionally identical to fossil jet kerosene. There are several possible routes to produce renewable drop-in kerosene based on different feedstocks and technology variants. Table I summarises their technology status.

The costs of alternative fuels are substantially higher today compared to fossil kerosene, with costs ranging between two and five times the price of conventional jet fuel (global average price paid at the refinery for aviation jet fuel in October 2019 was about US$600 per million tonne). The lowest alternative fuel costs today are associated with the most commercially mature route consisting of the large scale hydroprocessing of used cooking oils (UCOs), animal fats and raw vegetable oils (16).

The GHG emissions savings from renewable routes will generally be substantial, but vary, largely depending on the emissions associated with producing the raw materials used in their production. It is generally expected that savings will be between about 95% in the case of renewable electricity based routes and 65% for routes based on conventional crops, with savings from routes based on biomass wastes somewhere in that range (17). Electricity used to produce e-fuels is generally supplied through the grid. The renewability of this electricity needs to be guaranteed through accounting procedures which also need to assure that the same renewable electricity is not double-counted for other uses. In the case of fuels based on energy crops, it will be important to consider their sustainability with regard to land use change impacts (18).

2.1 Synthetic Paraffinic Kerosene Produced from Hydroprocessed Esters and Fatty Acids (HEFA-SPK)

2.2 Alcohol-to-Jet Synthetic Paraffinic Kerosene (ATJ-SPK)

2.3 Synthesised Isoparaffins Produced from Hydroprocessed Fermented Sugars (HFS-SIP)

2.4 Fischer-Tropsch Synthetic Paraffinic Kerosene (FT-SPK)

2.5 Pyrolysis and Upgrading

The APR process catalytically converts biomass-derived oxygenates (such as sugars, sugar alcohols and polyols) in an aqueous solution into hydrogen, CO₂ and a mixture of alkanes, acids, ketones and aromatics (24). A series of condensation reactions then lengthen the carbon chains in the mixture of hydrocarbons. This mixture then undergoes hydroprocessing, isomerisation and distillation. APR using conventional sugars is at TRL 5–6 as a result of pilot scale plants operated by Virent Inc, USA. APR derived bio-crude using lignocellulosic sugars has been produced and upgraded to bio-kerosene at laboratory scale (25). Aviation kerosene produced via APR is in Phase 2 of the ASTM International certification procedure and referred as hydro-deoxygenated synthetic kerosene (HDO‐SK) (14).

Unlike other reforming processes, APR operates in wet conditions which reduces the costs of dewatering certain feedstocks like sugars. However, this process has low selectivity to liquid hydrocarbons (high gaseous yields) and short catalysts lifetime due to deactivation and coking (20). These two characteristics make APR expensive from a capital and operational cost standpoint. APR is also gaining interest as a route for biochemicals production (26), which could lead to higher value products.

Hydrothermal liquefaction (HTL) is a process where biomass and water are heated at very high pressures to produce a bio-crude. The near and supercritical water acts as a reactant and catalyst to depolymerise the biomass. The bio-crude produced can then be upgraded similarly to the pyrolysis route. The higher molecular weight distribution makes HTL oil more suitable for diesel production, but gasoline and jet are possible adding hydrocracking steps. HTL is well suited to process very wet biomass (sewage sludge, manure, micro and macro algae), as well as some lignocellulosic feedstocks. Bio-crude production of HTL oils is currently at TRL 5–6 with small scale demonstration activities ongoing (27). Dedicated upgrading to jet fuel is at laboratory-scale (TRL 3–4). The upgrading of HTL oil in refineries is being tested as part of the Horizon 2020 4REFINERY project (28). This route has not entered the ASTM International certification procedure and is still in pre-qualification stage (14).

HTL oils typically have much lower water content, higher energy content, lower oxygen content and greater stability than pyrolysis oils, hence are expected to be cheaper to transport and require less extensive upgrading. It is expected that HTL oils could be used at high blends in refinery fluid catalytic cracking (FCC) units. With mild hydrodeoxygenation, it might be possible to co-process the bio-crude with fossil crude oil in the front end of existing oil refineries (29). Challenges of this route are the high pressure and corrosive conditions under which the process operates.

The PtL FT route produces liquid fuels by catalytically combining a carbon source with a hydrogen stream produced via electrolysis. This pathway requires three ‘feedstocks’: electricity, water and a concentrated source of CO₂. The maturity of the PtL FT route depends on the maturity of single components and the design configuration chosen, with some systems being demonstrated at small scale (TRL 5–6). High-temperature PtL employs solid oxide electrolysers (SOE), which are more efficient but less mature than other electrolysis technologies (for example, alkaline electrolysers) (30). CO₂ from concentrated sources like biogas upgrading, ethanol production or beer brewing or CO₂ waste streams from industrial processes are commercially available, but other sources, such as direct air capture, are at an earlier stage of development and commercialisation (TRL 6–7) (31). FT synthesis is a well-established process at large scale, but at the demonstration stage for small scale applications (TRL 6–7) (20). FT-SPK produced through PtL is certified under ASTM International as long as the FT synthesis is based on iron or cobalt catalysts (D7566 Annex 1, article A1.4.1.1).

Operating costs for this route can be very high depending on the cost of electricity. Specific capital costs are currently high as the technology is at the early demonstration stage, and the potential to reduce these through scaling and learning remains to be demonstrated (32). Technology developers are also working on different FT catalysts with different selectivities that could provide more direct routes to desired fuels and be more economically viable at relatively small scales. The technology also requires concentrated flows of CO₂, which might constrain the location of these plants in proximity to large industries. Despite being at very early stage with just a handful of active developers, PtL is a pathway attracting widespread interest as a result of its potential to produce fuels with very low GHG emissions and subject to less feedstock constraints and sustainability issues compared to bio-based fuels.

2.9 Demand and Supply Scenarios for Drop-in Kerosene Fuels

A transition to hydrogen in civil aviation requires major aircraft and infrastructure changes. However, the potential for hydrogen as a widespread clean energy source in the future also leads to interest in its use in aviation. In August 2019 the German government announced the ‘Leipzig Statement for the Future of Aviation’, proposing the introduction of a hydrogen in aviation strategy by the end of 2019 (46). Use of hydrogen, both as a source of propulsion power and on-board power, has the potential to reduce noise pollution, increase efficiency and reduce GHG emissions associated with the aviation sector as long as hydrogen is produced from a renewable source, from other potentially low carbon energy sources such as nuclear or from fossil sources with carbon capture and storage.

While hydrogen has a much higher gravimetric energy density than kerosene, its volumetric energy density is much lower and both characteristics are critical to airframe design and performance (Figure 1). Due to hydrogen’s low volumetric energy density, redesign of the airframe is required to accommodate the highly-insulated tanks required to store liquid hydrogen (LH₂) (47).

3.1 Hydrogen Turbofan

3.2 Hydrogen Fuel Cell Aircraft

Aviation electrification has been a trend since the 1960s, with many auxiliary systems increasingly electrified owing to the relative lightweight and higher efficiency compared to mechanical systems. Electric propulsion has also seen development since the 1970s, but so far it has been limited to demonstration or leisure activities (57). Electrically enhanced propulsion could provide significant benefits, including fuel and emissions savings and noise reduction, but technical challenges associated with battery energy and power density remain yet. Like automotive electrification, various degrees of electrification and different architectures are possible.

4.1 Hybrid Electric Aircraft

4.2 Full-Electric Aircraft

The SAF and propulsion options described in this review span across different levels of technical maturity, economic viability and current applicability to different types of aircraft. Table II provides a summary of these options, highlighting key technical, environmental and economic characteristics.

Renewable drop-in kerosene is an attractive decarbonisation option for aviation because it does not require modification of the aircraft airframe and engine and refuelling infrastructure. Today it is commercially produced in low volumes for use in commercial flights from a limited number of airports. Its production cost is currently significantly higher than the fossil kerosene price, representing the main challenge to its uptake, which will depend on strong policy support. While hydrogen is a very appealing fuel that can be derived from a range of renewable sources and produced from fossil sources with carbon capture and storage, its use in medium and long-haul aircraft requires a radical redesign of the engine and airframe, as well as the fuel supply chain, including on-the-ground storage and refuelling, leaving it a prospect for the long term.

Hybrid and full electric aviation are gaining traction with several projects and prototypes being developed to demonstrate the technology and trial new aircraft concepts, involving research organisations, small companies, as well as major aircraft manufacturers. Small full-electric planes (up to 10-seaters) are likely to see commercial deployment in the near term. But, the technical requirements of medium and long-haul aircraft (weight, seat capacity, speed and range requirements) cannot be met with current battery technology. Without a breakthrough in battery chemistry, electric propulsion is unlikely to be used in commercial aviation beyond the smaller short-haul flights. However, as technological progress is made, hybrid electric solutions could emerge for larger aircraft, furthering hybrid powertrain and airframe integration and contributing to the reduction of fossil kerosene use in aviation.

Hydrogen FCVs are establishing themselves in consumer and fleet markets worldwide, with 11,200 FCVs and 376 hydrogen refuelling stations (HRSs) open to the public and fleets by late 2018 (1). However, the lack of a convenient refuelling infrastructure remains a barrier to greater FCV diffusion.

There is robust discussion regarding network deployment of initial HRSs to address this (2–8), although there is not agreement on how best to geographically arrange stations to do so (9, 10). This area of literature began before the initial market diffusion of FCVs, and relied on surveys of drivers of conventional vehicles about hypothetical station scenarios, or of analogue populations of diesel or natural gas vehicle drivers, to predict FCV adoption and refuelling behaviour (11–15). Since the roll-out of HRSs and FCVs, recent studies have surveyed initial FCV drivers about their station usage (16), but a key outstanding research area is how prospective FCV adopters, accustomed to the ubiquity of gasoline stations, evaluate the spatial arrangement of the full network of HRSs when adopting FCVs. That is the primary emphasis of our National Science Foundation (NSF)-funded research project, which employed a mixed methods research design to address a set of related questions (Table I). In this short paper, we distil 15 high-level insights from these studies for regions and companies planning a rollout of HRSs and FCVs. We refer readers to current and future publications and presentations for greater detail on the methods and results than is possible here.

Our mixed-methods research design involved a combination of (a) revealed preference survey research, (b) qualitative ethnographic approaches that analyse consumers’ decision-making processes and language and (c) geodesign participatory planning (18–20).

We conducted this research in California and Connecticut. By November 2019 over 7700 FCVs had been sold or leased in California, supported by 42 public HRSs (21), allowing an opportunity to evaluate how recent early adopters evaluated FCVs and HRSs when they got their vehicles. Connecticut is one of eight Northeast US states with Hydrogen and Fuel Cell Development Plans updated in 2018 (22), making it a compelling location to evaluate stakeholder opinions and prospective FCV adoption.

2.1 Survey Research

2.2 Ethnographic Content Analysis and Decision Tree Modelling

2.3 Geodesign Workshop

We distil our findings from this mixed-methods approach into 15 primary lessons.

3.1 Motivations for Fuel Cell Vehicle Adoption

3.2 Fit Between Vehicle and Driver

3.3 Convenience to Home is the Most Important Factor

3.4 Perceived Convenience to Home Varies

3.5 Stations Near Work or On the Way Can Substitute for Near Home

3.6 Secondary Stations

3.7 Station Trade Areas

3.8 Station Reliability and Backup Stations

3.9 Secondary Vehicles

3.10 Convenience to Freeways

3.11 Planned Stations

3.12 Changing Refuelling Stations

3.13 New Stations After Experience

3.14 Demographics and Stakeholder Priorities for Placing Initial Stations

3.15 Sufficient Initial Number of Stations

Finalising the EDM for FCV adoption in California will require completing additional interviews, especially with drivers who seriously considered adopting an FCV but ultimately did not. In addition, we recently completed data collection for the stated preference survey in Connecticut, which prompts respondents to evaluate their willingness to get an FCV given three maps that show different pre-generated spatial arrangements of initial HRSs.

The consistent lesson is that these early FCV adopters are diverse in their motivations for wanting an FCV and in the list of stations and refuelling strategies they planned to use at the time of adoption. Drivers consider everything from lifestyle to image, and from incentives to station locations when deciding to get an FCV. A station “near home” is important to many drivers, but it is neither necessary nor sufficient for others. What is subjectively “near home” varies from minutes to over an hour away. Station reliability, secondary stations, freeway access and convenience to a variety of destinations all are important, especially while awaiting the opening of planned stations. Over time, drivers begin using stations they initially did not consider to support travel farther from home, with reliability and short detours also playing important roles.

The key implication is that stations should be located to serve not only ‘targeted’ nearby residents but also others who may visit or pass nearby regularly. Likewise, developers should also locate stations far from these neighbourhoods to benefit the wider travel of these residents and local travel of those who live elsewhere.