The interest in hydrogen has coincided with increased concern about climate change, and hydrogen has been earmarked as a suitable alternative to fossil fuel energy and products for decades. ‘Green’ hydrogen is produced via electrolysis using zero-carbon electricity, thus contributing no carbon emissions. Enthusiasm has also grown since the 2010s partly due to the increased presence of renewable electricity generation. The application of renewable hydrogen electrolysis is continually changing, and further research is required to understand and address uncertainties in weather, prices, and demands [1]. The relationship between electrolysis and renewable electricity is symbiotic; green hydrogen requires green electricity, but there also exists reverse benefits, including auxiliary grid services [2], energy storage [3], and the reduction of curtailment [4]. Running electrolysers with intermittent (renewable) electricity is difficult, and since electricity cost dominates the levelised cost of hydrogen (LCOH), optimal supply with dedicated control strategies is preferable [5]. The selection of electrolyser technology influences the electrical control strategy [6], as well as the cost. Intended delivery formats of hydrogen differ due to the variety of storage media that exist and will exist, such as in salt caverns [7], or chemically bonded in a easier-handling format such as ammonia borane [8]. Storage costs and the means of preparing hydrogen for storage will also contribute to the price of hydrogen, and must be appreciated as part of the overall scenario assessment.

There is little literature coverage on the utilisation of oxygen as a byproduct of green hydrogen production [9]. The available literature mainly covers the electrochemical fundamentals of electrolysis, including oxygen production as an afterthought [10,11], with little comment on the application outside of the electrolyser. The range of applications of oxygen in industry is large, but depends on the quality of the electrolytic byproduct oxygen. Medical-grade oxygen would require careful post-processing, and high-purity oxygen is being used increasingly in welding, metallurgy, and the chemical process industry [12]. It is worth noting that using the electrolysis of water to only produce oxygen is not economically feasible, as other technologies such as cryogenic air separation or pressure swing adsorption produce oxygen on a preferable scale and efficiency [13]. Thus, the discussion of oxygen is only logical as a byproduct of green hydrogen production, or reimagined as a dual hydrogen and oxygen production operation. However, in future electrolyser-heavy energy systems, an abundance of byproduct oxygen can benefit many other processes that were previously cost-prohibitive. There are considerable energy savings to be made by introducing higher concentrations of oxygen to processes such as electric arc welding, glass melting, and gas turbine electricity generation [14]. The economic-and climate-positives continue since energy is saved in not having to source the required oxygen from the use of conventional industrial practices, more so if it can be produced electrolytically on-site, saving on transportation energy expenditure. A study on integrated liquified electrolytic hydrogen and oxygen co-production has been performed [15], but this excluded the implementation of associated costs.

Medical-grade oxygen as a byproduct has been techno-economically assessed in hospital co-production scenarios [14], which shows it can be economically viable under certain conditions, with the hydrogen used for energy. This depends on the utilisation of the byproduct because if more hydrogen is needed than the equivalent-produced oxygen, this leads to byproduct waste and higher costs. To balance the system, grey hydrogen would have to be introduced. The temporal context of this research is important. Electrolyser stack prices have fallen rapidly since 2005, alongside improvements in system efficiency. The price of renewable electricity has decreased and its penetration has grown, both much faster than anticipated, which changes the operational costs and the dependance of electrolysers on system resilience. Lastly, the hydrogen and oxygen markets have changed; as anticipated, hydrogen demand has increased across all sectors but also, in the wake of the COVID-19 pandemic, the demand for medical-grade oxygen has increased, especially with new treatments such as hyperbaric oxygen therapy being investigated as a cure for ‘long covid’ [16]. A photovoltaic (PV) and electrolytic oxygen-based system with byproduct hydrogen for backup energy or sale for a hospital is now feasible [17], which can save the hospital money and guarantee a supply of renewable oxygen and energy, which would be especially favourable in remote locations, where significant cost is encountered sourcing external gas.

The generation of oxygen as a primary output has other unconventional applications. Solid oxide electrolysis of carbon dioxide has been hypothesised for generating living conditions on Mars [18]. While slightly outside the climate-focused energy system remit of this research, the application of generating oxygen for respiration in remote locations can be extended to Earth systems. Electrolytic oxygen for submarine respiration would be ideal in remote locations to perform, for example, marine research, underwater structure engineering, or covert military operations, again all with the benefit of byproduct hydrogen for energy generation. Investigating water electrolysis cogeneration in submarine and extra-terrestrial applications makes a refreshing change compared to climate-focused projects. The value placed on the gas products is much higher, and system resilience is a greater priority than in widely-connected grid applications, amongst many other factors. A diversity of hydrogen applications will inspire innovation. It is pleasant to see the use of novel electrolysis technologies in alternative roles where carbon abatement is only a side note.

The crux of electrolytic green hydrogen undoubtedly is the access to zero carbon energy. Fortunately, water electrolysis stands to compliment the operation of a renewable-intensive electricity grid. The benefits of electrolytic hydrogen is recognised in the (UK) Future Energy Scenarios 2022 report [19], by reducing curtailment and network congestion. To support this, in the most ambitious scenarios, there will be 50 GW of offshore wind by 2035 and up to 70 GW of solar-PV by 2050, in order to nourish a target of 5 GW of green hydrogen production in the 2030s, for use across multiple sectors including heating, transport, the electricity grid, and industry. The hydrogen would alleviate network congestion by providing an alternative vector to transport energy, since during high renewable generation and/or peak demand periods, a heavily-electrified energy system would suffer from constraints; network congestion will rise with increased electrification of transport and heating, as planned by the government. Decentralised hydrogen storage and electricity regeneration would make a positive impact to help address this. Hydrogen also supports renewables by absorbing otherwise curtailed oversupply from intermittent non-despatchable generators.

There is much to gain from merging many aspects of the hydrogen value chain in the outcome analysis. This is often the limit of modelling analysis, as the scope of inputs is reduced to avoid long computational times, and nebulous results. It has been recommended to increase the modelling scope to identify more synergies [20]. In a modelling scenario, the selected output modes of hydrogen and oxygen are to be carefully selected so as to address key objectives. The criteria for deciding the outputs is to represent typical delivery formats of the product hydrogen and (by)product oxygen that end users would typically encounter.

At the whole system level, post-processing costs play into the utilisation of hydrogen as an energy vector. Hydrogen is played off in scenario comparisons for a range of sectors, and it is normally compared to batteries for electricity storage options for the electricity grid, or in transport modes. The main issue with gaseous energy stores is the energy required for compression, in order to achieve volumetric energy densities that are similar to those of conventional solid or liquid fuels, or increased compression levels compared to methane compression standards in conventional use in the present day. The compression energy requirement as a percentage of the energy content of hydrogen has been measured as a function of achieved pressure [21]. With the typical useful pressure across a variety of technologies being 350 bar, this instantly removes approximately 5–12% of the energy storage efficiency depending on the compression process technology. Equivalently, this adds a significant energy expense both for operational expenditure (OPEX) and reduced energy input for electrolysis for non-dispatchable renewable sources, highlighting the importance of adequately modelling the process parameters and outputs at the system level for compression. The higher pressure requirements also lead to operational safety considerations that have been preventative of technology adoption, particularly due to the cost of novel materials to achieve (for example) 700 bar in transport fuel cell applications [22]. The variety of storage media relevant to future hydrogen scenarios is wide; for large applications, this includes custom salt caverns for large inter-seasonal heating fuel reserves [7], repurposing expunged natural gas wells in geological formations, and the linepack of a national gas transmission network [23]. The energetic cost of storing hydrogen as a component of another chemical [8] could also be considered.

With anticipated electrolytic hydrogen production targeted for 5 GW of production in 2030 [24], justifying the utilisation of a significant fraction of byproduct oxygen would require estimates of the market size. Electrolysis efficiency of 100% and a hydrogen energy density of 142.2 MJ/kg (HHV) results in 1.11 billion kg of hydrogen annually and, equivalently, 8.80 billion kg of oxygen. The Netherlands is estimated to use 2.5 billion kg per year [25] and through extrapolation based on differences in GDP, this could mean the UK is using around 13 billion kg of oxygen per year. Therefore, the full utilisation of byproduct electrolytic oxygen from a 5 GW national electrolyser fleet (assuming no losses) could address around 8.5% of an estimated oxygen market.

Hydrogen electrolysis has been studied in some contexts as a standalone generation source [26], but it is understood that a diversity of generation sources provides much higher capacity utilisation and a greater guarantee for having available power for electrolysis at a given time. For example, coupling cheap wind power with a less intermittent (but likely more expensive) source like geothermal power can give a very low levelised cost of electricity (LCOE) when combined with other processes such as hydrogen production, hot water, and freshwater production and cooling effects [27]. Another study investigated the advantages of combining nuclear generation with wind turbines for electrolysis, hot water, and electricity [28]. In the context of a national electricity grid, the total generation would be source-non-discriminating, and input power research has to consider the multitude of producers to assess the advantages of hybrid or multi-sourced power generation. When integrated into a hydrogen multi-modal value chain, further benefits would become apparent from system linking.

Instead of multi-generational solutions, green hydrogen production has also been hypothesised from dedicated power installations, such as offshore wind [26]. The best strategy for implementing this process was identified as low-temperature electrolysis due to the power variations from wind speed changes, and not using direct seawater electricity, since it is less efficient and has a large environmental impact. This favours dedicated or integrated marine hydrogen production, especially as the cost of offshore wind energy is still falling. Another study [29] showcases the best business case to be using wind generation, with the best daily production of hydrogen, and good efficiency. Geothermal power had a good daily yield but poor efficiency, and solar-PV had a low daily yield, albeit at a very low cost. This would supplement the scenarios that implement offshore wind as the preferred choice for green hydrogen, but there is also the argument that suggests hybridising the generation to increase capacity utilisation of the stacks. On the other hand, it is possible to implement dedicated renewable generation for green hydrogen production, given the geographical factor. The solar irradiance in Chile can deliver electricity costs from solar-PV of USD 21/MWh, resulting in a LCOH of USD 2.20/kg (2018) [30]. There were also comparisons between direct PV connection or using a power purchasing agreement, to see how the change in electricity cost plays against the increased capacity factor provided. It was found that the best power provision strategy depended on the generation technology; photovoltaic preferred a power purchase agreement, and concentrated solar power favoured the direct connection since it already benefits from higher capacity utilisation due to thermal energy storage coupling. Despite the low electricity prices, electricity cost was still the most sensitive contributing factor, followed by specific capital expenditure (CAPEX), and with water costs being almost negligible. Comparing the latter two PV cases [29,30] demonstrates the considerations of geography, but at least does not rule out dedicated generation assets for hydrogen.

There is constant discourse regarding the type of electrolysis technologies in the green hydrogen production chain. Commercial applications see heavy use of Alkaline and Proton Exchange Membrane Electrolysers (AEL and PEM). PEM electrolysis was introduced and developed partly to help alleviate the challenges facing AEL when coupled with intermittent supply from renewables, mainly the inability to operate at low current densities (partial loading) [31] and slower ramp up/down times, and it has been shown to be better performing for thermal efficiency [32], with follow-up research demonstrating the business cases against different renewable generation sources [29]. Additionally, for scaling green hydrogen, it is noted that AEL suffers from low current densities due to high ohmic losses [31] and the low operational pressure. Compression of hydrogen could be performed in line with electricity price and availability to save cost, and in line with demand side management for improved electrical system management, although this would require intermediate gas storage at lower pressures. The choice between AEL and PEM has been evaluated in previous studies and resulted in different technology selection in different scenarios. Thus, in future modelling-based research, the flexibility of technology choice must be considered. Operation of electrolysers with variable current affects plant lifetime [6]; this must be factored into scenario models and techno-economic assessments. System adjustments can be included to improve electrolyser performance, such as power smoothing with short-term storage with batteries or capacitors, or through control strategies such as maintaining part-load to achieve optimal economic conditions, all of which could be hypothesised and tested through modelling.

AEL is a more mature technology, which results in a higher capital cost for PEM [6]. Since the motivation to develop PEM was loosely dependent on the uptake of renewable energy, it is reasonable to assume that market interest, and hence reduced CAPEX, will fall following the trend of renewable energy growth in the new millennium, since the benefits of complimentary electrolysers in the decarbonised system will be realised. This will affect PEM CAPEX as a preferred candidate (theoretically) for intermittent power generation. A new project [33] is pioneering a 100 MW PEM plant coupled to the Hornsea 2 offshore wind farm. This plant could benchmark grid-scale electrolysis, providing many of the system benefits previously discussed. The green hydrogen will be directly used to decarbonise the Humber region’s industry, especially the Phillips 66 refinery. This should serve as a direct real-world application to certify the research, but many subsequent projects and studies would likely result from this to continue the expansion of green hydrogen production.

Outside of AEL and PEM there are other nascent technologies, including solid oxide electrolysis (SOEL) cells, which are currently pre-commercial and have not fallen enough in cost to be competitive as of yet [6]. SOEL could allow higher production efficiency and reversible solid oxide fuel cell (rSOFC) modes to provide flexible power generation from hydrogen storage and increase the capacity utilisation of the cells. rSOFC could have applications at a range of scales, from national-scale storage/peak generation to providing system resilience and auxiliary services to smaller decentralised grids similarly to batteries, but with the option to use the hydrogen for other end uses. Capillary-fed electrolysis [34] is hoping to take electrolytic hydrogen to the next step by simplifying the balance-of-plant through a different production mechanism, but this technology is yet to be verified at a demonstration level. The subsequent cost reduction would also be necessary to implement this into a real-world system.

Renewable power for electrolysis compounds the effect of electrode degradation, due to the non-constant and often low current densities, and the increased likelihood of stop–start cycles [35]. This phenomenon will impact the sensitivity of techno-economic parameters in the whole system electrolysis model, due to the consideration for the reduction of the lifetime of electrolysis stack, and reduced hydrogen production efficiency. The implementation of the contribution to overvoltage in the stack due to degradation would thus be a time-dependent and start–stop cycle-dependent term, based on empirical observations of prolonged stack operation.

The field literature focus is on stack behaviour modelling, but unfortunately for applications regarding system level analysis, hydrogen evolution efficiency cannot be reduced to the energy efficiency of the cell alone. For a case study using solar energy in Chile, the energy requirement per kilogram was calculated including auxiliaries to encompass the true LCOH [30], despite the fixed energy consumption in the stack removing the finer detail of the hydrogen evolution reaction (especially regarding inefficient low current operation). The auxiliary energy included in this study does account for the distribution, storage, and transport accommodations, such as the energy cost of liquefying or conversion to ammonia for intercontinental trade. Post-processing energy requirements are especially invaluable to the whole system perspective, as unpressurised hydrogen is virtually unusable in decarbonised energy futures. Post-processing of green hydrogen with non-green energy would obviously introduce a carbon footprint, so system-level control of compression in tandem with water electrolysis has to account for the two processes being run from green, and thus mostly intermittent, power supplies, and all energy-consuming auxiliary processes can be added on to the total energy demand with compression. Detailed modelling of auxiliary power consumptions had not been found in the literature review from [36], the most detail was found to be modelled empirically with a linear fit. The exception was the behaviour and inefficiency in the power converters. The implementation of power converters in electrolysis systems is situation dependent—indeed, direct coupling with solar-PV arrays bypasses the requirement for rectification since the system can run from a DC bus [37].