The concept of hydrogen production from electricity summarized under the term “power-to-gas” is not new, including various concepts of producing hydrogen from wind energy. Different concepts and projects are described in [8, 15, 16, 19–21].
Decentralized hydrogen production facilities are likely to be placed at fuel stations. These plants would be grid connected and could be used as grid-stabilizing facilities.
All grid-connected concepts are more likely to be implemented where intermittent renewable energy is already largely in place to store excess electricity and stabilize the grid.
Grid connection is state-of-the-art but also costly and a major cause of conversion and transmission losses. However, most “wind-to-gas” concepts refer to grid-connected wind farms. Considering the size of such plants to match Norway’s theoretical demand of hydrogen as a transportation fuel which is 450,000 tons, over 1,200 of the largest today produced electrolysers and over 4 billion liters of water per year would be needed.
Large areas would be necessary to build these plants onshore which could be compensated with offshore sited hydrogen production, using freely available sea water.
Furthermore, using offshore wind reservoirs solely to produce hydrogen could be a solution to make offshore wind energy exploitable in Norway. It would exclude the costs of grid connection and problems caused to the grid due to intermittency, and improve the acceptance of wind energy in the society due to these problems.
But at the same time new challenges related to producing hydrogen offshore arise, for example hydrogen transportation to shore and new technological requirements.
A 100 megawatt (MW) wind farm will constitute the baseline of this study.
Technical assessment of offshore hydrogen production
In this section, the technical necessities of offshore-based hydrogen production will be described. The most feasible cases will be selected based on median values of literature review.
A base case as the most realistic scenario will be created between best and worst case scenarios. Based on this, the yearly produced amount of hydrogen for each scenario will be calculated, as well as the specific energy demand and the efficiency.
Wind farm
The electric energy supply of this study will come from a 100 MW offshore wind park. The overall potential for exploitable offshore-generated electricity in Norway is estimated between 18 and 45 TWh per year. This number is based on a study provided by the Norwegian Water Resources and Energy Directorate that proposes 15 locations for offshore wind farms with magnitudes between 100 and 2000 MW [22].
As wind farms are usually developed over time, it is practicable to use a simple number as a base case. Mathur et al. have shown that the capacity of 100 MW represents the minimum capacity for economically feasible hydrogen production using offshore wind [23].
Distance from the shore is included as a cost factor in the economical assessment.
The area selected is close to the German offshore wind farm alpha-ventus, based on which operating data an output profile was created [25]Footnote 2 and extrapolated to the capacity profile of a 100 MW wind farm (Figs. 1, 2).
In discussion with CMR Prototech engineers and their technology manager [26] it has been agreed that a minimum of 5 % of the installed capacity is necessary for a stable system. Therefore, only power produced above 5 % of the installed capacity is taken into account. However, sufficient production is available throughout 75 % of the year.
Figure 3 shows the output profile of alpha-ventus for a full year of production.
Taking all electricity above 5 % of the installed capacity into account adds up to . This will be taken as the worst case scenario since it is likely that the actual available power is higher because the output of the analyzed offshore farm alpha ventus (60 MW) was in 2011 [27]. In addition, transmission losses of electricity are less when there is no grid connection. As seen in Table 2, full production hours are estimated to 4,050 h per year which results in total electric power output of and of usable electricity extrapolated to the capacity profile and excluding power under 5 %. This is taken as a best-case scenario.
Table 2 Specification of electrolysis
For the base case, the arithmetic mean between the worst and best-case values is taken and extrapolated to the output profile. This adds up to .
Hydrogen production platform
The platform where all systems for hydrogen production are based is the main uncertainty in economic terms. However, offshore platforms are state-of-the-art technology. The offshore oil and gas industry builds platforms where sophisticated machinery, such as chemical processing, drilling and other refinery processes as well as living quarters, is placed on the platform [29]. Most offshore wind farms need substations to collect the cables for grid connection and for power transformation and conversion [30, 31]. Offshore platforms are a well-known technology in Norway, where the biggest offshore gas platform Troll A has been built [32, 33]. Therefore, the technical feasibility is given.
Offshore wind output data describes the usable electricity including transformation and transmitting. Transmission losses occur only marginally as electricity is used locally. Li et al. [34] report high efficiencies for electrolyser power converters up to 95 %. Therefore, the estimated power output will be seen as the usable power for the system including current converting and transformation.
Electrolyser systems and analysis of electrolysis process
The electrolyser is the core system which makes it the main element to be examined. There are four principle ways of electrolysis shown in Table 2.
Brine electrolysis would be the most obvious system as it uses a sodium chloride (NaCl) water solution which is basically a form of concentrated sea water [35]. However, there was close to no data found on sea water electrolysis. [36] and [37] stated that due to impurities and the insufficient concentration of NaCl, brine electrolysis is possible but not in the focus of their research. It is, however, a standardized industry process to produce caustic soda with hydrogen as byproduct.
The most common electrolysis process is alkaline electrolysis which uses potassium hydroxide (KOH)—water solution, which would require transport and storage of KOH and is therefore not considered as feasible either.
The proton exchange membrane electrolysis cell (PEMEC) and the solid oxide electrolysis cell (SOEC) are electrolysis solutions which require only water as feed in. Further analysis will be done for these options. Specifications of the chosen electrolysers are shown in Table 3 in comparison with alkaline electrolysis.
Table 3 Efficiencies and specifications of electrolysis solutions
When examining the operation of electrolysers, the characteristics of offshore wind have to be considered. Offshore wind power is highly variable. However, according to the European Wind Energy Association (EWEA) it is not intermittent meaning irregular and unpredictable changes or start/stop intervals in power output on a minute or even second basis do not occur. Short-term variability (within the minute) is not an issue, while variations within the hour are significant [45].
The variability is the crucial factor for the dimensioning of an electrolysis-based hydrogen plant, as it requires electrolysers and auxiliary systems to be able to handle this variation and power converters to deliver the right voltage at different capacities with more or less same efficiency. Grid-connected electrolysis concepts have the advantage that electrolysers can run steadily at the same capacity and run throughout the whole year.
As mentioned in 2.1.1 sufficient electricity (more than 5 %) is produced 75 % of the time but the power output varies strongly in a range of 5–80 % of the installed capacity. The positive implication is that the wind farm can deliver energy throughout most of the year, so that annual production amounts of hydrogen could be predicted rather safely. Downtime is not so much of an issue as the handling of varying loads.
Siemens has recently introduced a megawatt electrolysis prototype (PEM) that can adapt changes in load within seconds and is able to process three times its nominal capacity for a period of time [38].
The electrolysers analyzed in this study are seen to have a nominal capacity of 50 % of the maximum available power. Change in capacity loads will affect efficiencies to some extent. According to Bartels et al. [19] this affects alkaline and PEM electrolysis most and can lead to dangerous conditions at very low capacities, because gaseous hydrogen and oxygen can evaporate through the membrane and create an explosive environment.
To reduce complexity of this study efficiencies of auxiliary processes are seen as set for each scenario. A factor for the decrease of efficiency due to capacity loads of the electrolyser is introduced as a range of 80–100 % where 50 % of the wind farms capacity is seen as the ideal capacity level of the electrolyser. The decision to take into account only electricity above 5 % of the minimum capacity is also in respect of varying efficiencies at different capacity loads. The ideal level of dimensioning has to be further researched and simulated.
Both chosen electrolysers use water as feed-in stream, but are different in process and specifications. The solid oxide electrolyser (SOEC) needs a steam generator and a high-capacity compressor is necessary as hydrogen is produced at atmospheric pressure. The proton exchange membrane electrolyser (PEM) works pressurized so the compressor for transportation requires less power and the input stream of the PEM electrolyser is fresh water. The SOEC ceramic material requires no noble and rare metals as the PEM, which makes it potentially cheaper.
The basic electrolysis reaction is
(1)
This reaction is endothermic and the required energy by the process is
where is the change of enthalpy, is the Gibbs free energy and represents the minimum electrical energy and is the process temperature and the entropy change and represents the minimum thermal energy demand [41, 46, 47].
ΔG can compensate the thermal energy demand and a change of temperature can lower the electric energy demand [48].
Based on the required cell voltage in the electrolyser can be determined with
where 2 is the number of electrons and F is the Faraday constant which is [49–52]. ΔH is usually the higher heating value (HHV) of hydrogen. The important difference here is as shown in Table 3 that the SOEC electrolyser can operate at a cell voltage below minimum cell voltage of the higher heating value of hydrogenFootnote 3 which is
(4)
and in some cases even below the cell voltage of the lower heating value which is
(5)
where is the molar mass of hydrogen. This shows how the increased temperature influences the energy need of hydrogen production and this is the reason why in theory efficiency above 100 % can be accomplished.
The amount of hydrogen produced can be calculated by
(6)
where is the available energy, V
C(V) is the cell voltage, F is Faraday’s constant as introduced above, M is the molar mass of hydrogen and η is the electrolyser efficiency.
The overall process efficiency can then be expressed when comparing the specific energy demand in to the thermodynamic minimum as referring to the higher heating value of
Water treatment
Both the SOEC and PEM electrolyser cannot be operated directly on sea water but need process water or boiler feed water quality, respectively, with a maximum of 0.5 ppm. Total dissolved units (TDS) [53]. Water treatment includes desalination and purification for the PEM and the SOEC and steam generation only for the latter. Desalination processes can be divided into electrical and thermal. Reverse osmosis (electrical) is the most commonly used technology, however, thermal processes such as multi-effect distillation and multi-stage flash distillation produce better quality and require less post-treatment for demineralization [54–56].
Post-treatment always includes chemical treatment in a resin polishing filter containing chemicals to bind remaining ions and other dissolved solids in the desalinated water. In the desalination process chemicals are also included to prevent scale [55, 57].
Using chemicals is undesirable as changing or refilling of the chemicals is inevitably more difficult on an isolated offshore platform. In the reviewed literature, there was no data available to technically quantify the necessary chemicals and the frequency of refilling. This is one additional challenge of the offshore approach that has to be further investigated.
Calculations done will be made based on the given efficiencies assuming that the whole process is described including the chemical treatment, necessary pumps and further equipment.
Thermal water treatment is the direct logical choice for the SOEC electrolyser since it requires steam. Reverse osmosis is slightly more efficient but it delivers lower quality (400 ppm TDS versus 5 ppm), and therefore requires more sophisticated post-treatment, which should be avoided. For these reasons thermal desalination will be chosen for both electrolysers, namely multistage flash distillation as it is mentioned for producing boiler feed water quality with [54] on a temperature range 70–120 °C.
The energy needed can be determined by multiplying the energy need per cubic meterFootnote 4 with the amount of water that is necessary which is
(8)
where is the production rate of hydrogen, M is the molar mass of hydrogen (), η is the efficiency of water use and is the molar mass of water ().
The PEM electrolyser system is less complex which leaves more electricity for the electrolysis process. The electrolysis itself is slightly less efficient and the main disadvantage is the use of expensive materials such as platinum. The distillated and demineralized water proceeds directly into the electrolyser, is split up into hydrogen and oxygen and then hydrogen proceeds into the compressor.
For the SOEC the water is preceded into an electric steam boiler which is state-of-the-art for other industry processes [58] and then conducted into the electrolyser. The energy needed for steam generation can be determined with
(9)
where Δh is the enthalpy change for water between ~100 °C and the desired temperature and η is the boiler efficiency that can be up to 99 % [59–61].
In the SOEC electrolyser there will be also a mass flow of air and some of the produced hydrogen to increase efficiency. To keep the electrolysis cell hot while the wind farm delivers insufficient energy wind power below 5 % can still be used. During zero electricity production times it is assumed that the SOEC system will consume some hydrogen to retain the heat for a faster process start up [26].
Leaving the electrolyser, hydrogen is separated from steam and then inducted into the compressor. In both process designs heat exchangers have been included to show potential of efficiency increase.
The advantages of the SOEC are the availability and price of materials and the increasing efficiency due to the process heat, though the auxiliary equipment also requires more of the energy available. The main disadvantage is the immaturity of the system in comparison to all other electrolyser solutions and the need of steam and a high-capacity compressor which makes the process design more complex [62, 63].
Compression and transportation of produced hydrogen
One main difference between the SOEC and the PEM electrolyser is that the SOEC operates at atmospheric pressure while the PEM electrolyser operates pressurized and is in some systems described to substitute a compressor completely.
The energy of compression can be calculated with
(10)
where n is the molar amount of hydrogen, is the universal gas constant, is the logarithm of the ratio of pressure after and before compression and η is the compressor efficiency. Compressor efficiency is around 70 % [64]. Handling different amounts of hydrogen due to variable capacity loads will reduce efficiency especially for the SOEC system. Therefore, efficiency of 50 % is used.
For the proposed 100 MW system it could be more feasible to transport the hydrogen by ship. Pipelines are a standard solution to transport gas from offshore platforms and are the only choice when the proposed project is scaled up. Subsea pipelines can also be used to deliver hydrogen to other countries as done in the case of natural gas [29].
Pipeline pressure is between 25 and 300 bar in different sources [29, 65, 66]. For simplification the selected pressure is 100 bar.
Economical assessment of offshore hydrogen production
The objective of the economical assessment is to see what price per kg of hydrogen for the proposed system will be.
Reviewed literature shows that costs of all components are proposed in a range of specific costs, e.g., or . The values from different studies are applied for three scenarios. Based on the maximum capacity of the components the total investment cost can be estimated.
Price ranges are consistent for most components. The main uncertainty is the platform cost. The estimation for this will be based on the values for platform cost of a wind farm presented in the offshore wind assessment for Norway [67] and compared to the costs of offshore oil rigs to see if the picture is realistic. The price ranges of electrolysers vary strongly between [20] and [39]. These varying costs have been put into three more realistically moderate-ranged scenarios.
Based on the collected data for all the components of the system the total investment and the annual costs have been estimated using the comparative cost method. The selected cost ranges of the different components for each scenario are presented in Table 4 and based on literature review from [20–23, 33, 34, 39, 43, 54, 65, 67–73].
Table 4 Price range of components
The wind cost can be modified to the specifications of the proposed system. According to [67] electrical equipment, substation and grid connection adds up to 20 % of the wind investment cost. These components of the wind farm are unneeded for this system. Therefore, 20 % has been subtracted from the wind farm investment. Green and Vasilakos [68] present a table by the European Environment Agency (EEA) with multiplying factors to adjust the cost to water depth and distance from shore. These have been added to the calculation to simulate different distances from shore. The 20 % basis has been used to estimate the platform and electrical equipment price. For this it was estimated that the platform will be 2.5 times the size and therefore 2.5 times the investment cost of 20 % of the wind farm total investment. This adds up to a widely spread price range. Costs for oil rigs are in a similarly wide range [74]. The pipeline costs have been added to the scenario with different distances to clarify the influence of transportation cost. The final annual costs have been carried out with and without transportation.
Figure 4 shows the influence on the total investment for each component of the system. The major cost component in both systems is the wind farm. The total investment in the base case is 716.15 M € for the PEM System and 640.85 M € for the SOEC system which is about 12 % less. This is mainly due to the influence of the size of the electrolyser and because auxiliary equipment is inexpensive when compared to wind farm and electrolyser. The implication is that even though the SOEC system needs a far more sophisticated overall system with steam generation and more powered compression, its total cost could be lowered.
The potential demand of hydrogen for Norwegian domestic transportation is 450,000 tons per year [17, 75]. Over 100 times the proposed systems size would be necessary to produce this amount of hydrogen.
With an average fuel consumption of 7.5 Liters per 100 km [76] and an average fuel price of 2.07 € (15.65 NOK) [77] per Liter, hydrogen produced as transportation fuel would be profitable if it can be produced on specific costs below
(11)
The present price of hydrogen in Norway per kg is 11.83 € according to [5].