PPA structure and transaction settlement
There are two main types of potential PPAs considered for a hydrogen production facility that satisfy the DA requirements:
-
a.
Physical PPA: In this arrangement, the corresponding link between the RES producer and the electrolysis plant, is not settled via the wholesale electricity market prices. In this arrangement, two variants are identified:
-
i.
Direct link without a connection to the external grid from either the RES or the electrolysis plant.
-
ii.
A utilization of the main grid to transport electricity from the RES asset to the electrolysis plant. The settlement of this arrangement is based on nominations of energy in the electricity market to the electrolysis plant. Energy nominated is not traded in the wholesale market quantities.
-
i.
-
b.
Virtual PPA: In this arrangement, the energy from renewables is sold to the wholesale market at a specific price in Euros per MWh equal to the market clearing price. The electrolysis plant procures from the market the required energy at the same MCP. Transactions in a virtual PPA between the RES producer and the electrolysis plant are the contract for fifference (CfD)43. It should be noted that while CfDs are generally linked to governmental subsidies, they are a financial instrument that can be applied to a variety of buyer/seller counterparties44. A financially settled Pay-as-Produced (PaP) at a fixed contract price (in Euros per MWh) would apply to each generated and metered MWh per market time unit (MTU) -usually an hour- and would require that the plant pays to the RES producer:
$${C\,f\,D}={E}_{{prod}}* ({{PPA}}_{{Strike\; price}}-{MCP})$$
(1)
where Eprod equals the energy that is produced by the RES in a specific hour, the PPAstrike price is the agreed fixed PPA price, and MCP is the market clearing price in that specific hour. To simplify the market operating model, we assume that the role of the distribution network operator is to manage the grid and provide meter readings, the role of the market operator is to produce associated Guarantees of Origin (GoOs) from RES, the role of the transmission system and network operator (TSO/TNO) is to operate the HV network and associated balancing markets. The role of the energy supplier is to provide settlement of the invoiced energy consumption of the electrolysis facility. PPA contract duration is assumed to be 10 years, Pay-as-produced and fixed price contract45.
For the purposes of this study, we consider that Pay as Consumed (PaC) PPAs do not ad hoc satisfy the additionality principle, since specific renewables must be included in the relevant contract, and thus we exclude them from further analysis. In all cases, the following actors are assumed with their defined roles:
-
RES producer: is the owner and operator of the RES plant;
-
Aggregator/trader: represents the RES asset in the electricity market and may utilize trading capabilities for market optimization;
-
Hydrogen producer: the offtaker of the renewable energy
A conceptual model of the energy interactions between the hydrogen production facility, the RES aggregator, and the RES assets that are included in a relevant PPA contract is depicted in Supplementary Fig. 2.
The energy produced by each RES asset n in hour h is defined as follows:
$$Epro{d}_{nh}={Pro}{{f}}_{nh}\ast Ca{p}_{n}\ast (1-{{degr}}_{re{s}_{ny}})$$
(2)
where,
\({{degr}}_{re{s}_{ny}}\) is the degradation factor applicable for RES asset n in year y,
\({{Prof}}_{{nh}}\) is the hourly production profile of RES asset n and
\({{Cap}}_{n}\) is the capacity of asset n
The different RES profiles are presented in detail in Table 3.
The hourly settlement correlation is modeled as follows:
$${{{{\boldsymbol{PPA}}}}}_{{{{\boldsymbol{h}}}}}= {\sum }_{1}^{h}{\sum }_{1}^{n}{{{PPA}}_{{Strike\; price}}}_{n}* {\sum }_{1}^{n}{{Eprod}}_{{nh}}\\ -a* {\sum }_{1}^{h}{\sum }_{1}^{n}\left({{Eprod}}_{{nh}}-{{Econs}}_{h}\right)* {{MCP}}_{h}$$
(3)
where,
\({{{PPA}}_{{Strike\; price}}}_{n}\) is the PPA contract strike price of RES asset n
\({{Eprod}}_{{nh}}\) is the production of RES asset n in hour h
\({{Econs}}_{h}\) is the energy consumed by the electrolysis unit in hour h
\({{MCP}}_{h}\) is the wholesale market clearing price in hour h
a is a binary variable that annotates the ability to perform market transactions.
$${{{\boldsymbol{a}}}}=\left\{\begin{array}{ll}0, & PPA\,case\,A,i\,(Direct\,link\,without\,connection\,to\,Grid)\hfill\\ 1, & PPA\,case\,Aii(Direct\,link\,with\,connection\,to\,Grid)\,or\,PPA\,case\,B(Virtual\,PPA)\end{array}\right.$$
(4)
The monthly settlement correlation is modeled as follows:
$${{{{\boldsymbol{PPA}}}}}_{{{{\boldsymbol{m}}}}}= {\sum }_{1}^{n}{{{PPA}}_{{Strike\; price}}}_{n}* {{Eprod}}_{{nm}}\\ -a* {\sum }_{1}^{n}\left({{Eprod}}_{{nm}}-{{Econs}}_{m}\right)* {{WAP}}_{m}$$
(5)
where
$$\begin{array}{cc}{{Eprod}}_{{nm}}={\sum }_{0}^{h}{{Eprod}}_{{nh}} & \forall h\in m\end{array}$$
(6)
Energy produced from RES asset n is aggregated on a monthly level (month m) by summing the energy produced on each hour h belonging to month m.
$$\begin{array}{cc}{{Econs}}_{m}={\sum }_{0}^{h}{{Econs}}_{h} & \forall h\in m\end{array}$$
(7)
Energy consumed from the electrolysis plant is the sum of energy consumed for each hour h belonging to month m
\({{WAP}}_{m}\) is the weighted average price of month m
a is a binary variable that annotates the ability to perform market transactions.
$${{{\boldsymbol{a}}}}=\left\{\begin{array}{ll}{{{\bf{0}}}}, & {{{\boldsymbol{PPA}}}}\,{{{\boldsymbol{case}}}}\,{{{\boldsymbol{A}}}},{{{\boldsymbol{i}}}}\,({{{\boldsymbol{Direct}}}}\,{{{\boldsymbol{link}}}}\,{{{\boldsymbol{without}}}}\,{{{\boldsymbol{connection}}}}\,{{{\boldsymbol{to}}}}\,{{{\boldsymbol{Grid}}}})\hfill\\ {{{\bf{1}}}}, & {{{\boldsymbol{PPA}}}}\,{{{\boldsymbol{case}}}}\,{{{\boldsymbol{Aii}}}}({{{\boldsymbol{Direct}}}}\,{{{\boldsymbol{link}}}}\,{{{\boldsymbol{with}}}}\,{{{\boldsymbol{connection}}}}\,{{{\boldsymbol{to}}}}\,{{{\boldsymbol{Grid}}}})\,{{{\boldsymbol{or}}}}\,{{{\boldsymbol{PPA}}}}\,{{{\boldsymbol{case}}}}\,{{{\boldsymbol{B}}}}({{{\boldsymbol{Virtual}}}}\,{{{\boldsymbol{PPA}}}})\end{array}\right.$$
Supplementary Fig. 3 presents a graphical representation of hourly settlement of a 120 MW PV RES asset versus a 30 MW electrolyzer. The aggregated excess energy is 13,372.8 MWhs. Supplementary Fig. 4 presents a graphical representation of monthly settlement of the same configuration. The aggregated excess energy is 2208.12 MWhs.
Electrolyzer model
The most common technologies for electrolysis technologies are alkaline (AEL) and polymer electrolyte membrane (PEMEL)46. The electrolyzer plant contains the electrolyzer stack, which is the core component responsible for Hydrogen production using energy, and the balance of plant (BoP) that includes water purification modules, gas separators, hydraulic pumps, compressors, and oxygen capture modules, as well as other electrical components such as rectifiers34. The electrolyzer unit operates under different states (cut off/ stand-by/production). In the cut-off state, the electrolyzer facility does not produce hydrogen under a certain threshold. The cut-off threshold has been acknowledged in the Electrolyzer model. Stand-by energy has been modeled as an increase in energy required to produce a unit of hydrogen. Concerning compressors, hydraulic piston systems are the most frequently employed due to their ability to handle high pressures, especially when compared to mechanical piston compressors, and their maturity in contrast to membrane compression47. These systems rely on electricity to increase the hydrogen pressure and distribute it to end-users or storage devices, and more than one compressor can be deployed. The energy required for the compressor units is integrated in the overall system-level energy demand.
Therefore, the core characteristics of the facility that are modeled in our framework are presented in Table 8.
Financial model
The LCOH metric value is formulated when the discounted sum of revenues equals the discounted sum of costs. The LCOH is calculated “as produced” at the metering point of the electrolysis plant with no additional storage or transportation cost included. The calculation of the LCOH takes into account any remaining value of the asset at the end of the PPA contract period and incorporating it into the overall TOTEX (Eq. 4).
$${LCOH}=\frac{{\sum }_{y=1}^{N}\frac{{{TOTEX}}_{y}}{{(1+r)}^{y}}-{\sum }_{y=1}^{N}\frac{{{TaxDepr}}_{y}}{{(1+r)}^{y}}}{\mathop{\sum }_{y=1}^{N}\frac{{{QH}2}_{y}}{{(1+r)}^{y}}}$$
(8)
where r stands for the discount rate, TOTEXy are the total (CAPEX and OPEX related) costs in year y, QH2y is the quantity of hydrogen produced in year y in kg, and TaxDepry is the tax depreciation value in year y8. For TaxDeprt calculations, a value of 10% of Electrolyzer CAPEX was used, spread across the 10-year effective electrolyzer lifetime (t). The 10% depreciation value does not apply to engineering, procurement, and construction (EPC) or any other CAPEX costs
The TOTEX component is the sum of both CAPEX and OPEX.CAPEX component is the sum of all costs including all related EPC costs. The development, BoP costs, and the actual cost of the electrolyzer unit are embedded in the CAPEX component, which represents the upfront expenses prior to the initiation of operations or relevant CAPEX during operations. It also includes the cost of land (CoL) if the facility is erected on an owned land. In case the land is rented, the relevant cost is allocated to the OPEX component. These expenses are dependent on the different suppliers or contractors deployed in each case and are presented in terms of Euros per MW installed in Supplementary Table 2. TOTEX, includes cost of water (CoW), CoE, CoL (in case the land upon which the facility is erected, is leased), personnel cost (CoP) and CoM including SLAs with the electrolyzer vendor, as well as any kind of reinvestment costs (CoR) in case of stack replacement minus any remaining value (\({{Value}}_{{rem}}\)) of the asset at the end of the period under evaluation, (\({Projec}{t}_{{life}}\)).
$${{TOTEX}}_{y}= \, {{CAPEX}}_{y}+{{CoW}}_{y}+{{CoL}}_{y}+{{CoM}}_{{ty}}+{{CoE}}_{y} \\ + {{{CoR}}_{y}+{{CoP}}_{y}-{{Value}}_{{rem}}}$$
(9)
$${{CoW}}_{y}={Q}_{H2y}* {w}_{e}* {{Water}}_{\euro}$$
(10)
$${{CoL}}_{y}={H}_{2}{\_}{land}* {{Land}}_{\euro}$$
(11)
$${{CoM}}_{y}=P* {{Maint}}_{\euro}$$
(12)
$${{CoP}}_{y}={Number}{\_}{of}{\_}{Employees}* {Employee}{\_}{Salary}$$
(13)
where \({{{{\rm{Water}}}}}_{\euro}\) is the unitary CoW in Euros per lt, \({{{{\rm{Land}}}}}_{\euro}\) is the unitary CoL expressed in Euros per 1000 m2/year, \({{{{\rm{Maint}}}}}_{\euro}\) is the cost of annual maintenance in Euros per MW installed. The CoW is assumed at 1 Euro per m3 (or 0.001 Euros per lt), based on the current industrial charges for water in most of Greek areas and the CoL at 250 Euros per 1000 m2/year, per market value for non-arable land in rural Greek areas.
The CoE is calculated as follows for hourly settlement:
$${{CoE}}_{y}= \, {{{{\boldsymbol{PPA}}}}}_{{{{\boldsymbol{h}}}}}{{{\boldsymbol{+}}}}\mathop{\sum }_{1}^{h}{PSO}* {{Econs}}_{h}+{\sum }_{1}^{h}{RES}* {{Econs}}_{h}\\ +{System}* {month}$$
(14)
Or for monthly settlement:
$${{CoE}}_{y}= \, {{{{\boldsymbol{PPA}}}}}_{{{{\boldsymbol{m}}}}}{{{\boldsymbol{+}}}}{\sum }_{1}^{m}{PSO}* {{Econs}}_{m}+{\sum }_{1}^{h}{RES}* {{Econs}}_{m}\\ +{System}* {month}$$
(15)
where
PSO are the Public Service Obligation regulatory charges imposed on the consumption of the facility and expressed in Euros per MWh, RES are the levy regulatory charges imposed on the consumption of the facility and expressed in Euros per MWh, system are the power network related regulatory charges expressed in Euros per MW per month, month is the number of months in period t, Unitary values for the regulatory charges can be found on Suplementary Table 3.
The amount of hydrogen produced can be calculated by the linear relationship shown in Eq. 15.
$${QH}{{2}}_{y}{=}{\sum }_{1}^{h}Econ{s}_{h}\ast \left(\frac{1}{e}\right)\ast (1-{degr})$$
(16)
degr represents the stack degradation rate, that results in less gas production per energy consumed. The discount rate (r) is estimated via the WACC for the investment calculated as follows48:
$${WACC}={Deb}{t}_{{ratio}}* {Deb}{t}_{{cost}}* (1-{Tax})+(1-{Deb}{t}_{{ratio}})* {Equit}{y}_{{cost}}$$
(17)
where Tax stands for the corporate tax rate (also called tax shield in terms of WACC calculation), \({Deb}{t}_{{ratio}}\) represents the ratio of debt to capital within an organization, the Debtcost stands for the yield to maturity on existing debt and Equitycost the required rate of return for equity in an investment. In the case of Greece, a corporate tax of 22% was utilized. The WACC resulting in this particular case study mounts to 11.44%.
The utilization of the electrolysis plant can be described as the ratio of the full hydrogen producing hours to the available hours in a specific period (usually a full calendar year)25,49.
$${Utilization\; Factor}\,( \% )=\frac{{\sum }_{y}{Operating\; hours}}{{\sum}_{y}{Total\; hours}}$$
(18)
where operating hours in period y are the equivalent full production hours (h) when the electrolysis facility produces green hydrogen and total hours in period t are the total available hours in given period (8760 for a full calendar year, 744 for a full calendar month, etc).
LCOH minimization
The aim of the optimization problem is to identify the necessary mix of assets and their respective capacity based on the production profile of each asset and their corresponding capacity that results to the lowest possible LCOH, while satisfying a minimum utilization factor for the facility, that implies a minimum amount of Hydrogen production restriction per year. The asset pool is presented in Table 3. The objective function selected is the LCOH, which can better represent the production costs of Hydrogen. The LCOH function as defined in Eq. 8 does not take into account the potential market selling price for Hydrogen itself.
As such, the fitness function of the optimization is reduced to the LCOH formula (Eq. 8), as follows:
$${\min }_{{{Cap}}_{1},{{Cap}}_{2},{{Cap}}_{3},{{Cap}}_{4},{{Cap}}_{5},{{Cap}}_{6},{{Cap}}_{7},{{Cap}}_{8}}{LCOH}$$
(19)
Subject to the following constraints:
-
\({{Cap}}_{n}\ge 0\)
-
\(0{\le {Eprod}}_{{nh}}\) \(\forall\) \(h\in {Projec}{t}_{{life}}\)
-
\({\sum }_{1}^{n}{{Eprod}}_{{nh}}\ge {{Econs}}_{h}\) \(\forall\) \(h\in {Projec}{t}_{{life}}\)
-
\({MinUT}{\le {Utilization\; Factor}}_{y}\le 100 \%\) \(\forall y\in {Projec}{t}_{{life}}\)
-
\({{{Thr}}_{{cut}}\le {Econs}}_{h}\le {P}_{{en}}\) \(\forall h\in {Projec}{t}_{{life}}\)
Where \({{Utilization\; Factor}}_{y}\) is the utilization factor (Eq. 3) in year y, \({MinUT}\) is the minimum allowed utilization factor across all operational years, that ensures a minimum Hydrogen amount production, and \({Projec}{t}_{{life}}\) is the estimated project lifetime.
We have modeled our framework using the expanded GRG optimization engine by frontline systems.
Curtailments impact assessment methodology
To assess curtailments impact, we have isolated additional energy that can be consumed from the grid (otherwise curtailed energy) as
$${{{E}_{{curt}}=P}_{{en}}-{Econs}}_{h}$$
(20)
Pen is defined as maximum energy that can be consumed on a specific hour at maximum capacity P, in Table 8. As \({{Econs}}_{h}\le {P}_{{en}}\) \(\forall h\in {Projec}{t}_{{life}}\) (minimization problem constraints), the \({E}_{{curt}}\) quantity is always greater than zero and represents the capacity of the electrolyzer unit to consume energy up to the maximum energy \({P}_{{en}}\), above the production capacity of the RES unit under the PPA contract, leveraging on grid-scale curtailments.
The electrolyzer can utilize this extra energy by a degree depending on other factors (such as plant availability). This degree is defined as capacity factor (c) and expressed in %. The utilized curtailed energy (\({{UE}}_{{curt}}\)) is
$${{UE}}_{{curt}}=c* {E}_{{curt}}$$
(21)
This additional energy creates a new consumed energy for hydrogen production defined as Econsh’, where,
$${{{{Econs}}_{h}{\prime} ={UE}}_{{curt}}+{Econs}}_{h}$$
(22)
The final consumption replaces original consumed energy (Econsh) in Eq. 16 and increases the produced hydrogen.
The CoE (Eq. 8) is amended as follows:
$${{CoE}}_{t}= {\sum }_{1}^{h}{\sum }_{1}^{n}{{PPA}}_{n}* {\sum }_{1}^{n}{{Eprod}}_{{nh}}+{\sum }_{1}^{h}{PSO}* {{Econs}}_{h}{\prime} \\ +{{Price}}_{{curt}}* {\sum }_{1}^{h}{{UE}}_{{curt\_h}}+{\sum }_{1}^{h}{RES}* {{Econs}}_{h}{\prime}\\ +{System}* {month}-a* {{Mkt}}_{h}$$
(23)
Where \({{Price}}_{{curt}}\) is the curtailment price (in Euros per MWh) that would be charged for the extra energy consumed.
Inputs
Electricity network charges
Regulated electricity charges are used to model the costs related to network charges. The values applicable to the Greek Transmission Network (Supplementary Table 3) are used, according to the 198/2023 Greek NRA50.
According to Greek Law 4951/2022, all electricity storage facilities, either standalone or combined with RES assets, are exempt from PSO and RES levy charges. Regarding system charges, storage stations are charged according to a specific methodology, approved by the NRA, to the extent that their demanded energy contributes to the system peak. If a similar methodology and discount approach is considered for hydrogen facilities, the discounted charges amount to 2200 Euros per MW per month installed with no energy-related charges. The potential discounted values for hydrogen facilities (if they are regarded as storage facilities would drive PSO and RES levy charges to 0 Euros per MWh and the system charges to 2200 Euros per MW per month.
Renewables
Renewable energy production profiles and locations were taken from historical values of operational plants from a renewable’s aggregator in Greece with more than 850 MW RES portfolio mix of solar and wind projects under management. Data were curated and cleansed in order to create hourly normalized profiles for each location. Generic RES production profiles for the Greek bidding zone were used from the Greek NRA51 as a benchmark which utilized the cost of new entry (CONE) methodology. Subsequently, a PPA price was estimated based on each asset’s performance vs the baseline generic equivalent, while taking into account the current level of wholesale electricity market prices. The technical characteristics and PPA prices of the RES Assets used are shown in Table 3. RES asset degradation was estimated by technology at 0.5% per year for wind farms52,53 and 0.3% per year for solar installations54,55.
Wholesale electricity market prices
Greek electricity market price scenarios are utilized in the reported analysis; the base reference scenario is based on the Greek NECP56, a low price scenario, and a high price scenario. MCP is modeled on an hourly basis throughout the period under consideration. Results of the simulation and aggregated day-ahead market, market prices for the decade 2030–2039 are presented in Supplementary Fig. 1. 2030 is the initial year where the temporal correlation condition will be applied on an hourly basis, since a transitional period with monthly temporal correlation is provisioned until 2030. Both monthly and hourly temporal correlation is further explored in the study scenarios, presented in Chapter III.
Electrolysis system inputs
A global tender exercise for a 30 MW electrolyzer facility took place during June–September 2023, as part of the EPHYRA project requirements. A total of nine different manufacturers responded (two PEMEL and seven AEL), providing a total of ten separate technical and commercial proposals. In order to enable technologies comparison, an “average PEMEL” and an “average AEL” electrolyzer have been generated, as averages of the data provided by each vendor. The information is summarized in Table 5 and is relatively aligned with the strategic research and innovation agenda by the clean hydrogen JU state-of-the-art data for 2020 and relevant 2024 targets35.
In addition to the global tender, detailed market research took place that consolidated input from various electrolyzer systems manufacturers based on publicly available data and proprietary quotes, in order to solidify the results of the global tender and ensure the consistency of the received offers. An additional reason behind the separate market research was to validate the outstanding outliers observed throughout the tender process in a number of metrics and KPIs.
Electrolyzer stack replacement takes place according to vendor specifications. A number of vendors provide a guarantee in years, while others in hours of operation. In this case study, a stack refurbishment is assumed for all electrolyzer models after 80,000 h of operation9. If the 80,000-h limit is not reached within the 10-year period, the remaining value of the equipment is calculated and its depreciated value counted against the CAPEX component of the LCOH. Under the current simulations, the trigger of 80,000 h of operation was not reached under any scenario; as such the CoR component of Eq. (4) is set to zero €. CoMs are considered as 2.5% of CAPEX based on8,9,57, for the simulation requirements. Regarding CoPs, a total of 8 employees are assumed with an average cost of 30,000 Euros per year/employee. The number of employees is suggested by the vendors proposals and is also supported by the MOH refinery operations. The cost assumption is based on the average wages for site engineers in Greece according to Randstad HR and the salaries tends 2024 report58. CoMs can be treated under two different approaches. The first approach provides for an annual fixed CoM proportional to the installed capacity and the CAPEX. The second approach provides for a non-static CoM, linked to the strain of the equipment and relevant utilization levels and Hydrogen production. The static approach is preferred in both vendor quotes as well as relevant literature8,9,57.
