OPTIMAL CO-PRODUCTION OF MARKET BASED POWER GRID SUPPORT AND
RENEWABLE FUELS OR CHEMICALS
Robert Weiss*, Lotta Kannari, Jari Pennanen, Teemu Sihvonen, Jouni Savolainen
VTT Technical Research Centre of Finland Ltd, Finland
*Corresponding author: email@example.com
This paper addresses the operational profitability and benefits of co-production and dynamics
in renewable Power-to-Fuel or Power-to-Chemical processes, when utilized to stabilize the power grid.
To reach climate neutrality, the transport sector must switch to climate-neutral or climate-positive fuels
or power. This can be done by a proper mix of electric vehicles where the power is based on renewable
electricity, hydrogen vehicles where the hydrogen is based on side-product hydrogen or renewable
power based processes, or by using in existing vehicle fleet renewable biofuels instead of fossil fuels.
For hydrogen-based or hydrocarbon-based chemical industry, similar change needs are obvious.
Today, renewable fuels can be produced with biomass based plants or in bio-refineries. If biomass
cannot be utilized or there is a lack of some sustainable chemicals like hydrogen in the bio-process,
new so called Power-to-X plants [1, 2 and 3] could be used to produce renewable fuels or chemicals of
non-biomass origin (X refers to the end product of the plant). These processes are utilizing hydro, solar
or wind power via electrolysis for production of the needed additional hydrogen fraction .
Optimized day- or week-ahead planning can increase considerably the value of renewable Power-to-X
plant operation. Such optimal operation can enhance commercial viability of the Power-to-X plants,
and in this way introduce these climate-positive, mid-sized and distributed units into a smart grid. In
this paper, we show examples on the value of such optimal dynamic operation together with the value
of the dynamics of key unit processes and intermediate storages, while at the same time reaching a
high share of sustainable solar and wind power in the end product.
Sustainable Power-to-X Concept
Power-to-X (PtX) plant’s main products can be typically fuels (H2, CH4, CH3OH, Fischer-
Tropsch-products), chemicals (NH3, light olefins C2H4, C3H6), or reduction agents (H2 for metals).
Side products are high purity oxygen and low grade waste heat from the electrolysis operation, as well
as process steam from some PtX processing units that are highly exotermic, like catalytic methanation.
These side products can generally only be economically utilized locally “over-the-fence”, and were
neglected in this analysis for generality reasons.
For sustainability and avoidance of fossil fuels, the PtX operations should avoid fossil fuel
based grid power, and instead utilize high enough content of sustainable renewable power. In our case
examples we target to utilize as much wind and solar power as possible which are intermittent. On the
other hand, parts of the PtX process may not be as flexible as the intermittent renewable power. Or the
demand for the end product itself might even be 24/7, stable, but completely inflexible industrial
demand. Finally, for sustainability the PtX plant should participate in power grid stabilization. This
operating environment, depicted in Fig. 1, render the production planning a challenging task,
necessitating the optimization approach adopted in this paper.
Figure 1 PtX concept and product examples.
Power-to-X Plant Operation and Dynamic Optimization Model
In our analysis, PtX power market contract handling as well as the physical production process
is formulated as a dynamic, linear optimization model described in detail in , which minimizes the
production cost. To our model, we added risk handling related to the power grid ancillary service as
well as the utilization of available intermittent solar and wind power as described in the next sub-
For our analysis, we configured a 4x1.25 MWe, state-of-the-art commercial PEM electrolyzer
system utilizing confidential manufacturer data. As illustrative examples, we present the following
example cases in our analysis: Case1 describes on-site electrolytic production of H2 as metal reduction
agent, with a stable but unflexible H2 demand. Case2 describes Power-to-Gas production of CH4 as
green synthetic natural gas for sustainable mobility, with a flexibility range 70-100% in the catalytic
methanation operation. In both cases, we investigate the impact of a 0-24h H2-buffer storage. The
optimization model schematic and case boundaries are depicted in Fig. 2.
Figure 2 Schematic of PtX scheduling and dynamic optimization model, with the boundaries of
example Case1 and Case2.
Markets for Wholesale Power and Ancillary Services
Deregulated electricity markets are generally divided into wholesale energy markets managed
by power exchanges, and ancillary service markets managed by Transmission System Operators
(TSO). Wholsesale energy is sold and purchased typically on international day-ahead markets for the
24 hours, see e.g.  for Nordic System, Baltic countries and Germany, or  for Germany, Austria,
France and Switzerland. Concerning the ancillary service markets, situation of the Nordic countries
compared to Netherlands, Germany and Poland (UCTE area) is well described in . We list some
examples on European ancillary service products in Table 1 below.
Table 1 Example on European Primary Frequency Control Schemes
Frequency Containment Reserve -
Normal Operation (FCR-N)
Also known as Primary Frequency Control or PRL
European 50Hz Grids: 49.9 – 50.1Hz or 49.8 – 50.2 Hz control band.
Generally a symmetric reserve which is continuously triggered.
Asymmetric bidding possible in some markets e.g. Denmark.
Day-ahead markets in e.g. Finland, Denmark, and Norway. Week-ahead
markets in Germany, Austria, Switzerland and The Netherlands.
Frequency Containment Reserve –
Scandinavian 50Hz Grids: 49.9 – 49.5 Hz (only disturbances).
Asymmetric, fast. Not often triggered.
Day-ahead markets in e.g. Finland, Denmark, and Norway.
In a power system with high penetration of intermittent renewable power production, grid
balancing support and ancillary service prices tend to be more and more volatile. As example on this
effect, we utilized the hourly markets of the Finnish primary frequency control products FCR-N and
FCR-D . In this market, a large number of Combined Heat and Power plants (CHP) are on-line in
the winter resulting in high power system intertia and a corresponding low need for additional FCR-N
and FCR-D. In contrast, these plants are generally off-line in the summer, with nuclear power plants in
maintenace, resulting in low intertia in the system and high prices for hourly additional FCR-N and
FCR-D. Examples on this price volatility is shown in Fig. 3 below.
Figure 3 Hourly prices for energy  and FCR  during winter and summer example weeks.
Uncertainty in Disturbances and Ancillary Service Activations, Robust Optimization Solution
Most ancillary services, including primary frequency control like FCR-N and FCR-D, are
capacity options with sometimes a strong uncertainty in the real need for their activation. The power
system can have prolonged periods of under- or overfrequency, leading even in symmetric freqency
control products (which theoretically should have 0 as net cumulative response) to long cumulative
distortions towards large up or down regulation amounts, as depicted in Fig. 4.
Figure 4 Left: FCR-N symmetric control scheme in Finland . Right: Example on the 24 hour
cumulative distortions up/down for the response of some primary frequency control schemes.
In our model, we handle such uncertainty with a robust optimization approach, where we solve
the conflict between the need for stable unit process conditions versus the ancillary services’ need for
capacity reservation, fast reactions as well as prolonged activation worst case distortions up/down. Our
ancillary service activation risk calculations utilize the budget uncertainty set methods from robust
optimization theory [10, 11]. Uncertainty sets were defined for combined maximum and duration
distributions of the ancillary service control response. These uncertainty set definitions were integrated
into the optimization model for the electrolyzer power consumption and H2 production, H2 storage as
well as H2 and power balances. This resulted in a min-max bi-level optimization problem which was
transformed into a LP model utilizing the duality of the problem. For the Finnish FCR-N and FCR-D
schemes, budget uncertainty sets were statistically identified based on the primary frequency control
scheme responses and cumulative worst cases identified from TSO frequency data, with a 0.1s
resolution for the entire year 2013. These identified uncertainty sets were utilized in our analysis.
Renewable Power and Dimensioning of a PtX Plant
Wind power from wind parks or larger areas can have short storm periods of very high power
output still representing only a small energy amount of the entire yearly wind power production, as
indicated in Fig 5. If a high annual wind energy share is desired for the PtX production, oversizing the
nominal wind power capacity should be considered. Using additional solar power can be an advantage,
since wind and solar power production are generally not strongly correlated. Also, in Northern
countries like Finland, winds can generally be stronger in autumn and winter while solar production is
concentrated to spring and summer. In our example, national wind power measurements from  and
local solar PV measurements from  where re-dimensioned to 133% and 100% nominal capacity of
the PtX plant, respectively, as shown in Fig. 5. Together, they resulted in a theoretical minimum
curtailment energy loss below 2% of the yearly solar and wind energy available.
Figure 5 PtX plant nominal size compared to available wind and solar power during one year.
In our optimization study, we assumed wind and solar power to be directly and locally
contracted on a yearly base, with only fixed costs but zero marginal costs when available. Forecasting
errors of available wind and solar power were assumed to be balanced on average to zero costs on the
balancing market. For hours when available wind or solar power where insufficient, the needed
additional power could be purchased from the wholesale market, subject to transmission fees and taxes
but no transmission congestions. For simplicity, we assumed that surplus wind and solar power not
used had to be curtailed at zero prices, not able to reach the market for various congestion reasons.
Results and Discussion
In the analysis below we define the part load level sizing of the electrolyzer as the ratio of
stable H2 demand divided by the electrolyzer nominal H2 capacity. Net power costs are defined as the
wholesale power cost + wind and solar power costs – income from FCR-N and FCR-D hourly ancillary
CASE 1: PtX with Stable but Inflexible H2 Demand
Net power costs seem to be clearly lower if the electrolyzer is sized to 50-70% partial load (i.e.
oversized) and larger H2 buffer size decreased clearly the net power costs at all part load levels, as
shown in Fig. 6 below. This finding applies both for the whole year analysis, as well as for the selected
example weeks displayed with strongly differing pricing in FCR ancillary service (see Fig. 3). The
FCR income was very high in the summer weeks, up to 3 US$/kgH2, if large H2-buffer space was
available. For smaller H2-buffers, our robust optimization restricted strongly the utilization of FCR
according to the identified uncertainty sets for FCR control activation and PtX plant dynamics.
Figure 6 Net power costs for various H2-buffer sizes during the example weeks. For a 24h H2-buffer,
the FCR income would have led to a net power income in the summer week.
Current state-of-the art commercial PEM and AEC electrolyzer technology can also be offered
with an overload option, meaning that the electrolyzer may run for a short time substantially over its
nominal capacity, but with less efficiency, and must thereafter cool down in the peripheral equipment
like cooling circuits and transformers. In our analysis, we evaluated a theoretical +60% overload with a
maximum utilization cycle of 0.5h overload + 0.5h cooling, which is quite close to current offerings on
the PEM markets. The overload option showed no benefit during the winter example week due to low
hourly FCR-N&D prices and absence of heavy peaks in wind power. However, during the summer
example week, the overload option could bring substantial savings but only if a large enough H2 buffer
was available storing the overproduction (because of inflexible H2-demand).
Figure 7 Electrolyzer overload option and power cost savings during the summer example week.
Calculating the hourly dynamic operation over an entire year, we analyzed the situation using a
24h H2-buffer and 60% overload option, with results displayed in Fig. 8. Almost all available wind and
solar power was used if the electrolyzer is sized to 100% i.e. full load. However, the wind and solar
power usage decreases if the electrolyzer is sized to part load, which was accelerated below a 70% part
load sizing. The reason for this utilization decrease was not the PtX plant flexibility, but the lack of H2
demand at part load sizing. The share of sustainable power-based H2 shows an opposite pattern: The
share is as low as 50% for full load sizing, but can grow to close to 90% for a 30% part load sizing,
being 65% at a 70% part load sizing. The reason for this utilization increase was the increased PtX
plant flexibility at part load.
Figure 8 Dynamic operation results for entire year, Case 1. Left Up: Utilization of Wind&Solar-
power, Share of Wind&Solar-based H2, Left Down: Net power costs, Right: Total production costs.
Based on estimates in  and confidential information from manufacturers, for current
commercial 1-1.5 MWe AEC or PEM electrolyzer we estimate a fixed cost range between 1500 to
3500 US$/kWe for CAPEX and 10 year O&M + insurance costs, which is indicated with a red bar in
Fig 8. If we use old “stranded” wind power at an O&M cost of 16 US$/MWh, the cost range would
result in total H2-prices 1.3-3.0 US$/kgH2, which are close to or even competitive with steam
reforming (avg. cost 1.7 US$/kg). When using new wind power at LCOE of 48US$/MWh,
competitiveness can be reached only if electrolyzer CAPEX is clearly below 1000 US$/kWe.
CASE 2: PtX for flexible CH4 production
Results were generally similar to Case1 and are displayed in Fig 9 below. Under-sizing the
methanation unit towards the electrolyzer with a ratio 0.7:1 reduced the utilization of available solar
and wind power, but increased the share of solar and wind based CH4 in the end product gas mixture.
The H2 buffer can be kept at a minimum size of 1-4 hours, and larger buffer does not affect results at
all, except if a longer term storage (week or larger) can be added. A more flexible methanation
(operation range 10-100% instead of 70-100%) increased by 10% the share of solar and wind based
CH4 in the end product gas mixture, when the methanation unit was sized 1:1 towards the electrolyzer.
Figure 9 Dynamic operation results for entire year, Case 2. Left: Utilization of Wind&Solar-power,
Share of Wind&Solar-based CH4, and Net power costs, Right: Total production costs.
Methanation CAPEX estimation was challenging, since small- and medium scale methanation is not
available on the market today. Using the fixed cost estimates for the electrolyzer as presented in
previous sub-chapter, and estimates in  for methanation, we can see from the results in Fig. 9 that
competitive CH4-prices compared to current US SPOT prices (< 3 US$/mmBTU) seem challenging to
reach. Higher CO2-allowance prices and/or cost efficiency improvements in the electrolyzer and
methanation manufacturing, and technology improvements could however change the picture.
PtX example cases for solar- and wind power-based H2 and CH4 production were shown for
Finnish climate, power market and grid conditions for fast ancillary services. Evidently, concurrent
sizing and operational planning is needed to find economical potential solutions and plant
configurations. Participation in power grid ancillary services, e.g. primary frequency control, is
essential both for plant economics and power system sustainability. Also, state-of-the-art PEM and
AEC electrolyzer offer capacity to temporarily overload during primary frequency control price peaks,
which is economically beneficial. We demonstrated the ability of robust optimization to handle the
activation uncertainty of fast ancillary services, which proved to be an essential part of the
Competitive H2-prices seem to be reachable for PtX compared to steam reforming, even for an
inflexible end customer. In contrary, the sustainable CH4 case seems to require technological advances
or more benefits for the reuse CO2 and avoidance of fossil fuels. Alternatively, dedicated markets have
to be found or created where conventional natural gas cannot be offered, e.g. bio-methane as a
sustainable transportation fuel.
1. Gahleitner, Gerda (2013), “Hydrogen from renewable electricity: An international review of
power-to-gas pilot plants for stationary applications”, International Journal of Hydrogen
Energy, 38(5), February, pp. 2039-2061.
2. Rönsch, Stefan, Schneider, Jens, Matthischke, Steffi, Schlüter, Michael, Götz, Manuel,
Lefebvre, Jonathan, Prabhakaran, Praseeth and Siegfried Bajohr (2015), “Review on
methanation – From fundamentals to current projects”, Fuel 166, pp. 276–296.
3. Götz, Manuel, Lefebvre, Jonathan, Mörs, Friedemann, McDaniel Koch, Amy, Graf, Frank,
Bajohr, Siegfried, Reimert, Rainer and Thomas Kolb (2015), “Renewable Power-to-Gas: A
technological and economic review”, Renewable Energy 85, January, pp. 1371–1390.
4. Weiss, Robert, Savolainen, Jouni, Peltoniemi, Pasi and Eero Inkeri (2016), “Optimal
scheduling of a P2G plant in dynamic Power, Regulation and Gas markets”, in: 10th
International Renewable Energy Storage (IRES 2016), Düsseldorf.
5. Nordpool (2016), http://nordpoolspot.com/
6. EPEX (2016), https://www.eex.com/en/products/power/power-spot-market
7. Grande, Ove S., Doorman, Gerard and Bjorn H. Bakken (2008), “Exchange of balancing
resources between the Nordic synchronous system and the Netherlands / Germany / Poland.”
SINTEF Energy Research, Technical Report.
8. Fingrid Oyj (2016), Finnish Transmission System Operator, Helsinki, www.fingrid.fi
9. HELEN Ltd (2016), https://www.helen.fi/sahko/kodit/aurinkosahko/suvilahti/
10. Ben-Tal, Aharon, El Ghaoui, Laurent and Arkadi Nemirovski (2009), “Robust Optimization”,
Princeton University Press.
11. Bertsimas, Dimitris, Brown David B. and Constantine Caramanis (2011), “Theory and
Applications of Robust Optimization”, SIAM Review 53(3):464-501.