You are on page 1of 6

NEIS Conference 2019 Â 19. – 20.09.

2019 in Hamburg

Transforming Fluctuating Wind Power to Base Load – How about


Power-to-Ammonia?
Marc Hölling
Hamburg University of Applied Sciences, Hamburg, Germany
marc.hoelling@haw-hamburg.de

Abstract
Due to the phase out of nuclear and coal-fired power plants in Germany, there will be a strong need for base load power
generation in the future. If renewable energies are used, the integration of energy storages will be mandatory. The most
common approach for long term storage of energy is Power-to-Gas, i.e. the production of hydrogen by electrolysis. Un-
fortunately, it shows disadvantages in energy density and storage pressures. Therefore, Power-to-Gas is compared to
batteries, methanation and a thermal energy storage. Additionally, Power-to-Ammonia is evaluated, in which the energy
is stored in the chemical form of ammonia. For the comparison, a 3 MW wind turbine is combined with the different
storage types and the overall efficiencies are simulated. The base load power outputs are determined in the range of
320…450 kW and the required storage sizes are about 100 MWh to guarantee full year base load. Thus, an energy storage
with a low power output and a high storage size will be most efficient. This task is well fulfilled by Power-to-Ammonia,
since it shows a high cost efficiency.

Keywords – Energy Storage, Power-to-Ammonia, Energy Costs, Hydrogen

1 Introduction
Under current regulations, the energy produced by a wind
turbine (WT) is sent to the electric grid and the grid opera-
tor has to take care of balancing the net. For future energy
systems without fossil fuels, wind turbines (or solar power)
should be able to generate base load on their own by means
of smart energy storages. The most common approach is
Power-to-Gas, i.e. the production of hydrogen by electrol-
ysis [1]. Other options are batteries, methanation [2], ther-
mal energy storage [3, 4] and Power-to-Ammonia. For a
fair comparison, a wind turbine is combined with different
storage types and main parameters like efficiency, base
load power, storage size, investment costs and electricity
generation costs are evaluated.
Figure 1 Duration curve of a 3 MW wind turbine (solid
line) and the corresponding average power (dotted line)
2 Wind turbine
For this study, a wind turbine with a maximum power out-
put of ܲ୫ୟ୶ = 3 MW is considered, which is reached at 3 Energy storage systems
wind speeds above ‫ = ݒ‬15 m/s. The wind conditions are 3.1 Efficiencies of energy storages
described with the Weibull equation, using ݇ = 2.5 and
‫ = ܣ‬8.6 m/s [5]. Real energy storages show conversion losses and thus have
an efficiency of ߟ < 100%. The efficiency is defined as
݇ ‫ ݒ‬௞ିଵ the ratio of work ܹୗ୲ to and from the storage system:
݄(‫= )ݒ‬ ήቀ ቁ ή exp(െ(‫ݒ‬/‫)ܣ‬௞ )
‫ܣ ܣ‬ ܹୗ୲,୭୳୲ ܲୗ୲,୭୳୲ ή ο‫ݐ‬୭୳୲
ߟ= =
The corresponding duration curve of this wind turbine is ܹୗ୲,୧୬ ܲୗ୲,୧୬ ή ο‫ݐ‬୧୬
given in Figure 1. It results in 1,770 full load hours, which
is in good agreement with the mean value of 1,760 h/a for Table 1 shows an overview of the selected storage types
Germany in 2017 [6]. The production of 5,310 MWh is including their efficiencies ߟ. Additionally, the efficiencies
equal to a mean power generation of ܲത୧ୢୣୟ୪ = 606 kW. for charging and discharging (i.e. ߟ୧୬ and ߟ୭୳୲ ) are included
With an ideal storage with an efficiency of ߟ = 100%, for dimensioning of the storages. A detailed description is
these 606 kW would be the base load power ܲതୠୟୱୣ . given in the subsequent sections, in which also the efficien-
cies are presented.

ISBN 978-3-8007-5152-5 240 © VDE VERLAG GMBH · Berlin · Offenbach


NEIS Conference 2019 Â 19. – 20.09.2019 in Hamburg

Table 1 Overview of different storage concepts and their


efficiencies with ߟ = ߟ୧୬ ή ߟ୭୳୲ (own simulations and liter-
ature)
ߟ ߟ୧୬ ߟ୭୳୲
Battery 96% 98% 98%
Hydrogen (20 bar) 49% 70% 70%
Hydrogen (200 bar) 47% 67% 70% Figure 2 Simplified process layout for a methanation pro-
Methane (50 bar) 26% 58% 45% cess including intermediate cooling [2]
Th. Energy Storage 23% 96% 24%
Power-to-Ammonia 31% 49% 63% The methane might be sent to the gas grid and thus an ex-
isting infrastructure can be used to avoid investments.
3.1.1 Battery storage From the gas grid it can be taken as a fuel for heating ap-
The highest efficiency can be achieved with batteries. plications or converted to electricity in a gas turbine
Though they are typically used for short term storage and (ߟ୭୳୲ = 35%) or in a combined-cycle plant (ߟ୭୳୲ = 55%).
are quiet expensive, they will be mentioned here as a refer- For reasons of comparison, a stand-alone storage option in
ence. The efficiency is taken as 98% for charging and dis- combination with a combustion engine (CE) with ߟେ୉ =
charging, which results in a total efficiency of 96%. 45% is chosen. It corresponds to the average efficiency of
a gas turbine and a combined-cycle plant. The total effi-
3.1.2 Power-to-Gas (hydrogen storage) ciency of the methanation storage is ߟ = 26%.
The production of hydrogen (Power-to-Gas) by means of The open question for the methanation process is the
electrolysers (EL) is a common approach for long term source of the carbon dioxide. It might come from cement
storage. The typical outlet pressure is about 20 bar and the plants, from biomass or from the combustion process itself
efficiency is taken as ߟ୉୐ = 70% 1. In times of low produc-
0F
(circular economy for the carbon dioxide). In this study, it
tion of the wind turbine, the hydrogen can be fed to a fuel is taken as free of charge and as a pure component at at-
cell (FC) to produce electrical power. The efficiency of the mospheric pressure. The energy/costs for separation and
fuel cell is taken as ߟ୊େ = 70%. The overall efficiency is cleaning are neglected. In reality, the efforts for the pro-
the product of the two steps, ߟ = ߟ୉୐ ή ߟ୊େ = 49%. If the duction of the carbon dioxide might have a big impact on
hydrogen is compressed to e.g. ‫݌‬ୗ୲ = 200 bar, the energy the efficiency and the economics.
demand for the compression (2 stages, 20 bar to 200 bar)
has to be taken into account. The total efficiency then de- 3.1.4 Thermal energy storage
creases to ߟ = 47%. Energy can also be stored as internal energy of a heat stor-
age. For the production of electricity, a high temperature
3.1.3 Methanation level is advantageous, which can be realized with a porous
The hydrogen from an electrolyser can also be used for a rock bed [3]. A schematic layout of the process is given in
chemical reaction. In the exothermic Sabatier process, hy- Figure 3.
drogen reacts with carbon dioxide to produce methane. The excess energy of the wind turbine is used in an electric
heater to produce hot air. This air flows through a porous
COଶ + 4 Hଶ ՞ CHସ + 2 Hଶ O rock bed and heats it up to more than 650°C. The efficiency
of the charging of the storage is set to ߟ୧୬ = 96%, as given
A methanation process, as proposed in [2], was simulated in [3].
with DWSIM [7], a freeware process simulation, see Fig- In times of low wind production, air is send through the
ure 2. rock bed and gets heated up to 650 °C to produce steam
The pressure level of the process is about 20 bar and the afterwards. It leaves the steam generator with a tempera-
temperature in the reactor varies between 220 °C and ture of 270 °C and is send back to the heat storage.
550 °C. For temperature control, the carbon dioxide is fed A simple Clausius-Rankine cycle (CRC) is simu-
to the process a different locations of the reactor. An effi- lated using DWSIM with a steam pressure of 85 bar, a su-
ciency of ߟ୫ୣ୲୦ = 58% was obtained for the production of per-heating temperature of 550 °C and a condensation
methane, including electrolysis and compression to 50 bar pressure of 400 mbar (air condenser). The isentropic effi-
at the outlet of the process. Though the methanation pro- ciency of the turbine is taken as 75%. With these assump-
cess shows a decrease of the efficiency, it shows an im- tions, the efficiency of the discharging is ߟ୭୳୲ = 24%.
provement in terms of energy density and storage options.

1
The efficiency includes auxiliary equipment like water
treatment and it is based on the low calorific value of hy-
drogen (241.8 kJ/mol).

ISBN 978-3-8007-5152-5 241 © VDE VERLAG GMBH · Berlin · Offenbach


NEIS Conference 2019 Â 19. – 20.09.2019 in Hamburg

kmol bar ଴.ହ 198,000 J/mol


݇Ԣ = 2.5714 ή 10ଵ଺ ଷ
ή exp ൬െ ൰
m h ܴήܶ

The outlet gas flow is cooled down to 30 °C by a combina-


tion of heat recovery and water cooling. Ammonia can be
separated as the liquid phase, since its vapor pressure is
11.6 bar at 30 °C. The gas phase (hydrogen / nitrogen) is
recycled to the inlet of the reactor. For storing of the am-
monia, the pressure is reduced to 20 bar. The DWSIM sim-
ulation model results in an efficiency of ߟ୧୬ = 49% for
Figure 3 Thermal energy storage with a porous rock bed charging of the storage.
and a simple Clausius-Rankine-Cycle (CRC) For the discharging of the storage (Figure 4 (b)), the am-
monia is taken from the tank and fed through the same plug
3.1.5 Power-to-Ammonia flow reactor with a pressure of only 2 bar and a temperature
of 500 °C, since these conditions favor a high conversion
Another well-known process in chemical engineering is the
of ammonia to hydrogen and nitrogen (endothermic reac-
Haber-Bosch process for the production of ammonia:
tion). The gas mixture is sent to a fuel cell to produce elec-
1 3 tricity. Since the hydrogen-rich gas contains about 25 vol-
Nଶ + Hଶ ՞ NHଷ % nitrogen, a gas stream has to be removed from the fuel
2 2 cell. It is burned with preheated air and used for heating of
the reactor and the ammonia. Assuming an efficiency of
Again, hydrogen is produced by an electrolyser and nitro-
ߟ୊େ = 70% for the fuel cell, the DWSIM simulation results
gen can be taken directly from the atmosphere. A huge ad-
is an efficiency of ߟ୭୳୲ = 63%. Thus the total efficiency of
vantage is the high energy density of the produced ammo-
nia. Figure 4 shows a simplified scheme of the Power-to- the Power-to-Ammonia process is ߟ = 31%.
Ammonia process.
3.2 Effect of the Storage Efficiencies on the
Base Load Power
For a storage system with losses, the real base load power
will be below the ideal value of 606 kW (ܲതୠୟୱୣ < ܲത୧ୢୣୟ୪ ).
In times of high production, the excess power is sent to the
storage, which is ܲୗ୲,୧୬ = ܲ െ ܲതୠୟୱୣ . In times of low pro-
duction, the base power is provided by the wind turbine and
the storage, ܲതୠୟୱୣ = ܲ + ܲୗ୲,୭୳୲ . For long periods, typically
one year (‫ = ݐ‬1 a), the energy balance must level out. Tak-
ing the total efficiency ߟ into account, this can be described
by:
Figure 4 Power-to-Ammonia process: (a) production of
ammonia (charging mode), (b) production of hydrogen for න max(0,
ᇣᇧᇧᇧᇧᇧᇧᇤᇧᇧᇧᇧᇧᇧᇥ max(0, ܲതୠୟୱୣ െ ܲ) d‫ݐ‬
ܲ െ ܲതୠୟୱୣ ) ή ߟ d‫ = ݐ‬න ᇣᇧᇧᇧᇧᇧᇤᇧᇧᇧᇧᇧᇥ
the fuel cell (discharging mode). ଵa ୡ୦ୟ୰୥୧୬୥ ୭୤ ୱ୲୭୰ୟ୥ୣ 1a ୢ୧ୱୡ୦ୟ୰୥୧୬୥ ୭୤ ୱ୲୭୰ୟ୥ୣ

In the charging mode (Figure 4 (a)), Hydrogen is partly Since the power generation ܲ is known from the duration
burned with compressed air. This is a simple way of re- curve, compare Figure 1, the base load ܲതୠୟୱୣ can be deter-
moving the oxygen from the feed gas and adjusting the mined iteratively with the efficiency as the remaining pa-
mole fraction of the feedgas to ‫ݔ‬ୌమ = 3 ‫୒ݔ‬మ . For a more rameter. Figure 5 gives the results for the 3 MW wind tur-
efficient process, pure nitrogen could be used from e.g. an bine with the different energy storages.
air separation unit or pressure swing adsorption. The pro-
duced water is removed by condensation and the pressure Even with low efficiencies down to ߟ ൎ 23%, the base
is increased to 155 bar. The feed gas is mixed with recycled load power is ܲതୠୟୱୣ > 320 kW, which is more than 50% of
hydrogen/nitrogen and heated up to 420 °C. The reaction ܲത୧ୢୣୟ୪ . This is because a big share of the power is taken di-
takes place in an isothermal plug flow reactor (PFR) and is rectly from the wind turbine and is not passed through the
modelled with the reaction kinetics according to Temkin storage.
und Pyzhev [8].

dܿ୒మ ‫ ୒݌‬ή ‫݌‬ୌଵ.ହమ ‫୒݌‬ୌ


‫=ݎ‬െ =݇ή మ െ ݇Ԣ ή ଵ.ହయ
d‫ݐ‬ ‫୒݌‬ୌయ ‫݌‬ୌమ
kmol 87,000 J/mol
݇ = 1.78954 ή 10ସ ଷ ଵ.ହ
ή exp ൬െ ൰
m h bar ܴήܶ

ISBN 978-3-8007-5152-5 242 © VDE VERLAG GMBH · Berlin · Offenbach


NEIS Conference 2019 Â 19. – 20.09.2019 in Hamburg

3.4 Cost Estimation for the Energy Storage


For the cost estimation specific project costs as well as spe-
cific equipment costs are used. For example, a wind turbine
can be installed for 1,500 €/MW including foundations,
erection, project management and grid connections. For the
cost estimation of the different storage concepts (hydrogen,
methane, ammonia, TES), specific costs for the equipment
are used, i.e. for reactors, compressors, turbines and the
storage (pressure vessels, tanks, rock beds). The total pro-
ject costs are obtained by the equipment costs and an addi-
tion of 100% for foundation, housing, automation, piping
etc. Table 3 gives an overview of the used specific costs.

Table 3 Specific costs for total project as well as for equip-


ment
Figure 5 Base load power for a 3 MW wind turbine de-
pending on the overall efficiency of the storage system Specific Total Costs
Wind turbine 1,500 €/kW
Battery 1,000 €/kWh
3.3 Dimensioning of the Energy Storage Specific Equipment Costs
The worst case for the energy storage is taken as one week Reactor 300,000 €/m³
with no wind production, i.e. the base load ܲതୠୟୱୣ is taken Electrolyser 800 €/kW
for ȟ‫ݐ‬୫ୟ୶ = 168 h from the storage. The maximum size Fuel Cell 1,000 €/kW
ܹ୫ୟ୶ of the storage can be determined for each concept: Combustion Engine 500 €/kW
Steam Turbine 600 €/kW
ܲതୠୟୱୣ ή ο‫ݐ‬୫ୟ୶
ܹ୫ୟ୶ = Rock Bed 25 €/kWh
ߟ୭୳୲ Pressure Vessel / Tank 10,000 €/tSteel
An alternative approach is the stochastic simulation of the For a 3 MW wind turbine in combination with a 103 MWh
fluctuating production and its effect on the charging level. battery storage the total costs will be in the range of
The results are nearly the same. Table 2 gives an overview 107.5 Mio. €. Here, the cost driver is the battery and the
of the size, the specific energy density and the required vol- power generation is only about 4% of the project costs.
ume (and weight) of the tanks/vessels. For the weight eval- For the other processes, the cost estimation is more com-
uation of the pressure vessels, steel is chosen as the con- plex since it consists of more parts. For the Power-to-Am-
struction material. The dimensions are limited to a maxi- monia process the cost estimation is given examplarily. For
mum diameter of 2 m for delivery by truck (e.g. hydrogen the other storage concepts the costs are summarized in Ta-
vessels for 20 bar). The other limit is a wall thickness of ble 4.
5 cm due to manufacturing of the vessels (e.g. hydrogen The Power-to-Ammonia process consists of an electrolyser
vessel for 200 bar). (ܲ୫ୟ୶ െ ܲത = 2.64 MW), a plug flow reactor (2 m³), a tank
(32 m³ / 15 t) and a fuel cell (ܲത = 360 kW). Thus the costs
Table 2 Required size in MWh, specific energy density in
for the main equipment are:
kWh/m³ and volume in m³ as well as weight in tsteel (where
suitable) for different storage concepts
x 2,640 kW ή 800 €/kW = 2.11 Mio. €
MWh kWh/m³ m³ / (tsteel) x 2 m³ ή 300,000 €/m³ = 0.6 Mio. €
Battery 103 60 1717 x 15 t ή 10,000 €/t = 0.15 Mio. €
Hydrogen (20 bar) 107 53 2019 (620) x 360 kW ή 1,000 €/kW = 0.36 Mio. €
Hydrogen (200 bar) 105 478 220 (735)
Methane (50 bar) 127 477 266 (207) The sum for the equipment costs is 3.22 Mio. €. With an
Th. Energy Storage 226 125 1808 addition of 100%, the total costs for the storage add up to
Power-to-Ammonia 97 3065 32 (15) 6.44 Mio. €. The costs of the wind turbine of 4.5 Mio. € in-
crease the total project costs to 10.94 Mio. €.
It can be seen, that most storage concepts are in the range
of 100 MWh. However, there is a huge difference in the 3.5 Electricity Generation Costs
volume/weight, which is crucial for the economic evalua- The minimum electricity generation costs can be obtained
tion. by the total project costs, the depreciation period and the
annual electricity generation. With a rather long deprecia-
tion period of 15 years, the generation costs (without inter-
est rates etc.) are given in Table 4 and in Figure 6.

ISBN 978-3-8007-5152-5 243 © VDE VERLAG GMBH · Berlin · Offenbach


NEIS Conference 2019 Â 19. – 20.09.2019 in Hamburg

Table 4 Total project costs, annual electricity production price increase of nearly 21 Ct/kWh or 70%. For a house-
in MWh/a and generation costs in €/MWh for different hold with 3000 kWh/a it would be equal to additional costs
storage concepts at a depreciation period of 15 years of 630 €/a. Due to the high significance and acceptance of
climate protection in the public, the costs seem appropriate.
Costs in MWh/a €/MWh
Mio.€
Battery 107.5 5,250 1,365 4 Conclusions
Hydrogen (20 bar) 21.9 3,916 373
Hydrogen (200 bar) 34.2 3,844 419 The integration of storage systems into renewable energies
Methane (50 bar) 15.0 2,987 336 could be a crucial milestone for the transformation of the
Th. Energy Storage 16.2 2,828 382 energy world. If politics decide to make energy storages
mandatory to any new installation, the supply of base load
Power-to-Ammonia 10.9 3,154 231
power could be guaranteed, even in a fossil-free future.
Without any incentives, the short term investments in stor-
The usage of batteries shows the highest generation costs,
ages will be low, even if their need is without any doubt.
as was expected. Batteries should be used as a short term
With the shutdown of fossil power plants the back-up ca-
energy storage and are not very cost efficient for the supply
pacities will decrease and might lead to a shortage in secure
of base load power. The other concepts show much lower
energy supply.
costs in the range of 230…420 €/MWh. Though the cost
In this study, the fluctuating power of a 3 MW wind turbine
estimation show some uncertainties, a cost advantage can
is transformed into base load by means of different energy
be identified for Power-to-Ammonia and for methanation.
storages. The evaluation depends on the efficiency of the
Both chemicals (ammonia and methane) show a high en-
storage, but also on the size/volume. A typical storage size
ergy density, which is very favorable for the investment
was found to be about 100 MWh and the base load power
costs of the tanks.
is the range of 320…450 kW. According to presented as-
sessment, Power-to-Ammonia is looking quite promising.
It shows the lowest generation costs of approx. 230 €/MWh.
Additionally, nitrogen is, other than carbon dioxide, avail-
able in the atmosphere in large concentrations.
The estimated cost increase due to the integration of stor-
ages might be acceptable for private households (or trade).
For any energy intensive industry, this might be a critical
threat and some kind of compensation should be imple-
mented.

5 References
[1] NOW-GmbH: IndWEDe – Industrialisation of water
electrolysis in Germany: chances and challanges of
sustainable hydrogen for mobility, electricity and he-
Figure 6 Electricity generation costs for the different stor- ating (in German: Studie IndWEDe – Industrialisie-
age concepts at a depreciation rate of 15 years. rung der Wasserelektrolyse in Deutschland: Chancen
und Herausforderungen für nachhaltigen Wasserstoff
For the assessment of this minimum generation costs it has für Verkehr, Strom und Wärme). https://www.now-
to be distinguished between industry and private house- gmbh.de/de, visited at April 25th, 2019.
holds. For energy intensive industries, like steel or alu- [2] Schaaf, T., Grünig, J., Schuster, M.R., Rothenfluh, T.,
mina, the relevant costs are determined at the European Orth, A.: Methanation of CO2 - storage of renewable
Energy Exchange (EEX). In 2018, the average costs for energy in a gas distribution system. Energy, Sustaina-
base load electricity were 44.40 €/MWh. Thus, the supply bility and Society. 4 (1), 2014, pp 1-14.
with base load from renewable energies would lead to a [3] Zavattoni, S.A., Barbato, M.C., Pedretti, A., Zanga-
tremendous cost increase. Due to today’s global economy, neh, G., Steinfeld A.: High temperature rock-bed TES
these costs cannot be charged to the customers and it would system suitable for industrial-scale CSP plant – CFD
lead a strong market distortion. analysis under charge/discharge cyclic conditions.
For a private household, the actual costs for electricity are Energy Procedia. 46, 2014, pp. 124-133.
about 30 Ct/kWh (300 €/MWh), which is in the same range [4] Jacob, R., Saman, W., Bruno, F.: Capital cost ex-
as some generation costs in Table 4. To allow for a little penditure of high temperature latent and sensible ther-
margin for the operator or for a shorter depreciation period, mal energy storage systems. AIP Conference Procee-
generation costs of 35 Ct/kWh (350 €/MWh) are assumed dings 1850, 2017.
for base load power. Additionally grid costs of approx. [5] Jarass, L., Obermair, G. M., Voigt, W.: Wind Energy
8 Ct/kWh and 19% of Value Added Tax have to be consid- (in German: Windenergie). Springer, Berlin, 2008.
ered. This results in electricity costs of 51 Ct/kWh. It is a

ISBN 978-3-8007-5152-5 244 © VDE VERLAG GMBH · Berlin · Offenbach


NEIS Conference 2019 Â 19. – 20.09.2019 in Hamburg

[6] Fraunhofer IEE: Wind Energy Report 2017 (Win-


denergiereport Deutschland 2017). http://windmoni-
tor.iee.fraunhofer.de/, visited on February 18th, 2019.
[7] DWSIM, dwsim.inforside.com.br, visited at April
25th, 2019.
[8] Temkin, I., Pyzhev, V.: Kinetic of Ammonia Synthe-
sis on Promoted Iron Catalyst, Acta Phy. Chem.
URSS, 12, 1940, pp. 327-356.

ISBN 978-3-8007-5152-5 245 © VDE VERLAG GMBH · Berlin · Offenbach

You might also like