Air Cathode

Electrolyte
Fuel Anode

Master of Science Thesis

H2 Techno-economic Assessment
H2O

O2– 2e– of Solid Oxide Fuel Cells and

Fuel-assisted Electrolysis CH4

Cells in Future Energy

Systems
H2(pore)
H2O(pore) 1 H
H 2e–
2
4
TPB 4
3 metal
2
electrolyte
O2–

Sander van Nielen

Techno-economic Assessment of Solid Oxide Fuel Cells and
Fuel-assisted Electrolysis Cells in Future Energy Systems

Sander van Nielen

s1159615, 4174755

MSc Industrial Ecology

Leiden University, TU Delft

Supervisors:
Dr. K. Hemmes – TU Delft, Faculty of Technology, Policy & Management (TBM)
Dr. L.M. Ramírez-Elizondo – TU Delft, Faculty of Electrical Engineering, Mathematics
and Computer Science (EWI)

November 29, 2016

 Abstract

There is an increasing awareness in society of the problems and risks associated with climate change.
In response, a transition from fossil to renewable energy sources is taking place. Meanwhile, there is an
ongoing trend of increasing electricity demand for various applications. There is a growing number of
electric appliances, as well as the electrification of heating, cooling, and transportation services. These
changes provide a challenge when it comes to matching electricity production and consumption. The
intermittent nature of wind and solar energy, combined with the large loads of heat pumps and BEVs,
reduces the flexibility and robustness of the system in the future.
Solid oxide cells (SOCs) can contribute to solving the balancing problem of the electricity grid. SOCs
perform electrochemical conversions and are developed for decentral energy conversion applications. A
solid oxide fuel cell (SOFC) uses a fuel to produce electricity, hydrogen, and heat. Solid oxide fuel-
assisted electrolysis cells (SOFECs) consume electricity and a fuel while producing hydrogen and heat.
Gaseous fuels like natural gas, hydrogen, and biogas can be used. SOFCs have been under investigation
for decades, and are mainly applied as backup power source. SOFEC technology, on the other hand, is
rather unknown. Only a dozen of articles has been published, dedicated to theoretical models and small-
scale lab tests. These SOCs are seen as promising technologies, because of the high energy efficiency,
flexible operation, and ability to produce hydrogen.
The aim of this research is to assess whether it is feasible to introduce SOFCs and SOFECs in the
(future) Dutch energy system. This research takes a broad perspective, by assessing the technological
components in relation to the energy system as a whole. In addition, multiple aspects that determine
the feasibility are assessed. These aspects are the technical possibilities and limitations of SOCs, their
cost-effectiveness under different assumptions, the environmental implications, and societal drivers and
barriers for the introduction in niche markets.
Existing literature shows that the commercial competitiveness of SOCs is currently limited because
of high investment costs and a limited lifetime. The manufacturing costs can be decreased by learn-
ing through research and development, and higher production volumes. To prolong the service life,
degradation of the electrodes at microscopic level should be minimised. Recent research indicates that
regular switching between fuel cell and (fuel assisted) electrolysis mode can also prevent degradation.
It is therefore recommended to further investigate how the service life is prolonged for a SOFC that is
operated regularly as SOFEC.
A thermodynamic 1-D cell model was used to describe the inputs and outputs of a SOFC and a
SOFEC. Based on this model, non-linear sets of equations with two degrees of freedom were derived for
both technologies. These equations approximate the energy conversion by SOCs for different current
densities and fuel utilisations. The level of complexity of these equations was suitable for the further
analysis of the operation of the technologies.
The derived equations were used to implement SOFC and SOFEC technologies in oemof, an energy
system optimisation framework. The framework was then used to model the energy system of the
Netherlands, including all major energy carriers, i.e. fossil fuels, electricity, heat, biomass, and hydrogen.
In the model, energy production, conversion, and consumption technologies are included, as well as
corresponding hour-by-hour patterns. An optimisation algorithm is used to find the optimal combination
of technologies and their optimal operation that results in a system that most effectively fulfils demand.
The objective is to minimise the total system costs, excluding subsidies or taxes. Costs related to GHG
emissions, carbon tax or emission certificates, are included. By doing so, it is the first optimisation
model of its kind for the Netherlands. Not all aspects of the real energy system are represented in the
model. Most notably, energy transport and demand response are not included.
The results clearly indicate that SOFCs can become an important part of the Dutch energy system
if the costs decrease sufficiently and a longer lifetime of the stacks is achieved. The cost-effectiveness
stems from the efficient co-production of three energy carriers and the low resource use. High GHG
emission taxes are favourable for the diffusion of SOFCs. Their installation results in an emission
reduction of around 5.2% (or 8 MtCO2 -eq./a) compared to a system without fuel cells. No evidence
was found for the influence of SOFCs on the competitiveness of renewable electricity sources. Under
some conditions, SOFCs can in fact squeeze solar and wind out of the electricity market.
If the costs of operating SOFCs remain high, then the installed capacity is limited to an extent
for which high capacity utilisations can be achieved. The production of hydrogen is maximised to
make expensive SOFCs profitable, indicating that this energy carrier is most important for a cost-
effective introduction. If costs can be reduced, SOFCs prove to be a suitable technology for electricity
production during periods of low renewable electricity production, while also responding to fluctuating
H2 demand. It becomes viable to operate the fuel cells with more flexibility when the technology is less
expensive. The influence of the demand on SOFC operation is also considerable. In a scenario with
a diverse demand for both heat, hydrogen, and electricity, a wide range of settings is used. In other
scenarios, different operating strategies were more optimal. Only two or three operating points were
used frequently, while other settings were applied seldom. For the requirements for SOFC technology,
this implies that the preferred technical specifications depend on how the energy demand will develop.
The role that SOFECs can play in the Dutch energy system is much more limited than that of
SOFCs. The conditions for cost-effectiveness are also stricter. A high level of renewable electricity
production and/or high GHG emission taxes are required, on top of large cost reductions for SOC
technology. Furthermore, its applicability is limited by the amount of biogas available. If SOFCs and
SOFECs with the same costs are both available, a SOFC is often preferred, since it allows for much
larger system cost reductions. The optimal operating strategy of SOFECs depends on their costs: if the
technology becomes less expensive, it is cost-effective to operate at lower current densities with higher
efficiency. Furthermore, SOFECs show a limited variation in the production rate over time. Unlike
regular electrolysers, SOFECs are hardly operated to prevent electricity excesses. By operating almost
continuously, a large fraction of biogas is utilised. A SOFEC profits from high GHG taxes, as it is
(almost) carbon neutral.
The societal aspects of the development and potential of SOC technology were assessed using the
multi-level perspective. This framework conceptualises a niche innovation as being nested structure
within a socio-technical regime and a socio-technical landscape. Within the dominant regime of energy
production and consumption, there is no fully developed solution to address the challenge of increased
fluctuations of electricity demand and intermittent production.
Considerable research efforts in the field of SOC technology are undertaken. SOFC have received
most attention, resulting in several pilot projects and growing production volumes. Few activities are
taking place around SOFECs. A possible strategy is to focus on SOFC applications first, suspending
efforts related to SOFECs. When lower electricity prices and lower SOC costs are achieved, it is
more viable to introduce a SOFEC. Actors involved in the development of SOFCs are active, with
several demonstration projects and market introductions in the area of micro-CHP applications. These
dynamics could result in an uptake in the existing regime structure without exploiting and exploring the
capability of SOFCs to co-produce hydrogen. Yet this study has shown the advantage of H2 production
for cost-effective SOFC applications.
It is deemed crucial that a user or market for the energy products of the SOFC or SOFEC is found.
Therefore, the energy carrier markets for inputs and outputs form the starting point for identifying
market niches, in which the SOC technology can be introduced first. Examples of identified niche
applications include a farmer who uses biogas from manure digestion and supplies H2 to trucks, or a
neighbourhood community that invests in a common installation. A SOC provides heat to households
in a neighbourhood, while H2 is used for fuel cell vehicles or to provide electricity at peak demand.
To convince potential users, these perceived barriers can be addressed or the advantages of SOCs can
be emphasised. The high efficiency, low emissions, and importance for local grid balancing of SOFCs
and SOFECs are arguments in favour of these technologies. In addition, the commitment of other
stakeholders in pilot projects or through Green Deal agreements can stimulate the adoption.

2
Highlights
• Co-production of H2 is important for the cost-effectiveness of SOFCs.
• SOC costs should decrease significantly to become an attractive addition to the (future) Dutch
energy system.
• SOFCs directly reduce GHG emissions and increase resource efficiency.
• SOFCs compete with renewable electricity sources and do not facilitate their introduction.
• SOFECs are economically less attractive than SOFCs, and their potential is limited by the avail-
ability of biogas.
• Ideal operating strategies for SOFC and SOFEC installations in the context of an energy system
were assessed, and shown to depend on the demand composition and technology costs.
• Markets or consumers for heat and hydrogen are a key factor for first applications of SOC tech-
nologies.

3
Contents

1 Introduction 13
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.2 Research questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3 Relevance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.4 Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2 Technical Aspects of Solid Oxide Cells 17

2.1 Chemical principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.1.1 Fuel cell types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.1.2 Solid oxide cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.1.3 Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.1.4 Fuel flexibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2 Materials used in SOFCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2.1 Anode – fuel electrode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.2.2 Cathode – oxygen electrode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.2.3 Electrolyte . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3 Material stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.1 Degradation research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.2 High temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.3 Sulfur tolerance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.4 Electrolysis mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.3.5 Material purity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.3.6 Materials for reversible cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4 System design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.4.1 SOFC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.4.2 External reforming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.4.3 Anode gas recycling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.4.4 Cathode gas recycling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.4.5 Stacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.4.6 SOFEC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.5 Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.6 System operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.6.1 Steam-to-carbon ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.6.2 Fuel utilisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.6.3 Air utilisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.7 Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.7.1 Superwind . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.7.2 Fuel-assisted electrolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.7.3 Reversible hydrogen production . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.7.4 Reversible fuel production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.7.5 Hybrid stack . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.7.6 Micro-CHP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

1
 2.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3 SOC Input-Output Relations 32

3.1 Cell performance modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.2 Operating conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3 Nernst losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.4 Mass and energy balances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.4.1 SOFC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.4.2 SOFEC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4.3 Scaling up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.5 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.6 Nonlinearity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

4 Energy System Modelling 41

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1.1 Energy systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1.2 Optimisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1.3 Path dependence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.2 Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.2.1 Simulation vs. Optimisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2.2 Problem definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2.3 Mapping of energy systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2.4 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2.5 Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2.6 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2.7 Modelling software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2.8 Models of the Netherlands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.3 Modelling decisions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.4 Model implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.4.1 Software sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.4.2 Energy system mapping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.4.3 User input . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.4.4 Implementation of SOC components . . . . . . . . . . . . . . . . . . . . . . . . 53
4.5 Levelised costs of energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

5 Assumptions & Scenarios 54

5.1 Goal & Scope definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.1.1 Goal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.1.2 Timesteps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.1.3 Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.1.4 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.2 Energy conversions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.3 Current energy demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
5.3.1 Electricity consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.3.2 Heat demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.3.3 Hydrogen consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.3.4 Natural gas consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.4 Demand Scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.4.1 Possible developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.4.2 Efficiency factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.4.3 Scenarios quantified . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.5 (Environmental) constraints to energy production . . . . . . . . . . . . . . . . . . . . . 63

2
 5.6 Supply patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.6.1 Fossil fuels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.6.2 Renewable sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.7 Demand patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.7.1 Electricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.7.2 Low-temperature heat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.7.3 Land transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.8 Final remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

6 Technology Options 71
6.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.2 Fossil fuels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.2.1 Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6.2.2 Gas powerplant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.2.3 Coal powerplant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.3 Renewable electricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.3.1 PV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.3.2 Wind turbines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.4 Energy storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.4.1 Lithium battery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.4.2 Compressed air energy storage . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.4.3 Flywheel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.4.4 Hot water tank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.4.5 Phase change materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.4.6 Hydrogen tank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.5 Biomass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.5.1 Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6.5.2 Biogas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.5.3 Green gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.5.4 Biodiesel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6.5.5 Biomass powerplant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.6 Fuel cells and hydrogen production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.6.1 SOEC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.6.2 SOFC and SOFEC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.6.3 PEMFC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.6.4 SMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.6.5 Methanation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.7 Heat-related technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.7.1 Solar thermal panels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.7.2 Boilers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.7.3 Heat distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.8 Greenhouse gas emissions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.8.1 Supply chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.8.2 Combustion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

7 Optimisation Results 79
7.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
7.2 Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
7.3 Nuclear energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
7.4 SOCs in the current energy system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
7.4.1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
7.4.2 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
7.4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

3
 7.5 Effect of the costs of SOCs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
7.5.1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
7.5.2 Installed capacities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
7.5.3 Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
7.5.4 SOFC operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
7.6 Comparison of concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7.6.1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7.6.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7.6.3 SOFEC operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
7.7 Effect of demand composition and GHG taxes . . . . . . . . . . . . . . . . . . . . . . . 94
7.7.1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
7.7.2 Effect of GHG taxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
7.7.3 SOFC operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
7.7.4 Sustainability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
7.8 Transition optimisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
7.8.1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
7.8.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
7.9 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103

8 Societal Aspects 104

8.1 Analytical framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
8.1.1 Socio-technical systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
8.1.2 Learning curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
8.1.3 Economies of scale . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8.1.4 Actor perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8.1.5 Multi-level perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8.2 Actor identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
8.3 Regime description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
8.4 SOC niche innovation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
8.4.1 Current status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
8.4.2 Potential system changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
8.5 Barriers & Drivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
8.5.1 Technology & infrastructure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
8.5.2 Markets & user preferences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
8.5.3 Policy & politics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
8.5.4 Cultural meaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
8.5.5 Industry structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
8.5.6 Knowledge base . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
8.6 Energy carrier markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
8.7 Niche strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
8.7.1 SOFEC niches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
8.7.2 SOFC niches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
8.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

9 Discussion 119
9.1 Sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
9.1.1 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
9.1.2 Cheaper batteries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
9.1.3 Cheaper PV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
9.2 Break-even analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
9.3 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
9.4 Cost reduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

4
10 Conclusion 124
10.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
10.2 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
10.2.1 Energy consumers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
10.2.2 Policy-makers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
10.2.3 System operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
10.2.4 SOC manufacturers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
10.3 Future research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
10.3.1 Model refinement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
10.3.2 Innovation systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

References 130

A Data 141
A.1 Conversion factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
A.2 Technology data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
A.2.1 Buses (Energy carriers) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
A.2.2 Energy (re)sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
A.2.3 Storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
A.2.4 Transformers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
A.3 Existing powerplants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

B Companies 145
B.1 SOFC manufacturers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
B.2 List of European SOFC research projects . . . . . . . . . . . . . . . . . . . . . . . . . . 145

C Source Code 148

C.1 Objective functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
C.2 SOC classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
C.3 SOC constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152

5
List of Figures

2.1 Components and operating principles of SOCs. . . . . . . . . . . . . . . . . . . . . . . 19

2.2 Schematic depiction of a SOFC anode material. . . . . . . . . . . . . . . . . . . . . . . 20
2.3 A possible SOFC system design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.4 A possible SOFEC system design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.5 Decrease in the local Nernst potential as a function of fuel utilisation. . . . . . . . . . . 26
2.6 Decrease in cell potential due to polarisation losses, as a function of current density. . . 26
2.7 Ternary diagram, showing the carbon deposition region for different fuel gas compositions. 27
2.8 Development directions of SOC technology. . . . . . . . . . . . . . . . . . . . . . . . . 29

3.1 Reversible cell voltage of a SOFC and a SOFEC as a function of Uf . . . . . . . . . . . . 35

3.2 Outputs of a SOFC as derived by 1-D numeric modelling. . . . . . . . . . . . . . . . . . 36
3.3 Input and outputs of a SOFEC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4 Partial load efficiencies of NG-fuelled combustion engine and gas turbine technologies. . 37

4.1 Energy conversion processes that connect electricity, hydrogen and natural gas. . . . . . 42
4.2 Representation of energy system components in a model. . . . . . . . . . . . . . . . . . 44
4.3 Structure of the energy system optimisation model, including inputs and outputs. . . . . 49
4.4 The Entity class, its subclasses, and their attributes as available in oemof. . . . . . . . 51

5.1 Energy flows through an energy system, including some examples of carriers. . . . . . . 55
5.2 Energy demand of the Netherlands in 2014. . . . . . . . . . . . . . . . . . . . . . . . . 56
5.3 Heat demand by the residential and industrial sector in the Netherlands in 2006. . . . . 57
5.4 Energy demand per scenario by each sector. . . . . . . . . . . . . . . . . . . . . . . . . 62
5.5 Capacity curve for wind turbines. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
5.6 The wind speed patterns for three consecutive days, and the resulting capacity factors. . 66
5.7 Average electricity production by PV for each month in 2015. . . . . . . . . . . . . . . . 66
5.8 Biogas production of a waste-water treatment plant in Denver. . . . . . . . . . . . . . . 67
5.9 The average gas demand per month in the Netherlands. . . . . . . . . . . . . . . . . . 67
5.10 Median electricity demand at residential BEV charging units in San Francisco. . . . . . . 68
5.11 Average consumption patterns of H2 an gasoline at fuel stations. . . . . . . . . . . . . . 69

6.1 Comparison of investment and O&M costs for SOFCs from different sources. . . . . . . 76

7.1 Electricity production by Plant 1 and 2 during two weeks, for three different demand
curves. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
7.2 System costs relative to the optimum value for a system with two powerplants. . . . . . 80
7.3 Optimised electricity production and consumption patterns in a simple energy system
during an average summer and winter week. . . . . . . . . . . . . . . . . . . . . . . . . 81
7.4 Optimal electricity production pattern when nuclear power generation is not constrained. 82
7.5 Energy sources for electricity production in the Netherlands under the current conditions. 83
7.6 Energy sources for heat production in the Netherlands under the current conditions. . . . 84
7.7 Levelised costs of electricity from SOFCs. . . . . . . . . . . . . . . . . . . . . . . . . . 86
7.8 Total annual production and installed capacities of transformers under different SOC
cost assumptions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

6
7.9 Hydrogen demand during an average summer and winter week, with SOFC properties
from scenario III. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
7.10 Change in system costs for scenario II–VI compared to scenario I. . . . . . . . . . . . . . 88
7.11 Primary energy use for scenario I–VI. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
7.12 Optimised hydrogen and electricity production, and fuel utilisation of SOFCs. . . . . . . 89
7.13 Heat map plots of optimal operation of SOFCs under different cost assumptions. . . . . 90
7.14 Annual production of H2 by different SOC technologies and different cost characteristics
of these technologies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
7.15 Change in total system costs for scenario I–VI compared to a system without SOCs, for
systems with different SOC types. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
7.16 Annual GHG emissions for scenario I–VI and systems with different SOC types. . . . . . 92
7.17 The frequency of operation at a given i for different SOFEC cost scenarios. . . . . . . . 93
7.18 Hydrogen production and consumption in a system with SOFECs. . . . . . . . . . . . . 93
7.19 Annual production and installed capacity for electricity, hydrogen, and heat production. . 96
7.20 Capacity utilisations of three powerplants and three hydrogen production technologies,
for different GHG emission taxes and two demand scenarios. . . . . . . . . . . . . . . . 96
7.21 Heat map plots of optimal operation of SOFCs under different demand scenarios and
GHG emission taxes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
7.22 Difference in GHG emissions between an optimal system with and without SOCs. . . . . 99
7.23 Share of renewable energy in the total primary energy consumption. . . . . . . . . . . . 99
7.24 Primary energy sources for the four demand scenarios and different GHG taxes. . . . . . 100
7.25 Total annual excess of biogas and electricity in scenario I–VI for different GHG taxes. . . 100
7.26 Annual production of electricity, hydrogen and heat, and installed capacities of generation
technologies during a transition path. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

8.1 Physical energy infrastructure and connected actors. . . . . . . . . . . . . . . . . . . . . 107

8.2 Number of articles published each year on SOFC and SOEC technology according to the
Scopus ® database. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
8.3 Annual fuel cell capacity shipped worldwide, in the period 2010–2015. . . . . . . . . . . 109

9.1 Sensitivity analysis of the assumed costs for Li batteries and PV. . . . . . . . . . . . . . 120
9.2 Reference systems for energy production by a SOFC or SOFEC or conventional alternatives.121
9.3 Break-even investment costs of SOFC when compared with a conventional supply chain. 122
9.4 Break-even investment costs of SOFEC when compared with a conventional supply chain.122
9.5 Possible cost reduction curve of the investment costs of SOFCs. . . . . . . . . . . . . . 123

7
List of Tables

3.1 Operating conditions assumed for SOFC and SOFEC operation. . . . . . . . . . . . . . 33

3.2 Inital amounts of gases at the fuel and air electrode of a SOFC, and at the fuel and
steam electrode of a SOFEC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

4.1 Open source renewable or hybrid energy system optimisation models. . . . . . . . . . . . 47

4.2 Tools for modelling the Dutch energy system at a national level. . . . . . . . . . . . . . 48
4.3 User input required for each step of the modelling process. . . . . . . . . . . . . . . . . 52

5.1 Average composition of gas from the Groningen gas field. . . . . . . . . . . . . . . . . . 58

5.2 Efficiency factors for the delivery of three useful energy forms. . . . . . . . . . . . . . . 60
5.3 New industrial boiler efficiencies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
5.4 Comparison of fuel economies of six vehicle types. . . . . . . . . . . . . . . . . . . . . . 61
5.5 The definition of four energy demand scenarios in terms of the demand for each energy
carrier. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.6 Three important GHGs and their conversion factors. . . . . . . . . . . . . . . . . . . . . 63
5.7 Annual availability of biomass for energy production in the Netherlands. . . . . . . . . . 63
5.8 Potential annual biogas production according to different sources. a . . . . . . . . . . . 64
5.9 Boundary conditions applied to the scale of implementation of various renewable energy
technologies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

6.1 Past and future fossil fuel prices in the Netherlands. . . . . . . . . . . . . . . . . . . . . 71

6.2 EU solid biomass supply costs per resource category. . . . . . . . . . . . . . . . . . . . 74
6.3 Costs and emissions for the production of biogas for local use and green gas. . . . . . . 74
6.4 Cost parameters and levelised cost of electricity for SOFCs. . . . . . . . . . . . . . . . . 75
6.5 Economic properties of industrial boilers. . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.6 Emissions related to fuel production/supply and conversion. . . . . . . . . . . . . . . . . 77
6.7 Emissions associated with the production and/or construction of energy conversion tech-
nologies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
6.8 GHG emissions released during the combustion of the most common fossil fuels in dif-
ferent applications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78

7.1 Properties of Plant 1 and 2, and their LCOE. . . . . . . . . . . . . . . . . . . . . . . . 80

7.2 Installed generation capacity of powerplants, in 2015 in the Netherlands. . . . . . . . . . 83
7.3 Energy sources for electricity and heat production in the Netherlands under the current
conditions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7.4 Capacity utilisation in three scenarios, based on the presently available powerplants. . . . 85
7.5 Assumed properties and values of parameters for each year in the transition. . . . . . . . 101

8.1 SOFC system production costs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

8.2 SOFC system production costs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
8.3 List of Dutch organisations that participated in EU-funded research projects on SOFCs. . 113
8.4 Potential niche markets for SOFCs and SOFECs and factors that influence chance of
successful implementation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

8
9.1 Definition and optimisation outcomes of scenarios featuring Li batteries with different
costs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
9.2 Definition and optimisation outcomes of scenarios featuring commercial PV with different
costs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

A.1 Construction year and installed capacities of powerplants that are currently in use in the
Netherlands. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

B.1 List of EU-funded research projects registered in Cordis. . . . . . . . . . . . . . . . . . . 145

9
Abbreviations
KNMI Royal Netherlands Meteorological In-
AC Alternating current
stitute
AGR Anode gas recycling
LCOE Levelised cost of energy
BAU Business as usual
LHV Lower heating value
BEV Battery electric vehicle
LP Linear programming
BoP Balance of plant
LSM Lanthanum strontium manganite
CAES Compressed air energy storage
LT Low-tempearture
CCGT Combined cycle gas turbine
MCFC Molten carbonate fuel cell
CF Capacity factor
MILP Mixed-integer linear programming
CGR Cathode gas recycling
MSW Municipal solid waste
CHHP Combined heat, hydrogen, and power
n-D n-dimensional
CHP Combined heat and power
NG Natural gas
CoP Coefficient of performance
OCV Open cell voltage
CRF Capital recovery factor
O&M Operation and management
CSP Concentrated solar power
PCM Phase change material
DC Direct current
PEMFC Proton exchange membrane fuel cell
DSO Distribution system operator
PtG Power-to-gas
ECN Energy Research Centre of the
PV Photovoltaic cell
Netherlands
R-SOC Reversible solid oxide cell
Eq. Equation
R&D Research and development
ETS European emission trading system
SCR Steam-to-carbon ratio
EU European Union
SOC Solid oxide cell
FC Fuel cell
SOEC Solid oxide electrolysis cell
FCEV Fuel cell vehicle
SOFC Solid oxide fuel cell
FCH-JU Fuel Cell and Hydrogen Joint Under-
SOFEC Solid oxide fuel-assisted electrolysis
taking
cell
GHG Greenhouse gas
TNO Dutch Organisation for Applied Sci-
HHV Higher heating value
entific Research
IC Installed capacity
TSO Transmission system operator
ICE Internal combustion engine
VAT Value-added taxes
IR-FC Internal-reforming fuel cell
WACC Weighted average cost of capital
IT Intermediate temperature
WWT Waste-water treatment plant
JRC Joint Research Centre
YSZ Yttrium-stabilised zirconia

Nomenclature

Symbol Meaning Unit

α Slope of Vrev as a function of Uf V
ASR Area specific resistance Ωcm2
A Active cell area cm2
Cfix Fixed O&M costs €/kW/a
Cfuel Fuel costs €/MWh
CGHG GHG emission costs €/MWh
Cinv Investment costs €/kW
Cvar Variable O&M costs €/MWh
CU Capacity utilisation
D Energy demand PJ/a
∆G Gibbs free energy change kJ/mol
∆H Enthalpy change kJ/mol
Ḣ Fuel flow rate J/s

10
Symbol Meaning Unit
∆c H Heat of combustion; HHV kJ/mol
∆S Entropy change kJ/mol/K
η Efficiency
Einv Initial GHG emission tCO2 -eq./kW
Evar Combustion GHG emission tCO2 -eq./MWh
F Faraday’s constant 96.485 × 103 C/mol
i Current density A/cm2
I Current A
J Energy flux (flow rate per unit of W/cm2
cell area)
KR Equilibrium constant of reaction R
L Lifetime a
νS,R Stoichiometric coefficient of mol/mol
species S in reaction R
n Stoichiometric coefficient of elec- mol/mol
trons in a redox reaction
ṅ Molar flow rate mol/s
P Electric power output We
Electric energy MWhe
p Pressure bar
pS Partial pressure of species S bar
Q̇ Heat flow rate J/s
R Gas constant 8.314 × 10−3 kJ/mol/K
r Discount rate
T Temperature K
T Time period length h
t Timestep
taxGHG GHG emission tax €/tCO2 -eq.
u Local fuel utlisation mol/mol
Ua Air utilisation mol/mol
Us Steam utilisation mol/mol
Uf Fuel utilisation mol/mol
v Wind speed m/s
V0 OCV under feed concentrations V
V◦ OCV under standard concentra- V
tions
∗
Vrev Local Nernst potential V
Vcell Cell potential V
Vrev Nernst potential V
ξR Extent of reaction R
y Fraction

Subscripts and superscripts

add Additional inv Investment, initial
c Energy carrier max Maximum
cur Curtailment out Output
e Component rated Rated
exc Excess sho Shortage
final Final energy tot Total
fix Fixed useful Useful energy
in Input var Variable

11
Acknowlegements

This thesis is the result of a a research process of almost a year. I take full responsibility for the contents
of the report, as I am the only author. However, I could have never completed the research successfully
without the help of a number of people. Here, I take the opportunity to thank these people.
First and foremost, I would like to thank my supervisors, Kas Hemmes and Laura Ramírez-Elizondo.
By providing me with guidance and feedback, they helped me to make the best of this thesis. Notwith-
standing their full agendas, they took the time to help me out – with or without appointment. My
thanks also go to the students from Sustaintable Energy Technology and Industrial Ecology with whom
I have shared many days together, studying and exchanging ideas. I appreciate the discussions and
the vital distraction now and then. Furthermore, I am grateful for Branco and Maarten, who were so
kind to review a part of my thesis. Their feedback has been useful for improving my writing. I could
also share several considerations with them. Finally, I would like to thank my parents. Not only were
they interested in discussing about this rather academic subject, they also provided mental support.
Sometimes, talking about what is on your mind helps a lot in sorting it out, whether it is related to
research or not!

12
Chapter 1

Introduction

1.1 Introduction
Scholars, citizens, societal organisations and politicians are increasingly aware of the problems and risks
associated with climate change. This awareness entails a growing sense of urgency to reduce greenhouse
gas (GHG) emissions. An important measure to reach the targets of GHG emission reduction is the
switch from fossil to renewable energy sources. Electric and thermal energy can be derived from wind
and solar energy, and other renewable sources that are locally available. Meanwhile, a continuous trend
towards a higher electricity demand is taking place. Fossil fuels are replaced by electricity as energy
carrier by the electrification of heating, cooling, and transportation services. The growing number of
electric appliances used by consumers also contributes to the trend.
These changes in production and consumption of electricity provide a challenge for the match
between the two. On the one hand, the intermittent nature of renewable electricity sources causes periods
of overproduction and shortage (Mulder, 2014). Fluctuations provide a challenge for conventional
powerplants that have a limited output adjustment speed. On the other hand, heat pumps and battery
electric vehicles (BEVs) provide considerable electric loads with the potential to cause major demand
peaks. The challenge consists of filling up shortages in production, and handling excesses to ensure
a balanced network at every moment in time. Solutions that are proposed include fossil fuel based
backup plants, energy storage, and demand-response mechanisms. The effectiveness of these solutions
depends on the configuration of the system. For example, backup plants work best with moderate or
low renewable energy penetration (Schenk et al., 2007). Energy storage options such as batteries are
attractive in systems with a high amount of renewable electricity production (Belderbos et al., 2015).
The efficiency and costs restrain application in the current system.
An alternative solution to the balancing problem is provided by solid oxide cells (SOCs). SOCs are
devices that perform electrochemical conversions. This developing technology is scalable from 1 kW to
several megawatts (Energistyrelsen & Energinet.dk, 2015), and is therefore suitable for decentral appli-
cations. Solid oxide fuel cells (SOFCs) use a fuel to produce electricity, hydrogen, and heat (Hemmes,
2004). Solid oxide fuel-assisted electrolysis cell (SOFECs) consume electricity and a fuel while produc-
ing hydrogen and heat (Cinti et al., 2016). By operating in response to the electricity market, excess
and shortage can be addressed by SOFCs and SOFECs. This would stimulate the growth of renewable
electricity production (Vernay et al., 2008). In this respect, the concept of a reversible SOC (R-SOC)
that can switch between fuel cell and fuel-assisted electrolysis mode is attractive. SOFECs and R-SOCs
have been investigated to a limited extent, so there is little operational knowledge of these technologies.
Given the ability of SOCs to co-produce hydrogen, they could play a role in a transition towards
a system with hydrogen as energy carrier. Besides, SOCs can use various gaseous fuels, among which
natural gas (NG), hydrogen, and biogas (Van Herle et al., 2004; Ud Din & Zainal, 2016). Therefore,
an opportunity to utilise a renewable fuel is provided.
In the past decades, large and growing efforts have been invested in the research and development
(R&D) of SOFCs (Mahato et al., 2015). Applications are found in specific situations, such as remote
areas or as subsidised installations for heat and power production in households (Hart et al., 2015). The

13
scientific knowledge of SOFECs, on the other hand, is much more limited. A number of articles has
been published, in which theoretical thermodynamic models are presented (Butler, 2010; Wijers, 2011;
Luo et al., 2014; Patcharavorachot et al., 2016). Some experiments on a lab-scale have been published
as well (Tao et al., 2006; Tao, 2007; Cinti et al., 2016).
Many questions related to the new SOC technology are yet unanswered, for example: At what
costs are they competitive? To what extent do SOCs stabilise the electricity grid? What role could or
should SOCs play on the energy market? How could they be operated in real-world market conditions?
How do the answers change in a future energy system? Conventional assessment methods fall short in
analysing SOCs, because of the complex input-output relations. The merit order is used to determine
how often an electricity source is operated, but is hard to apply for multi-generation technologies.
Hydrogen production alternatives can be compared by calculating the levelised costs of energy, but this
is not straightforward if variable electricity prices apply for an electrolyser. Furthermore, technologies
that have multiple inputs and/or outputs should somehow respond to the dynamics of all energy carrier
markets to which they are connected. To conclude, more sophisticated and system-oriented tools are
required to study SOCs.
A method that takes an integral approach, and is therefore able to answer most questions above,
is energy system optimisation. Computational optimisation has been applied to many problems, and is
also suitable for the assessment of energy systems. An energy system optimisation model can determine
the optimal combination of technologies and/or their optimal operation, based on a given objective.
An energy system is defined as the collection of all socio-technical structures related to energy within
a given geographic area. Energy systems include the supply chains of energy carriers. A typical supply
chain consists of energy generation, conversion, transport, distribution, storage, and consumption units.
Because of the wide variety of technical, financial, regulatory, institutional, and socio-cultural structures,
actors, interactions, and dynamics involved in energy systems, they are complex systems.
Given the complexity of energy systems, it is not straightforward to evaluate the performance of a
single component. The efficiency of the system as a whole does not only depend on the characteristics
of single components, but also on how well they complement each other. When this complexity is
acknowledged, the system should be assessed as a whole. To do so, computer models can be useful
tools. By analysing a model of an energy system, valuable insights can be gained. Energy models can
generate and quantify alternative energy system structures as part of the decision making processes of
policy makers and business strategists.

1.2 Research questions

SOFCs and SOFECs are promising emerging technologies. They are claimed to be efficient in producing
energy, flexible in their operation, and beneficial for the integration of renewable energy sources into
the energy system. The latter two benefits can only be assessed properly from a systems perspective.
Therefore, the aim of this research is to assess the sensibility of including SOFCs and SOFECs in a
future national energy system. The feasibility depends on the technical feasibility, the effects on the
energy system, its costs and its eco-efficiency, the required technical modifications of the energy system,
and the societal acceptance of the technology.
To summarize, the research question is formulated as

“Is it feasible to introduce SOFCs and SOFECs in the (future) Dutch energy system?”

The feasibility can be further specified by acknowledging that different aspects determine the feasibility.
These aspects are taken into account in the subquestions:
Technical aspects:

• Is it technically possible to operate SOCs in both normal and fuel-assisted electrolysis mode?
• What is the optimal operating strategy for SOCs?

Economic aspects:

14
 • Are SOCs a cost-effective addition to the current Dutch energy system?
• At what cost level are SOCs a cost-effective component of the future Dutch energy system?
• What effect do GHG emission taxes have on the economic feasibility?

Environmental aspects:

• How do SOCs affect the GHG emissions of the Dutch energy system?
• What changes in primary energy consumption result from the introduction of SOCs?

Societal aspects:

• How can the transition towards a system with SOCs take place in terms of technology replacement?
• What are potential niche markets, in which SOC technologies can be applied first?
• How can the development and diffusion of SOC technology in the Netherlands be promoted?

1.3 Relevance
SOFC and SOFEC are relevant for the field of Industrial Ecology, because of their contribution to
resource savings, emission reductions, and ultimately a more sustainable energy system. Besides, SOCs
can be operated using several fuel options and produce two or three energy carriers. These characteristics
make that the technology can be relevant for industrial symbioses or industrial ecosystems, in which
materials, energy, knowledge, and other resources are exchanged.
This research contributes to existing knowledge by deriving a model that describes the tri-generation
of heat, hydrogen, and power by a SOFC as well as for the fuel-assisted electrolysis of water by a SOFEC.
SOFC technology has been well-studied in the past, yet less development efforts are invested in its ability
to produce H2 . Research on fuel-assisted electrolysis is limited to a number of studies on theoretical
models and small-scale lab tests. Besides, the energy system of the Netherlands was modelled, while
including all energy carriers and production and consumption patterns with a resolution of one hour. This
integrated approach is relatively new, but is important to address the increasing conversions between
different energy carriers. Energy system optimisation is applied to assess the feasibility in terms of
economic and environmental aspects of SOFC and SOFEC technologies in the (future) energy system.
This will provide new insights in the applications of both. SOFECs and SOFCs producing H2 have
not been investigated in this respect yet. Given the uncertainties related to future developments, it is
relevant to determine the influences of SOC costs, GHG emission taxes, and demand composition on
the feasibility of these emerging technologies. No studies have been conducted yet to investigate the
ideal operating strategy of SOFCs and SOFECs. The optimisations of the present study allows to assess
the operation in the context of realistic energy carrier markets.

1.4 Outline
The body of this thesis is structured as follows. Chapter 2 focusses on the technical aspects of SOFCs
and SOFECs. By reviewing literature on the materials, operation and design of solid oxide cells, the
characteristics and possible applications are assessed. Based on the insights gained from Chapter 2,
Chapter 3 derives input-output relations that allow for the implementation of SOCs in optimisation
frameworks. Next, Chapter 4 discusses the theory and practice of energy system modelling. The theo-
retical framework is discussed, as well as existing energy system models. Also, the modelling approach
of this research is specified. Assumptions and scenarios related to the production and consumption of
energy in the model are elaborated in Chapter 5, whereas data for specific technologies are presented
in Chapter 6. In the next chapter, Chapter 7, optimisation outcomes are presented and discussed.
These results are the key to answering the research questions related to economic and environmental
aspects. Chapter 8 starts with the specification of an analytical framework based on literature study.
The concepts of this framework are then applied to determine drivers, barriers, and niche markets for
SOFC and SOFEC technology. The sensitivity and validity of the results are tested in Chapter 9. The

15
final chapter provides conclusions, implications, and recommendations that follow from this thesis. In
addition, possible directions for further research are recommended.

16
Chapter 2

Technical Aspects of Solid Oxide Cells

This chapter covers existing knowledge of the chemistry, materials, design, and operation of SOFCs.
This information is needed to understand what the technical possibilities and limitations are for the
application of SOFCs and SOFECs. Besides, the parameters that describe system operation and losses
are important for the mathematical modelling of the cells.

2.1 Chemical principles

2.1.1 Fuel cell types
A fuel cell (FC) is an electrochemical device that converts chemical energy of a fuel to electric power and
heat. The theoretical efficiency of a FC is higher than that of internal combustion engines. Fuel cells
consist of an anode, where the fuel is supplied, a cathode, where the oxidant is supplied (usually air),
and an electrolyte that transports ions between the electrodes. Various FC designs exist, differing in the
choice of fuel and electrolyte (Giddey et al., 2012). FC types are usually named after their electrolyte.
Hydrogen is the most common fuel.
Solid oxide fuel cells are named after their solid, ceramic, oxygen ion conducting electrolyte. SOFCs
operate at high temperatures, that can range from 650 to 1000 ℃ (Mathiesen, 2008). These high
temperatures entail several advantages. First, less or no precious metal catalysts such as platinum and
palladium are required to drive the reactions at the electrodes. Second, higher efficiencies can be reached
compared to low-temperature FCs. This is because electric losses, resulting from the internal resistance
of a FC, decrease with temperature (Standaert, 1998). Third, any ‘waste’ heat that is recovered from the
outflows has a high temperature and high exergetic value, allowing for reuse. Fourth, high-temperature
FCs are capable of sustaining an endothermic internal reforming reaction with heat produced by the
oxidation of fuel. If methane and steam are supplied to the anode, H2 and CO2 are formed. In that
case, the FC is referred to as internal reforming fuel cell (IR-FC). Some of the generated hydrogen can
be extracted from the cell as a product. By doing so, a very high efficiency for the conversion from fuel
to energy products is realised (see § 3.4.1).
The disadvantage of high-temperature FCs is that all materials, including those of the support
structure, have to be resistant towards high temperatures and temperature changes. Although the
temperature is high in SOFCs, it is not as high as in conventional combustion engines. Therefore
SOFCs cause, like other FCs, virtually no emissions of NOx and particulate matter (Giddey et al., 2012).
An advantage of FCs is their scalability. Even small stacks have a high efficiency, making them
attractive for distributed combined heat and power (CHP) generation, or household-scale CHP (micro-
CHP). Moreover, it is possible to build modular powerplants which can easily scale up when required
by adding extra stacks. Furthermore, the energy production rate of FCs can be adapted quickly to the
demand. An IR-FC installation can be ramped up in 15 minutes. A precondition for SOFCs is that
temperature gradients are minimized to prevent degradation (Fardadi et al., 2016). Therefore, long
start-up times are needed, unless the operating temperature can be maintained when the SOFC is not
operated.

17
2.1.2 Solid oxide cells
The operation of SOCs is based on the principles of electrochemistry, which describes the conversion
of electric energy to chemical energy and vice versa. A cell is composed of two electrodes, the anode
and cathode, and an electrolyte sandwiched in between. An electric circuit connects the cathode and
anode. Unlike batteries, the chemical reactants are not stored within the electrodes, but are continuously
supplied to the cell during operation. The operational principles of SOCs are depicted in Figure 2.1.
At the oxygen electrode of a SOFC, oxygen from air takes up electrons from the external circuit and
enters the electrolyte. The oxygen ions are transported between the electrodes via the electrolyte. At
the fuel electrode, gaseous fuel is supplied and oxidised by the oxygen ions. To close the electrochemical
cycle, an external load is connected to both electrodes.
A SOC can also be operated in reverse when steam is supplied to the fuel electrode. By applying
an external voltage, hydrogen and oxygen are produced from the electrolysis of water. In this mode of
operation, the cell is referred to as a solid oxide electrolysis cell (SOEC). Operating temperatures similar
to that of SOFCs are applied. In a SOEC, voltages higher than the decomposition potential of water
are required.
The high (over)potentials of a SOEC can be prevented in a third mode of operation, namely as solid
oxide fuel-assisted electrolysis cell. Fuel (e.g. methane or syngas) is fed to the fuel electrode, where
oxidation takes place. This reduces the operating voltage by about 1 V (Martinez-Frias et al., 2003),
since the thermodynamically unfavourable formation of oxygen is prevented. Therefore, the electric
power required to produce hydrogen from steam at the steam electrode decreases significantly.

2.1.3 Reactions
In a SOFC, chemical and electrochemical reactions are taking place. The operation of a SOFC can be
described by three reactions: R1–R3. The overall reaction is identical to regular combustion, with the
difference that fuel and air are never in direct contact as indicated by Figure 2.1a. The electrochemical
fuel oxidation process has a higher exergy efficiency. The electrochemical redox reaction R1 of H2
and O2 to form water, with a half reaction at each electrode (Figure 2.1a), is responsible for the power
generation. The hydrogen required for this reaction can be produced during two reactions: fuel reforming
(R2) and water-gas shift (R3). Other reactions occur in parallel, but these have lower rates (Wendel
et al., 2016) and are only relevant in detailed studies of kinetic effects at the fuel electrode (Hemmes,
2004). Reactions R2 and R3 together are referred to as steam methane reforming (SMR). The overall
result of the three reactions is that methane is fully oxidised to CO2 and H2 O (R4), provided that all
reactions are complete. For any practical application, R1 will never reach full conversion, as discussed
in § 2.6.2.
2 H2 + O2 −−→ 2 H2 O (R1)
CH4 + H2 O −−→ CO + 3 H2 (R2)
CO + H2 O −−→ CO2 + H2 (R3)
Overall: CH4 + 2 O2 −−→ CO2 + 2 H2 O (R4)
In a SOEC, the redox reactions proceed in the opposite direction compared to a SOFC. An applied
voltage provides the driving force for the two redox half-reactions at the SOEC electrodes (Figure 2.1b).
Oxygen is formed instead of consumed at the air electrode from O 2 – ions. The electrons released travel
through the external circuit to the steam electrode, where H2 is generated. In a SOFEC, H2 O is reduced
to H2 at the steam electrode, identical to the SOEC cathode reaction. Instead of O2 formation at the
anode, a fuel is oxidised (e.g. H2 or CO) in a reaction with O 2 – . The fuel electrode reaction results
in an overall cell reaction that is close to equilibrium, implying that an applied voltage close to zero is
sufficient to drive the production of H2 .

18
 Fuel electrode − Anode Depleted
Fuel/steam electrode − Cathode Hydrogen
Fuel Steam
2– – fuel Steam
H2 + O → H2O + 2e H2O + 2e– → H2 + O2–
– –
O2– Power O2– Voltage
produced applied
+
+
– 2– 2– –
½O2 + 2e → O Depleted O → ½O2 + 2e Enriched
Air Air
air air
Air electrode − Cathode Air electrode − Anode

(a) SOFC, in which a fuel, e.g. H2 , reacts electro- (b) SOEC, where electricity is used for electrolysis,
chemically with O2 to yield power. splitting H2 O into H2 and O2 .
Fuel electrode − Anode Depleted
Fuel
fuel
H2 + O2– → H2O + 2e–
+
O2– Voltage
applied –

H2O + 2e– → H2 + O2– Hydrogen

Steam
Steam
Steam electrode − Cathode

(c) SOFEC, where electrolysis is performed, but with

a fuel supplied at the anode as reducing agent.

Figure 2.1: Components and operating principles of solid oxide cells. Redox reactions at the
electrodes are shown for all operating modes. The three main components of SOCs are illustrated
in the figures. A cell has two porous electrodes, the anode and the cathode, which are separated
by a dense electrolyte.

2.1.4 Fuel flexibility

Many hydrogen-rich fuels can be used as feed for the fuel electrode of a SOFC or SOFEC. Apart from
methane, other hydrocarbons or hydrogen can also be used as a fuel. All of these compounds could
be derived from biomass in the future. In the Netherlands, a limited amount of organic wastes is
available, originating from agriculture, forestry, waste-water treatment, and households (Warmerdam
et al., 2011; Ros & Prins, 2014). On a European level, the amount of available biowaste is estimated
enough to run SOFCs with a total installed capacity of 12 GWe (Van Herle et al., 2004). Biomass
can be converted to biogas by for example pyrolysis, thermo-chemical gasification, or fermentation.
Combined with appropriate purification equipment and in some cases steam methane reforming, biogas
is compatible with SOFCs (Ud Din & Zainal, 2016).
The advantage of using biogas is that fossil fuels are not depleted, and only biogenic CO2 is emitted.
Disadvantages of biogas include the higher costs of the plant due to contaminant removal equipment,
and the limited supply of domestic biomass. Further issues and concerns arise if dedicated energy crops
are used. The consequences include deforestation, the interference with food supplies, and fertiliser use
(Ridjan et al., 2013).
SOFCs have been demonstrated to work well with CO-rich gases (Omosun et al., 2004; Penchini
et al., 2013). These gases can be derived from municipal solid waste (MSW) incineration plants and
steel production. Blast furnace gas, coke oven gas, or mixtures of these are presently used to generate
electricity in conventional turbines (CBS, 2015b). Syngas (a mixture of CO and H2 ) is another potential
fuel derived from industrial processes. Higher hydrocarbons can be used by SOFCs (Murashkina et al.,
2008).

2.2 Materials used in SOFCs

The technical limitations and focus points for current research are closely connected to the materials
that constitute the core parts of SOCs. Therefore, an overview of commonly used compounds for the

19
anode, cathode, and electrolyte (see also Figure 2.1) is given below. Most materials discussed here can
also be used in a SOFEC, as explained in § 2.3.6 in more detail.

2.2.1 Anode – fuel electrode

The fuel electrode is made of a porous material, connected to the electrolyte, as shown in Figure 2.2a. In
almost all current SOFCs (Mahato et al., 2015), nickel–yttria-stabilized zirconia (Ni–YSZ) is the anode
material. Ni–YSZ is a cermet, i.e. a material composed of both a ceramic and a metallic material. The
metal ensures a high ionic conduction, while the ceramic material makes sure the cermet remains stable
under high temperatures and provides ionic conductivity.
An important measure for the reactivity of the anode is the length of the triple-phase boundary
(TPB). This interface between the Ni, zirconia, and gas phase (see Figure 2.2b) is the catalytically
active part of the electrode. By optimising the TPB, high current densities and low losses can be
achieved. Therefore, a great deal of research efforts have been invested in techniques to synthesize
anode materials with longer TPBs. An evolution has occured from Ni patterns applied on the electrode,
through screen printed or coated Ni anodes, to impregnated skeletons and cermets (Mahato et al., 2015;
Irvine et al., 2016). The newest techniques give more precise control over the microstructure of the
materials, leading to better performance.

H2(pore)
electrolyte anode cermet
H2O(pore) 1 H
H2 H 2e–
2
4
H2O TPB 4

O2–
3 metal
2e–
2
electrolyte
O2–
= electrolyte (e.g. YSZ)
= metal (e.g. Ni)
(a) Schematic representation of a porous SOFC (b) Detail of the triple phase boundary (TPB), where the
cermet anode. metal, cermet, and gas phase are in contact.

Figure 2.2: Schematic depiction of a SOFC anode material. Reprinted from Brandon (2013).

2.2.2 Cathode – oxygen electrode

Metal oxides with a perovskite crystal lattice are important cathode materials. Perovskites have a
general formula of ABO3 , in which A and B are metal cations. The most common cathode material for
SOFCs is La1-x Srx MnO3-δ (LSM). The ionic conductivity of perovskites can be enhanced by doping the
material with acceptor cations, resulting in vacant oxygen sites. In the case of LSM, Sr 2+ ions replace
some of the lanthanum. The high operating temperatures increase electronic conductivity (Mahato et
al., 2015).

2.2.3 Electrolyte
The function of the electrolyte is to transport oxygen ions (O 2 – ) from the cathode to the anode, thereby
closing the electrochemical cycle of the fuel cell. A good electrolyte has high O 2 – conductivity, but
does not conduct electrons. YSZ has these properties at high temperatures. Ceria-stabilized zirconia
(CeSZ) is more attractive for intermediate temperatures (Marina et al., 2007).

20
 Several designs and geometries of SOCs have been developed. As for the cell design, there are
electrolyte-supported, anode-supported, and metal-supported cells. The geometry can be either planar
or tubular. The most common planar design geometry is characterised by the ’sandwich’ of anode,
cathode, and electrolyte materials. This design allows to easily stack multiple cells in a fuel cell stack
(Irvine et al., 2016).

2.3 Material stability

2.3.1 Degradation research
Better insights in how SOCs work and how they can be improved can be obtained through research on the
chemical properties and mechanisms of fuel cell materials. As mentioned in § 2.2.1, the electrochemical
interface or TPB is crucial for the performance of the cells. Damage done to the phase boundaries is the
main cause of the short lifetime of most SOCs. Therefore, it is important to understand the mechanisms
that cause passivation, activation, or degradation of the interfaces (Irvine et al., 2016). Degradation
persistently damages the electrode, whereas passivation can be reversed.
Research conducted on SOCs is usually divided into three phases. In the first phase, small tubular
button cells of typically 1-3 cm diameter are used. These cells allow researchers to investigate material
behaviour and combinations and optimize the compositions. Second, design alternatives are compared
using single cells with their intended dimensions (around 100 cm2 of active area). Finally, a stack of
cells is build and tested for contact issues and other problems occurring only at full system level (US
Fuel Cell Council, 2007).

2.3.2 High temperatures

While the high operating temperature provides several advantages (internal reforming, more efficient
electrolysis, redundancy of precious catalysts), it also poses a strict constraint on possible materials
to used. Minimum requirements for all SOC materials are stability under high temperatures and the
capability to withstand thermal stress caused by thermal gradients within the cell. This requirement
does not only apply to the electrodes and electrolyte, but also holds for housing and interconnecting
materials.
An example of a temperature-induced degradation effect is chromium poisoning. This effect occurs
at the air electrode, where chromium present in interconnect materials or in steel housing, blocks
the electrode active sites (Irvine et al., 2016). Gradients induce cracking of electrode materials. To
minimise detrimental effects of thermal gradients as a result of heating and cooling of the cell, steep
load changes should be avoided. In practical applications, the load-following capability of a SOFC is
therefore limited. Transport phenomena add to the inertia of the cell (Napoli et al., 2015). Thermal
management strategies (possibly using an external heat source) can increase the dynamic response of
SOFCs (Fardadi et al., 2016; Komatsu et al., 2014).
The transient behaviour of a SOFC stack in response to a change in current density was examined
by Komatsu et al. (2014). A sudden change results in an overshoot of the cell voltage, followed
by stabilisation within 30-60 minutes. A ramping rate of 3 We /s was reported for a 1 kWe SOFC
manufactured by Ceres Power (Brandon, 2013). This implies that a cell can switch between full and
zero load in 6 minutes. These figures indicate that the load-following operation limitations are more
important at short time intervals. For this reason, low-temperature fuel cells perform better in following
the electricity demand of as single household Napoli et al. (2015).

2.3.3 Sulfur tolerance

The fuel electrode is sensitive to sulfur poisoning. Even small traces of H2 S are adsorbed by nickel at
the TPB, resulting in a smaller active area. At higher H2 S concentrations or large overpotentials in
electrolysis mode, degradation occurs due to the formation of NiS (Irvine et al., 2016). This fact is
important, because untreated biogas and, to a lesser extent, NG, contain sulfur compounds. In biogas

21
generated from sewage sludge and other organic origins, H2 S and organic sulfur compounds are the
most abundant contaminant (McPhail et al., 2011). Sulfur concentrations depend on the origin of the
biomass source. Although naturally occurring sulfur compounds are removed from crude natural gas,
sulfur-containing odorants are added to the gas (De Wild et al., 2002). The odorants give the gas a
characteristic smell, allowing for the detection of potentially dangerous gas leaks. In the Netherlands,
the odorant used is tetrahydrothiophene, in a concentration of around 18 mg/m3 (GTS, 2014).
A higher tolerance towards sulfur poisoning can be achieved by replacing YSZ by scandia stabilised
zirconia (ScSZ) or gadolinia doped ceria (GDC) in the fuel electrode (Mahato et al., 2015). Experi-
mental results indicate that a Ni–GDC electrode remains stable during operation with biosyngas as fuel
(Ouweltjes et al., 2006; Ud Din & Zainal, 2016).

2.3.4 Electrolysis mode

The electrolysis mode seems to pose the largest challenges for electrode material stability. Typically,
SOECs show higher degradation rates than SOFCs (Graves et al., 2015). Some of the mechanisms are
discussed below.
A main source of degradation is the high overpotential applied during electrolysis. One possible
consequence is crack formation in the anode material due to the growth of oxygen bubbles. The
increasing pressure of these bubbles destabilises the electrode, resulting in disintegration or delamination
in the long run (Graves et al., 2015). Meanwhile, ZrO2 reduction can occur at the cathode. This
mechanism eventually leads to passivation of the interface too. Yet, cycling between fuel cell and
electrolysis mode can reverse the damage to a certain extent, since Zr can be oxidised again (Irvine et
al., 2016).
Finally, as a consequence of the high steam pressures typically applied in SOECs, Ni particles can
aggregate and coarsen at the Ni–YSZ electrode. This decreases the length of the TPB.

2.3.5 Material purity

Electrode materials should be able to withstand high temperatures, sulfur traces, and the reactants
flowing through it, as discussed in the previous sections. An additional prerequisite is that the materials
should not react in the presence of each other or under the applied redox atmospheres. Degradation
of the Ni–YSZ electrode seems to be heavily influenced by the type and level of impurities not only in
the fuel gas, but also in the electrode material itself. This could explain differences in performance of
several orders of magnitude as reported for otherwise identical electrodes (Irvine et al., 2016).

2.3.6 Materials for reversible cells

As discussed above, degradation and passivation occur after prolonged operation in a single mode.
Sometimes, this can be prevented by operating a fuel cell in a cyclic way, i.e. switching between SOFC
and SOEC or SOFEC mode. For instance, Graves et al. (2015) show that after reversible cycling between
electrolysis and fuel-cell mode, the microstructure of the electrodes is still intact. The test setup used
cycles of 5 h in each mode. After 4000 h of operation, the power output was almost identical to the
initial value (Graves et al., 2015).
The biggest challenge for the development of a R-SOC is to design electrode materials that are
both stable and efficient under a range of redox environments. The electrolysis mode is still a major
bottleneck (§ 2.3.4, Gómez and Hotza (2016)). New perovskite materials are investigated for their
performance. For example, a La0.3 Ca0.7 Fe0.7 Cr0.3 O3-δ air electrode (Molero-Sánchez et al., 2015) and a
La1-x Srx Cr1-y Mny O3 (LSCM) fuel electrode (Tao, 2007) were developed. Both were shown to be stable
and efficient in both fuel cell and electrolysis mode.
In 2016, a commercial, 50 kW R-SOC was installed by Sunfire (Fuel Cells Bulletin, 2016). This
module can help to find optimal operating parameters and to identify bottlenecks in robustness. Tao
(2007) has tested the use of SOCs in fuel-assisted electrolysis mode with positive results. Cells with
100 cm2 active area each were combined in short stacks of 2 to 10 cells. These stacks were able to

22
produce hydrogen in SOFEC mode at an overpotential of around 1 V lower than regular electrolysis.
Both CH4 , H2 , and CO were tested as fuel, but only at low current densities (<0.5 A/cm2 ).

2.4 System design

2.4.1 SOFC
Although the fuel cell is the core component of a combined heat, hydrogen, and power (CHHP) system,
other components are needed as well to complete the system. A possible layout of an entire system is
provided in Figure 2.3. Fuel gas entering the system is desulfurised before it is mixed with steam in the
pre-reformer. The gas mixture now consists of mostly H2 , H2 O, CO, CO2 , and some unreacted CH4 .
At the SOFC anode, H2 is oxidised, and the SMR reactions proceed further. Hydrogen is separated
from the cathode off-gas as a product. The remainder of the mixture is directed to the afterburner,
where it reacts with partly depleted air from the SOFC anode. Heat released during the combustion
reactions is exchanged with the pre-reformer. The temperature of the afterburner flue gases is decreased
in a series of heat exchangers that heat up reactant flows. To reuse the remaining waste heat, another
heat exchanger connected to an external circuit is included. Depending on the desired temperature of
the external heat demand, it is possible to position this heat exchanger before or after the others. For
maximum water and energy recovery, steam is condensed from the flue gas and re-used in the process.

Flue gas
Water
tank External heat demand

Hydrogen
Natural gas Anode
Pre-reformer
/ Biogas
Desulfuriser
Cathode
Afterburner
~ AC Power
=
Water Demineraliser

Air
Filter

Figure 2.3: A possible SOFC system design. Similar drawings are published in Arduino and
Santarelli (2016); Braun (2002); James et al. (2012); Scataglini et al. (2015).

The flowchart of a micro-CHP SOFC is very similar to the design in Figure 2.3. A difference is
that H2 leaving the cathode is not collected, but combusted in the afterburner together with the other
off-gases (Pfeifer et al., 2013).
Some designs deviate from the one given in Figure 2.3. The pre-reformer can be excluded, or exhaust
gas recycling of either the anode or cathode is included. Besides, the size of the stack is a relevant
design parameter. The considerations for each of these decisions are discussed in § 2.4.2–2.4.5.

2.4.2 External reforming

An important design decision is whether (partial) external reforming is used, or that the SMR reac-
tions (R2 and R3) occur internally, i.e. at the fuel electrode. In a fuel processor, external reformer or
pre-reformer (as shown in Figure 2.3), the preheated fuel mixture is brought into contact with a Ni
catalyst. Nickel, which is also present in the fuel electrode, catalyses SMR. Whether or not to include
a pre-reformer depends on several considerations. Firstly, the choice affects the thermal management
strategy. While the exothermic oxidation reaction (R1) releases heat, the reforming reaction R2 is highly

23
endothermic; ∆r H = 225.6 kJ/mol at 800 ℃(Wendel et al., 2016). On the one hand, this means that
an IR-FC needs less cooling, which is an advantage. On the other hand, this does result in large tem-
perature gradients in the cell, which accelerates degradation of the materials (Braun, 2002; Aguiar et
al., 2002). Secondly, the cell efficiency depends on the way of reforming. External reforming results
in a high hydrogen concentration at the inlet, which makes the fuel cell more efficient (Janardhanan
et al., 2007). However, the overall system efficiency decreases by applying external reforming. These
conclusions are supported by the model of Braun (2002). Braun shows that pre-reforming has only a
very small effect on the power production, because most methane is reformed within a small distance
from the fuel inlet of an IR-FC. Thirdly, the costs of the system are affected by the decision. An external
reformer requires additional investments and maintenance compared to internal reforming (Braun, 2002;
Janardhanan et al., 2007). Moreover, the reduced cooling requirements allow for smaller air pumping
equipment (Staffell & Green, 2013).

2.4.3 Anode gas recycling

Instead of using pure steam as input to the pre-reformer, anode gas recycling (AGR) can be applied.
This means that the anode exhaust gases, that have a high water content, are directly recycled. AGR
reduces heat transfer efforts, because less or no steam preheating is required. Besides, the system fuel
conversion is higher since unreacted CH4 and H2 are also recycled. Systems with AGR make less use of
the afterburner. Because hydrogen output is minimised by AGR, it is not suitable if hydrogen is seen as
a product.

2.4.4 Cathode gas recycling

Similar to AGR, cathode gas recycling (CGR) involves the recycling of hot fuel cell exhaust gases to
the inlet. In the case of the cathode, the recycled stream partly replaces fresh air feed. Less energy is
needed for air pumping and heating. A drawback is the lower oxygen concentration in the cathode feed,
which reduces the fuel cell efficiency.

2.4.5 Stacks
Fuel cells are combined in a stack, which is a series of connected cells. The stack voltage is equal to
the sum of the single cell voltages. The current is the same in each cell. Therefore, the number of cells
in a stack determine the voltage that is produced. The stack size also has implications for the operator
of the system. For maintenance or inspection, a single stack can be turned off, while the other stacks
continue to operate.

2.4.6 SOFEC
A SOFEC system contains the same elements as a SOFC, but there are four differences in the connections
between the components (see Figure 2.4). Air is no longer fed to the fuel cell, but directed directly to the
afterburner. Steam replaces air as the feed of the anode. The fraction of steam that is not converted to
hydrogen, is separated from the rest of the anode off-gas, and recycled to the water feed. The cathode
is still fed by a steam–methane mixture. The reacted mixture is conveyed to the afterburner, without
extracting any hydrogen present.
By comparing Figure 2.4 and 2.3, it can be concluded that in principle it is possible to design a
reversible SOFC/SOFEC system. Extra connections (the blue lines) are required, as well as several
switching mechanisms to open and close valves.

24
 Flue gas
Water
tank External heat demand

Natural gas Anode

Pre-reformer
/ Biogas
Desulfuriser Cathode
Afterburner
~ AC Power
=
Water Demineraliser Hydrogen

Air
Filter

Figure 2.4: A possible SOFEC system design. Blue lines indicate connections that are not present
in a SOFC system.

2.5 Losses
The low efficiencies of fuel combustion in regular engines is caused by the high entropy production
during the reaction. From a thermodynamic point of view, there are two ways to reduce the irreversible
losses of fuel oxidation: operate at higher temperature, or reduce the chemical driving force (Dunbar
and Lior in Braun (2002)). This is why SOFCs have such a high efficiency: the operating temperature
is high, and the oxidation reaction is controlled because oxidant and fuel are not in direct contact.
For conventional electrolysis of water, there are two main sources of losses: a high overpotential
stemming from the large activation energy of the oxygen reduction redox half reaction, and loss of
entropy (Cinti et al., 2016).
An important property of a SOFC is the current-voltage relation. Once this relation is obtained, the
inputs and outputs can be derived. The most general relation is

Vcell = Vrev − i · ASR (2.1)

In Eq. 2.1, ASR is the area-specific resistance, and the reversible voltage Vrev is the Nernst potential of
the reactants, defined as:

∆G pH2 O
!
Vrev = with ∆G = ∆H − T ∆S + RT ln p (2.2)
nF pH2 · pO2

As Eq. 2.2 suggests, Vrev depends on the concentration of species involved in the reactions. Moreover,
since the concentrations change as the fuel gas flows through the electrode, Vrev is a function of the
fuel utilisation. Towards the outlet of the electrode, the fuel gas is depleted of hydrogen, thus lowering
the cell potential compared to the initial OCV (see Figure 2.5). The losses that occur because of this
effect are called Nernst losses. Nernst losses are reversible and cannot be prevented. A first order
approximation of the reversible cell voltage is given by:

Vrev ≈ V0 − αUf (2.3)

V0 is the approximate OCV at the feed concentrations. This relation gives a rather accurate approx-
imation of the reversible SOFC voltage (Hemmes, 2004). Eq. 2.3 also applies to SOFECs, although
different values of V0 and α apply.
Irreversible losses are captured in the internal resistance term (i· ASR) in Eq. 2.1. The irreversible
overpotential is caused by three mechanisms (Park et al., 2012):

25
 Open cell voltage (V)

Fuel utilisation (%)

Figure 2.5: Decrease in the local Nernst potential as a function of fuel utilisation. The fuel is a
mixture of CH4 or H2 mixed with 3% steam. Measurements were performed at 700 ℃. Reprinted
from Kuhn and Kesler (2015).

Activation polarization is the result of kinetic effects. The collection of electrochemical reactions
taking place at both electrodes provide activation barriers that should be overcome. The Butler-
Volmer equation describes the resulting overpotential. To lower the activation losses, it is crucial
to have suitable catalysts and a large TPB. The active area can be increased if the microstructure
of the electrode materials is controlled (Irvine et al., 2016; Mahato et al., 2015).

Ohmic polarization originates from the finite ionic conductivity of the electrolyte. Oxygen ions
that migrate through the electrolyte experience a resistance. These losses can be minimised by
operating at higher temperatures, decreasing the electrolyte thickness, or increasing the number of
vacancies in the electrolyte material. The latter can be achieved by using dopants. The electrical
resistance of the electrodes can also contribute to ohmic losses, but is often small compared to
the electrolyte resistance.

Concentration polarization occurs because concentration gradients exist at the electrodes. Close

Standard potential V°
Non-standard
conditions Reversible cell voltage Vrev
Activation losses

Ohmic losses
Cell potential

Concentration
losses

Current density

Figure 2.6: Decrease in cell potential due to polarisation losses, as a function of current density
(Nagle, 2008).

26
 to the active sites (the triple-phase boundaries, see § 2.2.1), the reactant concentrations are
lower than in the bulk because of mass transport limitations. Especially at large current densi-
ties, diffusion of reactants limits reaction rates and cause concentration polarisation. To reduce
concentration losses, the electrodes, especially the fuel electrode, are made porous.
These three polarisation mechanisms can be further classified as anode and cathode polarisation, since
the losses are electrode-dependent.
Figure 2.6 indicates that each polarisation mechanism dominates the voltage loss in a distinct region
of current densities. As soon as current is drawn from the cell, activation losses occur. Ohmic losses
are responsible for the linear decrease of the voltage in the mid-range of i. Concentration losses induce
additional losses at high current densities (Wijers, 2011).

2.6 System operation

During operation of a SOFC or SOFEC, three parameters should be taken into account: the steam-to-
carbon ratio (SCR), and fuel and air utilisation. These parameters affect the performance of the fuel
cell in different ways, which are discussed below.

2.6.1 Steam-to-carbon ratio

Carbon deposition is an unwanted side reaction that can occur at the fuel electrode. This negatively
affects the cell performance, since solid carbon can block the active sites of the catalyst (Mahato et al.,
2015). It can be prevented by mixing steam with the fuel gas up to the point that carbon deposition
becomes thermodynamically unfavourable. If a perfect equilibrium between the chemical species is
assumed, the carbon deposition potential depends on the ratio of C, H, and O in the mixture. As
shown in the ternary diagram of Figure 2.7, the carbon deposition region also depends on temperature
and pressure. The possibility of carbon deposition decreases for high temperatures and low pressures
(Wendel et al., 2016). At atmospheric pressure and 800 ℃, a SCR of 2 should be sufficient to prevent
deposition (Mahato et al., 2015). Higher SCRs have a negative effect on the overall system efficiency,
because steam dilutes the hydrogen at the anode and because of the larger flows involved (Braun, 2002;
Janardhanan et al., 2007).

0 C
100
850 ℃
725 ℃
600 ℃ 20
80

40 Carbon
deposition 60
region
60
40

80
CH4 20
Fully
SC
oxidised
H100 1
2
R
0
0 20 H2O 40 60 80 100
O
Figure 2.7: Ternary diagram, showing the carbon deposition region for different fuel gas composi-
tions and temperatures. The pressure is 1 bar. The dashed line indicates a mixture of only CH4
and H2 O. SCR: steam-to-carbon ratio. Adapted from Wendel et al. (2016).

27
2.6.2 Fuel utilisation
The driving force for electricity generation in a fuel cell is the electrochemical potential of R1, as given
by Eq 2.2. This relation implies that the SOFC’s cell potential is higher for higher concentrations of H2
at the fuel electrode. If no H2 is present at all, then the cell potential is not defined. Reaching a fuel
utilisation of 100% is both unachievable and undesirable in practice, since the cell potential approaches
0 (see Figure 2.5). The sharp decline in OCV for high Uf is due to reversible losses stemming from
reactant concentration differences (see § 2.5).
The fuel utilisation is defined as the fraction of the fuel inflow that is converted to electrical energy.
In the case that CH4 is the only fuel supplied, Uf can be expressed as:

∆ṅCH4 CH4 − (ṅCH4 + 4 ṅH2 + 4 ṅCO )

1 out 1 out
ṅin out
Uf ≡ = (2.4)
ṅin
CH ṅin
CH
4 4

If Uf is low, a larger amount of CH4 leaves the fuel cell unreacted, reducing the efficiency. To put
it differently, a larger capacity of balance of plant (BoP) components is needed to achieve the same
electricity output of a system with higher Uf . At the same time, a higher amount of H2 is produced
at low Uf . But if Uf is lowered too much, the heat produced by the electrochemical reaction (R1) is
insufficient to sustain hydrogen formation (R2–R3). There is also an upper limit to the fuel utilisation.
If Uf is too high, the fuel cell is said to suffer from starvation, i.e. a lack of hydrogen to drive R1 (Fang
et al., 2015). Starvation negatively affects the cell voltage, and should therefore be prevented.
In this research, it is assumed that the region of safe operation lies between a Uf of 0.6 and 0.95. In
recent cell tests, significant starvation was observed at 90% Uf (Fang et al., 2015), so this is a positive
assumption.

2.6.3 Air utilisation

The concentrations of oxygen and hydrogen at both electrodes determine the driving force of the fuel cell
(Eq. 2.2). A higher excess of air results in a higher average oxygen concentration at the air electrode, and
therefore a higher cell voltage. SOFCs are often designed to operate at low air utilisations, since the air
flow also provides cooling to the fuel cell stack (Wendel et al., 2016). Similar to the fuel utilisation, there
is a trade-off between a low utilisation and BoP costs; large air flows require substantial investments in
pumps and heat exchangers (Strazza et al., 2015).
Air utilisation is defined as the ratio of oxygen consumed (defined by R4) to the supplied amount
of oxygen:
2 · ∆ṅCH4 2 · ṅin
CH4 Uf
Ua = = (2.5)
ṅin
O ṅin
O
2 2

2.7 Concepts
Building on the technology of SOCs, several applications have been proposed. The most well-known
application is the SOFC, followed by the SOEC. These and other concepts discussed in literature are
shown in Figure 2.8. Six concepts are explained in more detail below.
Other concepts not treated here are the integrated gasification fuel cell powerplants, tri-generation
of electricity, heating and cooling, and SOFCs applied in fuel cell electric vehicles (FCEVs) (Wachsman
et al., 2012).

2.7.1 Superwind
The Superwind concept (Vernay et al., 2008) builds upon the capability of IR-FCs to co-produce hydro-
gen. Not only SOFCs, but also molten carbonate FCs (MCFCs) can perform CHHP production. The
business model of Superwind is intended for wind park owners, who can complement the fluctuating
electricity production using a fuel cell. The fuel cell has extra operational flexibility, because it can

28
 Higher current density
SOEC ?

R-SOC ?

μ-CHP ?

SOC SOFC IR-SOFC ?

Lower temperature
IT-SOFC LT-SOFC

Hybrid ?
Higher sulfur tolerance
SOFEC ?

Figure 2.8: Development directions of solid oxide cell technology. Current and possible future appli-
cations are shown. R-SOC: reversible SOC, IR: internal reforming, IT: intermediate temperature,
LT: low temperature.

operate in different modes with other product ratios. The produced amounts of its three outputs can be
controlled by changing the fuel inflow and cell voltage (Hemmes et al., 2005). When the wind turbines
generate ample electricity, the fuel cell will switch to a mode with low fuel utilisation. This results in
a higher rate of hydrogen production relative to electricity. When less electricity is needed, the fuel
supply can be lowered, leading to a reduction in the sum of outputs. The concept is also applicable in
combination with other intermittent power sources.
Various studies have addressed the feasibility of Superwind in technological, social, and economical
terms. Vernay et al. (2008) have experimentally demonstrated that a MCFC is capable of operating
under a changing fuel supply and/or voltage. A technical limitation is the somewhat slow rate of
power output increase. Operation with high power output was not tested. Manné (2009) argues that
Superwind can be accepted by the public if it becomes better known, and if fears considering the safety
of hydrogen are taken away. From an economic point of view, the concept is most attractive in systems
with a moderate renewable energy market share and a local demand for hydrogen (Van Leeuwen, 2015).
The output of electricity is profitable because it prevents underproduction fines, or is produced during
peak demand hours (Vernay et al., 2008). Injecting hydrogen in the NG distribution network is less
profitable, since it then has no added value compared to NG. Instead, hydrogen can be sold to companies
or hydrogen fuel stations (Van Leeuwen, 2015). In spite of these business opportunities, the feasibility
study of Vernay et al. found that Superwind is not economically viable due to the high investment
costs of the fuel cell. The latter fact is also a barrier for investments and experiments by organisations
(Manné, 2009).

2.7.2 Fuel-assisted electrolysis

SOCs can be applied for electrolysis of water, thereby functioning as hydrogen source (Energinet.dk,
2011). In order to prevent high overpotentials due to the slow water reduction reaction, a fuel can be
supplied to the anode of what is now called a SOFEC (Cinti et al., 2016). The idea of producing hydrogen
by feeding electricity and methane to a SOFC was first described by Pham et al. (2000). This hydrogen
production method is more efficient than regular electrolysis (Ewan & Adeniyi, 2013). If electricity is
derived from coal powerplants, then it is not only more efficient, but also more environmentally friendly
to use a SOFEC instead of SOEC. Furthermore, SOFECs can be used for peak shaving in electricity
markets with a high penetration of intermittent power sources.
In theory, the electrolysis mode can be added to the options of the Superwind fuel cell. The current
knowledge of and experience with SOFCs that can also run in reverse is limited, since the technology is
in an early stage of development (Minh & Mogensen, 2013). More research is required to increase the

29
durability of these devices and find the ‘optimal’ design parameters.

2.7.3 Reversible hydrogen production

The reversible hydrogen production concept addresses electricity excesses by storing the energy in the
form of hydrogen. A R-SOC operates in electrolysis mode at low electricity prices, and in fuel cell mode
with H2 as fuel when electricity prices are high. This system would have a high storage capacity in the
form of H2 tanks. It is not yet cost competitive with other electricity storage systems, due to the low
round-trip efficiency from electricity to hydrogen and back. An important barrier for higher electrolysis
efficiency are high polarisation losses (Remick & Wheeler, 2011). Another challenge is provided by the
considerable amounts of heat consumed and generated during charging and discharging (Gençer et al.,
2014). As explained in § 2.4.2, the oxidation of H2 in a fuel cell is an exothermic reaction, implying
that heat is generated. In the reverse reaction, this amount of heat has to be provided. Thus, either
a heat storage system is required, or the cell stack has to be connected to a (high-temperature) heat
source and sink.
The largest R-SOC installation has a capacity of 50 kWe and was installed by Sunfire in 2016(Fuel
Cells Bulletin, 2016).

2.7.4 Reversible fuel production

The setup introduced above involves the transfer of large amounts of heat. This can be prevented
by using the reforming capability of SOFCs and performing co-electrolysis of CO2 and H2 O in SOECs
(Ridjan et al., 2013). These reactions will compensate for the (reverse) H2 oxidation reaction. For an
efficient co-electrolysis reaction, a CO2 source is needed. Wendel et al. (2016) propose a system where
the exhaust gases from the SOFC mode are stored, and used for this purpose. In fact, a closed system
can be designed, with storage tanks for fuel, exhaust gas, and oxidant (Bierschenk et al., 2011; Wendel
et al., 2016). Alternatively, underground storage can be considered (Jensen et al., 2015). Mathiesen
(2008) discusses SOECs and their ability to electrolyse water and CO2 . He emphasizes that the low
efficiency makes other energy storage options more attractive.

2.7.5 Hybrid stack

Tao (2007) developed a SOC that is functional in both SOFC and SOFEC mode. He also demonstrated
a hybrid stack. This is a SOC stack of which some cells operate in SOFC mode, and others in SOFEC
mode. It is thus possible to operate reversibly on a stack level, but also on a cell level. The mode of
operation for each individual cell is controlled by the choice of the cathode gas. Steam results in SOFEC
behaviour, and air results in SOFC mode. This is possible when the direction of the current is aligned
for the cells in a stack. A hybrid stack produces both hydrogen, heat and power, as long as there are
not too few cells in fuel cell mode. When most cells are operated in SOFEC mode, the stack requires
a supply of electricity to keep operating.

2.7.6 Micro-CHP
Since SOFCs can be downscaled without loss of efficiency, it is possible to design systems small enough
to supply one household with heat and power (Pfeifer et al., 2013). SOFC-based micro-CHP systems
are considered as an alternative for conventional boilers. The micro-CHP can be configured to respond
to electricity price signals (such as in smart grid configurations) or follow the heat demand of the
household. In IDA’s climate plan (from the Danish Society of Engineers) (Mathiesen et al., 2009), fuel
cells are introduced in decentralized and central CHP plants, as well as micro-CHP. Using NG, biogas,
or synthetic fuel, the fuel cells provide a flexible and reliable electricity source to complement renewable
electricity. In Kikuchi et al. (2014), the role of SOFCs as distributed CHP was suggested too.
Several large-scale pilot projects with micro-CHPs have been completed or are in progress in Europe.
Examples include SOFT-PACT (2011–2015), ene.field (2012–2017), PACE (2016–2021) (European
Commission, 2016), and Powermatching City (2009–2015) (Gerdes et al., 2014).

30
2.8 Conclusion
Strict requirements apply to the materials of which the electrodes, electrolyte, and housing of SOCs
are made. All components have to be resistant towards high temperatures and thermal stress, traces
of sulfur and other contaminants, and anticipated redox environments. Research has led to profound
improvements on these aspects. The largest issues remaining are related to the high overpotential in
electrolysis cells, degradation caused by sulfur, and stability in general over the lifetime of the stack.
SOFECs have the advantage of lower overpotentials, but the experimental demonstration of larger stacks
and long periods of time is still lacking.
To guarantee an efficiently operating SOC, it is important that polarisation losses are kept at a
minimum, also after several years of operation. For ohmic, concentration, and activation losses, some
mitigation strategies that can guide further research were discussed. The length of the TPB proves to
be an important parameter related to losses. Therefore, preservation of the electrode microstructure
throughout the lifetime of a SOC has a high priority when material selection is concerned.
It is known how SOFC systems can be designed and operated. Some modifications are needed to
obtain a SOFEC. The system layouts in Figure 2.3 and 2.4 – with pre-reformer but without AGR and
CGR – are assumed in the further modelling efforts of the next chapter. The operating conditions in this
chapter (SCR, Uf , Ua ) are important for the performance of a SOFC or SOFEC. In principle, reversible
SOFC/SOFEC systems can be manufactured and operated, but experience is limited to lab-scale tests.

31
Chapter 3

SOC Input-Output Relations

3.1 Cell performance modelling

To derive a mathematical description of the relation between inputs and outputs of SOFCs, it is useful
to have a model of a fuel cell. A range of mathematical models that describe SOCs have been proposed
in literature. These models differ in complexity and level of detail, depending on the research objective.
The similarity is their basis of physical principles. Examples of model types are static, dynamic, kinetic
and lumped models (Yadav et al., 2015). Static or steady-state models assume a constant operation that
does not change in time (Huo et al., 2008). Dynamic or transient models, on the other hand, incorporate
changes in e.g. fuel supply or temperature changes. Kinetic models have a bottom-up approach. They
intend to include all kinetic effects and transport phenomena that play a role at the electrodes. By
doing so, Trendewicz and Braun (2013) derived an input-output model for a SOFC as CHP source.
Lumped models are top-down oriented, and describe the fuel cell as if it were a black box. In a lumped
model, thermodynamic laws are applied at the level of the whole cell, thus reducing complexity. By
taking energy and mass conservation into account, the inputs and outputs can be calculated (Yadav et
al., 2015). This approach was used by Wendel et al. (2016) to model a R-SOC.
Fuel cell models can also be classified based on the number of dimensions of the model. Two- and
three-dimensional (2-D and 3-D) models allow to include transport phenomena. 0-D and 1-D modelling
can be applied for control purposes on a system-level. Lumped models are 0-D by definition.
To evaluate a model, a numerical or analytical (exact) approach can be adopted. Simpler models
can typically be solved using analytical methods. This is not always possible for more complex models.
To simplify a numerical model, the cell can be conceptually divided into sub-units to discretise the
problem (Hemmes, 2004). Finally, there is a great variety in the complexity of models. Dynamic and
more-dimensional models tend to cost more computation time. Still, these models are useful to assess
the potential for performance improvements and to explore different design options. Lumped or 0-D
models typically have a much shorter computation time. These models approximate the input-output
relations using a top-down approach and experimentally determined values.
Given the large amount of parameters that determine how a SOFC operates, it is inevitable to
assume a fixed value for some of them in further analysis. For example, Trendewicz and Braun (2013)
assume a fixed operating voltage and fuel utilisation in their economic assessment. In linear energy
system models (§ 4.2.4), the complexity of fuel cells is reduced to a component that converts fuel to
electricity and heat with a fixed efficiency. The model constructed by Steup (2014) features a fuel cell
that co-produces heat, hydrogen, and power, implemented with non-linear equations. Only the applied
voltage can be adapted during operation.

3.2 Operating conditions

As concluded in § 2.6, the selected operating conditions influence the performance of a SOFC or SOFEC.
For the input-output models presented in the following sections, a set of operating conditions is selected
and listed in Table 3.1. The composition of NG was derived by assuming that all hydrogen molecules in

32
 Table 3.1: Operating conditions assumed for SOFC and SOFEC operation.

Air composition 79% N2 , 21% O2

NG composition 95.81% CH4 , 4.19% N2
T 1073.15 K (800 ℃)
p 1.013 bar (1 atmosphere)
SCR 2
Ua (SOFC) 50%
Us (SOFEC) 25%
Uf 60–95% (SOFC); 80% (SOFEC)

NG from Groningen are present in the form of CH4 , and the remainder is N2 (see § 5.3.4). The inflow
to the fuel electrode is a mixture of NG and steam, for which the SCR is 2. This ratio is assumed to be
sufficient to prevent carbon deposition (see § 2.6.1). Normal air is used as oxidant flow in SOFCs. The
choices for the utilisation of air, steam, and fuel (Ua , Us , Uf ) are motivated in the following sections.
Furthermore, the area-specific resistance ASR is assumed to be independent of the current density.
This is a reasonable assumption, as long as i is not too high and not too low (see Figure 2.6). An
experimentally determined value of 0.45 Ω cm2 is used (Trendewicz & Braun, 2013; Tao et al., 2011)
for both SOFCs and SOFECs.

3.3 Nernst losses

As discussed in § 2.5, the reversible cell voltage is reduced due to Nernst losses, and the dependence on
Uf can be approximated by Eq. 2.3. To obtain a relation that describes the cell voltage as a function
of Uf , an analytical 1-D model of a fuel cell is used. The method is explained in this section.
To accurately describe reversible losses, the concentrations of each gas at each position in the fuel cell
should be known. An approximate solution can be obtained by dividing the fuel cell in hypothetical sub-
cells. For each of these sub-cells, the local Nernst potential can be calculated. The size of each sub-cell
is chosen in such a way that the change in Uf between them is fixed (Hemmes, 2004). Furthermore, it
is assumed that the reaction rates for R2 and R3 are identical and much larger than the redox reaction
rates. Therefore, SMR reactions have reached an equilibrium in each sub-cell. The corresponding
equilibrium equation is:
pCO2 · pH2 4
K2+3 = (3.1)
pCH4 · pH2 O2
The partial pressure of each species (pS) can be expressed in terms of the initial amount present, the
local fuel utilisation u, and the extent of reaction R2 ξ2 . The fact that ξ1 = 4u and the assumption
that ξ2 = ξ3 are used to derive Eq. 3.3, which gives the number of moles nS for each species.
nS (u)
pS(u) = P (3.2)
s ns (u)
nS (u) = nS (0) + ξ1 νS,1 + ξ2 νS,2 + ξ3 νS,3 = nS (0) + 4u νS,1 + ξ2 (νS,2 + νS,3 ) (3.3)

Here, νS,R is the stoichiometric coefficient of species S in reaction R. Using these expressions, Eq. 3.1
can be rewritten as:
(nCO2 (0) + ξ2 ) · (nH2 (0) − 4u + 4ξ2 )4
K2+3 = (3.4)
(nCH4 (0) − ξ2 ) · (nH2 O (0) − 2ξ2 )2 · 2ξ2
Since the value of K2+3 is known (177.7 at 800 ℃), Eq. 3.4 can be solved for ξ2 . It is advised to
use a numerical method to find the solution of this non-linear equation. Once the extent of reaction is
known, Eq. 3.2–3.3 can be utilised to calculate all concentrations in a sub-cell. Then, the local Nernst
potential Vrev
∗ follows from Eq. 2.2 for a SOFC and, from an analogous equation for a SOFEC:

∆G pH2 [s] · pH2 O[f]

 
Vrev = with ∆G = ∆H − T ∆S + RT ln (3.5)
nF pH2 O[s] · pH2 [f]

33
where the Nernst potential difference of the reactants and products at the fuel electrode [f] and steam
electrode [s] is calculated.

RT  pH2 (u) · pO2 (u) 

 q 

SOFC : ∗
Vrev (u) = V ◦ + ln (3.6)
4F pH2 O(u)
RT pH2 O[s](u) · pH2 [f](u)
 
SOFEC : ∗
Vrev (u) =V + ◦
ln (3.7)
4F pH2 [s](u) · pH2 O[f](u)

with V ◦ the (open) cell potential under standard concentrations.

To obtain the Nernst potential of the whole cell, the sub-cell potentials should be averaged:

1
Z Uf
Vrev (Uf ) = ∗
Vrev (u)du (3.8)
Uf 0

If it is assumed that Vrev

∗ has a linear dependence on u, then the integral can be solved, yielding Eq. 2.3

again:
∗
∂Vrev
Vrev (Uf ) ≈ V0 + α · Uf with 2α = (3.9)
∂u
The value of α and V0 can now be found by fitting Eq. 2.3 to Eq. 3.8.
To calculate the reaction progress from Eq. 3.4, the initial amounts presented in Table 3.2 were
used. The calculation is made on the basis of 1 mol CH4 , with the other amounts derived based on the
assumed conditions (Table 3.1). For the case of a SOFC, a maximum air utilisation of 50% is assumed,
whereas a steam utilisation of 25% is set for a SOFEC.

Table 3.2: Inital amounts of gases at the fuel and air electrode of a SOFC, and at the fuel and
steam electrode of a SOFEC.

Fuel electrode (both) Air el. (SOFC) Steam el. (SOFEC)

CH4 H2 O N2 H2 CO2 O2 N2 H2 O H2
nS (mol) 1 2 0.0765 0 0 4 15 16 0

Proceeding as outlined above, it was derived that α = −53 mV for a SOFC and α = −87 mV for
a SOFEC. Finding a linear fit for Vrev (Uf ) as determined from Eq. 3.8 in the range 0.60 ≤ Uf ≤ 0.95,
results in the following equations:

SOFC : Vrev (Uf ) = 1.069 − 0.053 · Uf (3.10)

SOFEC : Vrev (Uf ) = 0.116 − 0.087 · Uf (3.11)

These linear approximations are a very close to the real Vrev , as indicated by Figure 3.1.
Now that the reversible voltage is known, it is possible to calculate the voltage of a single cell from
Eq. 2.1:
Vcell = V0 − αUf − i · ASR (3.12)
This expression contains ASR, which accounts for all losses occurring due to the polarisation effects
discussed in § 2.5.

34
 1.08

1.04

Reversible voltage (V)

SOFC
1.00
0.16
SOFEC
0.12

0.08

0.04

0
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Fuel utilisation

Figure 3.1: Reversible cell voltage of a SOFC and a SOFEC as a function of Uf , calculated from
the local cell voltage Vrev
∗ . Linear approximations of V (U ) within the operating range are drawn.
f

3.4 Mass and energy balances

3.4.1 SOFC
Knowing Vcell , the next step is to relate it to the mass and energy flows through the cell. Since the
mass and energy balances of SOFCs and SOFECs differ, those of SOFCs are discussed first. The case
of SOFECs is treated in the next section.
Mass flows are expressed in units of moles per unit of cell area. The molar influx rate of methane
(JCH
in ), to begin with, can be related to the current density by recognising that eight electrons are
4
transferred per mole of CH4 in R4.

i = 8F Uf JCH
in
4
or (3.13)
i
in
JCH = (3.14)
4 8F Uf

Then, using Eq. 3.12 and 2.3, the power flux JPout equals:

JPout = i · Vcell = i · (V0 − αUf − i · ASR) (3.15)

The molar hydrogen outflux rate is based on the mole balance for H2 and Eq. 3.13. A correction
factor of 0.81 is applied here. This factor accounts for the fact that in equilibrium, the water gas shift
reaction (R3) is not complete, so ξ˙3 < JCH
in .
4

4i 0.81i 1
!
JHout = 0.81 · (1 − Uf ) · 4JCH
in
= 0.81 · (1 − Uf ) = · −1 (3.16)
2 4 8F Uf 2F Uf

The heat flux calculation is based on the energy balance over the fuel cell and afterburner (see Figure 2.3),
and assumes that any CH4 or CO remaining in the exhaust gas is fully oxidised in an afterburner (after
H2 has been removed).
JQout
= JCH
in
4
· ∆c HCH4 − JPout − JHout
2
· ∆c HH2 (3.17)
Here, ∆c Hs denotes the HHV of species s.
The resulting input-output relations resulting from the equations above are depicted in Figure 3.2.
The stackplots show that hydrogen production is mainly determined by Uf . Furthermore, the highest
electric efficiency occurs at low current densities. i has a lower limit, below which the SOFC system

35
is a net consumer of heat. In Figure 3.2, the maximum power point is indicated. If the current
density exceeds the value belonging to this point, the electricity output no longer increases. During
SOFC operation, the maximum power point should not be passed to prevent the unnecessary efficiency
reduction.
1.8 1.8
Heat

Energy output (W/cm²)

Energy output (W/cm²)

1.6 1.6
1.4 1.4 Hydrogen
1.2 1.2 Electricity
1.0 1.0
0.8 0.8
0.6 0.6
Maximum
0.4 0.4
power point
0.2 0.2
0 0
60% 65% 70% 75% 80% 85% 90% 95% 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3
Fuel utilisation Current density (A/cm²)

(a) At i = 0.9 A/cm . 2

(b) At Uf = 85%.

Figure 3.2: Outputs of a SOFC as modelled by Eq. 3.13–3.17. Note that, because of Eq. 3.17, the
sum of the outputs equals the input of fuel. α = 0.053 V, V0 = 1.069 V, and ASR=0.45 Ω cm2 .

3.4.2 SOFEC
Many of the relations derived for a SOFC also apply to a SOFEC. Eq. 3.14, 3.15 and 3.17 are still valid.
Note that Vcell – and therefore JP – is negative, reflecting the fact that a voltage is applied to instead of
drawn from the cell. Another difference is the compositions of gases. In a SOFEC, the steam electrode
gas is a mixture of H2 O and H2 . This results in a radically different Vcell compared to a SOFC. Any
H2 or unconverted CH4 in the fuel electrode flue gas is combusted in the afterburner (see § 2.4.6).
Hydrogen generated at the steam electrode is directly related to the current (similar to an electrolyser),
and provides the H2 output:
i
JHout = (3.18)
2 2F

2.5 2.5
Methane in, JCH4
Electricity in, –JP
Energy in-/output (W/cm²)

Energy in-/output (W/cm²)

2 2 Hydrogen out, JH2

Heat out, JQ

1.5 1.5

1 1

0.5 0.5

0 0
0.3 0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9 2.1 2.3
60% 65% 70% 75% 80% 85% 90% 95%
Fuel utilisation Current density (A/cm²)

(a) At i = 1.0 A/cm2 . (b) At Uf = 80%.

Figure 3.3: Input and outputs of a SOFEC, as described by Eq. 3.14, 3.15, 3.17, and 3.18. All
functions are plotted as positive values.

The internal resistance is equal to that of a SOFC, provided that it is a property of the electrode
and electrolyte materials. The fuel utilisation has a big influence on the amount of methane required
to produce a given volume of hydrogen, whereas the cell potential is hardly affected (see Figure 3.3a).
The SOFEC’s Uf was fixed at 80%, since this is the optimum (Wijers, 2011). A higher Uf might lead

36
to starvation, i.e. a lack of fuel, causing high overpotentials (Cinti et al., 2016; Fang et al., 2015).
Lower utilisations reduce the efficiency of the conversion from CH4 to H2 , as indicated by Figure 3.3a.
Therefore, deviating from the optimum value has no clear benefits. The calculated SOFEC inputs and
outputs are plotted in Figure 3.3b.

3.4.3 Scaling up
The balance of plant includes all auxiliary equipment needed for a correct operation of the SOC system.
Heat losses occur throughout the system and via the exhaust. Electricity is lost in the DC/AC (direct to
alternating current) inverter. Furthermore, electronic control systems, pumps, blowers, and compressors
use some electric power. An efficiency of 96% is assumed for the DC/AC inverter (James et al., 2012).
Parasitic loads are responsible for a reduction of 10% of the electricity output. Large contributors to
BoP energy consumption are compressors and pumps (James et al., 2012). Based on these assumptions,
it can be stated that the BoP efficiency ηBoP = 96% × 90% = 86.4%. The fraction of electricity that is
‘lost’ is converted to heat within the system, resulting in a lower power output of a SOFC, and a higher
power demand for a SOFEC. It is assumed that only 5% of all heat is lost from the system, while the
remainder of heat generated is transferred to an external heat demand.
The costs for a fuel cell system are very dependent on the scale of the system. A cost comparison
performed by (James et al., 2012) indicated that a 100 kW SOFC system is around 13 times as expensive
per kW as a 1 kW system. An other important factor for the production costs is the scale of the
manufacturing facility. Both effects are not taken into account in the model.
Unlike conventional powerplants, fuel cells become more efficient when operating at partial load. As
observed from Figure 3.2, a SOFC converts a higher fraction of the input energy to electricity at low
current densities due to the lower overpotential. This is an advantage over conventional powerplants,
such as gas turbines. Not only do gas turbines demand a minimum load of 40%, their efficiency at that
point is almost 10% lower than the full-load efficiency (Figure 3.4). Besides, emissions of CO and NOx
per kWhe tend to increase for lower loads (López, 2013).

60

50

40

30
Wärtsilä Simple Cycle Engine
Wärtsilä FlexiCycle Engine
20
Siemens SGT6-500F Simple cycle
GE 7FA,05 Simple cycle
Minimum load limits
10
Gas
Siemens SGT6-500F CCGT
Reciprocal FlexiCycle GE 7FA,05 CCGT
turbine
engines steam turbine
0
range
0 10 20 30 40 50 60 70 80 90 100

Figure 3.4: Partial load efficiencies of NG-fuelled combustion engine and gas turbine technologies
in a 25 ℃ environment. Reprinted from López (2013).

In § 3.4.1, the energy flows per unit of active area were derived. In this section, the flows of interest
are the total input or output of methane, electricity, hydrogen, and heat (ḢCH4 , P , ḢH2 , Q̇), all in
units of power. These will be expressed relative to the maximum input capacity for a collection of fuel
cell stacks. The power output, to begin with, depends on the voltage of the single cells. The power
output is determined using the equation for Vcell (Eq. 3.12):

P = Vcell · I = (V0 − αUf − i · ASR) · I (3.19)

37
The total current I is determined straightforward from the methane inflow (either molar ṅin
CH , or 4
energetic ḢCH4 ) and fuel utilisation.

ḢCH4
I = 8F Uf ṅin
CH4 = 8F Uf (3.20)
∆c HCH4

The current density and total current are related to each other by the total active area A. In turn,
A is determined by the maximum CH4 inflow allowed, which occurs at the maximum current density
imax and highest fuel utilisation. imax is derived as follows:
∂P
(imax ) = 0 (3.21)
∂i
∂P
= V0 − αUf,max − 2i · ASR (3.22)
∂i Uf =Uf,max
V0 − αUf,max
imax = (3.23)
2ASR
Using Eq. 3.23, A is calculated by inserting the maximum current and current density.

I 8F Uf ḢCH4 8F Uf,max ḢCH4 ,max

A= = =
i ∆c HCH4 · i ∆c HCH4 · imax
8F Uf,max · ḢCH4 ,max 2ASR
= (3.24)
∆c HCH4 · (V0 − αUf,max )

To find the ohmic loss term i · ASR in Eq. 3.19, the expressions for the total current and total area
(Eq. 3.20 and 3.24) are subsituted to obtain Eq. 3.25. The result is that the ohmic loss does not depend
on ASR any more. The only variables are ḢCH4 and Uf .

I · ASR ∆c HCH4 · (V0 − αUf,max ) 8F Uf ḢCH4 ∆c HCH4 · (V0 − αUf,max )

i · ASR = =I· = ·
A 16F Uf,max ḢCH4 ,max ∆c HCH4 16F Uf,max ḢCH4 ,max
Uf ḢCH4
= · 21 (V0 − αUf,max ) (3.25)
Uf,max ḢCH4 ,max

Although this substitution is useful for creating an input-output model, it should not be forgotten that
ASR remains an important parameter in particular with respect to costs (Tao, 2007). If ASR can be
reduced by 50%, then the same power can be produced using only half of the original cell area.
For a SOFEC, the maximum current density cannot be determined similar to that of a SOFC, because
the current-power relation does not have a maximum. So instead of deriving imax from the maximum
power point, it is set to 2.3 A. As a consequence, a different expression for the ohmic loss term appears.
By rearranging Eq. 3.24, we see that:
imax imax
i=I· = ḢCH4 · (3.26)
Imax ḢCH4 ,max

3.5 Implementation
The implementation of a SOFC and a SOFEC component in an optimisation model (see Chapter 4)
is based on the equations presented in § 3.4. The equations below define the inputs and outputs in
terms of I, i and Uf . I and i can be related to each other using Eq. 3.25 and 3.26 for SOFC and
SOFEC, respectively. The unknowns in these equations are ḢCH4 , P, ḢH2 , Q̇, Uf , I, i. For both SOFC
and SOFEC, there are 4 input-output relations and one I − i relation. Therefore, there are 7 − 5 = 2

38
degrees of freedom in both cases. Note that P and Q̇ are defined to be positive if power or heat is
produced by the SOFC or SOFEC.
∆c HCH
=I



ḢCH4 8F Uf
4

= I · (V0 − αUf − i · ASR) · ηBoP



P
(3.27)

SOFC




ḢH2 = I · 0.81
2F ∆c HH2 · ( Uf − 1)
1

= 0.95 · (ḢCH4 − P − ḢH2 )


Q̇

 ∆c HCH


ḢCH4 =I 8F Uf
4

= I · (V0 − αUf − ASR · I · Imax )


imax 1

P ·
(3.28)

SOFEC ηBoP


ḢH2 = I · 2F
1
∆c HH2
= 0.95 · (ḢCH4 − P − ḢH2 )


Q̇


Apart from the equalities listed in Eq. 3.27 and 3.28, an inequality constraint applies to restrict the
installed capacity. In energy converters, it is conventional to express the installed capacity is defined
in terms of energy output. However, this is not useful for flexible multi-output technologies, with their
fluctuating conversion efficiencies. Therefore, the installed capacity of SOFCs and SOFECs is expressed
relative to the fuel input, yielding the following inequality:

ḢCH4 ,e (t) ≤ IC e + IC e (3.29)

add
∀e, t

Here, ḢCH4 ,e (t) is the input of methane for component e at timestep t, and IC e and IC e
add
are the
available and maximum additional installed capacity.

3.6 Nonlinearity
The relations that were derived to describe the inputs and outputs of a SOFC (Eq. 3.27) are nonlinear.
Therefore, either the nonlinear problem has to be solved, or it has to be simplified to a linear problem. It
is not possible to retain the dependence of the outputs as a function of both I and Uf . This is because
the expressions for the input of methane, and the output of power and hydrogen (Eq. 3.14–3.16) all
contain terms featuring both of these variables.
A linear relation is obtained when the fuel cell operates at a fixed fuel utilisation, and when the
power–current relationship is approximated by a linear function.
Alternatively, a linear mixed-integer relation can be formulated. For example by defining a number
of states, characterised by their fixed fuel utilisation, between which the SOFCs can switch. Under
this approach, an integer variable determines the fuel utilisation of all SOFCs in the model. The
power–current relation is described by a slightly different linear equation for each state.
The fact that the efficiency of SOFCs decreases with their total output, allows for an easy linearisation
with respect to the capacity. If for example ten individually controlled stacks are installed, the highest
system efficiency is achieved when each of the stacks operates at 10 1
th of the total required load. Thus,
the output is simply ten times the output of a single unit with the same capacity factor. In other words,
an collection of many SOFCs has the same input-output relation as a single stack.
Provided that a linear problem can be solved faster, a trade-off has to be made between the accuracy
of the model and computation time. For this research, the decision was made to retain the complexity
of SOC operation, thereby accepting the additional time needed to solve the optimisation problems.

3.7 Summary
This chapter has employed a simple thermodynamic model to arrive at a set of equations that describe
the inputs and outputs of a SOFC and SOFEC, namely Eq. 3.27 and 3.28. These equations have two
degrees of freedom each, which is required to account for different ratios of outputs, and variation in the

39
total throughput. The energy flows are expressed as a function of density and fuel utilisation, although
other variables such as fuel input and cell voltage can also be chosen. This gives sufficient accuracy for
the implementation in an energy system model. Yet, the implication is that a non-linear problem has to
be solved.

40
Chapter 4

Energy System Modelling

4.1 Introduction
4.1.1 Energy systems
The aim of energy systems is to fulfil the need for energy by consumers. A good system solves the
mismatch between energy supply and demand at three levels: time, place, and energy carrier (Hemmes,
2009). The technical components of a supply chain are operated to overcome these mismatches. Trans-
portation networks address the spatial mismatch, storage deals with temporal variances, and converters
ensure the desired energy carrier is delivered. The ensemble of components in not static; replacements
and additions are occurring frequently.
The choice for a set of technologies is a trade-off between four parameters: affordability of energy,
security of supply, sustainability (both GHG emissions, land use, and resource use), and social accept-
ability (safety concerns, perception of health impacts, comfort, etc.) (Mancarella et al., 2016). How
these parameters are weighted is a normative judgement, and differs between actors. In the Netherlands,
the liberalisation of the energy market has been completed to ensure affordability. The risk of supply
failure is very low, which is managed by distribution system operators (DSOs) under strict regulation by
governments. The societal attitude towards the social and sustainability aspects are subject to change.
The acceptance of different sources of energy can increase or decrease within a few years time.
Currently, a transition towards more sustainable energy production is taking place. Mainly driven
by governmental and societal pressure, the unpredictability of fossil fuel prices, and an expected cost
reduction of emerging technologies, new energy production and consumption technologies are installed.
This development is expected to require a drastic restructuring of the energy system (Bokhoven et
al., 2015). Some of the entailed changes are a growing share of intermittent power sources and more
decentralized generation. To retain the resilience of the system under these changes, it is expected that
the energy production system will become increasingly complex in the future (Bourennani et al., 2015).
To conclude, it seems inevitable that the importance of the energy system society will increase, while it
is uncertain how it will develop.

4.1.2 Optimisation
Many choices are to be made regarding the design of energy systems. Multiple energy (re)sources are
available, even more options to convert various energy carriers, and a number of options to use final
energy to provide a certain service. This raises the question what combination of technologies is best
suited to provide the energy services needed. Finding an answer to this question is in the strategic
interest of all stakeholders involved in the energy supply chain. Energy users are best served by energy
carriers that meet their needs in a comfortable, reliable and inexpensive way. For energy producers and
distributors, the challenge is to know the demand for each energy form in time. Engineers who operate
and plan energy infrastructure are interested in reliable technologies that can be operated flexibly.
And policy makers are assigned to secure an affordable, reliable, sustainable, and safe energy supply
(Mancarella et al., 2016).

41
 NG Electricity NG H2 Electricity

Powerplant Powerplant
CO2 Electrolysis

H2 Methanation
Fuel cell
SMR SMR

SOFC

Figure 4.1: Schematic overview of energy conversion processes that connect electricity, hydrogen
and natural gas (NG) in the current system (left) and future options (right).

Most models that were created in the past only address one energy carrier. Electricity is studied
most often, because the balancing of the electricity grid is a delicate task. Models for other energy
carriers also exist, for example to optimise hydrogen infrastructure (Strachan et al., 2009). Regarding
the expected energy transition in the near future, interactions between the supply chains of different
energy carriers can no longer be neglected. The markets for electricity, hydrogen, gas, heat, and liquid
fuels will become increasingly interconnected (Bokhoven et al., 2015), creating the need for models
that cover the whole energy system. It is not clear yet whether hydrogen will become a major energy
carrier, but it could play a role as connecting element between methane and electricity. In Figure 4.1,
the potential role of hydrogen is illustrated. H2 can be produced by electrolysis of water from electricity,
or by steam methane reforming (SMR) from NG. Fuel cells or combustion engines can (re)generate
electricity. Methanation of CO2 , converts H2 to methane. It is also possible to perform co-electrolysis
of H2 O and CO2 to obtain methane directly (Ridjan et al., 2013).

4.1.3 Path dependence

Since energy infrastructure has a lifetime of several decades, decisions made in the past can have
consequences reaching far into the future. This effect is called path dependence, and may lead to lock-
ins, unfavourable system configurations that are hard to escape from. The existence of path dependence
is one of the reasons to use energy system optimisation when planning new investments in the system.
Since optimisation is an exploratory mode of research, it can generate unexpected system configurations.
This type of research is relevant for decision makers because of four more reasons (Schenk et al., 2007):

• The high investment costs related to the energy system make that a large amount of money can
be saved if a more optimal system is created.
• The long lifetime of energy technologies imply that decisions made now have implications for a
long time period.
• There is a danger of creating lock-in situations, for which it is hard to move away from.
• The rapid development of novel technologies can have far-reaching impacts on the system, so the
possible effects should be anticipated on.

When a future energy system configuration is determined by means of optimisation, a new question
raises immediately: how can the current system be transformed into the optimal future system? The
answer can be found by optimising longer time periods, taking the lifetime of existing installations into
account.

4.2 Literature review

For conventional energy systems, the design and operation are of limited complexity. These systems
are composed of generation technologies that produce one energy carrier, and sometimes waste heat

42
is recovered. The problem of operational dispatch in these systems can be solved by using the merit
order of the technologies. Demand is met by turning on as many plants as required, starting with those
with the lowest operating and fuel costs. The investment decision for new plants can be made based
on a LCOE analysis (4.5). Finding an optimum for a system containing multi-generation components
is more complicated. Whether or not to operate depends on the costs and revenues of all inputs and
outputs (Chicco & Mancarella, 2009). In the past decade, various models have been made that address
the challenge of finding the optimal composition of energy systems. Existing literature describing these
modelling efforts has been reviewed. The differences and similarities are discussed below. The focus is
put on three parts of the modelling work: problem definition, implementation, and scenario construction.
In addition, software tools used and relevant results are discussed.

4.2.1 Simulation vs. Optimisation

Considering energy system models, there is a big difference between simulation and optimisation – both
conceptually and computationally. Simulation models are intended to study the performance of a certain
system design. Each component of the energy system is represented in the model by a set of heuristics
or rules, according to which the component behaves. There is no overall objective, as optimisation only
occurs at the component level. It is possible that a simulation does not find the optimum solution,
for instance when the heuristics of two components are contradictory. Also, the existence of a better
heuristic is always present.
Simulations with pre-defined control strategies are useful to assess existing technologies, but they
do not provide insight in the best way to operate or design a system. Additionally, simulations are used
to predict system dynamics and the behaviour of its components. An example of a simulation tool
is EnergyPLAN (Lund, 2007). For emerging technologies like SOCs, optimisations are more relevant,
since the technology operation is determined by the algorithm. Optimisations help system developers
to investigate the effects of design decisions and control constraints (Hawkes et al., 2009).
An optimisation model searches for the conditions, i.e. the set of values of variables, that give
the best system performance. The performance is measured by the objective function, whose result is
lowest or highest for the optimisation solution (Frangopoulos, 2003). Three types of optimisation can be
distinguished: operation optimisation, system design optimisation, and system evolution optimisation.
For the first type, the power generation equipment is fixed, but the dispatch is to be optimised. For
example, Lozano et al. (2010) have studied a small system of one CHP plant that supplies heat and power
to a collection of buildings with a given heating, cooling, and electricity demand. Notice the difference
between operation optimisation and simulation: the former finds the optimal operating scheme, while the
latter gives a possible option. The second optimisation type aims to find the optimal set of components
and their optimal operation (Frangopoulos, 2003). The third type optimises the development of a
system in time (e.g. Aki et al., 2005). Questions related to long-term strategies for installing and
decommissioning energy infrastructure can be addressed. Usually, the operation is not part of the model
in system evolution optimisation.

4.2.2 Problem definition

The formulation of the problem generally consists of determining the objective, scope, and restrictions.
The objective specifies what should be optimised. This can for example be the total costs of the system,
the net present value, or the GHG emissions. It is also possible to have minimum power outages as
an objective, for which several indicators exist (Shahmohammadi et al., 2015). A single, or multiple
objectives can be selected.
The scope sets the spatial and temporal boundaries and the available options for power generation.
Ridjan et al. (2013) take a national scope with a few options for fuel supply, whereas Aboumahboub et
al. (2010) have built a global model and a model describing Europe and North-Africa. For larger scopes,
the importance of transport and associated losses increases. The temporal scope strongly depends on
the optimisation type. Operation optimisation is short-term oriented, and should therefore have small
timesteps (in the range of minutes to hours). On the other hand, system evolution optimisation has

43
both larger timesteps and timespans. Differences also exist in the technology options that are used.
The efficiency of either existing technology (e.g. Kamjoo et al., 2016; Kaviani et al., 2009) or expected
future technologies (Kikuchi et al., 2014; Mathiesen et al., 2009; Aki et al., 2005) can be assumed. If
the selected year is further in the future, technology assumptions become more uncertain.
Restrictions specify what conditions should be met by a solution. Most, but not all models feature the
boundary condition that the energy demand should be met. If this condition is not present, the modellers
have chosen to maximize the reliability of energy supply or set a lower bound for it (Shahmohammadi et
al., 2015). Lozano et al. (2010) add legal constraints. In operation optimisation models, the number and
size of components is fixed. But also in system design optimisations, it is possible to limit components
to a maximum or minimum size. A minimum can for example be included to account for existing
infrastructure (Belderbos et al., 2015). Upper bounds can reflect limitations for expansion, such as
the limited space available for solar cells. Restrictions can also be defined for the energy storage and
transport capacities. What restrictions apply in the model, depends on the assumptions made. These
will be discussed in more detail in Section 4.2.5.

4.2.3 Mapping of energy systems

In order to map a system, a simplified representation is needed. Only the relevant dimensions of the
real-world system are mapped. This reduces the complexity, while retaining sufficient detail to answer
the research questions.
Biberacher (2004) describes a system mapping approach with two types of entities: commodities
and processes. Commodities represent energy carriers. The conversion, transport, and storage of these
carriers is undertaken by processes. The mapping approach can be applied for visualisation as well, as in
Figure 4.2. This representation proves to be a powerful tool to conceptualise any existing or imaginary
energy system, including all energy forms and technologies of interest.

Bus 1 Bus 2 Bus 3

Source 1 C2
C1
Sink 1
Source 2

Sto Sink 2
C3
Source 3
Sink 3

Figure 4.2: Representation of energy system components in a model. Three types of components are
shown: sources, sinks, and converters. Storage is a special type of component. These components
are connected via buses. Each bus represents a distinct type of energy carrier.

4.2.4 Implementation
Optimising an energy system requires the selection of an algorithm that efficiently and correctly finds
the optimum. The solution space of the problem is determined by the set of variables that should
be optimised. Given the infinite number of possibilities, it is not feasible to calculate the solution by

44
exhaustive search. Therefore other, more sophisticated algorithms are needed (Geidl, 2007). For the
implementation phase, the problem definition serves an an input.
First of all, a suitable spatial resultion can be derived from the scope definition. For a global scope,
the world can be divided into regions. For a national model, smaller spatial units can be used. This
spatial differentiation allows to model energy transport and exchange, and local differences in energy
consumption and weather conditions. For small-scale cases, these factors do not play a role, so spatial
divisions are not needed. The MARKAL model can be linked to GIS to study spatial developments
(Strachan et al., 2009). The resolution in time has to be set too. Usually, timesteps of one hour are
applied, because this is the timescale at which data on wind speeds and solar irradiation (for modelling
solar and wind energy) are available (Aboumahboub et al., 2010; Kaviani et al., 2009). Shahmohammadi
et al. (2015) have reduced the computation time of their case-study by taking one representative weekday
and weekend day for each season.
Boundary conditions are important, because they formalize requirements and restrictions from the
problem definition. Basic constraints like technological and physical limits are included in every model.
Other conditions were discussed in the previous section. The objective function is the formalisation of
the objective. The algorithm aims to minimize or maximize the outcome of this function. In their review,
Chicco and Mancarella (2009) discuss the objective function used by several modellers. They note that
sometimes, the emission of CO2 is expressed in monetary value to obtain a single economic objective.
In the case of multiple objectives, there is a trade-off, and the algorithm is referred to as multi-objective
optimisation (Frangopoulos, 2003). This has been done by for example Kamjoo et al. (2016) and Aki et
al. (2005), who present their results as a Pareto set. For the case of Kamjoo, the Pareto set describes
the trade-off between the reliability of energy supply and the costs. Aki et al. present the trade-off
between CO2 and consumer costs. By doing so, the authors avoid judging the costs of power outages
and climate change respectively.
During model formulation, a trade-off has to be made between computational burden and model
accuracy. Real-world phenomena are often not fully linear, but in some cases a linear approximation
is justified. When the optimisation problem is formulated as a linear programming (LP) problem, the
solution can be found numerically. The ever increasing computation speed and capacity of computers
opens up the road to more detailed and complex models (Erdinc & Uzunoglu, 2012). Often, the problem
formulation contains a convex objective function, binary or integer variables, or nonlinear constraints.
Integer variables (as opposed to continuous variables) describe for example start-up costs or the number
of built installations. This results in a convex solution space. In these cases, numerical tools can still
be used, but it is not guaranteed that the global optimum is found (Geidl, 2007).
Another approach is mixed-integer LP (MILP), which is a mix of LP and other algorithms such as
branch & bound. It can be applied as long as both the objective function and the constraints are linear
(Mathworks, 2015). An example of a model that uses MILP is the Bing model (Steup, 2014; Steup
& Hemmes, 2015). In case of nonlinear problems, the solution can only be approximated, for which
a number of dedicated algorithms exists. Advanced bio-inspired algorithms are becoming increasingly
popular for solving non-linear problems. Among others, particle swarm optimisation, genetic algorithms,
and simulated annealing are in use (Chicco & Mancarella, 2009; Bourennani et al., 2015). Other
approaches that are promising, but have not yet been applied in energy system optimisation models, are
ant colony and artificial immune system algorithms (Erdinc & Uzunoglu, 2012).

4.2.5 Scenarios
Before a model can run, the input parameters should be quantified. When dealing with future energy
systems, high uncertainties are involved. This is because of the distant time horizon and long lifetimes,
and because of the unpredictable dynamics of the energy transition. Therefore, the construction of
explorative scenarios becomes important. Using scenarios, sets of input parameters are generated. The
aim of explorative scenarios is to cover a wide range of possible future developments (Börjeson et al.,
2006). Sørensen (2008) presents three steps to construct a renewable energy scenario:

1. Determine the end-use demand of energy by society.

45
 2. Determine the potential energy supply of available resources.
3. Match supply and demand using intermediate conversions, storage, transmission, and possibly
imports and/or exports.
Step 1 can be approached in two ways: taking the final energy or useful energy demand. Step
2 requires that energy production technologies are selected. Step 3 involves the definition of energy
infrastructure and the simulation of supply and demand time series.
Predictions and/or assumptions have to be made for a wide range of input parameters and to set
boundary conditions. Important factors as identified by existing literature are:
• The market share of renewable energy technologies (Belderbos et al., 2015).
• The supply of fossil fuels (Ridjan et al., 2013).
• Legislation in the form of taxes, subsidies, quota, obligations (Aki et al., 2005).
• Available technologies and their efficiencies. Both for generation, conversion, and storage of
energy (Mathiesen et al., 2009).
• The demand for specific energy carriers (Kikuchi et al., 2014; Ridjan et al., 2013; Sørensen, 2008),
depending on:
– population size,
– efficiency changes of existing technology, and
– spread of new technology, like BEVs, FCEVs, electrification (heat pumps, electric stoves).
For many countries, national energy scenarios have been made. For the Netherlands, ECN has
drawn up a possible energy future (Schoots & Hammingh, 2015). In Denmark, a ‘100% renewable
energy scenario’ was developed (Mathiesen et al., 2009). On a worldwide level, there are also several
energy scenarios. See for example the IEA Outlook (IEA, 2015) and the Sustainable Global Energy
Outlook (Teske et al., 2010).

4.2.6 Results
The most relevant outcomes of the reviewed papers are presented in this section. It should be noted
that the conclusions drawn from the case studies are not always valid for other cases, in particular when
assumptions and/or scope differ.
The acclaimed advantages of FCs for CHP production have been confirmed by several studies. For
example, Aki et al. (2005) show that micro-CHP installations with PEMFCs can be effective to reduce
CO2 emissions. The penetration of FCs is accelerated by a price reduction of the devices (possibly with
subsidies), and a lower price of NG relative to electricity. Kikuchi et al. (2014) also find that distributed
co-generation systems such as SOFCs in the residential sector are part of an optimal energy system.
Aboumahboub et al. (2010) demonstrate the merits of international high-voltage electricity trans-
mission. It is shown that transport infrastructure reduces the need for energy storage and vice versa.
However, no capacity limits were considered, possibly yielding suboptimal results. Two other researches
focussed more on storage technology. Power-to-gas technology (PtG, producing methane by electrol-
ysis) is preferred over batteries, when renewable energy has a share of 80% (Belderbos et al., 2015).
The researchers explain this observation by noting that gas is an attractive energy storage medium to
handle seasonal energy availability. Batteries remain more efficient for short-term storage. Meibom and
Karlsson (2010) calculated the configuration of an economically optimal energy system in 2060. Hydro-
gen storage was included in the solution even if there was no demand for it from transport. Under the
technology assumptions made, the direct conversion of biomass to hydrogen is not attractive. Instead,
biomass combustion followed by water electrolysis was selected. Hydrogen production could also serve
as demand response mechanism (Kikuchi et al., 2014): by constructing more electrolysis facilities, the
installed capacity of fossil-based powerplants can be reduced. No reports were found that described
an optimisation model with SOFCs with the ability to produce hydrogen or operate at different fuel
utilisations, as described in Section 2.7.1. or SOFECs
One of the first optimisation models with SOFCs was demonstrated by Hawkes et al. (2009). The
model had a very limited set of components, and focussed on the degradation of the cells and the ideal

46
moment of stack replacement. The impact of a number of operating constraints were also assessed.
Hawkes et al. (2006) used an optimisation model to show that the rate of current density change of a
SOFC stack in response to load changes is not important for the economic attractiveness. Therefore, a
long lifetime should be preferred over ramp-up ability during the design phase.

4.2.7 Modelling software

The continued interest in energy system planning has resulted in a wide range of modelling and optimi-
sation tools (Connolly et al., 2010). These software tools allow to model, optimise, and analyse (hybrid)
renewable energy systems. Examples of packages with many users include energyPRO, HOMER, and
LEAP. Connolly et al. (2010) identified 68 different tools, ranging from short-term dispatch optimisation
software at the level of individual houses (HOMER), to models for long-term investment and replace-
ment planning (Balmorel, MARKAL/TIMES). In recent years, an increasing number of free and open
source models have been added to this collection (see Table 4.1). The wide variety is a result of the
diversity in problem definitions (see § 4.2.2) and optimisation algorithms. Depending on the aim of a
research, a different model will be most suitable to use.
Although EnergyPLAN is a popular tool, it is not designed for optimisations. Instead, the user
should compose different sets of technologies, for which EnergyPLAN simulates the optimal dispatch
using heuristic methods (Connolly et al., 2010).

Table 4.1: Open source renewable or hybrid energy system optimisation models (Schlecht & Davis,
2016).

Name Programming language

Balmorel GAMS
Calliope Pyomo (Python)
DIETER GAMS
EMMA GAMS
Ficus Pyomo (Python)
Genesys C++
NEMO Python
Oemof Pyomo (Python)
OSeMOSYS GNU MathProg
PyPSA Pyomo (Python)
Renpass R
Temoa Pyomo (Python)
URBS Pyomo (Python)

The number and type of predefined technology options differs between the models. In some cases,
a selection of cost and performance data is provided, while other models leave the definition of these
characteristics to the user.

4.2.8 Models of the Netherlands

A limited number of models have specifically been designed to apply to the Dutch energy system.
Netbeheer Nederland (2016) has listed 17 of these tools, of which six can handle multiple energy
carriers on a national scale. The models that have a national scope are listed in Table 4.2. Most models
are not publicly available, but are used and maintained by consultants and research institutes. Two
example studies of the Netherlands are highlighted below.
Schenk et al. (2007) have designed a model that simulates the Dutch electricity system on an hourly
basis. The simulation is employed to study the effect of different installed wind turbine capacities
on the electricity network. The merit order approach is used to determine the operation strategy of
conventional powerplants in this model. The research shows that hydrogen production via electrolysis
can prevent start-up and shut-down costs or electricity excess. However, because of the energy losses

47
Table 4.2: Tools for modelling the Dutch energy system at a national level (Netbeheer Nederland,
2016).

Model Organisation(s) Carrier(s)

Vesta PBL Heat
CEGRID CE Delft Electricity
Smart Grid Scenario Model (DSSM) DNV GL Multiple
Energy Transition Model (ETM) Quintel Intelligence Multiple
ETMoses Quintel Intelligence; Alliander Multiple
PICO Geodan; TNO; Alliander; NRG31/Waifer; Multiple
Geodan; Ecofys; Esri
OPERA ECN Multiple

associated with electrolysis, the total primary energy use was only reduced if the installed wind energy
capacity exceeded 8 GWe in the Netherlands (Schenk et al., 2007).
This model has also been applied by De Boer et al. (2014) and Bellekom et al. (2012). The study
by Bellekom et al. (2012) assessed the role of BEVs in compensating the fluctuating production of
electricity from wind. It was concluded that charging BEVs at night contributes to a higher direct
consumption of wind energy. The importance of appropriate load management strategies to prevent
network problems was stressed. In De Boer et al. (2014), the effect of electricity storage options
(compressed air energy storage (CAES), Norwegian pumped hydro storage and PtG) to compensate
for wind energy fluctuations was simulated. The simulation heuristic did not always result in cost or
emission reductions at low wind penetrations. At high installed capacities of wind turbines, pumped
hydro storage was the most cost-effective technology to reduce electricity excesses and start-up costs
of powerplants.
The first study has investigated the system costs of residential heat supply (Van Melle et al., 2015).
The costs of energy distribution and local conversion to heat were analysed (not optimised) for different
scenarios. Central energy conversion was excluded. Demand data of a year with an extremely cold were
used to simulate high heat demands. The analysis has shown that most investments have to be made by
households, for insulation, heat pumps and micro-CHPs. Insulation plays an important role in reducing
the energy demand. The highest grid costs occur in full electric scenarios, in which case the electricity
grid has to be expanded.
The second study of the energy system in the Netherlands is about the optimal sustainable heat
provision (Schepers et al., 2015). The authors have divided all neighbourhoods in fifteen categories,
and determined the most cost-effective heat supply alternative for each category. The alternatives are
different combinations of energy generation, transport, and local conversion technologies. Instead of
performing simulations, the required installed capacity per household were estimated. The importance
of insulation is stressed here too. Although green gas is attractive for many locations, its availability
restricted. In urban areas, sources of geothermal and waste heat are exploited whenever available. For
rural areas, heat pumps are often the optimal choice.
While both studies are valuable because of the bottom-up approach of the energy demand, the
restricted optimisation ignores unexpected solutions. It might for example be attractive to equip some
households with electric boilers, that produce heat during cold days and/or oversupply of electricity.
This option is, however, not considered by both models. Another limitation is the limited scope: energy
demand by industry is excluded, as well as the demand of the transport sector. As discussed in § 4.1.2,
the different energy carriers cannot be properly studied in isolation, in particular in future energy systems.

4.3 Modelling decisions

To assess the economic and environmental effects of SOCs on the energy system, a system design
optimisation model will be used. This model calculates the optimal energy supply system for a given
future energy demand, and is based on a framework called oemof (oemof , 2016). The following

48
modelling decisions were made:

Objective: A cost objective is used, which means that the annualised costs of the total energy supply
system are minimised. GHG emissions are expressed in terms of costs by multiplication with an
emission tax to obtain a single objective function.

Scope: The national energy system of the Netherlands is considered. To get a good understanding of
the role of components with multiple inputs or outputs, not only the market for electricity, but
also for heat, hydrogen, and other fuels are included. Timesteps of one hour are used. No spatial
resolution smaller than the country is applied, so transportation limits are absent.

Restrictions: The final energy demand of all included energy carriers should be met. The only
exchanges with other countries are modelled as commodity markets for fossil fuels and biomass.
It is allowed to ‘dump’ excess electricity, biogas, hydrogen, or heat.

Model input: The set of available generation technologies consists of currently used technologies (like
fossil fuel powerplants and wind turbines), and some ‘new’ technologies like SOCs and electrolysers.
Furthermore, components for storage, and conversions of energy are included (e.g. batteries and
hydrogen tanks, electrolysers and heat pumps).

Model output: The calculations yield the amount of energy generating, converting, transporting, and
storing devices that should be installed to achieve the optimal situation. The throughput of each
component at each timestep is provided as well. From these data, detailed information on the
GHG emissions and costs of the system can be derived.

Scenarios: Multiple scenarios will be explored to find under what developments fuel cell systems are
favoured. The scenarios are constructed by making combinations of different demand compositions
and levels of GHG emission taxes.

Important elements of the model as mentioned above are schematically represented in Figure 4.3.
Each of the decisions is discussed and motivated further in Chapter 5.

Model input
Scenario
Demand
GHG tax Model output
profile
Optimal energy
system:
Available Optimization - Installed capacities
External data Scenario
technical algorithm - Operating times
components - GHG emissions
- Total system costs

Renewable Objective
production function
patterns

Figure 4.3: Structure of the energy system optimisation model, including inputs and outputs.

49
4.4 Model implementation
4.4.1 Software sources
The optimisation model implementation is based on the open-source modelling framework oemof
(oemof , 2016). The framework, written in Python and dependent on Pyomo, was extended with
additional functionalities to accommodate for SOFC and SOFEC components. The code is appended
in Appendix C.
The Center for Sustainable Energy Systems (ZNES) together with the Reiner Lemoine Institute (RLI)
in Berlin and the Otto-von-Guericke-University of Magdeburg (OVGU) have initiated the development
of an Open Energy System Modeling Framework (oemof). Their ambition is to create a free and open-
source software package that addresses the lack of transparency of many existing energy system models.
The fact that both the source code and a clear documentation are freely accessible, improvement,
debugging, and re-use of the software are stimulated (oemof , 2016). Oemof is under development,
and anyone can propose changes or additions to the framework via the online software sharing platform
GitHub.
Oemof is programmed in Python and uses several other Python packages. The most important
dependency is Pyomo, which supports formulating, solving, and analyzing optimisation models. To
solve an optimisation model, Pyomo passes the problem to a dedicated solver.
In this research, the GNU Linear Programming Kit (GLPK) solver was used for linear programming
problems (Makhorin, n.d.). IPOPT (an interior-point optimisation algorithm) was used as a solver for
non-linear problems (Wächter & Biegler, 2006). This software package depends on a number of libraries,
among which the Harwell Subroutine Library (HSL) (HSL, n.d.) and AMPL Solver Library (ASL) (Fourer
& Kernighan, 2002).

4.4.2 Energy system mapping

The system mapping that is applied in oemof, matches the concept of Figure 4.2. This concept dictates
the basic structure of any energy supply system within the framework.
The system consists of two types of entities: busses and components. A bus is an imaginary reservoir
of an energy carrier or resource. Components are entities that produce and/or consume energy carriers,
such as sinks, sources, and converters. Components are always connected to one or several buses, but
never to other components. The same holds for buses. To build an energy system, components and
buses are combined and connected. An energy system can be conceptualised as a bipartite directed
graph, in which all entities are interpreted as nodes and the directed edges are the in- and outputs
of the components. An energy system instance is described by the nodes and edges present, while a
possible solution is characterised by the weight of the edges (the energy flow) at each timestep.
In oemof, all entities are an instance of the class Entity or one of its subclasses. Figure 4.4 lists
the classes that are predefined in oemof. Most energy system components can be modelled as one of
these.

4.4.3 User input

The energy system optimisation in oemof is embedded in a sequence of steps, including model definition,
result processing, and everything in between. Most of these steps require the user to decide what input
is provided to the model, as indicated in Table 4.3. To clarify the process, a small code example is
provided below. This is the code used for Test 3 discussed in § 7.2.

# Import software libraries (not shown here)

# Define logger that allows to print status information

logger.define_logging()

# Read data from file and set time index

50
 Entity
+uid: string
+inputs: list
+outputs: list

Bus Component
+type: string -in_max: list
+price: float -out_max: list
+balanced: boolean -ub_out: list
-sum_out_limit: float -add_out_limit: float
+excess: boolean -capex: float
+shortage: boolean +lifetime: float
+excess_costs: float +wacc: float
+shortage_costs: float -crf: float
-opex_fix: float
-opex_var: float
-co2_var: float
-co2_cap: float

Source Transformer Transport Sink

+val: list -out_min: list
-in_min: list
-grad_pos: float
-grad_neg: float
DispatchSource -t_min_off: float Simple Simple
-t_min_on: float
-outages: float +eta: float
-input_costs: float -in_max: float
Commodity -out_max: float
-start_costs: float
-stop_costs: float
-ramp_costs: float
FixedSource -output_price: list
-eta_min: list

Simple CHP Storage

+eta: list +eta: list +cap_max: float
-cap_min: float
-add_cap_limit: float
VariableEfficiencyCHP +eta_in: float
+eta_out: float
TwoInputsOneOutput +cap_loss: float
+eta: list +c_rate_in: float
SimpleExtractionCHP +c_rate_out: float
+f: float
+eta_el_cond: float
+beta: float
+sigma: float

Figure 4.4: The Entity class, its subclasses, and their attributes as available in oemof. Attributes
indicated with a plus sign (+) must be specified when an instance of the class is created. Others
are optional attributes that are ignored or set to zero when not provided.

filename = ospath.normpath("/path/to/data/file.csv")
data = pd.read_csv(filename, sep="\t")
time_index = pd.date_range(‘1/1/2015’, periods=8760, freq=‘H’)

# Initialize the energy system and specify solver and objective

simulation = es.Simulation(
timesteps=range(len(time_index)), verbose=True, solver=‘glpk’,

51
 Table 4.3: User input required for each step of the modelling process.

Model step User input

1. Read energy production and consumption Specify data file(s)
patterns
2. Initialise the energy system Specify the solver algorithm, objective function,
and logging options
3. Create all entities that are part of the en- List the energy carrier buses, and whether excess
ergy system or shortage is allowed
List all components and their characteristics
(see Figure 4.4)
Specify which components’ installed capacities
can be increased
4. Calculate the optimal solution of the prob- –
lem
5. Process the results Functions to visualise, print, or save the results

objective_options={‘function’: ghg_objectives.minimize_cost, ‘co2_tax’: 0})

energysystem = es.EnergySystem(time_idx=time_index, simulation=simulation,

display=True)

# Set some or all components to investment mode

source.FixedSource.optimization_options.update({‘investment’: True})
transformer.Storage.optimization_options.update({‘investment’: True})

# Define entities and their characteristics

# Energy carriers
bele = Bus(uid="bele", type="el", excess=True)

# Sources
pv2 = source.FixedSource(uid="PV_comm", outputs=[bele], out_max=[0],
add_out_limit=29757, capex=660, opex_fix=6.591, lifetime=40,
wacc=0.08, co2_cap=1793.75e-3, val=data[‘solar’])

# Storage
battery = transformer.Storage(uid=‘Li_battery’, inputs=[bele],
outputs=[bele], eta_in=1, eta_out=0.85, cap_loss=6.94e-5,
opex_var=0.001, capex=50, lifetime=10, cap_max=1,
add_cap_limit=10e10, c_rate_in=1, c_rate_out=1)

# Demand
ele_dem = sink.Simple(uid="ele_dem_hh", inputs=[bele], val=data[‘el_dem_hh’]*3)

# Optimise the energy system

om = OptimizationModel(energysystem=energysystem)
energysystem.optimize(om)
logging.info(‘Finished!’)

# Plot and save the results

stackplots(energysystem.results, time_index, bele)
store_results (energysystem.results, time_index, "ResultsFile")

52
4.4.4 Implementation of SOC components
As depicted in Figure 4.4, each set of components that has different input-output relations is modelled by
a unique sub-class of Component. SOFCs and SOFECs each have their own sub-class. These classes do
not have additional parameters. The constraints that apply to instances of this class were implemented
in accordance with § 3.5. The code that defines the SOC classes and the corresponding constraints are
included in Appendix C.

4.5 Levelised costs of energy

Levelised cost of energy (LCOE) analysis is a rapid method to compare several options of energy
production. The idea is that all costs involved in energy production using a technology – investment
costs, fuel costs, fixed and variable operation and management (O&M) costs, and optionally taxes –
are expressed in terms of costs per unit of energy (see Eq. 4.1). This approach ensures that technology
options with very different cost structures can be compared on an equal basis, e.g. €/kWhe .
An important aspect of a LCOE calculation is the discounting of investment costs. In economic
analyses, the investment costs are annualised by multiplication with the capital recovery factor (CRF).
The CRF accounts for the fact that the investment costs are discounted to the future. It is calculated
by taking into consideration the expected lifetime of the technology and the discount rate r (Eq. 4.2).
The value of the discount rate is assumed equal to the weighted average cost of capital (WACC), which
is the interest rate for loans or the desired interest for investors (Bruckner et al., 2014). In general,
lower interest rates apply for conventional technologies, due to the higher reliability and demonstrated
performance. New and unconventional energy transformers, however, can guarantee the profitability to
a lesser extent, resulting in higher interest rates. In the present analysis, a WACC of 5% for proven, and
8% for unproven technologies is used.
The definition of LCOE is (Bruckner et al., 2014):

CRF · Cinv + Cfix Cfuel

LCOE = + Cvar + + CGHG (4.1)
8.76 · CU η
r r · (1 + r)L
with CRF = = (4.2)
1 − (1 + r)−L (1 + r)L − 1

and with Cinv , Cvar , and Cfix the investment costs, variable and fixed O&M costs, CU the capacity
utilisation, and L the lifetime of the technology.
The most used indicator is the levelised cost of electricity, but the levelised cost of hydrogen or any
other energy carrier can be calculated too (Chen, 2006). Furthermore, the levelised cost of electricity
storage can and has been assessed (Lazard, 2015b). The simplicity of LCOE analysis entails some
limitations. To calculate the variable costs, a CU has to be assumed. Therefore, LCOE analysis is most
useful to analyse a single technology in isolation from the rest of the energy system, or to evaluate
different options for a specific application.

53
Chapter 5

Assumptions & Scenarios

5.1 Goal & Scope definition

5.1.1 Goal
The goal of the optimisation model is to find a combination of technologies that most cost-effectively
meets the final energy demand throughout the year. The model should cover all relevant energy car-
riers, and be able to include any type of existing or future energy conversion, storage, and transport
technologies. The model is intended to be used to study SOCs specifically. It is applied in Chapter 7
to see whether fuel cells can make the Dutch energy sytem more efficient, and what the effects of fuel
cells on various aspects of the system are.

5.1.2 Timesteps
The dynamics of each energy carrier emerge at various temporal scales. For example, coal has to be
ordered weeks or months before it arrives at a powerplant, while a cloud affects the electricity output of a
solar panel within a minute. Since the model should cover all energy carriers, a timestep size in between
these extremes is best. Therefore, the model consists of 8760 timesteps, i.e. one year divided in steps
of one hour. This choice allows to include the effects of both short-term and long-term fluctuations.
A model with only eight representative days as in (as in Shahmohammadi et al., 2015) would greatly
reduce the computation time, but is not suitable to guarantee proper long-term performance of the
energy system. Considering the balancing function of energy systems (see § 4.1.1), the balancing is
not investigated at all timescales. Electricity grid balancing on the millisecond to minute scale is not
considered. For other energy carriers, the dynamics that play a role in timescales shorter than hours are
hardly important for the operational planning of the system.
No specific reference year is defined. But as explained in § 5.4, scenarios with an energy demand
that largely differs from the current situation are included.
The demand and supply patterns used are intended to represent an average year. Thus, no conclu-
sions can be drawn about the security of supply under extreme conditions, such as very cold periods
(which was investigated by Van Melle et al., 2015). On the other hand, a full year features a broad
range of weather conditions and combinations of solar and wind availabilities.

5.1.3 Constraints
A number of constraints is applied to the solution, such that it meets the requirements of being tech-
nically and thermodynamically possible. Constraints are associated with each entity.

5.1.4 Objectives
Two objectives are distinguished: cost minimisation and GHG emission minimisation. The former
objective has an economic focus, but can also include environmental aspects by means of a weighting

54
factor for GHG emissions. This weighting factor translates emissions to costs in order to obtain a single
objective value. In this thesis, it will be referred to as emission tax. Yet, it can also be interpreted as the
price that has to be paid on the European market for emission certificates within the emission trading
system (ETS), or as an expression of societal costs associated with climate change.
The total annualised costs are a sum of investments, fixed and variable O&M, curtailment costs of
dispatchable sources, costs for an excess or shortage of energy.
T X add
Investment costs: Cinv = IC e · CRFe · Cinv,e ∀e (5.1a)
8760 e
T X
Fixed O&M costs: Cfix = (IC e + IC add
e ) · Cfix,e ∀e (5.1b)
8760 e
Variable O&M costs: Cvar = Outpute (t) · Cvar,e (5.1c)
XX
∀e, t
e t
Curtailment costs: Ccur = Curtailmente (t) · Ccur,e (5.1d)
XX
∀e, t
e t
Excess costs: Cexc = Excesse (t) · Cexc,e (5.1e)
XX
∀e, t
e t
Shortage costs: Csho = Shortagee (t) · Csho,e (5.1f)
XX
∀e, t
e t
" #
GHG costs: CGHG = taxGHG · Einv,e + Outpute (t) · Evar,e (5.1g)
X X
IC add
e ∀e, t
e t
8760
Annual costs: Ctot = · (Cinv + Cfix + Cvar + Ccur + Cexc + Csho + CGHG ) (5.1h)
T
For storage components, the IC add factor is expressed in units of storage capacity.
Similarly, but with less terms, the GHG emissions can be calculated by summing the emissions during
construction and operation of components:

Construction: (5.2)
X
IC add
e · Einv,e ∀e
e
Operation: Outpute (t) · Evar,e (5.3)
XX
∀e, t
e t

5.2 Energy conversions

The energy system can be conceptualised as a series of energy conversions, as illustrated by Figure 5.1.
Inputs of the system are forms of primary energy: solar, nuclear, and chemical energy. The purpose
of the system is to provide energy in a useful form (‘useful energy’), such as light, sound, heating,
cooling, and work. The energy carrier that is bought by consumers (‘final energy’) often undergoes a
final energy conversion to obtain useful energy. The end-use of energy by society can be characterised
as the demand for either final or useful energy. In the first case, the demand for specific energy carriers
should be known. In the second case, the demand for services should be known, including the efficiencies
of the appliances that provide these services (Deng et al., 2012).

Final energy Useful energy

Wind Gaseous, solid, or liquid fuel Heating

Sun Electricity Cooling
Nuclear energy Hydrogen Light
Geothermal heat Steam Work
Fossil fuel

Figure 5.1: Energy flows through an energy system, including some examples of carriers.

55
 Determining the demand for useful energy can be a challenge, since statistics offices only record final
energy demand. If the final conversion options were included in the boundaries of the energy system,
demand would be optimised too. Implicitly, ideal consumer behaviour is assumed in this manner, because
consumers are ‘allowed’ to only use the devices that are most effective according to the optimisation
objective. Although companies and the government can to some extent stimulate desirable consumption
patterns, full consumer control would be an unrealistic assumption. From a computational point of view,
the final conversion options would increase the number of variables of the model, thereby increasing
the calculation time. Based on these considerations, the approach used here is to specify final energy
demand. Instead of optimizing the final conversion, leading to one optimal consumption pattern, multiple
possible consumption scenarios will be used (see § 5.4).
Throughout this thesis, amounts and flows of energy carriers will be quantified in terms of their
energy content. For fuels, the higher heating value (HHV) is used. See Appendix A.1 for the conversion
factors to calculate the mass of a given fuel.

5.3 Current energy demand

Before constructing scenarios of future energy demand, it is instructive to first analyse the current
consumption of energy carriers in the Netherlands. Official statistics for the year 2014 are used here
(CBS, 2016b). The national energy balance maps all energy-related flows. The final energy demand
per sector is the relevant part of the balance in this analysis. Energy consumption by the energy sector
itself and transmission losses are not included. Some fuels are used for applications other than energy
production, such as the synthesis of plastics and other chemicals from oil. These applications are
categorised as non-energy use. The data are visualised in Figure 5.2.

Figure 5.2: Energy demand of the Netherlands in 2014, grouped by energy carrier (left) and final
energy demand (right).

From Figure 5.2, a number of conclusions can be drawn. Oil and oil products represent the biggest

56
fraction of final energy demand, followed by NG. Electricity also has a big share of the demand. Because
of high conversion losses in power plants, the share of primary energy used for electricity production is
larger than the share in final energy.

5.3.1 Electricity consumption

Electricity is used predominantly for lighting and electric appliances, but some of it is used to produce
heat. Electric boilers, cooking stoves, microwaves, and heat pumps can perform this conversion. For
instance, around 1.75 GJ/a electricity was used for cooking in 2012 (Gerdes et al., 2014). 15% of
electricity consumed by households is used for heating: 3% for cooking, 7% for space heating, and 5%
for hot tap water) (Gerdes et al., 2014).

5.3.2 Heat demand

Heat demand is split into two categories: low-temperature (<100 ºC) and high-temperature (>100 ºC)
heat demand. It is assumed that households and businesses only consume low-temperature heat, and
industries only high-temperature heat. Concluding from the heat demand per temperature interval as
shown in Figure 5.3, this approach slightly overestimates the high-temperature heat demand.

700
600
Heat demand (PJ/a)

500
>1000ºC
400
750-1000ºC
300 500-750ºC
200 250-500ºC
100-250ºC
100 <100ºC
0
Industry Business and
consumers

Figure 5.3: Heat demand by the residential and industrial sector in the Netherlands in 2006, divided
in temperature intervals (S. Spoelstra, 2008).

5.3.3 Hydrogen consumption

Hydrogen does not occur in the energy statistics. This does not mean that it is not produced or
consumed. In fact, large quantities of hydrogen are consumed by the ammonia and petrochemical
industry (Sveshnikova, 2015).
Ammonia (NH3 ) is produced from H2 and N2 in the Haber-Bosch process. The total production of
ammonia in the Netherlands amounted to 2.6 Mt in 2014. Most of this is used for fertilizer production
(Beljaars, 2015). The required hydrogen gas originates from NG; 91 PJ of NG was consumed by Dutch
ammonia plants in 2014. Two thirds of this (61 PJ) is directly needed to produce hydrogen, while the
remainder is combusted to provide heat to drive the endothermic reaction (Beljaars, 2015). 44 PJ of
hydrogen would be sufficient to replace the methane feed. An additional amount of 17 PJ heat input is
then required to compensate for the reduced methane input.
In the petrochemical industry, hydrogen serves as a reactant for hydrocracking, a process used in
the production of oil-based fuels (Citizendium, 2013). However, this application falls outside the scope
of the analysis, since it can be classified as energy use by the energy sector. Therefore it is not included
as final demand, leaving the hydrogen demand calculated above as the single current demand.

57
5.3.4 Natural gas consumption
The composition of NG is assumed to be that of Groningen gas, which has a low caloric value. This
is the mixture that is delivered by the gas distribution network of the Netherlands. The composition is
given in Table 5.1.

Table 5.1: Average composition of gas from the Groningen gas field (Wikipedia, 2012).

Component volume% mol% mass%

Methane 81.30 81.29 69.97
Ethane 2.85 2.87 4.63
Propane 0.37 0.38 0.90
Butane 0.14 0.15 0.47
Pentane 0.04 0.04 0.16
Hexane 0.05 0.05 0.23
Nitrogen 14.35 14.32 21.52
Oxygen 0.01 0.01 0.02
Carbon dioxide 0.89 0.89 2.10

5.4 Demand Scenarios

Building on the insights from § 5.3, this section discusses what factors influence the energy demand,
and how to make scenarios based on these factors.

5.4.1 Possible developments

The final energy demand Dfinal,c of an energy carrier depends on the magnitude of a useful energy
demand Duseful and on the fraction yc of that demand derived from the carrier:

Dfinal,c ∝ Duseful · yc (5.4)

Both aspects are subjected to change over time. For example, the demand for rail transport as a form
of useful energy has grown (more passenger kilometres are consumed), while the form of final energy
has changed from coal to electricity. Both developments influence the demand for the energy carriers
coal to electricity. There mare many of such developments taking place, or expected to take off in the
future. By considering these developments, scenarios for future energy demand can be constructed.
A good starting point is provided by the six useful energy categories depicted in Figure 5.2. Below,
the available energy carrier options to meet each demand are discussed briefly, providing guidance with
respect to the values of yc .

Electric appliances
Currently, electricity constitutes about 15% of the final energy demand (see Figure 5.2). It is consumed
in all sectors of industry, business, and households. The majority of electricity is used to operate all
kinds of appliances, such as computers, machines, lights, etc. In these applications, electricity cannot
be replaced by other energy carriers.

Low-temperature heat
Households and business consume a considerable amount of NG to produce low-temperature heat. Some
heat is derived from oil and biomass combustion in stoves, electric heaters, and district heating systems.
Low-temperature heat has three main purposes: space heating, hot tap water delivery, and cooking. In

58
the future, NG could be replaced by electricity in all these applications. Electric or induction cooking
stoves, microwaves, heat pumps, and electric boilers can be installed in such a scenario. Other options
for future heating include direct heat from solar thermal panels, the expansion of district heating systems,
or stoves fired with wood. The attractiveness of these options depend on the specific local situation
(Schepers et al., 2015).

Aviation and shipping

The transport sector is dominated by several types of fossil fuels as energy carrier. Oil products are used
to propel air planes, ships, trucks, cars, and trains. Aviation and shipping are grouped in a separate
category, because there are very few commercially attractive alternative energy carriers available. This
is because it is hard to approach the energy density of kerosene and fuel oil currently used for these
modes of long-distance transport., and it is expected that these fuels will not be replaced in the coming
decades.

Road and rail transport

Road traffic is mostly relying on fossil energy carriers, whereas rail traffic is to a large extent electrified in
the Netherlands. can be electrified. Especially BEVs have the potential to drastically increase electricity
demand. Hydrogen is another possible future energy carrier for transport. Since it does not (yet)
function as an energy carrier, hydrogen is not included in official energy statistics. This can change if
the technology of FCEVs spreads.

High-temperature heat
The final energy demand by industry (excluding the energy industry) amounted to 1108 PJ in 2014
(about one third of the total). Around 415 PJ is used to produce heat for industrial processes. The
remainder of energy carriers are for electric appliances and non-energy uses. Figure 5.3 illustrates that
most heat is high-temperature heat, as opposed to the low-temperature heat demands discussed earlier.
A few options exist to obtain high-temperature heat: one can burn coal, oil, gas, or biomass in a furnace,
use concentrated solar power (CSP), or connect to an industrial district heating system. CSP is not as
attractive as in countries in the south, because the efficiency drops sharply when no direct sunlight is
available (MacKay, 2009).

Non-energy uses
Half of the final energy demand by industry is not needed for energy purposes, but to make products.
Most importantly, plastics and polymers are derived from oil. The synthesis of NH3 is another example.
And coal can be used as reducing agent in some processes.

Translation to scenarios
To keep the number of scenarios low, only developments with a big impact on the demand are considered
here. Out of the six useful energy categories discussed above, three are regarded as most relevant here:
low-temperature heat, high-temperature heat, and land transport. The mix of final energy forms used
for these demands is much more flexible than that of the other three.
The magnitude of all six demand categories (Duseful in Eq. 5.4) is likely to change over time too. On
the one hand, the globally growing population and affluence are drivers for increasing energy consump-
tion: more goods are manufactured and transported, longer distances are travelled, more buildings have
to be heated, etc. On the other hand, higher technological efficiencies, better insulation, and demateri-
alisation of the economy can induce lower energy demands (Chertow, 2000). For the present study, the
magnitude of the demand is of lesser importance. A higher or lower total demand presumably results
in accordingly scaled installed capacities of technologies. Therefore, no variations in the magnitude of
demand are included in the scenarios.

59
5.4.2 Efficiency factors
To calculate the final demand for an energy carrier starting from the fraction and magnitude of a useful
energy demand, the efficiency of the final conversion needs to be taken into account. For example, to
drive a given distance, a BEV requires less electricity than a gasoline car requires gasoline when both
carriers are expressed as energetic values. Therefore, Eq. 5.4 can be extended, yielding:

Dfinal,c = Duseful · yc · ηc (5.5)

In Eq. 5.5, ηc is the efficiency factor that applies to the conversion of energy carrier c to a given form
of useful energy. The assumed efficiencies are tabulated in Table 5.2. Using Eq. 5.5, it is possible to
calculate the useful energy demand in the current situation. What is more, the final energy demand for
a given energy carrier can also be calculated.

Table 5.2: Efficiency factors for the delivery of three useful energy forms.

Heat <100 ºC kJ/kJ Heat >100 ºC kJ/kJ Land transport kJ/km

Gas; condensing boiler 0.96 Oil; industrial furnace 0.8 Gasoline/diesel; ICE 2320
Electric; heat pump 4 Coal; industrial furnace 0.85 BEV 576
Hot water; district heat 1 Gas; industrial furnace 0.75 FCEV 130
Steam 1

The remainder of this section aims to explain the derivation of the efficiency factors from Table 5.2.

Low-temperature heat can be derived from three energy carriers: (natural) gas, electricity, and
hot water from a district heating system. For simplicity, only space heating is considered, since it
provides the largest heat demand from households and business.
Small boilers cannot convey all heat of combustion to the water that has to be heated. Part of the
energy is lost with the exhaust gases. The highest performance is achieved when condensing the steam
from the exhaust. By doing so, a residential gas-fired condensing boiler can have an efficiency of up to
96% of the HHV (Wikipedia, 2016b). This corresponds to 104% of the lower heating value (LHV).
Two types of heat pumps exist: ground source and air source heat pumps (De Swardt & Meyer,
2001). Although heat pumps can also cool a house, here only their heating capacity is considered.
Ground source heat pumps have a coefficient of performance (CoP) of 3.2, whereas air source heat
pumps operate at a CoP of 2.6 (Mathiesen et al., 2009). The Dutch heat pump association even
mentions a CoP of 4, with the expectation of 5 to 6 in the near future (DHPA, 2010). Therefore, heat
pumps are assumed to have an average CoP of 4, which would be a heat pump with energy class A/A+
(Daikin UK, 2012).

High-temperature heat is assumed to be produced by industrial boilers. Since the technologies

that are currently state-of-the-art will in the future become mainstream, the efficiency of new boilers is
used as efficiency factor. The data are provided in Table 5.3.

Table 5.3: New industrial boiler efficiencies (Van Wortswinkel & Nijs, 2010).

Coal Oil Gas Biomass

Full load efficiency 85% 80% 75% 70%
Low load efficiency 75% 72% 70% 60%

Land transport efficiencies are based on the fuel consumption of different passenger vehicles, see
Table 5.4. It is assumed that the relative efficiencies are comparable for rail transport.

60
Table 5.4: Comparison of fuel economies of six vehicle types (Verbeek et al., 2014). LPG: liquefied
petroleum gas; CNG: Compressed NG.

Vehicle Fuel Unit Energy density Fuel economy

(X) (MJ/X) (X/100 km) (MJ/100 km)
Conventional Gasoline L 34.8 6.67 232
Diesel L 38.6 5.48 212
LPG L 33.1 8.89 295
CNG kg 55 6.67 367
BEV Electricity kWhe 3.6 16 57.6
FCEV Hydrogen kg 130 1.0 130

5.4.3 Scenarios quantified

To cover the range of directions of demand development introduced in § 5.4.1, four scenarios are
constructed:

• Business as usual (BAU),

• Diversity,
• Full electric, and
• High hydrogen.

These scenarios differ in how the demand for each of the three ‘variable’ useful energy demand categories
is met. Table 5.5 gives a quantitative definition of each scenario, and the figures are plotted in Figure 5.4.
The calculations are made using Eq. 5.5. Further explanation of each scenario will now be given.

Table 5.5: The definition of four energy demand scenarios in terms of the demand (in PJ/a) for
each energy carrier. The amount of useful energy demand is specified in the leftmost column.

Useful energy Scenario Coal Hydrogen Heat Electricity Gas Oil

Low-temperature heat BAU 68.7 2.6 431
528.7 Diversity 176 44.1 169
Full electric 0 132.2 0
High Hydrogen 529 0 0
High-temperature heat BAU 17.8 115 91.4 72.4
348.1 Diversity 74.0 87.0 65.3 69.6
Full electric 17.8 115 91.4 72.4
High Hydrogen 17.8 115 91.4 72.4
Land transport BAU 0 1.2 434
438.5 Diversity 79.9 41.5 146
Full electric 0 124.6 0
High Hydrogen 239.6 0 0
Electric appliances All 346.5
Aviation & shipping All 706
Non-energy uses All 1.7 44.2 38.5 486

In BAU, the final energy mix as reported for 2014 is reproduced. This scenario therefore represents
a situation where very few changes in energy demand have taken place. The Full electric scenario is
characterised by a high degree of electrification. All low-temperature heat is derived from electricity,
and all road and rail transport is fully electrified too. High hydrogen is a scenario that features a high
demand for hydrogen, with all land transport equipped with fuel cells for propulsion. Both low-and
high-temperature heat demands ask for steam or hot water. This is expected to provide a sink for heat
produced by SOCs. To complete the list, the Diversity scenario represents a situation in which each

61
Annual demand (PJ) 700
600
500 Aviation,
400 shipping,
non-energy
300
Transport
200
Households
100 (Heat pumps)
0 Households
BAU

BAU

BAU

BAU

BAU

BAU

BAU
High hydrogen

High hydrogen

High hydrogen

High hydrogen

High hydrogen

High hydrogen

High hydrogen
Full electric

Full electric

Full electric

Full electric

Full electric

Full electric

Full electric
Diversity

Diversity

Diversity

Diversity

Diversity

Diversity

Diversity
Industry

Coal Electricity Natural gas LowT heat HighT heat Hydrogen Liquid fuel

Figure 5.4: Energy demand per scenario by each sector. The liquid fuel demand of aviation,
transport and non-energy uses adds up to 1192.2 PJ in all scenarios, but is not fully plotted. The
numbers belonging to this figure are displayed in Table 5.5.

useful energy demand is met by a balanced mix of energy carriers. This means that equal fractions yc
of each useful energy category are derived from the respective carrier options listed in Table 5.2.

62
5.5 (Environmental) constraints to energy production
In the present analysis, economic costs are used as indicator. Costs are important, because energy
carriers constitute a significant part of the economy. Therefore, the consequences of changes in the
energy supply system can be profound. Besides, providing affordable energy is one of the desired energy
system properties. By minimising the energy system costs, the extent of competitiveness of emerging
technologies is demonstrated. Yet there are many other aspects impacts related to energy supply.
Air pollution and climate change are arguably the most important impacts of fossil-fuel based energy.
For renewable technologies, the land use impact can be a limiting factor. Resource availability can
constrain large-scale application of some technologies (Kleijn & Van der Voet, 2010). Resources and
their distribution and scarcity also have social and geopolitical effects. Consequently, the choice for
specific energy sources is a trade-off between these impacts. Depending on the prioritisation of each
impact, a different set of technologies is preferred.
It falls outside the scope of this research to collect data on all impacts of all technologies and
develop a prioritisation method. Therefore, air pollutant emissions and resource constraints are not
considered. All GHG emissions are taken into account. Whenever data was available, the emission of
CO2 , CH4 , and N2 O during combustion or construction processes were assigned. Table 5.6 presents
the conversion factors applied to each GHG. To express these emissions in monetary terms, the total
amount is multiplied by a price per tonne CO2 -equivalent (tCO2 -eq.).

Table 5.6: Three important greenhouse gases and their conversion factors (IPCC, 2007).

CO2 CH4 N2 O
Conversion factor (tCO2 -eq./t) 1 25 298

In the case of biomass, it is highly uncertain what amount can be made available for energy uses.
Dedicated energy crop farming faces competition with food production. Organic residues from forestry
and agriculture are usually cheaper, but are a much more diffuse source of biomass. The extent to
which imports can fulfil the demand for biomass is another source of uncertainty. The amount of
biomass available for the Netherlands under different assumptions of the global production potential
and allocation method are shown in Table 5.7 (PBL, 2011). The lower end estimate of 470 PJ per year

Table 5.7: Annual availability of biomass for energy production in the Netherlands, based on
different allocation methods (PBL, 2011).

Global production
Allocation method 150 EJ 400 EJ
Based on population 284 PJ 760 PJ
Based on GDP 735 PJ 1960 PJ
50% of global trade based on GDP
plus 50% of domestic production 470 PJ 1080 PJ

is used. The biomass supply is assumed to consist of equal amounts of residues and energy crops, which
in in agreement with Lamers et al. (2015).
In principle, any form of biomass can be converted to biogas or green gas. Several techniques
are under development to perform the gasification process (IEA-ETSAP & IRENA, 2013a). Given
the limited amount of biomass, it makes sense to only use wet biomass streams, that are suitable for
gasification but not applicable for other uses such as biodiesel production (Van Melle et al., 2015).
Anaerobic digestion is typically used to convert moist biomass to biogas. A typical composition of the
combustible gas mixture is 50–70% CH4 , 20–35% CO2 , 0–10% H2 O, with a dependence on the source
(D’Appolonia, 2016). It is thus assumed that biogas is only derived from wet biomass streams such as
manure and waste-water treatment sludge (Bergsma & Croezen, 2011). Landfills and food processing
industries might also be a source of wet biomass (D’Appolonia, 2016). Imports are not considered,
given that transport is costly because of the high water content of suitable streams.

63
 The estimated future potential for biogas production of three different sources is compared in Ta-
ble 5.8. The range of 20–77 PJ is rather broad, indicating that this source of biomass involves a high
uncertainty. Much like the biomass residues and products described above, the potential that can be
realised depends on the development of e.g. collection systems. 48 PJ is used as the limit for annual
biogas production.

Table 5.8: Potential annual biogas production according to different sources. a

Source PJ bcm b NG equivalents

Bergsma and Croezen (2011) 30–48 1–1.5
Van Melle et al. (2015) 53 1.5
Schepers et al. (2015) 77 2.2
D’Appolonia (2016) 20 0.56 (0.92 bcm)
a Numbers in italics are calculated using an energy density of 35.17 MJ/m3 natural gas equivalent.
b Billion cubic meter. 1 bcm = 109 m3

The boundaries for the installed capacity of biomass sources, wind turbines, and solar thermal and
photovoltaic (PV) cells are summarised in 5.9.

Table 5.9: Boundary conditions applied to the scale of implementation of various renewable energy
technologies.

Energy source Production limit a Source

(PJ/a) (GW)
Biogas 48 1.52 (Bergsma & Croezen, 2011)
Biomass residues 235 7.45 (PBL, 2011)
Biomass products 235 7.45 (PBL, 2011)
Onshore wind 8 (PBL, 2011)
Offshore wind 34 (PBL, 2011)
Residential PV 52.5 (Veenstra, 2015)
Commercial PV 29.8 (Veenstra, 2015)
Solar thermal b 260 (PBL, 2011)
a Numbers in italics are calculated assuming a constant production.
b 400 km2 · 65% · 1 kW/m2

5.6 Supply patterns

5.6.1 Fossil fuels
It is expected that the three main fossil fuels that are used nowadays (natural gas, oil, and coal) will still
be available in the future. What is highly uncertain is the price at which these goods can be purchased
on the international energy market.
The Netherlands traditionally has a strongly developed gas infrastructure. The majority of the
resource is extracted from gas fields in Groningen and under the North Sea. After 2025, a sharp
decrease in domestic extraction is expected, because the reserves are almost depleted around that time
(Schoots & Hammingh, 2015). Developments in technology can possibly allow extraction after that
time, but with much lower rates. Consequently, a large dependence on imports seems inevitable for NG.
Imported gas is expected to be more expensive, partly because of the considerable energy losses during
transport.
For oil, the Netherlands is dependent on imports already. The price of oil on the international market
thus determines the economic attractiveness of using this fuel. Many factors influence the oil price, and
these complex interactions have resulted in unexpected price fluctuations in the past. In the future,

64
prices can rise as a consequence of growing demand or depletion of easily accessible reserves. If a rapid
shift towards other energy carriers is made, prices can drop because of overproduction.
Like oil, coal is also traded on a global market. Long-term predictions assume a modest growth in
the price levels (Schoots & Hammingh, 2015). Since the coal reserves are large and no unconventional
technologies are needed to extract it, the supply risks are limited. Combustion of coal yields the highest
emission of CO2 per Joule, so GHG taxation policies will have the strongest impact on coal use.
Gas, oil, and coal are commodities, and it is assumed that they are available whenever there is
a demand. Because of the assumption of a perfect match between demand and supply, storage is
neglected. In reality, there are a number of storage facilities along the fuel supply chain. However, the
costs of and losses during storage are negligible.

5.6.2 Renewable sources

For most renewable energy sources, the supply pattern is important for two reasons. Firstly, the amount
of a renewable energy source available for conversion depends on external factors. For example biomass
and solar radiation have predictable, seasonal variations in availability, while wind speed and cloudiness
introduce more irregular fluctuations in the yield of wind turbines and PV cells. Secondly, a majority of
renewables generate electricity, which cannot be readily stored. At the same time, a tight balance has
to be maintained between supply and demand of electricity at all times.

Wind
The output capacity of wind turbines can be related to the wind speed. Therefore, wind speed data for
two locations were retrieved from the Royal Netherlands Meteorological Institute (KNMI, 2005). Wind
measurement station K13 (on the North Sea) is used for offshore turbines, while measurement station
Lelystad will provide wind profiles for onshore turbines. For both locations, data from the year 2004 are
used. The reported data are adjusted wind speeds at a height of 10 m. To calculate wind speeds at the
height of a wind turbine, the wind power law is used (MacKay, 2009, p. 266):
v log(h/l)
= (5.6)
v0 log(h0 /l)
With v and v0 the wind speeds at height h and h0 , respectively. Parameter l is the roughness length, and
depends on the roughness of the surrounding land. For Lelystad l = 0.03 m, and for K13 l = 0.002 m
(KNMI, 2005).
The capacity factor (CF) of a wind turbine is the ratio of delivered power to the rated maximum
power output. To obtain the capacity factor at a given wind speed, the following relation is used:
3
P v

CF = = (5.7)
Pmax vrated
The rated output speed (the wind speed at which the power is rated; vrated ) for all wind turbines in
the model is 14 m/s (Bradbury, 2009). This means that the power production will not increase further
above that wind speed. The cut-in speed is set at 3.5 m/s. The cut-out speed (where the turbine is
shut down for security reasons) of 25 m/s is not reached in the dataset, so the drop of all power above
that speed is not modelled. These assumptions result in the capacity curve shown in Figure 5.5. When
the KNMI data are combined with Eq. 5.7, an example of the resulting pattern is shown in Figure 5.6.
In the Netherlands, the yearly averaged capacity factors were 21.5% in the period 2003–2007
(Boccard, 2009). For offshore installations, it can reach values as high as 25–30%. The approach
described above yields a capacity factor of 21.3% onshore and 46.5% offshore. Therefore, the offshore
wind production could be overestimated.

Solar
From Elia, the Belgian transmission system operator (TSO), data about the power generated by PV cells
during 2015 were obtained (Marsboom, 2013). The dataset represents all PV installations in Belgium

65
 14 0.9

12
} Lelystad Onshore
K13 (Northsea) Offshore } 0.8

Capacity factor (MW/MW)

1.2
Rated power Cut-out
point speed 0.7
1.0 10

Wind speed (m/s)

0.6
Capacity factor

0.8 8 0.5
0.6 0.4
6
0.4 0.3
4
Cut-in 0.2
0.2 speed
2
0.1
0
0 5 10 15 20 25 0 0
Wind speed (m/s) June 1 June 2 June 3

Figure 5.5: Capacity curve for wind Figure 5.6: The wind speed patterns for three consecutive days,
turbines. and the resulting capacity factors.

that are registered at Elia, and is considered to be representative for the pattern in the Netherlands. It
is applied to both PV and solar thermal panels.

2.5 January
February
2 March
Average solar power (MW)

April
May
1.5
June
July
1 August
September
0.5 October
November
December
0
0 4 8 12 16 20 24
Time of the day

Figure 5.7: Average electricity production by PV for each month in 2015. Data from Marsboom
(2013).

Biogas
The availability of biomass suitable for gasification (see § 5.5) is subjected to seasonal variations. The
bio-organisms responsible for methane production in digestion tanks cannot be controlled precisely.
Therefore, the output of biogas exhibits irregularities. Very few data on the production is made public.
Here, the pattern of a waste-water treatment plant in Denver is used, see Figure 5.8 (Trendewicz &
Braun, 2013). The yearly pattern is produced by repeating the patterns for a week in summer and in
winter 26 times each. The pattern is scaled to ensure that the monthly produced amounts match. The
monthly average production has a variability of less than 10% of the output, which is smaller than the
weekly variations.

66
 (a) Weekly variations during winter and summer. (b) Monthly variations.

Figure 5.8: Biogas production of a waste-water treatment plant in Denver. Reprinted from
Trendewicz and Braun (2013).

5.7 Demand patterns

For each of the demand categories and energy carriers (see § 5.4), a pattern for demand will be specified.
No demand-response mechanisms are included. The patterns described in this section serve as input
to the optimisation model (see Figure 4.3). To facilitate scaling of the data to a given annual energy
demand (in PJ), the patterns are normalised to an annual total of 1 PJ.

5.7.1 Electricity
Electricity consumption shows a rather regular pattern, with peaks in the morning and afternoon, and
troughs during night-time. The data for the full year 2015 are obtained from TenneT (TenneT, 2016).
It is assumed that businesses and households are responsible for the fluctuations, whereas the industry
and trains provide a constant load. To extract the load profile of the variable demand, 3973 MW is
subtracted from the total load on each hour of the day. The 3973 MW correspond to the averaged
electricity demand of the industrial and transport sector.

5.7.2 Low-temperature heat

The demand for low-temperature heat is currently fulfilled mostly by NG. No hourly consumption data
for gas were found. Therefore, the hourly demand pattern is assumed to be identical to that of electricity.
Since gas demand is highly dependent on the season (E.ON, 2014)(see Figure 5.9), a correction is made
for the monthly average demand.

18%
15%
12%
9%
6%
3%
0%
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Figure 5.9: The average gas demand per month in the Netherlands (E.ON, 2014).

67
5.7.3 Land transport
The demand for energy in transport consists mostly of privately owned cars. The charging or refuelling
behaviour of car owners is very dependent on the type of fuel. Conventional combustion engine cars
are exclusively refuelled at a gas station during a trip. BEVs are sometimes charged at the office or at
home (J. C. Spoelstra, 2014). But high-speed charging facilities at fuel stations are also an option. For
hydrogen cars, these three options are open too. If hydrogen in produced centrally, it makes sense to
refuel at fuel stations. If micro-CHP fuel cells are widespread, it is more likely that hydrogen cars are
refuelled at home or at the office. Either way, some sort of hydrogen storage and transport infrastructure
is required.
Refuelling at a gas station is done during a trip. Therefore, the demand is strongly coupled to the
traffic density at a given moment. In general, more fuel is needed during the day than during the night.
A demand curve of gasoline demand during an average week at a Chevron fuel station in the US (Chen,
2006) is used to represent fuel demand. The pattern is shown in Figure 5.11.
BEVs have not been used for a long time yet, but some studies have addressed the charging behaviour
of this vehicle type (e.g. Schey et al., 2012; INL, 2015; J. C. Spoelstra, 2014). Schey et al. show that
electricity pricing policies can strongly influence when vehicles are charged. For this study, it is assumed
that in the future, charging during the night is stimulated. The pattern of San Francisco (see Figure 5.10)
gives a good indication of how the demand pattern looks like in that case. The extremely steep increase
in demand at midnight is the result of consumers that schedule the start of charging at that time, when
they pay a lower tariff (Schey et al., 2012). This is seen as undesirable, and therefore it is assumed that
a more smoothened demand pattern is established in the future. Figure 5.10 depicts a modified pattern
with a parabolic shape of demand during the night. This pattern is used in this study.

1.2
Demand per charging station (kW)

Recorded demand
1
Spread-out demand
0.8

0.6

0.4

0.2

0
8 10 12 14 16 18 20 22 0 2 4 6 8
Time of the day

Figure 5.10: Median electricity demand at residential BEV charging units in San Francisco. The
recorded data were obtained during a 3-week measurement period in October 2011 (Schey et al.,
2012).

Similar to the research referenced above, the fuelling pattern of FCEVs has been studied. NREL
has an extensive monitoring programme of hydrogen production, distribution, and usage for transport.
The Composite Data Products that result from their analyses are publicly available (Sprik et al., 2015),
and will be used here. Figure 5.11 depicts the distribution of hydrogen demand and number of fuelling
events over the course of a day and a week. For comparison, the number of fuelling events at a regular
gas station is shown, too. Hydrogen demand is highest in the morning, while gasoline shows a demand
peak in the afternoon. A possible explanation is that FCEV owners try to minimise evaporation losses
of hydrogen when the car is idle during the night.

68
 20%
10%

8% 15%

6% 10%

4% 5%
2% 0%
day uesday nesday ursday y y
Frida Saturda Sunda
y
0% Mon T Wed Th
0 3 6 9 12 15 18 21 24
Hour of the day H2 fuellings Dispensed H2 Gasoline fuellings

Figure 5.11: Average consumption pattern of H2 at fuel stations during the course of a day and a
week. For comparison, the pattern of gasoline demand is shown, too. Adapted from Sprik et al.
(2015).

5.8 Final remarks

This chapter has specified the reasoning and data sources behind the assumptions used as input for
the model. The most important simplifications are summarised here. Note that the effect of each
simplification on the results depend on the configuration of the system that is modelled.

• It is assumed that technologies are available all the time. Scheduled maintenance or unexpected
failures and crashes are not considered.
• The model is mostly linear, with an exception for the SOC components. Due to the linearity,
the efficiency of transformers is constant, and independent of the capacity utilisation. As a
consequence, the efficiency of fossil fuel powerplants is overestimated in the model.
• Another implication of the linearity is that economies of scale (see also § 8.1.3 are not taken into
account. All costs have a linear dependence on the installed capacity or produced energy. The
effect of this simplification becomes significant if the installed capacity of a technology differs a
lot from the capacity for which the assumed costs hold.
• Transport losses are neglected, since the model has a country-wide spatial resolution. Therefore,
no differentiation between locally and centrally produced energy can be made. This assumption
affects the results in particular when an energy carrier has high transport losses, or when the
distance between production and demand are large.
• Because of the lacking spatial resolution, aggregated patterns for supply and demand are used.
Local effects, such as congestion on the electricity grid and differences in demand between regions,
are not taken into account.
• The demand patterns, as introduced in § 5.7 are fixed. No demand-response mechanisms or price
elasticity are taken into account.
• Fossil fuels are instantaneously available whenever there is a demand. This simplification removes
the need to include storage components to the model for fossil fuels. Only for isolated systems
with limited or intermittent access to continental or global commodity markets, this assumption
is debatable.
• The optimum configuration for the system as a whole is to be found. From the perspective of
individual organisations, the results can be sub-optimal.
• The solver has a perfect information position, since all patterns are known in advance. This is
not the case in reality, where weather forecasts and demand predictions are never fully accurate.
Due to this knowledge difference, there are backup power facilities or ‘spinning reserves’ in the
real world, but not in the model.
• If the full complexity of the energy system were to be modelled, several second-order effects should
be taken into account. To be precise, the costs of resources and technologies are not static, but
depend on the composition of the system. A higher demand for e.g. coal results in a higher coal
price. And the higher the installed capacity of a technical component, the lower the unit price

69
 will be (see also § 8.1.2–8.1.3).

Finally, it is important to note that the total system costs considered by the model. Therefore, energy
costs for consumers cannot be readily derived. The reason is that the amount consumers pay for an
energy carrier consists for a large part of service fees, distribution network connection costs, value-added
taxes (VAT), and other taxes (Schepers et al., 2015). These factors are dependent on policies and
contracts, instead of on technologies, which is why they are not taken into account for this research.
However, it can be expected that there is a positive correlation between system costs and energy
consumer prices.

70
Chapter 6

Technology Options

6.1 General
Each technology that is considered in the model has characteristic properties. Three types of properties
can be distinguised. Physical properties describe input-output relations such as efficiency, economic
properties have to do with the costs entailed, and environmental properties determine GHG emissions.
This chapter discusses the data inputs for each component, and the sources or calculations behind the
numbers. A full overview of the data is provided in Appendix A.2.
A large set of economic properties are obtained from Energistyrelsen and Energinet.dk (2015). This
is because the report contains an elaborate collection of cost predictions, and a high level of consistency
is reached by using one source. If not specified, then the costs are derived from this report. The
interest rate for conventional technologies is assumed 5%, whereas a rate of 8% is taken for emerging
technologies. See § 4.5 for a more elaborate discussion.
Environmental properties are discussed separately in § 6.8.

6.2 Fossil fuels

6.2.1 Sources
Hard coal, crude oil, and NG can be extracted domestically or imported. The costs of the fuels are
independent of the source, as they are traded on an (inter)national market. The prices of the fossil fuel
commodities as predicted by Schoots and Hammingh (2015) for 2030 are used, see Table 6.1. Since
the focus of this study is on the total system costs, the raw product prices are used, excluding taxes
and subsidies.
Table 6.1: Fossil fuel prices in the Netherlands, in the past and predicted for the future. The type
of fuels are North Sea Brent oil, Groningen gas, and steam coal, respectively. All prices apply to
the wholesale market (Schoots & Hammingh, 2015).

Oil price NG price Coal price

($/barrel) (€/GJ) (€/m ) (€/GJ) (€/t) (€/GJ)
3

2000 39 4.82 0.16 4.55 52 1.99

2015a 50 6.18 0.24 6.90 65 2.50
2020 89 11.00 0.28 7.96 81 3.10
2030 140 17.30 0.33 9.38 88 3.37
a Realisations for 2015 according to CBS (http://statline.cbs.nl).

71
6.2.2 Gas powerplant
Gas-fired powerplants are assumed to be of the type Combined Cycle Gas Turbine (CCGT). These plants
operate gas turbines, in which NG is combusted. It is possible to operate at part-load, but this results
in reduced efficiencies (see e.g. Figure 3.4), but this effect is not taken into consideration. The fixed
efficiency is 60% (Energistyrelsen & Energinet.dk, 2015).

6.2.3 Coal powerplant

Electricity can be generated from coal in advanced pulverised fuel powerplants. The efficiency is assumed
to be 43%, the highest efficiency of currently existing plants (Santoianni, 2015). This makes coal
powerplants less efficient than CCGTs, but better than most current coal powerplants.

6.3 Renewable electricity

6.3.1 PV
Several material options exist for the production of PV cells. The most abundantly used material is
high-purity silicon, which is included in PV panels as a mono- or multi-crystalline wafer. For simplicity,
only silicon-based PV panels are considered here. Because of significant cost differences for large and
small installations, differentiation between commercial and residential PV systems is applied. Utility-
scale systems are not included. Although these systems could be more cost-effective (Energistyrelsen &
Energinet.dk, 2015), the uncertainties related to the availability and costs of land are large.

6.3.2 Wind turbines

Electricity is generated from wind in large wind turbines. The height of these turbines has increased
in the past, and is expected to rise some more in the future, to a height of 150 m. The investment
and O&M costs are almost two times lower for onshore wind turbines (Energistyrelsen & Energinet.dk,
2015). On the other hand, there is more space and wind available offshore.

6.4 Energy storage

6.4.1 Lithium battery
Lithium-ion batteries are an important technology for energy storage in devices. The first commercial
versions of larger battery systems are entering the market, although the costs are still high. Data on the
characteristics of batteries are based on the Tesla Powerpack, a Li battery system intended for industrial-
scale (100 kWh) energy storage. The round-trip efficiency under ideal conditions is 92% (Lazard, 2015a).
However, during the assumed lifetime of 10 years, the performance will decrease. Therefore, the average
efficiency is assumed to be 85% (Energistyrelsen & Energinet.dk, 2015). Besides, a self-discharge rate
of 694 kW per MWh installed capacity is assumed, implying that 99.8% of the stored energy is lost
during one day.
The claims made with respect to the investment costs of the Powerpack might be exaggerated,
but do give an estimate of what battery storage can cost in the future. Thus, 250 $/kWh is used as
investment cost. O&M costs are estimated at 10 $/kW (Lazard, 2015a).
Many other battery types exist, and redox flow batteries are also being developed. These will not
be considered here.

6.4.2 Compressed air energy storage

CAES uses underground cavities, like depleted gas reservoirs and salt caverns to store air under high
pressure. Electric compressors are used to fill the storage. When the air is released, it can drive a turbine
to regenerate electricity. CAES is especially suited for the storage of large volumes of energy during long

72
terms (Lazard, 2015b). Since the compression of a gas is an exothermic process, heat is released during
the ‘charging’ process. It is considered unrealistic that this low-temperature heat is recovered. This
implies that an external heat source is required during expansion. Often, the output power is increased
by also burning NG to drive the turbine (Energistyrelsen & Energinet.dk, 2015). For simplicity, it is
assumed here that the stored energy can be recovered without using a fuel. In that case, an efficiency
of 43%
The investment costs for CAES are estimated at 171 $/kWh (Lazard, 2015a), which translates to
154 €/kWh.

6.4.3 Flywheel
A flywheel stores electricity as kinetic energy. It has a very fast response, so can be used to respond to
sudden peaks of demand or production, and for frequency regulation (Lazard, 2015a). This technology is
rather expensive, with O&M costs of 77 $/kWh and investment costs of 2070–9933 $/kWh. Therefore,
the merit of flywheels can only be assessed with models with high temporal resolution.

6.4.4 Hot water tank

The storage of low-temperature heat is needed especially in houses. A low-cost solution is to employ
sensible heat storage in a simple hot water tank. This technology costs only 3 €/kWh (IEA-ETSAP &
IRENA, 2013b). No operational costs are assumed during the lifetime of 20 years. Hot water can be
added to and extracted from the tank without loss. When stored, self-discharge with a rate of 13.6%
of the capacity is assumed.

6.4.5 Phase change materials

High-temperature heat can be more challenging to store, partly because of heat losses over longer
time periods. One of the promising options under development is latent heat storage in phase change
materials (PCMs). By using thermal energy to induce a phase change instead of only heating up a
material, the temperature difference between the PCM and its environment is much smaller. Therefore,
heat losses are reduced to 3% and longer storage periods are possible. Besides, PCMs generally have a
higher energy density compared to sensible heat storage options (IEA-ETSAP & IRENA, 2013b).
The investment costs of PCM thermal energy storage range between 10–50 €/kWh (IEA-ETSAP &
IRENA, 2013b). While the raw PCMs are relatively cheap, substantial costs are involved with the BoP.
Heat transfer technologies (heat exchangers) that provide sufficient (dis)charging capacity are expensive.
To allow for storage of high-temperature heat, either materials with a suitable phase change point or
extensive heat pump systems are needed (IEA-ETSAP & IRENA, 2013b). The heat exchange process
entails some losses; both charge and discharge are assumed to be 80% efficient.

6.4.6 Hydrogen tank

The purpose of a hydrogen tank is to store H2 for later use. Depending on the volume restrictions, the
pressure of the tank can be higher or lower. At (very) high pressure and if the temperature is lowered,
H2 can be stored in liquid form, resulting in a high energy density. Few information on the costs of
H2 storage could be found. The assumptions made are therefore uncertain. The lifetime of the tank is
assumed 15 years. The investment costs are 2.5 €/kWh, and the self-discharge rate is 7% per 24 hours
(Schoenung, 2011). The operating costs are set at a low level of only 0.01 €/kWh.

6.5 Biomass
6.5.1 Sources
Many sources of biomass exists, with large local variation in quantity and nature. Here, two classes of
biomass are distinguished: residues and products. Residues include organic byproducts from industries,

73
wastes from agriculture, forestry, gardens, and food waste. Products are energy crops that are grown
specifically for the application as energy carrier. The average costs of biomass residues and products are
shown in Table 6.2. A range of organic waste streams was analysed by Koppejan et al. (2009). Biomass
availability was discussed in § 5.5.

Table 6.2: Solid biomass supply costs for the European Union (EU), per resource category in € per
GJ energy content (Lamers et al., 2015).

Agriculture Forestry Average

Residues (€/GJ) 8.47–9.94 4.03–4.89 6.83
Products (€/GJ) 8.47–11.8 9.32–11.31 10.25

Not only the sources of biomass, but also the pathways towards usable energy products show a
large diversity (IEA-ETSAP & IRENA, 2013a). Here, only a limited range of conversion technologies is
included. The selection is made while taking the research questions into account.

6.5.2 Biogas
The production of biogas is assumed to proceed via the anaerobic digestion of moist biomass. Futher
details were given in § 5.5. The costs and emissions for the production of biogas and green gas (see
Table 6.3) were reported by Bergsma and Croezen (2011).

Table 6.3: Costs and emissions for the production of biogas for local use and green gas (Bergsma
& Croezen, 2011).

Investment costs (€/kW) Operating costs (€/MWh) GHG emission

Gasifier Gas cleaning Total O&M Electricity Total (kg CO2 -eq./MWh)
Biogas 888 88 977 5.95 12.01 17.95 -172
Green gas 888 162 1051 6.33 15.22 21.54 -172

Storage of biogas provides two problems. First, it takes a considerable storage volume because
methane is diluted with CO2 , leading to low energy densities. Second, traces of steam in the mixture
can react with non-metal oxides to form corrosive acids, which can damage pipelines and tanks (Wiser
et al., 2010). Therefore, a pretreatment is often applied when the biogas is not used immediately. In
such a process, harmful contaminants are removed. The costs of a pretreatment system range between
500–3000 $/kW (Trendewicz & Braun, 2013). However, costs of only 88 €/kW are possible according
to Bergsma and Croezen (2011).

6.5.3 Green gas

Biogas can be upgraded to the same quality as natural gas in the distribution network. It is then called
‘green gas’. Sulfur compounds, ammonia, steam, and CO2 are removed in the process. Desulfurisation,
which has to be applied to NG too, costs $4.71 per 1000 m3 of processed NG (James et al., 2012).
This translates to 105 €/GJ. Other values reported in literature are 162 €/kW (see Table 6.3 (Bergsma
& Croezen, 2011)) and 300 €/kW (Mathiesen et al., 2009). The numbers from Mathiesen et al. (2009)
were used: 300 €/kW investment costs, 56.25 €/kW fixed O&M costs, and an efficiency of 95%.

6.5.4 Biodiesel
Solid biomass can be converted to a liquid form, such as biodiesel or biogasoline. Liquid fuels are
most convenient for combustion processes. Because of the high energy density, they are the dominant
transport fuel. Here, the biomass-to-liquid (BtL) process is considered. This process can use a wide
range of organic products and residues to produce biodiesel (Energistyrelsen & Energinet.dk, 2015).
With an efficiency of 56%, these biomass streams are converted to fuel suitable for regular combustion

74
engines (Boerrigter, 2006). The costs are high with an investment cost of 3393 €/kW and fixed O&M
costs of 103 €/kW (Mathiesen et al., 2009).

6.5.5 Biomass powerplant

The ecomomic characteristics of a biomass powerplant are assumed to be identical to those of coal
powerplants (§ 6.2.3). This is because biomass and coal are both solid fuels, for which the combustion
process parameters are comparable. In the Netherlands, biomass is often co-fired in coal powerplants.

6.6 Fuel cells and hydrogen production

6.6.1 SOEC
A SOEC has heat and electricity as energy inputs. These are used to electrolyse water and produce
hydrogen as output. The ratio between the inputs is fixed, but the rate of electrolysis can be changed.
For each 100 J of electricity input, 20 J of heat with the same temperature as the cell (800–1000 ℃)
should be supplied. The process is assumed to have a heat loss of 10% (Mathiesen et al., 2009). Thus,
the input-output relations can be written mathematically as:
Q̇ = 1
5 · P in (6.1)
ḢH2 = 0.9 · (P in + Q̇) (6.2)
The cost parameters are as follows: Cinv = 590 €/kW, Cfix = 11.8 €/kW (2% of Cinv ), and
Cvar = 1.65 €/MWh (Mathiesen et al., 2009). The investment costs estimated by Ridjan et al. (2013)
are 210 €/kWe input, which is lower.

6.6.2 SOFC and SOFEC

Various sources have estimated current and future costs of SOFC systems. Differences in design choices,
system size, production rate, etc. result in a wide range of cost estimates. The fact that not all studies
include the costs of BoP manufacturing is responsible for some of the diversity encountered in general
(Staffell & Green, 2013). In Figure 6.1, the investment and O&M cost estimates of four sources
are compared. For both costs, there is an order of magnitude difference between the extremes. The
estimates made by Energistyrelsen and Energinet.dk (2015) for SOFCs in base-load mode are used to
make future cost predictions, shown in Table 6.4. The tabulated lifetimes correspond to the lifetime
of the installation including BoP. The costs associated with the replacement of the SOC stacks due to
degradation are included in the O&M costs.

Table 6.4: Cost parameters and levelised cost of electricity for SOFCs, based on (Energistyrelsen
& Energinet.dk, 2015).

0 I II III IV V VI
Properties expected in year 2015 2019 2020 2022 2026 2030 2050
Lifetime (a) 6 13 15 17 19 20 20
Variable O&M costs a (€/MWh) 13.75 6.41 5.5 4.02 2.93 2.2 1.65
Investment costs a (€/kW) 2750 1008 825 670 544 440 275
LCOE b (€/MWh) 227.15 104.38 93.64 85.66 79.81 75.75 70.88
a Per unit of NG input.
b Per unit of electricity. Assuming an electric efficiency of 55%, 90% capacity utilisation, 8%

discount rate, and neglecting taxes and the value of the co-products.

The economic properties of SOFECs are assumed to be identical to those of SOFCs. This is a
reasonable assumption since both installations can be produced and maintained in a similar manner.
The default fuel for SOFCs is NG, whereas SOFECs are assumed to be fuelled by biogas. No restriction
in terms of ramping rates are applied, because this does not play a role at the timescale of hours.

75
 Investment cost (€/We) Source – System size, year O&M costs (€/MWhe)

€3.43-6.76 Lazard (2015) – 2.4 MWe €27-45

€5.00 Energistyrelsen (2015) – 100 kWe, 2015a €25
€4.35 Trendewicz & Braun (2013) – 330 kWe €3-14
€2.09 Arduino et al. (2016) – 250 kWe €25
€2.00 Energistyrelsen (2015) – 100 kWe, 2015b €25
€0.50 Energistyrelsen (2015) – 100 MWe, 2050a €3
€0.40 Energistyrelsen (2015) – 500 kWe, 2030b €10
€0.34 Vernay et al. (2008) – 2010 €3

6 5 4 3 2 1 0 0 5 10 15 20 25 30 35 40 45

Figure 6.1: Comparison of investment and O&M costs for SOFCs from different sources. Amounts
in Dollars were converted to Euros using the average exchange rate in the year of the publication.
a Base-load operation. b Grid balancing operation.

6.6.3 PEMFC
Low-temperature PEMFCs can be used to convert hydrogen to electricity. Using data from James et
al. (2012), the efficiency is 35%, and the investment costs are 207 €/kW. Note that the theoretical
efficiency is lower than 100% due to thermodynamic limits. The oxidation of H2 is exothermic, so heat
losses are inevitable.

6.6.4 SMR
An alternative and most widely applied way to produce hydrogen is via steam methane reforming. The
economic characteristics of SMR are adapted from Diakov et al. (2011), in which technology costs for
2025 are predicted. The efficiency of the conversion of NG to H2 is assumed to be 80%. A lifetime of
20 years is assumed.

6.6.5 Methanation
The Sabatier process, also referred to as methanation, is a reaction between CO2 and H2 , in which
methane is produced:
CO2 + 4 H2 −−→ CH4 + 2 H2 O (R5)
An installation that performs this process has a lifetime of 20 years and costs 900 €/kW. The process
operates at 90% efficiency, while consuming some electricity (15 times less than the input of H2 ). The
fixed O&M costs amount 22.14 €/kW/a (Connolly & Mathiesen, 2014).
Methanation in combination with electrolysis of water (producing the required hydrogen) is called
PtG.

6.7 Heat-related technologies

6.7.1 Solar thermal panels
Solar energy can be used to directly heat water in solar thermal panels. These systems are often installed
on the roof of residential buildings. The investment costs associated with these panels are
£3300
= 943 £/kW
3.5 m2 × 1 kW/m2

76
(Milieucentraal, 2016). Future cost reductions are expected to reduce the investments by 50%, backed
by the estimate of 60% cost reduction in 2030 of Stryi-Hipp (2009). Thus, the investment costs are
assumed to be 471 €/kW. The operating costs are estimated at 10 €/a per system, or 2.20 €/kW per
year(Milieucentraal, 2016). A lifetime of 30 years is assumed (Energistyrelsen & Energinet.dk, 2015).

6.7.2 Boilers
The efficiencies of boilers fuelled with NG, coal, oil, or biomass are tabulated in Table 5.3.

Table 6.5: Economic properties of industrial boilers.

Coal Oil Gas Biomass

Investment costs (€/kW)
a 800 100 100 800
Variable O&M costs a (€/MWh) 5.40 – – 5.40
Fixed O&M costs a (€/kW/a) – 3.7 3.7 –
Economic lifetime (a)b 25–40; average 32.5
a (Energistyrelsen & Energinet.dk, 2015). b (Van Wortswinkel & Nijs, 2010).

6.7.3 Heat distribution

It is possible to convert high-temperature heat to low-temperature heat using heat exchangers or by
dilution. The investment costs associated with this process are those of district heating system construc-
tion. This is because only central high-temperature heat production options and local low-temperature
heat consumption are present in the model. Therefore, a heat distribution network is always required
to deliver diluted heat. The installation are amount 1309 €/kW, which is an average of different
neighbourhood types (Van Melle et al., 2015, p. 30).

6.8 Greenhouse gas emissions

6.8.1 Supply chain
The supply chain of fuels and energy conversion equipment should be taken into account when the
impact on climate change is considered. The variable GHG emissions for the supply chain of fossil
fuels, biomass and biogas are tabulated in Table 6.6. Biomass has a negative emission, since CO2 is
absorbed during its production. Two conversion processes are included in this table, whereas combustion
processes are covered later. In Table 6.7, the fixed GHG emissions that occur during the production and
construction of equipment needed for energy conversion. For solar and wind energy, these fixed emissions
at the beginning of the life cycle are most important. For conventional energy supply, emissions from
fuel combustion tend to dominate the life cycle impact.

Table 6.6: Emissions related to fuel production/supply and conversion.

Fuel/Process Emission Source

(kg CO2 -eq./MWh)
Coal 56.4 Edwards et al. (2014)
NG 56.8 Edwards et al. (2014)
Crude oil 50.4 Bergsma and Croezen (2011)
Biomass residues -238 Biomass Energy Centre (2004)
Biomass products -341 Biomass Energy Centre (2004)
Biogas -172 Bergsma and Croezen (2011)
Biogas upgrade 10.1
SMR 236 Diakov et al. (2011)
Biodiesel 183 Edwards et al. (2014)

77
Table 6.7: Emissions associated with the production and/or construction of energy conversion
technologies (Bruckner et al., 2014).

Technology Emissions
(kgCO2 -eq./GW)
Nuclear power 3996
Coal power 2886
NG power 4829
Wind onshore 591
Wind offshore 975
PV 1794
Solar thermal heat 38

The production of solar thermal panels is associated with an emission of 38 kgCO2 -eq./GW (Lamnatou
et al., 2015). This number was derived assuming a maximum solar irradiation of 1 kW/m2 , and 65%
efficiency.
For biogas production, a negative emission of −179.5 kg CO2 -eq./kWh is assumed for the digestion
process (Bergsma & Croezen, 2011). Some CH4 leaks away during the production and processing
of biogas. Expressed in CO2 -equivalents, there is a loss of 7.8 kgCO2 -eq./kWh biogas Bergsma and
Croezen (2011). This translates to a loss of 1.7% of biogas throughout the chain. For validation, this
figure is compared with the values from the Swedish voluntary agreement system for biogas producers
(Holmgren, 2012). Their agreement states that a maximum of 2.5% is lost during production, and 2.1%
during gas upgrading. The loss presented by CE Delft falls below this upper bound. Yet, the large gap
between the numbers could indicate that one of them is an under- or overestimation.
The process to upgrade biogas to green gas has an efficiency of 95% (§ 6.5.3). The gas that is lost
is assumed to be emitted as if it were combusted with emission of NG.

6.8.2 Combustion
The emissions of conventional electricity and heat sources were derived from Gómez et al. (2006). The
emissions of CO2 and other GHGs are listed in Table 6.8. For the combustion of biogas, the same
emissions per unit of calorific value of NG are assumed. For SOFCs and SOFECs, an emission of CO2
identical to that of NG combustion was assumed, along with zero emission of other GHGs.

Table 6.8: GHG emissions released during the combustion of the most common fossil fuels in
different applications (Gómez et al., 2006), per GJ of fuel. Emissions of individual GHGs are
expressed relative to the LHV of the fuel, whereas the total GHG intensity is expressed relative to
the HHV.a

Sector Fuel Assigned to CO2 CH4 N2 O Total emission a

(kg/GJ) (g/GJ) (g/GJ) (tCO2 -eq./MWh)
Energy Natural gas NG powerplant 56.1 1 0.1 184
industries Coking coal Coal powerplant 94.6 1 1.5 326
Other primary Biomass boiler 100 30 4 350
solid biomass
Manufacturing Coking coal Coal boiler 94.6 10 1.5 327
& construction Other primary Biomass boiler 100 30 4 350
solid biomass
Natural gas Gas boiler 56.1 1 0.1 184
Transport Gasoline Cars & aviation 70 3 0.6 241
a The HHV is assumed to differ from the LHV by a factor of 1.1 for gaseous fuels, and 1.05 for

solid or liquid fuels (Gómez et al., 2006).

78
Chapter 7

Optimisation Results

7.1 General
A number of experiments was conducted using the optimisation model described in § 4.4. By analysing
the resulting optimal systems, questions related to the economic competitiveness of SOCs and the effect
on the sustainability of the energy system can be answered. This chapter starts with three test cases to
demonstrate the correctness of the optimisation model in § 7.2. Afterwards, the results of experiments
addressing the effect of the costs of SOCs, the energy demand, and GHG emission taxes are presented.
The chapter finishes with an optimisation of the transition pathway towards a future energy system.
The total system cost objective is used. A price for GHG emission is taken into account, but the
emissions are never the only thing that is optimised. For each optimisation, the following information
will be provided, to specify the input data used for that specific experiment:

• GHG emission tax

• Demand scenario
• Special restrictions on components
• The costs associated with SOCs

The demand and production patterns are the same in each experiment, and are introduced in § 5.6–5.7.
The data cover one year, i.e. 8760 timesteps of one hour.

7.2 Tests
GHG tax Scenario Restrictions SOC costs
0 €/tCO2 -eq. Special (see below) Limited set of components –
For the test cases in this section, the complexity of the models is low because the number of components
is reduced. There is one energy source, i.e. natural gas. On the demand side, there is a 1 PJ/a electricity
demand by households or 1 PJ/a consumption by BEVs. A time period of two weeks is considered,
average summer and average winter week.

Test 1: Two identical powerplants

In this test model, there are two gas-fired powerplants that have to fulfil a demand for electricity only.
Both plants have the same properties. As a consequence, it is indifferent which powerplant is installed
and operated. The linear solver GLPK yields a solution with one of the two plants not used, while
the solution of IPOPT features the two plant types with identical installed capacities and operating
schemes. Both solutions are optimal given the assumptions made. These results are independent of the
electricity demand profile selected.

Test 2: Two powerplants with different costs

Two plants are available, with the only difference being the fixed and variable O&M costs (Table 7.1).

79
 Constant demand

Plant 1
Plant 2

400

System costs (k€/a)

300
Household demand
Electricity production (MW)

200

100

0
0 10 20 30 40 50 60 70 80 90 100
Capacity Plant 1 (MW)

BEV damand Figure 7.2: System costs relative to the op-

timum value for a system with two power-
plants. The costs were calculated by fixing
the installed capacity of Plant 1 at the given
vale, and the capacity of Plant 2 at a size
to complement Plant 1.

Figure 7.1: Electricity production by Plant 1 and 2

during two weeks, for three different demand curves.

Table 7.1: Properties of Plant 1 and 2, and their levelised costs of electricity (LCOE) for different
capacity utilisations. Fuel costs are included in the variable costs.

Lifetime Investment Variable O&M Fixed O&M LCOE (€/MWh) at CU

(a) (€/kW) (€/MWh) (€/kW) 25% 78% 100%
Plant 1 25 790 58.80 15 91.24 69.20 66.91
Plant 2 25 790 56.60 30 95.90 69.20 66.43

Further assumptions are 60% fuel efficiency, 5% discount rate, and NG costs of 33.78 €/MWh. If the
capacity utilisation is set to 78%, then the LCOE is identical for both plants. At higher CUs, Plant 2 has
the lower LCOE, while Plant 1 becomes more attractive at lower CU. In the optimal configuration, the
installed capacity of Plant 2 is at least as big as the lowest demand of the time period. It also supplies
each infinitesimal increase in demand above the base load, on the condition that the CU of that part of
the demand is smaller than or equal to 78%. The remaining electricity demand peaks are met by Plant
1.
The optimisation results for three demand curves are plotted in Figure 7.1. In accordance with the
expectation, Plant 2 is the only plant installed when there is a constant electricity demand. In case of
very variable demand, as for example from BEVs, Plant 1 is the dominant type, since a high CU is only
attained for a small fraction of the capacity. When the demand pattern of households is assumed, then
both plants have a comparable installed capacity. The latter case was studied further by restricting the
installed capacity of both plants. As shown by Figure 7.2, any deviation from the installed capacities in
the optimum case result in higher system costs.
This test shows that the basics of energy economics are accurately reproduced in the model, provided
that the installed capacities are continuous decision variables that can also assume non-integer values.
In addition, the optimality of the solution for the test problem was demonstrated.

80
Test 3: Storage and excess
Fixed sources, such as PV and wind, do not produce sufficient energy to meet demand all the time.
There are two options to guarantee all demand is met: temporary storage of energy, or the construction
of more generation facilities in combination with curtailment of excess energy.
6000 900
PV 800 PV
Production (MW)

Production (MW)
5000
Battery 700 Battery
4000 600
500
3000
400
2000 300
200
1000
100
0 0
Battery 800 Battery
Consumption (MW)

5000

Consumption (MW)
Excess Excess
4000 600 Demand
Demand
3000
400
2000

1000 200

0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Time (h) Time (h)

(a) 225 €/kWh investment costs, 9.015 €/MWh vari- (b) 225 €/kWh investment costs, no variable O&M
able O&M costs. costs.
700 350
600 PV 300 PV
Production (MW)
Production (MW)

Battery Battery
500 250
400 200
300 150
200 100
100 50
0 0
600 Battery 300 Battery
Consumption (MW)
Consumption (MW)

Excess Excess
500 250
Demand Demand
400 200
300 150
200 100
100 50
0 0
0 50 100 150 200 250 300 0 50 100 150 200 250 300
Time (h) Time (h)

(c) 50 €/kWh investment costs, 1 €/kWh variable (d) No investment costs, 1 €/kWh variable O&M
O&M costs. costs.

Figure 7.3: Optimised electricity production and consumption patterns in a simple energy system
during an average summer and winter week. Demand consists of 1 PJ/a electricity by households.
The characteristics of the components are those of a Li battery and commercial PV, except for the
cost characteristics of the battery as specified below the graph.

To test whether these options are possible within the model, a simple system with three available
components was constructed: PV, batteries, and electricity demand from households. Then, the optimal
solution was calculated four times, each time with different costs associated with the battery. As
Figure 7.3 shows, large-scale storage becomes attractive if the costs decrease significantly.

7.3 Nuclear energy

GHG tax Scenario Restrictions SOC costs
5 €/tCO2 -eq. Diversity None II, see Table 6.4
It turns out that nuclear energy is the most cost-efficient source of electricity. Without any restric-
tions on the number of nuclear powerplants, virtually all electricity is generated by these plants (see
Figure 7.4). In such a system, electricity has a very low price, in particular in summer. Electrolysis

81
 Summer Winter
20000

Production (MW)
15000
nuclear power
10000 onshore wind
gas powerplant
5000

0
20000
Consumption (MW)

industry
15000
households
heat pumps
10000
transport
5000
SOEC
electric boiler
0
Thu Fri Sat Sun Mon Tue Wed Thu Fri Sat Sun Mon Tue Wed

Figure 7.4: Optimal electricity production pattern when nuclear power generation is not con-
strained. The patterns are an average over a summer and winter week, for a diverse demand
scenario.

therefore becomes an attractive method of hydrogen production. Since the output of nuclear plants
cannot be regulated easily, the system features electric boilers that convert excess electricity to heat.
The resulting energy system configuration found here is far from the real-world situation. There
are also several objections to such a system, among which safety concerns, the dependence on uranium
producing countries, and the limited space for long-term storage of nuclear waste. To incorporate these
considerations into the model, a constraint is added in addition to those presented in § 5.5. All following
optimisations will be performed with 4 GWe as the maximum installed capacity of nuclear powerplants.

7.4 SOCs in the current energy system

7.4.1 Setup
GHG tax Scenario Restrictions SOC costs
5 €/tCO2 -eq. BAU see Table 7.2 VI, see Table 6.4
The question at hand is whether the (electricity) production system as it is currently in place, can
profit from the installation of SOCs. To model the current energy system, the assumptions made are
as follows. The installed capacity of electricity producing components in 2015 were obtained from
ENTSOE (2016) and are listed in Table 7.2. The capacity of MSW powerplants is added to that of NG
plants. Hydropower is not included in the model. It is assumed that the installed capacity of SMRs is
1402 MW, exactly enough to meet the demand for hydrogen. The demand for energy was set to the
values of 2014, as identified in § 5.3. The prices of fossil fuels were set to the 2015 values as listed
in Table 6.1. A model was constructed with these numbers included. Then, the system was optimised
three times with different constraints on the installed capacity of electricity producing components:
A The installed capacity is fixed
B The installed capacity is fixed, except for SOFCs and SOFECs
C The installed capacity is not fixed.
Supply and demand of energy carriers other than electricity and hydrogen are still part of the model,
but are not constrained and are not discussed here. Heat generation is not constrained, because there
is no data on the installed capacity of heat-related equipment.

82
Table 7.2: Installed generation capacity of powerplants aggregated by type, in 2015 in the Nether-
lands (ENTSOE, 2016).

Powerplant type Capacity (MW)

Natural gas 19914
Hard coal 5658
Onshore wind 3284
PV 1429
MSW 674
Nuclear 486
Biomass 398
Offshore wind 357
Hydro 38
Total capacity 32238

7.4.2 Validation
In case A, the optimisation problem reduces to an operation optimisation problem of electricity pro-
duction. This setup can serve to validate the optimisation results by comparing them with energy
production statistics. For this comparison, data on the real electricity and heat generation in 2014 and
2015 are presented in Table 7.3 and Figure 7.5–7.6. A major difference between statistical data and
results from A, is the larger use of coal for energy production. Virtually all heat1 originates from coal
in the optimised system, whereas NG is the primary source in reality. To begin with, this is because no
information about the installed capacity of NG furnaces was included in the model. Yet, the investment
costs of furnaces are small compared to the fuel costs. This points to a second reason for the difference:
in the model, coal is much less expensive than NG. Factors that are not accounted for are distribution
costs (which are higher for coal than for NG), ease of use (a coal stock has to be maintained), and
regulatory particulate matter emission limits.
The model correctly operates nuclear and coal powerplants as base load plants with 99–100% CU
(Table 7.4), which is consistent with reality. Furthermore, gas plants are employed as peaking plants,
as concluded from the CU of 38%. Interestingly, the installed capacity of gas powerplants is more than
twice as big as the capacity that is actually used in the model. This is mostly explained by the fact that
electricity production is shifted towards coal in case A compared to reality (Table 7.3). It is not clear
what causes this difference.
1
Heat derived from fuel combustion in households and businesses is not considered here.

400
350 400
Electricity production (PJ/a)

300 SOFC 350

Electricity production (PJ/a)

250 PV 300
Other fossil
200 250
Nuclear
150 200
Wind onshore
100 Wind offshore 150
Coal 100
50
Natural gas
50
0
A. 2015 B. SOCs C. all plants 0
System available available 2014 2015
(a) Results of optimisation. (b) Data from CBS (2015a).

Figure 7.5: Energy sources for electricity production in the Netherlands under the current condi-
tions.

83
 200
Heat production (PJ/a) 200
150

Heat production (PJ/a)

SOFC 150
100
Coal
100
Electricity
50
Other fossil
50
Natural gas
0
A. 2015 B. SOCs C. all plants 0
System available available 2014 2015
(a) Results of optimisation. (b) Data from CBS (2015a).

Figure 7.6: Energy sources for heat production in the Netherlands under the current conditions.

Table 7.4 indicates that none of the biomass powerplants is being used in the model. This reflects
the current situation, where electricity from biomass is only competitive if subsidies are granted, or if
the price difference between biomass and coal were smaller.

7.4.3 Results
Case B is an extreme case where SOFCs and SOFECs are assumed to have reached low costs. The results
show, that SOFCs have the potential to substitute a considerable fraction of the electricity currently
generated by gas powerplants. An annual electricity production of 26 PJ/a by SOFCs is optimal. As a
consequence, the CU of gas powerplants further decreases to 27%, while the CU of SOFCs is very high
for both electricity and hydrogen production (see Table 7.4). The impact of SOFCs on the hydrogen
market is even larger: over half of the SMR installations becomes redundant, while the remaining
installations are used as a backup facility to provide only 6% of the demand. The effect on primary
energy use compared to case A are significant. The total gas consumption (including non-energy uses)
decreases by 9 PJ/a, while the use of coal is even 34 PJ/a lower. The former is due to the high efficiency
of SOFCs, while the latter is because high-temperature heat from coal is partly replaced by heat from
SOFCs. SOFECs are not installed, indicating that the technology is too expensive to compete with
already installed SMR systems.
Under the assumption of case C, where any plant type can be installed, the following is observed.
The installed capacity of SOFCs is the same as in case B, but less electricity is produced. The reason
is that the capacity of nuclear powerplants doubles, now that investments are allowed. There are no
other changes in installed powerplant capacities. Together with the electricity production, the hydrogen
output of SOFCs also declines somewhat. Apparently, both products are strongly coupled in this system.
Consequently, the required capacity and the CU of SMRs increases with respect to case B.
From these results, conclusions can be drawn only for the hypothetical situation that SOFCs with
the costs expected in 2050 were available now. Under that assumption, it would be attractive to install
a substantial amount of them from a systems point of view. SOFCs can become the main source of
hydrogen for industry. SMRs are mostly replaced, and are used as backup technology. Furthermore,
SOFCs can substitute some electricity from gas powerplants and heat from coal boilers. However, the
current costs associated with SOCs are far from competitive, as will be shown in the next section.

84
Table 7.3: Energy sources for electricity and heat production in the Netherlands under the current
conditions.

Realisations Optimised system

(CBS, 2015a) A. B. SOCs C. All
Energy source 2014 2015 Fixed available available
Electricity Natural gas 184 166 104 62 56
Coal 104 140 198 197 196
Wind onshore 21 27 22 22 22
Wind offshore 5.2 5.2 5.2
Nuclear 15 15 15 15 32
Other fossil 16 15
Solar 2.8 4.0 5.3 5.3 5.3
Fuel oil 1.2 0.3
Hydro 0.4 0.3
Biomass
SOFC 43 34
Other
Total 343.5 366.9 350.4 350.4 350.4
Heat Natural gas 148 141
Other fossil 15 11
Electricity 12 12
Coal 3.1 4.8 183 157 163
Fuel oil 0.033 0.013
Solar 1.1 1.1 1.1
SOFC 26 20
Total 177.7 169.9 184.2 184.2 184.2

Table 7.4: Capacity utilisation in three scenarios, all assuming the present installed capacities of
powerplants. The capacity utilisation is calculated by dividing the average output by the maximum
output.

A. fixed B. SOCs available C. free

Electricity
Nuclear 100% 100% 100%
Biomass 0% 0% 0%
Coal 99% 99% 98%
Gas 38% 27% 26%
PV households 15% 15% 15%
Offshore wind 46% 46% 46%
Onshore wind 21% 21% 21%
SOFC 94% 91%
Hydrogen
SMR 100% 12% 19%
SOFC – 94% 89%

85
7.5 Effect of the costs of SOCs
7.5.1 Setup
GHG tax Scenario Restrictions SOC costs
20 €/tCO2 -eq. Diversity – I–VI, see Table 6.4
The current costs of SOCs are very high, and it is unlikely that they can compete with mature tech-
nologies at the present price level. However, the effects discussed in § 8.1.2 can cause major cost
reductions. The change in the potential role of SOCs in the energy system is assessed based on an
existing, approximately exponential, cost reduction prediction (Energistyrelsen & Energinet.dk, 2015).
The properties of SOFCs in the future are given in Table 6.4. The composition of the LCOE is plotted
in Figure 7.7.

€250 Emission taxes

Variable O&M
€200 Investment costs
Fuel costs
LCOE (€/MWh)

€150

€100

€50

€0
2015 2019 2020 2022 2026 2030 2050
0 I II III IV V VI

Figure 7.7: Levelised costs of electricity from SOFCs, based on the numbers in Table 6.4.

7.5.2 Installed capacities

Using the diverse demand scenario and a GHG tax of 20 €/tCO2 -eq., the energy system was optimised
for each of the sets of characteristics listed in Table 6.4 applied to both the SOFC and the SOFEC
component. The contribution of the available technology options to the total production of electricity,
hydrogen, and heat are show in Figure 7.8. As expected, the current costs prove to be too high to be
competitive. This is why the results of scenario 0 are not plotted; they are identical to those of scenario
I. With the predicted costs of 2020 (scenario II), SOFCs already start to play an important role. The
main driver for their application appears to be the demand for hydrogen, since SMR as a H2 source
is replaced. SOFCs can be more cost-effective than SMR because of the co-production of heat and
electricity. As a consequence, less coal-fired boilers and gas powerplants are installed, but also less wind
turbines.
With the property set of scenario III, all SMRs are substituted by SOFCs. After decreasing the costs
one step further (scenario IV), all coal and gas powerplants are replaced. Comparing scenario IV with
I (a system without SOFCs), 30% less wind turbines are installed. Furthermore, a large fraction of
coal heat plants is replaced. From these observations, it is deduced that the levelised costs of hydrogen
production by SOFCs are competitive earlier than the costs of electricity production. A further reduction
in costs, as seen under scenario V and VI, does not result in drastic shifts in production patterns.
SOFECs play a much smaller role than SOFCs do. The highest production of hydrogen by SOFECs is
1.75 PJ/a in scenario VI, corresponding to 317 MW installed capacity. The low application of SOFECs is
mostly due to the low amount of intermittent electricity production. Peaks in production are addressed
by electric boilers, whereas fuel-assisted and normal electrolysis require more periods of low electricity
prices. Under the assumptions made here, biogas is most cost-effectively used by upgrading it to green
gas, and hydrogen is derived from SOFCs.

86
 500
30
450
400
PEMFC 25
350 Offshore wind Offshore wind
Production (PJ/a)

Installed capacity (GW)

300 Natural gas 20 PEMFC
250 Coal Natural gas
SOFC 15 Coal
200
Onshore wind SOFC
150 Onshore wind
Nuclear 10
100 Nuclear
50 5
0
I II III IV V VI 0
SOC lifetime I II III IV V VI
(a) Electricity (b) Electricity
16
140
14

Installed capacity (GW)

120 12 SOFEC
Production (PJ/a)

100 SOFEC
10 SOFC
80 SOFC 8
60 SOEC SOEC
6
40 SMR 4 SMR
20 2
0 0
I II III IV V VI I II III IV V VI
(c) Hydrogen (d) Hydrogen
40
300 35
Installed capacity (GW)

250 30 SOFEC
Production (PJ/a)

200 SOFEC 25
SOFC
SOFC 20
150 15 Coal boiler
Coal boiler
100 10 Electric boiler
Electric boiler
50 5
0
0
I II III IV V VI I II III IV V VI

(e) Heat (f) Heat

Figure 7.8: Total annual production (left) and installed capacities (right) of transformers under
different SOC cost assumptions (see Table 6.4).

Summer Winter
8
H2 consumption (GW)

7
Industry
6
PEMFC
5
Transport
4
H2 tank
3
2
1
0
Thu Fri Sat Sun Mon Tue Wed Thu Fri Sat Sun Mon Tue Wed

Figure 7.9: Hydrogen demand during an average summer and winter week, with SOFC properties
from scenario III (Table 6.4).

87
 The annual production of hydrogen in scenario IV is around 3 PJ larger compared to scenario I (see
Figure 7.8c). Also, high peaks in H2 production by SOFCs occur. A closer look at the hydrogen demand
patterns in Figure 7.9 learns that this is because PEMFCs play a role in this system as peak electricity
production technology. These situations occur almost exclusively during weekdays in winter months,
when much electricity is required for among others heat pumps. In summer, there are some occasions
where PEMFCs are operated to address the electricity demand peak caused by BEV charging. It is
unexpected that this phenomenon does not occur in scenario VI, but this observation can be explained.
In scenario VI, the investment costs of SOFCs are sufficiently low to allow the installation of some
systems that are unused most of the time. These systems help to address the high electricity demand
peaks, making the PEMFCs redundant. In scenarios III-V, the installed capacity of SOFCs is lower. To
still meet electricity demand, the hydrogen output is increased and converted to electricity by PEMFCs.
Figure 7.8f shows that the installed capacity of coal boilers remains more or less constant in all
scenarios, while its production decreases. This indicates that SOFCs do provide heat, but it is mainly a
by-product. Therefore, the coal boilers remain necessary as load-following technology.

7.5.3 Costs
The cheaper the SOFCs are, the more of them are installed, and the lower the total system costs are.
This effect is illustrated by Figure 7.10. In scenario II and III, the energy system as a whole has higher
investment costs due to the SOFCs. O&M costs slightly increase, too. But the reduced emission tax
burden outweights these effects at a given point. In the case of a 20 €/tCO2 -eq. tax, the tipping point
lies between SOFC property sets I and II (Figure 7.10). Presumably, higher GHG taxes can make SOFCs
attractive at an earlier stage (see also § 7.7.2). Most system-wide O&M costs originate from fossil fuel
production. When less expensive SOFCs are available, they replace coal and gas powerplants in the
optimal energy system. The resulting replacement of coal by NG as energy input increases the resource
costs of the system. This is partly compensated by the shift from coal powerplants with high O&M
costs to SOFCs with lower O&M costs.
Consistent with the similarity in energy production patterns of scenario IV–VI, the emission taxes
are virtually identical for these scenarios. Further cost reductions originate from the lower technology
costs, while GHG emissions do not decrease further from scenario IV onwards (Figure 7.10).

3200
500 2800
System cost difference (M€/a)

V VI
Total production (PJ/a)

0 2400
I II III IV Renewables
-500 2000 Coal
1600 Natural gas
-1000
Nuclear
1200
-1500 Oil
Investment costs 800
-2000 O&M costs
Emission taxes 400
-2500
0
I II III IV V VI
Figure 7.10: Change in system costs for scenario
II–VI compared to scenario I. Figure 7.11: Primary energy use for scenario I–
VI.

The primary energy use of the six scenarios, depicted in Figure 7.11, is compared. A larger utilisation
of SOFCs has two effects. The total primary energy use is reduced, coal is replaced by NG as energy
sources. As a consequence of these changes, the annual GHG emissions decline by 34.6 MtCO2 -eq.
between scenario I and III. Both effects can be fully assigned to the efficiency of SOFCs, since only
SOC costs are different between the scenarios and SOFECs are only added at lower costs (Figure 7.8d).
The addition of SOFCs does not, however, promote the use of renewable resources. So from a financial
perspective, it does not become more attractive to operate intermittent energy sources because of

88
the diffusion of SOFCs. No conclusion can be drawn on the reverse effect: whether SOFCs are more
appealing if more renewable energy is produced.
It should be noted that the results presented in this section depend on the GHG emission tax that
is applied. The effect of this assumption is investigated further in § 7.7.

7.5.4 SOFC operation

Comparing the results of scenario II and V, there are more differences than the installed capacity of
SOFCs. The optimal operation of the fuel cells differs in the two cases, as illustrated by Figure 7.12 a
and b. In this representative part of the year, Uf changes much more for scenario V. It takes all values
within the allowed boundaries (0.60–0.95). In scenario II, however, Uf equals 0.60 most of the time.
This results in a high hydrogen production relative to electricity. The NG input of the SOFCs shows
a similar difference between the scenarios as Uf (not shown). For scenario II, the input is often high,
while it fluctuates much more in scenario V.

a) 16 100%
14
80%
Output (GW)

12
10 60%
8

Uf
6 40%

4
20%
2
0 0%
80%
b) 5
70%
4 60%
Output (GW)

50%
3
40%

Uf
2 30%
H2 20%
1 Electricity 10%
Fuel utilisation
0 0%
c) 8
7
Hydrogen output (GW)

6
5
4
3
2
1 SMR
SOFC
0
26 27 28 29 30 31 1 2 3 4 5 6 7 8 9 10
January February

Figure 7.12: Optimised operation of SOFCs. Fuel utilisation and hydrogen and electricity pro-
duction of SOFCs for Diverse demand scenario and costs a) V and b) II. c) Cumulative hydrogen
output of SOFCs and SMRs in scenario II.

89
 These observations can be explained by taking the installed capacity of SOFCs into account. The
hydrogen production curve of SOFCs in scenario V is dictated by demand, since SOFCs supply virtually
all hydrogen. In scenario II, supplementary hydrogen is produced by SMR (Figure 7.12c). This allows
to operate the SOFCs in the most profitable way: with high hydrogen production and high capacity
utilisation. This is not possible when the installed capacity is as high as in scenario V.

1.0 200
'CE-4.dat' matrix
0.8
150

i (A/cm2)
0.6
100
0.4
50
0.2

0.0 0
No SOFCs 60 65 70 75 80 85 90 95
(a) Scenario I, 0 GWe . (b) Scenario
Uf (%)II, 3.3 GWe .
1.0 160 1.0 120
'CE-3.dat' matrix 140 'CE-2.dat' matrix
0.8 0.8 100
120
80
i (A/cm2)

i (A/cm2)

0.6 100 0.6

80 60
0.4 60 0.4
40
40
0.2 0.2 20
20
0.0 0 0.0 0
60 65 70 75 80 85 90 95 60 65 70 75 80 85 90 95
(c) Scenario III, 10 GWe .
Uf (%) (d) Scenario
Uf (%)IV, 13 GWe .
1.0 100 1.0 90
'CE-1.dat' matrix 90 'CE-0.dat' matrix 80
0.8 80 0.8 70
70
60
i (A/cm2)

i (A/cm2)

0.6 60 0.6
50 50
0.4 40 40
0.4
30 30
0.2 20 0.2 20
10 10
0.0 0 0.0 0
60 65 70 75 80 85 90 95 60 65 70 75 80 85 90 95
Uf (%) Uf (%)

(e) Scenario V, 14 GWe . (f) Scenario VI, 15 GWe .

Figure 7.13: Heat map plots of optimal operation of SOFCs under different cost assumptions. The
electric installed capacity is reported below each figure. The intensity refers to the frequency (h/a)
of operation in each state.

To find out more about how SOFCs with different cost characteristics are operated, the operating
parameters (i and Uf ) were compared for scenarios I–VI. Figure 7.13 shows how often a SOFC operates
with each combination of i and Uf . These plots show that less expensive fuel cells operate in a more
flexible way, while expensive SOFCs are limited to only a few spots. This change in operation is related
to the installed capacity. At higher costs, the capacity of SOFCs is small, and a high production of H2
is possible. When larger number of SOFCs are installed while H2 demand is unchanged, the cells are
operated to produce more electricity.
Figure 7.13 also indicates that the current density in scenario II and III is limited, which results in
a higher efficiency. The single most used setting in scenario II is at Uf = 60% and i = 0.2 A/cm2 ,

90
entailing a low total output, but a high yield of hydrogen, and low heat production. In scenario IV–VI,
three areas of frequent operation can be discerned: high efficiency hydrogen production (low Uf , low i),
maximum power production (high Uf , intermediate i), and maximum total output (intermediate Uf ,
high i). At lower SOFC costs, operation between these spots occurs more often, indicating a behaviour
responsive towards market circumstances.
The current density does not exceed 0.9 A/cm2 in all scenarios, which is somewhat below the
maximum power point of 1.13 A/cm2 (Figure 3.2). This should not be seen as a guideline for SOFC
design. In fact, imax depends on the assumed ASR value (Eq. 3.23). New cell designs with a lower ASR
can make higher current densities possible. What can be concluded by comparison with Figure 3.2, is
that a further increase of i towards imax is not attractive due to the limited gain of hydrogen and power
output.

7.6 Comparison of concepts

7.6.1 Setup
GHG tax Scenario SOC technologies SOC costs
20/100 €/tCO2 -eq. Diversity SOFC I–VI, see Table 6.4
„ „ „ „ SOFEC „ „
20 €/tCO2 -eq. „ „ SOFC+SOFEC „ „
„ „ „ „ Biogas-fuelled SOFC „ „
„ „ „ „ NG-fuelled SOFEC „ „
„ „ „ „ Hybrid SOC „ „
In § 7.5, SOFCs and SOFECs compete with each other as hydrogen production technologies. To assess
the role of each of these technologies separately, the energy system was optimised in absence of the
competing technology. The other parameters settings were identical to those used in the previous
section.
In addition, three alternative concepts for the application of SOCs were evaluated. One of the
alternative setups is a SOFC with biogas as fuel. This is expected to be an advantage, since biogas is
less expensive than NG. A NG-fuelled SOFEC was also tested next to the setup with a biogas-fuelled
SOFEC. While it seems unattractive to replace the default and more intuitive biogas by NG to assist
electrolysis, the application of SOFECs is no longer bound by the availability of biogas in this case. A
third alternative is a hybrid SOC, which can switch between SOFC and SOFEC mode (§ 2.7.5).
Since it is expected that electrolysis is more attractive in systems with high renewable electricity
production, the optimisation of the system with a SOFC or SOFEC is performed for a moderate and a
high GHG emission tax.

7.6.2 Results
Analysis of the annual hydrogen production (Figure 7.14) confirms most expectations regarding SOFECs
stated above. It is not feasible to use NG as a fuel for a SOFEC, since it negatively affects the break-
even point. Competition with SOFCs indeed hinders the adoption of SOFECs. If the fuel cells are not
available, SOFECs can be selected to produce around 48 PJ H2 , thereby using a large fraction of the
available biogas. It is observed from Figure 7.14 that the cost reductions needed for SOFEC applications
are larger than those needed for SOFCs. Assuming that costs for SOFC and SOFEC stack production
and operation are tightly linked, SOFCs are the more attractive technology.
Additional evidence for the limited economic feasibility of SOFEC applications is provided in Fig-
ure 7.15, which plots the system cost reductions of the investigated scenarios. As would be expected,
larger cost reductions are observed when SOC technologies are less expensive. The installation of
SOFECs leads to negligible savings compared to SOFCs. SOFCs, on the other hand, reduce costs
because they are more efficient, resulting in less resource consumption and less emissions. The same
reduction in emission results in a larger saving at higher GHG taxes, which explains the significant
difference between the graph of 20 and 100 €/tCO2 -eq..

91
 140
Demand
120 SOFC
SOFEC

H2 production (PJ/a)
100 SOFC+SOFEC
80 biogas SOFC
NG SOFEC
60 Hybrid
40
20 €/t CO2-eq.
20 100 €/t CO2-eq.

0
I II III IV V VI
SOC costs

Figure 7.14: Annual production of H2 by different SOC technologies and different cost character-
istics of these technologies.

180

GHG emissions (Mt CO2-eq./a)

170
SOC costs
I II III IV V VI 160
0
SOFC 150
System cost change (M€/a)

-500 SOFEC 140

SOFC+SOFEC 130
-1000 biogas SOFC
120
NG SOFEC
-1500 110
Hybrid
100
-2000 20 €/t CO2-eq. 90
100 €/t CO2-eq. I II III IV V VI
-2500 SOC costs

Figure 7.15: Change in total system costs for scenario I– Figure 7.16: Annual GHG emissions for
VI compared to a system without SOCs, for systems with scenario I–VI and systems with different
different SOC types. SOC types.

The results show that using biogas for SOFCs is feasible, and already at higher SOFC costs compared
to the NG-fuelled concept. What is more, this application of biogas is more effective in reducing system
costs and GHG emissions than fuel-assisted electrolysis. A drawback linked to the use of biogas is that
the full potential of SOFCs cannot be realised. To conclude, using biogas to fuel SOFCs is attractive
for the first installations. At lower cell costs, the number of SOFC can be increased by also employing
NG.
Concerning the hybrid stack, a pattern very similar to systems with a SOFC emerges in Figure 7.15–
7.16. In fact, the curves of a system with only SOFCs, with SOFCs and SOFECs, and with a hybrid
SOC overlap in the plots. Further analysis learns that the hybrid stack is almost exclusively operated in
fuel cell mode. There are no conditions that are favourable for NG-assisted electrolysis, even when it
can be performed by an already installed SOFC. This implies that the capability of reversible operation
is not effective in the investigated energy system.
The GHG emission reductions induced by the different SOC options shows diverging results, in line
with the system cost. Figure 7.16 indicates that the use of biogas by SOFCs or SOFECs induces an
emission reduction of 7 and 2 MtCO2 -eq. respectively. The use of SOFCs, although consuming NG,
has a much larger effect. When a SOFC (or a hybrid stack operating in fuel cell mode) is present, the
emission of up to 50 MtCO2 -eq. can be prevented on a yearly basis. At least when compared to the
emissions of a system with 20 €/tCO2 -eq.. If the emissions are compared with a situation with high
renewable energy shares, an increase is observed, indicating that SOFCs squeeze wind and solar energy
out of the market. Under 100 €/tCO2 -eq. taxes, SOFECs have no noticeable effect on the GHGs

92
emitted.

7.6.3 SOFEC operation

The optimisation algorithm yields the optimal operation of SOFECs based on the assumptions of costs
and technical aspects. To gain insight in how a SOFEC is operated under the market conditions in
the model, the state of SOFECs during a one-year optimisation was analysed. Figure 7.17 shows the
number of hours that a given current density is used, for different SOFEC costs. It can be derived that
the maximum current density is 1.03 A/cm2 . At the high cost level of II, this current density is applied
often. In the other cases, i fluctuates around lower values. The lower the costs, the lower the optimal

1.0 1400
'FRES-SOFEC.dat' matrix
1200
0.8
1000
i (A/cm2)

0.6 800

600
0.4
400
0.2
200
0.0 0
II III IV V VI

Figure 7.17: The frequency (in h/a) of operation at a given i for different SOFEC cost scenarios.
A GHG tax of 100 €/tCO2 -eq. and Diverse demand scenario apply.

current density range. Given that the H2 production of SOFECs only increases (§ 7.14), a lower i
implies the total installed capacity is larger. This is because the higher efficiency of the SOFEC at lower
i can compensate for the higher initial investments. Higher investment costs require a higher annual
production to recover these costs.
Comparison with a SOEC indicates that both technologies are employed rather differently, as il-
lustrated by Figure 7.18. A SOEC is off for around two thirds of the time, and it is operated only
when renewable energy production is high. SOFECs, on the other hand, are always on and supply the
base-load of H2 demand. The variation in output of SOFECs is more limited, and SOFECs are never
turned off.
12000
12000 H2 tank
H2 tank
SMR
SMR
10000 10000 SOEC
SOEC
SOFEC SOFEC
H2 demand
H2 production (MWh)

8000 H2 demand 8000

6000 6000

4000 4000

2000 2000

0 0
25 Jan 27 Jan 29 Jan 31 Jan 2 Feb 4 Feb 25 Jan 27 Jan 29 Jan 31 Jan 2 Feb 4 Feb

(a) 20 €/tCO2 -eq. (b) 100 €/tCO2 -eq.

Figure 7.18: Hydrogen production and consumption during 11 days in an optimal energy sys-
tem with SOFECs with characteristics IV. Overproduction is not lost, but stored in H2 tanks or
consumed by PEMFCs.

93
7.7 Effect of demand composition and GHG taxes
7.7.1 Setup
GHG tax Scenario Restrictions SOC costs
5, 10, 20, 30, 40, 50, 75, 100 €/tCO2 -eq. variable – II, see Table 6.4
In this section, the influence of demand composition and GHG taxes on the optimal energy system are
investigated simultaneously. Both are expected to affect the system to a large extent. Energy demand,
because it determines what amount of each energy carrier has to be delivered. And GHG taxes can
shift the order of which technology is cheaper (merit order). It is expected that higher tax levels make
efficient technologies and carbon-poor energy sources more competitive. SOFCs are expected to be more
effective in some systems than in others. Some combinations of demand patterns of SOFC products
might be more favourable than others. For example, a scenario with alternating peaks in the demand
for hydrogen and electricity might provide opportunities. Yet it is hard to predict in which scenario the
patterns are in favour of SOFCs. It is expectable that more SOFECs are installed in systems with high
renewable electricity production, since there are high electricity production peaks that can be used.
The optimisation algorithm was run for 32 scenarios: all combinations of four demand scenarios
(§ 5.4) and eight emission tax levels between 5 and 100 €/tCO2 -eq.. The costs of SOCs were fixed
to those expected in 2020. The emission taxes works in advantage of less emitting technologies, while
carbon-intensive fuels become less competitive. By comparing the primary energy inputs of the optimal
system at different imposed taxes, the potential emission reductions and their cost-effectiveness can be
assessed.

7.7.2 Effect of GHG taxes

Electricity
As expected, the use of coal – the most carbon-intensive fuel – for heat and power generation deceases
rapidly for increasing taxes (Figure 7.19a and e). Coal powerplants are no longer favoured starting from
20 or 30 €/tCO2 -eq. (depending on the demand scenario), while a tax of at least 75 or 100 €/tCO2 -eq.
is required to replace all coal-fired furnaces. The electricity originating from coal is replaced by a mix of
technologies, among which PV and onshore and offshore wind. The latter technology has the highest
potential (see § 5.5), and a large part of this potential is also used. For onshore wind, the maximum
capacity of 8 GWe is reached in the majority of the scenarios (Figure 7.19b). Electricity derived from
NG initially profits from higher taxes, but becomes less attractive when CO2 is taxed even more. In
particular in the Full electric scenario, a high fraction of electricity is produced by NG powerplants and
SOFCs. The highest potential for gas powerplants is found around 30 to 40 €/tCO2 -eq., while the
maximum SOFC electricity production occurs around 75 €/tCO2 -eq..
The capacity utilisation for three powerplant types is plotted in Figure 7.20. The utilisation of
SOFCs is lies between 60%–80%, which is somewhat lower than that of coal powerplants. For gas
powerplants, the capacity utilisation is much lower and does not exceed 30%. A general trend can be
observed: when more SOFCs are installed, the capacity utilisation of the stacks decreases. In parallel,
their role as flexible backup technology becomes more important.

Heat
Heat derived from biomass has a very low GHG emission. Yet, the assumed costs of biomass technology
restrain the use of biomass boilers up to a 50 €/tCO2 -eq. tax (Figure 7.19e). At that point, biomass is
used to ensure heat supply during the winter months. Solar thermal is insufficient, and SOFCs are too
expensive to be installed only for the peaks in winter.
Considering electric boilers, it is observed that the installed capacity has the largest increase at low
taxes. The capacity is related to the amount of intermittent electricity production. Interestingly, the
total amount of heat produced by electric boilers becomes significant only at a tax level of 75 €/tCO2 -
eq.. Below that, the boilers function mainly as a mechanism to get rid of excess electricity, while

94
 700
PV household
600
Annual production (PJ)

PV commercial
500 Onshore wind
Offshore wind
400 SOFC
300 Coal
Natural gas
200 Biomass
100 PEMFC
Nuclear
0
100

100

100

100
10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75
5

5
BAU Diversity Full electric High hydrogen
GHG emission tax (€/t CO2 -eq.)

(a) Electricity production.

70
PV household
60 PV commercial
Installed capacity (GW)

50 Onshore wind
Offshore wind
40
SOFC
30 Coal
20 Natural gas
Biomass
10
PEMFC
0 Nuclear
100

100

100

100
10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75
5

BAU Diversity Full electric High hydrogen

GHG emission tax (€/t CO2-eq.)

(b) Installed capacities for electricity production.

300
Annual production (PJ)

250
200 SOEC
150 SOFEC
SOFC
100
SMR
50
H2 tank
0
100

100

100

100
10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75
5

BAU Diversity Full electric High hydrogen

GHG emission tax (€/t CO2 -eq.)

(c) Hydrogen production.

25
Installed capacity (GW)

SOEC
20 SOFEC
15 SOFC
SMR
10 H2 tank
5
0
100

100

100

100
10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75
5

BAU Diversity Full electric High hydrogen

GHG emission tax (€/t CO2-eq.)

(d) Installed capacities for hydrogen production.

95
 700
600
Annual production (PJ)

500 Solar thermal

400 Electric boiler
SOFEC
300
SOFC
200 Coal boiler
100 Biomass boiler

0
100

100

100

100
10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75
5

5
BAU Diversity Full electric High hydrogen
GHG emission tax (€/t CO2-eq.)

(e) Heat production.

100
90 Solar thermal
Installed capacity (GW)

80 Electric boiler
70
SOFEC
60
SOFC
50
NG boiler
40
Coal boiler
30
20 Biomass boiler
10
0
100

100

100

100
10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75
5

5
BAU Diversity Full electric High hydrogen
GHG emission tax (€/t CO2-eq.)

(f) Installed capacities for heat production.

Figure 7.19: Annual production and installed capacity for electricity, hydrogen, and heat produc-
tion. Scenarios are a combination of an emission tax and a demand scenario.

100% 100%
Electric output
Coal
80% 80%
SOFC
Capcity utilisation
Capcity utilisation

Gas
60% 60%

40% 40%
H2 output
SOFEC
20% 20%
SOFC
SMR
0% 0%
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
GHG emission tax (€/t CO2-eq.) GHG emission tax (€/t CO2-eq.)

(a) Diversity (b) Full electric

Figure 7.20: Capacity utilisations of three powerplants and three hydrogen production technologies,
for different GHG emission taxes and two demand scenarios.

96
structural use for low-emission heat provision becomes attractive under high taxes. The total electricity
production increases with higher GHG tax, because of the employment of electric boilers to convert
cheap electricity to heat.

Hydrogen
The hydrogen market is affected most by SOCs. For low GHG taxes, H2 is produced via SMR, but
SOFCs can produce it cheaper starting from 20 €/tCO2 -eq. (Figure 7.19c). SMRs are still installed,
and operate as backup facility. When there is an imminent excess of electricity, SOFCs are halted and
H2 production is taken over by SMR. SOFECs play a noticeable role only when taxes exceed 75 €/tCO2 -
eq.. In the High hydrogen scenario, the number of SOFECs is sufficiently large that almost all biogas
is used for fuel-assisted electrolysis. The High hydrogen scenario is also interesting because it shows
the limit of cost-effective hydrogen production by SOFCs. In this scenario, SMRs are responsible for an
important part of the hydrogen supply. Possibly, SOFCs are constrained by the market for electricity that
is co-produced. Even when operated to maximise hydrogen output, SOFCs still produce a considerable
amount of electricity. Therefore, at high shares of intermittent electricity production (as seen in the
systems with 100 €/tCO2 -eq.), SOCs that are operated as electrolysers can replace some SOFCs. The
results indeed show hydrogen production by SOEC and SOFEC components.
By comparing the results of the Diverse demand scenario with those in § 7.5, the effects of cost
reduction and GHG taxes can be compared. The increase from 10 to 20 to 30 €/tCO2 -eq. results
in comparable higher application of SOFCs as the change from characteristics I to II to III. So, a
10 €/tCO2 -eq. higher tax has a similar effect on the break-even point of SOFCs as a 9% lower LCOE.
The capacity utilisations of three hydrogen producing technologies, as plotted in Figure 7.20, show
different patterns for the demand scenarios. These difference have to do with the hydrogen consumption
of PEMFCs, as discussed in § 7.5.2. In the full electric scenario, the hydrogen demand peaks caused
by PEMFCs are large compared to the total demand. Therefore, SOFCs have a low capacity utilisation,
sometimes below 20%. In the diversity scenario, the capacity utilisation of SOFCs is around or above
40% for two reasons. There are SMRs to address hydrogen demand peaks, and the peaks occur more
often, due to the hydrogen demand from FCEVs.

7.7.3 SOFC operation

Heat map plots of SOFC operation in the scenarios treated here were made. A selection of these plots
is provided in Figure 7.21. The most remarkable observation is that a wide range of settings is used
in the Diversity scenario only. In the other scenarios, SOFCs are operated in one or a few settings
most of the time. Regardless of the demand scenario, the fuel cells are operated at low Uf for low
GHG taxes. At the point of 20 €/tCO2 -eq., just past the threshold value for SOFC introduction, Uf is
almost exclusively 60%. This observation supports the hypothesis that H2 production is a driver for the
installation of SOFCs.
At a tax of 75 €/tCO2 -eq. and BAU demand, two states occur most frequently: high power output
mode, and a high efficiency low i mode. A curve of sparsely visited states in between these two modes
is visible in Figure 7.21b. A similar pattern is seen in Figure 7.21f, although the high power mode is
used more often. This reflect the situation with high electricity demand in the Full electric scenario.
By contrast, in the scenario with High hydrogen demand and high GHG taxes (Figure 7.21h) modes
with low Uf dominate, in response to the need for hydrogen. The effects seen in the Diversity scenario
are very similar to what was discussed in § 7.5.4. The wide range of operating conditions used in the
Diversity scenario is partly due to the variability of the H2 demand.

7.7.4 Sustainability
GHG emission reductions
SOC technologies can cause direct GHG emissions reductions (by being more efficient than alternatives)
or induce indirect changes (by changing the attractiveness of renewable energy). To determine what

97
 1.0 4500 1.0 2500
'GE-20-BAU.dat' matrix 4000 'GE-75-BAU.dat' matrix
0.8 3500 0.8 2000
3000

i (A/cm2)
i (A/cm2)

0.6 0.6 1500

2500
0.4 2000 0.4 1000
1500
0.2 1000 0.2 500
500
0.0 0 0.0 0
60 65 70 75 80 85 90 95 60 65 70 75 80 85 90 95

(a) BAU,U20 €/tCO2 -eq. (b) BAU, 75 €/tCO2 -eq.

Uf (%)
f (%)
1.0 2500 1.0 140
'GE-20-Equal.dat' matrix 'GE-75-Equal.dat' matrix
0.8 2000 0.8 120
100
i (A/cm2)

i (A/cm2)
0.6 1500 0.6 80
0.4 1000 0.4 60
40
0.2 500 0.2
20
0.0 0 0.0 0
60 65 70 75 80 85 90 95 60 65 70 75 80 85 90 95
(c) Diversity, 20 €/tCO2 -eq.
Uf (%) (d) Diversity,
Uf (%)75 €/tCO2 -eq.
1.0 5000 1.0 5000
'GE-20-Ele.dat' matrix 4500 'GE-75-Ele.dat' matrix 4500
0.8 4000 0.8 4000
3500 3500
i (A/cm2)
i (A/cm2)

0.6 3000 0.6 3000

2500 2500
0.4 2000 0.4 2000
1500 1500
0.2 1000 0.2 1000
500 500
0.0 0 0.0 0
60 65 70 75 80 85 90 95 60 65 70 75 80 85 90 95

(e) Full electric,

Uf (%)20 €/tCO2 -eq. (f) Full electric,
Uf (%) 75 €/tCO2 -eq.

1.0 2500 1.0 3000

'GE-20-Hyd.dat' matrix 'GE-75-Hyd.dat' matrix
2500
0.8 2000 0.8
2000
i (A/cm2)
i (A/cm2)

0.6 1500 0.6

1500
0.4 1000 0.4
1000
0.2 500 0.2 500

0.0 0 0.0 0
60 65 70 75 80 85 90 95 60 65 70 75 80 85 90 95

Uf (%) Uf (%)

(g) High hydrogen, 20 €/tCO2 -eq. (h) High hydrogen, 75 €/tCO2 -eq.

Figure 7.21: Heat map plots of optimal operation of SOFCs under different demand scenarios and
GHG emission taxes.

98
emission reductions can be assigned to SOCs, the emissions of an optimal system with and without
SOFCs and SOFECs are compared. Figure 7.22 shows the GHG emission savings of a system where
SOCs are available relative to a baseline system without SOCs. The optimisation was performed for
the BAU and Diversity scenario under different emission taxes. The absolute emissions decrease in a
different pace in the different scenarios, due to demand definitions. Yet in both scenarios, the presence
of SOCs reduce the annual emissions by 5.2% on average, corresponding to around 8 MtCO2 -eq. This
amount is saved on top of the savings realised in the baseline case without SOCs.

18%
BAU with SOCs
16%
BAU

Renewable energy share

14%
Diversity with SOCs
12%
Diversity
10%
8%
GHG tax (€/t CO2-eq.)
5 10 20 30 40 50 75 100 6%
GHG emission difference

0% 4%
no SOCs
-4% 2%
average: -5.2% 0%
-8% 0 10 20 30 40 50 60 70 80 90 100
GHG tax (€/t CO2-eq.)
-12% BAU
-16%
Figure 7.23: Share of renewable energy in the
Diversity
total primary energy consumption, for a BAU
Figure 7.22: Difference in GHG emissions between and Diverse demand with and without SOCs
an optimal system with and without SOCs. and different GHG taxes.

Primary energy use

Only direct emission savings occur. As shown in Figure 7.23, the fraction of primary energy coming
from renewable sources does not systematically increase when SOCs are in use. The share of renewables
grows by 1.4% per €10 tax increase on average, independent of the demand scenario or the availability
of SOCs. So, it can be concluded that SOFCs do not make renewable energy more attractive from a
systems point of view. No evidence for indirect GHG emission reductions was found, consistent with
the finding in earlier sections that SOFCs do not facilitate renewable electricity production.
It is not surprising that the share of energy obtained from renewable sources rises for higher GHG
taxes, as confirmed by Figure 7.24. What was not foreseen, is that the highest share of biomass is
achieved in the high hydrogen scenario. Examination of Figure 7.19c learns that the biomass is used for
heat production. The production of biofuels does not become attractive, even at a tax of 100 €/tCO2 -
eq..
Focussing on electricity (Figure 7.24), the highest share of renewable electricity is found in the Full
electric scenario with the highest GHG tax. This scenario has an annual average of 49% renewable
electricity. In the other scenarios, this fraction is similar, but the absolute amount of electricity is much
smaller.

Excess
Dumping of excess energy is allowed for electricity, biogas, hydrogen and heat. The latter two energy
carriers are not wasted in any scenario. Excess electricity is clearly related to the amount of fluctuating
electricity sources installed. It amounts to 9 PJ/a in the Full electric scenario, as visualised in Figure 7.25.
Some biogas is dumped, indicating that the fluctuations in production are not always followed. This is
because it is not economical to oversize the equipment for upgrading to green gas, and cheap electricity
for fuel-assisted electrolysis is not always available.

99
Annual production (PJ) 3500 Biomass products
3000 Biomass residues
2500 Onshore wind
2000 Offshore wind
1500 PV household
PV commercial
1000
Biogas
500
Coal
0 Natural gas

100

100

100

100
10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75
5

5
Nuclear
BAU Diversity Full electric High hydrogen Oil products
GHG emission tax (€/t CO2-eq.)

(a)
800
Annual production (PJ)

700
600
500
400
300
200
100
0
100

100

100

100
10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75
5

5
BAU Diversity Full electric High hydrogen
GHG emission tax (€/t CO2-eq.)

(b)

Figure 7.24: Primary energy sources, a) total and b) renewable energy sources, for the four demand
scenarios and different GHG taxes.

10
Annual excess (PJ)

Electricity
8
Biogas
6
4
2
0
100

100

100

100
10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75

10
20
30
40
50
75
5

BAU Diversity Full electric High hydrogen

GHG emission tax (€/t CO2-eq.)

Figure 7.25: Total annual excess of biogas and electricity in scenario I–VI for different GHG taxes.

7.8 Transition optimisation

7.8.1 Setup
In § 7.4, the effect of introducing SOCs in the current energy system of the Netherlands was investigated,
while in § 7.5–7.7, a completely new future energy system was optimised. This section aims to bridge
the gap between the two, by sketching a pathway for the coming 35 years. Starting with the system
and parameters of 2015, the time period was divided in timesteps of 5 years. After each timestep,
the optimal energy system configuration and operation were determined as in the previous sections.
The parameters related to costs, resource and emission prices, and demand compositions applied in
each step evolve as given in Table 7.5. The assumptions with respect to resource prices are based on
Schoots and Hammingh (2015), and the SOC costs are loosely based on Energistyrelsen and Energinet.dk
(2015). The values of the other parameters in Table 7.5 are intended to represent a gradual, logistic
growth-like, evolution towards a system with low fossil fuel demands and high GHG emission taxes. The
only information exchange between subsequent optimisations concerns the installed capacities from the
previous timestep.

100
 Table 7.5: Assumed properties and values of parameters for each year in the transition.

2015 2020 2025 2030 2035 2040 2045 2050

SOC properties Investment (€/kW) 2750 1375 825 670 544 440 350 275
O&M (€/MWh) 13.75 8 5.5 4.02 2.93 2.2 1.8 1.65
Lifetime (a) 6 10 15 17 19 20 20 20
GHG tax (€/tCO2 -eq.) 5 10 20 35 55 75 86 90
Transport Gasoline 99% 94% 85% 67% 47% 23% 8% 1%
energy demand Electricity 1% 5% 12% 27% 38% 44% 47% 49%
mix Hydrogen 0% 1% 3% 6% 15% 33% 45% 50%
Low-T heat Total 100% 95% 90% 84% 77% 69% 60% 50%
demand mix NG 85% 80% 68% 51% 35% 21% 11% 5%
Electricity 2% 6% 15% 29% 40% 47% 52% 55%
Heat 13% 14% 17% 20% 25% 32% 37% 40%
Resource prices Oil 22.25 39.60 50.73 62.30 66.75 68.97 71.20 73.42
(€/MWh) NG 24.84 28.66 31.73 33.78 35.83 37.87 39.92 41.97
Coal 9.00 11.15 11.70 12.12 12.53 12.94 13.36 13.77

To accurately model a transition over the timespan of several decades, it is important to know when
powerplants are expected to be depreciated. Data from three sources (Wikipedia, 2016c; Davis et al.,
2015; CBS, 2016b) were combined, resulting in the overview of Appendix A.3. Powerplants fuelled with
coke oven gas, fuel oil or hydro-power sources were not included. These plants combined represent
1 GWe , or 3.5% of the country-wide installed capacity. The installed capacity of MSW plants were
added to the capacity of NG plants.

7.8.2 Results
In Figure 7.26, the annual production of electricity, hydrogen, and heat as found by the optimisations are
presented. In the first 10 years, few changes occur in the energy system. In 2030, more wind turbines
and PV panels are installed to replace old coal powerplants. The GHG tax of 35 €/tCO2 -eq. facilitates
this shift. Around 940 kW of SOFCs and 150 kW SOFECs (kW H2 output) are installed. Driven by the
increasing H2 demand, the installed capacity of SOFC and SOFEC installations expands rapidly in the
years after that. In 2040, also 8.6 PJ of H2 is produced by SOECs. With respect to heat production,
it is observed that electric boilers gain importance over the years. These boilers are operated to limit
electricity excesses.
This optimisation shows how the initial diffusion of SOFC technology is hindered by existing tech-
nology. Although in the setup used here, SOFCs are competitive in 2025, their introduction starts only
after that. SMR installations and powerplants that have not reached the end of their service life are
more cost-effective than investments in construction of new SOFCs. These results thus indicate that a
barrier for the adoption of SOFC technology exists, providing a dilemma. To be competitive at large
scale, substantial cost reductions are needed. Yet for these reductions to occur, more systems have to
be installed to gain experience. This dilemma and possible solutions are discussed in Chapter 8.

101
 600 60

500 50
PV commercial

Installed capacity (GW)

Annual production (PJ)

PV household
400 40
Onshore wind
300 Offshore wind 30
SOFC
200 Gas 20
Coal
100 Biomass 10
Nuclear
0 0
2015 2020 2025 2030 2035 2040 2015 2020 2025 2030 2035 2040

(a) Annual electricity production. (b) Installed capacities of powerplants.

12

140 10
Installed capacity (GW)

120
Annual production (PJ)

8
100
80 SOEC 6
SOFC
60 4
SOFEC
40 SMR
2
20
0 0
2015 2020 2025 2030 2035 2040 2015 2020 2025 2030 2035 2040

(c) Annual hydrogen production. (d) Installed capacities of H2 plants.

50
45
300
40
Installed capacity (GW)

250 35
Annual production (PJ)

Solar thermal 30
200
Electric boiler 25
150 SOFEC 20
SOFC
100 15
Coal boiler
10
50
5
0 0
2015 2020 2025 2030 2035 2040 2015 2020 2025 2030 2035 2040

(e) Annual heat production. (f) Installed capacities of heat generators.

Figure 7.26: Annual production of electricity, hydrogen and heat, and installed capacities of gen-
eration technologies during a transition path. Further assumptions: increasing emission taxes and
demand from FCEVs and BEVs, decreasing heat demand and SOC costs; see Table 7.5.

102
7.9 Conclusion
The results clearly indicate that SOFCs can become an important part of the Dutch energy system if
the costs decrease sufficiently and a longer lifetime of the stacks is achieved. An installed capacity of
7.5 GWe (High hydrogen, 50 €/tCO2 -eq.) to 15 GWe (Diverse demand, 20 €/tCO2 -eq.) is optimal.
SOFCs are cost-effective because of the co-production of three energy carriers, of which H2 is most
valuable. The high efficiency, resulting in a lower resource use, also contributes to the effectiveness of
SOFCs. The installation of SOFCs results an emission reduction of around 8 MtCO2 -eq./a compared
to a system without them. High GHG emission taxes promote fuel-efficient technologies, and therefore
favour the diffusion of SOFCs. SOFC are especially good to facilitate a transition from coal to NG, at
the same time decreasing the primary energy need. If GHG taxes are low, then renewable electricity
sources become less attractive compared to SOFCs. At very high GHG taxes, the balance shifts towards
renewables (see Figure 7.19b).
If the costs of operating SOFCs remain high, then the installed capacity is limited to an extent for
which high capacity utilisations and a high production of H2 can be achieved. If costs can be reduced,
SOFCs prove to be a suitable technology for electricity production during periods of low renewable
electricity production. Analysis of the operating parameters Uf and i indicated that expensive SOFCs
are operated to maximise H2 production, whereas cheaper cells are operated more flexible.
The conditions for which SOFECs can play a role in the Dutch energy system are more limited. A
high level of renewable electricity production and/or high GHG emission taxes are required, on top of the
cost reductions of SOC technology. Furthermore, the competition with SOFC and SOEC technologies
as a H2 source, and the limited availability of biogas further limit the application of SOFECs. If SOFCs
and SOFECs with the same costs are both available, a SOFC is often preferred, since it allows for
much larger system cost reductions. The operational dispatch of SOFECs shows a trend similar to
SOFC operation. For expensive installations, the H2 output is maximised, even though this entails a
lower efficiency. At lower costs, a higher installed capacity is operated at lower rates. In general, a
SOFEC is operated optimally with a high capacity utilisation. This has two important implications.
First, the variations in current density during operation are limited, which is favourable for the thermal
management. Second, a SOFEC shows a limited response to electricity prices. As opposed to SOECs,
they are therefore not ideal to prevent electricity excesses. Nonetheless, SOFECs are more favourable
under high GHG taxes. This is because the technology profits from low average electricity prices and it
uses (almost) carbon neutral biogas.
The optimisation results of the last section indicate that the diffusion of SOCs is hampered initially
by existing technologies. Over time, when some old installations are dismantled and SOC costs have
decreased, SOFCs and SOFECs become cost-effective.

103
Chapter 8

Societal Aspects

8.1 Analytical framework

8.1.1 Socio-technical systems
The energy system was introduced in § 4.1.1 as a complex system composed of many technical com-
ponents. Here, the perspective is broadened by acknowledging that also a wide variety of actors (or-
ganisations and individuals) is involved. The attitude of these actors, emerging network structures, and
interactions such as learning processes influence how the system will evolve. In particular when tran-
sitions or transformations are considered, these social effects are of big importance for the outcomes.
For SOC technology, this implies that social dynamics influence their development and diffusion in the
Netherlands. To be able to understand the possible future developments around SOCs, it is instructive
to use a number of concepts and an analytical framework. With respect to the future costs, learn-
ing curves and economies of scale are relevant concepts. The concepts of path dependence and actor
perspectives are linked to the social side of the transition. The following sections will elaborate on
these concepts. To analyse and asses the societal aspects of the adoption of SOFCs and SOFECs, the
multi-level perspective (MLP) will be used. This analytical framework will be introduced in § 8.1.5.

8.1.2 Learning curves

One of the shortcomings of the optimisation model, is that it does not address the effect of learn-
ing, which is one of the major sources of cost reduction for emerging technologies. Learning curves
describe the relation between cumulative installed capacity of a technology and the associated costs
(Rivera-Tinoco et al., 2012). For other energy technologies, a linear relationship is often found between
production cost and cumulative production on a double-logarithmic scale. This implies that the costs
have decreased by a fixed fraction each time the cumulative output doubles (Staffell & Green, 2013).
Experience can be gained through three processes, namely learning-by-doing (during production and in-
stallation processes), learning-by-searching (during R&D projects), and technology spillover (knowledge
coming from other sectors) (Rivera-Tinoco et al., 2012). The learning rate of SOFC systems has been
estimated by a few studies. Staffell and Green (2013) has determined the learning rate of micro-CHP
SOFCs in the Enefarm pilot project in Japan (2004–2012) at 20.6% for the fuel cell stack and 11.7% for
the BoP. Rivera-Tinoco et al. (2012) determined learning rates of the R&D phase in the range 13–17%,
and 1–27% for the pilot and early commercial phase.
For comparison, the LCOE produced by wind turbines and PV have decreases by 61% and 82%,
respectively, between 2009 and 2015 (Lazard, 2015a). These cost reductions are driven by the rapid
increase of installed capacities of both technologies.
Since learning is only one of the mechanisms by which costs decrease, it can be difficult to distinguish
its exact effect. Other cost reduction sources include economies of scale, automation, and raw material
prices (Rivera-Tinoco et al., 2012). The former effect is discussed in the next section.

104
8.1.3 Economies of scale
The more fuel cell systems a factory produces, the lower the costs of each individual system. So by
scaling up production, large cost reductions can be achieved. This can result in a positive feedback
loop: lower prices will stimulate demand, which makes further upscaling possible. Besides, larger SOC
systems are generally less expensive than small systems in a relative sense. The effect of economy of
scale on SOFC production was quantified by (James et al., 2012). Table 8.1 and 8.2 illustrate the
dependence of the fuel cell price on the production volume and the size of the fuel cell system itself.
The data in the first table show a higher scale advantage than the second.

Table 8.1: SOFC system production costs in Table 8.2: SOFC system production costs in
$/kWe (James et al., 2012). $/kWe , excluding sales markup (Batelle, 2014).

Production Fuel cell size (kWe ) Production Fuel cell size (kWe )
(systems/a) 1 5 25 100 (systems/a) 1 5 100 250
100 11,830 3,264 981 532 100 15,419 3,608 1,517 1,194
1,000 6,786 2,168 671 440 1,000 9,661 2,416 1,194 973
10,000 5,619 1,862 599 414 10,000 8,381 2,108 1,044 857
50,000 5,108 1,709 570 402 50,000 8,347 2,104 962 787

8.1.4 Actor perspectives

CE Delft has studied the transition pathways to a CO2 -emission free heat supply in the residential sector
(Schepers et al., 2015). Some conclusions regarding the transition in heat supply are also applicable
to the case of SOFCs. One important finding is related to the resistance to change of actors. In the
study, a number of barriers were identified that constrain a transition. These barriers relate to how the
system is perceived by the actors. First of all, stakeholders have a lack of information. It is not known
throughout the sector which strategy is financially and technically most optimal. Secondly, there is a
variety of perspectives among actors, in terms of scale and time horizon. Depending on their perception,
each actor sets different targets. Thirdly, the ideal moment of making certain ‘switches’ (e.g. replacing
a NG distribution network by a district heating system) is different for each stakeholder. These three
challenges can be addressed by a close cooperation of the actors involved, a clear division of roles and
responsibilities, and a shared vision of the aim of the transition (Schepers et al., 2015).

8.1.5 Multi-level perspective

The MLP was developed in the field of innovation studies, and it aims to guide the analysis of tech-
nological innovations and societal transitions (Geels, 2002, 2012). The framework distinguishes three
analytical levels: socio-technical landscape, socio-technical regimes and niche innovations. Together the
three levels can be understood as a nested hierarchy (Geels, 2002). The MLP acknowledges that the
success of a new technology depends on processes at all three levels (Geels, 2002). Therefore, a better
understanding of socio-technical changes can be achieved by assessing the dynamics of the landscape,
regime, and niche. These concepts are elaborated below.

Socio-technical landscape
The socio-technical landscape refers to the deep structure of society in which niches and regimes are
embedded. The landscape trends provides an environment and boundary conditions that are beyond the
direct influence of niche and regime actors (Geels & Schot, 2007). Several examples of such landscape-
factors are cultural and environmental patterns, economic paradigms, demography, wars and political
developments. Landscape characteristics typically change at a slow pace, although sudden unexpected
changes such as natural disasters can occur (Geels & Schot, 2007) or that. For the latter one, this
means that this change in the landscape can put pressure on the existing regime creating windows of
opportunity for niche innovations to break through into the regime and change (Geels, 2010).

105
Socio-technical regime
A regime is an established socio-technical system in which technologies, infrastructures, regulations,
user patterns, knowledge, and cultural discourses are aligned. The actors that are embedded in a regime
reproduce and maintain the rules and patterns of the regime. The stability of the regime is a result
of coordination and alignment of orientation between actors. Changes do occur, but innovations often
proceed slowly and in a predictable direction. Often, incremental improvements dominate (Geels, 2012).
In spite of its stability, a regime can face pressure arising from developments at the other two levels
(Geels & Schot, 2007)

Niche
Niches are the third and smallest level of the MLP. This is where radical innovations emerge and
develop (Geels, 2002). A niche is a protected space, which is shielded from selection pressures governing
dynamics at the regime level (Smith & Raven, 2012). Protection is provided by subsidies, demonstration
projects, geographic conditions, or user groups with special needs. Within a niche, actors can learn in
various dimensions, articulate expectations or visions, and build social networks (Kemp et al., 1998).
These processes are important for growth of the niche. By voicing criticism on the structural problems
of the regime and demonstrating alternative solutions, niche actors can exert pressure.
Both niches and regimes can be described by and are characterised by their nature in six socio-
technical dimensions (Verbong & Geels, 2007; Geels, 2012; Smith et al., 2005):

• Industry structure
• Technology and infrastructure
• Knowledge base
• Markets and user preferences
• Policies and politics
• Cultural meaning

In a stable socio-technical regime, these dimensions are well-aligned, whereas they are subject to a
dynamic co-evolution in niches. The characteristics that prevail in the regime form a selection pressure.
Within niches, the selection pressures are less strict, which can be explained by an effect of shielding
(Smith & Raven, 2012).

8.2 Actor identification

In Figure 8.1, the following (groups of) actors that are connected to the electricity and gas supply chain
are identified.

Energy producers are companies that exploit an energy (re)source. Several powerplant operators
are active in the Netherlands, such as Eneco, E.ON, and Essent. Some of them also own renewable
energy plants, such as wind parks. Fossil fuels are mostly imported. An exception is NG, which is
extracted from the domestic reserves by the Nederlandse Aardolie Maatschappij (NAM).

TSOs operate the national transmission systems. TSOs are monopolists, and are fully owned by the
government. TenneT is the TSO of the high-voltage electricity network. Gasunie is the TSO of the
high-pressure gas transmission network. Their task is to match supply and demand on both short and
long term. Besides, TSOs are responsible for the maintenance and management of the transmission
system infrastructure.

DSOs ensure that electricity and gas are distributed to all individual consumers. They operate,
maintain, and expand the distribution systems of both electricity and gas, each in a designated area. It
is enshrined in law that DSOs have to be independent organisations, without commercial connections

106
 Research Energy
Policy makers NGOs

Facilitators
institutes companies

Technology Watchdogs Aggregators Installation &

suppliers Inspectors Energy matchers maintenance
companies

Households
Energy
companies
Businesses
Owners

Energy Local energy

TSO DSO Industy
producers cooperatives
Infrastructure

CH4

Energy Transmission Distribution Distributed

Consumption
production network network generation

Figure 8.1: Physical energy infrastructure and connected actors. The figure applies to both elec-
tricity and natural gas supply chains. Illustration adapted from EDF Energy (2016).

with energy producers or retailers. The shareholders of all DSOs are municipalities and provinces. The
main interest of DSOs is to guarantee safe and reliable energy distribution.

Measurement and maintenance firms perform repair and construction works, and are respon-
sible for monitoring the consumption of energy users. These firms are typically part of the organisation
of a DSO. During construction works, external companies and suppliers can be involved.

Local energy initiatives comprise various actors that own or operate distributed energy genera-
tion technologies. Examples are farmers, waste-water treatment plants (WWTs) or waste treatment
companies that produce biogas or green gas for local use. Or solar farms and wind turbines owned by
a community in the form of a cooperative (Van Bruggen et al., 2016). The installations used have a
relatively small size.

Energy consumers include all entities that use energy. This diverse group of actors can be divided
in three groups: households, businesses, and industries. The interaction of these consumers with the
energy system depends on the type and amount of energy carriers they need and their demand pattern.
Furthermore, local energy production, storage, or demand response at the consumer is expected to
become increasingly important (Bokhoven et al., 2015).

Energy retailers are the intermediates between the energy wholesale market and the consumer
market. They negotiate contracts with both energy producers and consumers. A number of retailers is
also an energy producer itself.

107
Energiekamer is part of the Authority for Consumers and Markets (ACM), a governmental organ-
isation that monitors market entities. The Energiekamer monitors energy market players to make sure
that laws are adhered to.

Governmental bodies include the national administration, provinces, water boards, and municipal-
ities. Their policies and regulations provide market actors and consumers with incentives, restrictions,
and guidelines for their behaviour.

Interest groups and NGOs (non-governmental organisations) have the power to lobby and influ-
ence policies. They can play an important role in the mobilisation of resistance to a technology, or by
providing legitimacy.

Technology suppliers and manufacturers develop and market technologies that are part of the
energy system. For conventional technologies, suppliers are often large international companies. But
also for emerging techniques, it proves important for companies to achieve substantial economies of
scale (see § 8.1.3). An overview of SOFC technology suppliers is provided in Appendix B.1. Technology
suppliers are interested in creating a large market share for their products.

Research institutes provide knowledge and new technological opportunities, which can form the
basis of improved or new energy technologies. In the field of energy, the Organisation for Applied
Scientific Research (TNO), the Energy Research Centre of the Netherlands (ECN), and the universities
contribute significantly to the body of knowledge.

8.3 Regime description

In the MLP framework (§ 8.1.5), both niches and regimes can be characterised by taking six socio-
technical dimensions into account, i.e. industry structure, technology and infrastructure, knowledge
base, markets and user preferences, policies, and cultural meaning. This section focusses on the regime
level.
The regime of the existing energy system in the Netherlands uses a linear model for the industry
structure. Central production is connected with consumers via wholesale markets and energy retailers,
as depicted in Figure 8.1. Consumers pay a fixed price for the consumed energy plus a fixed fee for using
the transmission and distribution networks. Producers adapt to the demand in order to balance the
grid. The paradigm of economies of scale (“bigger plants are more efficient") further governs policies,
technologies, and knowledge base. R&D aims at gradual improvements. Energy is culturally seen as an
unconditionally available commodity.
According to the MLP framework, pressures on the socio-technical regime arise from landscape and
niche developments. Indeed, ongoing developments outside and within the energy system are exerting
pressure on the dominant regime. These pressures affect the dynamics within the regime, as discussed
below. It is relevant to discuss changes in the regime, since regime actors in turn influence niche
dynamics, treated in the next section.
An important factor is the continuous trend of increasing renewable electricity production, supported
by actors within the regime and several niches. As these fluctuating electricity sources become available,
the flexibility of the electricity market is reduced. In fact, adjustable fossil fuel plants are replaced by
sources with an uncontrolled output. The business case of remaining powerplants is undermined because
the capacity utilisation declines and it is expensive to ramp the plants up and down frequently (Van den
Bergh et al., 2013). From an infrastructural point of view, this means that other or new components
of the energy system should compensate for the loss of flexibility. Examples include storage, demand
response, etc. Taking a regulatory perspective, it might be needed to provide incentives for small actors
to provide flexibility. Besides, the role of large players can be redefined by explicating their tasks and
responsibilities under the new circumstances.

108
 Currently, incumbents agree that structural changes are necessary, as witnessed by their commitment
to the ‘Energieakkoord’ (Sociaal-Economische Raad, 2013) and a common position paper (Bokhoven
et al., 2015). It is currently unclear who should take actions. DSOs see the urgency of system change
most clearly, since they are aware of a changing energy demand and the impact on energy network
capacity requirements. Grid refurbishments carried out now should be ready for the energy system in
2050 (Schepers et al., 2015). The question is whether DSOs should be the only actors that deliver
flexibility services, and whether grid balancing fits in the role of a grid operator. Other organisations
that could also bear (part of) the responsibility are energy producers and energy cooperatives. Both
have operational experience and knowledge through their ownership of existing generation technologies.
Therefore, they might also operate energy storage or additional conversion systems (batteries, fuel
cells, PtG) to balance energy grids. Whereas energy producers have access to substantial resources to
undertake this action, local cooperatives might be better geared towards the management of distributed
technologies (Van Bruggen et al., 2016).

8.4 SOC niche innovation

8.4.1 Current status
Many resources are invested in research to improve SOC technology. An indication for this is given by
the growing number of publications with the subject of SOFCs and SOECs, as presented in Figure 8.2.
The R&D efforts have resulted in the sales of FCs growing almost every year (Figure 8.3). In 2015,
63.1 MW of SOFCs were sold. This amount is marginal compared to the terawatts of powerplants or
billions of boilers installed globally. Looking in more detail at the capacity installed annually, it becomes
clear that stationary FCs represent 60% of all FC types sold (Hart et al., 2015). Although small in
capacity, the number of FCs applied in transport applications is big. The market for FCEVs is expected
to continue growing in the near future (Hart et al., 2015), providing an increasing demand for hydrogen.
Additionally, buses, trains, military vehicles, and material handling vehicles fuelled by H2 have been
demonstrated.

350

300 75,6

250
63,1
Shipments (MW)

200 MCFC
24
SOFC
150 PAFC
PEMFC
100 179,6
50

0
2010 2011 2012 2013 2014 2015

Figure 8.2: Number of articles pub- Figure 8.3: Annual fuel cell capacity shipped worldwide, in
lished each year on SOFC and SOEC the period 2010–2015 (Carter et al., 2013; Hart et al., 2015).
technology according to the Sco- Amounts are decomposed into FC types: Molten carbonate
pus ® database. Reprinted from FC (MCFC), Solid oxide FC (SOFC), Phosphoric acid FC
Gómez and Hotza (2016). (PAFC), Proton exchange membrane FC (PEMFC).

Half of the twenty SOFC manufacturers are located in Europe, while the other firms are based in
Japan and the United States (Appendix B.1).

109
8.4.2 Potential system changes
Contrasting the characteristics of the existing regime with a system in which SOCs play a role, a number
of differences can be identified. The most important difference is that SOCs are a distributed generation
technology, for which other industry structures are more supportive than the current structure (Sauter &
Watson, 2007). Central energy producers and TSOs play a much smaller role in distributed generation
systems. This also affects retailers, as intermediate players between consumers and central producers.
And the role of DSOs and energy retailers changes considerably if a distributed energy production
structure is implemented.
New SOC technology involves new technology suppliers, and different requirements for the skills
and knowledge offered by installation and maintenance firm. In terms of policy and technology, SOFCs
and SOFECs profit when regulations and infrastructure accommodate for local grid balancing and
exchange. Markets that are of interest are those where user preferences are focussed on energy savings,
increased energy independence, and deployment of novel technology. User preferences with respect to
the deployment model of SOCs are not yet clear. Possibly, consumers take care of their own device,
or self-organise in local cooperatives to share responsibilities. For some consumers, the preference for
energy as a service can prevail, in which case energy retailers can offer their services to operate and
maintain SOC installations.
Finally, a certain alignment or cooperation with facilitating actors (see Figure 8.1) in the form
of regulations and complementary technology can provide support. If distributed generation becomes
the dominant practice in the energy system, a number of changes in the regulatory system should be
accomplished. Energy taxation legislation has to be adapted to the new situation, taking into account
the decline of tax incomes from energy retail. In addition, a coordination structure to control all
distributed generation technologies should be implemented to prevent problematic system dynamics.
For instance, price mechanisms or control via a central computer make sure that the energy networks
are balanced properly.

8.5 Barriers & Drivers

This section discusses the drivers and barriers related to the diffusion and use of SOCs. Drivers can be
internal strengths of the technology and niche, or opportunities provided by external regime or landscape
dynamics. The same is true for barriers, which can be internal weaknesses or external threats.

8.5.1 Technology & infrastructure

SOCs are made of cheap and abundant raw materials. Traces of rare earth elements (La, Sc, Y, Ce) are
needed for manufacturing, but the small quantities are unlikely to cause a scarcity, even at large-scale
application of the technology (Thijssen, 2011). The reasons for most current customers to install SOFC
products are the high reliability and low maintenance requirements. SOFCs are therefore an excellent
choice at remote areas or as a backup power source (Curtin & Gangi, 2015; Carter et al., 2013). In
applications for residential energy supply, the high efficiency conversion of fuel to high-value energy
products is an additional strength of SOC technology.
Technological barriers that have to be overcome are degradation issues of stack materials. The
effects of high temperatures and redox environments on materials within the cell are still considerable,
resulting in short lifetimes. Another material-related weakness is the sensitivity to impurities, which is
currently an important constraint for fuel selection. If SOCs with higher sulfur-tolerance are developed,
the opportunity for efficient biogas use without cleaning processes in SOFEC or SOFC applications
emerges. See also Chapter 2.
An important barrier is related to energy transport infrastructure. While the infrastructure to trans-
port NG and electricity to and from most locations is well-established, possibilities for heat, H2 , and
alternative fuel transport are much more limited. A fuel should be available at the location of the fuel
cell installation. NG meets this requirement, since gas pipelines are present on many places, providing
a reliable fuel source with a constant quality. As noted in § 2.1.4, biogas is another fuel option. Due to

110
the limited technical possibilities for storage and transport, it is advised to convert or use biogas at its
origin. Blast furnace or coke oven gas can serve as fuel too. Pipeline infrastructure for transport over
small distances exists in some areas, or can be constructed. Given the limited importance of hydrogen
in the current energy system, transportation infrastructure is limited. Users of small volumes of H2 are
supplied by trucks transporting compressed and/or cooled H2 . Large consumers usually own an on-site
H2 source. At fuel stations, H2 has to be compressed before it is transferred to FCEVs. Therefore, it can
be considered to compress the H2 at its origin, allowing transport by truck (Chen, 2006). A distribution
infrastructure to supply FCEVs is virtually non-existent.

8.5.2 Markets & user preferences

Although the high efficiency and cheap raw materials (see § 8.5.1) are financial drivers of SOC de-
velopment, there are also a number of weaknesses. The capital-intensive production process of SOCs
and the BoP (James et al., 2012), together with the limited quality in terms of service life, limit the
number of current market applications. Furthermore, significant expenses must be made due to network
connection fees paid to the DSO (Prag et al., 2016). Medium or large-scale installations require a
special connection to the gas distribution network, unless local biogas is used and a backup fuel source
is not desired. To deliver electricity to the grid, another grid connection fee has to be paid.
Market opportunities are provided by experiments with flexible electricity prices, to which SOCs can
respond by changing the ratio of outputs or operate at lower rates. A R-SOC has the additional advan-
tage of being able to switch from electricity producing to electricity consuming mode. In combination
with storage of H2 , a reversible cell can respond to fluctuating electricity prices. Both the response time
and operating window of the technology are important here.
In the future, market circumstances with high amounts of renewable electricity, combined with a
flexible pricing mechanism, can provide a window of opportunity for SOCs. The same hold for feed-in
tariffs, which could provide a substantial stimulant for the adoption of SOFCs.
Compared to conventional gas-fired technologies, SOCs have lower emissions, both because of the
higher efficiency and the virtual absence of non-CO2 emissions. Amongst some groups of users, there
is a willingness to pay more for energy-saving devices. A bigger opportunity might be provided by
technology frontrunners, for whom a novel device is an expression of their attitude, and costs are less
important (Sauter & Watson, 2007).

8.5.3 Policy & politics

International agreements concerning climate change mitigation signed by the Dutch government indicate
political commitment to stimulate technology with high energy efficiency and low emissions. Several
(inter)national and regional subsidy options for research on and implementation of fuel cells exist. Ex-
amples include EU funding for research projects, the SDE+ subsidy, Green Deal agreements, and local
stimulation policies. SOFEC projects can apply to existing subsidy programmes for renewable energy.
SOFC installations can be supported by schemes for industrial efficiency improvement. Grants from the
European Fuel Cell and Hydrogen Joint undertaking (FCH-JU) are available for both technologies. An-
other example of fuel cell-specific policy is the local stimulation of micro-CHP installations in Gelderland
(Energy Company, 2015). Within Green Deals, the government helps to remove obstacles by bringing
actors together and adapting legislation (Ministry of Economic Affairs, 2011).
Acquiring a permit for supplying electricity to the grid does not only involve paying a connection fee
(§ 8.5.2), but also a lengthy and bureaucratic application procedure (Prag et al., 2016). In addition,
SOC manufacturers, which are typically international companies, face a wide variety of standards and
(safety) regulations. On a national level, not all regulations accommodate for fuel cells. A total of 13
norms were identified that apply to micro-CHP systems (Prag et al., 2016). When hydrogen is involved,
an additional set of regulations apply to guarantee the safety of the installation.
Goals that guide the energy transition in the Netherlands were formulated in the 2013 agreement
‘Energy Agreement for Sustainable Growth’ (Sociaal-Economische Raad, 2013), which was supported
by many (non-)governmental organisations and companies. The agreement aims to achieve a ‘climate

111
neutral’ energy supply of the built environment in 2050. This goal cannot be achieved when NG fuelled
SOCs are installed at a large scale.

8.5.4 Cultural meaning

The cultural associations related to SOCs, and behavioural patterns can either constrain or support
their diffusion. A distributed SOC system is a way of local energy production that reduces the need
for centrally produced electricity. This idea appeals to several communities, and to energy production
companies that seek to increase legitimacy in their customer relations. The use of biogas can increase the
level of autarky (self-sufficiency) of a community. These considerations can prove to be an argument for
households or energy cooperatives to choose for a SOC. On the other hand, an unknown new technology
often faces the challenge of lacking awareness, proving its safety and demonstrating its merits to the
general public. The high temperatures and the risk of hydrogen leaks make that SOCs could be perceived
as an unsafe or risky technology.
The fact that SOFCs are a net emitter of CO2 if NG is used as fuel is a potential threat. Life cycle
assessment of SOFC use show that the largest impacts are linked to fuel supply (Strazza et al., 2015).
If GHG emissions are an important criterion, then solar thermal or PV panels combined with energy
storage could be a more attractive alternative for a household.

8.5.5 Industry structure

As discussed in § 8.3, the pressure to modify the structure of the energy supply chain is growing to
achieve a more sustainable energy provision. If the industry structure is transformed, opportunities for
the application of SOFC and SOFEC technology are created. SOFCs can fill gaps of low electricity pro-
duction from wind and solar energy. SOFECs can respond to peaks in renewable electricity production.
Furthermore, SOC installations are scalable sources of hydrogen, and can facilitate the development of
H2 as a transport fuel. Yet if the absence of hydrogen storage and dispensing infrastructure persists in
the future, the exploitation of the full (flexibility) potential of SOCs is hampered.
The most successful business models featuring SOCs can be developed if the energy supply chains
are decentralised more. This is because SOCs help to balance of production and consumption at a local
level. What is more, SOC installations can maximise the useful application of energy products when
they are distributed instead of located centrally. This requires an industry structure that differs from
that of the current regime. For example, district heating systems are not open to heat suppliers other
than the central CHP unit (Van Melle et al., 2015).
SOCs need a user of the produced hydrogen. If new applications for H2 appear, this provides an
opening for new technologies. This is because the existing industry structure does not provide an
adequate solution for the supply to these new users.

8.5.6 Knowledge base

The complexity of the production process of SOCs makes that only specialised companies succeed in
doing so. A considerable amount of theoretical and operational knowledge is vital for these firms. Since
no manufacturers are located in the Netherlands, the knowledge base for SOC technology is assessed
within the scope of Europe.
It is noted that several SOFC manufacturers are active in Europe (see Appendix B.1), all of which
drive knowledge development by investing in research. As indicated by Figure 8.2, the scientific com-
munity contributes to knowledge development of SOC technology. Also in the Netherlands, there is a
strong basis for R&D. Table 8.3 lists all organisations that have participated in research projects related
to SOFC technology.
The expectations of stakeholders influence how much resources are invested in R&D and on what
aspect the research is focussed. Currently, the majority of the European SOFC manufacturers see micro-
CHP as the biggest potential application. At least six firms are actively marketing SOFC systems aimed

112
Table 8.3: List of Dutch organisations that participated in EU-funded research projects (as reg-
istered in Cordis (European Commission, 2016)) started between 1990 and 2016, on the topic of
SOFCs. A full list of the projects in which these organisations participated is included in Ap-
pendix B.2.

Organisation Number of Start of most

projects recent project
ECN 18 2010
HyGear Fuel Cell Systems BV 6 2015
TNO 4 1993
Gastec NV 3 1998
Technische Universiteit Delft 2 2015
Technische Universiteit Eindhoven 2 2015
University of Twente 2 2003
Advanced Lightweight Engineering BV 1 2002
Brandstofcel Nederland BV 1 1998
Business Unit of TNO Built Environment and Geosciences 1 1993
Centre for Concepts in Mechatronics BV (CCM) 1 2005
DSM Resins NV 1 1994
Energy Matters BV 1 2015
Gasterra BV 1 2015
HOMA Software BV 1 2011
Micro Turbine Technology BV 1 2015
Netherlands Agency for Energy and the Environment 1 1998
(NOVEM)
Shell Hydrogen BV 1 2003
Universiteit Utrecht 1 2015

at small consumers: Sunfire, Viessmann1 , SOLIDPower, Ceres Power, Bosch and Elcogen (Hart et al.,
2015). This trend affects the setup of recent research projects, such as ene.field, SOFT-PACT and PACE
(European Commission, 2016), all aiming at the installation of SOFC-based micro-CHPs that replace a
conventional boiler. This approach is supported by the EU, which subsidises all three projects. Another
example of government support is the subsidy granted by the province of Gelderland to households that
install a micro-CHP (Energy Company, 2015).
While the observed mobilisation of resources positively affects the development of SOC technology,
the focus on micro-CHP instead of CHHP applications could hamper the latter. The pitfall exists that,
under influence of regime actors, SOCs are used as a direct replacement of conventional heating systems.
Consequently, less research efforts are invested in the capability to co-produce H2 , thereby cutting off
the road to larger system changes.

8.6 Energy carrier markets

In addition to the identification of markets and user preferences in § 8.5.2, this section provides an
analysis of the markets for energy carriers. When no or limited energy transport options exist, the
presence of a nearby source or demand becomes a boundary condition for the application of SOCs.
To identify market niches where the application of SOFCs results in a viable business case, a range
of cost aspects has to be considered. From the energy system perspective adopted in the optimisation
model, only investment, O&M, and fuel costs are important. For specific applications, also taxes and
sales prices have to be taken into account. These factors show large variations between cases. For
instance, a lower percentage of taxes has to be paid if larger amounts of NG are consumed. And small
1
Viessmann has acquired the SOFC manufacturer Hexis to safeguard its position in the fuel cell market (Hart et
al., 2015).

113
consumers are willing to pay more for the SOFC products, because the alternative energy sources are
more expensive for them than for large consumers (Van Leeuwen, 2016). The business case calculations
done by Van Leeuwen (2016) will not be repeated here. Instead, the emphasis is put on the qualitative
local market conditions of energy carriers and their role in creating a successful niche.

Natural gas: Large variations in NG costs exist. Bulk consumers pay much lower taxes per volume
of the fuel than small consumers. Due to several developments, the price is expected to rise in
the future (see Table 6.1).

Biogas: Possible sites for biogas-fuelled SOCs are at farms with a digestion tank, WWTs, or at organic
waste treatment plants. The fuel cell can be operated to match the production of biogas, while
maintaining some flexibility in the ratio of outputs. The price of biogas depends on the costs of
the biomass source (or biomass collection) and the generation equipment. The latter costs can
be reduced through learning and scaling up (Bergsma & Croezen, 2011), whereas both costs are
dependent on the quality of the biomass.

Industrial gas: Blast furnace or coke oven gas can be derived from steel plants, and syngas from
refineries. These industrial by-product gases are produced in large quantities in a continuous flow.
Fuel cells directly compete with conventional turbines to convert the energy of the energy-rich
gases.

Electricity: Because of its versatility, electricity is an energy carrier with a high value. The price is
time-dependent, since the possibilities for storage are limited and entail losses. The production
location is less important, since the extensive electricity grid allows for efficient transport. Elec-
tricity can be purchased from energy retailers. To sell electricity, a contract can be signed with
large industrial users, or grid connection costs and electricity fees have to be negotiated with
DSOs (see also § 8.5.2). In the future, renewable electricity sources can cause periods with very
low or high electricity prices.

Hydrogen: There is a limited number of applications for hydrogen. In § 5.3.3, H2 fuel stations and the
fertiliser and petrochemical industry were identified. In general terms, larger users expect lower
prices (Energinet.dk, 2011). A niche market not discussed before, are firms that need hydrogen in
small amounts. Applications are found in industries producing among others food, semiconductors,
electronics, and chemicals. Most of these users currently rely on centrally produced H2 , which is
transported by truck or pipeline (Energinet.dk, 2011). Current costs of H2 are substantially higher
in this niche compared to the bulk consumers. The price is influenced by the transport distance
from the central production site (Chen, 2006). Given that FCEVs are an emerging technology, the
hydrogen market for this purpose is not fully established yet. Therefore, competing H2 production
options and the corresponding prices are to some extent uncertain. What can be said is that H2
is produced either locally or centrally, can be transported by pipe or truck, and can be stored in
a tank or other medium. At fuel stations, hydrogen has to be compressed before it is transferred
to FCEVs. Therefore, it can be considered to compress the H2 at its origin, allowing transport by
truck (Chen, 2006).

Heat: Heat produced by SOCs can be most effectively used by industries that need high-temperature
heat. Heat exchangers can convert the heat to lower temperatures, although this results in a loss of
exergy. In that case, heat can be supplied to district heating systems. A third possible destination
are greenhouses, which need heat to maintain the indoor temperature during winter. Although
industrial and residential district heating systems are not open for competition (Van Melle et
al., 2015), SOCs will face compete to some extent with industrial sources of waste heat and
CHP plants. Heat users have local heat production technologies as alternative. The considerable
investment costs for heat transport infrastructure makes them attractive only when a cluster of
heat consumers is located nearby (Stryi-Hipp, 2009). This can be the case in industrial areas,
densely populated cities, or greenhouse areas.

114
8.7 Niche strategies
Building on the analysis of drivers, barriers, and energy carrier markets, niches that are favourable for
SOFC and SOFEC development can be identified. Niches that have a higher level of protection from
regime selection pressures are generally more successful. Therefore, this section focusses on configura-
tions that are passively shielded, and suggests options for active shielding.
An important decision parameter is the size of the system. This is a trade-off between investment
costs (which are lower for larger systems; § 8.1.3) and the distance to suppliers and consumers (which
tend to be higher for larger energy flows). The proximity is particularly important for biogas and industrial
gas sources, and heat consumers. Micro-CHPs are positioned at the extreme end of the spectrum, with
a size that matches the demand of a single household. While this seems attractive in terms of ownership
and grid connection procedures, there are some disadvantages. When operated to follow the demand of
heat or electricity of the household, only low capacity utilisations are attained. The optimisation results
have shown how SOFC operation at high capacity utilisation is cost-effective for the system as a whole
and for the SOFC installation. Therefore, it is a better strategy to install a somewhat larger fuel cell
for a block of houses, an apartment building, or (part of) a neighbourhood. By supplying a group of
households, the capacity utilisation is higher. Besides, the specific investment costs are lower, and the
financial risks are shared.
Another trade-off concerns the flexibility in operation. The mode of operation of the installation
can be adapted to match either the availability of resources, the need for products, or combinations
of these. The optimisation results from Chapter 7 have shown that a SOFC has sufficient flexibility to
respond to demand variations when NG is used as fuel. A source of flexibility not discussed before is
the use of NG as a backup fuel to replace biogas.

8.7.1 SOFEC niches

From the considerations above, it becomes clear that the operation of a SOFEC is dictated by the
availability of low-priced electricity and biogas or industrial gas from a nearby source. When located at
a steel plant, a SOFEC can convert blast furnace gas streams to H2 for a user in the same industrial area.
In this application, SOFEC faces strong competition on price with SMR. Protection can be provided
if the companies involved exempt the pilot project from profitability targets, and/or receive a subsidy
for industrial efficiency improvement. The practise of continuous operation of steel plants on the one
hand disables adaptive operation in response to electricity prices, but on the other hand ensures a high
capacity utilisation. Similar reasoning applies to the use of syngas by a SOFEC at refineries.
Another site for SOFECs could be at a farm or WWT with a digester. For this setup, a conflict
between the inputs arises. The SOFEC is either operated to profit from cheap electricity, or to operate
continuously and use all biogas. The optimal choice is to adhere to the latter strategy, even if this
means that some of the biogas is not used (§ 7.7.4). Therefore, a SOFEC provides a constant supply of
H2 and heat in this setup. In the case of a farm, trucks frequently deliver or collect goods. These trucks
can be fuelled by hydrogen, providing a reliable consumer. As noted in a case study for a brewery, this
project should be set up in collaboration with the involved transport companies (Van Leeuwen, 2016).
When also greenhouses are located near the farm, the SOFEC installation represents an extra source of
income.
What limits the attractiveness of this second option is the price of electricity, which is considerably
larger than in the model because of taxes. In particular in the near future, with modest renewable
electricity production, a SOFC can be a better alternative in terms of cost savings (see § 7.6). In the
optimisation results, SOFECs only marginally help to reduce the pressure on the electricity distribution
grid. Incentives to increase responsive operation can be provided by amplifying the price signals of the
wholesale market.
Taking these considerations into account, the best strategy for SOFEC development seems to be
diversification, i.e. by focussing on SOFC technology first. The knowledge and experience gained during
SOFC development are to a large extent applicable to SOFECs. After 10 or 20 years, the decreased
manufacturing costs provide SOFECs with a higher competitiveness. By that time, the share of renewable

115
electricity production is likely to be larger, providing additional incentives for its introduction.

8.7.2 SOFC niches

A possible location for a SOFC is at a fuel station with a nearby heat consumer. The installation is
sized to meet H2 demand by FCEVs. An incentive for entrepreneurs to install a SOFC system, is to gain
a market share in this developing market. Comparable to the market for BEV charging, investors are
willing to make losses in the first years, as long as the market is expected to grow rapidly (Van Gurp,
2016). Even if the growth in H2 demand does not meet expectations, the SOFC can still generate
income from electricity production by operating at high Uf . Because of the low variable O&M costs,
the installation has a low marginal electricity cost and therefore a high capacity utilisation.
As discussed for SOFECs, a SOFC can also be fuelled with blast furnace gas from a steel plant, or
biogas from a digester. Again, H2 can be used as truck fuel, while electricity is produced for individual
use or feed-in to the grid. In this setup, higher electricity prices are favourable, as they cause the SOFC
to pay of earlier. After all, electricity generated on-site becomes more attractive compared to taxed
external power. Due to the centralised nature of steel plants and the segregated location of farms, it
might not always be feasible to find a nearby destination for all heat produced.
Businesses with a small volume H2 demand, for whom higher prices are acceptable, a SOFC system
can be attractive. These users can save on transport by installing a small stack, operated to produce
H2 when needed. It is likely that most of these firms are located at industrial areas, so a customer for
the co-produced heat is nearby. Electricity can be used by the company itself, or delivered to the grid.
A profitable business case might not be a sufficient incentive for these actors, as the expenses on H2
are small relative to the total turnover. To overcome the resistance to change, the process of financing,
designing, installing, and certifying the system should be clear and straightforward. Installing a SOFC
can also be a means to emphasise the role of the company as a frontrunner, and to distinguish itself
from competitors.
A SOFC system can also be integrated in a local community energy system. Neighbourhoods that
are conscious about energy and have an ambition to reduce their environmental impact are suitable
communities (Van den Hil, 2015). People with this attitude are often willing to try new technologies
and are less focussed on (short term) financial benefits (Sauter & Watson, 2007). By collectively
investing in a SOFC installation, the members of of the community or cooperative have access to
locally produced heat, electricity, and hydrogen. This setup can complement local renewable electricity
and heat production. When some storage volume for heat and H2 is included, it can be operated
to primarily respond to electricity market fluctuations. A prerequisite for this niche is that a heat
transport infrastructure is present or can be installed, and that at least some people drive in FCEVs. A
lower carbon footprint can be achieved when a biogas source is near. Yet, there is a limited number
of neighbourhoods with a manure or waste-water digester in the proximity. To guarantee security of
supply NG can be used as backup. The chance for successful growth of the niche is increased if many
stakeholders are involved. DSOs, energy retailers, and municipalities can provide expertise and support,
as they also do in a dozen of other pilot projects (Gerdes et al., 2014).
Each of these applications can be facilitated by (local) governments through temporary subsidies
or a Green Deal agreement. A better alignment of the practice of SOFC use and legislation can be
accomplished by making a Green Deal. To convince policy makers of the need for stimulation, interest
groups or individual actors can lobby and bring SOC technology under the attention.

8.8 Conclusion
This chapter has provided an overview of the socio-technical system related to the energy supply chain
in the Netherlands. Learning effects and economies of scale determine how fast cost reductions of SOC
technology will go. The technology development and its potential were assessed from a social point
of view using the MLP. This framework features the concept of hierarchically nested levels, i.e. niche
innovation, socio-technical regime, and socio-technical landscape.

116
Table 8.4: Potential niche markets for SOFCs and SOFECs and factors that influence chance of
successful implementation.

Niche Boundary conditions Motivation/Barrier Suitable

sites
All Heat and H2 consumers in proximity.
Electricity grid connection.
Gas network connection or near other
fuel source.
SOFEC
Refinery or Subsidy to be competitive with SMR. High capacity utilisation. Many
steel plant
Farm or Arrangement with H2 -fuelled trucks. Generate income from Some
WWT by-product.
High electricity price.
SOFC
Fuel station Produce sufficient H2 to charge FCEVs. Gain market share in H2 for Many
transport market.
Farm or Biomass digester present. Generate income from Some
WWT H2 trucks frequently pass. by-product.
Difficult to find heat
users.
Small in- Meet own H2 needs. Save on H2 acquisition Some
dustrial H2 Guaranteed heat consumer. costs.
users Complicated
installation and
regulation process.
Energy Dedication of most neighbours. Energy independence. Limited
cooperative Presence of FCEVs and district heating Complement electricity
system. and heat production.
Support from other
stakeholders.
Sustainable As above. Sustainable energy provi- Few
community Located near a biogas source. sion.

Within the dominant regime of energy production and consumption, there is no fully developed solu-
tion to address the challenge of increased fluctuations of electricity demand and intermittent production.
From the perspective of regime actors, the roles should be redefined to provide flexibility and maintain
the balance on the electricity grid. From the perspective of the SOC niche innovations, there are several
drivers, but also barriers, that influence the development towards a role in a new regime. Few activities
are taking place around SOFECs. It does not need to be problematic that efforts are suspended, given
that SOFEC technology becomes more attractive in the future, with lower electricity prices and lower
SOC costs. The SOFC niche is active, with several demonstration projects and market introductions in
the area of micro-CHP applications. These dynamics could result in an uptake in the existing regime
structure without exploiting and exploring the capability of SOFCs to co-produce hydrogen.
It is deemed crucial that a user for the energy products of the SOFC or SOFEC is found. There-
fore, the energy carrier markets for inputs and outputs form the starting point for identifying market
niches, in which the SOC technology can be introduced first. In these niches, the SOC need not be
financially profitable from the beginning, since the involved actors are motivated by other aspects of
the technology. The potential niche markets that were identified are summarised in Table 8.4. This
table also lists expected motivations and barriers for the stakeholders. Barriers to be overcome include
the high investment costs, regulations for equipment, and procedures and costs associated with NG
and electricity grid connections. Drivers for the adoption of SOCs can be the high efficiency and low

117
emissions of SOFCs and SOFECs, policies such as Green Deal agreements, and commitment of other
stakeholders to learn from pilot projects.

118
Chapter 9

Discussion

9.1 Sensitivity analysis

9.1.1 Setup
GHG tax Scenario Restrictions SOC costs
30 €/tCO2 -eq. Diversity see Table 9.1 and 9.2 II, see Table 6.4
It should be kept in mind that the results obtained using the model – especially quantitative results
– are dependent on the set of assumptions made. While it is impossible to assess the sensitivity of
all assumptions listed in Chapter 6, the results described in Chapter 7 give incentive to investigate
two technologies in more detail. These technologies are Li batteries and PV (see also § 6.4.1 and
6.3.1, respectively). Both technologies have shown a rapid cost reduction in the past decade, and it
is assumed that this trend will continue (Lazard, 2015a, 2015b). Yet, even when assuming the costs
for these technologies expected in the future, their role is of limited importance. The installed capacity
is expected to be larger. Battery technology has a potentially important role in systems with large
amounts of intermittent electricity sources (Mulder, 2014). At most 9 GWe of PV is installed in the most
favourable scenario, but hardly any at a GHG tax of around 10 €/tCO2 -eq. (Figure 7.19b). Therefore,
the effect of the costs assumed for both technologies on the optimisation results were investigated. The
results are discussed below.

9.1.2 Cheaper batteries

As presented in Table 9.1, four scenarios with decreasing costs associated with batteries were optimised.
The change to cheaper batteries for electric energy storage has only a modest effect on the other energy
system components. The installed capacity of batteries increases from around 0.45 GWhe in the base
case to 5.3 GWhe for the B4 scenario (Figure 9.1a). Cheaper storage has a positive effect on the
penetration of PV, since the daily production peaks can be stored for later use. The results show that
cost-effective batteries can replace some peak production powerplants. The installed capacity of SOFCs,
PEMFCs, and hydrogen tanks marginally decreases with decreasing battery costs.

Table 9.1: Definition and optimisation outcomes of scenarios featuring Li batteries with different
costs.

Base case B1 B2 B3 B4
Investment costs (€/kWh) 225 175 150 125 100
Variable O&M costs (€/kWh) 9.015 1 0.1 0.05 0.01
Installed capacity (MWh) 453 1473 1851 2473 5294
Electricity production (TJ/a) 1.39 6.27 9.87 29.2 1916

Although the use of batteries is affected by the large cost reductions studied here, the effect on
SOFCs is small. It can be concluded that the two technologies do not compete, as they have different

119
functions in the energy system. Therefore, the conclusions with respect to SOFCs are valid also when
batteries are less expensive.

18 18
16 16
Installed capacity (GW)

Installed capacity (GW)

14 14
12 12 PV commercial
10 10 Onshore wind
8 8 SOFC
6 6 Offshore wind
4 4 PEMFC
2 2 Li battery
0 0
Base B1 B2 B3 B4 Base PV1 PV2 PV3
case Cheaper case Cheaper

(a) Change of costs of Li battery component, (b) Change of costs of PV commercial component, see Ta-
see Table 9.1. ble 9.2.

Figure 9.1: Sensitivity analysis of the assumptions for two emerging technologies. Further assump-
tions: 30 €/tCO2 -eq. emission tax, diverse demand scenario, SOFC available with costs expected
in 2020.

9.1.3 Cheaper PV
Commercial PV installations are cheaper than domestic ones. Since the limit to the installed capacity
of commercial PV is not reached, domestic installations are not present in an optimal system. The
effect of cheaper installations is considerable, as shown in Table 9.2 and Figure 9.1b. In the base case,
1.9 GWe of PV is installed, which increases to about 17 GWe in the PV3 scenario.

Table 9.2: Definition and optimisation outcomes of scenarios featuring commercial PV with different
costs. The lifetime is assumed constant at 15 a.

Base case PV1 PV2 PV3

Investment costs (€/kW) 660 600 500 400
Fixed O&M costs (€/MW) 6.591 6 5 4
Installed capacity (GW) 1.91 4.56 10.1 17.2
Electricity production (PJ/a) 8.94 21.4 47.2 80.7

Both the installed capacity and annual production of SOFCs decrease by about one third for decreas-
ing PV costs. PV replaces the electricity, SMR produces the remainder of hydrogen, and electric boilers
(indirectly using solar energy) fill the gap of decreasing heat production from SOFCs. Therefore, it is
possible that a lower number of SOFC installations is optimal in the future energy system, compared to
what was found in Chapter 7.

9.2 Break-even analysis

As can be concluded from Figure 7.7, the fuel costs are an important factor in the cost structure of
SOFCs and SOFECs. The price of these resources was specified either directly (e.g. for NG) or indirectly
by assuming costs of other technologies (e.g. for biogas). In either case, the results can be sensitive
towards changes in these assumptions. Since the combination of many input parameters plays a role
in establishing the energy prices, it laborious to assess each of them individually. Instead, the influence
of the fuel prices on the levelised costs of energy from SOFC and SOFEC installations are assessed
in a break-even calculation in this section. To do so, the levelised cost calculated for a SOFC and a
SOFEC are compared with the costs of an alternative combination of conventional technologies. The
comparison is made on a basis of equal energy outputs.

120
 A SOFC operated at Uf = 85% and i = 1.1 A/cm2 generates 0.156 MWh H2 , 0.337 MWhe , and
0.243 MWh heat per MWh NG consumed (Eq. 3.27). An alternative pathway, also based on NG as
energy source, is composed as depicted in Figure 9.2a. This alternative produces the same amounts of
energy carriers by means of SMR, NG boiler, and NG powerplant. The break-even point can be found
by comparing the levelised costs (expressed in Eq. 4.1) of the two alternative pathways:

LCOEconv = DH2 · LCOESMR + DP · LCOEPP + DQ · LCOEboil (9.1)

= LCOESOFC

Most parameters are assumed fixed: the properties of the conventional technologies were adapted from
Chapter 6, the CU is identical for all technologies, and for the SOFC we assume that Cvar = 5.5 €/MWh
and CRF=11.7%. Then, Eq. 9.1 can be rewritten to find the break-even investment costs of a SOFC:

Cinv,SOFC = 311 + (24.9 CNG + 11.9 taxGHG − 341) · CU (9.2)

156 kWh H2 SMR

NG production SOFC 337 kWh Electricity NG powerplant NG production

243 kWh Heat NG boiler

(a) SOFC

Biogas SMR
1030 kWh H2
production
SOFEC NG production

Electricity 330 kWh Heat NG boiler

(b) SOFEC

Figure 9.2: Reference systems for energy production by a SOFC or SOFEC (left) or conventional
alternatives (right).
.

This equation indicates that the capacity utilisation has a big influence on the break-even point.
Higher SOFC investment costs are acceptable if a high CU can be realised. Higher NG prices and GHG
taxes also positively affect the competitiveness of SOFCs. The effect of a 1 €/MWh increase of the NG
price has roughly the same effect as a tax increase of 2 €/tCO2 -eq.. The break-even investment costs
are plotted in Figure 9.3 as a function of CU and CNG . In the upper right corner of the plot represents
the most favourable set of conditions, where Cinv ≤ 1000 €/kW makes the SOFC competitive. This
value for the investment costs increases to about 1550 €/kW if a 50 €/tCO2 -eq. emission tax is applied.
It should be kept in mind that this analysis is about the full replacement of one system by another (the
alternatives in Figure 9.2a), by assuming an identical CU for both. Yet in practice, the alternatives can
coexist and complement each other. For example, one technology fulfils a base-load function, while
another operates for peak production. Consequently, SOFCs could already be introduced successfully
with higher costs than calculated here.
A similar approach is used to assess the break-even point of a SOFEC. When operated at Uf = 80%
and i = 1.3 A/cm2 , a SOFEC generates 1.03 MWh H2, 0.330 MWh heat, and consumes 0.433 MWhe
per MWh biogas consumed (Figure 9.2a). A combination of an SMR and a NG boiler that to-
gether produce the same amounts of energy form the alternative. Most parameters are assumed fixed:
taxGHG =0 €/tCO2 -eq., CU=1, the properties of the SMR and boiler were adapted again from Chap-
ter 6, and for the SOFEC we assume that Cvar = 5.35 €/MWh, CRF=11.7%, and electricity consumed

121
 Break-even investment costs (€/kWin)

10
40

50

60

70

80

90

00
0

0
40 80
Negative

Electricity price (€/MWhe)

70
35 40
NG price (€/MWh)

0€
/kW
80
60 0€
/kW
30 12
00
€/k
50 16 W
00
€/k
25 20 W
00
40 €/k
W

20 30
0 0.2 0.4 0.6 0.8 1 10 15 20 25 30
Capacity utilisation Biogas price (€/MWh)

Figure 9.3: Investment costs of SOFC at which Figure 9.4: Investment costs of SOFEC at
combined heat, hydrogen, and power production which combined heat and hydrogen production
is equally expensive as the alternative of using is equally expensive as the alternative of using
SMR, a NG boiler, and a NG powerplant. The SMR and a NG boiler. The capacity utilisation
GHG tax is 5 €/tCO2 -eq. is 1.

has no GHG emissions. As above, Cinv,SOFEC can be described by equating its LCOE to the alternative’s
LCOE:

LCOEconv = DH2 · LCOESMR + DQ · LCOEboil (9.3)

= LCOESOFEC
Cinv,SOFEC = 4212 − 75.0 Cbiogas − 32.5 CP + 26.4 taxGHG (9.4)

Of course, a higher price of biogas and electricity decreases the break-even investment costs. The effect
of both resources’ prices are shown in Figure 9.4. A larger amount of biogas is used by a SOFEC, so
the price of it is about twice as important as the electricity price. The biogas price estimated from
the data in Chapter 6 is 28 €/MWh, which is somewhat higher than the current NG price. This price
might decline due to efficiency improvements or through co-location with a cheap biomass source. Yet,
increased competition for biomass or -gas could increase the price. As illustrated by Figure 9.4, the
larger range of electricity prices provide opportunities. The higher the biogas price, the less electricity
should cost to generate profit. There is also a relation with CU: lower electricity prices are low only a
fraction of the time. Thus, the lower the required price, the lower the CU.

9.3 Validation
The calculations made in § 9.2 have demonstrated the influence of energy prices on the break-even
point of SOCs. Combined with the variety of price levels (§ 8.6) for energy consumers, it is evident that
the business case is dependent on the specific situation in which the SOC is implemented. To validate
the results of the break-even calculation and the break-even costs found in § 7.5, a comparison with
values obtained by other studies is made. No other optimisation models have studied the role of SOFCs
with hydrogen co-production or SOFECs yet. Therefore, assessments of SOFCs used as (micro-)CHP
are reviewed.
In the study of Pruitt et al. (2013), SOFCs for energy supply of various commercial buildings were
considered. A hotel in California benefits from a SOFC CHP installation if it costs less than 4500 $/kWe ,
and 2500 $/kWe for a hotel in Wisconsin. Whether GHG emission taxes support SOFC introduction
or not depends on the situation. A comparison of a SOFC and a microturbine for the distributed

122
generation of electricity was made by Strazza et al. (2015). Both alternatives have the same LCOE if
the investment costs of the SOFC are 2450 €/kWe for households and 2300 €/kWe for industrial users.
One reason why this break-even point is lower than that determined by Pruitt et al. (2013) is the fact
that heat is not considered as a product. Hauptmeier et al. (2016) have assessed a SOFC system fuelled
by biogas from a WWT. The results are highly dependent on the assumed electricity price and discount
rate. The break-even point of the investment costs lies at 4500 €/kWe for the best or 1900 €/kWe for
the worst case assumptions.

9.4 Cost reduction

For several experiments in Chapter 7, the SOC cost predictions from Energistyrelsen and Energinet.dk
(2015) were used for guidance. One finding was that SOFCs become competitive starting from an
investment cost of around 825 €/kW NG, or 1710 $/kWe . The question can be asked whether it is
realistic that these cost target is reached in 2020, or else when it will be reached. To get an idea of
the evolution of the investment costs in the future, the concept of learning rates introduced in § 8.1.2
is used. Considering the learning rate ranges found in literature (Rivera-Tinoco et al., 2012; Staffell &
Green, 2013), a rate of 18% is used for the calculations following in this section.

12000
2010
10000
2011
Costs ($/kWe)

8000
2012
6000 2013 2015
2014
4000
2025
2000 2020
2030
2040 2050
0
100 10 1
10 2
10 3
104 105
Cumulative installed capacity (MWe)

Figure 9.5: Possible cost reduction curve of the investment costs of SOFCs as a function of the
cumulative installed capacity. The learning rate is 18%.

In Figure 9.5, the SOFC investment costs are plotted against the cumulative installed capacity. As
a point of reference, the costs in 2014 are taken to be 5080 $/kWe (Lazard, 2015a). To assign the year
in which a cost level was reached in the past, the cumulative installed capacity in the period 2009–2014
was used (Hart et al., 2015). The capacity installed in the future was estimated by linear extrapolation
of the shipped amount of SOFCs in the past five years (see Figure 8.3).
When Figure 9.5 is compared with the numbers from § 9.3, one would expect that the first appli-
cations of SOFCs become competitive within the coming five years. On the other hand, it might not
be before 2040 that the point of 1710 $/kWe is reached.

123
Chapter 10

Conclusion

10.1 Conclusion
The aim of this research is to assess the feasibility of including SOFCs and SOFECs in a future national
energy system. Therefore, the technical feasibility, the effects on the energy system and its eco-efficiency,
the required technical modifications of the energy system, and the societal dynamics of stakeholders
were assessed. In this section, the answers to the research questions are summarised.

Is it technically possible to operate SOCs in both normal and fuel-assisted electrolysis

mode?
By reviewing existing literature, the technical possibilities of SOFC and SOFEC technologies were as-
sessed. SOFCs that co-produced heat and power have been extensively investigated at lab scale and
in demonstration projects. Commercial installations are deployed as micro-CHP or as auxiliary power
units. The possibility of hydrogen production by IR-SOFCs has been demonstrated in the lab. The
same is true for the hybrid or reversible SOFEC concept. A reversible SOFC/SOEC system has been
manufactured and employed at commercial scale.
A key limiting factor in the performance of SOCs is their limited lifetime. To address this issue,
degradation of the electrodes at microscopic level should be minimised. This can be achieved by using
other electrode materials or manufacturing methods. Recent research indicates that regular switching
between fuel cell and (fuel-assisted) electrolysis mode can also prevent degradation.

To obtain a representation of the input-output relations of SOFCs and SOFECs, a 1-D model was
used. In the SOFC model, hydrogen production was taken into account. Non-linear equations with two
degrees of freedom were derived for both technologies. These equations have a suitable level of detail
to represent the possible ratios of outputs and the freedom in the total throughput. In further analyses,
this has proven useful for the assessment of technology operation.
The derived equations were used to implement SOFC and SOFEC components in oemof, an energy
system optimisation framework. Then, the framework was used to model the energy system of the
Netherlands in its current and possible future configurations. The model covers all major energy carriers,
i.e. fossil fuels, electricity, heat, biomass, and hydrogen, allowing to assess technologies with multiple
outputs. It models the system on an hourly basis during one year, taking into account the dynamic
patterns of production and consumption. By doing so, it is the first optimisation model of its kind for
the Netherlands. Not all aspects of the real energy system are represented in the model. Most notably,
energy transport and demand response are not included.

What is the optimal operating strategy for SOCs?

The model was used to calculate the optimal set of technologies and their operation under different
scenarios of SOC costs, demand compositions, and GHG emission taxes. The results allowed to study
among other things the optimal operating strategy of SOFC and SOFEC installations.

124
 Flexibility in the operation of SOFCs proves to be useful for responding to fluctuating hydrogen and
electricity demand patterns. At low electricity demand, low fuel utilisations are used, whereas a setting
with high i and Uf is used if the electricity demand is relatively high. The three factors investigated
– emission tax, demand, SOFC costs – all influence the operation. When a small number of SOFCs is
installed (because of high costs), operation with high H2 output is optimal. At large-scale introduction,
SOFC operate to produce more electricity and to be more flexible.
In the Diversity scenario (with high demands for both heat, hydrogen, and electricity), a wide range
of settings is used. This requires a BoP design that can handle changes in mass and energy flows, plus
an advanced control system that responds appropriately to changing market conditions. This control
mechanism should take into consideration thermal gradients and temperature changes associated with
this. The operation of a SOFC in the other demand scenarios displayed a different pattern. Only two
or three operating points were used frequently, while other settings were applied seldom. Thus, the
composition of the energy demand determines how SOFCs are operated optimally.
For SOFECs, the optimal operating strategy is also characterised by a high H2 production. This is
achieved by operating around a fixed, optimal i. Degradation due to temperature changes is therefore
less of an issue in this application. There are also less restrictions on the ramping rate, since SOFECs
are ideally operated almost continuously, contrary to regular electrolysers.

Are SOCs a cost-effective addition to the current Dutch energy system?

Both SOFCs and SOFECs with the current characteristics are not competitive as an additional compo-
nent of the Dutch energy system. If significantly lower costs are assumed, SOFCs can play a role as
a replacement for SMRs and NG powerplants as source of hydrogen and electricity. SOFECs are not
effective in the current system.

At what cost level are SOCs a cost-effective component of the future Dutch energy
system?
Since the costs of SOFC and SOFEC technologies are expected to decrease in the future, it was inves-
tigated at what point they become competitive. This was done by adapting the economic properties
of SOCs in steps, and optimising their role in the Dutch energy system each time. The system was
optimised without pre-defined installed capacities of existing technology. The LCOE of SOFCs should
be reduced to 94–104 €/MWhe to be cost-effective in a future energy system with 20 €/tCO2 -eq.. This
implies that investment costs of below 1000 €/kWe are required. Another important finding is that the
production of H2 increases the profitability of SOFCs, as witnessed by the potential to replace a major
part of SMR installations.
SOFECs cannot realise significant energy system cost reductions, in particular in systems with limited
renewable electricity. On the one hand, a SOFEC faces competition with other H2 sources, and on the
other hand upgrading to green gas is a good alternative application for biogas. Very low investment
costs or high GHG emission taxes are required to make SOFECs attractive for the energy system. And
even then, only part of the hydrogen demand is met by a SOFEC.

What changes in GHG emissions and primary energy consumption result from the
introduction of SOCs?
A financial driver for the introduction of SOFCs is their resource-efficiency and low emission of GHGs.
The life-cycle savings on fuel compensate for the high investment costs of SOFCs compared with
competing technologies. SOFCs can support a transition from coal to NG as energy carrier. Additionally,
the primary energy use and emissions of the energy system decrease when SOFCs are introduced. An
average reduction in GHG emissions of 5% is achieved by adding SOFCs to the system. Yet, no
supporting effect of SOFCs on the share of renewable energy was found. SOFCs with sufficiently low
costs can even reduce the optimal capacity of solar and wind energy.

What effect do GHG emission taxes have on the economic feasibility?

Since SOFCs and SOFECs are more efficient and use less polluting fuels than alternative technologies,

125
higher GHG emission taxes are favourable for their deployment. SOFECs are almost carbon neutral due
to the use of biogas as a fuel. Consequently, a SOFEC profits from the stimulation of renewable energy
use. If NG were used, the technology is not attractive.
For SOFCs, a 10 €/tCO2 -eq. higher tax has a similar effect on the break-even point as a 9% lower
LCOE. Also, the competitiveness of SOFCs increases for higher NG prices because of the efficiency. Only
when very high taxes of around 100 €/tCO2 -eq. are applied, the attractiveness of SOFC technology
starts to decline. Therefore, SOFCs can contribute to the first steps of GHG emission reduction at low
societal costs. Taxation of either GHGs or fossil fuels directly are possible policies that promote SOFCs
and SOFECs.

How can the transition towards a system with SOCs take place in terms of technology
replacement?
Whether SOFC of SOFEC installations are a cost-effective addition to the energy system depends on the
technologies that are already installed. As a consequence, the introduction of the new technologies was
shown to occur somewhat later than expected purely based on costs. A growing demand for hydrogen
by new (distributed) applications can provide opportunities for the installation of SOCs.

What are potential niche markets, in which SOC technologies can be applied first?
The optimisation model only selects technologies if they contribute to the effectiveness of the energy
system. Therefore, no conclusion can be drawn for the economic feasibility of SOFC or SOFEC appli-
cations in specific market conditions or business cases. To give a better view on the possibilities for
introducing the first SOC installations, potential niches were studied.
These first market applications are established through arrangements of involved stakeholders, who
are interested in technology development and experimentation with less focus on short-term financial
profit. The market and infrastructure for the involved energy carriers provide important conditions
for possible applications. Electricity and NG can be exchanged with the distribution network at many
locations. Alternative fuels can be supplied by WWTs, farms, steel plants, or refineries. There are
often heat consumers nearby, but the transport infrastructure is not always present. Hydrogen can be
supplied to firms with a small volume demand, to FCEVs parked in residential areas, or FCEV fuel
stations. Examples of identified niche applications include a farmer who uses biogas from manure
digestion and supplies H2 to trucks, or a neighbourhood community that invests in a common CHHP
installation.

How can the development and diffusion of SOC technology in the Netherlands be
promoted?
The MLP framework was applied to analyse interactions and dynamics of the dominant regime and the
SOC niche innovation. Existing barriers include the high investment costs, divergence of regulations for
equipment, and procedures and costs associated with NG and electricity grid connections. To convince
potential users, these perceived barriers can be addressed or the advantages of SOCs can be emphasised.
The high efficiency, low emissions, and importance for local grid balancing of SOFCs and SOFECs are
arguments in favour of these technologies. In addition, the commitment of other stakeholders in pilot
projects or through Green Deal agreements can stimulate the adoption.

10.2 Recommendations
10.2.1 Energy consumers
The results of this study are of limited direct importance for consumers. This is because the total
system costs are considered, ignoring any taxes, service fees, subsidies, and (in)variability of energy
prices. Other studies, in which these factors were taken into account, suggest that the break-even
investment costs are higher than those calculated here (see § 9.3). It is likely that a SOFC is financially
attractive earlier than a SOFEC, in line with the total system results.

126
10.2.2 Policy-makers
As predicted by for example Schoots and Hammingh (2015), the NG reserves of the Netherlands will
be depleted to a large extent around 2030. Therefore, policy-makers can ask the question whether it is
worthwhile to invest resources in the development of SOFC technology, which relies on NG. In particular
micro-CHPs do not comply with the goal of an emission-free residential sector (Sociaal-Economische
Raad, 2013). There are several facts that make SOFCs fit to meet these policy goals. They can be
switched to biogas or green gas as a fuel. For larger installations, it is relatively easy to apply carbon
capture, since the flue gas has a very high CO2 concentration. And as long as fossil fuelled electricity
is produced, SOFCs are a highly resource-efficient option and can reduce the annual GHG emission by
about 8 Mt/a. That being said, SOFC technology does not support solar and wind energy production.
Therefore, GHG reduction policies cannot be replaced by the stimulation of SOFCs.
SOFCs are effective to achieve policy goals related to energy and resource efficiency. If cost reduc-
tions can be realised or subsidies are granted, the fuel cells can be applied for distributed CHHP for
companies resulting in lower environmental impacts.
SOFEC technology should be critically evaluated, given its low cost-effectiveness. It might be more
desirable to use the limited amount of biogas for applications other than fuel-assisted electrolysis. The
options of upgrading to green gas or combustion for heat production can be more attractive, depending
on the cost developments in the future.

10.2.3 System operators

DSOs should assess what organisations are willing and able to provide flexibility of consumption or
production as a service. By comparing SOFCs and SOFECs with other strategies to accommodate
for large amounts of intermittent electricity generation, a well-considered decision can be made. No
available technology is currently mature enough, but SOCs can become a feasible option if the learning
curve is continued. Under the assumptions made in this report, storage options (batteries and H2 tanks)
are best suited to address short-term fluctuations. SOCs, in particular SOFCs, are better geared towards
mitigating imbalances on the timescale of months and seasons.
If SOFCs or SOFECs are considered to be useful in this respect, DSOs can stimulate their instal-
lation by providing favourable distribution network connection procedures and contracts. Besides, it is
advised that a flexibility market that includes small consumers is established in cooperation with other
stakeholders. Such a market creates the boundary conditions for the development of innovative solutions
to maintain the electricity grid balance.

10.2.4 SOC manufacturers

SOC manufacturers benefit from a growing market share of SOC technologies. To achieve this, it
is crucial to sustain R&D efforts. Therefore, manufacturers should aim for close cooperation with
research institutes. As illustrated by Figure 2.8, various directions for future SOC applications have
been identified. Efforts can be made to achieve miniaturisation of the BoP (for micro-CHP), biogas
compatibility (for SOFEC), high conductivity at lower temperatures (for IT- and LT-SOFC), operation
responsive to fuel variations (for fuel independence), safer H2 handling equipment (for CHHP SOFC),
reversible operation (for R-SOCs), or stability under high overpotentials and high current densities (for
SOEC). Although divergence of R&D efforts can be beneficial, a common focus shared by multiple
actors fosters learning through interaction resulting in fast progress. Besides, it is attractive to invest
most in applications with the largest market potential. Based on the results of this research, some of
the future SOC applications listed above appear more promising than others.
To begin with, it was concluded that the co-production of hydrogen increases the competitiveness of
SOFC installations. Therefore, it is recommended that this capability is investigated and developed fur-
ther, contrary to the current industry focus on CHP applications. The development of low-temperature
SOFCs seems less attractive, since this creates the need for precious metal catalysts. Recent findings
that demonstrate how reversible operation can prevent degradation give incentive to research R-SOCs
further. It might not be economically attractive to switch between fuel cell and fuel-assisted electrolysis

127
mode, but if regular operation as SOFEC prolongs the service life of the stack, there is a different situ-
ation. Furthermore, this study has revealed the limited potential market niche of SOFECs, in particular
in the current system. It is recommended that development of this technology is not prioritised for
now, but can be reconsidered in the future. Similarly, SOECs can become effective in systems with
high renewable electricity production. SOC installations are preferably applied for distributed generation
to minimise energy transport losses and maximise product utilisation. It is advised to install systems
that supply several households. Smaller micro-CHP SOFCs are less attractive due to the relatively large
investment costs.

10.3 Future research

10.3.1 Model refinement
Within oemof, the possibilities for further refinement of the details of the energy system model are
infinite. Yet, the trade-off between model accuracy and calculation time should always be kept in mind.
The enumeration of simplifications in § 5.8 can provide a starting point for model refinements. Some
modifications to the model that are worth considering are discussed in this section.
To obtain a faster model, the linearisation options for SOC input-output relations proposed in § 3.6
could be considered. The effect of such a simplification has to be assessed by comparing the optimisation
results with those of the nonlinear model. The deviation is probably significant, in particular if the costs
of SOFCs are assumed to be low. In that case, nonlinear equations are needed to correctly describe the
flexible behaviour as shown in the various optimisations.
On the other hand, it is also possible to increase the detail of the input-output relations of other
components. The reduced efficiency at part load operation can be modelled by adding suitable (non-
linear) equations. Note that it should be specified how big all individual plants are for which efficiency
reduction applies. This allows solutions where one or more plants are halted, to allow the remaining
plants of the same type to operate at full load.
A useful refinement of the model can be the inclusion of energy transport infrastructure. This allows
to account for losses that occur during transport, and costs associated to the infrastructure. If transport
is included, the benefits of distributed energy conversion (close to the consumer) can be included better.
This approach is also relevant for SOC technology, given their suitability for small-scale applications.
To be able to model transport, it is necessary to apply some form of spatial quantisation to the model.
A very basic form of quantisation can be achieved by defining two ‘locations’: central and local. All
energy demand is located in the local space, while large-scale or remote generation occurs centrally.
Any local energy source does not rely on transport, because it is situated locally. More ‘realistic’
quantisations can also be imagined. The country can be divided in regions, provinces, municipalities, or
neighbourhoods, or in rectangles with desired dimensions. The more locations are added, the larger the
optimisation problem becomes in terms of data input and computation time. For each location, buses
and components, production and consumption, constraints and states have to be included.
The costs associated with the required transport infrastructure in a given scenario can also be esti-
mated by recognising that these costs depend on the peak demand. If the highest expected consumption
rate is known (for example by modelling a cold winter), the network capacity to distribute the energy
carrier of interest can be calculated. This approach was applied by Van Melle et al. (2015).
If spatial quantisation is applied, then location-specific demand and production patterns have to
be obtained or generated. Various options to obtain demand patterns of households are discussed in
Van den Hil (2015). In the case of wind turbine power production, the database of KNMI (KNMI,
2005) features wind speeds for a wide range of locations. There are several more opportunities for the
refinement of the input data. For example, the pattern for NG demand could be based on a dataset
with hourly resolution. For this study, only a gas demand pattern with monthly resolution was found
(see § 5.7.2).
Other data improvements can be implemented if more practical knowledge about new consumption
and production technologies becomes available. For example, the behaviour of BEV and FCEV owners
is not crystallised yet, and is therefore subject to change.

128
10.3.2 Innovation systems
To obtain a better understanding of the innovation niche around SOC technology, three aspects can be
investigated further.
First, the attitude of the general public and the preferences of potential users can be assessed.
This will provide insight into the differences between these groups, and into the perceived barriers for
the adoption of SOCs. Also, the insights can be used to develop applications of SOC technology that
match user preferences. Progress has been made in this area by Vernay et al. (2008) and Manné
(2009). One step further could be the identification of deployment strategies, as introduced by Sauter
and Watson (2007). They have explored possible company-consumer relations for distributed generation
technologies.
Second, it is possible to elaborate further on the role of companies in the development of SOCs.
The attitude towards SOC technology development, and the perception of the system and problems
faced can be mapped for both companies within the SOC niche and organisations that are part of the
socio-technical regime. These insights are valuable to better understand the dynamics and interactions
of the niche and regime level.
Third, as demonstrated in § 9.2, energy carrier costs in specific market niches are the key determining
factor for the viability of a business case including SOCs. Models or calculations that take into account
local fuel costs, distribution network connection fees, VAT, and other taxes can give a verdict on the
feasibility of SOCs in specific (niche) markets. An approach that can be taken is provided in planSOEC
(Energinet.dk, 2011). This report assesses the value that different consumers assign to hydrogen, based
on a cost calculation of the available alternatives.

129
References

Aboumahboub, T., Schaber, K., Tzscheutschler, P., & Hamacher, T. (2010). Optimization of the uti-
lization of renewable energy sources in the electricity sector. In Proceedings of the 5th iasme/wseas
international conference on energy & environment (Vol. 11, pp. 196–204).
Aguiar, P., Chadwick, D., & Kershenbaum, L. (2002). Modelling of an indirect internal reforming
solid oxide fuel cell. Chemical Engineering Science, 57 (10), 1665–1677. doi: http://dx.doi.org/
10.1016/S0009-2509(02)00058-1
Aki, H., Murata, A., Yamamoto, S., Kondoh, J., Maeda, T., Yamaguchi, H., & Ishii, I. (2005).
Penetration of residential fuel cells and CO2 mitigation–case studies in Japan by multi-objective
models. International Journal of Hydrogen Energy , 30 (9), 943–952. doi: 10.1016/j.ijhydene.2004
.11.009
Arduino, F., & Santarelli, M. (2016, jun). Total cost of ownership of CHP SOFC systems: Effect of
installation context. Energy Policy , 93 , 213–228. doi: 10.1016/j.enpol.2016.02.053
Batelle. (2014). Manufacturing cost analysis of 1 kW and 5 kW solid oxide fuel cell (SOFC)
for auxilliary power applications (Tech. Rep.). Columbus: Batelle Memorial Institute. Re-
trieved from http://energy.gov/sites/prod/files/2014/06/f16/fcto_battelle_cost
_analysis_apu_feb2014.pdf
Belderbos, A., Delarue, E., & D’haeseleer, W. (2015). Possible role of Power-to-Gas in future energy
systems (Working paper No. WP EN2015-09). Leuven: KULeuven Energy Institute, TME Branch.
doi: 10.1109/EEM.2015.7216744
Beljaars, M. (2015). Productie van minerale meststoffen in Nederland (Tech. Rep.). Den Haag:
Meststoffen Nederland.
Bellekom, S., Benders, R., Pelgröm, S., & Moll, H. (2012). Electric cars and wind energy: Two problems,
one solution? A study to combine wind energy and electric cars in 2020 in The Netherlands.
Energy , 45 (1), 859–866. doi: 10.1016/j.energy.2012.07.003
Bergsma, G. C., & Croezen, H. J. (2011). Kansen voor Groen Gas: Concurrentie Groen Gas met andere
biomassa opties (Tech. Rep.). Delft: CE Delft.
Biberacher, M. (2004). Modelling and optimisation of future energy systems using spatial and temporal
methods (Doctoral dissertation). University of Augsburg.
Bierschenk, D. M., Wilson, J. R., & Barnett, S. A. (2011). High efficiency electrical energy storage
using a methane–oxygen solid oxide cell. Energy Environmental Science, 4 (3), 944–951. doi:
10.1039/C0EE00457J
Biomass Energy Centre. (2004). Carbon emissions of different fuels. Retrieved
2016-05-02, from http://www.biomassenergycentre.org.uk/portal/page?_pageid=75%
2C163182&_dad=portal&_schema=PORTAL
Boccard, N. (2009). Capacity factor of wind power realized values vs. estimates. Energy Policy , 37 (7),
2679–2688. doi: 10.1016/j.enpol.2009.02.046
Boerrigter, H. (2006). Economy of Biomass-to-Liquids plants (Tech. Rep.). Petten: ECN. Retrieved
from https://www.ecn.nl/docs/library/report/2006/c06019.pdf
Bokhoven, T., Fennema, H., Grünfeld, H., Jurjus, A., Miesen, R., Molengraaf, P., . . . Wagenaar,
T. (2015). Nieuwe spelregels voor een duurzaam en stabiel energiesysteem. Overlegtafel En-
ergievoorziening(September).
Börjeson, L., Höjer, M., Dreborg, K.-H., Ekvall, T., & Finnveden, G. (2006). Scenario types and
techniques: Towards a user’s guide. Futures, 38 (7), 723–739. doi: 10.1016/j.futures.2005.12.002

130
Bourennani, F., Rahnamayan, S., & Naterer, G. F. (2015). Optimal Design Methods for Hybrid
Renewable Energy Systems. International Journal of Green Energy , 12 (2), 148–159. doi: 10.1080/
15435075.2014.888999
Bradbury, L. (2009). Wind turbine power ouput variation with steady wind speed. In WindPower
Program (chap. 14). Plymouth: PelaFlow Consulting. Retrieved 2016-03-25, from http://
www.wind-power-program.com/turbine_characteristics.htm
Brandon, N. (2013). Recent Developments in Solid Oxide Fuel Cells. In H2fc supergen. London.
Braun, R. J. (2002). Optimal design and operation of solid oxide fuel cell systems for small-scale
stationary applications (Doctoral dissertation). University of Wisconsin.
Bruckner, T., Fulton, L., Hertwich, E., McKinnon, A., Perczyk, D., Roy, J., . . . Wiser, R. (2014). Annex
III: Technology-specific cost and performance parameters. In S. Schlömer (Ed.), Climate Change
2014: Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment
Report of the Intergovernmental Panel on Climate Change (pp. 1329–1356). Cambridge; New
York: Cambridge University Press.
Butler, B. J. (2010). Energy balance in solid oxide fuel assisted electrolyzer cell modules (Master thesis).
University of Utah.
Carter, D., Ryan, M., Wing, J., Ryan, M., & Wing, J. (2013). The Fuel Cell Industry Review 2011,
2012, 2013. Fuel Cell Today , 56 (4), 1–56. doi: 10.1595/147106712X657535
CBS. (2015a). Elektriciteit en warmte; productie en inzet naar energiedrager [Data set]. Re-
trieved from http://statline.cbs.nl/Statweb/publication/?DM=SLNL&PA=80030NED&D1=
2,4-5&D2=1-2&D3=2-5,7-12&D4=16-17&HDR=T&STB=G1,G2,G3&VW=T
CBS. (2015b). Energiebalans; aanbod, omzetting en verbruik [Data set]. Retrieved 2016-03-10,
from http://statline.cbs.nl/Statweb/publication/?DM=SLNL&PA=83140NED&D1=a&D2=
0-1,11,34-50&D3=l&HDR=G2,G1&STB=T&VW=T
CBS. (2016a). Calorific value. Retrieved 2016-02-10, from https://www.cbs.nl/en-gb/our
-services/methods/definitions?tab=c#id=calorific-value
CBS. (2016b). Hernieuwbare elektriciteit; productie en vermogen [Data set]. Retrieved 2016-
09-26, from http://statline.cbs.nl/Statweb/publication/?DM=SLNL&PA=82610ned&D1=
6-7&D2=a&D3=a&HDR=G2&STB=T,G1&VW=T
Chen, T.-P. (2006). Hydrogen delivery infrastructure options analysis (Tech. Rep.). Washington D.C.:
Nexant.
Chertow, M. R. (2000, sep). The IPAT equation and its variants. Journal of Industrial Ecology , 4 (4),
13–29. doi: 10.1162/10881980052541927
Chicco, G., & Mancarella, P. (2009). Distributed multi-generation: A comprehensive view. Renewable
and Sustainable Energy Reviews, 13 (3), 535–551. doi: 10.1016/j.rser.2007.11.014
Cinti, G., Bidini, G., & Hemmes, K. (2016, jun). An experimental investigation of fuel assisted
electrolysis as a function of fuel and reactant utilization. International Journal of Hydrogen
Energy . doi: 10.1016/j.ijhydene.2016.05.205
Citizendium. (2013). Hydrocracking. Retrieved 2016-09-05, from http://en.citizendium.org/
wiki/Hydrocracking
Connolly, D., Lund, H., Mathiesen, B. V., & Leahy, M. (2010). A review of computer tools for analysing
the integration of renewable energy into various energy systems. Applied Energy , 87 (4), 1059–
1082. doi: 10.1016/j.apenergy.2009.09.026
Connolly, D., & Mathiesen, B. V. (2014). A technical and economic analysis of one potential pathway
to a 100% renewable energy system. International Journal of Sustainable Energy Planning and
Management, 1 , 7–28. Retrieved from http://journals.aau.dk/index.php/sepm/article/
view/497 doi: 10.5278/ijsepm.2014.1.2
Curtin, S., & Gangi, J. (2015). Fuel Cell Technologies Market Report 2014 (Tech. Rep.). Washington
D.C.: Fuel Cell and Hydrogen Energy Association. Retrieved from http://energy.gov/sites/
prod/files/2015/10/f27/fcto_2014_market_report.pdf
Daikin UK. (2012). Seasonal efficiency – Energy labelling. Retrieved 2016-05-23, from http://
www.daikin.co.uk/seasonal-efficiency/energy-label/

131
D’Appolonia. (2016). Analysis of market opportunities and legislative assessment (Tech. Rep.). Genova:
Bio-HyPP. Retrieved from http://www.bio-hypp.eu/wp-content/uploads/2016/05/D1.1
-Market-opportunities-and-legislative-assessment.pdf
Davis, C. B., Chmieliauskas, A., Dijkema, G. P. J., & Nikolic, I. (2015). Enipedia. Delft.
De Swardt, C. A., & Meyer, J. P. (2001, aug). A performance comparison between an air-source and a
ground-source reversible heat pump. International Journal of Energy Research, 25 (10), 899–910.
doi: 10.1002/er.730
De Wild, P., Nyqvist, R., & De Bruijn, F. (2002). Removal of sulphur-containing odorants from natural
gas for PEMFC-based micro-combined heat and power applications. In Fuel cell seminar. Palm
Springs.
De Boer, H. S., Grond, L., Moll, H., & Benders, R. (2014). The application of power-to-gas, pumped
hydro storage and compressed air energy storage in an electricity system at different wind power
penetration levels. Energy , 72 , 360–370. doi: 10.1016/j.energy.2014.05.047
Deng, Y. Y., Blok, K., & Van der Leun, K. (2012). Transition to a fully sustainable global energy
system. Energy Strategy Reviews, 1 (2), 109–121. doi: 10.1016/j.esr.2012.07.003
DHPA. (2010). Veel gestelde vragen over warmtepompen. Dutch Heat Pump Association. Retrieved
2016-05-23, from http://www.dhpa-online.nl/veel-gestelde-vragen/
Diakov, V., Ruth, M., James, B. D., Perez, J., & Spisak, A. B. (2011). Technical breakthrough points
and opportunities in transition scenarios for hydrogen as vehicular fuel (Tech. Rep.). Golden,
Colorado: NREL.
EDF Energy. (2016). Electricity infrastructure. Retrieved 2016-09-23, from https://www.edfenergy
.com/sites/default/files/generation-and-supply-electricity_3.png
Edwards, R., Larivé, J.-F., Rickeard, D., & Weindorf, W. (2014). Well-to-wheels analysis of future
automotive fuels and powertrains in the European context (4.a ed.; S. Godwin; et al., Eds.).
Luxembourg: Joint Research Centre (JRC). Retrieved from http://iet.jrc.ec.europa.eu/
about-jec/ doi: 10.2790/95629
Energinet.dk. (2011). planSOEC (Tech. Rep. No. May). Erritsø.
Energistyrelsen, & Energinet.dk. (2015). Technology Data for Energy Plants – Generation of Electricity
and District Heating, Energy Storage and Energy Carrier Generation and Conversion (Tech. Rep.).
Energy Company. (2015). HRe-ketels en brandstofcellen voor Gelderland. Retrieved from http://
energy-company.nl/hre-ketels-en-brandstofcellen-voor-gelderland/
ENTSOE. (2016). Installed Capacity per Production Unit. Retrieved from https://transparency
.entsoe.eu/generation/r2/installedGenerationCapacityAggregation/show
E.ON. (2014). Warme winter = lagere stookkosten. Retrieved 2016-03-22, from http://magazine
.eon.nl/media/11659061/20140428_Grafiek-energieverbruik-umbraco.jpg
Erdinc, O., & Uzunoglu, M. (2012). Optimum design of hybrid renewable energy systems: Overview
of different approaches. Renewable and Sustainable Energy Reviews, 16 (3), 1412–1425. doi:
10.1016/j.rser.2011.11.011
European Commission. (2016). Community Research and Development Information System. Retrieved
from http://cordis.europa.eu
Ewan, B. C. R., & Adeniyi, O. D. (2013). A Demonstration of Carbon-Assisted Water Electrolysis.
Energies, 6 (3), 1657–1668. doi: 10.3390/en6031657
Fang, Q., Blum, L., Peters, R., Peksen, M., Batfalsky, P., & Stolten, D. (2015). SOFC stack performance
under high fuel utilization. International Journal of Hydrogen Energy , 40 (2), 1128–1136. doi:
10.1016/j.ijhydene.2014.11.094
Fardadi, M., McLarty, D. F., & Jabbari, F. (2016). Investigation of thermal control for different SOFC
flow geometries. Applied Energy , 178 , 43–55. doi: 10.1016/j.apenergy.2016.06.015
Fourer, R., & Kernighan, B. W. (2002). AMPL: A Modeling Language for Mathematical Programming.
Duxbury Press.
Frangopoulos, C. A. (2003). Methods of energy systems optimization. In Summer school: Optimization
of energy systems and processes. Gliwice.

132
Fuel Cells Bulletin. (2016). Sunfire supplies Boeing with largest reversible solid oxide electrolyser/fuel
cell system (Vol. 2016) (No. 2). Elsevier. doi: 10.1016/S1464-2859(16)70002-2
Geels, F. W. (2002). Technological transitions as evolutionary reconfiguration processes: a multi-level
perspective and a case-study. , 31 , 1257–1274.
Geels, F. W. (2010, may). Ontologies, socio-technical transitions (to sustainability), and the multi-level
perspective. Research Policy , 39 (4), 495–510. doi: 10.1016/j.respol.2010.01.022
Geels, F. W. (2012). A socio-technical analysis of low-carbon transitions: introducing the multi-
level perspective into transport studies. Journal of Transport Geography , 24 , 471–482. doi:
10.1016/j.jtrangeo.2012.01.021
Geels, F. W., & Schot, J. (2007). Typology of sociotechnical transition pathways. , 36 (August 2003),
399–417. doi: 10.1016/j.respol.2007.01.003
Geidl, M. (2007). Integrated Modeling and Optimization of Multi-Carrier Energy Systems (Unpublished
doctoral dissertation). ETH Zurich.
Gençer, E., Al-musleh, E., Mallapragada, D. S., & Agrawal, R. (2014). Uninterrupted renewable
power through chemical storage cycles. Current Opinion in Chemical Engineering, 5 , 29–36. doi:
10.1016/j.coche.2014.04.001
Gerdes, J., Marbus, S., & Boelhouwer, M. (2014). Energietrends 2014 (Tech. Rep.). Petten: ECN,
Energie-Nederland en Netbeheer Nederland.
Giddey, S., Badwal, S., Kulkarni, A., & Munnings, C. (2012, jun). A comprehensive review of direct
carbon fuel cell technology. Progress in Energy and Combustion Science, 38 (3), 360–399. doi:
10.1016/j.pecs.2012.01.003
Gómez, D. R., Watterson, J. D., Americanohia, B. B., Ha, C., Marland, G., Matsika, E., . . . Treanton, K.
(2006). Stationary Combustion. In Guidelines for national greenhouse gas inventories (2nd ed.,
p. 47). IPCC. Retrieved from http://www.ipcc-nggip.iges.or.jp/public/2006gl/pdf/
2_Volume2/V2_2_Ch2_Stationary_Combustion.pdf doi: 10.1016/S0166-526X(06)47021-5
Gómez, S. Y., & Hotza, D. (2016). Current developments in reversible solid oxide fuel cells. Renewable
and Sustainable Energy Reviews, 61 , 155–174. doi: 10.1016/j.rser.2016.03.005
Graves, C., Ebbesen, S. D., Jensen, S. H., Simonsen, S. B., & Mogensen, M. B. (2015). Eliminating
degradation in solid oxide electrochemical cells by reversible operation. Nature Materials, 14 (2),
239–244. doi: 10.1038/nmat4165
GTS. (2014). Safety sheet: Aardgas (geodoriseerd) met Tetra Hydro Thiofeen (THT) (Tech. Rep.).
Groningen: Gasunie Transport Services.
Hart, D., Lehner, F., Rose, R., Lewis, J., & Klippenstein, M. (2015). The Fuel Cell Industry Review
2015 (Tech. Rep.). Lausanne: E4tech.
Hauptmeier, K., Penkuhn, M., & Tsatsaronis, G. (2016). Economic assessment of a solid oxide fuel cell
system for biogas utilization in sewage plants. Energy . doi: 10.1016/j.energy.2016.05.072
Hawkes, A. D., Aguiar, P., Hernandez-Aramburo, C. A., Leach, M. A., Brandon, N. P., Green, T. C.,
& Adjiman, C. S. (2006, jun). Techno-economic modelling of a solid oxide fuel cell stack for
micro combined heat and power. Journal of Power Sources, 156 (2), 321–333. doi: 10.1016/
j.jpowsour.2005.05.076
Hawkes, A. D., Brett, D. J. L., & Brandon, N. P. (2009, dec). Fuel cell micro-CHP techno-economics:
Part 1-model concept and formulation. International Journal of Hydrogen Energy , 34 (23), 9545–
9557. doi: 10.1016/j.ijhydene.2009.09.094
Hemmes, K. (2004). Fuel Cells. In R. E. White, B. E. Conway, & C. G. Vayenas (Eds.), Modern aspects
of electrochemistry (37th ed., p. 131). New York: Kluwer Academic; Plenum Publishers.
Hemmes, K. (2009). Towards multi-source multi-product and other integrated energy systems. Inter-
national Journal of Integrated Energy Systems, 1 (1), 1–15.
Hemmes, K., Patil, A., & Woudstra, N. (2005). Internal reforming SOFC system for flexible coproduction
of hydrogen and power. In 3rd international conference on fuel cell science, engineering and
technology (pp. 577–582). Ypsilanti, Michigan: ASME. doi: 10.1115/FUELCELL2005-74153
Holmgren, M. A. (2012). Methane emissions from biogas plants – The Swedish voluntary
agreement system and sustainability criteria. In Nordic biogas conference 2012. Retrieved

133
 from http://www.slideshare.net/oznut/methane-emissions-from-biogas-plants-the
-swedish-voluntary-agreement-system-and-sustainability-criteria
HSL. (n.d.). Oxford: STFC Rutherford Appleton Laboratory. Retrieved from http://www.hsl.rl.ac
.uk/
Huo, H.-B., Zhong, Z.-D., Zhu, X.-J., & Tu, H.-Y. (2008, jan). Nonlinear dynamic modeling for a
SOFC stack by using a Hammerstein model. Journal of Power Sources, 175 (1), 441–446. doi:
10.1016/j.jpowsour.2007.09.059
IEA. (2015). Energy and climate change. World Energy Outlook Special Report, 1–200. doi: 10.1038/
479267b
IEA-ETSAP, & IRENA. (2013a). Production of Liquid Biofuels: Technology Brief.
doi: 10.1111/j.1745-4514.2010.00447.x
IEA-ETSAP, & IRENA. (2013b). Thermal Energy Storage. Technology Brief.
INL. (2015). Plugged In: How Americans Charge Their Electric Vehicles (Tech. Rep.). Idaho National
Laboratory.
IPCC. (2007). Climate Change 2007: Mitigation. In B. Metz, O. Davidson, P. Bosch, R. Dave, &
L. Meyer (Eds.), Contribution of working group iii to the fourth assessment report of the ipcc.
Cambridge; New York: Cambridge University Press.
Irvine, J. T. S., Neagu, D., Verbraeken, M. C., Chatzichristodoulou, C., Graves, C., & Mogensen, M. B.
(2016). Evolution of the electrochemical interface in high-temperature fuel cells and electrolysers.
Nature Energy , 1 (1). doi: 10.1038/nenergy.2015.14
James, B. D., Spisak, A. B., & Colella, W. G. (2012). Manufacturing Cost Analysis of Stationary Fuel
Cell Systems (Tech. Rep.). Arlington: Strategic Analysis Inc.
Janardhanan, V. M., Heuveline, V., & Deutschmann, O. (2007, oct). Performance analysis of a SOFC
under direct internal reforming conditions. Journal of Power Sources, 172 (1), 296–307. doi:
10.1016/j.jpowsour.2007.07.008
Jensen, S. H., Graves, C., Mogensen, M. B., Wendel, C. H., Braun, R. J., Hughes, G., . . . Barnett,
S. A. (2015). Large-scale electricity storage utilizing reversible solid oxide cells combined with
underground storage of CO2 and CH4. Energy Environ. Sci., 8 (8), 2471–2479. doi: 10.1039/
C5EE01485A
Kamjoo, A., Maheri, A., Dizqah, A. M., & Putrus, G. A. (2016). Electrical Power and Energy Systems
Multi-objective design under uncertainties of hybrid renewable energy system using NSGA-II and
chance constrained programming. International Journal of Electrical Power and Energy Systems,
74 , 187–194. doi: 10.1016/j.ijepes.2015.07.007
Kaviani, A. K., Riahy, G. H., & Kouhsari, S. H. M. (2009). Optimal design of a reliable hydrogen-based
stand-alone wind/PV generating system, considering component outages. Renewable Energy ,
34 (11), 2380–2390. doi: 10.1016/j.renene.2009.03.020
Kemp, R., Schot, J., & Hoogma, R. (1998). Regime shifts to sustainability through processes of
niche formation: The approach of strategic niche management. Technology Analysis & Strategic
Management, 10 (2), 175–198. doi: 10.1080/09537329808524310
doi: 10.1080/09537329808524310
Kikuchi, Y., Kimura, S., Okamoto, Y., & Koyama, M. (2014). A scenario analysis of future energy
systems based on an energy flow model represented as functionals of technology options. Applied
Energy , 132 , 586–601. doi: 10.1016/j.apenergy.2014.07.005
Kleijn, R., & Van der Voet, E. (2010). Resource constraints in a hydrogen economy based on renewable
energy sources: An exploration. Renewable and Sustainable Energy Reviews, 14 (9), 2784–2795.
doi: 10.1016/j.rser.2010.07.066
KNMI. (2005). HYDRA MetaDataConnection (Tech. Rep.). DeBilt: Royal Netherlands Meteorological
Institute. Retrieved from http://projects.knmi.nl/hydra/cgi-bin/meta_data.cgi
Komatsu, Y., Brus, G., Kimijima, S., & Szmyd, J. (2014). The effect of overpotentials on the transient
response of the 300W SOFC cell stack voltage. Applied Energy , 115 , 352–359. doi: 10.1016/
j.apenergy.2013.11.017
Koppejan, J., Elbersen, W., Meeusen, M., & Bindraban, P. (2009). Beschikbaarheid van Nederlandse

134
 biomassa voor elektriciteit en warmte in 2020. (November). Retrieved from http://edepot.wur
.nl/51989
Kuhn, J., & Kesler, O. (2015, mar). Carbon deposition thresholds on nickel-based solid oxide fuel cell
anodes I. Fuel utilization. Journal of Power Sources, 277 , 443–454. doi: 10.1016/j.jpowsour.2014
.07.085
Lamers, P., Hoefnagels, R., Junginger, M., Hamelinck, C., & Faaij, A. (2015, jul). Global solid biomass
trade for energy by 2020: an assessment of potential import streams and supply costs to North-
West Europe under different sustainability constraints. GCB Bioenergy , 7 (4), 618–634. doi:
10.1111/gcbb.12162
Lamnatou, C., Notton, G., Chemisana, D., & Cristofari, C. (2015). The environmental performance of
a building-integrated solar thermal collector, based on multiple approaches and life-cycle impact
assessment methodologies. Building and Environment, 87 , 45–58. doi: 10.1016/j.buildenv.2015
.01.011
Lazard. (2015a). Lazard’s levelized cost of energy analysis. , 9.0 .
Lazard. (2015b). Lazard’s levelized cost of storage analysis. , 1.0 .
López, J. (2013). Combustion engine vs. gas turbine: Part load efficiency and flexibil-
ity. Retrieved 2016-01-01, from http://www.wartsila.com/energy/learning-center/
technical-comparisons/combustion-engine-vs-gas-turbine-part-load-efficiency
-and-flexibility
Lozano, M. A., Ramos, J. C., & Serra, L. M. (2010). Cost optimization of the design of CHCP
(combined heat, cooling and power) systems under legal constraints. Energy , 35 (2), 794–805.
doi: 10.1016/j.energy.2009.08.022
Lund, H. (2007). EnergyPLAN–Advanced Energy Systems Analysis Computer Model. Documentation
Version 7.0. Aalborg University, Denmark. Retrieved from http://www.energyplan.eu
Luo, Y., Shi, Y., Li, W., Ni, M., & Cai, N. (2014). Elementary reaction modeling and experimen-
tal characterization of solid oxide fuel-assisted steam electrolysis cells. International Journal of
Hydrogen Energy , 39 (20), 10359–10373. doi: 10.1016/j.ijhydene.2014.05.018
MacKay, D. J. C. (2009). Sustainable Energy–Without the hot air. Cambridge: UIT Cambridge.
Retrieved from withoutthehotair.com
Mahato, N., Banerjee, A., Gupta, A., Omar, S., & Balani, K. (2015). Progress in material selection
for solid oxide fuel cell technology: A review. Progress in Materials Science, 72 , 141–337. doi:
10.1016/j.pmatsci.2015.01.001
Makhorin, A. O. (n.d.). GNU Linear Programming Kit (GLPK). Moscow: GNU Project. Retrieved
from http://www.gnu.org/software/glpk
Mancarella, P., Andersson, G., Peças-Lopes, J. A., & Bell, K. R. (2016, jun). Modelling of inte-
grated multi-energy systems: Drivers, requirements, and opportunities. In 2016 power systems
computation conference (pscc). Genoa: IEEE. doi: 10.1109/PSCC.2016.7541031
Manné, D. J. (2009). The market introduction of Superwind in the Netherlands (Master thesis). TU
Delft.
Marina, O. A., Pederson, L. R., Williams, M. C., Coffey, G. W., Meinhardt, K. D., Nguyen, C. D., &
Thomsen, E. C. (2007). Electrode performance in reversible solid oxide fuel cells. Journal of The
Electrochemical Society , 154 (5), 452–459. doi: 10.1149/1.2710209
Marsboom, P.-J. (2013). Forecast and actual solar-PV power generation (Tech. Rep.). Elia.
Martinez-Frias, J., Pham, A.-Q., & Aceves, S. M. (2003). A natural gas-assisted steam electrolyzer
for high-efficiency production of hydrogen. International Journal of Hydrogen Energy , 28 (5),
483–490. doi: 10.1016/S0360-3199(02)00135-0
Mathiesen, B. V. (2008). Fuel cells and electrolysers in future energy systems (PhD, Aalborg University).
Retrieved from http://people.plan.aau.dk/~bvm/FinalWebVersion3.pdf
Mathiesen, B. V., Lund, H., & Karlsson, K. (2009). IDA’s climate plan 2050 – Background report.
Copenhagen V.
Mathworks. (2015). Mixed-Integer Linear Programming Algorithms. Retrieved 2016-01-02,
from http://nl.mathworks.com/help/optim/ug/mixed-integer-linear-programming

135
 -algorithms.html
McPhail, S. J., Aarva, A., Devianto, H., Bove, R., & Moreno, A. (2011). SOFC and MCFC: Com-
monalities and opportunities for integrated research. International Journal of Hydrogen Energy ,
36 (16), 10337–10345. doi: 10.1016/j.ijhydene.2010.09.071
Meibom, P., & Karlsson, K. (2010). Role of hydrogen in future North European power system in 2060.
International Journal of Hydrogen Energy , 35 (5), 1853–1863. doi: 10.1016/j.ijhydene.2009.12
.161
Milieucentraal. (2016). Kosten en besparing zonneboiler. Retrieved 2016-07-01,
from https://www.milieucentraal.nl/energie-besparen/energiezuinig-huis/
energiezuinig-verwarmen-en-warm-water/zonneboiler
Minh, N. Q., & Mogensen, M. B. (2013). Reversible solid oxide fuel cell technology for green fuel and
power production. Interface, 22 (4), 55–62. Retrieved from http://www.electrochem.org/dl/
interface/wtr/wtr13/wtr13_p055_062.pdf doi: 10.1149/2.F05134if
Ministry of Economic Affairs. (2011). Green Deal. Retrieved 2016-11-20, from https://www
.government.nl/topics/energy-policy/contents/green-deal
Molero-Sánchez, B., Prado-Gonjal, J., Ávila-Brande, D., Chen, M., Morán, E., & Birss, V. (2015).
High performance La0.3Ca0.7Cr0.3Fe0.7O3–δ air electrode for reversible solid oxide fuel cell
applications. International Journal of Hydrogen Energy , 40 (4), 1902–1910. doi: 10.1016/
j.ijhydene.2014.11.127
Mulder, F. M. (2014). Implications of diurnal and seasonal variations in renewable energy generation
for large scale energy storage. Journal of Renewable and Sustainable Energy , 6 (3), 33105. doi:
10.1063/1.4874845
Murashkina, A. A., Maragou, V. I., Demin, A. K., Pikalova, E. Y., & Tsiakaras, P. E. (2008). Hydrogen
production aided solid oxide electrochemical reformer fed with octane: A theoretical analysis.
Journal of Power Sources, 181 (2), 304–312. doi: 10.1016/j.jpowsour.2007.12.069
Nagle, A. (2008). Polarization curve of PEM Fuel Cell. Roads2HyCom. Re-
trieved from https://www.ika.rwth-aachen.de/r2h/images/thumb/c/cc/Polarization
_curve_neu2.jpg/450px-Polarization_curve_neu2.jpg
Napoli, R., Gandiglio, M., Lanzini, A., & Santarelli, M. (2015). Techno-economic analysis of PEMFC
and SOFC micro-CHP fuel cell systems for the residential sector. Energy and Buildings, 103 ,
131–146. doi: 10.1016/j.enbuild.2015.06.052
Netbeheer Nederland. (2016). Energietransitie Rekenmodellen. Retrieved 2016-09-30,
from http://nbn-assets.netbeheernederland.nl/p/32768//files/ETRM_Interactieve
-Infographic_A.pdf
oemof. (2016). Retrieved from https://github.com/oemof (Open Energy Modelling Framework)
Omosun, A., Bauen, A., Brandon, N., Adjiman, C., & Hart, D. (2004). Modelling system efficiencies
and costs of two biomass-fuelled SOFC systems. Journal of Power Sources, 131 (1–2), 96–106.
doi: 10.1016/j.jpowsour.2004.01.004
Ouweltjes, J. P., Aravind, P. V., Woudstra, N., & Rietveld, G. (2006). Biosyngas utilization in solid
oxide fuel cells with Ni/GDC anodes. Journal of Fuel Cell Science and Technology , 3 (4), 495–498.
doi: 10.1115/1.2349535
Park, J., Li, P., & Bae, J. (2012). Analysis of chemical, electrochemical reactions and thermo-fluid
flow in methane-feed internal reforming SOFCs: Part II-temperature effect. International Journal
of Hydrogen Energy , 37 (10), 8532–8555. doi: 10.1016/j.ijhydene.2012.02.109
Patcharavorachot, Y., Thongdee, S., Saebea, D., Authayanun, S., & Arpornwichanop, A. (2016). Per-
formance comparison of solid oxide steam electrolysis cells with/without the addition of methane.
Energy Conversion and Management, 120 , 274–286. doi: 10.1016/j.enconman.2016.04.100
PBL. (2011). Naar een schone economie in 2050: routes verkend.
Penchini, D., Cinti, G., Discepoli, G., Sisani, E., & Desideri, U. (2013). Characterization of a 100 W
SOFC stack fed by carbon monoxide rich fuels. International Journal of Hydrogen Energy , 38 (1),
525–531. doi: 10.1016/j.ijhydene.2012.09.060
Pfeifer, T., Nousch, L., Lieftink, D., & Modena, S. (2013, jan). System design and process layout for a

136
 SOFC micro-CHP unit with reduced operating temperatures. International Journal of Hydrogen
Energy , 38 (1), 431–439. doi: 10.1016/j.ijhydene.2012.09.118
Pham, A.-Q., Wallman, P. H., & Glass, R. S. (2000, apr). Supplying natural gas to anode side of
electrolyzer to lower electricity consumption by either providing an auxilliary source of hydrogen
resulting from syngas formation and then water-shifting or by burning out oxygen by-product on
anode. Google Patents. (US Patent 6,051,125)
Prag, C. B., Hallinder, J., Nielsen, E. R., King, A., Santarelli, M., Drago, D., & Tudoroiu-Lakavičė,
A. (2016). Non-economic barriers to large-scale market uptake of fuel cell based micro-CHP
technology: preliminary report. ene.field.
Pruitt, K. A., Braun, R. J., & Newman, A. M. (2013, nov). Establishing conditions for the economic
viability of fuel cell-based, combined heat and power distributed generation systems. Applied
Energy , 111 , 904–920. doi: 10.1016/j.apenergy.2013.06.025
Remick, R. J., & Wheeler, D. (2011). Reversible Fuel Cells Workshop Summary Report. Arlington:
U.S. Department of Energy.
Ridjan, I., Mathiesen, B. V., Connolly, D., & Duić, N. (2013). The feasibility of synthetic fuels in
renewable energy systems. Energy , 57 , 76–84. doi: 10.1016/j.energy.2013.01.046
Rivera-Tinoco, R., Schoots, K., & Van der Zwaan, B. (2012). Learning curves for solid oxide fuel cells.
Energy Conversion and Management, 57 , 86–96. doi: 10.1016/j.enconman.2011.11.018
Ros, J., & Prins, A. G. (2014, nov). Biomass: Wishes and limitations. Retrieved 2016-01-10, from
http://infographics.pbl.nl/biomass
Santoianni, D. (2015). Setting the benchmark: The world’s most efficient coal-fired power plants.
Cornerstone, 3 (1). Retrieved from http://cornerstonemag.net/setting-the-benchmark
-the-worlds-most-efficient-coal-fired-power-plants
Sauter, R., & Watson, J. (2007). Strategies for the deployment of micro-generation: Implications for
social acceptance. Energy Policy , 35 (5), 2770–2779. doi: 10.1016/j.enpol.2006.12.006
Scataglini, R., Mayyas, A., Wei, M., Chan, S. H., Lipman, T., Gosselin, D., . . . James, B. D. (2015).
A Total Cost of Ownership Model for Solid Oxide Fuel Cells in Combined Heat and Power and
Power-Only Applications (Tech. Rep.). Berkeley: Lawrence Berkeley National Laboratory.
Schenk, N. J., Moll, H. C., Potting, J., & Benders, R. M. (2007). Wind energy, electricity, and hydrogen
in the Netherlands. Energy , 32 (10), 1960–1971. doi: 10.1016/j.energy.2007.02.002
Schepers, B. L., Naber, N. R., Rooijers, F. J., & Leguijt, C. (2015). Op weg naar een klimaatneutrale
woningvoorraad in 2050 (Tech. Rep.). Delft: CE Delft. Retrieved from http://www.ce.nl/
?go=home.downloadPub&id=1638&file=CE_Delft_3A31_Klimaatneutrale_gebouwde
Schey, S., Scoffield, D., & Smart, J. (2012). A First Look at the Impact of Electric Vehicle Charging
on the Electric Grid in The EV Project. In International battery, hybrid and fuel cell electric
vehicle symposium (evs26). Retrieved from http://www.inl.gov/technicalpublications/
Documents/5394115.pdf
Schlecht, I., & Davis, C. (2016). Open models. Retrieved 2016-06-01, from http://wiki.openmod
-initiative.org/wiki/Open_Models
Schoenung, S. (2011). Economic analysis of large-scale hydrogen storage for renewable utility appli-
cations (Tech. Rep.). Albuquerque; Livermore: Sandia National Laboratories. Retrieved from
http://prod.sandia.gov/techlib/access-control.cgi/2011/114845.pdf
Schoots, K., & Hammingh, P. (2015). Nationale Energieverkenning 2015 (Vol. ECN-O-15-0; Tech.
Rep.). Petten: Energieonderzoek Centrum Nederland (ECN), Planbureau voor de Leefomgev-
ing (PBL), Centraal Bureau voor de Statistiek (CBS), Rijksdienst voor Ondernemend Neder-
land (RVO). Retrieved from http://www.cbs.nl/NR/rdonlyres/C0F1D74C-1EB1-4E3B-9A46
-ECA28ABFED7E/0/2015nationaleenergieverkenning.pdf
Shahmohammadi, A., Moradi-Dalvand, M., Ghasemi, H., & Ghazizadeh, M. S. (2015). Optimal Design
of Multicarrier Energy Systems Considering Reliability Constraints. IEEE Transactions on Power
Delivery , 30 (c), 878–886. doi: 10.1109/TPWRD.2014.2365491
Smith, A., & Raven, R. (2012). What is protective space? Reconsidering niches in transitions to
sustainability. Research Policy , 41 (6), 1025–1036. doi: 10.1016/j.respol.2011.12.012

137
Smith, A., Stirling, A., & Berkhout, F. (2005). The governance of sustainable socio-technical transitions.
Research Policy , 34 (10), 1491–1510. doi: 10.1016/j.respol.2005.07.005
Sociaal-Economische Raad. (2013). Energieakkoord voor duurzame groei (Tech. Rep.). Den Haag.
Retrieved from http://www.energieakkoordser.nl
Sørensen, B. (2008). A sustainable energy future: Construction of demand and renewable energy supply
scenarios. International Journal of Energy Research, 32 (5), 436–470. doi: 10.1002/er.1375
Spoelstra, J. C. (2014). Charging behaviour of Dutch EV drivers (Master thesis). Utrecht University.
Spoelstra, S. (2008). De Nederlandse en industriële energiehuishouding van 2000 tot en met 2006
(Tech. Rep.). Petten: ECN.
Sprik, S., Kurtz, J., Ainscough, C., & Peters, M. (2015). Next Generation Hydrogen Station Com-
posite Data Products (Tech. Rep.). Columbus: NREL. Retrieved from http://www.nrel.gov/
hydrogen/proj_infrastructure_analysis.html
Staffell, I., & Green, R. (2013). The cost of domestic fuel cell micro-CHP systems. International Journal
of Hydrogen Energy , 38 (2), 1088–1102. doi: 10.1016/j.ijhydene.2012.10.090
Standaert, F. R. A. M. (1998). Analytical fuel cell modelling and exergy analysis of fuel cells (Doctoral
dissertation). TU Delft.
Steup, B. J. (2014). Design and Optimisation of Renewable Energy Systems using Matrices (Master
thesis). TU Delft.
Steup, B. J., & Hemmes, K. (2015, sep). The Bing Model; Let the computer design the optimum
energy system configuration.
Strachan, N., Balta-Ozkan, N., Joffe, D., McGeevor, K., & Hughes, N. (2009). Soft-linking energy
systems and GIS models to investigate spatial hydrogen infrastructure development in a low-
carbon UK energy system. International Journal of Hydrogen Energy , 34 (2), 642–657. doi:
10.1016/j.ijhydene.2008.10.083
Strazza, C., Del Borghi, A., Costamagna, P., Gallo, M., Brignole, E., & Girdinio, P. (2015). Life Cycle
Assessment and Life Cycle Costing of a SOFC system for distributed power generation. Energy
Conversion and Management, 100 , 64–77. doi: 10.1016/j.enconman.2015.04.068
Stryi-Hipp, G. (2009). Solar heating and Cooling for a Sustainable Energy Future in Europe
(Revised) (Tech. Rep.). Brussels: European Solar Thermal Technology Platform (ESTTP).
Retrieved from http://www.estif.org/fileadmin/estif/content/projects/downloads/
ESTTP_SRA_RevisedVersion.pdf
Sveshnikova, A. (2015). Estimation of possibility to implement fuel cell technology for decentralized
energy supply in Russia (Master thesis, KTH Stockholm). Retrieved from http://kth.diva
-portal.org/smash/get/diva2:852241/FULLTEXT01.pdf
Tao, G. G. (2007, mar). A reversible planar solid oxide fuel-fed electrolysis cell and solid oxide fuel cell
for hydrogen and electricity production operating on natural gas/biomass fuels (Tech. Rep.). Salt
Lake City: Materials and Systems Research, Inc. (MSRI). Retrieved from http://www.osti.gov/
servlets/purl/934689-auwPAL/ doi: 10.2172/934689
Tao, G. G., Butler, B., & Virkar, A. V. (2011). Hydrogen and Power by Fuel-Assisted Electrolysis Using
Solid Oxide Fuel Cells. ECS Transactions, 35 (1), 2929–2939. doi: 10.1149/1.3570292
Tao, G. G., Virkar, A., & Garland, R. (2006). A reversible planar solid oxide fuel-assisted electrolysis cell
and solid oxide fuel cell for hydrogen and electricity production operating on natural gas/biogas.
In Fy annual progress report (pp. 24–28).
TenneT. (2016). Measurement data. Retrieved 18-3-2016, from http://www.tennet.org/english/
operational_management/export_data.aspx
Teske, S., Zervos, A., Lins, C., & Muth, J. (2010). Energy (r)evolution; A sustainable global energy
outlook (3rd ed.). Retrieved from www.energyblueprint.info
Thijssen, J. (2011). Solid oxide fuel cells and critical materials: A review of im-
plications (Tech. Rep.). Redmond: National Energy Technology Laboratory. Re-
trieved from https://www.netl.doe.gov/FileLibrary/research/coal/energysystems/
fuelcells/Rare-Earth-Update-for-RFI-110523final.pdf
Trendewicz, A., & Braun, R. J. (2013, jul). Techno-economic analysis of solid oxide fuel cell-based

138
 combined heat and power systems for biogas utilization at wastewater treatment facilities. Journal
of Power Sources, 233 , 380–393. doi: 10.1016/j.jpowsour.2013.01.017
Ud Din, Z., & Zainal, Z. A. (2016). Biomass integrated gasification-SOFC systems: Technology
overview. Renewable and Sustainable Energy Reviews, 53 , 1356–1376. doi: 10.1016/j.rser.2015
.09.013
US Fuel Cell Council. (2007). Introduction to Solid Oxide Fuel Cell Button Cell Testing (Tech.
Rep.). Washington D.C.. Retrieved from http://ftp.fchea.org/core/import/PDFs/
TechnicalResources/SOFCFG-ButtonCellTesting-07-015.pdf
Van den Bergh, K., Delarue, E., & D’haeseleer, W. (2013, may). The impact of renewable injections
on cycling of conventional power plants. In 2013 10th international conference on the european
energy market (eem) (pp. 1–8). IEEE. doi: 10.1109/EEM.2013.6607322
Van den Hil, E. (2015). Multi-objective optimisation of integrated community energy systems and
assessment of the impact on households (Master thesis). TU Delft.
Van Herle, J., Membrez, Y., & Bucheli, O. (2004). Biogas as a fuel source for SOFC co-generators.
Journal of Power Sources, 127 (1–2), 300–312. doi: 10.1016/j.jpowsour.2003.09.027
Van Leeuwen, I. (2015). Feasibility of combined heat, hydrogen and power production by molten
carbonate fuel cells.
Van Leeuwen, I. (2016). Feasibility of combined heat, hydrogen and power production by molten
carbonate fuel cells (Master thesis). TU Delft.
Van Wortswinkel, L., & Nijs, W. (2010). Industrial combustion boilers. Technology Brief (1).
Van Bruggen, A., O’Connor, M., Ding, Z., Gao, Y., Paulina, A., Rojas, G., & Starmans, Z. (2016). The
role of the citizen-oriented initiatives in the dutch energy transition: Four archetypes. Leiden.
Van Gurp, T. (2016). Fastned: Laadstations bereiken medio 2017 break-even punt.
Retrieved 2016-10-01, from http://www.tankpro.nl/ondernemen/2016/01/11/fastned
-laadstations-bereiken-medio-2017-break-even-punt/
Van Melle, T., Menkveld, M., Oude Lohuis, J., De Smidt, R., & Terlouw, W. (2015). De systeemkosten
van warmte voor woningen (Tech. Rep.). Utrecht: Ecofys.
Veenstra, A. (2015). Ruimte voor zonne-energie in Nederland 2020-2050 (Tech. Rep.). Utrecht: Holland
Solar.
Verbeek, R., Van Zyl, S., Van Grinsven, A., & Van Essen, H. (2014). Brandstoffen voor het wegverkeer.
kenmerken en perspectief (Tech. Rep.). Delft: CE Delft; TNO.
Verbong, G., & Geels, F. (2007). The ongoing energy transition: Lessons from a socio-technical, multi-
level analysis of the Dutch electricity system (1960–2004). Energy Policy , 35 (2), 1025–1037. doi:
10.1016/j.enpol.2006.02.010
Vernay, A.-L., Manné, D. J., Steenvoorden, G., & Hemmes, K. (2008). Superwind: A feasibility study
(Tech. Rep. No. NEOH02010). Delft: SenterNovem, TU Delft.
Wachsman, E. D., Marlowe, C. A., & Lee, K. T. (2012). Role of solid oxide fuel cells in a balanced
energy strategy. Energy & Environmental Science, 5 (2), 5498–5509. doi: 10.1039/c1ee02445k
Wächter, A., & Biegler, T. L. (2006). On the implementation of an interior-point filter line-search
algorithm for large-scale nonlinear programming. Mathematical Programming, 106 (1), 25–57.
doi: 10.1007/s10107-004-0559-y
Warmerdam, J., Yildiz, I., & Koop, K. (2011). Biomassapotentieel Provincie Utrecht (Tech. Rep.).
Utrecht: Ecofys.
Wendel, C. H., Kazempoor, P., & Braun, R. J. (2016, jan). A thermodynamic approach for selecting
operating conditions in the design of reversible solid oxide cell energy systems. Journal of Power
Sources, 301 , 93–104. doi: 10.1016/j.jpowsour.2015.09.093
Wijers, F. K. B. (2011). Modeling a solid oxide fuel-assisted electrolysis cell in Cycle Tempo (Master
thesis). TU Delft.
Wikipedia. (2012). Gronings gas. Retrieved 2016-04-22, from https://nl.wikipedia.org/wiki/
Gronings_gas
Wikipedia. (2016a). Energy content of fuel. Retrieved 2016-02-10, from https://en.wikipedia.org/
wiki/Fuel_efficiency#Energy_content_of_fuel

139
Wikipedia. (2016b). Hoogrendementsketel. Retrieved from https://nl.wikipedia.org/wiki/
Hoogrendementsketel
Wikipedia. (2016c). Lijst van elektriciteitscentrales in Nederland. Retrieved 2016-09-26, from https://
nl.wikipedia.org/wiki/Lijst_van_elektriciteitscentrales_in_Nederland
Wiser, J. R., Schettler, J. W., & Willis, J. L. (2010). Evaluation of Com-
bined Heat and Power Technologies for Wastewater Facilities (Tech. Rep.). Colum-
bus: Brown and Caldwell. Retrieved from http://www.cwwga.org/documentlibrary/121
_EvaluationCHPTechnologiespreliminary[1].pdf
Yadav, S., Singh, M. K., & Sudhakar, K. (2015). Modelling of a solid oxide fuel cell. International
Journal of Scientific & Engineering Research, 6 (4), 834–841.

140
Appendix A

Data

A.1 Conversion factors

Energy carrier Energy content Energy density Source

(MJ/kg) (MJ/L)
Biodiesel 39.9 35.1 (Wikipedia, 2016a)
Crude oil 41.9 38.5 (Wikipedia, 2016a)
Diesel 48 38.6 (Wikipedia, 2016a)
Electricity 3.6 MJ/kWhe
Heavy fuel oil 43.05 (CBS, 2016a)
Hydrogen 141.86 (Wikipedia, 2016a)
Jet fuel, kerosene 47 37.6 (Wikipedia, 2016a)
Liquefied hydrogen 130 9.3 (Wikipedia, 2016a)
Liquefied petroleum gas a 51 27.1 (Wikipedia, 2016a)
Natural gas 42.2 35.17 × 10−3 (CBS, 2016a)
Steam coal 24.9 (CBS, 2016a)
a 60% propane and 40% butane

A.2 Technology data

A.2.1 Buses (Energy carriers)
Energy carrier uid excess shortage balanced
Biogas bbiog True False True
Biomass bbio False False True
Coal bcoal False False True
Electricity bele True False True
Heat <100 ℃ blowT True False True
Heat >100 ℃ bhighT False False True
Hydrogen bhyd True False True
Natural gas bgas False False True
Oil bfuel False False True

141
 A.2.2 Energy (re)sources
Class uid outputs add_outcapex life- wacc opex_fix opex_var co2_var co2_cap
_limit time
bus MW €/kW a €/kW/a €/MWh kgCO2 -eq./MWh gCO2 -eq./MW
Biomass products Commodity bio_prod bbio 7452 36.90 -341.35
Biomass residues Commodity bio_res bbio 7452 18.88 -237.81
Coal mine Commodity coal_prod bcoal 12.72 56.412
NG source Commodity gas_prod bgas 33.78 56.772
Oil well Commodity oil_prod bfuel 51.61 50.4
Biogas FixedSource WWT bgas 1522 977 25 0.08 17.95 -172.145
Nuclear FixedSource nuclear bele 4000 3971 60 0.05 68.99 10.17 9.18
PV commercial FixedSource pv2 bele 29757 810 40 0.08 10.35 0 0 1793.75
PV household FixedSource pv1 bele 52500 810 40 0.08 10.35 0 0 1793.75
Solar thermal FixedSource solar_therm blowT 260000 471 30 0.08 1.428 38.46
Wind offshore FixedSource wind_offshore bele 34000 2120 30 0.08 15 0 975
Wind onshore FixedSource wind_onshore bele 8000 900 30 0.08 8 0 591.25

142
A.2.3 Storage
For all: class=storage.Simple; wacc=0.05; cap_max=0; add_cap_limit=+inf; c_rate_in=1; c_rate_out=1.
uid inputs/ capex life opex_fix opex_var cap_max add_cap eta_in eta_out cap_loss
outputs -time _limit
€/kWh a €/kW/a €/kWh MWh MWh MWh/h
Compressed air caes bele 154 20 1400 3.6 0 +inf 1 0.43
energy storage
Flywheel flywheel bele 5410 20 69.41 0 +inf 1 0.83
H2 tank h2tank bhyd 2.5 15 0.01 0 +inf 1 0.98 8.93 · 10−3
Hot water tank lowT_sto blowT 0.0015 20 0 +inf 1 1 1.36 · 10−3
Li-ion battery battery bele 225 10 9.01 1.26 0 +inf 1 0.85 6.94 · 10−5
PCM highT_sto bhighT 30 20 2100 0 +inf 0.8 0.8 0.03
 A.2.4 Transformers
For all: out_max=0, add_out_limit=+inf.
transformer.Simple class
uid inputs outputs capex lifetime wacc opex_fix opex_var co2_var co2_cap eta
€/kW a €/kW/a €/MWh kgCO2 -eq./MWh gCO2 -eq./MW
Alkaline electrolyser alkel bele bhyd 0 10 0.05 0 0 0 0.82
Biodiesel production biodiesel bbio bfuel 3393 20 0.05 102.1 183.1 0.56
Biomass boiler bio_boil bbio bhighT 800 32.5 0.05 5.4 499.3 0.70
Biomass powerplant pp_bio bbio bele 1890 40 0.08 61.6 2.2 728.2 6.92 0.43
CCGT pp_gas bgas bele 790 25 0.05 30 2.5 336.9 10.81 0.6
Coal boiler coal_boil bcoal bhighT 800 32.5 0.05 5.4 384.4 0.85
Coal powerplant pp_coal bcoal bele 1890 40 0.05 61.6 2.2 728.2 6.92 0.43
Electric boiler ele_boil bele bhighT 75 15 0.05 0.5 1
Heat distribution dilute bhighT blowT 1309 40 0.05 1
NG boiler gas_boil bgas bhighT 100 32.5 0.05 3.7 0.42 256.7 0.75
PEMFC pemfc bhyd bele 207 20 0.05 0.35
SMR smr bgas bhyd 322 20 0.05 11.48 0.68 235.6 0.80

143
SOFC sofc bgas bele; bhyd 275 20 0.08 1.65 183.6 36.72
SOFC sofc bgas bele; bhyd 275 20 0.08 1.65 183.6 36.72
Upgrade upgrade bbiog bgas 300 15 0.08 56.25 10.1 0.95
transformer.TwoInputsOneOutput class
uid inputs outputs capex lifetime wacc opex_fix opex_var co2_var co2_cap eta f a

€/kW a €/kW/a €/MWh kgCO2 - gCO2 -

eq./MWh eq./MW
Methanation hydrogenation bhyd; bele bgas 900 20 0.08 22.14 48.36 0; 0.9 15.75
SOEC soec bele; bhighT bhyd 590 20 0.08 11.8 1.65 0 36.72 0.9; 0.9 5
a This parameter represents the ratio of the first and the second input of the process.
A.3 Existing powerplants

Table A.1: Construction year and installed capacities (in MW) of powerplants that are currently
in use in the Netherlands, per fuel type.

Year built Biomass Coal MSW NG Solar Uranium Offshore wind Onshore wind
1935 80
1956 81
1958
1960 45
1969 44
1972 60
1974 92 485
1975 332
1976 332
1978 638
1980
1983 400
1985 103 6
1987 1353
1988 520
1989 520 282
1990 196 217 1
1991 1 33
1992 50 18
1993 80 1 30
1994 1230 130 21
1995 48 58 98
1996 62 1609 1 46
1997 7 924 1 28
1998 460 2 39
1999 3 47
2000 25 231.9 4 37
2001 8 38
2002 43 5 187
2003 6 20 233
2004 1017 4 170
2005 1.2 29 5.8 1 149
2006 1.75 2 108 229
2007 77 410 1 188
2008 59 800 5 120 280
2009 40 10 73
2010 40 2175 21 15
2011 2.4 63 870 59 79
2012 2575 220 117
2013 1320 377 280
2014 50 302 152
2015 2360 437 129 394
Total 139 4630 649 16649 1485 530 357 3031

144
Appendix B

Companies

B.1 SOFC manufacturers

The most important international SOFC manufacturers are listed below, in alphabetical order (Hart et
al., 2015).

• Acumentrics SOFC (USA)

• Adelan Ltd. (UK)
• Bloom Energy Corporation (USA)
• Ceramic Fuel Cells Ltd. (Australia)
• Ceres Power Ltd. (UK)
• Convion Ltd. (Finland)
• Delphi Automotive plc (USA)
• Elcogen AS (Estonia)
• FuelCell Energy, Inc. (USA)
• Versa Power Systems, Inc. (USA)
• Hexis AG (Germany)
• Kyocera (Japan)
• NexTech Materials Ltd. (USA)
• Protonex Technology Corporation (USA)
• Rolls-Royce plc (UK)
• Sunfire GmbH (Germany)
• SOLIDpower (Italy/Switzerland)
• Topsoe Fuel Cell A/S (Denmark)
• Toto Ltd. (Japan)
• Ultra Electronics AMI (USA)
• Ztek Corporation (USA)

B.2 List of European SOFC research projects

Table B.1: List of EU-funded research projects registered in Cordis (European Commission, 2016).
Projects involving SOFCs, started between 1990 and 2016, are listed.

Project Dutch partners Start End

ACCEPT Advanced Lightweight Engineer- 2002 2005
ing BV
ECN 2002 2005
ASOF TNO 1990 1991

145
Project Dutch partners Start End
Bio-HyPP Gasterra BV 2015 2019
Micro Turbine Technology BV 2015 2019
Technische Universiteit Eindhoven 2015 2019
BIOCELLUS ECN 2004 2007
Technische Universiteit Delft 2004 2007
BIOFEAT ECN 2003 2006
CORE-SOFC ECN 2001 2004
DESIGN HyGear Fuel Cell Systems BV 2011 2014
Development of 50 kW class SOFC system and ECN 1996 2000
components
Development of a novel partial oxidation reactor Gastec NV 1997 1999
for NG and integration into a micro-CHP SOFC
system
Development of a SOFC ECN 1990 1993
Development of an advanced 1 kW SOFC proto- TNO 1992 1993
type
Development of co fired ceramic cells for low tem- DSM Resins NV 1994 1996
perature 800C SOFC operation
ECN 1994 1996
Development of SOFC flat plate reactor with metal ECN 1993 1995
separator plate for low temperature operation
DIAMOND HyGear Fuel Cell Systems BV 2014 2017
FC CHAIN HyGear Fuel Cell Systems BV 2006 2010
FC-DISTRICT ECN 2010 2014
FELICITAS Centre for Concepts in Mecha- 2005 2008
tronics BV (CCM)
Technische Universiteit Eindhoven 2005 2008
FLAME-SOFC ECN 2005 2010
FlexiFuel-SOFC HyGear Fuel Cell Systems BV 2015 2019
Technische Universiteit Delft 2015 2019
Universiteit Utrecht 2015 2019
Framework for system optimisation studies of in- ECN 1994 1995
tegrated fuel energy systems
GREEN-FUEL-CELL ECN 2004 2008
High temperature fuel cells use for the next ten Brandstofcel Nederland BV 1998 2000
years
ECN 1998 2000
Gastec NV 1998 2000
Netherland Agency for Energy and 1998 2000
the Environment (NOVEM)
Improving durability of SOFC stacks ECN 1996 1998
INNO-SOFC Energy Matters BV 2015 2018
IT-SOFC Technology ECN 1996 1998
Gastec NV 1996 1998
LOTUS HyGear Fuel Cell Systems BV 2011 2014
Manufacturing techniques for components of flat ECN 1991 1994
plate SOFC reactors
New manufacturing technologies for advanced TNO 1990 1992
SOFCs
New SOFC materials and technology TNO 1993 1995
Proof of feasibility of composite plate technology Business Unit of TNO Built Envi- 1993 1994
ronment and Geosciences

146
Project Dutch partners Start End
SOFC600 ECN 2006 2010
SOFCNET ECN 2003 2005
Shell Hydrogen BV 2003 2005
University Of Twente 2003 2005
SOFT-PACT Homa Software BV 2011 2014
SUAV HyGear Fuel Cell Systems BV 2011 2015
Thin electrolyte layer for SOFCs and electrode ma- University Of Twente 1991 1993
terials

147
Appendix C

Source Code

This appendix displays the code that was added to oemof to allow for the optimisation of energy systems
with SOC technologies. Modifications and additions were made for three purposes. First, objective
functions were defined that take the emission of GHGs into account. Second, new classes of components
were created to describe SOFCs and SOFECs. Third, sets of constraints are implemented to support the
class definitions. These additions are displayed below. The full source code can be found on the following
web pages: https://github.com/oemof/oemof/ and https://github.com/SanderNielen/oemof.

C.1 Objective functions

# -* - coding : utf -8 -* -
"""
This module contains a greenhouse gas minimisation objective .

@author : Sander van Nielen ( sander . vannielen@planet . nl )

"""
i m p o r t pyomo . environ as po
i m p o r t oemof . solph as solph

from .. core . network . entities i m p o r t Bus

from .. core . network . entities . components i m p o r t transformers as transformer
from .. core . network . entities . components i m p o r t sources as source

try :
i m p o r t obj ecti ve_e xpres sion s as objexpr
except :
from . i m p o r t obj ecti ve_e xpres sion s as objexpr

d e f set_ref ( block ) :
""" Determine the type ( capacity / input / output ) of the first entity of
‘ block ‘ , and return it .
Detects the property ‘ cap_max ‘ and ‘ in_max ‘ to classify an entity .
"""
f o r e i n block . objs :
try :
i f e . cap_max i s not None :
r e t u r n ’ capacity ’
e x c e p t AttributeError : p a s s
try :
i f e . in_max i s not None :
r e t u r n ’ input ’
e x c e p t AttributeError : p a s s
r e t u r n ’ output ’

# @author : Sander van Nielen .

d e f minimize_ghg ( self , cost_objects = None , revenue_objects = None ) :
""" Builds objective function that minimises the total GHG emissions .

148
 Emissions to be included are :
Construction emissions
Operating / combustion emissions

Parameters
- - - - - - - ---
self : pyomo model instance
cost_objects : array - like list containing classes of
objects of which GHG emissions should be included in
terms of objective function ( optional )
revenue_objects : only included for compatibility reasons .
"""
expr = 0
ghg_blocks = cost_objects

i f ghg_blocks i s None :
ghg_blocks = [ s t r ( transformer . Simple ) ,
s t r ( transformer . CHP ) ,
s t r ( transformer . Vari able Effi cienc yCHP ) ,
s t r ( transformer . SimpleExtractionCHP ) ,
s t r ( transformer . TwoInputsOneOutput ) ,
s t r ( transformer . Sofc ) ,
s t r ( transformer . Sofec ) ,
s t r ( transformer . Hybrid ) ,
s t r ( transformer . Storage ) ,
s t r ( source . FixedSource ) ,
s t r ( source . Commodity ) ,
s t r ( source . DispatchSource ) ]
# add additional transformers if needed

blocks = [ block f o r block i n self . block_data_objects ( active = True )

i f not i s i n s t a n c e ( block ,
solph . optimization_model . OptimizationModel ) ]

f o r block i n blocks :
i f ( block . name i n ghg_blocks ) and block . objs : # if block has emissions
ref = set_ref ( block )

# variable emissions
expr += objexpr . add_co2_var ( self , block , ref )
# fixed emissions
# expr += objexpr . add_co2_fix ( self , block , ref )
# construction emissions
i f block . optim ization_options . get ( ’ investment ’ , False ) :
expr += objexpr . add_co2_cap ( self , block , ref )

self . objective = po . Objective ( expr = expr )

# @author : Simon Hilpert . Modified by Sander van Nielen .

d e f minimize_cost ( self , cost_objects = None , revenue_objects = None ) :
""" Builds objective function that minimises the total costs , including
costs related to GHG emissions .

Costs included are :

opex_var ,
opex_fix ,
curtailment_costs ( dispatch sources ) ,
annualised capex ( investment components ) ,
env_opex_var , env_capex
Parameters
- - - - - - - ---
self : pyomo model instance
cost_blocks : array like
list containing classes of objects that are included in

149
 cost terms of objective function
revenue_blocks : array like
list containing classes of objects that are included in revenue
terms of objective function
"""
from .. core . network . entities i m p o r t ExcessSlack , ShortageSlack

expr = 0
c_blocks = cost_objects
r_blocks = revenue_objects
tax = self . co2_tax

i f cost_objects i s None :
c_blocks = [ s t r ( transformer . Simple ) ,
s t r ( transformer . CHP ) ,
s t r ( transformer . SimpleExtractionCHP ) ,
s t r ( transformer . Vari able Effi cienc yCHP ) ,
s t r ( transformer . TwoInputsOneOutput ) ,
s t r ( transformer . Sofc ) ,
s t r ( transformer . Sofec ) ,
s t r ( transformer . Hybrid ) ,
s t r ( transformer . Storage ) ,
s t r ( source . FixedSource ) ,
s t r ( source . Commodity ) ,
s t r ( source . DispatchSource ) ]
i f revenue_objects i s None :
r_blocks = []

blocks = [ block f o r block i n self . block_data_objects ( active = True )

i f not i s i n s t a n c e ( block ,
solph . optimization_model . OptimizationModel ) ]

f o r block i n blocks :
i f block . name i n c_blocks and block . objs :
ref = set_ref ( block )

# variable costs
i f ref == ’ capacity ’:
expr += objexpr . add_env_opex_var ( self , g e t a t t r ( self ,
s t r ( transformer . Storage ) ) , ’ input ’ , tax )
expr += objexpr . add_env_opex_var ( self , block , ’ output ’ ,
tax )
else :
expr += objexpr . add_env_opex_var ( self , block , ref , tax )
# fixed costs
i f block != s t r ( source . Commodity ) :
expr += objexpr . add_opex_fix ( self , block , ref ) * l e n ( self .
timesteps ) /8760
# investment costs
i f block . optimiz ation_ options . get ( ’ investment ’ , False ) :
expr += objexpr . add_env_capex ( self , block , ref , tax ) * l e n ( self .
timesteps ) /8760
i f h a s a t t r ( block , ’ z_start ’) :
expr += objexpr . ad d_startup_costs ( self , block )

# revenues
i f block . name i n r_blocks :
expr += objexpr . add_revenues ( self , block , ’ output ’)

# costs for dispatchable sources

i f h a s a t t r ( self , s t r ( source . DispatchSource ) ) :
expr += \
objexpr . a dd_c urtai lmen t_co sts ( self ,
g e t a t t r ( self , s t r ( source . DispatchSource ) ) )

150
 # artificial costs for excess or shortage
i f h a s a t t r ( self , s t r ( ExcessSlack ) ) :
expr += objexpr . add _e xc ess _s la ck _co st s ( self ,
g e t a t t r ( self , s t r ( ExcessSlack ) ) )
i f h a s a t t r ( self , s t r ( ShortageSlack ) ) :
expr += objexpr . a dd _ s h o rt a g e _s l a c k_ c o s ts ( self ,
g e t a t t r ( self , s t r ( ShortageSlack ) ) )

self . objective = po . Objective ( expr = expr * 8760/ l e n ( self . timesteps ) )

C.2 SOC classes

from . i m p o r t Transformer
i m p o r t logging
i m p o r t numpy as np
c l a s s Sofc ( Transformer ) :
# Added by Sander van Nielen
"""
A SOFC transformer that produces three outputs

Parameters
- - - - - - - ---
eta : list
constant effciency for converting input into output . First element of
list is used for conversion of input into first element of
attribute ‘ outputs ‘. Second element for second element of attribute
‘ outputs ‘. E . g . eta = [0.3 , 0.4]

"""
o p t i m i z ation_options = {}

d e f __init__ ( self , ** kwargs ) :

s u p e r () . __init__ (** kwargs )

self . eta = kwargs . get ( ’ eta ’ , [ None , None , None ])

c l a s s Sofec ( Transformer ) :
# Added by Sander van Nielen
"""
A SOFEC consumes electricity and methane to produce H2 and heat .

Parameters
- - - - - - - ---
eta : list
constant effciency for converting input into output . First element of
list is used for conversion of input into first element of
attribute ‘ outputs ‘. Second element for second element of attribute
‘ outputs ‘. E . g . eta = [0.3 , 0.4]

"""
o p t i m i z ation_options = {}

d e f __init__ ( self , ** kwargs ) :

s u p e r () . __init__ (** kwargs )

self . eta = kwargs . get ( ’ eta ’ , [ None , None ])

c l a s s Hybrid ( Transformer ) :
# Added by Sander van Nielen
"""
A hybrid SOFC / SOFEC transformer can switch between two operation modes .

Parameters

151
 - - - - - - - ---
eta : list
constant effciency for converting input into output . First element of
list is used for conversion of input into first element of
attribute ‘ outputs ‘. Second element for second element of attribute
‘ outputs ‘. E . g . eta = [0.3 , 0.4]

"""
o p t i m i z ation_options = {}

d e f __init__ ( self , ** kwargs ) :

s u p e r () . __init__ (** kwargs )

C.3 SOC constraints

"""
The n o n l i n e ar_contraints module contains the pyomo constraints wrapped in
functions . These functions are used by the ’ _assembler - methods
of the Op timizationModel () - class .

The module frequently uses the dictionaries I and O for the construction of
constraints . I and O contain all components ’ uids as dictionary keys and the
relevant input input / output uids as dictionary items .

* Illustrative Example *:

Consider the following example of a chp - powerplant modeled with 4 entities

(3 busses , 1 component ) and their unique ids being stored in a list
called ‘ uids ‘:

>>> uids = [ ’ bus_el ’, ’ bus_th ’, ’ bus_coal ’,’ pp_coal ’]

>>> I = { ’ pp_coal ’: ’ bus_coal ’}
>>> O = { ’ pp_coal ’: [ ’ bus_el ’, ’ bus_th ’]}
>>> print ( I [ ’ pp_coal ’])
bus_coal

Sander van Nielen ( s . s . vannielen@student . tudelft . nl )

"""

i m p o r t inspect
i m p o r t logging
i m p o r t pyomo . environ as po
from pandas i m p o r t Series as pdSeries
from . i m p o r t pyomo_fastbuild as pofast

d e f add_ sofc_relation ( model , block ) :

""" Adds constraints for the relation between input and outputs of a
SOFC component .
Added by Sander van Nielen
"""
i f not block . objs o r block . objs i s None :
r a i s e ValueError ( " No objects defined . Please specify objects for \
which the constraints should be build " )

# Create fuel utilisation variable with bounds

block . fuel_util = po . Var ( block . uids , model . timesteps ,
within = po . NonNegativeReals , bounds =(0.6 ,0.95) )
fuel_util = block . fuel_util
add_out = g e t a t t r ( block , " add_out " , { obj . uid :0 f o r obj i n block . objs })

# Relation of outputs to input .

d e f power_out_rule ( block , e , t ) :
OCV =1.0691993282
alfa =0.0526705851

152
 R =0.45; Ri ={}
I = model . w [ model . I [ e ][0] , e , t ] * fuel_util [e , t ] /1.1534914168
f o r obj i n block . objs :
in_max = obj . in_max [0] + add_out [ obj . uid ]
Ri [ obj . uid ] = 1.1534914168* I * ( OCV - alfa *0.95) /(0.95*2* in_max )
out = ( OCV - alfa * fuel_util [e , t ] - Ri [ e ]) * I *0.96*0.9
r e t u r n ( model . w [e , model . O [ e ][0] , t ] == out )
d e f hyd rogen_out_rule ( block , e , t ) :
I = model . w [ model . I [ e ][0] , e , t ] * fuel_util [e , t ] /1.1534914168
out = I *1.199821776 * (1/ fuel_util [e , t ] - 1)
r e t u r n ( model . w [e , model . O [ e ][1] , t ] == out )
d e f heat_out_rule ( block , e , t ) :
# Energy balance : in - out = heat
heat = ( model . w [ model . I [ e ][0] , e , t ] -
model . w [e , model . O [ e ][0] , t ] - model . w [e , model . O [ e ][1] , t ])
r e t u r n ( model . w [e , model . O [ e ][2] , t ] == 0.95 * heat )

block . power_out = po . Constraint ( block . indexset , rule = power_out_rule ,

doc = " Ele_out = i *(0.119 + (1.14 -0.143) * U_f -0.2* i ) " )
block . hydrogen_out = po . Constraint ( block . indexset , rule = hydrogen_out_rule ,
doc = " H_2 , out = i *1.199*(1/ U_f -1) " )
block . heat_out = po . Constraint ( block . indexset , rule = heat_out_rule ,
doc = " Heat_out = 0.95*( CH4_in - H2_out - Ele_out ) " )

d e f a dd_ so fec_relation ( model , block ) :

""" Adds constraints for the relation between input and outputs of a
Solid Oxide Fuel - asissted Electrolysis Cell .
Added by Sander van Nielen
"""
i f not block . objs o r block . objs i s None :
r a i s e ValueError ( " No objects defined . Please specify objects for \
which the constraints should be build " )

# Assumed constants :
a = 0.08725 # alpha
R = 0.45 # ASR
Uf = 0.8 # Fuel utilisation
OCV = 0.11564 # Open cell voltage
add_out = g e t a t t r ( block , " add_out " , { obj . uid :0 f o r obj i n block . objs })

# Relation of outputs to input .

d e f power_in_rule ( block , e , t ) :
Ri = {}
I = model . w [ model . I [ e ][0] , e , t ] * Uf /1.1534914168
f o r obj i n block . objs :
Imax = ( obj . in_max [0] + add_out [ obj . uid ]) * Uf /1.1534914168
Ri [ obj . uid ] = R * I * 1.5 / Imax
ele_in = -( OCV -a * Uf - Ri [ e ]) * I /(0.9*0.96)
r e t u r n ( model . w [ model . I [ e ][1] , e , t ] == ele_in )
d e f hyd rogen_out_rule ( block , e , t ) :
Uf = 0.8
I = model . w [ model . I [ e ][0] , e , t ] * Uf /1.1534914168
r e t u r n ( model . w [e , model . O [ e ][0] , t ] == I *1.4812614519)
d e f heat_out_rule ( block , e , t ) :
heat = model . w [ model . I [ e ][0] , e , t ] + model . w [ model . I [ e ][1] , e , t ]
heat -= model . w [e , model . O [ e ][0] , t ]
model . w [e , model . O [ e ][1] , t ]. setlb (0)
r e t u r n ( model . w [e , model . O [ e ][1] , t ] == 0.95 * heat )

block . power_in = po . Constraint ( block . indexset , rule = power_in_rule ,

doc = " Ele_in = i *(0.119 - 0.143* U_f -0.45* i ) " )
block . hydrogen_out = po . Constraint ( block . indexset , rule = hydrogen_out_rule ,
doc = " H_2 , out = i *1.199*(1/ U_f -1) " )
block . heat_out = po . Constraint ( block . indexset , rule = heat_out_rule ,

153
 doc = " Heat_out = 0.95*( CH4_in - H2_out -
Ele_out ) " )

d e f a d d_ h y brid_relation ( model , block ) :

""" Adds constraints for the relation between input and outputs of a
Solid Oxide Cell with a SOFC and a SOFEC mode .
Added by Sander van Nielen
"""
i f not block . objs o r block . objs i s None :
r a i s e ValueError ( " No objects defined . Please specify objects for \
which the constraints should be build " )

# Create fuel utilisation variable with bounds

block . fuel_util = po . Var ( block . uids , model . timesteps ,
within = po . NonNegativeReals , bounds =(0.6 ,0.95) )
fuel_util = block . fuel_util
# Create variable for the fraction of cells operating in SOFC mode .
block . y = po . Var ( block . uids , model . timesteps ,
within = po . Binary ) # NonNegativeReals , bounds =(0 ,1) )

add_out = g e t a t t r ( block , " add_out " , { obj . uid :0 f o r obj i n block . objs })

# Relation of outputs to input as a linear combination of SOFC and SOFEC .

d e f power_out_rule ( block , e , t ) :
OCV =1.0691993282; alfa =0.0526705851; R =0.45; Ri ={}
I = model . w [ model . I [ e ][0] , e , t ] * fuel_util [e , t ] /1.1534914168
f o r obj i n block . objs :
in_max = obj . in_max [0] + add_out [ obj . uid ]
Ri [ obj . uid ] = 1.1534914168* I * ( OCV - alfa *0.95) /(0.95*2* in_max )
out = ( OCV - alfa * fuel_util [e , t ] - Ri [ e ]) * I *0.96*0.9 * block . y [e , t ]
r e t u r n ( model . w [e , model . O [ e ][0] , t ] == out )
d e f power_in_rule ( block , e , t ) :
alfa = 0.08725; OCV = 0.11564; R = 0.45; Uf = 0.8; Ri = {}
I = model . w [ model . I [ e ][0] , e , t ] * fuel_util [e , t ] /1.1534914168
f o r obj i n block . objs :
Imax = ( obj . in_max [0] + add_out [ obj . uid ]) * fuel_util [e , t ]
/1.1534914168
Ri [ obj . uid ] = R * I * 1.5 / Imax
ele_in = -( OCV - alfa * fuel_util [e , t ] - Ri [ e ]) * I /(0.9*0.96) * (1 - block . y [e , t
])
r e t u r n ( model . w [ model . I [ e ][1] , e , t ] == ele_in )
d e f hyd rogen_out_rule ( block , e , t ) :
I = model . w [ model . I [ e ][0] , e , t ] * fuel_util [e , t ] /1.1534914168
out = I * block . y [e , t ]*1.199821776 * (1/ fuel_util [e , t ] - 1)
out += I *(1 - block . y [e , t ]) *1.4812614519
r e t u r n ( model . w [e , model . O [ e ][1] , t ] == out )
d e f heat_out_rule ( block , e , t ) :
# Energy balance : in - out = heat
heat = model . w [ model . I [ e ][0] , e , t ] + model . w [ model . I [ e ][1] , e , t ]
heat -= model . w [e , model . O [ e ][0] , t ] + model . w [e , model . O [ e ][1] , t ]
model . w [e , model . O [ e ][2] , t ]. setlb (0)
r e t u r n ( model . w [e , model . O [ e ][2] , t ] == 0.95 * heat )

block . power_out = po . Constraint ( block . indexset , rule = power_out_rule ,

doc = " Ele_out = i *(0.119 + (1.14 -0.143) * U_f
-0.2* i ) " )
block . power_in = po . Constraint ( block . indexset , rule = power_in_rule ,
doc = " Ele_in = i *(0.03922 -0.143*0.8 -0.45* i ) "
)
block . hydrogen_out = po . Constraint ( block . indexset , rule = hydrogen_out_rule ,
doc = " H_2 , out = i *1.199*(1/ U_f -1) " )
block . heat_out = po . Constraint ( block . indexset , rule = heat_out_rule ,
doc = " Heat_out = 0.95*( CH4_in - H2_out -
Ele_out ) " )

154