The Interaction of a High RE With the Gas System Through PtG

KULeuven Energy Institute

TME Branch

WP EN2014-17

The Interaction Of A High Renewable Energy/Low Carbon Power System With The Gas System Through Power To Gas

J. Vandewalle, K. Bruninx, W. D'haeseleer

TME WORKING PAPER - Energy and Environment Last update: September 14

An electronic version of the paper may be downloaded from the TME website:

http://www.mech.kuleuven.be/tme/research/

1

THE INTERACTION OF A HIGH RENEWABLE ENERGY/LOW CARBON

POWER SYSTEM WITH THE GAS SYSTEM THROUGH POWER TO GAS

Jeroen Vandewalle1,2,3

, [email protected]

Kenneth Bruninx1,2,3

, [email protected] 1 University of Leuven (KU Leuven) Energy Institute – branch Applied Mechanics and Energy Conversion (TME),

Celestijnenlaan 300A P.O Box 2421, B-3001 Heverlee, Belgium, Phone: +32 16 32 25 11 2 EnergyVille (joint venture of VITO NV and KU Leuven), Dennenstraat 7, B-3600 Genk, Belgium

3 Supervisor: William D’haeseleer

1,2, [email protected]

Abstract – The increasing share of renewable electricity production leads to operational challenges in the

power sector. Storage will be needed, amongst other options, to ensure a safe and reliable operation of

the power system. The power to gas concept is interesting to store excess renewable power that

otherwise would be curtailed. The renewable methane can easily be stored in the gas network and used

later on. The gas network has a much larger storage capacity compared to current electricity storage

technologies. However, power to gas introduces extra couplings between the gas, power and carbon

system and it is not known what the effect of these new interactions could be. Therefore, we develop an

operational model that includes the gas, power and carbon (CO2) sector to analyse the effects of power to

gas on these sectors and the interactions between them. A case study with a high share of renewable

energy has been analysed to describe the qualitative effects. The results indicate that power to gas

partially transfers the downward pressure on energy prices and capacity and flexibility problems

introduced by renewable energy in the electricity sector to the gas sector. However, the effects of power

to gas are generally smaller than those of renewable energy production. If power to gas is to be deployed

at large scale, these effects should be considered.

Keywords: power to gas, system interactions, CO2 emissions, gas network

1. Introduction

The energy landscape has changed considerably and it is expected that it will continue doing so. The share

of renewable energy sources (RES), especially in the electric power sector, has increased significantly and

the current energy pathways indicate a further increase [1]. However, the variability and limited

predictability of RES result in new operational challenges for power system operators in maintaining the

system balance. Advanced operational techniques will be needed, amongst other options, such as

storage, to ensure a safe and reliable operation of the power system [2].

Several electricity storage options exist, e.g., compressed air energy storage (CAES), pumped hydro

storage (PHS), supercapacitors and batteries. However, these technologies have a limited storage capacity

[3]. An interesting option is the power to gas (PtG) technology [4] that converts excess renewable

electricity into renewable gas which can be stored easily in the gas network. Two types of renewable

gases can be produced with the PtG process. Firstly, electric power can be converted to hydrogen with an

efficiency of 54 – 77%. The injection of hydrogen into the natural gas network is limited, though, to

approximately 10% [5]. The hydrogen can be further processed to methane, lowering the total conversion

efficiency to 49 – 65%. The main advantage of producing renewable methane is that there are no

restrictions regarding the renewable methane content in the gas network. Furthermore, the potential to

store renewable methane in the national natural gas networks is large enough to cover several days of gas

2

demand in e.g. Germany and France [3,4,6]. The focus in this work is on the conversion to renewable

methane.

The PtG process requires a CO2 source for the methanation step, which can be captured e.g. from fossil

fuelled power plants or from the atmosphere. The latter option is less efficient and more costly, however.

The first option is especially interesting in a power system where the CO2 emissions are to be mitigated,

e.g. through carbon capturing (CC). However, CC is currently not a mature technology [7]. Operational

aspects have been studied with simulations but practical experiences are limited to pilot plants [8].

Therefore, in this research, an energy system will be assumed with mature CC technology. Furthermore, a

developed carbon system requires a CO2 transportation network, CO2 storage and a CO2 market. PtG can

recycle carbon of fossil origin to produce renewable methane, thereby lowering the need for natural gas

and consequently defossilising the gas system.

PtG is also still in the research phase. The currently available technology for the PtG process needs further

optimisation to cope with the fast fluctuating renewable power generation and to reduce the investment

costs. Moreover, the development of large-scale units will be required for PtG [3]. Thorough technology

reviews of PtG technologies can be found in [3] and [7]. For a review of PtG pilot plants, see [10]. For an

analysis of implementation strategies in the energy system, see [4], where different options are

investigated depending on the share of RES, the natural gas import level and the availability of biomass

for digestion or gasification. Furthermore, an analysis of the spatial distribution of PtG on the electric

power network can be found in [11]. Also, a method for the optimal dimensioning of power to gas is

demonstrated in [11].

Natural gas is considered to be an important fuel in the future energy system, especially in systems with a

high share of volatile RES generation [12–15]. Furthermore, it is generally assumed that the natural gas

system is robust enough to cope with the injection of renewable power methane. However, this has not

yet been studied in detail. Therefore, the question arises what the impact is of power to gas on the

natural gas sector, and more generally what the impact is on the energy system.

The aim of this work is to study the effects of power to gas on the energy system, comprised of the

natural gas, electricity and carbon sectors, and on the interactions between these sectors. The focus in

this work is on an electric power system with only wind and solar generation, supplemented with gas-

fired power plants, which are also the CO2 source for the power to gas process. As a high share of RES and

mature PtG and carbon capture technology are assumed, the case should be interpreted as a possible

future scenario around 2040 – 2050. An operational model will be used in order to accurately represent

the effects and interactions. However, as there are many uncertainties regarding such case with PtG and

CC in it, this study is limited to one specific case which is analysed qualitatively rather than quantitatively.

The paper is further organised as follows. Firstly, the simulation model, the used data and the PtG

dimensioning method are explained. Secondly, the results are discussed, indicating the effects of PtG on

the different sectors and on the interactions between the different energy sectors. The main finding is

that the issues caused by high shares of renewable power generation in the electricity sector are partially

translated to the gas sector through PtG. Finally, the conclusions and ideas for further work are

formulated.

3

2. Method

Firstly, the model is described generally. To maintain the overview, the model details and the

mathematical formulation are left out of this paper and can be found in [16]. Secondly, the case study is

introduced with the used data and parameters. At last, the determination of the capacity of power to gas

plants in the system is explained.

2.1 Model description

An operational model for is available using mixed integer linear programming (MILP), as given in [16]. The

objective function in the model is to minimise the total operational costs of providing all electricity and

gas. The energy system is approached with an optimisation of the operational cost from the whole system

perspective. Hence, no distinction is made between e.g. the owners of the wind turbines, power plants

and the PtG plants.

All gas and electricity domestic demands and renewable energy production profiles are assumed to be

known a-priori and are exogenous to the model. With domestic demand, we refer to the industrial,

commercial and residential demand. The model is deterministic. Analysing the impact of unpredictabilities

could further strengthen this research, but would lead too far for this work.

The illustration in Fig. 1 shows the three coupled subsystems in the model: the electricity, the gas and the

carbon (-dioxide) market. Each of them is coupled through price and demand signals to one or more other

subsystems.

Fig. 1. The electricity, gas and CO2 markets are coupled directly. In addition to the traditional coupling between the three

markets through the respective price and demand signals, power to gas introduces alternative couplings.

In the gas system, the demand for the domestic and electricity generation sectors has to be met at all

times by importing gas and by producing renewable methane (RM) with PtG. It is assumed that the gas

market is one single spot market with a known and fixed, linearly increasing supply curve. There is also

seasonal gas storage and gas flexibility available, both with a cost associated to it, such that the gas

demand and import profiles can be different. The physical networks of gas and electricity are not

modelled as such, but are implicitly part of the model through the supply curve and demand constraints.

As to gas flexibility, a simplified approach is used in this work. Flexibility allows an imbalance between the

gas supply and demand for every time step. This imbalance is then accumulated over the time and forced

4

to be zero at the end of each day, based on current practices in countries like e.g. Belgium [17]. The cost

for providing the required flexibility is related to the swing of the accumulated imbalance. In physical gas

networks, balancing is needed to match supply and demand, due to prediction errors of the demand and

gas dynamics. This is, however, not as critical as in the electric network where the demand and supply

have to be balanced precisely at every moment, whereas the gas network has an inherent source of

flexibility, the line-pack, and it usually provides enough flexibility to cope with the daily imbalances.

Furthermore, fast-cycling storages could also be used to match supply and demand. The actual

dispatching of gas is an economic trade-off between using and paying for sources of flexibility ex-ante,

which could include flexibility on the import market, or paying the transmission system operator (TSO) ex-

post for the caused imbalances. A detailed discussion of this complex matter would lead to far here and

we refer to [17] for more information.

In our model, electricity generation is provided by PV solar installations, wind turbines and power plants.

The domestic electricity demand has to be covered at all times. The power plants are all gas-fired (GFPP)

and as such they provide a primary link between the gas and the electricity sector. In this work, they are

included with a unit commitment formulation in the model, subject to techno-economic operational

constraints. For computational reasons, the power plants have been aggregated in this work, which still

yields sufficiently detailed results for our purposes; see [16]. The excess renewable electricity generation

is curtailed or used for the PtG process. Short term electricity storage like pumped hydro could also be

part of the power sector, but it is not included in this work.

The power plants are equipped with carbon capture (CC) facilities1 which consume electricity from the

power plant, lowering the power plants’ net output efficiency. CC plants provide a primary coupling

between the electricity and CO2 sectors. When CO2 emissions are not captured, an emission price is due,

which is further referred to as emission cost for the electric power sector. The captured CO2 can be used

in the PtG process, possibly after short term buffering; else, it is disposed of.

Adding power to gas to the model introduces new indirect linkages between the already coupled markets;

see Fig. 1. PtG consumes renewable excess electricity and CO2, while producing renewable methane. The

methane is injected in the network where it mixes with natural gas. In this model, the use of renewable

methane is not limited to power generation only, but it can also be used for domestic demand or storage

in the seasonal gas storage facility. Also, the PtG process requires water in the electrolysis step and

produces oxygen in the methanation step. The oxygen is pure and can be marketed.

The power to gas plants are modelled similar to the electric power plants with part-load characteristics,

start-up costs and techno-economic constraints such as minimum up/down times and maximum ramping

constraints.

The simulation horizon is a full year, which allows us to see the effects during all seasons. To account for

the intermittent behaviour of renewable electricity production, a time step of fifteen minutes is used in

the model, which is the reference time step in practice of the electricity sector. This high accuracy results

in a model that is computationally hard to solve for long time periods. Therefore, a rolling horizon

1 The terminology carbon (-dioxide) capture and storage (CCS) is widely used in literature. The plant itself, however,

only captures CO2 in this model and does not store it. Also, it should also be noted that carbon storage here can have

two meanings: firstly, there is long term storage, or disposal, of CO2 e.g. in abandoned gas fields. Secondly, there is

short term buffering of the CO2 to be used in the power to gas plants.

5

approach is implemented in which subsequently limited time intervals are solved; for more information

see [16].

2.2 Data and case

The scale of the data in the model represents a region of small country with an electric demand of 80

TWh 2 and a domestic gas demand of 140 TWh (i.e. without gas demand for electricity generation); see

Appendix A for more details. For simplicity, the import and export of electricity and the export of gas are

not considered in this work. The case study regards a hypothetical system with 100% renewable energy3,

divided 50%-50% over wind and PV solar electricity generation. The operational costs for wind and solar

are assumed to be zero. Historic domestic electricity and gas demand data, as well as solar and wind

production data, are obtained from the Belgian TSOs for electricity ELIA4 and gas Fluxys5. However, it is

not our goal to formulate conclusions concerning Belgium, the data only serves as a realistic input to the

model.

The prices of natural gas on the market are assumed to be in the range of 60 – 76 €/MWh, related linearly

to the gas import level, ranging from 0 to 80 GW. Though this is a strongly simplified model for a gas

market, it serves well for the aimed analysis. This gas price range is taken considering a higher price in the

EU than the US [6] and taking the high price scenario in [1] in 2040. The maximum import level has been

chosen such that it is not constraining the gas supplies in this case. Further work could include analysing

the effects of limits on the maximum import level.

The cost for providing gas flexibility depends on the daily swing of the accumulated imbalance. We

assume a flexibility cost of 2.5 €/MWh of gas imbalance per day, based on information from Fluxys.

Seasonal gas storage is included with a storage capacity of 7 TWh, an injection capacity of 3.25 GW and a

withdrawal capacity of 6.25 GW. It is assumed that these maximum in- and output constraints are not

affected by the storage status. The gas storage cost is 5 €/MWh. These figures are based on the Belgian

system, as published by Fluxys.

All GFPPs are equipped with a carbon capture (CC) plant. The operation of a CC plant depends mainly on

the input gas price and the emission costs. A basic model is used which allows part-load operation from

40 to 100%6. There are no dynamic effects of a CC plant taken into account such as restrictions on the

maximum ramping up/down rates. It is assumed that the maximum capture rate is 90% of the produced

emissions and that 10% of the generated electricity output is required to capture this amount.

Furthermore, the CC plants consumption scales linearly with the actual capture rate. The disposal of CO2 is

considered as one single entity with an unlimited storage capacity. In the application of the model for this

work, as a simplification, no distinction is made between long term storage—e.g. in salt caverns or

abandoned gas fields—and short term buffering for power to gas. No storage or transport costs for CO2

are accounted for in the operational model. The CO2 emission cost is proposed to be fixed at 100 €/tonCO2

in the studied case. The effect of the level of the emission cost will be discussed in the results section.

2 Different types of energy or power can be distinguished: electric, thermal or primary (natural gas), and these can

be indicated by subscripts e, th and pr, respectively. However, this will only be done in this text when it is not clear

from the context which type it concerns. 3 Percentage on a yearly energy basis, thus without matching demand and production in time (e.g., in TWh/year).

4 ELIA, see http://www.elia.be/en

5 Fluxys, see http://www.fluxys.com/?sc_lang=en

6 Minimum operating point stated in internal communication with power plant operator Electrabel.

6

The electric power and PtG plants are aggregated in this work. The most important parameters are the

conversion efficiencies, being 56% for the power plants (gas → electricity) and 65% for the PtG plants

(electricity → gas), based on the higher heating value of natural gas and renewable methane. For more

information about the power and PtG plant models and their operational parameters, see Appendix B and

[16].

2.3 Determination of the installed power to gas capacity

It is out of scope to provide an accurate figure for the optimal capacity of PtG in the system due to the

numerous uncertainties related to investment costs and conversion efficiencies of the plants, and future

energy and CO2 prices, which are affecting the competitiveness of PtG.

In order to determine a reasonable capacity of PtG plants in the system, the following, simplified,

dimensioning approach is used, based on the model input data for loads and generation profiles

described above. Firstly, the production cost of renewable methane is estimated, and subsequently, the

required number of operating hours is estimated to produce renewable methane that is competitive with

natural gas in the regular market.

The production cost is expressed as a function of the running hours of the plant, accounting for the

assumed investment and O&M costs (see Table 1) and the running costs of the plant (water, oxygen, CO2

and electricity). No planning and construction costs are accounted for. In this dimensioning analysis, CO2 is

assumed to be available at 10 €/tonCO2, which includes transportation from the power plants to the PtG

plant and local buffering. Pure oxygen is available as a by-product of the methanation process and can be

sold. We use 70 €/tonO2 for the dimensioning, based on [4]. The water needed to produce renewable

methane is approximately 0.150 m³/MWh and the cost of water is assumed at 0.7 €/m³.

Table 1. Assumed investment and O&M costs of an electrolyser and a methaniser

Electrolyser Methaniser

Investment costs €/MWinput 750 000 50 000

O&M costs % of inv. cost 4 10

Depreciation period Year 20 20

Discount rate % 7 7

The production cost of renewable methane is shown in Fig. 2. As a form of sensitivity analysis, the

production costs are shown for three different values of the input electricity cost (indicated by the three

different markers), and for two values of oxygen (dotted and solid lines). Firstly, it can be seen that the

number of operating hours will be determining for the competitiveness of renewable methane with

natural gas. The grey zone indicates the current reference prices of gas in Europe (25 – 40 €/MWh, [6])

and illustrates the high number of operating hours that would be needed to make competitive renewable

methane. Furthermore, and as expected, it can be seen that the price of electricity has a very high impact

on the production costs. In the further work, we will assume electricity is free because it is excess

renewable energy. Currently, zero or even negative prices exist. It is unclear how this will evolve with the

growing share of renewables. Whether or not this is economically viable from the perspective of the

owners of the RES installations is a different discussion that will not be dealt with here. In this work, we

7

assume a minimum gas price on the market of 60 €/MWh; see point a in Fig. 2, which makes the

minimum required number of operating hours 1600 h (point b).

Fig. 2. Production costs of the renewable methane are highly

dependent on the number of operating hours. Furthermore,

the value of oxygen (line style) and the electricity price

(markers) have an impact. For a certain reference price on the

market (a), and certain running costs, the required number of

operating hours (b) can be found in order to make competitive

renewable methane.

Fig. 3. Dimensioning of the power to gas plant. The minimal

required number of operating hours (NOH) is found from the

economic analysis of the renewable methane production cost

(Fig. 2). With the residual load duration profile in this figure,

the corresponding capacity of power to gas plants can be

found, which is approx. 7 GWe in this case.

Secondly, the maximum capacity of PtG plants that can be installed with profitability can be found on the

load duration curve7 of the residual load (electric demand minus electric renewable generation), see Fig.

3. This is indicated by the intersection (c) of the minimal required operating hours (NOH) and the residual

load line. The PtG input capacity in this case would be approximately 7 GWe.

Note that with the current gas prices (grey area in Fig. 2) a very high duration of the excess renewable

electricity generation would be required to produce competitive renewable methane, even with the most

optimistic values for running and investment costs. However, as the shares of RES are currently far below

the situation depicted in Fig. 3 (100% RES on energy basis), there seems to be no profitable business case

for power to gas in current energy systems.

3. Results

Firstly, the effects of PtG are discussed per individual sector. This facilitates the understanding of the

effects of PtG on the interactions between the sectors, discussed in the last part of the results. The effects

per sector are discussed on two different time scales. The long term results indicate the general

conclusions and the short term results illustrate the effects. The short term effects are all shown for one

specific day with high wind and high solar production. The characteristics of this day will be shown on the

figure of the electric dispatch further on (Fig. 5).

The main findings are that PtG leads to a certain extent to a shift from the electricity sector to the gas

sector of the problems related to high shares of renewables. In particular, the results indicate that the

7 In a load duration curve, the data is put in descending order. All time information is lost, however.

0 1000 2000 3000 4000 50000

20

40

60

80

100

120

140

Number of operating hours (h)

Ma

kin

g p

rice

of

ga

s (

EU

R/M

Wth

)

0 EUR/MWh

e

25 EUR/MWhe

50 EUR/MWhe

70 EUR/toxygen

10 EUR/toxygen

Current Ref.

0 1000 2000 3000 4000 5000 6000 7000 8000 9000−40

−30

−20

−10

0

10

20

Duration (hours)

Ele

ctr

ic p

ow

er

(GW

)

Load

Residual load

Power to gas

a

b

c

Cap. PtG

NOH

8

gas import costs are lowered while the costs for capacity and flexibility increase. However, the effects of

PtG are secondary compared to the impact of renewable electricity generation.

3.1 Impact on the electric power sector

Firstly, the long term effects indicate the general conclusions and trends, which are subsequently

illustrated by the short term effects.

3.1.1 Long term effects on the electric power sector

The main effects of PtG on the power sector are analysed over a whole year, see the load duration curves

in Fig. 4. Load duration curves of the domestic electric demand, gas-fired power generation, RES

generation, electricity consumption for PtG and curtailment are shown8. The areas circumscribed by the

load duration curves then represent annual energy demand or production. Though the aim is a qualitative

analysis, some figures are given below to facilitate the interpretation of the case.

Fig. 4. Load duration representation of the electric energy production and consumption over a whole year with power to gas

in the system. Even though the high RES share, still 30% of the demand needs to be covered by gas-fired power plants. Even

with power to gas, the curtailed energy is still high and the curtailment peak is extreme.

The annual electricity demand is 77.8 TWh of which 53.7 TWh is directly provided by RES. An excess RES

production of 16.7 TWh is used in PtG plants and 13.4 TWh is curtailed. The residual load, covered by gas-

fired power plants is 26.8 TWh of which 2.68 TWh has been used to generate electricity for the carbon

capture plants.

Though RES accounts for 100% of the electric demand on an energy basis, the production and demand are

not balanced, resulting in a still relatively high residual load coverage (31%) from GFPPs. Of the 30.1 TWh

8 Normally, in a load duration curve, the data is represented in a decreasing order with increasing duration.

However, the load duration curves for PtG and curtailment should be interpreted in the opposite direction of

duration.

0 1000 2000 3000 4000 5000 6000 7000 8000 9000−40

−30

−20

−10

0

10

20

Duration (hours)

Load &

Pro

duction (

GW

)

Renewable production

Residual load

Power to gas

Curtailment

Gas-fired generation

Power to gas

Curt.

RES

Electric load curve

9

of excess renewable energy, about 55% can be used in PtG. This amount is limited because of the limited

capacity of PtG. Adding more PtG plants would not have been beneficial in this case from the viewpoint of

the PtG plant owner because the number of operating hours would be too low to recover the investment

costs. While PtG leads to less curtailment, there is still 45% of excess renewable energy that needs to be

curtailed, with a peak of approx. 30 GW.

Note that after the conversion of excess renewable electricity with PtG to methane (efficiency 65%), and

re-conversion of that methane to electric power (efficiency 50%), 5.43 TWh of the residual load could be

covered indirectly by renewable electricity9. This would increase the share of renewable electric energy

from 69% to 76%.

Regarding the power plants, the high amount of RES reduces the number of operating hours drastically

while still a high installed capacity is required. In fact, enough capacity should be available in case there is

no wind and no sun. It is stated that this situation could lead to problems regarding having sufficient

generation capacity when the electricity market is not functioning properly, especially with flawed

regulation rules [18], which is often referred to as the missing money problem. In fact, this has to do with

the non-proper representation of scarcity in the electricity price.

Regarding electric power generation itself, power to gas has no impact on the load duration curve of the

power plant electricity generation in this case. Without PtG, the only difference in Fig. 4 would be that the

PtG area would be curtailment. However, power to gas could possibly have an impact on the individual

dispatch of power plants in systems with a more complex generation mix and other price and cost

scenarios. Also, the curtailment of RES could be slightly different when power plants with other dynamic

constraints are in the system. However, the peak power generation would stay the same. This has not

been further analysed in this work.

3.1.2 Short term effects on the electric power sector

The electricity dispatch on the specific day is shown in Fig. 5. During the night, wind power alone is not

sufficient to cover the entire demand and the residual load is covered by gas-fired power generation.

During the day, part of the excess electric power is converted to renewable methane with the power to

gas process, the rest is curtailed. Though the figure suggests curtailment of solar power, this is only a

representation. The actual curtailment could be a mix of wind and solar, depending on the curtailment

costs. In this model, the curtailment costs for both wind and solar were assumed to be zero. Also note the

highly variable demand for residual load coverage by the electric power plants. GFPPs are believed to

have an important role in such situations because of their abilities of flexible generation [15].

Furthermore, the cost to generate an extra unit of electricity has been studied. The marginal electricity

cost (MEC) of the specific day is shown in Fig. 6 where three cases are compared: (i) no renewable

production, (ii) renewable electricity generation without PtG and (iii) renewables with PtG. Whenever

there is electricity generation by GFPPs, the MEC is set by the gas cost corrected by the power plants’

efficiency. As stated before, the gas costs increase with the gas demand. As the gas demand is highest

without RES, because of the gas demand for power generation, the MEC will also be highest; see the

upper curve in Fig. 6. With RES production, the MEC is lower because less gas is used which leads to a

9 However, the renewable methane mixes with natural gas and can also be used to cover domestic gas demand.

10

lower gas demand and consequently a lower marginal gas cost (MGC). The MGC is studied in detail in

Section 3.2.2.

Fig. 5. Electricity dispatch on a day with high renewable

production. During the night, wind energy alone is not

sufficient to cover the demand and the residual load is

covered by gas-fired power plants. During the day, part of

the excess power is converted in power to gas plants, the

rest is curtailed.

Fig. 6. Marginal electricity cost (MEC). Renewable energy

lowers the MEC. During the day, solar production is so high

that curtailment is needed. Consequently, the MEC are zero

then. With power to gas, a small side effect occurs, increasing

the MEC because the cost of an extra unit of electricity

corresponds to less renewable methane production.

During the day, the MEC goes to zero when there is RES curtailment, because curtailment is free of charge

in this model. This situation is marked with b in Fig. 6. When PtG is part of the system, another effect can

be observed at the beginning (around 6:45 – 8:00) and the end (18:00 – 19:15) of the solar production

period: the MEC is then not zero, marked with a. This is because renewable methane is produced. Hence,

a unit of extra electricity would result in less production of renewable methane, which has a certain

market value. When the excess renewable power production exceeds the PtG capacity (8:15 – 17:45),

however, the MEC falls to zero again, as there is curtailment. Other effects, marked with c are related to

operational constraints and are not relevant in this analysis.

3.1.3 Conclusions electric power sector

The main effect of PtG on the electric power sector is that the curtailment is lowered. However, still a

large portion of renewable excess energy has to be curtailed, with a high peak power. Furthermore, when

there is PtG operation and no curtailment, the marginal electricity costs are not related to RES but to

natural gas through the value of renewable methane on the gas market. This is in fact an interaction

effect and will be touched upon again in Section 3.4.

3.2 Impact on the gas sector

The effects are analysed regarding the gas imports and gas flexibility for both the long and the short term

below. No significant effects were observed regarding the seasonal gas storage dispatch. Hence, this is

not included in this analysis.

3.2.1 Long term effects on the gas sector

0 6 12 18 240

5

10

15

20

25

30

35

40

Time (hours)

Ele

ctr

icity d

ispatc

h (

GW

)

Gas−fired gen.

Wind

Solar

Curtailment

Demand

for Power to gas

0 6 12 18 240

50

100

150

200

250

Time (hours)M

arg

inal ele

ctr

icity c

osts

(E

UR

/MW

h)

No RES

RES, no PtG

RES + PtG

c

aa

b

11

The first long term effect on the gas sector is the gas import level throughout the year. This is analysed by

the load duration curve of the gas import in Fig. 7. The import level is generally lowered considerably by

RES, and to a smaller extent by PtG. Note that the load duration diagram loses all time information,

hence, the actual difference with and without RES, and with and without PtG, depends on the actual day.

(a) Whole year

(b) Detail of peak demand

Fig. 7. Load duration profile of the gas import. The gas import level is generally lowered substantially by RES and even

more by power to gas. Still a very high maximum import capacity is needed to cover the demand, especially when taking

into account the risk of no RES and no PtG production at cold, dark and windless days.

An important effect is that still a high import capacity of the network is required, even though the mean

gas demand is generally much lower than without RES. Especially, when taking the cold spell10 into

account, the import capacity with RES (and PtG) should be at least the same as without RES. The reduced

imports with RES put more pressure on the investment of the gas infrastructure, and PtG further

aggravates this situation. This could be seen as a partial transfer of the capacity problem in the power

sector where RES lower the number of electric power plant operating hours drastically such that it

becomes hard to maintain a profitable power generation. The lowered full load hours in the gas network

should lead to a higher capacity cost per unit of transported gas in order to recover the investment costs.

This is not necessary a problem if the capacity costs can be settled properly, but it could be an issue if e.g.

regulations are obstructing. It should be noted that capacity in the electric power sector refers to the

generation capacity while capacity in the gas sector refers to the import and transport capacity. Hence,

with a capacity problem, we refer to a problem that may complicate the recovery of investments, or

hamper the investment in new infrastructure to provide the required capacity.

Also, the demand for gas flexibility is affected by PtG. The daily demands for gas flexibility throughout the

year are put in a load duration diagram, see Fig. 8. The daily demand for gas flexibility is calculated by the

daily swing of the accumulated flexibility that is used. It can be seen that the demand for flexibility is

generally increased substantially in the presence of RES. With PtG in the system, the flexibility demand is

even higher. This is not necessarily a problem, as long as the safe operation limits of the network are not

exceeded. However, this could be a problem if not enough flexibility is available. Furthermore, sufficient

gas flexibility has to be available locally, as gas has a limited travelling speed. This increase of the demand

10

The term cold spell refers to two cold weeks without sun in the winter where all power has to be generated with

gas-fired power plants in our case.

0 1000 2000 3000 4000 5000 6000 7000 8000 90000

10

20

30

40

50

60

Duration (hours)

Gas Im

port

s (

GW

)

No RES

RES, no PtG

RES + PtG

100 200 300 400 500

35

40

45

50

55

60

Duration (hours)

Gas

Import

s (G

W)

No RES

RES, no PtG

RES + PtG

Detail

12

for flexibility has to be seen in parallel with the increased demand for flexible electric power generation in

a system with a high share of RES.

Fig. 8. Load duration of the swing of accumulated gas flexibility demand. RES increases the demand for flexibility, PtG

increases it further. This suggests that the gas system balancing will require a greater effort with power to gas.

3.2.2 Short term effects on the gas sector

Firstly, the effects on the gas import profile are analysed on the same specific day as Figures 5 and 6; see

Fig. 9. As explained for the long term effect on the gas sector, the impact of RES is larger than the impact

of PtG. Recall from the electricity generation dispatch (see Fig. 5) that there is almost no gas demand from

power plants during this specific day because of the high RES generation, except for a small part during

the night. Hence, most imported gas in Fig. 9 is related to the domestic gas demand. PtG is operating

during the daytime, which leads to a drop in the gas supplies.

An interesting effect can be noted in Fig. 9; there is an inversed peak of the import profile. Usually, in

current gas systems, the peak occurs during the day. However, because of RES, and even more because of

PtG, the import profile is higher during the night than during the day in this case study. This inversion

occurs during sunny days, such as the specific day shown here. During dark days, depending on the actual

wind generation profile, the peak occurs still mostly during the day. A similar situation has been observed

in the electricity transmission network since 2012, where the lowest demand occurs during the day in

summer while peak demand occurs during the day in winter [19].

Following, the marginal gas costs (MGC) are analysed in Fig. 10. Recall that the MGC is directly related to

the level of the gas import (see Fig. 9) because of the assumed linear supply curve. It results that RES has a

downward pressure on gas import prices, and PtG strengthens this effect. In the studied case, the MGC is

never below 60 EUR/MWh. However, more severe situations have been observed in a simulation without

a domestic gas demand. Such simulation has shown moments where the MGC is zero. These moments

occur during days without gas imports, which have a high RES generation and PtG production. The

marginal unit of gas is then related to PtG, which has zero operational costs in this model, instead of the

gas import market. Though it may not be realistic to assume zero operational costs for PtG, these results

show that situations can occur where PtG sets the gas price on the market.

0 50 100 150 200 250 300 350 4000

10

20

30

40

50

60

70

80

Duration (days)

Gas fle

xib

ility

dem

and (

GW

h)

No RES

RES, no PtG

RES + PtG

13

Finally, the gas flexibility usage is analysed during the specific day. The flexibility is used to allow a

difference between the import profile, as shown in Fig. 9, and the demand profile, as shown in Fig. 11. An

economic trade-off is made between flexibility costs and costs on the import market which are related to

the variability of the import profile 11 12. The resulting accumulated flexibility of the specific day is shown

in Fig. 12. Recall that flexibility costs are related to the swing of the accumulated flexibility. For this

particular day, RES has no major impact on the flexibility demand while PtG has a high impact. This is

related to the higher variability of the gas demand profile with PtG than without, as can be seen in Fig. 11.

Fig. 9. Gas import profile on a specific day with high

renewable (RES) production. The highest impact comes from

RES, as the demanded gas for power plants is reduced. With

power to gas, the gas demand during the day is further

lowered.

Fig. 10. Marginal gas costs are directly related to the level of

gas imports. RES has a downward pressure on the gas costs,

power to gas strengthens this effect.

Fig. 11. Gas demand profile on the specific day. The presence

of renewable energy and power to gas lead to a more

variable demand.

Fig. 12. Accumulated flexibility usage on the specific day. The

impact of RES is limited for this particular day, while the

impact of PtG is high. The costs are related to the daily swing.

11

Because there is a quadratic relation between the demanded volume on the import market and the cost for that

volume, it will always be cheaper to have an import profile that is smooth and spread out over the time [16]. 12

The use of flexibility is the reason why the import (Fig. 9) with and without PtG do not match exactly at night.

0 6 12 18 240

5

10

15

20

25

30

35

40

45

50

Time (hours)

Ga

s I

mp

ort

s (

GW

)

No RES

RES, no PtG

RES + PtG

0 6 12 18 2460

62

64

66

68

70

72

74

76

78

80

Time (hours)

Marg

inal gas c

osts

(E

UR

/MW

h)

No RES

RES, no PtG

RES + PtG

0 6 12 18 240

5

10

15

20

25

30

35

40

Time (hours)

Ga

si m

po

rt d

em

an

d w

ith

ou

t fle

xib

ility

(G

W)

No RES

RES, no PtG

RES + PtG

0 6 12 18 24−15

−10

−5

0

5

Time (hours)

Gas F

lexib

ility

Dem

and (

GW

h)

No RES

RES, no PtG

RES + PtG

swin

g

14

3.2.3 Conclusions gas sector

With PtG in the system, still a high gas import capacity is still required to ensure supplies at moments

without or with a low RES production while the annual volume of imported gas is lowered considerably.

This could increase the costs of gas related to the import and transport infrastructure. Furthermore, the

demand for flexibility increases, partially because of RES and partially because of PtG, which could

increase the flexibility-related costs of gas. On the other hand, RES and PtG pose a downward pressure on

the gas costs at the import market. Moreover, in certain situations, renewable methane can be put on the

gas market at marginal production costs which can be substantially lower than the normal gas supply

prices. Hence, the results suggest a partial transfer from the electric to the gas sector of the energy cost

and capacity and flexibility issues related to high shares of RES. The effects of RES on the gas system are

higher than the effects of PtG, though, except regarding the gas flexibility demand.

3.3 Impact on the CO2 sector

The CO2 emissions are not affected by PtG in this case study. This is because the demand for gas-fired

generation stays the same, with or without power to gas. The CO2 capture rate is also the same, whether

natural gas or renewable methane is used, as the carbon contents and thus the emission costs are the

same for both. Hence, the CO2 emissions are equal with or without PtG. These observations apply for this

particular system studied with only gas-fired power generation and may be different with a more complex

fuel mix and/or other emission costs.

Note that emission costs are indeed applicable to renewable methane as the CO2 is still of fossil origin.

Not assigning emission costs to electric power generation with renewable methane would lead to

emissions in the atmosphere of fossil CO2. These emissions would be delayed one step, though, as the CO2

was captured in the previous step from power generation with natural gas and released in the next step

when the renewable methane is combusted. Hence, PtG should be seen as a way of recycling carbon and

thereby mitigating the need for fossil fuels.

The CO2 storage requirements, on the other hand, are affected by power to gas because of the CO2

consumption in the Sabatier process. The effects of PtG are analysed below for the long and the short

term.

3.3.1 Long term effects on the CO2 sector

Looking at the yearly results of the long term CO2 storage requirements in Fig. 13, we see that there is

always an excess of CO2. In other words, long term storage is always needed in this case study. However,

due to PtG, less CO2 will have to be stored. This could lower the long term storage costs.

15

Fig. 13. Long term CO2 storage (disposal) requirements. RES

drastically lowers the need for long term storage. Power to

gas further lowers this need.

Fig. 14. CO2 storage requirements of the specific day. The

presence of RES drastically lowers the end of the day (=long

term) storage demand. Power to gas further lowers this

demand. Furthermore, a short term buffer will be needed

because CO2 capture at the power plants and CO2 demand of

the PtG plants are not matched in time.

3.3.2 Short term effects on the CO2 sector

The results of the specific day are depicted in Fig. 14. Without RES, a large amount of CO2 would need to

be stored because all electric power generation would be carbon based. With RES, CO2 is only produced

during the night, because, during the day time, wind and solar production are more than enough to

provide the demanded electricity. When PtG is present, the final disposal requirement is even zero for

this day. It has to be noted that a (short term) buffer will be necessary to cope with the unbalanced

capture and usage of CO2.

The marginal cost of making available one unit of CO2 (MCC) has also been analysed. It has been found

that it is always zero in this case. Because the emission cost is high enough, carbon capture is always

deployed. The captured amount of CO2 is higher than the demanded amount for PtG, hence, there is

always an excess and the cost for an additional unit of CO2 is thus zero. More generally, this MCC is a

lower limit for the price of CO2 that a PtG plant would pay and it is related to storage costs. This lower

limit can even be negative when long term storage costs are higher than short term buffering costs. E.g., if

the costs for the long term storage and corresponding transport are 10 EUR/tonCO2, and the costs for local

transport and short term buffering are 5 EUR/tonCO2, the avoided costs by using CO2 in the PtG process are

5 EUR/ton. Hence, the lower limit of the CO2 price would be −5 EUR/tonCO2. Hence, the MCC is

determined by the CO2 storage costs, but only when there is an excess of CO2. In case of scarcity, the MCC

is also determined by the costs for carbon capturing, which will be explained further in the next section.

There is also an upper limit for the CO2 price, determined by the market value of renewable methane.

3.3.3 Conclusions CO2 sector

PtG enables carbon recycling in the energy system because captured CO2, from gas-fired power

generation, is converted to renewable methane, thereby mitigating the need for fossil fuels. The need for

long term CO2 storage is lowered by RES and further by PtG. There is a need for short term buffering to

match the capture and consumption of CO2. The market value of CO2 is limited by the CO2 storage costs,

the value of renewable methane and in certain cases by carbon capturing costs.

0 50 100 150 200 250 300 350 4000

0.5

1

1.5

2

2.5

3

3.5x 10

7

Time (days)

Carb

on S

tora

ge (

ton)

No RES

RES, no PtG

RES + PtG

0 6 12 18 24−1

0

1

2

3

4

5

6

7

8

9x 10

4

Time (hours)

Carb

on S

tora

ge (

ton)

No RES

RES, no PtG

RES + PtG

16

3.4 Effects of power to gas on the interactions between the gas, electricity and carbon systems

The previous results have indicated several effects of PtG on different sectors. However, there are also

effects of PtG on the interactions between these sectors. They are described below.

Firstly, the interactions between the gas and the electricity system are affected by PtG. PtG enables an

important function in a high RES system, namely, storing excess renewable electricity as renewable

methane in the gas network. However, as such, the balancing issues from the electric power sector are

partially passed on to the gas sector. Furthermore, as indicated in 3.2, flexibility and capacity related costs

in the gas system may increase because of PtG. In turn, these costs could be passed on to the electricity

sector by the gas-fired power generation. There is also another effect that could make electricity more

expensive; when PtG is the marginal unit, the marginal electricity price is related to the value of

renewable methane on the gas market and not the generation costs of RES.

Secondly, PtG affects the interactions between the gas and the CO2 sectors. As pointed out in Section 3.3,

carbon capture is always deployed because the emission costs are high enough, hence, there is always an

excess of CO2. However, carbon-based generation and/or the emission costs could be lower. In particular,

there are emission costs which are low enough to not trigger carbon capture without PtG, but high

enough to trigger it when PtG is in the system. This is because CO2 is a commodity for PtG, enabling the

production of renewable methane which has a value on the gas market. Thus, carbon capture can be

deployed if the costs of capturing can be covered by producing renewable methane. The marginal CO2

costs (MCC) are then determined not only by the CO2 storage costs but also by the costs to compensate

for carbon capturing.

4. Conclusions

The power to gas concept is argued to be an interesting method for storing surplus electric renewable

energy in the form of renewable hydrogen or methane. The renewable methane can be stored in the gas

system, which has a large capacity compared to electrical storage. However, power to gas introduces new

couplings between the gas, the electricity and the CO2 systems which have not been analysed.

In our work, we have demonstrated that power to gas indeed affects the gas, the electric power and the

CO2 sector, and even the interactions between these sectors.

An operational model is used that shows the interactions between the gas, electricity and CO2 market

regarding the flows and prices. A case with high renewable energy shares and only gas-fired electric

power generation has been studied. Even though the results are limited to the single case analysed, some

general conclusions can be made regarding the impact of power to gas.

The main findings are:

- The known capacity and flexibility issues, and downward pressure on the energy prices in the

electricity sector due to a high share of renewable electricity production are partially passed on to

the gas sector. Renewable electric energy reduces the average demand for gas in power

generation, while still a high transport and import capacity of the gas network is needed. This may

put more pressure on the gas infrastructure investments. This effect is further aggravated in the

presence of power to gas. Furthermore, the results indicate that more flexibility will be needed in

the gas network with PtG. This could increase capacity and flexibility related costs in the final gas

17

price. On the other hand, gas import prices could be lowered by power to gas. An extreme case

even shows renewable methane injection at marginal production costs, which could be well

below the market price of natural gas.

- Another finding is that power to gas can increase the marginal electricity costs. This happens

when there is power to gas operation but no curtailment. Marginal electricity costs are then

indirectly coupled to the gas market through the market value of renewable methane.

- There are certain situations where power to gas can put renewable methane on the gas market at

marginal production costs, which can be substantially below the normal market prices of natural

gas.

- The presence of power to gas could enable more carbon capture as it lowers the critical value of

CO2 emission costs that trigger carbon capture, coupled again through the market value of

renewable methane.

- Power to gas lowers the need for the disposal of carbon dioxide in underground disposal sites.

Therefore, it can be concluded that power to gas would lower the associated storage costs. On

the other hand, short term buffering is needed to match carbon capture and usage in the power

to gas process.

These effects should be kept in mind when designing a system with a high share of renewables and power

to gas. The impact of power to gas is mostly lower than the impact of renewable generation, though,

except for the case of flexibility.

Further work

The model is implemented and operationally optimised as one whole energy system. It would be

interesting, however, to see how different actors in the system react to power to gas.

Further analysis is needed regarding the sensitivity to the assumed parameters such as the amount of

wind and solar in the system and the operational parameters. Also, the effect of the dimensioning of the

power to gas plants need to be studied further.

The impact on physical gas and electricity networks could be analysed by using the simulation data as

input to the network models, or the networks could be integrated in the model itself.

It would also be interesting to study the impact of the unpredictabilities with a stochastic model. PtG

could provide short term flexibility in the electric power sector.

Acknowledgements

The research of this work has been funded by the research project ‘local intelligent networks and energy

active regions’ (LINEAR) supported by the Flemish agency for innovation through science and technology

(IWT).

18

References

[1] U.S. Energy Information Administration (EIA), Annual Energy Outlook 2014, 2014.

[2] K. Bruninx, K. Van Den Bergh, E. Delarue, W. D’haseleer, Optimization and allocation of spinning

reserves in a low carbon framework Optimization and allocation of spinning reserves in a low

carbon framework, Submitted to IEEE Power Systems. (2014).

https://www.mech.kuleuven.be/en/tme/research/energy_environment/Pdf/wpen2014-08.pdf.

[3] DNV KEMA, Final Report Systems Analyses Power to Gas, Technology Review, Tech. Report. (2013).

[4] M. Sterner, Bioenergy and renewable power methane in integrated 100% renewable energy

systems, PhD Thesis, University of Kassel. (2009).

[5] D. Haeseldonckx, Concrete transition issues towards a fully-fledged use of hydrogen as an energy

carrier, PhD Thesis, KU Leuven. (2009).

[6] DG Energy, Quarterly report on European Gas Markets, second quarter 2013, Tech. Report. (2013).

[7] E. Blomen, C. Hendriks, F. Neele, Capture technologies: Improvements and promising

developments, Energy Procedia. 1 (2009) 1505–1512. doi:10.1016/j.egypro.2009.01.197.

[8] D.Y.C. Leung, G. Caramanna, M.M. Maroto-Valer, An overview of current status of carbon dioxide

capture and storage technologies, Renewable and Sustainable Energy Reviews. 39 (2014) 426–443.

doi:10.1016/j.rser.2014.07.093.

[9] G. Benjaminsson, J. Benjaminsson, R.B. Rudberg, Power-to-Gas – A technical review, Tech. Report.

(2013).

[10] G. Gahleitner, Hydrogen from renewable electricity: An international review of power-to-gas pilot

plants for stationary applications, International Journal of Hydrogen Energy. 38 (2013) 2039–2061.

doi:10.1016/j.ijhydene.2012.12.010.

[11] M. Jentsch, T. Trost, M. Sterner, Optimal Use of Power-to-Gas Energy Storage Systems in an 85%

Renewable Energy Scenario, Energy Procedia. 46 (2014) 254–261.

doi:10.1016/j.egypro.2014.01.180.

[12] EU Commission, EU Commission Task Force for Smart Grids Expert Group 4 Smart Grid aspects

related to Gas, 2011.

[13] E. Cristostomi, A. Franco, M. Raugi, G. Giunta, The smart gas grid: state of the art and perspectives,

in: 4th IEEE PES Innovative Smart Grid Technologies Europe (ISGT Europe), Ocotober 6-9, 2013,

Copenhagen, 2013: pp. 1–5.

[14] I. Vos, The Impact of Wind Power on European Natural Gas Markets, International Energy Agency -

Working Paper. (2012).

[15] L. Hirth, The market value of variable renewables, Energy Economics. 38 (2013) 218–236.

doi:10.1016/j.eneco.2013.02.004.

19

[16] J. Vandewalle, K. Bruninx, W. D’haeseleer, A mixed-integer linear operational model of a coupled

electricity , natural gas and carbon energy system with power to gas, TME Working Paper. (2014).

https://www.mech.kuleuven.be/en/tme/research/energy_environment/PublicationsEnergyanden

vironment/.

[17] N. Keyaerts, Gas balancing and line-pack flexibility concepts and methodologies for organizing and

regulating gas balancing in liberalized and integrated EU gas markets, PhD Thesis, KU Leuven.

(2012).

[18] P. Cramton, S. Stoft, The Convergence of Market Designs for Adequate Generating Capacity,

(2006). http://www.cramton.umd.edu/papers2005-2009/cramton-stoft-market-design-for-

resource-adequacy.pdf.

[19] Elia, System- en marktoverzicht 2012, (2012). http://www.elia.be/~/media/files/Elia/publications-

2/system-and-market-overview/systeem-en-markt_nl_2013-07-30.pdf.

[20] DNV KEMA, Study on LT-ST Markets in Gas, Tech. Report. (2013).

Appendix – Additional information

A. Input data domestic gas and electricity demands

The domestic gas and electricity demands are shown in Fig. A1 and Fig. A2, respectively. The start of the

data corresponds to the 1st of May. The annual domestic gas demand is 142 TWh. The gas demand

depends on the outside temperature, as a lot of gas is used for heating. The minimum gas demand is 5.4

GWth and occurs during summer, the maximum is 41 GW. The annual electric demand is 77.8 GW and the

minimum is 5.58 GW and the maximum 12.8 GW. The electricity demand is less dependent on the outside

temperature.

Fig. A1. Domestic gas demand.

Fig. A2. Domestic electricity demand.

0 100 200 300 4000

10

20

30

40

50

Time (days)

Dom

estic g

as d

em

and (

GW

th)

0 100 200 300 4000

2

4

6

8

10

12

14

Time (days)

Dom

estic e

lectr

ic d

em

and (

GW

e)

20

B. Power plant and power to gas plant characteristics

Plant plant parameters are based on typical CCGTs and GTs, see Table B1. Concerning power to gas plants,

however, there are no operational data available. Hence, a hypothetical plant is assumed, roughly based

on available knowledge from existing small-scale technology in [3]. We assume a rated input of 500 MWe

and a rated efficiency of 65% referring to the gross heating value of the output renewable methane.

Minimum operating points of 5% are stated in [3], however, this would lead to very low efficiencies.

Therefore, we assume the minimum operating point at 25%. When the energy requirements to heat the

plants are 10% of the maximum input power, the minimum efficiency is then 35%. Further data regarding

the dynamics of the plant is assumed, see Table B1.

The model in this work uses aggregated plants, based on the characteristics in Table B1. Though there is a

small loss of information due to the aggregation, it is still sufficiently detailed to yield good results. For

more information, see [16].

Table B1. Used parameters for the high detail power generation and power to gas plants. (a) referring to lower heating value,

(b) assuming 60 €/MWhth of input natural gas for one hour at minimal output, (

c) referring to higher heating value of methane,

(d) assuming 10% of rated power is always needed to keep the plant at the right operating temperature, (

e) assuming a start-up

requires two hours of warm-up at 10% of rated power and 100 €/MWe.

CCGT GT PtG

Rated output MWe 500 100 Rated input MWe 500

Minimum output MWe 175 30 Minimum input MWe 125

Rated efficiency a % 58 40 Rated efficiency

c % 65

Minimum efficiency % 46 25 Minimum efficiency d % 35

Start-up cost b € 23 000 3 600 Start-up cost

e € 10 000

Maximum ramp up %/min. 6 10 Maximum ramp up %/min. 10

Maximum ramp down %/min. 6 10 Maximum ramp down %/min. 10

Minimum up time h 2 0.5 Minimum up time h 4

Minimum down time h 1 0.5 Minimum down time h 2