Economic assessment of a power-to-substitute-natural-gas process including high-temperature steam electrolysis

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

Economic assessment of a power-to-substitutenatural-gas process including high-temperature steam electrolysis Myriam De Saint Jean a,b, Pierre Baurens a, Chakib Bouallou b,*, Karine Couturier a a

Univ. Grenoble Alpes, CEA, Liten, 17 Rue des Martyrs, 38054, Grenoble, France Mines ParisTech, PSL e Research University, CES Centre efficacite energetique des systemes, 60 Bd St Michel, 75006, Paris, France b

article info

abstract

Article history:

Power-to-Substitute-Natural-Gas (SNG) processes are studied since they can offer solutions

Received 18 November 2014

for renewable energy storage and transportation. In the present study, an original Power-

Received in revised form

to-SNG process combining high-temperature steam electrolysis and CO2 methanation is

10 March 2015

economically assessed. This evaluation is based on experimental data describing the

Accepted 15 March 2015

electrolyser performance and its degradation during long-term operation. These data are

Available online 11 April 2015

obtained on a commercial single cell. Tests are originally conducted by imposing voltage and steam conversion rate. The main conclusions of this experimental work are that two

Keywords:

fields of current density evolution are evidenced in the tested conditions: a first transient

Substitute natural gas

one where the current density absolute value decreases very quickly, followed by a second

Power-to-Gas

field where the evolution of current density achieves a steady state. The influence of the

High-temperature steam-electrol-

operating point seems not significant once this state is achieved.

ysis

The process assessed here has been developed in a previous work where a precise

SOEC

framework matching with a wide range of applications was defined. The design of the

Performance degradation

process main units and the calculation of matter and energy fluxes are based on this work.

Economic assessment

To go further, the economic assessment of this process is proposed. In this study, various scenarios of initial performance of the electrolyser are considered and integrate the performance degradation associated. The SNG production cost is estimated between 211 and 570 V/MWhHHV according to the scenario and the hypotheses considered. The two major items accounting for this high cost are the purchase cost of the electrolyser and its performance degradation. It also evidenced that none of the other items can be neglected. Through a sensitivity analysis, it is shown that an increase of plant annual availability and capacity makes the SNG production cost decrease. Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

* Corresponding author. Tel.: þ33 1 691 917 00; fax: þ33 1 691 945 01. E-mail addresses: [email protected] (P. Baurens), [email protected] (C. Bouallou). http://dx.doi.org/10.1016/j.ijhydene.2015.03.066 0360-3199/Copyright © 2015, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.

6488

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

Introduction Power-to-Gas processes are considered as a possible and interesting solution to integrate efficiently wind and solar renewable resources into the current energy mix. This solution aims at using power to convert water into hydrogen via electrolysis [1], and storing the obtained fuel until having a high power consumption period when it would be reconverted into electricity. A further step could be considered with the conversion of the electrochemical hydrogen into methane thanks to the Sabatier reaction. This allows to produce Substitute Natural Gas (SNG) and then, it becomes possible to store the obtained fuel by injection into the natural gas grid [2,3]. Methane, which is the main compound of SNG, has several advantages over hydrogen which deal with volumetric energy content and safety concerns. In addition to these physical data, there is no limit for SNG injection into the gas grid whereas if hydrogen is produced, it can be injected into the grid in the limit of 6vol% for instance in the French grid [4], the current factual fraction being under 2vol%. Grid for transportation and distribution already exists in Europe, allowing to store and deliver natural gas and its substitute. Producing SNG is chosen here since, on top of the advantage of accessing the grid, it is a versatile compound, which can be used to generate thermal energy, chemicals, fuels for mobility and finally electrical energy, as illustrated in Fig. 1. Added to the social acceptability of natural gas currently observed, these arguments make the solution of Power-to-SNG interesting and relevant for further investigations. Units required to produce SNG were previously defined and modelled [3,5]. They consist of High-Temperature Steam Electrolysis (HTSE), methanation with the Sabatier reaction and finally gas purification. The original Power-to-SNG process including these units were simulated and results are presented elsewhere. This previous work allowed to calculate matter and energy fluxes of the process in an accurate framework matching to a wide range of applications. Components as electrolyser and methanation reactors were also designed in this work.

As a following study of Power-to-SNG modelling and simulation, the present work focuses on the economic assessment of such a process. As no other detailed Power-toSNG process simulation has been conducted, no economic assessment could have been done on this detailed basis. However, in Ref. [6], several Power-to-SNG processes from literature including low temperature electrolysis technologies are described and economically compared. Authors made hypotheses when data were missing, particularly concerning the electrolyser lifespan. In the same way, a German study concerning a Power-to-SNG process including low temperature electrolysis is carried out in Ref. [7] but it mainly focuses on the economic aspect and does integrate neither detailed technology specifications nor process behaviours. As a result of this study, low-temperature Power-to-Gas processes are difficult to be economically profitable under the German current regulations. To determine if high-temperature Power-toSNG processes are economically profitable, it is important to rely on detailed data of design and architecture and to consider, in a first step, the energy consumption and the SNG production observed for a given process, to evaluate in a second step, the production cost of SNG. Concerning the electrolyser, no data were found in literature reporting performance and evolution during long-term operation for electrolysis cells operating under conditions describing the Power-to-SNG process; that is to say at high steam-conversion rate at the thermoneutral voltage, these two parameters being constant during operation. However, data of performance and ageing must be integrated into the economic assessment. The performance is linked to the hydrogen production per surface area, and to the plant investment cost, whereas the time evolution of performance describes cells ageing and has to be included in the calculation of the plant operating cost. Experimental work is carried out to obtain data qualifying the electrolyser performance and its degradation under various working points in terms of temperature and steam conversion rate at the thermoneutral voltage. Experiments are conducted on commercial single cells. Results from long-term tests are used to propose mathematical

Fig. 1 e Power-to-SNG concept as a solution for electrical energy storage and transportation, [3].

6489

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

functions which describe the evolution with time of the hydrogen production according to the cell working point. Afterwards, the Power-to-SNG process is described and simulation results will be presented. Finally, all these results are gathered into the economic assessment to take into account a representative electrolyser performance and a degradation in accordance with the electrolyser working point used in the process simulation. The SNG production cost is then calculated and a sensitivity analysis is proposed to determine how it evolves according to the economic and context assumptions.

Experimental on SOECs Experiments were performed on a commercial single cell used as a lab reference [8,9]. Even though results are used to model a stack, single cell tests were implemented for test bench availability concerns and for simplification of test and results interpretation.

Tested cell and instrumentation The tested cells are hydrogen electrode supported. The H2 electrode, or cathode, is a NiO-8YSZ cermet (nickel oxide NiO þ 8vol% Yttria Stabilized Zirconia YSZ) with a thickness of 500 mm. The electrolyte, having a thickness of 5 mm, is 8YSZ. The O2 electrode, or anode, is made of Strontium-doped Lanthanum Cobaltite (LSC) and has a thickness of 20 mm, with a diffusion barrier layer of Gadolinia Doped Ceria (GDC) applied between YSZ and LSC. Cells are circular with a 5.0 cm outer diameter. The active area, matching with the O2 electrode of 2.0 cm in diameter, is 3.14 cm2. The cell is settled in an alumina housing ensuring gas distribution, see Fig. 2. A ceramic glass seal is laid on the edge of the cell for gas tightness between the two compartments. Gases are collected on the cathode side while they are directly released in the furnace on the anode side. A gold grid and a nickel grid are used as contact layers on the anode and cathode sides respectively. Four wires welded onto the grids allow to measure voltage independently from current application. A mechanical pressure of around 1 kg cm2 (z 0.1 MPa) is applied on the upper part of the alumina housing to improve electrical contacts. Temperature is measured by a thermocouple set in the alumina housing, below and as near as possible to the cell geometrical center.

Table 1 e Experimental conditions. Pure oxygen on anode side, H2/H2O ¼ 1/9 on cathode side. Temperature (K) 1073 1073 973

Uop (V)

SC (%)

Number of tests

Utn ¼ 1.289 Utn ¼ 1.289 Utn ¼ 1.283

75 45 45

2 1 2

Experimental procedures Once the cell is settled as described in Fig. 2, temperature is increased at 1 K min1. A thermal treatment at temperature higher than 1073 K is implemented for sealing and electrical contact purposes. During this step, anode and cathode are fed with air and nitrogen respectively. The mechanical load is applied when temperature is maximal. After that, temperature is set at 1073 K for cermet reduction. To do so, a flow-rate of 12 NmL min1 cm2 of nitrogen is gradually replaced by hydrogen.1 Once the Open Circuit Voltage is stabilised, tightness is checked through OCV value and the ohmic resistance of the set-up is determined by electrochemical impedance spectroscopy (EIS). Then, gas composition is set at the ratio H2/ H2O of 1/92 used in the following study. Detailed procedures and intermediate results can be found in Ref. [10]. Long-term tests are performed under potentiostatic control and steam conversion rate is kept constant on the overall test duration, leading to a gradual decrease of the inlet flowrate on the cathode side. To do so, the conventional galvanostatic control system with constant flow-rates is modified into an original control system by adding two functions and set-points: voltage and steam conversion rate. Steam conversion rate SC is calculated thanks to equation (1) where Vm is the molar volume, j is the current density in A cm2, F the Faraday constant, xH2 O the steam content of the cathodic flow· rate ncath expressed in NmL min1 cm2. SC ¼

60 Vm jjj ·

2F xH2 O ncath

(1)

Conditions applied for long-term tests are summarised in Table 1. In each case, the anodic sweep gas is pure oxygen and the cathodic composition is set with a ratio H2/H2O equalling 1/9. Cell is operated at 1073 K at the thermoneutral voltage for an easy thermal management at stack level. This operating temperature is traditionally used in high-temperature electrochemistry and usually leads to a good compromise between performance and durability with standard SOEC materials.

Experimental results and extrapolations The initial cathodic flow-rate matching with the selected operating conditions is determined and the initial polarisation curves of Fig. 3 are obtained. It is noteworthy that OCV value obtained in each case is really close to the theoretical value of 0.912 V at 973 K and 0.875 V at 1073 K, under these gaseous conditions. It is observed that for tests carried out at SC ¼ 45%, evolution between j and U is linear close to the thermoneutral 1

Fig. 2 e Experimental set-up.

2

Normal conditions: 273.15 K, 101,325 Pa. All ratios are molar ratios.

6490

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

Table 2 e a and b coefficients to be used with equation (2) for interpolation functions. Temperature (K) SC (%) a b R2

Fig. 3 e Initial polarisation curves for different temperatures, flow-rates and SC values at Utn. Cathodic · flow-rate ncath is expressed in NmL min¡1 cm¡2, with a ratio H2/H2O of 1/9, and pure oxygen sweep is used on the anode side.

voltage whereas it is not the case for tests at SC ¼ 75%, j being close to the limiting current. It is also noticeable that for both tests carried out at 1073 K, SC ¼ 75%, polarisation curves are similar while for tests at 973 K, SC ¼ 45%, a gap between the obtained curves is observed. It appears that the flow-rate for the test reported in black is nearly 9% lower than typical values obtained with this kind of cells. However, as the OCV and the ohmic resistance belong to the valid range, confirming good tightness and electrical contact, a cell dispersion is suspected.

1073

1073

973

973

75

75

45

45

45

0.842 0.387 0.994

0.937 0.356 0.991

0.743 0.301 0.973

0.376 0.670 0.962

1.124 0.633 0.988

After having qualified the initial performance with polarisation curves, long-term test is started with conditions specified in Table 1 and using the original control module. · Initial values for j and ncath are identical to these determined thanks to the polarisation curves, indicating that the control module responds well. Time evolutions of j are presented in Fig. 4a. Temperature is manually maintained at the nominal test value during these tests. Due to the decrease of j, irreversibilities and Joule effect in wires lower, implying an increase of the furnace set-point to compensate. Temperature evolution is depicted in Fig. 4b. In Fig. 4a, curves describing tests at 1073 K and SC ¼ 75% are similar and j drops with time to a more steady state regime. The shape of the curve obtained at 1073 K with SC ¼ 45% shows the same trend but with a more pronounced initial decrease. For this test, the initial current density being very high, irreversibilities are important too, and temperature adjustment is more important. It seems that the higher is the initial current density, the more pronounced is the first j evolution, at a set temperature. EIS measurements suggest that the causes of this performance decrease could be a degradation of the electrolyte ionic conductivity or the formation of a resistive phase (see Ref. [10] for details). Curves obtained at 973 K also show a j decrease but with a lower evolution. This smaller slope could be simultaneously due to temperature and initial j effects. More tests are needed to confirm this statement. However, the shape of the two curves under the same conditions is different. The cell previously identified with a lower initial performance thanks to the polarisation curves shows a lower initial current density again but also a smaller slope of j. These data are used to determine the hydrogen production variation when cells are operated during a long period. The obtained signals are mathematically studied to propose extrapolated functions. The function giving the best fit is determined by interpolation and is described by equation (2). The coefficients a and b as well as the correlation coefficient R2 are reported into Table 2 for time t expressed in kh in equation (2). jðtÞ ¼

Fig. 4 e Evolution of j and temperature during long-term tests.

1073

1 at þ b

(2)

Extrapolated functions are plotted in Fig. 5 where it is observed that a steady state is achieved after 1.3e2.0 kh. The beginning of this steady state, where j variation would be more stable and limited, is fixed when the second derivative of j is lower than 0.2. This specific time is denoted t0 and is identified by a star symbol on curves of Fig. 5. The corresponding value of j is calculated as well as the value obtained after a one year working period (8.8 kh). The mean degradation for this working period, denoted Dj/j, is also reported in Table 3.

6491

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

Table 4 e Gas specifications for the French natural gas grid [4]. 3

H quality

L quality 9.5e10.5 11.8e13.0

HHV W

kWh/Nm kWh/Nm3

10.7e12.8 13.4e15.7

Composition

%vol mg/Nm3

CO < 2, CO2 < 3, H2 < 6 H2O < 55

elements of the process are reminded here, as well as some energy and matter main flux characteristics. Fig. 5 e Extrapolated j(t) curves.

Description

Once steady state is reached, the extrapolated current density relative degradation Dj/j at 1073 K and SC ¼ 75% is 9%/ kh, which is similar to the degradation observed with SC ¼ 45%, all other things being equal. For long-term operation, decreasing steam conversion rate could not show any influence on the relative degradation in these tests. For the tests at 973 K, j relative degradation is between 8.4%/kh and 9.1%/kh, which is also very close to the previous values obtained at 1073 K, whatever SC value. Then, temperature does not seem to influence the extrapolated relative degradation obtained by this method. Even though the extrapolated current density degradation is not influenced by temperature and SC once steady state has been reached, the current value is slightly different according to the operating conditions, the trend being similar to what was observed on the polarisation curves. j(t0) values belong to a range extending between 0.56 and 0.41 A cm2. The signal in black, dealing with a low initial performance and low degradation is an exception, where j(t0) is between 54% and 110% higher than the other values calculated. From a general point of view, it seems that whatever the operating conditions, current density values tend to be into a range between 0.6 and 0.4 A cm2 at the beginning of steady state and once in this state, the extrapolated relative degradation is identical whatever the conditions applied. Thus, from an industrial point of view, it may become interesting to operate the electrolyser at high steam conversion rate, since it does not influence its lifespan in the studied range and it allows optimisation of the steam management at process level.

Power-to-SNG process The Power-to-SNG process assessed in this work was described, modelled and simulated previously [3]. Main

The aim of the process consists in producing SNG satisfying specifications for injection into the gas grid. These specifications are detailed in Table 4 and concern the Higher Heating Value HHV, the Wobbe index W and the maximum content in CO2, CO, H2 and H2O. Considering that SNG must be pressurised at 0.4 MPa at least to be injected into the grid, the process is expected to work at a high enough pressure to achieve 0.4 MPa for SNG at the process oulet. Three main steps are required to produce SNG from electrical power, water and CO2, as illustrated by Fig. 6. The first one consists in producing hydrogen from water into an electrolyser. High-Temperature Steam Electrolysis including Solid Oxide Electrolysis Cells (SOEC) is selected for this work. Then, hydrogen is used with additional CO2 to feed the methanation reactors. CO2 is assumed pure and coming from an industrial capture process. Another possible source could be biogas. The gas exiting the methanation unit contains too much water, unreacted H2 and CO2 to meet injection specifications and needs to be purified into the last unit. High-Temperature Steam Electrolysis (HTSE) unit aims at producing H2 from water. It includes an electrolyser made of SOECs where electrochemical reaction R1 occurs. H2 OðgÞ #H2 þ½O2

(R1)

This unit is depicted in Fig. 7 where the full Power-to-SNG process is illustrated. Due to the pressure constraint, the unit working pressure is between 1.1 MPa and 0.7 MPa, and the operating temperature is 1073 K. Incoming liquid water is firstly pumped to 1.1 MPa before being turned into superheated steam. Steam is partially heated-up to the electrolyser working temperature by streams exiting the electrolyser and by an electrical superheater to compensate the temperature pinch. SOECs are fed by steam containing 10% of hydrogen on the cathodic side, the H2-rich cathode outlet stream being partially recycled to the cathode inlet thanks to a gas ejector

Table 3 e Characteristics of steady state and j degradation for a one-year working period. Temperature SC t0 j(t0) j(t0þ8.8) Dj/j

K

1073

1073

1073

973

% kh A cm2 A cm2 %/kh

973

75

75

45

45

45

1.9 0.50 0.11 9.0

1.9 0.47 0.096 9.0

2.0 0.56 0.12 8.9

1.3 0.86 0.22 8.4

1.6 0.41 0.081 9.1

6492

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

Fig. 6 e Process and process environment, pressure and temperature are reported as pressure (MPa) j temperature (K), [3].

for this purpose. The cathode outlet gas containing 77.5% of H2 and 22.5% of H2O, the stream is cooled-down until the residual steam content into the gas is 0.5%. On the anodic side, pure oxygen is produced by the electrolysis reaction. This stream is cooled-down and then compressed to 1.8 MPa to be stored for a further valorisation. Electrolyser modelling is based on mass and energy balances. The electrochemical reaction kinetics is considered through a function obtained experimentally linking the cathodic inlet flow-rate to the current density, that is to say to the hydrogen production density, at 1073 K and at the thermoneutral voltage, with a pure oxygen sweep on the anode side and a cathodic composition H2/H2O equalling 1/9. This was presented in detail and modelled elsewhere [3]. The

incoming flow-rate is set so that the steam conversion rate in electrolyser is 75% at the thermoneutral voltage. The hydrogen produced by HTSE unit is used to hydrogenate carbon dioxide and form methane thanks to reaction R 2 into the methanation unit. CO2 þ 4H2 #2H2 O þ CH4

(R2)

Thermodynamic studies have shown that this reaction is unique when H2/CO2 equals 4 and temperature is lower than 750 K [11]. Low temperature and high pressure promote CO2 conversion and CH4 production. However, this reaction requires a catalyst to increase kinetics. Thus, operating conditions must be adapted to this constraint and temperature must be high enough for catalyst activation. It must also be

Fig. 7 e High-temperature Power-to-SNG process. Pressure and temperature are reported as pressure (MPa) j temperature (K), arrows: heat supply/removal, mixed dotted lines: electrical consumption.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

considered that this reaction is highly exothermic 1 (DH0273K ¼ 163 kJ mol ). The kinetic law considered here and the reactor modelling are described in detail in Ref. [3]. The reactor inlet temperature selected as a trade-off is 573 K. Reactors are assumed adiabatic. High CO2 conversion into the unit is achieved by using four reactors alternating with cooling steps. To avoid a dramatic temperature increase in reactor R1, damageable for its catalyst bed, outlet gases are recycled at 75% and only 95% of the stoichiometric quantity of CO2 feed reactor R1, the remaining quantity feeding directly reactor R2. In these conditions, outlet temperature of reactors R1, R2 and R3 are respectively 819 K, 723 K and 649 K, inlet pressure of reactor R1 being 0.68 MPa. Between reactors R3 and R4, the water produced by the reaction is partially removed by condensation at 392 K. The quantity removed is adapted to avoid both carbon deposition and carbon monoxide production in the following reactor, this limit being determined thermodynamically. Outlet temperature of reactor R4 is 633 K the pressure is 0.5 MPa. The exiting stream contains 40% of water, which is removed again by condensation at 308 K. This unit converts 96.8% of CO2 and heat recovered during cooling steps after each reactor is used to vaporise water in HTSE unit. The reaction of methanation is not total and the gas produced must be purified to lower CO2, H2 and H2O contents. To do this, the first step in the purification unit consists in removing CO2 with an amine scrubbing process. The resulting gas goes through a membrane device where H2 is removed and finally, the gas is dried by cooling down and water condensation [3]. The recovered CO2 and H2 are recycled toward the adequate methanation unit incoming stream, see Fig. 7. The exiting SNG contains then 97.3% of CH4, 1.35% of CO2, 1.30% of H2 and 53 mg/Nm3 of water and satisfies specifications for grid injection.

Simulation hypotheses Concerning the design of heat-exchanger cascade, one heatexchanger ensures a temperature gap between the inlet and the outlet of 150 K when the maximal temperature is lower than 873 K, this gap becoming 100 K for a higher maximal temperature. Each item induces a pressure loss of 0.02 MPa. A pinch analysis module is integrated into the simulation and determines the minimal hot and cold utilities required by the process. Heat fluxes are integrated with the temperature pinches specified in Table 5. The electrical power consumption to produce these utilities is then determined. Cold utility is provided by refrigeration units. Two different-temperature units are proposed, 217 K and 270 K, showing an electrical-to-thermal efficiency of 1.4 and 1.73 respectively. Cold source is ambient air, at 303 K.

Table 5 e Pinch values according to the temperature field for the determination of the ideal thermal integration, data are in K. Field T  373 373 < T  873 873 < T

DT 10 30 50

6493

In the selected configuration, the process does not required any hot utility except the electrical superheater dedicated to the steam superheating from 1023 to 1073 K due to the pinch value at this temperature level. The isentropic efficiency of compressors is assumed to be 80%, the mechanical efficiency being 95%. Concerning the water pump, it is assumed that the volumetric efficiency is 95%. The AC/DC converter used to feed SOECs with direct current is assumed to have an efficiency of 92%. Utilities required by the process as thermal energy or mechanical work are assumed to be entirely provided from electricity. The electrical consumption of the process is split according to the following balance of plant: Pel,HTSE is the AC power required for electrolysis, Pel,heat and Pel,cold are the electrical equivalents of the heat demand and cold demand respectively. Pel,heat is the equivalent electrical power consumed by the steam superheater. Pel,mech is the electrical consumption for the supply of the mechanical work required in the process. The efficiency hHHV is the ratio between the higher calorific value of the SNG flow-rate noted QSNG and the total AC electrical power input and is calculated with formulae 3. hHHV ¼

Q SNG HHVSNG Pel;HTSE þ Pel;heat þ Pel;cold þ Pel;mech

(3)

Simulation results The process simulation gives the results summarised in Table 6. The produced SNG meets the gas grid injection requirements in terms of composition, HHV and W values, this latter being 14.4 kW h/Nm3. The process efficiency hHHV achieves 75.8% which is between 8 and 14 points higher than low temperature Power-to-Gas process efficiency [6,12]. Some flow-rates are reported as they will be used to determine the cost of the associated equipment. The total catalyst volume Vcat required in the methanation reactors is also reported for this purpose.

Economic assessment To determine the production cost of SNG, each equipment required in the process is designed to estimate the capital cost. Equipment life span must also be considered to determine the operating cost. General economic assumptions concerning the Power-to-SNG plant are summarised in Table 7. Availability of cheap electricity matching to the plant annual availability comes from a study on renewable energy deployment and storage presented in Ref. [13]. To make to the economic assessment of this Power-to-SNG process, the production cost of SNG CSNG is determined with equation (4), this requiring the knowledge of the capital cost Cinv and the annual operating cost Cop for the considered plant. Pt is the production of year t expressed in MWhHHV and t is the discount rate. Ph CSNG ¼

t

i  Cinv;t þ Cop;t ð1 þ tÞt i Ph Pt ð1 þ tÞt t

(4)

6494

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

Table 6 e Results of the Power-to-SNG process simulation. QSNG Pel,HTSE Pel,heat Pel,cold Pel,mech hHHV

Nm3/h kWAC kW kW kW %

67.2 879 9.9 52.6 15.0 75.8

Table 7 e General economic assumptions. Construction period Project life Plant annual availability Discount rate t

y y % %

kWh/Nm3 kg/d kg/d kg/d kg/d m3

HHV CO2 capture H2 capture O2 production Net H2O consum. Vcat

2 30 51, i.e. 4500 h/y 10

10.79 60 16.8 4560 2568 1.56

operating cost, as well as the cost of the pressure vessels containing the stacks. The optimal maximum age is obtained when the sum of these costs is minimum. Values characterizing each scenario are reported in Table 8. The reported current density values take into account pressure and stack effects [3]. Parameter Sextra denotes the surface area to add to S in order to maintain hydrogen production considering performance degradation.

Electrolyser stacks management strategy Capital cost It is assumed that stacks forming the electrolyser integrate 100 cells, recent stacks of 25 cells [8,9] and 50 cells [14] having been successfully implemented. Each cell is assumed with an active area of 10  10 cm2 and a stack has an active area of 1 m2. As shown in Experimental results and extrapolations section, hydrogen production decreases sensibly during long-term uses under the operating conditions selected. A specific strategy must be set up to maintain H2 production level. The strategy chosen here consists in adding extra stacks to increase the active area and to age stacks homogeneously and continuously. This strategy allows to keep constant parameters considered constant in the process simulation. To take into account the SOECs performance degradation, data obtained experimentally are used. Results from Experimental results and extrapolations section help to determined the cell current density and the associated degradation. Due to the large value range observed for these parameters, several scenarios are studied. The first one, the risky scenario, integrates the initial value of the current density and its time evolution is determined from this point thanks to the proposed extrapolated functions. The second one, the current scenario, uses the value of the current density at the beginning of the steady state and its time evolution is determined with the extrapolated functions from this point. In the last scenario, called ideal prospective, the current density is the same as in the current one and it is assumed that there is no degradation. For each scenario, the optimal maximum age of stacks is determined considering the stack capital cost and the stack

Capital cost is determined by the Chauvel method [15]. This method allows to determine the investment for a plant knowing the cost of the required installed equipments. This method uses the capital cost breakdown which is illustrated in Table 9 where values of the economic parameters are also indicated. Each parameter is described thereafter and its value is selected according to the context of this study. In the Chauvel method, investment in limit of the production units Ilpu matches with the installed cost of equipments, including primary and secondary materials, development expenditures, assembly operations and indirect costs of building. Usually, provision for contingency is added. A rate of 30% of installed cost is applied in the current study. General services include administration, industrial store, and maintenance workshop. In first approach, they are assessed to be 15% of Ilpu. Engineering cost includes costs for the feasibility studies, project monitoring, supervision and acceptance of the work for instance. Generally the value is between 7 and 30% of Ilpu with an average at 12%. For highly industrial countries, cost linked to the stock of spare parts can be neglected since they can be supplied in reasonable delivery time. Contractor cost is related to operating license and contractual documents. It is set to 5% of the unit investment. Other cost components like initial expense, working capital and start-up expense are neglected in this study. All these sub-costs are added to form the final capital cost, or investment, as shown in Table 9.

Table 8 e Scenario data to maintain the hydrogen production of the electrolyser. j is given at the beginning of the scenario. Scenario S j Degradation Sextra Max. age

2

m A cm2 % y

Risky

Current

Ideal prospective

25 2.51 Initial, important 571 1

115 0.547 Steady state, low 12 1

115 0.547 ∅ 0 30

6495

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

Table 9 e Capital cost breakdown of the total investment on industrial plant. Total investment Cinv Capital cost of depreciable properties Fixed capital Unit investment Ilpu P

Inst. cost +30%

General services

Engineering

Spare parts

Contractor cost.

Interest construc.

Initial expense

Start-up expense

Working capital

15% Ilpu

12% Ilpu

∅

5% Units inv

Construc. period

∅

∅

∅

To determine the installed cost of required equipments, correlations from literature and prices determined at CEA Liten are used. They are reported in Table 10. If data are not expressed in 2012 currency, the CEPCI index is used to take into account price and service inflation. The production cost of electrochemical stack is estimated at 1890 V/m2 by CEA Liten internal studies. The assessed stacks integrate cells that are similar in materials and structure to the ones used so far. A profit margin, set at 10% of this cost, is added to obtain the selling price. Installation and additional material are considered according to ref. [15] by multiplying it by 2.91. Installed cost for stacks is then 6071 V/m2. When a stack is only replaced, replacement cost includes selling price and stack installation and is 3686 V/m2. Due to high pressure and high temperature working conditions, stacks must be included into pressure vessels. The number of vessels is linked to the maximum working time of a stack. All stacks contained in one vessel are installed or replaced at the same time. So, the number of required pressure vessels equals the maximal age for stacks plus one, this additional vessel helping the maintenance operation and allowing to anticipate the installation of new stacks in replacement of old ones. The volume of each pressure vessels depends on the number of stacks inside. Each stack has a square base measuring 0.12  0.12 m and is 0.5 m long. Including hightemperature heat-exchangers and piping, the total volume for a stack and its utilities is set at 0.03 m3. The void fraction into the pressure vessel, dedicated to the thermal insulating and the inlet piping, is set at 60%. Correlation for the installed cost is based on supplier data and is indicated in Table 10. Cost for thermal insulation to add is the half of pressure vessel cost. Other installed costs are taken from literature or from internal data and are presented in Table 10. Installed cost for the AC/DC inverter is based on literature data.

Heat exchanger network cost is based on the cost obtained in a HTSE process study carried out at CEA Liten, where the electrolyser working point is similar to the one used in this work. For each compressor required in the Power-to-SNG process, a correlation from literature is used. This is not applied to the recycling compressor since its cost is already included in the data used to describe the methanation reactors. Installed cost for methanation reactors is obtained in a study where syngas methanation in three adiabatic reactors is considered [18]. Gas is cooled-down between each reactor and 83% of the gas is recycled around the first reactor. Heatexchangers and recycling compressor are included in the reported cost. Amine scrubbing is characterised by the amount of CO2 captured per hour, noted Q capt CO2 as well as the filtering membrane cost is linked to the amount of H2 captured per day, noted Q capt H2 . Data concerning the cooling unit describe its capital cost and not the installed cost. Thus, all cost categories as presented in Table 9 are already included in the cited value. Then, this value will be added to the all other equipment capital costs to obtain the total plant capital cost. Finally, installed cost for instrumentation and control is set at 10% of the sum of the installed costs. Capital cost for a Power-to-SNG plant with a production capacity of 725 kWHHV is estimated at 6.99 MV or 9.63 V/WHHV for the risky scenario, 6.18 MV or 8.52 V/WHHV for the current scenario and 5.13 MV or 7.07 V/WHHV for the ideal prospective scenario. Capital cost distribution is detailed in Fig. 8. The two major equipments in terms of capital cost are the electrolyser stacks with a share between 26% and 28% and the pressure vessels, with a share between 14% and 25%. The purification unit including the amine scrubbing, the cooling unit and the filtering membrane represents 12.3% of the investment in the

Table 10 e Cost correlations in V2012 for process equipments. Exponent values come from Ref. [15]. Equipment Pressure vessel AC/DC inverter Heat exchangers Compressor Methanation reactors Amine scrubbing Filtering membrane Cooling unit

Correlation 3

3

Reference 0.6

15 10 (Volume (m )/0.0785) 160  P(kWAC) 84 103 (Q H2 (kg/d)/100)0.65 267 103 (P(kWel)/445)0.67 82.5 106 (P(MWhHHV)/1770)0.65 49.5 106 (Qcapt CO2 (t/h)/408)0.65 380.3 103 (Qcapt H2 (kg/d)/100)0.60 13.6 106 (P(MWth)/8.29)0.70

CEA Liten [16] CEA Liten [17] [18] [19] [20] [21]

6496

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

Fig. 8 e Investment repartition for a Power-to-SNG plant producing 725 kWHHV.

first scenario, 13.9% in the second one and 16.8% in the last one. This share is quite similar to the methanation reactor share which evolves between 14% and 19% of the investment cost.

Operating cost Operating cost includes energy and matter consumptions, stack replacement and operating charges as maintenance, taxes and insurance, general expenses and work force. Values for energy and matter prices and assumptions for operating charges are summarised in Table 11. Electricity price includes only the grid access, tax, transmission and various other charges [22] since the Power-toSNG plant aims at consuming electricity when it is massively produced by renewable resources. Deionised water is required to feed the electrolyser. Methanation catalyst volume equalling 1.56 m3 is assumed to be fully replaced every two years and its cost is provided in Refs. [23], this value being in accordance with some supplier data giving a cost of 230 V/kg. These data are for nickel based alumina pellet catalyst having a bulk density of 750 kg/m3. Amine make-up is required with the rate of 3 kg per ton of captured CO2 (European project ReCO2) and amine costs 1520 V/t [19]. Regular maintenance is set at 4% of the investment except stacks per year. Stacks are considered with a regular replacement which cost includes mounting charges and purchase cost. This

replacement cost equals 6.07 and 4.67 MV/year for the risky and current scenarios respectively and is zero in the ideal prospective scenario. Work force is assumed to be 0.2 worker full time, this matching to an annual cost of 22.9 kV. Finally, operating cost is assessed at 6.63 and 5.20 and 0.482 MV/y for each of the three scenarios. Cost distribution is illustrated in Fig. 9. Annual operating cost represents 95% of the investment for the risky scenario, this part becoming 84% and 9.4% for the current and ideal prospective scenarios respectively. Except stack replacement and electricity for the electrolyser, the annual operating cost amounts at around 7% of the investment, which is in accordance with values reported in literature for catalytic processes [23].

SNG production cost SNG production cost is determined from capital and operating costs as mentioned in equation (4) and values are reported in Fig. 10. O&M share includes operating cost items except stack replacement cost and electricity supply. SNG production cost is maximal at 567 V/MWhHHV in the case of the risky scenario, intermediate at 494 V/MWhHHV for the current scenario and minimum at 304 V/MWhHHV for the ideal prospective scenario. By comparison of the current scenario and the ideal prospective scenario, which are similar except concerning the stack performance ageing consideration, the part of the cost linked to the stack degradation is assessed at 190 V/MWhHHV

Table 11 e Hypotheses for the calculation of the plant operating cost. Energy and matter Electricity Deionised water Catalyst Amine

V/MWh V/m3 kV/m3 V/t

Operating charges 25 10 187.5 1520

Maintenance Tax. & ins. General exp. Work force

% (Cinv eCinv,stack) % Unit inv. % Unit inv. worker

4 2 1 0.2

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

6497

Fig. 9 e Operating cost distribution for the Power-to-SNG plant producing 725 kWHHV.

in the case of the current scenario, which is 38% of the SNG production cost. In each case, the capital cost represents between 38% and 52% of the production cost. In the case of the risky and current scenarios, the O&M share is estimated between 23% and 25% respectively, this becoming 37% for the last scenario. The values of SNG production cost are high enough, this being a consequence of hypotheses on plant availability and capacity, as shown by the following sensitivity analysis.

Sensitivity analysis In this sensitivity analysis, several parameters linked to the economic assumptions are varied in the case of the current

scenario. The economic impact of the purification unit is also evaluated. Parameters are detailed in Table 12 with pessimistic and optimistic values around the reference value used so far. Results are illustrated in Fig. 11. It is noticeable that the discount rate and the stack selling price have a strong influence on the SNG production cost. When the discount rate is set at 5% and 15%, the production cost varies respectively from 15% to 17%. For a reduced stack selling price, set at 1100 V/m2, this matching with a very large stack production plant and a profit margin equalling zero, SNG cost decreases by 20% and for a selling price of 2750 V/m2, corresponding to a profit margin of 45% instead of 10%, the increase is evaluated at 13%. SNG purification costs 42 V/MWhHHV i.e. 8.4% of the production cost.

Fig. 10 e SNG production cost distribution for the Power-to-SNG plant producing 725 kWHHV.

6498

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

Table 12 e Parameter values used in the sensitivity analysis. Parameter Discount rate Construction period Project life Stack max. age SNG purification Stack selling price Catalyst cost Electricity cost O2 valorisation CO2 cost

Pessimistic

Reference

Optimistic

15 3 20 2 e 2750 250 50 e 50

10 2 30 1 With 2079 187.5 25 0 0

5 1 40 e Without 1100 150 0 20e70 50

% y y y e V/m2 kV/m3 V/MWh V/t V/t

Fig. 11 e Results of the sensitivity analysis for the Power-to-SNG plant producing 725 kWHHV in the case of the current scenario. Variations on matter and energy prices as catalyst and electricity costs imply variations of þ2.7% and 2.1% and of þ6.2% and 6.2% respectively for the selected values. In cases where the O2 produced by electrolysis is sold at 70 or 20 V/t, the gain on the production cost is 3.7 and 1.1% respectively. Concerning CO2 consumption, in cases where its consumption costs 50 V/t, the loss on the production cost is 1.8% and if its consumption is paid at 50 V/t, the gain is 1.8%. Using a stack during two years before replacing it increases the cost by 3%, this being linked to the extra stacks installed to compensate the performance degradation. This result shows that stack and pressure vessel prices are so important that it is economically favourable to install few stacks and to replace them more often under the present assumptions. SNG production cost varies by ±1.7% when construction period varies by ±1 year and it increases by 3% when project life is set at 20 years instead of 30. Increasing project life to 40 years implies a cost reduction of 1.6%. To focus on the influence of electricity cost, Fig. 12 shows the evolution of the SNG production cost in function of the electricity cost in the case of the current scenario. It also illustrates the influence of the plant annual availability. The highest cost is obtained for the lowest annual availability and for an electricity price of 50 V/MWh, and is 186% higher than the cost obtained with the reference case and equalling 494 V/ MWhHHV. Production cost is minimal for a fully available plant and for free electricity. Under these conditions, the cost is reduced by 50%.

The last step of this analysis consists in studying the influence of plant capacity. Fig. 13 reports variations in the case of the current and the ideal prospective scenarios. When plant capacity increases up to 150 MWHHV, SNG production cost decreases very quickly in a first step before achieving a plateau. This is a consequence of the equipment cost expressions (see Table 10), the higher is the capacity, the lower is the relative cost. If plant capacity is set at 150 MWHHV instead of 725 kWHHV, SNG production cost is 365 V/MWhHHV, corresponding to a gain of 26% in the case of the current scenario.

Fig. 12 e Influence of electricity cost and plant annual availability on CSNG for a Power-to-SNG plant producing 725 kWHHV in the case of the current scenario.

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

Fig. 13 e Influence of plant capacity on CSNG for the current and the ideal prospective scenarios.

For the ideal prospective scenario, SNG production cost becomes 211 V/MWhHHV when plant capacity is 150 MWHHV.

6499

long-term degradation will lower cost of SNG production. This is one of the main improvements of this application. Through the sensitivity analysis, it is shown that an increase of plant annual availability and of plant capacity makes SNG production cost decreasing by 50% and 26%, respectively, the lowest cost obtained in this analysis being 211 V/MWhHHV. In this first economic assessment of a Power-to-SNG process based on experimental performance for HTSE, it is shown that to produce a more competitive SNG, the selling price of stack must be as low as possible, the plant availability and the plant capacity must be as high as possible. The stack management strategy must take into account investment and replacement cost to determine the most profitable stack maximal age and to provide a closer-to-reality SNG production cost. High improvements on stack performance durability would help to improve economics of Power-to-SNG processes. These conclusions are also linked to the regulatory frame since economic incentives could modify the results obtained here.

Conclusion A Power-to-SNG process is economically evaluated. This work is based on a previous work dealing with definition, modelling and simulation of the process [3]. The main units required in this application are a High-Temperature Steam Electrolysis unit, a methanation unit and a gas purification unit. Electrolyser modelling is based on experimental data specially obtained for this purpose. As a main simulation result, the process efficiency achieves 75.8%, which is higher than the efficiency of low temperature Power-to-Gas processes. This value addresses H2O and CO2 conversion into SNG, ready for the grid injection and does not include CO2 capture energy cost. To provide data for stack performance and ageing in accordance with the operating point chosen for the simulation part, experimental work was carried out. This work was done on a single cell with an original control strategy which sets operating voltage and steam conversion rate. Experimental results are mathematically processed to propose longterm extrapolations to then model a long-term stack behaviour for the economic assessment. The main conclusions of this experimental work are that the current density curve shows two fields: a first transient one of a few hundreds of hours, for which current density varies very quickly, followed by a second field for which current density evolution achieves a steady state. In this field, variations are lower and current density values are contained into a tighter range for all working points tested. The influence of the working point is not significant once this state is achieved. As a consequence, variations of steam conversion rate and operating temperature seem to have neither positive nor negative effect on electrolyser lifespan in the tested range. Afterwards, considering energy and matter fluxes obtained thanks to the process simulation, the process economic assessment is carried out according to various scenarios based on the previous experimental results. SNG production cost evolves between around 300 and 570 V/MWhHHV according to the scenario considered, this high cost being linked to the cost of electrolysis stacks and pressure vessels and being also highly dependent on stack performance degradation. Concerning SOECs, improvements on performance and

references

mes d’e lectrolyse a  [1] Petipas F. Conception et conduite de syste rature alimente s par des e nergies renouvelables haute tempe rieure des mines [Ph.D. thesis]. France: Ecole nationale supe de Paris, MINES ParisTech; 2013. [2] Gahleitner G. Hydrogen from renewable electricity: an international review of power-to-gas pilot plants for stationary applications. Int J Hydrogen Energy 2013;38:2039e61. [3] De Saint Jean M, Baurens P, Bouallou C. Parametric study of an efficient renewable power-to-substitute-natural-gas process including high-temperature steam electrolysis. Int J Hydrogen Energy 2014;39:17024e39.  [4] Tilagone R. Gaz naturel e Energie fossile. Techniques de nieur BM2591; 2004. l’inge [5] De Saint Jean M, Baurens P, Bouallou C. Study of the efficiency of a HT power-to-gas process. Chem Eng Trans 2014;39:391e6. [6] Benjaminsson G, Benjaminsson J, Boogh Rudberg R. Powerto-gas e a technical review, technical report. Svenskt Gastekniskt Center AB; 2013. [7] Reichert F. Wind-to-gas-to-money? Economics and perspectives of the power-to-gas technology [Master’s thesis]. Denmark: Department of Development and Planning, Aalborg University; 2012. [8] Di Iorio S, Petitjean M, Petit J, Chatroux A, Gousseau G, Aicart J, et al. SOE stack activities at CEA, B1307. In: 11th European SOFC and SOE Forum; 2014 [Lucerne, Switzerland]. [9] Reytier M, Di Iorio S, Chatroux A, Petitjean M, Cren J, Mougin J. Stack performances in high temperature steam electrolysis and co-electrolysis. In: 20th World Hydrogen Energy Conference (WHEC); 2014.  nerge tique et e valuation [10] M. De Saint Jean. Etude e conomique d’une boucle de stockage - de stockage d’e nergie e lectrique d’origine renouvelable sur me thane de synthe  se a  e lectrochimique re versible SOEC l’aide d’un convertisseur e rieure des SOFC [Ph.D. thesis]. France: Ecole nationale supe mines de Paris, MINES ParisTech; 2014. [11] Gao J, Wang Y, Ping Y, Hu D, Xu G, Gu F, et al. A thermodynamic analysis of methanation reactions of carbon oxides for the production of synthetic natural gas. RSC Adv 2012;2:2358e68.

6500

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 0 ( 2 0 1 5 ) 6 4 8 7 e6 5 0 0

[12] Zuberbu¨hler U., Specht M., Power-to-gas: experience with demo and pilot plants, up to 99 % methane, 8th IRES Conference. Session G3.  L, Girardeau P, Chammas M, Tarel G, [13] Renaud A, Fournie  nergies, Chiche A, et al. Etude sur le potentiel du stockage d’e technical report, Ademe. 2013. [14] Mineta S., Mizuki K., Yoshida Y., Sugita S., Kobayashi R., Performances of cost-reduced planar type SOFC stacks with anode-supported cell, in: solid oxide fuel cells 13 (SOFC-XIII), volume 57, ECS Trans., pp. 343e350. valuation [15] Chauvel A, Fournier G, Raimbault C. Manuel d’e conomique des proce  de s, editions Technip. 2001. e [16] Fu Q, Mabilat C, Zahid M, Brisse A, Gautier L. Syngas production via high-temperature steam/CO2 co-electrolysis: an economic assessment. Energy & Environ Sci. 2010;3:1382e97. [17] Park SH, Lee YD, Ahn KY. Performance analysis of an SOFC/ HCCI engine hybrid system: system simulation and thermoeconomic comparison. Int J Hydrogen Energy 2013;39:1799e810.

[18] Gribik A, Patterson M, Boardman R. Nuclear-integrated substitute natural gas production analysis, technical report. United States Department of Energy INL; 2010. [19] Abu-Zahra MR, Niederer JP, Feron PH, Versteeg GF. CO2 capture from power plants part II. A parametric study of the economical performance based on mono-ethanolamine. Int J Greenh Gas Control 2007;1:135e42. [20] Acquaviva J. High-performance, durable, palladium alloy membrane for hydrogen separation and purification. 2009. http://www.hydrogen.energy.gov/pdfs/review09/pd_07_ acquaviva.pdf. [21] O'Brien DJ, Roth LH, McAloon AJ. Ethanol production by continuous fermentationepervaporation: a preliminary economic analysis. J Membr Sci 2000;166:105e11. [22] Jørgensen C, Ropenus S. Production price of hydrogen from grid connected electrolysis in a power market with high wind penetration. Int J Hydrogen Energy 2008;33:5335e44. [23] Grond L, Schulze P, Holstein J. Systems analyses power to gas: a technology review, Technical Report. DNV KEMA Energy & Sustainability; 2013.