CHEMICAL ENGINEERING TRANSACTIONS
VOL. 61, 2017
A publication of
The Italian Association
of Chemical Engineering
Online at www.aidic.it/cet
Guest Editors: Petar S Varbanov, Rongxin Su, Hon Loong Lam, Xia Liu, Jiří J Klemeš
Copyright © 2017, AIDIC Servizi S.r.l.
ISBN 978-88-95608-51-8; ISSN 2283-9216
Biogas as a Renewable Feedstock for Green Ethylene
Production via Oxidative Coupling of Methane: Preliminary
Feasibility Study
Alberto T. Penteado*, Mijin Kim, Hamid R. Godini, Erik Esche, Jens-Uwe Repke
Process Dynamics and Operations Group - Technische Universität Berlin, Sekretariat KWT9, Straße des 17. Juni 135,
10623, Berlin, Germany
albertopenteado@mailbox.tu-berlin.de
A preliminary feasibility study for the use of biogas as feedstock for the oxidative coupling of methane process
aiming at green ethylene production is carried out. An economic assessment is performed based on literature,
market and process simulation data, with the uncertainties being considered through Monte Carlo simulations.
It is shown that the proposed process is economically interesting under a wide range of scenarios. The
challenges and opportunities for the implementation of the process are highlighted to guide further studies.
1. Introduction
Biogas or landfill gas are methane-rich gas mixtures also containing carbon dioxide and lesser amounts of
contaminants. It is produced in the anaerobic treatment of effluents or energy crops and often combusted in
biogas engines for heat and power generation after minor prior treatment. Through the removal of CO2, biogas
can alternatively be upgraded into biomethane, which can then be transported via regular natural gas pipelines
after a final adjustment of the dew point and heating value and odorization (Sun et al., 2015).
The Oxidative Coupling of Methane (OCM) consists in the catalytic oxidation of methane and has been
investigated since the 1980’s (Keller and Bhasin, 1982) as a promising path for utilizing natural gas, and more
recently shale gas, as a precursor for chemical building blocks such as ethylene. Different reaction paths have
been proposed, but it is generally accepted that the desired selective oxidation path leads methane to be
adsorbed in the catalyst surface and converted into the intermediary ethane (Arndt et al., 2012). Ethane reacts
further, either on the catalyst surface or in the gas phase, to form the main product ethylene. The unselective
oxidation of hydrocarbons into carbon monoxide and carbon dioxide, as well as the reforming and water-gas
shift reactions complement the highly exothermic reaction system. Previous contributions including modelling
(Esche et al., 2012) and mini-plant experiments (Godini et al., 2014) regarding the OCM reaction section have
been published in related literature. The separation after the reaction consists of a compression and water
removal, CO2 removal, and distillation and it is crucial for the economic performance of the process. The CO2
removal section have also been assessed based on optimisation studies (Esche et al., 2015), as well as through
mini-plant experimentation (Stünkel et al., 2012). The process flow diagram for the OCM process is depicted in
Figure 1. Herein, biogas is considered as an alternative feedstock for the OCM process.
The conversion of biogas into chemicals has not yet been extensively investigated and would enable another
option for the biogas monetization and to produce fuels and chemicals through greener paths. This would allow
for a distributed production of base chemicals, especially in countries with abundant biomass resources and
limited fossil resources. While some research groups have already investigated the reforming of biogas to
produce syngas (Chen et al., 2017), no publication is found dealing specifically with the use of biogas as a
feedstock for the OCM process for ethylene production. The aim of this contribution is to assess the feasibility
and potential for the utilization of biogas as a raw material for a bio-based OCM process or Bio-OCM. The scope
of this study is limited to the techno-economic feasibility and other relevant aspects such as health, safety, and
environment shall be addressed in the future once more detailed information is available.
DOI: 10.3303/CET1761096
Please cite this article as: Penteado A.T., Kim M., Godini H.R., Esche E., Repke J.-U., 2017, Biogas as a renewable feedstock for green
ethylene production via oxidative coupling of methane: preliminary feasibility study, Chemical Engineering Transactions, 61, 589-594
DOI:10.3303/CET1761096
589
Figure 1: Process flow diagram of the OCM process
2. Methods
A biorefinery in Brazil producing 151x103 m³/y of anhydrous bioethanol and 1.2x106 m³ of vinasse, which is a
liquid effluent, is taken as a base case. The vinasse is treated anaerobically in sequencing batch reactors with
immobilized biomass yielding 2,200 Nm³/h of biogas with 70 mol% CH4 and 30 mol% CO2 (Albanez et al., 2016).
Such compositions are also found in other biogas sources (Rasi et al., 2007). The OCM plant nominal capacity
results in 800 t/y of ethylene, which is tiny when compared to naphtha or ethane cracking plants.
2.1 Biogas treatment
Besides methane and carbon dioxide, different contaminants in varying concentrations are expected to be
present in the raw biogas. The nature and quantity of the contaminants vary significantly among literature
sources, but typical compositions encountered might include up to 4,000 ppm H2S, 15 ppm Siloxanes, 100 ppm
NH3, and chlorinated compounds (Sun et al., 2015). Other sources mention lower values for those, but also the
presence of up to 287 mg/m³ of aromatics (Rasi et al., 2007). There are several different techniques that can
be used to treat biogas prior to its use depending on the level and nature of contaminants, production scale,
availability of utilities, and application requirements. A review of major available processes is presented in (Sun
et al., 2015). The most energy intensive and costly step is often the removal of the CO2, given its high
concentration in the biogas (30-50 mol%). This step is often called “upgrade into biomethane”. However, for
the OCM reactor, the CO2 contained in biogas can be used as a diluent gas, avoiding the need for introducing
another diluent such as nitrogen. Therefore, the raw biogas is only treated for minor contaminants and the
biogas upgrade is omitted.
A contaminant of major concern is hydrogen sulphide (H2S), which is a toxic, malodorous, and sour gas, which
causes equipment corrosion. Within the biogas industry, bio-scrubbers using Thiobacilli bacteria are widely used
to convert H2S into elemental sulphur and sulphate (Ramírez-Sáenz et al., 2009). This is an effective method
providing low operational costs and the use of harmless educts and products in the relevant production scales.
A disadvantage is the sensitivity of the microorganisms towards fluctuations in temperature and H2S
concentrations. Therefore, a constant efficient removal cannot be guaranteed (Krischan et al., 2012). An
approximate cost of 0.1-0.25 EUR/kg H2S is suggested for this method (Sun et al., 2015).
To achieve a higher H2S removal (< 1 ppmv), adsorption on metal oxides or impregnated activated carbon is
applied. This is an effective method for deep H2S removal, but if the H2S content in the feed gas is high, frequent
replacement of adsorption material leads to high operation cost. The cost of this method is estimated to be
around 3.85 €/kg H2S (Sun et al., 2015). Physical or chemical absorption are not considered herein, as they
would simultaneously remove the CO2, increasing the energy requirement unnecessarily.
Other contaminants of concern are siloxanes. These can cause erosion and deposit of silica compounds, being
a major concern for micro-turbine applications. Adsorption with activated carbon is a common method for their
removal. Concentrations below 0.87 ppm can be achieved at a cost of 81-113 €/kg (Sun et al., 2015).
The selection of appropriate biogas treatment and upgrade methods should consider operational and capital
investment costs as well as specific requirements for the gas application and should, therefore be done on a
case by case basis. Since the specific gas quality requirements for the OCM have not yet been investigated, it
is assumed that a deep removal of contaminants is required. Different biogas compositions from varied sources,
e.g. livestock farms or landfill gas, are then used and the cost range of 0 to 0.028 USD/Nm³ for the treated
biogas is estimated based on treatment costs collected and published by (Sun et al., 2015) for a clean and a
heavily contaminated stream.
Biogas
REACTION
Oxygen
COMPRESSION
H2O
H2O
CO2 REMOVAL DISTILLATION
Ethylene
BFW
HPS
CO2
CHP
Flue Gas
Heat & Power
Lights
Recycle (H2, CH4, CO)
Combustion Air
CO2
Ethane
590
2.2 Oxidative coupling of methane process
The OCM process is assessed based on a simulation model implemented in the software Aspen Plus v9
(Aspentech, 2017). The flowsheet is developed based on previous works of the research group with the natural
gas-based OCM process (Salerno et al., 2016), which have been experimentally validated in an OCM mini-
plant. Other process flowsheets considering alternative separation methods, such as pressure and temperature
swing adsorption, need to be considered in the future as they might bring advantages on the smaller production
scales. While heat integration is performed to some extent during the synthesis of the simulation flowsheet, a
more rigorous analysis needs to be performed once more flowsheet structures are available and consolidated.
Two adiabatic fixed bed reactors in series are applied. The reactors are modelled as kinetic plug flow reactors,
using kinetics from (Stansch et al., 1997). A dilution gas, commonly nitrogen, is applied in the reactor to deal
with the reaction heat release and to avoid the formation of explosive atmospheres between oxygen and
methane. The CO2 contained in the biogas is used in this application instead. Therefore, pure oxygen (95 mol%)
instead of air is fed to both reactors as the limiting reactant. The cost for the air separation unit is included in the
oxygen supply price. The achieved ethane and ethylene yield is 15 %, which is lower than the current state of
the art (>20 %) achieved experimentally with a more complex fixed bed membrane reactor (Godini et al., 2014).
The reactors’ outlet streams are used to pre-heat the feed and to generate high pressure steam.
The reactor outlet gas, containing H2, N2, CH4, C2H4, C2H6, CO, CO2, and H2O, is quenched, compressed to
11 bar, and cooled to condense most of the water. It is then fed to the CO2 removal section, wherein a hybrid
membrane and absorption-based process (Penteado et al., 2016) is used. The bulk of the CO2 content is
removed using a cascade of polyimide membranes, while the remaining part is removed via
absorption/desorption with a 30 wt% aqueous monoethanolamine solution. A CO2 rich stream is gained as a by-
product and its utilization is out of the scope of this contribution, but of significant importance given the amount
that is produced. The CO2 free gas is further compressed to 35 bar and fed to the distillation section.
The main process stream is pre-cooled in regeneration exchangers and separated in a series of two columns.
The demethanizer column produces a methane-rich stream at the top. This stream is partially recycled back to
the OCM reactor or used for cogeneration of heat and power. The bottom product contains ethylene and ethane,
which are separated in the C2 Splitter achieving above 99.99 mol% pure ethylene and an ethane stream that is
either recycled to the reaction section or sold as a by-product.
2.3 Cost estimations
Based on the simulation results, utilities’ consumption and cost rates are estimated using the default utilities
contained in the program’s databank. For this purpose, cooling water, electricity, medium and light pressure
steam, as well as refrigeration utilities are employed. The process exports electricity and heat and credit is taken
for those. The default costs for the utilities in the program, which reflect U.S. Gulf Coast values, are given a
range from 0.7 to 1.3 for the economic evaluation step. Similarly, the costs for the raw materials and products
are given a certain price range as shown in Table 1. For simplicity and lack of information on non-commercial
technologies, make-up of amine solution and cryogenic utility, and catalyst and membrane material replacement
are neglected at this stage. The equipment cost is estimated using the tool Aspen Process Economic Analyzer,
as well as a tool previously developed within our research group based on an industrial, pilot, and mini-plant
equipment data bank (Lühe, 2012). Other plant investment costs are estimated based on the equipment cost
via multiplication factors (Bejan et al., 1996), which are also given a lower and upper bound. The costs for
installation, piping, automation, electric, civil, service facilities, land/terrain, engineering & procurement,
construction site, contingency, maintenance are estimated this way, leading to the Fixed Capital Investment
(FCI). By adding the start-up cost and the working capital, the Total Capital Investment (TCI) is calculated.
3. Economic analysis
A year-by-year cash flow analysis is then performed for the base case Bio-OCM plant. An operating-life of 30 y
is assumed, which is typical for chemical plants (Bejan et al., 1996). The total capital investment is divided into
30 % for the first year (2017) and 70 % for the second. Plant operation and revenues start in January 2019. A
linear depreciation until zero salvage value during the entire project duration is assumed. The assumed ranges
for the main economic variables are shown in Table 1. The costs for raw material, operation & maintenance,
depreciation and return to investors escalate based on the inflation rate. Due to development of alternative
technologies and depletion of fossil resources, the price for the main product ethylene escalate based on the
nominal escalation rate, which is slightly higher than the inflation. The break even for the project is determined
based on the accumulated cash flow during the project years. The project’s net present value is then calculated
by Eq(1), where the NPV is the net present value, the FCI is the fixed capital investment, SU is the start-up
costs, WC is the working capital, ieff is the effective interest rate, z is the operation year, and NCFz is the net
cash flow on the zth year of operation. A positive NPV indicates that the project is economically feasible.
591
Table 1: Cost / price for utilities, raw materials, and products and main economic assumptions
Item Price Range Units Item Range
Electricity 0.05 0.09 USD / kWh Inflation rate (ri) 0.02 0.09
MP Steam 1.8x10-6 2.6x10-6 USD / kJ Real escalation rate (rR) 0.03 0.06
LP Steam 1.5x10-6 1.7x10-6 USD / kJ Nominal escalation rate (1+ri) ∙ (1+rR) - 1
Cooling Water 1.7x10-7 2.54x10-7 USD / kJ Effective interest rate (ieff) ri 0.15
Refrigeration 6.0x10-6 1.1x10-5 USD / kJ Total income tax 0.10 0.30
Treated biogas 0 0.028 USD / Nm³ Operation factor 0.85 0.98
Oxygen 95 % 0.050 0.074 USD / Nm³ Yearly operating hours 7,446 8,585
Ethane 46.74 96.06 USD / t Plant operating life 30 y
Ethylene 700 1,400 USD / t
Carbon dioxide 3 5 USD / t
𝑁𝑃𝑉 = 0.3 ∙ 𝐹𝐶𝐼 +
0.7𝐹𝐶𝐼 + 𝑆𝑈 + 𝑊𝐶
(1 + 𝑖𝑒𝑓𝑓)
1
+ ∑
𝑁𝐶𝐹𝑧
(1 + 𝑖𝑒𝑓𝑓)
1+𝑧
30
𝑧=1
(1)
A total of 2,000 simulations are then conducted with the variables taking random values within the given ranges.
For this purpose, the randomize function of Microsoft Office Excel 2016 is used. The net present value, total
annualized cost, and the breakeven are recorded. The results of this Monte Carlo simulation in terms of average
values and frequency distributions are presented in the results section.
4. Results and discussion
A first analysis is conducted based on the simulation results by only considering raw materials, products and
utility costs. Based on the given bounds, the worst scenario occurs at the maximum biogas, oxygen, cooling
water and refrigeration costs and the lowest possible ethylene, ethane, electricity, and steam prices. This results
in a plant cash flow of -8.07 USD/h. The best scenario results in a plant cash flow of 181.08 USD/h. Therefore,
the process can be profitable and the capital investment cost and other operation costs need to be estimated.
The total installed equipment cost is estimated at 2.1x106 USD. The most expensive equipment is the gas and
steam turbines at the cogeneration section. The use of gas engines could provide a less capital-intensive
solution at the expense of a lower efficiency. A cost estimation for a gas engine cogeneration system could not
be obtained at this stage and, therefore the prices for turbines available in Aspen Plus are used. Heat
exchangers represented the second most expensive equipment type, followed by membrane modules,
compressors, columns, and reactors. The use of CO2 as dilution gas also implied that the reaction and CO2
removal are the most capital-intensive steps, while the reaction and distillation were the most capital-intensive
ones while utilizing nitrogen dilution in the natural gas-based OCM (Salerno et al., 2016).
The results of the Monte Carlo simulation, namely the frequency distributions for the Net Present Value and
Breakeven for the base case, are shown in Figure 2. The Net Present Value is positive in 77 % of the cases
with an average value of 6.8x106 USD. This indicates that the process is likely to be feasible in a wide range of
different scenarios regarding technological, market, and economic aspects. The standard deviation is however
± 9.3x106 USD, which is to be expected given the uncertainties assumed in the simulations. The payback period
is relatively high with an average of 15 ± 4 y, since the capital expenses for the plant installation are relatively
high compared to the net cash flow. This could be overcome by an increased production scale, which could help
to dilute the installation costs.
Figure 2: Frequency distribution of the Net Present Value and the Break Even for the base case
0%
2%
4%
6%
8%
10%
12%
14%
16%
-12 -6 0 5 11 16 22 28 33 39 45
F
re
q
u
e
n
c
y
(
%
)
NPV (x106 USD)
0%
2%
4%
6%
8%
10%
12%
14%
8 10 12 15 17 19 21 23 26 28 30
F
re
q
u
e
n
c
y
(
%
)
Break Even (y)
592
Two further studies are then conducted. The first one considers the introduction of an additional Ethane
Dehydrogenation (EDH) reactor, which is placed downstream to the OCM reactor. It allows more ethane to be
converted into ethylene, thus increasing the revenue. This reactor is not added to the simulation flowsheet, but
a per pass ethane conversion of 60 % and a 60 % ethylene selectivity are assumed based on previous works
(Salerno et al., 2016). The total raw material and utility costs are increased by 10 % and the equipment cost is
assumed to be similar to the OCM reactors. The ethane recycle allows for an overall 100 % ethane conversion.
This configuration greatly improves the NPV to an average of 24.6 ± 16.8 x106 USD, with 99.95 % of positive
values, and the payback period is reduced to 10 ± 2 y.
Another study is conducted considering a higher production scale still including the EDH reactor. The scale of
biogas plants is likely to grow in the future, notably in developing countries, based on modernization of
agriculture, large-scale livestock farms, and urbanization. For instance, China is already experiencing a
reduction in the number of household biogas plants and an increase in medium to large scale plants with a net
increase in the country’s biogas output (Wang et al., 2016). A very large biogas plant being commissioned in
Tønder, Denmark is taken as another base point for this analysis. The plant treats liquid manure, household
waste, and energy crops from 120 farmers, producing approximately 7,800 Nm³/h of biogas (Gudmundsson,
2016). Equipment cost are scaled using a simple 6/10th rule, while raw materials and utility costs as well as
product revenue scale-up linearly. A further extrapolation is also performed for theoretical biogas plants
producing 20,000, 50,000, and 100,000 Nm³/h of biogas. The resulting increase in the NPV is presented in
Figure 3. The payback time is reduced to 8 ± 2 y at the largest scale.
Figure 3: Sensitivity analysis for the production scale
5. Conclusions
This study is a first assessment for the potential use of biogas as a renewable feedstock for green ethylene
production via Oxidative Coupling of Methane. Technical aspects are considered at a superficial level and
through many assumptions to allow for a first estimation of the economic potential of this process. While the
accuracy and significance of these results should not be overestimated, it can be concluded that this process is
economically attractive under different conditions. The introduction of another reactor for the dehydrogenation
of ethane can significantly increase ethylene production, leading to a higher revenue that increases the project’s
net present value and reduces the time for breakeven. Furthermore, it is also shown that the process can benefit
from larger production scales, which should be a major challenge considering that biogas is produced in rather
small scales when compared to natural gas or shale gas.
The proposed Bio-OCM process needs to be investigated in more detail for its further development.
Experimental investigations for the operation of a mini-plant OCM reactor with biogas-like feed compositions
are to be carried out. Kinetic models might need to be revised or fitted to the new operating conditions. The
effect of eventual biogas contaminants, e.g. catalyst poisoning, also needs to be considered in detail, also to
allow for the synthesis of a suitable biogas treatment section. The downstream separations also need to be
revised, especially taking into consideration particularities of this application such as the high inlet CO2
concentration and the reduced production scale. Energy and material integration aspects are also to be
considered in the process synthesis stage, allowing, for example, for an integration with the endothermic dry
reforming of methane to produce syngas. At this stage, the sustainability of the proposed process chain is also
to be assessed in more detail through carbon footprint calculations and Life Cycle Analysis and compared to
the equivalent fossil routes for ethylene production via naphtha or ethane cracking. From an economic
standpoint, it would be reasonable to compare the performance of the Bio-OCM process with those of a
standalone biogas CHP plant and with a biomethane producing plant.
0
100
200
300
400
500
600
700
800
0 20 40 60 80 100
N
P
V
(
x
1
0
6
U
S
D
)
Biogas Flow (x1,000 Nm³/h)
593
Acknowledgments
Alberto Penteado gratefully acknowledges the support from CAPES – Brazil (BEX 11946/13-0). Financial
support from the Cluster of Excellence Unifying Concepts in Catalysis by the German Research Foundation
(DFG EXC 314) is also gratefully acknowledged.
References
Albanez R., Chiaranda B. C., Ferreira R. G., França A. L. P., Honório C. D., Rodrigues J. A. D., Ratusznei S.
M., Zaiat M., 2016, Anaerobic Biological Treatment of Vinasse for Environmental Compliance and Methane
Production, Applied Biochemistry and Biotechnology, 178(1), 21-43.
Arndt S., Otremba T., Simon U., Yildiz M., Schubert H., Schomäcker R., 2012, Mn–Na2WO4/SiO2 as catalyst
for the oxidative coupling of methane. What is really known?, Applied Catalysis A: General, 425-426, 53-61.
Aspentech, 2017, Aspentech Products – Aspen Plus, , accessed:
30.05.2017.
Bejan A., Tsatsaronis G., Moran M., 1996, Thermal Design & Optimization, John Wiley & Sons Inc., New York
/ Chichester / Brisbaine / Toronto / Singapore, ISBN: 978-0-471-58467-4.
Chen X., Jiang J., Li K., Tian S., Yan F., 2017, Energy-efficient biogas reforming process to produce syngas:
The enhanced methane conversion by O2, Applied Energy, 185(Part 1), 687-697.
Esche E., Arellano-Garcia H., Biegler L.T., Wozny G., 2012, Two-Dimensional Modeling of a Packed-Bed
Membrane Reactor for the Oxidative Coupling of Methane, Chemical Engineering Transactions, 29, 1537-
1542.
Esche E., Müller D., Song S., Wozny G., 2015, Optimization during the process synthesis: enabling the oxidative
coupling of methane by minimizing the energy required for the carbon dioxide removal, Journal of Cleaner
Production, 91, 100-108.
Godini H.R., Xiao S., Kim M., Holst N., Jaso S., Görke O., Steinbach J., Wozny G., 2014, Experimental and
model-based analysis of membrane reactor performance for methane oxidative coupling: Effect of radial
heat and mass transfer, Journal of Industrial and Engineering Chemistry, 20, 1993-2002.
Gudmundsson E., 2016, Ramboll Biogas Tønder - the largest biogas plant in the Nordics,
, accessed: 21.02.17.
Keller G. E., Bhasin M. M., 1982, Synthesis of ethylene via oxidative coupling of methane: I. Determination of
active catalysts, Journal of Catalysis, 73(1), 9-19.
Krischan J., Makaruk A., Harasek M., 2012, Design and scale-up of an oxidative scrubbing process for the
selective removal of hydrogen sulfide from biogas, Journal of Hazardous Materials, 215-216, 49-56.
Lühe C., 2012, Modular cost estimation in support of system planning for bidding and early basic engineering
phase (in German), PhD Thesis, Technische Universität Berlin, Germany.
Penteado A., Esche E., Salerno D., Godini H.R., Wozny G., 2016, Design and Assessment of a Membrane and
Absorption Based Carbon Dioxide Removal Processs for Oxidative Coupling of Methane, Ind. Eng. Chem.
Res., 55 (27), 7473-7483.
Ramírez-Sáenz D., Zarate-Segura P.B., Guerrero-Barajas C., García-Peña E.I., 2009, H2S and volatile fatty
acids elimination by biofiltration: Clean-up process for biogas potential use, Journal of Hazardous Materials,
163(2-3), 1272-1281.
Rasi S., Veijanen A., Rintala J., 2007, Trace compounds of biogas from different biogas production plants,
Energy, 8, 1375-1380.
Salerno D., Godini H.R., Penteado A., Esche E., Wozny G., 2016, Techno-Economic Evaluation of an Oxidative
Coupling of Methane Process at Industrial Scale Production, Computer Aided Chemical Engineering, 38,
1785-1790.
Stansch Z., Mleczko L., Baerns M., 1997, Comprehensive Kinetics of Oxidative Coupling of Methane over the
La2O3/CaO Catalyst, Ind. Eng. Chem. Res., 36(7), 2568–2579.
Stünkel S., Illmer D., Drescher A., Schomäcker R., Wozny G., 2012, On the design, development and operation
of an energy efficient CO2 removal for the oxidative coupling of methane in a miniplant scale, Applied
Thermal Engineering, 43, 141-147.
Sun Q., Li H., Yan J., Liu L., Yu Z., Yu X., 2015, Selection of appropriate biogas upgrading technology - a review
of biogas cleaning, upgrading and utilisation, Renewable and Sustainable Energy Reviews, 51, 521-532.
Wang X., Lu X., Yang G., Feng Y., Ren G., Han X., 2016, Development process and probable future
transformations of rural biogas in China, Renewable and Sustainable Energy Reviews, 55, 703-712.
594