Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

Computers and Chemical Engineering 165 (2022) 107948

Contents lists available at ScienceDirect

Computers and Chemical Engineering


journal homepage: www.elsevier.com/locate/cace

Power-to-X: A review and perspective


Matthew J. Palys, Prodromos Daoutidis ∗
Department of Chemical Engineering and Materials Science, University of Minnesota, Twin Cities, Minneapolis, MN 55455, United States of America

ARTICLE INFO ABSTRACT

Keywords: In the past decade, producing chemicals from renewable energy for use as fuel has gained considerable interest.
Power-to-x Renewable hydrogen production (PtH) is the backbone of this power-to-x concept, while further conversion
Energy storage to methanol (PtM) or ammonia (PtA) serves to increase energy density. In this article, we review production
Systems engineering
and utilization technologies for PtH, PtM, and PtA in the context of the energy and transportation sectors.
Hydrogen
Specifically, each technology’s basic operating principals, state of development, energy efficiency, dynamic
Methanol
Ammonia
flexibility, and deployment outlook is discussed. We also review recent process systems engineering research
of PtH, PtM, and PtA. At the process level, this research largely aims to improve economics through optimal
synthesis and design of novel processes as well as coupled real-time optimization and control for dynamic
operation. At the facility or supply chain level, combined capacity planning and scheduling to optimally use
intermittent renewable resources is the major focus.

1. Introduction needs to be made. Despite considerable growth over the last decade,
wind and solar generation together meet roughly 10% of global elec-
Since the beginning of the 21st century, development of renewable tricity demand (International Energy Agency, 2022a) while EVs make
electricity generation technologies has been extensive, with a view up only 1% of the global civilian vehicle fleet (International Energy
on decoupling energy supply from fossil fuels. In the last 10 years,
Agency, 2021), with even smaller market penetrations for heavier
deployment of these technologies, primarily wind turbines and PV
transportation such as trucks, trains, or ships. The intermittent nature
arrays, has increased seven-fold (International Energy Agency, 2022a).
A renewables-dominant energy supply for power, heat, and cooling is of renewables is a major barrier to more widespread implementa-
critical in keeping global warming below 1.5 ◦ C to avoid the worst tion. Wind and solar generation outputs can vary on wide range of
impacts of anthropogenic climate change identified by the IPCC (In- timescales, from nearly instantaneous to seasonal. In a renewables-
tergovernmental Panel on Climate Change, 2018). The energy supply dominant energy mix, this generation must be balanced with demands
sector is the single largest source of global CO2 emissions, responsible across these different timescales using energy storage systems (ESS).
for 35% in 2018 (Intergovernmental Panel on Climate Change, 2018). Batteries are well-suited to short-term generation-demand balanc-
Furthermore, it has been the largest contributor to emissions growth
ing, specifically within a day, and make up the majority of globally
since 2010, largely due to coal combustion in developing countries (In-
deployed energy storage (Sivaram et al., 2018). However, battery ESS
tergovernmental Panel on Climate Change, 2018). Replacing coal with
renewables could ensure continued energy access and the associated have relatively low energy densities (kWh/kg) and thus high energy
quality of life without negative climate effects. Alone, decarbonizing capacity costs of at least $250/kWh, resulting in large and expensive
energy supply will not be sufficient for meeting the 1.5 ◦ C goal. Using systems for energy storage over longer durations (Guerra, 2021). Other
renewables to electrify transportation has widespread implications to common energy storage technologies include those classified as thermal
this end as this sector is responsible for 25% of global emissions (Inter- (e.g. molten salts) or mechanical (e.g. pumped hydro, compressed
governmental Panel on Climate Change, 2018). Electric vehicles (EVs) air) storage. Each of these technologies has their own advantages
are one way to facilitate this. Their global deployment has risen from
and drawbacks and generally offer energy capacity costs in the $100-
20,000 in 2010 to 7.2 million today (International Energy Agency,
250/kWh range (Guerra, 2021). Broadly, these costs are still prohibitive
2021).
The global momentum towards renewable energy supply and trans- for high capacity, long durtion storage applications (Sepulveda et al.,
portation electrification is undeniable, but significant progress still 2021). Technologies for decarbonizing transportation have largely been

∗ Corresponding author.
E-mail address: daout001@umn.edu (P. Daoutidis).

https://doi.org/10.1016/j.compchemeng.2022.107948
Received 2 March 2022; Received in revised form 18 June 2022; Accepted 29 July 2022
Available online 11 August 2022
0098-1354/© 2022 Elsevier Ltd. All rights reserved.
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

Table 1
Comparison of hydrogen, methanol, and ammonia energy properties (Dias et al., 2020).
Hydrogen Methanol Ammonia
Energy density (kWh/kg) 33.3 5.6 5.2
Boiling point at 1 bar (◦ C) −252.8 64.7 −33.3
Storage pressure at 25 ◦ C (bar) 300–700a 1 10
Volumetric energy density (kWh/L) 0.5-2.3a 4.6 4.3
a
Hydrogen can be stored as a high pressure gas or as a cryogenic liquid, resulting in a range of energy
densities. This is discussed in SubSection 2.2.

limited to batteries. However, their low energy density limits the full- energy. For instance, since solar energy from PV arrays is not produced
charge driving distance (range) in battery electric vehicles (BEVs) while overnight, a solar-based power-to-x project could accommodate this by
precluding long-haul trucking or shipping almost entirely. only operating chemical production while solar energy is available and
Producing fuels from renewable electricity which can later be used oversizing the involved chemical processes to meet production targets
for stationary energy production or transportation is one approach (e.g., installing double the capacity and operating half of the time) or
to overcoming the limitations of electrical energy storage. This idea by using batteries or other direct electrical energy storage which are
was termed ‘‘power-to-x’’ in Germany in the early 2010s (Sterner and charged during the day and discharged overnight to allow for steady-
Specht, 2021). Practically, power-to-hydrogen (PtH) is the backbone
state chemical production. These design considerations inevitably lead
of the power-to-x concept (Siemens, 2020). Hydrogen can be produced
to higher capital intensities compared to traditional chemical manu-
from renewable energy via electrolysis with water as the only feedstock,
facturing which operates with continuously available feedstock at high
making its deployment quite geographically flexible. On a weight basis,
capacity factors to minimize capital intensity ($/product).
hydrogen is very energy dense (see Table 1). Hydrogen can be used
directly as fuel for stationary energy generation or transportation, pro- Thus, efforts to reduce individual process costs and to optimize
ducing only water as byproduct when used as fuel in well-designed fuel the sizing of each process in power-to-x systems are both of utmost
cell or combustion systems. As a result, hydrogen has received extensive importance. The viability of power-to-x projects can also be affected
attention from academic, corporate, and governmental perspectives at by the energy efficiency of each involved process, but improving this
a global scale in the context of energy and transportation (Solomon and efficiency can often be in conflict with reducing capital cost. At the
Banerjee, 2006; Bockris, 2013). Hydrogen can also be further processed systems level, sizing each involved process is non-trivial as is illustrated
to other fuels to improve storage or utilization characteristics. Power- even in the simple example in the previous paragraph. The lowest
to-methanol (PtM) and power-to-ammonia (PtA) are two promising cost system design will depend on a number of factors, including the
examples of such processing routes which are largely motivated by location and scale of the project, the types of renewable energy to be
improving volumetric energy densities (see Table 1). PtM involves used, and the costs of each involved processes. Operational aspects of
combining electrolytic hydrogen with carbon dioxide (CO2 ). This CO2 power-to-x must also be carefully considered. Some existing processes
can be obtained from a variety of sources and electrically-powered pro- which could theoretically be used for power-to-x are not well-suited
cesses, either as an industrial byproduct or captured directly from air.
to frequent and/or large changes in their production rates for safety,
PtM is thus well-suited for coupling with carbon capture installations,
efficiency, or performance degradation reasons. Such processes must
though subsequent use of this methanol to produce energy would emit
thus be re-designed or retrofitted with dynamic flexibility as a key
CO2 as a byproduct. Considerable past exploration has been devoted to
consideration or otherwise must be deployed in systems which can ac-
methanol as a fuel for both fuel cell and combustion systems (Hagen,
1977; Bromberg and Chang, 2010). PtA involves combining electrolytic commodate these limitations. Process systems engineering is uniquely
hydrogen with nitrogen electrically separated from air. Ammonia does suited to systematically address these challenges across multiple tempo-
not emit carbon when used to produce energy, instead releasing water ral and spatial scales with a view on economic competitiveness. Thus, in
along with nitrogen-containing molecules, some of which must be addition to discussing relevant technologies, we review the literature on
abated. Further, PtA relies only on water and air as feedstock, giving optimal design, scheduling, and control of PtH, PtM, and PtA processes,
it similar geographical versatility as PtH. For these reasons, PtA in systems, and networks for energy and transportation.
energy applications has recently gained traction (Giddey et al., 2017; The rest of this article is structured as follows. Sections 2 through 4
Valera-Medina et al., 2018). respectively describe the enabling technologies and recent systems
In this paper, we review PtH, PtM, and PtA in the context of energy engineering research for PtH, PtM, and PtA. Section 5 describes recent
supply and transportation. Specifically, we describe technologies for investigations into integrated power-to-x systems which combine en-
production of these chemicals using renewables and for their subse- ergy vectors and/or use cases to exploit synergies and reduce costs.
quent use in energy supply or transportation applications. Hydrogen Section 6 summarizes the state of development of each power-to-x
storage methods are also specifically reviewed. This discussion focuses
pathway and provides important directions for future research.
on each technology’s state of development as well as their relative ad-
vantages and drawbacks in power-to-x systems. Detailed reviews have
previously been conducted for specific technologies in PtH (Schmidt 2. Power to hydrogen
et al., 2017; Acar et al., 2021; Yue et al., 2021), PtM (Dieterich et al.,
2020; Mbatha et al., 2021), and PtA (Valera-Medina et al., 2021; Afif
PtH relies on the use of electrical energy to power electrolysis, sep-
et al., 2016; Yapicioglu and Dincer, 2019) systems. The goal of this
arating water into its constituent hydrogen and oxygen molecules. This
paper is to synthesize information about power-to-x production and
hydrogen can be stored as a high pressure gas, a cryogenic liquid, or in
utilization processes and analyze them in a comparative context.
Economic considerations are a main barrier to competitive power- metal hydrides and then fed to fuel cells or combustion technologies for
to-x adoption, particularly the high associated capital investment. energy generation. Further, PtH serves as the precursor to electrifying
These elevated costs arise because power-to-x systems using intermit- the production of chemicals such as methanol and ammonia as dis-
tent wind and solar resources cannot operate at steady state. These cussed respectively in Sections 3 and 4. Fig. 1 outlines the PtH-relevant
systems instead require time-varying operation of at least some subset technologies described in this section and the interconnections between
of oversized process units, additional energy storage units, or a combi- them. Detailed descriptions of these hydrogen production, storage, and
nation of both to meet production targets with intermittently available utilization technologies are provided in Sections 2.1 through 2.3.

2
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

Fig. 1. Conceptual technology outline for electrified hydrogen production and utilization in energy and transportation.

Table 2 moderate share of electrolysis deployment, with some predictions that


Comparison of electrolysis technology investment costs.
PEMEC will be the dominant technology by 2030 (Schmidt et al., 2017).
AEC PEMEC SOEC SOEC is at a much earlier stage of development and has mostly been
Present-day ($/kW) 850 1,000 3,500 limited to demonstration efforts (Zhang et al., 2015). These systems
Projected 2030 ($/kW) 500 600 1,900
operate at temperatures of 650–1000 ◦ C and can achieve efficiencies
as high as 100% of hydrogen LHV (Laguna-Bercero, 2012). SOEC
uses ceramics such as Yttria stabilized Zirconia (YSZ) as electrolyte
2.1. Hydrogen production technology: Electrolysis through which O2− ions flow from a Ni on YSZ cathode to Perovskite-
type lanthanum strontium (LSM) on YSZ anode. The capital cost of
Electrolysis uses an electrical current to separate water: these systems is currently quite high, but reductions are expected with
their continued development (Kilner et al., 2012). However, high tem-
2H2 O → 2H2 + O2 𝛥𝐻 = 571.66 kJ∕mol (1) perature operation presents a number of challenges. Primarily, SOEC
Electrolysis cells consist of an anode and a cathode separated by an materials degrade faster than in AEC or PEMEC, leading to lifetimes less
electrolyte. The oxidation and reduction half reactions respectively than 10,000 h. Furthermore, the dynamic flexibility of SOEC is similar
occurring at the anode and cathode depend on the specific electrol- to or worse than AEC, with operating lower bounds above 30% and
ysis technology. Three main classes of electrolysis are used for PtH: startup times of up to an hour (Hauch et al., 2020).
alkaline (AEC), proton electrolyte membrane (PEMEC), and solid oxide
(SOEC) (Schmidt et al., 2017). Table 2 provides estimated present-day 2.2. Hydrogen storage technology
and 2030 capital costs for each electrolysis technology. At present, AEC
is the most widely deployed (International Energy Agency, 2022b). Hydrogen storage is of key importance for the viability of PtH.
Both AEC and PEMEC operate at 50–80 ◦ C and can achieve efficiencies At ambient conditions, hydrogen is the least dense of all elements at
as high as 85% based on the lower heating value (LHV) of hydrogen. 0.89 kg H2 /m3 ; the resultant low energy density is untenable. Storage
In AEC, OH− anions pass through an aqueous alkaline hydroxide elec- technologies primarily serve to increase this volumetric energy density.
trolyte, most commonly potassium hydroxide (KOH). Nickel or nickel The key technological considerations are the cost of the storage system
alloys are used in both the cathode and anode (Zeng and Zhang, and the energy required to transform hydrogen into its stored form.
2010). AEC has the lowest associated present-day capital cost due In transportation applications, the weight of the storage system is
to its high technical maturity, avoidance of expensive noble metals, also crucial. This section focuses on three types of hydrogen storage
operating lifetimes of up to 90,000 h (Bertuccioli et al., 2014). The technologies: compressed gas, cryogenic liquid, and metal hydrides.
main drawback of AEC is limited dynamic flexibility, with an operating The key characteristics of each are summarized in Table 3. Hydrogen
lower bound of 10%–40% of installed capacity and more detrimentally, can also be ‘‘stored’’ in the form of chemical bonds — its use as a
startup times on the order of 60 min (Lehner et al., 2014). reactant for methanol and ammonia synthesis is respectively discussed
PEMEC achieves similar efficiencies as AEC while offering consid- in the subsequent Sections 3 and 4.
erably better dynamic response characteristics, with a lower operating Compressed gas is the most developed form of hydrogen storage.
bound of 0%–10%, millisecond response times, and startup times less Practically, this entails ambient temperature hydrogen compressed to at
than 20 min (Carmo et al., 2013). PEMEC can also operate at pressures least 200 bar and up to 1000 bar, with many storage vessels for energy
up to 200 bar, making it well suited for gaseous hydrogen storage or and transportation applications rated at 700 bar (Barthélémy et al.,
other high pressure applications. PEMEC uses a solid polymer mem- 2017). Compressing hydrogen to 700 bar consumes approximately 10%
brane electrolyte such as Nafion through which protons (H+ ) flow of the contained energy (based on LHV). State-of-the-art high pressure
from anode to cathode. PEMEC anodes use platinum (Pt) or platinum– vessels rely on two separate components. They generally use (i) a
palladium (Pt–Pd) while cathodes employ ruthenium oxide or iridium composite shell of carbon or glass fiber embedded in resin to provide
oxide. The use of these rare earth metals leads the capital cost of mechanical integrity without excessive thickness or weight and (ii) a
PEMEC to be approximately double that of AEC. At present, PEMEC metal or polymer liner in contact with the gas to act as a hydrogen
has shorter operating lifetimes than AEC, specifically in the range of permeation barrier (Tzimas et al., 2021). Compressed hydrogen storage
20,000–60,000 h. However, PEMEC’s flexibility and resultant ability to investment costs are $10–15/kWh based on LHV (James and Houchins,
better couple with intermittent renewables has driven industrial efforts 2016). State-of-the-art hydrogen-fueled passenger vehicles store 5–7 kg
to reduce cost and increase lifetime. At present, PEMEC makes up a H2 in a full tank at 700 bar to give a maximum driving distance (range)

3
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

Table 3
Comparison of hydrogen storage technologies.
Compressed Cryogenic Metal hydride
Pressure (bar) 200–1,000 4 20 (Releasea : 1–5)
Temperature (◦ C) 25 −253 25 (Releasea : 200–300)
Energy density (kWh/L) 0.5–1.8 2.3 1.4–1.8
Energy lossb (%) 10 35–40 15
Storage cost ($/kWh) 10–20 5 1
a
In a storage context, energy loss refers to the additional energy required on a percentage basis to store a
given quantity of hydrogen.
b
Metal hydride storage materials require decreased pressure and increased temperature to release the stored
hydrogen.

of 400–600 km (Lipman et al., 2018). Salt caverns can be used for Table 4
Comparison of fuel cell investment costs.
stationary storage at larger scales. Salt is inert with respect to hydrogen
AFC PEMFC SOFC PAFC
and thus limits its diffusion from storage. Salt cavern storage generally
occurs at pressures up to 200 bar, thus requiring less energy for hydro- Present-day ($/kW) 700 1,600 2,200 2,900
Projected 2030 ($/kW) 400 950 1,200 2,500
gen compression (Ozarslan, 2012). This storage method becomes more
economical than vessel-based storage if more than 20 ton H2 are to be
stored and the specific investment is roughly an order of magnitude
lower ($1–1.5/kWh) over 100 ton H2 storage volume (Ahluwalia et al., 2.3. Hydrogen utilization technology
2010). The main drawback of salt cavern storage is their geographical
limitation in that these caverns may not exist in desirable areas for Fuel cells are the most commonly considered technology for produc-
PtH. Further, hydrogen transportation to and from these caverns is ing energy from hydrogen. As with electrolysis discussed in Section 2.1,
nontrivial, generally making their use uneconomical for even moderate fuel cells consist of an anode and a cathode with an ion-permeable
distances between the caverns and production/utilization sites. electrolyte between them. Hydrogen is delivered to the anode, oxygen
Cryogenic liquid storage involves cooling to −253◦ with mild com- (generally air-contained) is delivered to the cathode, and electrons
pression to about 4 bar (Barthélémy et al., 2017). This allows for which are abstracted at either the anode of cathode (depending on
energy density twice that of compressed gas while avoiding safety the technology used) flow through an external circuit to generate
concerns associated with high pressure. However, storage at these con- electricity while producing water as a byproduct. Water formation is
ditions consumes approximately 35% LHV of the contained hydrogen. an exothermic reaction which can be exploited in combined heat and
More recent efforts have utilized temperatures of −238 ◦ C to −163 ◦ C power (CHP) applications. Conceptually, fuel cell reactions are the
and pressures of 50–700 bar to achieved so-called ‘‘cyro-compressed’’ reverse of those taking place in electrolysis. As such, many of the
storage (Yanxing et al., 2019). This can reduce storage energy con- major types of hydrogen fuel cells fall under the same classification as
sumption to roughly 25% of LHV but at additional compression cost. the electrolysis technologies discussed in Section 2.1, namely alkaline
Ultimately, liquid or cyro-compressed hydrogen storage is not well (AFC), proton electrolyte membrane (PEMFC), and solid oxide (SOFC).
suited to stationary energy applications. Few hydrogen-fueled vehicles The reader is thus directed to that Subsection for detailed discussion
employ cryogenic storage either due to the requirement of advanced of the anode, cathode, and electrolyte materials along with the elec-
onboard cooling/insulation systems which add additional cost. Overall, trochemistry basics. The discussion here focuses on operational aspects
the most likely applications for cryogenic storage are the ones in which of these fuel cell technologies, especially in comparison to each other.
high volumetric densities are of utmost importance, even at the expense Phosphoric acid fuel cells (PAFC) are also introduced and discussed in
of energy losses. this section. Table 4 provides estimated present-day and 2030 capital
Reversible storage in metal hydrides is based on the exothermic costs for each hydrogen fuel cell technology discussed in this section.
hydrogen dissociation onto a metal surface and subsequent absorption AFC operates at 70–100 ◦ C and can achieve state-of-the-art electri-
into the metal (Barthélémy et al., 2017). Then, the absorbed hydro- cal efficiency around 60% and CHP efficiencies of approximately 80%.
gen is endothermically released by increasing the temperature. The As with electrolysis, alkaline technologies require the least investment
suitability of metals for hydride storage depends largely on thermody- due to technical maturity and low cost materials, but suffer from
namics, both in terms of the required absorption-release temperature limited dynamic operating range and slower startup. PEMFC achieves
swing, which dictates the energy penalty for storage, and the maximum similar electrical efficiencies of 50%–60% but CHP efficiencies up to
amount of hydrogen which can be stored, which determines the total 90%. These systems operate either in a 80–100 ◦ C low-temperature
weight of the storage system. Magnesium hydride (MgH2 ) has been regime or a 200 ◦ C high-temperature regime (Yue et al., 2021). Higher
identified as a promising storage candidate. Mg can store more than operating temperatures can increase CHP efficiency and simplify the
7% hydrogen by weight, about three times the amount stored in management of water product which will vaporize over 100 ◦ C. How-
typical metal hydrides (Motyka, 2015). This storage can be achieved ever, this vaporization also causes membrane dehydration, increasing
at pressures of below 20 bar and is inexpensive from a raw materials degradation rates and decreasing lifetime (Mekhilef et al., 2012). Low
perspective, resulting in investment costs below $1/kWh (Eigen et al., temperature PEMFCs are also better suited to dynamic operation owing
2007). Micron-scale Mg powders doped with approximately 1 mol% to faster startup. The main drawback of PEMFC is the expensive materi-
catalyst can enable charge and discharge on the order of minutes (Leng als used in the anode and cathode. Both AFC and PEMFC are best suited
et al., 2006), and cyclic stability has been demonstrated for over 2000 to small-scale stationary applications. Their maximum single module
cycles (Luo et al., 2019). The main drawback of MgH2 storage is the outputs are approximately 100 kW and 250 kW respectively (Mekhilef
heat required to achieve release temperatures of 200–300 ◦ C (Sakin- et al., 2012). If their desired total output is higher, ‘‘numbering up’’
tuna et al., 2007). Onboard vehicle use of metal hydrides is unlikely due to increase total capacity does allow for some economies of volume
to these high temperatures as well as high total storage system weight cost savings on balance of plant components. PEMFC are the most
and the resulting detriment to fuel economy. However, their ability to commonly considered technology for fuel cell vehicles due to their
store hydrogen at lower pressures makes their use at fueling stations availability in outputs below 1 kW, low temperatures, lightweight
intriguing from a safety perspective. components, and dynamic operating capabilities (Wang et al., 2005).

4
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

PAFCs employ a phosphoric acid electrolyte held inside a porous ICE generators are less efficient than turbines and have higher per
matrix through which protons (H+ ) travel, with both anode and cath- unit capital costs. Gasoline-fueled ICE is the ubiquitous passenger ve-
ode using Pt coated on carbon microfibers (Kirubakaran et al., 2009). hicle engine concept, so hypothetically these could be retrofitted or
PAFC operate between 180–200 ◦ C, with electrical and CHP efficiencies replaced with hydrogen-fed engines with the appropriate modifications
of 40%–45% and 80% respectively. PAFCs have been mostly considered as discussed above. Hydrogen-fed ICE or ICE generators are about 80%–
for energy generation, with maximum single module power output in 85% as efficient as gasoline ICE or diesel generators (White et al.,
the range of 400–1000 kW; economies of scale enable lower per kW 2006), but these power losses can be mitigated by using higher engine
capital investment than PEMFC (Mekhilef et al., 2012). However, the compression ratios (Akal et al., 2020). Hydrogen ICE have considerably
use of expensive Pt in the anode and cathode still results in higher lower efficiency than fuel cells, resulting in higher fuel consumption
capital investment than for AFC. Dynamic flexibility is also major draw- and lower range in a vehicle context (Candelaresi et al., 2021).
back of PAFC, with startup times as long as 4 h because considerable
preheating is required before operation (Acar et al., 2021). Nonetheless, 2.4. Control and real-time operation
PAFCs are commercially available, though interest in them has waned
with the continued development of AFC and PEMFC. A major focus of recent systems engineering research in PtH systems
SOFCs involve considerably higher temperatures than the previously is control and real-time operation of hydrogen production, storage,
discussed fuel cell systems, operating in the range of 600–1000 ◦ C. Elec- and (potentially) utilization processes in the face of intermittent re-
trical and CHP efficiencies are 35%–45% and up to 90% respectively. newables. The operational decision-making for these systems must
High temperature operation leads to a number of distinct advantages account for the dynamics of each constituent technology as they inter-
and drawbacks. High temperature waste heat makes these fuel cell act with intermittent renewable availability. Torreglosa et al. (2014)
technologies especially attractive in CHP applications. Further, SOFCs proposed a two-layer hierarchical control strategy for a standalone
can operate reversibly in both electrolysis and fuel cell modes (Mo- wind-PV power system which uses both batteries and hydrogen for
gensen et al., 2019); this could enable further cost savings in stationary energy storage. In their formulation, the primary control layer deter-
energy applications by reducing the number of required process units. mines the cost optimal storage charge and discharge states based on
However, high temperatures accelerate degradation of cell compo- renewable generation, power demand, and the current states of each
nents which worsens performance and/or shortens lifetime (Acar et al., storage technology relative to their operating limits. The dispatch cost
2021). High temperature operation also limits dynamic flexibility, as it calculation accounts for the effects of charge/discharge cycling and
results in startup and shutdown times on the order of hours (Oryshchyn operating hours on storage technology lifetime. The secondary layer
et al., 2018). SOFC are available in large modules, with maximum sizes uses cascade control with individual PI controllers for electrolysis,
of 1000 kW (Mekhilef et al., 2012). This makes them well suited to fuel cells, and batteries. Case studies for 25 years of operation using
large scale energy supply applications or perhaps in long haul trucking one hour sampling for primary control showed both prolonged battery
or marine shipping settings where startup times, system size/weight, and fuel cell lifetimes and effective setpoint tracking and disturbance
and high temperatures are less detrimental than in passenger vehicles. rejection. Garcia et al. (2013) proposed a parallel control strategy for
Hydrogen can also be combusted for energy supply or transporta- a hybrid battery-hydrogen system. A utilization cost optimization layer
tion applications. Conventional combustion systems must be modified was used to determine battery or hydrogen storage dispatch for charg-
for use with hydrogen due to its higher flame speed and resulting ing or discharging while considering degradation of each component.
higher combustion temperature than conventional fuels (Ilbas et al., The cost layer did not explicitly account for battery state-of-charge
2006). These high temperatures increase formation of nitrous oxide (SOC) and hydrogen storage inventory. This was accomplished through
from nitrogen in the air used for combustion. Mitigating these NOx fuzzy logic control which overrode the utilization cost layer to ensure
emissions is essential for local air quality and because on a mass basis allowable storage operation. The fuzzy logic controller used the battery
NOx is at least 250 times more potent than CO2 as a greenhouse SOC, hydrogen inventory, and net power (generation–demand) as input
gas (U.S. Environmental Protection Agency, 2022). Design alterations with three, three, and five membership functions respectively. A set
such as the use of advanced hydrogen-mixing devices (Cappelletti and of rules for each permutation of these membership functions then
Martelli, 2017) or sequential combustion chambers with intermediate determined which control mode was active and then implemented
temperature control (Bothien et al., 2019) have been investigated to charge/discharge if fuzzy logic control was chosen. Compared with a
this end. Higher temperatures coupled with the increased amount of binary fully rule-based control strategy via a 25 year hourly resolution
steam in the combustion products can also result in increased corrosion simulation, the proposed fuzzy logic approach enabled close to 30%
and degradation of system components, though this can be mitigated reduction in utilization cost and enables zero renewable curtailment.
through material changes to system components (Stefan et al., 2021). Trifkovic et al. (2013) proposed a multilevel control architecture
Turbines are the state-of-the-art for large scale power generation with for hydrogen ESS consisting of a supervisory layer which determines
fossil fuels. These generate power by converting kinetic energy from binary operation modes and set-points for electrolysis/hydrogen com-
a moving gas, either steam raised in a boiler using the fuel’s heat pression and fuel cells and model predictive control (MPC) for each unit
of combustion or the combustion product gases themselves. Turbines based on linearized state–space models. Two MPCs were designed for
using pure hydrogen as fuel have demonstrated electrical efficiencies each unit, specifically with linearization at 10% and 90% rated power
of 35% (Cappelletti and Martelli, 2017; Goldmeer, 2019), comparable operating points, to account for their highly nonlinear behavior. This
that of simple combustion turbine systems. Combined cycle gas turbines allowed for the mitigation of controller-model mismatch by selecting
(CCGT) achieve electrical efficiencies as high as 60% by using waste the MPC closer to the setpoint from supervisory control. Case studies
heat from a direct combustion turbine to drive a secondary steam- performed for 1 kW order of magnitude residential demands showed
powered turbine. Major manufacturers have set a 2030 target to deliver that this dual MPC strategy significantly improved electrolysis and fuel
100% hydrogen-fed CCGTs, with investment costs projected to be 10% cell efficiency while also reducing electrolysis startup and shutdown
higher than conventionally fueled systems (European Turbine Network, cycling. Trifkovic et al. (2014) then used this multilevel control
2019; Bothien et al., 2020). formulation as the basis for proactive energy management strategy
Internal combustion engine (ICE) generators can also be used for based on real time optimization (RTO). The RTO level determined
energy supply as well as transportation. In a stationary context, they are unit commitments and setpoints to maximize renewable utilization and
generally used as ‘‘peaking’’ plants in grid operations during periods of hydrogen production while minimizing electrolysis startups over a daily
high demand or for emergency backup or remote power applications. receding horizon with hourly resolution. Forecasts for power demand,
ICE generators startup times are much less than turbines. However, wind speed, solar irradiance as well as current unit states were used

5
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

as RTO input data. The optimal setpoints were then achieved using the signals from the different markets; the value of energy became higher
dual MPC formulation proposed in their previous work (Trifkovic et al., at later stages, so corrective action through the cascade improved
2013). A notable feature of the 2014 work of Trifkovic et al. was the economics. They also compared the MIQP and MILP formulations and
inclusion of a battery and the ability to adjust its operation at the unit found similar grid economics but found that accounting for degradation
control level to resolve the difference in forecasted and realized de- resulted in less battery and hydrogen storage cycling.
mand and renewable availability. This proactive strategy was simulated Cau et al. (2014) used an E-MPC framework to manage uncertainty
for a year of energy system operation. The optimal control strategy in renewable generation and demand forecasts. They used scenario
favored consecutive hours of electrolysis and fuel cell operation to mini- trees with multi-hour to describe the evolution of this uncertainty
mize their cycling, charging or discharging the batteries to achieve this. over a day. The E-MPC was formulated as a MILP, using piecewise
Torreglosa et al. (2015) used MPC to improve upon their previous work linear functions to describe the efficiency of batteries, electrolysis, and
by replacing the PI controllers (Torreglosa et al., 2014). They employed fuel cells. They performed case studies for a representative week from
a centralized (linear/nonlinear) MPC formulation which determined each season, using a 24 h receding horizon with hourly resolution.
power charge and discharge for battery and hydrogen storage options For each representative week, a tree with 15 scenarios was obtained
based on net power availability. This strategy achieved 15% higher sys- using scenario reduction technique. Compared to a binary rule-based
tem efficiencies compared to the use of individual PI controllers. Nair formulation, E-MPC resulted in cost savings through more efficient
and Costa-Castelló (2020) also used MPC for control islanded energy hydrogen utilization. However, the time required to solve the E-MPC
storage system which included batteries, a reversible hydrogen-fed fuel approached an hour in many instances; this temporal proximity to
cell, and a dispatchable fossil fueled generator. They formulated the the decision resolution could cause problems with implementation.
MPC problem as a mixed integer quadratic programming (MIQP) to Cano et al. (2015) used fuzzy logic control for power management
penalize battery and fuel cell cycling. They used 5 min control intervals under uncertainty in a wind-PV-battery-hydrogen ESS. They described
in a receding horizon and performed case studies for horizon lengths net load (renewable generation–demand) as a stochastic process and
ranging from 3 h to 24 h. Prediction horizons longer than 6 h only used the current value, forecasted difference in successive periods,
marginally improved control performance while drastically increasing and forecasted variance in the next period along with the current
computational time. Their MIQP MPC formulation outperformed a battery state-of-charge as input variables to the fuzzy controller. The
heuristic power management scheme based on fuzzy logic, providing controller output was the electrolysis power consumption or fuel cell
less battery degradation due to less time at SOC above 80%, smoother power generation. Triangular and trapezoidal membership functions
reversible fuel cell setpoint variation, less curtailed PV power, and less were used for the input–output mapping. They applied this framework
dispatchable fossil generation to case studies for the same hourly resolution representative weeks
Recent research has also used economic model predictive control (E- as Cau et al. (2014), and found their approach resulted in fewer
MPC) for in the context of PtH and hydrogen ESS. Pereira et al. (2015) electrolysis startups and less fuel cell degradation with much faster
used such a formulation for a PV-battery-hydrogen (in metal hydride)
computation.
system for diurnal energy storage that could exchange energy with the
neighboring utility. The economic objective was to maximize the profit
2.5. Capacity planning
from this exchange and included the cost of deviating from day-ahead
energy bids (the quantity and duration of which were parameters in
their work) as well as degradation costs for ESS components. They There has been extensive research on capacity planning of PtH sys-
used a linear state–space model for battery and hydrogen storage tems which both produce and use hydrogen for energy storage. In this
dynamics based on the average step response over 300 simulations of a review, we classify capacity planning for power-to-x systems as prob-
nonlinear process model they developed previously. The initial state lems in which multiple process units are sized simultaneously. These
of the system was left as a free variable, and they instead imposed problems often incorporate unit scheduling decisions at the design stage
periodic constraints. They simulated E-MPC implementation on the as to allow unit sizing to explicitly account for temporal variability
original nonlinear model for five days of operation, using a 30 min in renewable generation. Much of the earlier PtH capacity planning
sampling time over a 24 h receding horizon. Each set of optimal work focused on optimization of standalone (most generally) wind-
control actions were computed in less than 4 min and convergence to PV-electrolysis-storage-fuel cells systems with case studies for single
optimal periodic trajectories was observed even in the face of changing locations (Garcia and Weisser, 2006; Dufo-López and Bernal-Agustín,
economic parameters, for example, if energy sale was no longer allowed 2008; Zhou et al., 2008; Kaviani et al., 2009; Shabani et al., 2010;
after 48 h of operation. Mendes et al. (2016) developed a two layer Erdinc and Uzunoglu, 2012; Jallouli and Krichen, 2012; Castaneda
control formulation for a wind-PV-battery-hydrogen (in metal hydride) et al., 2013; Vasallo et al., 2013; Lacko et al., 2014; Kalinci et al.,
system using both E-MPC and MPC. The E-MPC layer determined opti- 2015; Bakhtiari and Naghizadeh, 2018; Zhang et al., 2018). Most of
mal control actions every five minutes, while the MPC layer ensured these efforts used heuristic power management rules parameterized on
real-time setpoint tracking. The control problems were formulated input data to reduce the decision space (Garcia and Weisser, 2006;
as mixed integer linear programs (MILPs), with constraints limiting Zhou et al., 2008; Shabani et al., 2010; Erdinc and Uzunoglu, 2012;
electrolysis and battery switching included in the E-MPC model. Com- Castaneda et al., 2013). Other research has used meta-heuristic algo-
putational experiments were performed for ‘‘sunny’’, ‘‘cloudy’’, and rithms (e.g particle swarm optimization or simulated annealing (Dufo-
‘‘windy’’ representative days, with the optimal results in each requiring López and Bernal-Agustín, 2008; Kaviani et al., 2009; Bakhtiari and
two electrolysis-fuel cell cycles to mitigate degradation of these units. Naghizadeh, 2018; Zhang et al., 2018) or parallel simulation of many
Garcia-Torres and Bordons (2015) used a four level cascade E-MPC fixed design alternatives (Lacko et al., 2014; Vasallo et al., 2013; Kalinci
formulation for a grid-connected wind-PV-hydrogen-battery ESS to link et al., 2015) for computational tractability. The solution approaches
market operations over different timescales. The considered markets applied in these works enabled high scheduling resolution, for example,
(with relevant control resolutions/horizons) were daily (1 h/24 h), a yearlong horizon with hourly granularity. One notable recent work
interday (1 h/24→9 h), deviation management (1 h/24→5 h), and is that of Rullo et al. (2019) who used a genetic algorithm solu-
regulation service (10min/3 h), where the interday and deviation man- tion approach for a bilevel MINLP which combined capacity planning
agement market horizons shrank throughout the day. Each E-MPC layer with scheduling based on E-MPC over a 12 h receding horizon. This
was formulated as an MIQP to account for battery, electrolysis, and fuel approach allowed for explicit accounting of process dynamics during
cell degradation costs. An implementation case study showed shifting scheduling and resulted in 10% reduction in levelized cost of energy
optimal battery SOC and hydrogen storage levels in response to price compared to heuristic power management rules (Rullo et al., 2019).

6
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

Research conducted since 2018 has used MILP optimization to onshore and offshore wind generation as well as installation and oper-
solve these capacity planning problems. Mallapragada et al. (2020) ation of electrolysis, high pressure and underground hydrogen storage,
performed a U.S. national survey of optimal PV-based hydrogen produc- hydrogen and direct electric boilers at both industrial and distribution
tion costs, determining the size and schedules of PV arrays, batteries, scales, and transmission and distribution infrastructure for power, heat,
electrolysis, and high pressure storage to meet continuous hydrogen and hydrogen. They performed a case study where great Britain was
production targets. They found that hydrogen production costs below divided into 16 characteristic zones and employed a hierarchical, non-
$2.5/kg H2 were best met through over-sized electrolysis which dy- uniform time aggregation approach with variable-length multi-hour
namically tracks solar generation, minimizing local battery capacity. periods per season to efficiently represent the hourly resolution yearly
These designs were most favorable with a reduction in electrolysis input data. They found that hydrogen was the key to the profitability
capital cost as well as low cost salt cavern hydrogen storage. Marocco of this network in reducing the cost of seasonal heat storage.
et al. (2021) and Kotzur et al. (2018) applied MILP capacity planning
respectively to wind- and solar-based isolated hydrogen energy storage 3. Power to methanol
systems, determining optimal sizing of this renewable generation along
with batteries, hydrogen production, storage, and fuel cell technolo- PtM extends the PtH concept discussed in Section 2 by combining
gies. Both found renewable generation investment to be the main cost hydrogen with CO2 to synthesize methanol. Methanol is easily stored
drivers of these energy systems. Additionally, optimal hydrogen storage as liquid and can be used to produce energy via decomposition to
capacity was an order of magnitude higher than battery capacity, hydrogen, direct use in fuel cells, or combustion. Fig. 2 outlines the
indicating its suitability for longer storage durations. Mehrjerdi (2020) PtM-relevant technologies described in this section and the intercon-
and Gabrielli et al. (2019) incorporated solar generation and energy nections between them. The CO2 for PtM can be obtained from a variety
demand uncertainty in their capacity planning for PV-hydrogen sys- of sources. It can be almost pure as an industrial byproduct, captured
tems, with Gabrielli et al. focusing on CHP supply. Mehrjerdi employed from combustion flue gas, or directly captured from air. CO2 capture
a stochastic MILP formulation using 10 scenarios for the PV capac- technologies are not discussed in this paper; for more information, the
ity factor and power demand, using four season-representative days reader is referred to a recent review by Bui et al. (2018). Detailed
with hourly resolution for operating decisions (96 scheduling periods). descriptions of these methanol production and utilization technologies
Gabrielli et al. instead used robust optimization, sampling over 1000 are provided in Sections 3.1 and 3.2.
input data scenarios to find temporal profiles which gave the highest
annual electrical and thermal demands and the lowest solar availability. 3.1. Methanol production technology
They subsequently used these worst-case input data to optimize sizing
and dispatch, employing a scheduling decision partitioning approach Catalytic methanol synthesis was invented in the 1920s (Kastens
where binary operating decisions were made over 48 hourly resolution et al., 1948) and is the prevailing industrial approach today. It uses
representative design days and these were linked via constraints to a feedstock which contains hydrogen, CO, and CO2 (termed ‘‘syngas’’)
continuous operating decisions made for every hour in the selected to produce methanol in the following network of reactions (Klier et al.,
‘‘robust’’ representative years. 1982; Sahibzada et al., 1998):
Given the extensive state of capacity planning research for single-
location hydrogen systems, recent investigations have also incorpo- 2H2 + CO ⇌ CH3 OH 𝛥𝐻 = −90.6 kJ∕mol (2)
rated a spatial element to enable the design of PtH networks. Ogbe
et al. (2019) considered the augmentation of existing natural gas 3H2 + CO2 ⇌ CH3 OH + H2 O 𝛥𝐻 = −49.4 kJ∕mol (3)
supply chains with renewable hydrogen, optimizing the locations of
electrolysis-based hydrogen production and injection into the existing
2H + CO2 ⇌ H2 O + CO 𝛥𝐻 = 41.6 kJ∕mol (4)
pipeline under maximum hydrogen concentration constraints. They
used a tractable stochastic formulation with 16 scenarios to account The state-of-the-art process pressure and temperature ranges are
for uncertainty in the purity of hydrogen produced via electrolysis and 50–100 bar and 200–300 ◦ C respectively (Haid and Koss, 2001). The
the total gas demand. The author-identified limitation of this study was primary reactor design consideration is economically removing the heat
not considering temporal variability in energy availability for hydro- formed by this exothermic synthesis reaction to enable high equilibrium
gen production due to the limitations on tractability inherent to the conversion to methanol while avoiding formation of unwanted side
stochastic formulation. He et al. (2021) optimized design and operation products which are promoted by higher temperatures (Mbatha et al.,
of a network of hydrogen production with truck and pipeline trans- 2021). The dominant and technically well-established reactor configu-
portation to demand sites, with these trucks and pipeline potentially ration is a multi-tubular bed which operates close-to-isothermally via
acting as storage in addition to stationary high pressure vessels. They cooling through vaporizing water. Other configurations include a coil
performed a case study for a six zone network in the Northeast U.S. wound heat exchanger with catalyst on the shell side, radial flow reac-
using 20 representative weeks for operation. They found that the ability tors, or multi-stage adiabatic reactors with inter-bed cooling (Bozzano
to flexibly use hydrogen transportation as storage reduces stationary and Manenti, 2016). Slurry, trickle bed, and fluidized bed reactors
hydrogen storage capacity by approximately 60% and hydrogen supply have been proposed as means to mitigate mass and heat transport
costs by $0.2/kg H2 . Heras and Martín (2021) used a time-invariant limitations, but have not been implemented at appreciable production
hydrogen energy network model employing magnesium for low cost scales (Hansen and Hojlund, 2008). Single pass conversions tend to
hydrogen storage. Since the use of Mg for low cost hydrogen storage is be in the 35%–60% range; reactor effluent is cooled and separated
a novel concept, Heras and Martin first performed detailed nonlinear so that unreacted syngas can be recycled. After synthesis, methanol
optimization of the storage and discharge processes in their work. They must be separated and purified. The product of the above-described
then optimized the location and scale of these technologies along with synthesis approaches is generally termed ‘‘raw’’ or ‘‘crude’’ methanol.
wind and solar generation, hydrogen production via electrolysis, and Purification of this raw methanol generally first employs flash sep-
fuel cells for power generation across the country of Spain. Optimized aration at 5–10 bar to remove any dissolved light gases, and then
electricity costs were found to be as low as $0.1/kWh if byproduct heat one or more distillation columns depending on the intended methanol
and O2 from the hydrogen storage process can be sold. Samsatli and use. Energy or fuel grade methanol requires only a single column
Samsatli (2019) developed a hydrogen-based heating network MILP with methanol as the distillate and higher boiling point components as
model for Great Britain. This model optimized the addition of new the bottoms (Bertau et al., 2014). Methanol synthesis process depends

7
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

Fig. 2. Conceptual technology outline for electrified methanol production and utilization in energy and transportation.

strongly on economies of scale, with its estimated capital investment to SOEC for hydrogen production. Practical operation of co-electrolysis
being (Demirhan et al., 2020): requires 5–10 mol% hydrogen co-feeding with water and CO2 in order
𝑥 to mitigate cathode oxidation (Cinti et al., 2016). This hydrogen can
𝐶𝐴𝑃 𝐸𝑋(𝑀𝑀$) = 1.8( 𝑀𝑒𝑂𝐻 )0.65 (5) be obtained through recycle from methanol production or a separate
1000
smaller electrolysis unit.
where 𝑥𝑀𝑒𝑂𝐻 is the ton/y nameplate capacity of the CO hydrogenation
Considerable research has also been dedicated to developing cata-
methanol synthesis process. From an energy perspective, this process
lysts and processes uniquely suited to direct CO2 hydrogenation. Cata-
requires approximately 0.25–0.5 MWh/ton MeOH.
lysts for CO2 hydrogenation tend to consist of multiple components,
Process intensification has been used in an attempt to reduce this
for example promoting CuO-ZnO-Al2 O3 with zirconium, gallium, or
cost and/or energy consumption. Membrane (Gorbe et al., 2018) and
yttrium. These catalysts are tuned to maximize CO2 conversion and mit-
sorption-enhanced (Terreni et al., 2019) reactors have been inves-
igate water formation (Dang et al., 2019; Guil-López et al., 2019). The
tigated with a view on removing methanol from the reaction zone
emergence of these multi-component catalysts has enabled methanol
and thus mitigating equilibrium limitations on conversion. Reactive
synthesis from direct CO2 hydrogenation with higher selectivity and
distillation has also been investigated for methanol production as an
fewer impurities than methanol production from syngas (Pontzen et al.,
intensified solution to reduce investment (Ghosh and Seethamraju,
2011). The less exothermic nature of CO2 hydrogenation relative to
2019). methanol production from syngas potentially allows for simpler reactor
The commercial synthesis catalysts use copper oxide (CuO) or zinc and process designs from a heat transfer/integration perspective — this
oxide (ZnO) supported on alumina (Al2 O3 ). Equilibrium conversions is promising both from a capital cost perspective and for the dynamic
using syngas feed are considerably higher than for direct hydrogenation flexibility of CO2 hydrogenation.
of CO2 (see Eq. (3).1) (Pontzen et al., 2011). As such, two stage
processes for PtM have been developed for CO2 abatement where 3.2. Methanol utilization technology
CO2 is first converted at least partially to CO and then the methanol
synthesis proceeds as described in the previous paragraph. One such Methanol can be reformed to hydrogen for use in fuel cells or
approach is to use the water gas shift (WGS) reaction (see Eq. (3).1) combustion technologies as discussed in Section 2.3. It is worth not-
for this CO2 transformation. This entails the use of an auxiliary iron ing that PAFCs can effectively handle more impurities in the feed;
catalyst bed operating at around 10 bar and 600–700 ◦ C (Joo et al., historically, this was a main driver in their development for use with
1999), with these high temperatures favoring CO formation because hydrogen sourced from hydrocarbons, and in present day would allow
WGS is endothermic. The resultant water is separated and additional for less stringent pre-reforming for use with MtP compared to PEMFC
hydrogen can be added to achieve conventional syngas compositions and AFC. Hydrogen for use in more established fuel cell technologies
before methanol synthesis. Co-electrolysis is an alternative which is less can be produced from methanol via four routes: steam reforming
developed than the combination of electrolysis to produce hydrogen (MSR), partial oxidation (POM), autothermal reforming (ATRM), and
coupled with water gas shift, but offers promise with respect to energy decomposition (MD). The first three are analogous to routes used for
efficiency and overall process cost reduction by eliminating the WGS hydrogen production from methane. This paper provides a high level
reactor (Andika et al., 2018). In a co-electrolysis cell, water and CO2 are overview of hydrogen production from methanol. A review by Garcia
fed to the cathode where they are reduced to hydrogen and CO to give et al. (2021) provides a more detailed discussion. Producing hydrogen
the desired syngas composition. The resulting O2− ions pass through from methanol results in CO and/or CO2 byproducts. In a hydrogen fuel
a solid oxide electrolyte where they combine to O2 at the anode. Co- cell context, CO must be separated from reactor effluent to avoid cell
electrolysis systems operate at 600–1000 ◦ C and use a Ni/YSZ cathode, degradation. CO can also be used to obtain additional hydrogen via
a LSM/YSZ anode, and a YSZ electrolyte (Zheng et al., 2017), similar WGS, but this adds cost. SMR produces considerably lower CO than

8
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

the other three routes, which makes coupling with fuel cells attrac- 2016). Additionally, retrofitting existing combustion systems to be com-
tive (Iulianelli et al., 2014). It is also the most well-established and patible with methanol generally only requires small investment (A.D.I.
having the highest specific yields since both methanol and water pro- Analytics, 2017). Long-term methanol turbine operation has also been
duce hydrogen. However, this endothermic process requires heating for successfully demonstrated at the 50 MW scale (Chudnovsky et al.,
both isothermal operation and steam generation. This results in higher 2015).
energy consumption and slower process dynamics. POM and ATRM ICE technologies can also be made compatible with methanol (Ver-
offer more dynamic flexibility and are respectively heat producing helst et al., 2019). The engine components must be used to mitigate the
and thermo-neutral with the aforementioned drawback of additional corrosive nature of methanol relative to other non-alcohol fuels (Yuen
CO production. POM entails exothermic methanol combustion with et al., 2010). methanol’s lower LHV, higher heat of vaporization, and
well-below stoichiometric amounts of O2 ; this ratio can be adjusted higher autoignition temperature also makes engine startup more dif-
for dynamic operation. ATRM combines the MSR and POM reactions, ficult. To overcome this, electrical heating can be used to increase
adjusting feed water and oxygen flowrates to balance their endothermic fuel temperature (Bergström et al., 2007) or high pressure injection
and exothermic heats of reaction. MD requires no additional feed and ports can be used improve fuel atomization. For stationary power
can give complete conversion into hydrogen and CO. As with SMR, this generation, methanol ICE gensets have been commercialized (Shmuli,
is an endothermic reaction. It has thus been suggested to spatially and 2021) but fuel cells are more widely considered due to their higher
temporally integrate MSR or MD and hydrogen combustion systems to efficiency. Pure methanol-fueled vehicles have been developed with
provide this required heat Wilson (2009). higher demonstrated efficiency than gasoline engines (Nguyen et al.,
Low temperature direct methanol fuel cells (DMFC) are specifically 2018). Additionally, methanol offers a lower flammability risk than
designed for methanol-based energy generation (Demirci, 2007; Ku- gasoline. Methanol has also gained considerable interest as a marine
likovsky, 2008). DMFC systems operate at 30–100 ◦ C and, similar to fuel as it can enable fewer CO2 and sulfur emissions than the currently
PEMFC, use polymer membranes as a proton conducting electrolyte and used heavy fuel oils while also being easier and less costly to store and
operate at 30–100 ◦ C. These systems are fed with a liquid mixture of handle than liquefied natural gas (Svanberg et al., 2018). The promise
methanol and water rather than H2 and O2 . DMFCs can achieve electri- of methanol as a marine fuel has led to a number of demonstration
cal efficiencies 40% and CHP efficiencies of 80%, while offering better projects in recent years (Hobson and Marquez, 2018).
dynamic flexibility than other direct-fed fuel cells due to their lower
operating temperature. The investment cost and capacity availability of 3.3. Process design and operation
DMFC are similar to that of PEMFC (Müller and Stolten, 2011), without
considering the cost of methanol reforming needed for use with PEMFC. One focus of recent research relevant to PtM has been intensification
Thus, DMFC are superior for methanol fueling in small stationary or and optimal design of syngas-fed methanol synthesis. Arora et al.
portable energy applications as well as for passenger vehicles. However, (2018) proposed a sorption-enhanced methanol production process to
direct methanol fueling of vehicles would result in highly spatially counter equilibrium limitations. A dynamic model was developed for
dispersed CO2 emissions. Methanol can also be directly fed to SOFC due this process which is inherently transient due to the four-step periodic
to their high operating temperatures, which enable internal reforming operation of the reaction–sorption column. This model was used for
of methanol to H2 . Molten carbonate fuel cells (MCFC) also enable constrained grey-box multi-objective optimization for methanol syn-
internal reforming. These systems operate at 600–700 ◦ C and can thesis process retrofit, that is, augmenting catalyst with sorbent in
achieve electrical and CHP efficiencies up to 50% and 85% respectively. an existing reaction vessel. This enabled up to an 8% increase in
MCFC systems use an electrolyte of sodium, potassium and/or lithium methanol yield and decrease in raw material consumption at similar
molten carbonates suspended in a ceramic matrix. The electrolyte- production costs. Monjur and Hasan (2021) proposed a methanol
traversing anion is CO2− 3
(Sundén, 2019). MCFCs must in fact be used production process based on a membrane reactor which served to
with carbon-based fuels such as methanol because they require the mitigate equilibrium while also allowing for steady-state production.
presence of CO2 to this CO2− 3
ion. The anode and cathode both use The process model used detailed reaction kinetics and surrogate models
nickel, avoiding expensive precious metals. At present, MCFCs are more for reaction and membrane thermodynamics. The lowest cost process
established than SOFC and require roughly 40%–50% less investment, design was determined using an NLP formulation, with the optimally
but their respective capital costs are projected to reach parity in the designed reactor operating at lower temperature and pressure than
future (International Energy Agency, 2022b). However, MCFCs have the conventional synthesis process. Overall, the use of a membrane
shorter lifetime due to the corrosive nature of electrolyte (Acar et al., reactor was shown to reduce compression investment and electricity
2021). As discussed in Section 2.3, high temperature operation limits consumption by approximately 65% each.
dynamic flexibility which restricts deployment of SOFC and MCFC to There has also been recent research into processes inherently better
certain stationary or long haul transportation applications. Nonetheless, suited to PtM, specifically direct CO2 hydrogenation and co-electrolysis.
their large available module sizes above 1000 kW and, for MCFC, Borisut and Nuchitprasittichai (2020) analyzed multiple flowsheet con-
mature state of development makes them attractive. figurations for CO2 hydrogenation via optimization of neural networks.
Methanol is also suitable for direct combustion and this has been The training data for these networks was obtained via latin hypercube
an active topic of research since the 1980s (Von KleinSmid et al., sampling of flowsheet results from different reactor-recycle configura-
1981; Weir et al., 1981; GE, 2001). Methanol can be used for power tions simulated over a range of operating conditions. Ultimately, the
generation in turbines, wherein the differences between methanol and lowest cost configuration used two CO2 hydrogenation reactors in series
methane necessitate a number of modifications to system design and/or to maximize methanol yield without the additional energy consumed
operation (Murray and Furlonge, 2009). Methanol contains 60% less by recycle. Kenkel et al. (2021) used superstructure optimization for
energy than CH4 on a mass basis, so more fuel and larger fuel feed CO2 hydrogenation flowsheets, considering multiple alternative CO2
infrastructure are required for a given energy demand. Methanol is sources and electrolysis technologies. Their model was formulated as
more viscous, so either additional engine lubricants are required or a MILP with piece-wise linearization for equipment costs. The lowest
components like fuel pumps need to be constructed from specific ma- cost solution captured CO2 from refinery fluegas and used low pressure
terials. Methanol also has a higher auto-ignition temperature, so either alkaline electrolysis. The maximum carbon removal design addition-
combustion enhancing additives are needed or advanced combustion ally powered CO2 direct air capture using methanol waste heat. This
strategies like exhaust gas scavenging or fuel boiling can be used. Well- increased methanol production costs by 10%.
designed methanol combustion turbines can actually achieve higher Much of the process design research for co-electrolysis has focused
electrical and CHP efficiencies than those use methane (Räuchle et al., on energy optimization. Léonard et al. (2016) and Chaniago et al.

9
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

(2019) developed simulation models for methanol production based on formulation which minimized methanol production costs over a four
co-electrolysis. These models used detailed kinetic and thermodynamic representative day operating period using 5 min control intervals.
models for the co-electrolysis, methanol synthesis reactors, and distilla- This control approach was used to compare steady-state methanol
tion. Léonard et al. (2016) used pinch analysis to reduce process heat production enabled by batteries and dynamic methanol production.
consumption by 55% and improve energy efficiency by 12%. Chaniago Oversized and dynamically operated methanol production resulted in
et al. (2019) included heat recuperators before the co-electrolysis unit lower methanol production costs by avoiding large battery installations.
and methanol synthesis reactors in their process flowsheet to further For the dynamically flexible approach to be viable, a partial capacity
reduce energy consumption. They employed a vortex search optimiza- operation lower bound and maximum ramp rate of 25% and 25%/h
tion algorithm for optimal process design and by doing so, improved respectively were identified.
overall energy efficiency by 16%. Al-Kalbani et al. (2016) performed
an energy optimization-based comparative assessment of CO2 hydro- 3.4. Capacity planning
genation and solid oxide co-electrolysis. They developed flowsheet
models for both processes and optimized operating conditions and heat Capacity planning and scheduling for PtM is less well-explored than
integration configurations to maximize energy efficiency. They found that for PtH. PtM systems are a newer topic of study and furthermore,
the overall energy efficiency of the co-electrolysis process to be double capacity planning for these systems can be more computationally dif-
that of the hydrogenation process. However, the co-electrolysis process ficult because of the additional processes involved and the complex
did require more heat due to its high operating temperature, so the
dynamics of methanol synthesis and distillation-based separation. As
methanol production setting which informs the heating method (steam
such, much of the work-to-date in this area uses simplifying assump-
vs. electrical) could affect process selection. Zhang and Desideri (2020)
tions to reduce the decision space. Rivarolo et al. (2016) considered
developed a heat integrated flowsheet model for methanol production
steady-state methanol production enabled by hydrogen buffer storage
based on co-electrolysis and used it for multi-objective optimization to
and grid power purchases, thus only scheduling the hydroelectric,
minimize methanol production cost and maximize energy efficiency.
wind, and solar power generation as well as electrolysis. They for-
They used a MINLP formulation to determine optimal process operating
mulated sizing optimization as an NLP and linked this to a genetic
conditions and to select and size the hot and cold utilities. The pareto-
algorithm for hourly energy management via a penalty function. Chen
optimal multi-objective solution achieved minimum production cost
et al. (2021) performed similar analysis, parameterizing wind, solar,
and maximum LHV energy efficiencies of $530/ton methanol and 72%
and external power import fractions to study their effect on methanol
respectively, with co-electrolysis investment and operation being the
production costs. They considered steady-state methanol production
main cost driver.
and developed a high throughput NLP capacity planning model with
Lonis et al. (2019) examined together methanol production and
heuristic power and hydrogen management algorithms based on renew-
energy utilization via reversible solid oxide cells (R-SOFC). They devel-
able power availability. Case studies for Kramer Junction (CA), USA
oped detailed steady-state process models for the R-SOFC in both elec-
and Norderney, Germany showed lowest cost methanol production with
trolysis and fuel cell modes, a CO2 hydrogenation-based methanol syn-
solar and wind dominant power supply respectively. The results also in-
thesis reactor, a distillation column for methanol purification, and ther-
dicated that interday hydrogen storage capacities (with approximately
mal storage to integrate methanol production and utilization modes.
18 h duration) optimized the tradeoff between high hydrogen storage
Thermal storage was modeled as an isothermal phase change material
costs and lessening the dependency on external power purchases.
which was melted by the heat generated by power generation in fuel
cell mode and then solidified when heat was withdrawn for electrolysis Martín and Grossmann (2017) developed a multi-period MINLP
and methanol synthesis. They assumed production-utilization mode flowsheet optimization model for methanol production using variable
switching every 6 h; methanol and thermal storage units were sized wind and solar energy and CO2 from algae growth. For computational
accordingly. Whole-system energy optimization via pinch analysis gave tractability, extensive operating decisions (e.g. material and energy
roundtrip power-to-power efficiency of 33%. flows) were optimized with monthly resolution based on wind and solar
Recent work has also focused on better understanding and quanti- availability whereas intensive decisions (e.g. temperature, pressure)
fying the dynamics of methanol synthesis based on CO2 hydrogenation. were optimized once for a given design and assumed to be constant over
Varela et al. (2020) developed a detailed dynamic model for CO2 each month. In their work, hydrogen and methanol production rates
hydrogenation to understand the effect of feed flow variations caused were coupled, so their installed capacities could be treated as a single
by renewable intermittency. This analysis focused only on a reactor- decision. They found that using PV arrays for power generation gave
flash separation process, the end product of which is ‘‘crude’’ methanol. the lowest methanol production costs. They (Martín, 2016) extended
They studied independently the effect of changes in H2 and CO2 con- this optimization approach to a two-stage MILP stochastic model to
centrations and found an operating window with up to 20% and 5% determine optimal capacity planning and dispatch of combined wind-
respective step changes in these concentrations allowed new steady and solar-powered methanol production using CO2 from local power
states to be reached within 90 s. Cui et al. (2022) performed detailed plants. The capacity decisions were the number of wind turbines, PV
dynamic simulations of renewable methanol production to understand arrays, and electrolysis units. Surrogate models developed using results
process performance characteristics with production rates reduced to from (Martín and Grossmann, 2017) defined the size of methanol syn-
50% of installed capacity over ramp times of 15 min to 2 h. They thesis units as a function of the number of electrolysis units installed as
specifically developed detailed dynamic models of methanol synthe- well as optimal intensive operating variables. Dispatch decisions were
sis and distillation with PI controllers tuned via Ziegler–Nichols and again made with monthly resolution and 27 scenarios were considered
Tyreus–Luyben rules to adjust production based on feed rate changes. each month cumulatively for uncertainty in wind generation, solar
Steady-state energy efficiencies at full and half capacity operation were generation, and the sale price of excess generated power.
found to be within 2%, with slower ramping reducing the magnitude Chen and Yang (2021) more explicitly analyzed the effects of
of temporal efficiency fluctuations. The detailed simulation results were methanol production flexibility by optimizing system operation with
used to develop nonlinear autoregressive exogenous surrogate models. higher resolution. They developed a capacity planning and scheduling
Huesman (2020) analyzed the effect of PtM dynamics on solar model which enables (in general) decoupled operation of electrolysis,
methanol production costs through an optimal control lens. They mod- methanol synthesis, and methanol distillation. Storage for H2 and
eled electrolysis, methanol synthesis, and methanol distillation each raw (pre-purified) methanol could be installed to accommodate the
as a first order process, giving the overall methanol production pro- varying dynamic ranges and ramping capabilities of each process unit.
cess third order dynamics. This process model was used in an E-MPC Further, they included a hydrogen-fed fuel cell for backup power

10
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

Fig. 3. Conceptual technology outline for electrified ammonia production and utilization in energy and transportation.

generation. They formulated this model as an LP to enable a full year, was still less expensive to use than methanol because it does not require
hourly resolution horizon for scheduling decisions. This formulation CO2 transportation.
was enabled by considering fully flexible electrolysis and a 10% lower
operating bound for methanol synthesis while methanol distillation was 4. Power to ammonia
constrained to operate at steady-state. They found that combining over-
sized and dynamically operated electrolysis and methanol synthesis PtA is another extension of PtH which involves combining hydrogen
processes reduced methanol production costs by up to 35%. with nitrogen separated from air. As with methanol, ammonia can
easily be stored as liquid and then used to generate power and/or
Bique et al. (2018) and d’Amore and Bezzo (2020) used time
heat via fuel cells or combustion technologies. Fig. 3 outlines the PtA-
invariant capacity planning models to analyze PtM from a supply chain
relevant technologies described in this section and the interconnections
perspective, and formulated MILPs to determine the optimal location
between them. Detailed descriptions of these ammonia production and
and sizing of carbon capture to provide CO2 , electrolysis for hydro-
utilization technologies are provided in Sections 4.1 and 4.2.
gen production from curtailed renewables, and CO2 hydrogenation
processes for methanol production. Bique et al. (2018) performed 4.1. Ammonia production technology
a case study for northwest Germany where they considered 18 can-
didate production sites for methanol or its feedstocks. They found Synthetic ammonia production currently relies on the Haber–Bosch
that economically optimal supply chains co-located electrolysis and (HB) process, which has been in operation since the 1920s (Smith et al.,
methanol production at sites of highest curtailed renewable availability 2020). The process combines hydrogen and nitrogen in a 3:1 molar
while transporting CO2 from the power plants at which it is captured. ratio to produce ammonia in an exothermic reaction:
d’Amore and Bezzo (2020) considered the entirety of Europe in a
3H2 + N2 ⇌ NH3 𝛥𝐻 = −91.8 kJ∕mol (6)
future scenario with 50% emissions reduction through a combination
of CO2 utilization and sequestration. In their work, the optimal sup- This synthesis reaction occurs at high pressures and temperatures
ply chain configuration transported CO2 captured at many nodes to of 150–250 bar and 350–450 ◦ C respectively. Reaction equilibrium
few hydrogen-methanol production sites selected based on their low limits single pass conversion of ammonia to 15%–20%, so unreacted
electricity costs. The ability to produce methanol for use in the energy hydrogen and nitrogen are recycled after being separated from the
sector decreased overall supply chain costs by 4% relative to CO2 ammonia product via condensation which occurs at synthesis pressure
sequestration alone. and temperatures below 0 ◦ C. State-of-the-art implementations of the
Tso et al. (2019) compared methanol to hydrogen as an energy HB process tend to use multiple catalyst beds in series with inter-
carrier. They formulated a time-invariant MILP supply chain model cooling to improve reaction conversion and reduce the amount of gas
which minimized the levelized cost of energy (LCOE) for fixed storage being recycled, ultimately to mitigate synthesis loop pressure drop and
durations by determining the optimal location and capacity of wind reduce refrigeration requirements for condensation. This is achieved
and solar generation, hydrogen or methanol production, hydrogen- or either through indirect cooling with a heat exchange network inter-
methanol-to-power technologies and, in the case of methanol produc- twined with the beds or direct quench cooling where low temperature
feed gases are partitioned between each bed’s inlet, which addition-
tion, pipelines to provide CO2 from industrial sources. For a one month
ally serves to increase synthesis rates by reducing ammonia concen-
storage duration, the LCOE using hydrogen was 35% less than that
tration (Khademi and Sabbaghi, 2017). As with methanol produc-
using methanol because hydrogen’s higher roundtrip efficiency resulted
tion, economies of scale affect the capital investment required for the
in less required investment in renewable generation and because CO2
HB synthesis process (Bartels, 2008; Morgan, 2013; Bañares-Alcántara
transportation costs were avoided. When energy must be stored for
et al., 2015):
three months, the economic advantage of hydrogen was reduced to 14%
𝑥
because higher energy capacity costs are incurred. However, hydrogen 𝐶𝐴𝑃 𝐸𝑋(𝑀𝑀$) = 3.4( 𝑁𝐻3 )0.6 (7)
1000

11
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

where 𝑥𝑁𝐻3 is the ton/y nameplate capacity of the HB synthesis electrolytes) can reduce capital costs. Additionally, the modular nature
process. Electric implementations of the HB process – those which of electrosynthesis cells enables production scale flexibility without
are not energy integrated with upstream SMR-based H2 production higher capital intensity. Further, these systems are more resilient to
– consume approximately 0.5–0.75 MWh/ton NH3 (Morgan, 2013; O2 or water impurities than catalytic ammonia synthesis. Overall,
Bañares-Alcántara et al., 2015). electrosynthesis is likely a decade from commercialization (MacFarlane
The first synthesis catalysts were composed primarily of iron (Fe) et al., 2020).
(Liu, 2014), with promoted iron (specifically magnetite or wustite) now
the industrial standard. Over the last 25 years, promoted ruthenium 4.1.1. Nitrogen production technology
catalysts with appreciably high kinetic rates at lower temperatures Nitrogen for PtA can be obtained from technologies which are
and pressures have been developed (Rosowski et al., 1997). Lower extensively used in the air separation business. Three different types
temperature shifts the synthesis reaction equilibrium toward the am- of processes are generally used: cryogenic distillation (CRD), pressure
monia product, potentially allowing higher single pass conversion to swing adsorption (PSA), and membranes. Each is reviewed briefly here.
ammonia. Lower pressure can reduce capital costs and energy required Cryogenic distillation produces nitrogen as distillate and O2 as bottoms,
for feed compression. However, lower pressure synthesis has limited relying on the difference in their respective boiling points of −196 ◦ C
utility when paired with condenser-based separation, which relies on and −183 ◦ C. The process operates at a temperature slightly below the
high pressures in addition to cryogenic temperatures for ammonia boiling point of O2 and uses at least two columns in series operating at
liquefaction. A desire for lower pressure ammonia production has descending pressures from 6 bar to 1.5 bar to obtain ultra high nitrogen
led to the development of alternative separations. One approach is purity above 99.9% (Bocker et al., 2013). Cryogenic distillation is
direct replacement of the condenser with a bed of adsorbent such tightly heat integrated to achieve energy consumption around 0.11
as a zeolite (Beach et al., 2018; Lucero et al., 2021) or supported MWh/ton N2 , less than PSA or membranes, despite its operation at
metal halides for chemisorption (Malmali et al., 2016; Wagner et al., very low temperatures (Bocker et al., 2013). Unfortunately, this heat
2017). This direct replacement enables comparable ammonia produc- integration also limits process dynamic flexibility. It requires hours
tion rates to condenser-based HB with an order of magnitude reduction for startup and has a fairly narrow dynamic range, with operating
in synthesis pressure down to 10–30 bar and separation at ambient lower bounds above 60% of installed capacity (Miller et al., 2008).
temperatures or even higher. Well-designed ammonia production at Nonetheless, this process is the industrial standard for large scale air
these milder operating conditions can reduce capital costs relative to separation at scales above 150,000 ton N2 /year. In fact, the Linde
HB synthesis because high pressure and low temperature separation Ammonia Concept already uses cryogenic distillation to obtain nitrogen
are the major cost drivers of that process. Some very recent work has feedstock (Pattabathula and Richardson, 2016).
analyzed intensified ammonia synthesis-separation processes such as a PSA is a cyclic separation process which uses at least two beds of
catalytic membrane reactor (Zhang et al., 2021) or a sorption-enhanced O2 -selective carbon molecular sieve absorbent (Step and Petrovichev,
reactor (Smith and Torrente-Murciano, 2021). Intensified synthesis and 2002). PSA operates isothermally at ambient temperature with pressure
separation may lead to further lowering of capital intensity by reducing cycling between 6–8 bar for adsorption and 1 bar for desorption. PSA
the number of required process units, but inherent loss in design units are compact and inherently modular, so from a cost perspective
degrees of freedom necessitates additional investigation, for example, they are well-suited for small production scales below 7500 ton N2 /day.
to ensure adequate conversion to ammonia or energy efficiencies. They also have good dynamic flexibility, with startup times less than
Electrochemical ammonia synthesis (electrosynthesis) which re- an hour and a lower operating bound around 30% (Ivanova and Lewis,
places the HB process and instead produces ammonia directly from 2012). The main drawbacks of PSA relative to cryogenic distillation are
water and nitrogen at ambient pressure is a more ambitious alternative more expensive scale-up for larger production targets, lower nitrogen
that has also been an active topic of recent research. These processes purity around 99.8%, and higher energy consumption of 0.22–0.31
produce ammonia at the cathode of an electrosynthesis cell, with MWh/ton N2 (Bocker et al., 2013).
the specific nitrogen reduction reaction and accompanying oxidation Membrane-based air separation uses O2 -permeable polysulfone hol-
reaction at the cathode depending on the electrolyte (Shipman and low fiber (Lababidi, 2000) or carbon molecular sieve (Campo et al.,
Symes, 2017). Aqueous electrolytes have been used, but both ammonia 2010) membranes. This separation process occurs at ambient temper-
synthesis rates and selectivity are too low to be practical at this ature and is driven by a pressure difference between the feed and the
stage, largely due to a competing hydrogen evolution reaction (HER) permeate, which respectively operate at 6 to 25 bar and atmospheric
accelerated by the water contained in the electrolyte (Suryanto et al., pressure. Membranes for nitrogen production are a less developed
2019). Hydrophobic ionic liquid electrolytes have also been studied technology than cryogenic distillation or pressure swing adsorption.
with a view on improving ammonia selectivity by minimizing HER. This process offers better dynamic flexibility than cryogenic distillation
Their associated energy efficiencies are as high as 60% but they are and simpler operation than PSA because it is continuous in nature.
limited by low synthesis rates (Zhou et al., 2017; Suryanto et al., However, increased pressure is required to achieve nitrogen purity
2018). On the other hand, high temperature (200–500 ◦ C) molten salt above 99.5%, which can result in energy consumption as high as 0.65
electrolytes improve synthesis rates by two orders of magnitude but MWh/ton N2 (Bocker et al., 2013). Post-membrane separation is also
suffer from energy efficiencies below 35% (Licht et al., 2014; Yang required due to the presence of argon. Additionally, the investment cost
et al., 2020). Solid-state proton conducting membrane electrolytes are of this process depends almost linearly on the quantity of membrane
promising, being able to achieve the highest recorded electrosynthesis required and thus the production scale; this lack of economies of scale
rates while operating below 100 ◦ C (Garagounis et al., 2014). However, means that membranes may only be well-suited to nitrogen production
these membranes can degrade over time when in contact with ammonia below 750 ton N2 /day (Smith and Klosek, 2001).
due to their acidic nature (Hongsirikarn et al., 2010), so strategies
are needed to mitigate this. Anode and cathode material selection is 4.2. Ammonia utilization technology
far from consensus at the current stage of development, but systems
studied in the literature have used a host of different metals (Ni, Fe, Ammonia can be decomposed to hydrogen and nitrogen in the
Pt, Pd, Ru, Re, Rh). Ultimate material selection will partially depend reverse of the synthesis reaction. This can produce hydrogen to feed the
on electrolyte compatibility while also requiring resolution of a trade- fuel cell technologies discussed in Section 2.3. Ammonia concentrations
off between cost and productivity, especially if precious metals are to on the order of 1 ppm can cause irreversible damage to alkaline or
be used. The potential benefits of this approach are plentiful (MacFar- PEM fuel cells, resulting in efficiency loss and shorter lifetimes (Halseid
lane et al., 2020). Ambient pressure and low temperatures (for some et al., 2006), so decomposition and/or purification must be complete.

12
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

The endothermic ammonia decomposition reaction occurs at or near from biomass. Sánchez and Martín (2018a) developed an NLP model
atmospheric pressure and requires high temperatures of 500–700 ◦ C. for optimal design of a wind- and solar-power PtA facility using alkaline
Reactors for ammonia decomposition are still actively being designed; electrolysis, cryogenic distillation for nitrogen production, and both
concepts include fixed bed reactors with membranes for hydrogen direct and indirect cooling synthesis reactor configurations. They found
removal (Abashar, 2018), or catalytic microchannel reactors (Chiuta lowest renewable ammonia production costs of roughly $1000/ton NH3
and Bessarabov, 2018). Heat integrating ammonia decomposition with which used PV arrays and direct reactor cooling, with main cost drivers
waste heat from hydrogen-fed power generation (Kim and Kwon, 2011) being PV arrays and electrolysis. They also extended this analysis to
or other co-located processes is crucial for energy efficiency. The most account for modular availability of electrolysis and nitrogen production
well-studied decomposition catalysts are composed of supported Ni and at small scale (Sánchez and Martín, 2018b), modeling and selecting
Pt (Chellappa et al., 2002), while Ru has been identified as the most between CRD, PSA and membrane separation for nitrogen production,
active (Lamb et al., 2019). Efforts to reduce cost of decomposition and ultimately determining the production scale ranges over which
catalysts have led a move away from use of transition metals to each technology was optimal. Wang et al. (2020) analyzed a PtA
sodium (David et al., 2014) or lithium (Hunter et al., 2016). system which used a battery and hydrogen buffer storage to modulate
Ammonia can also be internally reformed in SOFC (Farhad and short-term variability in renewable generation, notably developing a
Hamdullahpur, 2010) to allow for its direct use, as previously discussed combined optimal design and control methodology which included a
for methanol. Ammonia-fed SOFC can achieve comparable efficiencies proportional load tracking strategy for electrolysis and P or PI temper-
to use with other fuels, and this has been demonstrated with over ature control on cryogenic distillation which produces nitrogen and the
1000 h of operation (Okanishi et al., 2017; Siddiqui and Dincer, 2018). ammonia synthesis reactor. Optimal design was performed considering
Direct ammonia fuel cells (DAFC) have increasingly gained attention as four representative days of wind generation parameterized into 100
their lower operating temperatures of 80–120 ◦ C potentially make them control intervals.
better suited to dynamic operation (Zhao et al., 2019). These are largely In our previous work (Palys et al., 2018), we developed a dynamic
still at the experimental stage, but the most promising recently studied model of the inherently transient absorbent-enhanced ammonia synthe-
cells use a Pt-based anode and a precious metal-free cathode where sis process discussed in Section 4.1 to enable MINLP optimal design
ammonia and air are respectively fed and a OH− exchange membrane to minimize synthesis costs. We found that this alternative process
electrolyte (Gottesfeld, 2018). This DAFC system achieved ammonia-to- required less investment than traditional process at production scales
power efficiencies as high as 40% with expected capital cost reduction up to 75,000 ton NH3 /year and was increasingly advantageous at small
due to the absence of precious metals in the cathode. production scales, but used 3x more energy due to endothermic desorp-
Power and heat can also be generated from ammonia combus- tion of ammonia. These results were motivation for continued discovery
tion, though fundamental differences relative to carbon-based fuels of absorbent cycling conditions which were less energy intensive. Lin
necessitate system design and operational changes. Ammonia suffers et al. (2020) performed similar comparative analysis for the condenser
from lower flame speed and a higher auto-ignition temperature than and absorbent processes at a production scale of 20,000 ton NH3 /year.
other fuels. This can be alleviated through addition of a combus- They pursued heat integration of the exothermic ammonia synthesis
tion promoter (Frigo and Gentili, 2013). Ammonia blended with 5%– reaction and absorption, and endothermic desorption which ultimately
10% hydrogen has been shown to maximize efficiency without added brought synthesis loop energy consumption to comparable levels as the
emissions (Mørch et al., 2011). Pure ammonia mixtures can also be condenser process.
preheated via exothermic combustion to reduce or eliminate the need There have also been similar efforts to analyze and optimize pro-
for blending during continuous, non-startup operation (Chen et al., cesses which use ammonia for energy production. Chachuat et al.
2022). Another challenge is that ammonia combustion dramatically (2005) developed a detailed steady-state model for a system with
increases NOx emissions. These can be removed via post-combustion an ammonia decomposition reactor followed by hydrogen-fed SOFC
selective catalytic reduction (SCR) of NOx with ammonia, which react and used a MINLP formulation to optimize its design, achieving a
to form nitrogen and water. The SCR unit adds cost to the combustion maximum energy densities above 1 MWh/ton NH3 while meeting a
system. Further, the additional use of ammonia, while not requiring nominal power demand. Wang et al. (2017) considered a 100 MW
storage of other chemicals, decreases the macroscopic ammonia-to- PtA system which employed a reversible SOFC which can operate in
power efficiency (Boretti, 2012). Combustion-based power generation both electrolysis and fuel cell modes along with dedicated ammonia
using ammonia has been studied and demonstrated using both gas synthesis and decomposition reaction systems. Energy integration of
turbines (Tsujimura and Iki, 2016; Kobayashi et al., 2019) and ICE the constituent processes in this system was pursued; exothermic heat
systems (Reiter and Kong, 2011; Patil et al., 2014; Kane and Northrop, from ammonia synthesis was used to heat steam for the SOFC in
2021). ammonia production or ‘‘charging’’ mode while the heat required for
endothermic ammonia decomposition was obtained from the SOFC
4.3. Process design and operation effluent in power generation or ‘‘discharging’’ mode. Optimal design of
this system at the 100 MW scale enabled overall efficiencies as high as
There has been research focused on process design and/or synthesis, 72% and levelized cost of energy as low as $0.24/kWh. Sánchez et al.
which is generally characterized by complex nonlinear models for (2021a) solved several NLPs to optimally design ammonia-to-power
steady-state or minimally time-varying (for example, monthly) ammo- process alternatives. Potential constituent processes included a cat-
nia production. Some of this work considered conventional condenser- alytic membrane reactor for ammonia decomposition, nitrogen/argon
based ammonia synthesis and focused on optimization of its design and recovery to monetize these combustion inerts, gas turbines, selective
operating conditions as well as design and/or selection of precursor catalytic reduction to remove NOx produced by ammonia combustion,
hydrogen and nitrogen production. Demirhan et al. (2019) developed a and either secondary hydrogen combustion or recovery and recycle
MINLP model for superstructure optimization of ammonia production, using a membrane. They determined ammonia-to-power efficiencies in
considering as feedstock a combination of renewable electricity (with the range of 30 to 45% and levelized power costs as low as $0.23/kWh
constant availability), biomass, and natural gas and multiple competing at the 100 MW scale.
technologies for hydrogen production (e.g. PEMEC and AEC). They Another key topic of recent research on PtA is determining how to
optimized the ammonia cost over a range of production scales, loca- operate its constituent subsystems in the face of renewable intermit-
tions, and GHG emissions restrictions. They found present day PtA to tency and variable energy demands. To date, this topic has received
be approximately two times more expensive than ammonia produced less focus. Allman and Daoutidis (2018) formulated a MILP optimal

13
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

scheduling model for dynamic ammonia production to minimize op- lowest production costs. They found that renewable ammonia produced
erating costs based on a 24-hour receding horizon. They considered at the 35,000 ton NH3 /year scale can be cost competitive in four
power sourced from both wind generation and a utility with time-of-use high renewable potential South America locations if ammonia synthesis
pricing. Dynamic flexibility was enabled through oversized hydrogen can operate at 20% partial loads. Allman et al. (2019) overcame
and nitrogen production with buffer storage as well as the oversized computational tractability limitations by embedding their previously
ammonia synthesis to enable some turn-down, though conservative developed scheduling framework from Allman and Daoutidis (2018) in
partial operation and ramp constraints were enforced. The iterative a PtA capacity planning model. Specifically, they first determined opti-
use of this scheduling formulation over representative weather years mal schedules for many fixed feasible designs and developed surrogate
allowed for a survey of different U.S. Midwest locations, ammonia models to describe annual operating costs based on these scheduling
production scales, and degrees of nitrogen and hydrogen oversizing to results. These surrogates were used to inform a much smaller time-
determine which factors had the largest impact on production costs. Xu invariant NLP design model, enabling computational tractability while
et al. (2020) developed a 24-hour optimal scheduling model for PtA- incorporating this more complex scheduling formulation.
based energy storage in a CHP microgrid setting. Dynamic flexibility Nayak-Luke et al. (2018) extended their previous capacity planning
was enabled through a battery, hydrogen buffer storage, and ammonia analysis with rule-based scheduling to determine optimal renewable
storage, and the process models included degradation effects to assess ammonia production costs for 534 locations using hourly resolution
the trade-off between dynamic production and component lifetime. wind and solar availability data in each and optimizes the power
This model was formulated as a MINLP, but practically decomposed generation fractions in addition to system design. They found lowest
into NLP and MIQP subproblems for computational tractability. Ac- global renewable ammonia production costs of $473/ton NH3 today
counting for degradation during scheduling resulted in less frequent and $310/ton NH3 in 2030. Fasihi et al. (2021) performed a similar
charge/discharge and on/off cycling for the battery and electrolysis global renewable ammonia production study, creating an optimal cost
respectively by using more ammonia for CHP. atlas over a 0.45◦ by 0.45◦ spatial grid. They formulated a capacity
planning model which sized every technology in the PtA system and
4.4. Capacity planning considered dynamic flexibility in the form of batteries, hydrogen buffer
storage, hydrogen-fed gas turbines and dynamic ammonia production.
PtA capacity planning is inevitably more computationally intensive Their scheduling model was formulated as a LP, with fully flexible
than for PtH because of the additional process units for nitrogen and
operation assumed for all processes except ammonia synthesis which
ammonia production as well as the possibility of decoupling hydrogen
was constrained to run at or above 50% partial capacity and with
and ammonia production by using hydrogen buffer storage. As such
hourly ramp up and down limits of 2% and 20% respectively. Their LP
these models rely on less detailed descriptive models for operational
scheduling formulation allowed the use of representative year data with
characteristics (e.g. energy consumption, dynamic lower bounds, and
hourly time steps. They also solved their model using technology cost
ramping capabilities) of process units than in Section 4.3 so that
projections from today to 2050, finding lowest global ammonia produc-
generally these problems can incorporate unit scheduling to account
tion costs of $440/ton NH3 today and $260/ton NH3 in 2050, with the
for variable renewable power availability or time-varying electricity
cost reduction stemming from less expensive renewable generation and
prices, as discussed in more detail in Section 2.5. Beerbühl et al.
electrolysis.
(2015) reduced the design decision space in the context of PtA from
More recently, MILP formulations have been used for PtA capacity
grid power by assuming that electrolysis was fully dynamically flexi-
planning. Osman et al. (2020) considered continuous ammonia produc-
ble while ammonia synthesis must run throughout the year. Further,
tion and optimized the selection, sizing, and scheduling of the upstream
ammonia synthesis capacity was fixed in their work as was nitrogen
electrical and thermal energy, hydrogen, and nitrogen production and
production capacity because no nitrogen buffer is assumed, so sizing
storage infrastructure to enable this constant operation when powered
decisions were only made for electrolysis and hydrogen buffer storage.
with solar, both through PV arrays and concentrated solar power.
This simplification allowed them to formulate a tractable convex NLP
which accounted for partial load operation effects on the efficiency They used a full year of hourly resolution data as input to the ca-
of electrolysis and ammonia synthesis. Nayak-Luke et al. (2018) and pacity planning model. They found present-day LCOA of $718/ton
Armijo and Philibert (2020) did consider sizing of ammonia synthe- NH3 , with future LCOA of $450/ton NH3 through projected technology
sis, along with electrolysis and hydrogen buffer storage as well as cost reductions. Wang et al. (2021a) performed a first of its kind
other units specific to their individual work. Both of their capacity capacity planning study for ammonia production using offshore wind.
planning formulations used scheduling heuristics for computational Their model allowed for a high degree of design flexibility, considering
tractability. Nayak-Luke et al. (2018) additionally optimized sizing decoupled hydrogen and nitrogen production with buffer storage as
of fuel cells for backup power and considered power purchases from well as batteries, hydrogen fuel cells, and ammonia ICE generators for
existing renewable generation. They developed electrical power and local energy storage. They formulated this model as a MILP and aggre-
hydrogen management algorithms which consisted of a set of decision gated wind speed input data into one hourly resolution representative
rules depending on renewable availability from existing wind and solar week for each month (2016 total scheduling periods) for computational
generation. They performed case studies for Lewrick, Scotland using tractability. They compared onshore ammonia production connected
five years of historical hourly resolution data to determine the effect of to offshore wind by underwater power cables with offshore ammonia
wind/solar generation fractions, renewable electricity costs, and ammo- production, and found offshore production to be less expensive. They
nia synthesis partial operation lower bound and ramping capabilities on also found that optimal designs used both hydrogen and nitrogen buffer
optimized ammonia production costs. They found that partial operation storage as well as hydrogen fuel cells to balance wind generation with
lower bounds and ramp rates of 20% and 6%/hour respectively are synthesis energy demands.
needed for cost competitive renewable ammonia. Armijo and Philibert Less work has focused on explicit capacity planning for systems
(2020) additionally determined optimal sizing of PV arrays and wind which both produce and use ammonia for energy. As with methanol,
turbines while also scheduling the purchase of ‘‘firming’’ power from Tso et al. (2019) comparatively analyzed ammonia energy storage to
the local utility. They used a simulation-based optimization approach hydrogen using a time-invariant supply chain optimization model. They
where they calculated ammonia production costs over a range of solar-, found that hydrogen storage gave 40% lower LCOE for 1 month dura-
wind-, and synthesis-to-electrolysis ratios. The size of hydrogen buffer tions due its higher roundtrip efficiency while ammonia and hydrogen
storage was computed during simulation from this relative oversizing are roughly equivalent for 3 month durations due to hydrogen’s higher
and the ‘‘optimal’’ configuration was identified as the one which gave storage cost. In our previous work (Palys and Daoutidis, 2020), we

14
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

compared the economics of using ammonia for standalone renewable Optimal use of hydrogen and ammonia for renewable CHP resulted in
energy storage to those of using hydrogen. We developed a MILP ca- energy supply costs only 3% more expensive than those at historically
pacity planning model to minimize the levelized cost of energy supply low conventional prices in both Mahaka, Hawaii and Northwest Arctic
by determining the size and schedule of each unit in the full power- Borough, Alaska. From a methodological point-of-view, Allen et al.
to-x-to-power system including renewable generation, and considered (2021) proposed a solution acceleration strategy for these types of ca-
both hydrogen and nitrogen buffer storage in ammonia production. For pacity planning and scheduling problems based on linear programming
computational tractability, we aggregated hourly resolution input data relaxation. This approach was shown to reduce computational time
into 672 variable length optimal scheduling periods. We performed by two orders of magnitude, potentially making complex formulations
case studies for 15 U.S. cities which comprehensively represent climate- tractable in the future, for example, those with less time aggregation
demand regions throughout the country and found that using hydrogen or more competing technologies.
gave slightly lower LCOE in locations with high solar potential and/or From a supply chain perspective, the previously described work of
less seasonal generation-demand mismatch, while ammonia provided Tso et al. (2019) considered simultaneous use of hydrogen, methanol,
up to 30% LCOE reduction in locations with high wind potential and and ammonia as energy carriers in Texas. In the optimal supply chain
larger seasonal storage requirements. for three month storage duration, all three carriers are used, with
spatial synergies resulting in 26% LCOE reduction compared to using
5. Integrated power-to-x systems ammonia alone, the best case for a single carrier. Some hydrogen was
produced since it provides the highest roundtrip efficiency, but only
The potential of power-to-x can be improved by exploring the pos- in areas of the highest PV generation potential because solar’s strong
sibility of systems which consider parallel power-to-x processes and/or diurnal (rather than seasonal) characteristic mitigates hydrogen storage
demands arising from different sectors. The synergies offered by such volumes. Methanol was produced near CO2 sources because PtM is
integration can be exploited to improve the economics and thus likeli- more efficient than PtA while ammonia was produced in other locations
hood of adoption. Optimization is a powerful tool to aid in maximizing to reduce total CO2 transportation costs. Sánchez et al. (2021b) also
this economic benefit, especially with the increased decision space that analyzed renewable energy storage supply chains which used methanol
comes with more constituent processes and demands. Further, certain and ammonia. They optimized these supply chains through a decompo-
synergies may not be readily apparent or may even counteract physical sition technique which first optimized the supply chain configuration
intuition. For this reason, rigorous MI(N)LP formulations are preferable and technology selection using monthly variation for wind and solar
to intuition-based approaches that use decision rules which are speci- generation and then optimized the capacity and scheduling of the
fied a priori or other heuristics. In this section, we review recent work chosen processes using an hourly resolution representative week per
on integrated power-to-x systems falling under two categories: Hybrid season. They performed case studies for the Leon province of Spain.
energy systems and sector coupling. Their results corroborate those of Tso et al. (2019), with methanol
produced only near CO2 sources and renewable ammonia produced
5.1. Hybrid energy systems
elsewhere since it only requires air and water as raw materials. Han
and Kim (2019) studied a transportation energy system which supplies
Recent work on renewable energy storage facilities has considered
carbon-containing fuels for ICE vehicles, power for electrical vehicles,
co-optimization of power-to-x pathways. In our previous work on com-
and hydrogen for fuel cell vehicles in 15 regions in South Korea. They
paring the optimal economics of hydrogen and ammonia for electrical
developed a MILP model for capacity planning and monthly schedul-
energy storage (Palys and Daoutidis, 2020), we also optimized their
ing with multiple investment intervals to reflect changes in vehicle
synergistic use in the 15 studied U.S. locations. We found that systems
mix. This optimization revealed a spatial synergy in that renewable
which use both outperform those which use only one storage option,
hydrogen produced in every region was used for fuel supply, while also
taking advantage of the more efficient hydrogen storage for short-term
being transported to centralized CO2 -to-fuel facility in regions with the
balancing and ammonia’s lower cost for seasonal storage. Sánchez et al.
cheapest land.
(2022) performed a similar analysis for 14 different regions of Spain
with batteries, hydrogen, and ammonia available as energy storage
media. They developed a MILP capacity planning and scheduling using 5.2. Sector coupling
one hourly resolution representative week per season. The optimized
fully renewable ESS in each region used all three storage technologies Sector coupling is well-suited to power-to-x due the utilization
to achieve the best overall tradeoff between efficiency and storage versatility of its products. Hydrogen and methanol both have strong
cost in balancing intermittent wind and solar resources with demand. demands in the chemical production sector, while ammonia is tradi-
Demirhan et al. (2020) proposed to optimize connected energy facilities tionally used as fertilizer. Combining these sectors with energy supply
in high renewable resource and high demand regions, considering can reduce costs and make more complete use of renewable energy
liquefied hydrogen, methanol, and ammonia to enable a low cost because, in general, chemical commodities inherently have monthly
interconnection. They formulated a combined capacity planning and or seasonal demand resolution, in contrast to energy demands which
scheduling MILP and aggregated input data into 12 season-representing are pseudo-instantaneous. This means that chemicals can largely be
hourly resolution days which are repeated a variable number of times. produced when renewable generation exceeds energy demands. Recent
They performed a case study analyzing the coupling of Texas and work has explored this possibility. Bødal et al. (2020) analyzed cou-
New York City (NYC), looking to replace 10% of NYC’s annual power pling the electrical power and fuel cell vehicle sectors using renewable
demand. They found that using chemicals for energy transportation hydrogen. They used an LP model for capacity planning and schedul-
reduces overall costs by up to 50% compared to segregated systems and ing across multiple generation and demand nodes with transmission
that using both hydrogen and methanol or hydrogen and ammonia as between them. They performed a case study for Texas and found that
energy carriers gives 10% cost reduction than using hydrogen alone. the dynamic flexibility of PEM electrolysis reduced energy prices by
We more recently extended our hydrogen–ammonia electrical energy acting as demand response in its hydrogen fuel production schedule.
systems analysis to renewable CHP systems in remote locations (Palys Further, some power generated via hydrogen fuel cells acted as a
et al., 2021). These locations are characterized by high conventional means of longer duration energy storage than batteries. Hydrogen also
energy costs. We considered fully renewable energy systems which allowed spatial balancing of power and fuel demands via transportation
could produce hydrogen and ammonia and use them for power gen- pipelines, with electrolysis mostly occurring at nodes with high renew-
eration with heat cogeneration or dedicated heat production in boilers. able potential. Allen et al. (2022) considered a similar Texas-based

15
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

system, with the major difference that they considered sector-agnostic alternative to battery electric vehicles in electrifying the transportation
cryogenic hydrogen demand. They formulated a planning and schedul- sector, especially for long haul or heavy transportation. PtH offers the
ing MILP in which investment decisions were made every five years and highest roundtrip efficiency and it the most well-developed pathway of
proposed a graph-based approach to solve optimal scheduling. Their the three discussed in this review. Hydrogen production via electrolysis
results showed that liquified hydrogen was produced almost exclusively requires only water as feedstock and its costs are expected to fall
with renewable energy that would otherwise be curtailed. Andika et al. as demand for electrolysis units increases in the coming decade. On
(2019) analyzed a wind and solar energy supply system which could the utilization side, flexible PEM fuel cells are still expensive and
also produce and export methanol to improve its economic viability. combustion technologies that can use 100% hydrogen are not yet
Their considered system included batteries for demand balancing and available at commercial scale. As with electrolysis, high demand for
methanol synthesis based on CO2 hydrogenation. They used an NLP these technologies should reduce costs and enable the overcoming
capacity planning and scheduling model with two years of hourly reso- of operational challenges with additional research and development.
lution renewable generation and demand data. They found that PtM is However, the high cost of hydrogen storage resulting from its low
able to monetize otherwise curtailed renewable energy, decreasing total energy density at ambient conditions is a barrier with a less clear
system costs without affecting ability to meet power demands. Salmon solution.
and Bañares-Alcántara (2021) considered ammonia production with The promise of PtM and PtA lies in decreasing these storage costs via
coupled use of variable renewable electricity and variable-priced grid liquid storage at moderate conditions. Methanol can be stored as liquid
electricity as well as the possibility of selling excess generated renew- at ambient pressure and temperature. Furthermore, its production in
able power to the grid. They used a MILP capacity planning model with a PtM setting sequesters CO2 . To this end, direct CO2 hydrogenation
one year of representative hourly resolution wind availability, solar and co-electrolysis processes are still being developed to improve en-
availability, and electricity price data to perform case studies for six ergy efficiency and reduce capital costs. From a utilization standpoint,
Australian states. They found that renewable ammonia production with methanol decomposition to hydrogen, combustion, or use in fuel cells
consumer–supplier grid interaction gave 5% average LCOA reduction is well established, but the feasibility of mitigating CO2 emissions
across Australia, while also reducing grid power supply emissions in in distributed use cases (e.g. passenger vehicles, ships) is an open
four of the six states. question. Ammonia also offers high energy density without the burden
Some recent work has examined coupling the energy supply and of CO2 emissions upon use. It can be produced using only water to
agriculture sectors. We conceived of a farm-scale system which used supply hydrogen and air for nitrogen. Interest in PtA has led to the de-
only renewable energy to meet power demands and make hydrogen velopment of new synthesis processes with the aim of lowering capital
and ammonia for use as fertilizer, tractor fuel, and energy storage (Palys investment and improving dynamic flexibility. Ammonia lags behind
et al., 2019). We optimized the design of this system using a MILP for- hydrogen and methanol from a utilization standpoint. Decomposition
mulation with optimal scheduling for hourly resolution representative technologies for hydrogen production or direct-use fuel cells are far
week per month. As in other previously described results, hydrogen from being commercially available. Ammonia combustion seems to
and ammonia were used synergistically for energy storage. Further, have the shortest path to commercialization, with ongoing development
combining energy and agriculture demands led to an increase in hy- of techniques to overcome ammonia’s undesirable low flame speed and
drogen and ammonia production capacity factors, ultimately giving high auto-ignition temperature while also mitigating NOx emissions.
better utilization of capital investment and lower overall costs than Increased development and deployment of power-to-x technologies
considering energy and agriculture systems separately. Wang et al. should lead to cost reductions via multiple avenues. One such avenue
(2021b) extended this system to include explicit nitrogen and water is the development of specific methanol and ammonia synthesis tech-
management with a view on decreasing nitrogen pollution in surround nologies well-suited to using intermittent renewables, namely direct
bodies of water. They also considered urea production from renewable CO2 hydrogenation and low pressure ammonia synthesis processes with
ammonia and captured CO2 , which has less propensity for leaching dynamic flexibility in the near term and electrochemical synthesis pro-
than ammonia. They developed surrogate models to describe inherently cesses in the longer term. The inherent design and operational features
nonlinear soil water and nitrogen dynamics, and embedded these in of these processes should enable lower cost implementations in specific
a capacity planning and scheduling model in which input renewable power-to-x settings than the current state-of-the-art CO hydrogenation
generation, demand, and precipitation data was aggregate into five and HB synthesis processes. Along a different vein, continued deploy-
hourly resolution representative weeks repeated a variable number of ment of existing technologies such as electrolysis and fuel cells should
times. They found that nitrogen pollution can be decreased by 57% reduce capital intensity through economies of volume. Electrolysis and
with only an 8% increase in net present cost. Ikäheimo et al. (2018) fuel cells are inherently modular. Their cell stacks and much of the
considered a much larger spatial scale, determining the optimal design required auxiliary equipment (e.g. power electronics) are fabricated in
of energy and fertilizer supply networks in Northern Europe which an optimized manufacturing environment and then transported to the
only used renewable energy. They performed capacity planning and location of their intended use. This results in potential advantages such
scheduling with hourly resolution for yearlong horizon with tractability as increased efficiency of labor, for example, as well as better use of ma-
enabled by an LP formulation. In their results, renewable ammonia was terials of construction when many modules are being fabricated at once
used as a fertilizer as well as to store and transport energy. Further, they before deployment. The per-module capital cost can be reduced via
found shared use of ammonia pipelines for both fertilizer and energy this manufacturing approach. This concept is well suited to power-to-
carrier purposes depending on the time of year, as fertilizer demands x systems which use spatially distributed renewable resources because
are seasonal. cost savings are still achieved by cumulative capacity increases across
multiple locations, in contrast to the economies of scale traditionally
6. Conclusions observed in the chemical production industries.

This article provides a review of power-to-hydrogen (PtH), power- 7. Process systems engineering perspective
to-methanol (PtM), and power-to-ammonia (PtA) through the lens of
decarbonizing the energy supply and transportation sectors. Collec- Process systems engineering has an important role to play in the
tively, power-to-x provides a path to reducing the cost of balancing economic viability of power-to-x, wherein the involved processes face
intermittent renewable energy generation with demands, especially challenges that conventional chemical production does not. The in-
storing such energy over long durations. Power-to-x also provides an termittency of the input renewable energy necessitates time-varying

16
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

chemical production which can lead to operational challenges as well Notation


as oversizing to meet production targets leading to higher capital
AEC Alkaline electrolysis cell
intensity. The cited work in this review has made strides in alleviating
AFC Alkaline fuel cell
these challenges through design of alternative methanol and ammonia ATRM Autothermal reforming of methanol
production processes, development of coupled scheduling and control (B)EV (Battery) electric vehicle
approaches for dynamic operation, and capacity planning of whole CCGT Combined cycle gas turbine
power-to-x systems or supply chains that use intermittent renewables. CHP Combined heat and power
However, a number of vistas for future research into these processes CRD Cryogenic distillation
and systems remain. DAFC Direct ammonia fuel cell
DMFC Direct methanol fuel cell
At the process level, continued work in optimal design of methanol E-MPC Economic model predictive control
and ammonia synthesis is needed. For methanol, the research-to- ESS Energy storage system
date indicates that a continued focus on CO2 hydrogenation or co- HB Haber–Bosch
electrolysis processes is prudent, specifically analyzing and optimizing HER Hydrogen evolution reaction
their investment costs and balancing this with their promising energy ICE Internal combustion engine
efficiency characteristics. For ammonia, a focus on electrochemical LCOE Levelized cost of energy
LHV Lower heating value
synthesis processes will aid in determining cost- and energy-optimal
LSM Lanthanum strontium
material selection and operating condition regimes with the aim of
MCFC Molten carbonate fuel cell
accelerating their economical deployment. Methanol and ammonia MD Methanol decomposition
synthesis processes, both the existing state-of-the-art and the novel ones MI(N)LP Mixed integer (non)linear program
mentioned above, have complex nonlinear dynamics that can manifest MIQP Mixed integer quadratic program
themselves unintuitively during time-varying operation. These must MD Methanol decomposition
be understood so as to quantify dynamic flexibility limitations, effects MPC Model predictive control
MSR Methanol steam reforming
on energy efficiency, and long term catalyst performance. As such,
PAFC Phosphoric acid fuel cell
dynamic characterization studies like those by Varela et al. (2020)
PEMEC Proton electrolyte membrane electrolysis
and Cui et al. (2022) for methanol synthesis by CO2 hydrogenation PEMEC Proton electrolyte membrane fuel cell
should be continued and expanded to methanol production based on POM Partial Oxidation of Methanol
co-electrolysis. Such studies should also be conducted for condenser- R-SOFC Reversible Solid Oxide Fuel Cell
based, separation-enhanced, and electrochemical ammonia synthesis PSA Pressure Swing Adsorption
approaches. Better understanding of synthesis process dynamics will PtA Power-to-Ammonia
then enable studies on real-time operation and control of PtM and PtA, PtH Power-to-Hydrogen
PtM Power-to-Methanol
especially under uncertainty; such research can be in part inspired by
RTO Real Time Optimization
the existing work of that kind for PtH (Torreglosa et al., 2014)-(Cano
SCR Selective Catalytic Reduction
et al., 2015), though considering methanol or ammonia synthesis in SOC State of Charge
addition to electrolysis in PtM or PtA systems will undoubtedly add SOEC Solid Oxide Electrolysis
complexity. SOFC Solid Oxide Fuel Cell
At the systems level, capacity planning models that simultaneously WGS Water Gas Shift
YSZ Yttrium Stabilized Zirconia
optimize the facilities or supply chains that both produce and use
methanol and/or ammonia should continue to be developed, especially CRediT authorship contribution statement
integrating these with hydrogen and other technologies better suited
for short duration storage. These models should consider multiple can- Matthew J. Palys: Conceptualization, Investigation, Writing – orig-
didate technologies for the same function (i.e. conventional vs. novel inal draft, Writing – review & editing. Prodromos Daoutidis: Supervi-
synthesis, fuel cells vs. combustion) to contextualize the benefits and sion, Conceptualization, Writing – review & editing, Funding acquisi-
drawbacks of each. Another interesting consideration is that of fuel- tion.
flexible operation, for example, allowing time-varying composition of
Declaration of competing interest
methanol–hydrogen or ammonia–hydrogen mixtures fed to compati-
ble fuel cells or combustion systems in the scheduling optimization. The authors declare the following financial interests/personal rela-
Future research should also attempt to incorporate multiple years of tionships which may be considered as potential competing interests:
renewable generation data in capacity planning formulations in or- Matthew J. Palys reports financial support was provided by Xcel Energy
der to better understand the true value of long duration storage to Inc.
mitigate year-over-year changes in renewable energy availability. For
Acknowledgments
power-to-x sector coupling studies, optimization models with multiple
decision-makers will be crucial for realistic representation of tradi-
This work was funded by a generous gift from Xcel Energy.
tionally separate entities (e.g. utilities, chemical producers) acting in
their own self-interest. Capacity planning models incorporating the References
considerations discussed above will inevitably be larger and more
complex. Such research will thus require the continued development Abashar, M., 2018. Ultra-clean hydrogen production by ammonia decomposition. J.
King Saud Univ.-Eng. Sci. 30 (1), 2–11.
of optimization solution approaches like that presented in Allen et al. Acar, C., Beskese, A., Temur, G.T., 2021. Comparative fuel cell sustainability assessment
(2022) to improve tractability. with a novel approach. Int. J. Hydrogen Energy 47 (1), 575–594.

17
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

A.D.I. Analytics, 2017. Methanol for power generation: A white paper. Bothien, M., Breuhaus, P., Griebel, P., Kaplan, B., Laagland, G., Stuttaford, P., Kunte, P.,
https://www.adi-analytics.com/wp-content/uploads/2017/09/ADI-MTP-White- (European Turbine Network), 2020. Hydrogen gas turbines. https://www.etn.
Paper-Sep-2017-vf.pdf. global/wp-content/uploads/2020/02/ETN-Hydrogen-Gas-Turbines-report.pdf.
Afif, A., Radenahmad, N., Cheok, Q., Shams, S., Kim, J.H., Azad, A.K., 2016. Bothien, M.R., Ciani, A., Wood, J.P., Fruechtel, G., 2019. Toward decarbonized power
Ammonia-fed fuel cells: a comprehensive review. Renew. Sustain. Energy Rev. 60, generation with gas turbines by using sequential combustion for burning hydrogen.
822–835. J. Eng. Gas Turbines and Power 141 (12), 121013.
Ahluwalia, R., Hua, T., Peng, J., Kumar, R., 2010. System level analysis of hydrogen Bozzano, G., Manenti, F., 2016. Efficient methanol synthesis: Perspectives, technologies
storage options. In: Proceedings of the 2008 US DOE Hydrogen Program Annual and optimization strategies. Prog. Energy Combust. Sci. 56, 71–105.
Merit Review. Bromberg, L., Chang, W., (M.I.T. - Sloan Automotive Laboratory), 2010. Methanol as
Akal, D., Öztuna, S., Büyükakın, M.K., 2020. A review of hydrogen usage in internal an alternative transportation fuel in the US: Options for sustainable and/or energy-
combustion engines (gasoline-Lpg-diesel) from combustion performance aspect. Int. secure transportation. afdc.energy.gov/files/pdfs/mit_methanol_white_paper.pdf.
J. Hydrogen Energy 45 (60), 35257–35268. Bui, M., Adjiman, C.S., Bardow, A., Anthony, E.J., Boston, A., Brown, S., Fennell, P.S.,
Al-Kalbani, H., Xuan, J., García, S., Wang, H., 2016. Comparative energetic assessment Fuss, S., Galindo, A., Hackett, L.A., et al., 2018. Carbon capture and storage (CCS):
of methanol production from CO2: Chemical versus electrochemical process. Appl. The way forward. Energy Environ. Sci. 11 (5), 1062–1176.
Energy 165, 1–13. Campo, M., Magalhães, F., Mendes, A., 2010. Separation of nitrogen from air by carbon
Allen, R.C., Baratsas, S.G., Kakodkar, R., Avraamidou, S., Demirhan, C.D., Heuberger- molecular sieve membranes. J. Membr. Sci. 350 (1–2), 139–147.
Austin, C.F., Klokkenburg, M., Pistikopoulos, E.N., 2022. A multi-period integrated Candelaresi, D., Valente, A., Iribarren, D., Dufour, J., Spazzafumo, G., 2021. Compar-
planning and scheduling approach for developing energy systems. Optim. Control ative life cycle assessment of hydrogen-fuelled passenger cars. Int. J. Hydrogen
Appl. Methods http://dx.doi.org/10.1002/oca.2866. Energy 46 (72), 35961–35973.
Allen, R.C., Baratsas, S.G., Kakodkar, R., Avraamidou, S., Powell, J.B., Heuberger, C.F., Cano, M.H., Kelouwani, S., Agbossou, K., Dubé, Y., 2015. Power management system
Demirhan, C.D., Pistikopoulos, E.N., 2021. An optimization framework for solv- for off-grid hydrogen production based on uncertainty. Int. J. Hydrogen Energy 40
ing integrated planning and scheduling problems for dense energy carriers. (23), 7260–7272.
IFAC-PapersOnLine 54 (3), 621–626. Cappelletti, A., Martelli, F., 2017. Investigation of a pure hydrogen fueled gas turbine
Allman, A., Daoutidis, P., 2018. Optimal scheduling for wind-powered ammonia burner. Int. J. Hydrogen Energy 42 (15), 10513–10523.
generation: Effects of key design parameters. Chem. Eng. Res. Des. 131, 5–15. Carmo, M., Fritz, D.L., Mergel, J., Stolten, D., 2013. A comprehensive review on PEM
Allman, A., Palys, M.J., Daoutidis, P., 2019. Scheduling-informed optimal design for water electrolysis. Int. J. Hydrogen Energy 38 (12), 4901–4934.
systems with time-varying operation: A wind-powered ammonia production case Castaneda, M., Cano, A., Jurado, F., Sánchez, H., Fernández, L.M., 2013. Sizing
study. AIChE J. 65 (7), e16434. optimization, dynamic modeling and energy management strategies of a stand-
Andika, R., Kim, Y., Yun, C.M., Yoon, S.H., Lee, M., 2019. Design of a renewable alone PV/hydrogen/battery-based hybrid system. Int. J. Hydrogen Energy 38 (10),
energy system with battery and power-to-methanol unit. Korean J. Chem. Eng. 36 3830–3845.
(1), 12–20. Cau, G., Cocco, D., Petrollese, M., Kær, S.K., Milan, C., 2014. Energy management
Andika, R., Nandiyanto, A.B.D., Putra, Z.A., Bilad, M.R., Kim, Y., Yun, C.M., Lee, M., strategy based on short-term generation scheduling for a renewable microgrid using
2018. Co-electrolysis for power-to-methanol applications. Renew. Sustain. Energy a hydrogen storage system. Energy Convers. Manage. 87, 820–831.
Rev. 95, 227–241. Chachuat, B., Mitsos, A., Barton, P.I., 2005. Optimal design and steady-state opera-
Armijo, J., Philibert, C., 2020. Flexible production of green hydrogen and ammonia tion of micro power generation employing fuel cells. Chem. Eng. Sci. 60 (16),
from variable solar and wind energy: Case study of Chile and Argentina. Int. J. 4535–4556.
Hydrogen Energy 45 (3), 1541–1558. Chaniago, Y.D., Qyyum, M.A., Andika, R., Ali, W., Qadeer, K., Lee, M., 2019.
Arora, A., Iyer, S.S., Bajaj, I., Hasan, M.F., 2018. Optimal methanol production via Self-recuperative high temperature co-electrolysis-based methanol production with
sorption-enhanced reaction process. Ind. Eng. Chem. Res. 57 (42), 14143–14161. vortex search-based exergy efficiency enhancement. J. Cleaner Prod. 239, 118029.
Bakhtiari, H., Naghizadeh, R.A., 2018. Multi-criteria optimal sizing of hybrid renewable Chellappa, A., Fischer, C., Thomson, W., 2002. Ammonia decomposition kinetics over
energy systems including wind, photovoltaic, battery, and hydrogen storage with Ni-Pt/Al2 O3 for PEM fuel cell applications. Appl. Catalysis A: General 227 (1–2),
𝜀-constraint method. IET Renew. Power Gener. 12 (8), 883–892. 231–240.
Bañares-Alcántara, R., Dericks, G., Fiaschetti, M., Grünewald, P., Lopez, J.M., Tsang, E., Chen, C., Yang, A., 2021. Power-to-methanol: The role of process flexibility in the
Yang, A., Ye, L., Zhao, S., 2015. Analysis of islanded ammonia-based energy storage integration of variable renewable energy into chemical production. Energy Convers.
systems. University of Oxford. Manage. 228, 113673.
Bartels, J., 2008. A feasibility study of implementing an ammonia economy. Chen, C., Yang, A., Bañares Alcántara, R., 2021. Renewable methanol production: Un-
Iowa State University: https://dr.lib.iastate.edu/entities/publication/0d44bb27- derstanding the interplay between storage sizing, renewable mix and dispatchable
0317-4e6f-901c-3d709853a435. energy price. Adv. Appl. Energy 2, 100021.
Barthélémy, H., Weber, M., Barbier, F., 2017. Hydrogen storage: recent improvements Chen, Y., Zhang, B., Su, Y., Sui, C., Zhang, J., 2022. Effect and mechanism of
and industrial perspectives. Int. J. Hydrogen Energy 42 (11), 7254–7262. combustion enhancement and emission reduction for non-premixed pure ammonia
Beach, J., Kinter, J., Welch, A., (Starfire Energy), 2018. Rapid ramp reactor up- combustion based on fuel preheating. Fuel 308, 122017.
date. https://www.nh3fuelassociation.org/2018/12/08/rapid-ramp-nh3-prototype- Chiuta, S., Bessarabov, D.G., 2018. Design and operation of an ammonia-fueled
reactor-update/. microchannel reactor for autothermal hydrogen production. Catalysis Today 310,
Beerbühl, S.S., Fröhling, M., Schultmann, F., 2015. Combined scheduling and capacity 187–194.
planning of electricity-based ammonia production to integrate renewable energies. Chudnovsky, B., Reshef, M., Keren, M., Baitel, S., 2015. Successful implementation of
European J. Oper. Res. 241 (3), 851–862. methanol firing at 50 MW gas turbine for long term operation. In: ASME Power
Bergström, K., Nordin, H., Königstein, A., Marriott, C., Wiles, M., 2007. Abc - alcohol Conference. Vol. 56604, American Society of Mechanical Engineers, V001T03A015.
based combustion engines. Challenges and opportunities. In: Aachen Colloquium Cinti, G., Discepoli, G., Bidini, G., Lanzini, A., Santarelli, M., 2016. Co-electrolysis of
Automobile and Engine Technology. Vol. 16, American Society of Mechanical water and CO2 in a solid oxide electrolyzer (SOE) stack. Int. J. Energy Res. 40 (2),
Engineers, pp. 1031–1071. 207–215.
Bertau, M., Offermanns, H., Plass, L., Schmidt, F., Wernicke, H.-J., 2014. Methanol: Cui, X., Kær, S.K., Nielsen, M.P., 2022. Energy analysis and surrogate modeling for the
The Basic Chemical and Energy Feedstock of the Future. Vol. 1, Springer. green methanol production under dynamic operating conditions. Fuel 307, 121924.
Bertuccioli, L., Chan, A., Hart, D., Lehner, F., Madden, B., Standen, E., 2014. Study d’Amore, F., Bezzo, F., 2020. Optimizing the design of supply chains for carbon capture,
on development of water electrolysis in the EU. https://www.fch.europa.eu/sites/ utilization, and sequestration in Europe: A preliminary assessment. Front. Energy
default/files/study%20electrolyser_0.pdf. Res. 190. http://dx.doi.org/10.3389/fenrg.2020.00190.
Bique, A.O., Nguyen, T.B., Leonzio, G., Galanopoulos, C., Zondervan, E., 2018. Integra- Dang, S., Yang, H., Gao, P., Wang, H., Li, X., Wei, W., Sun, Y., 2019. A review of
tion of carbon dioxide and hydrogen supply chains. In: Computer Aided Chemical research progress on heterogeneous catalysts for methanol synthesis from carbon
Engineering. Vol. 43, Elsevier, pp. 1413–1418. dioxide hydrogenation. Catalysis Today 330, 61–75.
Bocker, N., Grahl, M., Tota, A., Häussinger, P., Leitgeb, P., Schmücker, B., 2013. David, W.I., Makepeace, J.W., Callear, S.K., Hunter, H.M., Taylor, J.D., Wood, T.J.,
Nitrogen. In: Ullmann’s Encyclopedia of Industrial Chemistry. John Wiley & Sons, Jones, M.O., 2014. Hydrogen production from ammonia using sodium amide. J.
Ltd, pp. 1–27. http://dx.doi.org/10.1002/14356007.a17_457.pub2. Am. Chem. Soc. 136 (38), 13082–13085.
Bockris, J.O.M., 2013. The hydrogen economy: Its history. Int. J. Hydrogen Energy 38 Demirci, U.B., 2007. Direct liquid-feed fuel cells: Thermodynamic and environmental
(6), 2579–2588. concerns. J. Power Sources 169 (2), 239–246.
Bødal, E.F., Mallapragada, D., Botterud, A., Korpås, M., 2020. Decarbonization synergies Demirhan, C.D., Tso, W.W., Powell, J.B., Heuberger, C.F., Pistikopoulos, E.N., 2020. A
from joint planning of electricity and hydrogen production: A Texas case study. Int. multiscale energy systems engineering approach for renewable power generation
J. Hydrogen Energy 45 (58), 32899–32915. and storage optimization. Ind. Eng. Chem. Res. 59 (16), 7706–7721.
Boretti, A.A., 2012. Novel heavy duty engine concept for operation dual fuel H2 –NH3 . Demirhan, C.D., Tso, W.W., Powell, J.B., Pistikopoulos, E.N., 2019. Sustainable ammo-
Int. J. Hydrogen Energy 37 (9), 7869–7876. nia production through process synthesis and global optimization. AIChE J. 65 (7),
Borisut, P., Nuchitprasittichai, A., 2020. Process configuration studies of methanol pro- e16498.
duction via carbon dioxide hydrogenation: Process simulation-based optimization Dias, V., Pochet, M., Contino, F., Jeanmart, H., 2020. Energy and economic costs of
using artificial neural networks. Energies 13 (24), 6608. chemical storage. Front. Mech. Eng. 6, 21.

18
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

Dieterich, V., Buttler, A., Hanel, A., Spliethoff, H., Fendt, S., 2020. Power-to-liquid via Hongsirikarn, K., Goodwin Jr., J.G., Greenway, S., Creager, S., 2010. Influence of
synthesis of methanol, DME or Fischer–Tropsch-fuels: A review. Energy Environ. ammonia on the conductivity of Nafion membranes. J. Power Sources 195 (1),
Sci. 13 (10), 3207–3252. 30–38.
Dufo-López, R., Bernal-Agustín, J.L., 2008. Multi-objective design of PV–wind–diesel– Huesman, A., 2020. Integration of operation and design of solar fuel plants: A carbon
hydrogen–battery systems. Renew. Energy 33 (12), 2559–2572. dioxide to methanol case study. Comput. Chem. Eng. 140, 106836.
Eigen, N., Keller, C., Dornheim, M., Klassen, T., Bormann, R., 2007. Industrial Hunter, H.M., Makepeace, J.W., Wood, T.J., Mylius, O.S., Kibble, M.G., Nutter, J.B.,
production of light metal hydrides for hydrogen storage. Scr. Mater. 56 (10), Jones, M.O., David, W.I., 2016. Demonstrating hydrogen production from ammonia
847–851. using lithium imide–powering a small proton exchange membrane fuel cell. J.
Erdinc, O., Uzunoglu, M., 2012. A new perspective in optimum sizing of hybrid Power Sources 329, 138–147.
renewable energy systems: Consideration of component performance degradation Ikäheimo, J., Kiviluoma, J., Weiss, R., Holttinen, H., 2018. Power-to-ammonia in future
issue. Int. J. Hydrogen Energy 37 (14), 10479–10488. North European 100% renewable power and heat system. Int. J. Hydrogen Energy
European Turbine Network, 2019. The gas turbine industry commitments to drive 43 (36), 17295–17308.
Europe’s transition to a decarbonised energy mix. https://www.euturbines.eu/wp- Ilbas, M., Crayford, A., Yılmaz, I., Bowen, P., Syred, N., 2006. Laminar-burning
content/uploads/2020/11/EUTurbines_press_release_on_the_Commitments.pdf. velocities of hydrogen–air and hydrogen–methane–air mixtures: An experimental
Farhad, S., Hamdullahpur, F., 2010. Conceptual design of a novel ammonia-fuelled study. Int. J. Hydrogen Energy 31 (12), 1768–1779.
portable solid oxide fuel cell system. J. Power Sources 195 (10), 3084–3090. Intergovernmental Panel on Climate Change, 2018. Special report: Global warming of
Fasihi, M., Weiss, R., Savolainen, J., Breyer, C., 2021. Global potential of green 1.5◦ C - Summary for policymakers. https://www.ipcc.ch/sr15/chapter/spm/.
ammonia based on hybrid PV-wind power plants. Appl. Energy 294, 116170. International Energy Agency, 2021. Global electric vehicle outlook 2020. https://www.
Frigo, S., Gentili, R., 2013. Analysis of the behaviour of a 4-stroke Si engine fuelled iea.org/reports/global-ev-outlook-2020.
with ammonia and hydrogen. Int. J. Hydrogen Energy 38 (3), 1607–1615. International Energy Agency, 2022a. Global energy review 2021. https://www.iea.org/
Gabrielli, P., Fürer, F., Mavromatidis, G., Mazzotti, M., 2019. Robust and optimal design reports/global-energy-review-2021/renewables.
of multi-energy systems with seasonal storage through uncertainty analysis. Appl. International Energy Agency, 2022b. Hydrogen. https://www.iea.org/reports/hydrogen.
Energy 238, 1192–1210. Iulianelli, A., Ribeirinha, P., Mendes, A., Basile, A., 2014. Methanol steam reforming for
Garagounis, I., Kyriakou, V., Skodra, A., Vasileiou, E., Stoukides, M., 2014. Electro- hydrogen generation via conventional and membrane reactors: A review. Renew.
chemical synthesis of ammonia in solid electrolyte cells. Front. Energy Res. 2, Sustain. Energy Rev. 29, 355–368.
1. Ivanova, S., Lewis, R., 2012. Producing nitrogen via pressure swing adsorption. Chem.
Garcia, G., Arriola, E., Chen, W.-H., De Luna, M.D., 2021. A comprehensive review of Eng. Progress 108 (6), 38–42.
hydrogen production from methanol thermochemical conversion for sustainability. Jallouli, R., Krichen, L., 2012. Sizing, techno-economic and generation management
Energy 217, 119384. analysis of a stand alone photovoltaic power unit including storage devices. Energy
Garcia, P., Torreglosa, J.P., Fernandez, L.M., Jurado, F., 2013. Optimal energy manage- 40 (1), 196–209.
ment system for stand-alone wind turbine/photovoltaic/hydrogen/battery hybrid James, B.D., Houchins, C., (Strategic Analysis Inc.), 2016. 700 Bar type IV H2 pressure
system with supervisory control based on fuzzy logic. Int. J. Hydrogen Energy 38 vessel cost projections. https://energy.gov/sites/default/files/2016/09/f33/fcto_h2_
(33), 14146–14158. storage_700bar_workshop_2_james.pdf.
Garcia, R.S., Weisser, D., 2006. A wind–diesel system with hydrogen storage: Joint Joo, O.-S., Jung, K.-D., Moon, I., Rozovskii, A.Y., Lin, G.I., Han, S.-H., Uhm, S.-J.,
optimisation of design and dispatch. Renew. Energy 31 (14), 2296–2320. 1999. Carbon dioxide hydrogenation to form methanol via a reverse-water-gas-shift
Garcia-Torres, F., Bordons, C., 2015. Optimal economical schedule of hydrogen-based reaction (the CAMERE process). Ind. Eng. Chem. Res. 38 (5), 1808–1812.
microgrids with hybrid storage using model predictive control. IEEE Trans. Ind. Kalinci, Y., Hepbasli, A., Dincer, I., 2015. Techno-economic analysis of a stand-alone
Electron. 62 (8), 5195–5207. hybrid renewable energy system with hydrogen production and storage options.
GE, 2001. Position paper: Feasibility of methanol as gas turbine fuel. https: Int. J. Hydrogen Energy 40 (24), 7652–7664.
//www.cmsgx.com/wp-content/uploads/2015/12/GE-White-Paper-Feasibility- Kane, S.P., Northrop, W.F., 2021. Thermochemical recuperation to enable efficient
of-Methanol-as-a-gas-turnbine-fuel.pdf. ammonia-diesel dual-fuel combustion in a compression ignition engine. Energies
Ghosh, S., Seethamraju, S., 2019. Feasibility of reactive distillation for methanol 14 (22), 7540.
synthesis. Chem. Eng. Process.-Process Intensif. 145, 107673. Kastens, M.L., Dudley, J., Troeltzsch, J., 1948. Synthetic methanol production. Ind. Eng.
Giddey, S., Badwal, S., Munnings, C., Dolan, M., 2017. Ammonia as a renewable energy Chem. 40, 2230–2240.
transportation media. ACS Sustain. Chem. Eng. 5 (11), 10231–10239. Kaviani, A.K., Riahy, G., Kouhsari, S.M., 2009. Optimal design of a reliable hydrogen-
Goldmeer, J., (G.E. Power), 2019. Power to gas: Hydrogen for power generation. based stand-alone wind/PV generating system, considering component outages.
https://www.ge.com/content/dam/gepower/global/en_US/documents/fuel- Renew. Energy 34 (11), 2380–2390.
flexibility/GEA33861%20Power%20to%20Gas%20-%20Hydrogen%20for% Kenkel, P., Wassermann, T., Rose, C., Zondervan, E., 2021. A generic super-
20Power%20Generation.pdf. structure modeling and optimization framework on the example of bi-criteria
Gorbe, J., Lasobras, J., Francés, E., Herguido, J., Menéndez, M., Kumakiri, I., Kita, H., Power-to-Methanol process design. Comput. Chem. Eng. 150, 107327.
2018. Preliminary study on the feasibility of using a zeolite a membrane in a Khademi, M.H., Sabbaghi, R.S., 2017. Comparison between three types of ammonia
membrane reactor for methanol production. Sep. Purif. Technol. 200, 164–168. synthesis reactor configurations in terms of cooling methods. Chem. Eng. Res. Des.
Gottesfeld, S., 2018. The direct ammonia fuel cell and a common pattern of 128, 306–317.
electrocatalytic processes. J. Electrochem. Soc. 165 (15), J3405. Kilner, J.A., Skinner, S.J., Irvine, S., Edwards, P., 2012. Functional Materials for
Guerra, O.J., 2021. Beyond short-duration energy storage. Nat. Energy 6 (5), 460–461. Sustainable Energy Applications. Elsevier.
Guil-López, R., Mota, N., Llorente, J., Millán, E., Pawelec, B., Fierro, J.L.G., Navarro, R., Kim, J., Kwon, O., 2011. A micro reforming system integrated with a heat-recirculating
2019. Methanol synthesis from CO2 : A review of the latest developments in micro-combustor to produce hydrogen from ammonia. Int. J. Hydrogen Energy 36
heterogeneous catalysis. Materials 12 (23), 3902. (3), 1974–1983.
Hagen, D.L., 1977. Methanol as a fuel: A review with bibliography. SAE Trans. Kirubakaran, A., Jain, S., Nema, R., 2009. A review on fuel cell technologies and power
2764–2796. electronic interface. Renew. Sustain. Energy Rev. 13 (9), 2430–2440.
Haid, J., Koss, U., 2001. Lurgi’s Mega-Methanol technology opens the door for a new Klier, K., Chatikavanij, V., Herman, R., Simmons, G., 1982. Catalytic synthesis of
era in down-stream applications. In: Studies in Surface Science and Catalysis. Vol. methanol from CO2 : IV. The effects of carbon dioxide. J. Catalysis 74 (2), 343–360.
136, Elsevier, pp. 399–404. Kobayashi, H., Hayakawa, A., Somarathne, K.K.A., Okafor, E.C., 2019. Science and
Halseid, R., Vie, P.J., Tunold, R., 2006. Effect of ammonia on the performance of technology of ammonia combustion. Proc. Combust. Inst. 37 (1), 109–133.
polymer electrolyte membrane fuel cells. J. Power Sources 154 (2), 343–350. Kotzur, L., Markewitz, P., Robinius, M., Stolten, D., 2018. Impact of different time
Han, S., Kim, J., 2019. A multi-period MILP model for the investment and design series aggregation methods on optimal energy system design. Renew. Energy 117,
planning of a national-level complex renewable energy supply system. Renew. 474–487.
Energy 141, 736–750. Kulikovsky, A., 2008. Direct methanol–hydrogen fuel cell: The mechanism of
Hansen, J., Hojlund, P., 2008. Methanol synthesis. In: Handbook of Heterogeneous functioning. Electrochem. Commun. 10 (9), 1415–1418.
Catalysis. Nielsen. Lababidi, H., 2000. Air separation by polysulfone hollow fibre membrane permeators
Hauch, A., Küngas, R., Blennow, P., Hansen, A.B., Hansen, J.B., Mathiesen, B.V., Mo- in series: Experimental and simulation results. Chem. Eng. Res. Des. 78 (8),
gensen, M.B., 2020. Recent advances in solid oxide cell technology for electrolysis. 1066–1076.
Science 370 (6513). Lacko, R., Drobnič, B., Mori, M., Sekavčnik, M., Vidmar, M., 2014. Stand-alone
He, G., Mallapragada, D.S., Bose, A., Heuberger, C.F., Gençer, E., 2021. Hydrogen renewable combined heat and power system with hydrogen technologies for
supply chain planning with flexible transmission and storage scheduling. IEEE household application. Energy 77, 164–170.
Trans. Sustain. Energy. Laguna-Bercero, M., 2012. Recent advances in high temperature electrolysis using solid
Heras, J., Martín, M., 2021. Multiscale analysis for power-to-gas-to-power facilities oxide fuel cells: A review. J. Power Sources 203, 4–16.
based on energy storage. Comput. Chem. Eng. 144, 107147. Lamb, K.E., Dolan, M.D., Kennedy, D.F., 2019. Ammonia for hydrogen storage; A review
Hobson, C., Marquez, C., (Methanol Institute), 2018. Renewable methanol report. https: of catalytic ammonia decomposition and hydrogen separation and purification. Int.
//www.enerkem.com/wp-content/uploads/2019/01/MetInst_MethanolReport.pdf. J. Hydrogen Energy 44 (7), 3580–3593.

19
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

Lehner, M., Tichler, R., Steinmüller, H., Koppe, M., 2014. Power-to-gas: Technology Nguyen, D.-K., Van Craeynest, T., Pillu, T., Coulier, J., Verhelst, S., 2018. Downsizing
and Business Models. Springer. potential of methanol fueled DISI engine with variable valve timing and boost
Leng, H., Ichikawa, T., Hino, S., Hanada, N., Isobe, S., Fujii, H., 2006. Synthesis and control. In: WCX™ 18: SAE World Congress Experience. SAE International.
decomposition reactions of metal amides in metal–N–H hydrogen storage system. Ogbe, E., Mukherjee, U., Fowler, M., Almansoori, A., Elkamel, A., 2019. Integrated
J. Power Sources 156 (2), 166–170. design and operation optimization of hydrogen commingled with natural gas in
Léonard, G., Giulini, D., Villarreal-Singer, D., 2016. Design and evaluation of a high- pipeline networks. Ind. Eng. Chem. Res. 59 (4), 1584–1595.
density energy storage route with CO2 re-use, water electrolysis and methanol Okanishi, T., Okura, K., Srifa, A., Muroyama, H., Matsui, T., Kishimoto, M., Saito, M.,
synthesis. In: Computer Aided Chemical Engineering. Vol. 38, Elsevier, pp. Iwai, H., Yoshida, H., Saito, M., et al., 2017. Comparative study of ammonia-fueled
1797–1802. solid oxide fuel cell systems. Fuel Cells 17 (3), 383–390.
Licht, S., Cui, B., Wang, B., Li, F.-F., Lau, J., Liu, S., 2014. Ammonia synthesis by N2 Oryshchyn, D., Harun, N.F., Tucker, D., Bryden, K.M., Shadle, L., 2018. Fuel utilization
and steam electrolysis in molten hydroxide suspensions of nanoscale Fe2 O3 . Science effects on system efficiency in solid oxide fuel cell gas turbine hybrid systems.
345 (6197), 637–640. Appl. Energy 228, 1953–1965.
Lin, B., Wiesner, T., Malmali, M., 2020. Performance of a small-scale haber process: A Osman, O., Sgouridis, S., Sleptchenko, A., 2020. Scaling the production of renewable
techno-economic analysis. ACS Sustain. Chem. Eng. 8 (41), 15517–15531. ammonia: A techno-economic optimization applied in regions with high insolation.
Lipman, T.E., Elke, M., Lidicker, J., 2018. Hydrogen fuel cell electric vehicle perfor- J. Cleaner Prod. 271, 121627.
mance and user-response assessment: Results of an extended driver study. Int. J. Ozarslan, A., 2012. Large-scale hydrogen energy storage in salt caverns. Int. J.
Hydrogen Energy 43 (27), 12442–12454. Hydrogen Energy 37 (19), 14265–14277.
Palys, M.J., Daoutidis, P., 2020. Using hydrogen and ammonia for renewable energy
Liu, H., 2014. Ammonia synthesis catalyst 100 years: Practice, enlightenment and
storage: A geographically comprehensive techno-economic study. Comput. Chem.
challenge. Chin. J. Catalysis 35 (10), 1619–1640.
Eng. 136, 106785.
Lonis, F., Tola, V., Cau, G., 2019. Renewable methanol production and use through
Palys, M.J., Kuznetsov, A., Tallaksen, J., Reese, M., Daoutidis, P., 2019. A novel
reversible solid oxide cells and recycled CO2 hydrogenation. Fuel 246, 500–515.
system for ammonia-based sustainable energy and agriculture: Concept and design
Lucero, J.M., Crawford, J.M., Wolden, C.A., Carreon, M.A., 2021. Tunability of
optimization. Chem. Eng. Process.-Process Intensif. 140, 11–21.
ammonia adsorption over NaP zeolite. Microporous Mesoporous Mater. 324,
Palys, M.J., McCormick, A., Cussler, E., Daoutidis, P., 2018. Modeling and optimal
111288.
design of absorbent enhanced ammonia synthesis. Processes 6, 91.
Luo, Q., Li, J., Li, B., Liu, B., Shao, H., Li, Q., 2019. Kinetics in Mg-based hydrogen
Palys, M.J., Mitrai, I., Daoutidis, P., 2021. Renewable hydrogen and ammonia for
storage materials: Enhancement and mechanism. J. Magnes. Alloys 7 (1), 58–71.
combined heat and power systems in remote locations: Optimal design and
MacFarlane, D.R., Cherepanov, P.V., Choi, J., Suryanto, B.H., Hodgetts, R.Y., scheduling. Optim. Control Appl. Methods http://dx.doi.org/10.1002/oca.2793.
Bakker, J.M., Vallana, F.M.F., Simonov, A.N., 2020. A roadmap to the ammonia Patil, A., Laumans, L., Vrijenhoef, H., 2014. Solar to ammonia–via Proton’s NFuel units.
economy. Joule 4 (6), 1186–1205. Procedia Eng. 83, 322–327.
Mallapragada, D.S., Gençer, E., Insinger, P., Keith, D.W., O’Sullivan, F.M., 2020. Can Pattabathula, V., Richardson, J., 2016. Introduction to ammonia production. CEP
industrial-scale solar hydrogen supplied from commodity technologies be cost Magazine 2, 69–75.
competitive by 2030? Cell Rep. Phys. Sci. 1 (9), 100174. Pereira, M., Limon, D., de la Peña, D.M., Valverde, L., Alamo, T., 2015. Periodic
Malmali, M., Wei, Y., McCormick, A., Cussler, E.L., 2016. Ammonia synthesis at reduced economic control of a nonisolated microgrid. IEEE Trans. Ind. Electron. 62 (8),
pressure via reactive separation. Ind. Eng. Chem. Res. 55 (33), 8922–8932. 5247–5255.
Marocco, P., Ferrero, D., Martelli, E., Santarelli, M., Lanzini, A., 2021. An MILP Pontzen, F., Liebner, W., Gronemann, V., Rothaemel, M., Ahlers, B., 2011. CO2-based
approach for the optimal design of renewable battery-hydrogen energy systems methanol and DME–efficient technologies for industrial scale production. Catalysis
for off-grid insular communities. Energy Convers. Manage. 245, 114564. Today 171 (1), 242–250.
Martín, M., 2016. Methodology for solar and wind energy chemical storage facilities Räuchle, K., Plass, L., Wernicke, H.-J., Bertau, M., 2016. Methanol for renewable energy
design under uncertainty: Methanol production from CO2 and hydrogen. Comput. storage and utilization. Energy Technol. 4 (1), 193–200.
Chem. Eng. 92, 43–54. Reiter, A.J., Kong, S.-C., 2011. Combustion and emissions characteristics of
Martín, M., Grossmann, I.E., 2017. Optimal integration of a self sustained algae based compression-ignition engine using dual ammonia-diesel fuel. Fuel 90 (1), 87–97.
facility with solar and/or wind energy. J. Cleaner Prod. 145, 336–347. Rivarolo, M., Bellotti, D., Magistri, L., Massardo, A., 2016. Feasibility study of methanol
Mbatha, S., Everson, R., Musyoka, N.M., Langmi, H.W., Lanzini, A., Brilman, W., production from different renewable sources and thermo-economic analysis. Int. J.
2021. Power-to-methanol process: A review of electrolysis, methanol catalysts, Hydrogen Energy 41 (4), 2105–2116.
kinetics, reactor designs and modelling, process integration, optimisation, and Rosowski, F., Hornung, A., Hinrichsen, O., Herein, D., Muhler, M., Ertl, G., 1997.
techno-economics. Sustain. Energy Fuels. Ruthenium catalysts for ammonia synthesis at high pressures: Preparation,
Mehrjerdi, H., 2020. Peer-to-peer home energy management incorporating hydrogen characterization, and power-law kinetics. Appl. Catalysis A: General 151 (2),
storage system and solar generating units. Renew. Energy 156, 183–192. 443–460.
Mekhilef, S., Saidur, R., Safari, A., 2012. Comparative study of different fuel cell Rullo, P., Braccia, L., Luppi, P., Zumoffen, D., Feroldi, D., 2019. Integration of sizing and
technologies. Renew. Sustain. Energy Rev. 16 (1), 981–989. energy management based on economic predictive control for standalone hybrid
Mendes, P.R., Isorna, L.V., Bordons, C., Normey-Rico, J.E., 2016. Energy management renewable energy systems. Renew. Energy 140, 436–451.
of an experimental microgrid coupled to a V2G system. J. Power Sources 327, Sahibzada, M., Metcalfe, I., Chadwick, D., 1998. Methanol synthesis from CO/CO2 /H2
702–713. over Cu/ZnO/Al2 O3 at differential and finite conversions. J. Catalysis 174 (2),
Miller, J., Luyben, W.L., Belanger, P., Blouin, S., Megan, L., 2008. Improving agility of 111–118.
Sakintuna, B., Lamari-Darkrim, F., Hirscher, M., 2007. Metal hydride materials for solid
cryogenic air separation plants. Ind. Eng. Chem. Res. 47 (2), 394–404.
hydrogen storage: a review. Int. J. Hydrogen Energy 32 (9), 1121–1140.
Mogensen, M., Chen, M., Frandsen, H., Graves, C., Hansen, J., Hansen, K., Hauch, A.,
Salmon, N., Bañares-Alcántara, R., 2021. Impact of grid connectivity on cost and
Jacobsen, T., Jensen, S., Skafte, T., et al., 2019. Reversible solid-oxide cells for
location of green ammonia production: Australia as a case study. Energy Environ.
clean and sustainable energy. Clean Energy 3 (3), 175–201.
Sci. 14 (12), 6655–6671.
Monjur, M.S., Hasan, M.F., 2021. Computer-aided process intensification of natural gas
Samsatli, S., Samsatli, N.J., 2019. The role of renewable hydrogen and inter-seasonal
to methanol process. AIChE J. e17622.
storage in decarbonising heat–comprehensive optimisation of future renewable
Mørch, C.S., Bjerre, A., Gøttrup, M.P., Sorenson, S.C., Schramm, J., 2011. Ammo-
energy value chains. Appl. Energy 233, 854–893.
nia/hydrogen mixtures in an SI-engine: Engine performance and analysis of a
Sánchez, A., Castellano, E., Martín, M., Vega, P., 2021a. Evaluating ammonia as
proposed fuel system. Fuel 90 (2), 854–864.
green fuel for power generation: A thermo-chemical perspective. Appl. Energy 293,
Morgan, E.R., 2013. Techno-economic feasibility study of ammonia plants powered by
116956.
offshore wind. (Ph.D. thesis). University of Massachusetts Amherst. Sánchez, A., Martín, M., 2018a. Optimal renewable production of ammonia from water
Motyka, T., (Savannah River National Laboratory), 2015. Metal hydrides. https://www. and air. J. Cleaner Prod. 178, 325–342.
energy.gov/sites/default/files/2015/02/f19/fcto_h2_storage_summit_motyka.pdf. Sánchez, A., Martín, M., 2018b. Scale up and scale down issues of renewable ammonia
Müller, M., Stolten, D., (Forschungszentrum Jülich), 2011. Competitiveness of plants: Towards modular design. Sustain. Prod. Consum. 16, 176–192.
DMFC-systems. https://www.fz-juelich.de/SharedDocs/Downloads/IEK/IEK- Sánchez, A., Martín, M., Zhang, Q., 2021b. Optimal design of sustainable power-to-fuels
3/Flyer/Wirtschaftlichkeitsbetrachtung_DMFC%20(E).pdf?__blob=publicationFile. supply chains for seasonal energy storage. Energy 234, 121300.
Murray, R., Furlonge, H., 2009. Market and economic assessment of using methanol Sánchez, A., Zhang, Q., Martín, M., Vega, P., 2022. Towards a new renewable power
for power generation in the Caribbean region. J. Assoc. Prof. Eng. Trinidad Tobago system using energy storage: An economic and social analysis. Energy Convers.
38 (1), 88–99. Manage. 252, 115056.
Nair, U.R., Costa-Castelló, R., 2020. A model predictive control-based energy man- Schmidt, O., Gambhir, A., Staffell, I., Hawkes, A., Nelson, J., Few, S., 2017. Future cost
agement scheme for hybrid storage system in islanded microgrids. IEEE Access 8, and performance of water electrolysis: An expert elicitation study. Int. J. Hydrogen
97809–97822. Energy 42 (52), 30470–30492.
Nayak-Luke, R., Bañares-Alcántara, R., Wilkinson, I., 2018. ‘‘Green’’ ammonia: Impact Sepulveda, N.A., Jenkins, J.D., Edington, A., Mallapragada, D.S., Lester, R.K., 2021.
of renewable energy intermittency on plant sizing and levelized cost of ammonia. The design space for long-duration energy storage in decarbonized power systems.
Ind. Eng. Chem. Res. 57 (43), 14607–14616. Nat. Energy 6 (5), 506–516.

20
M.J. Palys and P. Daoutidis Computers and Chemical Engineering 165 (2022) 107948

Shabani, B., Andrews, J., Watkins, S., 2010. Energy and cost analysis of a solar- Vasallo, M.J., Bravo, J.M., Andújar, J.M., 2013. Optimal sizing for UPS systems based
hydrogen combined heat and power system for remote power supply using a on batteries and/or fuel cell. Appl. Energy 105, 170–181.
computer simulation. Sol. Energy 84 (1), 144–155. Verhelst, S., Turner, J.W., Sileghem, L., Vancoillie, J., 2019. Methanol as a fuel for
Shipman, M.A., Symes, M.D., 2017. Recent progress towards the electrosynthesis of internal combustion engines. Prog. Energy Combust. Sci. 70, 43–88.
ammonia from sustainable resources. Catalysis Today 286, 57–68. Von KleinSmid, W., Schreiber, H., Klapatch, R., 1981. Methanol combustion in a 26-MW
Shmuli, L., (D.O.R. Group), 2021. M-100 technology: Using methanol gas turbine. In: Turbo Expo: Power for Land, Sea, and Air. Vol. 79634, American
as a clean alternative fuel. https://www.dor-group.com/db/content/ Society of Mechanical Engineers, V003T10A007.
f15497c7e1bcfeba8c985e8b42a49b16.pdf?langid=2. Wagner, K., Malmali, M., Smith, C., McCormick, A., Cussler, E., Zhu, M., Seaton, N.C.,
Siddiqui, O., Dincer, I., 2018. A review and comparative assessment of direct ammonia 2017. Column absorption for reproducible cyclic separation in small scale ammonia
fuel cells. Therm. Sci. Eng. Prog. 5, 568–578. synthesis. AIChE J. 63 (7), 3058–3068.
Siemens, 2020. Power-to-x: The crucial business on the way to a carbon-free world. Wang, H., Daoutidis, P., Zhang, Q., 2021a. Harnessing the wind power of the ocean
https://www.siemens-energy.com/global/en/offerings/technical-papers/download- with green offshore ammonia. ACS Sustain. Chem. Eng. 9 (43), 14605–14617.
power-to-x.html. Wang, G., Mitsos, A., Marquardt, W., 2017. Conceptual design of ammonia-based energy
Sivaram, V., Dabiri, J.O., Hart, D.M., 2018. The need for continued innovation in solar, storage system: System design and time-invariant performance. AIChE J. 63 (5),
wind, and energy storage. Joule 2 (9), 1639–1642. 1620–1637.
Smith, C., Hill, A.K., Torrente-Murciano, L., 2020. Current and future role of Haber– Wang, G., Mitsos, A., Marquardt, W., 2020. Renewable production of ammonia and
Bosch ammonia in a carbon-free energy landscape. Energy Environ. Sci. 13 (2), nitric acid. AIChE J. 66 (6), e16947.
331–344. Wang, H., Palys, M., Daoutidis, P., Zhang, Q., 2021b. Optimal design of sustain-
Smith, A., Klosek, J., 2001. A review of air separation technologies and their integration able ammonia-based food–energy–water systems with nitrogen management. ACS
with energy conversion processes. Fuel Process. Technol. 70 (2), 115–134. Sustain. Chem. Eng. 9 (7), 2816–2834.
Smith, C., Torrente-Murciano, L., 2021. Exceeding single-pass equilibrium with Wang, C., Zhou, S., Hong, X., Qiu, T., Wang, S., 2005. A comprehensive comparison of
integrated absorption separation for ammonia synthesis using renewable fuel options for fuel cell vehicles in China. Fuel Process. Technol. 86 (7), 831–845.
energy—Redefining the Haber-Bosch loop. Adv. Energy Mater. 11 (13), 2003845. Weir, W., VonKleinSmid, W., Danko, E., (Southern California Edison Company), 1981.
Solomon, B.D., Banerjee, A., 2006. A global survey of hydrogen energy research, Test and evaluation of methanol in a gas turbine system. https://www.osti.gov/
development and policy. Energy Policy 34 (7), 781–792. servlets/purl/6618572.
Stefan, E., Talic, B., Larring, Y., Gruber, A., Peters, T.A., 2021. Materials challenges White, C., Steeper, R., Lutz, A., 2006. The hydrogen-fueled internal combustion engine:
in hydrogen-fuelled gas turbines. Int. Mater. Rev. http://dx.doi.org/10.1080/ A technical review. Int. J. Hydrogen Energy 31 (10), 1292–1305.
09506608.2021.1981706. Wilson, M.S., 2009. Methanol decomposition fuel processor for portable power
Step, G.K., Petrovichev, M., 2002. Pressure swing adsorption for air separation and applications. Int. J. Hydrogen Energy 34 (7), 2955–2964.
purification. Chem. Pet. Eng. 38 (3), 154–158. Xu, D., Zhou, B., Wu, Q., Chung, C.Y., Li, C., Huang, S., Chen, S., 2020. Integrated
Sterner, M., Specht, M., 2021. Power-to-gas and power-to-X - The history and results modelling and enhanced utilization of power-to-ammonia for high renewable
of developing a new storage concept. Energies 14 (20), 6594. penetrated multi-energy systems. IEEE Trans. Power Syst. 35 (6), 4769–4780.
Sundén, B., 2019. Hydrogen, Batteries and Fuel Cells. Academic Press. Yang, J., Weng, W., Xiao, W., 2020. Electrochemical synthesis of ammonia in molten
Suryanto, B.H., Du, H.-L., Wang, D., Chen, J., Simonov, A.N., MacFarlane, D.R., salts. J. Energy Chem. 43, 195–207.
2019. Challenges and prospects in the catalysis of electroreduction of nitrogen to Yanxing, Z., Maoqiong, G., Yuan, Z., Xueqiang, D., Jun, S., 2019. Thermodynamics anal-
ammonia. Nat. Catalysis 2 (4), 290–296. ysis of hydrogen storage based on compressed gaseous hydrogen, liquid hydrogen
Suryanto, B.H., Kang, C.S., Wang, D., Xiao, C., Zhou, F., Azofra, L.M., Cavallo, L., and cryo-compressed hydrogen. Int. J. Hydrogen Energy 44 (31), 16833–16840.
Zhang, X., MacFarlane, D.R., 2018. Rational electrode–electrolyte design for effi- Yapicioglu, A., Dincer, I., 2019. A review on clean ammonia as a potential fuel for
cient ammonia electrosynthesis under ambient conditions. ACS Energy Lett. 3 (6), power generators. Renew. Sustain. Energy Rev. 103, 96–108.
1219–1224. Yue, M., Lambert, H., Pahon, E., Roche, R., Jemei, S., Hissel, D., 2021. Hydrogen energy
Svanberg, M., Ellis, J., Lundgren, J., Landälv, I., 2018. Renewable methanol as a fuel systems: A critical review of technologies, applications, trends and challenges.
for the shipping industry. Renew. Sustain. Energy Rev. 94, 1217–1228. Renew. Sustain. Energy Rev. 146, 111180.
Terreni, J., Trottmann, M., Franken, T., Heel, A., Borgschulte, A., 2019. Yuen, P.K.P., Villaire, W., Beckett, J., 2010. Automotive Materials Engineering Chal-
Sorption-enhanced methanol synthesis. Energy Technol. 7 (4), 1801093. lenges and Solutions for the Use of Ethanol and Methanol Blended Fuels. SAE
Torreglosa, J., García, P., Fernández, L., Jurado, F., 2014. Hierarchical energy manage- Technical Paper, https://www.sae.org/publications/technical-papers/content/2010-
ment system for stand-alone hybrid system based on generation costs and cascade 01-0729/.
control. Energy Convers. Manage. 77, 514–526. Zeng, K., Zhang, D., 2010. Recent progress in alkaline water electrolysis for hydrogen
Torreglosa, J.P., García, P., Fernández, L.M., Jurado, F., 2015. Energy dispatching based production and applications. Prog. Energy Combust. Sci. 36 (3), 307–326.
on predictive controller of an off-grid wind turbine/photovoltaic/hydrogen/battery Zhang, X., Chan, S.H., Ho, H.K., Tan, S.-C., Li, M., Li, G., Li, J., Feng, Z., 2015. Towards
hybrid system. Renew. Energy 74, 326–336. a smart energy network: The roles of fuel/electrolysis cells and technological
Trifkovic, M., Marvin, W.A., Daoutidis, P., Sheikhzadeh, M., 2014. Dynamic real-time perspectives. Int. J. Hydrogen Energy 40 (21), 6866–6919.
optimization and control of a hybrid energy system. AIChE J. 60 (7), 2546–2556. Zhang, H., Desideri, U., 2020. Techno-economic optimization of power-to-methanol
Trifkovic, M., Sheikhzadeh, M., Nigim, K., Daoutidis, P., 2013. Modeling and control with co-electrolysis of CO2 and H2 O in solid-oxide electrolyzers. Energy 199,
of a renewable hybrid energy system with hydrogen storage. IEEE Trans. Control 117498.
Syst. Technol. 22 (1), 169–179. Zhang, W., Maleki, A., Rosen, M.A., Liu, J., 2018. Optimization with a simulated
Tso, W.W., Demirhan, C.D., Lee, S., Song, H., Powell, J.B., Pistikopoulos, E.N., 2019. annealing algorithm of a hybrid system for renewable energy including battery
Energy carrier supply chain optimization: A Texas case study. In: Computer Aided and hydrogen storage. Energy 163, 191–207.
Chemical Engineering. Vol. 47, Elsevier, pp. 1–6. Zhang, Z., Way, J.D., Wolden, C.A., 2021. Design and operational considerations of
Tsujimura, T., Iki, N., (National Institute of Advanced Industrial Science and Tech- catalytic membrane reactors for ammonia synthesis. AIChE J. 67 (8), e17259.
nology), 2016. Gas turbine power generation with a methane-ammonia gas Zhao, Y., Setzler, B.P., Wang, J., Nash, J., Wang, T., Xu, B., Yan, Y., 2019. An efficient
mixture and 100% ammonia. https://www.aist.go.jp/aist_e/list/latest_research/ direct ammonia fuel cell for affordable carbon-neutral transportation. Joule 3 (10),
2016/20160412/en20160412.html. 2472–2484.
Tzimas, E., Filiou, C., Peteves, S., Veyret, J., (European Commission Joint Research Zheng, Y., Wang, J., Yu, B., Zhang, W., Chen, J., Qiao, J., Zhang, J., 2017. A review
Center - Institute for Energy), 2021. Hydrogen storage: State-of-the-art and future of high temperature co-electrolysis of H2O and CO2 to produce sustainable fuels
perspective. https://www.core.ac.uk/download/pdf/38614035.pdf. using solid oxide electrolysis cells (SOECs): advanced materials and technology.
U.S. Environmental Protection Agency, 2022. Greenhouse gas emissions: Understand- Chem. Soc. Rev. 46 (5), 1427–1463.
ing global warming potentials. https://www.epa.gov/ghgemissions/understanding- Zhou, F., Azofra, L.M., Ali, M., Kar, M., Simonov, A.N., McDonnell-Worth, C., Sun, C.,
global-warming-potentials. Zhang, X., MacFarlane, D.R., 2017. Electro-synthesis of ammonia from nitrogen at
Valera-Medina, A., Amer-Hatem, F., Azad, A., Dedoussi, I., De Joannon, M., Fernan- ambient temperature and pressure in ionic liquids. Energy Environ. Sci. 10 (12),
des, R., Glarborg, P., Hashemi, H., He, X., Mashruk, S., et al., 2021. Review on 2516–2520.
ammonia as a potential fuel: from synthesis to economics. Energy Fuels 35 (9), Zhou, K., Ferreira, J., De Haan, S., 2008. Optimal energy management strategy
6964–7029. and system sizing method for stand-alone photovoltaic-hydrogen systems. Int. J.
Valera-Medina, A., Xiao, H., Owen-Jones, M., David, W.I., Bowen, P., 2018. Ammonia Hydrogen Energy 33 (2), 477–489.
for power. Prog. Energy Combust. Sci. 69, 63–102.
Varela, C., Mostafa, M., Ahmetovic, E., Zondervan, E., 2020. Agile operation of
renewable methanol synthesis under fluctuating power inputs. In: Computer Aided
Chemical Engineering. Vol. 48, Elsevier, pp. 1381–1386.

21

You might also like