Assessing Uncertainties of Life-Cycle CO₂ Emissions Using Hydrogen Energy for Power Generation

Abstract

Hydrogen and its energy carriers, such as liquid hydrogen (LH₂), methylcyclohexane (MCH), and ammonia (NH₃), are essential components of low-carbon energy systems. To utilize hydrogen energy, the complete environmental merits of its supply chain should be evaluated. To understand the expected environmental benefit under the uncertainty of hydrogen technology development, we conducted life-cycle inventory analysis and calculated CO₂ emissions and their uncertainties attributed to the entire supply chain of hydrogen and NH₃ power generation (co-firing and mono-firing) in Japan. Hydrogen was assumed to be produced from overseas renewable energy sources with LH₂/MCH as the carrier, and NH₃ from natural gas or renewable energy sources. The Japanese life-cycle inventory database was used to calculate emissions. Monte Carlo simulations were performed to evaluate emission uncertainty and mitigation factors using hydrogen energy. For LH₂, CO₂ emission uncertainty during hydrogen liquefaction can be reduced by using low-carbon fuel. For MCH, CO₂ emissions were not significantly affected by power consumption of overseas processes; however, it can be reduced by implementing low-carbon fuel and waste-heat utilization during MCH dehydrogenation. Low-carbon NH₃ production processes significantly affected power generation, whereas carbon capture and storage during NH₃ production showed the greatest reduction in CO₂ emission. In conclusion, reducing CO₂ emissions during the production of hydrogen and NH₃ is key to realize low-carbon hydrogen energy systems.

1. Introduction

Hydrogen energy (hydrogen and its energy carriers such as liquid hydrogen (LH₂), methylcyclohexane (MCH), and ammonia (NH₃)) are attracting attention for realizing carbon neutrality and preventing global warming. Hydrogen does not emit carbon dioxide (CO₂) during the utilization phase and can be produced with low-carbon emissions through water electrolysis, or by combining natural gas reforming and coal gasification processes with carbon capture and storage (CCS) technology. Therefore, energy-related CO₂ emissions can be drastically reduced by replacing fossil fuel combustion with low-carbon hydrogen utilization. According to the International Energy Agency’s roadmap to reach net-zero global emissions by 2050 [1], global hydrogen and hydrogen-based fuel use will expand from 87 Mt in 2020 to 528 Mt in 2050, and 98% of hydrogen in 2050 will be produced from low-carbon methods.

Innovation is essential for establishing a hydrogen economy. Japan is one of the leading countries in research, development, and demonstration (RD&D) of hydrogen technologies. In December 2017, the Japanese government formulated the Basic Hydrogen Strategy [2] which provides a shared vision for the public and private sectors to realize a hydrogen society by 2050. Considering this approach, the Strategic Road Map for Hydrogen and Fuel Cells was revised in March 2019. This scheme presented future milestones and action plans to achieve the new targets indicated in the Basic Hydrogen Strategy. For both the strategy and road map, Japan set the future price of hydrogen at 30 JPY/Nm³ by 2030 and 20 JPY/Nm³ by 2050. To achieve the target price, hydrogen would need to be imported from overseas renewable electricity or be produced from cheap, unused energy resources with CCS. Japan plans to develop commercial-scale hydrogen supply chains by 2030 to procure 300 kt of hydrogen annually. Liquid hydrogen energy carriers (e.g., LH₂, MCH, and NH₃) are suitable for international transportation due to their advantage of easy handling and high energy storage compared to gaseous hydrogen. Demonstration projects for the hydrogen supply chains using liquid hydrogen energy carriers is underway [3,4,5]. On the other hand, the use of hydrogen energy in the electricity sector has attracted attention from the perspective of its end use [6,7,8]. Replacing existing thermal power generation with hydrogen energy generation is expected to reduce carbon emissions as well as serve as a regulated power supply and backup power source required for expanding renewable energy [2]. As hydrogen energy generation consumes a large quantity of hydrogen constantly, it is positioned as a technology that will be commercialized by 2030, in parallel with the construction of an international supply chain. A demonstration experiment conducted in Kobe in April 2018 achieved the world’s first cogeneration of heat and electricity using a gas turbine that exclusively burns hydrogen [9]. Among the energy carriers, NH₃ can be directly combusted to generate electricity. NH₃ co-firing experiments in coal-fired power plants [10] and the development of low nitrogen oxide (NO_x) NH₃ gas turbines [11] are in progress.

Evaluation of hydrogen technologies is also important for establishing a low-carbon hydrogen society. The world’s first ministerial-level meeting on realizing a hydrogen society, termed the Hydrogen Energy Ministerial Meeting, was held in Tokyo on 23 October 2018. The Chair’s summary of the meeting, known as the Tokyo Statement [12] stresses the importance of investigating and evaluating the potential of hydrogen to reduce CO₂ and other pollutants in realizing a hydrogen society. As mentioned above, hydrogen does not emit CO₂ during the utilization phase; however, CO₂ emissions are generated by the resource and utility inputs (electricity, fuel, water, etc.) during the production, storage, and transportation stages of hydrogen and energy carriers. Therefore, to demonstrate the environmental superiority of hydrogen energy, it is necessary to evaluate the emissions of the entire supply chain through a life-cycle inventory (LCI) analysis. LCI analysis for hydrogen supply chains has been conducted in many studies on the well-to-tank greenhouse gas (GHG)/CO₂ emissions of hydrogen for mobility use [13,14,15,16,17,18,19]. In terms of LCI analysis of other applications, Ozawa et al. [20] analyzed life-cycle CO₂ emissions of hydrogen and NH₃ power generation, and Chisalita et al. [21] conducted a cradle-to-gate environmental assessment of European NH₃ production considering various hydrogen supply chains.

Although these studies could provide some implications for establishing low-carbon hydrogen supply chains, they rarely deal with the uncertainties of parameters related to supply chains. Since most technologies that compose hydrogen supply chains are in the development and demonstration stages, there may be a large degree of uncertainty about the environmental value of the entire supply chain at the practical application stage. Such analysis can provide researchers and policymakers with useful feedback that can help in robust decision-making regarding RD&D of hydrogen technologies. LCI analysis often involves assessing the probable impact of several variables in the target supply chains on life-cycle GHG/CO₂ emissions through Monte Carlo simulations [22]. However, LCI analysis that considers the uncertainties in life-cycle GHG/CO₂ emissions from hydrogen supply chains is limited to studies that focus on the use of hydrogen for smobility [13,19].

In most LCI studies, specific values are assumed for supply chain parameters [14,15,16,17,18,20,21]. However, essential technologies that compose hydrogen supply chains for power generation are still in the development and demonstration stages, and we recognize that their specification (e.g., energy input and CO₂ emissions intensity) at the practical application stage should have a certain degree of uncertainty, which have not been fully considered in previous research. This study aims to evaluate uncertainties regarding the environmental benefits of hydrogen in the electricity sector by applying the Monte Carlo method for LCI analysis. While hydrogen energy is attracting attention in the context of lowering the carbon emissions of energy systems in Japan, its environmental effects can range depending on the future performance of elements that make up hydrogen supply chains. The purpose of this study is to identify and understand the important factors involved in electricity generation using hydrogen energy by conducting a stochastic LCI analysis using Monte Carlo simulations. The variations in the parameters related to the operation of hydrogen supply chains were collected by literature surveys, and the uncertainties involved in life-cycle CO₂ emissions from hydrogen and NH₃ power generation were estimated. Please note here that the life-cycle CO₂ emissions attributed to capital goods necessary to operate the hydrogen supply chain were out of the scope of this study. The results obtained are compared with those of coal- and liquefied natural gas (LNG)-fired power generation, and the effect of the use of hydrogen energy on reducing CO₂ emissions over the entire life-cycle is discussed.

2. Method and Assumptions

2.1. Outline

Figure 1 illustrates the research methodology of this study. Firstly, the system boundaries are defined to determine which unit processes to be included in this study. Secondly, inventory data related to the operation of the hydrogen supply chains are collected by literature survey. This study focuses on the life-cycle CO₂ emissions due to the operation of the supply chains and the emissions for capital goods are out of scope. Thirdly, based on the collected data, the probability distributions of energy input and CO₂ emission intensity are set for each process in the supply chains. Fourthly, according to the given probability distributions, Monte Carlo simulations are performed to calculate the life-cycle CO₂ emissions of hydrogen and NH₃ power generation stochastically. Finally, the calculated life-cycle CO₂ emissions are compared with those of thermal power generation to discuss the environmental benefits of hydrogen energy for power generation.

Figure 2 shows the power generation technologies and energy supply chain envisioned in this study. Respectively, both hydrogen and NH₃ power generation (co-firing and mono-firing) are the subjects of analyses, whereas coal- and LNG-fired power generation are used for comparison. The hydrogen supply chain sequentially consists of hydrogen production from renewable energy sources overseas; production of energy carriers; storage at the loading port (LH₂ or MCH); international transportation; storage at the unloading port; and finally, hydrogen restoration. In contrast, the NH₃ supply chain consists of its production from renewable energy or natural gas overseas; storage at the loading port; international transportation; storage and transportation at the unloading port; and re-evaporation. The supply chain for coal and LNG consists of mining and pre-processing; domestic transportation; and international transportation. The CO₂ emission reduction effect of applying CCS technology was also evaluated for NH₃ production from natural gas and coal- and LNG-fired power generation.

Regarding cost analyses of hydrogen and energy carriers [23,24], Australia (AUS), Norway (NOR), and the United Arab Emirates (UAE) were selected as the hydrogen-producing countries; and the UAE as the NH₃-producing country. The one-way distances between NOR, AUS, the UAE, and Japan were considered as 20,000 km; 10,000 km; and 12,000 km, respectively.

The LCI analysis conducted in this study considers the entire life-cycle of products and services (the series of processes from resource extraction to manufacturing, distribution, use, and disposal), and quantifies the consumption of resources and utilities, as well as the emission of environmentally hazardous substances in each process. Life-cycle CO₂ emissions, $E$, which is the result of the analysis, can be obtained using Equation (1):

$$E={\displaystyle \sum }_{i,j}\left(c_{i,j}\times e_i\right)$$

(1)

where $c_{i,j}$ is the consumption of input $i$ in process $j$, and $e_i$ is the CO₂ emission intensity due to the consumption of $i$. In this study, the probability distribution of each parameter was set according to the procedure provided in Appendix A to perform an uncertainty analysis of CO₂ emissions from overseas renewable energy generation and by using the energy consumption of each process as variable factors. The Japanese LCI Inventory Database for Environmental Analysis (IDEA) v2.2 [25]—the largest life-cycle inventory database in Japan which has been developed by National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Japan—was used for domestic processes to estimate the CO₂ emission intensity of consumption by other utilities. It has been developed since 2008 by the Safety Science Research Division of the National Institute of Advanced Industrial Science and Technology and is packaged with data of ~3800 processes based on the Japanese Standard Industrial Classification. To estimate the CO₂ emission intensity of overseas processes, the 2015 statistics of the International Energy Agency [26] were used for electricity inputs and direct CO₂ emission factors were used as fuel inputs, while IDEA v2.2 values were used for non-energy inputs.

2.2. Overseas Renewable Energy Power Generation

In this study, onshore wind (NOR and AUS) and solar photovoltaics (PVs) (AUS and the UAE) were assumed to be renewable energy resources. Life-cycle CO₂ emissions per kWh of electricity at the transmission end were calculated on the assumption that the basic components for constructing a power plant are procured in the hydrogen- and NH₃-producing countries, whereas other components (nacelle and hub of wind turbine, solar panels, etc.) are manufactured in Japan and exported to hydrogen- and NH₃-producing countries. The probability distribution for this is listed in Appendix A

Table A1. Uncertainty parameters for renewable power generation.

Parameter [Unit]	PDF	Mean	SD	Reference(s)
CO2 emissions for wind power in AUS [g-CO2/kWh]	Normal	12.4	1.2	[20,23,27]
CO2 emissions for solar PV in AUS [g-CO2/kWh]	Normal	39.2	3.9	[20,23,27]
CO2 emissions for wind power in NOR [g-CO2/kWh]	Normal	12.1	1.2	[20,23,27]
CO2 emissions for solar PV in UAE [g-CO2/kWh]	Normal	39.9	4.0	[20,23,27]

AUS, Australia; NOR, Norway; PDF, probability density function; PV, photovoltaic; SD, standard deviation; UAE, United Arab Emirates.

. Inputs to the power plant ($x_{i,j}$) were based on data from Ozawa et al. [20]. The facility utilization rate, in-service rate, and service life of the power plant were set assuming overseas wind and sunshine conditions [20,23,27]. The specifications of the power plant are listed in

Table 1. Specifications of renewable power plants.

Parameter [Unit]	Wind	Solar PV
Capacity [MW]	40	10
Capacity factor [%]	35	20
Auxiliary power ratio [%]	10	3
Lifetime [year]	30	30

. Parameters listed in Table 1 were set by collecting data from literature surveys [20].

2.3. Hydrogen Supply Chain

For the “base case” (denoted as Base in the figures) of this study, it was assumed that only the hydrogen production process uses renewable energy power generation, while the electricity input to the other overseas processes uses grid electricity. However, for countries such as AUS and the UAE, where the CO₂ emission intensity of grid electricity is large, the emission from the entire supply chain can be decreased by substituting overseas processes with available renewable energy electricity into the grid electricity input. Therefore, in the analysis, it was also considered that for the hydrogen supply chain from AUS and the UAE, the electricity necessary to operate the overseas processes is from the same renewable energy sources as renewable hydrogen production—denoted the “low-carbon case” (LC) in the figures. It was also assumed that the electricity input into domestic processes was supplied by hydrogen power generation.

2.3.1. Hydrogen Production from Renewable Energy

It is expected that hydrogen produced from renewable energy should play a major role in the future hydrogen economy. In 2050, more than 300 Mt of green hydrogen will be produced by water electrolysis to achieve global carbon-neutrality by 2050 [1]. In this study, hydrogen is produced from onshore wind or solar PV power via alkaline or solid polymer water electrolysis. The probability distribution of electricity consumption was set as shown in Appendix A

Table A2. Uncertainty parameters for H₂ supply chain.

Parameter [Unit]	PDF	Mean	SD	Min.	Max.	Reference(s)
Electricity input for hydrogen production via water electrolysis [kWh/Nm3-H2]	Normal	4.88	0.80			[28,29]
Electricity input for liquefaction of gaseous hydrogen to produce LH2 [kWh/Nm3-H2]	Normal	0.906	0.244			[30,31,32,33,34,35]
Electricity input for LH2 storage at loading port [kWh/Nm3-H2]	Normal	0.055	0.006			[18]
LH2 Boil-off rate during ocean transport [%/day]	Triangular	0.30		0.20	0.40	[30,31]
Electricity input for LH2 storage at unloading port [kWh/Nm3-H2]	Normal	0.017	0.002			[18]
Electricity input for MCH production [kWh/t-MCH]	Triangular	40.68		7.54	93.30	[30,31,36]
Electricity input for MCH storage at loading/unloading port [kWh/t-MCH]	Normal	0.915	0.120			[30,31]
Electricity input for MCH dehydrogenization [kWh/Nm3-H2]	Normal	0.310	0.058			[30,31,36]
Electricity input for TOL storage at loading/unloading port [kWh/t-TOL]	Normal	0.915	0.120			[30,31]

LH₂, liquid hydrogen; MCH, methylcyclohexane; PDF, probability density function; PV, photovoltaic; SD, standard deviation; TOL, toluene.

based on published data [28,29], and the consumption of pure water was estimated to be 0.804 kg/Nm³-H₂ based on the theoretical value.

2.3.2. LH₂ Supply Chain

Figure 3 shows the supply chain for the use of LH₂ as an energy carrier. Hydrogen gas is liquefied to produce LH₂, which is stored in insulated tanks at the loading site, and then transported from the hydrogen-producing country to Japan by an LH₂ tanker (capacity: 160,000 m³ at cruising speed: 16 knots). Because detailed information on the energy consumption performance of the LH₂ tanker during navigation was not available, its CO₂ emission intensity was estimated with reference to LNG tanker transport in IDEA v2.2, it was assumed that heavy fuel oil (HFO) and LH₂ boil-off gas were used as marine fuel. Here, CO₂ emissions associated with the transportation of LH₂ by tankers consider the unladen voyage of the tanker from Japan to the hydrogen-producing country and its return from the hydrogen-producing country to Japan laden with LH₂. The imported LH₂ is stored in insulated tanks at the unloading port and supplied to the power plant. The probability distributions of electricity consumption due to hydrogen liquefaction and LH₂ storage, and the LH₂ boil-off rate due to international transportation were set based on published data as shown in Appendix A Table A2 [18,30,31,32,33,34,35,36].

2.3.3. MCH Supply Chain

Figure 4 shows the supply chain for the use of MCH as an energy carrier, and

Table 2. Specifications of H₂ supply chain using MCH.

Process	Parameter [Unit]	Value
MCH production	Reaction yield of H2 addition to TOL [%]	99.8
	H2 consumption rate [%]	97.9
MCH dehydrogenation	Conversion rate [%]	95.0
	Selectivity [%]	99.9
	H2 yield [%]	90.0
TOL replenishment	Annual replenishment rate of TOL [%]	3.0

lists the specifications for each process (“Value” in Table 2,

Table 3. Specifications of NH₃ supply chain.

Process	Parameter [Unit]	Value
NH3 production from renewable hydrogen	Hydrogen loss rate during NH3 synthesis [%]	4.9
	NH3 loss rate during NH3 liquefaction [%]	0.1
CO2 capture	Rate of CO2 capture [%]	90
	Monoethanolamine input [kg/t-CO2 captured]	2.41
	Sodium hydroxide input [kg/t-CO2 captured]	0.13
	Activated carbon input [kg/t-CO2 captured]	0.08
CO2 compression	CO2 leakage from CO2 compressor [t-CO2/MW/year]	23.2
CO2 transport	CO2 leakage rate from pipelines [%]	0.00367

,

Table 4. Specifications of coal supply chain.

Process	Parameter [Unit]	Value
Coal mining and pretreatment	Electricity input [kWh/t-coal]	13.2
	Diesel input [liter /t-coal]	3.88
	Gasoline input [liter /t-coal]	0.0309
	Steel input [kg/t-coal]	0.0170
	Explosive input [kg/t-coal]	2.81
	Cement input [kg/t-coal]	0.0108
	Rubber input [kg/t-coal]	0.000238
	CH4 emissions during coal mining [kg-CO2/t-coal]	121
Coal inland transport	Diesel input [liter/t-coal]	5.98
Coal ocean transport	Heavy fuel oil (HFO) input [kg/t-coal]	8.74

and

Table 5. Specifications of LNG supply chain.

Process	Parameter [Unit]	Value
Natural gas mining	CH4 emissions [kg-CO2/t-natural gas]	60.7
LNG production	CO2 emissions [kg-CO2/t-natural gas]	99.0
	LNG input [kg/t-LNG]	139
LNG ocean transport	LNG input [kg/t-LNG]	30.4
	HFO input [kg/t-LNG]	13.2

means the specific values assumed for the parameters of each process in the supply chains). Parameters listed in Table 2 were set by collecting data from literature surveys [30,37,38].

MCH is produced from hydrogen and toluene (TOL) and stored in cone-roof tanks at the loading port. TOL is produced from oil refining and naphtha reforming in Japan and is transported to a hydrogen-producing country by chemical tankers. The MCH produced using TOL is then transported from a hydrogen-producing country to Japan by chemical tankers. The petroleum product tanker transport of IDEA v2.2 was used to evaluate the CO₂ emission intensity of the chemical tankers. The imported MCH is stored in cone-roof tanks at the unloading port and then converted back to hydrogen and TOL by dehydrogenation. Hydrogen is supplied to the power plant, and the TOL is transported to the hydrogen-producing country for reuse. The heat required for dehydrogenation is assumed to be supplied by the combustion of HFO. TOL degraded by side reactions (formation of benzene and methane by demethylation reaction, paraffin by ring-opening reaction, cyclopentane by isomerization, and alkyl biphenyls by dimerization [28,37]) was assumed to be replenished. The probability distributions of electricity consumption due to MCH production, storage, dehydrogenation, and TOL storage were set as shown in Appendix A Table A2 based on published data [20,26].

2.4. NH₃ Supply Chain

Figure 5 shows the NH₃ supply chain, and Table 3 lists the specifications for each process. Parameters listed in Table 3 were set by collecting data from literature surveys [39,40,41].

2.4.1. NH₃ Production from Renewable Energy

NH₃ is synthesized and liquefied by the Haber–Bosch process from hydrogen produced by alkaline or solid polymer water electrolysis using solar PV power. Nitrogen was obtained via air separation. The consumption of electricity and pure water for hydrogen production was set as described in Section 2.3.1. The probability distributions of electricity consumption for nitrogen and NH₃ production were set as shown in Appendix A

Table A3. Uncertainty parameters for NH₃ supply chain.

Parameter [Unit]	PDF	Mean	SD	Min.	Max.	Reference(s)
Electricity input for nitrogen separation from air [kWh/Nm3-N2]	Triangular	0.249		0.195	0.380	[43,44,46]
Electricity input for NH3 production from renewable hydrogen [kWh/kg-NH3]	Triangular	0.457		0.177	0.849	[39,42,45,46]
Natural gas input as a fuel for NH3 production from natural gas [MJ-NG(HHV)/kg-NH3]	Normal	*a	*b			[25]
Electricity input for NH3 production from natural gas [kWh/kg-NH3]	Normal	*a	*b			[25]
Steam input for CO2 capture [GJ/t-CO2 captured]	Normal	2.48	0.70			[40,41,47,48,49,50,51]
CO2 emissions for steam input for CO2 capture [g-CO2/kg-steam]	Triangular	*c		0	*d	[25]
Electricity input for CO2 capture [kWh/t-CO2 captured]	Normal	23.6	2.4			[40,41]
Electricity input for CO2 compression [kWh/t-CO2 captured]	Normal	127.8	12.8			[40,41]
Electricity input for CO2 injection [kWh/t-CO2 captured]	Normal	7.0	0.7			[40,41]
Electricity input for NH3 storage at loading port [kWh/t-NH3]	Normal	0.83	0.08			[20]
Electricity input for NH3 storage at unloading port [kWh/t-NH3]	Normal	2.77	0.28			[20]
Electricity input for NH3 distribution [kWh/t-NH3]	Normal	0.57	0.06			[20]
Electricity input for NH3 re-evaporation [kWh/t-NH3]	Normal	0.022	0.002			[20]

*^a Data retrieved from the NH₃ production process of IDEA v2.2. *^b 10% of the data retrieved from the NH₃ production process of IDEA v2.2. *^c 50% of the data retrieved from steam production by the HFO-fired boiler of IDEA v2.2. *^d Data retrieved from steam production by the HFO-fired boiler of IDEA v2.2. HFO, heavy fuel oil; HHV, higher heating value; NG, natural gas; PDF, probability density function; SD, standard deviation.

based on published data [39,42,43,44,45,46].

2.4.2. NH₃ Production from Natural Gas

For the process of NH₃ production from natural gas, the input/output relationship was established based on the NH₃ production process of IDEA v2.2 from AIST, Tsukuba, Japan. In this process, natural gas is used as a feedstock for NH₃ and as a fuel for the reactor. The probability distributions of the utility natural gas and electricity consumption were set as shown in Appendix A Table A3.

Application of CCS is effective to mitigate CO₂ from NH₃ production process from natural gas. CCS consists of CO₂ capture, compression, pipeline transportation, and injection processes. Utility-derived CO₂ was recovered using the chemical absorption method (recovery rate: 90%). On the other hand, feedstock-derived CO₂ is emitted with high density from the NH₃ synthesis process; hence, a recovery process is not necessary. The recovered utility-derived CO₂ is compressed together with all the feedstock-derived CO₂, which is transported by pipeline to a CO₂ reservoir (transport distance: 50 km) and injected into the reservoir. The probability distributions of steam consumption due to CO₂ capture, electricity consumption due to CO₂ capture, compression, and injection were set as shown in Appendix A Table A3 based on published data [25,40,41,47,48,49,50,51]. As for the CO₂ emission intensity of steam required for chemical absorption, the probability distribution of Appendix A Table A3 was set with a minimum value of 0 (assuming steam can be supplied by waste heat) and the maximum value given by steam production from the HFO-fired boiler of IDEA v2.2 from AIST, Tsukuba, Japan.

2.4.3. NH₃ Storage, Transportation, and Regasification

The produced NH₃ is stored in tanks at the loading port and transported from the NH₃-producing country to Japan by NH₃ tankers (capacity: 38,000 m³ at a cruising speed of 17 knots). For the CO₂ emission intensity of the NH₃ tanker, the petroleum product tanker transportation of IDEA v2.2 was used by considering CO₂ emissions during round trips between Japan and the NH₃-producing country. After storage in tanks at the unloading port, the imported NH₃ is transported by pipeline to the power plant, where it is re-gasified and used as fuel for power generation. The probability distribution of electricity consumption due to NH₃ storage and transportation was set as shown in Appendix A Table A3, based on published data [20].

2.5. Fossil Fuel Supply Chain

We calculated CO₂ emissions from the coal and LNG supply chains based on data from Ozawa et al. [20] and statistics on import destinations [52]. In this analysis, coal-fired and LNG-fired power plants are considered commercialized technologies, and the uncertainty of the fossil fuel supply chain is not considered.

Figure 6 shows the coal supply chain, and Table 4 lists the specifications for each process. Coal mined overseas first undergoes pre-processing. Methane emissions from coal mines and CO₂ emissions from consumables during mining (steel, explosives, etc.) were also considered. The coal is then transported by rail (fueled by diesel) to the port, where it is transported by coal carriers to power plants in Japan. For international transportation, CO₂ emissions for the round trip between Japan and the coal-producing country were considered, and the fuel was assumed to be HFO.

Figure 7 shows the LNG supply chain, and Table 5 lists the specifications for each process. First, natural gas is extracted from overseas gas fields and liquefied. Consideration is also given to methane emissions from natural gas mining and CO₂ emissions from crude gas during liquefaction. LNG is then transported by a tanker to a power plant in Japan. For international transportation, CO₂ emissions for the round trip between Japan and the gas-producing country were considered, and the fuel was assumed to be boil-off gas and HFO.

2.6. Power Generation and CCS

Table 6. Specifications of thermal power plants.

Parameter [Unit]	H2–LNG Co-Firing	H2 Mono-Firing	NH3–Coal Co-Firing	NH3–LNG Co-Firing	NH3 Mono-Firing	Coal	LNG
Co-firing ratio of H2/NH3 [%] (on HHV basis)	20	-	20	20	-	-	-
Capacity [MW]	1000	1000	1000	1000	1000	1000	1000
Capacity factor [%]	-	70	70	70	70	70	70
Efficiency (HHV) [%]	53.0	51.3	39.6	53.0	53.0	39.6	53.0
Auxiliary power ratio without CCS [%]	2.08	2.08	5.06	2.08	2.08	5.06	2.08
Electricity generated without CCS [MWh/year]	6,004,454	6,004,454	5,821,721	6,004,454	6,004,454	5,821,721	6,004,454
Auxiliary power ratio with CCS [%]	-	-	-	-	-	36.89	15.03
Electricity generated with CCS [MWh/year]	-	-	-	-	-	3,869,743	5,210,228

HHV, higher heating value.

shows the specifications of the hydrogen and NH₃ power generation processes that were analyzed, and the thermal power generation process used for comparison. Five types of power generation were examined: hydrogen–LNG co-firing, hydrogen mono-firing, NH₃–coal co-firing, NH₃–LNG co-firing, and NH₃ mono-firing. The co-firing ratio of hydrogen and NH₃ to existing fuels was set at 20% (based on a high heating value) in all cases. The efficiency of hydrogen mono-fired power generation was set to be lower than that of LNG-fired power generation [53]. The efficiencies of hydrogen co-firing power generation and NH₃ mono-firing/co-firing power generation were set to be the same as those of LNG-fired power generation. Thermal power generation was assumed to be ultra-supercritical coal-fired power generation (600 °C class) and LNG combined-cycle power generation (1500 °C class), and CO₂ emissions from coal ash disposal were also considered [23].

Table 7. Specifications of carbon capture and storage (CCS) from thermal power plants [40,41].

Process	Parameter [Unit]	Value
CO2 capture	CO2 capture rate [%]	90
	Steam input [GJ/t-CO2 captured]	3
	Electricity input [kWh/t-CO2 captured]	23.6
	Monoethanolamine input [kg/t-CO2 captured]	2.41
	Sodium hydroxide input [kg/t-CO2 captured]	0.13
	Activated carbon input [kg/t-CO2 captured]	0.08
CO2 compression	CO2 leakage from CO2 compressor [t-CO2/MW/year]	23.2
CO2 transport	HFO input for CO2 transport by ship [kg/t-CO2 transported]	1.316
CO2 injection	HFO input for CO2 injection [kg/t-CO2 injected]	3.666

lists the specifications for the CCS in thermal power plants (“Value” in Table 7 means the specific values assumed for the parameters of each process in the supply chains). Parameters listed in Table 7 were set by collecting data from literature surveys [40,41]. The emitted CO₂ is liquefied after being collected using the chemical absorption technique, transported by ship to a CO₂ reservoir (transport distance: 50 km), and injected into the reservoir. The electricity consumption required for CO₂ capture, liquefaction, and injection was assumed to be supplied by power generation by increasing the internal consumption rate.

2.7. Uncertainty Analysis

Uncertainty analysis aims at quantifying the variability of the calculation results attributed to the variation of the parameter input, and it helps robust decision making under uncertainty. As an example of uncertainty analysis, some financial studies focus on the effect of mitigating commodity pricing volatility in supply chains for supply chain risk management [54,55].

In terms of environmental profile of supply chains, there are two methods for dealing with uncertainty in the LCI analysis. One method is to conduct sensitivity analyses. The International Organization for Standardization (ISO) 14044 guidelines suggest that sensitivity analyses should include a wide range of factors to determine the influence of variations in assumptions, methods, and data [56]. An alternative option, the uncertainty analysis, employs probabilistic simulations based on the Monte Carlo method to evaluate the combined influence of multiple uncertain factors on the results. Here, probability distributions were assumed for the input parameters of the system. Repeated calculations with different input values yield a probability frequency distribution of the total GHG emissions from the entire system [57].

In this study, we performed a Monte Carlo simulation to calculate the life-cycle CO₂ emissions of hydrogen and NH₃ power generation. We treated energy consumption ($c_{i,j}$) and CO₂ emission intensity ($e_i$) for each process in the supply chains as uncertain parameters and calculated the life-cycle CO₂ emissions ($E$) stochastically using Equation (1). Based on the variations in the parameters collected by literature survey, probability distributions were assumed for each uncertain parameter, as shown in Appendix A Table A1, Table A2 and Table A3. Either normal distribution or triangular distribution was applied for the parameters’ probability density function. If a normal distribution was applied for a parameter $x$, the probability distribution of $f\left(x\right)$ was defined as Equation (2):

$$f\left(x\right)=\frac{1}{\sqrt{2\pi {\sigma }^2}}\exp \left\{-\frac{{\left(x-\mu \right)}^2}{2{\sigma }^2}\right\}$$

(2)

where $\mu$ and $\sigma$ were the mean and standard deviation of $x$, respectively.

If a triangular distribution was applied, $f\left(x\right)$ was defined as Equation (3):

$$\begin{gathered} f\left(x\right)=\frac{2\left(x-\alpha \right)}{\left(\beta -\alpha \right)\left(\mu -\alpha \right)}\hspace{1em}\mathrm{for}\alpha \le x<\mu \\ f\left(x\right)=\frac{2}{\left(\beta -\alpha \right)}\hspace{1em}\mathrm{for}x=\mu \\ f\left(x\right)=\frac{2\left(\beta -x\right)}{\left(\beta -\alpha \right)\left(\beta -\mu \right)}\hspace{1em}\mathrm{for}\mu <x\le \beta \end{gathered}$$

(3)

where $\alpha$ and $\beta$ were the minimum and maximum values of $x$, respectively. In Appendix A, the probability distributions of the uncertain parameters assumed in this study are described in detail. Monte Carlo simulations were performed using Oracle’s Crystal Ball, and the number of iterations was set to 100,000.

3. Results and Discussion

We conducted an LCI analysis of both hydrogen and NH₃ power generation (co-firing and mono-firing), while considering the uncertainties, to identify and understand the factors that are important for realizing low-carbon generation using hydrogen energy in the electric power sector. Hydrogen was assumed to be produced from renewable energy sources in AUS, NOR, and the UAE and stored and transported using LH₂ and MCH as carriers, while NH₃ was assumed to be produced from renewable energy or natural gas in the UAE. Parameters related to the supply chain were collected from the literature, and the Japanese life-cycle inventory database IDEA was used to calculate emissions. The probability distributions of each parameter were established using the energy consumption and CO₂ emissions of each process in the supply chain as variables, and the uncertainties of life-cycle CO₂ emissions were calculated stochastically using Monte Carlo simulation. We compared the life-cycle CO₂ emissions of hydrogen and NH₃ power generation obtained from the Monte Carlo simulations with those of thermal power generation, as shown in Figure 8, Figure 9 and Figure 10. The bars represent the mean of the results, and the error bars represent the 99.7% confidence interval (corresponding to 3σ if a normal distribution is assumed). Probability distributions of the life-cycle CO₂ emissions of hydrogen and NH₃ power generation are also illustrated in Appendix B Figure A1, Figure A2, Figure A3, Figure A4, Figure A5, Figure A6 and Figure A7 (see Appendix B).

3.1. Hydrogen Power Generation

Figure 8 shows life-cycle CO₂ emissions of hydrogen power generation when LH₂ is used as the energy carrier. The leftmost seven bars illustrate the emissions of hydrogen–LNG co-firing power generation (hydrogen co-firing ratio: 20%). If hydrogen is produced in AUS and the UAE and the base case is selected (grid electricity is used for overseas processes other than hydrogen production), the range of emission variation from hydrogen–LNG co-firing power generation includes those from LNG-fired power generation without CCS, which indicate that the two power generations are comparable with each other. On the other hand, in the low-carbon cases of AUS and the UAE (where all electricity consumption in overseas processes is covered by renewable energy) and in the case of hydrogen production in NOR, emissions are lower than those from LNG-fired power generation without CCS, indicating that the co-firing of hydrogen can reduce CO₂ emissions from LNG power plants. NOR has low emissions in the base case owing to the low CO₂ emission intensity of its grid electricity. It is also confirmed that emissions from hydrogen–LNG co-firing power generation is higher than those from coal- or LNG-fired power generation with CCS in all cases.

The differences between the cases become even more pronounced with hydrogen mono-firing. When hydrogen is produced in AUS and the UAE and the base case is selected, the range of emission variation is very large, and the emissions can reach those from coal-fired power generation in the worst case. The main reason for this is the effect of electricity consumption on the hydrogen liquefaction. As shown in Appendix A Table A2, the probability distribution of electricity consumption for hydrogen liquefaction has a large standard deviation. If grid electricity from AUS and the UAE (having high CO₂ emission intensities) is used for hydrogen liquefaction, the uncertainty attributed to this process will affect the emissions of the entire supply chain.

This uncertainty can be reduced by selecting a low-carbon case. In the low-carbon case of solar PV in AUS or the UAE, CO₂ emissions from electricity consumption are reduced by using renewable energy for hydrogen liquefaction, while life-cycle CO₂ emissions are lower than those from coal-fired power generation with CCS and comparable to LNG-fired power generation with CCS. If hydrogen is produced from wind power in AUS and the low-carbon case is chosen, or if hydrogen is produced in NOR, emissions will be lower than those of LNG-fired power generation with CCS.

Thus, if LH₂ is used as the energy carrier, the life-cycle CO₂ emissions of hydrogen power generation can be reduced by using renewable energy and grid electricity with low CO₂ emission intensity for hydrogen liquefaction.

Figure 9 shows life-cycle CO₂ emissions of hydrogen power generation when MCH is used as the energy carrier. In the case of hydrogen–LNG co-firing power generation (at a hydrogen mixed combustion ratio of 20%), the range of emission variation is comparable to those of LNG-fired power generation without CCS, except in cases where hydrogen is produced by wind power in AUS. Even in the AUS wind cases, the superiority to LNG-fired power generation is limited, indicating that the effect of CO₂ emission reduction of hydrogen co-firing is smaller when MCH is used as the energy carrier rather than LH₂. On the other hand, a comparison of hydrogen mono-firing with LNG-fired power generation without CCS shows that the emissions from hydrogen mono-firing are lower when hydrogen is produced by wind power in AUS and NOR. Especially if hydrogen is produced by wind power in AUS and the low-carbon case is selected, emissions is comparable to those from coal-fired power generation with CCS.

A comparison of the base case with the low-carbon case shows no significant difference in terms of emissions, which indicates that the effect of low-carbon electricity consumption in the overseas process is small because the power consumption is smaller than that of LH₂. In addition, a large amount of heat is required during MCH dehydrogenation, and CO₂ emissions from heat consumption account for 0~45% of the total. If the fuel for the dehydrogenation reaction is changed from HFO to city gas, CO₂ emissions from heat consumption can be reduced by ~20%. These results suggest the possibility of reducing CO₂ emissions by using a low-carbon fuel or implementing waste-heat utilization in the MCH dehydrogenation process.

3.2. NH₃ Power Generation

Figure 10 shows life-cycle CO₂ emissions of NH₃ power generation. The result for NH₃–coal co-firing (at an NH₃ co-firing rate of 20%) shows that the range of emission variation is lower than those from coal-fired power generation without CCS in all cases, indicating that NH₃ co-firing is effective in reducing CO₂ emissions from coal power plants. In contrast, when comparing NH₃–LNG co-firing (at an NH₃ co-firing rate of 20%) and LNG-fired power generation without CCS, the emissions from NH₃–coal co-firing are lower if CCS is implemented, or when solar PV power is used in NH₃ production process. This result indicates that environmental benefits of NH₃ co-firing in LNG power plants can be gained when a low-carbon NH₃ production process is selected.

For NH₃ mono-firing, it is observed that the range of emission variation without CCS during NH₃ production from natural gas is higher than those of LNG-fired power generation without CCS, while the range of emission variation with CCS is lower than those of LNG-fired power generation with CCS. The latter result implies that the effect of CO₂ emission reduction of CCS during NH₃ production is greater than that of CCS during power generation, because the NH₃ production process emits highly pure CO₂ derived from the raw material and CCS is easier to implement than in thermal power generation. These results show that the use of CCS in the NH₃ production process can reduce the life-cycle CO₂ emissions of NH₃ power generation.

3.3. Summary of Results

Figure 11 and

Table 8. Representative results on life-cycle CO₂ emissions of H₂ and NH₃ power generation.

Power Generation	H2/NH3-Producing Country	Energy Source	Energy Carrier	Case	Life-Cycle CO2 Emissions Mean (99.7% CI) [g-CO2/kWh]
H2–LNG co-firing	AUS	Wind	LH2	LC	357.3 (353.4–362.2)
			MCH	LC	405.9 (396.5–415.8)
		PV	LH2	LC	376.5 (363.5–392.5)
			MCH	LC	413.4 (401.5–426.3)
	NOR	Wind	LH2	Base	359.9 (355.9–364.7)
			MCH	Base	416.6 (407.0–426.8)
	UAE	PV	LH2	LC	377.2 (363.9–393.3)
			MCH	LC	427.4 (414.1–443.4)
H2 mono-firing	AUS	Wind	LH2	LC	56.9 (36.2–82.3)
			MCH	LC	289.7 (247.2–342.4)
		PV	LH2	LC	156.5 (89.1–240.3)
			MCH	LC	407.9 (310.2–532.6)
	NOR	Wind	LH2	Base	70.1 (49.4–95.2)
			MCH	Base	355.3 (306.3–416.0)
	UAE	PV	LH2	LC	160.7 (91.3–244.1)
			MCH	LC	421.5 (322.4–545.2)
NH3–coal co-firing	UAE	Natural gas	NH3	w/o CCS	919.9 (911.6–928.3)
				w/CCS	786.7 (782.2–793.6)
		PV	NH3		824.1 (801.8–849.5)
NH3–LNG co-firing	UAE	Natural gas	NH3	w/o CCS	465.3 (459.2–471.4)
				w/CCS	368.1 (364.8–373.2)
		PV	NH3		395.4 (379.2–413.9)
NH3 mono-firing	UAE	Natural gas	NH3	w/o CCS	579.9 (550.0–610.5)
				w/CCS	98.7 (82.2–123.7)
Coal				w/o CCS	949.3
				w/CCS	278.1
LNG				w/o CCS	432.1
				w/CCS	150.4

AUS, Australia; CI, confidence interval; NOR, Norway; LC, low-carbon; PV, photovoltaic; UAE, United Arab Emirates.

summarize representative results on life-cycle CO₂ emissions of hydrogen and NH₃ power generation. The results show that life-cycle CO₂ emissions of hydrogen mono-firing power generation using LH₂ should be lower than those of LNG-fired power generation with CCS if hydrogen is produced from wind power in AUS and the low-carbon case is chosen, or if hydrogen is produced in NOR. Since electricity consumption for hydrogen liquefaction has a wide range of uncertainty, the total emissions can be decreased by utilizing low-carbon electricity for this process. Therefore, large-scale renewable power plants are required for supplying electricity to hydrogen production and liquefaction processes in hydrogen-producing countries. The results also indicate that life-cycle CO₂ emissions of NH₃ mono-firing power generation should be lower than those of LNG-fired power generation with CCS if CCS is available for NH₃ production process in the UAE. Since the CO₂ from NH₃ synthesis process has a high density, CO₂ capture from NH₃ production process is more efficient than post-combustion CO₂ capture from power plants.

This study focuses on the environmental benefit of using hydrogen and NH₃ produced from renewable energy or natural gas as a power generation fuel. The LCI and its uncertainty analysis techniques used in this study can also be applied to other technology value chains where technological parameters are available. Moreover, the Monte Carlo method can be applied for uncertainty analysis of procurement cost for any technology value chain by assuming investment cost, operation and management cost, and commodity prices for each process.

4. Conclusions

This study evaluated uncertainties regarding the environmental benefits of hydrogen use in the electricity sector by applying the Monte Carlo method for LCI analysis. The results show that the life-cycle CO₂ emissions of hydrogen power generation using LH₂ is influenced by the uncertainty of electricity consumption during hydrogen liquefaction, whereas using low-carbon power sources (such as renewable energy for this process) can reduce uncertainty and CO₂ emissions. In the case of using MCH, while CO₂ emissions are not significantly affected by the power consumption of overseas processes, it is suggested that CO₂ emissions be reduced by implementing low-carbon fuel and waste-heat utilization during MCH dehydrogenation. The life-cycle CO₂ emissions of NH₃ power generation are significantly affected by a low-carbon NH₃ production process, and in particular, CCS during NH₃ production has a greater CO₂ emission reduction effect than CCS during power generation.

An assessment of the entire hydrogen energy supply chain provides important insights for realizing a significant reduction in the carbon emissions of hydrogen energy. In this study, we focused on the production of hydrogen and NH₃ using renewable energy or natural gas. In the future, it may be necessary to conduct similar analyses of lignite gasification and hydrogen production using domestic resources to examine various possibilities for realizing a low-carbon energy system through hydrogen energy. Similarly, we note that conducting life-cycle analysis to include economic, social, and other environmental impacts as a part of multiple criteria analyses will also be increasingly important for the deliberation of hydrogen’s potential role as an energy medium [58,59,60,61].