Quantitative Assessment of Coal Phaseouts and Retrofit Deployments for Low-Carbon Transition Pathways in China’s Coal Power Sector

Abstract

Accelerating the low-carbon transition of China’s coal-fired power sector is essential for advancing national sustainability goals and fulfilling global climate commitments. This study introduces an integrated, data-driven analytical framework to facilitate the sustainable transformation of the coal power sector through coordinated unit-level retirements, new capacity planning, and targeted retrofits. By combining a comprehensive unit-level database with a multi-criteria evaluation framework, the analysis incorporates environmental, technical, and economic factors into decision-making for retirement scheduling. Scenario analyses based on the China Energy Transformation Outlook (CETO 2024) delineate both baseline and ideal carbon neutrality pathways. Optimization algorithms are employed to identify cost-effective retrofit strategies or portfolios, minimizing levelized carbon reduction costs. The findings reveal that cumulative emissions can be reduced by 10–14.9 GtCO₂ by 2060, with advanced technologies like CCUS and co-firing contributing over half of retrofit-driven mitigation. The estimated transition cost of 6.2–6.7 trillion CNY underscores the scale of sustainable investment required. Sensitivity analyses further highlight the critical role of reducing green hydrogen costs to enable deep decarbonization. Overall, this study provides a robust and replicable planning tool to support policymakers in formulating strategies that align coal power sector transformation with long-term sustainability and China’s carbon neutrality commitments.

1. Introduction

Global efforts to combat global warming are increasingly focused on carbon reduction, with the projection of fossil fuel-related carbon emissions reaching a record of 37.4 billion tonnes (Gt) in 2024 [1]. Among them, coal-related emissions account for over 40% of global fossil fuel CO₂ emissions over the past two decades, highlighting the ongoing reliance on coal for energy. China, emerging as a leading emitter, contributes around 32% of total emissions in 2024, as illustrated in Figure 1. However, its cumulative share in the global carbon budget since 1750 remains around 15%, which is lower than that of the EU27 and the United States [1].

Figure 1. Global Fossil Fuel CO₂ Emissions: (a) by fuel type, (b) by country, (c) cumulative emissions by country, and (d) by sector in China. The data presented in Panels (a–c) are sourced from the Global Carbon Budget 2024 [2], while Panel (d) is derived from the EDGAR Community GHG database [3]. IAS stands for international aviation and shipping. The sectors in Panel (d) follow the classification used in EDGAR. The black lines in the panels correspond to the Y-axes on the right-hand side.

Figure 1. Global Fossil Fuel CO₂ Emissions: (a) by fuel type, (b) by country, (c) cumulative emissions by country, and (d) by sector in China. The data presented in Panels (a–c) are sourced from the Global Carbon Budget 2024 [2], while Panel (d) is derived from the EDGAR Community GHG database [3]. IAS stands for international aviation and shipping. The sectors in Panel (d) follow the classification used in EDGAR. The black lines in the panels correspond to the Y-axes on the right-hand side.

As the pursuit of China’s ‘Dual Carbon Goals’ takes center stage, its energy system is undergoing a profound transition. Currently, coal-related emissions account for more than 40% of China’s energy-related emissions [3]. The share of fossil fuels in the energy mix is expected to decline from over 80% to only 20%, while non-fossil energy sources are expected to rise to 80% by mid-century [4]. Despite a declining share of installed capacity in coal-fired power from over 60% to less than 40% by 2023 [5], coal-fired power, historically the backbone of China’s rapid economic growth, remains the largest emitter. Although its contribution to total electricity generation has fallen from 80% to 60% [6] due to the gradual shift towards non-fossil fuel sources in the energy structure, coal power continues to dominate China’s energy landscape. Recent regional power shortages since 2021 have highlighted the intricate challenges of advancing this transition, particularly during peak demand periods. These underscore the necessity for China to reduce its reliance on coal and to develop a new energy system that integrates more non-fossil fuels while maintaining grid stability. On this matter, facilitating an orderly and rational transition of coal-fired power is critical as an integral element of China’s overarching energy strategy and global decarbonization.

The low-carbon transition in China’s coal power sector is pivotal for global sustainability, as China now represents over 56% of global coal power capacity and nearly 93% of new installations since 2024 [7], thereby contributing over 10.5% of worldwide carbon emissions. Progress in coal phase-down, large-scale adoption of advanced retrofit technologies, and accelerated deployment of renewables could enable China to achieve cumulative emission reductions of 10–15 GtCO₂ by 2060 [8]. Therefore, this transition is not only a domestic imperative but also a linchpin for global decarbonization and the evolution of a sustainable energy future.

To support China’s low-carbon transition, the government has launched a series of policy measures to drive the low-carbon transition of coal power, mainly following two strategies: restructuring coal power capacity by phasing out outdated units and introducing advanced units to ensure grid stability, and deploying low-carbon retrofit technologies. Accordingly, the coal power fleet can be classified into retired, newly constructed, and operational units, with the operational units further divided based on their retrofit status. Assessing the transition pathways of each category is essential for accurately evaluating coal power’s mitigation potential.

Most studies on decarbonization of coal power emphasize primarily either phaseout-driven or retrofit-driven carbon reduction strategies independently, and few have quantitatively integrated both approaches at the unit level. Previous analyses on coal power retirements have explored key factors, including operational lifespans [9], policy-induced uncertainties [10], health impacts [11], technical considerations [12], economic implications [13], and environmental constraints [14]. However, projections of China’s coal power transition are contingent on scenario assumptions and technology adoption pathways [15]. The rapid growth of renewable energy sources alongside recent hydropower shortages underscores the impracticality of abrupt, large-scale coal retirements due to renewable intermittency. Thus, current transition strategies emphasize shifting coal power gradually from a dominant to a more flexible role, reflecting recent policy-driven developments such as coal unit constructions tailored for flexible peak shaving [16] and the introduction of emerging retrofit technologies [17,18]. Thereby, an updated analysis incorporating these evolving policies and practical trends is thus necessary.

Research focused on retrofit-driven mitigation has predominantly evaluated the mitigation potential achieved through the implementation of carbon reduction technologies (CRTs), with particular focuses on advanced technologies such as carbon capture and storage (CCS) [19] and bio-energy with carbon capture and storage (BECCS) [20,21]. Cost-effectiveness assessments have covered extensive CRT portfolios, such as 32 [22] and 25 technologies [23], while frequently lacking dynamic, temporal, detailed projections. Our prior study [23] provided foundational insights into optimal CRT portfolios but similarly omitted the temporal perspective. Thus, this study includes the analysis by optimizing year-by-year unit-level technology portfolios, offering a temporally and dynamically informed view of China’s coal power transition toward carbon neutrality.

This study aims to develop an integrated framework to coordinate unit retirement, new construction, and retrofitting strategies, quantitatively assessing emission reduction pathways and cost-effectiveness under different transition scenarios. Grounded in updated unit-level databases and guided by scenario-based constraints on capacity, emission intensity, and carbon budgets, the framework evaluates retirement mechanisms and the mitigation potential of CRTs, providing detailed insights into low-carbon transition pathways. Based on comprehensive, up-to-date scenario projections from the China Energy Transformation Outlook 2024 (CETO) [8], the analysis investigates two transition scenarios, namely the Baseline Coal Neutrality Scenario (BCNS), prioritizing stability with relatively modest climate action, and the Intensive Coal Neutrality Scenario (ICNS), reflecting aggressive energy transformation and reduced CRT costs.

Overall, this analytical framework identifies optimal, year-by-year technology portfolios for each phase of China’s coal power transition, projecting future capacity structures, retrofit deployment, and investment needs. The study further investigates key criteria influencing retirement decisions and assesses the implications of advancing low-carbon technologies on achieving carbon neutrality. Ultimately, this study provides actionable, sustainability-oriented insights to inform policymakers and stakeholders in advancing China’s coal power sector toward long-term carbon neutrality goals and supporting national sustainable development.

2. Analytical Methods

The analytical framework, illustrated in Figure 2, provides a structured approach to guide the low-carbon transition of China’s coal power sector through three core components. First, the phaseout trajectory analysis schedules unit retirements based on the latest unit-level database (Section 2.1), employing a multi-criteria evaluation method detailed in Section 2.2.1. Second, annual construction of new units (Section 2.2.2) is estimated in line with scenario-specific constraints regarding total capacity and phased emission intensity targets, as elaborated in Section 2.3. Third, retrofit optimization identifies cost-effective annual retrofit portfolios for both existing and newly built units, using a simulated annealing algorithm to minimize the levelized cost of carbon reduction (LCOC), as discussed in Section 2.2.3. Collectively, these components enable integrated planning across coal power capacity structures, retirement schedules, retrofit strategies, and carbon mitigation pathways.

2.1. Data Input

2.1.1. Unit-Level Coal Power Database

To enable comprehensive evaluation and support detailed analyses of phaseout strategies and technology retrofits in China’s operational coal power fleet, an updated unit-level coal-fired power database as of 2024 has been developed. This database integrates data from a variety of authoritative sources, including the China Electricity Council’s (CEC) annual operational and benchmarking reports covering 1367 units updated in 2024 [5], the Global Coal Power Tracker database [24], plant-specific operational data, and publicly available bidding documents. Data for units below 300 megawatts (MW) have been sourced from the 2022 coal-fired power database presented by Wang et al. [23], after removing retired units for accuracy. A list of data sources is provided in

Table A2. List of Data Source (accessed on 13 June 2020).

Data	Source Link	Citation
EDGAR	https://edgar.jrc.ec.europa.eu/country_profile/CHN	[3]
Global Coal Power Tracker	https://globalenergymonitor.org/projects/global-coal-plant-tracker	[24]
bp statistical review of world energy Year 2022	https://www.bp.com/en/global/corporate/energy-economics.html	[77]
Statistical review of world energy Year 2024	https://assets.kpmg.com/content/dam/kpmg/az/pdf/2024/Statistical-Review-of-World-Energy.pdf	[78]
World Energy Outlook 2024	https://www.iea.org/reports/world-energy-outlook-2024	[78]
Global Carbon Budget 2024	https://globalcarbonbudget.org/gcb-2024	[2]
China Energy Statistical Yearbook 2022	-	[6]
China’s Electric Power Industry Statistical Yearbook 2010–2023	-	[79]
Annual Development Report of China’s Power Industry Year 2024	-	[5]
China Energy Statistical Yearbook 2022	-	[6]
China Energy Transformation Outlook Year 2024	https://www.cet.energy/	[8]
Decarbonization Technology Database	-	[23]

of Appendix C.

The database encompasses three categories:

• Static data: geographical location, installed capacity, cooling types, etc.
• Operational data: annual utilization hours, coal consumption rates, power generation, auxiliary power consumption, boiler heat consumption, etc.
• Technological parameters: retrofit application matrix indicating the implementation status of retrofit technologies.

All parameters are collected at the unit level, forming the foundation for the subsequent analysis. To facilitate a detailed assessment, 25 unit types have been classified based on combustion technology, cooling type, and operational configuration, as summarized in

Table 1. Classification of Types of Units.

Marker	Installed Capacity	Combustion Technology	Cooling Type	Function
U1	1000 MW	USC	WC	CHP
U2		USC	WC	PC
U3		USC	AC	CHP
U4		USC	AC	PC
U5	600 MW	USC	WC	CHP
U6		USC	WC	PC
U7		USC	AC	CHP
U8		USC	AC	PC
U9		SCP	WC	CHP
U10		SCP	WC	PC
U11		SCP	AC	CHP
U12		SCP	AC	PC
U13		SBC	WC	CHP
U14		SBC	WC	PC
U15		SBC	AC	CHP
U16		SBC	AC	PC
U17	300 MW	SCP	WC	CHP
U18		SCP	WC	PC
U19		SCP	AC	CHP
U20		SCP	AC	PC
U21		SBC	WC	CHP
U22		SBC	WC	PC
U23		SBC	AC	CHP
U24		SBC	AC	PC
U25	below 300 MW	-	-	-

Note: The abbreviations capture both the technical attributes, namely ultra-supercritical (USC), supercritical (SCP), and subcritical (SBC), and key operational attributes, such as cooling systems (water cooling, WC; air cooling, AC) and plant configuration (pure condensation, PC; combined heat and power, CHP).

.

The current database contains details for 4708 coal power units, covering approximately 99% of the actual total installed capacity in China (1190 GW) [25]. Coverage rates for annual coal consumption and utilization hours are about 98.5% and 82%, respectively. Missing data were imputed using the provincial averages for units of the same type. Accordingly, China’s estimated coal power generation totals 5.39 TWh in 2024, with a deviation of −1.2% compared to the actual generation, with estimated coal consumption of 1.55 Gt standard coal. The average coal consumption rate is approximately 305.9 g of standard coal equivalent per kilowatt-hour (gce/kWh). Correspondingly, total carbon emissions from the coal-fired generation are estimated at 4.44 Gt, resulting in an average carbon intensity of 822.7 g of CO₂ per kilowatt-hour (gCO₂/kWh). Figure 3 illustrates the unit-level carbon intensity distribution.

The largest share of China’s coal power fleet comprises units ranging from 600 MW to 900 MW, together representing roughly 37% of the total capacity and responsible for emitting around 1.83 GtCO₂ annually. Units in categories U2, U21, and U25 are identified as major emission sources, together accounting for approximately 40% of total emissions, highlighting their importance in future emission reduction strategies.

2.1.2. Decarbonization Technology Database

Implementation of CRTs is pivotal in directly reducing coal consumption through improvements in the efficiency of coal-fired power generation, forming a key component of most policies aimed at promoting the development of a low-carbon coal power system [17,26]. Through a comprehensive review of over 380 cases and studies referenced in our previous study [23], the technology database used in this study encompasses eight types of technologies for energy-saving and efficiency enhancement retrofits (T1–T8), as applied in coal power operational systems, as summarized in

Table 2. Key Technology Characterization in the Decarbonization Technology Database.

System	Representative Technology	Marker	Capacity Level	Initial Investment (CNY/kW)	Reduction in Coal Consumption Rate (gce/kWh)	Reduction in Auxiliary Rate (%)	Reference
S1: Boiler System	Boiler Combustion Optimization 1	T1	300 600 1000	8–9.1 11.7–19 6.4–8	1–1.5 1.2–20 0.5–0.8	- - -	[18,27,28,29]
S2: Steam System	Flow Path Transformation	T2	300 600 1000	166.7–200 140–150 165–180	8–15 8–12 7–10	- - -	[27,30,31]
	Optimization of Cold End	T3	300 600 1000	166.7–200 140–150 165–180	8–15 8–12 7–10	- - -	[27,32,33]
S3: Auxiliary System 2	Fan System Speed Control	T4	300 600 1000	18–25 20–25 17–20	- - -	30–40%	[18,27]
	Pump System Retrofit	T5	300 600	36–42 38.3–45	- - -	30–40%	[18,27,34,35]
S4: Full System	Flue gas waste heat recovery	T6	300 600 1000	36.7–45 36.8–50 30.9–44.6	1–2.5 1.2–3 1–2	- - -	[18,27,28,29,36]
	Heating System Retrofit	T7	300 600 1000	71.5–105 58–66.7 90–105	10–15 7–13 6–10	- - -	[37,38,39,40]
	Unit parameter upgrade 3	T8	300 600	167.7–181.8 161.7–170	7.5–12.5 6.5–11	- - -	[18,28,29,41]

¹ Boiler combustion optimization technology enhances combustion efficiency and reduces fuel consumption and pollutant emissions by adjusting combustion parameters such as the fuel-to-air ratio, combustion temperature, and combustion duration. It typically includes measures such as upgrading combustion control systems, pre-treating fuel, retrofitting burners, and optimizing air pre-heaters. ² Auxiliary equipment retrofit involves upgrading and modifying auxiliary equipment in power plants or other industrial facilities to improve overall system performance and efficiency, including pumps, fans, compressors, coolers, and more. Its mitigation performance represents the decrease rate relative to the original auxiliary equipment rates. ³ Unit parameter upgrade indicates the optimization and improvement of units’ key parameters, leading to their enhanced overall performance and efficiency. These parameters include temperature, pressure, flow rate, power, etc., upgrading that can significantly improve the operational effectiveness of the units.

.

Furthermore, the latest low-carbon action plan for the coal power sector, released in July 2024 [42], underscores the importance of retrofitting both existing and newly built plants with three emerging technologies, CCUS, biomass co-firing, and green ammonia co-firing, categorized as T9–T11 in the technology database. These represent either source-based strategies, replacing coal with alternative fuels, or end-based strategies that directly remove carbon from the disposal process.

Among these, CCUS is primarily deployed as point-to-point initiatives with small-scale cluster projects in operation, typically achieving emission reductions of less than 0.1 million tons/CO₂ [43] and lacking stable revenue mechanisms [44]. However, widespread CCUS deployment in coal power by 2060 could lower carbon intensity by 78–109 gCO₂/kWh, with costs diminishing by an estimated 10–20% for each doubling in cumulative installed capacity [43]. Biomass co-firing, with the earliest initiatives dating back to 2005, has advanced at a relatively modest pace and primarily focuses on units of 600 MW or larger [45], yet its recent deployment has been hindered by challenges such as limited availability of compressed biomass feedstock and reduced financial support. Green ammonia co-firing with coal, recommended as one of the key retrofits in 2024 [17], remains in limited use because of high costs associated with green ammonia, with current levelized costs surpassing 4000 CNY per ton [46].

Table 3 summarizes the key technology characteristics assumed for CCUS, biomass co-firing, and green ammonia co-firing up to 2060, all of which are intended for units of 600 MW or larger. While biomass co-firing is already relatively mature, CCUS and green ammonia retrofits are projected to reach maturity around 2035, with associated costs declining by 60% by 2035 and 40% by 2045 relative to current levels. For CCUS retrofits, the primary technological parameter is annual capture capacity, currently on the order of 50–100 million tons per year, with a future trajectory toward full capture. The operational and maintenance (O&M) costs for CCUS, segmented into capture, transportation, and storage costs, are derived from Yuan et al. [20] and supported by values from Wang et al. [23]. These estimates are primarily based on inland conditions and thus do not fully reflect regional cost variations, particularly for coastal deployments.

The co-firing ratios for biomass and green ammonia are assumed to increase incrementally over time. Green ammonia co-firing, although nascent, exhibits considerable promise for decarbonizing coal power plants. However, high green hydrogen prices inflate the cost of green ammonia currently. As hydrogen-production technologies advance, the cost of green ammonia is expected to decline markedly, thereby lowering the expense of ammonia co-firing retrofits. Improved hydrogen-production technologies are projected to cut green-ammonia costs from 689 USD/t to 464 USD/t by 2030 (67% of current) and 295 USD/t by 2050 (42%) [47]. A comprehensive discussion on reducing costs is provided in Section 4.2.

Table 3. Key Technology Characterization in T8–T11.

Technology	Marker	Target Period	Capture Rate 1	Co-Firing Rate 1	Initial Investment (CNY/kW) 2	O&M (CNY/tCO2)	Citation
CCUS	T9	2025–2035	100 million tons 3	-	3417–3925	380–500 4	[20,48]
		2036–2045	90% of total emissions	-	2050–2355	228–300
		2046–2060	100% of total emissions	-	1367–1570	152–200
Biomass-Cofiring	T10	2025–2035	-	10%	573.3	74.7 5	[23,48]
		2036–2045	-	20%
		2046–2060	-	30%
Ammonia-Cofiring	T11	36–54.5	-	10%	49–60	3087–3773 6	[47,49]
		2036–2045	-	20%	49–60	1852–2023
		2046–2060	-	35%	49–60	1234–1509

¹ With ongoing advancements in emerging technologies, these parameters are expected to improve over the targeted time horizons. ² Apart from biomass co-firing, all other technology costs are assumed to decline by 60% by 2035 and 60% by 2045 relative to current levels. ³ For CCUS, the primary technological parameter is the annual capture scale, which is projected to increase over time. At present, most projects capture on the order of 50–100 million tons/yr, with a future outlook toward full capture. ⁴ Projected CCUS O&M costs are based on estimates from Yuan et al. [20], Jia-Hai et al. [48]. ⁵ Biomass co-firing O&M costs include expenditures for compressed biomass feedstock (relative to reduced coal consumption) and equipment operation, with coal priced at 800 CNY/t and compressed biomass at 700 CNY/t. ⁶ Ammonia co-firing O&M costs comprise expenses for green ammonia feedstock, equipment operation, and savings from reduced coal consumption, with coal priced at 800 CNY/t and green ammonia at 3200 CNY/t. The initial investment and O&M are assumed to be not influenced by the co-firing rates. The cost of green ammonia will influence the O&M cost, while not the initial investment, assuming that the initial investment could not vary with time.

Table 3. Key Technology Characterization in T8–T11.

Technology	Marker	Target Period	Capture Rate 1	Co-Firing Rate 1	Initial Investment (CNY/kW) 2	O&M (CNY/tCO2)	Citation
CCUS	T9	2025–2035	100 million tons 3	-	3417–3925	380–500 4	[20,48]
		2036–2045	90% of total emissions	-	2050–2355	228–300
		2046–2060	100% of total emissions	-	1367–1570	152–200
Biomass-Cofiring	T10	2025–2035	-	10%	573.3	74.7 5	[23,48]
		2036–2045	-	20%
		2046–2060	-	30%
Ammonia-Cofiring	T11	36–54.5	-	10%	49–60	3087–3773 6	[47,49]
		2036–2045	-	20%	49–60	1852–2023
		2046–2060	-	35%	49–60	1234–1509

¹ With ongoing advancements in emerging technologies, these parameters are expected to improve over the targeted time horizons. ² Apart from biomass co-firing, all other technology costs are assumed to decline by 60% by 2035 and 60% by 2045 relative to current levels. ³ For CCUS, the primary technological parameter is the annual capture scale, which is projected to increase over time. At present, most projects capture on the order of 50–100 million tons/yr, with a future outlook toward full capture. ⁴ Projected CCUS O&M costs are based on estimates from Yuan et al. [20], Jia-Hai et al. [48]. ⁵ Biomass co-firing O&M costs include expenditures for compressed biomass feedstock (relative to reduced coal consumption) and equipment operation, with coal priced at 800 CNY/t and compressed biomass at 700 CNY/t. ⁶ Ammonia co-firing O&M costs comprise expenses for green ammonia feedstock, equipment operation, and savings from reduced coal consumption, with coal priced at 800 CNY/t and green ammonia at 3200 CNY/t. The initial investment and O&M are assumed to be not influenced by the co-firing rates. The cost of green ammonia will influence the O&M cost, while not the initial investment, assuming that the initial investment could not vary with time.

2.2. Methodology

In this study, the coal power transition pathways encompass projections of both capacity structure and potential carbon emissions. Specifically, the projection of coal power capacity structure involves phasing out outdated units and scheduling the construction of new units, discussed separately in Section 2.2.1 and Section 2.2.2. Based on the resulting capacity structure and the phased carbon-intensity constraints for the coal power sector, annual carbon emissions are subsequently estimated. Any emission gap between these estimated levels and the target levels is closed through the deployment of CRTs, discussed in Section 2.2.3. Collectively, these processes form a comprehensive analytical framework for transitioning the coal power sector under scenarios, allowing for detailed predictions of coal-fired capacity compositions, annual retrofit schedules, and quantification of associated carbon emissions up to 2060.

The framework’s initial input contains the unit-level coal power database, the decarbonization technology database, scenario constraints (e.g., detailed projections of coal power capacity composition, generation forecasts, and carbon-intensity targets), and various fixed parameters (see

Table A3. List of Key Parameters used in the study.

Factor	Value	Unit	Factor	Value	Unit
${\beta }_U$	25 (units below 300 MW)	Years	Coal Heat Value	29.3	MJ/kg
	35 (300-MW units, 600-MW units)		Biomass Heat Value	14.65	MJ/kg
	40 (1000-MW units)		NH3 Heat Value	16.9	MJ/kg
Heating Raduis Constraints	30 before 2030	km	Emission Factor	2.77	tCO2/tce
	50 (2030–2035)		Initial Temperature	1000	Simulated Annealing Algorithm
	80(2035–2040)		Final Temperature	1 × 10−3
Score Assignment	Details in Table 4	-	Number of Iterations	1000
New Construction Ratio	40:55:5 (1000 MW:600 MW:300 MW)	-	Cooling Rate	0.95
Discount Ratio	0.08(r)	-	Advanced Coal Consumption Rate	273	gce/kWh
Technology Lifetime	30	Years	Capacity investments	see Table A1	CNY/MW
Technology Investment	see Table 3	CNY/MW

in Appendix C). The simulation framework was implemented in MATLAB R2024b, with visualizations generated using Origin and Python Version 3.10.

2.2.1. Analysis of Phaseout Trajectory

Phasing out outdated units has been a cornerstone of China’s coal power policy strategies since 2006 [10], with consistent emphasis in government-issued policies and action plans on the systematic decommissioning of obsolete coal power capacity [25,26,42,50]. In this study, projected unit-level phaseouts up to 2060 proceed via two mechanisms: natural phaseout and policy-driven phaseout. Natural phaseout imposes a strict constraint based on predefined operational lifespans, whereas policy-driven phaseout targets inefficient, outdated units that remain even after natural phaseout, determined by multiple criteria. Phaseout decisions are made on a year-by-year, unit-level basis, incorporating factors such as remaining lifespan, coal consumption rates, combustion technology, and capacity, in accordance with available data.

Natural retirement refers to the decommissioning of outdated units upon reaching the end of their operational lifespans, determined by start-year data and established design standards. The global retirement trends of coal power units as of 2023, illustrated in Figure A1, inform the operational lifespans for various capacity levels [24]. With designed lifespans of 30 years [12] or 35 years [51], the following operational lifespan constraints are adopted. Units below 300 MW are assigned a lifespan of 25 years, those at 300 MW and 600 MW for 35 years, and those at 1000 MW for 40 years. This approach ensures that retirement trajectories align with empirical data and design standards, forming the foundation for subsequent phaseout projections.

Currently, the policy-driven retirement primarily targets inefficient units below 300 MW. This mechanism is expected to expand to cover units above 300 MW. In this study, a policy-driven phaseout mechanism for large-scale units (above 300 MW) is activated when the capacity that remains after natural phaseout still exceeds the target coal-power capacity for a given period. This mechanism prioritizes the decommissioning of less efficient, high-consumption units.

For units with capacities below 300 MW, the phaseout decisions follow heating-radius constraints, a policy instrument limiting the operation of small, less efficient units within a defined radius of larger heating sources (i.e., CHP units above 300 MW). The heating radius constraint, initially set at 15 km in 2018 [52], was expanded to 30 km in 2023 [53]. Initially set to 15 km in 2018 [52], the heating radius was extended to 30 km by 2023 [53], and the latest policy of October 2024 further mandates full enforcement of the 30-km constraint by 2025 [17]. Figure A2 illustrates the effects of expanding these radius constraints on the remaining capacity below 300 MW. As shown, applying a 100-km constraint would result in the retirement of around 113 GW, covering 72.6% of all such units. Extending the radius to 150 km would increase the phaseout to around 120 GW, contributing 77.1% of the total. Beyond this threshold, nearly 35.7 GW of units remain unaffected, indicating that these units are either sufficiently remote or uniquely located such that they fall outside any plausible heating radius constraint. In this study, progressively stricter enforcement is assumed: 30 km before 2030, 50 km by 2035, and 80 km by 2040, respectively.

For units above 300 MW, phaseout priorities are established based on a weighted assessment of technical, economic, and environmental indicators, including coal consumption rate, remaining lifespan, combustion technology, and capacity. Informed by prior studies on score-based phaseout priorities [13,51,54], these criteria are aggregated into a policy-driven phaseout mechanism at the unit level.

Table 4. Unit-level Criteria for Policy-Driven Phaseout.

Aspects	Index	Data Type	Details	Score Assignment	Weight
Environmental	Emissions	Quantitative	Emissions per generation	[0,1]	25%
Technical	Utilization hours	Quantitative	Utilization hours relative to theoretical maximum value of 8760	[0,1]	15%
	Age factor	Quantitative	Ratio of remaining lifespan to predefined value	[0,1]	15%
	Combustion technology	Categorical	USC	1	5%
			SPC	0.75
			SBC	0.5
			Other	0.25
	Capacity level	Categorical	≥1000 MW	1	10%
			≥600 MW	0.67
			≤300 MW	0.33
	Usage type	Categorical	Self-Use	0.5	5%
			CHP	1
			Power	1
Economic	Profitability	Quantitative	Strand Assets per generation	(0,1)	25%

details the indices used to guide these decisions.

Carbon intensity (emissions per unit of electricity generated) represents the environmental performance dimension and is given a 25% weight. Due to data limitations, other environmental indicators (e.g., air pollutants and water usage) are not yet incorporated but are expected to be included in future database expansions.

Technical attributes collectively contribute a total of 50%, encompassing utilization hours (15%), age factor (15%), combustion technology (5%), capacity level (10%), and usage type (5%). These indicators enable a rigorous evaluation of operational efficiency and technological advancement at the unit level. For instance, a young 1000-MW CHP unit equipped with USC technology would receive a relatively higher score than older 300-MW units using SBC. The usage type criterion distinguishes among self-use, CHP, and power-only units, with CHP systems prioritized for their enhanced system-level benefits.

The economic dimension, weighted at 25%, is represented by stranded assets. These unrecovered values derive from initial investment ($R^{II}$), unpaid bank loans ($R^{IBL}$), and lost equity returns ($R^{ERE}$) incurred through early retirement [13,55], as formulated in Equation (1). Lifespan constraints ${\beta }_U$ (Section 2.2.1) determine the portion of investment or equity that remains unrecovered at retirement. It is assumed that investments are recovered over a 20-year period, with an 8:2 ratio of bank loan to corporate equity financing [56]. Investment and return data are sourced from China Electric Power Planning & Engineering Institute (EPPEI) [57], Zhao et al. [58].

$$R_i^{SA}=\left\{\begin{array}{cc}{R}_i^{IBL}\times {\displaystyle \frac{15-T_i}{15}}+R_i^{II}\times {\displaystyle \frac{20-T_i}{20}}+R_i^{ERE}\times {\displaystyle \frac{{\beta }_U-T_i}{{\beta }_U}},\hfill & \hfill \mathrm{for}{T}_i<15\\ R_i^{II}\times {\displaystyle \frac{20-T_i}{20}}+R_i^{ERE}\times {\displaystyle \frac{{\beta }_U-T_i}{{\beta }_U}},\hfill & \hfill \mathrm{for}15\le T_i<20\\ R_i^{ERE}\times {\displaystyle \frac{{\beta }_U-T_i}{{\beta }_U}},\hfill & \hfill \mathrm{for}20\le T_i<{\beta }_U\\ 0,\hfill & \hfill \mathrm{for}{T}_i\ge {\beta }_U\end{array}\right.$$

(1)

in which ${\beta }_U$ represents the predefined lifespan for different types of units, mentioned in Section 2.2.1.

By combining environmental, technical, and economic measures, this phaseout approach ensures that any excess capacity, beyond the targets after natural phaseout, is retired based on balanced considerations of efficiency, cost, and emissions, thereby facilitating a more equitable and effective transition in China’s coal power sector.

2.2.2. Analysis of New Construction

China’s rapid economic expansion continues to drive growth in electricity demand, highlighting coal-fired power’s role in stabilizing supply and compensating for variability in renewable generation [59]. This importance has been reaffirmed by widespread electricity shortages since 2020 [60]. Although China has pledged to halt new overseas coal projects and prohibits building coal plants solely for electricity generation domestically, new domestic coal-fired capacity is still being constructed to meet demand growth and support renewable integration. As reported, two-thirds of global coal capacity additions in 2023 occurred in China [24]. These dynamics present significant challenges and opportunities for advancing sustainable energy transitions, making the effective management of coal phaseout and low-carbon retrofitting essential for achieving national and global sustainability objectives.

Projections of future coal power capacity expansion are imposed as scenario-based constraints on total coal-fired capacity. Year-by-year ceilings, derived from phaseout model outputs, limit further coal expansion. Drawing on coal power construction in 2023 and 2024, new construction is split 40:55:5 among 1000-MW, 600-MW, and 300-MW units. By integrating phaseout trajectories with new construction estimates, the study provides a dynamic evolution of China’s coal fleet, capturing both retirements and new constructions. This approach forms the basis for subsequent retrofit optimization, ensuring that efficiency improvements and emission reductions are accurately accounted for in future capacity planning.

2.2.3. Analysis of Retrofit Portfolio

China’s commitment to clean energy has led it to become the world’s largest source of renewable power. Within this broader transition, approximately 740 GW of domestic coal-fired capacity has already undergone energy-saving and carbon-reducing retrofits as part of the “Three Reform Linkage” initiative Research Office of the State Council [61], which includes energy-saving and carbon reduction retrofits, heating retrofits, and flexibility retrofits. Building on the projected evolution of coal power capacity, this study evaluates the additional carbon reduction potential that can be achieved through the strategic deployment of retrofit technologies. Selective retrofitting not only enhances the coal fleet’s alignment with long-term decarbonization goals but also ensures that the transition remains cost-effective and supportive of China’s sustainability objectives.

The core methodology involves optimizing annual technology portfolios to meet mandated carbon-intensity targets at minimal cost, measured via the levelized cost of carbon reduction (LCOC). Carbon reductions for each technology are captured by an environmental indicator, CRP, representing each unit’s carbon reduction potential.

$${\displaystyle CR{P}_{P_k}=\sum _u\sum _{T_i}CR{P}_{P_k,u,T_i}}$$

(2)

where u denotes the unit number and i corresponds to each CRT (1–11). Equations (3)–(6) quantify reductions from lower coal consumption in power generation (for T1–T3, T6–T8), in auxiliary operations (for T4–T5), and from co-firing and CCUS retrofits (for T9–T11).

$$CR{P}_{P_k,u,T_i}=\left\{\begin{array}{lll}{C}_u\times U{H}_{p_k,u}\times {\gamma }_{p_k,u,T_i}\times \Delta {\gamma }_{U_j,T_i}\times \kappa ,\hfill & i=1–3,6–8\hfill & \hspace{1em}\hspace{1em}\mathrm{(3)}\hfill \\ C_u\times U{H}_{p_k,u}\times \Delta {ϵ}_{U_j,T_i}\times {\gamma }_{p_k,u,T_i}\times \kappa \hfill & i=4,5\hfill & \hspace{1em}\hspace{1em}\mathrm{(4)}\hfill \\ C_u\times U{H}_{p_k,u}\times {\gamma }_{p_k,u,T_i}\times \kappa \times {\eta }_{Cap}\hfill & i=9\hfill & \hspace{1em}\hspace{1em}\mathrm{(5)}\hfill \\ C_u\times U{H}_{p_k,u}\times {\gamma }_{p_k,u,T_i}\times {\eta }_{Cap,T_i}\times {\displaystyle \frac{H_{coal}}{H_{fuel,T_i}}}\times \kappa \hfill & i=10,11\hfill & \hspace{1em}\hspace{1em}\mathrm{(6)}\hfill \end{array}\right.$$

Here, $C_u$, $U{H}_{p_k,u}$, and ${\gamma }_{p_k,u,T_i}$ represent the installed capacity, utilization hours, and coal consumption rate of production of unit u in period $p_k$.

Cost-effectiveness of retrofit portfolios is evaluated using the total annual investment $I{T}_{P_k}$, which incorporates both capital expenses (amortized) and operation and maintenance (O&M) costs:

$$\begin{array}{c}\hfill I{T}_{P_k}=\left\{\begin{array}{cc}\hfill & \sum _u\sum _{T_i}\left[C_u\times \left(I{n}_{U_j,T_i}\times {\displaystyle \frac{r{(1+r)}^{t_{T_i}}}{{(1+r)}^{t_{T_i}}-1}}+\Delta O{M}_{U_j,T_i}\right)\right]i=1-8\hfill \\ \hfill & \sum _u\sum _{T_i}\left[C_u\times \left(I{n}_{P_k,U_j,T_i}\times {\displaystyle \frac{r{(1+r)}^{t_{T_i}}}{{(1+r)}^{t_{T_i}}-1}}\right)+\Delta O{M}_{P_k,U_j,T_i}\times CR{P}_{P_k,u,T_i}\right]i=9-11\hfill \end{array}\right.\end{array}$$

(7)

Here, r represents the discount rate, set at 0.08 [23], and In and OM stand for the initial, operational, and maintenance costs of technologies. The LCOC ($LCO{C}_{P_k}$) is thus,

$$LCO{C}_{P_k}={\displaystyle \frac{I{T}_{P_k}}{CR{P}_{P_k}}}$$

(8)

An optimization algorithm is employed to identify the most cost-effective carbon mitigation strategies. The objective is to minimize LCOC for each period while fulfilling the mitigation target. This is carried out using a simulated annealing (SA) algorithm, which efficiently navigates the solution space to reach optimal or near-optimal retrofit strategies. The algorithm begins with the baseline retrofit status of each unit as of 2024, represented by a binary technology application matrix. A value of ‘1’ indicates that a given retrofit has already been implemented, while ‘0’ denotes absence. This matrix, reflecting the most current status, serves as the baseline for future analyses, though it carries inherent uncertainties.

During optimization, the SA algorithm generates candidate solutions by randomly modifying the retrofit assignments of selected units. Each candidate’s LCOC is calculated using the model’s cost and emissions framework. Solutions reducing LCOC are always accepted, while those that increase LCOC may still be accepted probabilistically, with the acceptance probability decreasing exponentially as the algorithm progresses (controlled by the temperature parameter, found in Table 3). This mechanism enables effective exploration of the solution space and helps the algorithm avoid local optima. Based on this, the model generates binary (0–1) matrices for each target year, outlining the planned deployment of CRTs. This ensures that retrofit strategies are not only cost-effective but also aligned with long-term carbon reduction goals.

2.3. Scenario Setup

China’s coal power is expected to remain indispensable in meeting the near-term electricity demand, ultimately being phased out and retained at minimal capacity to serve as backbones through 2060 as the nation advances toward a low-carbon future. As of 2024, China’s total coal-fired power capacity has surpassed 1190 gigawatts (GW), with an additional 420 GW under development [24]. By 2040, coal power is anticipated to evolve into a flexible resource, with its capacity dropping to around 800 GW or less [62], further followed by a complete phase-out by 2060 [8]. Under the 1.5 °C climate target, China’s power sector faces a carbon budget of over 71 Gt CO₂ [63]. Further, recent estimates of China’s total electricity demand now surpass 20,000 TWh by 2060 [8,64], compared to earlier forecasts of around 15,000 TWh [63,65]. This study thus incorporates updated power-sector projections and policy developments to examine pathways as the scenario setup for the exploration of China’s coal power transition, providing a timely assessment of the country’s phaseout strategies and low-carbon energy trajectory.

The scenarios adopted from the CETO [8] are employed in this study, closely reflecting China’s latest development situation and aligning with its updated nationally determined contribution targets (NDCs) and carbon neutrality commitments [66]. The CETO projection is backed by authoritative sources, including the National Development and Reform Commission (NDRC) and the National Energy Administration (NEA). These scenarios embed key constraints, such as carbon intensity targets, renewable deployment goals, etc. [8], which are consistent with China’s 2030 and 2060 climate pledges.

Table 5. Scenario description.

Scenarios	Description
BCNS [8]	Overall, carbon emissions in China will peak by 2030, and carbon neutrality by 2060 on schedule. Coal power reaches carbon peaking before 2030 and neutrality before 2060. Reflects current trends in the energy system. Total electricity demand is projected to reach 20,000 TWh by 2060. Total installed capacity in China is projected to be around 10,530 GW. Carbon intensity projected to decline by 75% in 2035 and 50% in 2040 relative to the 2024 level. The maximum retrofit cost values presented in Table 2 and Table 3 are applied in projecting the technology portfolios. Detailed capacity structure and projected generation figures are provided in Figure 4.
ICNS  [8]	Overall, China’s carbon emissions will peak and reach neutrality sooner than in the BCNS. Total electricity consumption is projected to reach 22,000 TWh by 2060. Total installed power capacity is expected to reach 11,820 GW. Coal power will gradually shift from baseload operation to regulation and backup, with decommissioning occurring over time, ultimately reaching zero operational units around 2050. Carbon intensity projected to decline by 50% in 2035 and 30% in 2040 relative to the 2023 level. The retrofit costs are lower than those employed under the BCNS. The minimum retrofit cost values presented in Table 2 and Table 3 are applied. Detailed capacity structure and projected generation figures are provided in Figure 4.

summarizes these scenarios.

Comparable scenario frameworks have been developed by major international institutions. For instance, the International Energy Agency (IEA) outlines the Announced Pledges Scenario (APS), Stated Policies Scenario (STEPS), and the Net Zero Emissions by 2050 (NZE) Scenario [67] to explore global low-carbon pathways under varying policy assumptions. The EU’s NDCs, including a 13% energy efficiency improvement and a 45% renewable share by 2030, serve as constraints in regional energy transition projection [68]. Additionally, the UNFCCC’s Climate Action Pathways [69] define sector-specific milestones toward mid-century decarbonization, while IRENA’s World Energy Transitions Outlook [70] presents a roadmap emphasizing renewable energy, electrification, and system-level innovation aligned with the 1.5 °C goal.

Both scenarios derive from the latest projections of the CETO, wherein coal-fired units rapidly decrease in utilization, with progressive retirements and advanced retrofits enabling net-zero emissions by around 2050. Figure 4 presents the detailed capacity structures in the power sector and generation trends in both scenarios, highlighting the interplay between renewable expansion and coal retirement. These scenarios serve as a framework for exploring how shifts in capacity, alongside key carbon-intensity targets, can propel China’s coal power sector toward an accelerated decarbonization pathway and sustain its long-term transition to a low-carbon energy system.

3. Results

3.1. Analysis of Capacity Composition and Phaseout Schedule

Natural retirement alone initiates an accelerated capacity decline around 2033, exceeding 10 GW annually. This sharp decline is primarily attributable to China’s relatively young fleet of coal-fired units, causing substantial capacity to retire once these units reach their designed lifespans. By 2037, the remaining operational capacity is projected to fall below 1000 GW, indicating substantial progress in the phase-out of coal power.

Retirements follow a clear size sequence. Smaller units with capacities below 600 MW will predominate retirements through 2045, followed by larger ones with capacities of 600 MW and above from 2046 onward, culminating in the retirement of units with capacities of 1000 MW or larger by 2050. By 2053, all units below 300 MW will be fully retired, leaving the active fleet to consolidate within higher-capacity categories. By 2060, only 10% of the units currently in operation remain, with units of 1000 MW or larger comprising over 70% of the total capacity. Overall, absent policy acceleration, small units (below 300 MW) commissioned between 2000 and 2020, totaling 387.9 GW, would theoretically remain in operation and gradually phase out by 2045. Around 888 units (above or equal to 300 MW) with a total capacity of 411.1 GW installed between 2000 and 2010 are expected to retire by 2050, while a further 520.8 GW commissioned in 2010–2020 would retire by 2060. Nonetheless, recent policy interventions have significantly accelerated retirements, promoting earlier and more aggressive decommissioning of smaller or outdated units.

Policy interventions have significantly accelerated the coal power phaseout process, as illustrated in Figure 5b,c. The analytical framework incorporates both current and forthcoming policy measures, referred to as the policy-driven phaseout mechanism, to capture the substantial impact of regulatory action. Relative to the natural retirement alone, policy mandates under BCNS impose a relatively moderate level of additional retirements, whereas ICNS adopts a more aggressive and expedited schedule, as shown in Figure 5c. Between 2035 and 2040 in ICNS, capacity declines more rapidly than under BCNS, largely due to mandated retirements of 600-MW units since 2036 and 1000-MW units starting from 2046. Consequently, by 2040, around half of the currently operational units are retired under BCNS, around 600 GW, while ICNS spurs even earlier retirements of 300-MW units. By 2060, both scenarios leave only newly built capacity, amounting to roughly 299.5 GW under BCNS and 287 GW under ICNS (Figure 5d). In total, 824.8 GW and 777.9 GW of capacity are subject to policy-driven retirement under BCNS and ICNS, respectively, compared to 199.4 GW and 182.4 GW under natural retirement.

These phaseout trajectories align with China’s carbon neutrality objectives and its long-term sustainability agenda by prioritizing the systematic retirement of smaller, older, and less efficient units. The multi-criteria approach used in this phaseout mechanism ensures that the residual fleet increasingly consists of larger, more advanced technologies, thereby supporting both carbon mitigation and energy efficiency goals. By setting clear phaseout benchmarks, the framework offers a strategic pathway for policy formulation aimed at accelerating decarbonization, reducing resource intensity, and fostering a cleaner and more resilient power system. A more detailed discussion of the weighting and prioritization methods employed can be found in Section 4.1.

3.2. Analysis of Carbon Emissions

The analytical framework used in this study enables quantification of the mitigation potential associated with both unit phaseouts and technology retrofits in China’s power sector. The carbon budget from coal-fired power generation ($Cu{m}_{C{O}_2}$) up to 2060 can be estimated as follows:

$$Cu{m}_{C{O}_2}=\sum _{p_k=2025}^{2060}{E}_{C{O}_2,p_k}=\sum _{p_k=2025}^{2060}({\gamma }_{coal,p_k}\times C_{coal,p_k}\times U{H}_{coal,p_k}\times \kappa )$$

(9)

Here, $E_{c{o}_2,p_k}$ denotes emissions from coal power in year $p_k$, $C_{coal}$, and $U{H}_{coal}$ represent total installed capacity and average utilization hours, respectively. ${\gamma }_{coal,P_k}$ is the average coal consumption rate. $\kappa$ is the emission factor for coal, expressed in grams of carbon dioxide equivalent per gram of standard coal. Thus, carbon emissions from coal power are primarily driven by installed capacity, utilization rates, and generation efficiency.

To evaluate an upper bound of carbon emissions from coal power, a baseline scenario is defined in which electricity demand is met exclusively by coal power, without contributions from alternative energy sources. This provides a theoretical maximum for carbon emissions in the power sector, illustrated by the solid grey lines in Figure 6. This fossil-intensive pathway yields cumulative emissions of 377.3 GtCO₂ (BCNS) and 416.6 GtCO₂ (ICNS) from 2025–2060, based on projected electricity generation of 19987 TWh and 22575 TWh by 2060, respectively. Given the estimated carbon budget for China’s power sector (78–130 GtCO₂) [71], this results in a mitigation gap of 247.3–338.6 GtCO₂, highlighting the scale of emissions reductions required for the coal power sector to support national carbon neutrality targets.

Accordingly, the mitigation from the baseline to actual emission is attributed to three mechanisms: reduction in capacity (M1), decrease in utilization hours (M2), and improvements in coal consumption efficiency (M3). Given the interdependencies among these factors, a simplified sequential decomposition approach is applied. Baseline emissions are calculated using Equation (9), assuming that all projected electricity generation is supplied by coal. Emissions attributed to M1 are estimated by adjusting capacity from the baseline to actual levels while holding utilization hours and efficiency constant. M2-related emissions are then derived by applying changes in utilization hours (from the 2024 level to projected values) to the adjusted M1 capacity.

Under constant 2024 utilization hours and scenario-specific capacity pathways (see Figure 4), capacity reduction (M1) accounts for 69.9% and 73.1% of the carbon budgets under the BCNS and ICNS baselines, respectively. Details on renewable deployment under each scenario are provided in Lyu et al. [8], as depicted in Figure 7. As coal transitions from baseload to a backup role, reductions in utilization hours (M2) yield an additional 41.7–54.2 GtCO₂, or approximately 12.5% of the baseline emissions. Efficiency improvements (M3), encompassing the phaseout of outdated units and deployment of retrofit technologies, contribute a further 10–14.9 GtCO₂, representing 2.3–3.7% of the baseline. Notably, advanced retrofits, including CCUS (M3.4) and co-firing (M3.5), constitute the majority of these reductions, contributing 52.9–61.3% of total mitigation under M3, with peak deployment occurring between 2035 and 2039. Collectively, the cumulative mitigation caused by the reductions in the utilization hours (M2) and the retrofit deployment (M3) is estimated at 51.7–70.1 GtCO₂, accounting for 50.4–56.4% of the carbon emission budget in China’s coal power sector. This aligns with the cumulative carbon emission reduction requirement under the 2 °C goal proposed by Wang et al. [72] (over 50%) and is consistent with the estimated cumulative mitigation of 50.4–64.2% reported by Wang et al. [73].

This decomposition highlights the dominant role of structural transformation (M1 and M2), complemented by technological improvements (M3), in closing the emissions gap and aligning China’s coal power sector with its carbon neutrality goals.

3.3. Analysis on Deployment of Technology Retrofits (T1–T8)

This study categorizes retrofit technologies into two groups: eight mature technologies (T1–T8), expected to be scaled up in the near term, and three emerging technologies, anticipated to be deployed more broadly over the longer term. Figure 8 presents the projected retrofit coverage (by capacity) across unit types for T1–T8 from 2025 to 2032 under BCNS, with corresponding technology descriptions provided in Table 3.

Among T1–T8, retrofits in the boiler and steam systems (T1, T2, and T3) are expected to be widely implemented in U9–U16 unit types, primarily 600 MW units, by 2029. This optimization analysis provides annual or period-specific retrofit recommendations for different unit types. Specifically, T3 (cold end optimization) is projected for extensive adoption in U3 units, which are 1000 MW units with AC and CHP. Adoption rates for T1 and T2 are expected to exceed 70% by 2030, contributing to 564 MtCO₂ in cumulative emission reductions.

From 2030 onward, retrofit deployment accelerates significantly. By 2033, more than 90% of applicable capacity is projected to be retrofitted, with full coverage expected by 2034. In 2032 alone, retrofits are expected to reduce emissions by 0.893 GtCO₂, with cumulative reductions from T1–T8 reaching 2.48 GtCO₂ by 2033. The LCOCs under BCNS and ICNS exhibit temporal variation between 2025 and 2033, ranging from 30.4 to 70.6 CNY/tCO₂ and 39.5 to 84.2 CNY/tCO₂, respectively.

3.4. Analysis of Transition Costs

As demonstrated in the preceding analysis, carbon mitigation in China’s coal power sector is primarily driven by two pathways, namely the substitution of coal with low-carbon power alternatives and the deployment of retrofit technology portfolios (T1–T11). Correspondingly, transition costs are defined as the cumulative investments required from the present onward for new capacity constructions (M1) and technology retrofits (M3), based on investment parameters outlined in

Table A1. Investment Parameters of different power sources [74].

Power	2025	2030	2035	2040	2045	2050	2055	2060
Coal	4.12	4.03	3.95	3.87	3.81	3.73	3.66	3.59
Natural Gas	2.23	2.14	2.09	2.06	2.03	2.00	1.96	1.93
Nuclear	14.77	14.37	13.90	13.42	12.94	12.44	11.94	11.43
Hydropower	14.21	14.43	14.64	14.86	15.09	15.32	15.54	15.77
Wind	6.73	6.06	5.59	5.30	5.00	4.70	4.40	4.09
Solar	4.87	3.82	3.31	3.16	3.01	2.86	2.70	2.55
Biomass	8.65	8.53	8.38	8.23	8.07	7.91	7.74	7.57

. These estimates exclude regional economic variation, carbon pricing, and investments in transmission and storage infrastructure.

Annual transition costs for year $p_k$ ($T{C}_{p_k}$) are estimated following the approach in Wang et al. [74]:

$$T{C}_{p_k}=I{P}_{p_k}+I{T}_{p_k}$$

(10)

where $I{P}_{p_k}$ represents capital investments in newly installed power generation and $I{T}_{p_k}$ denotes expenditures associated with technology retrofits, using Equation (7). The investment used for the construction of new capacity is further disaggregated as

$$I{P}_{p_k}=\sum _{n=1}^{6}I{P}_{p_k,n}=\sum _{n=1}^6(UI{P}_{p_k,n}\times C_{power,n})$$

(11)

Here, $UI{P}_{p_k,n}$ is the unit investment cost for the $n^{th}$ power source in a given year $p_k$, as demonstrated in Table A1. $C_{power,p_k,n}$ is the total capacity of newly built units, derived from the scenario-based projections in capacity provided by [8].

As depicted in Figure 9, the cumulative transition costs are projected to reach around 6.2–6.7 trillion CNY by 2060 under BCNS and ICNS. Of this total, 64–66% is allocated to the investments in alternative power sources, reflecting the dominance of wind and solar in capacity expansion projections before 2045. Between 2025 and 2050, the investments in wind and solar power account for 78% (BCNS) and 85% (ICNS) of the total power investments.

Annual transition costs peak between 2036 and 2045, averaging 1.6–1.8 trillion CNY over five years. Within the coal sector, cumulative retrofit costs (M3) are estimated at 2.2–2.3 trillion CNY. For mature retrofits (T1–T8), cumulative investments total approximately 55 (ICNS) and 135 (BCNS) billion CNY, with average annual expenditures of 7.8 (ICNS) and 18.7 (BCNS) billion CNY during 2025–2029.

As found, CCUS deployment expands significantly post-2035, ultimately comprising 65–76% of M3-related cumulative costs. This underscores the pivotal role of CCUS in shaping the retrofit cost structure, particularly between 2035 and 2045. Although green ammonia co-firing is also a high-cost zero-carbon option, its contribution to total transition costs remains modest due to the declining number of eligible units after 2040 (only 710 GW remaining) and persistently high ammonia costs. Despite projected O&M cost reductions, green ammonia remains economically less competitive over the projection period. A more detailed assessment of retrofit cost trajectories and the implications of declining green ammonia costs is presented in Section 4.2.

4. Discussion

4.1. Weights in the Policy-Driven Phaseout

The unit-level criteria for policy-driven phaseout are detailed in Table 4. These targeted retirements are introduced to address the capacity gap between post–natural retirements and the scenario-based projections of total coal power capacity. The assessment framework integrates seven indicators from the perspective of environmental, technical, and economic aspects. The weighting scheme, as outlined in Section 2.2.1, is adapted from the methodologies proposed by Cui et al. [9,51] and supports the retirement prioritization at the unit level. The underlying principle is to systematically phase out older, less efficient, and technologically outdated units, thereby enhancing the overall efficiency and emissions profile of the remaining fleet.

To test the robustness and reliability of the assessment framework, a sensitivity analysis is performed to evaluate how variations in the weight configurations influence unit-level performance scores and, by extension, the prioritization of units for retirement and associated mitigation potential. According to the capacity trajectory illustrated in Figure 5, the policy-driven retirements are expected to commence in 2036, focusing on the units that remain after natural retirements. This year marks a critical turning point where natural phaseouts alone are no longer sufficient to meet the projected coal capacity reduction targets. Consequently, 2036 is selected for sensitivity analysis as it represents the first year in which the policy mechanism is actively engaged at scale, with the largest volume of units subject to evaluation under multi-criteria retirement scoring. By 2036, the operational fleet is estimated to consist of 1996 units with a total installed capacity of 954.9 GW, of which 71.5 GW is supposed to be phased out to align with the scenario-based capacity constraints.

To assess the uncertainties induced by weight assignment in this policy-driven mechanism, a unit-by-unit uncertainty analysis is conducted using the Monte Carlo method. Specifically, the relative bias of unit scores is employed here and defined as the deviation between the score under a randomly generated weight set and the baseline score (Table 4), normalized by the baseline score itself. This metric enables quantification of the sensitivity of scores at the unit level and thus reveals how sensitive unit rankings are to changes in weight assignments and how such variations influence the distribution of carbon mitigation potential across the fleet.

As illustrated in Figure 10a, the mean relative bias of unit-level scores is 0.11 across a variety of weight configurations, translating into an estimated deviation of 1.1 MtCO₂ in mitigation potential. Correspondingly, the BCNS scenario projects 176.1 MtCO₂ in emissions from the policy-driven retirement in 2036, with an uncertainty of around 0.63%. This accounts for only 0.013% of total mitigation in 2036, indicating that the framework is relatively robust to changes in weight assignments.

Further disaggregation by capacity level, shown in Figure 10b, indicates that units with 1000 MW capacity are more sensitive to the variation in weight variation, exhibiting a mean bias error of 0.14. By contrast, the average bias is 0.079 for 300 MW units and 0.12 for 600 MW units. These findings suggest that larger units, which are more responsive to technical or economic criteria, tend to receive higher scores, introducing slightly greater uncertainty into mitigation outcomes.

Taking China’s two largest coal power provinces as examples, in-depth case studies were conducted for Shandong and Inner Mongolia, each with total installed capacities exceeding 100 GW. Both provinces exhibit similar characteristics, with over 15% of capacity in units below 300 MW and around 10% in 1000-MW units, below the national average. Sensitivity analysis of retirement variations indicates that Shandong and Inner Mongolia exhibit mean relative biases of 0.19 and 0.182, corresponding to estimated deviations of −0.256 and −0.749 MtCO₂, respectively. Correspondingly, the BCNS scenario projects 7.1 and 21.5 MtCO₂ for Shandong and Inner Mongolia in emissions from the policy-driven retirement in 2036, with associated uncertainties of approximately −3.38% and −3.5%. This discloses that the weighting mechanism can lead to region-specific variability in retirement and emissions outcomes, which should be emphasized when refining the regional research.

4.2. Variations in O&M Costs of Green-Ammonia Retrofit

As indicated in Table 3, the O&M costs associated with green-ammonia co-firing retrofits are projected to decrease over time, primarily due to technological maturation and increased deployment scale. A key driver of this cost reduction is the predicted decline in green hydrogen production costs, which currently constitute more than 75% of the total. To evaluate how variations in O&M costs affect mitigation outcomes, a quantitative analysis was conducted across three phases, namely, the initial maturation phase (2025–2035), the widespread implementation phase (2035–2045), and the near-carbon-neutrality phase (2045–2060).

Using BCNS as a baseline, a parametric analysis encompassing over 100,000 combinations of relative O&M cost reductions for green ammonia retrofits across three transition phases, as depicted in Figure 11a. For each combination, a monotonic decline in O&M costs is assumed for the periods 2025–2035, 2035–2045, and 2045–2060, which are assigned to follow absolute downward trends, generating a scenario setup matrix that captures a broad range of potential cost trajectories. These extensive simulations enabled the quantification of cumulative mitigation potential and the corresponding O&M expenditures up to 2060. All results were derived using the analytical framework detailed in Section 2.2.3. This approach provides systematic insights into how retrofit cost trajectories influence long-term decarbonization outcomes, identifying critical thresholds that can inform targeted policy interventions and investment strategies aligned with sustainability objectives.

As found, the cumulative mitigation remains unaffected by declining O&M costs until the widespread implementation phase (2035–2045) reaches approximately 832 tCO₂/CNY (Figure 11b). Notably, reductions in O&M costs during the near-carbon-neutrality phase alone (2045–2060) do not alter cumulative mitigation levels. At this late stage, other cost-effective retrofits (e.g., CCUS and biomass co-firing) have already been maximally deployed, leaving green ammonia co-firing as the primary remaining retrofit option essential for achieving carbon neutrality. Nonetheless, cumulative O&M expenditures vary significantly with cost reductions, as indicated by the increasingly green shading in Figure 11e,f, underscoring the critical role of cost competitiveness during the widespread implementation phase.

Furthermore, the analysis suggests that if green ammonia technology fully matures and O&M costs decrease dramatically to around 832 tCO₂/CNY, the optimal technology portfolios recommended from 2035 onwards will shift considerably. Under such conditions, cumulative mitigation from green ammonia co-firing could increase substantially, from 2.62 GtCO₂ under the BCNS baseline up to ideally 21.8 GtCO₂ by 2060.

5. Conclusions

This study presents an integrated analytical framework to support the low-carbon transition of China’s coal-fired power sector. By incorporating unit-level data, the framework can refine the decision-making process for coal retirements, new constructions, and the deployment of low-carbon retrofits. These facilitate the transition toward a more efficient, low-carbon coal power fleet while ensuring alignment with long-term decarbonization goals.

The phaseout mechanism accounts for both natural retirements and updated policy constraints, including heating radius and operational lifespan limits. It employs a multi-criteria evaluation based on environmental, technical, and economic indicators to prioritize unit retirements to fill in the gap between the projected capacity constraints and the capacity after the natural retirement. This structured approach ensures rational and phased retirements, mitigating the risks of abrupt shutdowns and maintaining system stability. Sensitivity analysis on weight assignments confirms the robustness of the phaseout strategy, providing a minor uncertainty for the total emission, around 0.013%.

Meanwhile, new construction is projected under scenario-specific capacity constraints and emission intensity targets. Scenario constraints are based on projections from CETO 2024 [8], with BCNS reflecting a moderate baseline trajectory up to 2060 following the up-to-date situations, while ICNS indicating a relatively more aggressive decarbonization pathway. Optimization algorithms identify cost-effective carbon reduction strategies across transition phases, providing near-term forecasts for LCOCs and determining optimal retrofit portfolios that maximize carbon reduction while minimizing costs.

The phaseout of outdated units and deployment of retrofit technologies collectively contribute around 10–14.9 GtCO₂ in cumulative mitigation. Total transition costs are estimated at 6.2–6.7 trillion CNY by 2060, with 64–66% allocated to the investments in alternative power sources, primarily wind and solar. Advanced retrofits, particularly CCUS and co-firing, account for 52.9–61.3% of retrofit-driven mitigation, with peak deployment occurring between 2035 and 2039. The impact of green ammonia O&M cost variability underscores the importance of reducing green hydrogen production costs, particularly through cheaper renewable electricity.

Overall, this framework offers a robust, data-driven toolset for integrating phaseout, retrofit, and capacity planning, providing actionable insights to advance China’s transition toward a sustainable, low-carbon power system. To further support the sustainability goals, the establishment of a coordinated retirement mechanism is recommended, particularly for small-scale units, to enhance sector efficiency and environmental targets. Our findings underscore the pivotal role of reducing green hydrogen production costs in enabling widespread adoption of ammonia co-firing, a key pathway for deep decarbonization. Targeted government subsidies and policy support should be directed toward pilot and demonstration projects to accelerate technology maturation. The large-scale deployment of CCUS will also require innovative financing, including blended public–private investment mechanisms and dedicated green finance instruments. Additionally, future research should explore the integration of circular economy principles and cross-sectoral synergies [75,76], particularly examining how data-driven digital transformation and enhanced supply-chain carbon transparency can optimize both environmental and economic performance.

While this study provides valuable guidance for policymakers and stakeholders, future research should further address uncertainties related to technology costs, the broader economic impacts of coal phaseout, regulatory risks, particularly associated with emerging technologies such as CCUS and ammonia co-firing, and market dynamics to strengthen the sustainability and global relevance of China’s transition strategies. In particular, incorporating spatial heterogeneity in CCUS and co-firing deployment, such as regional variations in transport and storage infrastructure, and biomass resource availability, will be critical for enhancing the precision, feasibility, and policy relevance of China’s coal transition strategies.