4

Mechanisms for inter-industrial collaboration in China’s energy transformation

Xiaoli Zhao, Chuyu Sun, Hongjun Zhang and Yarui Deng

Climate change has imposed widespread and profound impacts on ecosystems, human health and socioeconomic systems, becoming one of the most pressing global environmental challenges. Mainstream scientific research has reached a broad consensus that the continuous rise in greenhouse gas (GHG) emissions is the principal driver of global warming (Gu et al. 2009; Lashof and Ahuja 1990; Mohajan 2017). In 2021, Syukuro Manabe and Klaus Hasselmann were awarded the Nobel Prize in Physics for their pioneering work in climate system modelling (von Storch and Heimbach 2022; Navarra 2023). Their research, grounded in physical modelling, quantitatively established the causal relationship between GHG emissions—especially carbon dioxide—and global temperature rise, thereby strengthening the scientific foundation for mitigation-focused climate governance. According to the Intergovernmental Panel on Climate Change (IPCC 2014), 80 per cent of global GHG emissions originate from energy-related activities, including power generation, heating, transportation and industrial processes. Given the dominant role of the energy system in global emissions, its deep structural transformation has become a fundamental pathway to achieving emission reductions. This low-carbon transition requires the coordinated advancement of multiple pathways, including replacing fossil fuels with renewable energy, improving energy system efficiency and optimising end-use energy consumption patterns.

Against this backdrop, China faces a series of complex challenges in its energy transition. In terms of energy structure, coal has long held a dominant position, resulting in carbon emission intensity significantly higher than the global average. At the same time, total energy consumption continues to rise, placing increasing pressure on the scale, dispatch complexity and safe operation of the energy system. China’s inter-regional energy supply and demand are seriously mismatched and there are constraints on grid access and load matching. Renewable energy on the supply side is developing rapidly, but stable and large-scale grid connection capacity on the demand side is insufficient, and problems such as wind and solar abandonment continue to exist. By the end of 2024, China’s installed capacities of wind and solar power had reached 1.407 terawatts and 886 gigawatts, respectively, accounting for 42.03 per cent of total installed power generation capacity (Ding 2025). However, the average utilisation rates of wind and solar power generation are only 95.3 per cent and 96.8 per cent, respectively (National Renewable Energy Consumption Monitoring and Early Warning Centre 2025). This supply–demand mismatch not only reduces the efficiency of renewable energy utilisation, but also increases the burden on conventional energy sources for peak regulation during the transition period. In September 2021, 21 provinces in China experienced electricity shortages and power restrictions were imposed on residents in the three north-eastern provinces (Peng and Xu 2021). This crisis revealed the lack of an effective mechanism to coordinate the phase-out of fossil fuel energy and the growth of renewables, leading to structural instability risks. Moreover, international experience shows that poorly managed structural adjustment in the energy sector can trigger significant socioeconomic disruptions. From 1985 to 2005, coal industry employment in the United Kingdom fell from 221,000 to 7,000 people, with many affected regions still struggling with structural unemployment, inadequate social support and a lack of alternative job opportunities. In the United States, fossil energy employment dropped by more than 30 per cent after 2008 and, despite government intervention, many regions face persistent economic decline (Sainato 2020). These examples demonstrate that the energy transition is not only a matter of technological substitution, but also a systemic challenge involving multiple sectors and governance levels.

Given these challenges, coordinated planning across power supply and demand sectors is essential to push the low-carbon transition of energy structure (Englberger et al. 2021; Huang et al. 2019; Eldeeb et al. 2018; Beuse et al. 2021). Among China’s end-use sectors, transport and heating play a critical role in achieving carbon neutrality. The transport sector accounts for approximately 10 per cent of China’s total carbon dioxide emissions, ranking third after the power and industrial sectors (MEE 2022). The heating sector, especially in northern China, also poses significant challenges. The integration of renewable electricity into heating systems—such as through power–heat coupling or advanced combined heat and power (CHP) retrofits—offers a promising pathway but requires systemic coordination across sectors.

Given the high emission intensity and decarbonisation difficulty of transportation and heating, this chapter selects these two sectors as representative domains to investigate coordination mechanisms between terminal energy-consuming sectors and renewable energy. Accordingly, it focuses on mechanisms for inter-industrial collaboration in China’s energy transformation and presents two empirical case studies. The first explores the integration of renewable energy into the transport sector through photovoltaic–energy storage–charging station (PV–ES–CS) systems, evaluating their economic and environmental performance considering location and scale. The second builds a power–heat coupling model using Beijing as a case study, assessing how renewable energy can support regional heating system decarbonisation. Through these two cases, this chapter aims to reveal the underlying logic and institutional requirements of inter-industrial coordination and offer policy-oriented recommendations for China’s deep energy decarbonisation.

Development pathways for coupled photovoltaic–energy storage–charging station

The coupled PV–ES–CS is an important approach for promoting the transition from fossil energy to low-carbon energy consumption. The use of electric vehicles (EVs) and the installation of distributed rooftop photovoltaics (PV) can form a feedback loop (Kaufmann et al. 2021), which is an efficient approach to integrating distributed PV and EV charging systems (Denholm et al. 2013). In addition, distributed PV is intermittent and volatile, thus requiring energy storage (ES) devices (Omran et al. 2011; Zhang et al. 2019). However, the integrated charging station (CS) is underdeveloped. One of the key reasons for this is that there has been little evaluation of its economic and environmental benefits. Based on the electricity load of different types of buildings and data about EV charging stations in Beijing, we analyse the economic and environmental benefits of integrated charging stations developed over different scales using the capacity optimisation model.

First, we provide a detailed introduction to the PV–ES–CS system, then construct the optimal capacity allocation method of this system and analyse its economic and environmental benefits with the goal of profit maximisation. In the case study, we aggregated several datasets for calculation, including solar radiation data for Beijing and charging data from 21 EV charging stations (covering seven types of buildings such as hospitals and residences). PV–ES–CS locations in different types of buildings determine the different characteristics of a building’s electricity load and EV charging rules, which lead to the heterogeneity of the economic and environmental benefits of PV–ES–CS. We analysed the economic and environmental benefits of different scales of PV–ES–CS in different locations.

The PV–ES–CS we discuss comprises distributed PV power generation modules, ES modules, EV charging modules and seven other power user modules such as hospitals and teaching buildings that are connected to the power grid (Figure 4.1). Besides solar PV, the PV–ES–CS system can buy electricity from power grids at the daily valley price. The electricity from the PV–ES–CS system is used not only for EV charging, but also for other uses within buildings. Because the electricity price for building power users is cheaper than for EV charging, the PV–ES–CS gives priority to charging EVs and then to supplying electricity to building users. The grid system could supply electricity to building users (Sabadini and Madlener 2021), but the electricity prices of ‘PV–ES–CS to building users’ are lower than the prices of ‘grid to building users’ (through peak–valley spread arbitrage, the PV–ES–CS could provide a cheaper electricity price than the grid system), thus building users would prefer to get electricity from PV–ES–CS than from a grid system.

Figure 4.1: Structure of the PV–ES–CS model

Source: Authors’ schema and analysis.

Methods

Objective function

Based on the maximum net present value (NPV) model of CS (Yang et al. 2020), the objective function in this chapter is to maximise the NPV of the PV–ES–CS (Equation 4.1). NPV is calculated by Equations 4.2 to 4.4. In Equation 4.2, NPV is maximised with the decision variables a, denoting the capacity of distributed PV; b, denoting the capacity of ES; and c, the initial investment in the CS (including equipment purchase costs, related facility construction costs and land utilisation costs; c is a positive integer). The NPV equals the discounted annual profit minus the initial investment in distributed PV (capacity a kilowatt), energy storage (capacity b kilowatt hour) and c charging piles, where P_pv, P_s, P_evc,c and P_evc,1 represent the investment costs of distributed PV, ES, each charging pile and land, respectively. The land use of the charging pile is indicated by the symbol neil. Y is the lifecycle of a PV–ES–CS.

Equation 4.1

max NPV

Equation 4.2

The annual profit is calculated by Equation 4.3 as the sum of the hourly profit for the year, HP(t, d, y), minus the annual distributed PV operation and maintenance cost, C_pvm.

Equation 4.3

The annual profit in year y is the summation of hourly profit in year y. As shown in Equation 4.3, the sum of the hourly profit is expressed as , which means the sum of hourly profit in year y from t = 1–T and d = 1–D.

Equation 4.4

The hourly profit is shown in Equation 4.4, comprising three parts. The first is charging income, which is calculated by multiplying the charging capacity of the EV by the electricity price per kilowatt hour. The electricity price consists of the electricity price and the service price, denoted as . The second part is the income from the PV–ES–CS supplying electricity to nearby buildings—that is, multiplying the energy consumption of the building by the commercial electricity price, denoted as . The third part is the cost of purchasing electricity from the grid for the PV–ES–CS, denoted as . The hourly profit is calculated by EV charging income plus nearby buildings’ charging income minus the cost of the PV–ES–CS.

Constraints

The constraints of this model include the following four aspects.

1) Investment constraints

Equation 4.5

The construction of the PV–ES–CS requires investment in distributed PV, ES and CS. Equation 4.5 indicates that the total investment in a PV–ES–CS comprises a, kilowatt capacity distributed PV; b, kilowatt hour ES; and c, charging piles; and does not exceed the fund, MI, owned by the investor.

2) Constraints on power balance

Equation 4.6

Equation 4.6 indicates that the difference between the next period of electricity of the PV–ES–CS and the current period of electricity must be equal to the difference between the power generation of distributed PV, ; the power generation of large power grids, ; the actual charging capacity of EVs, ; and the actual power consumption of users, .

3) Capacity constraints

Equation 4.7

Equation 4.8

Equations 4.7 and 4.8 indicate that the ES capacity per hour cannot exceed the ES capacity of the PV–ES–CS (simplified to b).

Equation 4.9

Equation 4.9 indicates that the power generation of the distributed PV used by the PV–ES–CS cannot exceed its power generation in the current period. The power generation of the distributed PV is calculated by Equation 4.10, in which a represents the installed capacity of the distributed PV; represents solar radiation per hour; represents the photoelectric conversion efficiency; and represents the conversion coefficient between the installed distributed PV capacity and the area of the distributed PV panel.

Equation 4.10

Equation 4.11

Equation 4.12

Equations 4.11 and 4.12 indicate that the charging capacity of EVs in a certain hour cannot exceed the EV power demand, and the remaining power, in the ES.

Equation 4.13

Equation 4.14

Under the limited ES capacity of the PV–ES–CS and the charging priority of EVs over large power users, Equations 4.13 and 4.14, respectively, represent the fact that the power purchased by large power users from the PV–ES–CS cannot exceed the power consumption of the large power users, , and the remaining power in the ES, .

4) Constraints on charging and discharging

Equation 4.15

Equation 4.16

Equation 4.17

Equation 4.15 indicates that the actual charge of EVs cannot exceed the maximum discharge of the PV–ES–CS hourly . Equation 4.16 indicates that the charging power of the large power users cannot exceed the remaining discharge power of the PV–ES–CS, , as EVs and large power users share the output line of the PV–ES–CS during the given period. Equation 4.17 indicates that the electricity purchased by the PV–ES–CS from the large power grid cannot exceed the maximum charging capacity of the power station, .

Model for calculating economic benefits and carbon dioxide emissions reduction

The rate of return on investment (ROI) in a PV–ES–CS can be calculated by Equation 4.18, based on the average annual profit divided by the total investment (Yang et al. 2020).

Equation 4.18

Carbon dioxide emissions reduction is used here to measure the environmental benefits of the PV–ES–CS (Pu et al. 2021), which are calculated by multiplying the use of large-scale grid electricity with the carbon dioxide emissions per kilowatt hour of the large grid—that is, the electricity directly used by the PV–ES–CS from distributed photovoltaics multiplied by the large grid’s carbon dioxide emissions per kilowatt hour (Equation 4.19).

Equation 4.19

In Equation 4.19, represents the carbon dioxide emissions per kilowatt hour of electricity from China’s regional power grid and PVQ represents the consumption of the distributed PV power generation.

Equation 4.20 expresses the distributed PV curtailment rate—an index commonly used to measure the utilisation efficiency of distributed PV.

Equation 4.20

In Equation 4.20, PVQ represents the consumption of the distributed PV power generation, PV.

Case study

Study area

The world’s largest PV–ES–CS in an urban centre was built in China’s capital, Beijing, in 2019 (He 2019). Beijing is both the political and the cultural centre of China (Shang et al. 2021), and building new projects there has greater influence and facilitates the promotion of PV–ES–CS. Therefore, Beijing is chosen for our case study.

In Beijing, the electricity loads of each PV–ES–CS are closely related to their location. PV–ES–CS near different types of buildings have different charging curves for EVs and loads for electricity users, which result in different economic and environmental benefits. We selected typical productive and non-productive buildings from the data of the Ministry of Housing and Urban–Rural Development (2020) to roughly divide installation locations of PV–ES–CS into seven categories: office buildings, factories, teaching buildings, hotels, shopping malls, hospitals and residences.

Data

To calculate the economic and environmental effects of the PV–ES–CS in different locations and at different scales, it is necessary to calculate the distributed PV power generation and EV charging load of the PV–ES–CS.

1) Power generation prediction for distributed PV

The long short-term memory (LSTM) network, a time-series prediction model based on the recurrent neural network (RNN) (Shang et al. 2021), is commonly used to predict distributed PV power generation and performs well in long-term prediction scenarios (Luo et al. 2021b; Ibrahim et al. 2021), as it does not have gradient explosion or disappearance problems over the long-term RNN learning process (Hochreiter and Schmidhuber 1997). The LSTM structure contains four neural network (NN) layers, which are the forgetting gate layer, input gate layer, renew values layer and output gate layer. The four-layer NN helps the LSTM network avoid the long-term dependency problem.

An LSTM model is used in this study. The model is fed with Beijing’s hourly solar radiation data for 27,024 hours from 2017 to January 2020 to regress and predict future solar radiation in MATLAB Version 2020a. With an interval of 100 data points, 70 per cent of the data are used for training the model and the other 30 per cent for validation. A comparison of real and predicted solar radiation data is shown in Figure 4.2, which indicates that the predicted results are credible. The hourly solar radiation data predicted by MATLAB for 20 years are recorded as RA(24, 365, 20), then the distributed PV power generation is , where a represents the installed capacity of the distributed PV, represents the conversion efficiency of the distributed PV and represents the conversion rate between the installed capacity and distributed PV panel area.

Figure 4.2: Real versus predicted data

Source: Data from the US National Oceanic and Atmospheric Administration (www.noaa.gov).

2) Electricity load prediction of charging pile

For conventional EV charging pile load analysis, the charging and discharging behaviour of EVs is generally simulated through data such as those from a ‘Family Travel Survey Report’, as the total load of the charging pile is accumulated from the bottom up. However, China lacks such surveys; it is difficult to fully cover the charging and discharging behaviour of EVs in Beijing through questionnaire data. Studies have shown that the power remaining when EVs drive into a charging pile is random (Luo et al. 2011)—that is, the charging power is independent of the charging start time. The electric load model of CS is constructed in this study through a probability analysis of the hourly EV charging pile discharge data obtained for Beijing. The hourly discharge amount of the charging pile when EV charging behaviour occurs is determined, then whether there is an EV charging in that hour is judged by analysing the EV charging start time.

The hourly electricity load of the charging pile of the PV–ES–CS can be obtained using the above method. Let , … be the hourly charging capacities of given EVs. The Arena simulation tool is used to fit the probability distribution function of hourly charging capacity, F(x). The charging time of EVs can be analysed from the existing data and represented by , which is a variable from 0 to 1, where means that there is no EV charging at that moment and means that there are EVs charging at that moment; is related to the number of EVs. The frequency of doubles as the number of EVs doubles. The charging and discharging of each charging pile in the PV–ES–CS is shown in Equation 4.21.

Equation 4.21

In Equation 4.21, t represents time on the day, d, of the year, y, while represents the probability distribution to which the unit charge of the n charging pile is subject, and c is the number of charging piles that must be optimised.

Hourly charging and discharging data from 21 public EV charging stations in Beijing from 2019 to 2020 are collected with the cooperation of State Grid EV Service Limited, which covers all administrative districts and seven building categories in Beijing. Each charging station has 20 to 30 charging piles. Excluding the abnormalities in the original data—such as EVs that do not charge when plugged in or EVs plugged and unplugged immediately without charging—650,255 valid data points remained. The charging data for each building type are accumulated before calculating the charging capacity per hour; then Arena software is used to generate a distribution function obeyed by the hourly charging capacity. As shown in Figure 4.3f, taking a hospital as an example, bars represent the hourly charge distribution of 8,760 hours and the curve represents the fitted distribution function. The data description item is the descriptive statistics of the CS near the hospital. There are 49,694 effective data points for hospitals in this sample. The average hourly charge value (kilowatts) is 21.5, the maximum value is 58.5 and the minimum value is 0.003. The hourly curve for hospitals is fitted with nine common distributions (for example, β distribution and normal distribution), then the distribution function with the smallest error is selected. The distribution functions of other building types are shown in Figures 4.3a to 4.3e and Figure 4.3g, as well as Table 4.1.

Figure 4.3: Hourly charging capacity distribution curve of charging station

Source: Data from State Grid EV Service Limited.

Table 4.1: Hourly charging capacity distribution of various building types (kilowatts)

	Office building	Factory	Teaching building	Hotel	Shopping mall	Hospital	Residence
Data description
Valid data	43,598	23,826	33,534	62,603	4,226	49,694	95,660
Mean	23.5	25.4	19.3	20	15.8	21.5	21.9
Standard deviation	11.8	13.2	13	11.2	12.8	11.5	11.2
Maximum	59.8	59.4	58.9	57	101	58.5	89.3
Minimum	0.055	0.008	0.002	0.003	0.028	0.003	0.001
Errors under various distribution functions
Beta	0.001	0.001	0.002	0.001	0.003	0.001	0.003
Normal	0.001	0.001	0.002	0.001	0.021	0.001	0.001
Triangular	0.001	0.003	0.002	0.003	0.026	0.003	0.013
Gamma	0.002	0.003	0.003	0.003	0.003	0.003	0.005
Erlang	–	0.004	0.007	0.007	0.002	0.004	0.005
Uniform	0.011	0.006	0.010	0.010	0.050	0.010	0.023
Lognormal	0.005	0.007	0.011	0.007	0.007	0.008	0.012
Exponential	0.014	0.011	0.010	0.010	0.002	0.012	0.020
Weibull	–	0.011	0.008	0.012	0.160	0.002	0.006
Optimal distribution	Beta	Beta	Triangular	Beta	Exponential	Beta	Normal
Distribution charge	60 × beta (2.03, 6.87)	60 × beta (1.61, 6.28)	Tria (0, 2.21, 59)	57 × beta (1.73, 3.21)	expo (15.8)	59 × beta (1.73, 3.07)	norm (21.9, 11.2)
Hourly	3.17	2.26	–	–	–	–	–

Source: Authors’ calculations.

In this study, indicates charging (or not charging) at a certain time. The solid line in Figure 4.4f represents the charging frequency of a CS near a hospital in 2019, the dotted line represents the charging situation in 2020, the coloured lines represent the number of charging EVs in an hour for each charging pile and the black line represents the simulated charging number. The simulation curves fit well for all types of buildings analysed in this study, which validates the proposed model. The charging frequency simulations of CS near other building types are shown in Figures 4.4a–e and Figure 4.4g.

Figure 4.4: Real and simulated charging times for charging station (24 hours)

Source: Data from State Grid EV Service Limited.

3) Other parameters selection

According to the ‘Specification for Photovoltaic Power Generation System Performance’ issued by China’s National Energy Administration (NEB 2020), the usual lifespan of a distributed PV module is 25 years. For ES modules, this chapter selects as the object the lithium iron phosphate battery—a common battery in PV–ES–CS; its configuration costs US$300 per kilowatt hour and the operation and maintenance cost is US$0.30 per kilowatt hour. The lithium iron phosphate battery has a lifespan of 10.91 years (Eldeeb et al. 2018). As a new business model, the PV–ES–CS does not have a unified computing lifecycle standard. If the lifecycle is set to 10 or 15 years, compared with 20 years, the distributed PV part of the PV–ES–CS cannot be fully utilised. If the lifecycle is set to 25 years, three more ESs will be needed, among which the service period of the third ES could be about three years only. In this case, the income and rate of return of the PV–ES–CS for 20 years of service are higher than that for 25 years of service. The lifecycle of the PV–ES–CS is set to 20 years accordingly.

The electricity consumption of power users is taken from Ye (2018), with 1 denoted as U(24, 365, 20). The characteristics of users’ electricity consumption on a typical day are shown in Figure 4.5. The EV charging fee, ; service fee, ; and Beijing peak and valley electricity price, , are shown in Figure 4.6 (Beijing Municipal Commission of Development and Reform 2020b). The exchange rate is from the Bank of China, where the exchange rate was RMB6.5249 to US$1 on 1 January 2021. Other parameters and sources used in this work are listed in Table 4.2.

Figure 4.5: Large users’ electricity consumption (typical day)

Source: Ye (2018).

Figure 4.6: Peak-to-valley tariff in Beijing

Source: Beijing Municipal Commission of Development and Reform (2020b).

Table 4.2: Abbreviated symbols and values

Symbol	Definition	Value	Data sources
	Cost of distributed PV	RMB3,380/kW	China PV Industry Association (www.chinapv.org.cn)
	Cost of ES	US$300/kWh	Liu et al. (2021)
	Cost of each charging pile	RMB533,000/pile	Yang et al. (2020)
	Land cost of charging pile	RMB1,920,000/group	Yang et al. (2020)
	Charging fee of EV (RMB/kWh)	Figure 4.6	Beijing Municipal Commission of Development and Reform (2020b)
	Service fee for EV charging (RMB/kWh)	Figure 4.6	Beijing Municipal Commission of Development and Reform (2020b)
	Change in peak-to-valley in Beijing (RMB/kWh)	Figure 4.6	Beijing Municipal Commission of Development and Reform (2020b)
	Annual operation and maintenance costs of distributed PV power generation	RMB54/kW/year	China PV Industry Association (www.chinapv.org.cn)
	Conversion efficiency of distributed PV	22.8	China PV Industry Association (www.chinapv.org.cn)
	Conversion rate of distributed PV installed capacity and the panel area	10 sq m/kW	Field research
	Discount rate	8 per cent annually	Wang et al. (2018b)
	Marginal emission factor	0.9419 tCO2/MWh	Ministry of Ecology and Environment

Source: Authors’ compilation.

Economic and environmental impact of PV–ES–CS in different locations and scales

There are obvious differences in the economic and environmental effects of the PV–ES–CS near different types of buildings. The return on investment of the PV–ES–CS calculated under different scales by building type is shown in Table 4.3. The distribution of PV, ES and CS can be identified when the maximum return on investment is achieved. The distributed PV installed capacity, ES capacity and number of charging piles are all non-zero, indicating that the return on investment of the PV–ES–CS is better than that of distributed PV power generation (a ≠ 0, b = c = 0), ES power station (b ≠ 0, a = c = 0), EV charging station (c ≠ 0, a = b = 0), PV ES power station (a, b ≠ 0, c = 0) or ES charging station (b and c ≠ 0, a = 0).

The economic and environmental effects of PV–ES–CS at different locations and scales can be summarised as follows.

Table 4.3: Maximum return on investment for PV–ES–CS

	Teaching building	Hotel	Shopping mall	Hospital	Residence	Office building	Factory
Maximum investment return rate	12.58%	11.23%	12.81%	13.92%	1.42%	9.81%	4.29%
Investment (RMB million)	26	16	39	41.5	8.5	13	13.5
NPV (RMB thousand)	24,396	12,591	37,955.7	44,954.7	–1,094.6	8,363.6	530.1
Distributed PV curtailment rate	21.92%	27.39%	26.50%	18.68%	36.99%	31.70%	29.20%
CO2 emissions reduction (thousand tonnes)	9.36	5.33	13.66	16.30	0.40	3.67	4.46
a (kW)	2,440.91	1,493.87	3,785.68	4,082.30	128.84	1,095.02	1,282.69
b (kWh)	5,258.67	2,888.29	8,206.03	8,728.05	184.27	2,312.60	2,246.25
c	5	5	5	5	5	5	5

Source: Authors’ calculations.

1) Investment return ratio of PV–ES–CS is highest in the hospital scenario and lowest in the residential scenario

Figure 4.7 shows that if there is sufficient investment, the return on investment of the PV–ES–CS falls by building type from high to low as follows: hospitals, shopping malls, teaching buildings, hotels, office buildings, factories and residences. According to the rate of return on investment, the PV–ES–CS in different scenarios can be classified into three gradients. The first gradient has the highest return on investment—namely, hospitals and shopping malls; the second is hotels, teaching buildings and office buildings; and the third gradient is factories and residences, which have the lowest return on investment. Among them, a CS near a residence has no investment value because the rate of return is lower than the current deposit interest rate of the Bank of China.

Figure 4.7: Return on investment of various PV–ES–CS with different levels of investment

Notes: On the one hand, a PV–ES–CS system should contain at least 1 kW PV for RMB3,380 (China PV Industry, 1 kWh ES for RMB1,957.47) (Liu et al. 2021) and 1 charging pile for RMB2.45 million (Yang et al. 2020)—that is, the smallest investment should be more than RMB3 million. On the other hand, the largest investment should be lower than RMB50 million; the return on investment of a PV–ES–CS system above this amount is less than zero. Therefore, we set the investment constraint of PV–ES–CS between RMB3 million and RMB50 million for all scenarios.

Source: Authors’ calculations.

2) Developing PV–ES–CS near hospitals is more conducive to carbon emissions reduction

The carbon emissions reduction of the PV–ES–CS is related to the curtailment ratio of distributed PV. Under the same investment level, a lower curtailment rate results in a larger reduction of carbon dioxide emissions. With sufficient investment, the PV–ES–CS falls into descending order of carbon dioxide emissions reduction per unit of investment (Figure 4.8): hospitals (3.87 kg/RMB), shopping malls (3.46 kg/RMB), factories (3.30 kg/RMB), hotels (3.24 kg/RMB), teaching buildings (3.18 kg/RMB), office buildings (2.61 kg/RMB) and residences (0.57 kg/RMB), which is basically consistent with the economic ranking. When the diurnal power load of the building is close to the distributed PV power generation curve and nocturnal power consumption is relatively high, carbon dioxide emissions are reduced to a greater extent by the PV–ES–CS. Buildings use a large amount of electricity from the ES at night, leaving enough space for consuming distributed PV during the day and reducing the rate of curtailment. When the building load curve during the day is closer to the distributed PV power generation curve, the distributed PV curtailment rate is relatively low. Hospitals show the largest nocturnal power consumption and diurnal power load curve closest to the distributed PV power generation curve. Therefore, building a PV–ES–CS near a hospital can maximise the use of the distributed PV, thereby reducing carbon dioxide emissions from the large power grid.

Synergistic development of heating system decarbonisation and renewable energy increase

Chinese policymakers are actively exploring the path to heating system decarbonisation. For instance, the Fourteenth Five-Year Plan for Renewable Energy Development, jointly issued in 2022 by nine national departments (NDRC 2022), states that the utilisation scale of geothermal heating, biomass heating, biomass fuel and solar thermal for non-power generation will exceed 60 million tonnes of standard coal equivalent by 2025. And the Implementation Opinions on Further Accelerating the Application of Heat Pump Systems and Promoting Clean Heating issued by the Beijing Municipal Commission of Development and Reform in 2019 provided that, by 2022, the utilisation area of the new heat pump system in Beijing would reach 20 million square metres, and the cumulative utilisation area 80 million square metres, accounting for about 8 per cent of the total heating area, to substantially enhance the application of the heat pump system in the city. However, the city’s heating system decarbonisation has not been achieved because the heating area provided by renewable energy remains limited, and the primary energy supply structure of the heating system is still dominated by fossil fuels (NBS 2005). Given that renewable energy power generation will dramatically grow in the future, it is crucial to step onto the path to achieve the synergistic development of renewable energy generation and heating system decarbonisation.

Figure 4.8: Carbon dioxide emissions reduction per unit of investment in PV–ES–CS

Source: Authors’ calculations.

Beijing sits on the North China Plain. As China’s capital, it plays a leading role in national carbon peaking and carbon neutrality initiatives (Huang et al. 2020; Beijing Municipal People’s Government 2022). De-coalification has basically been realised in Beijing, however, the city is still under enormous pressure to cut carbon dioxide emissions from natural gas and oil usage. Beijing has implemented several policies aimed at reducing carbon dioxide emissions, such as its Fourteenth Five-Year Plan for Ecological and Environmental Protection, issued in 2021 (Beijing Municipal People’s Government 2021). It stated that, by 2025, the total carbon dioxide emissions in the city would be more than 10 per cent lower than the peak in 2020, which was 148 megatonnes (Huang et al. 2020; Zhang et al. 2019; Beijing Municipal People’s Government 2017a; Beijing Municipal Bureau of Statistics 2021b), and the proportion of renewable energy consumption would be about 14 per cent.

The current energy system in Beijing mainly comprises the power, heating and transport sectors. Its local power supply system is dominated by natural gas generation, which amounted to 39.7 terawatt hours in 2020, accounting for 87.1 per cent of local power generation, whereas renewable energy, such as wind, PV, hydropower and waste incineration power generation, accounted for only 9.87 per cent of local power generation (CEC 2021), as shown in Figure 4.9. Although Beijing’s power system has moved away from coal, supply is still dominated by fossil fuels, with a low proportion of renewable energy. Furthermore, as shown in Figure 4.10, only 40 per cent of the electricity consumed in Beijing in 2020 came from local electricity production, with the remaining demand met by imported electricity. As a result, increasing the amount of local renewable energy is important to decarbonise the power and heating systems in Beijing.

Beijing’s heating system, which uses natural gas as the primary energy input, is separated into two segments: district heating and individual heating. The district heating area accounts for 73.67 per cent of the total heating area (MoHURD 2021; Beijing Municipal Commission of Development and Reform 2020a). Gas-fired combined heating and power meets 9 per cent of the district and 6.89 per cent of the individual heating demand. The analysis of Beijing’s heating structure shows that the city has established a clean heating system. However, due to the technical structure based on gas-fired boiler heating, additional steps must be taken to replace natural gas with renewable energy to achieve complete decarbonisation of the system.

Figure 4.9: Beijing’s local power supply structure in 2020

Source: Calculated based on data for power generation by region in CEC (2021).

Figure 4.10: Beijing’s energy balance in 2020 (terawatt hours)

Sources: Based on data in MoHURD (2021); CEC (2021); Beijing Municipal Bureau of Statistics (2021a).

Future energy system design in Beijing

Key technical paths for heating system decarbonisation

The technical paths to heating system decarbonisation in the existing policy documents mainly include ground-source heat pumps, solar thermal, ‘coal-to-electricity’ and ‘coal-to-natural gas’ central heating pilot projects and power plant waste heating (Beijing Municipal People’s Government 2017a), which must be fully considered when setting the first technical path scenario in this chapter. This scenario serves as a baseline for other scenarios for heating decarbonisation. Considering Beijing’s industrial structure, resource endowment and infrastructure construction, the natural gas demand in the heating system should be further reduced based on the technical paths mentioned in the existing policies during the design of the heating system decarbonisation path (Beijing Municipal People’s Government 2021). Therefore, it is necessary to replace natural gas heating with cleaner electricity, which is the second technical path scenario in this chapter—that is, replace traditional gas heating with electric technologies (such as air-source heat pumps). In addition, in the context of China’s dual carbon targets, more renewable energy should replace fossil-based energy and cross-sector application of energy technologies should be employed (Luo et al. 2021a). So, it is necessary to analyse how cross-sectoral energy technologies will help achieve synergistic development of the electricity and heating sectors in the low-carbon transition. The third technical scenario for heating system decarbonisation in this chapter is the synergistic development scenario, which is mainly based on improvements in the traditional technology for generating combined heat and power—that is, replacing fossil fuels with renewable energy.

As an efficient production technology, combined heat and power (CHP) can not only achieve the synergistic development of power and heating systems, but also has great potential in reducing energy costs and carbon dioxide emissions (Zhu et al. 2021), and is widely used in the design of the district heating system decarbonisation path (Jimenez-Navarro et al. 2020; Aunedi et al. 2020; Olympios et al. 2020). However, CHP essentially burns fossil fuels for energy production, making it impossible to fully eliminate carbon dioxide emissions. How to replace fossil fuel with renewable energy in CHP affects not only carbon dioxide emissions reduction but also large-scale penetration of renewable energy in the future. As a result, this chapter focuses on the CHP generation of solar thermal power plants for future development. When intermittent renewable electricity is generated, the excess power is transferred to an electric heater to heat salt, which is transported to a high-temperature molten salt tower. The high-temperature steam generated by the molten salt is used to heat water and drive a steam turbine unit to generate electricity. A thermal power plant’s steam turbine generator can be employed as a peak-shaving power station for wind or PV power generation (Wu et al. 2017). The new development modes for CHP generation technology can fully utilise the infrastructure resources of a thermal power plant and accomplish the synergistic development of renewable energy and heating system decarbonisation. Finally, it is worth noting that the development of wind and hydropower requires vast natural resources and land; in contrast, the development of distributed photovoltaics is more flexible in terms of location. Therefore, the scenarios in this study focus on the synergistic development of large-scale applications of heat pumps and CHP in combination with distributed photovoltaics.

Forecasting the future energy demand in Beijing

The dynamic trend of Beijing’s future energy demand must be analysed to design the city’s large-scale renewable energy system. Beijing’s power consumption was 114 terawatt hours in 2020 (CEC 2021) and is expected to reach 268.76 terawatt hours by 2050 based on a 2.9 per cent annual growth rate (Beijing Municipal People’s Government 2017a). Electricity imported to Beijing from other provinces accounts for 30.3 per cent of future total energy consumption, including that generated by coal, oil and natural gas, and is expected to reach 127.96 terawatt hours in 2050 (Beijing Municipal People’s Government 2017a).

The heating demand in Beijing is determined by both heating area and heating supply per unit area. The total heating area in the city was 895 square kilometres in 2020 and 1,000 square kilometres in 2022 (Beijing Municipal Commission of Development and Reform 2019). If the heating area in Beijing continues to grow at its current rate, it will reach 4,725.82 square kilometres by 2050. Thus, the total heating demand will reach 203.03 terawatt hours (TWh) by 2050, considering future improvements in buildings’ energy savings (Beijing Municipal People’s Government 2022). The dynamic changes in future heating demand in Beijing are listed in Table 4.4.

Table 4.4: Forecast future heating demand in Beijing

Year	Heating area (sq km)	Heating demand per unit area (TWh/sq km)	Total heating demand (TWh)
2020	895	0.0808	72.35
2030	1,558.50	0.0655	102.05
2050	4,725.82	0.0430	203.03

Note: It is assumed that the heat supply per unit area in Beijing will fall by 10 per cent every five years in the future.

Source: Authors’ forecasts.

Description of future energy transition scenarios in Beijing

This chapter proposes three development modes for Beijing’s future power and heating system based on the city’s energy supply structure and important decarbonisation and transition technologies, including the business-as-usual (BAU) scenario, the electricity substitution (ELS) scenario and the synergistic development (SD) scenario. The operation and transition of Beijing’s power and heating systems in the three scenarios are simulated year by year from 2020 to 2050. The design and key assumptions of the three scenarios are described in the subsections that follow.

1) Business-as-usual (BAU) scenario

Figure 4.11 depicts the BAU scenario, which is similar to Beijing’s energy system in 2020. We analyse the technical path for Beijing’s future heating system decarbonisation under existing policy conditions. In the BAU scenario, Beijing’s heating system mainly uses fossil fuel and various renewable energy technologies, such as ground-source and air-source heat pumps. Beijing’s future electricity demand, heating demand and renewable energy development under the BAU scenario will be the benchmarks for developing future energy system transition pathways and will be determined primarily by policy documents such as Beijing’s Energy Development Plan during the Thirteenth Five-Year Plan Period, Thirteenth Five-Year Plan for New and Renewable Energy Development and Fourteenth Five-Year Plan for Ecological and Environmental Protection (Beijing Municipal People’s Government 2017a, 2017b, 2021).

In the BAU scenario, the replacement of coal-fired and oil-fired units in the power system will be completed by 2025. The installed capacity of gas turbines will rise at a rate of 3.3 per cent until it reaches the peak in 2025 (Beijing Municipal People’s Government 2017a, 2021). The scale of wind and hydropower development remains unchanged, whereas the installed capacity of photovoltaic and biomass power generation is expected to rise at the same rate as in 2020. The future electricity supply structure in the BAU scenario is presented in Table 4.5. The heating system is divided into district heating and individual heating. Table 4.6 lists the future heating supply structure in Beijing. Since the electrification transition pathway for the transport sector is not considered in this chapter, the rate of change in electricity consumption in the transport sector is assumed to be in line with overall energy consumption.

Figure 4.11: Conceptual diagram of Beijing’s energy system

Source: Authors’ schema.

Table 4.5: Beijing’s electricity supply structure under the business-as-usual scenario

Year	Coal (MW)	Oil (MW)	Natural gas (MW)	Wind (MW)	Hydro (MW)	PV (MW)	Biomass (MW)	Import (TWh)
2020	770	200	10,000.00	190	990	560.00	370.00	68.59
2030	0	0	11,762.55	190	990	4,011.84	644.97	84.44
2050	0	0	11,762.55	190	990	167,975.67	1,959.80	127.96

Source: Authors’ forecasts.

Table 4.6: Beijing’s heating structure under the business-as-usual scenario (square kilometres)

Year	District heating area				Individual heating
	Gas boiler	Gas CHP	Waste incineration	Ground-source heat pump	Gas boiler	Gas CHP	Air-source heat pump
2020	457.52	59.37	23.32	35.00	154.42	16.23	65.00
2030	538.16	69.83	50.19	124.68	181.64	19.09	574.92
2050	538.16	69.83	232.43	378.07	181.64	19.09	3306.61

Notes: The replacement of coal-fired and oil-fired heating boilers and CHP generation units in the fossil energy heating technology in district heating will be completed by 2025. Beijing’s Fourteenth Five-Year Plan for Ecological and Environmental Protection (Beijing Municipal People’s Government 2021) is the source of the future development trend for gas heating. The heating area of ground-source heat pumps accounts for 8 per cent of the overall heating area and remains unchanged. The heating area of waste incineration technology is computed using the installed capacity, utilisation hours and thermal efficiency of biomass thermal power plants. The future trend for fossil fuel heating in individual heating is consistent with that for district heating, and the remaining individual heating demand is met by air-source heat pumps.

Source: Authors’ forecasts.

2) Electricity substitution (ELS) scenario

Unlike the BAU scenario, under which gas dominates the heating structure, the ELS scenario emphasises the use of heat-pump technology to replace gas heating. This substitution mode affects the structure of the heat and power supplies. After electrical heating replaces gas heating, such as gas boilers and gas CHP generation, additional electricity generation technologies—for example, renewable and imported electricity—will be used to also replace gas-generated power. Therefore, the electricity and heating systems must be rebuilt based on the BAU scenario.

Replacing gas heating with air-source heat pumps will increase power consumption. Consequently, Beijing’s power consumption will reach 280.34 terawatt hours in 2050. Except for the replacement of fossil energy power production technology, the changes in other power production technologies are consistent with the BAU scenario. Furthermore, the ELS scenario focuses on decarbonising the energy system when gas heating is replaced with electric heating. Thus, it ignores the influence of transport electrification. The future development mode of the transport sector is assumed to be consistent with the BAU scenario.

The assumptions for the future total heating area and the heating demand per unit area are consistent with the BAU scenario. The trend for fossil fuel heating before 2025 was consistent with the BAU scenario. After 2025, the area for gas heating will decline until 2050, when natural gas consumption in the heating sector is completely replaced with electricity. The trends for the other heating technologies are consistent with the BAU scenario. Table 4.7 presents Beijing’s future heating structure under the ELS scenario.

Table 4.7: Beijing’s future heating structure under the electricity substitution scenario (square kilometres)

Year	District heating area				Individual heating
	Gas boiler	Gas CHP	Waste incineration	Ground-source heat pump	Gas boiler	Gas CHP	Air-source heat pump
2020	457.52	59.37	23.32	35	154.42	16.23	65
2030	430.53	55.86	50.19	124.68	145.31	15.27	736.66
2050	0	0	232.43	378.07	0	0	4,115.33

Source: Authors’ forecasts.

3) Synergistic development (SD) scenario

Although under the ELS scenario fossil fuels are replaced in the heating system, the electricity system will continue to consume a certain amount of fossil fuels to ensure stability. Therefore, not all the electricity consumed by heat pumps comes from renewable sources, preventing the ELS scenario from achieving 100 per cent renewable power and heating systems. We therefore propose the synergistic development (SD) scenario under which the energy consumption of CHP units will gradually shift from the combustion of fossil fuel to renewable energy. It is assumed that the individual heating development model under the SD scenario is consistent with the ELS scenario. The heating area of ground-source heat pumps and waste incineration in district heating systems is comparable with the BAU scenario. In terms of fossil fuel technology, it is expected that thermal power units will be retrofitted, replacing fossil fuel combustion with renewable generation based on the operation of the BAU scenario and the technical principle of CHP generation. Under the SD scenario, zero consumption of fossil fuel in the heating system can be achieved by 2050. Therefore, district heating is the same under the SD and BAU scenarios, while individual heating is comparable with that under the ELS scenario. The key difference is that gas-fired units are powered by renewable energy generation rather than by fossil fuels in the SD scenario. Figure 4.12 depicts the power and heating system operation modes under the SD scenario.

Methods

Energy system simulation tool: EnergyPLAN

The EnergyPLAN model is used to simulate the various sectors of the energy system in Beijing, and the model’s accuracy is validated using the current state of Beijing’s energy system. The EnergyPLAN model has several advantages as a bottom-up energy system analysis platform: 1) the model is suitable for the simulation of energy systems at various spatial scales; 2) the model can simulate the hourly technical operations of the energy system; 3) the model has advantages over existing institutional frameworks by enabling the adjustment of multiple operational methods and reducing the exogenous impact of the current system on the high proportion of renewable energy in the future; and 4) the model is highly efficient and can quickly compare the technical alternatives of various energy systems and adjust the operating strategy of the energy system in real time.

Figure 4.12: Synergistic development of renewable energy and fossil fuels

Source: Authors’ schema.

Figure 4.13: Schematic diagram of the EnergyPLAN model and the structure of Beijing’s electricity and heating sectors

Source: Authors’ schema.

The EnergyPLAN model is primarily utilised in the following areas to accomplish the strategic design of large-scale energy systems through technical and economic analysis of possible technology combinations: 1) determination of the transition pathway for a high-proportion renewable energy system (Hansen et al. 2019; Connolly et al. 2011, 2016; Sáfián 2014; Nielsen et al. 2011; Thellufsen et al. 2020; Kwon and Østergaard 2013); 2) the widespread impacts of certain technology on the overall operation of the energy system (Zhang et al. 2019; Askeland et al. 2019; Groppi et al. 2019; Lund and Salgi 2009; Thellufsen et al. 2019; Hedegaard et al. 2012); 3) evaluation of different energy policies and development strategies (Xiong et al. 2015; Ma et al. 2014); and 4) interconnection of various energy systems (Huang et al. 2020; Askeland et al. 2019; Alves et al. 2020; Drysdale et al. 2019). In addition, the EnergyPLAN model can be used to design energy systems at different levels, including urban (Zhang et al. 2019; Luo et al. 2021a), regional (Yuan et al. 2020; Huang et al. 2020) and national (Hansen et al. 2019; Xiong et al. 2015).

Figure 4.13 depicts the entire structure of the three decarbonisation scenarios for Beijing’s power and heating systems using the EnergyPLAN toolbox. The inputs of the model involve all aspects of the energy system, which can be divided into five categories: energy demand, development scale of energy production and storage technology, cost, energy system operating strategy and carbon dioxide emissions. In addition, since renewable energy is intermittent and unstable, the model requires time-series data on energy demand and renewable power production at an hourly resolution, facilitating the matching of energy supply and demand in systems with variable power sources. In terms of energy system output, the EnergyPLAN model employs various indicators to evaluate the operational status of the energy system under different technology combination modes, including carbon dioxide emissions, excess power production, socioeconomic costs, proportion of renewable energy and primary energy structure. Although the EnergyPLAN model fails to integrate renewable energy generation into CHP under the SD scenario, the operation of thermal power plants before and after the transition is merely the difference in fuel supply rather than a change in power generation technology. As such, EnergyPLAN can be used to indirectly simulate the energy supply after the thermal power plants’ transition under the SD scenario. The detailed steps are as follows: 1) operation of the thermal power units is assumed to be dependent on the burning of fossil fuels under the SD scenario; 2) the renewable energy generation required to replace fossil fuel combustion in CHP units can be calculated based on the outputs obtained from the model—the power supply heat of the thermal power unit, the proportion of retrofitted gas-fired units, the power generation efficiency of the retrofitted CHP unit and the utilisation efficiency of renewable energy in the CHP generation unit (Zhang et al. 2019).

Model validation

In this section, the simulated power and heating supply structure of Beijing’s energy system in 2020 was compared with the real data from Beijing in 2020 to validate the EnergyPLAN model (Table 4.8). The real data for power production were obtained from the China Electricity Council (CEC 2021), the district heating production data were from the Ministry of Housing and Urban–Rural Development (MoHURD 2021) and the individual heating production data were the difference between the total heating area and the district heating area in 2020. The results show that the relative error between the real data and the EnergyPLAN model simulation results is less than 2 per cent. Evidently, the EnergyPLAN model can accurately simulate Beijing’s energy system and can be used to simulate the city’s future energy system transition scenarios.

Table 4.8: Validation of the model of Beijing’s energy system in 2020 (terawatt hours)

Indicator	Actual data	EnergyPLAN simulation	Difference (%)
Electricity
Thermal	41.1	41.01	0.22
Hydro	1.1	1.1	0
Wind	0.4	0.4	0
PV	0.6	0.6	0
Waste incineration	2.3	2.31	0.43
Renewables share of electricity	3.95%	3.9%	1.28
Heating
District boiler	43.01	43.01	0
District CHP	5.58	5.58	0
Individual heating	19.05	19.04	0.05

Source: Authors’ validations.

Cost data

It is necessary to consider local resource endowments, the matching of energy supply and demand, system operating stability and the costs of various combinations of technologies—that is, the total annual cost of the energy system—to identify the optimal decarbonising strategy for the future energy system in Beijing. The EnergyPLAN model divides the total annual cost of the energy system into four components: investment cost, operation and maintenance (O&M) cost, fuel cost and external electricity prices. The technology lifecycle and interest rates should also be considered in the total system cost (Yuan et al. 2020). Most of the cost data come from the China model cost dataset (Luo et al. 2021a; Xiong et al. 2015) and the EnergyPLAN model cost database; fuel price information comes from Luo et al. (2021a) and fuel handling and other O&M costs come from Sorknæs (2020). Appendix 4.1 provides detailed cost statistics for various technologies, fuel price data and predictions of future costs for various technologies and gasoline prices.

Results and discussion

The impact of different heating decarbonisation paths on renewable energy integration

The impact of the decarbonisation of different heating sectors on renewable energy integration differs substantially between the BAU, ELS and SD scenarios. Figure 4.14 demonstrates that the local renewable energy share in total electricity and heating demand in Beijing is the lowest under the BAU scenario (72.57 per cent), without considering the imported power supply structure. Under the ELS and SD scenarios, the proportion of local renewable energy is more than 80 per cent and 100 per cent, respectively, of total electricity and heating demand in Beijing. Although fossil energy is fully replaced with renewable energy under the ELS scenario, thermal power units must produce power to match supply and demand in the power system in real time. Therefore, the electricity and heating demands cannot be met by 100 per cent renewable energy integration under the ELS scenario. Under the SD scenario, however, the operating mode of the retrofitted thermal power unit is essentially a fuel supply change rather than technical substitution. As a result, even if many thermal power production technologies are utilised under the SD scenario, renewable energy could meet 100 per cent of the electricity and heat demand.

In addition, although the ELS scenario achieves a high proportion (more than 80 per cent) of locally generated renewable energy, imported electricity still accounts for more than half of the electrical supply in Beijing—greater than under the SD scenario (see Table 4.9). The reason for this is that renewable energy cannot fill the gap resulting from the phasing out of gas-fired units in terms of the hourly balancing of energy supply and demand, so imported power is needed to achieve this balance. However, we are unable to determine the amount of imported decarbonised electricity. Therefore, we assume that the share of imported electricity in future power and heating systems is as low as possible to reduce carbon dioxide emissions. The SD scenario outperforms the ELS scenario in terms of both renewable energy integration and power supply structure.

Figure 4.14: Shares of locally generated renewable energy in Beijing’s electricity and heating systems

Source: Authors’ projections.

Table 4.9: Proportion of imported electricity under the business-as-usual, electricity substitution and synergistic development scenarios (%)

	2025	2030	2035	2040	2045	2050
BAU scenario	57.87	56.12	56.74	56.98	52.99	47.09
ELS scenario	57.87	57.65	62.76	65.04	62.10	55.41
SD scenario	50.58	45.21	43.71	43.77	41.89	38.68

Source: Authors’ projections.

Environmental sustainability of different heating decarbonisation paths

The impacts of different heating decarbonisation paths on carbon dioxide emissions are shown in Figure 4.15. The carbon dioxide emissions calculated in the EnergyPLAN model are the sum of carbon dioxide emissions from the use of coal, oil and natural gas in Beijing’s energy system. The carbon dioxide emissions from using these fossil fuels are obtained by multiplying the consumption of each fuel by its emission coefficient. The amount of fossil fuel consumption is the direct output of the EnergyPLAN model and the carbon dioxide emission coefficient of each kind of fossil fuel comes from the database of the EnergyPLAN model, which has been widely used in the emissions accounting for China’s regional energy system (Yuan et al. 2020; Zhang et al. 2019; Luo et al. 2021a; Xiong et al. 2015). The results show that the carbon dioxide emissions arising from fossil fuels are lowest under the SD scenario, as its goal is to eliminate carbon dioxide emissions from natural gas combustion in power and heating systems. Under the ELS scenario, in which natural gas is gradually replaced with electricity in the heating system, the carbon dioxide emissions in the energy system cannot be fully eliminated since a small number of gas-fired units are required to maintain the system’s stable operation. Thus, the SD scenario is the most environmentally friendly scenario.

Figure 4.15: Carbon dioxide emissions from fossil fuel combustion in electricity and heating systems

Source: Authors’ projections.

The above analysis only considers carbon dioxide emissions resulting from the combustion of fossil fuels. Given that the electricity supply structure in Beijing is dominated by imported electricity, as shown in Table 4.9, it is necessary to further analyse whether imported and exported electricity affect carbon dioxide emissions in the electricity and heating systems. Fortunately, in addition to carbon emissions from fossil fuel combustion, the EnergyPLAN model has an output indicator for carbon dioxide emissions, which represents the system’s emissions from imported electricity and exportable excess electricity production (EEEP). This indicator can be used to analyse how imported and exported electricity affect the environmental benefits of the electricity and heating systems. Taking the BAU scenario as an example, when the system requires imported electricity, carbon dioxide emissions in the system would exceed the emissions resulting from fossil fuel combustion in Beijing, as shown in Figure 4.16. This indicates that some of the imported electricity comes from fossil fuel power generation, which will have negative environmental impacts on the electricity and heating systems. However, the growth rate of carbon dioxide emissions (pentagram) gradually slows when the EEEP emerges, and carbon dioxide emissions from imported and exported electricity will peak in 2037 and then decline rapidly. These results indicate that although the export of renewable energy reduces the carbon dioxide emissions of the external electricity system, considering that this electricity is produced locally from renewable energy, it is still viewed as an environmental benefit from the local electricity system. The mechanisms of the impacts of imported and exported electricity on carbon dioxide emissions under the ELS and SD scenarios are the same as under the BAU scenario.

Figure 4.16: Carbon dioxide emissions from electricity and heating systems with and without consideration of imported electricity and exportable excess electricity production under the business-as-usual scenario

Note: The left axis represents carbon dioxide emissions and the right axis represents imported electricity and exportable excess electricity production.

Source: Authors’ projections.

Table 4.10 presents the values of carbon dioxide emissions under the three scenarios. Emissions under the SD scenario change from positive to negative in 2047 when imported and exported electricity are considered. This suggests that, when local fossil fuel consumption is reduced, renewable energy can not only eliminate local carbon dioxide emissions, but also reduce emissions from external electricity systems. By 2050, the imported and exported electricity will generate 10.55 megatonnes of carbon dioxide emission reductions in Beijing under the SD scenario, which is equivalent to 30.85 terawatt hours of coal consumption reductions. In contrast, the carbon dioxide emission reductions and coal consumption reductions are only 1.59 megatonnes and 4.67 terawatt hours, respectively, under the ELS scenario. The main reason for this difference is that the ELS scenario requires more imported electricity and produces more negative environmental effects than the SD scenario. A similar conclusion is drawn when comparing the relevant results under the BAU scenario and the SD scenario. In sum, the SD scenario is superior to both the ELS and the BAU in terms of carbon dioxide emissions reduction resulting from imported and exported electricity.

Table 4.10: Carbon dioxide emissions from imported and exported electricity under the BAU, ELS and SD scenarios (megatonnes)

	Year
Scenario	2020	2021	2022	2023	2024	2025	2026	2027	2028	2029	2030
BAU	26.153	26.202	26.307	26.42	26.555	26.721	26.985	27.247	27.526	27.806	28.082
ELS	26.153	26.202	26.307	26.42	26.555	26.721	26.716	26.726	26.763	26.809	26.861
SD	26.153	25.451	24.756	24.045	23.32	22.601	21.857	21.104	20.359	19.611	18.879
	Year
	2031	2032	2033	2034	2035	2036	2037	2038	2039	2040	2041
BAU	28.349	28.602	28.837	29.037	29.187	29.291	29.307	29.232	29.025	28.669	28.122
ELS	26.912	26.964	27.004	27.014	26.991	26.922	26.871	26.547	26.197	25.7	25.02
SD	18.17	17.496	16.86	16.249	15.659	15.07	14.439	14.104	12.961	12.046	10.974
	Year
	2042	2043	2044	2045	2046	2047	2048	2049	2050
BAU	27.333	26.248	24.813	23	20.894	18.655	16.411	14.196	12.039
ELS	24.105	22.9	21.351	19.431	17.219	14.752	12.003	9.021	5.832
SD	9.691	8.138	6.264	4.027	1.503	–1.155	–3.832	–6.488	–9.09

Source: Authors’ projections.

Economic affordability of different heating decarbonisation paths

The cost-effectiveness of various heating decarbonisation paths for electricity and heating systems is critical to determine whether power and heating systems using high proportions of renewable energy are feasible in the future. Table 4.11 demonstrates that the total system cost under the ELS and SD scenarios is 8.88 per cent and 18.08 per cent, respectively, higher than under the BAU scenario. Furthermore, the cost in the SD scenario is 8.44 per cent higher than under the ELS scenario. Although the cost of fossil energy consumption is zero under the SD scenario, the cost of retrofitting thermal power plants is relatively high, as is the power generation cost using renewables. The higher cost under the ELS scenario is the result of the installation of air-source heat pumps. However, the cost increase is much lower than that of the thermal power plant retrofit under the SD scenario. The cost-effectiveness of the SD scenario can be improved in the future by reducing renewable energy generation costs, on the one hand, and improving the efficiency of retrofitted CHP units on the other. Table 4.11 lists the cost structure of future heating systems with high proportions of renewable energy integration. The results show that most of the cost of the heating system under the SD scenario is attributed to the thermal power plant retrofit rather than the fuel cost.

Table 4.11: The total cost and cost components of the energy system under the three scenarios (RMB billion)

		2025	2030	2035	2040	2045	2050
BAU scenario	Total energy system cost	117.88	160.13	204.73	257.87	325.80	402.93
	Investment cost	43.51	70.94	105.14	148.39	207.89	278.72
	Fixed O&M cost	18.81	24.04	30.78	39.49	51.74	66.41
	Variable O&M cost	55.56	65.16	68.81	70.00	66.18	57.80
ELS scenario	Total energy system cost	117.88	172.64	225.32	284.34	355.72	438.72
	Investment cost	43.51	84.35	129.17	180.66	246.41	327.22
	Fixed O&M cost	18.81	26.38	34.96	45.04	58.32	75.23
	Variable O&M cost	55.56	61.91	61.19	58.63	51.00	36.28
SD scenario	Total energy system cost	188.31	228.57	286.92	341.36	404.42	475.76
	Investment cost	106.59	139.94	191.52	240.63	299.85	369.11
	Fixed O&M cost	33.06	39.38	49.85	59.73	71.80	86.03
	Variable O&M cost	48.52	49.01	45.23	40.63	32.37	20.17

Source: Authors’ projections.

Conclusions and policy implications

This chapter examined two representative cases of inter-industrial coordination in China’s energy transition: photovoltaic–energy storage–charging station (PV–ES–CS) and the synergistic development of renewable energy and heating systems in Beijing. Despite differences in technical architecture and application scenarios, both cases illustrate the crucial role of coordinated planning between renewable energy supply and power demand in improving the economic viability of decarbonisation and reducing carbon emissions.

The study demonstrates that, first, the PV–ES–CS development model is superior to the independent development of distributed PV, ES power stations, EV charging stations, combined distributed PV and ES power stations and combined ES power stations and EV charging stations. The PV–ES–CS also has stronger environmental benefits because distributed PV can improve environmental value while ES is favourable for minimising power fluctuations by cutting peaks and filling valleys, which can improve energy efficiency. Second, PV–ES–CS systems show significant differences in economic and environmental benefits depending on the location in which they are constructed. Considering only the return on investment, the preferred places for PV–ES–CS construction should be hospitals, shopping malls, teaching buildings, hotels, office buildings, factories and residences, in this order; this is because the closer the load curve of the building is to the generation curve of solar PV in the daytime, and the lower is the power consumption at night, the higher will be the return on investment in a PV–ES–CS system. Considering only the carbon dioxide emission reductions per unit of investment, the preferred places for PV–ES–CS construction are hospitals, shopping malls, factories, hotels, teaching buildings, office buildings and residences, in this order; this is because the closer the load curve of the building is to the generation curve of solar PV in the daytime, the higher are the carbon dioxide emission reductions of a PV–ES–CS system. Hence, considering both the return on investment and carbon dioxide emission reductions per unit of investment, the most preferred places for PV–ES–CS construction are hospitals, and the last, residences. Third, the economic and environmental effects of PV–ES–CS systems are basically the same—that is, high economic benefits often accompany high environmental benefits. The economic benefits of PV–ES–CS are closely related to the utilisation rate of distributed PV. A low PV curtailment rate has high economic benefits and carbon dioxide emissions reduction equal to the distributed PV amount multiplied by the grid’s carbon dioxide emissions factor.

Bejing’s power–heat coupling model reveals that, first, the SD scenario, under which the CHP incorporating renewable energy generation occurs synergistically with heating system decarbonisation, can provide the efficient utilisation of renewable energy and the elimination of carbon dioxide emissions from fossil fuel combustion. Specifically, this scenario achieves 100 per cent renewable energy use to meet electricity and heating demands while reducing emissions from fossil fuel combustion to zero by 2050. Carbon dioxide emissions still occur under the ELS scenario because gas-fired units cannot be completely replaced. In addition, imported and exported electricity will generate 10.55 megatonnes of carbon dioxide emission reductions under the SD scenario, while under the ELS scenario the emission reductions are only 1.59 megatonnes. Furthermore, the overall economic costs under the ELS and SD scenarios are 8.16 per cent and 15.31 per cent higher, respectively, than those under the BAU scenario. Second, as the proportion of renewable energy in the power and heating systems increases, there will be a surplus of renewable electricity produced that exceeds demand. Surplus renewable energy can be used locally or exported to obtain revenue. It is found that the local use of surplus renewable energy can promote decarbonisation of the heating system and the synergistic development of renewable energy. Third, the cost of fossil fuel consumption accounts for less than 14.34 per cent of the total cost in the electricity and heating systems in 2050, suggesting that the cost of fossil fuel consumption is no longer the most important factor impacting the cost–benefit of future electricity and heating systems. Instead, the investment cost of the sophisticated renewable energy infrastructure required to facilitate heating decarbonisation, which accounts for more than 69.17 per cent of the total cost of the electricity and heating systems in 2050, will be the most important factor affecting the cost–benefit.

Based on the two case studies, several policy recommendations are put forward.

1. Promote integrated infrastructure planning and PV–ES–CS deployment: Policymakers should strengthen spatial and investment coordination across power generation, transport and heating sectors to ensure that infrastructure is aligned with renewable generation profiles and demand characteristics. In particular, the construction and development of PV–ES–CS stations should be increasingly promoted.

2. Prioritise flexible demand-side deployment: Incentives should be provided for installing PV–ES–CS systems in locations with favourable load profiles, such as hospitals and commercial buildings, and for electrification of heat demand using smart control technologies.

3. Implement differential incentive policies to unlock system value. For large-scale (investment above RMB41.5 million in this study) PV–ES–CS stations, incentive policies should be implemented to increase the peak and valley electricity price difference; for small-scale systems (investment below RMB13 million in this study), incentives will improve the number of EV charging stations; for all scales of PV–ES–CS systems, the cost reduction of ES plays an important role in promoting the development of PV–ES–CS stations.

4. Facilitate the progressive retrofitting of fossil fuel–based CHP units: To accelerate the decarbonisation of heating systems, it is essential to promote the gradual conversion of coal-fired and gas-fired CHP plants into renewable-powered thermal units. By changing the primary energy source rather than the entire technological system, this approach enables the continued use of existing infrastructure while aligning power and heat production with carbon neutrality goals. Such retrofitting also supports the integrated planning and operation of power and heat systems, improving overall system flexibility and decarbonisation efficiency.

5. The subsidy policies or investment strategies for energy corporations should focus on decreasing capital investment in infrastructure construction to maximise the long-term cost–benefit of high-proportion renewable energy integration and heating system decarbonisation.

Future research

As China’s energy transition enters a new phase, marked by high shares of renewables and complex cross-sector interactions, the need for robust mechanisms of inter-industrial coordination becomes increasingly urgent. However, existing academic and policy research falls short in several key areas that are critical for scaling up and institutionalising such coordination. First, existing studies tend to concentrate on bilateral interactions between sectors, lacking a systemic, multisectoral and multilevel coupling perspective. Second, most modelling approaches are based on static optimisation, which fails to capture dynamic features such as load flexibility, market price interactions and dispatch uncertainty. Third, institutional and policy mechanisms supporting coordination are often fragmented—particularly in cross-sectoral investment alignment, carbon price transmission and regionally integrated governance. These gaps point to the urgent need for a more holistic and dynamic framework to understand and guide inter-industrial coordination.

To address these limitations, it is necessary to develop a multidimensional research framework that systematically characterises the mechanisms and features of different types of industrial synergy. This framework is organised around six key dimensions, each targeting a distinct aspect of coordination:

1. Horizontal synergistic development, which focuses on the systemic coupling of renewable energy with sectors such as transportation, heating and energy storage.

2. Vertical synergistic development, which examines the coordination of renewable energy deployment and the electrification of energy-intensive industries.

3. Spatial synergistic development, which addresses the regional mismatch between renewable energy production and consumption, along with mechanisms for cross-regional dispatch and market integration.

4. Temporal synergistic development, which explores the synchronisation of renewable energy expansion and the phase-out of traditional fossil fuels to ensure system stability.

5. Technological synergistic development, which investigates how advances in key low-carbon technologies, pathway selections and cost evolution affect the potential for coordination.

6. Institutional synergistic development, which centres on institutional safeguards, including the design of flexible power systems, market mechanisms and cross-sectoral policy integration.

Together, these six dimensions offer a comprehensive structure for future research and modelling efforts, enabling more robust evaluation of coordination strategies under various energy transition scenarios.

Building on this framework, future research must also overcome critical technical and methodological challenges. Limitations in data granularity, algorithmic integration and spatial resolution constrain the accuracy of coordination modelling. Leveraging advanced tools such as geographic information systems, big-data analytics and machine learning can enhance the identification, simulation and validation of synergistic mechanisms. Furthermore, energy system transformation is not only a technical or institutional endeavour; it also has deep social implications. Issues such as changes in employment structure, local governance capacities and public behavioural responses are vital to consider. Hence, a comprehensive analytical framework should incorporate social coordination, facilitating an integrated approach that links technology, institutions and society to support decarbonisation goals.

Beyond methodological innovations, future research should also aim to elevate the theoretical foundations and practical relevance of coordination mechanisms. At the theoretical level, multisector coordination can be conceptualised as a complex systems coupling process. Relevant theories such as complexity science, institutional embeddedness theory and network evolution theory can help unpack the drivers, interaction pathways and boundary conditions of such coordination. This would enable a shift from empirical identification towards theoretical abstraction and mechanistic interpretation—contributing to a more generalisable theory of synergistic energy transition. At the practical level, research must focus on enhancing the policy relevance of modelling outcomes. This includes linking model outputs with specific policy objectives, improving the evaluability of policies, optimising implementation pathways and establishing a scenario-based comparative framework for coordinated strategy design.

References

Alves, M., R. Segurado, and M. Costa. 2020. ‘On the Road to 100% Renewable Energy Systems in Isolated Islands.’ Energy 198: 117321. doi.org/10.1016/j.energy.2020.117321.

Askeland, Kristine, Kristina N. Bozhkova, and Peter Sorknæs. 2019. ‘Balancing Europe: Can District Heating Affect the Flexibility Potential of Norwegian Hydropower Resources?’ Renewable Energy 141: 646–56. doi.org/10.1016/j.renene.2019.03.137.

Aunedi, Marko, Antonio Marco Pantaleo, Kamal Kuriyan, Goran Strbac, and Nilay Shah. 2020. ‘Modelling of National and Local Interactions Between Heat and Electricity Networks in Low-Carbon Energy Systems.’ Applied Energy 276: 115522. doi.org/10.1016/j.apenergy.2020.115522.

Beijing Municipal Bureau of Statistics. 2021a. 2020 Beijing Energy Balance Sheet. Beijing: China Statistics Press.

Beijing Municipal Bureau of Statistics. 2021b. ‘Gross Regional Product.’ Beijing Statistical Yearbook 2020. Beijing: China Statistics Press.

Beijing Municipal Commission of Development and Reform. 2019. Implementation Opinions on Further Accelerating the Application of Heat Pump Systems and Promoting Clean Heating. Beijing: Beijing Municipal Commission of Development and Reform.

Beijing Municipal Commission of Development and Reform. 2020a. ‘In the Winter of 2020, Beijing’s Clean Heating Area Exceeded 98%, and Many Projects Were Approved to Weave a Dense Heating Network.’ Press release. Beijing: Beijing Commission of Municipal Development and Reform.

Beijing Municipal Commission of Development and Reform. 2020b. Notice of Beijing Municipal Development and Reform Commission on Adjusting the Sales Electricity Price in Beijing. [2020], no. 1708 (30 November). Beijing: Beijing Municipal Commission of Development and Reform. www.beijing.gov.cn/zhengce/zhengcefagui/202012/t20201201_2153982.html.

Beijing Municipal People’s Government. 2017a. Beijing Energy Development Plan during the Thirteenth Five-Year Plan Period. Beijing: Beijing Municipal People’s Government.

Beijing Municipal People’s Government. 2017b. Thirteenth Five-Year Plan for New and Renewable Energy Development. Beijing: Beijing Municipal People’s Government.

Beijing Municipal People’s Government. 2021. Fourteenth Five-Year Plan for Ecological and Environmental Protection: Promoting the Large-Scale Application of Local Photovoltaics. Beijing: Beijing Municipal People’s Government.

Beijing Municipal People’s Government. 2022. Beijing Energy Development Plan during the Fourteenth Five-Year Plan Period. Beijing: Beijing Municipal People’s Government.

Beuse, Martin, Bjarne Steffen, Mathias Dirksmeier, and Tobias S. Schmidt. 2021. ‘Comparing CO₂ Emissions Impacts of Electricity Storage Across Applications and Energy Systems.’ Joule 5, no. 6: 1501–20. doi.org/10.1016/j.joule.2021.04.010.

China Electricity Council (CEC). 2021. 2020 Power Industry Statistics Compilation. Beijing: Statistics and Data Centre of the China Electricity Council.

Connolly, D., H. Lund, and B.V. Mathiesen. 2016. ‘Smart Energy Europe: The Technical and Economic Impact of One Potential 100% Renewable Energy Scenario for the European Union.’ Renewable and Sustainable Energy Reviews 60: 1634–53. doi.org/10.1016/j.rser.2016.02.025.

Connolly, D., H. Lund, B.V. Mathiesen, and M. Leahy. 2011. ‘The First Step Towards a 100% Renewable Energy System for Ireland.’ Applied Energy 88, no. 2: 502–7. doi.org/10.1016/j.apenergy.2010.03.006.

Denholm, Paul, Michael Kuss, and Robert M. Margolis. 2013. ‘Co-Benefits of Large Scale Plug-In Hybrid Electric Vehicle and Solar PV Deployment.’ Journal of Power Sources 236: 350–6. doi.org/10.1016/j.jpowsour.2012.10.007.

Ding, Yiting. 2025. ‘Installed Capacity of Wind and Solar Power in China Exceeds 1.4 Billion Kilowatts.’ People’s Daily, 26 January. www.gov.cn/lianbo/bumen/202501/content_7001210.htm.

Drysdale, David, Brian Vad Mathiesen, and Susana Paardekooper. 2019. ‘Transitioning to a 100% Renewable Energy System in Denmark by 2050: Assessing the Impact from Expanding the Building Stock at the Same Time.’ Energy Efficiency 12: 37–55. doi.org/10.1007/s12053-018-9649-1.

Eldeeb, Hassan H., Samy Faddel, and Osama A. Mohammed. 2018. ‘Multi-Objective Optimization Technique for the Operation of Grid Tied PV Powered EV Charging Station.’ Electric Power Systems Research 164: 201–11. doi.org/10.1016/j.epsr.2018.08.004.

Englberger, Stefan, Archie C. Chapman, Wayes Tushar, Tariq Almomani, Stephen Snow, Rolf Witzmann, Andreas Jossen, and Holger Hesse. 2021. ‘Evaluating the Interdependency Between Peer-to-Peer Networks and Energy Storages: A Techno-Economic Proof for Prosumers.’ Advances in Applied Energy 3: 100059. doi.org/10.1016/j.adapen.2021.100059.

Groppi, D., D. Astiaso Garcia, G. Lo Basso, and L. De Santoli. 2019. ‘Synergy Between Smart Energy Systems Simulation Tools for Greening Small Mediterranean Islands.’ Renewable Energy 135: 515–24. doi.org/10.1016/j.renene.2018.12.043.

Gu, Chaolin, Tan Zongbo, and Liu Wan. 2009. ‘Progress in Research on Climate Change, Carbon Emissions and Low-Carbon Urban Planning.’ Urban Planning Forum, no. 3: 38–45.

Hansen, Kenneth, Brian Vad Mathiesen, and Iva Ridjan Skov. 2019. ‘Full Energy System Transition Towards 100% Renewable Energy in Germany in 2050.’ Renewable and Sustainable Energy Reviews 102: 1–13. doi.org/10.1016/j.rser.2018.11.038.

He, Jijiang. 2019. ‘Beijing Dahongmen Jimei DC Photovoltaic Storage and Charging Integrated Project Creates Multiple Firsts in Various Fields.’ Polaris Power News Network, 28 April. chuneng.bjx.com.cn/news/20190428/977571.shtml.

Hedegaard, Karsten, Brian Vad Mathiesen, Henrik Lund, and Per Heiselberg. 2012. ‘Wind Power Integration Using Individual Heat Pumps—Analysis of Different Heat Storage Options.’ Energy 47, no. 1: 284–93. doi.org/10.1016/j.energy.2012.09.030.

Hochreiter, Sepp, and Jürgen Schmidhuber. 1997. ‘Long Short-Term Memory.’ Neural Computation 9, no. 8: 1735–80. doi.org/10.1162/neco.1997.9.8.1735.

Huang, Pei, Marco Lovati, Xingxing Zhang, Chris Bales, Sven Hallbeck, Anders Becker, Henrik Bergqvist, Jan Hedberg, and Laura Maturi. 2019. ‘Transforming a Residential Building Cluster into Electricity Prosumers in Sweden: Optimal Design of A Coupled PV–Heat Pump–Thermal Storage–Electric Vehicle System.’ Applied Energy 255: 113864. doi.org/10.1016/j.apenergy.2019.113864.

Huang, Xiaodan, Hongyu Zhang, and Xiliang. Zhang. 2020. ‘Decarbonising Electricity Systems in Major Cities Through Renewable Cooperation—A Case Study of Beijing and Zhangjiakou.’ Energy 190: 116444. doi.org/10.1016/j.energy.2019.116444.

Ibrahim, Ibrahim Anwar, Slaiman Sabah, Robert Abbas, M.J. Hossain, and Hani Fahed. 2021. ‘A Novel Sizing Method of a Standalone Photovoltaic System for Powering a Mobile Network Base Station Using a Multi-Objective Wind Driven Optimization Algorithm.’ Energy Conversion and Management 238: 114179. doi.org/10.1016/j.enconman.2021.114179.

Intergovernmental Panel on Climate Change (IPCC). 2014. Climate Change 2014: Mitigation of Climate Change. Working Group III Contribution to the IPCC Fifth Assessment Report. Cambridge: Cambridge University Press. www.ipcc.ch/site/assets/uploads/2018/02/ipcc_wg3_ar5_full.pdf. doi.org/10.1017/CBO9781107415416.

Jimenez-Navarro, Juan-Pablo, Konstantinos Kavvadias, Faidra Filippidou, Matija Pavičević, and Sylvain Quoilin. 2020. ‘Coupling the Heating and Power Sectors: The Role of Centralised Combined Heat and Power Plants and District Heat in A European Decarbonised Power System.’ Applied Energy 270: 115134. doi.org/10.1016/j.apenergy.2020.115134.

Kaufmann, Robert K., Derek Newberry, Chen Xin, and Sucharita Gopal. 2021. ‘Feedbacks Among Electric Vehicle Adoption, Charging, and the Cost and Installation of Rooftop Solar Photovoltaics.’ Nature Energy 6, no. 2: 143–9. doi.org/10.1038/s41560-020-00746-w.

Kwon, Pil Seok, and Poul Alberg Østergaard. 2013. ‘Priority Order in Using Biomass Resources—Energy Systems Analyses of Future Scenarios for Denmark.’ Energy 63: 86–94. doi.org/10.1016/j.energy.2013.10.005.

Lashof, Daniel A., and Dilip R. Ahuja. 1990. ‘Relative Contributions of Greenhouse Gas Emissions to Global Warming.’ Nature 344, no. 6266: 529–31. doi.org/10.1038/344529a0.

Liu, Xin, Xinran Li, Zhuangxi Tan, Huang Jiyuan, Cui Yiwei, and Zhang Bingyu. 2021. ‘Economic Comparison of Optimal Strategies Based on Different Types of Energy Storage Batteries Participating in Primary Frequency Regulation.’ High Voltage Engineering: 1–9. chn.oversea.cnki.net/kcms/detail/detail.aspx?filename=GDYJ202204018&dbcode=CJFQ&dbname=CJFDLAST2022&uniplatform=NZKPT.

Lund, Henrik, and Georges Salgi. 2009. ‘The Role of Compressed Air Energy Storage (CAES) in Future Sustainable Energy Systems.’ Energy Conversion and Management 50, no. 5: 1172–9. doi.org/10.1016/j.enconman.2009.01.032.

Luo, Shihua, Weihao Hu, Wen Liu, Xiao Xu, Qi Huang, Zhe Chen, and Henrik Lund. 2021a. ‘Transition Pathways Towards a Deep Decarbonization Energy System—A Case Study in Sichuan, China.’ Applied Energy 302, no. 117507: 1–14. doi.org/10.1016/j.apenergy.2021.117507.

Luo, Xing, Dongxiao Zhang, and Xu Zhu. 2021b. ‘Deep Learning Based Forecasting of Photovoltaic Power Generation by Incorporating Domain Knowledge.’ Energy 225: 120240. doi.org/10.1016/j.energy.2021.120240.

Luo, Z., Zechun Hu, Yonghua Song, Xia Yang, Kaiqiao Zhan, and Junyang Wu. 2011. ‘Study on Plug-In EVs Charging Load Calculating.’ Automation of Electric Power Systems 35, no. 14: 36–42.

Ma, Tao, Poul Alberg Østergaard, Henrik Lund, Hongxing Yang, and Lin Lu. 2014. ‘An Energy System Model for Hong Kong in 2020.’ Energy 68: 301–10. doi.org/10.1016/j.energy.2014.02.096.

Ministry of Ecology and Environment (MEE). 2022. China’s Policies and Actions for Addressing Climate Change (2022). Beijing: Ministry of Ecology and Environment of the People’s Republic of China. english.mee.gov.cn/Resources/Reports/reports/202211/P020221110605466439270.pdf.

Ministry of Housing and Urban–Rural Development (MoHURD). 2021. Urban Construction Statistical Yearbook 2020. Beijing: Ministry of Housing and Urban–Rural Development of the People’s Republic of China.

Mohajan, Haradhan Kumar. 2017. ‘Greenhouse Gas Emissions, Global Warming and Climate Change.’ In Proceedings of the 15th Chittagong Conference on Mathematical Physics, Jamal Nazrul Islam Research Centre for Mathematical and Physical Sciences (JNIRCMPS), Chittagong, Bangladesh. Volume 16.

National Bureau of Statistics of China (NBS). 2005. China Statistics Yearbook 2004. Beijing: China Statistics Press.

National Development and Reform Commission (NDRC). 2022. Fourteenth Five-Year Renewable Energy Development Plan. Beijing: National Development and Reform Commission.

NDRC, National Energy Administration, Ministry of Finance, Ministry of Natural Resources, Ministry of Ecology and Environment, Ministry of Housing and Urban–Rural Development, Ministry of Agriculture and Rural Affairs, China Meteorological Administration, and National Forestry and Grassland Administration. 2022. Fourteenth Five-Year Plan for Renewable Energy Development. Beijing: National Development and Reform Commission. www.ndrc.gov.cn/xxgk/zcfb/ghwb/202206/t20220601_1326719.html?code=&state=123.

National Energy Board (NEB). 2020. NB/T 10394-2020: Specification for Photovoltaic Power Generation System Performance. National Energy Administration Announcement No. 5 of 2020. Beijing: National Energy Board. www.chinesestandard.net/PDF/English.aspx/NBT10394-2020.

National Renewable Energy Consumption Monitoring and Early Warning Centre. 2025. National New Energy Grid Connection and Consumption in 2024. Beijing: National Renewable Energy Consumption Monitoring and Early Warning Centre. mp.weixin.qq.com/s/2SJA4s8afIDoe0g3AWSuNw.

Navarra, Antonio. 2023. ‘Syukuro Manabe: Recipient of Nobel Prize in Physics 2021.’ In Oxford Research Encyclopedia of Climate Science. Oxford: Oxford University Press. doi.org/10.1093/acrefore/9780190228620.013.930.

Nielsen, Steffen, Peter Sorknæs, and Poul Alberg Østergaard. 2011. ‘Electricity Market Auction Settings in a Future Danish Electricity System with a High Penetration of Renewable Energy Sources—A Comparison of Marginal Pricing and Pay-as-Bid.’ Energy 36, no. 7: 4434–44. doi.org/10.1016/j.energy.2011.03.079.

Olympios, Andreas V., Antonio M. Pantaleo, Paul Sapin, and Christos N. Markides. 2020. ‘On the Value of Combined Heat and Power (CHP) Systems and Heat Pumps in Centralised and Distributed Heating Systems: Lessons from Multi-Fidelity Modelling Approaches.’ Applied Energy 274: 115261. doi.org/10.1016/j.apenergy.2020.115261.

Omran, Walid A., M. Kazerani, and M.M.A. Salama. 2011. ‘Investigation of Methods for Reduction of Power Fluctuations Generated from Large Grid-Connected Photovoltaic Systems.’ IEEE Transactions on Energy Conversion 26, no. 1: 318–27. doi.org/10.1109/TEC.2010.2062515.

Peng, Danni, and Xu Dawei. 2021. ‘China’s Power Shortage Pain.’ China Newsweek, 14 October. mp.weixin.qq.com/s/zb6aiSPNG723pFd-HSEQLA.

Pu, Yanru, Peng Wang, Yuyi Wang, Wenhui Qiao, Lu Wang, and Yuhu Zhang. 2021. ‘Environmental Effects Evaluation of Photovoltaic Power Industry in China on Life Cycle Assessment.’ Journal of Cleaner Production 278: 123993. doi.org/10.1016/j.jclepro.2020.123993.

Sabadini, Felipe, and Reinhard Madlener. 2021. ‘The Economic Potential of Grid Defection of Energy Prosumer Households in Germany.’ Advances in Applied Energy 4: 100075. doi.org/10.1016/j.adapen.2021.100075.

Sáfián, Fanni. 2014. ‘Modelling the Hungarian Energy System—The First Step Towards Sustainable Energy Planning.’ Energy 69: 58–66. doi.org/10.1016/j.energy.2014.02.067.

Sainato, Michael. 2020. ‘The Collapse of Coal: Pandemic Accelerates Appalachia Job Losses.’ The Guardian, 29 May. www.theguardian.com/us-news/2020/may/29/coal-miners-coronavirus-job-losses.

Shang, Wen-Long, Jinyu Chen, Huibo Bi, Yi Sui, Yanyan Chen, and Haitao Yu. 2021. ‘Impacts of COVID-19 Pandemic on User Behaviors and Environmental Benefits of Bike Sharing: A Big-Data Analysis.’ Applied Energy 285: 116429. doi.org/10.1016/j.apenergy.2020.116429.

Thellufsen, J.Z., H. Lund, P. Sorknæs, P.A. Østergaard, M. Chang, D. Drysdale, S. Nielsen, S.R. Djørup, and K. Sperling. 2020. ‘Smart Energy Cities in a 100% Renewable Energy Context.’ Renewable and Sustainable Energy Reviews 129: 109922. doi.org/10.1016/j.rser.2020.109922.

Thellufsen, Jakob Zinck, Steffen Nielsen, and Henrik Lund. 2019. ‘Implementing Cleaner Heating Solutions Towards a Future Low-Carbon Scenario in Ireland.’ Journal of Cleaner Production 214: 377–88. doi.org/10.1016/j.jclepro.2018.12.303.

von Storch, Hans, and Patrick Heimbach. 2022. ‘Klaus Hasselmann: Recipient of the Nobel Prize in Physics 2021.’ In Oxford Research Encyclopedia of Climate Science. Oxford: Oxford University Press. doi.org/10.1093/acrefore/9780190228620.013.931.

Wang, Mudan, Xianqiang Mao, Yubing Gao, and Feng He. 2018b. ‘Potential of Carbon Emission Reduction and Financial Feasibility of Urban Rooftop Photovoltaic Power Generation in Beijing.’ Journal of Cleaner Production 203: 1119–31. doi.org/10.1016/j.jclepro.2018.08.350.

Wu, Y., X. Zhang, H. Wang, and J. Sun. 2017. ‘Molten Salt Heat Storage and Supply Technology Based on Heating Using Abandoned Wind Power PV Power or Off-Peak Power.’ Energy 22: 93–99.

Xiong, Weiming, Yu Wang, Brian Vad Mathiesen, Henrik Lund, and Xiliang Zhang. 2015. ‘Heat Roadmap China: New Heat Strategy to Reduce Energy Consumption Towards 2030.’ Energy 81: 274–85. doi.org/10.1016/j.energy.2014.12.039.

Yang, Meng, Lihui Zhang, and Wenjia Dong. 2020. ‘Economic Benefit Analysis of Charging Models Based on Differential EV Charging Infrastructure Subsidy Policy in China.’ Sustainable Cities and Society 59: 102206. doi.org/10.1016/j.scs.2020.102206.

Ye, X. 2018. ‘Economic and Environmental Value Analysis of Replacing Coal by Gas in Buildings.’ Thesis, China University of Petroleum, Beijing.

Yuan, Meng, Jakob Zinck Thellufsen, Henrik Lund, and Yongtu Liang. 2020. ‘The First Feasible Step Towards Clean Heating Transition in Urban Agglomeration: A Case Study of Beijing–Tianjin–Hebei Region.’ Energy Conversion and Management 223: 113282. doi.org/10.1016/j.enconman.2020.113282.

Zhang, Hongyu, Li Zhou, Xiaodan Huang, and Xiliang Zhang. 2019. ‘Decarbonizing a Large City’s Heating System Using Heat Pumps: A Case Study of Beijing.’ Energy 186: 115820. doi.org/10.1016/j.energy.2019.07.150.

Zhu, Shunmin, Guoyao Yu, Kun Liang, Wei Dai, and Ercang Luo. 2021. ‘A Review of Stirling-Engine-Based Combined Heat and Power Technology.’ Applied Energy 294: 116965. doi.org/10.1016/j.apenergy.2021.116965.

Appendix 4.1 Cost datasets of the future energy system

The cost datasets used for the economic analysis of the future energy system using the EnergyPLAN model were primarily derived from Luo et al. (2021a), although this provided data only for 2016, 2030 and 2050. Therefore, the economic analysis was based on the following assumptions. First, it was assumed that the costs remained constant from 2016 to 2019, changed every five years from 2020 to 2030 and the rate of change remained constant. Second, this study assumed a constant rate of change in fuel prices between 2016 and 2030 and 2030 and 2050. Third, the interest rate remained unchanged at 8 per cent (Yuan et al. 2020). The cost and fuel price data for different technologies in 2020 are listed in Appendix Tables A4.1 and A4.2.

Table A4.1: Costs including investments, fixed operation and maintenance costs and operation lifetime for different technologies in 2020

Technology	Investment cost (RMB million/MW)	Fixed operation and maintenance costs (% of investment cost)	Lifetime (years)
Coal power plant	4.3	4.0	25
Natural gas power plant	3.7	4.0	25
Biomass CHP plant	9.0	4.0	40
Natural gas CHP plant	4.9	4.0	25
Biomass CHP plant	9.9	4.0	40
Pump turbine	8.0	1.5	50
Wind power	9.6	2.0	25
PV power	22.0	2.0	30
Natural gas boiler	0.78	3.0	20
Ground-source heat pump	4.0	1.0	25

Table A4.2: Assumed fuel prices in 2020 (RMB/gigajoule)

Year	Coal	Fuel oil	Diesel	Petrol	Natural gas	LPG	Biomass
2020	23.19	56.33	99.85	106.3	58.34	69.91	40.33