Exploration of CO₂ emission reduction pathways: identification of influencing factors of CO₂ emission and CO₂ emission reduction potential of power industry

Abstract

Low-carbon development of China's power sector is the key to achieving carbon peaking and carbon neutrality goals. Based on the logarithmic mean divisor index (LMDI) model, considering the carbon transfer caused by inter-provincial electricity trading, this paper analyzes the influencing factors of CO₂ emissions in the provincial power sector and uses K-means clustering method to divide 30 provinces into four categories to analyze the differences in regional carbon emission characteristics. In addition, by establishing different development scenarios, the carbon emission trends and emission reduction potentials of each cluster under different emission reduction measures from 2020 to 2040 are studied, in order to explore the differentiated emission reduction paths of each cluster. The results show that the contribution of influencing factors shows great differences in different provinces. Trends in CO₂ emissions vary widely across scenarios. In the reference scenario, the CO₂ emissions of each cluster will continue to increase; in the existing policy scenario, the total power industry will peak at 6.1Gt in 2030; in the advance peak scenario that puts more emphasis on the development of advanced technologies and renewable energy under the clean development model, the carbon emission peak will be brought forward to 2025, and the peak will be reduced to 5.2Gt. Finally, differentiated emission reduction paths and measures are proposed for the future low-carbon development of different cluster power industries, providing theoretical reference for the deployment of provincial-level emission reduction work, which is of great significance to the global green and low-carbon transformation.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1007/s10098-022-02456-1.

Introduction

Global warming is a major problem faced by all mankind, and low-carbon emission reduction has become an important challenge for sustainable development in the future. CO₂ is the largest supplier of the greenhouse effect and has become a key target of governance. Taking 2019 as an example, the world emits 36.4 billion tons of CO₂, and China’s CO₂ emissions are 10.2 billion tons, accounting for 28% of global emissions. As the world's largest emitter of greenhouse gases, China plays a vital role in mitigating global climate change. The combustion of fossil energy, especially coal, is the main source of CO₂ emissions. In order to reduce carbon emissions, many countries restrict the use of coal and explore the supply of energy from alternatives to coal, such as wood (Jandačka et al. 2017), methane (Mardoyan and Braun 2014) and biofuels (Maroušek 2014). According to the World Energy Statistics Yearbook 2020, China's coal, natural gas and oil consumption accounted for 52, 8 and 15% of the world's total consumption, respectively, so it is clear that China's first priority to control CO₂ emissions is to control coal consumption. About half of the coal in China is used to generate electricity, and 43% of China's total CO₂ emissions in 2018 came from thermal power (about 90% of which is coal power), making it the largest source of CO₂ emissions. However, China's resource endowment of “rich in coal, poor in oil and gas” makes it difficult for the power industry to leave coal in the short term. In the 14th Five-Year Plan Outline for ecological civilization construction (2021–2025), China proposed to resolutely implement the requirements of carbon peaking and carbon neutralization. Therefore, the power industry is facing the dual challenge of increasing supply and reducing emissions. Carbon peaking in the power industry is an essential prerequisite for China to achieve the carbon emission peaking target, and decarbonization of the power industry is the key to achieving a low-carbon society (Zhang et al. 2021). It should be pointed out that carbon emission characteristics of the power sector vary significantly between provinces (Wen and Li 2020). Therefore, it is critical to explore the carbon emission trajectory and peak time of the provincial power industry in order to formulate scientific and reasonable carbon emission reduction policies.

The analysis of influencing factors has been used to study the changes of carbon emissions in various industry. The main methods to study influencing factors are SDA method (Luo et al. 2020), GDIM method (Yan et al. 2019a, b), IDA method (De Oliveira-De 2019) and Kaya method (Yang et al. 2020). In view of the significant regional and provincial differences, some literatures further discussed regional and provincial influencing factors. Specifically, Wang et al. (2018) analyzed the various influencing factors on the carbon emissions per unit of electricity from 1995 to 2014 in combination with the geographical differences of different regions. Based on the LMDI decomposition results, Liao et al. (2019) divided 30 provinces into five categories, and put forward emission reduction suggestions according to the characteristics of each category. Tian et al. (2021) used LMDI decomposition to analyze the differences in carbon intensity in different sectors and regions in China, and the results showed that energy and economic structure were the main factors for the differences in carbon intensity. Lin et al. (2019) quantified the impact of socioeconomic factors and population density on carbon emissions from the transportation industry based on the LMDI model, but ignored the transboundary issue of carbon dioxide generated during transportation. Based on the carbon emission data of the power sector from 2005 to 2016, He et al. (2020) constructed a spatial correlation network of carbon emissions in various provinces, and analyzed the spatial correlation characteristics and influencing factors. Many scholars have also studied the influencing factors in a single province or city, such as Shanghai (Wei et al. 2020), Baoding (Zhang et al. 2019), Beijing (Zhang et al. 2020), Shandong (Wang et al. 2017), Guangdong (Xu et al. 2021) and so on.

Research on CO₂ emission reduction potential and peak path in power sector has also achieved corresponding results. In the past 5 years, there have been many studies on the carbon emission reduction potential generated by the independent effects of factors such as the level of technology (He et al. 2021), carbon emission power structure (Cui et al. 2021) or carbon reduction policies (Chen and Chen 2019). Yang et al. (2021) set 3 different technology levels, and used the STRIPAT model to analyze the impact of technical factors on CO₂ reduction in 6 sectors in China. Others referred to national and industry environmental protection policies, planning reports and technological development requirements to construct different scenarios for the possible economic structure, technical level, and energy consumption level in the future, and explored the potential for CO₂ emission reduction under different scenarios (Yu et al. 2020). Yang et al. (2020) combined the LMDI method to decompose the influencing factors of CO₂ emissions from 1996 to 2016 and discussed the influence degree of each factor. Considering the carbon sink technology, Demetriou and Hadjistassou (2021) developed four different energy structure scenarios and analyzed the reduction potential with a top-down approach. Carbon dioxide storage mostly refers to the high-pressure sealing of carbon dioxide into deep waste mines. The current cost of high-pressure storage in China is estimated to be 500 yuan per ton of CO₂. However, the capture cost of biochar is relatively low, which is produced from natural biomass such as wood (Marousek and Gavurova 2022) or agricultural waste such as biogas (Marousek and Trakal 2022). It is currently an economically viable way to sequester carbon dioxide and can contribute to carbon emission reduction.

The methods for forecasting peak carbon emissions mainly include the environmental Kuznets model (Jiang et al. 2019), scenario analysis (Wang et al. 2021) and IPAT model (Wen and Li 2019; Yin et al. 2022). Scholars used the above different methods to study the peak and peak time of carbon emissions in China, and drew different conclusions. Based on carbon emissions data of 8 sectors in China from 1995 to 2017, Fang et al. (2022) investigated the environmental Kuznets curve hypothesis of eight sectors, and predicted the peak time of each sector. Meng et al. (2017) used log-linear equations in a mixed model to predict the value of variables and set up five scenarios, and their results suggest that China's power sector will not reach the peak emission by 2030. Hernández and Fajardo (2021) set up three scenarios to estimate carbon emissions and carbon intensity in 2050 using the LEAP model. Based on the regional perspective, Tang et al. (2018) divided six regions based on geographical location, assessed the impact of technical factors and energy consumption on CO₂ emissions from the regional power sector and studied the time of carbon peak. Chang et al. (2022) used scenario analysis to evaluate three carbon emission reduction scenarios from the perspectives of social equity, emission reduction efficiency and forest carbon sink. The results show that under the 2030 carbon emission target, the marginal CO₂ emission reduction cost is 2315–5387 yuan.

In summary, a lot of research has been carried out on the influencing factors, emission reduction potential and peak size of carbon emissions in China, but there are still the following shortcomings in the current studies: (1) There are many studies that analyze the influencing factors of carbon emissions in the power industry from the national level or a single region, province and city, while there are fewer studies that explore regional differences in each region. (2) The existing literature analyzes the regional CO₂ emissions mainly according to the geographical location or the surface characteristics, ignoring the potential characteristics of carbon emissions. (3) There are more analyses of the peak time and peak size of the entire power industry, while only a few scholars have studied the carbon peak paths of different clusters based on the carbon emission characteristics of the power industry. Under the background of climate change, how to put forward targeted measures to promote the peak value according to the characteristics of regional carbon emissions and avoid the "flooding" of energy-saving and emission-reduction policies is an urgent problem to be solved. In response to previous research deficiencies, this paper adopts LMDI method to quantitatively analyze the influence degree of main factors of China's power industry from 2000 to 2019. Combined with the decomposed influencing factors, it is hypothesized that there are regional differences in the influencing factors of carbon emissions, and cluster analysis is carried out on 30 provinces according to the decomposition results from 2015 to 2019. According to China's medium and long-term economic development and energy demand, the change of carbon emissions from power production in 2020–2040 is divided into 3 scenarios: reference scenario, existing policy scenario and advance peak scenario. The carbon emission trends of the whole power industry and different clusters under different scenarios are predicted, and the emission reduction potential is analyzed. Based on the scenario analysis results, the contribution of each factor to future CO₂ emission reduction is analyzed. Finally, targeted measures are made to promote peak attainment in the power sector. It can provide a reference for policy makers to formulation of provincial carbon emission reduction policies, which is of great significance to global climate change.

Methods and dataset

Estimation of CO₂ emissions from power industry

In the power sector, the consumption of fossil energy is a major contributor to carbon emissions. Accordingly, this paper selects a total of 22 fossil energy sources, including coal, petroleum and natural gas. The carbon emissions of the power industry are measured through the measurement model proposed by the United Nations Intergovernmental Panel on Climate Change in 2006 (Yan et al. 2019a, b). The fuel parameters and emission coefficients involved are shown in Table 1, and the calculation formula is as follows:

$$C=\sum _{i=1}^3{C}_i=\sum _{i=1}^3{E}_i\times {\text{EF}}_i=\sum _{i=1}^3{E}_i\times {\text{NCV}}_i\times {\text{CEF}}_i\times {\text{COF}}_i\times 44/12$$ 1

where $C$ is CO₂ emission, i is the energy type, $E_i$ is the consumption of energy i (physical quantity), ${\text{EF}}_i$ is the carbon emission factor of energy i, ${\text{NCV}}_i$ is the average low heating value of energy i, ${\text{CEF}}_i$ denotes the carbon content per unit heat generated of energy i, ${\text{COF}}_i$ denotes the carbon oxidation rate of energy i, and 44/12 is the molecular weight ratio of CO₂.

Table 1.

Fuel parameters and emission factors

Energy name	Average low calorific valuea (kJ/kg or kJ/m3)	Conversion coefficient of standard coala (kg/kg or kg/m3)	Carbon content per unit calorific valueb (tC/TJ)	Carbon oxidation rateb (%)
Raw coal	20,934	0.7143	26.37	0.94
Washed coal	26,377	0.9000	25.41	0.93
Other coal washing	8374	0.4286	25.41	0.93
Coal products	20,908	0.6000	33.60	0.90
Gangue	8374	0.2857	25.80	0.93
Coke	28,470	0.9714	29.50	0.93
Coke oven gas	18,003	0.6143	13.58	0.99
Blast furnace gas	3768	0.1286	70.80	0.99
Converter gas	7945	0.2714	49.60	0.99
Other gas	5227	0.3571	12.20	0.99
Other coking products	28,435	1.3000	29.50	0.93
Crude oil	41,868	1.4286	20.10	0.98
Gasoline	43,124	1.4714	18.90	0.98
Kerosene	43,124	1.4714	19.60	0.98
Diesel oil	42,705	1.4571	20.20	0.98
Fuel oil	41,868	1.4286	21.10	0.98
Petroleum coke	31,947	1.0918	27.50	0.98
Liquefied petroleum gas	50,242	1.7143	17.20	0.98
Refinery Gas	46,055	1.5714	18.20	0.98
Other petroleum products	41,031	1.4000	20.00	0.98
Natural gas	38,979	1.3300	15.30	0.99
Liquified natural gas	51,498	1.7572	17.20	0.98

^aData are collected from China Energy Statistical Yearbook in 2019 and General Principles for Calculation of Comprehensive Energy Consumption GB/T 2589–2020

^bData are collected from Guidelines for Compilation of Provincial Greenhouse Gas Inventory

LMDI decomposition method

LMDI decomposition method is widely used in the analysis of influencing factors, which has the characteristics of residual-free and wide applicability (Jiang et al. 2020). This paper establishes a decomposition model based on LMDI method to analyze the influencing factors of CO₂ emission change in the power sector. The change of CO₂ emission generated by power generation can be decomposed into:

$$\begin{array}{c}C=\sum _{i=1}^3{C}_i=\sum _{i=1}^3\frac{C_i}{E}\times \frac{E}{\text{TP}}\times \frac{\text{TP}}{\text{EG}}\times \frac{\text{EG}}{\text{PD}}\times \text{PD}\\ =\sum _{i=1}^3{\text{FS}}_i\times \text{EI}\times \text{PS}\times \text{ET}\times \text{PD}\\ \end{array}$$ 2

where $C_i$ is the CO₂ emission of the energy type $i$, $E$ is the fossil energy consumption, $\text{TP}$ is the thermal power generation, $\text{EG}$ is the total power generation, and $\text{PD}$ is the total power demand. ${\text{FS}}_i$, $\text{EI}$, $\text{PS}$ and $\text{ET}$ are represent the fuel structure, energy efficiency, power generation structure, and electricity trade, respectively. The decomposition formula for the change in CO₂ emissions can be described as follows:

$$\mathrm{\Delta }C=C_i^T-C_i^t=\mathrm{\Delta }{C}_{\text{FS}}+\mathrm{\Delta }{C}_{\text{EI}}+\mathrm{\Delta }{C}_{\text{PS}}+\mathrm{\Delta }{C}_{\text{ET}}+\mathrm{\Delta }{C}_{\text{PD}}$$ 3

where $\mathrm{\Delta }{C}_{\text{FS}}$, $\mathrm{\Delta }{C}_{\text{EI}}$, $\mathrm{\Delta }{C}_{\text{PS}}$, $\mathrm{\Delta }{C}_{\text{ET}}$ and $\mathrm{\Delta }{C}_{\text{PD}}$ are the changes of FS, EI, PS, ET, and PD that affect the CO₂ emission change from year t to year T, respectively, and $C_i^T$ and $C_i^t$ are the CO₂ emission of energy type $i$ in year T and t, respectively.

$\mathrm{\Delta }{C}_{\text{FS}}$, $\mathrm{\Delta }{C}_{\text{EI}}$, $\mathrm{\Delta }{C}_{\text{PS}}$, $\mathrm{\Delta }{C}_{\text{ET}}$ and $\mathrm{\Delta }{C}_{\text{PD}}$ are calculated as follows, where $w$ is the log mean weight:

$$\mathrm{\Delta }{C}_{\text{FS}}=\sum _{i=1}^{3}w\times ln\left(\frac{{\text{FS}}_i^T}{{\text{FS}}_i^t}\right)$$ 4

$$\mathrm{\Delta }{C}_{\text{EI}}=\sum _{i=1}^{3}w\times ln\left(\frac{{\text{EI}}_{}^T}{{\text{EI}}_{}^t}\right)$$ 5

$$\mathrm{\Delta }{C}_{\text{PS}}=\sum _{i=1}^{3}w\times ln\left(\frac{{\text{PS}}_{}^T}{{\text{PS}}_{}^t}\right)$$ 6

$$\mathrm{\Delta }{C}_{\text{ET}}=\sum _{i=1}^{3}w\times ln\left(\frac{{\text{ET}}_{}^T}{{\text{ET}}_{}^t}\right)$$ 7

$$\mathrm{\Delta }{C}_{\text{PD}}=\sum _{i=1}^{3}w\times ln\left(\frac{{\text{PD}}_{}^T}{{\text{PD}}_{}^t}\right)$$ 8

$$w=\frac{C_i^T-C_i^t}{ln\left(C_i^T/C_i^t\right)}$$ 9

K-means clustering analysis method

The K-means clustering algorithm is based on Euclidean distance and is one of the most common statistical clustering methods. It can be clustered according to the spatial position of each object to be clustered, and the individuals with similar distances, that is, with similar characteristics, can be clustered into one class. First define the number of clusters k, and then assign the dataset into k clusters. Calculate the distance of the data point to the cluster center and redistribute the dataset into the closest cluster. And repeat the above steps through multiple iterations, and finally get the result of clustering.

Scenario analysis

Scenario analysis is a commonly used multivariate forecasting method to study the possible outcomes of a combination of factors. Scenario analysis has been widely used in carbon emission projections in recent years, which can predict the changing trend of carbon emission in the future. Through the multi-scenario setting, that is, simulating different paths of future development, the corresponding variable change speed parameters can be formulated, and the carbon emission trend under different scenarios can be predicted, and the CO₂ emission reduction potential can be obtained through comparative analysis.

Data sources

In this paper, the physical quantities of all kinds of fossil fuels in 30 provinces from 2000 to 2019 are obtained from the China Energy Statistical Yearbook. In addition, the total power generation, thermal power generation and power consumption of each province from 2000 to 2019 are from the China Statistical Yearbook. The average low calorific value and reduced standard coal coefficient of fossil fuels come from China Energy Statistical Yearbook in 2019 and General Principles for Calculation of Comprehensive Energy Consumption GB/T 2589–2020, and the carbon content and carbon oxidation rate of unit calorific value come from Guidelines for Compilation of Provincial Greenhouse Gas Inventory. It is assumed that the fuel parameters and emission coefficient are constant in the time span analyzed in this paper.

Results and discussion

Historical carbon emission calculation results

Figure 1 shows the change in CO₂ emissions from the power sector from 2000 to 2019, showing an overall upward trend, from 1.0 Gt in 2000 to 4.8Gt in 2019.

Fig. 1.

CO₂ emissions in China's power industry from 2000 to 2019

From 2000 to 2010, total CO₂ emissions increased from 1.0 Gt to 3.2 Gt, and most of the growth rates were above 10% and fluctuating. The reason behind this is that in the initial stage of the electricity market reform, the cost advantage of coal power was obvious, far lower than the cost of clean energy power generation, which promoted the growth of CO₂ emissions.

From 2011 to 2019, under the background of economic development, the power industry developed rapidly and production capacity expanded rapidly. However, driven by measures such as energy emission control, the growth rate of total emissions slowed down and was basically controlled at around 5%, from 3.7 to 4.5 Gt.

In general, since the fuel structure was dominated by coal, the CO₂ emissions from power industry mainly came from coal-fired power generation, accounting for 98.4% of the total. The proportion of CO₂ emissions from oil-fired power generation showed an overall downward trend, falling to 0.20% of the total by 2019. The proportion of CO₂ emissions from gas-fired generation has been increasing, reaching 2.01% of the total by 2019 and surpassing oil-fired generation.

Analysis of decomposition results

Based on the LMDI method, the CO2 emission factors of the power industry from 2000 to 2019 are decomposed. Figure 2 shows the contribution of each influencing factor to carbon emissions. The period from 2000 to 2019 is subdivided into four time periods, and the decomposition results of these four periods are analyzed in depth in this paper.

Fig. 2.

LMDI decomposition results of influencing factors from 2000 to 2019

Between 2000 and 2005, the industry's CO₂ emissions increased by 1.0Gt, with a growth rate of 98.10%. Energy efficiency effect, electricity trade effect and power demand scale effect contributed 98.58% increase in total CO₂ emissions, which was slightly offset by changes in fuel structure (− 0.43%) and power generation structure (− 0.05%).

Between 2005 and 2010, emissions increased by 1.1Gt, with a growth rate of 55.82%. The power demand had a reduced impact on CO₂ emissions (63.63%). During this period, the energy efficiency and power generation structure were optimized, and the impact on carbon emission reduction increased by − 9.77 and 4.21%, respectively.

From 2010 to 2015, CO₂ emissions of the industry increased by 0.5Gt, a growth rate of 16.36%. The impact of the electricity demand was further reduced (32.57%), and coupled with the fuel structure effect and the electricity trade effect, the positive influencing factors led to a 37.77% increase in total. The impact of energy efficiency and power generation structure effect on carbon emission reduction continued to increase, which together offset the impact of 21.42%.

Between 2015 and 2019, CO₂ emissions increased by 0.8Gt, with a growth rate of 21.71%. ${\text{FS}}_i$, $\text{EI}$, $\text{PS}$, $\text{ET}$ and $\text{PD}$ contributed 3.20, − 3.31, − 6.30, 1.79 and 26.34%, respectively. The results show that the electricity demand factor has dominated CO₂ emissions over the past 20 years, however the contribution is gradually decreasing as more of the new electricity demand is met by clean energy. The most critical factor in reducing CO₂ emissions was the improved of energy efficiency, followed by the optimization of power generation structure. Since the coal-dominated fuel mix has not been improved, the fuel mix effect has a very limited impact on carbon emissions. Changes in fuel structure and electricity trade also led to a slight increase.

Figure 3 shows the contribution of each province's influencing factors to carbon emissions. It can be seen that the contribution of each province's carbon emission influencing factors to carbon emissions varies greatly and changes dynamically over time, which verifies the research hypothesis of this paper. Except for the power demand scale effect that promoted carbon emissions in all provinces, other influencing factors had positive or negative effects on CO₂ emissions in all provinces. Between 2000 and 2005, the main driver of provincial CO₂ emissions was power demand scale effects. Electricity trade caused inter-provincial carbon transfer, in some provinces affected by “north–south power supply” and “west–east power transmission”, the power trade effect played a key role. For example, Guangdong, Shandong, Shanghai, Hebei and other power outsourcing provinces have thus reduced part of CO₂ emissions, while Guizhou, Yunnan, Shanxi, Shaanxi and other power outsourcing provinces have increased CO₂ emissions. The power generation structure effect reduced CO₂ emissions in Sichuan, Hubei, Zhejiang and other provinces with high level of clean energy utilization. Energy efficiency significantly contributed to Shandong's CO2 emissions (6.30%). The impact of fuel structure on CO₂ emissions in each province, whether positive or negative, was quite limited.

Fig. 3.

Relative effects of the influencing factors by province from 2000 to 2019

During the period from 2005 to 2010, the power demand effect remained the major factor driving CO₂ emissions in various provinces. Jiangsu's power demand effect was the most significant, resulting in the national emission growth of 11.00%. After the implementation of the policy of eliminating inefficient generating units (Qin et al. 2020), the energy efficiency effect and power generation structure effect reduced part of CO₂ emissions, contributing − 18.45 and − 5.80% to national emissions, respectively. The influence of power trade continued to increase in the provinces with “north–south power supply” and “west–east power transmission”. Due to the continuous increase in power demand, a large amount of coal-fired units had been invested, which increased the impact of fuel structure effects on CO₂ emissions in various provinces, with a contribution rate of 8.77%.

Between 2010 and 2015, the contribution of the power demand effect to provincial CO₂ emissions further increased. Shandong's power demand effect was considered the most significant driver, contributing 31.70% of the national emissions growth. The energy efficiency effect made a significant contribution to the emission reduction, with Shandong and Hebei offsetting 10.19 and 8.79% of the national CO₂ emissions, respectively. Power generation structure effect (Sichuan, etc.), fuel structure effect (Guizhou, etc.) and electricity trade effect (Henan, etc.) were the key factors of CO₂ emission reduction in some provinces.

From 2015 to 2019, among provinces, the power demand effect had the largest pulling effect on CO₂ emissions. For example, the contribution rate of power demand effect in Inner Mongolia and Shandong Province to the national CO₂ emissions is 16.36 and 11.47%, respectively. The power demand effect was offset by the energy efficiency effect in most provinces (Jiangsu, etc.) and the power generation structure effect in most provinces (Shandong, etc.), reducing the national CO₂ emissions by −21.73 and − 27.90%, respectively. The carbon emission reduction of Zhejiang (− 2.45%), Shaanxi (–2.06%) and Jiangsu (− 1.43%) were significantly affected by electricity trade effect. Fuel structure promoted carbon emissions in Jiangsu, Hebei, Beijing and Guangdong.

In general, the influencing forces varied by province, showing different effects across the four phased studies. Therefore, carbon reduction strategies should vary according to the specific conditions of each region.

Analysis of inter-provincial clustering and regression results

Cluster analysis

Faced with severe air pollution, the Chinese government has put forward a series of emission reduction measures, but there is still a big gap from the goals of carbon peak and carbon neutralization. Compared with the carbon emission reduction plans of other countries, such as the implementation of carbon sink projects in British agriculture and animal husbandry, the implementation of green building concept in Germany, and the development of a complete carbon tax system in Finland, the low-carbon transformation of China's power industry needs more targeted emission reduction plans. Based on the K-means clustering method, taking 5 influencing factors and CO₂ emissions of provinces from 2015 to 2019 as clustering variables, 30 provinces in China are divided into four clusters, in order to obtain more targeted emission reduction strategies. Figure 4 shows the clustering results of provinces, and Fig. 5 shows the carbon emission characteristics of different clusters.

Fig. 4.

China's provincial classification and influencing factors effect from 2015 to 2019

Fig. 5.

Influencing factors effect of each cluster from 2015 to 2019

Cluster 1 contributed 27.17% to the growth of carbon emissions, so the low-carbon transition of cluster 1 is urgent. Different from other clusters, all influencing factors of cluster 1 were positively pulling CO₂ emissions, in which the expansion of power demand scale played a leading role, and power generation structure and electricity trade also promoted CO₂ emissions. It is suggested to eliminate inefficient coal-fired power plants or introduce advanced technology to gradually improve fuel efficiency and optimize power structure.

Among the four clusters, cluster 2 caused the least increase in carbon emissions, accounting for only 4.30% of the national emission increment from 2015 to 2019. Growth in electricity demand contributed most to the cluster's CO₂ emissions (13.14%), but was largely offset by energy efficiency effects (− 7.81%) and generation mix effects (− 8.10%).

Cluster 3 covers four provinces, which contributed to a smaller increase in national carbon emissions, accounting for 9.35%. Different from other clusters, the effect of electricity trade on provinces in this cluster played a promoting role in carbon emission reduction. Having to purchase electricity from outside because local power generation could not meet demand, which resulted in carbon leakage and thus reduced CO₂ emissions in the provinces.

Cluster 4 contributed 59.18% from 2015 to 2019, resulting in the largest increase, so cluster 4 is the key area for emission reduction. The expansion of power demand played a leading role (69.57%) in the CO₂ emissions of provinces in the cluster, and the change of fuel structure also promoted CO₂ emissions. Energy efficiency and power generation structure offset a total of 20.82% of the increase. In the cluster, Jiangsu, Shandong and Hebei are major industrial provinces, and the high proportion of heavy industries and the high coal consumption are the fundamental reasons for the large contribution of these provinces to emissions. Therefore, provinces in cluster 4 need to introduce more advanced production technologies, improve industrial efficiency and encourage the development of new technology industries to achieve a low-carbon transition.

In terms of emission reduction measures, the economic feasibility of implementing different clusters varies. From the perspective of economic feasibility, compared with cluster 3 and cluster 4, most of the provinces in cluster 1 and cluster 2 are economically underdeveloped areas, which means that the implementation of carbon emission reduction policies in the power industry in these two clusters is more challenging and may face economic pressure. Therefore, the region should provide appropriate financial incentives, such as subsidies for the introduction of new technologies, which can provide targeted financial assistance for the promotion of carbon emission reduction policies in the power industry.

Fitting analysis of cluster carbon emissions

China's unique resource endowment has prompted China's energy consumption to be heavily biased toward coal. The proportion of coal consumption represents China's energy structure to a certain extent, and the proportion of coal input in the power generation process greatly affects the cleanliness of production. Therefore, in this paper, the proportion of coal consumption is used to characterize the structure the fuel structure in the actual research. The formula is obtained:

$$C=f(\text{PC,EI,PS,ET,PD})$$ 10

The logarithm of the formula can be obtained:

$$lnC={\alpha }_{1}ln\text{PC}+{\alpha }_{2}ln\text{EI}+{\alpha }_{3}ln\text{PS}+{\alpha }_{4}ln\text{ET}+{\alpha }_{5}ln\text{PD}+{\alpha }_0$$ 11

where, $\text{PC}$ represents the proportion of coal and its products in fossil energy consumption; $\text{EI}$ represents energy efficiency; $\text{PS}$ represents power generation structure; $\text{ET}$ and $\text{PD}$ represent electricity trade and power demand, respectively.

Firstly, time series data of cluster 1 are tested by unit root test, and as shown in Table 2 *** and ** denote the original assumed significance level of the existence of the unit root at a point is 1 and 5%, respectively. As can be seen from the test results, not all variables are stable. Each variable is integrated of order 2 and can be cointegration test. In this paper, the Engle-Granger two-step method is used to conduct cointegration test. Firstly, regression is performed for variables, and then unit root test is performed for residual items. If it is stable, it indicates that there is a cointegration relationship between variables. The stability test results of the residual term are shown in Table 3. It can be seen from the results that the residual term is stable and there is a co-integration relationship between variables, thus obtaining the co-integration equation:

$$\begin{array}{c}ln{C}_1=-0.387ln\text{PC}+0.961ln\text{EI}+0.943ln\text{PS}\\ +0.698ln\text{ET}+1.072ln\text{PD}-1.952\\ \end{array}$$ 12

Table 2.

Results of unit root test

Variables	ADF test	P–P test	Variables	ADF test	P–P test	Variables	ADF test	P–P test
$lnC$	−2.029	−2.423	$d(lnC)$	−3.34**	−3.341**	$d(lnC,2)$	−5.70***	−12.1***
$ln\text{PC}$	−3.944**	−3.943**	$d(ln\text{PC})$	−7.26***	−16.3***	$d(ln\text{PC,}2)$	−9.07***	−23.4***
$ln\text{EC}$	−0.925	−0.803	$d(ln\text{EC})$	−4.97***	−5.18***	$d(ln\text{EC},2)$	−7.17***	−18.4***
$ln\text{PS}$	−3.217**	−3.217**	$d(ln\text{PS})$	−5.93***	−5.93***	$d(ln\text{PS},2)$	−9.58***	−23.0***
$ln\text{ET}$	−2.114	−2.115	$d(ln\text{ET})$	−4.32**	−4.32**	$d(ln\text{ET},2)$	−7.90***	−10.9***
$ln\text{PD}$	−3.60**	−3.25**	$d(ln\text{PD})$	−2.049	−1.979	$d(ln\text{PD},2)$	−4.84***	−8.39***

Table 3.

Stationary test of residual term

Residual term	ADF test	P–P test
Resid01	− 5.356***	− 5.919***

Similarly, the cointegration equations of clusters 2, 3 and 4 can be obtained:

$$\begin{array}{c}ln{C}_2=-3.327ln\text{PC}+0.858ln\text{EI}+0.932ln\text{PS}\\ +1.073ln\text{ET}+1.007ln\text{PD}-1.522\\ \end{array}$$ 13

$$\begin{array}{c}ln{C}_3=-0.243ln\text{PC}+0.869ln\text{EI}+0.645ln\text{PS}\\ +0.786ln\text{ET}+0.946ln\text{PD}-1.032\\ \end{array}$$ 14

$$\begin{array}{c}ln{C}_4=-9.399ln\text{PC}+0.868ln\text{EI}+0.573ln\text{PS}\\ +0.801ln\text{ET}+1.054ln\text{PD}-1.965\\ \end{array}$$ 15

The following conclusions can be drawn from the equations for each of the above clusters:

All the coefficients in the cointegration equation of each cluster are different, which further verifies the research hypothesis that there are regional differences in the influencing factors of carbon emissions. The synergistic efficiencies of $\text{ln EC}$, $\text{ln PS}$, $\text{ln ET}$ and $\text{ln PD}$ of the four clusters are all positive, while the coefficient of $lnPC$ is negative, which is consistent with the social and economic development.

The coefficients of coal consumption ratio are − 0.387, − 3.327, − 0.243 and − 9.399 in the four clusters, respectively, indicating that the relationship with carbon emissions is negative. In addition, the coefficients of cluster 1 and cluster 3 are larger, which is due to the fact that the fuel structure of these two clusters has not been adjusted significantly in the past 20 years, resulting in that the carbon emission level is less affected by the proportion of coal. The proportion coefficient of coal in cluster 4 is the smallest, which indicates that the change of fuel structure in this cluster has the greatest impact on carbon emissions, and the optimization and adjustment of fuel structure will be conducive to the carbon emission reduction of this cluster. However, most of the provinces in this cluster are coal power provinces, which means that it is difficult to change the current situation that coal power occupies the dominant position in a short time.

The coefficient of energy efficiency and power generation structure is positive, indicating that improving energy efficiency and optimizing power generation structure will significantly improve carbon emission reduction potential. Cluster 1 has the largest coefficient of these two factors, indicating that the cluster can effectively promote carbon emission reduction by shutting down inefficient units, introducing advanced technologies, and vigorously developing renewable energy power generation. The generation structure coefficient of cluster 4 is the smallest, so the power generation structure of cluster 4 needs to be vigorously adjusted to play a role in carbon emission reduction.

The coefficient of electricity trade and power demand is positive, indicating that carbon leakage caused by inter-provincial and regional power exchange will increase the carbon emission of power output province, and the expansion of the scale of electricity demand will also promote the increase of carbon emissions. Cluster 2 has more power exporting provinces, and its electricity trade coefficient is the largest, indicating that inter-provincial and regional power exchange will lead to the increase of carbon emission of the provinces in the cluster. The newly increased power demand of cluster 3 is more satisfied by clean energy, resulting in that the impact of power demand on CO₂ emissions of this cluster less than that of other clusters, and the power demand coefficient of cluster 3 is the minimum. Therefore, it is recommended to promote power substitution in this cluster.

Analysis of CO₂ emission reduction potential

Scenario hypothesis

In order to clarify the carbon emission trend of China's power industry before 2040, this paper sets up three different development scenarios, namely reference scenario, existing policy scenario and advance peak scenario. Each scenario represents a corresponding future development policy planning path of China's power industry. It should be noted that in 2020, due to the impact of the COVID-19, the expansion of power demand scale slows down, and the electricity demand growth rate for that year is set at 3.1% for each cluster in all 3 scenarios.

The reference scenario assumes that China's power sector will not take any additional carbon reduction measures from 2019. In the reference scenario, the demand for electricity continues to grow with the development of the economy. From 2021 to 2040, the average annual growth rate of terminal power consumption of the four clusters will reach 6.85, 4.25, 6.93 and 6.61%, respectively. Power generation structure, the share of coal in fuels, energy efficiency and electricity trade will remain unchanged at 2019 levels.

The existing policy scenario are set based on the future development policies and planning documents promulgated by China’s power industry, including the 14th Five-Year Plan for Energy Development, the 14th Five-Year Plan for Renewable Energy Development, and the 2035 Visionary Goals Outline. Considering the increasingly severe resource and environmental pressures, China should actively take measures such as demand-side management to control energy consumption during this period. And gradually increase the proportion of low-carbon energy power generation, constantly eliminate backward coal power units and adding high-efficiency coal power units.

The advance peak scenario uses more aggressive policies than the existing policy scenario to accelerate the low-carbon transition in the power sector, and would set stricter targets for the control of total energy consumption, which is a relatively optimistic low carbon development scenario. The progress of industrial structure will continue to optimize the power demand structure, and the proportion of electricity used by the tertiary industry and residents will continue to rise. Compared with existing policy scenarios, the proportion of low-carbon energy generation in China will be further improved, and a part of small thermal power will guarantee power supply in the regional grid.

To explore the carbon reduction potential of power sector, this paper first predicts the future trend of the variables in Eqs. (12)-(15). This paper sets the parameters of coal proportion, energy efficiency, power generation structure, electricity trade and power demand scale in China from 2020 to 2040 by referring to the optimal scenario design of power decoupling by Wang et al. (2021) and Dong et al. (2021), and the parameter setting results are shown in Table 4.

Table 4.

Scenario parameters of China's power industry by cluster from 2020 to 2040

	Reference scenario				Existing policy scenario				Advance peak scenario
	2025	2030	2035	2040	2025	2030	2035	2040	2025	2030	2035	2040
Cluster 1
$\text{PC}$(%)	0.994	0.994	0.994	0.994	0.994	0.973	0.949	0.903	0.964	0.927	0.904	0.881
$\text{EI}$(%)	0.293	0.293	0.293	0.293	0.291	0.285	0.283	0.277	0.288	0.281	0.279	0.272
$\text{PS}$(%)	0.608	0.608	0.608	0.608	0.608	0.578	0.522	0.472	0.572	0.491	0.401	0.327
$\text{ET}$(%)	1.129	1.129	1.129	1.129	1.135	1.146	0.151	1.158	1.079	1.134	1.135	1.135
$\text{PD}$(100 billion kWh)	1.817	2.531	3.524	4.908	1.557	1.859	1.953	2.053	1.666	1.839	2.030	2.242
Cluster 2
$\text{PC}$(%)	0.994	0.994	0.994	0.994	0.994	0.994	0.984	0.974	0.995	0.995	0.993	0.988
$\text{EI}$(%)	0.282	0.282	0.282	0.282	0.279	0.276	0.275	0.273	0.276	0.274	0.273	0.272
$\text{PS}$(%)	0.479	0.479	0.479	0.479	0.451	0.428	0.387	0.332	0.446	0.403	0.361	0.310
$\text{ET}$(%)	1.056	1.056	1.056	1.056	1.062	1.072	1.078	1.083	1.025	1.040	1.046	1.051
$\text{PD}$(100 billion kWh)	2.167	2.669	3.287	4.047	2.070	2.373	2.495	2.662	2.160	2.326	2.506	2.700
Cluster 3
$\text{PC}$(%)	0.994	0.994	0.994	0.994	0.982	0.972	0.953	0.934	0.976	0.952	0.928	0.905
$\text{EI}$(%)	0.261	0.261	0.261	0.261	0.259	0.258	0.256	0.253	0.259	0.257	0.254	0.252
$\text{PS}$(%)	0.721	0.721	0.721	0.721	0.639	0.578	0.496	0.426	0.601	0.490	0.399	0.309
$\text{ET}$(%)	0.814	0.814	0.814	0.814	0.839	0.851	0.864	0.877	0.790	0.794	0.798	0.802
$\text{PD}$(100 billion kWh)	2.276	3.181	4.446	6.215	1.944	2.319	2.438	2.562	2.059	2.273	2.510	2.638
Cluster 4
$\text{PC}$(%)	0.999	0.999	0.999	0.999	0.998	0.997	0.996	0.995	0.998	0.998	0.997	0.996
$\text{EI}$(%)	0.304	0.304	0.304	0.304	0.302	0.300	0.297	0.294	0.302	0.299	0.296	0.293
$\text{PS}$(%)	0.860	0.860	0.860	0.860	0.762	0.689	0.573	0.467	0.762	0.590	0.456	0.353
$\text{ET}$(%)	1.097	1.097	1.097	1.097	1.130	1.147	1.165	1.182	1.033	1.038	1.043	1.048
$\text{PD}$(100 billion kWh)	3.855	5.310	7.314	10.07	3.340	3.986	4.294	4.626	3.451	3.885	4.289	4.736

Analysis of CO₂ emission reduction potential

By substituting the parameters of each scenario into Eqs. (12)–(15), the CO₂ emissions of each cluster and the whole country before 2040 under each scenario are obtained as shown in Fig. 6. In general, the CO₂ emission trends vary widely among scenarios due to the different intensity of policy measures adopted.

Fig. 6.

Prediction and comparison of CO₂ emissions from 2000 to 2040 under different scenarios of each cluster

Under the reference scenario, CO₂ emissions of each cluster will continue to increase. Due to the absence of active policies to combat climate change, the demand for electricity will rise significantly, but the fuel structure, energy efficiency and power generation structure will not be improved, and the technological level is stagnant, leading to the CO₂ emissions of cluster 1–4 to 2040 reaching 3.2, 1.8, 2.6 and 9.3 Gt, respectively, and the total national emissions are 1.7 Gt. This indicates that carbon emissions from China's power sector will not peak before 2030 if measures are not taken.

Under the existing policy scenario, according to the promulgated energy plan, the power industry will significantly reduce CO₂ emissions, and the CO₂ emission in 2040 will be reduced by 1.1 Gt compared with the reference scenario. The entire industry will peak at 6.1 Gt in 2030, and carbon emissions will gradually decrease from 2030. The four clusters will peak in 2030, 2029, 2030 and 2032, respectively, with the peak values of 1.1, 0.9, 0.9 and 3.2Gt.

In the advance peak scenario, more stringent power demand side management measures are implemented, and the power supply structure is further improved. China's power sector will peak at 5.2Gt in 2025, 5 years earlier than the existing policy scenario, and with 0.9 Gt lower peak carbon emissions. CO₂ emissions of each cluster will peak at 0.9, 0.8, 0.8 and 2.7 Gt in 2025, 2024, 2025 and 2027, respectively. Compared with the existing policy scenario, the total CO₂ emissions in 2040 will be reduced by 20.52% under the advance scenario.

Analysis on contribution of influencing factors of CO₂ emission

In order to understand the contribution of influencing factors to CO₂ emission in each cluster under the four scenarios, the LMDI model constructed above is used to decompose carbon emissions in each scenario, and the emission reduction contribution of each influencing factor is shown in Table 5.

1. Fuel structure ($\mathrm{\Delta }FS$)

   In the existing policy scenario, the total contribution of four clusters’ fuel structure optimization to CO₂ emissions is 1.1Gt, among which cluster 4 contributes the largest amount of 0.8Gt; in the advance peak scenario, the total contribution of four clusters to CO₂ emissions is 1.4Gt, among which cluster 4 contributes the largest amount of 1081.56Mt. In the future, with the improvement and promotion of carbon capture technology and the installation of CCUS equipment in coal-fired units, carbon emissions will be further reduced, which will play a more important role in CO₂ emission reduction.

2. Energy efficiency ($\mathrm{\Delta }\text{EI}$)

   Improvements in thermal power generation efficiency will continue to be one of the most important factors contributing to CO₂ reductions. From 2019 to 2040, in the existing policy scenario and the earlier peak scenario, the improvement of thermal power generation energy efficiency in the four clusters contributes 0.2 and 0.2Gt to CO₂ emission reduction, respectively. However, as the efficiency of power generation continues to improve in the future, the rate of technological progress will gradually decrease. Meanwhile, due to the existence of carbon lock-in effect of thermal power generation, the emission reduction contribution of energy efficiency may gradually decrease.

3. Power generation structure ($\mathrm{\Delta }\text{PS}$)

   In the existing policy scenarios, the contribution of low-carbon energy generation to CO₂ emissions in the four clusters reached − 0.2, − 0.3, − 0.4, and − 1.6Gt, respectively, contributing − 17.08, − 22.91, − 32.30, and − 130.05%, far exceeding 4.34, − 8.10, − 6.52, − 17.61% in 2000–2019; in the advance peak scenario, due to the more emphasis on low-carbon energy development, the impact of power generation structure to CO₂ emissions reaches − 0.4, − 0.3, − 0.6 and − 2.1Gt, which is the most effective CO₂ emission reduction factor. With the problems of traditional energy depletion and increased pressure on environmental protection, China will accelerate the development of renewable energy generation, and the emission reduction effect of low-carbon energy power generation will be further highlighted.

4. Electricity trade ($\mathrm{\Delta }\text{ET}$)

   In terms of relative quantity, the contribution of power trade effect to CO₂ emission of power industry is small. In the existing policy scenario, from 2019 to 2040, the power trade effect of the four clusters contributed 1.70, 1.58, 4.59 and 15.97%, respectively. However, in terms of absolute amount, the power trade effect also makes a great contribution to emission reduction. The cumulative CO₂ emissions from the electricity trading effect for the power sector in 2019–2040 amount to 0.3Gt and − 0.1Gt in the existing policy scenario and the advance peak scenario, respectively. In the future, with the completion of digital power grid and modern power grid, the power trade effect will make a continuous and stable contribution to CO₂ emission reduction.

5. Power demand ($\mathrm{\Delta }\text{PD}$)

   According to the decomposition results, from 2019 to 2040, the factor that contributes the most to CO₂ emissions in each scenario is the terminal power demand. From 2019 to 2040, the power demand effect under the three scenarios will increase by 0.1, 0.3 and 2.4Gt of CO₂, respectively. This means that the continuous guidance of electricity conservation through power demand-side management and active promotion of energy-saving products will become one of the most effective means of reducing CO₂ emissions in the power industry in the future.

Table 5.

Analysis of contribution of influencing factors from 2019 to 2040

Scenario		$\mathrm{\Delta }\text{PC}$(Gt)	$\mathrm{\Delta }\text{EI}$(Gt)	$\mathrm{\Delta }\text{PS}$(Gt)	$\mathrm{\Delta }\text{ET}$(Gt)	$\mathrm{\Delta }\text{PD}$(Gt)	$\mathrm{\Delta }C$(Gt)
Reference scenario	Cluster 1	0.000	0.000	0.000	0.000	2.4	2.4
	Cluster 2	0.000	0.000	0.000	0.000	1.0	1.0
	Cluster 3	0.000	0.000	0.000	0.000	1.9	1.9
	Cluster 4	0.000	0.000	0.000	0.000	7.0	7.0
Existing policy scenario	Cluster 1	0.08	−0.05	−0.2	0.2	0.4	0.2
	Cluster 2	0.08	−0.2	−0.3	0.02	0.3	0.1
	Cluster 3	0.1	−0.02	−0.4	0.05	0.4	0.1
	Cluster 4	0.8	−0.08	−1.6	0.2	1.4	0.8
Advance peak scenario	Cluster 1	0.08	−0.06	−0.4	0.01	0.4	0.01
	Cluster 2	0.04	−0.03	−0.3	−0.004	0.3	0.03
	Cluster 3	0.2	−0.02	−0.6	−0.01	0.4	−0.05
	Cluster 4	1.1	−0.09	−2.1	−0.1	1.3	0.09

Drawing from the above, the decomposition results of the factors affecting carbon emissions in China's power industry are fuel structure, energy efficiency, power generation structure, power trade and power demand. Compared with the LMDI decomposition results with socioeconomic indicators as the main influencing factors, this paper pays more attention to the direct influencing factors of carbon emissions in the power industry, and considers the impact of carbon leakage caused by power trade in various regions, which is helpful to formulate targeted emission reduction policies. According to the analysis results of influencing factors, power generation structure and energy efficiency are the two most important factors for the low-carbon transformation of the power industry. In the scenario analysis, the emission reduction measures under different scenarios have a significant impact on CO₂ emissions. Due to the active policies adopted in fuel structure, energy efficiency, power generation structure, power trade and other aspects, compared with the reference scenario, the entire power industry will peak in 2030 and 2025 under the existing policy scenario and the advance peak scenario, respectively, and the peak value under the advance scenario is lower. The results show that exploring differentiated emission reduction paths in different clusters will help achieve the goal of peaking the power industry by 2030. This study has some limitations that are worth noting. This paper focuses on the carbon emission analysis of China's power production, and does not consider the carbon emissions of power transmission and use. If there are reliable data, this point deserves attention in the future. In future research, it is necessary to analyze the carbon emissions in the whole process of power industry from production, transportation to use. In addition, the economic cost analysis of policy implementation is also the direction that needs to be worked on in the future. Researchers can also use other methods and samples (data of power industry in other countries, even data of other industries) to obtain new insights.

Conclusions and recommendations

This study decomposes the influencing factors of carbon emissions in China's power industry, verifies the research hypothesis that there are regional differences in the influencing factors of carbon emissions through cluster analysis and carbon emission regression model fitting analysis, and finally discusses the carbon emission trend and emission reduction potential of the power industry under different development modes through scenario setting, revealing the following results. First, different cluster carbon emission models (Eqs. 12–15) confirm the long-term relationship between various influencing factors and CO₂ emissions, which indicates that carbon emission reduction policies should be differentiated based on regional characteristics. Secondly, the active countermeasures in the scenario analysis are effective in advancing the peak time of carbon emissions and reducing the peak of carbon emissions. Compared with the reference scenario, the current policy scenario and the advance peak scenario for the entire power industry will reach the peak value in 2030 (6.1Gt) and 2025 (5.2Gt), respectively. And in the advance peak scenario, the power industry in each cluster can achieve the goal of peaking by 2030 in the 14th Five-Year Plan. The findings of this research can serve as a guide for other nations with high carbon characteristics and regional differences to transition their power industries to be low-carbon.

Combined with the research results, this paper proposes the following countermeasures for the future low-carbon development of the power industry:

1. On the Power production side. The first is to adjust the power generation structure. The large proportion of thermal power is the main problem on the power generation side, and the production of thermal power depends on coal, which is more significant in clusters 3 and 4. Therefore, controlling coal consumption is an effective way to reduce carbon emissions. We should constantly increase the proportion of clean energy power generation such as wind energy and photovoltaic to reduce the space for thermal power generation, with the proportion of thermal power generation falling to 46% in 2030 and 32% in 2040. After 2030, thermal power units should be phased out, and some of them can be converted into peak-shaving power sources or emergency backup power sources. Second, to improve the comprehensive utilization efficiency of energy. In particular, cluster 1 and cluster 4 need to strive to reduce energy consumption, accelerate the transformation of coal saving and consumption reduction through the development and introduction of new technologies to improve the energy conversion efficiency, and eliminate and shut down small and medium-sized generator sets with outdated technologies that cannot be retrofitted before 2030.The third is to develop and promote carbon capture and storage technology. Considering the technical and economic costs, in order to improve the level of decarbonization technology, it is suggested to implement it first in provinces with higher economic level in cluster 3 and cluster 4.

2. Power transmission and distribution side. There is a need to improve the grid connection issue of clean energy power generation. Cross-provincial and cross-regional power transactions should prioritize the development of low-cost clean energy power, especially cluster 2, which is a major power export province. It is proposed to actively promote the construction of large-scale clean energy power generation transmission channels by 2030, with the goal of prioritizing clean energy delivery and constantly improving power grid mutual aid and supply guarantee capacity. It is necessary to improve the distribution network's carrying capacity to accept new energy, as well as the transmission network's intelligent grid structure, in order to promote trans-provincial transmission of new energy and increase new energy consumption capacity.

   It is suggested to actively to promote the construction of large-scale new energy power generation transmission channels by 2030, and increase the external supply of low-carbon energy in areas with rich clean energy resources, so as to promote the realization of optimized allocation of large-scale energy resources.

3. Power consumption side. The first is to optimize the industrial structure. In particular, cluster 4 has a large number of industrial provinces, and it is recommended to curb the expansion of energy-intensive industries and actively develop modern service industries. Second, vigorously expand the area of replacing electric energy, promote the development of transportation electrification, tap the potential of replacing kilns and boilers in industrial production, and implement rural electrification upgrading projects to increase the proportion of electric energy in terminal consumption. Third, cultivate awareness of energy conservation. The power demand effect is the most important factor driving the increase of carbon emissions in the power industry, and cluster 4 is the most significant. Therefore, controlling power consumption through demand response and energy-saving transformation means of power demand side management is crucial to the low-carbon development of the power industry. It is recommended to implement economic measures such as reasonable industry price differentials and tiered electricity prices to reduce electricity consumption and waste, and to promote the application of energy-efficient products, equipment and technologies in high-energy-consuming industries such as steel and non-ferrous metals.

Supplementary Information

Below is the link to the electronic supplementary material.

Authors’ contributions

All authors contributed to the study conception and design. Conceptualization, supervision, writing—reviewing and editing were performed by WW. The first draft of the manuscript was written by QT. Material preparation and data collection were performed by BG. All authors read and approved the final manuscript.

Availability of data and materials

The dataset used and/or analyzed during the current study are available in the National Bureau of Statistics of China. http://www.stats.gov.cn/tjsj/ndsj/

Declarations

Conflict of interest

The authors declare no conflict of interest.