Carbon Emission Intensity of Final Electricity Consumption: Assessment and Decomposition of Regional Power Grids in China from 2005 to 2020

Abstract

Electricity consumption is a crucial source of carbon emissions. Many studies have focused on the emissions relevant to electricity generation, yet it is also important to evaluate the carbon intensity of final electricity use for the emission assessment of various industries. Considering fossil fuel consumption, regional exchange, and transmission loss, this study proposed a method to assess the emission intensities of electricity generation, supply, and use. Based on statistical data, the carbon intensities of North, Northeast, East, Central, Northwest, and South China power grids were evaluated during the period from 2005 to 2020, and the driving factors including energy structure, energy intensity, clean production, supply structure, and power loss effects were discussed using Logistic Mean Divided Index (LMDI) decomposition. An overall descending trend was observed for the carbon intensity, and the national average carbon intensities of electricity generation, supply, and use were 0.599, 0.599, and 0.622 kgCO_2e/kWh, respectively. The carbon intensity of South China gird was the lowest at 0.425 kgCO_2e/kWh in 2020. Moreover, the energy intensity effect and clean production effect were identified as the most influential factors of carbon intensity. The proposed method and results can be helpful to assess carbon emissions relevant to final electricity use.

1. Introduction

As one of the largest greenhouse gas emitters, China contributed approximately 31% of the energy-related emissions worldwide in 2021 [1]. Despite the challenge in economic development, the Chinese government has committed to reaching a CO₂ emission peak before 2030 [2], which demonstrates China’s determination to undertake emission reduction responsibilities. With the progress of urbanization and industrialization, electricity generation and demand have continuously grown over the past decades [3]. In 2021, the installed power generation capacity was 2.38 billion kW in China according to China Statistical Yearbook [4]. At the same time, the total electricity generation and consumption reached 8.53 and 8.33 trillion kWh, respectively, and coal-fired thermal electricity generation is still widely used in China [5]. Consequently, the electricity industry accounts for nearly 40% of the national carbon emissions [6,7]. Therefore, it is crucial to reduce carbon emissions from the electricity industry.

Accurate calculation of carbon emissions from the electricity industry is the foundation for achieving emission reduction. Previous studies have made attempts to investigate the total carbon emissions from electricity generation and the carbon intensity of electricity. For example, Ari and Koksal [8] investigated the carbon emissions from the Turkish electricity sector based on the fossil fuel combustion of power plants. Cao et al. [9] calculated the carbon emissions and energy consumption of distributed generation and proposed an approach to predict the carbon intensity of provincial power grids. Lindner et al. [10] considered electricity transfer and compared the carbon emissions of China’s provincial power grid from consumption and production perspectives. Harmsen and Graus [11] proposed a method to estimate the marginal carbon intensity of electricity, and compared the carbon reduction from electricity saving with the average carbon intensity of electricity generation. Ang and Su [12] investigated the carbon intensity in electricity production globally. They proposed an indicator named “aggregate carbon intensity”, which was defined as the ratio of total carbon emissions from electricity generation to the total electricity production. The indicator has been widely used in recent studies [13,14].

Based on the assessment of carbon emissions from electricity generation using historical data, potential approaches to reduce the carbon emissions from electricity generation, especially thermal power generation were explored and discussed. Considering the balance between economic development and environmental concerns, clean power generation technology such as the application of renewable energy has been regarded as a significant pathway toward the low-carbon electricity industry [15,16,17]. In this context, many studies have made efforts to predict the potential emission reduction from using clean power generation technology, which usually considered economic factors, technical improvements, and management measures [18,19,20,21]. Moreover, other than analyzing the benefits of replacing thermal electricity with renewable energy, previous studies have also concentrated on identifying and quantifying the driving factors of carbon emissions from electricity. Among the various decomposition models, Logistic Mean Divided Index (LMDI) decomposition has been widely applied [22,23,24].

Table 1. A summary of studies on carbon emissions of the electricity sector.

Source	Period	Region	Indicator	Driving Factor
Wang et al. [13]	1995–2014	China	Aggregate carbon intensity of electricity generation	Geographic distribution, thermal power proportion, utilization efficiency, and energy composition
Liu et al. [14]	2000–2014	China	Aggregate carbon intensity of electricity generation	Emissions coefficient, energy mix in thermal power generation, thermal generation efficiency, and clean energy penetration
Chen et al. [25]	2003–2015	China	Electric carbon productivity	Environmental efficiency, industrial structure, economic efficiency, economic scale, population scale, and the ratio of total population to electric CO2 emissions
Liao et al. [26]	2005–2015	China	CO2 emissions from electricity generation	Energy efficiency, electricity structure, electricity trade and electricity consumption scale
Chen et al. [27]	2003–2017	China	CO2 emissions from thermal electricity generation	Energy consumption structure, fossil energy conversion efficiency, power production structure, and total power generation
Peng and Tao [28]	1980–2014	China	Aggregate carbon intensity of electricity generation	Technological innovation and structural adjustment
Wang et al. [29]	1995–2012	Shandong Province, China	CO2 emissions from electricity generation	Electricity power production, power production structure, energy conversion efficiency, energy mix effect, and emission factor
Steenhof and Weber [30]	1990–2008	Canada	Carbon emissions from electricity generation	Generation mix, fossil fuel, generation efficiency, and emission intensity
Ang and Goh [31]	1990–2013	Association of Southeast Asian Nations	Aggregate carbon intensity of electricity generation	Fossil fuel share, fossil fuel mix, and generation efficiency

summarizes the studies on carbon emissions of the electricity sector and relevant driving factors [13,14,25,26,27,28,29,30,31]. With respect to electricity generation in China, the main driving factors considered in previous research can be classified as follows: (1) social and economic factors, including gross domestic product, population, electricity consumption per capita, and electricity generation scale [25,26,27]; (2) energy structure factors including share of clean power generation and fossil fuel types for thermal power generation [28,29]; and (3) technical factors, including electricity production efficiency, line loss rate, and geographic distribution [13,14]. Moreover, a few studies also investigated the driving factors of carbon intensity. Ang and Goh [31] studied the aggregate carbon intensity of electricity generation and suggested that thermal efficiency is a key factor to reduce carbon intensity from a global perspective. Wang et al. [13] analyzed the driving forces of aggregate carbon intensity in China and the results indicated that the proportion of clean power generation was also an influential factor.

Although many studies have been conducted to investigate the carbon emissions of electricity generation and relevant driving factors, there are still some knowledge gaps. First, the aggregate carbon intensity of electricity generation was evaluated on global, national, and provincial scales, which can be useful for policy-making on the low-carbon electricity industry. However, from the consumer’s view, it is also important to know the quantity of carbon emissions relevant to electricity use. In this context, it is necessary to investigate the carbon intensity of electricity consumption. Second, previous studies have concentrated on the driving factors of carbon emissions and carbon intensity from electricity generation, yet relatively few studies have considered the driving factors of the carbon intensity of electricity consumption. Hence, with consideration of electricity generation, regional power exchanges, and transmission loss, this study aims to propose a method to evaluate the carbon emission intensity of electricity consumption for the regional power grid in China and explore its driving factors, which covers the period from 2005 to 2020. The assessed transmission loss can provide fundamental data for electricity consumers to evaluate carbon emissions. The remainder of this paper is organized as follows: Section 2 introduces the method for evaluating the carbon emission intensity and LMDI decomposition; Section 3 describes the data on electricity generation and use, fossil fuel consumption for thermal power generation, and emission factors of fuel combustion; Section 4 presents and discusses the results; and Section 5 concludes the main findings and limitations of this study.

2. Methodology

2.1. Carbon Intensity Assessment

2.1.1. System Boundary

The climate change department of the Ministry of Ecological Environment of China divides the national power grids into the North, Northeast, East, Central, Northwest, and South grids [32].

Table 2. Geographical scopes of regional power grids in China.

Regional Grid	Geographical Scope
North China (NC)	Beijing, Tianjin, Hebei, Shanxi, Shandong, and Inner Mongolia
Northeast China (NEC)	Liaoning, Jilin, and Heilongjiang
East China (EC)	Shanghai, Jiangsu, Zhejiang, Anhui, and Fujian
Central China (CC)	Henan, Hubei, Hunan, Jiangxi, Sichuan, and Chongqing
Northwest China (NWC)	Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang
South China (SC)	Guangdong, Guangxi, Yunnan, Guizhou, and Hainan

summarizes the geographical scope of each regional grid. Tibet, Taiwan, Hong Kong, and Macao were not studied because of data scarcity.

Various energy sources, including thermal power, nuclear power, solar power, hydropower, and wind power, are currently applied for electricity generation. For thermal power generation, coal, oil products, and gaseous fuels are used for energy transformation, and, accordingly, fossil fuel combustion is the critical source of carbon emissions for electricity generation [13]. However, for the abovementioned clean power generation, although supply-chain carbon emissions from the equipment production, operation maintenance, and disposal processes are identified from a life cycle perspective, the energy transformation process produces almost no carbon emissions [33]. To maintain the consistency of the system boundary between thermal and clean power generation systems, this study considered only the energy transformation process for the assessment of carbon intensities.

With respect to the carbon intensity associated with regional power grids, the fundamental concept of this indicator can be described as follows [12]:

$$E{F}_i=\frac{C_i}{E_i}$$

(1)

where EF_i, C_i, and E_i are the carbon intensity, total carbon emissions, and gross electricity generation, supply, or final consumption, respectively, of power grid i.

As shown in Figure 1, the above description indicates that the carbon intensity of electricity can have multiple definitions. First, E_i may represent the generated, supplied, and final consumed electricity, and the relevant carbon intensities can be the aggregate carbon intensity of electricity generation, the carbon intensity of electricity supply, and the associate carbon intensity of final electricity consumption, respectively. The electricity production includes both thermal electricity and electricity generated from clean energy; the electricity supply includes locally generated electricity and imported electricity for local use; and the electricity consumption included corresponds to the regional final electricity consumption, which excludes the transmission loss from the electricity supply. The clear energy in this study refers to renewable energy (such as solar, wind, and hydro energy) and nuclear power. Second, C_i may be quantified using production-based and consumption-based methods. For the production-based method, carbon emissions of imported electricity from other regional grids or abroad are not included, and carbon emissions of exported electricity are not excluded. However, for the consumption-based method, the opposite is true. In this study, the carbon intensity of final electricity consumption was analyzed using the consumption-based accounting method, which could reflect the real level of carbon emissions from final energy use (such as energy use of the industry and building sectors).

2.1.2. Assessment Method

As shown in Figure 2, the carbon intensities of electricity generation, supply, and final use are evaluated according to the procedure as follows:

(1)

Obtain the data on electricity generation for each regional power grid considering both clean and thermal power generation.

(2)

Obtain the data on fossil fuel consumption for thermal power production, and calculate the relevant carbon emissions.

(3)

Assess the aggregate carbon intensity of electricity generation based on the gross electricity generation and carbon emissions.

(4)

Evaluate the electricity supply, which encompasses the electricity generated for local use and the electricity imported from other regional grids.

(5)

Evaluate the carbon emissions relevant to electricity supply and assess the relevant carbon intensity.

(6)

Subtract transmission loss from the electricity supply to obtain the final electricity use.

(7)

Evaluate the carbon emissions relevant to final electricity use and assess the associate carbon intensity.

According to Intergovernmental Panel on Climate Change (IPCC) Guidelines for National Greenhouse Gas Inventories [34], carbon emissions from fossil fuel combustion can be calculated as the product of energy consumption and relevant emission factor. The emission factor of a certain type of fossil fuel is generally measured as follows:

$$E{F}^{\mathrm{e}}=\left(\frac{44}{12}{C}_{\mathrm{C}}{R}_{\mathrm{O}}+C_{{\mathrm{CH}}_4}{\mathrm{GWP}}_{{\mathrm{CH}}_4}+C_{{\mathrm{N}}_2\mathrm{O}}{\mathrm{GWP}}_{{\mathrm{N}}_2\mathrm{O}}\right)Q_{\mathrm{cv}}$$

(2)

where EF^e is the emission factor of the fuel, C_c is the carbon content per unit calorific value, R_o is the oxidation rate of the fuel, $C_{{\mathrm{CH}}_4}$ and $C_{{\mathrm{N}}_2\mathrm{O}}$ are the methane (CH₄) and nitrous oxide (N₂O) emitted during fuel combustion, respectively, ${\mathrm{GWP}}_{{\mathrm{CH}}_4}$ and ${\mathrm{GWP}}_{{\mathrm{N}}_2\mathrm{O}}$ are the global warming potential (GWP) of CH₄ and N₂O, respectively, and $Q_{\mathrm{cv}}$ is the net calorific value.

Based on the above method, carbon emissions from thermal power generation in each year can be calculated as follows:

$$C_i^{\mathrm{thm}}={\displaystyle \sum _{j=1}^{J}E{F}_j^{\mathrm{e}}\cdot Q_{ij}}={\displaystyle \sum _{p=1}^i{\displaystyle \sum _{j=1}^{J}E{F}_j^{\mathrm{e}}\cdot Q_{pj}}}$$

(3)

where Q_ij and Q_pj are the consumption of fuel j used for energy transformation in power grid i and province p, respectively, $E{F}_j^{\mathrm{e}}$ is the emission factor of fuel j, and J is the total number of fuel types.

Moreover, considering that carbon emissions from clean electricity production are assumed to be zero [13,14,33], the total carbon emissions for electricity generation in power grid i, $C_i^{\mathrm{pro}}$ are equal to $C_i^{\mathrm{thm}}$. Hence, the aggregate carbon intensity of electricity generation in power grid i, $E_i^{\mathrm{pro}}$ can be estimated as:

$$E{F}_i^{\mathrm{pro}}=\frac{C_i^{\mathrm{pro}}}{E_i^{\mathrm{pro}}}=\frac{C_i^{\mathrm{pro}}}{E_i^{\mathrm{thm}}+E_i^{\mathrm{c}\ln }}$$

(4)

where $E_i^{\mathrm{pro}}$ is total electricity generation in power grid i, $E_i^{\mathrm{thm}}$ and $E_i^{cln}$ are the thermal and clean electricity generation in power grid i, respectively. Notably, the electricity generated by a power station can be divided into (1) electricity supply to the power grids and (2) electricity used by the operation of the power station. Herein, $E_i^{\mathrm{pro}}$, $E_i^{\mathrm{thm}}$, and $E_i^{\mathrm{cln}}$ are net electricity generation accessed to the power grids that the electricity demanded by the station itself have been excluded.

Considering the power exchange, the electricity generation in power grid i can be divided into three components, including electricity supplied for local use, electricity supplied to other regional grids in China, and electricity exported abroad. In this context, the gross electricity supply in power grid i, $E_i^{\mathrm{supply}}$ is calculated as follows:

$$E_i^{\mathrm{supply}}=E_i^{\mathrm{pro}}+{\displaystyle \sum _{q=1,q\ne i}^6\left(E_{iq}^{\mathrm{pro},\mathrm{in}}-E_{qi}^{\mathrm{pro},\mathrm{out}}\right)}+{\displaystyle \sum _{r=1}^R\left(E_{ir}^{\mathrm{pro},\mathrm{im}}-E_{ri}^{\mathrm{pro},\mathrm{ex}}\right)}$$

(5)

where $E_{iq}^{\mathrm{pro},\mathrm{in}}$ represents the electricity supplied to power grid i by power grid q, $E_{qi}^{\mathrm{pro},\mathrm{out}}$ represents the electricity supplied to power grid q by power grid i, $E_{ir}^{\mathrm{pro},\mathrm{im}}$ represents the electricity imported from country or region r abroad, $E_{ri}^{\mathrm{pro},\mathrm{ex}}$ represents the electricity exported to country or region r abroad, and R is the total number of involved countries and regions.

Considering consumption-based carbon emission accounting, the total carbon emissions for electricity supply in power grid i, $C_i^{\mathrm{supply}}$ can be assessed using Equation (6), and the relevant carbon intensity for electricity supply, $E{F}_i^{\mathrm{supply}}$ can be estimated using Equation (7).

$$C_i^{\mathrm{supply}}=E{F}_i^{\mathrm{pro}}\cdot \left(E_i^{\mathrm{pro}}-{\displaystyle \sum _{q=1,q\ne i}{E}_{qi}^{\mathrm{pro},\mathrm{out}}}-{\displaystyle \sum _r{E}_{ri}^{\mathrm{pro},\mathrm{ex}}}\right)+{\displaystyle \sum _{q=1,q\ne i}E{F}_q^{\mathrm{pro}}\cdot E_{iq}^{\mathrm{pro},\mathrm{in}}}+{\displaystyle \sum _{r}E{F}_r^{\mathrm{pro},\mathrm{im}}\cdot E_{ir}^{\mathrm{pro},\mathrm{im}}}$$

(6)

$$E{F}_i^{\mathrm{supply}}=\frac{C_i^{\mathrm{supply}}}{E_i^{\mathrm{supply}}}$$

(7)

where $E{F}_q^{\mathrm{pro}}$ and $E{F}_r^{\mathrm{pro},\mathrm{im}}$ represent the carbon intensities of electricity generation in power grid q and country or region r abroad, respectively.

It should be emphasized that the imported and exported electricity abroad were not considered in this study for two main reasons. First, the power exchange between China and other countries (regions) is relatively small compared to the gross power supply nationwide. Second, aggregate carbon intensities of electricity generation abroad are difficult to obtain. Therefore, the above method of estimating the carbon intensity of electricity supply is simplified as follows:

$$E{F}_i^{\mathrm{supply}}=\frac{E{F}_i^{\mathrm{pro}}\cdot \left(E_i^{\mathrm{pro}}-{\displaystyle \sum _{q=1,q\ne i}{E}_{qi}^{\mathrm{pro},\mathrm{out}}}\right)+{\displaystyle \sum _{q=1,q\ne i}E{F}_q^{\mathrm{pro}}\cdot E_{iq}^{\mathrm{pro},\mathrm{in}}}}{E_i^{\mathrm{pro}}+{\displaystyle \sum _{q=1,q\ne i}^6\left(E_{iq}^{\mathrm{pro},\mathrm{in}}-E_{qi}^{\mathrm{pro},\mathrm{out}}\right)}}$$

(8)

For conciseness and clarity, the numerator (namely, $C_i^{\mathrm{supply}}$) in Equation (8) can be further rewritten as follows:

$$C_i^{\mathrm{supply}}={\displaystyle \sum _{k=1}^{6}E{F}_k^{\mathrm{pro}}\cdot E_{ik}^{\mathrm{pro}}}$$

(9)

where $E{F}_k^{\mathrm{pro}}$ is the aggregate carbon intensity of electricity generation in power grid k, $E_{ik}^{\mathrm{pro}}$ represents the electricity supplied to power grid i by power grid k, and when k = i, the term $E{F}_k^{\mathrm{pro}}\cdot E_{kk}^{\mathrm{pro}}$ equals to the term $E{F}_i^{\mathrm{pro}}\cdot \left(E_i^{\mathrm{pro}}-{{\displaystyle \sum }}_{q=1,q\ne i}{E}_{qi}^{\mathrm{pro},\mathrm{out}}\right)$.

Furthermore, the electricity for final consumption is lower than the electricity supply considering the loss during power transmission, and the loss rate is defined as:

$${\lambda }_i=\frac{E_i^{\mathrm{loss}}}{E_i^{\mathrm{supply}}}=\frac{E_i^{\mathrm{supply}}-E_i^{\mathrm{use}}}{E_i^{\mathrm{supply}}}$$

(10)

where ${\lambda }_i$ is the loss rate of electricity transmission in power grid i, $E_i^{\mathrm{use}}$ is the final electricity consumption in power grid i, and $E_i^{\mathrm{loss}}$ is the corresponding gross electricity loss, which can be calculated as the difference between electricity supply and final consumption.

Hence, the associate carbon intensity of final electricity consumption, $E{F}_i^{\mathrm{use}}$ can be obtained by correcting $E{F}_i^{\mathrm{supply}}$ as follows:

$$E{F}_i^{\mathrm{use}}=\frac{E{F}_i^{\mathrm{supply}}\cdot E_i^{\mathrm{supply}}}{E_i^{\mathrm{use}}}=\frac{E{F}_i^{\mathrm{supply}}}{1-{\lambda }_i}$$

(11)

Moreover, combining Equations (3), (4), (7), (9) and (11), the above equation can be rewritten as:

$$E{F}_i^{\mathrm{use}}={\displaystyle \sum _{k=1}^6{\displaystyle \sum _{j=1}^J\frac{E{F}_j^{\mathrm{e}}\cdot Q_{kj}\cdot E_{ik}^{\mathrm{pro}}}{E_k^{\mathrm{pro}}\cdot E_i^{\mathrm{use}}}}}={\displaystyle \sum _{k=1}^6{\displaystyle \sum _{j=1}^J\frac{C_{ikj}^{\mathrm{use}}}{E_i^{\mathrm{use}}}}}$$

(12)

where $C_{ikj}^{\mathrm{use}}$ represents the contribution of fuel j consumption in power grid k to the implied carbon emissions of final electricity consumption in power grid i.

2.2. LMDI Decomposition

Based on the additive LMDI method suggested by Ang [35], the abovementioned associate carbon intensity of final electricity consumption is decomposed as follows:

$$\begin{array}{ll}E{F}_i^{\mathrm{use}}& ={\displaystyle \sum _k{\displaystyle \sum _{j}E{F}_j^{\mathrm{e}}\cdot \frac{Q_{kj}}{Q_k}\cdot \frac{Q_k}{E_k^{\mathrm{thm}}}\cdot \frac{E_k^{\mathrm{thm}}}{E_k^{\mathrm{pro}}}\cdot \frac{E_{ik}^{\mathrm{pro}}}{E_i^{\mathrm{pro}}}}}\cdot \frac{E_i^{\mathrm{pro}}}{E_i^{\mathrm{use}}}\\ & ={\displaystyle \sum _k{\displaystyle \sum _{j}E{F}_j^{\mathrm{e}}\cdot E{S}_{kj}\cdot E{I}_k\cdot C{U}_k\cdot S{S}_{ik}}}\cdot P{L}_i\end{array}$$

(13)

where Q_kj is the consumption of fuel j in terms of standard coal in power grid k, Q_k is the comprehensive energy input for electricity transformation in power grid k, $E_k^{\mathrm{thm}}$ is the thermal power generation in power grid k, and $E_k^{\mathrm{pro}}$ is the total electricity generation in power grid k.

Furthermore, Equation (13) shows that $E{F}_i^{\mathrm{use}}$ is decomposed using six influencing factors: $E{F}_j^{\mathrm{e}}$ is the emission factor of fuel j, incorporating emission factor effect; $E{S}_{kj}=\frac{Q_{kj}}{Q_k}$ is the contribution of fuel j to the total fuel consumption for thermal energy generation in power grid k, incorporating the production energy structure effect; $E{I}_k=\frac{Q_k}{E_k^{\mathrm{thm}}}$ is the ratio of total fuel consumption to thermal electricity generation in power grid k, incorporating the production energy intensity effect; $C{U}_k=\frac{E_k^{\mathrm{thm}}}{E_k^{\mathrm{pro}}}$ is the proportion of thermal electricity generation in total electricity generation in power grid k, incorporating the clean energy utilization effect; $S{S}_{ik}=\frac{E_{ik}^{\mathrm{pro}}}{E_i^{\mathrm{pro}}}$ is the proportion of electricity supplied by power gird k in the total electricity supply in power grid i, incorporating the energy supply structure effect; and $P{L}_i=\frac{E_i^{\mathrm{pro}}}{E_i^{\mathrm{use}}}$ is the ratio of final electricity consumption to total electricity supply in power grid i, incorporating power transmission loss effect.

Generally, emission factors of fuels are considered as constant values. Hence, the change in $E{F}_i^{\mathrm{use}}$ between the adjacent two years t and t − 1 can be decomposed into five parts as shown in Equation (14), and the individual effect by each influencing factor can be calculated using Equations (15)–(20).

$$\Delta E{F}_i^{\mathrm{use}}=E{F}_i^{\mathrm{use}}\left(t\right)-E{F}_i^{\mathrm{use}}\left(t-1\right)=\Delta E{F}_{ES,i}+\Delta E{F}_{EI,i}+\Delta E{F}_{CU,i}+\Delta E{F}_{SS,i}+\Delta E{F}_{PL,i}$$

(14)

$$\Delta E{F}_{ES,i}={\displaystyle \sum _k{\displaystyle \sum _{j}W\left[E{F}_{ikj}^{\mathrm{use}}\left(t\right),E{F}_{ikj}^{\mathrm{use}}\left(t-1\right)\right]\ln \frac{E{S}_{kj}\left(t\right)}{E{S}_{kj}\left(t-1\right)}}}$$

(15)

$$\Delta E{F}_{EI,i}={\displaystyle \sum _k{\displaystyle \sum _{j}W\left[E{F}_{ikj}^{\mathrm{use}}\left(t\right),E{F}_{ikj}^{\mathrm{use}}\left(t-1\right)\right]\ln \frac{E{I}_k\left(t\right)}{E{I}_k\left(t-1\right)}}}$$

(16)

$$\Delta E{F}_{CU,i}={\displaystyle \sum _k{\displaystyle \sum _{j}W\left[E{F}_{ikj}^{\mathrm{use}}\left(t\right),E{F}_{ikj}^{\mathrm{use}}\left(t-1\right)\right]\ln \frac{C{P}_k\left(t\right)}{C{P}_k\left(t-1\right)}}}$$

(17)

$$\Delta E{F}_{SS,i}={\displaystyle \sum _k{\displaystyle \sum _{j}W\left[E{F}_{ikj}^{\mathrm{use}}\left(t\right),E{F}_{ikj}^{\mathrm{use}}\left(t-1\right)\right]\ln \frac{U{S}_{ik}\left(t\right)}{U{S}_{ik}\left(t-1\right)}}}$$

(18)

$$\Delta E{F}_{PL,i}={\displaystyle \sum _k{\displaystyle \sum _{j}W\left[E{F}_{ikj}^{\mathrm{use}}\left(t\right),E{F}_{ikj}^{\mathrm{use}}\left(0\right)\right]\ln \frac{P{L}_i\left(t\right)}{P{L}_i\left(t-1\right)}}}$$

(19)

$$W\left[E{F}_{ikj}^{\mathrm{use}}\left(t\right),E{F}_{ikj}^{\mathrm{use}}\left(t-1\right)\right]=\frac{E{F}_{ikj}^{\mathrm{use}}\left(t\right)-E{F}_{ikj}^{\mathrm{use}}\left(t-1\right)}{\ln E{F}_{ikj}^{\mathrm{use}}\left(t\right)-\ln E{F}_{ikj}^{\mathrm{use}}\left(t-1\right)}$$

(20)

where $\mathsf{\Delta }E{F}_i^{\mathrm{use}}$ represents the total change in $E{F}_i^{\mathrm{use}}$ during the period from year t − 1 to year t, and $\mathsf{\Delta }E{F}_{ES,i}$, $\mathsf{\Delta }E{F}_{EI,i}$, $\mathsf{\Delta }E{F}_{CU,i}$, $\mathsf{\Delta }E{F}_{SS,i}$, and $\mathsf{\Delta }E{F}_{PL,i}$ are the corresponding changes in $E{F}_i^{\mathrm{use}}$ caused by the production energy structure, production energy intensity, clean energy utilization, energy supply structure, and power transmission loss effects, respectively.

Moreover, to avoid failure of calculating logarithms in the decomposition, zero values were replaced with small numbers (10⁻²⁰), as suggested by Ang and Liu [36].

3. Data Collection and Processing

3.1. Electricity Generation, Supply, and Consumption

This study covers a research period from 2005 to 2020. Electricity generated from both thermal power transformation and clean energy sources of each province are collected from regional energy balance sheets in China Energy Statistical Yearbook (CESY) [37], and the electricity generation in each power grid was calculated according to the geographical scope as shown in Table 2. Raw data on electricity exchanged between different power grids are obtained from China Power Statistical Yearbook (CPSY) [38]. Based on the electricity generation and exchange data, the electricity supply of each power grid was calculated according to Equation (5). Moreover, the final electricity consumption was obtained from CESY to estimate transmission loss rates using Equation (10). Electricity generation, supply, and final consumption are summarized in

Table 3. Electricity generation, supply, and final consumption (10⁶ MWh) from 2005 to 2020.

Source	Power Grid	Year
		2005	2006	2007	2008	2009	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020
Generation	NC	625	723	847	879	938	1067	1175	1246	1327	1371	1462	1550	1619	1841	1927	1978
	NEC	196	212	233	243	248	278	298	305	319	326	322	346	358	392	411	429
	EC	574	667	755	807	858	983	1099	1141	1198	1194	1223	1303	1379	1465	1500	1507
	CC	502	553	640	673	732	837	947	973	1058	1105	1119	1167	1224	1326	1350	1394
	NWC	189	218	257	282	301	374	466	525	608	672	709	736	820	925	1012	1113
	SC	414	482	548	590	626	714	787	838	911	966	1010	1057	1121	1217	1297	1352
Supply	NC	618	715	837	868	926	1059	1191	1262	1345	1393	1511	1600	1664	1885	1977	2058
	NEC	192	209	231	238	241	270	288	294	301	304	304	325	336	357	366	377
	EC	613	702	799	859	912	1040	1148	1210	1299	1337	1370	1467	1570	1685	1749	1805
	CC	455	510	586	618	679	789	917	925	984	996	1010	1074	1134	1257	1299	1342
	NWC	189	215	254	278	298	360	424	482	569	625	630	642	701	772	818	848
	SC	434	504	573	612	648	738	803	854	923	978	1021	1050	1115	1211	1287	1344
Final use	NC	596	693	809	840	896	1025	1164	1224	1304	1352	1471	1560	1596	1831	1923	2008
	NEC	183	200	226	233	235	258	276	279	285	288	295	312	317	335	347	359
	EC	575	661	753	812	863	978	1082	1143	1238	1274	1310	1405	1505	1624	1690	1743
	CC	424	477	548	579	630	738	862	871	913	925	940	1019	1080	1174	1217	1265
	NWC	183	203	239	262	280	341	408	464	547	601	606	626	683	745	789	817
	SC	409	468	534	574	611	696	759	803	867	920	965	998	1065	1152	1230	1290

.

3.2. Fossil Fuel Consumption

For thermal power generation, coal, liquid fuels, and gaseous fuels are usually used, yet for clean power generation, primary energy such as solar and wind energy is used. In this study, only the fossil fuels used for power transformation were considered for carbon emission assessment, and the upstream energy consumption for the construction of power generation facilities was not included. Fuel consumption for provincial power generation in China was obtained from regional energy balance tables in CESY [37]. Sixteen kinds of fossil fuels were considered including raw coal, cleaned coal, other washed coal, briquettes, gangue, coke oven gas, blast furnace gas, converter gas, crude oil, diesel oil, fuel oil, petroleum coke, refinery gas, other petroleum products, natural gas, and liquefied natural gas. The original provincial data were merged into the fuel consumption of six power grids according to Table 2. The total fossil fuel consumption for electricity generation measured by standard coal equivalent (ce) is shown in

Table 4. Fossil fuel consumption for electricity generation (10⁶ tce) from 2005 to 2020.

Year	Power Grid
	NC	NEC	EC	CC	NWC	SC
2005	237.9	77.6	169.9	129.9	54.4	111.8
2006	268.0	86.5	193.1	148.8	61.6	126.7
2007	304.3	90.1	214.6	164.5	72.2	141.9
2008	325.2	95.2	223.0	157.3	78.2	133.8
2009	333.3	93.5	232.1	161.4	82.7	144.6
2010	386.3	102.3	262.8	184.8	101.0	162.8
2011	448.5	114.0	304.7	213.3	132.7	187.6
2012	456.9	112.1	306.4	189.5	145.4	178.3
2013	442.6	106.5	323.2	208.5	161.4	183.5
2014	440.4	111.6	306.4	192.8	172.9	166.0
2015	459.9	106.8	300.2	182.0	177.4	155.4
2016	474.7	106.1	308.4	179.1	182.0	157.5
2017	495.6	111.2	330.6	185.2	200.1	179.1
2018	557.3	111.1	344.4	205.2	216.4	190.6
2019	570.6	117.3	340.0	205.7	233.8	197.2
2020	569.5	113.8	326.3	192.0	257.5	203.2

. Notably, “1 tce” refers to 1 ton of standard coal equivalent which equals 7000 kcal when expressed in calorific value [38]. This standard coal equivalent can be used to generalize the unit of fossil fuel consumption, and, therefore, it is possible to compare the energy use for electricity generation in different years.

3.3. Emission Factor of Fossil Fuels

Emission factors of the abovementioned fossil fuels were evaluated based on Equation (2), and the results are presented in

Table 5. Emission factors of fossil fuels.

Fuel Type	Unit	Carbon Content (tC/TJ)	Oxidation Rate (%)	Greenhouse Gas Emission			Net Calorific Value (kJ/Unit)	Emission Factor (kgCO2e/Unit)
				CO2 (t/TJ)	CH4 (t/TJ)	N2O (t/TJ)
Raw coal	kg	26.4	94	90.99	0.001	0.0015	20908	1.912
Cleaned coal	kg	25.4	98	91.27	0.001	0.0015	26344	2.416
Other washed coal	kg	25.4	98	91.27	0.001	0.0015	19969	1.831
Briquette	kg	33.6	90	110.88	0.001	0.0015	15472	1.722
Gangue	kg	25.8	98	92.71	0.001	0.0015	8363	0.779
Coke oven gas	m3	12.1	100	44.37	0.001	0.0001	17354	0.771
Blast furnace gas	m3	70.8	100	259.60	0.001	0.0001	3763	0.977
Converter gas	m3	49.6	100	181.87	0.001	0.0001	7945	1.445
Crude oil	kg	20.1	98	72.23	0.003	0.0006	41816	3.031
Diesel oil	kg	20.2	98	72.59	0.003	0.0006	42652	3.107
Fuel oil	kg	21.1	98	75.82	0.003	0.0006	41816	3.181
Petroleum coke	kg	27.5	98	98.82	0.003	0.0006	31947	3.165
Refinery gas	kg	18.2	98	65.40	0.001	0.0001	46055	3.015
Other petroleum product	kg	20.0	98	71.87	0.003	0.0006	40980	2.955
Natural gas	m3	15.3	99	55.54	0.001	0.0001	38931	2.164
Liquefied natural gas	kg	17.2	98	61.81	0.003	0.0006	51434	3.192

. The sources of carbon content, oxidation rate, and net calorific value are as follows: (1) the values of carbon content and oxidation rate were obtained from Guidelines for Provincial Greenhouse Gas Inventories [39]; (2) the methane and nitrous oxide emission were obtained from IPCC Guideline for National Greenhouse Gas Inventories [33]; and (3) the net calorific values of fuels were obtained from CESY, which was consistent with the source of fuel consumption data.

4. Results and Discussion

4.1. Carbon Intensity of Electricity in Regional Power Grids

The aggregate carbon intensity of electricity generation in each power grid was evaluated according to Equation (4), and the results are shown in Figure 3. An overall descending trend was observed for the carbon intensity in all power grids, especially during the period from 2011 to 2016. The carbon intensities of electricity generation in 2005 were 1.024, 1.063, 0.801, 0.693, 0.761, and 0.708, respectively, in the NC, NEC, EC, CC, NWC, and SC power grids. The values decreased to 0.825,0.757,0.603,0.396,0.617, and 0.407, respectively, in 2020. The national average carbon intensity changed from 0.837 in 2005 to 0.599 in 2020, and the total decrease rate reach 28.5%. Moreover, based on the relevant values and trends, the carbon intensity of power generation in the six regional grids can be divided into three groups as follows: (1) the first group includes the NC and NEC grids, of which the carbon intensity was relatively high; (2) the second group includes the EC and NWC grids, of which the carbon intensity was closed to the national average level in recent year, whereas the changes were relatively gentle in the assessment period; and (3) the third group includes the CC and SC grids, of which the carbon intensity was relatively low and the decrease rate was relatively high (42.7% on average from 2005 to 2020).

Considering power exchanges among the regional grids, the carbon intensity of electricity supply in each power grid was evaluated according to Equation (8). The difference in carbon intensities between electricity generation and supply was analyzed, and the results are shown in Figure 4. For the NEC, NWC, and SC grids, the electricity supplied by other grids was not considerable, and, therefore, the carbon intensities of electricity generation and supply were similar. For the CC grid, the carbon intensity of electricity supply was higher than that of electricity generation, and the difference between them increased in recent years. The reason was that the CC grid used more electricity supplied by the NWC gird, of which the aggregate carbon intensity of electricity generation was relatively high. For similar reasons, the carbon intensities of electricity supply were lower than those of electricity generation in the NC and EC grids.

Based on the analysis of electricity supply and transmission loss, the associate carbon intensity of final electricity consumption was evaluated according to Equation (11), and the results are summarized in

Table 6. Associate carbon intensity of final electricity consumption (kgCO_2e/kWh).

Year	Power Grid
	NC	NEC	EC	CC	NWC	SC	Average
2005	1.061	1.111	0.854	0.742	0.787	0.752	0.883
2006	1.031	1.156	0.838	0.766	0.795	0.747	0.879
2007	1.007	1.073	0.817	0.746	0.794	0.734	0.857
2008	1.039	1.081	0.789	0.669	0.787	0.642	0.826
2009	1.003	1.054	0.771	0.640	0.779	0.647	0.805
2010	1.032	1.048	0.771	0.654	0.762	0.644	0.811
2011	1.065	1.087	0.777	0.666	0.789	0.671	0.832
2012	1.036	1.056	0.747	0.586	0.767	0.607	0.788
2013	0.953	0.966	0.761	0.607	0.737	0.579	0.761
2014	0.909	0.998	0.707	0.539	0.714	0.493	0.709
2015	0.891	0.950	0.681	0.502	0.694	0.438	0.680
2016	0.870	0.890	0.655	0.470	0.677	0.425	0.656
2017	0.889	0.914	0.658	0.461	0.670	0.452	0.662
2018	0.879	0.847	0.654	0.481	0.648	0.446	0.658
2019	0.857	0.843	0.638	0.476	0.641	0.432	0.644
2020	0.834	0.794	0.616	0.435	0.641	0.425	0.622

. The associate carbon intensity of final electricity consumption in SC grid was the lowest at 0.425 kgCO_2e/kWh in 2020, and the highest value at 0.879 kgCO_2e/kWh occurred in the NC grid. Descending trends were observed in the associate carbon intensities, especially after 2011. Moreover, the difference between the carbon intensities of electricity generation and use are compared in Figure 5. There was a fluctuation in the difference between the two emission factors, yet an overall descending trend was observed in the difference, which indicated that the transmission loss was better controlled in recent years.

4.2. LMDI Decomposition of the Associate Carbon Intensity of Electricity Use

Detailed results of LMDI decomposition for each regional power gird are demonstrated in Figure 6. As indicated in Figure 6, the assessment period was divided into three stages: (a) Stage 1 including a 5-year period from 2005 to 2010; (b) Stage 2, including a 5-year period from 2010 to 2015; and (c) Stage 3, including a 5-year period from 2015 to 2020. The characteristics of the five driving factors varied considering different regional grids and stages.

For the NC gird (Figure 6a), the decreases of the associate carbon intensity in the three stages were 0.029, 0.141, and 0.057 kgCO_2e/kWh, respectively. All driving factors except for the energy structure effect indicated positive influences on the decrease during the assessment period. In Stage 1, the change in the associate carbon intensity was not significant, and the energy intensity effect and clean production effect contributed equally. In Stage 2, the energy intensity effect dominated the change with a contribution of nearly 70%, and the associate carbon intensity experienced a rapid decrease. In Stage 3, the application of clean power generation was promoted, and the most influential factor became the clean production effect.

For the NEC grid (Figure 6b), the decreases of the associate carbon intensity in the three stages were 0.063, 0.098, and 0.156 kgCO_2e/kWh, respectively. The energy structure effect showed an inhibiting influence on the changes during the assessment period. Moreover, the supply structure effect was zero because the electricity demand was self-sufficient by the NEC grid itself. In Stages 1 and 2, both the energy intensity effect and clean production effect contributed positively to the reduction of associate carbon intensity, and their contribution were similar. In Stage 3, an approximately 16.5% decrease was realized, of which the influence of the clean production effect was more considerable than the energy intensity effect.

For the EC grid (Figure 6c), the decreases of the associate carbon intensity in the three stages were 0.083, 0.090, and 0.065 kgCO_2e/kWh, respectively. In Stage 1, the only considerable factor was the energy intensity effect. In Stage 2, while the energy structure effect was negligible, the other driving factors showed a positive impact on the reduction of associate carbon intensity. In Stage 3, the energy intensity effect and the clean production effect continued to play active roles, yet the energy structure effect and supply structure became inhibiting factors.

For the CC grid (Figure 6d), the decreases of the associate carbon intensity in the three stages were 0.088, 0.152, and 0.067 kgCO_2e/kWh, respectively. In all stages, the influence of the energy intensity effect was significant, which led to a 0.25 kgCO_2e/kWh reduction of associate carbon intensity, yet the energy structure effect slightly inhibited the decrease. In Stages 2 and 3, the clean production effect also became an important factor, with a total contribution of 0.107 kgCO_2e/kWh to the reduction of associate carbon intensity.

For the NWC grid (Figure 6e), the decreases of the associate carbon intensity in the three stages were 0.025, 0.068, and 0.053 kgCO_2e/kWh, respectively. The energy intensity effect played a positive role in Stages 1 and 2, but its influence in Stage 3 was very limited. The clean production effect was an influential factor only in Stage 3. For the same reason as the NEC grid, the impact of the supply structure effect was zero. Overall, although the associate carbon intensity of the NWC gird was relatively low in 2005, the relevant decrease trend was not as considerable as other regional grids.

For the SC grid (Figure 6f), the decreases of the associate carbon intensity in the three stages were 0.108, 0.206, and 0.013 kgCO_2e/kWh, respectively. In Stages 1 and 2, the energy intensity effect and clean production effect showed strong influences, they contributed 0.121 and 0.209 kgCO_2e/kWh reduction to the reduction of associate carbon intensity, respectively. The associate carbon intensity of electricity use has already reached a relatively low value of 0.438 kgCO_2e/kWh in 2015, and in Stage 3, the change in the associate carbon intensity is not considerable.

Overall, the results of LMDI decomposition indicated that the associate carbon intensities of electricity use decreased by 0.227, 0.317, 0.237, 0.307, 0.146, and 0.327 kgCO_2e/kWh, respectively, in the NC, NEC, EC, CC, NWC, and SC grids during the period from 2005 to 2020. As shown in

Table 7. Driving factors in the changes of associate carbon intensity of electricity use (kgCO_2e/kWh).

Power Grid	Energy Structure Effect	Energy Intensity Effect	Clean Production Effect	Supply Structure Effect	Power Loss Effect	Total Change
NC	0.059	−0.130	−0.133	−0.012	−0.011	−0.227
NEC	0.060	−0.174	−0.206	0.000	0.002	−0.317
EC	0.018	−0.148	−0.082	−0.003	−0.021	−0.237
CC	0.041	−0.250	−0.105	0.011	−0.005	−0.307
NWC	0.007	−0.095	−0.062	0.000	0.003	−0.146
SC	0.022	−0.122	−0.218	−0.001	−0.008	−0.327

, the decreases in associate carbon intensity can mainly be attributed to the energy intensity effect and clean production effect. The influence of the supply structure effect and power loss effect varied in different regional grids, yet they both showed very limited impacts on the results. The energy structure effect was an inhibiting factor in the decrease in associate carbon intensity, especially in the NC and NEC grids. the reasons for this result can be as follows: (1) the share of coal in all fossil fuels for thermal power generation slightly increased in some regions, and (2) in other regions, coal-fired electricity generation was reduced and the share of blast furnace gas and converter gas increased. However, their net calorific values are relatively low, and more carbon dioxide will be emitted when provided with the same amount of thermal energy for power generation. Despite the inhibiting factors, the actual decreases in the associate carbon intensities were 0.286, 0.380, 0.255, 0.360, 0.157, and 0.349 kgCO_2e/kWh, respectively, in the NC, NEC, EC, CC, NWC, and SC grids, of which the contribution of the energy intensity effect and clean production effect was more than 95%. It should be noted that the emission factors of fossil fuels were considered time-invariant, and the relevant influence is not considered in this study.

5. Conclusions

This study proposed a comprehensive method to assess the carbon intensities of electricity generation, supply, and use in the six regional power grids of China. The emission factors were evaluated covering the period from 2005 to 2020, and an LMDI decomposition model was developed to investigate the influence of driving factors including energy structure effect, energy intensity effect, clean production effect, supply structure effect, and power loss effect. The main findings of this study are as follows:

(1)

The associate carbon intensities of electricity use were 0.834, 0.794, 0.616, 0.435, 0.641, and 0.425 kgCO_2e/kWh, respectively, in the NC, NEC, EC, CC, NWC, and SC grids in 2020. The national average carbon intensities of electricity generation, supply, and use were 0.599, 0.599, and 0.622 kgCO_2e/kWh, respectively.

(2)

The carbon intensities of electricity generation, supply, and use experienced an overall descending trend during the assessment period. Total decreases of the associate carbon intensity of electricity use reached 0.227, 0.317, 0.237, 0.307, 0.146, and 0.327 kgCO_2e/kWh, respectively, in the NC, NEC, EC, CC, NWC, and SC grids. A national average 30% decrease was observed by comparing the results in 2020 and 2005.

(3)

Although the driving factors and trends of the changes in carbon intensity varied among different regional power grids in China, the energy intensity effect and clean production effect were determined as the most influential factors for the decrease in the carbon intensity, which can be mainly attributed to the efforts in upgrading production technology and equipment and promoting clean energy utilization.

The proposed method provided a helpful approach to evaluating the carbon intensity of electricity in regional power grids. The results of carbon intensities for electricity generation, supply, and use during the assessment period can be widely applied to assess the carbon emissions relevant to electricity use in China. It is worth noting that the proposed method was developed based on statistical data. Considering the changing production ways and final demands of electricity, the carbon intensity can be evaluated based on the latest data to ensure applicability and timeliness.