Cooking/hot water in households usually relies on coal stoves or gas stoves, RC, WH, RH and MO. The calculation of its end-use energy consumption is complex due to the existence of a mixed supply of multiple energy sources. All natural gas and LPG consumed by households are considered to be used for cooking and hot water. And the energy end-use for cooking activities needs to be converted by the combustion efficiency of the gas stove.

In urban areas where natural gas is not fully penetrated, coal is assumed to be the fuel source for household cooking. This can be calculated based on the penetration of urban gas and the combustion efficiency of coal stoves. The energy consumption of other appliances, RC, WH, RH and MO are based on the penetration of the appliance, the average annual energy consumption and the number of households. The energy end-use for cooking/hot water in urban households is calculated by the following formula (2), formula (3), formula (4), formula (5), formula (6). E t C H = E t , C C H + E t , L C H + E t , G C H + E t , E C H (2) E t , L C H = E t , L ∙ η L C H (3) E t , G C H = E t , G ∙ η G C H (4) E t , C C H = E t , G C H + E t , L C H ∙ 100 − r t r t ∙ η c k (5) E t , E C H = e t , W H ∙ n t , W H + e t , R C ∙ n t , R C + e t , M O ∙ n t , M O + e t , R H ∙ n t , R H 100 ∙ m t (6)

Where, E t , C C H , E t , L C H , E t , G C H , E t , E C H denote the Coal, LPG, Natural Gas and Electricity consumption of Cooking/hot water in year t (PJ), η C C H , η G C H , η L C H denote the combustion efficiency of coal, LPG, Natural Gas for Cooking/hot water (%), r t denotes the penetration rate of natural gas in t years (%). e t , W H , e t , R C , e t , M O , e t , R H denote the average annual energy consumption of WH, RC, MO and RH in year t (PJ/unit); n t , W H , n t , R C , n t , M O , n t , R H denote the penetration rate of WH, RC, MO and RH (unit/100 households), m t denote the number of urban households in year t (households).

The heating approaches in Chinese urban households mainly include centralized thermal heating, decentralized heating using coal or gas, and air conditioner heating. Centralized heating is an important heating method in northern China and can be estimated based on the total heat consumption and the combustion efficiency of the equipment. Gas heating will be ignored in this study due to its low penetration rate. Decentralized heating mainly considers coal consumption, which is estimated by the salvage value of coal consumed in the kitchen/hot water. Air conditioner heating is calculated based on the penetration rate of air conditioners, the average annual energy consumption, the number of households and residence area. The energy consumption of the Heating can be calculated by formula (7). E t H = E t , C H + E t , H H + E t , E H = E t , C − E t , C C H + E t , H + e t , A C ∙ n t , A C 100 ∙ m t ∙ S t (7)

Where, E t , C H , E t , H H , E t , E H denote the Coal, Heat and Electricity consumption of Heating in year t, (PJ). e t , A C denote the per unit area heating energy consumption of AC in year t, (PJ/m²). n t , A C denote the penetration rate of AC in t years, (unit/100 households).

Household cooling, lighting and power consume a large amount of electricity. Among them, cooling energy consumption comes from the use of AC and EF in hot weather, and lighting energy consumption can be deduced from the penetration of lamps, per unit area of energy consumption and residential area. Power mainly contains the energy consumption of four types of common household end-use devices: TV, R, WM, and C. The energy consumption of household appliances is also derived from the average annual energy consumption, the penetration rate of the appliances, and the number of households.

The calculation of cooling, lighting and power energy consumption in urban households is shown in formulas (8), formulas (9), formulas (10). E t C = e t , A C ∙ n t , A C + e t , E F ∙ n t , E F 100 ∙ m t ∙ S t (8)

Where, e t , A C , e t , E F denotes the per unit area cooling energy consumption of AC, FE in year t, (PJ/m²). n t , A C , n t , E F denotes the penetration rate of AC, EF in t years, (unit/100 households). E t L = e t , F ∙ n t , F + e t , S ∙ n t , S ∙ m t ∙ S t (9)

Where, e t , F , e t , S denote the per unit area energy consumption of fluorescent lamps and energy-saving lamps in year t, (PJ/m²). n t , F , n t , S denote the penetration rate of fluorescent lamps, and energy-saving lamps in year t, (unit/100 households).

S t denotes the residence area in year t. E t P = e t , T V ∙ n t , T V + e t , C ∙ n t , C + e t , W M ∙ n t , W M + e t , R ∙ n t , R 100 ∙ m t (10)

Where, e t , T V , e t , C , e t , W M , e t , R denote the average annual energy consumption of TVs, C, WM and R in year t, (PJ/unit).

n t , T V , n t , C , n t , W M , n t , R denote the penetration rate of TVs, C, WM and R in year t, (unit/100 households).

In terms of energy consumption for regional heating, the data in the statistical yearbook only reflects the central heating of urban households in each province. It is not possible to confirm the usage habits of household appliances during heating and cooling. Considering the differences in climate, we introduced the number of degree days as a condition for calculating the electric energy consumption for urban household heating and cooling in each province. Degree days is a general indicator to reflect climate change human comfort, mainly divided into Heating Degree Days (HDD_18,°C day) and Cooling Degree Days (CDD_26, °C day). HDD₁₈ and CDD₂₆ can objectively evaluate the energy demand for heating and cooling in urban households (Bhatnagar et al., 2018) (K. Lee et al., 2014). The formula is as follows (11) and (12).

The end-use energy consumption results for centralized and decentralized heating, cooking/hot water, lighting, and power for urban households in each province we can still obtain from formulas (2), formulas (3), formulas (4), formulas (5), formulas (6), formulas (9), formulas (10). H D D 18 = ∑ k = 1 n T B − T , T B = 18 ℃ (11) C D D 26 = ∑ k = 1 n T B − T , T B = 26 ℃ (12)

Where, T B denote the threshold temperature, T denotes the average temperature.

The formula for calculating the electric energy consumption of heating and cooling in urban households in each province is as follows (13) (14). E i H = e A C H ∙ n i , A C ∙ m i ∙ S i ∙ H D D H i / H D D H C (13) E i C = ( e A C C ∙ n i , A C + e E F C ∙ n i , E F ) ∙ m i ∙ S i ∙ C D D C i / C D D C C (14)

Where, E i H , E i C denotes electric energy consumption of heating and cooling in each province, (PJ). e A C H , e A C , C e E F C denote the per unit area energy consumption of AC, EF in year t, (PJ/m²). n i , A C , n i , E F denote the penetration rate of AC and EF in urban households of province. m i , the number of households; S i denote residential area in urban household of province. H D D H i , C D D C i denote the average number of heating/cooling degree days in each province, (°C day). H D D H C , C D D C C denote the average number of heating/cooling degree days in national, (°C day).

In this paper, the Statistical Yearbook Data is an important support for conducting the study. The national urban household energy consumption data for 2010 - 2019 are taken from the national energy balance sheet in the China Energy Statistical Yearbook, it includes national scale and provincial scale. Moreover, the data on urban population, the number of households, the penetration rate of household appliances, per capita income and GDP, and residential area involved in the promotion of the study were obtained from the China Statistical Yearbook, as well as from the statistical yearbooks of 30 provinces (excluding Tibet, Hong Kong, Macao, and Taiwan).

Other data relevant to the study, including data related to energy fuel combustion efficiency, power of household appliances, and data related to climatic conditions, were obtained mainly from other energy statistical reports, social publications, and government public information websites. Some important data and sources of access are shown in Table 2.

In estimating the electricity consumption of household appliances, the average annual electricity consumption (AAEC) is a very important indicator, but it is not directly available. In the (Murata et al., 2008) study, a method is proposed that can calculate this indicator using the power of the appliance as well as the annual usage time. For other related data, such as per unit area energy consumption of lamps, per unit area energy consumption of air conditioners, etc., the main reference is (Jiang and Wu, 2010). The data sources are presented in Table 2. Table 3 shows the average annual electricity consumption of all household appliances, as well as the per unit area electricity consumption.

In Supplementary Table A1, the abbreviations for the names of the provinces are detailed, as well as the results of HDD₁₈, CDD₂₆ for urban areas in province for 2019, calculated based on formula (13)(14). The 30 provinces were divided into 7 regions based on geographical distribution to explore the impact of climate and economic development on household energy consumption.

The decomposition results of cooking/hot water energy consumption of urban households in different regions are detailed in Figure 5. The column chart indicates the variability of total end-use energy and the structure of energy use, while the point distribution indicates the variability of average household energy use.

In terms of total regional consumption, Central has the highest household end-use energy, accounting for about 22% of the national total, followed by Southwest and North with 17% and 16%, respectively, and Northeast with the lowest share of total energy use at 6%. In terms of provincial performance, GD has the highest total energy use, and almost 90% of Southeast’s cooking/hot water energy use is from GD. The next is SC, with a total cooking/hot water energy use of 216 PJ, accounting for about 9% of the national total. In the second tier, HEN, SD, JS, have a total cooking/hot water energy use of almost all over 150 PJ. This is closely related to the highly dense population in urban areas. In contrast, HN and NX have very little energy consumption.

Areas with higher average household energy consumption level are typically concentrated in provinces characterized by relatively slower urbanization development and smaller household members. This phenomenon is most pronounced in the provinces of the Northwest region. QH exhibits a remarkably high average household cooking consumption of 23.5 GJ/household, XJ and SAX follow closely behind. SC is the only province that ranks high in both total energy consumption and average household cooking consumption. Its average household cooking consumption is 13.24 GJ/household, ranking second only to QH and XJ.

The energy consumption structure for household cooking and hot water is complex. With the popularization of urban gas appliances and the improvement of infrastructure, natural gas and LPG have become the most important energy sources for household cooking activities, accounting for over 80% of the total energy consumption. Electricity follows next, representing 13% of the energy mix. In North and Yangtze River Delta, the household cooking energy structure consists of natural gas and electricity, and natural gas consumes nearly one-third of the national total. LPG is the primary energy supply for cooking activities in the Southeast region, and its total consumption is the highest of all regions. The Southwest region consumes the highest amount of coal for cooking in the country, accounting for approximately 36% of the national total.

The “high in the north and low in the south” is an important regional difference in household heating energy consumption in urban areas of China. As shown in Figure 6, Northwest, Northeast and North are the regions with high concentration of heating energy consumption. Due to the high geographical latitude, the winter temperature is low and lasts for a long time, and the demand for heating is high in order to maintain a comfortable indoor living environment. Among them, the heating energy consumption of HLJ, SD, IM and LN far exceeds that of other provinces. The heating energy consumption of the four provinces accounts for almost half of the national total. Among them, HLJ has the highest total heating energy consumption of 338.16 PJ. IM has the highest average household energy consumption at 50.36 GJ/household. Compared to other high-latitude provinces, SD has a relatively suitable climate and temperature environment in winter, but a large number of urban households leads to a high demand for heating.

In terms of the energy structure of heating, urban centralized thermal heating is a type of heating unique to northern China, accounting for more than 70% of energy use. Central areas mainly have decentralized heating using coal as fuel, such as heating stoves. The Yangtze River Delta, Southwest and Southeast have a subtropical monsoon climate with low demand for winter heating and rely mainly on air conditioning, with over 70% of air conditioning heating coming from these regions.

From Figure 7, it is evident thathousehold cooling energy consumption is mainly concentrated in the southern regions and areas characterized by a high degree of economic development. This concentration is attributable to several factors.

In regions with relatively cooler summer temperatures, households exhibit a lower demand for cooling, resulting in significantly reduced total cooling energy consumption in the northeastern and northwestern regions compared to other areas. Conversely, GD positioned at a lower latitude and experiencing hotter and more extended summers, demonstrate notably higher cooling energy consumption compared to other provinces. It is worth noting that, unlike household heating, the central provinces also show a higher demand for cooling energy consumption.

Regarding average household energy consumption levels, the results for the northeast and northwest regions closely mirror those observed for total energy consumption. In the more developed provinces of ZJ, SH, JS and BJ, the average household energy consumption level surpass those in regions at the same latitude. In addition to climatic factors, this disparity may be attributed to a superior material infrastructure and a greater emphasis on achieving a higher standard of comfortable living, potentially contributing to the elevated cooling energy consumption in these areas.

Figure 8 shows the differences in energy consumption for power and lighting within urban households. Concerning the regional distribution of total power energy consumption, the provinces are ranked as follows: Central, Yangtze River Delta, North, Southeast, Southeast, Southwest and Northeast, in descending order. The Yangtze River Delta region, being China’s most economically developed and robust region, significantly ahead of other regions in terms of average household energy consumption. Aside from this, the average household energy consumption levels in other provinces are relatively uniform. The balanced development of urbanization is gradually diminishing the disparities in economic income and consumption levels among regions.

The lighting energy consumption of urban households is directly related to the residential area. SH and BJ, as China’s first-tier cities, have a high degree of urbanization leading to a denser population distribution and the per capita residential area much lower than the national average residential level. This factor primarily contributes to their lower average household lighting energy consumption levels.

Cooking/hot water usage constitute essential components of daily energy consumption activities within Chinese households, with the rich culinary culture encouraging households to dedicate relatively more time to cooking (Zheng et al., 2014). In the study conducted by (Lee and Song, 2022b), the household size of and household income are identified as pivotal factors influcing cooking energy consumption. However, this study does not show the determinants related to the spatial differences affecting household cooking energy consumption. This may be due to the inconsistency in the division of end-use energy consumption and the study regions. Additionally, in the study by (Lee and Song, 2022b), only the energy consumption of cooking appliance was considered, and data on the consumption of fossil fuels such as coal and natural gas were not included.

Cooking/hot water is indispensable as the basis of productive household life. In terms of provinces, cooking/hot water energy consumption is significantly higher in SC and GD, which is strongly related to differences in cooking habits and behaviors of residents. This view was also confirmed in a study by (Hager and Morawicki, 2013).

In the analysis of influencing factors, HDD₁₈, CDD₂₆ and GPR show trends that affect cooking/hot water consumption. This indicates that climatic conditions influence regional variations in hot water demand (Meireles et al., 2022). The popularity of gas can make household cooking/hot water activities more convenient and promote energy consumption (NDRC, 2022). PSH exhibits a positive correlation with PECH, indicating that in cities with a higher proportion of single-person households, household cooking activities are more dispersed, resulting in increased energy consumption intensity.

Similar to cooking/hot water, the household characteristics is an important factor in power consumption. PSH shows a positive correlation with the level of household energy consumption, while an increase in HS demonstrates a significant negative correlation with per capita energy consumption. Single-person households often imply higher per capita energy consumption. This also suggests that the reduction in household size and the increase in the number of single-person households have led to the growth in total households, thereby promoting the increase in energy consumption.

The reasons for changes in family characteristics are multiple. Uneven economic development and population mobility are also significant factors contributing to this phenomenon. In regions such as ZJ, SH, GD, rapid economic development has attracted migrant workers. Single-person households account for nearly 30% of the total households in these areas, with household sizes lower than the national average. In contrast, in the northeastern regions, the outflow of a large number of employed individuals has exacerbated population aging, leading to an increase in single elderly households living alone (Liu et al., 2020). This is an important factor causing household downsizing. With social changes and shifts in family values, traditional notions of “four generations under one roof” and “three generations living together” have been abandoned. Additionally, the rise of individualistic thinking has contributed to smaller average household sizes. According to the data from the seventh National Population Census (NBS, 2021), the average household size in China dropped to 2.62 people, breaking the pattern of the “three-person household.”

Economic development and the level of urban development also have a significant impact on household power consumption (Huang and Matsumoto, 2021). PCI and PDI contribute to the growth in energy consumption with correlation coefficients of 0.798 and 0.81, respectively. This phenomenon is more pronounced in highly urbanized areas such as BJ and SH. Areas with higher levels of urban development tend to represent higher household incomes, and higher income elasticity promotes an increase in the types, penetration, and utilization of household appliances (Lei et al., 2022), which also changes the lifestyle and consumption habits of residents, and resulting in higher consumption demand. This also reflects how increasing the share of energy-efficient appliances is an important initiative to address environmental impacts in the era of household electrification (Baldini et al., 2018).

In the correlation coefficient analysis, a significant positive correlation was found between HDD₁₈ and PEH. Climate variation plays a crucial role in explaining the disparity in energy consumption for heating in urban households. Centralized thermal heating systems are prevalent in northern Chinese cities due to low winter temperatures and prolonged winter seasons, aimed at ensuring a comfortable indoor living environment (M. Fan et al., 2020), However, this feature also leads to high energy consumption levels.

Household cooling energy consumption exhibits a significant positive correlation with CDD₂₆ and a negative correlation with HDD₁₈. In general, regions with higher CDD₂₆ tend to exhibit increased demand for household cooling energy consumption. Apart from climate-related factors, urban development and household income levels are also important factors contributing to disparities in cooling energy consumption (Al-Hinti and Al-Sallami, 2017). The findings reveal that both PGDP and PDI display a significant positive correlation with PEC, with correlation coefficients of 0.728 and 0.676, respectively. Cooling energy consumption is primarily calculated based on electricity usage for air conditioning during the summer season, and the ownership rate of air conditioning in households depends on the level of urban development and household income in different regions. Typically, economically advanced areas characterized by high levels of urbanization, there is also higher purchasing power for air conditioning, resulting in higher cooling energy consumption in households. In terms of household characteristics, PSH also shows a positive correlation with cooling energy consumption, indicating that regions with a higher proportion of single-person households tend to experience higher per capita cooling energy consumption. This also explains the concentration of cooling energy consumption in the Yangtze River Delta region.