1. Introduction

2. Research Process

2.2. Construction of Evaluation Models

The construction of the evaluation model is divided into six stages, as follows:

2.2.1. Indicator Factor Sorting

Given the diverse factors influencing rural building energy consumption, which necessitated a comprehensive analysis of multiple indicators, an evaluation index system framework was established through literature analysis.This framework comprised four levels: the target level, the criterion level, the sub-criterion level, and the factor level. An innovative comprehensive evaluation metric, Low Carbon Intensity (LCI), is proposed to be utilized for quantifying the level of energy consumption in buildings. Consequently, the target level was defined as LCI, and the criterion level was summarized into three aspects: Cleanliness of Energy (C), Energy Efficiency of Buildings (E), and Self-Discipline of Residents (S), based on the energy’s inherent attributes, the spatial carrier of energy usage, and the energy implementer. This led to the CES model being defined. The framework of the index system is illustrated in Figure 2.

The preliminary evaluation index system was derived by the research team through two rounds of brainstorming sessions for selecting indicator factors. Subsequently, the indicator factors underwent optimization through two rounds of Delphi methodology. In the first round, focused primarily on open-ended consultations, the experts were presented with the preliminarily drafted indicators and their corresponding explanations. A total of 12 experts from relevant fields were invited to provide feedback on the evaluation system. The recognition rate for the criterion level and sub-criterion level of the evaluation index system achieved 100% among the experts.

The second round of expert consultations utilized the Likert scale method to assign scores to each indicator. Initially, the Kendall’s W coefficient was employed to evaluate the degree of coordination among experts’ judgments across all indicators. Subsequently, the weighted average score, weighted standard deviation, and weighted coefficient of variation for each indicator were calculated, serving as the basis for indicator screening. The calculation results consistently indicated a high degree of coordination and consensus among experts’ judgments, with the consistency test being successfully passed. Through these two rounds of expert consultation methods, a final evaluation index system comprising 8 sub-criterion levels and 26 factor levels was established, with a distinction made between objective and subjective indicators for each. The evaluation index system is presented in the corresponding columns of Table 2.

2.2.2. Construction of a Model for the Mutual Influence Relationship between Indicators

The ANP (Analytic Network Process) methodology primarily categorizes system elements into two hierarchical levels: the control layer and the network layer. The control layer encompasses indicators of the target level and the criterion level, whereas the network layer comprises indicators of the sub-criterion level and the factor level. The interdependent relationships among these indicators were determined through expert consultations and questionnaire surveys. A total of 12 experts were consulted via questionnaires, and when the number of experts who perceived a correlation between two indicators was equal to or greater than one, it was determined that there existed an influence relationship between those two indicators; otherwise, no influence relationship was assumed. Based on the dependencies and feedback relationships among the indicators, a network structure model diagram (Figure 3) was constructed using the Super Decisions (yaanp) software.

2.2.3. Construction of Judgment Matrix

Based on the indicator network hierarchy diagram, judgment matrices were formulated to evaluate the superiority and inferiority of indicators at both the control layer and the network layer. With considerations given to feasibility and representativeness, six experts were selected from the previously mentioned twelve, who were highly relevant to the field of human settlements environment, to ensure their corresponding professional competence and credibility. These experts were then tasked with conducting pairwise comparisons of factor importance using Saaty’s 1-9 scale. Among them, four experts self-assessed their familiarity with the indicators as 1 (fully familiar), while the remaining two assessed their familiarity as 0.75 (moderately familiar). Subsequently, the judgments on superiority/inferiority were weighted and averaged according to Formula (2.1), where Q_i represents the weighted average superiority/inferiority value, Cs_i denotes the expert’s self-assessed familiarity with the study, and n refers to the total number of experts consulted. This process was completed in the past.

$$Q_i={\displaystyle \frac{{Cs}_i}{{\sum }_{i=1}^{i=n}{Cs}_i}}$$

(2.1)

The criteria-level indicators, being independent of each other, had their judgment matrixes established based on a direct superiority/inferiority degree. In contrast, the interplay among the sub-criteria levels necessitated the adoption of a combined approach of direct and indirect superiority/inferiority degrees for establishing their judgment matrixes. As a result, 8 cluster judgment matrixes and 87 node judgment matrixes were ultimately derived. Table 1 presents solely the judgment matrix for the sub-criteria level under the energy supply and demand criteria level, with additional relationship tables provided in the appendix.

2.2.4. Consistency Check of Judgment Matrix

The logical coherence of the decision-makers’ inputs within the judgment matrix was validated through the consistency check. The degree of consistency among the pairwise comparison matrices, known as the consistency ratio (CR), was utilized as a form of feedback for the experts to have their judgment matrixes revised. The derivation of CR involved the calculation of the maximum eigenvalue (λ_max), the consistency index (CI), and the random consistency index (RI). λ_max was computed by utilizing the linear algebra library functions in programming software, while the formulas for CI and CR, as presented in Equations (2.2) and (2.3) respectively, where n represented the order of the matrix, were applied. When CR ≤ 0.1, it was determined that the judgment matrix had passed the consistency check. According to the calculations that were performed, the consistency ratio for the judgment matrix presented in Table 1 was found to be 0.0276, which was less than 0.1 (similarly, all judgment matrixes had their consistency ratios determined to be less than 0.1).

$$\mathrm{C}\mathrm{I}={\displaystyle \frac{{\lambda }_{max}-n}{n-1}}$$

(2.2)

$$\mathrm{C}\mathrm{R}={\displaystyle \frac{CI}{RI}}$$

(2.3)

Table 1. Energy Supply and Demand cluster judgement matrix under the criterion layer.

Table 1. Energy Supply and Demand cluster judgement matrix under the criterion layer.

Energy Supply and Demand	Architectural Design	Envelope Structure	Energy Use	Energy Sustainable	Awareness Management	Behavior Management
Architectural Design	1	2	1/2	1/2	3	2
Envelope Structure	1/2	1	1/2	1/2	2	2
Energy Use	2	2	1	1	2	2
Energy Sustainable	2	2	1	1	3	2
Awareness Management	1/3	1/2	1/2	1/3	1	1
Behavior Management	1/2	1/2	1/2	1/2	1	1
consistency test: λmax: 6.174113;CR=0.0276<0.1

2.2.5. Calculation of Indicator Weights

After the expert scoring results for the judgment matrices were obtained and their consistency was verified, the data for the matrices was entered into yannp program. The input was facilitated through a “questionnaire format,” which captured the individual experts’ scoring results for the judgment matrices. Subsequently, the judgment outcomes of the superiority and inferiority for 8 cluster judgment matrices and 87 node judgment matrices were derived. By selecting the appropriate options within YAAHP, the unweighted supermatrix, weighted supermatrix, and limit supermatrix were calculated. The final determined weight results were then presented in the corresponding weight columns of Table 2.

Table 2. Comprehensive evaluation index system of energy consumption of rural residential buildings.

Table 2. Comprehensive evaluation index system of energy consumption of rural residential buildings.

Criterion layer	weight	Sub-canonical layer	weight	Factor layer	weight	Normalized values
						q∈[80,100]	q∈[60,80)	q∈[40,60)	q∈[20,40)	q<20
Energy Cleanliness (C)	0.559	Energy Supply and Demand C1	0.071	Clean Energy Demand Satisfaction C11 (subjective)	0.041	Satisfaction with clean energy demand is high	Satisfaction with clean energy demand is relatively high	Satisfaction with clean energy demand is average	Satisfaction with clean energy demand is relatively low	Satisfaction with clean energy demand is low
				Energy Price Stability C12 (subjective)	0.016	Energy prices are stable	Energy prices are relatively stable	Energy prices vary in general	Energy prices are relatively highly volatile	Energy prices are highly volatile
				Energy Subsidies and Satisfaction C13 (subjective)	0.013	Residents are highly satisfied with energy subsidies	Residents are more satisfied with energy subsidies	Residents’ satisfaction with energy subsidies is average	Residents’ satisfaction with energy subsidies is relatively low	Residents’ satisfaction with energy subsidies is low
		Energy Use C2	0.285	Electricity Consumption per capita C21	0.064	Electricity consumption Q∈ [646,727]	Electricity consumption Q∈ (727,808]	Electricity consumption Q∈ (808,889]	Electricity consumption Q∈ (889,970]	Electricity consumption Q∈ (970,1051]
				Gas Consumption per capita C22	0.041	Gas consumptionG∈[72,81]	Gas consumption G∈ (81,90]	Gas consumption G∈ (90,99]	Gas consumption G∈ (99,108]	Gas consumption G∈ (108,117]
				Proportion of Energy Use from Commodities C23	0.074	commodity energy use/total energy ×100per cent
				Percentage of Clean Energy Use C24	0.105	Total clean energy usage/total energy usage×100per cent
		Energy Sustainable C3	0.204	Biomass Energy Utilisation C31	0.102	It meets the requirements of biogas digester on-site use and has a high frequency of use	It meets the requirements of biogas digester on-site use and the frequency of use is average	It meets the requirements of biogas digester on-site use and is used less frequently	Does not meet the requirements for use or does not use modern biomass energy	Conventional biomass energy is used
				Solar Energy Systems C32	0.102	30 points for solar thermal equipment, 30 points for solar photovoltaic equipment, and 20 points for setting up a sunshine room, and the cumulative score is calculated
Building Energy Efficiency (E)	0.297	Architectural Design E1	0.098	Building Site Selection E11	0.043	According to the definition of the rationality of building site selection in relevant specifications, five main conditions are established to determine the evaluation criteria for building site selection based on the number of buildings
						5 conditions are met	4 conditions are met	3 conditions are met	2 conditions are met	0-1 conditions are met
				Building Orientation E12	0.016	The growth rate of energy consumption is 0per cent-3per cent, corresponding to the direction	The growth rate of energy consumption is 3per cent-6per cent, corresponding to the direction	The growth rate of energy consumption is 6per cent-9per cent, corresponding to the direction	The energy consumption growth rate of 9per cent-12per cent corresponds to the direction	The energy consumption growth rate is greater than 12per cent, corresponding to the direction
				Architectural Space Layout E13	0.025	Floor height 2.7≤h≤3.0	Floor height 3.0<h≤3.3	loor height 3.0<h≤3.3	Floor height 3.6<h≤3.9	Floor height 3.9<h≤4.2
				Building form Factor E14	0.013	0.35≤Tx≤0.45	0.45<Tx≤0.55	0.55<Tx≤0.75	0.75<Tx≤0.95	0.95<Tx≤1.2
		Envelope Structure E2	0.131	Thermal Performance of Exterior Walls E21	0.041	0.6≤Km≤1.0	1.0<Km≤1.4	1.4<Km≤1.8	1.8<Km≤2.2	2.2<Km≤2.6
				Thermal Performance of Exterior Windows E22	0.041	1.4≤Kw≤2.4	2.4<Kw≤3.4	3.4<Kw≤4.4	4.4<Kw≤5.4	5.4<Kw≤6.4
				Thermal Performance of Roofing E23	0.022	0.8≤Kr≤1.4	1.4<Kr≤2.0	2.0<Kr≤2.6	2.6<Kr≤3.2	3.2<Kr≤4.0
				External Shading Measures E24	0.027	2.0≤L≤2.7	1.5<L≤2.0	1.0<L≤1.5	0.5<L≤1.0	0<L≤0.5
		Building Material E3	0.068	Building Materials Localization Ratio E31	0.026	City-wide use of building materials/total use of building materials×100per cent
				Utilization Rate of Environmentally Friendly Construction Materials E32	0.042	Green building materials used/total building materials used×100per cent
Resident Self-discipline (S)	0.144	Awareness Management S1	0.042	Widespread Awareness of Low Carbon S11 (subjective)	0.016	Residents have a high level of low-carbon knowledge	Residents have a relatively high level of low-carbon knowledge	Residents’ low-carbon knowledge is average	Residents’ understanding of low-carbon knowledge is relatively low	Residents’ low-carbon knowledge is low
				Acceptance of Low-Carbon Living S12 (subjective)	0.011	The main low-carbon lifestyles are green consumption, food conservation, residential energy-saving renovation, energy-saving household appliances, garbage classification, and clean travel
						Meet 5-6 items	Meet 4 items	Meet 3 items	Meet 2 items	Meet 0-1 items
				Responsiveness to Low-carbon Construction S13 (subjective)	0.015	The village residents are supportive of infrastructure construction	Residents are in favour of the development of rural infrastructure and hardware	Residents generally support the construction of rural infrastructure and hardware	The construction of rural infrastructure is less supported by residents	Residents do not support the construction of rural infrastructure
		Behavior Management S2	0.101	Proportion of Equipment Designed to Save Energy S21	0.036	Number of energy-saving devices in the dwelling/Total number of devices in the dwelling×100per cent
				Implementation Rate of Energy-saving Measures S22	0.037	The number of energy-saving behaviors achieved by residents/10×100per cent
				Waste Recycling S23 (subjective)	0.015	Utilise household waste to its full potential	A significant proportion of household waste is utilised.	Household waste is partly utilised	A small quantity of domestic waste is utilised	Household waste is not utilised
				Indoor Air Quality Discipline S24	0.013	The cumulative score is calculated by assigning 30 points for indoor planting of green plants, 30 points for indoor air purifiers, and 20 points for window ventilation.

2.2.6. Index Classification Criteria and Determination of LCI

The scoring of objective indicators was conducted using the linear interpolation method based on national or local standards, prevailing regulations, and statistical yearbooks. Data sources encompassed field measurements and observations, while questionnaires were utilized to gather information and document the subjective perceptions of respondents. Subsequently, the subjective indicators were quantified through the application of fuzzy mathematics theory. The standardized values corresponding to the scoring criteria for specific indicators are presented in Table 2.

LCI was used to measure energy consumption levels and has a certain functional relationship with the indicators of the three criteria layers from a composite perspective:

$$LCI=F\left(\mathrm{C},\mathrm{E},\mathrm{S}\right)$$

(2.4)

The correlation between the three criterion levels of C, E, and S and the sub-criterion levels was described by Equation (2.1). Formula (2.1) can be written as $LCI=F\left(f_C\right(C_i),f_E(E_i),f_S(S_i\left)\right)$, where Ci, Ei, and Si represent the indicators of the subcriteria layer, and i represents the number of indicators of the subcriteria layer. The process of pushing secondary functions is as follows:

$$C=f_C\left(C_i\right)=\sum _{i=1}^{i=3}{w_{Ci}C}_i=\sum _{i=1,j=1}^{i=3,j=m}{w}_{ij}{C}_{ij}$$

(2.5)

$$E=f_E\left(E_i\right)=\sum _{i=1}^{i=3}{w_{Ei}E}_i=\sum _{i=1,j=1}^{i=3,j=n}{w}_{ij}{E}_{ij}$$

(2.6)

$$S=f_S\left(S_i\right)=\sum _{i=1}^{i=3}{w_{Si}S}_i=\sum _{i=1,j=1}^{i=3,j=l}{w}_{ij}{S}_{ij}$$

(2.7)

The weight coefficients of each indicator are represented by “w”. “j” represents the number of factor levels, while “m”, “n”, and “l” represent the number of indicators in the subcriteria level. Formula (2.8) shows the functional relationship between LCI and the criterion layer, where 0.559, 0.297, and 0.144 are the indicator weight coefficients of the criterion layer, obtained from the weight calculation method described earlier.

$$LCI=0.559C+0.297E+0.144S$$

(2.8)

3. Results

3.2. Building Energy Efficiency (E) Sub-Svaluation Results

The Building Energy Efficiency (E) index includes multiple factors. Differences were present in the overall LCI scores, with each factor having been assigned a unique score. Figure 9 showed the LCI scores for each factor, with the scores exhibiting marked diversity, spanning from 56.24 to 75.97, and averaging 65.78. As had been analyzed earlier, the score of the C24 indicator was notably high. Notably, the LCI score of Thermal Performance of Exterior Walls (E21) index was particularly singled out, having been ranked last among all factors with a score of 56.24, significantly falling below the 14.5% average level of the overall score when considered as a percentage. This low scoring not only underscored the shortcomings of external wall thermal performance under the prevailing evaluation framework but also pointed to the obstacles it confronted in terms of energy conservation, emission reduction, and enhancing building energy efficiency.

Figure 9. LCI scores for each factor.

Figure 9. LCI scores for each factor.

The exterior walls of rural buildings surrounding Chengdu encompassed clay solid brick walls, sintered hollow brick walls, sintered porous brick walls, and concrete hollow blocks, each having distinct thermal conductivity coefficients of 1.89W/(m•K), 0.63W/(m•K), 1.26W/(m•K), and 0.315W/(m•K), respectively. Figure 10 illustrated the periodic temperature fluctuations within the wall surface, contingent upon these varying thermal conductivity coefficients, revealing notable differences. Notably, concrete hollow blocks, due to their minimal thermal conductivity, restricted indoor air heat dissipation, resulting in a more pronounced decrease in room temperature. In contrast, solid clay bricks, possessing the highest thermal conductivity, facilitated greater heat dissipation from indoor air.

Figure 11 shows the proportion of exterior wall types in rural buildings around Chengdu. The findings had revealed that only 6.3% of those rural buildings had employed concrete hollow blocks for their exterior walls, whereas the proportion of clay solid brick walls had stood as high as 40.9%. Consequently, the overall LCI score of the E21 indicator in rural Chengdu had been merely 56.24. Pidu District had stood out as a notable example, where the majority of buildings in Jinbai Village had been self-constructed by villagers over an extended period. As a result of the high thermal conductivity of these exterior walls and their tendency to have absorbed more internal heat, the LCI score of the E21 index in this area had been low, reaching only 42.19.

As depicted in Figure 12, the spatial distribution of LCI scores for the E indicator had shown a notable decline from the southwestern regions towards the northeast by then. Specifically, the pinnacle of 81.35 for the E indicator’s LCI score had been achieved in Liyuan Demonstration Village, located within Shuangliu District. In stark contrast, the lowest point of 51.79 had been recorded in Yituan Village, situated in Xindu District. This disparity had primarily stemmed from the fact that some villages in Xindu had endured protracted construction periods, coupled with a severe degradation of their overall architectural integrity. Notably, the prevalence of solid brick walls, known for their inferior insulation properties, and outer windows constructed from either single-layer plastic steel or wooden materials, both of which had contributed significantly to heat loss, had been found in these villages. Consequently, the average LCI score for the E index in Xindu District had hovered at a mere 59.00 by that time, underscoring the urgency that had arisen to enhance energy efficiency.

In stark juxtaposition, Liyuan Village in Shuangliu District had stood as a beacon of sustainability. The local government had embraced a proactive approach, having integrated greening and comfort considerations into every facet of building design, construction, and operation. This holistic methodology had yielded a commendable LCI score of 68.89 for the E index within the region, attesting to the potential that had been realized for sustainable development and improved energy performance in rural areas.

4. Recommendations

Conclusions

References

1. Jiashi Han, Lei Zhang,Yang Li. Hotspots. flaws and deficiencies of research on rural energy upgrading:A review. Energy Strategy Reviews 38 (2021) 100766. [CrossRef]
2. Donglin Zhang, Yong Ding, Lingxiao Fan, Xiangting Jiang. New index for a comprehensive evaluation of building energy performance through spatial and temporal dimensions.Energy & Buildings. [CrossRef]
3. Zhengxuan Liu,Chenxi Yu, Queena K Qian, Ruopeng Huang, Kairui You,Henk Visscher , Guoqiang Zhang. Incentive initiatives on energy-efficient renovation of existing buildings towards carbon–neutral blueprints in China: Advancements, challenges and prospects. Energy & Buildings 296 (2023) 113343. [CrossRef]
4. Geoffrey.K.F. Tso, JingJing Guan. A multilevel regression approach to understand effects of environment indexs and household features on residential energy consumption, Energy 66 (2014) 722–731, . [CrossRef]
5. Keith J. Baker, R.Mark Rylatt. Improving the prediction of UK domestic energy-demand using annual consumption-data, Appl. Energy 85 (6) (2008) 475–482. [CrossRef]
6. Shva.Saadatian, Nuno.Simoes, Fausto Freire. Integrated environmental, energy and cost lifecycle analysis of windows: optimal selection of components, Build. Environ. 188 (2021) 107516. [CrossRef]
7. Chanita. Mano, Atthakorn Thongtha. Enhanced thermal performance of roofing materials by integrating phase change materials to reduce energy consumption in buildings, J. Renew. Mater. 9 (3) (2021). [CrossRef]
8. Yanling Wang, Fang Wanga, Haiyan Wanga. Influencing factors regression analysis of heating energy consumption of rural buildings in China. Procedia Engineering 205 (2017) 3585–3592. [CrossRef]
9. Cristina Marincu, Daniel Dan, Ligia Moga. Investigating the influence of building shape and insulation thickness on energy efficiency of buildings. Energy for Sustainable Development 79 (2024) 101384. [CrossRef]
10. Xiaoliang Wang, Yi Wu, Xinyi Dong, Manshu Liu, Bo Lei, Xianmin Mai. Optimization of global energy consumption of buildings based on photothermal coupling effect of exterior windows in Qinghai-Tibet plateau. Journal of Building Engineering 85 (2024) 108710. [CrossRef]
11. Issa Bosu, Hatem Mahmoud, Shinichi Ookawara, Hamdy Hassan. Applied single and hybrid solar energy techniques for building energy consumption and thermal comfort: A comprehensive review. Solar Energy 259 (2023) 188–228. [CrossRef]
12. Lingyan Li, Wangming Sun, Wei Hu, Yongkai. Sun. Impact of natural and social environmental factors on building energy consumption: based on bibliometrics, J. Build. Eng. 37 (2021) 102136, . [CrossRef]
13. Abdeen Mustafa Omer. Energy, environment and sustainable development, Renew. Sustain. Energy Rev. 12 (9) (2008) 2265–2300, . [CrossRef]
14. Jihui Yuan, Zhichao Jiao, Xiong Xiao, Kazuo Emura, Craig Farnham. Impact of future climate change on energy consumption in residential buildings: A case study for representative cities in Japan. Energy Reports 11 (2024) 1675–1692. [CrossRef]
15. Ivan Julio Apolonion Callegas, Raquel Moussalem Apolonio, Emeli Lalesca Aparecida da Guarda, Luciane Cleonice Durante, Karyna de Andrade Carvalho Rosseti, Filipa Roseta and Leticia Mendes do Amarante. Bermed earth-sheltered wall for low-income house: thermal and energy measure to face climate change in tropical region, Appl. Sci. 11 (1) (2021), . [CrossRef]
16. Mohammadjavad Mahdavinejad, Hassan Bazazzadeh, Fatemeh Mehrvarz,Umberto Berardi, Tahereh Nasr, Somayeh Pourbagher, Siamak Hoseinzadeh. The impact of facade geometry on visual comfort and energy consumption in an office building in different climates. Energy Reports 11 (2024) 1–17. [CrossRef]
17. Qiaoni Wei, Qifen Li, Yongwen Yang, Liting Zhang, Wanying Xie. A summary of the research on building load forecasting model of colleges and universities in North China based on energy consumption behavior: A case in North China. Energy Reports 8 (2022) 1446–1462. [CrossRef]
18. Jiaojiao Duan, Nianping Li, Jinqing Peng, Qingqing Liu, Ting Peng,Sheng Wang. Clustering and prediction of space cooling and heating energy consumption in high-rise residential buildings with the influence of occupant behavior: Evidence from a survey in Changsha, China. Journal of Building Engineering 76 (2023) 107418. [CrossRef]
19. Douglas Roschildt Hax, Rodrigo Karini Leitzke, Antonio César Silveira Baptista da Silva ,Eduardo Grala da Cunha. Influence of user behavior on energy consumption in a university building versus automation costs. Energy & Buildings 256 (2022) 111730. [CrossRef]
20. Xiaoxiao Xu, Hao Yu, Qiuwen Sun, Vivian W.Y. Tam. A critical review of occupant energy consumption behavior in buildings:How we got here, where we are, and where we are headed. Renewable and Sustainable Energy Reviews 182 (2023) 113396. [CrossRef]
21. Dasheng Lee, Chin-Chi Cheng. Energy savings by energy management systems: a review, Renew. Sustain. Energy Rev. 56 (2016) 760–777, . [CrossRef]