Multi-Objective Optimization of Envelope Structures for Rural Dwellings in Qianbei Region, China: Synergistic Enhancement of Energy Efficiency, Thermal Comfort, and Economic Viability

Abstract

In China, retrofitting rural dwellings is a crucial step toward enhancing living conditions and lowering energy waste. One of the most important ways to enhance building performance is to retrofit the building envelope. The Qianbei Region’s (Northern Guizhou Province, China) rural dwellings are the subject of this study. It identifies the persistent issue of inadequate thermal comfort in local rural dwellings through indoor thermal environment measurements and questionnaire surveys. Using a parametric modelling tool (Rhino-Grasshopper-Ladybug Tools), multi-objective optimization was performed using a non-dominated sorting genetic algorithm (NSGA-II), with the types of external windows, walls, and roof insulation as optimization variables, and building energy consumption (E), annual thermal discomfort hours (TDT), and life cycle cost increment (ΔLCC) as optimization objectives. After the retrofitting, the building’s energy consumption was reduced from the baseline value of 96.41 kWh/m² to 42.40 kWh/m² (a 56% reduction), and the annual duration of thermal discomfort decreased from 6173 h to 5078 h (a 17.7% decrease). This resulted in a positive economic return, with a cost saving of ΔLCC = −56,329.87 CNY. The research proposes a scientific method for the energy-saving retrofitting of rural dwellings in the Qianbei Region, which also serves as a guide for the optimization of building performance in comparable climate zones.

1. Introduction

1.1. Background

China possesses extensive rural regions and a significant rural populace. In 2023, 477 million individuals in China reside in rural areas, representing 33.84% of the total population [1]. The Ministry of Housing and Urban-Rural Development projects that by 2030, the per capita housing area in rural regions will reach 45 m², with the total living area expected to reach approximately 22 billion m² [2]. Most rural dwellings in China are constructed by residents and local workers, often without adequate energy-saving measures or thermal insulation designs. Consequently, these buildings exhibit elevated energy consumption and inadequate living comfort [3]. Within the framework of the current dual carbon strategy and rural revitalization, the reduction of energy consumption in buildings and the enhancement of living comfort are critical considerations in the construction and retrofitting of rural dwellings in China. Studies indicate that the primary factor influencing the indoor thermal environment and energy consumption in rural dwellings is the building envelope [4]. Despite the rapid growth of the per capita economy in rural regions, the majority of China’s rural areas remain underdeveloped and significantly below the national average [5]. Under current economic conditions, energy-saving retrofits for existing dwellings, which cost approximately 200–350 yuan/m², are more viable than new construction at 800–1200 yuan/m², resulting in an initial investment reduction of over 50% [6]. Optimizing retrofit costs, energy consumption, and comfort in the retrofitting of rural dwellings is essential for achieving the rural emission reduction target. This effort is projected to lead to an annual reduction of the equivalent of 100 million tons of carbon dioxide emissions while also enhancing living quality [7].

1.2. Indoor Thermal Environment and Energy Consumption of Rural Dwellings

The indoor thermal environment of rural dwellings significantly influences residents’ quality of life and the energy consumption of buildings. Research indicates that this issue primarily arises from inadequacies in the design of the building envelope and construction methods [8]. Numerous rural structures were constructed without adherence to scientific energy-saving design standards, leading to thermal performance of walls, roofs, and windows that significantly fail to meet current energy-saving regulations. External walls may exhibit inadequate insulation or insufficient thickness, roof insulation may demonstrate poor thermal resistance, and the air tightness and insulation performance of windows may not satisfy fundamental standards. The identified factors contribute directly to the instability of the indoor thermal environment and the low comfort levels in rural dwellings [9]. The economic conditions of residents also constrain the indoor thermal environment of rural dwellings. The per capita disposable income of rural residents constitutes merely 41.85% of that of urban residents, thereby limiting the advancement and execution of building envelope retrofitting [1]. The inadequate configuration of heating and cooling systems further worsens the thermal environment. In winter, heating in rural dwellings primarily relies on coal stoves or traditional fireplaces, which exhibit significant energy inefficiency and contribute substantially to indoor air pollution. In summer, due to the low penetration rate of air conditioning equipment, rural dwellings predominantly depend on natural ventilation. Many rural dwellings exhibit suboptimal design and inadequate ventilation, leading to significant issues of indoor overheating during the summer months [8].

Numerous studies indicate that energy consumption in rural dwellings is on the rise, with a trend suggesting it may even exceed that of urban dwellings. Research data from the Building Energy Research Center (BERC) at Tsinghua University indicates that the energy consumption of typical rural dwellings amounted to 229 million tons of standard coal in 2020, representing 22% of China’s total building energy consumption [10]. Tsinghua University conducted a comprehensive survey of rural residents in China, revealing significant potential for energy conservation in these areas. The findings suggest that building performance can be enhanced through the optimization of building envelopes and layouts [4]. Ugwoke et al. [11] analyzed low-cost buildings in developing countries, proposing strategies for retrofitting the building envelope and enhancing energy efficiency, thereby illustrating the potential for energy savings. Chicherin et al. [12] demonstrated that enhancing the thermal insulation of walls can lead to a 14.4% reduction in annual energy consumption based on their study of the coordinated optimization of building envelopes and heating systems in rural Kazakhstan. Cao et al. [13] demonstrated that substantial enhancements in building performance can lead to energy savings exceeding 50% via comprehensive retrofits of the building envelope. The aforementioned studies indicate that retrofitting measures for the building envelope can decrease energy consumption and enhance indoor thermal comfort, thereby significantly improving residents’ quality of life.

1.3. Research on Rural Dwellings in Guizhou Province

The indoor thermal environment and energy use of rural dwellings in Guizhou have become a focal point of extensive academic research in recent years. Liu et al. [14], Li et al. [15], Wu et al. [16], Luo [17], Liu [18], and Li [19] selected the traditional Miao dwellings in Southeast Guizhou, the traditional Dong dwellings in Dimen Village, the Miao stilt houses in Southeast Guizhou, the Tujia traditional dwellings, the Dong dwellings in Southeast Guizhou, and the slate-roofed dwellings in Central Guizhou as their respective research objects. Physical measurements and questionnaire surveys were conducted to evaluate the thermal environment and energy consumption characteristics of these building forms, and optimization strategies for building envelopes and natural ventilation were proposed. The energy-saving retrofit of Guizhou’s existing dwelling buildings was the subject of additional attention by Wang et al. [20]. They suggested specific technical solutions, including composite walls, the use of lightweight insulation materials, the addition of a roof insulation layer, and the improvement of the airtightness of external doors and windows after analyzing the external building envelope’s inadequate thermal performance. Zhou [21] employed BIM technology to offer creative solutions for the renovation of dwelling buildings in the Qianbei Region, while Xiao [22] used field research to highlight the types and features of dwelling buildings in the Qianbei Region. Through a survey of traditional rural dwellings in the Qianbei Region, Liu et al. [23] examined the thermal comfort and adaptability of the residents and suggested methods for improving the indoor environment using an enhanced APMV model. These studies offer a multitude of empirical data and design optimization techniques to help Guizhou’s various dwelling building types adapt to the city’s complex climate. The provincial government of Guizhou has also recently released pertinent policies that highlight the significance of energy-saving retrofits of existing buildings, particularly the upgrading of the building envelope, which has emerged as a key strategy to enhance building thermal performance and lower energy consumption [24,25].

1.4. Multi-Objective Optimization of Building Performance

The process of determining appropriate variable values to enable the objective function to reach its minimum or maximum value is referred to as multi-objective optimization, which originated in mathematics [26]. The rapid development of digital design and building performance simulation software in recent years has facilitated the extensive application of multi-objective optimization techniques in addressing challenging performance balance issues across various building disciplines, such as building energy consumption. Key findings from representative studies on multi-objective optimization of rural dwellings conducted both domestically and internationally in recent years are summarized in Table 1.

Multi-objective optimization strategies are much more common than single-objective algorithms in building energy efficiency design, according to foreign research [27,28]. A new avenue for low-carbon design has been opened by numerous studies [29,30,31,32] that have put forth multi-objective optimization frameworks that accomplish the coordinated optimization of building envelope performance between economic and energy efficiency indicators. The effectiveness of multi-performance coordinated optimization using simulation tools in conjunction with optimization algorithms has been confirmed by domestic studies [33,34,35,36], which demonstrate that the thermal insulation performance of the building envelope of rural dwellings has a significant impact on building energy consumption, indoor comfort, and life cycle costs. However, the promotion of these technologies in rural areas has been hindered by their high retrofit costs [37]. A viable path for technology adaptation under cost constraints has been provided by some studies [38,39,40,41,42], which have suggested retrofit strategies through multi-scenario simulation. These studies show that low-cost technologies can be compatible with energy efficiency targets and that there are workable solutions for synergies between economy and energy efficiency.

Table 1. Representative studies on the multi-objective optimization of rural dwellings both at home and abroad.

Reference	Year	Region	Parameters	Simulation Tools	Optimization Algorithms	Simulation Objectives
Attia S. et al. [27]	2013	Review	Building envelope	MATLAB™	ACOA, NSGA	Zero energy buildings
Longo S. et al. [28]	2019	Review	Building envelope	CFD, DragonFly	NSGA-II	Cost-optimal, Low-energy
Lizana J. et al. [29]	2016	Southwest of Europe	Roof, wall insulation	DOE, EnergyPlus	Multi-criteria assessment	Energy efficiency
Ehsan A. et al. [30]	2012	Portuguese	Envelope parameters	MATLAB	Tchebycheff	Energy saving, Retrofit cost
Jermyn D. et al. [31]	2016	Toronto, Canada	Building envelope	EnergyPlus	Estimated using a hybrid method	Deep energy retrofits, Cost estimation
Naderi E. et al. [32]	2020	Six different climatic regions of Iran	Shading control strategy	JEPl, EnergyPlus	NSGA-II	Thermal comfort, Visual comfort, Building energy
Zhu et al. [33]	2011	China (multi-climate)	Thickness, Material of insulation layers	EnergyPlus	Analytical Formula Method	Life-cycle, Reduced heating, Cooling demand
Duan et al. [34]	2024	Cold regions, China	Geometry, Envelope	OpenStudio, EnergyPlus, Radiance	Octopus, Genetic Algorithm	Energy demand, Thermal comfort, Daylighting
Wang et al. [35]	2023	Southeastern coastal areas of China	Envelope parameters	DesignBuilder, EnergyPlus	NSGA-II	Energy-saving, Comfort
Molake et al. [36]	2023	Kezhou, China	Flat skylight, Clerestory window	Rhino-Grasshopper, EnergyPlus	NSGA-II	Daylight performance, Thermal comfort, Energy-saving
Gao et al. [38,39]	2022	Hebei, China	Envelope design variables	Rhino-Grasshopper	NSGA-II	Low-carbon retrofit, Renovation cost
Zhang et al. [40]	2024	Liaoning, China	Thickness of insulation layers	Grasshopper, EnergyPlus	NSGA-II	Renovation cost, Energy-saving
Liu et al. [41]	2024	China	Envelope design variables	Grasshopper	NSGA-II	Energy consumption, Economic efficiency
Zhai et al. [42]	2025	Severely cold region of China	Envelope parameters	EnergyPlus	NSGA-II	Energy consumption, Thermal comfort, Cost-effectiveness

Table 1. Representative studies on the multi-objective optimization of rural dwellings both at home and abroad.

Reference	Year	Region	Parameters	Simulation Tools	Optimization Algorithms	Simulation Objectives
Attia S. et al. [27]	2013	Review	Building envelope	MATLAB™	ACOA, NSGA	Zero energy buildings
Longo S. et al. [28]	2019	Review	Building envelope	CFD, DragonFly	NSGA-II	Cost-optimal, Low-energy
Lizana J. et al. [29]	2016	Southwest of Europe	Roof, wall insulation	DOE, EnergyPlus	Multi-criteria assessment	Energy efficiency
Ehsan A. et al. [30]	2012	Portuguese	Envelope parameters	MATLAB	Tchebycheff	Energy saving, Retrofit cost
Jermyn D. et al. [31]	2016	Toronto, Canada	Building envelope	EnergyPlus	Estimated using a hybrid method	Deep energy retrofits, Cost estimation
Naderi E. et al. [32]	2020	Six different climatic regions of Iran	Shading control strategy	JEPl, EnergyPlus	NSGA-II	Thermal comfort, Visual comfort, Building energy
Zhu et al. [33]	2011	China (multi-climate)	Thickness, Material of insulation layers	EnergyPlus	Analytical Formula Method	Life-cycle, Reduced heating, Cooling demand
Duan et al. [34]	2024	Cold regions, China	Geometry, Envelope	OpenStudio, EnergyPlus, Radiance	Octopus, Genetic Algorithm	Energy demand, Thermal comfort, Daylighting
Wang et al. [35]	2023	Southeastern coastal areas of China	Envelope parameters	DesignBuilder, EnergyPlus	NSGA-II	Energy-saving, Comfort
Molake et al. [36]	2023	Kezhou, China	Flat skylight, Clerestory window	Rhino-Grasshopper, EnergyPlus	NSGA-II	Daylight performance, Thermal comfort, Energy-saving
Gao et al. [38,39]	2022	Hebei, China	Envelope design variables	Rhino-Grasshopper	NSGA-II	Low-carbon retrofit, Renovation cost
Zhang et al. [40]	2024	Liaoning, China	Thickness of insulation layers	Grasshopper, EnergyPlus	NSGA-II	Renovation cost, Energy-saving
Liu et al. [41]	2024	China	Envelope design variables	Grasshopper	NSGA-II	Energy consumption, Economic efficiency
Zhai et al. [42]	2025	Severely cold region of China	Envelope parameters	EnergyPlus	NSGA-II	Energy consumption, Thermal comfort, Cost-effectiveness

In conclusion, prior research on the indoor thermal environment, rural thermal energy consumption, and multi-objective optimization of rural dwellings has achieved significant progress. However, systematic investigations into the types, characteristics, and current status of rural dwellings in the Qianbei Region remain insufficient. Moreover, economic factors related to energy-saving retrofits and thermal comfort for rural dwellings in the Qianbei Region have not been adequately considered. Therefore, this study must adopt a multi-objective collaborative optimization approach to explore retrofit strategies and accurately identify the retrofit needs of rural dwellings in the Qianbei Region based on field research. This work is theoretically, methodologically, and practically innovative.

1.5. Research Objectives

This study aims to examine the indoor thermal environment and energy consumption of rural dwellings in the Qianbei Region, considering local climatic attributes and economic conditions. A scientifically valid energy-saving retrofit strategy is suggested to attain synchronized optimization of energy usage, comfort, and cost-effectiveness. A closed-loop framework of “environmental perception-model iteration-decision optimization” is established to offer theoretical support for energy-efficient retrofits of rural dwellings in the Qianbei Region and to serve as a reference for the design and sustainable retrofitting of rural dwellings in analogous climatic conditions. The primary objectives are as follows:

• Methodically categorize the architectural types and present conditions of representative rural dwellings in the Qianbei Region and elucidate their distribution patterns and structural variations.
• Methodically identify the deficiencies in the thermal performance of the building envelope and elucidate their primary causes by analyzing the indoor thermal environment characteristics indicated by actual measured data and questionnaire surveys conducted in winter and summer.
• A multi-objective optimization algorithm was employed to investigate the overall influence of design variables for retrofitting the building envelope on performance and economics. The algorithm aimed to attain an optimal equilibrium among energy consumption, occupant comfort, and financial viability.

2. Research Region and Object

2.1. Geography and Climate

The Qianbei Region encompasses Zunyi City and its surrounding counties and districts in the northern part of Guizhou Province, China [43]. Situated at the intersection of the Sichuan, Chongqing, and Guizhou provinces, this area centers on Zunyi City and forms part of the northern edge of the Yunnan–Guizhou Plateau (Figure 1). Geographically, Zunyi City spans from 27°8′ N to 29°12′ N latitude and from 105°36′ E to 108°13′ E longitude, covering a total area of approximately 30,762 km². The region exhibits a humid, subtropical, monsoon climate.

According to the “Code for Thermal Design of Civil Buildings” GB 50176-2016 [44], the Qianbei Region falls within the hot summer and cold winter zone (Zone IIIA). The annual average temperature in this area is approximately 16 °C. July is the warmest month, with an average temperature of 26 °C and occasional peaks up to 35 °C. January is the coldest month, with an average temperature of 4 °C and minimum temperatures approaching 0 °C. Annual precipitation ranges between 1100 and 1300 mm, predominantly occurring during the summer months. The climate is characterized by high humidity and relatively mild winters (Figure 2).

2.2. Characteristics of Rural Dwellings in Qianbei Region

The research selected Luoqiao Village in Zunyi City as the object of investigation. Luoqiao Village is situated in the hilly and plateau mountainous regions of Guizhou Province, with 720 households and 3080 people. There are 675 rural dwellings, which are distributed in a scattered manner along the terrain. Based on the field investigation of the rural dwellings in Luoqiao Village, Zunyi City, this study categorized them into three types according to the construction period, building materials, and architectural form (

Table 2. Dwelling characteristics of Luoqiao Village.

Category	First Type of Dwellings	Second Type of Dwellings	Third Type of Dwellings
Construction Year	Before 1990	1990–2015	After 2015
Representative Photos
Heating Method	Wood or Coal	Coal or Electric Stove	Electric Stove
Heating Tools
Structural Materials	Board Wall and Pier-Beam Structure	Brick and Concrete Structure	Concrete Frame Structure
Thermal Performance	No insulation layer, and severe heat loss	Partial exterior wall tiles, insufficient insulation	Exterior wall tiles, energy-saving windows
Existing Problems	Rotten, leaky, poor ventilation	Cold in winter and hot in summer, high energy consumption	High construction cost, low popularity
Retrofit Potential	Low (Need to preserve the style)	High (Exterior wall insulation + Roof retrofit)	Low (Advanced technology but high cost, limited retrofit benefits)

). These dwelling types reflect the evolution of local rural dwellings, from traditional to modern types.

The first type encompasses 127 rural dwellings featuring a long construction history, small area, and low floor height. The roof forms are diverse, and the building materials are mainly derived from local trees. As remnants of traditional dwellings in the Qianbei Region, their wooden board and beam structure and pitched roof forms embody regional architectural culture. However, they have significant deficiencies in modern thermal performance and functionality. Due to a long-term lack of maintenance, problems such as the decay of wooden beams and columns, cracked walls, and roof leaks are prevalent, resulting in cold and damp living environments and poor ventilation. Additionally, due to the outflow of the rural population and the phenomenon of the “hollowing out” of houses, most of these rural dwellings are now vacant, and due to their age and lack of maintenance, the remaining quantity is scarce.

The second type consists of 512 rural dwellings. These dwellings are typically two to three stories high with flat roofs, and their floor plans better meet the requirements of modern living patterns, featuring strong durability and resistance to wind and rain. Nevertheless, the high thermal conductivity of brick–concrete materials in the hot summer and cold winter climate of the Qianbei Region creates a “cold storage in winter–heat storage in summer” double deterioration effect. Most of the rural dwellings only have simple treatments, such as tiling on the main facade. In contrast, other facades remain exposed and untreated. This simplified construction reduces costs but exacerbates the indoor thermal environment issues of the dwellings.

The third type comprises 36 rural dwellings. Compared with the second type, they have significant enhancements in terms of the floor plan, architectural appearance, wall materials and construction, roof and ceiling treatment, etc. These dwellings extensively adopt new materials and technologies, such as high-performance energy-saving windows and insulated walls. These significantly improve the building’s airtightness and insulation performance, effectively reducing heat loss in winter and indoor temperature rise in summer and enhancing thermal comfort and energy efficiency. Due to their high construction costs and the fact that villagers have increasingly favored settling in cities in recent years, the number of existing third-type rural dwellings is relatively small.

2.3. Profile of the Research Object

Among the three types of dwellings, the first type was constructed long ago, and the majority of these structures have significantly deteriorated and require the preservation of their original architectural style. Interventions ought to prioritize protective repair. The third type of dwelling, constructed relatively recently, has already integrated energy-efficient technologies and their overall condition is satisfactory, and the residents exhibit a low inclination to retrofit. The second type of dwelling, constructed of brick–concrete, is of the highest quantity and has significant retrofitting potential. Among them all, the second type costs the lowest for retrofitting in terms of economy. Thus, the second type of dwelling was chosen as the subject of the study.

The sample building was constructed in 2005. It features a two-story brick–concrete structure with a total floor area of 232 m², a flat roof, and an orientation of 45° southwest. The indoor clear height is 3.0 m on the first floor and 2.9 m on the second floor. The building utilizes wooden doors and 5 mm single-pane glass windows, with some windows showing signs of damage. A detailed profile of the sample building is provided in

Table 3. Overview of sample building structures.

Part	South Facade Wall	Roof	Slab
Real Photos
Structure	10 mm white tiles + 10 mm cement mortar + 120 mm red bricks + 10 mm cement mortar + 2 mm plastering mortar	20 mm cement mortar + SBS waterproof membrane + 20 mm cement mortar leveling layer + reinforced concrete cast-in-place slab + 2 mm plastering mortar	Common floor tiles + 20 mm cement mortar + 20 mm cement mortar leveling layer + cement slurry + reinforced concrete cast-in-place slab + 2 mm plastering mortar
Part	Other Exterior Walls	Floor	Doors and Windows
Real Photos
Structure	120 mm red bricks + 10 mm cement mortar + 2 mm plastering mortar	20 mm cement mortar leveling layer + cement slurry + compacted natural soil	80 mm wooden door + single-layer wooden frame with 5 mm single glass

.

3. Methodology

This study adopts a three-stage progressive framework of “measurement–simulation–optimization” to systematically explore the multi-objective synergy mechanism for rural dwelling envelope retrofits in the Qianbei Region (Figure 3). First, through an integrated analysis of thermal environment field measurements and both subjective and objective evaluation data, a comprehensive dataset is established that includes building physical parameters (planar and elevational morphology, construction patterns), microclimate dynamic characteristics (temperature, humidity, wind velocity, and mean radiant temperature), and residents’ thermal comfort perception. Subsequently, a dynamic building simulation model is developed using the parametric modeling toolchain (Rhino-Grasshopper-Ladybug/Honeybee) combined with the EnergyPlus energy consumption engine for annual performance simulation. Cross-validation between measured data and simulation results is performed based on Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) metrics to ensure the accuracy and reliability of the model. Finally, with envelope design parameters as decision variables and building energy consumption (E), annual thermal discomfort hours (TDT), and Life Cycle Cost Increment (ΔLCC) as optimization objectives, the NSGA-II algorithm is applied to generate the Pareto optimal solution set. The Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) multi-criteria decision-making method is then utilized to identify the envelope retrofit scheme best suited to the climatic conditions of the Qianbei Region.

3.1. Field Measurement and Questionnaire Survey of Sample Buildings

3.1.1. Measurement Time

Based on the local climate conditions, this study was conducted during the hottest month (25–26 July 2024) and the coldest month (27–28 January 2025). To ensure the accuracy and representativeness of the test data, the indoor–outdoor temperature difference during the measurement period was maintained at least 50% of the designed temperature difference [45].

3.1.2. Measurement Content

In accordance with Fanger’s thermal comfort theory [46], the key physical parameters influencing indoor thermal comfort include air temperature, relative humidity, wind velocity, and mean radiant temperature (Figure 3-step 1). Therefore, the measurements primarily focused on these four parameters. The measurement instruments and their respective precisions are detailed in

Table 4. Testing instruments and accuracy.

Measurement Parameter	Testing Instrument	Testing Range	Testing Accuracy	Recording Method	Interval
Outdoor Air Temperature	TES-1360A	−20~60 °C	±0.8 °C	Manual	1 h/Time
Outdoor Relative Humidity		10~95%	±3%	Manual	1 h/Time
Indoor Air Temperature	JT-IAQ-50	−20~120 °C	±0.5 °C	Automatic	5 min/Time
Indoor Relative Humidity		0~100%	±1.5%	Automatic	5 min/Time
Indoor Wind Velocity		0.05~2 m/s	±0.03 m/s	Automatic	5 min/Time
Indoor Black Globe Temperature		−20~120 °C	±0.5 °C	Automatic	5 min/Time

. Specifically, the mean radiant temperature was calculated using Equation (1). All testing instruments were calibrated before the experiment and adhered to the relevant requirements outlined in the “Evaluation Standard for Indoor Thermal and Humid Environment of Civil Buildings GB/T 50785-2012” [45,47] and the “Standard for Testing Methods of Building Thermal Environment JGJ/T347-2014”. During the measurement period, all room doors and windows were kept in a natural open state (doors opened at 7:30 a.m. and closed at 9:00 p.m., with windows continuously open), and no active equipment (e.g., air conditioners or fans) were utilized. Additionally, the activities of occupants within the measured samples remained largely consistent.

$${\mathrm{t}}_{\mathrm{m}\mathrm{r}}={[{\left({\mathrm{t}}_{\mathrm{g}}+273\right)}^4+{\displaystyle \frac{1.1\times {10}^8{\mathrm{V}}^{0.6}}{\mathsf{\epsilon }\mathrm{D}}}\left({\mathrm{t}}_{\mathrm{g}}-{\mathrm{t}}_{\mathrm{a}}\right)]}^{0.25}-273$$

(1)

where t_a is the dry-bulb temperature (°C); t_g is the globe temperature (°C); V is the wind velocity (m/s); $\mathsf{\epsilon }$ is the globe emissivity ($\mathsf{\epsilon }$= 0.97); and D is the black globe diameter (D = 0.05 m).

Given that these rural dwellings lack artificial cooling or heating sources such as air conditioners, residents rely on natural ventilation and behavioral adjustments to adapt to climatic conditions. This study, therefore, referenced the evaluation criteria for buildings without artificial cooling or heating sources specified in GB/T 50785-2012 [45]. The Adapted Predicted Mean Vote (APMV) was deemed more suitable for objectively reflecting the actual thermal sensation experienced in different types of dwellings under natural environmental conditions. The calculation formula for APMV is presented in Equation (2).

$$\mathrm{APMV}={\displaystyle \frac{\mathrm{PMV}}{1+\mathsf{\lambda }·\mathrm{PMV}}}$$

(2)

where PMV is the Predicted Mean Vote; and λ is the adaptive coefficient for hot summer and cold winter regions (λ = 0.21 when PMV ≥ 0, and λ = −0.49 when PMV < 0). The APMV value ranges from −3 to +3, representing varying levels of thermal comfort.

3.1.3. Measurement Method

The measurement method strictly complies with the standards of GB/T 50785-2012 and JGJ/T347-2014. Outdoor measurement points are strategically positioned to avoid direct sunlight and air vents, maintaining a distance of 1.2 m from the exterior envelope. Indoor measurement points are primarily located in the living room and bedrooms. Given that indoor occupants predominantly adopt sitting or lying postures, the instruments are placed at a height of 1.5 m above the ground for testing [47]. All measurement points are situated near the seating areas of the occupants without interfering with their daily activities. To accurately reflect the thermal environment conditions of living spaces under normal living conditions and ensure the authenticity and representativeness of the test data, residents are required to maintain their regular routines, including sleep schedules and window opening/closing habits, during the measurement period. The floor plan of the sample building and the distribution of measurement points are presented in Figure 4.

3.1.4. Questionnaire Survey

Based on the ASHRAE Standard 55-2023 [48] and GB/T 50785-2012 [45], while considering the characteristics of dwelling buildings in the Qianbei Region and the living habits of local residents, the questionnaire content was systematically designed (Figure 5). The questionnaire comprehensively covers the following aspects: basic information (gender, age, population structure, income level, building area, construction year, building type, window frame and glass material, etc.), behavioral habits (active measures taken by residents to regulate indoor temperature and humidity, common cooling methods, etc.), and subjective perceptions (indoor thermal sensation, temperature preference, humidity sensation, humidity preference, thermal comfort, and acceptability). Specifically, thermal sensation, humidity sensation, and thermal comfort are evaluated using a 7-point scale; acceptability is assessed using a 5-point scale; and temperature preference and humidity preference are rated using a 3-point scale. To ensure the objectivity and reliability of the subjective judgments regarding the surrounding thermal environment when completing the questionnaire, each participant is required to sit quietly for 20 min before the survey begins during the test.

3.2. Building Performance Simulation and Validation

3.2.1. Simulation Method

This study used the Rhino-Grasshopper-Ladybug/Honeybee-EnergyPlus integrated tool chain to complete the building performance simulation analysis (Figure 3-step 2), and the specific process is as follows:

• Geometric modeling: Based on measured data, Rhinoceros 8 was used to accurately construct a 3D geometric model of rural dwellings in the Qianbei Region, covering components such as walls, doors, and windows, to provide a spatial benchmark for subsequent analysis;
• Parametric design: The Grasshopper visual programming platform was used to dynamically adjust the variable parameters of the building envelope, enabling it to quickly generate hundreds of retrofit schemes and automatically transfer the parameters to the performance simulation module;
• Climate data analysis: Using Ladybug to analyze typical meteorological data for the Zunyi area for a typical year. The climate data comes from the EPW Map website. The Zunyi EPW weather file used is CHN_GZ_Zunyi.577130_CSWD.epw [49];
• Energy model construction: Honeybee was used to construct the building energy simulation model and connect it to the simulation engine. The building operating parameters shown in

  Table 5. Building operating parameters.

  Parameters	Settings
  Simulation period	From 1 January to 31 December
  Population density	25 m2/p
  Calculated number of air changes for winter heating	1.0 H−1
  Metabolic rate per inhabitant for home activities	Sitting/Sleeping	2.45 mL/(kg·min)
  	Standing/Relaxing	3.5 mL/(kg·min)
  	Cooking	6.475 mL/(kg·min)
  	Cleaning the room	6.475 mL/(kg·min)
  Lighting density per area	5.0 W/m2
  Heating, ventilation, and Air Conditioning parameters	Winter heating temperature	18 °C
  	Air conditioning temperature in summer	26 °C
  	Cooling and heating coefficient of performance	2.5

  were set with reference to the “General Code for Energy Efficiency of Buildings and Utilization of Renewable Energy GB55015-2021” [50];
• Dynamic simulation: The energy simulation engine OpenStudio and EnergyPlus were called to perform hourly simulations for 8760 h throughout the year to simulate the annual energy consumption and thermal comfort of the building [51].

3.2.2. Simulation Verification

Before the simulation, the model requires verification. Initially, the research utilizes the online EPW editor to import the original meteorological file (EPW format), identifies the test period, and substitutes the recorded hourly meteorological data with the relevant fields to create a modified meteorological file [52]. Subsequently, the revised file is imported into the Ladybug platform to simulate the hourly indoor temperature of the sample building during winter and summer, respectively. Upon completion of the simulation, the calculation results are extracted and meticulously synchronized with the measured data by timestamp for error analysis. The error analysis employs MAE and RMSE for comparison, with RMSE being extensively utilized for simulation validation and particularly appropriate for assessing building performance simulations across diverse natural operating conditions. In indoor temperature simulation, an MAE and RMSE error range of 1 °C to 2 °C is deemed reasonable and acceptable. The equations for determining MAE and RMSE are as follows:

$$\mathrm{MAE}={\displaystyle \frac{{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}}|{\mathrm{x}}_{\mathrm{m},\mathrm{i}}-{\mathrm{x}}_{\mathrm{r},\mathrm{i}}|}{\mathrm{n}}}$$

(3)

$$\mathrm{R}\mathrm{M}\mathrm{S}\mathrm{E}=\sqrt{{\displaystyle \frac{{\displaystyle \sum _{\mathrm{i}=\mathrm{1}}^{\mathrm{n}}}{\left({\mathrm{x}}_{\mathrm{r},\mathrm{i}}-{\mathrm{x}}_{\mathrm{m},\mathrm{i}}\right)}^2}{\mathrm{n}}}}$$

(4)

where ${\mathrm{x}}_{\mathrm{m},\mathrm{i}}$ is the i-th simulated value; ${\mathrm{x}}_{\mathrm{r},\mathrm{i}}$ is the i-th measured value; n is the sample size, and n = 48.

3.3. Multi-Objective Optimization

3.3.1. Multi-Objective Optimization Approaches

Multi-objective optimization is a theoretical framework and methodology that systematically examines and effectively optimizes multiple conflicting objectives. In multi-objective optimization, various objectives frequently conflict or are incomparable, indicating that a solution may excel in certain objectives while underperforming in others. Consequently, it is challenging to define the overall outcome through a singular optimal solution. In contrast to the “optimal solution” in single-objective optimization, the essence of multi-objective optimization is that no singular solution can simultaneously optimize all objectives. It is essential to identify a solution that reconciles various objectives, thereby creating a collection of non-dominated optimal solutions, referred to as the Pareto non-dominated solution set [53].

The optimization and evaluation of multiple objectives in retrofitting the building envelope of existing rural dwellings require the selection and integration of various measures within a comprehensive retrofit framework. In this study, the parameters associated with the building envelope retrofit are used as variables in the fitness function, while the energy consumption, thermal comfort, and economic viability of existing rural dwellings constitute the three-dimensional optimization objectives. Multi-objective optimization algorithms facilitate the execution of three essential processes: pre-processing, search, and decision-making. Initially, during the pre-processing phase, the performance optimization goals and variables are delineated, and a parametric building information model (Ladybug & Honeybee) is constructed to furnish data support for the ensuing optimization analysis. In the search phase, the performance simulation engine (OpenStudio 3.9.0 & EnergyPlus 25.1.0) iteratively resolves the objective function, while the optimization software (Wallacei 2.7) integrates with the simulation tool to execute a global search via its built-in algorithm, producing a Pareto non-dominated solution set. Subsequently, in the decision-making phase, this solution set undergoes analysis and evaluation through a multi-attribute decision-making method, culminating in the selection of the optimal solution. The fundamental mathematical equations for the multi-objective optimization algorithm are as follows:

$$\left\{\begin{array}{c}\min \mathrm{y}=\mathrm{F}(\mathrm{x})={[{\mathrm{f}}_1(\mathrm{x}),{\mathrm{f}}_2(\mathrm{x}),\dots {\mathrm{f}}_{\mathrm{m}}(\mathrm{x})]}^{\mathrm{T}}\\ \mathrm{s}.\mathrm{t}.{\mathrm{g}}_{\mathrm{i}}(\mathrm{x})\le 0,\mathrm{i}=1,2,\dots ,\mathrm{q}\\ {\mathrm{h}}_{\mathrm{j}}(\mathrm{x})=0,\mathrm{j}=1,2,\dots ,\mathrm{p}\end{array}\right.$$

(5)

where $\mathrm{x}=({\mathrm{x}}_1,\cdots {\mathrm{x}}_{\mathrm{m}})\in \mathrm{X}\subset {\mathrm{R}}^{\mathrm{m}}$ denotes the n-dimensional decision variable, $\mathrm{X}$ represents the n-dimensional decision space; $\mathrm{y}=({\mathrm{y}}_1,\cdots {\mathrm{y}}_{\mathrm{m}})\in \mathrm{Y}\subset {\mathrm{R}}^{\mathrm{m}}$ denotes the m-dimensional objective vector, and $\mathrm{Y}$ represents the m-dimensional objective space. The objective function $\mathrm{F}(\mathrm{x})$ defines m mapping functions from the decision space to the objective space.${\mathrm{g}}_{\mathrm{i}}(\mathrm{x})⩽0 (\mathrm{i}=1,2,\cdots \mathrm{q})$ specifies q inequality constraints, and ${\mathrm{h}}_{\mathrm{j}}(\mathrm{x})=0 (\mathrm{j}=1,2,\cdots \mathrm{p})$ specifies p equality constraints.

3.3.2. NSGA-II Algorithm

Genetic algorithms are optimization algorithms based on the natural evolution process. By simulating fundamental biological processes such as genetic inheritance, mating, and mutation, they conduct population-based optimization searches to identify optimal solutions. These algorithms perform natural selection according to the fitness of each solution (i.e., the value of the objective function), progressively retaining superior solutions while introducing new characteristics through crossover and mutation operations to explore improved solutions. Genetic algorithms are widely applied in architectural optimization design due to their ability to handle variables with multiple distribution states, support parallel computing, evaluate multiple optimization objectives, and avoid becoming trapped in local optima [54].

In this study, multi-objective optimization was implemented using the genetic algorithm via the Wallacei plugin on the Grasshopper platform (Figure 3-step 3), where the Wallacei-X plugin employs NSGA-II as the primary evolutionary algorithm [55]. NSGA-II was developed by Srinivas and Deb in 1994 as an improvement over the original Non-dominated Sorting Genetic Algorithm (NSGA). It has become a critical tool for solving multi-objective optimization problems, with its core enhancement being the introduction of the crowding distance comparison operator. This operator replaces the traditional fitness-sharing strategy, eliminating the need to specify a sharing radius [56]. Additionally, the crowding distance comparison operator ensures a uniform distribution of solutions, enabling individuals in the Pareto non-dominated solution set to span the entire solution domain while maintaining solution diversity. Due to its excellent convergence, uniform solution distribution, and ease of implementation, NSGA-II has been extensively adopted in building performance optimization research [57,58].

In this study, a total of 31,000 variable combinations were optimized. Due to the prohibitively high computational cost of full-scale calculations, the NSGA-II algorithm’s intelligent search strategy was adopted. Based on relevant studies and recommendations from the Wallacei-X tool [59,60], the population size was set to 50, the number of generations to 100, the crossover probability to 0.9, and the mutation probability to 0.01. Convergence to the Pareto front was achieved with only 5000 evaluations (approximately one-sixth of the solution space), and the total computation time was controlled within 7 days. The medium-sized population (50) effectively balanced the diversity of the solution set and computational efficiency (each iteration took approximately 2 min), thereby avoiding local optima. The setting of 100 generations ensured convergence while considering computational efficiency and resource constraints, making it suitable for low-dimensional building optimization scenarios.

3.3.3. Optimization of Objective Function Setting

The retrofit of rural dwellings aims to improve building performance while providing residents with an enhanced living environment, thereby effectively promoting the retrofit’s viability and implementation. This study selects building energy consumption (E), annual thermal discomfort hours (TDT), and Life Cycle Cost Increment (ΔLCC) as sub-objective functions for optimization, enabling a comprehensive evaluation of the performance changes post-retrofit.

• Sub-objective Function f₁(x) = E

In this study, E is defined as the sum of the annual heating energy consumption (E_H) and cooling energy consumption (E_C) of rural dwellings, serving as a key indicator for evaluating energy consumption in rural buildings. Energy consumption associated with hot water, electrical appliances, and lighting is excluded from the calculation, as the optimization variables do not influence it. By employing the Rhino-Grasshopper platform and utilizing the Honeybee plugin to invoke the EnergyPlus engine, the annual energy consumption of rural dwellings is simulated. A smaller E value signifies superior energy-saving performance of the building. The objective function can be expressed as follows:

$${\mathrm{f}}_1(\mathrm{x})=\mathrm{E}={\mathrm{E}}_{\mathrm{H}}+{\mathrm{E}}_{\mathrm{C}}$$

(6)

where E_H notes the building’s annual heating energy consumption (kWh/m²); and E_C denotes the building’s annual cooling energy consumption (kWh/m²).

2.

Sub-objective Function f₂(x) = TDT

While reducing building energy consumption through optimization measures, potential adverse effects on indoor thermal comfort during non-heating and non-cooling seasons must be considered. Therefore, this study adopts TDT as an evaluation index for indoor thermal environment quality. Since the NSGA-II algorithm minimizes the objective function during optimization, the evaluation index is transformed from comfortable time to uncomfortable time.

For the annual indoor thermal environment of rural dwellings, two operational modes are defined based on living habits and door/window opening conditions: a closed ideal climate room during the heating season and a naturally ventilated room during the non-heating season. According to GB/T 50785-2012 [45], thermal discomfort during the heating season is evaluated using PMV, while APMV is applied for the non-heating season. This study adopts the secondary thermal comfort index (−1 ≤ PMV/APMV ≤ 1) as the criterion for calculating the annual thermal discomfort time. The objective function can be expressed as follows:

$${\mathrm{f}}_2(\mathrm{x})=\mathrm{T}\mathrm{D}\mathrm{T}={\displaystyle \frac{{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}}\mathrm{T}\mathrm{D}{\mathrm{T}}_{\mathrm{i}}}{\mathrm{n}}}$$

(7)

where TDT denotes the average cumulative annual indoor thermal discomfort hours across all rooms in rural dwellings (h); TDT_i denotes the annual indoor thermal discomfort hours of the i-th room (h); and n denotes the total number of rooms.

3.

Sub-objective Function f₃(x) = ΔLCC

To evaluate the economic benefits of the retrofit plan, it is essential to comprehensively consider the initial investment, energy-saving gains, and maintenance costs over the entire life cycle. Therefore, ΔLCC is adopted as the evaluation index for the sub-objective function f₃(x) [61,62]. ΔLCC is defined as the difference between the life cycle cost after retrofit (LCC_post) and the life cycle cost before retrofit (LCC_pre), which directly reflects the economic feasibility of the retrofit plan. The objective function is expressed as follows:

$$\Delta \mathrm{L}\mathrm{C}\mathrm{C}=\mathrm{L}\mathrm{C}{\mathrm{C}}_{\mathrm{post}}-\mathrm{L}\mathrm{C}{\mathrm{C}}_{\mathrm{pre}}$$

(8)

$$\mathrm{L}\mathrm{C}{\mathrm{C}}_{\mathrm{pre}}={\displaystyle \sum _{\mathrm{t}=1}^{\mathrm{T}}\frac{{\mathrm{C}}_{\mathrm{energy},\mathrm{pre},\mathrm{t}}}{{(1+\mathrm{r})}^{\mathrm{t}}}}$$

(9)

$$\mathrm{L}\mathrm{C}{\mathrm{C}}_{\mathrm{post}}={\displaystyle \sum {\mathrm{C}}_{\mathrm{i}}}+{\displaystyle \sum _{\mathrm{t}=1}^{\mathrm{T}}\frac{{\mathrm{C}}_{\mathrm{energy},\mathrm{post},\mathrm{t}}+{\mathrm{C}}_{\mathrm{maintenance},\mathrm{t}}}{{(1+\mathrm{r})}^{\mathrm{t}}}}$$

(10)

$$\mathrm{P}\mathrm{W}\mathrm{F}={\displaystyle \frac{1-{(1+\mathrm{r})}^{-\mathrm{T}}}{\mathrm{r}}}$$

(11)

where ΔLCC: increment in life cycle cost after retrofit (CNY); LCC_pre: life cycle cost before retrofit (CNY), i.e., the present value of heating and cooling energy consumption costs for the sample building prior to retrofit; LCC_post: life cycle cost after retrofit (CNY), i.e., the sum of the present value of the initial investment cost of the retrofit measures and the heating and cooling energy consumption costs throughout the life cycle; C_i: initial investment cost of the i-th retrofit measure (CNY); C_energy,pre,t: heating and cooling energy consumption cost in the t-th year before retrofit (CNY); C_{energy,post,t}: heating and cooling energy consumption cost in the t-th year after retrofit (CNY); C_{maintenance,t}: maintenance or replacement cost in the t-th year (CNY); PWF (Present Worth Factor): calculated based on relevant research methods and parameter values [63], PWF = 13.59; r: discount rate, used to convert future costs into present value (typically referencing interest rates); and T: analysis period (unit: years).

If ΔLCC > 0, it suggests that the retrofit plan is economically unfeasible; if ΔLCC < 0, it indicates that the retrofit plan is economically feasible, with smaller ΔLCC values (larger absolute negative values) corresponding to greater economic benefits over the life cycle.

3.3.4. Optimization Variables

In the Qianbei Region, villagers’ understanding of house construction primarily stems from the building practices of neighboring households and recommendations by local artisans. Due to information isolation and economic constraints, local residents have limited familiarity with energy-saving standards and related construction schemes, leading to inadequate implementation of these standards in rural dwellings. Consequently, there is considerable potential for enhancing both building energy efficiency and indoor thermal environment performance. When retrofitting existing rural dwellings without compromising structural integrity or functional layout, design optimization can be achieved mainly through the addition of insulation layers to walls and roofs, as well as the replacement of windows with energy-efficient alternatives. Based on the living habits, energy demands, and economic capacity of local residents, this study defines optimization variables for the building envelope retrofit (

Table 6. Optimization variable parameter settings.

Name		Variable Type	Range	Step Size	Unit Price
Type of Thermal Insulation Materials	EPS Board 1	Discrete Variable	0.039 W/(m2·K)	—	400 CNY/m3
	XPS Board 2		0.030 W/(m2·K)		420 CNY/m3
	RW Board 3		0.041 W/(m2·K)		380 CNY/m3
	PU Board 4		0.024 W/(m2·K)		600 CNY/m3
Thickness of Thermal Insulation Layer	Exterior walls	Continuous Variable	20–150 mm	10 mm	—
	Roof
Type of Exterior Windows	5 mm Low-E Glass, PVC Frame 5	Discrete Variable	3.4 W/(m2·K))	—	280 CNY/m2
	5 mm Clear Glass + 9A (Air Gap) + 5 mm Clear Glass, Standard Aluminum Frame		3.0 W/(m2·K)		360 CNY/m2
	5 mm Clear Glass + 9A (Air Gap) + 5 mm Clear Glass, PVC Frame		2.8 W/(m2·K)		390 CNY/m2
	5 mm Clear Glass + 9A (Air Gap) + 5 mm Clear Glass, Thermal Break Aluminum Frame		1.9 W/(m2·K)		480 CNY/m2
	5 mm Low-E Glass + 9A (Air Gap) + 5 mm Clear Glass, PVC Frame		2.0 W/(m2·K))		450 CNY/m2
Indoor Ceiling	9 mm Gypsum Board	Discrete Variable	0/1 (Boolean Value)	—	40 CNY/m2

¹ EPS is the Expanded Polystyrene, ² XPS is the Extruded Polystyrene, ³ RW is the Rock Wool, ⁴ PU is the Polyurethane, ⁵ PVC is the Polyvinyl chloride.

). The cost estimation for the retrofit measures is derived from the “Guizhou Province Construction Project Cost Information”, Issue 01, 2025 [64].

4. Results and Analysis

4.1. Analysis of Field Test Results

4.1.1. Outdoor Test Results

Figure 6 illustrates the characteristic diurnal variations of outdoor climate parameters in the Qianbei Region during winter and summer. In summer, outdoor air temperature exhibits pronounced diurnal fluctuations, with a mean daily temperature of 27.8 °C and a diurnal amplitude of 11.2 °C. Relative humidity shows an inverse relationship with temperature, peaking at 88% (03:00) due to radiative cooling at night and decreasing to 42% (15:00) during daytime solar heating, forming a distinct diurnal alternation pattern between “high-temperature/low-humidity” and “low-temperature/high-humidity” phases. The combined effects of low temperature and high humidity characterize winter climatic conditions. Thermally, the mean daily temperature is 4.8 °C with a diurnal amplitude of 10.8 °C. Minimum temperatures drop to −0.6 °C (05:00), while the daytime maximum of 10.2 °C (16:00) remains below the ASHRAE-defined thermal comfort threshold (18 °C). Geometrically, the average relative humidity is 75%, with near-saturation levels (90% RH) persisting for six nocturnal hours (23:00–05:00), confirming the region’s “cold-humid” winter climate signature. As shown in Figure 6, both seasonal temperature-humidity combinations deviate significantly from the ASHRAE Standard 55 comfort envelope (operative temperature: 18–25 °C; relative humidity: 40–60%).

4.1.2. Indoor Test Results

Figure 7 illustrates the pronounced seasonal variations in the indoor thermal environment of rural dwellings in the Qianbei Region. In summer, indoor temperatures exhibit significant fluctuations, peaking at 37.8 °C (14:00), well above the upper limit of the comfort zone (28 °C), with a diurnal temperature range of 13.9 °C. The high-temperature period (>30 °C) lasts for over 8 h, combined with an average relative humidity ranging from 82.1% to 98.5%, creating a dual stress of “high temperature and high humidity”. In winter, indoor temperatures remain below the lower comfort threshold (16 °C) throughout the day, with nighttime lows reaching 10.9 °C. Concurrently, relative humidity fluctuates between 82.1% and 90.7%, with near-saturation levels at night exacerbating the risk of condensation on building envelopes. Field measurements indicate that natural ventilation efficiency is low in summer (average daily wind speed of 0.6 m/s, with only 12% of the time meeting the minimum comfort criterion of 0.3 m/s). In winter, tightly sealed doors and windows result in negligible wind speeds (≤0.2 m/s), contributing to issues such as heat and humidity retention in summer and air stagnation.

4.1.3. Results of the Questionnaire Survey

A total of 103 and 131 questionnaires were distributed to local dwellings in summer and winter, respectively, with effective recovery rates of 93.2% (n = 96) and 95.4% (n = 125). The results (Figure 8) reveal a significant seasonal disparity in residents’ thermal sensation distributions. During summer, sensations are predominantly within the range of 0 (neutral) to +2 (hot), whereas in winter, they cluster between −2 (cold) and 0 (neutral). Regarding humidity perception, summer responses are primarily concentrated between 0 (neutral) and +1 (slightly dry), while winter responses span from −1 (slightly wet) to +1 (slightly dry). The survey further indicates that residents exhibit greater sensitivity to low winter temperatures compared to high summer temperatures. Additionally, the elderly group demonstrates significantly higher climate tolerance than the younger cohort, highlighting population heterogeneity in thermal comfort evaluation. Overall, residents report higher levels of comfort and satisfaction with the summer environment but express notable dissatisfaction with the indoor conditions during winter.

4.1.4. Analysis of Indoor Thermal Comfort

The thermal comfort of local dwellings was evaluated using empirical measurements and survey data. During the measurement period, observations of the occupants revealed that their indoor activities were predominantly sedentary, which led to the application of a metabolic rate of 1.2 met (light activity in a sitting posture) for the PMV calculation. Statistical analysis (Figure 9) indicated a thermal resistance of approximately 0.5 clo in summer and 1.3 clo in winter for the clothing of residents. The environmental parameters (temperature, humidity, wind speed, and mean radiant temperature) were derived from empirical measurements.

Figure 10 illustrates the diurnal variation characteristics of APMV in rural dwellings in the Qianbei Region: a unimodal distribution during summer, peaking at 1.8 (significant overheating) at 15:00, with 23.5% of the time exceeding 1.0, and a trough value of −1.2 (overcooling) at 07:00. The intense daily temperature variations in summer result from the thermal inertia of the valley’s topography. However, during the early morning (08:00–10:00) and evening (17:00–19:00), it will attain a relatively comfortable condition; in winter, owing to the restricted local heating capacity of the electric stove, the APMV will persistently remain negative (−0.8 to −0.4) throughout the day, with the minimum value occurring at 07:00. Analysis of thermal defects indicates that during summer, the thermal mass of the brick–concrete exterior wall results in elevated temperatures, while insufficient insulation in winter intensifies nocturnal heat loss. The interplay of these two factors results in a substantial deviation of the APMV from the adaptive comfort zone (−0.5 ≤ APMV ≤ 0.5).

4.2. Verification and Analysis of Simulation Results

Figure 11 illustrates the recorded and projected indoor temperatures during winter and summer. The MAE and RMSE between the simulated and measured indoor air temperatures in the sample building are computed using Equations (4) and (5), respectively. The findings indicate that the MAE is 1.06 °C and the RMSE is 1.38 °C during summer conditions, while the MAE is 1.39 °C and the RMSE is 1.66 °C in winter conditions. The existing errors fall within an acceptable range, satisfying the accuracy criteria for indoor temperature simulation, thereby demonstrating that the simulation software and parameter configurations are both scientific and dependable.

4.3. Analysis of Multi-Objective Optimization Results

4.3.1. Pareto Optimal Solutions and Distribution in Objective Space

Figure 12a illustrates the nonlinear trade-off relationship among E, TDT, and ΔLCC within a three-dimensional objective space framework. Analyzing the Pareto front to identify the optimal solution with superior energy efficiency (Figure 12b) demonstrates the viability of multi-objective collaborative optimization within defined energy efficiency parameters. Figure 13a–c illustrate the two-dimensional trade-off relationships between two variables within the three-dimensional objective space, respectively. Figure 13d–f illustrate the algorithm’s convergence from the initial generation (First Gen) to the final generation (Last Gen) through the variation in the standard deviation of E, TDT, and ΔLCC.

The parallel coordinate plot in Figure 14 elucidates the multi-dimensional trade-off mechanism among E, TDT, and ΔLCC. E, TDT, and ΔLCC exhibit significant negative correlations, indicating that the adoption of energy-saving technologies and improvements in thermal environment quality come at the cost of higher initial investment. Additionally, E and TDT show positive correlations, suggesting that energy consumption is optimized concurrently with the reduction in thermal discomfort duration during the optimization process.

4.3.2. Comparative Analysis of the Performance of Optimal Solution Transformation Schemes

Table 7. Multi-objective optimization results.

Objective		E (kWh/m2)	Rank	TDT (h)	Rank	ΔLCC (CNY)	Rank	Retrofit Cost (CNY)
Sample building		96.41	-	6173	-	-	-	-
Scheme 1	Optimal Solution for E 1	42.40	1/5000	5092	588/5000	−29,566.94	4894/5000	47,073.58
Scheme 2	Optimal Solution for TDT 2	42.87	501/5000	5078	1/5000	−27,966.58	4969/5000	48,007.11
Scheme 3	Optimal Solution for ΔLCC 3	45.23	1979/5000	5183	2003/5000	−56,329.87	1/5000	16,297.20
Scheme 4	The average value of the fitness ranking’s optimal solution	43.95	1366/5000	5139	1571/5000	−56,158.32	324/5000	18,274.56

¹ E is the building energy consumption, ² TDT is the thermal discomfort time, ³ ΔLCC is the Life Cycle Cost Increment.

presents the objective function values for each specialized optimal solution, along with their rankings within the solution set. This highlights the dynamic equilibrium characteristics of the Pareto solution set in balancing energy consumption, comfort, and economy.

• The retrofit scheme 1 for achieving the optimal E value of the building envelope incorporates 150 mm rock wool board on the exterior wall, 150 mm XPS board on the roof, 5Low-E + 9A + 5 transparent glass with a PVC frame, and a gypsum board ceiling. This scheme achieves the lowest E value of 42.40 kWh/m², resulting in a 56.0% energy saving compared to the sample building (E = 96.41 kWh/m²). Its TDT is 5092 h (ranked 588/5000), representing a 17.5% reduction from the sample building. The ΔLCC is −29,566.94 CNY (ranked 4894/5000), indicating potential economic benefits. However, the retrofit cost is 47,073.58 CNY, underscoring that achieving high energy efficiency requires a significant initial investment.
• The retrofit scheme 2 for achieving the optimal TDT value of the building envelope comprises 150 mm rock wool board on the exterior wall, 150 mm XPS board on the roof, 5Low-E + 9A + 5 transparent glass with a broken-bridge aluminum frame, and a gypsum board ceiling. This scheme achieves the lowest TDT value of 5078 h, reducing it by 17.7% compared to the sample building (TDT = 6173 h). It demonstrates the best thermal comfort performance among all schemes. Its E value is 42.87 kWh/m² (ranked 501/5000), close to Scheme 1 (42.40 kWh/m²), maintaining high energy efficiency. The ΔLCC is −27,966.58 CNY (ranked 4969/5000), with a retrofit cost of 48,007.11 CNY, highlighting that enhancing thermal comfort also requires a substantial initial investment.
• The retrofit scheme 3 for achieving the optimal ΔLCC value of the building envelope involves 20 mm XPS board on the exterior wall, 110 mm XPS board on the roof, 5Low-E glass with a PVC frame, and no ceiling. This scheme achieves the optimal ΔLCC value of −56,329.87 CNY, with a retrofit cost of only 16,297.20 CNY. Its E value is 45.23 kWh/m² (ranked 1366/5000), achieving a 53.1% energy saving compared to the sample building. The TDT is 5183 h (ranked 2003/5000), improving thermal comfort by 16.0%. This scheme realizes economic optimization through moderate reductions in thermal performance, reflecting a typical “low-cost, low-performance” Pareto frontier characteristic.
• Based on the TOPSIS multi-objective decision-making model, the retrofit scheme 4 was selected based on the optimal Average of Fitness Ranks (where no other solution outperforms it across all objectives). The balanced optimal solution corresponds to the 34th individual of the 89th generation (Figure 15). This scheme employs a strategy of 20 mm XPS board on the exterior wall, 150 mm XPS board on the roof, 5Low-E glass with a PVC frame, and no ceiling, aiming to balance E, TDT, and ΔLCC. Its E value is 43.95 kWh/m² (ranked 1366/5000), representing a 54.4% reduction compared to the sample building. The TDT is 5139 h (ranked 1571/5000), reducing it by 16.8%. The ΔLCC is −56,158.32 CNY (ranked 324/5000), with a retrofit cost of 18,274.56 CNY. This solution ranks 1148.33 in terms of fitness (average), illustrating a dynamic balance among the three-dimensional objectives of energy efficiency, comfort, and economy.

4.3.3. Comparative Analysis of Energy Consumption and Thermal Comfort Performance Before and After Retrofit in the Sample Building

Figure 16 and Figure 17 present a comparative analysis of E and TDT between the envelope retrofit scheme based on the balanced optimal solution and the sample building. In terms of energy consumption, the monthly energy intensity of the renovated building exhibits a systematic reduction. Specifically, in July, the peak cooling demand month, energy consumption decreased from 11.02 kWh/m² under baseline conditions to 5.73 kWh/m², representing a 47.9% reduction. In January, dominated by heating demand, energy consumption dropped from 16.45 kWh/m² to 8.89 kWh/m², achieving a 45.9% reduction. After cumulative annual calculations, the comprehensive energy-saving rate of the retrofit scheme reaches 54.4%. Regarding thermal comfort, the PMV values of the sample building during summer (June to August) primarily range from “hot” (PMV > +1.0) to “warm” (+0.5 < PMV ≤ +1.0), while in winter (December to February), they are concentrated within the range of “cold” (−1.5 < PMV ≤ −1.0) to “cool” (−0.5 < PMV ≤ 0). Post-retrofit, the PMV distribution significantly converges toward the neutral range (−0.5 ≤ PMV ≤ +0.5), reducing the annual duration of thermal discomfort from 6173 h for the sample building to 5139 h, a decrease of 16.8%.

5. Discussion

The study investigated strategies and techniques for systematically addressing the intricate conflicts among “energy consumption, thermal comfort, and affordability” in the retrofitting of building envelopes in rural dwellings within the Qianbei Region through multi-objective optimization methods. Nonetheless, in promoting these initiatives locally, it is essential to devise differentiated implementation strategies tailored to regional variances and specific priorities. These may be categorized into the following primary domains:

• Priority of energy efficiency (prevalence of energy conservation awareness): This option targets the lowest energy consumption (42.40 kWh/m², 56% more energy efficient than the baseline building), incurring higher initial retrofit expenses (47,073.58 CNY), yet yielding substantial long-term energy savings (ΔLCC = −29,566.94 CNY). It is appropriate for regions with elevated electricity tariffs or fuel expenses. The advantages of energy conservation can rapidly offset the retrofit expenses. Integrating energy-saving technologies at the outset of a building’s entire life cycle planning is advisable to prevent the extra expenses associated with piecemeal retrofits in the future.
• Priority of comfort (motivated by health requirements): This solution targets the minimal TDT (5078 h, representing a 17.7% decrease). The upfront retrofit expense is substantial (48,007.11 CNY), yet the long-term energy savings are considerable (ΔLCC = −27,966.58 CNY). It can diminish long-term energy consumption costs (42.87 kWh/m²) by decreasing the frequency of heating and cooling system usage, while enhancing residents’ health and quality of life, thereby indirectly lowering medical and living expenses. It is appropriate for mountainous regions characterized by cold, humid winters or hot, humid summers. It must prioritize user experience and be implemented initially in households of the elderly and dwellings of left-behind children. The conventional notion of prioritizing economy over comfort must be dispelled by enhancing the intuitive thermal environment, such as by decreasing the indoor temperature differential by 5–8 °C.
• Priority of economy (short-term cost constraints): This solution seeks optimal cost-effectiveness (ΔLCC = −56,329.87 CNY). The retrofit expense amounts to 16,297.20 CNY, with energy savings of 53.1% (45.23 kWh/m²) and a 16.0% reduction in TDT. This strategy compromises energy efficiency and thermal comfort, making it appropriate for impoverished countries or villages with constrained budgets, prioritizing the fulfilment of fundamental living requirements. During the initial phases of technology promotion, economical solutions are employed to foster trust and engagement among residents, highlighting “affordable immediate advantages” (e.g., a 3–5 year return on investment).
• Balanced strategy (dynamic adaptive orientation): The balanced solution (E = 43.95 kWh/m², a 54.4% reduction; TDT = 5139 h, a 16.8% reduction; ΔLCC = −56,158.32 CNY) achieves multi-dimensional synergistic optimization between energy consumption, comfort, and economy (mean fitness ranking 1148.33), with a moderate retrofit cost (18,274.56 CNY), to achieve a global balance of comprehensive performance. This strategy is relevant to two contexts: firstly, as a universal benchmark framework for retrofitting; secondly, in extensive government-initiated retrofits, where consistent technical standards and subsidy policies are essential to mitigate the risk of technological fragmentation.

6. Conclusions

This study explores the coordinated optimization path for retrofitting the building envelopes of rural dwellings in the Qianbei Region based on a “measurement-simulation-optimization” research framework. Through actual measurements of the thermal environment in winter and summer and questionnaire surveys, the adaptive thermal comfort index (APMV) was quantitatively analyzed. A dynamic simulation model of the building was constructed using a parametric modeling tool chain (Rhino-Grasshopper-Ladybug/Honeybee). Finally, E, TDT, and ΔLCC were used as optimization objectives for multi-objective optimization of rural dwellings in the Qianbei Region. The main conclusions are as follows:

• The thermal deficiencies of the building envelopes of rural dwellings in the Qianbei Region are obvious. Indoor temperature and humidity in summer are high (peak 37.8 °C/98.5% RH), and in winter, they are low (minimum 10.9 °C/90.7% RH). During the test period, thermal discomfort accounted for more than 70% of the time. Questionnaire surveys showed that the local residents’ demand for retrofits showed significant seasonal differences. They were relatively satisfied with the indoor thermal environment in summer, but expressed a strong willingness to retrofit the indoor thermal environment in winter.
• The Pareto optimal solution set shows that retrofitting the building envelope can achieve a maximum reduction in E of 56% (96.41 → 42.4 kWh/m²), a reduction in TDT of 17.7% (6173→5078 h), and a ΔLCC of −56,329.87 CNY (net profit). The use of low-cost measures such as XPS boards and 5Low-E glass can achieve a positive economic return without policy subsidies.
• The retrofit scheme that balances the optimal solution achieves a dynamic balance between E (43.95 kWh/m²), TDT (5139 h), and ΔLCC (−56,158.32 CNY). Compared to the reference building, it has a 54.4% lower E, 16.8% lower TDT, and a negative ΔLCC (net benefit). This solution balances economy and performance improvement, and is suitable for ordinary farmers who have no clear preferences but need comprehensive improvement. It can be used as a benchmark for multi-objective coordinated optimization.

The retrofit of rural dwellings in the Qianbei Region needs to focus on “low cost and high efficiency”, and give priority to the promotion of mature technologies such as insulation and energy-saving exterior windows. Combining the climate characteristics of “cold and damp winters, hot and humid summers” and the economic level of rural areas in the Qianbei Region, and through the coordinated optimization of “energy conservation and consumption reduction, comfort improvement, and economic viability”, a low-cost retrofit strategy with stronger regional adaptability is proposed, which verifies the scientific nature of the closed-loop logic of “environmental perception-model iteration-decision optimization”, and provides a replicable technical path for the green retrofit of rural dwellings in high-humidity climate zones.