Climate Adaptation of Folk House Envelopes in Xinjiang Arid Region: Evaluation and Multi-Objective Optimization from Historical to Future Climates

Abstract

Under intensifying global warming and extreme climate events, the climate adaptability of folk houses in Xinjiang’s arid regions faces critical challenges. However, existing studies predominantly focus on traditional folk houses under current climate conditions, neglecting modern material hybrids and long-term performance under future warming scenarios. This study develops a data-driven framework to assess and enhance building envelope performance across historical-to-future climate conditions (2007–2021 TMY data, 2024 observations, and 2050/2080 SSP3–7.0 projections) using the entropy-weighted TOPSIS method and NSGA-II algorithm. Analyzing rammed earth, brick–wood, and brick–concrete folk houses in Kashgar, Hotan, Kuqa, and Turpan, the optimization targets thermal discomfort hours (TDHs), heating energy consumption (HEC), and net present value (NPV). The results demonstrate optimized solutions achieve 30–60 year climate resilience, reducing HEC by 51.54–84.76% (43.02–125.78 kW·h/m²·a) compared to baseline buildings, TDH by 15–52.93% (301–1236 h) in arid Zone A and by 5.54–10.8% (208–352 h) in the extreme hot-arid Zone B (Turpan), and NPV values by CNY 31,000–85,000. Rammed earth constructions demonstrate superior performance in Zone A, while brick–concrete exhibits optimal extreme hot-arid adaptability, and brick–wood requires prioritized retrofitting. The findings advocate revising China’s design standards to address concurrent winter overcooling and summer overheating risks under future warming. This work establishes a climate-resilient optimization paradigm for arid-region folk houses, advancing energy efficiency and thermal comfort.

1. Introduction

Driven by economic and demographic expansion, anthropogenic greenhouse gas emissions have reached unprecedented levels over at least the past 800,000 years [1]. As documented by the Intergovernmental Panel on Climate Change (IPCC), global warming is unequivocal, with post-1950s climatic alterations exhibiting unprecedented magnitudes across millennia [2]. This trajectory not only reshapes natural ecosystems but also directly endangers human health: extreme heat events in temperate zones are intensifying in frequency, severity, and duration, with projections indicating substantial increases under continued warming [3]. Elevated indoor temperatures exacerbate public health risks, including mortality, while impairing residents’ productivity and cognitive performance [4,5].

In response to this global crisis, decarbonization has emerged as an international priority. China’s “Dual Carbon” strategy (peaking emissions by 2030 and achieving carbon neutrality by 2060) imposes urgent demands on the construction sector, which accounts for 20% of national energy consumption. Within this framework, folk houses—representing substantial building stock with high energy-saving potential—constitute a critical focus for emission mitigation. However, since the 1950s, the global arid zone area has doubled due to climate warming [6]. China’s arid zones, spanning 25% of its terrestrial area and representing the Northern Hemisphere’s highest-latitude drylands [7], concentrate over 90% of regional socioeconomic wealth within merely 5% of oasis zones [8]. Xinjiang’s arid territories (1.3869 million km², 83.3% of the region) encompass half of China’s total drylands, with over 50% of oases experiencing desertification. Local populations historically developed climate-resilient architectures through passive design strategies (e.g., earth thermal mass utilization and courtyard shading). Yet, with accelerating economic development and intensifying climate warming, local folk houses increasingly fail to meet contemporary comfort standards, urgently requiring scientifically informed guidance for their retrofitting and modernization.

Research on folk houses in global arid zones exhibits distinct regional concentration across three primary areas: West Asia, North Africa, and China. In West Asia, scholars have intensively investigated traditional architectures in Iran, Oman, and Turkey: M. Khalili uncovered passive strategies balancing summer energy efficiency and thermal comfort in Iranian arid-region buildings [9]; Sanaz Saljoughinejad and Siavash Rashidi Sharifabad systematized spatial hierarchies of climate-responsive design in Iranian vernacular housing [10]; Rana Soleymanpour demonstrated traditional dwellings’ superior indoor comfort through climate-adaptive design compared to modern counterparts [11]; Nazanin Nasrollahi et al. quantified the microclimatic impacts of geometric parameters in Iranian courtyards using ENVI-met 4.0 beta, 2015 simulations [12]; Alalouch et al. identified traditional wisdom’s energy-saving potential in Omani dwellings [13]; Ayhan Bekleyen and Yahya Melikoğlu revealed thermal regulation through high-capacity earth materials in Şanlıurfa’s windcatchers [14]; Gülmüş et al. classified 47 Haran earthen houses in Şanlıurfa into six typologies, systematically evaluating environmental comfort through ClimateStudio (2020) thermal–visual simulations and SolidWorks (2024) CFD ventilation analysis [15]; and Masoud NASOURI et al. optimized solar-integrated traditional designs for enhanced Iranian dwelling performance [16]. North Africa studies focus on North African innovations: Marwa Dabaieh et al. examined optimal cooling through Egypt’s reflective arched roofs [17]; Belkhir Hebbal et al. quantified 2.34 m-depth underground architecture, reducing Algerian cooling loads by 4.5 kWh/m² [18]; and Khechiba et al. confirmed traditional buildings’ dual energy–comfort advantages in Ouargla [19]. China’s Turpan-focused research reveals the following: Haiyan Yan documented spatiotemporal adaptation strategies [20]; Yang Liu modeled residents’ thermal resilience [21]; Chang Sha quantified semi-open spaces’ temporal thermal complementarity [22]; Junkang Song and Wanjiang Wang achieved multi-objective low-carbon retrofitting [23]; Wenfang He et al. decoded historical climate-buffering spatial layering [24]; and Weiqin Gou synergized cultural preservation with functional upgrades in rammed-earth renovations [25]. Collectively, these studies demonstrate arid-zone folk houses’ unique environmental adaptation system through spatial configuration, material properties, and passive strategies, offering critical insights for sustainable contemporary design.

A systematic review of extant arid-zone dwelling studies (

Table 1. Comparison of related research on folk houses in arid zones.

Research Subject	Climate Data Types	Study Areas (Region: Country-City)
Traditional folk houses	TMY climate data	West Asia: Turkey—Şanlıurfa [15]
		China—Turpan [24,25]
	Observed climate data	North Africa: Algeria—Ghardaïa [26]
		West Asia: Iran—multiple cities [9], Iran—multiple cities [10], Iran—Shiraz [12], Turkey—Şanlıurfa [14]
		China-Turpan [20,21,22]
Modern folk houses	TMY climate data	North Africa: Algeria—Ouargla [17], Egypt—Cairo [18]
		West Asia: Oman—multiple cities [13], Iran—Bushehr [16], Oman—multiple cities [27], Iran—multiple cities [28]
		China—Turpan [23]
Traditional folk houses vs. modern folk houses	Observed climate data	North Africa: Algeria—Ouargla [19]
		West Asia: Iran—multiple cities [11]

) reveals a predominant focus on static climate analyses of traditional typologies, with limited attention paid to modern counterparts or comparative evaluations. For instance, while Khalili et al. demonstrated 18–25% summer energy savings via passive cooling in Iranian vernacular buildings, their reliance on historical climate data neglects future climatic uncertainties, compromising long-term resilience assessments [9]. Geographically, Chinese research disproportionately emphasizes Turpan’s extreme hot-arid region (Yan et al.; Liu et al.), overlooking climatic diversity across Xinjiang’s other arid regions (e.g., Kuqa and Hotan) [20,21,22,23,24,25]. This spatial bias leaves critical knowledge gaps regarding non-extreme hot-arid zones in Xinjiang.

Therefore, through a comparative analysis of the relevant literature, this study identifies the research gaps in existing studies as follows:

• Temporal narrowness: Existing methods for optimizing thermal comfort and energy demand rely solely on the Typical Meteorological Year (TMY) and current climate data, raising concerns about their feasibility under future climate scenarios.
• Geographic imbalance: Geographic overemphasis on Turpan’s extreme hot-arid climate neglects the moderate hot-arid conditions predominant in most of Xinjiang, creating research gaps for folk houses in other sub-regions.
• Type bias: Most related studies emphasize traditional construction while inadequately addressing modern-material folk houses now proliferating across the region.

To bridge the knowledge gaps revealed through prior investigations, this study pioneers the following contributions:

• Temporal scope expansion: This study analyzes climate impacts across historical (2007–2021 TMY data), current (2024 observed data), and future periods (2050/2080 SSP3–7.0 projected data), enabling comprehensive performance evaluation and optimization under evolving climate conditions.
• Geographic and type diversity: The investigation covers rammed earth, brick–wood, and brick–concrete folk houses in Kashgar, Hotan, Kuqa, and Turpan, ensuring widely applicable research conclusions and outcomes.
• Integrated optimization framework: Combining the entropy-weighted TOPSIS method with the NSGA-II algorithm, this study achieves efficient multi-objective optimization to ensuring climate resilience in the optimized solutions.

2. Materials and Methods

The methodology framework is shown in Figure 1.

The methodological framework comprises three stages: (1) Case study description—Case study sites and folk houses were identified through the literature review and field investigations. Construction techniques, structural technologies, and climate data were collected, followed by a summarization of folk houses’ forms and tectonic features. Future climate projections were subsequently generated for the sites. (2) Simulation—Rhino–Grasshopper models calibrated by monitoring data were used to simulate thermal discomfort hours (TDHs) and heating energy consumption (HEC) using the TMYx (2007–2021), 2024 observed, and 2050/2080 projected datasets. The climate adaptability of building envelopes was evaluated through comparative analysis with two controls: national-standard reference building HEC and Climate Consultant 6.0-derived optimal passive strategy TDHs. (3) Multi-objective optimization—global sensitivity analysis guided variable selection for NSGA-II-based Wallacei optimization, minimizing TDH/HEC while maximizing the net present value (NPV), followed by entropy-weighted TOPSIS evaluation to determine the optimal solution by ranking Pareto solutions.

2.1. Case Study Description

2.1.1. Study Sites and Folk House Typology

In the arid regions of Xinjiang, traditional folk houses have evolved through climatic adaptation, utilization of local materials, and cultural practices, resulting in four distinctive architectural typologies (Figure 2): the Ayiwan-style in Hotan, the Pixiayiwan-style in Kashgar, the high trellis-style in Turpan, and the hybrid-style in Kuqa. These four cities—Hotan, Kashgar, Turpan, and Kuqa—epitomize centuries-old architectural paradigms shaped by environmental constraints and socio-cultural traditions [29].

As major administrative and economic hubs, these cities collectively influence over 80% of oasis settlements in Xinjiang’s arid zone, making their building practices representative of broader regional conditions. Geographically, their locations span Xinjiang’s arid belt from west to east (Kashgar to Turpan) and north to south (Kuqa to Hotan), ensuring coverage of diverse microclimates. Due to this comprehensive spatial and typological representation, these four cities are selected as case study sites for this research.

With the advancement of China’s New Rural Construction initiatives, modern materials (concrete and clay bricks) create brick–concrete/wood hybrids. He Ping Wen et al. (2016) classified rural houses into four structural types: earth–wood, brick–wood, brick–concrete, and masonry (seismic-resistant), based on 253 regional samples in Xinjiang [30]. Their 2021 study further categorized historic urban/suburban houses into earth–wood, brick–wood, and brick–concrete types using 42 historic urban/suburban samples in Xinjiang, with both studies reporting predominantly single-story structures [31].

This study is based on a systematic review of prior research conducted by the same team, and field surveys of nearly 500 dwellings in representative villages across four regions—Turpan, Aksu, Kashgar, and Hotan—carried out by the authors and research team members, revealed prevalent envelope materials: raw earth bricks and solid/perforated clay bricks. Roofs predominantly use wood-frame dense beams or reinforced concrete; windows combine glass–wood and aluminum–plastic composites; and doors are wooden with clay brick floors. Spatial layouts (linear, L-shaped, and U-shaped) show 91% south-facing orientations [32].

This study categorizes Xinjiang folk houses into three types: rammed earth (Type A), brick–wood (Type B), and brick–concrete (Type C).

Table 2. The typical plan layout and exterior envelopes of folk houses.

Architectural Typology	Typical Plan Layout	Building Envelope Components	Construction Component Name
Pixiayiwan-style		Roof	Wooden truss with dense beam flat roof
			Reinforced concrete roof
High trellis-style		Wall	Rammed earth wall
			Solid clay brick wall
			Perforated clay brick wall
Ayiwan-style		Window	Wooden-framed glass window
			Aluminum–plastic composite window
Hybrid-style	-	Ground	Clay brick ground
			Cement ground

details regional layouts and envelope specifications. Furthermore, a representative rammed earth folk house in Kashgar was selected as the case study. On-site measurements and indoor thermal environment monitoring were conducted in July 2024, with the case building’s model and floor plan illustrated in Figure 3.

2.1.2. Climate Data Sources

The 2007–2021 Typical Meteorological Year (TMY) data for four regions were sourced from EnergyPlus; the 2024 observed data compiled hourly from the China Weather Network include the dry bulb temperature, precipitation, wind direction/speed, air pressure, relative humidity, visibility, cloud cover, dew point temperature, and radiation components (shortwave, direct, diffuse, and total horizontal).

Some tools have been developed based on statistical downscaling methods to predict future weather data under different emission scenarios. For example, CCWorldWeatherGen can provide weather files based on the SRES A2 scenario [33]. WeatherShift can generate weather files according to the RCP scenarios (RCP4.5 and RCP8.5) [34], while Meteonorm generates weather files based on the RCP2.6, RCP4.5, and RCP8.5 scenarios. Future Weather Generator is a free and open-source tool that can generate future weather data based on the latest emission scenarios released by the IPCC-SSP (SSP1–2.6, SSP2–4.5, SSP3–7.0, and SSP5–8.5) [35], and this tool has been widely used in the field of future meteorological data prediction [36,37,38]. Therefore, this study chose this tool to predict the weather data for 2050 and 2080. The four SSP scenarios reflect different climate mitigation targets: SSP1–2.6 (low emission), SSP2–4.5 (medium emission), SSP3–7.0 (medium-high emission), and SSP5–8.5 (high emission) [39]. Among them, the SSP3–7.0 scenario, a medium-high emission scenario, predicts that global carbon dioxide emissions will roughly double by 2100 compared to current levels, and the average temperature will rise by 3.6 degrees Celsius by the end of this century, representing a regional competitive path [35].

The SSP3–7.0 scenario was selected for future meteorological predictions due to escalating extreme climate events in China and globally. This medium-high emission pathway provides actionable insights for building energy reduction—demonstrating effectiveness here implies applicability to low-emission scenarios, aligning with China’s energy conservation commitments and supporting sustainable development goals.

2.2. Simulation

The typical folk house in the case study was modeled using Grasshopper 1.0+ Ladybug Tools 1.8.0 (Rhino 7.0) to simulate the HEC and non-heating season indoor temperatures (converted to TDH; see Section 2.2.5). These metrics evaluated the building envelope climate adaptability in Xinjiang’s arid region under historical (2007–2021), current (2024), and future (2050/2080 SSP3–7.0) meteorological data, with parameters calibrated against actual usage patterns. Figure 3 illustrates the model and layout, with detailed parameters described below.

2.2.1. Model Validation

To ensure simulation accuracy, field monitoring was performed from 1 to 31 July 2024, using three JT2013 Wet Bulb Globe Temperature (WBGT) Index Meters (Beijing Jantytech Co. Ltd., Beijing, China) installed 1.1 m above floor level in case building rooms (Figure 3), avoiding direct sunlight and recording indoor temperatures at 10 min intervals. Model validation employed Pearson’s correlation coefficient (R), the regression line slope (k), and the root mean square error (RMSE) derived from the July 2024 measurements. High accuracy is indicated when R approaches 1, while k and RMSE approaching 0 signify superior precision, calculated via Equations (1)–(3).

$$\mathrm{P}\mathrm{e}\mathrm{a}\mathrm{r}\mathrm{s}\mathrm{o}\mathrm{n}\mathrm{’}\mathrm{s} \mathrm{R}={\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}\left(\mathrm{M}\mathrm{e}\mathrm{a}{\mathrm{s}}_{\mathrm{i}}+\overline{\mathrm{M}\mathrm{e}\mathrm{a}\mathrm{s}}\right)\left(\mathrm{S}\mathrm{i}\mathrm{m}+\overline{\mathrm{S}\mathrm{i}\mathrm{m}}\right)}{(\sqrt{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\left(\mathrm{M}\mathrm{e}\mathrm{a}{\mathrm{s}}_{\mathrm{i}}+\overline{\mathrm{M}\mathrm{e}\mathrm{a}\mathrm{s}}\right)}^2})(\sqrt{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\left({\mathrm{S}\mathrm{i}\mathrm{m}}_{\mathrm{i}}+\overline{\mathrm{S}\mathrm{i}\mathrm{m}}\right)}^2}}}$$

(1)

$$\mathrm{S}\mathrm{i}\mathrm{m}=\mathrm{k}·\mathrm{M}\mathrm{e}\mathrm{a}\mathrm{s}+\mathrm{i}\mathrm{n}\mathrm{t}$$

(2)

$$\mathrm{R}\mathrm{M}\mathrm{S}\mathrm{E}={\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}\left(\mathrm{M}\mathrm{e}\mathrm{a}{\mathrm{s}}_{\mathrm{i}}+{\mathrm{S}\mathrm{i}\mathrm{m}}_{\mathrm{i}}\right)}{\mathrm{n}}}$$

(3)

In Equations (1) and (3), n represents the total hours in July (744 h), and i denotes each hourly interval during the study period. In Equation (2), k (slope) and int (intercept) define the coefficients of the hypothesized linear regression model.

2.2.2. Meteorological Parameter Settings

Four sets of meteorological data were simulated: 2007–2020 Typical Meteorological Year (TMY) data representing historical conditions; 2024 observed data reflecting current weather patterns; and 2050/2080 projected data under SSP3–7.0 scenarios for future weather conditions over the next 30–60 years.

To comprehensively evaluate and optimize the building envelopes of folk houses in the Xinjiang arid region, four typical regions—Turpan, Kuqa, Kashgar, and Hotan—were selected as simulation locations, with the case study building situated in Kashgar City.

The four typical regions chosen for this study are categorized under the “Cold Region” according to the current national standard “Code for thermal design of civil building” (GB 50176-2016) [40]. Specifically, Turpan falls into Cold Zone B (extremely hot and arid), while the others are in Cold Region A (moderate arid and hot).

As shown in Figure 4, the annual outdoor temperature trends across these cities reveal a consistent warming pattern. Compared to historical data (2007–2020), the 2024 average yearly temperatures increased by 2.0 °C in Turpan, 0.2 °C in Kuqa, 0.2 °C in Kashgar, and 0.8 °C in Hotan. By 2050, temperatures are projected to rise further relative to 2024 levels: +1.9 °C (Turpan), +1.8 °C (Kuqa), +1.6 °C (Kashgar), and +1.6 °C (Hotan). By 2080, the increases reach +3.8 °C (Turpan), +3.5 °C (Kuqa), +3.3 °C (Kashgar), and +3.4 °C (Hotan), confirming an overall warming trend in the Xinjiang arid region.

2.2.3. Construction Parameters

The case study building is a rammed earth structure. To comprehensively assess and optimize performance, brick–wood and brick–concrete configurations were added to the same model for comparative analysis. Envelope thermal parameters were determined via heat flux meter measurements and calibrated against national standards, with detailed specifications provided in

Table 3. Folk houses’ external envelope parameters.

Type	Envelope Components	Construction Component Name	Construction Details (from Exterior to Interior)	U-Values (W/m2·K)
Type A (rammed earth folk houses)	Roof	Wooden truss with dense beam flat roof	200 mm wheat straw–clay plaster + 100 mm reed–straw layer + Φ80 mm semicircular log battens	0.352
	Wall	Rammed earth wall	20 mm wheat straw–clay plaster + 500 mm adobe brick wall + 20 mm wheat straw–clay plaster	1.14
	Window	Wooden-framed glass window	Single-pane glass	4.53
	Door	Wooden door	-	2.15
	Ground	Clay brick ground	compacted plain soil layer + 60 mm solid bricks	1.44
Type B (brick–wood folk houses)	Roof	Wooden truss with Dense beam flat roof	200 mm wheat straw-clay plaster + 100 mm reed-straw layer + Φ80 mm semicircular log battens	0.352
	Wall	Solid clay brick wall	20 mm cement mortar + 370 mm solid bricks + 20 mm cement plaster	1.54
	Window	Wooden-framed glass window	Single-pane glass	4.53
	Door	Wooden door	-	2.15
	Ground	Clay brick ground	compacted plain soil layer + 60 mm solid bricks	1.44
Type C (brick–concrete folk houses)	Roof	Reinforced concrete roof	30 mm wheat straw–clay plaster + 2.5 mm polymer-modified waterproofing membrane + 30 mm C20 fine aggregate concrete + 30 mm crushed aerated concrete + 110 mm EPS insulation board + 100 mm reinforced concrete	0.326
	Wall	Perforated clay brick wall	20 mm cement mortar + 240 mm/370 mm perforated bricks + 20 mm cement plaster	1.203
	Window	Aluminum–plastic composite window	Single-pane glass	4.4
	Door	Wooden door	-	2.15
	Ground	Cement ground	compacted plain soil + 60 mm concrete leveling course + 20 mm cement mortar	1.45

.

2.2.4. Building Heating, Ventilation, and Internal Heat Gain Parameter Settings

Local government regulations define heating seasons, such as Kashgar’s October 25–March 15 period (2024), with heating temperatures and ventilation frequencies regulated by the General code for energy efficiency and renewable energy application in buildings (GB 55015-2021) [41]. Folk house parameters—internal heat gain, ventilation rates, and occupancy schedules—were configured per GB 55015-2021 and residents’ actual habits [41]. Non-heating season ventilation schedules followed outdoor temperatures, excluding mechanical cooling in simulations. The detailed inputs are listed in

Table 4. Building heating, ventilation, and internal heat gain parameter settings.

Input Parameters	Type	Value	Period
Heating	Heating temperature	18 °C	Heating season (1:00~24:00)
Ventilation	Ventilation air changes per hour during heating season	0.5 ac/h	Heating season (1:00~24:00)
	Natural ventilation in non-heating seasons	3 ac/h	Transitional seasons (10:00~22:00) Summer (1:00~24:00)
Internal gains	People	25 m2/people	Year-round (1:00~24:00)
	Lighting	5 W/m2	Year-round (1:00~24:00)
	Equipment	3.8 W/m2	Year-round (1:00~24:00)

.

2.2.5. Evaluation Metrics

The four typical regions studied are classified as “Cold Zones” under national standards, requiring buildings to meet winter insulation needs, with some areas requiring summer heat prevention. Field research shows heating provision in Xinjiang’s arid regions during winter, while summer relies solely on natural/minimal mechanical ventilation without a cooling supply. Thus, this study employs the building HEC and non-heating season TDH as evaluation metrics.

For HEC, the metric used is the energy consumed per unit of building floor area, termed the annual average heating energy consumption density (HEC), with the unit being kW·h/(m²·a). The calculation formula is as follows:

$${\mathrm{Q}}_{\mathrm{C}}=\sum _{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{Q}}_{\mathrm{C}\mathrm{i}}/\sum _{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{A}}_{\mathrm{i}}$$

(4)

In Equation (4), Q_Ci represents the HEC for each room, in kW·h; and A_i represents the area of each room, in m². The lower the value of Q_C, the smaller the annual average heating energy density, indicating better energy-saving performance of the folk house. The HEC is compared with a baseline building established based on the GB 55015-2021, with the simulation parameters for the baseline building set according to the GB 55015-2021.

The adaptive comfort model is the most suitable evaluation method for hot climate conditions. This model posits that occupants are not passive recipients of environmental conditions but active participants in shaping their thermal comfort. Therefore, this study employs the adaptive thermal comfort model from the ASHRAE Standard 55 [42] to assess TDH in folk houses. The optimal indoor comfort temperature (T_op) is calculated based on the running mean outdoor dry-bulb temperature (T_rm) using Equations (5) and (6):

$$\mathrm{U}\mathrm{p}\mathrm{p}\mathrm{e}\mathrm{r} 80\mathrm{\%} \mathrm{a}\mathrm{c}\mathrm{c}\mathrm{e}\mathrm{p}\mathrm{t}\mathrm{a}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y} \mathrm{l}\mathrm{i}\mathrm{m}\mathrm{i}\mathrm{t} {\mathrm{T}}_{\mathrm{O}\mathrm{P}}=0.31{\mathrm{T}}_{\mathrm{r}\mathrm{m}}+21.3$$

(5)

$$\mathrm{L}\mathrm{o}\mathrm{w}\mathrm{e}\mathrm{r} 80\mathrm{\%} \mathrm{a}\mathrm{c}\mathrm{c}\mathrm{e}\mathrm{p}\mathrm{t}\mathrm{a}\mathrm{b}\mathrm{i}\mathrm{l}\mathrm{i}\mathrm{t}\mathrm{y} \mathrm{l}\mathrm{i}\mathrm{m}\mathrm{i}\mathrm{t} {\mathrm{T}}_{\mathrm{O}\mathrm{P}}=0.31{\mathrm{T}}_{\mathrm{r}\mathrm{m}}+21.3$$

(6)

In Equations (5) and (6), T_rm represents the 7-day running mean dry-bulb temperature (unit: °C) before the calculation date. By determining whether the hourly indoor temperature falls within the 80% thermal comfort acceptability range, the TDH over a specified period can be derived, reflecting indoor thermal discomfort during non-heating seasons.

Climate Consultant 6.0, developed by the University of California Energy Institute [43], is a building climate adaptability analysis software that generates bioclimatic charts (Figure A1) and recommends region-specific passive design strategies. By inputting EPW weather data files, the software identifies the most effective passive strategy combinations (e.g., shading and natural ventilation) and provides tailored design guidelines for specific locations. The TDH values under optimal passive strategies (obtained from Climate Consultant 6.0) serve as the baseline for comparison (Figure A1). Lower TDH indicates better indoor thermal comfort.

2.3. Multi-Objective Optimization

Based on the analysis of commonly used multi-objective optimization algorithms in the related literature, this study summarizes their performance and applicability, as detailed in

Table A1. Performance and application scenarios of common multi-objective optimization algorithms.

Algorithm	Applicable Objectives	Computational Efficiency	Diversity Maintenance	Parameter Sensitivity	Limitations
GA	Single or multiple custom objectives	Medium	Low	Medium	High computational cost, limited optimization ability for multi-objective problems
PSO	Simple single or multiple objectives	High	Medium	Low	Easily falls into the local optimum
MOACO	Discrete multiple objectives	Low	High	High	Low computational efficiency, sensitive parameters
NSGA-II	2–3 conflicting objectives	High	High	Medium	High-dimensional target performance degradation
NSGA-III	≥3 high-dimensional objectives	Medium	High	High	Three objective efficiency degradation
MOEA/D	Decomposable multi-objective (≥2)	Medium	Medium	High	Heavy-weight vector design

[44,45,46]. Among these, NSGA-II demonstrates superior performance in optimizing 2–3 conflicting objectives, achieving a balance between computational efficiency and Pareto front quality. Other algorithms (e.g., NSGA-III for high-dimensional objectives and MOEA/D for large-scale problems) exhibit limitations in three-objective scenarios. Therefore, the NSGA-II algorithm was selected for multi-objective optimization of folk houses.

Developed by Deb et al. in 2000, NSGA-II is a Pareto-based evolutionary algorithm that resolves multi-objective optimization challenges by simultaneously optimizing computational efficiency, solution accuracy, and population diversity [47]. It improves upon earlier genetic algorithms by reducing complexity, increasing computational speed, and enhancing optimization precision [47]. Yao et al. used genetic algorithms to optimize the layout of prefabricated building sites, improving solution quality and reducing time [48]. Therefore, this method has been widely applied in building performance optimization [49,50,51,52,53]. Academic contributions addressing these fields are documented below. Sudhir Kumar Gupta et al. used NSGA-II to optimize the carbon emissions and cooling load of traditional folk houses in the northeastern region of India [54]. Letiane Benincá* et al. used the NSGA-II algorithm to conduct multi-objective optimization of the shape and orientation of two multi-family residential buildings in the southern region of Brazil, considering both cooling and heating needs, resulting in the optimal solar 180 orientation [55].

This study employs the NSGA-II-based Wallacei to perform optimization of the target values (TDH, HEC, and NPV). The parameters are configured as follows: population size of 50, maximum iterations of 100, crossover probability of 0.8, and mutation probability of 0.2. Forty-eight multi-objective optimization groups were executed across four typical regions, three folk house types, and four climate scenarios, amounting to 120,000 simulations. By adjusting the parameters of the screened building envelope variables (Section 3.3.2), the algorithm optimizes the target values, including TDH, HEC, and NPV.

2.3.1. Optimization Objectives

The evaluation of folk houses in Section 3.1 revealed significant discrepancies between the HEC and TDH values and those of the control group. Both HEC and TDH demonstrated substantial optimization potential. Therefore, this study establishes three optimization objectives: (1) minimizing non-heating-season TDH, (2) reducing HEC, and (3) maximizing NPV (economic feasibility). Definitions of non-heating-season TDH and HEC are provided in Section 2.2. NPV, a critical economic evaluation metric, determines the feasibility of an investment: projects with NPV > 0 are considered feasible, while those with NPV < 0 are rejected. Higher NPV values within a defined payback period indicate superior economic performance. NPV is calculated using Equation (8) [56]:

$$\mathrm{N}\mathrm{P}\mathrm{V}=-\mathrm{I}+\sum _{\mathrm{n}=0}^{{\mathrm{I}}_{\mathrm{S}}}{\displaystyle \frac{{\mathrm{S}}_{\mathrm{n}}}{{\left(1+\mathrm{r}\right)}^{\mathrm{n}}}}$$

(7)

$${\mathrm{S}}_{\mathrm{n}}={\mathrm{S}}_0{\left(1+\mathrm{p}\right)}^{\mathrm{n}}$$

(8)

$$\mathrm{r}=-0.015\ln {\mathrm{I}}_{\mathrm{S}}+0.106$$

(9)

In Equations (7)–(9), I is the initial investment (retrofit cost) in CNY; I_s is the remaining building service life, set to 30 years (more extended periods reduce accuracy [57]); r is the benchmark discount rate (%); S_n is the energy-saving revenue in year n (CNY), assumed constant over the calculation period; S₀ is the initial energy-saving revenue (CNY); and p is the annual energy price escalation rate (4%).

2.3.2. Optimization Variable Screening and Ranges

(1) Optimization Variable Screening. A total of 18 building envelope parameters were preliminarily selected (ranges in

Table 5. Sensitivity analysis variables and ranges of values for the envelope of folk houses.

Envelope Components	Sensitivity Analysis Variables	Range of Values
Roof	U-value (UR)	0.2–4 W/m2·K
	Outer surface solar radiation absorption coefficient (SRA_R)	0–0.95
Wall	U-value of the north-facing wall (UW_N)	0.25–2.1 W/m2·K
	U-value of the south-facing wall (UW_S)
	U-value of the east-facing wall (UW_W)
	U-value of the west-facing wall (UW_E)
	Outer surface solar radiation absorption coefficient of the north-facing wall (SRA_W_N)	0–0.95
	Outer surface solar radiation absorption coefficient of the south-facing wall (SRA_W_S)
	Outer surface solar radiation absorption coefficient of the east-facing wall (SRA_W_E)
	Outer surface solar radiation absorption coefficient of the west-facing wall (SRA_W_W)
Window	U-value of the north-facing window (UWin_N)	1.2–4.6 W/m2·K
	U-value of the south-facing window (UWin_S)
	North-facing window-to-wall ratio (WWR_N)	0.1–0.5
	South-facing window-to-wall ratio (WWR_S)	0.1–0.3
	SHGC of north-facing window (SHGC_N)	0.3–0.7
	SHGC of south-facing window (SHGC_S)
Door	U-value (W/m2·K)	1.5–2.5 W/m2·K
Ground	U-value (W/m2·K)	0.4–2 W/m2·K

) to identify key variables influencing HEC and TDH. A Latin Hypercube Sampling (LHS) method generated 1000 parameter samples for building performance simulations. Sensitivity analysis using Standardized Regression Coefficients (SRCs) and Standardized Rank Regression Coefficients (SRRCs) ranked the parameters’ impacts on HEC and TDH, with the most influential variables retained for multi-objective optimization.

SRC and SRRC are widely applied in building performance analysis, where SRC is suitable for linear models and SRRC for nonlinear, non-monotonic models [58,59,60,61,62]. This study integrates sensitivity indices to quantify parameter impacts on optimization objectives and ranks parameters using a composite sensitivity index across multiple regions and climate scenarios. The composite sensitivity percentage (SAP) is calculated as follows:

$${SAP}_{(x,y,z)}={\displaystyle \frac{\left|{SA}_{(x,y,z)}\right|}{{\sum }_{x=1}^n{SA}_{(x,y,z)}}}$$

(10)

In Equation (10), SA is the total sensitivity value across indices; x is the input parameter; n is the total number of influencing factors; y is the output objectives, including heating energy consumption per unit floor area and total thermal discomfort hours during non-heating seasons; and z is the sensitivity analysis index (SRC/SRRC).

(2) Optimization Variable Ranges. Key factors influencing the optimization objectives were identified as final variables through the methodology. Variable ranges were established based on national standards and field survey data, with optimization strategies, costs, and material specifications detailed in

Table 6. Optimization variables and ranges of values for the envelope of folk houses.

Optimization Variable	Optimization Strategy	The Range for Type A	The Range for Type B	The Range for Type C	Optimization Cost
SRA_R	(a) High reflectance coating. (b) Reflective coating. (c) Conventional coating.	(a) 0.1–0.25 (b) 0.25–0.65 (c) 0.65–0.95			(a) 100 CNY/m2 (b) 50 CNY/m2 (c) 20 CNY/m2
Thickness of roof insulation layer (TR)	(a) Wheat straw insulation layer. (b) EPS insulation layer.	25/55–105 mm	25/55–105 mm	15–80 mm	(a) 380 CNY/m2 (b) 450 CNY/m2
Thickness of north-facing wall insulation layer (TW_N)		95–150 mm	105–160 mm	95–150 mm
Thickness of south-facing wall insulation layer (TW_S)
Thickness of east-facing wall insulation layer (TW_E)
Thickness of west-facing wall insulation layer (TW_W)
Thickness of ground insulation layer (TG)		75–105 mm
UWin_N	New energy-saving window	1.2–1.8 W/m2·K			For window cost, see Table A2
UWin_S
SHGC_S		0.3–0.7
WWR_N	-	0.1–0.3
WWR_S		0.1–0.5

.

2.3.3. Entropy-Weighted TOPSIS Comprehensive Evaluation Method

The TOPSIS method ranks alternatives by the proximity to ideal solutions with minimal sample requirements. Initially developed for multi-criteria decision-making, it evaluates alternatives by measuring distances from the positive ideal solution (optimal criteria performance) and negative ideal solution (poorest performance). However, conventional TOPSIS depends on subjective weight assignments. To overcome this limitation, this study integrates the entropy weight method to determine criterion weights through data variability analysis objectively, improving evaluation accuracy. The entropy-weighted TOPSIS method proves applicable to multi-scenario optimization across the entire building lifecycle, encompassing structural selection during design phases, energy efficiency management in operational stages, and cost–benefit analysis for retrofitting projects. This method avoids iterative computations and efficiently generates decision outcomes for small-to-medium-scale problems, making it well-suited for rapid decision-making in real-world engineering practices. Consequently, the entropy-weighted TOPSIS method has gained extensive application in decision-making systems [63,64,65,66,67,68,69].

Preparing the matrix first, this involves using the initial data to prepare a matrix with m indicators and n objects, as follows:

Step 1: Construct the Decision Matrix

Using the initial data, a matrix is prepared with m indicators and n objects, as follows:

$$\mathrm{X}=\left(\begin{array}{ccc}{\mathrm{x}}_1& \cdots & {\mathrm{x}}_{1\mathrm{m}}\\ ⋮& \ddots & ⋮\\ {\mathrm{x}}_{\mathrm{n}1}& \cdots & {\mathrm{x}}_{\mathrm{n}\mathrm{m}}\end{array}\right)$$

(11)

Step 2: Calculate the Entropy Value of Indicators

The entropy value e_i for the ith indicator is calculated as follows:

$${\mathrm{e}}_{\mathrm{i}}={\displaystyle \frac{-1}{\ln \mathrm{n}{\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{p}}_{\mathrm{i}\mathrm{j}}\ln {\mathrm{p}}_{\mathrm{i}\mathrm{j}}}}$$

(12)

The difference coefficient g_i for the ith indicator is calculated as follows:

$${\mathrm{g}}_{\mathrm{i}}=1-{\mathrm{e}}_{\mathrm{i}}$$

(13)

The weight w_i for the ith indicator is calculated as follows:

$${\mathrm{w}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{g}}_{\mathrm{i}}}{{\sum }_{\mathrm{i}=1}^{\mathrm{m}}{\mathrm{g}}_{\mathrm{i}}}}$$

(14)

Step 3: Calculate the Distance of Each Alternative from the Positive and Negative Ideal Solutions

$${\mathrm{D}}_{{\mathrm{i}}^{+}}=\sqrt{\sum _{\mathrm{j}=1}^{\mathrm{m}}{\left({\mathrm{x}}_{\mathrm{i}\mathrm{j}}+{\mathrm{A}}_{{\mathrm{i}}^{+}}\right)}^2}$$

(15)

$${\mathrm{D}}_{{\mathrm{i}}^{-}}=\sqrt{\sum _{\mathrm{j}=1}^{\mathrm{m}}{\left({\mathrm{x}}_{\mathrm{i}\mathrm{j}}+{\mathrm{A}}_{{\mathrm{i}}^{-}}\right)}^2}$$

(16)

Aj+ and Aj− are the positive and negative ideal solutions for the jth indicator, respectively, and D_i+ and D_i− represent the distances of the evaluation object from the optimal and worst solutions, respectively.

Step 4: Calculate the Relative Closeness C_i for Each Alternative

$${\mathrm{C}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{D}}_{{\mathrm{i}}^{-}}}{{{\mathrm{D}}_{{\mathrm{i}}^{+}}+\mathrm{D}}_{{\mathrm{i}}^{-}}}}$$

(17)

The larger the value of C_i, the better the alternative.

3. Results

The results are presented in three subsections:

(1) Validation results: This section includes the validation outcomes of the simulation model.

(2) Simulation results: This section presents the analysis of the heating energy consumption (HEC) and thermal discomfort hour (TDH) performance of the three vernacular dwelling types under multiple climate scenarios.

(3) Multi-objective optimization results: This section encompasses the sensitivity analysis of optimization variables, the Pareto-optimal solutions for HEC, TDH, and net present value (NPV), and the most balanced Pareto-optimal solutions identified using the entropy-weighted TOPSIS method.

3.1. Validation Results

To enhance the accuracy of the experimental conclusions, simulation results were validated against field-measured data by adjusting the ventilation parameters to align simulated and measured values. As shown in Figure 5, the k-values for the tested rooms were 1.06, 0.97, and 1.03, respectively. The Pearson’s correlation coefficients (Rs) reached 0.84342, 0.80552, and 0.8195, while the RMSEs were 2.36478 °C, 1.81574 °C, and 1.95151 °C, respectively. These metrics confirm strong agreement between the simulated and measured data, demonstrating the model’s reliability and validity for subsequent analyses.

3.2. Simulation Results

3.2.1. Thermal Discomfort Hours’ Simulation Results

After completing the heating energy consumption simulation, the TDH of folk houses in four typical regions during the non-heating season were further simulated using the historical to future four different scenario meteorological data on the GH + Ladybug platform, with the results showing different trends in various regions. The TDH simulation results are shown in Figure 6.

Turpan region: For Type A folk houses, the entire non-heating season TDH under the 2080 climate scenario increased by 33.46%, with a 47.19% increase in summer TDH and a 14.94% increase in transition season TDH. Type B folk houses saw an increase of 31.04% in the entire non-heating season TDH, with a 42.63% increase in summer and a 14.27% increase in the transition season. Type C folk houses experienced a 36.7% increase in the entire non-heating season TDH, with a 45.98% increase in summer and a 22.84% increase in the transition season. The optimal passive strategies resulted in a 24.4% increase in the entire non-heating season TDH, with a 32.8% increase in summer and a 13.8% decrease in the transition season.

Kuqa region: Type A folk houses saw a decrease of 17.92% in the entire non-heating season TDH under the 2080 climate scenario, with a 17.59% increase in summer TDH and a 39.62% decrease in transition season TDH. Type B folk houses experienced a 19.66% decrease in the entire non-heating season TDH, with a 21.41% increase in summer and a 43.05% decrease in the transition season. Type C folk houses had a 20.79% decrease in the entire non-heating season TDH, with a 19.6% increase in summer and a 44.96% decrease in the transition season. The optimal passive strategies resulted in a 28% decrease in the entire non-heating season TDH, with a 78.3% increase in summer and a 66.4% decrease in the transition season.

Kashgar region: Type A folk houses saw a decrease of 24.47% in the entire non-heating season TDH under the 2080 climate scenario, with a 15.47% increase in summer TDH and a 47.87% decrease in transition season TDH. Type B folk houses experienced a 24.65% decrease in the entire non-heating season TDH, with a 10.4% increase in summer and a 46.03% decrease in the transition season. Type C folk houses had a 23.65% decrease in the entire non-heating season TDH, with a 13.73% increase in summer and a 44.77% decrease in the transition season. The optimal passive strategies resulted in a 23.8% decrease in the entire non-heating season TDH, with a 21% increase in summer and a 49.3% decrease in the transition season.

Hotan region: Type A folk houses saw a decrease of 6.78% in the entire non-heating season TDH under the 2080 climate scenario, with a 44.22% increase in summer TDH and a 40.54% decrease in transition season TDH. Type B folk houses experienced an 11.62% decrease in the entire non-heating season TDH, with a 36.43% increase in summer and a 44.18% decrease in the transition season. Type C folk houses had a 10.84% decrease in the entire non-heating season TDH, with a 31.31% increase in summer and a 41.39% decrease in the transition season. The optimal passive strategies resulted in a 13.6% decrease in the entire non-heating season TDH, with a 17.8% increase in summer and a 7.9% increase in the transition season.

Overall, as climate warming intensifies in the future, the non-heating season TDH of folk houses in the arid region of Xinjiang shows a significant trend of improvement in the transition season and an increase in summer. Still, the extremely hot-arid region of Turpan shows an overall increasing trend in TDH during the non-heating season. This is consistent with the expectation of high summer temperatures due to climate warming. The climate adaptability of all folk houses during the non-heating season is generally improving. It rapidly declines in summer, likely leading to frequent indoor overheating phenomena.

Type A folk houses generally show better climate adaptability than others in all four regions, with the best overall performance in Kashgar. Still, adaptability collapses in extremely hot-arid areas (Turpan). Type C folk houses show moderate climate adaptability during the non-heating season, with the best performance in Kashgar and Kuqa, but relatively better performance in extremely hot-arid areas. Type B folk houses also show moderate climate adaptability during the non-heating season, with similar gaps to other folk houses. However, in the extremely hot-arid region of Turpan, the climate adaptability of any folk house relying solely on the building’s inherent properties will fail in high temperatures. In addition, the optimal passive strategies in all regions show significant potential for climate adaptability, demonstrating a notable advantage in optimizing thermal comfort during the non-heating season in the arid region of Xinjiang, especially in terms of performance improvement in the transition season; however, the improvement in climate adaptability is relatively tiny under intensified summer high temperatures, particularly in extremely hot-arid areas.

3.2.2. Heating Energy Consumption Simulation Results

After model validation, the HECs of four typical regions were simulated using historical to future meteorological data from four different scenarios on the GH + Ladybug platform, and the results all pointed to a decreasing trend of HEC to varying degrees. The HEC simulation results are shown in Figure 7.

Turpan region: The HEC gap of Type A folk houses compared to the reference building narrowed from 63.74% under the historical climate scenario to 55.04% under the 2080 climate scenario, with a decrease of 35.11%, indicating limited improvement in climate adaptability. The HEC gap of Type B folk houses compared to the reference building narrowed from 108.5% to 98.15%, with a decrease of 34.87%, but still 98.15% higher than the reference building in 2080, indicating the worst climate adaptability. The HEC gap of Type C folk houses compared to the reference building narrowed from 83.12% to 73.13%, with a decrease of 33.68%, close to that of the reference building (31.5%), indicating medium climate adaptability.

Kuqa region: The HEC gap of Type A folk houses compared to the reference building narrowed from 72.94% under the historical climate scenario to 66.02% under the 2080 climate scenario, with a decrease of 24.86%, indicating a slight improvement in climate adaptability. Compared to the reference building, the HEC gap of Type B folk houses narrowed from 122.21% to 112.83%, with a decrease of 25.04%, indicating the worst climate adaptability in this region. The HEC gap of Type C folk houses compared to the reference building narrowed from 92.08% to 85.12%, with a decrease of 24.57%, indicating medium climate adaptability in this region.

Kashgar region: The HEC gap of Type A folk houses compared to the reference building narrowed from 76.9% under the historical climate scenario to 55.32% under the 2080 climate scenario, with a decrease of 40.97%, indicating a significant improvement in winter climate adaptability. Compared to the reference building, the HEC gap of Type B folk houses narrowed from 131.84% to 99.84%, with a decrease of 42.05%, and the large HEC gap with the reference building indicates the worst winter climate adaptability in this region. The HEC gap of Type C folk houses compared to the reference building narrowed from 94.48% to 72.30%, with a decrease of 42.97%, indicating the most significant improvement in winter climate adaptability.

Hotan region: The HEC gap of Type A folk houses compared to the reference building narrowed from 66.59% under the historical climate scenario to 50.71% under the 2080 climate scenario, with a decrease of 37.64%, having the smallest HEC gap and a relatively significant improvement in climate adaptability. Compared to the reference building, the HEC gap of Type B folk houses narrowed from 116.19% to 97.05%, with a decrease of 37.18%, still having a large HEC gap with the reference building, indicating the worst winter climate adaptability in this region. The HEC gap of Type C folk houses compared to the reference building narrowed from 84.25% to 71.25%, with a decrease of 35.94%, indicating the most significant improvement in winter climate adaptability.

Overall, as the future climate warms, the HEC of folk houses in the arid region of Xinjiang shows a significant decreasing trend, consistent with the expectation that climate warming will reduce winter heating demand. The climate adaptability of various folk houses in winter is generally improving. Type A folk houses typically show better winter climate adaptability in all four regions, with HEC closest to the reference building and a relatively small decrease in HEC, performing best overall in the Hotan region. Type C folk houses show medium winter climate adaptability, with HEC relatively close to the reference building and a minor decrease in HEC, showing the best sensitivity to winter climate change in the Turpan region, with a decrease closest to the reference building. Type B folk houses show poor winter climate adaptability, with the most significant HEC gap from the reference building but the largest decrease in HEC, indicating the highest sensitivity to winter climate and the most significant potential for energy savings. It is worth noting that the HEC reduction of reference buildings is generally less than that of various folk houses. The HEC reduction of reference buildings in various regions is stable at 21%~32%, indicating a relatively balanced adaptability design of national energy-saving standards for climate change during the heating season and higher sensitivity of folk houses to climate change during the heating season.

The superior winter climate adaptability of rammed earth folk houses (Type A) in Hotan can be attributed to their high thermal mass and traditional passive design strategies. Adobe walls (500 mm thickness with wheat straw–clay plaster) provide effective heat storage, buffering diurnal temperature fluctuations—a feature validated by Yan et al. in their thermal comfort surveys of Turpan dwellings [20]. In contrast, brick–concrete houses (Type C) exhibit better extreme heat resilience in Turpan due to modern insulation materials (e.g., 110 mm EPS roof insulation, U-value = 0.326 W/m²·K), which reduce heat transfer under prolonged high-temperature exposure. This aligns with Nasouri et al. [16], who reported EPS insulation’s effectiveness in reducing cooling loads in Iran’s Bushehr region by 23%.

3.3. Multi-Objective Optimization Results

3.3.1. Sensitivity Analysis Results

After completing the simulation of 16 groups of variable parameter samples extracted by the LHS method under four different climate scenarios for four regions (each group containing 1000 samples, totaling 16,000 simulation samples), and conducting sensitivity analysis using SRC and SRRC for 18 variables and integrating the sensitivity indices, the ranking of the influence of the 18 input variables on heating energy consumption and thermal discomfort is shown in Figure 8. When using LHS, the analysis index results of SRC and SRRC were similar. By evaluating the influence of each parameter on heating energy consumption and thermal discomfort using the SRC and SRRC sensitivity analysis methods, and averaging the influence obtained from various climate scenarios and regional indicators, the results are arranged in descending order and shown in the last column of Figure 8. For subsequent multi-objective optimization, variables with average influence below a certain threshold were removed, including precisely six parameters: SHGC_S, SRA_W_E, SRA_W_S, SRA_W_N, SRA_W_W, and UD.

3.3.2. Optimization Results

This study employs the NSGA-II algorithm for multi-objective optimization, running the algorithm under four climate scenarios in four regions. After 50 generations of iterative optimization, a converged Pareto front solution set containing 2500 solutions was obtained. Each of the 50 generations produced 2500 sets of calculation data per group, resulting in 48 multi-objective optimization groups (four regions with three types of folk houses and four climate scenarios), amounting to 120,000 sets of calculation data. The Pareto front solutions of the last generation, consisting of 2400 sets, are shown in Figure 9.

3.3.3. Optimal Solution for TDH

As shown in

Table 7. Pareto-optimal TDH solution set.

Region	Year	Type-Indv	TDH/h	Type-Indv	TDH/h	Type-Indv	TDH/h
Kashgar	2007–2021	A-2	1167	B-1	1211	C-1	1185
	2024	A-2	1046	B-1	1077	C-2	1031
	2050	A-4	1233	B-2	1205	C-4	1162
	2080	A-2	1371	B-4	1328	C-3	1294
Turpan	2007–2021	A-3	2288	B-4	2280	C-1	2215
	2024	A-2	2771	B-2	2751	C-4	2745
	2050	A-1	3259	B-2	3248	C-4	3234
	2080	A-1	3548	B-2	3522	C-3	3505
Kuqa	2007–2021	A-4	1157	B-3	1236	C-1	1210
	2024	A-2	1143	B-1	1155	C-5	1323
	2050	A-1	1389	B-3	1377	C-3	1328
	2080	A-2	1397	B-2	1428	C-2	1401
Hotan	2007–2021	A-2	1214	B-1	1191	C-5	1147
	2024	A-2	1135	B-4	1156	C-5	1125
	2050	A-2	1410	B-1	1384	C-5	1291
	2080	A-4	1701	B-5	1655	C-2	1634

, in terms of TDH, under the climate scenarios of TMY (2007–2021), most regions are mainly distributed between 1100 and 1250 h, while in the extremely hot-arid region (Turpan), the distribution is between 2200 and 2300 h. Under the climate scenarios of the observed meteorological year 2024, most regions are distributed between 1000 and 1200 h, and in Turpan, between 2700 and 2800 h. Under the predicted meteorological year 2050 climate scenarios, most regions are distributed between 1200 and 1400 h, and in Turpan, 3200 and 3260 h. Under the predicted meteorological year 2080 climate scenarios, most regions are mainly distributed between 1300–1650 h, and in Turpan, between 3500 and 3550 h.

After optimization, under the climate scenarios of TMY (2007–2021), the reduction in the optimal indicators compared to the original building indicators in most regions ranges from 43.59% to 53.65%, and in Turpan, from 18.72% to 22.66%. Under the climate scenarios of the observed meteorological year 2024, the reduction in most regions ranges from 37.92% to 53.88%, and in Turpan, 11.30% to 15.59%. Under the predicted meteorological year 2050 climate scenarios, the reduction in most regions ranges from 20.83% to 39.54%, and in Turpan, from 7.41% to 10.47%. Under the predicted meteorological year 2080 climate scenarios, the reduction in most regions ranges from 15.2% to 32.55%, and in Turpan, from 5.56% to 8.83%.

3.3.4. Optimal Solution for HEC

As can be seen from

Table 8. Pareto-optimal HEC solution set.

Region	Year	Type-Indv	HEC/kW·h/(m2·a)	Type-Indv	HEC/kW·h/(m2·a)	Type-Indv	HEC/kW·h/(m2·a)
Kashgar	2007–2021	A-4	23.321	B-3	25.395	C-3	26.916
	2024	A-1	19.91	B-2	22.261	C-3	22.307
	2050	A-2	13.781	B-1	16.638	C-5	15.532
	2080	A-3	10.877	B-3	12.582	C-5	11.522
Turpan	2007–2021	A-2	27.607	B-2	29.266	C-2	32.539
	2024	A-1	23.551	B-1	25.672	C-1	28.345
	2050	A-4	19.035	B-1	21.155	C-2	20.694
	2080	A-2	15.486	B-4	16.039	C-4	16.638
Kuqa	2007–2021	A-2	30.972	B-2	32.032	C-2	34.198
	2024	A-3	27.976	B-3	32.17	C-4	20.924
	2050	A-6	22.445	B-1	24.934	C-1	26.593
	2080	A-3	18.389	B-3	19.772	C-3	20.878
Hotan	2007–2021	A-4	16.592	B-4	18.389	C-3	19.496
	2024	A-3	15.947	B-1	18.205	C-4	19.403
	2050	A-1	11.522	B-4	13.596	C-1	13.642
	2080	A-1	9.033	B-3	9.955	C-4	10.14

, in terms of HEC, under the climate scenarios of TMY (2007–2021), most regions are distributed between 16.592 and 34.198 kW·h/(m²·a); under the observed meteorological year 2024 climate scenarios, most regions are mainly distributed between 15.947 and 32.17 kW·h/(m²·a); under the predicted meteorological year 2050 climate scenarios, most regions are primarily distributed between 11.522–26.593 kW·h/(m²·a); and under the predicted meteorological year 2080 climate scenarios, most regions are mainly distributed between 9.033 and 20.878 kW·h/(m²·a).

Under the climate scenarios of TMY (2007–2021), the reduction range of the optimal index and the original building index in most regions is between 76.53% and 88.64%. Under the observed meteorological year 2024 climate scenarios, the reduction range in most regions is between 78.24% and 88.65%. Under the predicted meteorological year 2050 climate scenarios, the reduction range in most regions is between 80.27% and 89.17%. Under the predicted meteorological year 2080 climate scenarios, the reduction range in most regions is between 81.19% and 90.21%.

3.3.5. Optimal Solution for NPV

As seen from

Table 9. Pareto-optimal NPV solution set.

Region	Year	Type-Indv	NPV/CNY	Type-Indv	NPV/CNY	Type-Indv	NPV/CNY
Kashgar	2007–2021	A-1	50,950	B-2	78,641	C-2	62,141
	2024	A-3	53,664	B-3	81,798	C-5	65,669
	2050	A-3	57,909	B-3	85,871	C-26	70,485
	2080	A-1	62,287	B-5	88,979	C-1	74,022
Turpan	2007–2021	A-1	59,541	B-3	74,751	C-3	60,343
	2024	A-3	61,321	B-5	84,110	C-3	64,656
	2050	A-3	66,485	B-5	88,952	C-1	69,061
	2080	A-4	68,721	B-1	91,347	C-5	72,998
Kuqa	2007–2021	A-3	59,783	B-1	87,144	C-3	66,536
	2024	A-1	58,949	B-23	86,373	C-3	75,624
	2050	A-2	64,379	B-2	91,690	C-2	71,787
	2080	A-1	67,179	B-4	96,445	C-4	76,641
Hotan	2007–2021	A-1	40,779	B-3	62,966	C-1	46,491
	2024	A-4	41,552	B-3	64,176	C-2	47,418
	2050	A-5	45,725	B-2	67,097	C-3	51,978
	2080	A-3	49,363	B-40	70,662	C-3	55,951

, regarding NPV, under the climate scenarios of TMY (2007–2021), most regions are distributed between CNY 40,000 to 88,000. Under the observed meteorological year 2024 climate scenarios, most regions are distributed between CNY 41,000 and 87,000. Under the predicted meteorological year 2050 climate scenarios, most regions are distributed between CNY 45,000 and 92,000. Under the predicted meteorological year 2080 climate scenarios, most regions are distributed between CNY 49,000 and 97,000.

3.3.6. Evaluation Results of the Entropy-Weighted TOPSIS Method

When evaluating the 50 solutions on the Pareto front for various types of folk houses in different regions under various climate scenarios, this study employed the entropy-weighted TOPSIS method for a comprehensive evaluation to select the Pareto optimal solution set with balanced three-dimensional objectives, thereby further screening the best optimization scheme. The best optimization scheme numbers and performance for various types of folk houses in different regions under various climate scenarios are shown in

Table 10. The optimal solution set is based on the entropy-weighted TOPSIS method.

Region	Year	Type-Indv	TDH/ h	HEC/kW·h/(m2·a)	NPV/ CNY	Type-Indv	TDH/ h	HEC/kW·h/(m2·a)	NPV/CNY	Type-Indv	TDH/ h	HEC/kW·h/(m2·a)	NPV/CNY
Kashgar	2007–2021 2024 2050 2080	A-4	1447	23.321	46,560	B-3	1430	25.395	74,709	C-5	1289	33.092	54,626
		A-2	1046	65.631	31,488	B-6	1088	43.6	74,481	C-37	1440	22.399	60,830
		A-33	1235	45.536	43,927	B-2	1205	44.937	72,658	C-3	1694	16.5	66,160
		A-3	1741	10.877	56,862	B-2	1329	48.854	72,667	C-7	1300	34.981	63,764
Turpan	2007–2021 2024 2050 2080	A-2	2553	27.607	56,765	B-4	2280	64.248	57,844	C-1	2215	64.11	46,179
		A-1	2934	23.551	56,779	B-1	2909	25.672	79,237	C-1	2853	28.345	59,229
		A-1	3259	43.416	51,518	B-27	3317	23.966	80,637	C-10	3318	21.155	62,801
		A-3	3549	39.314	59,142	B-4	3585	16.039	85,596	C-30	3507	45.536	60,530
Kuqa	2007–2021 2024 2050 2080	A-4	1157	65.631	37,359	B-2	1697	32.032	83,597	C-1	1210	59.27	55,980
		A-3	1794	27.976	55,849	B-4	1155	69.456	68,823	C-4	1792	20.924	70,338
		A-17	1397	57.888	51,080	B-3	1377	60.837	77,091	C-1	1810	26.593	67,761
		A-19	1415	49.776	56,868	B-7	1443	55.491	80,006	C-2	1401	51.988	64,061
Hotan	2007–2021 2024 2050 2080	A-3	1524	17.929	36,344	B-4	1484	18.389	59,748	C-8	1428	19.726	41,110
		A-6	1447	16.177	36,187	B-1	1439	18.205	57,217	C-18	1375	19.634	41,264
		A-3	1500	16.17	52,655	B-3	1388	40.973	54,776	C-4	1296	35.535	42,691
		A-2	1705	29.589	36,582	B-5	1655	42.494	52,083	C-10	1638	26.593	45,038

. Overall, each set of optimization schemes below has significant optimization effects on heating energy consumption, indoor thermal comfort, and economic performance.

From Table 10, it can be observed that based on the entropy-weighted TOPSIS method, the optimal solutions in most regions under the climate scenarios of TMY (2007–2021) are mainly distributed as follows: HEC between 18 and 60 kW·h/(m²·a), TDH between 1200 and 1700 h, and NPV between CNY 36,000 and 84,000, with Turpan’s TDH distributed between 2750 and 2950 h. Under the observed meteorological year 2024 climate scenarios, HEC in most regions is mainly distributed between 15 and 70 kW·h/(m²·a), TDH between 1000 and 1800 h, and NPV between CNY 31,000 and 74,000, with Turpan’s TDH distributed between 2750 and 2950 h. Under the predicted meteorological year 2050 climate scenarios, HEC in most regions is mainly distributed between 16 and 65 kW·h/(m²·a), TDH between 1200 and 1800 h, and NPV between CNY 40,000 and 77,000, with Turpan’s TDH distributed between 3200 and 3350 h. Under the predicted meteorological year 2080 climate scenarios, HEC in most regions is mainly distributed between 10 and 50 kW·h/(m²·a), TDH between 1300 and 1800 h, and NPV between CNY 36,000 and 72,000, with Turpan’s TDH mainly distributed between 3500 and 3600 h.

4. Discussion

4.1. Strategies and Policy Proposal for Improving Climate Adaptability

Gradual alleviation of overcooling risks with future climate changes is observed in the arid region of Xinjiang. Section 3.2’s analysis confirms narrowed HEC gaps between folk houses and reference buildings, indicating strong envelope performance in mitigating indoor overcooling under climate warming scenarios. For example, folk houses in Hotan demonstrate significantly enhanced winter climatic adaptability, with heating energy consumption (HEC) reduced from 77.798–101.718 kW·h/(m²·a) to an optimized 9.033–10.14 kW·h/(m²·a).

Non-heating season analysis reveals escalating overheating risks, particularly in Turpan. Type A folk houses show summer overheating gaps increasing from 59.47% (historical) to 76.69% (2080), with transitional season gaps rising from 50.13% to 51.65%. Similar trends occur in Type B (summer: 71.89% to 84.56%; transitional: 51% to 51.66%) and Type C (summer: 61.93% to 77.95%; transitional: 37.72% to 48.68%) houses.

Climatic adaptability challenges for folk houses in Xinjiang’s arid region have shifted from sole winter overcooling to dual overheating and overcooling under warming scenarios. The increased summer thermal discomfort hours and reduced TDH optimization range (particularly in extreme hot-arid zones) highlight the summer climatic vulnerability of building envelopes. Post-optimization analysis reveals declining efficacy: Turpan’s thermal discomfort hour reductions decreased from 15.2–32.55% (2007–2021 baseline) to 5.56–8.83%, while other regions declined from 43.59–53.65% to 20.83–39.54%. These trends indicate insufficient envelope adaptability to mitigate indoor overheating risks, which are projected to intensify with future climate changes in extreme hot-arid zones.

Current building policies in Xinjiang’s arid region prioritize winter insulation but neglect summer thermal comfort and overheating mitigation. The urgent need to balance winter insulation with summer heat prevention in folk houses requires updated design strategies and evaluation standards under global climate change. Observed diminishing TDH optimization efficacy demonstrates that envelope improvements alone cannot adequately address high-temperature challenges in warming scenarios. International research confirms passive solutions like ventilation and shading effectively alleviate residential overheating [69].

Figure 10 ranks thermal comfort optimization strategies for Kashgar’s 2080 summer (June–August) as follows: indirect evaporative cooling, shading, high heat storage with night ventilation, direct evaporative cooling, high heat storage, natural ventilation, and passive solar heating (high/low heat storage). Despite evaporative cooling’s theoretical dominance, its water-intensive nature (annual precipitation < 100 mm vs. evaporation rate 3× precipitation) renders it unsuitable for Xinjiang’s arid conditions. Shading thermal mass with night and natural ventilation achieves 36% summer comfort improvement—consistent with prior findings [69]. These results inform climate-adaptive design standard suggestions for Xinjiang’s arid region (

Table 11. Suggestions for revising the standards and specifications related to summer heat insulation in the arid regions of Xinjiang.

Standards	Codes	Content	Recommendations
Code for thermal design of civil buildings (GB 50176-2016) [40]	4.1.2	“Cold Zone A—Should meet the thermal insulation design requirements, heat protection design need not be considered”.	Cold Zone A: Add “Should meet thermal insulation design requirements, with consideration for natural ventilation and shading design”.
		“Cold Zone B—Should meet the thermal insulation design requirements, it is advisable to meet the heat insulation design requirements, with consideration for natural ventilation and shading design”.	Cold Zone B: “Enhance thermal insulation design requirements, prioritize natural ventilation and shading design”.
	4.3.2	The architectural design in regions with hot summers and warm winters, as well as hot summers and cold winters, must meet the summer heat protection requirements. The architectural design in Cold Zone B should consider the summer heat protection requirements.	Cold Zone B: Revise to “Must comply with summer heat protection requirements.” Cold Zone A: Revise to “Should consider summer heat protection requirements”.
	6.3.1	The product of the solar heat gain coefficient of the transparent envelope and the summer building shading coefficient should be less than the limit values specified in the table.	Add “Limit values for the product of Solar Heat Gain Coefficient (SHGC) and Summer Shading Coefficient (SC) of transparent building envelopes in cold regions”.
Xinjiang Local Standards—Design standard for energy efficiency of residential buildings in severe cold and cold zones (XJJ001-2021) [70]	3.03	The selection of indoor thermal environment calculation parameters should comply with the following regulations: 1. The calculated indoor heating temperature in winter should be set to 20 °C; 2. The calculated air change rate for heating in winter should be set to 0.5 h−1.	Add “Defined indoor cooling temperature calculation methods and maximum allowable summer ventilation rates (air changes per hour) for cold regions”.
	4.2.2	In Cold Zone B (Zone 2B), the solar heat gain coefficient for exterior windows in summer should not exceed the limit values specified in 4.2.2 of the standard, and the solar heat gain coefficient for summer skylights should not be greater than 0.45. (No requirements for Cold Zone 2A.)	Add “Specific Solar Heat Gain Coefficient (SHGC) limits for windows in Cold Subzone 2A”.
	4.2.4	Buildings in Cold Zone B should preferably have horizontal shading installed on the south-facing windows (including the transparent parts of balconies).	Revise “Cold Zone B” to “Cold Region”.
		When there is building shading, the exterior windows and skylights in Cold Zone B should consider the effect of shading.	Revise “Cold Zone B” to “Cold Region”.

), balancing water conservation and overheating mitigation under intensifying warming trends.

4.2. Sensitivity Analysis Discussion

The article employs two types of sensitivity analysis, SRC and SRRC, to investigate the impact of variable parameters of the building envelope on the performance objectives of folk houses in the arid regions of Xinjiang. The following can be seen from Figure 9:

The roof thermal parameters UR and SRA_R significantly affect HEC and TDH. Among them, UR has the highest sensitivity coefficient and exerts the most significant influence on HEC and TDH, which is similar to the research findings of Yang Liu [71]. Under different climate scenarios, the sensitivity coefficient of UR to HEC remains relatively stable, while the sensitivity coefficient to TDH gradually decreases with climate warming.

Wall U-values (UW) show orientation-dependent impacts: south-facing walls have maximum HEC sensitivity (3.5% < |sensitivity| < 9%), while west walls show minimal effects. In the Turpan area, the east wall has the most significant impact, which is consistent with the research results of Yang Liu [71]. TDH sensitivity remains low (|sensitivity| < 3%) for all orientations. West-wall U-values (UW_W) exhibit the most significant TDH sensitivity increase by 2080, followed by the north (UW_N) and east (UW_E) walls. South-wall sensitivity remains stable, suggesting orientation-specific climate interactions.

Wall solar absorption coefficients (SRAs) minimally affect HEC (|sensitivity| < 2%) and slightly affect TDH (|sensitivity| < 3.5%), with east walls marginally outperforming west walls. No significant temporal trends emerge despite regional fluctuations.

Window U-values moderately influence HEC, showing south-facing sensitivity declines versus north-facing increases under warming. For TDH, the impact is minimal (|sensitivity| < 2%), with low sensitivity coefficients that remain relatively stable under different climate scenarios.

The SHGC value of windows has a relatively small effect on HEC (|sensitivity| < 3%). For TDH, the impact is more significant, with higher sensitivity coefficients (1.5% < |sensitivity| < 6%). The south-facing windows’ sensitivity increases gradually with climate warming.

The WWR value has a relatively small impact on HEC, with minimal differences between the north and south orientations and low sensitivity coefficients (|sensitivity| < 3.5%). For TDH, the impact is more significant, with higher sensitivity coefficients (0.5% < |sensitivity| < 16.5%). The sensitivity of windows exhibits varying trends across different regions under the influence of climate warming.

Ground U-values (UG) strongly affect TDH (11.5% < |sensitivity| < 34.5%), with Turpan/Hotan showing sensitivity increases by 2080. HEC sensitivity remains stable regionally (1% < |sensitivity| < 6%).

Door U-values (UD) demonstrate negligible impacts on HEC (|sensitivity| < 1%) and TDH (|sensitivity| < 1.5%), confirming their low optimization priority.

The analysis demonstrates that roof thermal parameters (UR and SRA_R) exert the most substantial influence on heating energy consumption (HEC) and thermal discomfort hours (TDH) in Xinjiang’s arid region folk houses, with UR showing dominant effects. Wall U-values (UW) significantly affect energy consumption but minimally impact thermal discomfort. Window parameters (U-value, SHGC, and WWR) moderately influence both indicators, demonstrating greater sensitivity in TDH regulation. Ground U-values (UG) substantially affect both HEC and TDH. Optimization of folk house envelopes in this region should prioritize these parameters, particularly roof thermal properties, while considering climate scenarios, regional variations, and parameter characteristics to achieve concurrent improvements in HEC and TDH.

4.3. Pareto-Optimal Solution Set Based on the Dual Entropy-Weighted TOPSIS Method

This study aims to enhance climate adaptability in Xinjiang’s arid region folk houses through building envelope optimization, particularly under climate warming scenarios. The NSGA-II algorithm was applied across multiple climate scenarios, yet Section 3.3 only derived single-scenario optimal solutions per building type in each region. To obtain long-term climate-adaptive envelope parameters, the Section 3.3 balanced solutions underwent secondary evaluation via entropy-weighted TOPSIS, identifying region-specific optimization parameters for individual folk house types. The optimized exterior envelope parameters of regional folk houses’ optimal equilibrium solutions under 2050/2080 climate scenarios were simulated in the 2024 observational meteorological year model to quantify heating energy consumption (HEC), thermal discomfort hours (TDHs), and net present value (NPV).

The specific steps are as follows:

• The optimized exterior envelope parameters of regional folk houses’ optimal equilibrium solutions under 2050/2080 climate scenarios were simulated in the 2024 observational meteorological year model to quantify HEC, TDH, and NPV.
• Then, both parameter sets’ triple-indicator outputs were evaluated against 2024 baseline data using entropy-weighted TOPSIS, measuring Euclidean distances to positive and negative ideal solutions.
• The relative closeness of the two solutions were calculated under the 2024 observational meteorological year scenario to obtain an optimized parameter for the exterior envelope structure with long-term climatic adaptability.

After the above steps, this study obtained 12 sets of optimal equilibrium solution sets with 30~60 years of climatic adaptability, and the results are shown in

Table 12. The optimal solution set is based on the dual entropy-weighted TOPSIS method.

Region	Year	Type-Indv	TR /m	SRA_R	TG /m	WWR_N	UWin_N / W/m2·K	TW_N /m	WWR_S	UWin_S /W/m2·K	SHGC_S	TW_S /m	TW_E /m	TW_W /m
Kashgar	2050	A-33	0.099	0.88	0.077	0.1	1.32	0.143	0.11	1.21	0.52	0.13	0.109	0.125
	2050	B-2	0.104	0.94	0.096	0.1	1.23	0.106	0.12	1.27	0.69	0.111	0.141	0.144
	2080	C-7	0.077	0.88	0.082	0.1	1.25	0.127	0.14	1.34	0.46	0.113	0.11	0.128
Turpan	2080	A-3	0.071	0.9	0.083	0.1	1.56	0.116	0.22	1.61	0.7	0.113	0.146	0.13
	2080	B-4	0.095	0.8	0.076	0.1	1.47	0.106	0.5	1.2	0.69	0.133	0.133	0.139
	2050	C-10	0.076	0.91	0.082	0.1	1.24	0.131	0.5	1.2	0.7	0.128	0.144	0.13
Kuqa	2050	A-17	0.086	0.93	0.086	0.1	1.67	0.142	0.1	1.45	0.69	0.113	0.12	0.097
	2050	B-3	0.061	0.95	0.091	0.1	1.21	0.153	0.11	1.25	0.63	0.114	0.109	0.129
	2080	C-2	0.073	0.92	0.09	0.1	1.24	0.135	0.1	1.23	0.69	0.097	0.137	0.122
Hotan	2080	A-2	0.103	0.62	0.103	0.1	1.28	0.121	0.11	1.44	0.44	0.144	0.108	0.137
	2080	B-5	0.069	0.49	0.086	0.1	1.28	0.149	0.13	1.33	0.31	0.109	0.109	0.143
	2080	C-10	0.076	0.52	0.075	0.1	1.26	0.114	0.2	1.37	0.69	0.128	0.126	0.132

and

Table A3. Supplement to optimal solution set based on the dual entropy-weighted TOPSIS method.

Region	Year	Type-Indv	TDH/h	HEC/kW·h/(m2·a)	NPV/CNY
Kashgar	2050	A-33	1142	58.717	35,714
	2050	B-2	1099	57.15	65,048
	2080	C-7	1088	52.127	53,430
Turpan	2080	A-3	2798	56.044	37,887
	2080	B-4	2907	25.487	79,709
	2050	C-10	2820	33.691	52,730
Kuqa	2050	A-17	1228	69.041	44,131
	2050	B-3	1200	73.005	69,509
	2080	C-2	1155	72.913	51,023
Hotan	2080	A-2	1264	51.435	38,264
	2080	B-5	1281	62.404	39,677
	2080	C-10	1228	48.578	31,340

.

Analysis of Table 12 and Table A3 reveals that entropy-weighted TOPSIS-optimized folk houses under the 2024 climate scenario demonstrate a heating energy consumption (HEC) ranging from 16 to 60 kW·h/(m²·a), thermal discomfort hours (TDHs) of predominantly 1200–1700 h, and net present values (NPVs) spanning CNY 31,000–80,000. Turpan exhibits exceptional TDH values between 2798 and 2907 h due to its extreme hot-aridity.

Comparative analysis of Table 10, Table 12 and Table A3 reveals that under 2024–2050/2080 forecasted climate scenarios, the equilibrium solutions achieve TDH reductions of 15–52.93% in most regions and 5.54–10.8% in the extremely hot-arid Turpan region. HEC reductions reached 51.54–84.76%, while NPV ranged between CNY 31,000 and 85,000.

TR analysis shows distinct patterns across folk house types: Type A roofs range 0.071–0.103 m with dispersed distribution, while SRA_R spans 0.88–0.93 regionally, decreasing to 0.62 in Hotan. Type B demonstrates greater TR variability (0.061–0.104 m) with SRA_R values of 0.8–0.94 regionally and 0.49 in Hotan. Type C exhibits tightly clustered TR (0.073–0.076 m) and consistent SRA_R ranges (0.88–0.92 regionally, 0.52 in Hotan), reflecting climate-responsive design adaptations.

TW analysis identifies distinct thickness patterns: TW_N 0.106–0.153 m, TW_S 0.097–0.144 m, TW_E 0.108–0.146 m, and TW_W 0.097–0.144 m. While these orientations exhibit dispersed insulation distributions, Type C folk houses show targeted western wall optimization, with a narrower TW_W range of 0.122–0.132 m, reflecting climate-responsive design prioritization.

External window analysis reveals distinct thermal performance patterns across Xinjiang’s folk houses: north window U-values (UWin_N) range 0.12–0.132 W/(m²·K) regionally, versus 1.24–1.56 W/(m²·K) in Turpan, while south windows (UWin_S) span 0.12–0.145 W/(m²·K) generally and 1.2–1.61 W/(m²·K) in Turpan. Solar heat gain coefficients (SHGC_S) cluster tightly at 0.63–0.7 in Turpan/Kuqa but disperse broadly (0.31–0.69) in Kashgar/Hotan. Window-to-wall ratios show regional specialization—WWR_N remains 0.1 universally, whereas WWR_S concentrates at 0.1–0.2 regionally but maximizes toward 0.5 in Turpan. This minimization–maximization trend in Turpan aligns with Song et al.’s [23] multi-objective optimization findings for local folk houses.

For the ground, TG is relatively diffuse in all types of folk houses in various regions, mainly distributed between 0.075 and 0.103 m.

Analysis of optimized envelope parameters reveals distinct concentration patterns in key thermal variables—Hotan’s SRA_R values, Type C folk houses’ TR and TW_W parameters, alongside the UWin_N, WWR_N, and WWR_S values, exhibit clustered distributions. Other envelope parameters demonstrate dispersed ranges from regional climatic variations and typological differences. Future designs should prioritize these prominent climate-sensitive parameters while conducting localized adaptive adjustments for other variables through sensitivity analyses across multiple climate scenarios.

4.4. Decision Support Framework

To assist policymakers and architects in prioritizing optimization solutions under diverse objectives, a two-step decision-making framework is suggested.

4.4.1. Step 1: Objective Prioritization

Economic-driven selection: For regions with limited budgets (e.g., rural Xinjiang), prioritization should be given to solutions with optimal solutions for NPV (Table 9). For instance, Type B folk houses in Turpan (B-4: NPV = CNY 85,596) should balance retrofit costs with long-term energy savings.

Thermal comfort-driven selection: In extreme hot-arid zones (e.g., Turpan), prioritization should be given to solutions with TDH reduction >10% (e.g., Type C-30: TDH = 3507 h, 5.56% reduction).

Balanced approach: The entropy-weighted TOPSIS closeness coefficient (Equation (17)) should be used to rank solutions, ensuring trade-offs between NPV, HEC, and TDH.

4.4.2. Step 2: Climate Scenario Adaptation

For short-term retrofits (e.g., 2024–2035), solutions optimized under the 2024 observed climate (e.g., Kashgar Type A-2: HEC = 65.63 kW·h/m²·a) should be adopted.

For long-term resilience (e.g., 2050–2080), solutions validated across multiple scenarios (Table 12), such as Hotan Type B-5 (TDH = 1655 h under 2080 SSP3–7.0), should be selected.

4.5. Global Comparative Insights and Socioeconomic Implications

This study’s findings resonate with global efforts to enhance climate resilience in arid regions, while also demonstrating localized socioeconomic impacts.

International adaptability: Compared to similar studies in global arid regions, this research demonstrates broader optimization potential. For instance, the HEC reduction range (51.54–84.76%) surpasses the 35–60% reported for Iranian vernacular houses [9], likely due to the integrated NSGA-II and entropy-weighted TOPSIS framework that simultaneously optimizes energy, comfort, and cost. In Algeria’s Ouargla, traditional buildings achieved 28% lower cooling energy than modern counterparts [19], whereas our optimized solutions reduced Turpan’s TDH by 5.54–10.8% under 2080 SSP3–7.0—highlighting the value of multi-temporal climate data in addressing future overheating risks. Notably, the “30–60 year climate resilience” metric proposed here advances the field beyond static TMY-based analyses, offering a dynamic adaptation paradigm for policymakers.

Economic and health benefits: The optimization outcomes hold direct socioeconomic significance for Xinjiang’s rural communities. Based on local coal prices (≈800 CNY/ton) and average house area (120 m²), the 84.76% HEC reduction equates to annual savings of CNY 1320–1950 per household—substantial for low-income regions where energy costs are a critical expenditure burden, particularly in rural areas with limited access to modern energy infrastructure [23]. Meanwhile, reducing TDH by 52.93% in Zone A could mitigate heat-related health risks, as studies in temperate climates have demonstrated that mortality risk escalates by 2.1–3.8% per °C above region-specific heat thresholds [4]. These benefits align with China’s rural revitalization goals, demonstrating how climate-adaptive retrofits can concurrently enhance energy equity and public health.

Policy synergy: This study directly aligns with multiple United Nations SDGs. The reduction in heating energy consumption (HEC) by 51.54–84.76% supports SDG 7 (Affordable and Clean Energy) by lowering dependence on fossil fuels for winter heating. Simultaneously, the mitigation of thermal discomfort hours (TDHs) by 15–52.93% contributes to SDG 3 (Good Health and Well-being), reducing heat-related morbidity in vulnerable populations. Furthermore, the optimized retrofitting strategies for vernacular architectures advance SDG 11 (Sustainable Cities and Communities) by preserving cultural heritage while enhancing climate resilience. These synergies highlight the potential of integrated building optimization to address interconnected environmental, social, and economic challenges in arid regions.

4.6. Limitations and Future Research Directions

4.6.1. Limitations

The study’s reliance on the SSP3–7.0 high-emission scenario for climate projections, while aligned with current extreme weather trends, overlooks potential low-carbon pathway impacts on building adaptability. By excluding comparative analysis across emission scenarios (e.g., SSP1–2.6), the research may underestimate policy and technological influences on Xinjiang folk houses’ climate resilience. Future work should integrate multi-scenario climate data to dynamically assess arid-region adaptations.

Economic evaluations omitted critical dynamic factors: climate-driven energy demand shifts (e.g., rising cooling loads offsetting heating savings) and technological cost reductions for retrofits like high-reflectivity coatings. These gaps in the net present value (NPV) framework necessitate empirical validations to improve economic forecasting accuracy for envelope optimization strategies.

Findings reveal that envelope optimization alone proves inadequate to address extreme heat in extremely hot-arid regions under 2080 projections, failing to resolve indoor overheating in these zones compared to historical climate data. Subsequent research should develop multi-technology passive design frameworks, coupling envelope upgrades with complementary strategies to enhance occupant comfort under escalating thermal stress.

4.6.2. Future Research Directions

Future studies should expand the scope of climate adaptability research in three directions:

• Dynamic climate scenario analysis: Incorporating SSP1–2.6 (low-emission pathway) and SSP5–8.5 (high-emission pathway) scenarios would clarify how policy interventions (e.g., carbon taxation) influence building performance. This could inform adaptive design standards under varying decarbonization trajectories.
• Socio-cultural acceptance: Field surveys and participatory workshops are needed to assess residents’ preferences for retrofitting strategies. For example, replacing traditional rammed earth walls with modern insulation may face resistance due to cultural attachment. A socio-technical approach could balance innovation and heritage preservation.
• Cross-regional benchmarking: Comparative studies with other arid zones (e.g., the Atacama Desert or the Arabian Peninsula) would validate the universality of the proposed framework and identify region-specific optimization priorities.

5. Conclusions

This study developed a climate adaptability evaluation and multi-objective optimization framework for folk house envelopes in Xinjiang’s arid region by integrating historical, current, and future climate data. For the first time, a long-term climate-resilient optimization paradigm was proposed, addressing the limitations of traditional studies that rely on single-period climate data and overlook modern materials and summer overheating risks in hot-arid zones. The key findings include the following:

• Climate adaptability of folk houses: Rammed earth dwellings demonstrate optimal climate adaptability in most Xinjiang arid regions. Brick–concrete structures exhibit superior summer performance in extreme hot-arid zones, while brick–wood constructions require prioritized optimization due to their poorest performance.
• Algorithm optimization effectiveness: The NSGA-II algorithm achieves efficient multi-objective optimization. Under historical-to-future climate scenarios, heating energy consumption (HEC) decreases by 76.5–90.2%, thermal discomfort hours (TDHs) reduce by 5.6–53.7% in most regions and by 5.6–22.7% in extreme hot-arid zones, and net present values (NPVs) reach CNY 40,000–97,000.
• Decision-making analysis: Solutions selected through the dual entropy-weighted TOPSIS method demonstrate 30–60 years of climate resilience. Optimized solutions achieve 51.5–84.8% HEC reduction and 15–52.9% TDH reduction while ensuring economic feasibility, providing differentiated strategies for Xinjiang’s sub-regions.
• Policy recommendations: Current standards neglect summer overheating protection and should be revised to incorporate “dual-season adaptability” requirements. Mandatory shading, night ventilation, and high thermal mass strategies could improve summer thermal comfort by 36% in Kashgar.
• Sensitivity analysis: Roof and ground thermal parameters (UR, SRA_R, and UG) dominate energy–comfort tradeoffs. Window properties (SHGC and WWR) significantly regulate TDH.

Innovative contributions include the following: (1) a cross-temporal climate data-driven optimization method spanning historical–current–future periods, overcoming limitations of static TMY data; (2) the revelation of performance reversal between traditional and modern materials under extreme heat, providing dynamic climate adaptation insights; and (3) a coupled decision model integrating dual entropy-weighted TOPSIS with NSGA-II to balance energy, comfort, and economic goals.

The limitations include the following: (1) reliance on SSP3–7.0 high-emission scenarios without low-carbon scenario comparisons; (2) exclusion of cooling demand growth and dynamic technology costs in economic analysis; and (3) optimization was limited to envelope structures without passive technology integration. Future work should expand to multi-scenario analyses and explore synergistic designs combining envelope optimization with courtyards and renewables.

This research provides scientific tools for enhancing climate resilience in Xinjiang’s folk houses, with methodologies applicable to global arid regions. By quantifying historical–future climate interactions, it reveals evolutionary pathways for vernacular architecture under climatic challenges, offering a paradigm for sustainable design that harmonizes cultural heritage preservation with climate adaptation. The findings directly inform revisions to China’s Green Building Evaluation Standards and guide sustainable construction in Belt and Road arid zones, demonstrating the integration of cultural conservation with low-carbon goals. The work holds theoretical and practical significance for achieving dual-carbon targets in rural-urban building transitions.