Evaluating the Microclimatic Performance of Elevated Open Spaces for Outdoor Thermal Comfort in Cold Climate Zones

Abstract

Improving outdoor thermal comfort is a critical objective in urban design, particularly in densely built urban environments. Elevated semi-open spaces—outdoor areas located beneath raised building structures—have been recognized for enhancing pedestrian comfort by improving airflow and shading. However, previous studies primarily focused on warm or temperate climates, leaving a significant research gap regarding their thermal performance in cold climate zones characterized by extreme seasonal variations. Specifically, few studies have investigated how these spaces perform under conditions typical of northern Chinese cities like Xi’an, which is explicitly classified within the Cold Climate Zone according to China’s national standard GB 50176-2016 and experiences both severe summer heat and cold winter conditions. To address this gap, we conducted field measurements and numerical simulations using the ENVI-met model (v5.0) to systematically evaluate the microclimatic performance of elevated ground-floor spaces in Xi’an. Key microclimatic parameters—including air temperature, mean radiant temperature, relative humidity, and wind velocity—were assessed during representative summer and winter conditions. Our findings indicate that the height of the elevated structure significantly affects outdoor thermal comfort, identifying an optimal elevated height range of 3.6–4.3 m to effectively balance summer cooling and winter sheltering needs. These results provide valuable design guidance for architects and planners aiming to enhance outdoor thermal environments in cold climate regions facing distinct seasonal extremes.

1. Introduction

Rapid urbanization has led to a substantial increase in energy consumption across many cities, contributing to the degradation of the urban climate and environment—particularly in developing countries [1]. Among the resulting challenges, the urban heat island (UHI) effect—characterized by elevated surface and air temperatures in urban areas compared to their rural counterparts—has attracted significant research attention [2]. The UHI effect not only exacerbates thermal discomfort but also intensifies energy demand for indoor cooling during hot periods [3]. For instance, studies indicate that energy consumption in urban areas is approximately 81% higher than in rural regions [4], underscoring the urgent need for effective UHI mitigation strategies.

As urban residents increasingly engage in outdoor activities such as cycling, jogging, and walking, the quality of the outdoor thermal environment has become a key factor influencing urban livability. A thermally comfortable outdoor setting encourages longer periods of outdoor stay, potentially reducing dependence on mechanical cooling and lowering overall indoor energy use.

Outdoor thermal comfort is commonly defined as “the condition of mind that expresses satisfaction with the outdoor thermal environment” [5]. Since the 1980s, interest in outdoor thermal comfort has grown, especially concerning pedestrian experiences in urban plazas, squares, and street canyons. Thermal comfort in outdoor settings is governed by the principles of human thermo-physiology and the body’s heat balance with its surroundings. This exchange of heat—via radiation, evaporation, conduction, and convection—is influenced by both human-related factors (e.g., activity level, clothing insulation) and environmental parameters that shape the thermal context (Figure 1) [6].

Vital physical parameters—such as air temperature, mean radiant temperature, wind velocity, and relative humidity—are interrelated and form the basis for calculating outdoor thermal comfort.

Numerous indices have been developed to quantify thermal comfort; over 165 indicators have been identified in previous research [7]. Among them, widely recognized indices include the Standard Effective Temperature (SET*) [8] and Predicted Mean Vote (PMV) [9]. However, these two indices were originally developed for indoor environments and are generally not suitable for outdoor applications due to the dynamic and complex nature of outdoor climates. While the PMV index may have limited outdoor applicability as an indicator for the transition from thermal discomfort to physiological heat stress.

The Universal Thermal Climate Index (UTCI) [10] and Physiological Equivalent Temperature (PET) [11] are examples of such indices. In this study, PET—derived from the Munich Energy Balance Model for Individuals (MEMI) [12]—was adopted due to its strong suitability and established validity in outdoor thermal comfort research.

Extensive research has been conducted on strategies to enhance outdoor thermal comfort, focusing on factors such as urban canyon geometry and orientation [2,13,14,15], vegetation coverage [16], building density [17], and surface thermal properties of pavements and façades [18,19,20]. For instance, increasing the aspect ratio (H/W) of urban canyons has been shown to reduce maximum daytime thermal stress by up to 4 °C PET [21], while the introduction of vegetation, such as trees, may result in a PET reduction of approximately 3 °C during daytime hours [22]. Likewise, materials with high albedo for pavements and low albedo for vertical surfaces have been found to improve local thermal conditions [23].

However, most of these studies have been concentrated in warm European climates [24,25,26,27], and only more recently has research emerged in various climate zones across China, including hot-summer and cold-winter climates [28,29,30], hot-summer and warm-winter regions [31,32,33,34,35], and severely cold areas [36,37,38]. Despite these developments, studies specifically addressing thermal comfort in cold climate zones as defined by the Chinese national standard GB 50176-2016 remain limited [39], even though such regions experience intense heat stress during the summer (Figure 2). Moreover, existing studies have primarily focused on traditional outdoor environments such as parks [40,41,42], streets [1], squares [43,44], and campuses [45], while research on the microclimatic performance of semi-open areas beneath buildings is still scarce.

One architectural approach that creates such semi-open spaces is the “elevated design,” which places part of a building above ground level on stilts or columns, generating shaded public areas below. Commonly seen in high-density cities like Hong Kong, this design provides sheltered outdoor space that can mitigate solar exposure on hot days. Studies have shown that wind velocity tends to be higher in elevated open spaces [46], and thermal simulations have demonstrated that these areas can exhibit PET reductions of up to 15 °C compared to adjacent open areas [47]. However, previous investigations of elevated buildings have primarily been conducted in hot-summer and warm-winter regions of southern China [46,47,48,49,50].

Xi’an—a city explicitly classified within the Cold Climate Zone according to GB 50176-2016—is experiencing increasingly frequent and intense summer heatwaves, becoming one of China’s “new top ten hottest furnaces.” Although Xi’an lies within the Cold Climate Zone, the city faces distinct seasonal extremes, including notably hot summers and cold winters. Thus, urban design strategies for Xi’an must concurrently address both summer heat stress and winter thermal discomfort to create genuinely climate-responsive urban environments. In this study, we aim to evaluate the thermal performance of elevated open spaces in Xi’an and identify optimal elevation parameters through a combination of on-site measurements and ENVI-met simulations. The findings are intended to inform design guidelines for elevated educational buildings and other semi-open structures in cold climate zones experiencing pronounced seasonal extremes, similar to Xi’an.

2. Materials and Methods

2.1. Climate Conditions of the Research City in China

Xi’an is geographically situated between 107.4° E and 109.49° E longitude and 33.42° N and 34.45° N latitude. According to the Thermal Design Code for Civil Building (GB50176-93) [39], the city is classified within the cold climate zone of China (Figure 3). Xi’an experiences four distinct seasons: a mild spring, a hot summer, a cool autumn, and a cold winter [51].

Meteorological data from 1971 to 2011 indicate that July has the highest average monthly temperature at approximately 27 °C, with a maximum recorded temperature of 31.7 °C. Conversely, January exhibits the lowest mean monthly temperature at −0.4 °C, with minimum temperatures reaching −3.4 °C. The annual mean relative humidity ranges between 57.2% and 77.4% [52].

2.2. The Analyzed Site

The field measurements were conducted on the Weishui Campus of Chang’an University, located in Xi’an, China. The test site was positioned within a corridor beneath an elevated building, featuring approximate dimensions of 15 m in length, 8.4 m in width, and 3.5 m in height (Figure 4).

In this study, one measurement point was selected, located beneath the elevated structure (referred to as the elevated test point) (Figure 5 and Figure 6). Meteorological parameters collected at this point were used for validating the ENVI-met simulation results. Boundary conditions for the ENVI-met simulations were derived from local meteorological data obtained from a nearby official weather station, ensuring representative and accurate input conditions for the model.

2.3. The Experimental Design

Field measurements were conducted separately during the summer and winter seasons, from 8 to 10 July 2020, and 4 to 6 January 2021, respectively, between 9:00 a.m. and 6:00 p.m. each day. These periods represent the typical hottest and coldest days of the year in Xi’an [53] (

Table 1. The meteorological conditions during the measured days.

Measured Date	Mini Temperature	Max Temperature
8 July 2020	26 °C	36 °C
9 July 2020	25 °C	37 °C
10 July 2020	22 °C	29 °C
4 January 2021	−2 °C	3 °C
5 January 2021	−8 °C	1 °C
6 January 2021	−9 °C	0 °C

).

The monitored microclimatic parameters included air temperature (Ta), relative humidity (Rh), wind velocity (Va), and globe temperature (Tg). Air temperature and relative humidity were recorded using HOBO data loggers (accuracy: ±0.2 °C for air temperature and ±2.5% RH for relative humidity; resolution: 0.02 °C and 0.05% RH, respectively). Wind velocity was measured using the Kestrel 5500 anemometer [54], a fan-type device with an accuracy of ±3% of the reading or ±0.1 m/s, a measurement range of 0.6–40 m/s, and a resolution of 0.1 m/s. This anemometer provides reliable readings primarily above 0.6 m/s and may exhibit reduced accuracy at lower wind speeds typically encountered in outdoor thermal comfort assessments. Globe temperature was measured using the HD32.2 WBGT Index Instrument [54] (accuracy: ±0.1 °C; range: −10 to 100 °C; resolution: 0.1 °C) (

Table 2. The detailed introduction of the instruments in this study.

Instrument	Parameter	Accuracy	Range	Resolution
Kestrel	Wind velocity	±3% of reading or ±0.1 m/s	0.6–40 m/s	0.1 m/s
HOBO	Air temperature Relative humidity	±0.2 °C ±2.5%	−40 °C ± 70 °C 0–100%	0.02 °C 0.05%
HD32.2 WBGT Index	Globe temperature	$\pm$0.1 °C	−10–100 °C	0.1 °C

).

We explicitly selected the 150 mm globe thermometer due to its higher measurement accuracy in quantifying radiant heat exchange, as extensively documented in the literature [54]. Although the 150 mm globe exhibits a comparatively longer response time than the 50 mm version, its improved accuracy in outdoor thermal comfort assessments justified this choice. All instruments were mounted at a standard pedestrian height and securely fixed to the ground. Measurements were recorded at one-minute intervals throughout the monitoring periods to capture detailed temporal variations.

2.4. The Introduction of the ENVI-Met Modeling

ENVI-met is a three-dimensional, non-hydrostatic microclimate simulation platform based on the fundamental principles of thermodynamics and fluid dynamics. It is specifically designed to evaluate the complex interactions among soil, vegetation, air, and buildings within urban environments. Figure 7 presents a detailed overview of the ENVI-met model layout used in this study. The simulation framework comprises several sub-models, including a one-dimensional (1D) boundary layer model, a three-dimensional (3D) soil model, and a 3D atmospheric model. The 1D model simulates vertical atmospheric processes from the ground surface (0 m) up to 2500 m in height. The 3D domain is discretized into grid cells, with spatial resolution defined in the model input files. To enhance computational accuracy near the pedestrian level, the lowest five vertical layers are refined using a stretching method, where the vertical increment is set to ∆z = 0.2∆z.

In terms of resolution, ENVI-met supports a spatial grid size ranging from 0.5 to 10 m, and a temporal resolution from 1 to 10 s, allowing for flexible and high-fidelity simulations of urban microclimates [55].

Previous studies have extensively utilized ENVI-met to assess outdoor thermal environments and human thermal comfort in urban contexts [2,16,21,54,55,56,57,58]. A significant advancement in this regard is the release of a critical version of ENVI-met, which introduced the capability to import 3D models from Rhinoceros. This update addresses the limitations of earlier versions that lacked the flexibility to model complex urban forms, thereby making the tool more accessible and practical for architects and urban designers.

In the context of urban design and planning, it is essential to consider seasonal variations in solar radiation—particularly the need to block excessive solar gain during summer while permitting beneficial radiation during winter. This study leverages the latest ENVI-met features to simulate and analyze the thermal environment in semi-open spaces beneath elevated buildings during both summer and winter seasons.

A key climatic variable in this context is the solar altitude angle, which differs significantly between seasons. To take advantage of this, designers commonly incorporate architectural elements such as overhangs, awnings, and shutters, which serve to mitigate overheating in summer and enhance solar heat gain in winter.

Similarly, when designing elevated buildings, the thermal performance of the open space beneath the structure must also account for seasonal thermal variability. Identifying an optimal height range for these semi-outdoor spaces can substantially improve outdoor thermal comfort and contribute to reducing building energy consumption year-round (Figure 8).

2.5. The Framework of This Study

Figure 9 illustrates the overall research framework, which consists of three main steps:

• Field Measurements: On-site microclimatic data were collected during two distinct periods—8–10 July 2020 (summer) and 4–6 January 2021 (winter)—to capture seasonal variations in thermal conditions.
• Model Validation: The ENVI-met simulation results were validated against the field measurements to ensure the reliability and accuracy of the model outputs.
• Analysis and Conclusions: A detailed analysis of the simulation results was conducted to derive key findings, identify optimal design parameters, and formulate conclusions relevant to thermal comfort optimization in elevated open spaces.

3. Analyzed Results

3.1. The ENVI-Met Boundary Conditions

Field measurements collected over six consecutive days—covering both summer and winter seasons—were used to drive the ENVI-met simulations via the Simple Forcing option. The initial and boundary conditions applied in the simulations are detailed in

Table 3. Initial settings of this study.

Parameters		Value
Material properties (elevated surface)	Tile	Roughness length	0.01
		Albedo	0.5
		Emissivity	0.9
Boundary conditions	Air temperature	Table 4, Table 5, Table 6, Table 7, Table 8 and Table 9
	Relative humidity
	Turbulent model	Kinetic energy (TKE) model
	Case-1 (Summer)
	Wind speed(m/s)	1.5 m/s
	Wind direction (°)	145
	Grid in dx (m)	1
	Grid in dy (m)	1
	Grid in dz (m)	0.5
	Number of x grid	200
	Number of y grid	200
	Number of z grid	30
Simulation	Starting day	8–10 July 2020
	Starting time	0:00 a.m.
	Total simulation time	72 h
	8 July	24 h
	9 July	24 h
	10 July	24 h
Boundary conditions	Case-2 (Winter)
	Wind speed (m/s)	1.3 m/s
	Wind direction (°)	45
	Grid in dx (m)	1
	Grid in dy (m)	1
	Grid in dz (m)	0.5
	Number of x grid	200
	Number of y grid	200
	Number of z grid	30
Simulation	Starting day	4–6 January 2021
	Starting time	0:00 a.m.
	Total simulation time	72 h
	4 January	24 h
	5 January	24 h
	6 January	24 h

,

Table 4. Initial air temperature and relative humidity in simulation (8 July).

Time	Air Temperature (°C)	Relative Humidity (%)	Time	Air Temperature (°C)	Relative Humidity (%)
0:00 a.m.	24.6	67.3	12:00 a.m.	35.4	38.8
1:00 a.m.	24	71	1:00 p.m.	36	37.5
2:00 a.m.	24.5	68.5	2:00 p.m.	36.3	36.1
3:00 a.m.	25.6	63.9	3:00 p.m.	36.9	34.5
4:00 a.m.	26.8	62	4:00 p.m.	36.1	34.9
5:00 a.m.	28.9	59.8	5:00 p.m.	35.2	35.9
6:00 a.m.	29.1	58.7	6:00 p.m.	34.2	38.8
7:00 a.m.	30.2	58.1	7:00 p.m.	32.5	45.6
8:00 a.m.	31	54.8	8:00 p.m.	29.8	48.3
9:00 a.m.	31.6	54.5	9:00 p.m.	27	52.8
10:00 a.m.	33.4	44.9	10:00 p.m.	25.9	61.8
11:00 a.m.	34.6	42.3	11:00 p.m.	25.1	62.5

,

Table 5. Initial air temperature and relative humidity in simulation (9 July).

Time	Air Temperature (°C)	Relative Humidity (%)	Time	Air Temperature (°C)	Relative Humidity (%)
0:00 a.m.	27	65.1	12:00 a.m.	34.8	39.8
1:00 a.m.	26.3	71.1	1:00 p.m.	35.8	37.6
2:00 a.m.	26.8	68.3	2:00 p.m.	36.5	36
3:00 a.m.	27.6	66.5	3:00 p.m.	36.9	35.5
4:00 a.m.	27.9	62.1	4:00 p.m.	36.6	36.2
5:00 a.m.	28.3	59.5	5:00 p.m.	36.1	38.1
6:00 a.m.	28.9	58.1	6:00 p.m.	35.3	39.1
7:00 a.m.	29.4	56.3	7:00 p.m.	34.1	40.5
8:00 a.m.	29.9	55.1	8:00 p.m.	32.3	42.1
9:00 a.m.	30.6	53.2	9:00 p.m.	30.8	48.6
10:00 a.m.	33.3	44.9	10:00 p.m.	28.9	55.1
11:00 a.m.	33.8	40.9	11:00 p.m.	27.8	61.2

,

Table 6. Initial air temperature and relative humidity in simulation (10 July).

Time	Air Temperature (°C)	Relative Humidity (%)	Time	Air Temperature (°C)	Relative Humidity (%)
0:00 a.m.	24.1	72.2	12:00 a.m.	29.9	61.2
1:00 a.m.	23.2	75.1	1:00 p.m.	30.1	58.9
2:00 a.m.	24.2	73.1	2:00 p.m.	31.2	58
3:00 a.m.	24.9	71.5	3:00 p.m.	32	57
4:00 a.m.	25.6	69.9	4:00 p.m.	31	57.9
5:00 a.m.	26.9	68.1	5:00 p.m.	30.5	58.9
6:00 a.m.	27.1	67	6:00 p.m.	30	59.3
7:00 a.m.	27.8	66.1	7:00 p.m.	28.5	60.2
8:00 a.m.	28.3	65.1	8:00 p.m.	27.1	63.1
9:00 a.m.	28.6	64.8	9:00 p.m.	26.5	65.6
10:00 a.m.	28.9	63.9	10:00 p.m.	25.4	67.8
11:00 a.m.	29	62.7	11:00 p.m.	24.8	69.1

,

Table 7. Initial air temperature and relative humidity in simulation (4 January).

Time	Air Temperature (°C)	Relative Humidity (%)	Time	Air Temperature (°C)	Relative Humidity (%)
0:00 a.m.	−2.1	51.2	12:00 a.m.	2.91	44.08
1:00 a.m.	−2.5	53.1	1:00 p.m.	3.70	42.96
2:00 a.m.	−1.9	50.8	2:00 p.m.	3.99	44.12
3:00 a.m.	−1.5	49.7	3:00 p.m.	4.40	44.43
4:00 a.m.	−1.1	49.6	4:00 p.m.	5.16	42.96
5:00 a.m.	−0.63	49.1	5:00 p.m.	4.32	44.31
6:00 a.m.	0.16	48.8	6:00 p.m.	3.27	47.41
7:00 a.m.	0.95	48.5	7:00 p.m.	2.51	48.1
8:00 a.m.	1.23	48.4	8:00 p.m.	1.68	48.93
9:00 a.m.	1.77	47.99	9:00 p.m.	0.91	49.21
10:00 a.m.	1.88	47.94	10:00 p.m.	−0.85	49.6
11:00 a.m.	2.32	44.83	11:00 p.m.	−1.63	49.9

,

Table 8. Initial air temperature and relative humidity in simulation (5 January).

Time	Air Temperature (°C)	Relative Humidity (%)	Time	Air Temperature (°C)	Relative Humidity (%)
0:00 a.m.	−3.9	39.5	12:00 a.m.	5.52	26.92
1:00 a.m.	−4.5	39.9	1:00 p.m.	5.52	26.47
2:00 a.m.	−3.5	39.1	2:00 p.m.	6.41	26.36
3:00 a.m.	−2.6	38.8	3:00 p.m.	6.40	26.89
4:00 a.m.	−1.7	38.5	4:00 p.m.	6.43	26.49
5:00 a.m.	−0.86	38	5:00 p.m.	6.87	29.36
6:00 a.m.	0.35	36.1	6:00 p.m.	5.31	31.36
7:00 a.m.	1.15	33.5	7:00 p.m.	3.85	33.2
8:00 a.m.	2.35	31.5	8:00 p.m.	2.56	34.8
9:00 a.m.	3.47	30.17	9:00 p.m.	1.68	36.12
10:00 a.m.	3.32	29.76	10:00 p.m.	0.98	36.85
11:00 a.m.	4.53	27.88	11:00 p.m.	−1.12	38.5

and

Table 9. Initial air temperature and relative humidity in simulation (6 January).

Time	Air Temperature (°C)	Relative Humidity (%)	Time	Air Temperature (°C)	Relative Humidity (%)
0:00 a.m.	−8.3	32.1	12:00 a.m.	0.91	24.1
1:00 a.m.	−9.3	33.5	1:00 p.m.	0.95	23.9
2:00 a.m.	−7.1	30.4	2:00 p.m.	1.10	23.1
3:00 a.m.	−5.1	29.7	3:00 p.m.	1.48	22.8
4:00 a.m.	−4.2	29	4:00 p.m.	1.25	21.5
5:00 a.m.	−3.5	28.15	5:00 p.m.	0.69	21.4
6:00 a.m.	−2.4	25.3	6:00 p.m.	0.30	21.63
7:00 a.m.	−0.8	25.1	7:00 p.m.	−1.36	26.5
8:00 a.m.	0.1	24.9	8:00 p.m.	−3.51	27.85
9:00 a.m.	0.80	24.34	9:00 p.m.	−4.15	28.91
10:00 a.m.	0.85	25.40	10:00 p.m.	−5.89	29.9
11:00 a.m.	0.80	22.79	11:00 p.m.	−7.1	30.5

. These parameters were derived from the on-site field data and supplemented by meteorological records obtained from local weather stations [52].

3.2. The Validation Between the Measured and Simulated Data

The accuracy of the simulation was assessed using a set of statistical metrics commonly applied in outdoor thermal environment studies. Specifically, we evaluated the model performance using Root Mean Square Error (RMSE), Systematic Root Mean Square Error (RMSEs), Unsystematic Root Mean Square Error (RMSEu), and the coefficient of determination (R²). These metrics were calculated at a representative point located beneath the elevated building. To ensure the reliability of the simulation results, the following criteria are generally recommended: RMSE and RMSEs should approach 0, RMSEu should approximate RMSE, and R² should be close to 1 [59]. These indicators provide a robust basis for evaluating the model’s predictive capability. As discussed earlier, thermal indices have become widely adopted for predicting outdoor human thermal comfort. Among them, Mean Radiant Temperature (MRT)—also denoted as TMRT—plays a crucial role. MRT is defined as the uniform temperature of an imaginary enclosure in which the radiant heat exchange between a human body and the surrounding surfaces is equivalent to that in a real, non-uniform environment. The calculation of MRT follows the methodology specified in ISO 7726 [60].

$$T_{\mathit{MRT}}={{[(Tg+273)}^4+{\displaystyle \frac{1.10\times Va\times {10}^8}{\mathsf{\epsilon }\times {\mathrm{D}}^{0.4}}}(\mathrm{Tg}-\mathrm{Ta}\left)\right]}^{{\displaystyle \frac{1}{4}}}-273$$

(1)

where Tg is the globe temperature (°C), Ta is the air temperature (°C), Va is the wind speed (m/s) at the selected point, D is the globe diameter (0.05 m in this study), and ε is the emissivity of the globe (0.95 for a black globe).

The PET index in ENVI-met is calculated based on several key variables, including Ta, Rh, Va, and TMRT. Therefore, validating the accuracy of these simulated parameters is a critical step in ensuring the reliability of the PET output. Figure 10 and

Table 10. The quantitative evaluation of the ENVI-met performance on the evaluation of summer in RMSE, RMSEu, and RMSEs.

Parameter	RMSE	RMSEu	RMSEs
Ta	1.22 °C	1.98 °C	1.03 °C
Rh	6.86%	7.02%	4.85%
Va	0.11 m/s	0.08 m/s	0.065 m/s
TMRT	3.12 °C	3.26 °C	1.89 °C

present the comparison between measured and simulated data for the summer period. The simulation of Ta demonstrated high precision across all measured days, with an average R² of 0.9008 and a RMSE of 1.22 °C. Similarly, Rh and TMRT showed strong agreement with measured values, with R² values of 0.8573 and 0.8653, and RMSE values of 6.86% and 3.12 °C, respectively—indicating consistent model performance.

Although the simulation accuracy for Va was relatively lower than for other parameters, the model still provided a reliable estimation of wind conditions, consistent with previous validation results [16].

For the simulated microclimate in winter (Figure 11,

Table 11. The quantitative evaluation of the ENVI-met performance on the evaluation of winter in RMSE, RMSEu, and RMSEs.

Parameter	RMSE	RMSEu	RMSEs
Ta	0.88 °C	1.21 °C	0.69 °C
Rh	7.65%	8.3%	6.59%
Va	0.19 m/s	0.13 m/s	0.11 m/s
TMRT	1.91 °C	2.09 °C	1.96 °C

), the validation results show a similarly high level of accuracy as observed in the summer. The R² values for Ta, Rh, Va, and TMRT were 0.814, 0.819, 0.6754, and 0.9633, respectively. Correspondingly, the RMSE values were 1.22 °C, 6.86%, 0.11 m/s, and 3.12 °C, respectively. These results confirm that the simulation model performs reliably in winter conditions, particularly for Ta and TMRT, while Va exhibits slightly lower predictive accuracy.

These findings confirm the model’s applicability in supporting climate-responsive design strategies for elevated structures in cold climate urban environments.

3.3. The PET Values of the Existing Scenario

As previously discussed, the thermal comfort index PET has become widely adopted for evaluating outdoor thermal environments. A prior study [61] (

Table 12. The PET values in the cold climate zone of northern China [61].

Thermal Stress	Thermal Sensation	PET (°C)
Extreme cold stress	Very Cold	<−4
Strong cold stress	Cold	−4~8
Moderate cold stress	Cool	8~16
Slightly cold stress	Slightly Cool	16~22
No thermal stress	Neutral	22~28
Slight heat stress	Slightly Warm	28~32
Moderate heat stress	Warm	32~38
Strong heat stress	Hot	38~44
Extreme heat stress	Very Hot	>44

) established a PET-based thermal perception scale specifically tailored for the cold climate zone of northern China. This 9-point classification system includes the categories: Very Cold, Cold, Cool, Slightly Cool, Neutral, Slightly Warm, Warm, Hot, and Very Hot, each corresponding to a specific PET range. This scale is employed in the present study to assess outdoor thermal comfort conditions beneath elevated buildings.

In ENVI-met, the default summer settings define sunrise at 6:00 a.m. and sunset at 7:00 p.m. The simulated PET distribution for the summer scenario is presented in Figure 12. As shown, all three curves exhibit similar trends, with PET values increasing after sunrise and peaking around 2:00 p.m., followed by a gradual decline. During the core observation period from 9:00 a.m. to 6:00 p.m., the site consistently falls within the ‘Hot’ to ‘Very Hot’ PET categories, indicating significant thermal stress throughout most of the daytime period.

Figure 13 illustrates the variation of PET during the winter season. Throughout the daytime period, the entire site predominantly remains within the ‘Cool’ category, indicating moderate thermal discomfort. In contrast, during nighttime hours, PET values drop significantly, falling into the ‘Cold’ and ‘Very Cold’ categories, reflecting the harsh thermal conditions characteristic of winter nights in cold climate zones.

4. Discussion

As summarized in the preceding sections, further discussion is warranted regarding the thermal comfort characteristics of the semi-open space beneath elevated buildings and the corresponding design implications. A key structural component in such configurations is the supporting column, which plays a critical role not only in the load-bearing function but also in shaping the local microclimate. Therefore, the discussion begins by evaluating the influence of column dimensions and placement on the surrounding thermal environment.

4.1. The Impact of the Columns on the Surrounding Thermal Environment

Due to the default grid constraints in ENVI-met, the minimum allowable length and width for model elements is 0.5 m, and all dimensions must be in multiples of this unit. Based on this limitation, a series of simulation cases was constructed using columns with varying widths to specifically evaluate their influence on the surrounding thermal environment, independently of seasonal conditions. In total, five scenarios were modeled, incorporating column widths of 0.5 m, 1.0 m, 1.5 m, 2.0 m, and 2.5 m. All other parameters and boundary conditions were held constant, explicitly to isolate the effect of column dimension alone, rather than assess seasonal differences or variations. The corresponding simulation results, solely focused on verifying the influence of column size, are presented in

Table 13. Simulated PET results under varying column widths.

Column Width	PET
0.5 m
1 m
1.5 m
2 m
2.5 m

and Figure 14.

The results indicate that PET values increase slightly as column width increases; however, the differences are marginal. Specifically, when the column width increases from 0.5 m to 2.5 m, the maximum PET difference observed is only 0.6 °C. This minimal variation clearly suggests that column width itself has a negligible impact on the thermal comfort of the surrounding environment. Therefore, within the design constraints of typical elevated buildings, column dimension can be considered a secondary factor in thermal performance optimization.

4.2. The Impact of the Bottom of the Elevated Building on the Surrounding Thermal Environment

Following the analysis of column width, it is also necessary to explore the broader impact of column-supported elevated structures on the thermal environment beneath buildings. To this end, two hypothetical cases were developed based on the same base geometry (

Table 14. The different hypothesis cases.

Base Case	Case-1	Case-2

).

Case 1 represents a theoretical scenario in which the first floor is completely removed, resulting in a fully open ground surface without any overhead structure. Although this configuration does not occur in real-world architecture, it serves as a comparative reference to isolate the effect of the elevated structure itself. Case 2 simulates a typical elevated ground floor supported by columns—reflecting a standard design commonly used in high-density urban developments.

The simulated PET results for both cases are presented in

Table 15. The simulated result of different hypothesis cases.

Case	PET
Case-1
Case-2

. The comparison provides insights into how the presence of elevated structures and their supporting columns influences microclimatic conditions and thermal comfort at the pedestrian level.

The simulation results for Case 1 and Case 2 reveal highly similar thermal conditions in the selected area. Specifically, Ta and PET values in Case 1—where the elevated floor is entirely absent—are nearly identical to those in Case 2, which features a typical column-supported elevated design. This finding suggests that the presence of supporting columns has minimal influence on the local thermal environment and outdoor thermal comfort beneath elevated buildings. The comparative results are illustrated in Figure 15.

4.3. The Difference Between the Overall Elevated Design and the Partial Elevated Design

In addition, the elevation ratio—defined as the proportion of the ground floor area that is elevated—plays an important role in influencing the thermal environment and outdoor comfort. In this section, four scenarios were modeled with elevation ratios of 20%, 50%, 80%, and 100%, respectively, to examine their impact on microclimatic conditions.

Table 16. The simulated result of different hypothesis cases.

Case	PET
10%-case
40%-case
70%-case
100%-case

and Figure 16 present the distributions of Ta and PET across the four cases. The comparison helps to reveal how varying degrees of elevation affect thermal performance beneath the building structure.

The simulation results clearly indicate that increasing the elevation ratio leads to an amplification of thermally comfortable areas, primarily due to the enhanced shading effect provided by the elevated structure. As shown in Figure 16, the minimum, average, and maximum values of Ta and PET exhibit similar trends across the four scenarios.

Among the tested cases, the 40% and 70% elevation ratios result in slightly higher Ta and PET values compared to the 10% and 100% cases. This can be attributed to reflected radiation from surrounding building surfaces, which partially offsets the shading benefit. Nevertheless, the 100% elevation case provides the most consistently comfortable thermal environment overall, as the continuous shading significantly reduces solar exposure at the pedestrian level.

To further clarify the relationship between the elevation ratio and the surrounding thermal environment, a numerical correlation was established between the elevation ratio and the average PET, as illustrated in Figure 17. The analysis reveals that the relationship follows a second-degree polynomial trend, rather than a simple linear correlation. The fitted equation for PET is expressed as Equation (2):

PET = −22.5x² + 24.85x + 22.295

(2)

R² = 0.9963

This result suggests that thermal comfort initially deteriorates as the elevation ratio increases. However, when the ratio exceeds approximately 55%, PET values begin to decrease, indicating an improvement in thermal comfort due to increased and more continuous shading effects.

4.4. The Estimation of the Optimal Height of the Elevated Building

Given that China is located in the Northern Hemisphere, local architectural design standards emphasize maximizing natural daylighting within buildings. As a result, east–west building orientations are generally preferred. In addition, to accommodate pedestrian circulation and recreational activities, the minimum required height for open spaces beneath elevated buildings is set at no less than 2 m, and the ideal length-to-width ratio is recommended to be 2:1 [62].

Based on these regulatory considerations, combined with the minimum spatial resolution constraint of 0.5 m in ENVI-met modeling, a total of eight simulation scenarios were developed. These scenarios feature elevated space heights of 2.0 m, 2.5 m, 3.0 m, 3.5 m, 4.0 m, 4.5 m, 5.0 m, 5.5 m, and 6.0 m, respectively. The plan dimensions of the teaching building were limited to a maximum length of 50 m and a width of 10 m to align with standard design practice. For each scenario, meteorological data from the hottest day (9 July 2020) and the coldest day (6 January 2021) were used as initial boundary conditions for ENVI-met simulations.

To effectively compare the thermal environment across all height configurations, box plots were used to visualize the distribution of PET values throughout the elevated space. The summer simulation results reveal a negative correlation between the elevation height and thermal comfort: as the elevation height increases, PET values rise, indicating a decline in thermal performance. For instance, the 6.0 m case recorded a peak PET of 40.5 °C, while the 2.0 m case showed a maximum PET of 37.5 °C. According to the thermal comfort classification in Table 12, this range corresponds to a shift in perceived thermal conditions from the ‘Warm’ to the ‘Hot’ category (Figure 18).

In contrast, winter simulation results exhibit a similar pattern of PET change with height but differ in comfort implications. As shown in Figure 19, the 2.0 m case provides the least comfortable thermal environment in winter, while the 6.0 m case yields the highest thermal comfort. Based on the comfort classification presented in Table 12, all cases generally fall within the ‘Cold’ category, but the taller elevated spaces improve solar penetration and air circulation, enhancing thermal perception.

Earlier studies have shown that elevated building designs can enhance thermal comfort during summer by increasing airflow and providing effective shading. As previously stated, the present study focuses on further investigating the thermal performance of the entire space beneath elevated buildings, offering data-driven insights to support climate-responsive architectural design.

Based on the PET classification in Table 12, thermal comfort levels are grouped into three categories:

• Unsuitable: Very Cold, Cold, Very Hot, Hot
• Fairly Suitable: Cool, Warm
• Suitable: Slightly Cool, Neutral, Slightly Warm

Given the inherent heat in summer, achieving the Neutral comfort stage is unrealistic. Therefore, this study designates the Slightly Warm and Warm categories as acceptable levels of outdoor thermal comfort for the summer season.

To identify the optimal elevated height for summer conditions, we applied regression analysis to the simulation data. As shown in Figure 20, the relationship between elevated height and average PET in summer follows a linear trend, expressed as:

PET = 2.0033x + 29.387

(3)

R² = 0.9884

With a coefficient of determination R² = 0.9884, indicating a highly significant correlation. Based on this equation and the PET thresholds defined in Table 12, the ideal elevation height for achieving acceptable summer thermal comfort falls between −0.73 m and 4.3 m. However, considering practical construction constraints and regulatory guidelines [62], the feasible and recommended elevation height range is 2.0 m to 4.3 m.

A similar method was applied to evaluate thermal comfort during winter. For this season, Slightly Cool and Cool are considered acceptable comfort levels. The corresponding regression equation, depicted in Figure 21, is:

PET = 4.998x − 10.045

(4)

R² = 0.87

The corresponding regression depicted in Figure 21 has a coefficient of determination R² = 0.87, also indicating a meaningful fit. According to this model, the optimal elevated height for winter lies between 3.6 m and 6.4 m.

By comparing the results from both summer and winter simulations, it becomes clear that achieving thermal balance in elevated open spaces requires careful calibration. In the cold climate zone of Northern China—where mitigating heat stress in summer and retaining warmth in winter are both essential—the recommended optimal year-round elevation height falls within the range of 3.6 m to 4.3 m.

It is valuable to contextualize the PET results obtained in this study by comparing them with similar investigations from other cold-climate regions in China. Recent research conducted in Harbin and Beijing has reported typical PET comfort thresholds ranging between 18–23 °C for summer and around 4–8 °C for winter [63]. These regional benchmarks align closely with the optimal ranges identified by our findings.

Furthermore, considering that Xi’an is geographically situated near the boundary between the “Cold” and “Hot summer–Cold winter” climate zones according to the Chinese national climate classification (GB 50176-2016) [39], this location inherently introduces a complexity that may affect the generalizability of the results. To address this complexity explicitly, we have further compared our outcomes with recent studies conducted in cities explicitly characterized by the “Hot summer–Cold winter” climate zone, such as Wuhan and Nanjing. These comparisons reveal that our recommended optimal elevated heights and spatial design thresholds are consistent with the findings from these mixed climate cities. This consistency is particularly evident regarding the architectural challenge of balancing shading effectiveness during hot summers with solar access in cold winters, reinforcing the robustness and practical relevance of our proposed design recommendations.

Moreover, previous studies specifically focusing on elevated or semi-open spaces beneath buildings in similar climatic conditions have demonstrated that these architectural configurations typically improve outdoor thermal comfort by approximately 5–10 °C compared to fully exposed outdoor spaces [64]. Such findings further substantiate and reinforce our conclusion that properly designed elevated building spaces can significantly enhance pedestrian comfort in regions experiencing distinct seasonal extremes.

Integrating these insights with national building design standards and modular dimension guidelines, our study refines the proposed optimal elevation thresholds to 3.6 m, 3.9 m, and 4.2 m. These recommended values constitute a balanced solution that simultaneously meets thermal efficiency requirements and practical feasibility constraints, thus providing valuable and directly applicable guidelines for the climate-adaptive design of elevated buildings in regions characterized by transitional or mixed climatic conditions.

5. Conclusions

This study systematically evaluated the microclimatic performance of elevated ground-floor semi-open spaces beneath buildings in the cold climate zone of Northern China, using field-validated ENVI-met simulations. Based on comprehensive quantitative analyses, the following clear conclusions and practical implications were derived:

(1) Structural dimensions and thermal comfort implications:

Our results indicated that variations in supporting column widths (ranging from 0.5 m to 2.5 m) had minimal impact on surrounding thermal conditions, with PET differences remaining within a negligible range (maximum of 0.6 °C). This finding implies that architects and structural engineers have substantial flexibility in selecting column dimensions, allowing decisions to prioritize structural requirements, aesthetic considerations, or functional needs over minor thermal comfort influences.

(2) Optimal elevated ratio for thermal comfort:

For buildings oriented in an east–west configuration—typical in Northern China for daylight optimization—the relationship between elevation ratio and thermal comfort (PET) was found to be nonlinear, following a clear parabolic trend. Specifically, PET initially increased (indicating reduced comfort) as elevation ratios increased, reaching a peak discomfort of around 55%. Beyond this critical threshold, further increases in elevation ratio significantly improved thermal comfort due to enhanced shading effectiveness. These findings offer valuable guidance for architectural and urban planning practice, recommending higher elevation ratios (≥55%) to effectively enhance outdoor pedestrian comfort in urban environments.

(3) Recommended year-round elevated height:

Our analysis demonstrated that optimal elevation heights varied seasonally, with an ideal range of 2.0 m to 4.3 m for summer thermal comfort and 3.6 m to 6.4 m for winter comfort. Considering both seasonal extremes together, the recommended year-round elevation height range is 3.6 m to 4.3 m. Moreover, consistent with modular construction principles and national building design standards, specific elevation heights of 3.6 m, 3.9 m, and 4.2 m were identified as particularly advantageous, balancing thermal efficiency, construction practicality, and regulatory compliance. These clear and actionable design thresholds can be directly applied by practitioners aiming to optimize thermal comfort in architectural and urban projects.

Overall, this study contributes meaningful evidence-based recommendations for climate-responsive design practice. Although conducted specifically in Xi’an, the methodological framework and validated findings possess considerable potential for broader application across similar cold-climate urban settings. Future research is encouraged to expand these analyses into other urban locations, climate contexts, and additional seasons (spring and autumn), further enhancing the robustness, relevance, and global applicability of our findings.

6. Limitations and Outlook

This study has provided a detailed evaluation of the microclimatic performance of elevated semi-open spaces beneath buildings in Xi’an using traditional statistical methods and ENVI-met simulations. Despite the valuable insights gained, several limitations should be explicitly acknowledged. Firstly, the current analysis was conducted at a single urban location, limiting the generalizability of our conclusions. Secondly, the assessment focused exclusively on two seasonal extremes (summer and winter), excluding transitional seasons (spring and autumn), which could offer additional meaningful insights into thermal comfort dynamics.

Additionally, the methodological approach employed in this study primarily relies on conventional statistical methods, which, although robust and widely accepted, have yet to fully incorporate the advantages of emerging technologies. With rapid developments in artificial intelligence (AI) and advanced computational frameworks, there are significant opportunities to enhance the precision, scalability, and automation capabilities of environmental modeling in urban design research.

In future research, we propose expanding our analytical scope to multiple urban locations and comprehensive seasonal assessments across diverse climatic zones, thus improving the robustness and broad applicability of the findings. Moreover, we plan to integrate AI-driven approaches by establishing cross-platform linkages between Python 3.12.6 and ENVI-met 5.0. This innovative integration will allow advanced computational analyses, automated optimization, and adaptive scenario simulations of thermal performance in semi-open outdoor spaces. Ultimately, these future methodological enhancements will facilitate more intelligent, efficient, and scalable strategies for climate-responsive architectural design.