Effect of High-Rise Residential Building Layout on the Spatial Vertical Wind Environment in Harbin, China

: Reasonable building height distribution in urban residential areas is conducive to smoother vertical airﬂow exchange and promotes sustainable development. This paper studies the inﬂuence of building layouts on vertical ventilation in high-rise residential areas in Harbin, China, and discusses typical building layout objectives. The ideal area of 220 × 220 m was determined using statistical analysis and speciﬁcation requirements, and seven typical layouts were deﬁned based on the distribution of building heights. The computation ﬂuid dynamics (CFD) simulation was veriﬁed using wind tunnel testing to improve the accuracy of the Phoenics simulation software. Wind speed, wind pressure, and the Universal Thermal Climate Index (UTCI) in residential areas distributed at different heights were analyzed and evaluated. The results indicated that the Phoenics simulation parameter settings, veriﬁed via wind tunnel testing, could achieve reasonable simulation results and different height distribution modes had an impact on the changes in wind speed and wind pressure. The equal-height layout that conformed to the row spacing of the buildings and the layout of taller buildings on the east side could provide higher comfort, and did not require enclosed and downwind layouts. This research can inform the green and livable design of residential buildings and provide a new perspective for the construction of high-rise residential areas in cold cities.


Introduction
With China's rapid urbanization [1], the urban canopy is increasing, with the increase in high-rise buildings having a more prominent impact on the urban wind environment [2,3]. The footprint of urban residential areas is the largest both on a spatial and population scale. The impact of wind on daily life in this area mainly relates to comfort [4], safety [5,6], and cleanliness [7]. A suitable wind environment can accelerate the evaporation of moisture, thereby reducing the temperature and humidity on the skin, thus increasing comfort. Wind that exceeds a certain velocity picks up debris and hinders outdoor activities, including unbalancing pedestrians, which can result in falls and injuries. Moreover, an appropriate wind environment can accelerate the diffusion of pathogens and pollutants in the air [8]. Under the premise that COVID-19 will persist, ensuring clean air in the residential canopy area will be beneficial to human health [9]. The focus of this study was predominantly on pedestrian elevation, and the evaluation and improvement relating to the wind environment was carried out on the vertical plane range [10,11].
Presently, field measurements, wind tunnel tests, and numerical simulations are being used in other research to study the wind environment. Large amounts of urban spatial data can be obtained by measuring air velocity in the field, however the boundary conditions and parameters cannot be controlled, the influencing factors are complex, and the data are difficult to reproduce, leading to a weak universality of influence. Wind tunnel testing is one (1) Whereas previous wind environment research has focused on the effect of pedestrian heights, this study has focused on analyzing the wind environment on the vertical plane and to clarify the factors influencing its variation. (2) To increase the reliability of the data by combining wind tunnel testing and numerical simulation, to research and select the method of stratified sampling to construct typical non-uniform layout conditions in Harbin, and to conduct simultaneous wind environment simulations to verify the CFD model. The accurate Phoenics CFD program, after meeting the required standard, performed the simulation and extracted data from high-rise residential areas. (3) With reference to previous studies on urban residential areas, air velocity, wind pressure, wind pressure difference, and Universal Thermal Climate Index (UTCI) were selected as parameters to describe the wind environment. The variation law of the wind environment on the vertical plane and the effect on comfort levels were evaluated. Because it identifies a suitable layout for construction developments, it can provide reference for the future spatial planning and construction of residential areas in cold regions.
This paper is presented as follows: Section 2 provides details of the research subject and the working methods of numerical simulation and wind tunnel verification. Section 3 compares the wind tunnel testing and numerical simulation results and the data are analyzed. Section 4 draws conclusions and suggestions based on the analysis of the data.

High-Rise Residential Area Data
The data for this study was mainly derived from the residential building data (2019) in Harbin's major built-up areas, which recorded 32,495 residential buildings, obtained from the Baidu Map database and verified by field observation. Major built-up areas in Harbin were screened based on the criteria for high-rise buildings defined in accordance with the National Unified Standard for Civil Building Design (GB50352-2019) and the Code for Fire Protection of Building Design (GB50016-2014) [25,26]. In addition, in accordance with the relevant provisions of the Residential Design Code (GB50096-2011) and the Urban Residential Planning and Design Standard (GB50180-2018) [27,28], the high-rise residential area in this study refers to the residential area with a building height exceeding 10 floors (greater than 30 m). As indicated in Figure 1, there are 5918 buildings in total.
Buildings 2022, 12, x FOR PEER REVIEW 3 of 31 (3) With reference to previous studies on urban residential areas, air velocity, wind pressure, wind pressure difference, and Universal Thermal Climate Index (UTCI) were selected as parameters to describe the wind environment. The variation law of the wind environment on the vertical plane and the effect on comfort levels were evaluated. Because it identifies a suitable layout for construction developments, it can provide reference for the future spatial planning and construction of residential areas in cold regions. This paper is presented as follows: Section 2 provides details of the research subject and the working methods of numerical simulation and wind tunnel verification. Section 3 compares the wind tunnel testing and numerical simulation results and the data are analyzed. Section 4 draws conclusions and suggestions based on the analysis of the data.

High-Rise Residential Area Data
The data for this study was mainly derived from the residential building data (2019) in Harbin's major built-up areas, which recorded 32,495 residential buildings, obtained from the Baidu Map database and verified by field observation. Major built-up areas in Harbin were screened based on the criteria for high-rise buildings defined in accordance with the National Unified Standard for Civil Building Design (GB50352-2019) and the Code for Fire Protection of Building Design (GB50016-2014) [25,26]. In addition, in accordance with the relevant provisions of the Residential Design Code (GB50096-2011) and the Urban Residential Planning and Design Standard (GB50180-2018) [27,28], the high-rise residential area in this study refers to the residential area with a building height exceeding 10 floors (greater than 30 m). As indicated in Figure 1, there are 5918 buildings in total.  Moreover, the field survey found that high-rise residential buildings in Harbin were distributed in 126 residential areas, each with at least two or more such buildings. The distribution was generally scattered, but in new areas of the city high-rise residential buildings were more widely distributed. The average floor area was 8.2 ha, as shown in Figure 2a, and most high-rise residential complexes covered 0-5 ha. The number of residential areas with between five and 10 buildings was 49. Residential high-rise buildings under construction had to be redesigned because the change in the city's laws led to a change of monomer architecture and layout form. In accordance with the latest state regulations (GB50180-2018), which are based on comprehensive research results, and which contain the relevant specifications, the maximum height limit for high-rise buildings was been restricted to 80 m. As demonstrated in Figure 2b, the main heights of the residences were between 18 and 24 storys. Moreover, the field survey found that high-rise residential buildings in Harbin were distributed in 126 residential areas, each with at least two or more such buildings. The distribution was generally scattered, but in new areas of the city high-rise residential buildings were more widely distributed. The average floor area was 8.2 ha, as shown in Figure 2a, and most high-rise residential complexes covered 0-5 ha. The number of residential areas with between five and 10 buildings was 49. Residential high-rise buildings under construction had to be redesigned because the change in the city's laws led to a change of monomer architecture and layout form. In accordance with the latest state regulations (GB50180-2018), which are based on comprehensive research results, and which contain the relevant specifications, the maximum height limit for high-rise buildings was been restricted to 80 m. As demonstrated in Figure 2b, the main heights of the residences were between 18 and 24 storys.

Mock Object
To reflect the influence of building height and layout on the vertical wind environment on building surfaces, the model needed to be arranged in groups, so that the vertical influence of the wind on each building in the model was as comprehensive as possible. First, in accordance with the latest national standard (GB50180-2018), the spatial scale of the study was defined at the neighborhood block level, with an area of 2-4 ha. In addition, based on the area with the largest distribution of residential land in Harbin, indicated in Figure 2a, it was decided that the length and width of the simulation scale should be 220 m and 2 ha, respectively. Second, based on the number of buildings included in a single group of residential land in Harbin as indicated in Figure 2a, the number of simulated buildings chosen was 9. Finally, based on the height distribution of residential buildings in Harbin, as indicated in Figure 2b, the heights of two typical buildings were set at 18 and 24 storys. The length, width, and height of the individual buildings were set in accordance with the GB50096-2011 specifications. A single building was 60 m in length, 15 m in width, and 3 m in height. The building spacing was selected in accordance with the specifications of Harbin's Urban and Rural Planning Regulations for sunlight and fire spacing [29]; that is, where the height of the building on the south side was H, the relative spacing of the longitudinal wall was 1.2 H, and that of the gable was 20 m. The specific spacing of buildings is indicated in Figure 3.
Because Harbin is situated in a cold climate, the residential buildings have higher requirements for sunshine exposure, and the low height of winter sun results in the need to space building further apart to accommodate the requirements for sunshine space. To improve land use efficiency, developers tend to construct low-height buildings on the south side, and high-height buildings on the north side, which was also reflected in the survey results. In this study, regional layout characteristics were considered to reduce the

Mock Object
To reflect the influence of building height and layout on the vertical wind environment on building surfaces, the model needed to be arranged in groups, so that the vertical influence of the wind on each building in the model was as comprehensive as possible. First, in accordance with the latest national standard (GB50180-2018), the spatial scale of the study was defined at the neighborhood block level, with an area of 2-4 ha. In addition, based on the area with the largest distribution of residential land in Harbin, indicated in Figure 2a, it was decided that the length and width of the simulation scale should be 220 m and 2 ha, respectively. Second, based on the number of buildings included in a single group of residential land in Harbin as indicated in Figure 2a, the number of simulated buildings chosen was 9. Finally, based on the height distribution of residential buildings in Harbin, as indicated in Figure 2b, the heights of two typical buildings were set at 18 and 24 storys. The length, width, and height of the individual buildings were set in accordance with the GB50096-2011 specifications. A single building was 60 m in length, 15 m in width, and 3 m in height. The building spacing was selected in accordance with the specifications of Harbin's Urban and Rural Planning Regulations for sunlight and fire spacing [29]; that is, where the height of the building on the south side was H, the relative spacing of the longitudinal wall was 1.2 H, and that of the gable was 20 m. The specific spacing of buildings is indicated in Figure 3.
Because Harbin is situated in a cold climate, the residential buildings have higher requirements for sunshine exposure, and the low height of winter sun results in the need to space building further apart to accommodate the requirements for sunshine space. To improve land use efficiency, developers tend to construct low-height buildings on the south side, and high-height buildings on the north side, which was also reflected in the survey results. In this study, regional layout characteristics were considered to reduce the need for special working conditions. Therefore, the south-side height characteristic was excluded, and seven building height distribution modes were selected: as the control group, the seven building height distribution modes relating to uniform building height (UBH) and non-uniform building height distribution modes of the north side height (1-N), east side height (1-E), west side height (1-W), east and north side height (2-EN), west and north side height (2-WN), and east and west side height (3-EWN) are shown in Figure 3. Finally, to reduce the influence of irrelevant factors, all buildings were laid out in rows in a north-south direction.
Buildings 2022, 12, x FOR PEER REVIEW 5 of 31 need for special working conditions. Therefore, the south-side height characteristic was excluded, and seven building height distribution modes were selected: as the control group, the seven building height distribution modes relating to uniform building height (UBH) and non-uniform building height distribution modes of the north side height (1-N), east side height (1-E), west side height (1-W), east and north side height (2-EN), west and north side height (2-WN), and east and west side height (3-EWN) are shown in Figure  3. Finally, to reduce the influence of irrelevant factors, all buildings were laid out in rows in a north-south direction.

Boundary Conditions
Common CFD simulation software includes Airpak, Pheonics, Wind Perfect, Star-CD, and Fluent. Pheonics has simple and intuitive parameter settings, can automatically

Boundary Conditions
Common CFD simulation software includes Airpak, Pheonics, Wind Perfect, Star-CD, and Fluent. Pheonics has simple and intuitive parameter settings, can automatically generate and adjust grids, and makes use of rich calculation models, thereby performing Buildings 2022, 12, 705 6 of 31 efficient simulations. Its FLAIR module is specially designed for the simulation of the wind environment to inform the grouping and distribution of outdoor buildings. Numerous studies in recent years have verified its accuracy and reliability in wind environment simulation [18,30]. Therefore, this study has used the Pheonics 2019 software based on the Reynolds average equation and Sketchup2018 to generate a 3D model of the simulated area, as shown in Figure 3.

Air Velocity Setting
In the simulation, the vertical air flow level at the entrance usually appeared lower, however regional terrain can also influence this, thereby producing a corresponding change. The influence of friction because of terrain results in a decrease in air velocity and can also produce the vertical air exchange function. Air velocity decreases with an increase in height, and is not affected by surface roughness in the atmospheric boundary layer; as such, free air flow is restored. As shown in Figure 4, this study utilized the Harbin air velocity, wind direction, and atmospheric pressure data obtained from The National Climatic Data Center (NCDC) of the United States [31]. Data collected from 1956 to 2020 at 3 h intervals were used. The average wind speed in Harbin is 3.23 m/s, the average atmospheric pressure is 100,553.8 Pa, and the prevailing wind direction is SSW. The above data were manually entered into the Phoenics wind environment attribute interface and are shown in Figure A1 under Appendix A.
uildings 2022, 12, x FOR PEER REVIEW 6 of 31 generate and adjust grids, and makes use of rich calculation models, thereby performing efficient simulations. Its FLAIR module is specially designed for the simulation of the wind environment to inform the grouping and distribution of outdoor buildings. Numerous studies in recent years have verified its accuracy and reliability in wind environment simulation [18,30]. Therefore, this study has used the Pheonics 2019 software based on the Reynolds average equation and Sketchup2018 to generate a 3D model of the simulated area, as shown in Figure 3.

Air Velocity Setting
In the simulation, the vertical air flow level at the entrance usually appeared lower, however regional terrain can also influence this, thereby producing a corresponding change. The influence of friction because of terrain results in a decrease in air velocity and can also produce the vertical air exchange function. Air velocity decreases with an increase in height, and is not affected by surface roughness in the atmospheric boundary layer; as such, free air flow is restored. As shown in Figure 4, this study utilized the Harbin air velocity, wind direction, and atmospheric pressure data obtained from The National Climatic Data Center (NCDC) of the United States [31]. Data collected from 1956 to 2020 at 3 h intervals were used. The average wind speed in Harbin is 3.23 m/s, the average atmospheric pressure is 100,553.8 Pa, and the prevailing wind direction is SSW. The above data were manually entered into the Phoenics wind environment attribute interface and are shown in Figure A1 under Appendix A.

Mean Air Velocity Distribution
Because of the influence of surface friction on air velocity, the variation law of average air velocity in the atmospheric boundary layer with the change of height can be expressed in two forms: exponential law and logarithmic law. The exponential law is derived from the actual measurement results, and the logarithmic law is derived from the boundary layer theory. In this study, the exponential law was used to calculate the change in air velocity near the surface, and the formula is as follows: In this formula, Z0 is the standard height above ground, which is generally 10 m. The incoming flow velocity at this height is U0, Z is the variable height from the ground, u is the air velocity at this height, and α is the roughness coefficient of the ground [32]. The subject of this study falls within the standard definition of urban building density as defined in the Thermal Environment Design Standard for Urban Residential Areas (JGJ286-

Mean Air Velocity Distribution
Because of the influence of surface friction on air velocity, the variation law of average air velocity in the atmospheric boundary layer with the change of height can be expressed in two forms: exponential law and logarithmic law. The exponential law is derived from the actual measurement results, and the logarithmic law is derived from the boundary layer theory. In this study, the exponential law was used to calculate the change in air velocity near the surface, and the formula is as follows: In this formula, Z 0 is the standard height above ground, which is generally 10 m. The incoming flow velocity at this height is U 0 , Z is the variable height from the ground, u is the air velocity at this height, and α is the roughness coefficient of the ground [32]. The subject of this study falls within the standard definition of urban building density as defined in the Thermal Environment Design Standard for Urban Residential Areas (JGJ286-2013); as such, the roughness coefficient(α) is 0.22 [33]. Specific parameters were manually entered into the Phoenics wind environment attribute interface, as shown in Figure A1.

Domain Size
The size of the computational domain could affect the reliability of the simulation experiment results. In accordance with the requirements of wind tunnel tests, the clogging rate of the computational domain should be kept below 3%, which is generally defined as the ratio of the frontal area of working conditions to the vertical cross-sectional area of the computational domain [34,35]. Relevant experimental cases have demonstrated that the lateral and upper boundaries should be greater than or equal to 5 H, where H is the maximum building height and therefore the outflow boundary should be set at the rear of the building to at least 10 H and consistent with the wind direction [36,37]. In addition, the height determination of the computational domain is also affected by different terrain type distribution. Based on the defined parameters and combined with wind direction, the respective longitudinal (Y), transverse (X), and vertical (Z) extents for this study were calculated to be 1297 m, 1300 m, and 432 m, respectively, as shown in Figure A2 under Appendix A. The dimensions on the XY plane are indicated in Figure A3 under Appendix A, and are 10 H in the eastern and northern directions, and 5 H in the western and southern directions. As shown in Figure A4 under Appendix A, the height on the Z plane is 5 H.

Grid Size and Independence
Grids of different scales are required for computing domains and core areas. To clearly express the vertical wind speed distribution, the Z-axis grid was set to a building height of 3 m, as shown in Figure A5c under Appendix A. As shown in Figure A5a,b, two grids were used for the X-axis and Y-axis, 15 m in the computational domain and 5 m in the core area, respectively, to meet the requirements of the horizontal and vertical proportions of the grid cells. The mesh setting was refined towards the core area and subdivided using SPARSOL to ensure statistical accuracy ( Figures A2 and A5). Grid independence verification indicated that the difference in simulation results was less than 5% when the grid scale was doubled, hence the research grid met the requirements without requiring further encryption.

Solution Methods and Convergence Condition
The standard K-ε model was used to calculate turbulence. This model has a reasonable prediction capability in a large turbulence range and is widely used in practical engineering [38]. To ensure accuracy and convergence of the results, it is generally required that the simulation residual of all working conditions reaches 10 −4 or less [39,40]. The Phoenics settings required for this study are shown in Table 1. As shown in Figure A6 under Appenidx A, the number of simulation iteration steps for the seven layout methods was 20,000 and the residual error was maintained below 10 −5 .

Validation of the CFD Simulation
One of the purposes of this study was to verify the influence of a non-uniform layout of high-rise buildings on the vertical wind environment. Therefore, the layered sampling method was adopted to sample the height of buildings in Harbin and conduct modeling on the non-uniform layout, to create the wind tunnel test conditions shown in Figure 5. The same operating conditions used in the preceding CFD simulation were used and the results were compared to the relevant town planning specifications (JGJ/T338-2014 and JTG/T3360-01-2018) [41,42]. The results can be considered to concur if the air velocity error between the numerical simulation and wind tunnel test at the same position is less than 10% within the same boundary parameters. Specific verification results are provided in Section 3.1.

Validation of the CFD Simulation
One of the purposes of this study was to verify the influence of a non-uniform layou of high-rise buildings on the vertical wind environment. Therefore, the layered samplin method was adopted to sample the height of buildings in Harbin and conduct modelin on the non-uniform layout, to create the wind tunnel test conditions shown in Figure 5 The same operating conditions used in the preceding CFD simulation were used and th results were compared to the relevant town planning specifications (JGJ/T338-2014 an JTG/T3360-01-2018) [41,42]. The results can be considered to concur if the air velocity erro between the numerical simulation and wind tunnel test at the same position is less tha 10% within the same boundary parameters. Specific verification results are provided i Section 3.1.

Simulation Results Evaluation Criteria
The wind environment needed to be simulated for the abstract models of seve height distribution modes using the parameters described above and then the Para View5.10 software was used to export and process the air velocity and wind pressure data When evaluating air velocity, the Beaufort Wind Scale correlated the influence of ai velocity on people, as shown in Table 2 [43]. As a city typical of colder regions, Harbin i required to meet the green building evaluation standard (GB/T50378-2019) for pedestria zone air velocity of less than 5 m/s in winter [44]. Low wind speed can lead to poor ai circulation and long-term accumulation of pollutants, thereby affecting people's comfor and health. Research that observed the effects of air velocity on the human body, as show in Table 2, indicates that regions with air velocity lower than 1 m/s or 0.5 m/s can be de fined as calm wind zones [45,46]. Table 2. Evaluation standard for the correlation between air velocity and comfort.

Wind Level Air Velocity Range (m/s) Effects on the Human Body
Tolerable limit for onshore wind 6 6.7 < V ≤ 8.6 Difficulty walking and holding an umbrella When evaluating wind pressure, barring the first row of windward buildings, th wind pressure difference between the windward side and the leeward side of each build ing should not be greater than 5 Pa [47]. It is also necessary to ensure that more than 75%

Simulation Results Evaluation Criteria
The wind environment needed to be simulated for the abstract models of seven height distribution modes using the parameters described above and then the ParaView5.10 software was used to export and process the air velocity and wind pressure data.
When evaluating air velocity, the Beaufort Wind Scale correlated the influence of air velocity on people, as shown in Table 2 [43]. As a city typical of colder regions, Harbin is required to meet the green building evaluation standard (GB/T50378-2019) for pedestrian zone air velocity of less than 5 m/s in winter [44]. Low wind speed can lead to poor air circulation and long-term accumulation of pollutants, thereby affecting people's comfort and health. Research that observed the effects of air velocity on the human body, as shown in Table 2, indicates that regions with air velocity lower than 1 m/s or 0.5 m/s can be defined as calm wind zones [45,46]. Table 2. Evaluation standard for the correlation between air velocity and comfort.

Wind Level Air Velocity Range (m/s) Effects on the Human Body
Tolerable limit for onshore wind 6 6.7 < V ≤ 8.6 Difficulty walking and holding an umbrella When evaluating wind pressure, barring the first row of windward buildings, the wind pressure difference between the windward side and the leeward side of each building should not be greater than 5 Pa [47]. It is also necessary to ensure that more than 75% of concrete buildings have a 1.5 Pa wind pressure difference between the windward side and the leeward side [48]. The above evaluation was based on pedestrian height. In a natural ventilation state, the comfort of each floor of a high-rise building is related to the pedestrian height. Therefore, the evaluation of air velocity and wind pressure is also applicable in the vertical direction. That is, the acceptable air velocity is 1-5 m/s, and the acceptable wind pressure difference ranges from ±1.5-5 Pa.
The effects on thermal comfort caused by ventilation were evaluated in this study using the Universal Thermal Climate Index (UTCI) to express the comfort effect of different vertical wind speed distributions on people who open windows for ventilation [49,50]. The formula is: where R h is the air humidity, T mrt is the mean radiation temperature, V a is the wind speed, and T d is the air temperature [49]. The UTCI calculation was performed using the Grasshopper script. The Harbin T d data originated from the statistical data of the NCDC [31] where the average temperature is 4.67 • C, the suitable human R h humidity is 50% [50], and the T mrt is the average radiation temperature of 24 • C [51]. As shown in Table 3, different UTCI ranges represent cold or hot regions and this research was carried out by modifying the range in line with the characteristics of cold regions [52].

CFD Simulation Verification Results
As shown in Figure 6, the air velocity data extracted from the wind tunnel test is presented as a line graph with the power function fitted for processing followed by performing an error calculation using the data generated from the Phoenics simulation. The point selection error and curve (CFD simulation curve and wind tunnel fitting curve comparison) errors were distributed in the range of −4-9%, which are all lower than the 10% specification requirement. Therefore, the reliability of the CFD simulation data was found to be high, and subsequent simulation analyses could be performed. The specific processing method began with the extraction of air velocity and wind pressure data from the grid nearest to the building surface. Based on the number of grids covered by the building surface, the average air velocity and wind pressure at the corresponding height was taken based on a building unit height of the number of storys. Finally, the distribution of windward and leeward air velocity, windward and leeward wind pressure, and wind pressure differences at each building was formed. Buildings 2022, 12, x FOR PEER REVIEW 10 of 31  Figure 7 and Table 4 show the proportion of air velocity distribution on building surfaces in seven distribution patterns. The simulation parameters were set according to the parameters outlined in Table 1. Figure 8 indicates the air velocity distribution on the building surface and the main vertical sections between buildings #2, #5, and #8 in seven distribution patterns.    Figure 7 and Table 4 show the proportion of air velocity distribution on building surfaces in seven distribution patterns. The simulation parameters were set according to the parameters outlined in Table 1. Figure 8 indicates the air velocity distribution on the building surface and the main vertical sections between buildings #2, #5, and #8 in seven distribution patterns.  Figure 7 and Table 4 show the proportion of air velocity distribution on building surfaces in seven distribution patterns. The simulation parameters were set according to the parameters outlined in Table 1. Figure 8 indicates the air velocity distribution on the building surface and the main vertical sections between buildings #2, #5, and #8 in seven distribution patterns.

Windward Air Velocity
It is evident from the statistical results that the main air velocity distributions of the seven distribution modes were between 2.1 and 3.4 m/s and the distribution ratio of comfortable air velocity in all distribution modes could reach more than 90%. However, there were 2.97-4.19% air velocities that were statistically greater than 5 m/s, which were all located at the upper and side gables of the buildings. There are no windows in these parts of the buildings, therefore these velocities would have little impact on the overall comfort.
According to the statistical results for each building shown in Figure A7, the average air velocity of buildings #1 and #4 on the windward side under the roof had a low variation range, and rapidly recovered to more than 5 m/s on the top floor of the building. The air velocities at buildings #7, #8, and #9, on the main windward side, changed with the increase of height. The air velocity at building #8 on the east side, which was high (1-E), decreased with the increase in height under the influence of the corner flow zone on the east side, with a decrease of 1.43 m/s. The air velocity at buildings #8 and #9, which were high (1-W) on the west side, decreased by 0.8 m/s and 1.63 m/s under the influence of the corner flow zone on the west side.
The buildings on the non-windward side were sheltered by the front and side buildings, and the average air velocity change mainly decreased with the increase in height. Among them, three buildings on the east side had a layout pattern with taller buildings, 1-E, 2-EN, and 3-EWN, and a slight decrease and then rise in air velocity was observed at building #6. The air velocity at buildings #2 and #3, which were downwind, increased when the building on the west side was high, and the increase was found to be significantly correlated to the increase of height.

Leeward Air Velocity
As the statistical results demonstrated, the main leeward air velocity distribution of the seven distribution modes was 0.1-1.0 m/s, which falls within the low air velocity category. Each distribution mode had a windless area of at least less than 0.67% and these spaces were prone to poor air circulation. The proportion of air velocity in the 1-5 m/s zone of comfort was 33.21-35.63%, and the two downwind modes, 1-N and 2-EN, of the building layout experienced a better ventilation effect on the leeward side of the building, accounting for 59.06 and 58.36%, respectively.
According to the statistical results shown in Figure A8, the average leeward air velocity of the nine buildings was strongly consistent with the change in height. Most of the buildings had low initial air velocity, and the air velocity increased slightly with the increase in height. The air velocity at the roof position was close to or exceeded the

Windward Air Velocity
It is evident from the statistical results that the main air velocity distributions of the seven distribution modes were between 2.1 and 3.4 m/s and the distribution ratio of comfortable air velocity in all distribution modes could reach more than 90%. However, there were 2.97-4.19% air velocities that were statistically greater than 5 m/s, which were all located at the upper and side gables of the buildings. There are no windows in these parts of the buildings, therefore these velocities would have little impact on the overall comfort.
According to the statistical results for each building shown in Figure A7, the average air velocity of buildings #1 and #4 on the windward side under the roof had a low variation range, and rapidly recovered to more than 5 m/s on the top floor of the building. The air velocities at buildings #7, #8, and #9, on the main windward side, changed with the increase of height. The air velocity at building #8 on the east side, which was high (1-E), decreased with the increase in height under the influence of the corner flow zone on the east side, with a decrease of 1.43 m/s. The air velocity at buildings #8 and #9, which were high (1-W) on the west side, decreased by 0.8 m/s and 1.63 m/s under the influence of the corner flow zone on the west side.
The buildings on the non-windward side were sheltered by the front and side buildings, and the average air velocity change mainly decreased with the increase in height. Among them, three buildings on the east side had a layout pattern with taller buildings, 1-E, 2-EN, and 3-EWN, and a slight decrease and then rise in air velocity was observed at building #6. The air velocity at buildings #2 and #3, which were downwind, increased when the building on the west side was high, and the increase was found to be significantly correlated to the increase of height.

Leeward Air Velocity
As the statistical results demonstrated, the main leeward air velocity distribution of the seven distribution modes was 0.1-1.0 m/s, which falls within the low air velocity category. Each distribution mode had a windless area of at least less than 0.67% and these spaces were prone to poor air circulation. The proportion of air velocity in the 1-5 m/s zone of comfort was 33.21-35.63%, and the two downwind modes, 1-N and 2-EN, of the building layout experienced a better ventilation effect on the leeward side of the building, accounting for 59.06 and 58.36%, respectively.
According to the statistical results shown in Figure A8, the average leeward air velocity of the nine buildings was strongly consistent with the change in height. Most of the buildings had low initial air velocity, and the air velocity increased slightly with the increase in height. The air velocity at the roof position was close to or exceeded the incoming air velocity of 3.23 m/s. An important trend was observed at buildings #5 and #6 with two layouts, 1-N and 2-EN, of the north height and east north height. In both cases, the initial air velocity was in the appropriate range and decreased with the increase in height with the air velocity being highest at the top. Figure 9 and Table 5 show the distribution ratio of building surface wind pressure under seven distribution patterns. Figure 10 shows the wind pressure on the building surface and main vertical sections, between buildings #2, #5, and #8, under seven distribution modes. The simulation parameters were set according to the parameters outlined in Table 1. incoming air velocity of 3.23 m/s. An important trend was observed at buildings #5 and #6 with two layouts, 1-N and 2-EN, of the north height and east north height. In both cases, the initial air velocity was in the appropriate range and decreased with the increase in height with the air velocity being highest at the top. Figure 9 and Table 5 show the distribution ratio of building surface wind pressure under seven distribution patterns. Figure 10 shows the wind pressure on the building surface and main vertical sections, between buildings #2, #5, and #8, under seven distribution modes. The simulation parameters were set according to the parameters outlined in Table 1.

Windward Wind Pressure
The significance of single-side wind pressure statistics was influenced by interior corridor layout where ventilation of all rooms cannot be guaranteed by the pressure difference on the windward and leeward sides of the building itself. Therefore, pressure on one side of the building could also improve ventilation and increase spatial comfort.
As evidenced by the overall wind pressure distribution presented in Figures 9a and 10, the high wind pressure distribution on the windward side was mostly concentrated on the windward buildings of the seven layout modes and the high pressure distribution was also present at the lateral windward buildings. The distribution of wind pressure in the acceptable range was above 78% with the highest being 85.83% (UBH) and the lowest 78.66% .
According to the average windward wind pressure of buildings shown in Figure A9, the wind pressure of the windward buildings in the first row, buildings #7, #8, and #9, changed considerably. The wind pressure decreased rapidly and became negative as the building height approched the top floor. Notable observations were made in the case of building #8 with east side high (1-E) and west side high (1-W), and building #9 with west side high (1-W) and east and north side high (2-EN). The initial wind pressure was negative and the variation range was small. The wind pressure at the center-row buildings #4, #5, and #6 varied slightly. The initial wind pressure was −1.91-1.0 Pa, and the maximum negative wind pressure was observed on the top floor of the building. The rear buildings #1, #2, and #3 were substantially affected by the layout and the initial wind pressure changes. Building #2, west side high (1-W), had the largest initial wind pressure, which was 5.37 and 4.42 Pa respectively. The wind pressure changes were the most elevated at a height above 42 m.

Leeward Wind Pressure
According to the results presented in Figure 9b and Table 4, the wind pressure on the leeward side was principally negative, and its distribution ranged from 62.46-81.87%, which was within the acceptable range. The acceptable wind pressure distribution of the north side high (1-N) and east and north high (2-EN) was 66.5% and 62.46%, respectively, with few distributions. This indicated that the layout of high-rise buildings downwind was not conducive to creating appropriate leeward wind pressure. The negative wind pressure was greater than 5 Pa for all layout patterns, and the distribution proportion were 18.04-36.85%. The high proportion was observed in the north side high (1-N) and east and north high (2-EN) areas, which were 32.03% and 36.85%, respectively. Figure A10 presents the average wind pressure on the leeward side of the buildings. The wind pressure of the seven layout modes was less affected by the change in height and was mainly stable. The center row buildings #4, #5, and #6 varied substantially with the layout pattern of high-rise buildings on the rear and lateral rear (1-N and 2-EN).

Wind Pressure Difference
According to the results presented in Figure 9c and Table 4, overall, a wind pressure difference of 1.5 Pa before and after 75% could be achieved for all the seven distribution modes, and the distribution ratio ranged from 79.62-84.23%. However, the optimum wind pressure distribution of each distribution mode was not more than 38%, and ranged from 23.77-37.16%. At the same time, there were areas with a wind pressure difference greater than 5 Pa, accounting for 43.90-58.86%. In these areas, it was necessary to strengthen windproofing at the building entrance to prevent heat loss as a result of air egress and to reduce the wind pressure differences to stabilize indoor and outdoor ventilation to meet human comfort needs.
As indicated in Figure A11, the initial average wind pressure difference between the front and rear of the first row of windward buildings #7, #8, and #9 was large, with a range of 8. 31-10.12 Pa. An important observation was made in the case of building #8, where the pressure was high on the east (1-N) and high on the west (1-W), and building #9, which was high on the west (1-W) and high on the east and north (2-EN). The wind pressure difference was negative and small, and the variation range was small as height increased. The center row buildings #4, #5, and #6 had a small wind pressure variation range. Building #4 faces the wind laterally and the average distribution of wind pressure difference for the building was slightly higher. Buildings #5 and #6 had an average wind pressure difference within the acceptable comfort zone. The layout influenced the initial wind pressure difference between rear row buildings #1, #2, and #3, which changed substantially. The roof wind on front row buildings influenced the average wind pressure differences at buildings #1 and #2 with the layout pattern of rear-row taller buildings (1-N and 2-EN) changing substantially with the increase in height. Building #2 and #3, which are located in the west side high (1-W) layout, are blocked by the west side high-rise buildings, showed the largest initial wind pressure difference of 9.80 and 9.30 Pa, respectively, and the variation range of wind pressure difference was the largest from a height of 42 m. Figure 11 indicates the UTCI proportion distribution of each building surface under the seven distribution modes, and the statistical value of the excluded interval was 0. Table 6 presents the overall distribution proportion of the UTCI interval on the building surface of the seven distribution modes.  According to the results presented in Table 4, UTCI comfort on the leeward side was better than that on the windward side, with a difference of 9-19.3%. The largest difference occured in the layout mode with the height on the west side. The amount of space with high degrees of comfort exceeded 74.9% and on the leeward side was more than 92%, indicating that natural ventilation could improve the spatial extent of comfort. However, Figure 11. Proportion distribution of UTCI interval of seven building distribution patterns. (a) Windward UTCI no thermal stress zone, (b) leeward UTCI no thermal stress zone, (c) windward UTCI slight/moderate cold stress zone, and (d) leeward UTCI slight/moderate cold stress zone. According to the results presented in Table 4, UTCI comfort on the leeward side was better than that on the windward side, with a difference of 9-19.3%. The largest difference occured in the layout mode with the height on the west side. The amount of space with high degrees of comfort exceeded 74.9% and on the leeward side was more than 92%, indicating that natural ventilation could improve the spatial extent of comfort. However, 14.9-25% of the space was located in a slightly colder stress area, which needed to be mitigated by installing mechanical ventilation or window shielding.

UTCI Thermal Comfort Assessment
Based on the UTCI proportional distribution of buildings, presented in Figure 11a,b, on the windward side, the comfort zone, +3.5 • C to +21 • C of each building had a distribution proportion higher than 60%, and buildings #1, #2, and #4 were most affected by height, with a variation range of 18.77%, 32.73%, and 27.35%. Buildings #3, #5, and #6 had the highest proportion of comfort zones, and all layouts could improve comfort. Windward buildings #7, #8, and #9 had a higher proportion of comfort distribution, 69.5-82.9%, with limited layout effects. The comfort zones on the leeward side were all above 87.4%, and only building #2 was substantially affected by the layout change, which changed by 11.38%.
As shown in Figure 11c,d, the distribution proportion of the slight/moderate cold stress factor on the windward side was mostly below 30%. Buildings #1, #2, and #4 were substantially affected by height change and building #1 had the highest distribution variation range of 36.6-55.4%. The distribution proportion of the slight/moderate cold stress factor on the leeward side was less than 13% and only building #2 was significantly affected by the layout.

Layout Rationality Judgment
According to the ranking logic shown in Table 7, by studying the distribution of wind speed, wind pressure, and UTCI parameters of 9 buildings in 7 layout modes, the distribution percentages of different ranking intervals were calculated respectively, and then the ranking and summation were carried out to form the wind environment and UTCI parameters of the 9 buildings, as shown in Figure A12. On this basis, the overall comfort level of seven distribution modes has been ranked. Table 7. Sort range and sort description of wind environment and UTCI parameters.

Sort Range Sort Description
Air velocity (m/s) 0 < V ≤ 0.1 Lower interval percentage correlates to higher sorting.

Universal Thermal Climate
Index ( • C) +39 to +45 Lower interval percentage correlates to higher sorting. +33 to +39 +21 to +33 +3.5 to +21 Higher interval percentage correlates to higher sorting. +3.5 to −4 Lower interval percentage correlates to higher sorting. −4 to −11 −11 to −18 Figure 12 summarizes the order of single parameters of air velocity and wind pressure in seven distribution modes. Considering the windward air velocity, the layout of uniform building height (UBH), east side high (1-E) and west and north side high (2-WN) is more suitable. Considering the leeward air velocity, the layout of north side high (1-N), east and north side high (2-EN) and east side high (1-E) had an improved ventilation effect, which would be conducive to the diffusion of pollutants. pressure in seven distribution modes. Considering the windward air ve of uniform building height (UBH), east side high (1-E) and west and n WN) is more suitable. Considering the leeward air velocity, the layout o (1-N), east and north side high (2-EN) and east side high (1-E) had an imp effect, which would be conducive to the diffusion of pollutants. Evaluation ranking of the UTCI-based comfort indicated that t comfort spaces on the windward side in the uniform building height (UB (E-1), and north side high (N-1). Considering the leeward results, in the l side high (W-1), west and north side high (2-WN), east-west side high (3 side high (1-N), the comfortable space is higher.
The ranking of air velocity and wind pressure for each build summated to obtain the rationality order of the wind environment, n height (UBH) > east side high (1-E) > uniform west and north side high side high (1-N) > west side high (1-W) > east-west and north side high (3 north side high (2-EN).

Discussion
The effect of wind velocity in a high-rise residential area in Harbin by observing baseline situation data of the typical building height to ide patterns, on which abstract modeling was performed. The accuracy of th settings and output results were verified via wind tunnel tests and the w pressure, UTCI, and other data under different height distribution mode With respect to the windward wind pressure statistics, the uniform building height (UBH), the east side high (1-E), and the west side high (1-W) demonstrated the ideal wind pressure effect and were beneficial to indoor natural ventilation on the one side. Based on leeward wind pressure results, the ideal unilateral wind pressure could be obtained at the west and north side high (2-WN), uniform building height (UBH), and the east side height (1-E). With respect to the wind pressure difference results, the west and north side high (2-WN), east-west height (1-W), north side high (1-N), and the west side height (1-W) were more conducive to natural ventilation of the building.
Evaluation ranking of the UTCI-based comfort indicated that there were more comfort spaces on the windward side in the uniform building height (UBH), east side high (E-1), and north side high (N-1). Considering the leeward results, in the layout of the west side high (W-1), west and north side high (2-WN), east-west side high (3-EWN) and north side high (1-N), the comfortable space is higher.
The ranking of air velocity and wind pressure for each building layout was summated to obtain the rationality order of the wind environment, namely: building height (UBH) > east side high (1-E) > uniform west and north side high (2-WN) > north side high (1-N) > west side high (1-W) > east-west and north side high (3-EWN) > east and north side high (2-EN).

Discussion
The effect of wind velocity in a high-rise residential area in Harbin was investigated by observing baseline situation data of the typical building height to identify distribution patterns, on which abstract modeling was performed. The accuracy of the CFD simulation settings and output results were verified via wind tunnel tests and the wind speed, wind pressure, UTCI, and other data under different height distribution modes were calculated to explore the relationship between these data and building height spatial layouts. This will assist in providing a spatial layout reference for the future construction of high-rise residential areas in Harbin, which will improve the comfort of living space.
This study analyzed a 3D layout model of high-rise residential buildings in Harbin relative to the wind environment and thermal comfort changes to determine the most suitable building spatial layout. However, this study did not perform correlation analysis between multi-story residential space, i.e., height less than 30 m [19], and other land use space [38]. In the future, additional 3D models could be added to analyze the wind environment and comfort state in other land use spaces, so that the research would have wider applicability.
With the assistance of wind tunnel tests, the vertical CFD simulation, based on the Phoenics model, of high-rise buildings can improve the accuracy of experimental results relating to grid scale, turbulence models, and iterative processes and can also analyze the vertical effect on high-rise buildings. The wind environment for each spatial layout could also be clarified. However, the selected meteorological conditions for research were too simple. In the future, more detailed meteorological data, such as winter and summer variation [54,55], or monthly variation [56], is needed to obtain more detailed information on the vertical wind environment.
In this study, UTCI was used to study thermal comfort levels resulting from the vertical wind environment in high-rise residential areas in Harbin. Compared with previous studies [50,52], UTCI was incorporated and the relationship between thermal comfort of space and the vertical wind environment was evaluated. The current limitation is that UTCI adaptation to the change in regional climate still needs to be explored further to identify the potential requirements in other cities in cold regions.

Conclusions
With respect to high-rise residential buildings, the wind environment at different heights is equally important for people's comfort levels. Therefore, study of the vertical wind environment distribution at high-rise residential buildings based on a suitable layout will be beneficial to the improvement of the comfort levels in residential spaces. Assessing the high-rise residential areas in Harbin as the study subject, the wind environment conditions and the distribution of wind velocity, wind pressure, and UTCI on the building surface have been studied by considering different height modes to improve spatial comfort and to inform suitable building layout. The study reached the following conclusions: (1) Seven typical height distribution patterns relating to the current high-rise residential communities in Harbin were statistically determined, and included: UBH, 1-N, 1-E, 1-W, 2-EN, 2-WN, and 3-EWN. The results can be used as a reference model for similar simulation and urban planning schemes in the future. (2) The CFD simulation of the wind environment can be verified by wind tunnel testing to improve the rationality of the simulation settings and the accuracy of the simulation results. Wind environmental indicators, including air velocity and wind pressure, are distributed along the vertical plane of buildings and the trend changes with height. Buildings in the upwind direction are strongly consistent, while buildings in the downwind direction have weak regularity in their changing trend. (3) The sorting results from seven typical height distribution patterns demonstrated that the contour height parallel to row spacing is the most conducive to the creation of comfortable wind environment space, however, this option is restricted by the plot ratio or site conditions. The northwest height layout pattern is also conducive to the improvement of the vertical wind environment. The selected layout in the case of tall single-row buildings should be on the east or north side. The layout of buildings with high east, west and north sides, and east-north side height, is not conducive to spatial comfort, which should be avoided in future development planning. (4) In the UTCI comfort evaluation of seven typical height distribution modes, the comfort zone on the windward side reaches 79%, and on the leeward side exceeds 92%. The different distribution locations of high-rise buildings have an impact on the comfort level in different buildings, and the same requirements regarding layout and wind environment should be applied.
Considering future urban development, it is necessary to formulate corresponding norms and standards, clarify the position of high-rise buildings in the landscape, and ensure correct building spacing. To obtain adequate ventilation and comfort, the installation of inward-opening windows, which push away from the bottom, and reinforcement and blocking of roof windows to block air flow, should be considered for buildings located in areas that may result in discomfort.

Conflicts of Interest:
The authors declare no conflict of interest. The funder had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

Appendix A
Considering future urban development, it is necessary to formulate corres norms and standards, clarify the position of high-rise buildings in the landscape sure correct building spacing. To obtain adequate ventilation and comfort, the ins of inward-opening windows, which push away from the bottom, and reinforcem blocking of roof windows to block air flow, should be considered for buildings l areas that may result in discomfort.