3.1. Flow Structures
As shown in
Figure 4 (right) (H = 2D; S = 1.125D), the pedestrian-level dimensionless wind speeds, measured by Irwin probes at the wind tunnel, are presented as velocity contours. Colors represent the level of wind speed, as the closer the color is to blue, the lower the wind speed; this is followed by the green, with the closer to red, the higher the wind speed. When the wind flows downstream after passing through the passage of high-rise buildings, the channeling effect increases the wind velocity in the street canyon (marked by symbol ❶). At the intersections of L3/S2 and L3/S3, the higher (0.5) and lower (0.2) dimensionless velocities can be found, respectively. This phenomenon can be verified in the oil film test (marked by ❷ and ❸, respectively).
Figure 5 shows the pedestrian-level Vel (dimensionless wind speed) contours in the study area in the absence of high-rise buildings in front of the study area, and in the presence of high-rise buildings with S/D values of 0.375, 0.75 and 1.875 in front of the study area, respectively.
Figure 5a shows the Vel distribution in the street canyons in the absence of high-rise buildings in front of the study area. It shows that there was a relatively high Vel (approximately 0.7) in the upstream area (marked by ❶) of downwind street canyon L1. In the downstream direction, the Vels at the intersection of S2 and L1 (marked by ❷) and the intersection of S3 and L1 (marked by ❸) were approximately 0.4–0.6. Further down the downstream direction, the Vel at location ❹ was less than 0.2. In the transverse street canyons, the approaching wind moved towards corner ❺ due to the presence of high-rise buildings. At this location, because the buildings were relatively short, the corner strong-wind region was not pronounced. Some of the approaching wind moved to the transverse street canyon ❻ via the outer side of the study area.
Figure 5b shows the wind-field distribution in the presence of high-rise buildings with an S/D value of 0.375. Because the high-rise buildings were arranged in a relatively dense pattern, their effects on the downstream study area were similar to those of a resistance flow [
6]. It shows a strong-wind region located at corner ① on the outer side of the high-rise buildings. Due to the resistance from the high-rise buildings, the Vel in downwind street canyon L1 was even lower than that in
Figure 5a. Additionally, the Vels in the transverse street canyons S1–S4 were also lower than those in
Figure 5a. There was different cause for the formation of low-speed circulation in the transverse street canyons. The horizontal circulations (red dotted circles in
Figure 5b) in the transverse street canyons between L2 and L1 were formed from airflows ② and ③, which whirled into the street canyons. The circulations in the transverse street canyons between L1 and the central axis of symmetry were formed from airflow ④, which whirled into the transverse street canyons. On the other hand, after the approaching flow passed over the top of the high-rise buildings, a downwash airflow ⑤ was formed, which met with the stream-wise flow that moved in the opposite direction on the third-row street canyon (S4–L1). This led to a relatively low Vel at this location.
As the distance between the high-rise buildings increased to 0.75D (
Figure 5c), there was an increase in the airflow passages (S). It became easier for the external wind to smoothly flow into the downwind street canyon L1 after passing between the high-rise buildings. Consequently, there was a wind field reduction at corner ❶. However, the Vel in the downwind street canyon ❷ was higher than the previous two cases. As the S increased, this phenomenon became increasingly pronounced, and could even lead to the formation of instant gusts that affect pedestrians. The downstream wind fields were less significantly affected by the high-rise buildings in front. In these wind fields, the Vels were relatively low. The flow characteristics in the case where S/D = 0.75 were similar to those in the previous scenario, where S/D = 0.375. However, there were two differences: (1) in the case where S/D = 0.75, after the approaching flow passed over the top of the high-rise buildings, a downwash airflow formed in the downstream area and merged with the relatively high-speed airflow in downwind street canyon L1, with these two airflows moving in different directions met and stagnating at a point further downstream outside the study area; and (2) in the case where S/D = 0.75, vertical airflow circulations were formed in the second- and third-row transverse street canyons (S2 and S3) between L1 and the central axis of symmetry, as shown by the blue dotted circles in
Figure 5c.
As S gradually increased to a value greater than the width of the downstream street canyons (e.g., S/D = 1.875 (
Figure 5d), the outdoor wind immediately encountered the downstream buildings after entering the passage between the high-rise buildings, and thus moved outward (towards location ①) along the transverse street canyon. Another airflow ② moved directly towards the downwind street canyon L1. This airflow began to move outward at location ③ after flowing to the second-row transverse street canyon S2. Due to the shelter effect, low-speed circulations were formed in the central regions of the transverse street canyons (between L1 and the central axis of symmetry). Due to the circulation at the corner of the high-rise building in front, there was a horizontal airflow circulation (④) in the central region of the first-row transverse street canyon S1. In contrast, there was primarily a vertical circulation (⑤ and ⑥, respectively) in the central region of each of the second- and third-row transverse street canyons (S2 and S3).
Based on the above analysis, similar flow field morphologies were observed when S/D = 0.375 and 0.75. Due to the obstruction by the high-rise buildings, the airflow speed in the downwind street canyon L1 was relatively low. In comparison, the wind field in the outer street canyon L2 consisted of a rapid circulation formed by the movement of the airflow undercut by the high-rise buildings in front, and was stronger than the flow field in the downwind street canyon L1. As a result, the airflow in street canyon L2 mainly entered the transverse street canyon, and formed a horizontal circulation at this location. When S/D = 1.125, S was close to the width of the downwind street canyon. Thus, the flow field morphology in this scenario was relatively stable. As S increased to 1.5D and 1.875D, the shelter effect of the high-rise buildings on the downwind street canyon L1 weakened. As a result, after entering the passage between the high-rise buildings and generating a channel effect, the wind field entered the downstream street canyon L1 at an increased speed. On the other hand, there was also an increase in the area of the weak-wind regions in street canyon L2. As a result, after whirling into the transverse street canyons, the strong wind field in street canyon L1 moved to street canyon L2.
In summary, when S/D ≤ 0.75, this study shows an edge effect-dominated airflow field. In this scenario, the airflow entered the transverse street canyons mainly through street canyon L2. However, the downwind field in street canyon L1 was stronger than the transverse flow fields. As a result, the flow field that moved from street canyon L2 to the transverse street canyons cannot enter street canyon L1. When S/D ≥ 1.5, it shows a stream-wise dominated airflow field. The airflow characteristics differed from those of the edge effect-dominated flow field. When 0.75 < S/D < 1.5, the flow field structure was between those of the previous two scenarios.
3.2. Effects of H
Figure 6a shows the airflow velocity (Vel) distribution in the study area at various S/D values when H/D = 1 (H/D = 1 means that the upstream high-rise buildings and the buildings in the downstream building cluster were of the same height). Due to the lack of airflow at the corner of the high-rise buildings and the high-rise buildings’ channel effect, the Vels in
Figure 6a were relatively low, overall.
Figure 6b,c show the Vel distribution when H/D = 3 and 7, respectively. As H increased, the strength of the circulation at the corner gradually increased. At a small S, a wind field with a Vel over 0.8 appeared at the corner. As S increased, due to the strengthening of the channel effect, a high-wind speed region appeared in the upstream area of the downwind street canyon.
Additionally, as demonstrated in
Figure 6a (S/D = 0.375), at a relatively low H, the area of the regions where the Vel of the wind field was less than 0.3 in the street canyons was relatively small. H was the largest in
Figure 6c. When the high-rise buildings were densely arranged (S/D = 0.375), the shelter effect affected the Vels in the street blocks, and the Vel was less than 0.3 in a large area. As S gradually increased, Vel gradually increased in the street canyons. A notable strong-wind region appeared in
Figure 6c (S/D = 1.5 and 1.875).
3.3. Variation of the Velocity in the Downwind Street Canyons
Figure 7a shows that when S/D = 0.375 (represented by symbol O), because the three high-rise buildings (H/D = 7) were in close proximity to each other, they exerted a shelter effect on the downstream street canyons. As a result, the Vel generated in downwind street canyon L1 was relatively low. As S/D increased to 0.75, Vel in street canyon L1 increased somewhat, in particular between street canyon measuring points 1 and 6. As S/D increased to 1.125, 1.5, and 1.875, Vel at the entrance of street canyon L1 was near 1. As S increased, Vel in the upstream area of street canyon L1 increased, whereas Vel in the downstream area of street canyon L1 was relatively low, and there was no significant difference between the scenarios.
Here, the effect of L (i.e., the distance between the high-rise buildings and street block S1) is considered. When L/D = 0.75 and 1, the Vel variation in the downwind street canyons exhibited relatively similar trends. However, when L/D = 1.5 (
Figure 7c), Vel increased suddenly at measuring point 5, and then decreased sharply further downstream. This high local Vel occurred because the street canyon formed by the buildings between street canyons S2 and S3 caused further flow field channelization.
Figure 8 shows the Vel variation in street L2 on the outer side of the street canyons. At a relatively small S, the airflow in street L2 was dominated by the corner circulation. As a result, Vel in street L2 was higher than that in other cases. When L/D = 0.75 and 1 (
Figure 8a,b), the variation of Vel in street L2 exhibited similar trends. However, when L/D = 1.5 (
Figure 8c), a local low Vel appeared at measuring points 3 and 4, as a result of the flow field structure (see
Section 3.2 for details).
3.4. Assessing the Urban Design Specification
Figure 9 shows the relationships of S and H with Vel (U
i/U
h, calculated for all measurement points) from gathering all the measured data.
Figure 9a shows that the Vel variation trend differed between H/D = 1–2 and H/D = 3–7. When H/D = 1–2, Vel varied relatively gently between 0.3 and 0.35. In comparison, Vel varied relatively sharply when H/D = 3–7. The greater the H or S was, the greater too was the Vel in the downstream street canyons.
Figure 9b shows that when H/D ≥ 3, the Vel curve transitions at an S/D value of 1.125. When S/D > 1.125, multiple strong-wind regions might be formed. Under this condition, it is necessary to evaluate whether pedestrian-level gusts could be generated in certain regions or at certain street corners. When S/D < 1.125, multiple weak-wind regions might be formed. Under this condition, it is necessary to evaluate ventilation issues resulting from poor outdoor airflow.
The urban design specification for the area on the north side of Banqiao Jiangcui in New Taipei City, Taiwan stipulates that the total width of the facades of the buildings along the river bank should not exceed 70% of the width of the construction site submitted for approval, i.e., S/D ≥ 1.125. In Taiwan, however, the H/D values for high-rise buildings generally range from 3 to 7.
Figure 9 demonstrates that the aforementioned urban design specification is suitable, and can allow relatively high wind speeds in the residential building cluster downstream of the high-rise buildings. However, several streets would still be located in strong-wind regions. Therefore, it is necessary to evaluate whether pedestrian-level strong winds can be generated. Additionally, when H/D = 3–7, on average, there were low Vels at 36–15% of the measuring points. Thus, it is necessary to note that some streets are in weak-wind regions. The wind maps in
Figure 6b,c can be used to identify the strong- and weak-wind regions.