Page 122 - JCAU-6-4

Journal of Chinese
Architecture and Urbanism Indoor wind environment in buildings for Qinghai-Tibet plateau of China

Where: calculations. The number of grids used in simulations is
• v: Horizontal wind speed at height h from the ground a crucial factor in determining the accuracy of the results.
(m/s) PHOENICS divides grids using its built-in PARSOL
• v : Horizontal wind speed at height h (m/s) method. After repeated adjustments, we determined the
1
1
• h : Reference height (10 m) grid count range that did not affect the calculation results
1
• α: Power exponent determined by surface roughness. and met the calculation standard of 20 cm × 20 cm.
Following a strategy of local refinement, the total grid
3.4.2. Grid partition count was set at 1,000,000.
In this simulation, non-uniform grids were utilized, with 3.4.4. Model accuracy verification
denser grids placed closer to the building. The number
of grids was adjusted during computation based on the Inaccurate simulations may cause flawed conclusions,
relationship between grids and computational results which could lead to inappropriate and potentially
to ensure accuracy. A locally refined grid model was harmful actions for the preservation of heritage buildings
employed in this study (Figure 5), where grid density (Huerto-Cardenas et al., 2020). Therefore, we conducted
decreased gradually in some positions, with a variation a thorough verification of the model’s accuracy in this
rate of 0.8, ensuring a grid size of 20 cm × 20 cm. study. First, we adjusted the grid quantity and evaluated
the model’s accuracy by comparing the airflow under
3.4.3. Determining the number of grids different grid quantities. The results indicate that as the
To determine the optimal number of grids, we conducted grid quantity increased, the airflow variation remained
tests by varying the number of grids and comparing the relatively stable, suggesting a certain level of accuracy in
airflow rates at different grid counts, while keeping other the model.
parameters constant. The results indicate that with fewer Second, we validated the simulation results by
than 500,000 grids, increasing the number of grids leads comparing them with actual measured data. We collected
to a proportional increase in airflow rates. However, when meteorological data from the actual site and used it as
the number of grids exceeds 500,000, further increases do input parameters for the simulation. Subsequently, the
not significantly affect the airflow rates. Therefore, for this simulated results were compared and analyzed against
study, a grid count exceeding 500,000 ensures accurate the measured data (Table 4). Across this comparison, we
assessed the model’s accuracy and reliability in simulating
Table 3. Values of surface roughness coefficient the wind environment.
Categorize Landform Ground The airflow simulation of DSPH was conducted using
roughness PHOENICS software under the specified conditions.
coefficient Wind speeds at various window openings were calculated
A Offshore sea, island, coast, lake 0.12 and compared with the measured indoor wind speeds
B Field, village, jungle, hill, small and 0.16 to verify the model’s accuracy and the correctness of the
medium-sized cities, suburban of large cities input parameters. The comparison results showed that
with sparse houses when the outdoor wind speed was 1 m/s, the simulated
C Densely built-up urban downtowns 0.2 and measured indoor wind speeds were highly consistent,
D Densely built-up urban areas with taller 0.3 with only minor discrepancies at the second decimal place,
buildings indicating that the model is highly accurate.

A B

Figure 5. Grid refinement scheme diagram. (A) Global mesh. (B) Locally amplified mesh. Source: Drawings by Zhong (created with PHOENICS)

Volume 6 Issue 4 (2024) 8 https://doi.org/10.36922/jcau.2396