Coupling BES / CFD for natural ventilation

1. Why ?

In Building Energy Simulation (BES) softwares, pressure coefficients are generally given by correlations according to the angle of incidence and the shape of the building [Swami & Chandra 1988] or tabulated values that are unsuitable for most real-life situations: for example, Figure 1 below shows the values used by some BES softwares to determine the pressure coefficients C_p .

Figure 1: Pressure coefficient tables used by default in DesignBuilder/EnergyPlus (left : free field, right : in environment with obstructions of size equivalent to half of the building considered)

The calculated natural ventilation flows are therefore very likely to be incorrect if they are estimated from the default values (see also our article on reducing uncertainties in natural ventilation).

2. Workaround

First, the pressure coefficients per façade element must be computed, depending on the wind orientation and magnitude (an illustration of the pressure field on a building is given in Figure 2).

Figure 2: Pressures on a building's façades for a given wind direction.

Each pressure coefficient must then be assigned to the correct façade element. To achieve this, we go through the file defining the BES problem (" *.idf " file in EnergyPlus) and replace the pressure coefficient for each opening and each wall, in order to calculate the flow rates related to infiltrations.

Figure 3: Scheme of the coupling developed between the different tools used to compute the pressure coefficients

This technique has two main advantages:

  • Compared to a direct use of the flowrates computed with CFD, this preserves the thermal buoyancy effects related to the temperature difference between inside and outside.
  • The number of pressure coefficients per wind direction is increased: by default they are given every 45 degrees, while we take a maximum of 30 degrees.

The question is then: how many wind directions should we simulation for a better prediction of wind-induced natural ventilation? Next section gives an insight about this topic.

3. Influence of angular discretization

Figure 4 below shows the natural ventilation rate in a largely glazed train station atrium. The different lines plotted correspond to an increasing number of wind directions simulated: for instance "4 directions" means one simulation is done every 90° and "24 directions" means one simulation every 15°.

Figure 4: Comparison of the natural ventilation rate for a test case, depending on the number of wind directions simulated (n=4, 8, 12, 24) for the determination of pressure coefficients.

One can see that from 8 directions (id est \Delta\theta=45, the difference between the curves gets lower. The difference between 12 and 24 simulations seems relatively low and would lead to using 12 directions only. [to be continued...]