Wind tunnel study of natural ventilation of building integrated photovoltaics double skin façade

The paper presents a wind tunnel experimental analysis of a small-scale building model (1:30). The objective of the study is to determine the wind influence on the ventilation of a double skin facade channel (DSF) and the cooling effect over integrated photovoltaic panels. The tests were achieved by conceiving and implementation of an experimental program using a wind tunnel with atmospheric boundary layer. The effect of the wind over the ventilation of the horizontal channels of double skin facades is evaluated for different incident velocities. The results are generalized for the average steady state values of the velocities analysed. The experimental results put in evidence the correlation between the reference wind velocity and the dynamics of the air movement inside the double skin facade. These values are used to determine the convective heat transfer and the cooling effect of the air streams inside the channel upon the integrated photovoltaic panels. The decrease of the photovoltaic panels temperature determines a raise of 11% in efficiency and power generated.


Introduction
The complex effect of the wind over buildings is most often studied using wind tunnel simulations. The analysis is focused on the dynamic loads on the buildings facades [1,2], snow accumulation on roofs [3] or natural ventilation of the indoor environment [4]. When buildings are exposes to the wind, some aerodynamic effects such as the bar, Venturi, stack, corner, slipstream or pyramid effect are generated [5]. The air velocity and direction has a random character, which represents the most important problem for studying its action on the buildings. Typically, for designing the wind effect, the mean values for velocity and direction are used. The probabilistic theories are used to determine the effects of the fluctuations, by knowing the statistical distributions specific of the wind velocities and pressures. The standard meteorological measurements are made using anemometers. The continuous and discontinuous recorded values are used for determining the 10 minutes average values, that is used to obtain the characteristic basic wind velocity [5]. Usually, the basic wind velocity, v b , includes both the direction and season effects and it is measured at 10 m height, with 2% probability of annual exceeding the mean [5]. The mean wind velocity, v m (z), is also based on the v b value, but varies with the height from the ground, z, with the roughness of the terrain and orographic particularities of the site [16]. The basic wind pressure, p b , can be determined using the basic wind velocity, v b : where, ρair density [kg/m 3 ]. Usually, the mean wind speed maintains its dynamic and random character, while the spatial fluctuations of the instantaneous velocity have also an important role. The instant velocity at the height z above the ground has two components: where, v med (z)mean velocity on time interval τ [m/s]; v'(z)fluctuating velocity [m/s]. The mean value is a characteristic of the wind profile, while the fluctuating component is used for measuring the turbulence intensity and quantifies the maximum and minimum velocities. The peak velocity pressure, q p (z), becomes relevant for designing the wind action, because it includes the effects of the random wind speed and pressure fluctuations. The peak pressure is obtained by using an exposure factor, which increases the basic velocity pressure, p b , about 3,4…4 times, proportional to the height above the ground and the turbulence intensity. The wind velocity and direction have a spatial variationin horizontal and vertical planes. The urban areas are characterized by lower wind velocities, but increased turbulence intensities, while the open fields are characterized by higher wind velocities and lower turbulence intensity - Fig. 1 [5]. The wind effect on the ventilation of double skin façades is studied in literature [6][7][8][9][10]18]. Recent studies are also focused on the building integrated photovoltaics system [1,[11][12][13][14]. These studies provide both qualitative and quantitative data concerning the influence of the wind on the natural circulation of the air inside double skin facades and cooling effect of airstreams [14]. The action of the wind causes a dynamic pressure, which is increasing with the height [5], and has a favourable effect for the ventilation of double skin facades. When the wind velocity is 10 m/s or higher, the flow becomes neutrally stable. Most often, the wind action upon the façades of the building determines a series of local pressures and suctions. It is recommended to use simulations in wind tunnels with turbulent boundary layer in order to determine the wind action on the buildings [16]. These tests are distinguished by the similarity requirement for the model and prototype [16]. The main target of the studies is to identify the similarity model and the criteria necessary for simulation and also in detecting the similarity criteria that may be passed by, due to their irrelevance [5]. Usually, the analysis made in wind engineering, showed that the inertial forces prevail, and the flow is a turbulent one, at very high Reynolds numbers (Re). Also, for objects with sharp edges, the similarity is ensured by self-modelling, being insensitive to the Re numbers [5]. Some studies regarding the simulations in wind tunnel showed that a self-modelling condition is accomplished when Re numbers reach values of (0.5...1.2) . 10 5 [5]. The self-modelling of the airflow over the bluff bodies simplifies physical models and avoid the inconveniences in satisfying the similarity requirements [15]. The current analysis is focussed in determining the wind effect over the ventilation of the vertical channels of double skin façades. The results for velocities are processed as average steady state values of 4 series of measurement of 30 seconds, resulting a total time of 2 minutes for each case. The frequency of measurement was set to 1000 values/second. The study was achieved by conceiving and implementing an experimental program in the atmospheric boundary layer wind tunnel.

Experimental setup
A reduced scale model (1:30) of a ventilated double skin façade is experimentally tested in a wind tunnel. A real building is a 3 stores building, 13.5 m height and 12 m width is modelled. The outer layer of the double skin façade system is realized from photovoltaic (PV) panels. The purpose of the study consists in determining the cooling effect of the air circulation when wind is incident on the façade of the building [5]. During the study, it is obtained with acceptable probability the air speed inside double skin façade, near the PV panels, during hot season. Also, a series of correlations between reference velocity and the velocities inside the channel are established.  The model represents a building equipped with ventilated double skin façade system. The external glazing is considered to be replaced with photovoltaic panels, in order to achieve a building integrated photovoltaic system (BIPV). The present configuration of the double skin façade consists in three horizontal compartments on the full height and width of one façade and the air circulation due to wind action is supposed to cool the PV panels. The wind tunnel testing program is based on the assumption that for the full-scale buildings the air circulates horizontally, determined by the temperature gradient and wind action, avoiding the vertical overheating inside the channel. The main components and geometrical details of this configuration are displayed in Table 1 and Fig. 3 -Fig. 4.

Results
The following results are based on the assumption that in case of this type of the building model, with sharp edges, it can be approximated that by assuring the geometric similarity, then the dynamic similarity is also performed. This hypothesis is available if the Reynolds numbers are superior to the turbulent flow limit. In order to obtain a general view of the phenomenon, taking into account the self-modelling, the measured values are represented as dimensionless velocity coefficients: These coefficients quantify the wind effect over the natural circulation of the air through the double skin façade channel. For reference velocities higher than 3 m/s, the self-modelling of the process is noticed and the coefficient determined using Eq. 3 tend to become constant, Fig 7. The self-modelling conditions were achieved and the results for the other velocities were neglected. Therefore, the c v coefficients are the mean of these three velocities. Their values are very approximate, with differences compared to the mean of less than 30%. The Re numbers to which the experiment was accomplished are synthesized in Table 2. It was noted that the values determined for reference velocities above 3 m/s confirm the previous assumption that the selfmodelling is achieved for the studied model.  The experimental results showed that the distribution of measured velocities is very close to the theoretical law, Fig. 8, with small differences for low level, at heights of maximum 5 cm from the horizontal plane. The match ratio of these distributions is presented in Table 3, for the reference velocity of 5 m/s. From 10 cm height above, the match is between 0.94 and 1.19 for the three reference velocities. This correspondence between experimental values and theoretical ones is very important and validates the results of the study. where, d -horizontal measurement (distance from the axis of the wind tunnel) [cm]; h -vertical measurement points (height above the floor of the wind tunnel) [cm].
The theoretical exponential law of vertical distribution of the wind velocities is: ; z g -gradient height [m] (z g = z 10 = 10 m); α -coefficient of terrain. The averaged values for c v coefficients are represented in Fig. 9, in different measuring points. The results regarding the c v coefficients of the velocities inside the channels conducted to the analysis of the convective heat transfer coefficients obtained in the case of a real building, with the same geometrical characteristic as the modelled one. Based on the c v values, the convective heat transfer coefficients, h c , inside double skin façade channel, near PV panels were calculated. Taking into account the annual average velocities in real conditions, the convective heat transfer coefficients were calculated for reference velocities between 0,5 m/s and 3 m/s. Therefore, for this configuration of the double skin façade, the h c coefficients reach the values according to Table 4. The analysis of the cooling impact over the integrated photovoltaic panels is based on the temperature effect over their efficiency [17]. A typical efficiency of the photovoltaic effect of 16 % is considered, for solar radiation of 1000 W/m 2 . Given that in the present study the photovoltaics integration is realized on vertical surface (exterior layer of the double skin façade), a 500 W/m 2 nominal solar radiation can be considered [17]. In order to simplify the interpretation of the results, all data are related to a 1 m 2 monocristaline PV panel. The power produced by the PV panel, at 500 W/m 2 and 25 °C is about 73.4 W/m 2 . The influence of cooling is calculated by taking into consideration an overall effect of the temperature over the efficiency of -0.45%/C [17].
The solar radiation effect used in calculation has a value of 1,66%/100 W/m 2 at 500 W/m 2 . Therefore, for the base case, when the PV panel is integrated in the façade without ventilation, an operating temperature of 75 °C can be achieved. This temperature reduces the maximum power generated by PV panel to 77.5 % of the nominal one. In Table 5 is presented the effect of the cooling the PV panel for different reference velocities.

Conclusions
The experimental results have revealed the effect of the wind over the air circulation and velocities inside double skin façade channel in the near vicinity of the photovoltaic panels. The air flow inside the double skin façade is predominantly horizontal, which represents an important aspect in order to avoid the vertical overheating and extreme thermal draft effect during summer.
The results, quantified as c v coefficients, showed that the average velocity of the air near the exterior glazing of the double skin façade is dependent on the reference velocity, v ref , and has the same order of magnitude. The averaged values of the coefficients, c v , in the median area of the façade, reach values between 1.11…1.25. For the inlet and outlet sections, these coefficients are laying between 0.59…0.91. The double skin façade system is very suitable for building integrated photovoltaic solutions. Therefore, a combination of aesthetics and efficiency is achieved. The cooling effect of natural ventilation of channel by wind action revealed a possible increase of the generated