Field tests to study the pressure equalization on air permeable façade elements

ICBEST 14 June 9, 14, Aachen, Germany Field tests to study the pressure equalization on air permeable façade elements Carine A. van Bentum 1, Ivo Kalkman 1 and Chris P.W Geurts 1, 1 TNO, Department Structural Dynamics, Delft, The Netherlands Eindhoven University of Technology, Department of the Built Environment, The Netherlands Corresponding author: Carine A. van Bentum, carine.vanbentum@tno.nl Abstract The net wind load on the outer layer of a façade consisting of multiple layers is determined by the external pressure and the cavity pressure. For an air-permeable outer layer, the cavity pressure is dependent on the extent to which the external pressure equalizes over the openings. Minimizing the pressure difference between external pressure and cavity pressure has beneficial effects with respect to the amount of fasteners needed for the elements, thus saving material and avoiding unnecessary penetrations of the thermal insulation layer. Current rules to determine the loads on such structures are very crude and conservative, and are based on very little experimental evidence. A full scale experiment has been set up on a 158 m high tower in Rotterdam, the Netherlands, in order to study pressure equalization in a ventilated cavity. The main interest of this study is the net wind load on the outer layer of the façade. The first results show that pressure equalization occurs very fast and that the cavity pressures generally follow the external pressures rather well. 1 Introduction Façades often consist of multiple layers. The inner layer is generally airtight and stiff, thus constituting a barrier for wind and water. Thermal insulation is usually present in front of this layer, which is then covered by an air-permeable layer that serves as a rain screen. Between the inner layer and outer layer, there usually is a ventilated cavity. The wind-induced pressure differences over the rain screen are determined by the external pressures and the pressures inside the cavity which they cause through openings in the rain screen. Minimizing this pressure difference decreases the risk of water leakage and also helps to minimize the local loads on the façade elements. Current rules to determine the loads on such structures are based on very little experimental evidence. Some guidance using very crude rules is given in EN 1991-4 (5) [1]. Gerhardt and Kramer (1983) [] carried out wind tunnel tests on pressure equalization in different scale models of buildings with a relatively small height to width ratio. From the ratio of measured peak pressures it was concluded that pressure equalization takes place on much smaller timescales than the typical gust duration, relatively independently of façade permeability. Equalization was found to be optimal when the flow resistance through the panel openings is small and the flow resistance for cavity flow is large. When cavities on different sides of a building are in direct contact with each other, large differential pressures over the outer layers can be found in corner areas, more or less independent of façade permeability. Gerhardt and Janser (1994) [3] performed wind tunnel tests on buildings with different dimensions, cavity depths and façade permeabilities. Both compartmentalized and noncompartmentalized cavities were studied. They compared their results with full scale experiments. As is the case for closed cavities, the maximum time-averaged force on the façade panels for compartmentalized cavities was found to occur for a 1 degree angle of attack. The corresponding time-averaged pressure coefficient reduction factor was between 15 and 5% for the investigated systems. However, for non-compartmentalized cavities, the peak pressures in corner areas were found to increase by to 5%. Inculet and Davenport (1994) [4] deduced from wind tunnel tests that

Van Bentum et al Field tests to study the pressure equalization on air permeable façade elements /9 pressure equalization is difficult to achieve in corner areas because equalization is incomplete for fast fluctuations (>1 Hz). Although compartmentalization has a beneficial effect on the differential pressures over the façade panels they state that residual values of 1 Pa or more are practically unavoidable. One of the first full-scale measurements of the flow velocity in a cavity was performed by Popp et al.[5] in 198. From this experiment, it was concluded that the flow velocity mainly depends on the geometry and size of the cavity openings; the velocity and direction of the wind; and any temperature differences inside the cavity. Van Schijndel and Schols (1998) [6] studied pressure equalization at normal inflow over a façade panel located in the middle of the facade of the Eindhoven University of Technology main building at a height of 38 m. For the investigated geometries, the reduction in peak load due to pressure equalization lies between 5 and 95 percent for façade panels at overpressure. These values coincide reasonably well with full scale experiments performed by Ganguli and Dalgliesh (1988) [7], who found reductions in peak loads varying between 5 and 6 percent, depending on the position on the panel. Geurts et al. (5) [8] found a pressure equalization factor between.3 and.5 for a typical brick wall with a ventilated cavity. According to a review article by Suresh Kumar () [9] it can be concluded from measurements performed by Straube and Burnett that pressure equalization is feasible on timescales larger than 5 minutes. The equalization of gusts, however, is limited. The authors attributed this to the spatial fluctuations in external pressure rather than temporal fluctuations. In contrast with these findings, Brown et al. (1991) [1] concluded from measurements on a high-rise building that equalization performance is good for both static and dynamic pressures as long as the ventilation openings are large and the cavity volume is small. A research project is currently undertaken in the Netherlands to better understand the basic mechanisms of pressure equalization over air permeable façade elements. As part of this project, a field test has been set up on a 158 m high building in Rotterdam and both CFD calculations and wind tunnel tests have been performed. This paper deals with the field test. Experimental set-up A full scale experiment has been set up on a high-rise residential building situated in Rotterdam, The Netherlands. This building, called the New Orleans, is 158 m high and is situated next to the former harbor area (Figure 1). Prevailing winds are from the south-west over a relatively open fetch. The façade is clad with slabs of natural stone. These typically have a size of 1 m each. The openings between the elements are about 1 mm wide. The air cavity has a depth of (nominally) mm, and the thermal insulation consists of stone wool. At a height of approximately 1 m above the ground, the façade elements have been equipped with pressure transducers (Figure ). The external pressures are measured near the openings between the elements, and the internal pressures are measured in the cavity behind the façade elements. A reference pressure vessel is situated inside the building, to which all differential pressure transducers are connected. An anemometer is installed on top of the tower at a height of.5 m above roof level, which is the maximum allowable height due to architectural restrictions. Unfortunately, this position is in the lee of the building for most wind directions. Therefore, data from a nearby meteorological station (Rotterdam airport) were also used in the data analysis.

Van Bentum et al Field tests to study the pressure equalization on air permeable façade elements 3/9 Figure 1. Left: View of the full scale building (tower at right hand side), photo by Tom Kroeze. Right: Detail of the building façade, showing the thermal insulation, the air cavity and the natural stone façade cladding. B Northwest A Southwest Wind direction 4 Wind direction 135 Northeast C Southeast D Figure. Plan of the building with the pressure tap positions (blue) and the anemometer (red). 3 Measurement data The data presented in this paper were measured between April 1st 1 and November 7th 13 and therefore spans more than 1.5 years of continuous measurements. The total dataset contains 5,936 records of 1-minute measurements. From this dataset, the data under neutral conditions have been selected by applying a velocity criterion. Since the velocity measurements are affected by the building, velocity measurements taken at the KNMI (Royal Dutch Meteorological Institute) meteorological station at Rotterdam airport were also used to select data. The following data selection procedure was followed: Firstly, all velocities measured at the top of the tower were corrected for the influence of the building. This was done by consulting wind tunnel measurements which were performed on a scale model of the New Orleans tower. In these measurements, velocity data were obtained with a thermistor at the position of the anemometer for every wind direction using 15 degree intervals. The ratio between disturbed and undisturbed velocities was used to correct the wind measurements at the site of the New Orleans tower. An important conclusion is that all velocities which were measured at the anemometer position were lower than the undisturbed velocities. Correction of measured velocities to undisturbed velocities therefore lead to an increase in velocity and consequently to a higher number of samples which met subsequent selection criteria. A threshold velocity of 1 m/s measured at the top of the tower was then applied to the adjusted data.

Van Bentum et al Field tests to study the pressure equalization on air permeable façade elements 4/9 For each of the records which met the 1 m/s criterion given above, KNMI data were obtained. The data which were used for this purpose were the 1-minute mean velocities at 1 m height for the last 1 minutes of each hour. An additional threshold of 6 m/s for the KNMI velocity data was specified. Data measured at the New Orleans tower within a timeframe in which the velocities at the meteorological station did not pass this threshold were deemed inconsistent and were therefore removed from the dataset. The total number of selected records in the measuring period is 7689. Figure 3 shows a comparison between the wind velocities measured at the KNMI meteorological station at Rotterdam airport scaled to 16m height and the wind velocities at the New Orleans Tower. Figure 4 gives the number of samples per wind direction that were selected for further data analysis. 35 3 New Orleans tower KNMI Rotterdam 1-minute mean wind speed [m/s] 5 15 1 5 1 3 4 5 6 7 Selected sample number Figure 3. Wind velocities (1-minute means at 16m) Red: meteorological station Rotterdam airport Black: New Orleans Tower 9 8 Number of samples per wind direction [-] 7 6 5 4 3 1 5 1 15 5 3 Angle of attack ( o ) Figure 4. Numbers of samples per wind direction which meet the selection criteria ( wind coming from North)

Van Bentum et al Field tests to study the pressure equalization on air permeable façade elements 5/9 4 Data analysis Pressures were sampled at Hz. For each wind direction where any data were available, pressure coefficients C p (θ,t) were derived according to: p( θ, t) pref ( θ, t) C p ( θ, t) = (1) 1 ρu ref where p(θ,t) is the measured external or cavity pressure for wind direction θ and time t, p ref (θ,t) the reference pressure measured inside the building, ρ the density of air and U ref the velocity measured on top of the New Orleans tower. In addition, for each wind direction where at least 16 records were available, peak pressure coefficients were derived from an extreme value analysis, according to the rules described in the Dutch CUR Recommendation 13 (5) [11], which is based on work by Cook (198) [1]. The peak pressure coefficients are determined from a Gumbel distribution that is fitted to the pressure coefficients: Cθ=U, +.9 1/a () where U p,36 and a p are the mode and standard deviation of the Gumbel distribution. The value of.9 which is given in Eq. () deviates from the value of 1.4 which was derived by Cook. The background for this higher value was described by Van Staalduinen and Vrouwenvelder (1993) [13] based on the wind climate in the Netherlands and corresponds to a 5% probability of non-exceedence of the peak pressure coefficient. Using the time signals of external pressures p e and cavity pressures p c, time series of the differential pressure were derived. The resulting differential pressure coefficients C pdiff (θ,t) are given by: C p, diff pe ( θ, t) pc ( θ, t) ( θ, t) = (3) 1 ρu ref Negative values correspond to a force vector pointing outwards from the façade while positive values imply a force towards the façade. Using Eq. (), peak differential pressure coefficients are also derived. The pressure equalization coefficient C eq is defined as the ratio between the peak differential pressure and the external peak pressure. C eq Cˆ Cˆ p, diff = (4) p, e 5 Results 5.1 Pressures Figure 5 shows the mean external pressure coefficients according to Eq. (1) for all wind directions. Full scale data are only shown for wind directions which contain at least 16 samples. The mean values of the pressure coefficients at the building are compared with a wind tunnel experiment on a scaled model. The mean pressure coefficients of both measurements differ

Van Bentum et al Field tests to study the pressure equalization on air permeable façade elements 6/9 significantly. For the wind directions of 15 degrees an offset in full scale pressure coefficients is visible. A possible explanation is the reference pressure which is determined differently for both measurements. For the full scale measurements the reference pressure is measured inside the building, where a small suction can generally be found. For the wind tunnel measurements, a free stream value measured in the approach flow is used. From Eq. (1), it can be seen that this difference might result in higher values for the pressure coefficients which are obtained from the full scale measurements than those which are found in the wind tunnel tests. For the other wind directions, the differences between full scale measurement and wind tunnel experiment vary per side. Sides with suction show higher values for the mean pressure coefficients in full scale than in the wind tunnel; sides with an overpressure show lower values. Although the reason for this discrepancy cannot be determined from these data, a possible explanation might be an inadequacy of the velocity corrections outlined in section 3. o 15 o 3 o 45 o 6 o 75 o 9 o 15 o 1 o 135 o 15 o 165 o 18 o 195 o 1 o 5 o 4 o 55 o 7 o 85 o 3 o 315 o 33 o 345 o A B C D A Figure 5. Mean external pressure coefficients per wind direction. Black: full scale measurements. Red: wind tunnel results Figure 6 shows both the peak external pressures (black lines) and the peak cavity pressures (red lines) for the minima. On the third façade section (D-A), the cavity pressure coefficients are unknown due to blocked measurement tubes. The cavity pressures in the other sections follow the external pressures very well. Despite this, the peak differential pressures over the outer layer are still appreciable (Figure 7).

Van Bentum et al Field tests to study the pressure equalization on air permeable façade elements 7/9 o 15 o 3 o 45 o 6 o 75 o 9 o 15 o 1 o 135 o 15 o 165 o 18 o 195 o 1 o 5 o 4 o 55 o 7 o 85 o 3 o 315 o 33 o 345 o A B C D A Figure 6. Peak pressure coefficients (minima) per wind direction. Black: external pressures. Red: cavity pressures o 15 o 3 o 45 o 6 o 75 o -.5 -.5 -.5 -.5 -.5 -.5.5.5.5.5.5.5 9 o 15 o 1 o 135 o 15 o 165 o -.5 -.5 -.5 -.5 -.5 -.5.5.5.5.5.5.5 18 o 195 o 1 o 5 o 4 o 55 o -.5 -.5 -.5 -.5 -.5 -.5.5.5.5.5.5.5 7 o 85 o 3 o 315 o 33 o 345 o -.5 -.5 -.5 -.5 -.5 -.5.5.5 A B C D A.5.5.5.5 5. Pressure equalization Figure 7. Differential pressure coefficients (minima) per wind direction. An example pressure trace measured at the peak of a severe storm which hit the Netherlands on October 8th 13 is shown in Figure 8. The cavity pressure coefficients (in red) generally coincide very well with the external pressure coefficients (in black), indicating that the differential pressure is low. It can be concluded that for this geometry the typical timescale of pressure equalization is much shorter than the timescale shown in this figure (~ 1s). The largest excursion which is observed for the differential pressure within the timeframe shown here is 75 Pa: only % of the maximum excursion for the external pressure of 369 Pa. Pressure equalization coefficients are plotted for two different wind directions in Figure 9, together with peak external and differential pressure coefficients. The presented pressure equalization coefficients are determined for the minimum pressures, which are relevant for the wind loads on the

Van Bentum et al Field tests to study the pressure equalization on air permeable façade elements 8/9 fasteners. Although in some areas the equalization coefficient is as low as., it reaches values larger than in others, corresponding with an extreme differential pressure which is twice the extreme external pressure. This might in part be attributed to the metric for pressure equalization Eq. (3) which is used in this research, which becomes singular for zero peak external pressure. When sides with high external suction (side B-C and D-A for a 135 degrees wind direction and side B-A for a 4 degrees wind direction) are considered, values between. and 1 are found for measurement points in the middle of the facade. One measurement point, at side B and the closest to the corner, gives a pressure equalization coefficient of 1.3. Due to practical reasons, no measuring points could be located in the corner zones of the building. It is expected that in these zones, pressure equalization coefficients higher than 1 are more likely to occur. Figure 8. Example pressure trace section measured on the middle of the north-western façade at the peak of a severe storm on October 8th 13. 3.5 135 degrees 3.5 4 degrees 1.5 1.5 1 1.5.5 -.5 -.5.5.5.5 A B C D A A B C D A.5 Figure 9. Pressure equalization coefficients of the minima for a wind direction of 135 degrees (left) and 4 degrees (right). Black: external peak minimum coefficients. Blue: differential peak minimum coefficients. Red: pressure equalization coefficients.

Van Bentum et al Field tests to study the pressure equalization on air permeable façade elements 9/9 6 Conclusions A full scale experiment has been set up on a 158 m high tower in Rotterdam, the Netherlands, in order to study pressure equalization in a ventilated cavity without compartmentalization. The main interest of this study was the net wind load on the outer layer of the façade. First results show that pressure equalization occurs very quickly and that the cavity pressures are able to follow the external pressures very well. Both conclusions are in agreement with earlier findings by other researchers. The pressure equalization coefficients found in areas with high external suction are between. and 1.3. 7 Acknowledgement The authors would like to gratefully acknowledge the sponsorship of Royal HaskoningDHV, Blitta, Centrum Natuursteen, Kenniscentrum Gevelbouw and Vesteda in this research. References [1] CEN, EN 1991-4 (5); Eurocode; Actions on structures; wind actions. [] Gerhardt, H. J., C. Kramer. (!983) Wind loads on permeable building facades. Journal of Wind Engineering and Industrial Aerodynamics, 11, 1. [3] Gerhardt, H.J., F. Janser. (1994). Wind loads on permeable facades. Journal of Wind Engineering and Industrial Aerodynamics 53, 37 48. [4] Inculet, D.R., A.G. Davenport. (1994). Pressure equalized rainscreen: A study in the frequency domain. Journal of Wind Engineering and Industrial Aerodynamics 53, 63 87. [5] Popp, W. E., Mayer, and H. Künzel. (198). Untersuchungen über die Belüftung des Luftraumes hinter vorgesetzten Fassadenbekleidung aus kleinformatigen Elementen. Forschungsbericht B Ho /8, Fraunhofer Institut für Bauphysik. [6] Schijndel, A.W.M. van, S.F.C Schols. (1998). Modeling pressure equalization in cavities. Journal of Wind Engineer-ing and Industrial Aerodynamics 74-76, 641 649 [7] Ganguli, U., W. A. Dalgliesh. (1988). Wind pressures on open rain screen walls. Place Air Canada. J. Struct. Eng., 113(3), 64 656. [8] Geurts, C.P.W., P.W. Bouma, A. Aghaei. (5). Pressure equalization of brick masonry walls, in: Proceedings of the 4th European-African Conference on Wind Engineering, published on CD rom, Prague. [9] Suresh Kumar, K.. (). Pressure equalization of rainscreen walls: a critical review. Building Environ., 35(), 161 179, [1] Brown, W.C., M.Z. Rousseau, W.A. Dalgliesh. (1991). Field testing of pressure equalized rainscreen walls. Exterior Wall Symposium, Precast Concrete, Masonry and Stucco, 59-69. [11] CUR Aanbeveling 13 'Windbelasting op (hoge) gebouwen' (5). Stichting CUR, Gouda [1] Cook, N.J., J.R. Mayne. (198). A refined working approach to the assessment of wind loads for equivalent static design, JWEIA, 8, 1537 [13] Staalduinen, P.C. van, A. Vrouwenvelder. (1993). In situ bepaling van de vormfactor van bouwwerken en onderdelen daarvan, B-9-738, TNO, the Netherlands.