Download The influence of wind direction on natural ventilation

The influence of wind direction on natural ventilation: Application to a large
semi-enclosed stadium
T. van Hooff1,2, B. Blocken3.
1
PhD student, Building Physics and Systems, Eindhoven University of Technology, Eindhoven,
The Netherlands, [email protected]
2
PhD student, Laboratory of Building Physics, Katholieke Universiteit Leuven, Leuven,
Belgium
3
Associate Professor, Eindhoven University of Technology, Eindhoven, The Netherlands,
[email protected]
ABSTRACT
Natural ventilation is a commonly applied way in building engineering to ensure a healthy and
comfortable indoor climate. In this paper CFD simulations of the natural ventilation of a large
semi-enclosed stadium in the Netherlands are described. Simulations are performed to assess the
air exchange rate for a total of eight wind directions. The CFD model consists of both the
complex stadium geometry and the urban environment in which the stadium is located. A grid
sensitivity analysis is conducted, furthermore, validation of the CFD model is performed using
full-scale 3D wind velocity measurements. Comparison of the calculated air exchange rates
showed that the wind direction has a significant effect on the air exchange rate; differences up to
100% were found for the air exchange rate, which can be explained by the position and size of
the buildings upstream of the stadium.
INTRODUCTION
Natural ventilation is nowadays more and more applied in buildings to maintain a healthy and
comfortable indoor climate. The driving force for natural ventilation can be wind or buoyancy,
but usually a combination of these forces is present. Wind speed and wind direction can
influence the amount of natural ventilation to a large extent, but numerical studies of natural
ventilation in which the wind conditions are varied are limited. A previous study concerning the
influence of wind direction on the air exchange rate (ACH) was conducted by Horan and Finn
[1] who examined the air exchange rate for four wind directions and found significant
differences in ACH. Jiang and Chen [2] performed Large Eddy Simulations of fluctuating wind
directions, but their study emphasized on short term small fluctuations of wind direction.
In this paper the influence of wind direction on the air exchange rate of a large multifunctional
sports stadium during summer conditions is investigated using Computational Fluid Dynamics
(CFD). Sports stadia are increasingly being used for other purposes, such as concerts,
conferences and other activities. One of these multifunctional sports stadia, and subject of this
study, is the Amsterdam ‘ArenA’ stadium in the Netherlands. The ArenA is equipped with a
retractable roof construction. When the roof is closed, ventilation can only occur through
relatively small openings in the building envelope. No HVAC systems are present in the stadium,
and therefore the air exchange rate depends on natural ventilation through the roof and openings
in the building envelope, being the only means to ensure indoor air quality and thermal comfort.
A specific feature of this research is the combined and simultaneous simulation of the wind flow
in the complex urban environment around the stadium with the air flow inside the stadium.
DESCRIPTION OF STADIUM AND SURROUNDINGS
SURROUNDINGS
The urban area considered in this study is part of the city of Amsterdam, which is located in the
north-west of the Netherlands. The city of Amsterdam and its surroundings are located on very
flat terrain; height differences are limited to less than 6 m. The stadium is surrounded by medium
and high rise office buildings, and buildings with an entertainment function, such as a cinema
and a concert hall. The height of the current surrounding buildings varies from 12 m to a
maximum of 95 m for the “ABN-AMRO” office building (Fig. 1a) located on the southwest side
of the ArenA.
The aerodynamic roughness length y0 of the surroundings, which is needed for the CFD
simulations, is determined from the updated Davenport roughness classification [3]. The area on
the north side of the ArenA can be classified as “closed terrain” due to the urban character that is
present in a radius of 10 km upwind. The estimated y0 for this area is 1.0 m (Fig. 1b). The south
side area of the ArenA is not as rough as the north side due to the presence of agricultural and
natural areas and can be characterised with an y0 of 0.5 m (Fig. 1b).
Figure 1: (a) Amsterdam ArenA and its surroundings. The two arrows indicate the ArenA stadium and the
highest building in its proximity (ABN-AMRO: 95 m). (b) Terrain surrounding the stadium with a radius of 10
km and estimated aerodynamic roughness length y0. The white square represents the computational domain
used in this study.
STADIUM
The Amsterdam ArenA is a so-called oval stadium (Fig. 2a). The roof is dome shaped and can be
closed by moving two large panels with a projected horizontal area of 110 x 40 m2 (L x W). The
roof consists of a steel frame, largely covered with semi-transparent polycarbonate sheets, while
steel sheets are applied at the edge of the roof until a distance of 18 m from the gutter. The stand
consists of two separate tiers and runs along the entire perimeter. Figures 2a-c show a detailed
plan view and the two cross-sections αα’ and ββ’. The exterior stadium dimensions are 226 x 190
x 72 m3 (L x W x H). The stadium has a capacity of 51,628 seated spectators and an interior
volume of about 1.2 x 106 m3.
Figure 2: (a) Horizontal cross-section of stadium. The arrows indicate the four large openings (gates) in the
corners of the stadium. (b) Cross-section αα’; (c) cross-section ββ’. The four measurement positions () for
air temperature and CO2 concentration inside the stadium are indicated Dimensions in m.
The ArenA is one of the many multifunctional stadiums that have been built in Europe during the
last two decades. Apart from sports events, they also host a wide variety of other activities, such
as concerts, conferences and festivities. For this purpose, many of these stadia are equipped with
a roof construction that can be opened and closed depending on the weather conditions and the
type of event. However, they are generally not equipped with HVAC systems to control the
conditions of the very large indoor air volume (up to 106 m3), which is also the case for the
ArenA. Indoor air quality problems can occur for the configuration with closed roof because of
the large number of spectators and insufficient natural ventilation. During the summer,
overheating can be an additional problem. In absence of HVAC systems, natural ventilation is
the only means to ensure indoor air quality. Natural ventilation can occur through the openings
that are present in the envelope of the stadium. The ArenA has several of such openings. The
semi-transparant roof is the largest potential opening. During concerts and other festivities
however, which are usually held in the summer period, the roof is closed most of the time to
provide shelter for the spectators and the technical equipment. When the roof is completely open,
it is the largest opening (4,400 m2) in the stadium envelope. When it is closed, natural ventilation
of the stadium can only occur through a few smaller openings. The four gates in the corners of
the stadium (Fig. 2a) together form the second largest (potential) opening (4 x 41.5 m2). They are
open most of the time, but are sometimes closed during concerts to limit noise nuisance for the
surroundings. Additionally, two relatively narrow openings are present in the upper part of the
stadium. The first opening is situated between the stand and the steel roof construction, and runs
along the entire perimeter of the roof (Fig. 3a). The total surface area of this opening is 130 m2.
The other opening is situated between the fixed and movable part of the roof (Fig. 3b). This
opening is only present along the two longest edges of the stadium. The total surface area of this
opening is about 85 m2. Of these openings, only the roof and gates can be opened/closed. In the
basic current configuration analysed in this study, the roof is closed, and all other openings are
open.
Figure 3: (a) Ventilation opening between the stand and the roof construction and (b) between the fixed and
the movable part of the roof.
FULL-SCALE MEASUREMENTS
AIR EXCHANGE RATE
To assess the natural ventilation in the stadium, CO2 measurements were performed at four
different locations (Fig. 2a-c), and converted to air exchange rates using the concentration decay
method:
ACH 
ln C  0   ln C  1 
1   0
(1)
With ACH = air exchange rate in h-1, C(τ0) is the concentration at time 0 in ppm, C(τ1) the
concentration at τ1 in ppm and (τ1- τ0) the time between the two measurements. The CO2
concentration at the four positions was measured during three consecutive evenings on which
concerts took place: June 1st until June 3rd, and were made after each concert, when CO2
concentrations had reached a maximum level caused by the attendants. During the
measurements, the potential (y0 = 0.03 m) daily averaged wind speed U10 measured by the
KNMI at Schiphol airport on these three evenings was about 3.5 m/s and the wind direction on
all three days was about 40° from north. The outdoor temperature during the concerts on all
evenings was about 19˚C on average, and the indoor air temperature was about 26˚C. Because of
the similar conditions, the calculated air exchange rate on these three nights was averaged. Table
1 shows that the average air exchange rate for all four measurement positions is about 0.7 h-1 on
these evenings, whereas the minimum air exchange rate according to ASHRAE Standard 62-1
[4] should be at least 1.5 h-1.
Table 1: Average air exchange rate at four positions measured after three concerts.
Position
North, first tier (NE1)
North, second tier (NE2)
Southeast, first tier (SE1)
Air exchange rate (h-1)
0.65
0.74
0.69
Southeast, second tier (SE2)
0.61
WIND VELOCITY
For CFD validation purposes, the 3D wind velocity in and around the stadium was measured in
the period September-November 2007, on days with strong winds (reference wind speed above 8
m/s). Measurements were made with ultrasonic anemometers, positioned on mobile posts, at a
height of 2 m above the ground. The measurement positions included the four openings in the
corners of the stadium. Reference wind speed (Uref; meas) was measured on top of a 10 m mast on
the roof of the 95 m high ABN-AMRO office building, which is the highest building in the
proximity of the stadium (Fig. 1). The data were sampled at 5 Hz, averaged into 10-minute
values and analysed. The measurement results will be reported together with the simulation
results in the validation section.
INDOOR THERMAL CONDITIONS
To analyse the indoor conditions, full-scale measurements were also made of indoor and outdoor
air temperature and relative humidity. Furthermore, the irradiance of the sky was measured to
investigate the influence of solar radiation on the indoor air temperature. These measurements
showed that the indoor temperature depends strongly on the solar radiation, and can exceed the
outdoor temperature by up to 6°C (Fig. 4), which indicates that the natural ventilation is not
capable of removing enough warm air during the day. The CO2 measurements already showed
that the air exchange rate during and after the three concert evenings does not meet the ASHRAE
requirements.
Figure 4: The measured air temperature a inside and outside the stadium and the measured irradiance E, on
a sunny day (July 18, 2007) and a cloudy day (July 20, 2007)
CFD SIMULATIONS: COMPUTATIONAL MODEL AND PARAMETERS
MODEL GEOMETRY AND COMPUTATIONAL DOMAIN
The computational model of the stadium reproduces its geometrical complexity with high
resolution, down to details of 0.02 m. This is required to accurately model the flow through the
narrow ventilation openings (Fig. 3a,b). Because data with such high resolution are not available
from GIS and/or city databases, the construction drawings of the stadium were used. The
buildings that are situated in a radius of 500 m from the stadium are modelled explicitly, but only
by their main shape. Buildings that are located at a greater distance are modelled implicitly, by
imposing an increased equivalent sand-grain roughness height kS and roughness constant CS at
the bottom of the computational domain. These values are based on the aerodynamic roughness
length y0 of the terrain in and beyond the computational domain and on the relationship between
kS, CS and y0 for the specific CFD code [5].
The computational domain has dimensions L x W x H = 2,900 x 2,900 x 908.5 m³. The
maximum blockage ratio is 1.6%, which is below the recommended maximum of 3% [6-7].
Franke et al. [6] also state that the distance from the building to the side, to the inlet and to the
top of the domain should be at least five times the height of the building and the distance from
the building to the outlet should be fifteen times the height. Since the stadium is 72 m high and
the smallest distance to the inlet of the domain is 1,130 m, these requirements are also fulfilled.
GRID
The computational grid consists of 5.6 million prismatic and hexahedral cells. The grid is a
hybrid grid; it is partially structured and partially unstructured. Special attention was paid to the
precise modelling and high grid resolution of the ventilation openings of the stadium. A high grid
resolution is used in the proximity of these openings in order to accurately model the flow. A
grid sensitivity analysis was performed with grids containing 3.0 million, 5.6 million and 9.2
million cells. The 5.6 million grid was found to provide fairly grid-independent results. Some
parts of the computational grid are displayed in Figure 5a,b.
Figure 5: (a) View from north showing part of the computational grid on the surfaces of the stadium and its
surroundings. (b) Bird-eye view of the geometry and grid on the southeast side of the stadium, illustrating
details such as the roof gutter which is modelled in detail for the air flow through the ventilation opening shown
in Fig. 3a.
BOUNDARY CONDITIONS
At the inlet of the domain a logarithmic wind speed profile is imposed with an aerodynamic
roughness length y0 of 0.5 m and 1.0 m, depending on the wind direction, and a reference wind
speed U10 of 10 m/s. The corresponding turbulent kinetic energy and the turbulent dissipation
rate profiles are also imposed at the inlet.
The roughness of the bottom of the domain is taken into account by imposing appropriate values
for the sand-grain roughness ks and the roughness constant Cs which are calculated using
Equation 2 [5]:
kS 
9.793 y0
CS
(2)
To avoid the use of excessively large cells near the ground when using the default values for Cs,
the sand-grain roughness ks in the Fluent 6.3.26 code has been taken equal to 0.7 m and in order
to achieve horizontal homogeneity of the approach-flow mean wind speed profile in this
situation, the value for Cs is set equal to 7 with a user defined function (UDF). More information
on this matter is provided in [5,8]. For the ground surface in the direct vicinity around the
explicitly modelled buildings and the stadium, y0 = 0.03 m is taken, which is imposed by setting
kS = 0.59 m and CS = 0.5. The temperature of the inlet air is set to 20ºC. Zero static pressure is
set at the outlet of the domain and the top is modelled as a slip wall (zero normal velocity and
zero normal gradients of all variables). To roughly take into account the increasing air
temperature inside the stadium because of solar irradiation, estimated surface temperatures are
imposed on several surfaces inside the stadium. These surface temperatures vary from 22ºC to
50ºC. Note that the intention of these simulations was only to compare the performance of
different ventilation configurations. It was not intended to model the exact transient thermal
behaviour of the stadium under transient meteorological conditions.
OTHER COMPUTATIONAL PARAMETERS
The 3D steady RANS equations are solved in combination with the realizable k-ε turbulence
model using the commercial CFD code Fluent 6.3.26 [9]. The realizable k-ε turbulence model is
chosen because of its general good performance for wind flow around buildings [10]. Pressurevelocity coupling is taken care of by the SIMPLE algorithm, pressure interpolation is standard
and second order discretisation schemes are used for both the convection terms and the viscous
terms of the governing equations. The Boussinesq approximation is used for thermal modelling.
CFD SIMULATIONS: VALIDATION AND RESULTS
VALIDATION
CFD validation for the stadium is based on the wind speed measurements mentioned previously.
The measured wind speed at the locations in the four gates is divided by the reference wind
speed measured on top of the ABN-AMRO office building. This wind speed ratio is also
calculated with CFD and both ratios are compared. The validation is performed for two wind
directions, and for a closed and opened roof, depending on the configuration of the roof during
the measurements. For brevity, results are only shown for a wind direction  of 228˚ and a
closed roof, since the configuration with the closed roof is the most interesting one from a
ventilation point of view. Figure 6 compares simulated and measured mean wind speed,
indicating a good agreement for the wind speed ratios (Fig. 6a) and for the wind direction in the
gates (Fig. 6b), except for gate D. The CFD simulation predicts a flow parallel to the opening of
gate D, whereas the measurements showed flow directed into the stadium. Overall, a fair to good
agreement is obtained for the simulations, and the stadium model is used to evaluate different
ventilation configurations.
Figure 6: Comparison between numerical and experimental results in the four gates A, B, C and D, for closed
roof and reference wind direction φ of 228º. (a) wind speed ratio U/Uref; (b) local wind direction φ.
RESULTS
Simulations are performed with eight wind directions:  = 16°, 61°, 106°, 151°, 196°, 241°, 286°
and 331°, all perpendicular or under an angle of 45° to the symmetry axis of the stadium . For
wind directions  between 286° and 16°, no immediate buildings are present upstream of the
stadium, as opposed to  between 61° and 241°. The most, and also the largest, buildings are
situated upstream of the stadium for a wind direction of 196° (Fig. 1a). The reference wind speed
U10 at 10 m height at the inlet is 10 m/s for all simulations. For each ventilation configuration,
the simulated mass flow rates through each opening are used to determine the ACH with Eq. (3)
[11].
ACH 
Q  3600
V
(3)
where Q is the volumetric air flow rate into the enclosure (m3/s) and V the volume of the
enclosure (m3).
The results of the calculations are shown in Figure 7 and indicate that the ACH indeed strongly
depends on the wind direction.The air exchange rate for a wind direction of 16° for example is
twice as high as for a wind direction of 196°, which is the wind direction with the lowest air
exchange rate. This low air exchange rate for  = 196º can be explained by the presence of a
group of large buildings upstream of the stadium. These buildings provide some shelter from
wind. Figure 8 shows the contours of the wind speed ratio U/U10 around the ArenA for  = 196º
(SSW) in horizontal planes at heights of 10, 20, 40 and 60 m above the ArenA deck, at which the
lowest openings (gates) are situated. The lower values around the ArenA indicate that the
stadium is indeed situated in the wake of the office buildings, causing the lower ACH for this
wind direction, which is the prevailing wind direction in the Netherlands. Note that these values
are much higher than the measured ACH because of the much higher wind speed and different
meteorological conditions as during the measurements.
Figure 7: Air exchange rates obtained with CFD for eight wind directions (U10 = 10 m/s). Large differences
are present for the various wind directions.
DISCUSSSION
In this study the air exchange range of a large multifunctional stadium has been calculated using
CFD simulations. Further research is needed on several aspects of the study that has been
performed.
First of all, the CFD simulations in this study were performed steady-state, concerning both the
flow and the heat transfer. Transient thermal simulations can be performed with CFD in the
future for a more detailed thermal analysis. Another possibility is to use a Building Energy
Simulation tool to simulate the transient thermal behaviour in a coupled approach with CFD
simulations for the air flow pattern. This coupling will be subject of future research by the
authors. Transient CFD simulations will also be performed to study the influence of pulsating
flow and large eddies on the air exchange between the building interior and the external flow.
Although the validation study showed a good agreement between the measurements and the
RANS simulations, it would be interesting to compare the air exchange rates obtained with
RANS simulations with results of transient simulations that do take into account time-dependent
flow properties [12,13]. Transient simulations will be performed using Large Eddy Simulation
(LES) and/or Detached Eddy Simulation (DES). Secondly, the balance between wind and
buoyancy as driving forces for natural ventilation will be subject of future research. CFD
simulations will be performed to gain more knowledge on the interaction of both driving forces.
Figure 8: Contours of wind speed ratio U/U10 in four horizontal planes, for φ = 196° (SSW) and U10 = 10 m/s;
at (a) 10 m; (b) 20 m; (c) 40 m and (d) 60 m above the ArenA deck. The lower wind speed ratios around the
stadium indicate that it is situated in the wake of the surrounding buildings, which explains the lower air
exchange rates for this wind direction.
CONCLUSIONS
Coupled CFD simulations are performed to assess natural ventilation in a large semi-enclosed
stadium for a range of wind directions. The following conclusions are made:

Measurements have shown that the air exchange rate of the stadium, with its roof
closed, was insufficient to avoid overheating during summer. Furthermore, the air
exchange rate measured after three concerts was about 0.7 h-1 during the
measurement period, this is only half of the recommended air exchange rate by
ASHRAE.

Wind speed measurements have been used to validate the CFD model of the
stadium and its surroundings and a good agreement has been found.

A grid sensitivity analysis has been performed that has shown that a grid with 5.6
million cells is adequate for this study.

The results demonstrate the importance of modelling the surrounding urban
environment for natural ventilation analysis. Furthermore, this study shows the
need to perform simulations with a range of wind directions, especially if one
wants to take into account the effect of the urban environment and building
asymmetry on natural ventilation.
ACKNOWLEDGEMENTS
The first author is currently a PhD student funded by both Eindhoven University of Technology
in the Netherlands and the Katholieke Universiteit Leuven in Flanders, Belgium. The work in
this paper is a result of his master thesis project at Eindhoven University of Technology. The
measurements reported in this paper were supported by the Laboratory of the Unit Building
Physics and Systems (BPS). Special thanks go to Ing. Jan Diepens, head of LBPS, and Wout van
Bommel, Ing. Harrie Smulders and GeertJan Maas, members of the Laboratory of the Unit BPS,
for their important contribution. The authors also want to thank Martin Wielaart, manager at the
Amsterdam ArenA, for his assistance during the measurements.
REFERENCES
[1] J.M. Horan, D.P. Finn, Sensitivity of air change rates in a naturally ventilated atrium space subject
to variations in external wind speed and direction, Energy Build. 40(8) (2008) 1577-1585.
[2] Y. Jiang, Q. Chen, Effect of fluctuating wind direction on cross natural ventilation in buildings from
large eddy simulation, Build. Environ. 37(4) (2002) 379-386.
[3] J. Wieringa, Updating the Davenport roughness classification, J. Wind Eng. Ind. Aerodyn. 41(1)
(1992) 357-368.
[4] ASHRAE, ASHRAE Standard 62.1-2004, Ventilation for acceptable indoor air quality, American
Society of Heating, Refrigerating and Air-Conditioning Engineers, Atlanta, GA, USA 2004.
[5] B. Blocken, T. Stathopoulos, J. Carmeliet, CFD simulation of the atmospheric boundary layer: wall
function problems, Atmos. Environ. 41(2) (2007a) 238-252.
[6] J. Franke, A. Hellsten, H. Schlünzen, B. Carissimo, (Eds.), Best practice guideline for the CFD
simulation of flows in the urban environment. COST Office Brussels, ISBN 3-00-018312-4, 2007.
[7] Y. Tominaga, A. Mochida, R. Yoshie, H. Kataoka, T. Nozu, M. Yoshikawa, T. Shirasawa, AIJ
guidelines for practical applications of CFD to pedestrian wind environment around buildings, J. Wind
Eng. Ind. Aerodyn. 96(10-11) (2008) 1749-1761.
[8] B. Blocken, J. Carmeliet, T. Stathopoulos, CFD evaluation of wind speed conditions in passages
between parallel buildings—effect of wall-function roughness modifications for the atmospheric
boundary layer flow, J. Wind Eng. Ind. Aerodyn. 95(9-11) (2007b) 941-962.
[9] Fluent Inc., Fluent 6.3 User’s Guide. Fluent Inc., Lebanon, 2006.
[10] J. Franke, C. Hirsch, A.G. Jensen, H.W. Krüs, M. Schatzmann, P.S. Westbury, S.D. Miles, J.A.
Wisse, N.G. Wright, N.G., Recommendations on the use of CFD in wind engineering. In: J.P.A.J. van
Beeck (Eds.), Proceedings of the International Conference on Urban Wind Engineering and Building
Aerodynamics. COST Action C14, Impact of Wind and Storm on City Life Built Environment. Von
Karman Institute, Sint-Genesius-Rode, Belgium, 2004.
[11] ASHRAE, Handbook of Fundamentals. American Society of Heating, Refrigerating and AirConditioning Engineers, Inc., Atlanta, USA, 2005.
[12] Y. Jiang, Q. Chen, Study of natural ventilation in buildings by large eddy simulation. J. Wind Eng.
Ind. Aerodyn. 89(13) (2001) 1155–1178.
[13] N.G. Wright, D.M. Hargreaves, Unsteady CFD simulations for natural ventilation, Int. J. Vent. 5(1)
(2006) 13-20.