See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267679056

Modelling of A Windcatcher for Natural Ventilation

Article

CITATIONS READS

11 472

2 authors:

Ali Mohammad Abbas Abbas Elmualim

The Ohio State University University of South Australia
42 PUBLICATIONS 626 CITATIONS 69 PUBLICATIONS 1,617 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Abbas Elmualim on 08 February 2016.

The user has requested enhancement of the downloaded file.

 Modelling of A Windcatcher for Natural Ventilation
Abbas Ali Elmualim
School of Construction Management & Engineering
The University of Reading
United Kingdom

Abstract
Interest in employing new natural ventilation technologies for the provision of a comfortable internal
environment has been growing in northern Europe for the past number of years. These new
technologies have the potential of reducing energy consumption and the release of destructive
emissions into the environment. The windcatcher system is one of these novel technologies which has
been installed in many buildings in the UK. A windcatcher is a device, installed on a roof, which
utilises both stack and dynamic wind pressures to deliver ventilation air to a building. Wind tunnel,
explicit, implicit and CFD modelling techniques were used to assess the potential of windctacher
systems. Wind tunnel and smoke visualisation tests were conducted on full-scale model of a square
section windcatcher. Pressure coefficients distribution, internal air speed and volumetric air flow were
measured for various wind speeds and directions. The Cp’s were correlated with CFD results and then
used in the explicit and implicit models to further validate the air flow measurements due to
windcatcher. It is concluded that windcatchers will contribute towards the ventilation of spaces. The
flow rate through the windcatcher increases with the increase in wind speed and is dependent on wind
direction. The results obtained using explicit, AIDA and CFX code correlate relatively well with the
experimental results at lower wind speeds and with wind incident at angles of 0-15°.
1. Introduction
Increased global warming and deterioration of the ozone layer have stimulated interest in the use of
renewable energy systems. Natural ventilation is increasingly being employed in modern buildings to
minimize energy consumption and the release of harmful emissions to the environment. Innovative
natural ventilation techniques such as the windcatcher and solar chimney have facilitated the effective
use of natural ventilation in a wide range of buildings for increasing the ventilation rate. In addition to
bringing energy savings, these environmentally friendly technologies also help create healthier
interiors for occupants. The windcatcher systems were employed in buildings in the Middle East for
more than three thousands years. They are known by different names in different parts of the region
(Bahadori 1994, Battle, Zanchetta et al. 2000, Elmualim and Awbi 2002). They were traditionally
constructed from wood-reinforced masonry with openings at height above the building level ranging
from 2 m to 20 m. Taller towers are capable of capturing winds at higher velocities and with less dust
(Bahadori 1994). In the modern design of windcatchers, the two ventilation principles of wind tower
and passive stack are combined in one design around a stack that is divided into two halves or four
quadrants/segments with the division running the full length of the stack. Recently, the windcatcher
systems are increasingly being installed in buildings around the UK. In most of these modern
installations the windcatchers terminate at the ceiling level with four quadrants acting as
supply/extract for the air. Some of these buildings have had great attentions in the recent years, such
as the Queen’s Building at Demonfort University, the Inland Revenue building in Nottingham, a
shopping mall building, Solihull, the Jubilee Campus at Nottingham University and the BRE office of
the future (Elmualim and Awbi 2002). Although windcatchers are becoming more favourable with
architects, services engineers, and researchers, their performance is yet to be evaluated for application
in temperate climate such as the UK. In this paper wind tunnel, CFD, implicit and explicit modelling
were used to evaluate the performance of a commercial windcatcher.
2. CFD and Wind Tunnel Modelling
2.1 Wind tunnel experimental set-up and measurement procedures
The wind tunnel experimental investigation was carried out using an open working section wind
tunnel located at the laboratories of the Building Services Research and Information Association,
BSRIA, Crowthorne, UK. The wind tunnel is purpose designed for testing natural ventilation devices.
It consists of three main sections; one circular section with 1.250 m diameter and two octagonal
fanned shaped sections (2x2m) connected together with a total length of 17.5 m approximately. The
open working section of the wind tunnel, where the windcatcher was positioned is about 3.6 m long.
The wind tunnel was built on metal frame and raised approximately 3 meters above ground floor
level. The tunnel is powered by a variable speed axial fan located at the entrance to the tunnel.
2.1.1 Experimental set-up
A square glass reinforced plastic (GRP) windcatcher of side 500 mm and length of 1.5 m was
connected to a cubic model test room of side 2.48 m (15.25 m3 in volume). The model room was
erected beneath the wind tunnel and silicon sealant and duct tapes were used to seal the test room from
the laboratory. The windcatcher was divided into four equal quadrants each with an area of 0.063 m2.
The quadrants ran the full length of the windcatcher. The windcatcher and the model test room were
positioned centrally within the wind tunnel in order to maintain a uniform wind profile over the
windcatcher (Figure 1). Tests were performed for nominal wind speeds of 0.5, 1, 2, 3, 4, 5 and 6 m/s.
The wind direction was varied in the range of 0 – 45° with an interval of 15° by rotating the
windcather on a turn table (Figure 2).

Figure 1: Experimental test set-up

Figure 2: Experimental test set-up cases summary

2.1.2 Measurement procedure
Upstream static and total pressures were measured using a pitot-static tube. Internal and external
pressure taps were fixed to the windcatcher in the windward, leeward and inside the room to allow the
measurement of static and stagnation pressures. Each windcatcher quadrant was fitted with an internal
pressure tap and three external pressure taps at the top, at the end of louvers and at the middle of the
duct connecting windcatcher to the room. Four additional external pressure taps were fixed to
quadrant 1 to investigate the variation of pressure on the side faces. The pressure taps were connected
to a selection box (Furness FC091MKII) and micromanometer (Furness FC0510) using plastic tubes
of 2mm diameter. The selection box and micromanometer were connected to a PC to facilitate data
entry. The air flow through the windcatcher was measured at the centre of each quadrant at the same
height as the pressure tap using a TSI digital hot wire anemometer (model No. 8357-M-GB). The
manometer was set to read the pressure every 20 seconds. Ten readings were taken sequentially for
every pressure and wind velocity port with time interval of one minute. The recorded readings were
then averaged to give the measured value. All pressure measurements were referenced to the dynamic
pressure upstream using the reference velocity in the wind tunnel. The pressure coefficient is
calculated using:
p − p∞
Cp = 2
0.5 ρv∞
where p is the static pressure on the surface of windcatcher, p∞ is the upstream static pressure and
2
0.5 ρv∞ is the dynamic pressure of the wind. Smoke visualisation tests were carried out using a
Dräger Cumulus hand-held smoke generator, CE smoke generator model RG-100 and smoke pellets.
2.2 CFX CFD Modelling
CFX CFD commercial code, Windows Environment, was used to evaluate the pressure coefficients
distributions and air flow around and through the windcatcher to the room. For the pressure coefficient
investigation, a geometrical representation of a solid model representing the windcatcher inside a wind
tunnel was used. The square windcatcher model was rotated along its axis to simulate wind direction
conditions of 0°, 15°, 30 and 45° exactly as in case 1-4 using BSRIA wind tunnel testing. CFX code
provides an automated mesh generation process. The mesh can then be modified, refined or adapted,
around critical areas of interests or critical values in the simulation. A second order scheme is used
across the simulation as it is expected to achieve better results than first order schemes (AEA
Technology 2002). Various turbulence models including κ-ε, RNG (ReNormalised Grouping),
Reynolds Stress Model and QI (Quasi-isotropic) Reynolds Stress Model were investigated with
solution convergence criteria of Root Mean Square value for the residuals of 0.001. To investigate the
solution independency from the grid many meshes were generated. Meshes with 85,000 elements,
285,000 and 2.3m elements were used with the latter used in obtaining the CFX results presented in
this paper. The QI Reynolds Stress turbulence model was used in this investigation.
3. Results and Discussion
Smoke visualisation tests were carried out prior to each test to establish the supply and extract
quadrants of the windcatcher. The results for the pressure coefficients showed, for all cases, variation
in the measured pressure coefficients with wind speed especially in the low wind speeds (in the range
0.5 to 2 m/s). Table 1 gives the measured and predicted pressure coefficients with wind direction
measured at the middle of the louvers. For cases 1 to 3 (wind directions of 0°, 15° and 30°), positive
pressure coefficients were recorded at the windward side of the windcatcher and negative values at the
other three sides. The results, which were confirmed by smoke visualisation tests, indicate that
quadrant 1 was acting as an air supply while the other three quadrants, 2, 3, and 4, were acting as air
extracts. It was observed in the three cases that pressure coefficients at quadrant 3 were marginally
higher than in quadrants 2 and 4.
The measured pressure coefficients in case 4 with the wind direction at 45°, hitting the windcatcher at
the junction line between quadrant 1 and 2, show that quadrant 1 and 2 were acting as air supply with
positive pressure coefficients recorded at each face with a higher Cp at quadrant 1 than in quadrant 2.
Negative pressure coefficients were measured at quadrants 3 and 4 and hence rendering them as the
air extract quadrants. The value of the positive pressure coefficient for quadrant 1 in this case was
found to be approximately 45% lower than in case 1 with the wind at 0°.
The measured Cp’s in the wind tunnels were compared with the CFD results and other experimental
results from Farija (1996) and BSI BS 6399-2 (BSI BS 6399-2 1998).The pressure coefficients
obtained using CFD correlate reasonably well with the experimental results at the stagnation face. It
was observed that the CFD predictions lie in the middle of the experimental data. At the stagnation
point CFD predictions gives lower Cp values than the experimental data. The predicted CFD Cp value
at that point was 0.84 compared with values of 0.83, 0.86 and 1.00 for wind tunnel, BSI BS6399-2
(1998) and Farija (1996) respectively. There was large variation of the measured Cp’s at the side and
the leeward of the windctcher. Such pressure variations between experimental and CFD predictions
were also reported by (Farija 1996, Karakatsanis, Bahadori et al. 1986). Despite the limitation of the
wind tunnel tests on full-scale windcathcer at BSRIA, it is the nearest representation of real
conditions. CFD predictions represent ideal conditions for the testing. The limitation of both
modelling techniques should be taken into account when analysing pressure coefficients. However,
both tools provide an interesting tool for modelling of windcatchers
Table 1: Comparison between measured and predicted pressure coefficients

Wind at 0° Wind at 15° Wind at 30° Wind at 45°

Quadrant
Wind Wind Wind Wind
CFX CFX CFX CFX
tunnel tunnel tunnel tunnel

Face 1 0.83 0.84 0.83 0.60 0.70 0.38 0.51 0.30

Face 2 -.034 -0.55 -0.22 -0.53 -0.20 -0.65 0.11 0.20

Face 3 -0.10 -0.44 -0.10 -0.48 -0.10 -0.46 -0.20 -0.58

Face 4 -0.33 -0.55 -0.30 -0.34 -0.23 -0.10 -0.20 -.058

Figure 3 shows the measured and predicted volumetric airflow through the windcatcher quadrants for
a wind direction of α = 0°, case 1. In this case, quadrant 1 was identified as the supply while
quadrants 2, 3 and 4 were acting as the extract quadrants. The airflow increases with external wind
speed and no short circuiting was observed. A volumetric airflow of 0.101 m3/s (101 l/s) was achieved
through the main supply quadrant for an average wind speed of 3 m/s. The flow reached a volume of
approximately 0.135 m3/s (135 l/s) for an external wind speed of 5.4 m/s. The measured and predicted
results for this case (case 1) correlate well particularly at lower wind speeds of 1-3 m/s. AIDA
(Liddament 1996) gives slightly higher results than the other methods when using the Cp’s measured
in the wind tunnel. By applying the Cp’s measured using CFX in AIDA much higher air flow were
obtained. This suggests that the variation in the Cp’s directly affect the calculated air flow rate which
emphasises the importance of establishing accurate Cp’s for windcatchers.
Figure 4 shows comparison between the measured volumetric air flow using wind tunnel tests and
CFD modelling for various wind speeds for all the cases, 1-4. Variation between the measured and
predicted air flow results were observed at incidence angle of 30 and 45° and higher wind speed of 3-
6 m.
The CFD code gives reasonably air flow rate compared with wind tunnel results but not Cp’s. There
was inherent difficulty in establishing the Cp’s in wind tunnel due to the measured low pressures and
also in the inaccuracy of the manufacturing of the windcatcher. CFX was valuable in showing the
flow pattern in and around the windcatcher and inside the test room. It provides information not
available when using other predictions methods namely wind tunnel, explicit and implicit models.

Wind tunnel wind at 0° CFX-Wind at 0°-ave

Explicit Method AIDA wind at 0°
AIDA-CFX-Cp-0°
0.300

0.250
Air flow rate (Q) m3/s

4
0.200 1
Case 1
0.150 3

2
0.100

0.050

0.000
0 1 2 3 4 5 6 7
Wind speed m/s

Figure 3: Measured air flow through the windcatcher system for case 1 (α = 0°) for various
prediction models

Wind tunnel wind at 0° Wind tunnel wind at 15° Wind tunnel wind at 30°
Wind tunnel wind at 45° CFX-Wind at 0°-ave CFX-Wind at 30°-ave
CFX-Wind at 45°-ave CFX-Wind at 15°-ave

0.300

0.250 4
Air flow rate (Q) m3/s

1
0.200 3
Case 1

2
α
0.150

0.100

0.050

0.000
0 1 2 3 4 5 6 7
Wind speed m/s

Figure 4: Variation of air flow with wind speed and wind direction using wind tunnel and CFD
modelling
4. Conclusions
Several prediction models were used to assess the performance of the windcatcher for natural
ventilation purposes in buildings. Wind tunnel and smoke visualisations tests, explicit, AIDA implicit
model and CFX CFD code were used in this investigation. The experimental testing in the wind tunnel
was carried out for a square sections windcatchers (500 x 500 mm and a height of 1.5 m). The
windcatchers was connected to a test room in open section wind tunnel at BSRIA. The tests were
conducted for various wind speeds and directions. Air flow, pressure coefficient and internal air speed
were measured. The wind tunnel configuration was then simulated using commercial CFX CFD codes
to investigation the suitability of such model in predicting the ventilation rates due to windcatcher
applications. The measured pressure coefficients and pressure losses for windctachers were then
applied into the explicit and AIDA models to establish the air flow.
The windcatcher performance depends greatly on the direction and speed of the wind in relation to the
windcatcher quadrants. In all cases examined, the wind tunnel tests showed that the ventilation rate
increases with wind speed and slightly decreases with the increase of the angle of wind direction in
the range between 0-45°, particularly at lower wind speeds. The results obtained using explicit, AIDA
and CFX code correlate relatively well with the experimental results at lower wind speeds and with
wind incident at angles of 0-15°. However, CFX gives higher results at wind incident angle of 30 -
45°. This was attributed to the difficulty in accurately measuring the air flow due to the geometrical
and meshing complications for the simulated cases in CFX. The optimum position for the windcatcher
for maximum ventilation is when the wind is incident directly on to a complete quadrant (α = 0o).
Experimental investigation using wind tunnel testing is vital in establishing the performances of
windcatchers for natural ventilation in buildings particularly on large-scale models. Experimental
results have much credibility in the research community as well as the industry. However, they are
expensive to acquire and demanding in terms of time and personnel. The wind tunnel testing is
indispensable in measuring Cp and pressure losses required by the widely used explicit and implicit
models. The CFD modelling results should be considered with care particularly in the absence of
experimental data.
5. Acknowledgment
The author would like to acknowledge the support of BSRIA, Ansys CFX and Monodraught Ltd in
carrying out this investigation.
6. References
AEA Technology (2002) CFX-5 User Manual, Harwell, UK: AEA Technology, www.cfx.aea.com.
Bahadori, M N (1994) Viability of Wind Towers in Achieving Summer Comfort in the Hot Arid
Regions of the Middle-East. . Renewable Energy, 5(5-8), 879-892.
Battle, G S, Zanchetta, M and Heath, P (2000) Wind Towers and Wind Driven Ventilation. In: Ed:
Sayigh, A (Ed.), World Renewable Energy Congress VI (WREC2000), Brighton. Elsvier Science
Ltd432-437.
BSI BS 6399-2 (1998) Loading for Buildings, Part 2: Code of practice for wind loads, London: British
Standard Institution.
Elmualim, A A and Awbi, H B (2002a) Wind Tunnel and CFD Investigation of the Performance of
Windcatcher Ventilation Systems. International Journal of Ventilation, 1(1), 53-64.
Farija, G M (1996) Wind Induced Natural Ventilation for Wind Tower Houses in Maritime -Desert
Climates with Special Reference to Bahrain, PhD, Construction Management and Engineering, the
University of Reading.
Karakatsanis, C, Bahadori, M N and Vickery, B J (1986) Evaluation of pressure coefficients and
estimation of air flow rates in buildings employing wind towers. Solar Energy, 37(5), 363-374.
Liddament, M W (1996) A Guide to Energy Efficient Ventilation. University of Warwick, UK:
International Energy Agency, AIVC, Air Infiltration and Ventilation Centre.

View publication stats