Sie sind auf Seite 1von 25

See

discussions, stats, and author profiles for this publication at:


https://www.researchgate.net/publication/245189703

A ventilated courtyard as a passive


cooling strategy in the warm
humid tropics
ARTICLE in RENEWABLE ENERGY SEPTEMBER 2003
Impact Factor: 3.48 DOI: 10.1016/S0960-1481(03)00012-0

CITATIONS

READS

46

746

3 AUTHORS, INCLUDING:
Indrika Rajapaksha
University of Moratuwa
19 PUBLICATIONS 95 CITATIONS
SEE PROFILE

Available from: Indrika Rajapaksha


Retrieved on: 06 April 2016

Renewable Energy 28 (2003) 17551778


www.elsevier.com/locate/renene

A ventilated courtyard as a passive cooling


strategy in the warm humid tropics
I. Rajapaksha a,, H. Nagai b, M. Okumiya c
a

Department of Architecture, Graduate School of Engineering, Nagoya University, Furou-cho,


Chikusa-Ku, Nagoya 464-8603, Japan
b
Division of Environmental Engineering and Architecture, Graduate School of Environmental Studies,
Nagoya University, Furou-cho, Chikusa-Ku, Nagoya 464-8603, Japan
c
Center for Integrated Research in Science and Engineering, Nagoya University, Furou-cho, ChikusaKu, Nagoya 464-8603,Japan
Received 1 February 2002; accepted 27 December 2002

Abstract
The paper investigates the potential of a courtyard for passive cooling in a single storey
high mass building in a warm humid climate. The inclusion of an internal courtyard in building
design is attributed to the optimization of natural ventilation in order to minimize indoor
overheating conditions. However, the efficiency of this strategy greatly depends on the design
details of the building composition in providing appropriate airflow pattern to the courtyard.
From the results of thermal measurements, a significant correlation between wall surface temperatures and indoor air temperatures is evident. A reduction of indoor air temperature below
the levels of ambient is seen as a function of heat exchange between the indoor air and high
thermal mass of the building fabric. However, this behavior is affected by indoor airflow
patterns, which are controlled through the composition between envelope openings and the
courtyard of the building.
From a computational analysis, several airflow patterns are identified. A relatively better
indoor thermal modification is seen when the courtyard acts as an air funnel discharging indoor
air into the sky, than the courtyard acts as a suction zone inducing air from its sky opening.
The earlier pattern is promoted when the courtyard is ventilated through openings found in
the building envelope. The computational simulation utilizing the standard k- turbulent model
with isothermal condition agrees closely with the measurements taken from the field investigation.
2003 Elsevier Science Ltd. All rights reserved.
Corresponding author. Fax: +81-617-3219-6629.
E-mail addresses: indrikaFr@hotmail.com (I.
Rajapaksha).

Rajapaksha);

upendraindi@hotmail.com

0960-1481/03/$ - see front matter 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0960-1481(03)00012-0

(I.

1756

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

Keywords: Airflow pattern; Opening composition; Passive cooling; Tropical courtyard; Warm humid
climate

1. Introduction
The concerns over global warming and the need for reduction of high emissions
of greenhouse gases demand the utilization of passive strategies for indoor climate
modification in promoting comfortable indoor environments. In such systems, the
energy required for indoor climate modification is obtained from natural renewable
sources but limitations for its applicability arrive through the prevailing climate [1].
In warm humid tropics, overheated building interiors are common due to solar
penetration through the buildings envelope and windows. Under these conditions,
and with the presence of high levels of relative humidity, the utilization of natural
ventilation is recommended as a suitable cooling option for the prevention of
overheating and indoor climate modification. Thus, design strategies to optimize
natural ventilation and maximize its effect with the thermal mass in reducing the
maximum of indoor temperature in buildings is of great importance.
In high-mass buildings with windows open and in the shade all day, the indoor
daytime temperature is consistently below that of outdoors [2]. Given the properties
of thermal mass of the building fabric with effectively shaded windows, the airflow
pattern to and within the building has an impact on indoor air temperature reduction.
Effectiveness of this process is dependent on the optimum exposure of the building
to the air movement that can enhance a greater surfaceair contact [3]. Thus, dealing

Nomenclature
Uin(z)
A
Uh
Ut
z
n
h

a
r
Cm
L

y
k

mean wind speed at height z


constant for log law
mean wind speed at reference height h
shear velocity
height above ground level
kinematic viscosity
reference height
Karmans constant (0.4)
exponent characteristic of terrain roughness
density
constant for two equation turbulent model
Length scale
rate of dissipation of turbulence energy
distance from the wall to the first cell of the grid
kinetic energy of turbulence

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

1757

with wind driven ventilation behavior and its effects on indoor air temperature
becomes important in such climates. This study aims to illustrate a passive cooling
approach that has been used to exploit the optimum use of wind-driven natural ventilation in preventing indoor overheating conditions and lowering maximum indoor
temperature below the ambient level. The illustration is focused on a particular building type that has an internal courtyard by investigating the performance and effectiveness of design strategies used in a single-storey high mass courtyard house found in
a suburb of Colombo, Sri Lanka.
The study investigates the effects of the airflow pattern to the courtyard on the
exposure of the building interior to the air movement in enhancing optimum surface
air contact and thus indoor climate modification. The goal is to provide some guidelines for manipulating courtyard-building compositions to improve indoor air movement and lower the maximum indoor air temperatures in warm humid climates.
The monitoring of thermal behavior within the case study building utilized five
different boundary configurations (shown in Table 1). The monitoring was conducted
on site from the 12 April to 3 May 2001. Boundary configurations of the building
were changed through closing and opening the major airflow access points found
within the building envelope. Each boundary configuration (opening composition)
showed different levels of indoor air temperature modifications and indoor airflow
rates. This paper discusses the results of the thermal performance of the building. It
was investigated using two identified different boundary configurations that resulted
in minimum and maximum indoor air temperature modifications and corresponding
airflow rates out of five boundary configurations as mentioned above. Later, the
airflow patterns and airflow rates were simulated using computational analysis (CFD)
Table 1
Different boundary configurations and corresponding values
Opening composition

Air temp modification (T)


Courtyard

Courtyard (case 1)
Op1,2 and Cy (case 2)
Op3,4 and Cy (case 3)
Op1,2,3 and Cy (case 4)
Op1,2,3,4 and Cy (case 5)

Interior zone

Day

Night

day

0.34
1.04
0.68
0.96
0.74

1.38
0.92
1.18
0.91
1.6

0.42
1.4
0.52
1.5
0.62

T indoor air temperature modification


T ToTi
To, daytime mean outdoor ; Ti, daytime mean indoor temperature
Ti courtyard daytime mean of Tcy1
Ti interior zones daytime mean of 1 / 9

(Ta1,Ta2,Tb,Tc,Td,Te1,Te2,Tf1,Tf2, refer Fig. 4A)

i=1

(1)

1758

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

and simulated airflow rates were compared with the calculated values obtained from
field investigations.
2. Ambient climate
The building investigated in this study is built within a suburb located 20 km from
Colombo, Sri Lanka (then Ceylon), at 6.5 N latitude and about 15 m above sea level.
The ambient climate is characterized by high temperatures and relatively high levels
of humidity. The mean monthly temperatures range from 27.2 C in November to
30.3 C in April; relative humidity varies from 7080% during a typical year.
While seasonal variations through temperature do not exist, there are two recognizable seasons due to wind and rainfall. The northeast monsoon occurs from June to
August and southwest monsoon occurs from December to February. The remaining
inter-monsoon represents the dry season while heavy rainfall is observed in monsoon
periods. Clear and fine weather with tropical thunderstorms brings slight daytime
wind and rain in the late afternoons. Also, the daily pattern of the dry season has
diurnal temperature ranges of 78 C.
3. Comfort temperature and required design strategies
3.1. Ambient climatic conditions in relation to Givonis bioclimatic chart for
Colombo
Fig. 1 shows Givonis bio-climatic chart for Colombo. The chart represents the
typical mean daily maximum and mean daily minimum values of relative humidity

Fig. 1.

Givonis building bio-climatic chart applied to Colombo, Sri-Lanka.

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

1759

and temperature for different months of the year, 2001. To extend the comfort zone
for higher humidity and higher temperatures, higher air velocities are required. This
is quiet extensive given the relative humidity around 7080%.
3.2. Neutrality temperatures in relation to dry bulb temperatures for a typical
year (2001)
The work of Humphreys [4] and its continuation by Auliciems [5] give the neutrality temperature as a function of monthly mean outdoor temperature. The neutrality
temperatures for Colombo are within 27.328.5 C for the year 2001. Fig. 2 gives
the neutrality temperature and comfort zone, which is defined as 2 C above and
below the neutrality temperature in relation to dry bulb temperature and still air conditions.
From the collective evidence of 11 different studies by others, Szokolay [6] confirms (taking into consideration of humidity, temperature, radiation balance, activity
level and clothing levels) that an air velocity of 1 m/s is capable of extending the
upper limits of comfort dry bulb temperatures by 3.7 C under warm humid conditions. This is applicable to situations where occupants wear light clothing (less
than 0.5 clo) for activity levels less than 1 met. The mean daytime wind velocities
in Colombo (still conditions during night) are around 0.9 m/s and thus the available
indoor air velocities are even below this value. At the highest level of applicability,
an air velocity of 0.9 m/s can extend the comfort zone up to 32.0 C.
3.3. Typical daily weather pattern in April (a month of the hottest period)
A closer look at the ambient climatic conditions during February to May in Colombo reveals that daytime dry bulb temperatures remain above the extended upper
limits of dry bulb temperature for most part of the daytime hours (refer Fig. 3).
Therefore, a strategy to reduce daytime maximum indoor air temperature, in addition
to the provision of indoor airflow, is required.

Fig. 2. Seasonal pattern of neutrality temperature and comfort zone in relation to dry bulb temperature
of Colombo, Sri Lanka, 2001.

1760

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

Fig. 3. Daily pattern of dry bulb temperature and relative humidity of Colombo, Sri Lanka (during the
very hot month of April).

4. Field investigation
4.1. The investigated building
The case study building presented in this study employs a number of passive
design strategies to modify the warm ambient climate in order to minimize thermal
discomfort from overheating conditions. The inclusion of a courtyard, high mass
building fabric, high timber ceilings and wide roof eaves are of significant importance.
The 230m2 rectangular building form that is designed for a single family accommodates a central rectangular courtyard measuring 3.7 and 8.1 m as the main design
strategy. The use of an internal central courtyard is a dominating feature in traditional
buildings found in Sri Lanka. The courtyard occupies approximately one tenth of
the buildings footprint, is bounded by a living area to the west and bedrooms to
the east. It can be visually opened to the outdoor environment through two axes
passages, which are perpendicular to each other.
By locating the rectangular courtyard at the center, the plan depth of the surrounding indoor spaces is minimized for better cross ventilation potential. With four major
openings (Op1, Op2, Op3 and Op4 shown in Fig. 4A and Fig. 5) placed in the
envelope, the courtyard maintains direct contacts with the external environment. This
enhances cross ventilation to the building and thus the exposure of high mass brick
walls to the incoming air movement. Opening 1, which is located at the north end
of the longitudinal axis, acts as the main entrance to the house.
The louvered door measuring 4.3 m2 at this opening can promote slight ventilation
even if the door is closed during the night. The other opening on the south end of
the longitudinal axis is a large vertical grill measuring 3.6 m2 (opening 2), which
promotes permanent ventilation during the day. In addition, openings 3 and 4 (each
measuring 3.6 m2) are located on the east and west ends of the cross axis. One third
of the courtyard floor is covered with water. Though there are other openings on the
building envelope (doors and windows), they remained closed during the study period. The courtyard dominates within the building section and is the tallest space

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

1761

Fig. 4. The enclosed courtyard house and measurement points in the field investigation.

Fig. 5.

3D form of the building showing opening composition.

within it (Fig. 4B). Having introduced openings at relatively lower levels within the
envelope, the potential for airflow due to stack effect during the night is encouraged.
The high-mass brick building has high internal walls with low external walls (22cm thickness including cement and lime plaster) to the east and west. Through this
design, east and west orientations are provided with shaded verandahs. Wide roof
eaves on all four sides protect all the windows and doors (openings) from solar
radiation throughout the day. The floor of the building is finished with clay terracotta
tiles while the roof is covered with half round clay tiles (Thermal Conductivity:
8.6 105 Wcm1 C1, Absorptivity: 0.650.74, [7]) with a timber ceiling underneath. Insulation is not found throughout the building.

1762

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

4.2. Experimental setup


4.2.1. Surrounding environment
The building is located on a large plot of land with tall trees. The immediate
surroundings of the building have not been affected by any other building. Uneven
terrain conditions with canopies of tall trees create a 0.25 terrain roughness similar
to that of a typical semi-urban type condition.
4.3. Boundary configurations (opening composition)
The courtyard building was monitored for 10 days from 12 April to 3 May 2001
with five different boundary configurations with regard to composition of major
openings found within the buildings envelope (shown in Table 1). Each boundary
configuration was investigated for two different days. However, this paper aims to
present the thermal performance of the building under the two of most significant
boundary configurations that demonstrated the worst and the best indoor climate
modification.
The rest of the boundary configurations also exhibited different levels of climate
modifications, but varied within the results of the two most significant cases
presented in this paper. The two cases that are presented in this paper are referred
to as case 1 and case 2. Case 1 was conducted on the 22 April and case 2 on the
18 April 2001.
In case 1 during the daytime, openings in the external envelope were kept closed
and the building was allowed to ventilate only from the courtyard top. In case 2,
the openings found at the end of longitudinal axis were kept open allowing daytime
ventilation through opening 1 and 2 in addition to the courtyards top.
During nighttime in both cases, opening composition remained similar with opening 2 and courtyards top open, while opening 1 was closed.
4.4. Data collection
Under each boundary configuration air temperature and humidity measurements
were taken at 5-min intervals using sensors, which recorded day and night (daytime
9:00 to 18:00, nighttime 18:00 to 9:00). Indoor wind velocities were taken at 1-s
sampling time and 1 min intervals using hot wire anemometers. Wall temperatures
were measured at 5-min intervals day and night using 0.5 mm copper constant thermocouples glued to the wall surfaces and covered with paper tapes.
Meteorological conditions were recorded at a weather station built on the site
using temperature and humidity sensors with recorders protected with 12 plate radiation shields, cup anemometer and a vane with a data logger. Measurements were
later averaged for hourly values.
4.5. Instrumentation
Tabai Espec data logger with RS 10/11 sensors were used for air temperature and
relative humidity measurements. The accuracy for temperature measurements was

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

1763

stated as 0.3 C from 545 C and 3% over 1090% relative humidity. Further,
sensors were shielded against direct and reflected solar radiation from surrounding
surfaces using Campbell Scientific Gill 41004, multi plate (12 plates) radiation shields.
Kanomax model 6511 Hot-wire anemometers were used for indoor air velocity
measurements. These were sensitive to air velocities down to 0.05 m/s. Campbell
Scientific 03001 cup anemometer and wind vane (Youngs wind sentry wind set)
coupled with CR 21X data logger was used for outdoor wind speed and direction
measurements. The outdoor cup anemometers were initially calibrated in the wind
tunnel using hot-wire anemometer probes. High correlation (r = 0.99) was observed.
Eto denki EF 5020A Thermodac data logger with type-T thermocouples (with an
accuracy of 0.2 C over 050 C range) was used for wall surface temperatures.
All equipments were recommended with research-grade accuracy.
4.6. Measurement locations
Measurements were recorded in the courtyard and selected indoor spaces. Air
temperature, relative humidity and wind velocities were measured at 1.1 m(Tcy1),
2.5 m(Tcy2) and 3.5 m(Tcy3) heights at the center of the courtyard (refer Fig. 4B).
Indoor spaces were measured for air temperature and humidity at human body level:
1.1 m from the floor level (Fig. 4A).
Wall temperatures were measured at the center of the main exterior surfaces of
the building (W6 W10) and the five interior wall surfaces that surround the courtyard (W3, W4), longitudinal axis (W1, W2) and the family living (W5) at 1.1 m from
the ground (refer Fig. 4A).
Wind velocity measurements were recorded at 0.6 m(V1), 1.2 m(V2) and 1.8 m(V3)
heights at each exterior opening of the longitudinal axis (Op.1 and 2, refer Fig. 4C).
The weather station installed on the site recorded climatic conditions at three levels: 1.1, 5.5 and 10 m ensuring freeness from wind shadows due to built or natural
structures (refer Fig. 14A).
4.7. Criteria for the assessment of data
The assessment of measurements is mainly concerned with the effectiveness of
the courtyard in lowering the air temperatures of its own and surrounding indoor
spaces below the level of daytime ambient temperature.
4.8. Ambient site climate
The ambient weather pattern was observed as similar throughout the study period
and even several days before the study. The maximum ambient temperature of 32.8
C was recorded between 13:0015:00 h, while the minimum temperature of 25.5
C was recorded in the morning at 4:007:00 h. Daytime humidity remained between
6070% with nighttime humidity at 99%.
Prevailing winds were from the southwest during the daytime and from the nor-

1764

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

theast and east at nighttime. Mean daytime wind speeds were of 1.2 and 1.5 m/s at
5 and 10 m heights, respectively, with almost still wind conditions (or 0.2 m/s)
at night.

5. Results and discussion of thermal investigation


Table 1 shows the average values of the modified measured air temperatures in
the courtyard and interior zones. The mean daytime modification of all interior zones
is presented as a single value. A positive value represents a reduction of air temperature below the ambient level while a negative value represents an overheated condition.
In each case, the results indicate a variation of air temperature modification in the
courtyard as well as in the interior zones. The difference is not correlated with the
uniformity of the ambient weather condition. Therefore, the best performance results
from case 2, and the worst performance results from case 1 with regard to cooling
are assessed in the following discussion.
5.1. Courtyard air temperature
The hourly air temperatures (TCy1) at the body level within the courtyard were
monitored and compared with the ambient values. Fig. 6 shows the air temperature
patterns within the courtyard for the two different boundary configurations together
with ambient temperature patterns.
Temperature pattern within the courtyard has followed the ambient pattern. A
reduction of courtyard air temperature below the ambient level is observed in both
cases 1 and 2 as expected in high-mass night ventilated buildings, yet, a clear variation in the values and pattern in case 1 and 2 was seen.
The maximum courtyard air temperature of 32 C was recorded with the maximum
ambient temperature of 32.7 C at the same time by 14:00 h in case 1. The modification of air temperature is only 0.7 C. During case 2, courtyard maximum air

Fig. 6. Ambient temperatures and courtyard air temperatures (Tcy1) at 1.1 m in case 1 and case 2.
Daytime maximum temperature in case 2 is lower by 2 C and seen 2 h later the maximum ambient
recorded.

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

1765

temperature decreased 2 C below the ambient maximum by 15:00 h, 2 h later than


the ambient maximum was recorded.
These results indicate a clear reduction of courtyard air temperatures during the
daytime in case 2.
The time lag between ambient and courtyard maximum values indicates an effect
of exposure of high mass brick walls to the airflow. The ambient weather conditions
(air temperature, wind direction, wind velocity, humidity and rainfall) remained similar in both cases. These results indicate that a relatively greater reduction of air
temperatures in case 2 correlate with the daytime indoor ventilation pattern (the only
difference in these two cases).
In Case 1 during the early nighttime, courtyard temperatures remained 0.52.0 C
above the ambient levels. The nighttime temperature difference in case 2 is only
0.51.0 C. These results indicate that maximum nighttime temperature in case 2 is
lower by 0.8 C than in case1 at 19:00 h. In both cases, the early daytime courtyard
temperatures closely followed the ambient between 8:00 and 9:00 h.
5.2. Vertical temperature profile
Fig. 7 shows the vertical temperature profile in the courtyard for case 1 and 2.
Results indicate a similar pattern of stratification during day and night in both cases.
Nevertheless, maximum daytime temperature (33.4 C) at the top level in the courtyard increased above the ambient in case 1, while the maximum temperature at the
top-level remained below the maximum ambient in case 2.
Also, the differences representing lowering of air temperatures (below the ambient)
at the bottom and middle levels in the courtyard are greater in case 2 than in case
1. A density difference due to air temperature is essential for stack effect to take
place. Results obtained for vertical thermal stratification indicate that temperatures
in the courtyard are below the temperature levels at the major openings found in the
external envelope during daytime. Thus, the density levels at the openings (1 and

Fig. 7. Vertical temperature profile in the courtyard in case 1 and case 2. Thermal stratification is shown
in both case 1 and 2. Maximum temperature at the top level is higher than the maximum ambient in case
1. Relatively lower levels of air temperatures recorded in the courtyard in case 2.

1766

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

2) are lower than the courtyard preventing the potential for stack effect during the
daytime.
Vertical temperature profile indicates a heat gain from the courtyards top in case
1 and 2. However, relatively lower levels of air temperatures at bottom, middle and
the top levels in case 2 indicate that the effect of heat gain from the top opening is
moderated due to the potential of upwind forced ventilation in the case 2, as compared to case 1 (refer Section 8 for the airflow pattern).
5.3. Wall surface temperature
The hourly internal wall surface temperatures in the three main areas (area next
to opening 1, courtyard and family living on the west) were compared against each
other and the external wall surface temperatures (north-W7, south-W10, east-W8 and
W9, and west-W6 walls) at the human body level.
All internal walls in each area were monitored and later averaged for a mean
value. Fig. 8, shows that mean surface temperatures of the courtyard walls (W3 and
W4) and the axis walls (W1 and W2) remained above against each other depending
on the airflow pattern. The mean surface temperatures in the living area remained
constant with the airflow patterns.
External surface temperatures remained constant on both days while daytime
internal wall surface temperatures (W5) decreased by 1 to 3.4 C well below the
external values. Nevertheless, relatively higher surface temperatures (W3 and W4) in
the courtyard in case 1 indicate heat absorption and thus the availability of incoming
outdoor air from the courtyard. On the other hand, higher surface temperatures in
the axis walls in case 2 exhibit heat absorption, and thus the availability of incoming
air through the axis.
In case 1, the maximum value of 29.8 C in the courtyard was recorded at 14:00
h without showing a time lag with the mean maximum of the external walls. Contrary
to this behavior, in case 2, a time lag for the courtyard surface temperature was seen.
The maximum courtyard surface temperature of 28.8 C was below the value

Fig. 8.
level.

Wall surface temperatures of three main areas (courtyard, axis and adjoining living) at body

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

1767

recorded in case 1 by 1 C, and recorded at 18:00 h with a time lag of 4 h. Such


results indicate a better interaction between high mass and airflow in case 2 in lowering the courtyards surface temperatures.
5.4. Indoor air temperatures and thermal comfort
The family living area on the west adjoining the courtyard is considered as a space
where the thermal environment is directly influenced by the courtyard due to the
absence of any partitioning between them. Thus, indoor air temperatures (mean of
Te1 and Te2) of this zone were monitored and compared with the courtyard (mean
of Tcy1 and Tcy2) and ambient values in order to investigate the influence of the
courtyard on the indoor environment.
Fig. 9 shows that the indoor air temperatures remained below the courtyard levels
during the daytime in case 1, as well as in case 2. However, during case 2, (except
in late afternoon) indoor air temperatures were relatively below the levels indicated
in case 1 while following parallel with the courtyard temperatures. This behavior
correlates with the pattern of courtyard air temperatures than the ambient. The ambient temperatures remained constant during both periods.
The maximum indoor air temperature recorded in case 2 was 29.8 C at 16:00 h.
This is a clear modification from ambient air temperatures of 30 C and 32.7 C
recorded between the same period of 11:00 and 16:00 h. The indoor relative humidity
during the daytime remained around 70% with an indoor air velocity of 0.4 m/s.
In this study thermal comfort is assessed using a new index PMV, which has
been proposed for warm humid climates. In warm humid tropics as air temperatures
and humidity levels are high, higher mean velocities are essential for promotion of
thermal comfort. However, the model predicting the percentage of people dissatisfied
by Fanger et al. [8] is applicable for a limited range of air temperatures and air
velocities, within 2026 C and 0.050.5 m/s, respectively.
Thus a new index PMV [9], which has been proposed for hot and humid environ-

Fig. 9. Air temperatures of courtyard, internal zone and outdoor with internal wall surface temperature
in family living zone.

1768

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

ment, replacing the operating temperature (To) of Fangers comfort equation with
standard effective temperature (SET) is used in the assessment of thermal comfort.
Activity levels are taken as 1.1 met (Fig. 16) (55 W/m2 for sedentary activity),
with tropical clothing of 0.4 clo. Mean radiant wall surfaces and indoor air velocity at
human body height of the family living area was considered as parameters of PMV.
Case 1 with a PMV vote of 2.2 was found as having the worst thermal conditions
with warm unpleasant indoor environment throughout the daytime. Contrary, the best
thermal conditions with comfortable (PMV vote of 0.2) indoor environment were
found in case 2. More details are published elsewhere [10]. However, the real problem occurs when the relative humidity moves to higher levels during the early
nighttime where indoor air temperatures remain around 2829 C. This necessitates
the need for a slight breeze during night for comfort.
5.5. Airflow due to stack effect at night
Typical ambient weather data for Colombo indicates almost still wind conditions
during the night. Within this context, areas between the courtyard and axis openings
(1 and 2) were monitored for air movement. These measurements indicate the availability of ventilation between 0.20.4 m/s in areas close to the openings during night
hours in case 2, where the building was allowed to ventilate through the envelope
opening in addition to the courtyard. Air movement during the night was not
observed in case 1.
Fig. 10 presents the average air temperatures in the courtyard and areas close to
openings 1 and 2 in the envelope. A variation of pressure distribution due to temperature difference of more than 1 C is visible during the nighttime in case 2. Relatively
lower temperatures at the openings can enhance a slight breeze due to stack effect
only in the night. This is evident with indoor air velocities recorded against almost
still ambient wind conditions in the nighttime in case 2 (Fig. 11). The benefit of
such airflow is important for thermal comfort in the early night hours along with
relatively higher levels of humidity. It was also found that such differences of temperatures needed to activate stack airflow are not visible when the building is ventilated from the courtyard only. The indoor areas show an overheated condition minim-

Fig. 10. Air temperatures in the courtyard and at openings 1 and 2.

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

1769

Fig. 11. Air velocities at openings 1 and 2 compared with ambient and courtyard at body level in case
1 and case 2.

izing the temperature differences with the courtyard and thus minimizing the
potential for stack effect.
5.6. Airflow rates during daytime
Table 2 shows airflow rates calculated for the daytime using measured daytime
air velocities at the openings of each opening composition, together with corresponding air volumes. The volumetric airflow rate is proportionate to the area of openings.
The airflow rate that corresponds with case 2 (the best thermal modification) is
around 1.5 air changes per h. Similar thermal behavior was evident in case 4, in
which the rectangular courtyard is ventilated through three openings including Op1
and Op 2 of the longitudinal axis. Thus, the effective airflow range is from 1.5 to
2.0 air changes h1 [11].
Relatively higher airflow rates and higher air volumes are visible when the building is ventilated through all four openings during the daytime. However, the air
temperature modification is relatively low in this case when compared to the cases
discussed above.

Table 2
Different boundary configurations with corresponding daytime mean airflow rates and air volumes
Opening composition

Air volume (m3/h)

Flow rate, air changes h1

Courtyard (case 1)
Op1, 2 and Cy (case 2)
Op3 and 4 (case 3)
Op1, 2 and 3 (case 4)
Op1, 2, 3 and 4 (case 5)

84
630
535
714
1050

0.2
1.5
1.2
1.7
2.5

Volume of the building, 420m3.

1770

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

6. Computational fluid dynamics (CFD) analysis


Results obtained from the thermal investigation indicate that modification of
indoor air temperatures in a naturally ventilated high mass courtyard building is
affected by the pattern of airflow within the building. The pattern of indoor airflow
was allowed by manually changing the opening composition. The maximum
reduction of courtyard and indoor air temperatures is seen in case 2 where the courtyard is ventilated through openings 1 and 2 found in the building envelope. The
following section investigates the indoor airflow patterns and rates using numerical simulation.
Although there are varieties of turbulence models, the use of k- two equation
turbulence model [12] in predicting airflow properties is commonplace because it
is based on the assumption of isotropic turbulence. Other turbulence models with
anistropic properties are superior to k- model [13], but their utilization as a design
tool in predicting airflow characteristics is not always appropriate due to complexity.
Thus in this study the flow analysis package a-Flow [14] and the standard k- two
equation model was implemented to evaluate the turbulent flow. This software solves
the NavierStokes equations based on finite difference method for three-dimensional
turbulent airflow.
Results taken from the k- model for daytime airflow pattern and rates were compared with the results taken from the field measurements to demonstrate the applicability of the boundary configurations in airflow predictions for design purposes.
6.1. Limitations and purpose
Ventilation occurs when openings are available at points exposed to different levels of air pressure [15]. Temperature difference between indoors and outdoors is the
main force, which generates a pressure gradient for thermal buoyancy influenced
ventilation. Conventional wisdom suggests that for airflow to be induced by thermal
buoyancy there must be two sets of openings at a vertical distance or apart, an
appreciable difference between indoor and outdoor temperatures and it is necessary
that the indoor temperatures are higher than the outdoor temperatures.
Computer fluid dynamics analysis used in this paper investigates daytime airflow
patterns and airflow rates of two thermal situations: a significant cool condition in
the courtyard and a slightly cool condition in the courtyard. In the first condition,
the courtyard and attached indoor spaces show relatively lower temperature than the
outdoor temperatures. In the second condition, the openings found in the external
envelope remain closed during the daytime. Thus, there is least tendency for daytime
ventilation due to thermal buoyancy influence and the CFD simulation aims to model
indoor daytime airflow patterns due to wind driven boundary conditions.
In residential buildings of warm humid climates, pressure differences due to wind
pressure are normally much larger than those due to stack effect and have a dominant
effect on airflow rates [16]. The airflow rates due to buoyancy influence is rather
small under summer conditions, in comparison with the flow which can be induced
in a naturally ventilated building even by a very light wind [15]. Therefore, the

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

1771

simulation aims to calculate daytime airflow rates due to wind-forced ventilation


only, assuming the results from wind pressure boundary conditions give a reasonable
accuracy. In selecting a program for the use in the simulation, a reasonable computation time and cost were also considered as an initial criterion.
CFD simulation in respect to multizone complex buildings (for example the case
study appears in this paper) using either temperature or wind pressure zones as the
boundary conditions is extensively time-consuming and costly. Therefore, the simulation is performed within the limitations of wind pressure boundary conditions.
6.2. Grid system
The grid system used in this study is an orthogonal grid based on the Cartesian
coordinate system (x, y, z coordinate axes), with non-uniform spacing. Application
of the fine grid was limited to the openings, courtyard void and boundary surfaces.
Dimensions of the cube surrounding the centrally placed building model were with
350 m each for x and y coordinate axes (plan area = 122 500 m2) and 60 m for z
coordinate axis. The surrounding environment is approximately 20 times the plan
area and 12 times the height of the building model. Larger surrounding geometry
was utilized to minimize calculation errors.
The grid is composed with 119 129 60 cells in x, y and z coordinate axes,
respectively. Thus there are 92 1060 total cells of which 40% of the cells are incorporated in the building model (88 102 40 cells along x, y and z axes,
respectively). Geometry of the grid is shown in Fig. 12. Minimum and maximum
cell sizes vary in 0.11 to 0.3m, 0.15 to 0.3 and 0.12 to 0.16m in x, y and z coordinate axes, respectively. Dimensions of the grid and cells on x, y, and z coordinate
axes are given in Table 3.
Table 4 shows the governing equations specified for surrounding and wall boundaries [14]. Figs. 13 and 14 show the boundaries of inlet, outlet and sky. With reference to the wind direction (southwest) obtained from the field investigation, the south

Fig. 12.

Orthogonal Cartesian grid with centrally placed building model.

1772

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

Table 3
Grid and cell dimensions
Grid characteristics

Surrounding grid
Total No. of cells
Bldg model cells
Cell dimensions

Coordinate axes
x

350
119
88
Min. 0.11
Max. 0.3

350
129
102
Min. 0.15
Max. 0.3

60
60
40
Min. 0.12
Max. 0.16

Table 4
Boundary conditions
Inlet boundary

Uin(z) = Uh(z / h)1 / 4


Kin(z) = 1 / 2(0.05Uin(z))2
/2
ein(z) = Cmk3in(z)
/L

Outlet boundary
Sky
Wall boundary (log law)

L = h / 2; h = 5.0m; Uh = 1.5m/ s
Free slip
Free slip
U / Ut = 1 / log(Uty /n) + A
K = U2t / Cm1 / 2; e = U2t / y
Ut = (t / r)1 / 2; = 0.4; A = 5.5

Fig. 13. Simulated building model. (a) Plan with wind direction and specified Inlet and outlet boundaries;
(b) Three-dimensional form.

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

1773

Fig. 14. (a) Vertical profile of wind speed at inlet boundary in relation to the building model with
surrounding boundary condition. (b) Wind velocity measurement points of the field investigation.

and west sides act as inlet boundaries. The corresponding wind profile was derived
from power-law formula. Log-law was used to model the wall boundary conditions.
6.3. Building model
Fig. 13 illustrates the plan and 3D form of the simulated building model. The
geometry of the building was realistically modeled, utilizing the dimensions of the
experimented building (refer Fig. 4A). The 3D model incorporates an open central
courtyard, internal layout with two axes perpendicular to each other, floor levels,
gable roofs and main axial openings. The dimensions of the building were, length:
18.6 m, width: 12.2 m and height: 4.8 m. Opening 1 was 4.3 m2 and other axial
openings (Op.2, 3,4) had an area of 3.6 m2 each, the courtyard opening was 18 m2.
The internal spaces were arranged at two ground levels differed by 0.9 m. Thus,
openings 2 and 4 are placed at 0.9 m above the ground level where openings 1 and
3 are found. For simulation purposes, walls, roofs and the ground of the building
were considered as impermeable.
6.3.1. Simulated cases
The building was simulated for indoor airflow behavior under five different boundary-opening compositions as shown in Table 1. This paper presents the airflow rates
of all cases and detailed airflow analysis of the two cases related to the thermal
investigation presented above.
6.4. Natural wind profile
Turbulence characteristics and vertical profile of wind speed are influenced by the
type of terrain and are a function of terrain roughness. The measurements taken with
respect to ambient wind during the field investigation (Fig. 14B) were used in deriv-

1774

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

ing the terrain roughness. Average velocity was recorded at 5 and 10 m heights from
the ground and was 1.5 and 1.8 m/s, respectively. The vertical profile of wind speed
was given by power law according to the following formula (refer Fig. 14A):
Uin (z) / Uh (Z / h)a

(2)

Thus, the investigated site represents a terrain roughness (a) of 0.25 with a turbulent intensity of 5%. In the presence of different boundary conditions from open flat
country to cities, the terrain roughness varies from 0.16 to 0.33, respectively. Terrain
roughness of 0.25 represents an urban characteristic [17]. Daytime wind direction
was observed from the southwest direction at an angle of 45 to the building facade
(Fig. 13A).

7. Results and discussion of the computational analysis


7.1. Comparison of airflow rates
Daytime airflow rates calculated from the results of the field investigation were
compared with CFD. The comparison is made for all cases mentioned in the section
of flow rates from field investigations. Fig. 15 shows the correlation between the
values calculated from the numerical simulation and the field investigations. The
correlation coefficient obtained was 0.96, indicating a reasonable accuracy of CFD.
7.2. Airflow pattern
When the 3D building model is allowed to ventilate through openings 1 and 2 as
found in the building envelope and located with the courtyard on the longitudinal
axis, it was found that the courtyard functions as a low pressure zone. In addition,
other openings in the envelope function as suction zones inducing airflow and thus
optimizing the exposure of high mass walls with incoming air. The air entering from

Fig. 15. Comparison of daytime airflow rates.

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

Fig. 16.

1775

Simulated airflow patterns (horizontal distribution) at human body height (1.1 m).

the openings 1 and 2 travels through the indoor spaces and finally discharges into
the sky through the courtyard (Fig. 17, case 2). It is also seen that a portion of air
travels into the adjoining living space before it discharges through the courtyard.
The results show that the courtyard does not admit any airflow from its sky opening
but acts as an upwind air funnel to discharge the indoor airflow into the sky.
When the 3D building model is allowed to ventilate from the courtyard while
other openings in the envelope remain closed, the courtyard acts as a suction zone
and admits airflow from the free flow above the building model (Fig. 17, case 1).
Detailed results indicate that air entering through the courtyard tries to get discharged
back into the sky through the same opening, creating a vortex within its top section.
A part of the air travels into the adjoining spaces but stagnates as openings in the
envelope are closed. However, intensity of this airflow within the indoor spaces is
lower than that of the previous case.
The above results clearly indicate two different airflow patterns within the
courtyard/building and the differences are correlated to the opening compositions.
A detailed presentation of airflow patterns can be found elsewhere [18].

Fig. 17. Vertical distribution of the simulated airflow patterns, case 1 with A top vortex; case 2 with
an Up wind air funnel.

1776

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

7.3. Wind velocities


Wind velocities through the full height of the courtyard were Wind velocities
through the full height of the courtyard were calculated using mean values of 3D
cells available in the courtyard. The courtyard was simulated with 12 000 such cells
with each level being composed of 300 cells.
An increase of air velocity is seen with the height of the courtyard in case 2 where
the courtyard acts as an upwind air funnel. Fig. 18, shows the calculated full vertical
wind velocity profile compared with the measured wind velocities (daytime mean)
at three levels. Calculated values of 0.32, 0.37 and 0.43 m/s at three levels (1.1 m;
Vc1, 2.5 m; Vc2, 3.5 m; Vc3) where field measurements were taken show a close
agreement with the measured values.
Also, a sudden increase of velocity is visible at 0.9 m where the bottom level of
opening 2 is designed. Such behavior indicates the combined effect of both openings
1 and 2. Calculated velocities at the openings indicate higher values at opening 2
(Fig. 19). Thus, the increase of air velocities at 0.9 m and levels above 2.5 m exhibit
the effect of opening 2 on the air velocities. Furthermore, the calculated and measured velocities at each opening demonstrate a close agreement between the values
(Fig. 19).
The vertical wind velocity profile in case 1 indicates a static behavior up to 3.5
m. The velocities below this level are lower than 0.2 m/s and relatively lower than
in case 2. The increase of velocities (Fig. 18) above 3.5 m indicates the effect of

Fig. 18.
and 2.

Simulated and measured daytime vertical wind velocity profile within the courtyard for case 1

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

Fig. 19.

1777

Simulated and measured daytime wind velocity profile at opening 1 and 2.

vortex type airflow from the courtyards top opening. However, its effect on middle
and lower levels is relatively low.

8. Conclusion
The computational simulation utilizing the standard k- turbulent model with isothermal condition agrees closely with the measurements taken from the field investigation.
The results obtained from the thermal investigation and the computational (CFD)
analysis explores the potential of a courtyard for passive cooling in a single storey
high mass residential building located in a warm humid climate. The effect of the
courtyard for mass-air heat exchange and thus lowering the daytime indoor air temperatures below the corresponding levels of shade ambient temperature is correlated
with the indoor airflow pattern. This correlation indicates that the potential of the
courtyard to act as a passive cooling strategy is a function of the indoor airflow
pattern. Thus, relatively better indoor thermal modification is seen when the courtyard acts as an air funnel discharging indoor air into the sky, rather than the courtyard
acts as a suction zone inducing air from its sky opening, as suggested by conventional knowledge.
Although the airflow rates are proportionate to the opening areas, the levels of
indoor thermal modifications are not proportionate to airflow rates. The maximum
airflow rate does not give the optimum thermal modification. The optimum airflow
rate that corresponds with the best thermal modification of more than 1 C below

1778

I. Rajapaksha et al. / Renewable Energy 28 (2003) 17551778

the ambient level is found between 1.52.0 ACH. Given the ambient wind climate,
and when the measured effective ventilation rate is established, CFD simulations
could be used as an appropriate design tool in deriving the corresponding opening
composition.
The implication of this conclusion on architectural design gives useful guidelines
in designing naturally ventilated high mass residential buildings with internal courtyards in warm humid climates. Thus, the potential of courtyards to act as passive
cooling can be correlated with a building composition in terms of airflow rate and
pattern.

References
[1] Givoni B. Passive and low energy cooling of buildings. New York: Van Nostrand Reinhold, 1994.
[2] Givoni B. Effectiveness of mass and night ventilation in lowering the indoor daytime temperatures
Part I: 1993 experimental periods. Energy and Building 1993;1993(28):2532.
[3] Boutet T. Controlling air movement. USA: McGraw Hill, 1987.
[4] Humphreys MH. Outdoor temperatures and comfort indoors. Building Research and Practice
1978;6(2):92105.
[5] Auliciemes A. In: ANZAAS Conference. Global differences in indoor thermal requirements. 1997.
[6] Szokolay SU. Thermal comfort in warm humid tropics. ANZAScA 1997:711.
[7] Haccoun A, Barcohen A. Thermal properties of building materials: A compilation. Division of Desert
Engineering, Ben-Gurion University of the Negev, Research and Development Authority, 1974.
[8] Fanger PO, Melikov AK, Hanazawa H et al. Turbulence and draft: the turbulence of airflow has a
significant impact on the sensation of draft. ASHRAE Journal 1989;31(7):1823.
[9] Kindangen J. Window and roof configurations for comfort ventilation. Building Research and Information 1997;25(4):21825.
[10] Rajapaksha I, Nagai H, Okumiya M. Indoor airflow behavior for thermal comfort in a courtyard
house in warm humid tropics. In: Proceedings Indoor Air 2002, Monterey, USA. 2002. p. 10727.
[11] Rajapaksha I, Nagai H, Okumiya M. Indoor thermal modification of a ventilated courtyard house
in the tropics. International Journal of Architectural Institute of Japan, Journal of Asian Architecture
and Building Engineering 2002;1(1):8497.
[12] Launder BE, Spalding JL. Mathematical models of turbulence. New York: Academic Press, 1972.
[13] Heiselberg P, Murakami S, Roulet C. Ventilation of large enclosures in buildings, Energy conservation in buildings and community systems: Annex 26. Denmark: Kolding, 1998.
[14] Alfa flow users guide, Fujitsu Corporation, Japan, 1992
[15] Givoni B. Man climate and architecture. London: Applied Science Publishers, 1976.
[16] Aynsley RM. Tropical housing comfort by natural airflow. Building Research and Practice
1980;8(4):24252.
[17] Awbi HB. Ventilation of buildings. London: Spon, 1991.
[18] Rajapaksha I. Passive cooling design strategies for tropical courtyard buildings, Doctoral thesis,
Department of Architecture, Nagoya University, Japan.

Das könnte Ihnen auch gefallen