Applied Thermal Engineering 40 (2012) 358e372

Contents lists available at SciVerse ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Numerical investigations of buoyancy-driven natural ventilation in a simple

atrium building and its effect on the thermal comfort conditions
Shafqat Hussain, Patrick H. Oosthuizen*
Department of Mechanicals and Materials Engineering, Queen’s University, 130 Stuart, Kingston, ON, Canada K7L3N6

a r t i c l e i n f o a b s t r a c t

Article history: In the present study use of solar-assisted buoyancy-driven natural ventilation in a simple atrium building
Received 19 October 2011 is explored numerically with particular emphasis on the thermal comfort conditions in the building.
Accepted 10 February 2012 Initially various geometric configurations of the atrium space were considered in order to investigate
Available online 17 February 2012
airflows and temperature distributions in the building using a validated computational fluid dynamics
(CFD) model. The Reynolds Averaged NaviereStokes (RANS) modelling approach with the SST-keu
Keywords:
turbulence model and the Discrete Transfer Radiation Model (DTRM) was used for the investigations.
Numerical investigations
The steady-state governing equations were solved using a commercial CFD solver FLUENTÓ. From the
Atrium building
Atrium designs
numerical results obtained, it was noted that an atrium space integrated with a solar chimney would be
Comfort analysis a relatively better option to be used in an atrium building. In the geometry selected, the performance of
the building in response to various changes in design parameters was investigated. The produced
airflows and temperature distributions were then used to evaluate indoor thermal comfort conditions in
terms of the thermal comfort indices, i.e. the well-known predicted mean vote (PMV) index, its modi-
fications especially for natural ventilation, predicted percent dissatisfied (PPD) index and Percent
dissatisfied (PD) factor due to draft. It was found that the thermal conditions in the occupied areas of the
building developed as a result of the use of solar-assisted buoyancy-driven ventilation for the particular
values of the design parameters selected are mostly in the comfortable zone. Finally, it is demonstrated
that the proposed methodology leads to reliable thermal comfort predictions, while the effect of various
design variables on the performance of the building is easily recognized.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction accounted for in the final design. Good atrium design will maximize
the natural environment to minimize energy consumption. Atriums
In recent years energy efficiency demands and environmental can be configured in an infinite number of ways, but atrium
concerns have prompted building designers to reconsider natural configurations should be always a reasoned response to the
ventilation in summer, solar heating in winter and the use of climatic and life safety goals. Typical atrium configurations may be
daylighting to save energy consumed by buildings. The advanced completely surrounded by building elements or may be partially
technology of highly glazed atriums is currently being incorporated enclosed. They may be top lit, side lit or a combination of both. The
in the design of large modern buildings in order to take advantage configuration of the atrium will dictate many of the features of
of day lighting, solar heating and buoyancy-driven natural venti- the atrium. The shape and geometry of an atrium depend upon the
lation. Atrium spaces, when properly integrated with the building intended functions of the adjoining occupied portions of the
design, complement the building’s functionality, provide vibrant building. There are several simple and complex basic configurations
space, provide daylighting deep into the building interior and also of an atrium space. The configuration selected by an individual
save energy. There have been several studies that support this view designer is a function of (among other issues) personal taste, life
e.g. see Refs. [1e6]. safety issues, proposed uses of both the atrium and adjoining
Atrium design is complex and creates unique interrelationships spaces, impact the atrium is wished to have climatically.
between various parameters that must be understood and Natural ventilation in atria buildings can be achieved with solar-
driven, buoyancy-induced airflows through the use of an atrium
* Corresponding author. Tel.: þ1 613 533 2573.
space, a solar chimney channel or a combination of both. In the past
E-mail addresses: shafqat55@yahoo.com (S. Hussain), oosthuiz@me.queensu.ca the use of solar chimneys or an atrium space in buildings has been
(P.H. Oosthuizen). examined i.e., see Refs. [7e14,38]. However, the integration of the

1359-4311/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2012.02.025
 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372 359

solar chimney with an atrium space in multi-storied buildings mean thermal sensation vote of the occupants of the naturally
needs further investigation in order to explore the use of solar- ventilated building in a warm climate. The expectancy factor is
assisted buoyancy-driven natural ventilation to reduce energy estimated to vary between 1 and 0.5. If the weather is warm all year
consumption. Currently computational fluid dynamics (CFD) tech- or most of the year, the expectancy factor may be 0.5. In regions
niques are being increasingly employed for predicting building with only a brief period of warm weather during the summer, the
airflows and testing natural ventilation strategies, e.g., see expectancy factor may be 0.9e1. It is 1.0 for the air-conditioned
Refs. [15e18]. With the recent advances in computing power, the buildings. The aforementioned models follow the ASHRAE
process of creating a CFD model and analyzing the results has thermal comfort scale [27]. For draft assessments, the Percentage
become much less labour-intensive, reducing the time and there- Dissatisfied (PD) index [27] may also be used. In order to quantify
fore the cost. Rundle et al. [19] carried out a systematic validation of thermal comfort in naturally ventilated buildings, these models
a commercial CFD code against experimental measurements. They may be integrated with a CFD model. In order to study indoor
noted that CFD can be used successfully to simulate the heat thermal comfort variations with respect to the architectural
transfer and fluid flow in atria geometries and provided recom- elements changes, various geometrical configurations may be
mendations regarding turbulence and relative heat transfer tested.
modelling. Joseph and Sam [20] examined the sensitivity of the The present study was undertaken to consider the various
energy performance of the office buildings in Hong Kong and geometric configurations of the atrium space in a simple prototype
indentified input design parameters and analysed annual building three-storied atrium building and to investigate the effects of the
energy consumptions. It was noted that building simulations are atrium design changes on the buoyancy-driven natural ventilation
useful for assessing thermal response of buildings and data flow rate and temperature distributions inside the building. This
variability. study addresses natural ventilation assessments for both the
The application of computational fluid dynamics (CFD) to physical phenomenon and the effects of the various architectural-
analyse the thermal comfort conditions in atrium buildings has design factors on the thermal comfort conditions in the building.
been studied i.e., see Refs. [21e24]. Standards and fundamentals of For different geometric and solar parameters (input variables)
thermal comfort have been published e.g., see Refs. [25e29]. considered, the airflow pattern is predicted through numerical
Thermal comfort is defined as “the condition of mind that expresses simulation using a CFD model that has been developed and vali-
satisfaction with the thermal environment” in ISO 7730 [29]. The dated in previous studies i.e., see Refs. [30e34]. Outdoor prevailing
most common thermal comfort model used is the traditional Pre- conditions are imposed as inlet boundary conditions. The numer-
dicted Mean Vote (PMV) index model for air-conditioned buildings ical results are then introduced to a special post-processing algo-
[28]. The thermal comfort indices PMV (Predicted Mean Vote) and rithm, which calculates the thermal comfort indices (PMV, PMVNV,
PPD (Predicted Percentage of Dissatisfied) of Fanger [28] are PPD, and PD) in the occupied region of the building. The details of
accepted as ISO 7730 and are calculated by empirical equations. The the study, the methodology, and the results are presented in the
PMV equation uses a steady-state heat balance for the human body. subsequent sections.
This equation was derived empirically and the thermal sensation
vote indicates the personal deviation from the heat balance 2. Atrium design configurations
[3(cold) to þ3(hot); a seven point scale, 0 ¼ neutral (optimum)].
The PPD equation indicates the variance in the thermal sensation of Previously we studied the actual atrium space in the Engi-
the group of persons exposed to same conditions. Dissatisfaction neering Building of Concordia University using a CFD model and the
with the thermal environment, discomfort, was defined for those CFD predictions were validated against the experimental data
who voted cool (2), cold (3), warm (þ2) or hot (þ3). Under available i.e., see Refs. [30e34]. In the present study we have
optimal thermal conditions (PMV ¼ 0) a minimum 5% dissatisfied is simulated a simple three-storied atrium building with various
found assuming identical activity level, clothing and environmental configurations of the atrium space (see Fig. 1) by following the
conditions. The PMV and PPD are calculated from six basic vari- design approach developed by Holford and Hunt [35] analytically
ables: activity, clothing, air temperature, air velocity, mean radiant for a simple multi-storied atrium building. Important dimensions
temperature (MRT) and humidity. The values of the variables for and areas of the building design (Case-1) are given in Table 1.In
the activity (metabolic rate) and clothing (ensemble insulation) are order to utilize the CFD model from the previous studies [30e34],
determined using ASHRAE Fundamentals. the simulated building was assumed to be located and oriented
However, it has been proven that the PMV index is inadequate in exactly as the actual Engineering Building of Concordia University,
the case of naturally ventilated buildings [28], and an optional i.e., 35 degrees west of south in the region of Montreal, Canada with
thermal comfort model, also known as “adaptive comfort standard” the facade glass wall facing southwest. The atrium exhausts are
is implemented in the new revised American Society of Heating, located on the highest point in the atrium and the storey inlets and
Refrigerating and Air-conditioning Engineers (ASHRAE) Standard outlets are located on the side walls of the rooms (see Fig. 1) which
55 [27], together with the PMV index. This model is used to were sized on each floor using the design curves developed by
determine thermal comfort in exclusively naturally ventilated Holford and Hunt [35]. The effective atrium outlet opening area was
spaces. However, it has limitations [28], such as the fact that its selected equivalent to the inlet opening areas for all floors.
applicability is restricted to mean monthly temperatures from 10  C
to 33  C, occupants must be engaged in near sedentary activity, and 3. Numerical solution procedures
only information of global thermal discomfort in large spaces is
provided. Fanger and Toftum [28] introduced the extended PMVNV 3.1. CFD model
comfort model which is more suitable for naturally ventilated
buildings and can be considered consistent with the traditional The airflow patterns and temperature distributions in the
PMV model. The PMVNV index represents an extension of the atrium building are governed by the conservation laws of mass,
traditional PMV model to account for the occupants ‘expectancy momentum and energy. The mathematical model applied includes
factor’ according to their habitat, as well as to the estimated activity, the numerical techniques to solve the continuity, NaviereStokes
i.e. the metabolic rate is reduced under hot and humid conditions (NeS), and energy for 3D, turbulent flow. The general form of the
[28]. The expectancy factor is multiplied with PMV to reach the momentum, pressure, turbulent kinetic energy, turbulent energy
360 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372

Fig. 1. Basic design configurations of an atrium space considered.

dissipation, and enthalpy (temperature for constant heat capacity) about natural ventilation but it requires excessive computational
equations can be expressed in the general form as follows: resources for practical applications. A simplified approach to
overcome such restrictions is the steady-state assumption, as most
 
vðrfÞ vðrui fÞ v vf phenomena take place at almost steady-state conditions over long
þ ¼ Gf þ Sf (1) periods of time. Furthermore, during daytime cycles in real build-
vt vxi xi vxi
ings, temperature changes occur but the steady-state assumption is
where (f) ¼ (u), (v), (w), (P), (k), (3 ), (h, T) , respectively and Sf considered to be valid over long periods of time. Thus, in order to
represents the source terms including pressure terms, thermal recognize the potential time-averaged value of temperatures and
source terms, etc., as appropriate for the variable (f) being solved. velocities, the solution was obtained using the Reynolds averaged
The assumptions made for the problem are: (a) single phase, governing equations for steady, incompressible, three-dimensional
steady-state flow for a Newtonian fluid, (b) heat transfer at the and turbulent flow. In dealing with the buoyancy forces in the
walls by either conduction or radiation is neglected, except glazing momentum equations the Boussinesq approach was adopted, i.e., it
walls (c) steady-state atmospheric conditions, (d) wind velocity is was assumed that the fluid properties are constant except for the
assumed zero and (e) air is considered as a mixture of dry air and density change with temperature which gives rise to the buoyancy
water vapour with constant relative humidity. forces, these being dealt with a linear relation between the density
Because natural ventilation is a phenomenon of random nature change and the temperature change. In addition, the dissipation
due to the constant changes of external weather conditions, any term in the energy equation was neglected due to the low velocities
mathematical model applied for the prediction of natural ventila- involved.
tion should include the dynamic nature of the external conditions. When the first-order fluid parameters are of main concern (e.g.
In applying a CFD method one should ideally use a time-dependent mean temperature and flow rate) rather than turbulent fluctuation
approach, which, however, would require knowledge of the details, two equation eddy-viscosity turbulence models are
time-dependent variations of the boundary conditions used. This generally thought suitable for modelling indoor flows. As noted
technique would provide very detailed and useful information from our previous studies [30e34] for the evaluation of turbulence
models and from literature survey, the SST-keu turbulence model
along with a radiation model DTRM was considered suitable for
Table 1 present CFD simulations. The resultant steady-state governing
Basis areas and dimensions of the prototype building (Case-1). equations were solved using the commercial CFD solver FLUENTÓ.
Areas and dimensions Pressure coupling was treated using the SIMPLE algorithm. The
Atrium height 16.00 m second-order upwind scheme was used to discretize the
Atrium width 5.00 m momentum, turbulent kinetic energy, dissipation rate and energy
Atrium depth 6.00 m conservation equations. The body force weighted scheme was used
Room height 4.00 m
to discretize pressureevelocity coupling. These were solved in
Room width 6.00 m
Room depth 6.00 m a segregated manner. Convergence was considered to have been
Ground floor air supply (net) area 0.40 m2 reached when the energy residual was less than 0.1% and the flow
First floor air supply (net) area 0.51 m2 variables residuals varied by 1% over the last 100 iterations. The
Second floor air supply (net) area 0.82 m2 under relaxation factors for pressure, density, momentum, turbu-
Atrium floor air supply (net) area 0.40 m2
Atrium outlet opening (net) area 2.40 m2
lence kinetic energy, turbulence dissipation rate, turbulent
viscosity, energy 0.3,1,2,0.8,0.8,1,0.9 respectively were used to get
 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372 361

the solution converged. Simulations were performed on a window Table 2

desktop PC with Intel Quad Core Processor EM64T Family 6 Model Solar irradiation and outside conditions for at time of 16:00 on July 10, 2010.

15 Stepping 11 Genuine Intel w3.0 Ghz each processor, 8 GB Ram Sun direction vector x y z
1333 Hz, Operating system is Windows XP Professional 64 bit SP2 0.54 0.84 0.06
and required approximately 24 h for the grid selected.
Sunshine fraction 1
Direct normal solar irradiation (at earth’s surface) 863
3.1.1. Boundary conditions [W/m2]
External climatic conditions affect architectural design of Diffuse solar irradiation e vertical surface [W/m2] 123
Diffuse solar irradiation e horizontal surface 109
naturally ventilated buildings. The internal airflow pattern is the
[W/m2]
result of interaction between the indoor and outdoor environment. Outside heat transfer coefficient [W/m2  C] 7.4
Especially in the case of natural ventilation, outdoor conditions Outside air temperature [ C] 29.7
strongly affect the indoor airflow pattern, and thus affect the
thermal sensation of the occupants. In the present study the focus
was to investigate the use of solar-assisted buoyancy-driven natural
3.2. Radiation model
ventilation, therefore numerical investigations were performed
under fixed steady-state outdoor climatic conditions with ambient
In order to calculate the local solar heat gain in the building
temperature of 29.7  C, relative humidity of 44 at 13:00 h on July
through the glass surfaces, the position of the sun in the sky must
15, 2010 which represents the peak summer time for the location of
be known, and then the intensity of incident total solar radiation
region considered while wind velocity was assumed zero. The
and the thermal radiation exchange between each building enve-
ventilation system of the building was considered based on the
lope element and the surroundings (sky, ground and neighbouring
only buoyancy-driven natural ventilation induced by the heat gains
buildings) must be calculated. This results in a source term that is
from the solar radiation and other heat sources present in the
added to the energy (temperature) equation (Eq. 1). In the present
building. The heat sources were assumed to be located in the centre
work, solar calculator was used for calculation of the incident solar
of each floor to match with the assumptions of the mathematical
radiation. Total solar radiation incident on a sloped surface is:
models developed by Hunt and Holford [35].
In practice airflow inside the connected spaces is mixed by It ¼ ID þ Ids þ Idg (3)
conduction, convection and radiation heat transfer effects. In this
work, conduction and radiation effects were only considered for the
where It is the total incident shortwave radiation, ID is the direct
glazing facade walls while all the other walls were assumed to be
radiation, Ids the sky diffuse radiation, and Idg is the ground
adiabatic to meet the assumptions of the mathematical models
reflected radiation to the surface. Shortwave radiation incident
developed by Hunt and Holford [35]. All boundaries of the domain,
upon a surface is considered to be converted to a thermal source,
except the façade glass surfaces, ventilation openings, heat source,
depending on the local absorption coefficient (a) of the material,
were modelled as no-slip (uj ¼ 0) wall boundaries with zero heat
which equals emissivity (e) for grey surfaces in equilibrium.
flux. The mixed thermal boundary conditions were used for the
To account for long-wave thermal radiation exchange among the
façade glass surfaces. One of the most demanding aspects of heat
surfaces within the building, radiation intensity transport equations
transfer through the envelope is the evaluation of the convection
(RTEs) are solved. FLUENT offers five radiation models; Discrete
heat transfer coefficient. There are many different correlations in
Transfer Radiation Model (DTRM); P-1 Radiation Model; Rosseland
the literature to determine the external heat transfer coefficient for
Radiation Model; Surface to Surface (S2S) Radiation Model; and
buildings. The paper by Palyvos [36] summarizes and outlines the
Discrete Ordinates (DO) Radiation Model. From our previous studies
different correlations found in literature. On the basis of thirty
it was noted that DTRM radiation model is more suitable for the
available linear correlations, he comes up with the following
present study. The main assumption followed in the DTRM model is
“average” correlation to calculate the heat transfer coefficient (hc)
that radiation leaving surface element in a specific range of solid
for windward surfaces:
angles can be approximated by a single ray. It uses a ray-tracing
algorithm to integrate radiant intensity along each ray and is
hc ¼ 4Vw þ 7:4 (2)
where Vw, is wind velocity. With zero wind velocity the value of
external heat transfer coefficient equals to 7.4 W/m2 K was used.
The optical properties of the glazing surfaces (semi transparent)
used in our previous studies [30e34] of an existing atrium building
of Concordia University, in Montreal with solar transmittance of
36% and absorptivity of 17.5% were used. The modelling of the
glazing walls was simplified as a single glazing with effective
thermal conductivity of 0.0626 W/m2 K with a total overall thick-
ness 24 mm. The radiation exchange between the facade and the
sky was also taken into account. The sky temperature was calcu-
lated to be 28.9  C using the Mills [37] correlation, Tsky ¼ [3 skyTout
4 1/4
]
where the emissivity of the sky (3 sky) for the daytime was calculated
to be 0.90 using the relation, 3 sky ¼ 0.727 þ 0.0060Tout with an
ambient temperature Tout of 29.7  C. The heat sources were
modelled as a no-slip wall boundary (2  2 m) located in the centre
of each floor with a constant heat flux equivalent to 132 W/m2 on
the atrium floor and 205 W/m2 on the floors of each room. A
constant relative pressure of 0 Pa was imposed across the room
inlets and the atrium outlet. Fig. 2. Mesh structure (Case-1) for CFD simulations (Mesh2).
362 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372

Fig. 3. Prediction of vertical temperature profile (a) and vertical velocity profile (b) in the centre of rooms for different mesh densities, where 0 m, 4 m and 8 m define the ground,
first and second floor levels respectively.

relatively simple model and increases accuracy by increasing 3.3. Mesh dependency test
number of rays while applies to a wide range of optical thicknesses.
The solar calculator was used to find the sun’s location in the sky Three mesh densities with hexahedral cells were investigated:
with the given inputs of time, date and the global location. Solar Mesh1 (415 k cells), Mesh2 (812 k cells, see Fig. 2) and Mesh3
irradiation and outside conditions at 13:00 on July 15, 2010 are (1235 k cells). The grids used are highly non-uniform characterized
shown in Table 2. by high-nodes density near solid walls and more cells are located

Table 3
Buoyancy-driven natural ventilation volume flow rates on each floor and temperature ( C) values in the centre of each room on left-hand side of the building.

Atrium designs Glazing area (m2) Volume flow rate (m3/s) in each room Temperature ( C) at 1.1 m from floor of each room

Ground floor First floor Second floor Atrium floor Ground floor First floor Second floor Atrium floor
Case-1 80 0.37 0.37 0.35 0.72 32.62 32.36 31.87 32.62
Case-2 154 0.44 0.45 0.42 0.83 33.64 33.43 32.97 33.79
Case-3 108 0.43 0.42 0.40 0.78 33.56 33.48 33.05 33.65
Case-4 90 0.42 0.42 0.39 0.81 33.32 33.09 32.70 33.32
Case-5 118 0.42 0.42 0.40 0.80 33.61 32.39 32.68 33.93
Case-6 286 0.51 0.52 0.49 1.00 35.28 34.79 35.65 34.79
 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372 363

where more velocity and temperature gradients are expected e.g., storey from 0.41 m2 to 0.51 m2. Similarly for Cases-D, E and F, the
near walls, ventilation openings and the area potentially occupied integration of atrium space with a solar chimney was tested and it
by the thermal plume in order to capture the variation in the was noted that this option also affects the ventilation flow rates in
airflows in these areas. The minimum cell size near each wall was the upper storey which was also compensated by increasing the
selected, leading to yþ values 10 in order to apply SST-keu area of inlets from 0.41 m2 to 0.51 m2. To observe the effect of solar
turbulence model. The simulation results shown in this sub- heat flux, the contours of solar heat flux on the facade and walls of
section are based on the conditions for achieving the same venti- the atrium for geometry Case-A are shown in Fig. 4. In order to
lation flow rate in each storey for Case-1 geometry. The vertical observe the airflow patterns and temperature distribution for all of
temperature and velocity profiles in the centre of the left-hand side the cases considered, the qualitative results indicating the
rooms along the height of the building using three mesh densities temperature and velocity contours in the middle plane parallel to
are shown in Fig. 3a and b respectively. It can be seen from Fig. 3, the facade glass surface along the height of the building are shown
there is very small difference less than 1% among the results in Figs. 5 and 6respectively.
obtained using the three meshes. Considering both accuracy and From the results obtained, it was found that all the geometries
computational time, it was decided that Mesh2 (812 k hexahedral considered have shown good performance. However, taking into
cells) is fine enough to accurately predict volume flow rate, airflow account other factors i.e., simplicity, night ventilation, winter
patterns and temperature distributions in the building. It took heating, overheating during summer and heat loss, the geometry of
about 10,000 iterations for the cases simulated with a mesh size of Case-5 integrated with a combination of atrium and solar chimney
812,000 cells to meet the convergence criteria. was selected as a base case for reference for further investigations
to seek the effect of various geometric and solar parameters on the
4. Results and discussion performance of the atrium building.

4.1. Validation of the CFD model 4.3. Parametric study

The CFD model used in this work was validated by comparing Due to the dynamic nature of climatic conditions and the effect
the CFD predictions against the experimental measurements in our of various architectural-design elements, selection of the specifi-
recent studies during the evaluation of various turbulence models cations for energy-efficient buildings is highly uncertain. Thus, any
for the prediction of the airflow and temperature distributions in mathematical approach requires a parametric study to quantify the
atria and validation of numerical modelling of conditions in an effect of various parameters to explore and understand the mech-
atrium space with a hybrid ventilation system in the Engineering anism controlling the performance of the building. To cover all the
building of the Concordia University, Montreal, and an the atrium climatic parameters and perform optimization is beyond the scope
building in Ottawa, Canada i.e., see Refs. [30e34]. A close agree- of the present study. Hence in this study attention is focused only
ment was found between the CFD predictions and experimental on the use of solar energy to induce buoyancy-driven natural
measurements which successfully demonstrate the ability of the ventilation in an atrium building and to investigate the effect of the
CFD model used to accurately predict three-dimensional buoyancy-
driven displacement ventilation flows in multi-storey spaces con-
nected to a common atrium.

4.2. Selection of geometry

Various geometrical configurations of the atrium space of

a simple atrium building designated as Cases-A, B, C, D, E and F.
shown in Fig. 1 were simulated to select a more appropriate
geometry and explore the use of solar-assisted buoyancy-driven
natural ventilation in summer. A series of CFD simulations were run
for all of the cases considered under the same boundary conditions
at 13:00 h on 15th July 2010 with an ambient temperature of
29.7  C i.e., a typical hot day and time during summer in the region
considered. For the comparison of various geometries, the
buoyancy-driven natural ventilation volume flow rates at air inlets
and temperature ( C) values at 1.1 m above the floors in the centre
of each room on the east-end side of the building and in the atrium
space were calculated and the results are shown in Table 3. It is seen
that due to the symmetry of the building, the results for the east
and west-end rooms are approximately the same and with the
increase of the glazing area, the ventilation volume flow rate and
temperature values increase as a result of more solar heat gain.
Cases-A and B have similar geometry but Case-B has more glazing
area and has shown better ventilation flow rates. Similarly Case-F
with higher glazing area has shown higher flow rates and
temperature values. Cases-C, D and E have shown approximately
similar performance. However, it was observed that for Case-C with
a dome-shaped roof and where the location of the exhausts is
lowered by 2 m, the ventilation flow rate in the upper storey is
reduced considerably due to a reduction in the buoyancy forces. Fig. 4. Contours of solar heat flux on the facade and walls of the atrium space for Case-
This was compensated by increasing the inlet size of the upper A geometry.
364 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372

main parameters influencing the performance of the atrium space, ambient temperature of 29.7  C and a relative humidity of 44% in
solar chimney and their solar interactions. The most influential the region of Montreal, Canada. At the air flow inlets, volume flow
parameters can be catagorised as geometrical (e.g. atrium height, rates were calculated and local values of temperatures in the
atrium width, chimney height and width), construction (e.g. type of centre of each occupied area at 1.1 m above each floor were
glass) and solar (e.g. solar intensity, radiation properties and determined in response to each input parameter. The results
shading). The parameters selected were varied one by one using obtained for the geometric and solar parameters considered are
three values of each parameter in a certain range with some vari- shown in Tables 4 and 5 where numbers 1, 2, 3 and 4 represent
ations from the basic design specifications selected to determine the ground floor, the first floor, the second floor and the atrium
the sensitivity of the building’s performance against each param- floor respectively. From the results obtained, it was observed that
eter. Some indicative results are presented in this section regarding for the variation in values of the input parameters in the range
the effect of these parameters on the performance of the building. selected, there is an increase or decrease in the air flow rates from
The effect of each input parameter was examined in terms of four 2 to 10% and a small change in the values of temperatures was
output performance parameters: ventilation volume flow rate at noticed. The increase of glazing area has a positive effect on the
each inlet, temperature value in the centre of each occupied space volume flow rates but also causes an increase in the inside
at 1.1 m above each floor, predicted mean vote (PMV) and predicted temperature values. The height of air flow inlets from floors at
percent dissatisfied (PPD) (thermal comfort indices) for seated 0.6 m has a relatively better effect on the volume flow rates but
activity in the centre of occupied zones. there is not any significant difference in temperature values. The
intensity of the solar heat flux is the motive force for the opera-
4.3.1. Calculation of volume flow rates and temperatures tion of the solar chimney and atrium configuration and is thus the
A series of CFD simulations was run to obtain results for the most determinant factor for the building performance. The air
atmospheric conditions at 13:00 h on 15th July 2010 with an flow rate was found to rise with the increase in solar heat flux. The

Fig. 5. Temperature contours in the middle plane parallel to the facade glass surface for all the cases considered along the height of the building.
 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372 365

Fig. 6. Velocity contours in the middle plane parallel to the façade glass surface for all the cases considered along the height of the building.

glazing type (glazing properties and shading) also affects the clothing (ensemble insulation) were determined from ASHRAE
volume flow rate, increase in the values of the glazing properties Fundamentals [25]. Air temperatures and air velocities were
(transmissivity and absorptivity) was found to increase the determined from CFD simulations. Mean radiant temperatures
induced airflow rates of the buoyancy-driven ventilation and were calculated using the relationship between dry bulb temper-
corresponding changes in the temperature distributions in the ature (DBT) and mean radiant temperature (MRT) [26]. The indices
building. used in the present study are the traditional PMV index and the
extended index PMVNV [28], the last one being more appropriate
4.3.2. Calculation of the thermal comfort indices for the present case of naturally ventilated building, as stated in
The performance indicators for the analysis of the various Introduction. The PMVNV model is based on the expectancy factor,
geometric and solar parameters in terms of the thermal comfort e, which is assumed to depend on the habitat of the occupant. For
indices (i) predicted mean vote (PMV) and (ii) predicted percentage further details about the extended PMV model, the reader is
dissatisfied (PPD) are calculated in this section. The PMV and PPD referred to Fanger and Toftum [28]. In the present calculations,
are calculated from six basic variables: activity, clothing, air the expectancy factor 0.8 was estimated for the region under
temperature, air velocity, mean radiant temperature (MRT) and consideration.
humidity for air-conditioned building [25]. In the present work, the The values of activity, metabolic rate (W/m2) for seated a person,
values of the variables for the activity (metabolic rate) and clothing (clo) and relative humidity were selected 0, 60, 0.5 and 44%
366 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372

Table 4
Effect of geometric variations on the volume flow rates and inside temperature values at 1.1 m from each floor.

Geometric input parameters Output parameters

Parameters Values Volume flow rate (m3/s) Average temperatures ( C) at 1.1 m from floor

1 2 3 4 1 2 3 4
Depth of atrium 6 ma 0.42 0.42 0.40 0.80 33.31 32.39 32.68 33.93
8m 0.44 0.44 0.42 0.82 33.42 32.64 32.86 33.96
10 m 0.47 0.47 0.45 0.85 33.53 33.27 33.06 34.21

Width of atrium 4m 0.40 0.41 0.40 0.74 32.24 33.03 33.1 33.47
5 ma 0.42 0.42 0.40 0.80 33.61 32.39 32.68 33.93
6m 0.45 0.46 0.43 0.78 32.9 33.45 33.34 33.55

Chimney height 2m 0.33 0.32 0.31 0.62 31.6 33.16 33.16 31.69
4 ma 0.42 0.42 0.40 0.80 33.61 32.39 32.68 33.93
6m 0.45 0.47 0.50 0.85 33.38 32.99 31.95 33.45

Chimney width 1m 0.41 0.41 0.39 0.76 33.42 32.15 32.43 33.76
2 ma 0.42 0.42 0.40 0.80 33.61 32.39 32.68 33.93
3m 0.44 0.45 0.43 0.82 33.82 33.46 33.14 34.21

Glazing area 68 m2 0.37 0.36 0.35 0.68 32.71 32.43 32.15 33.17
86 m2 0.40 0.40 0.39 0.76 33.45 32.09 32.51 33.70
118 m2a 0.42 0.42 0.40 0.80 33.61 32.39 32.68 33.93

Location of inlets 0 ma 0.42 0.42 0.40 0.80 33.61 32.39 32.68 33.93
(height from 0.6 m 0.42 0.42 0.40 0.80 33.66 33.47 33.36 33.88
each floor) 1.2 m 0.42 0.41 0.39 0.76 32.8 33.41 34.15 31.57

Ambient temperature 25  C. 1- Ground floor, 2- First floor, 3- Second floor, 4- Atrium floor.
a
Base case.

respectively and air temperatures and velocities (V) in the area of 4.3.3. Effect of various parameters on the performance of the
interest were determined from CFD simulations to calculate the building
values of PMV and PPD for the evaluation of the thermal comfort 4.3.3.1. Effect of atrium depth and width. From the results obtained
conditions in the occupied areas of all the floors in the building shown in Table 4 it is seen that with the increase of atrium depth by
with respect to three values of each parameter considered. The 2 m, the volume flow increases only by 4% and there is a small
values of the PMV and PPD (%) were calculated in the centre of the increase in temperature values at all floors. Similarly an increase in
occupied area at the height of 0.6 m from each floor for seated atrium width by 1 m causes a 4% increase in volume flow rates.
activity using the JAVA applet ISO 7730 computer program [27]. To Table 6 shows the effect of the atrium depth and width on the
calculate the PMVnv and PPDnv for natural ventilation, the expec- PMVnv and PPDnv values in centre of the occupied area in each floor
tancy factor 0.8 was used [28]. The results obtained for the PMVnv of the building for seated activity. It is seen that with the selected
and PPDnv indices in each floor for seated activity are shown in values of the atrium depth and width, the value of PMVnv was
Tables 6 and 7. The effect of each parameter on the building’s found to be in the range 0.20 to 0.50 (corresponding to neutral
performance in terms of PPD (%) is presented graphically in Figs. 7 comfortable) for the ground, first and atrium floors and 0.54
and 8 and is discussed in the next section. to 0.65 (slightly cool, acceptable) for the second floor. The PPDnv

Table 5
Effect of climatic variations on the volume flow rates and inside temperature values at 1.1 m from each floor

Climatic input parameters Output parameters

Parameters Values Volume flow rate (m3/s) Average temperatures ( C) at 1.1 m above floor

1 2 3 4 1 2 3 4
Solar intensity 7:00 h 0.45 0.43 0.41 0.83 33.44 32.69 32.39 34.47
13:00 ha 0.42 0.42 0.40 0.80 33.61 32.39 32.68 33.93
18:00 h 0.35 0.34 0.32 0.61 32.53 31.79 31.74 32.38

Transmissivity 0.16 0.39 0.36 0.33 0.70 32.43 31.63 31.17 32.84
0.36a 0.42 0.42 0.40 0.80 33.61 32.39 32.68 33.93
0.56 0.48 0.48 0.46 0.89 33.63 33.09 33.05 34.20

Absorptivity 0.075 0.42 0.40 0.38 0.76 33.81 32.76 32.51 33.85
0.175a 0.42 0.42 0.40 0.80 33.61 32.39 32.68 33.93
0.375 0.45 0.44 0.42 0.82 33.62 32.87 32.46 35.45

Emissivity 0.4 0.45 0.43 0.40 0.82 32.51 32.75 32.42 34.56
0.8a 0.42 0.42 0.40 0.80 33.61 32.39 32.68 33.93
1.0 0.35 0.34 0.32 0.62 32.55 31.83 31.7 32.53

Shading Blinds fully open 0.42 0.42 0.40 0.80 33.61 32.39 32.68 33.93
Blinds half-open 0.38 0.37 0.36 0.71 32.58 31.89 31.78 33.0
Blinds closed 0.33 0.31 0.30 0.60 31.6 30.73 30.36 32.45

1- Ground floor, 2- First floor, 3- Second floor, 4- Atrium floor.

a
Base case.
 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372 367

Table 6
Calculations of PMV and PPD for geometric parameters at 0.6 m above each floor in the centre of each occupied area for seated activity.

Input parameters Performance parameters

Parameters Values PMVNV (at 0.6 m above floor) (3 cold to þ3 hot) PPDNV (%) (at 0.6 m above floor)

1 2 3 4 1 2 3 4
Depth of atrium 6m 0.32 0.37 0.65 0.20 7.3 8.0 14.1 6.5
8m 0.30 0.34 0.60 0.20 7.1 7.8 14.0 6.5
10 m 0.28 0.31 0.54 0.23 7.0 7.6 14.0 6.7

Width of atrium 4m 0.39 0.42 0.60 0.48 8.0 8.4 12.5 9.8
5m 0.32 0.37 0.65 0.20 7.3 8.0 14.1 6.5
6m 0.50 0.37 0.56 0.37 11 7.9 11 7.9

Chimney height 2m 0.41 0.45 0.56 0.41 8.2 9.5 11 8.2

4m 0.32 0.37 0.65 0.20 7.3 8.0 14.1 6.5
6m 0.35 0.43 0.70 0.20 7.9 8.2 15.5 5.7

Chimney width 1m 0.45 0.48 0.75 0.42 8.4 9.7 16.2 8.4
2m 0.32 0.37 0.65 0.20 7.3 8.0 14.1 6.5
3m 0.24 0.32 0.56 0.30 6.2 7.7 11 7.3

Glazing area 68 m2 0.42 0.50 0.70 0.38 9.2 10.1 15.7 8.2
107 m2 0.37 0.40 0.70 0.28 7.7 9.1 15.5 6.7
118 m2 0.32 0.37 0.65 0.20 7.3 8.0 14.1 6.5

Location of inlets 0m 0.32 0.37 0.65 0.20 7.3 8.0 14.1 6.5
(height from 0.6 m 0.24 0.36 0.52 0.16 6.4 8.0 10.9 5.4
each floor) 1.1 m 0.34 0.46 0.37 0.20 7.8 9.5 8.1 5.7

(%) values indicate that only 6.5e9.8% people are not satisfied with 1 m causes an increase only by 4% at all floors. Table 6 shows the
the thermal conditions on the ground, first and atrium floors but on effect of the chimney height and width on the PMVnv and PPDnv
the second floor 11e14% people are dissatisfied with the thermal values in the centre of the occupied area of each floor of the
conditions. Overall thermal conditions prevailing in the building building for seated activity. It is seen that with the selected values
with buoyancy-driven ventilation within the selected range of of the chimney height and width, the value of PMVnv was found to
atrium depth and width are satisfactory. From these values, it be in the range 0.20 to 0.48 (corresponding to neutral
was noted that relative atrium depth of 6 m and a width 5 m are comfortable) for the ground, first and atrium floors and 0.65 to
feasible for the design specifications of the atrium space under 75(slightly cool, acceptable) for the second floor. The PPDnv (%)
consideration. values indicate that 6.5e9.7% people are not satisfied with the
thermal conditions on the ground, first and atrium floors but on the
4.3.3.2. Effect of chimney height and width. From Table 4 it can be second floor of 16% people are dissatisfied with the thermal
seen that with the increase of chimney height from 2 m to 4 m, the conditions. For the selected values of chimney height and width,
volume flow rate increases by 30% on all floors and from 4 m to 6 m thermal conditions prevailing in the building due to buoyancy-
caused an increase by 7% on the ground, first and atrium floors and driven ventilation are satisfactory. From these values, it was
30% on the second floor while there is a small increase in temper- noted that relative chimney height of 6 m and width of 2 m are
ature values at all floors. Whereas an increase in chimney width by more feasible for the design specifications of the chimney used.

Table 7
Calculations of PMV and PPD for geometric parameters at 0.6 m above each floor in the centre of each occupied area for seated activity.

Input parameters Output parameters

Parameters Values PMVNV (at 0.6 m above floor) PPDNV (at 0.6 m above floor)

1 2 3 4 1 2 3 4
Solar intensity 7:00 h 0.33 0.46 0.68 0.30 7.8 9.5 14.3 7.2
13:00 h 0.32 0.37 0.65 0.20 7.3 8.0 14.1 6.5
18:00 h 0.50 0.64 0.75 0.50 10.6 14.2 16.2 10.1

Transmissivity 0.16 0.46 0.61 0.75 0.40 9.5 12.5 17.2 8.3
0.36 0.32 0.37 0.65 0.20 7.3 8.0 14.1 6.5
0.56 0.31 0.44 0.60 0.24 7.7 9.2 14.0 6.2

Absorptivity 0.075 0.48 0.56 0.45 0.40 5.8 10.6 11.1 8.3
0.175 0.32 0.37 0.65 0.20 7.3 8.0 14.1 6.5
0.375 0.30 0.43 0.67 0.02 7.3 9.2 14.3 5.0

Emissivity 0.4 0.32 0.45 0.67 0.24 7.7 8.4 14.3 6.4
0.8 0.32 0.37 0.65 0.20 7.3 8.0 14.1 6.5
1.0 0.52 0.65 0.76 0.48 10.9 14.2 17.2 10.2

Shading (blinds) Open 0.32 0.37 0.65 0.20 7.3 8.0 14.1 6.5
Half-open 0.56 0.68 0.73 0.43 11.0 14.2 16.2 8.2
Closed 0.61 0.75 0.80 0.47 12.6 16.5 18.5 9.5
368 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372

a PPD (%) values for seated persons in the centre of b PPD (%) values for seated persons in the centre of
each room each room
16 16
Atrium width 4m Atrium depth 6m
Atrium width 5m Atrium depth 8m
14 Atrium width 6m 14
Atrium depth 10m

PPD (%) values

PPD (%) values

12 12

10 10

8 8

6 6

4 4
Ground floor First floor Second floor Atrium floor Ground floor First floor Second floor Atrium floor

c PPD (%) values for seated persons in the centre of d PPD (%) values for seated persons in the centre of
each room
each room

16 18
Chimney height 2m Chimney width 1m
Chimney height 4m Chimney width 2m
Chimney height 6m 16 Chimney width 3m
14

14
PPD (%) values

12 PPD (%) values

12
10
10
8
8

6 6

4 4
Ground floor First floor Second floor Atrium floor Ground floor First floor Second floor Atrium floor

e PPD (%) values for seated persons in the centre of

each room
f PPD (%) values for seated persons in the centre of
each room
16 18
Inlets above floor 0m Glazing area 68 m2
Inlets above floor 0.6m Glazing area 107 m2
14 16
Inlets above floor 1.2m Glazing area 118 m2
PPD (%) values

PPD (%) values

14
12

12
10
10
8
8

6 6

4 4
Ground floor First floor Second floor Atrium floor Ground floor First floor Second floor Atrium floor

Fig. 7. Effect of the atrium width (a) atrium depth (b) chimney width (c) chimney height (d) location of inlets (e) and glazing area (f) on PPD (%) values for seated persons in the
centre of each occupied floor of the building.

4.3.3.3. Effect of glazing area and location of inlets. Table 4 shows only 5.4e9.4% people are not satisfied with the thermal conditions
that with the increase of glazing area in the selected range, the on the ground, first and atrium floors but on the second floor
volume flow rate increases by 4% and there is small increase in 8.1e14% people are dissatisfied with the thermal conditions. For the
temperature values at all floors while location of inlets within the selected values of the glazing area and location of inlets, the
selected range has no any considerable effect on the volume flow thermal conditions prevailing in the building with buoyancy-
rates. Table 6 shows the effect of the glazing area and location of driven ventilation within the selected range were found
inlets on the PMVnv and PPDnv values in centre of the occupied area satisfactory.
of the each floor of the building for seated activity. It is seen that
within the selected values of the glazing area and location of inlets, 4.3.3.4. Effect of the solar intensity and shading. The intensity of the
the value of PMVnv is found to be in the range 0.16 to 0.46 solar heat flux is the motive force for the operation of the solar
(corresponding to neutral comfortable) for the ground, first and chimney and atrium configuration and is thus the most determi-
atrium floors and 0.37 to 0.65 (neutral and slightly cool, nant factor for the building performance. The air flow rate was
acceptable) for the second floor. The PPDnv (%) values indicate that found to rise with the increase in solar heat flux. From Table 5 it is
 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372 369

a PPD (%) values for seated persons in the centre of each b PPD (%) values for seated persons at the centre of
room each room
18
At 7:00 hr
At 13:00 hrs Blinds open Blind half open Blinds closed
16
At 18:00hrs 20
PPD (%) values

14 18

PPD (%) values

12 16
14
10
12

8 10
8
6
6

4 4
Ground floor First floor Second floor Atrium floor Ground floor First floor Second floor Atrium floor

c PPD values for seated persons in centre of each room d PPD (%) values for seated persons at the centre of
each room
18
Emisivity 0.4 18
Emissivity 0.8 Transmissivity 0.16
16 Transmissivity 0.36
Emissivity 1.0 16
Transmissivity 0.56
14

PPD (%) values

PPD (%) values

14

12
12

10
10

8 8

6 6

4 4
Ground floor First floor Second floor Atrium floor Ground floor First floor Second floor Atrium floor

e PPD (%) values for seated persons in the centre of

each room
16
Absorptivity 0.075
Absorptivity 0.175
14 Absorptivity 0.375

12
PPD (%)

10

4
Ground floor First floor Second floor Atrium floor

Fig. 8. Effect of the solar intensity (a) shading (b) emissivity (c) transmissivity (d) and absorptivity (e) on PPD (%) values for seated persons in the centre of each occupied floor of the
building.

seen that the volume flow rate calculated at 13:00 h is decreased by floors but on the second floor 14% people are dissatisfied with the
approximately 20% when calculated at 18:00 h at all floors. Simi- thermal conditions. The PPDnv (%) values calculated at 18:00 h
larly the volume flow decreases by almost 24% when blinds posi- indicate that 10e14% people are not satisfied with the thermal
tioned inside the glass surfaces are closed. Table 7 shows the effect conditions on the ground, first and atrium floors but on the second
of the solar intensity on the PMVnv and PPDnv values in centre of the floor 16% people are dissatisfied with the thermal conditions.
occupied area of the each floor of the building for seated activity. It From Table 5 it is seen that that when blinds are closed, the
is seen that for the solar intensity at 7:00 h and 13:00 h, the value of volume air flow rate is reduced by 20%. Table 7 shows the effect of
PMVnv is found to be in the range 0.20 to 0.32 (corresponding to the blind positions on the PMVnv and PPDnv values in centre of the
neutral comfortable) for the ground, first and atrium floors occupied area of the each floor of the building for seated activity. It
and 0.68 (slightly cool, acceptable) for second floor while at is seen that for the blinds open, value of PMVnv is found to be in the
18:00 h the value of PMVnv is found to be 0.50 to 0.64 (slightly range 0.20 to 0.37 (corresponding to neutral comfortable) for
cool, acceptable) for ground, first and atrium floors and 0.75 ground, first and atrium floors and 0.65 (slightly cool, acceptable)
(slightly cool, acceptable) for the second floor. The PPDnv (%) values for second floor while when blinds are half-open or closed, the
indicate that 6.5e9.5% people are not satisfied with the thermal value of PMVnv is found to be 0.43 to 0.47 (corresponding to
conditions at 7:00 h and 13:00 h, on the ground, first and atrium neutral comfortable) for atrium floor and 0.56 to 0.80 (slightly
370 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372

cool, acceptable) for ground, first and second floors. The PPDnv (%) and 44% respectively with an expectancy factor of 0.8 in the
values indicate that 6.5e14% people are not satisfied with the calculations of PMVNV and PPDNV indices and the results are shown
thermal conditions on all the floors when blinds are open while in Table 8aed. From the results obtained, it is seen that that for
when blinds are half-open or closed, 9% people are not satisfied in selected points in the occupied area on the ground, first and atrium
the atrium space and 11e18% are not satisfied on other floors. floors the value of PMVnv is found to be in the range 0.20 to 0.50
(corresponding to neutral comfortable) for the ground, first and
4.3.3.5. Effect of the glazing properties. Glazing properties have atrium floors and 0.65 to 0.72 (slightly cool, acceptable) for the
remarkable effect on the performance of the building. From Table 5 it second floor. The PPDnv (%) values indicate that only 6.5e11%
was noted that with the increase in the value of transmissivity by people are not satisfied with the thermal conditions on the ground,
0.20, the flow rate increases by approximately 15%. With the increase first and atrium floors but on the second floor 14e16% people are
of absorptivity in the selected range, the volume flow rate increases dissatisfied with the thermal conditions. It was noticed that overall
by 4% while increase of emissivity from 0.8 to 1.0 causes decrease in thermal conditions prevailing in the building due to use of solar-
volume flow rate approximately 16%. Table 7 shows the effect of assisted buoyancy-driven ventilation are in the comfortable zone.
glazing properties on the PMVnv and PPDnv values in centre of the
occupied area of the each floor of the building for seated activity. It is 4.4.2. Discomfort due to draft
seen that for the values transmissivity, absorptivity and emissivity To calculate the risk of thermal discomfort due to draft in the
0.16, 0.075 and 1.0 respectively, the calculated values of PMVnv are building, the thermal comfort empirical equation, Percentage of
found to be in the range 0.40 to 0.75 (slightly cool, acceptable) for Dissatisfied (PD) of Fanger [28] is normally used which is accepted
all the floors. For other values of the glazing properties the values of as ISO 7730 [29].
PMVNV are found to be in the range 0.02 to 0.45 (corresponding to
neutral comfortable) for ground, first and atrium floors while 0.67 PDðPercent DissatisfiedÞ ¼ ð34  Ta ÞðV  0:05Þ0:62
for second floors. Similarly for the transmissivity, absorptivity and
ð0:37V$Tu þ 3:14Þ (4)
emissivity values of 0.16, 0.075 and 1.0 respectively the PPDnv (%)
values are found to be 8e17% for all the floors and for other values of This equation requires the values of three parameters; velocity
the glazing properties PPDNV values are 5e9.2% for ground, first and (V), air temperature (Ta) and turbulence intensity (Tu) to be known.
atrium floors while 14% for second floors. These parameters were evaluated at a height of 0.6 m, which covers
From these values, it was noted that the glazing material the head, neck, and shoulder region of a seated person. Fig. 9 shows
properties; transmissivity, absorptivity and emissivity with values the contours of the three required parameters obtained through
of 0.36, 0.175 and 0.8 respectively used for the reference case are CFD simulations over a horizontal plane at 0.6 m above each floor in
feasible for the design specifications of the glazing material. the occupied areas of the building. Using the values of the velocity,
air temperature and turbulence intensity from the contours shown
4.4. Evaluation of thermal comfort conditions in the building (Case-5) in Fig. 9 Eq. (4) was solved and the resulting contours of Percent
dissatisfied (PD) factor over the region considered are displayed in
4.4.1. Calculation of thermal comfort indices Fig. 10.
The thermal comfort conditions in the building for geometry From Fig. 10 it is seen that near the air flow inlets of the building,
Case-5 were evaluated using the thermal comfort indices. The due to both low temperatures and high velocities, the PD values are
values of the air temperatures and velocities (V) in the area of higher. However, in the whole region of the occupied area of each
interest were determined from CFD simulations. The values of the floor, the PD values lie in the range of 2e7%. The calculated values of
PMVNV and PPDNV (%) were calculated in the occupied area along the PD indicate that due to draft the thermal conditions in the
the horizontal plane at nine points (x ¼ 1, 3, 5 and z ¼ 1, 3, 5) for occupied area of the rooms and atrium space ventilated by solar-
seated activity at 0.6 m above each floor using the JAVA applet assisted buoyancy-driven airflow are in comfortable zone. The PD
ISO 7730 computer program. The values of activity, metabolic rate values show that less than 7% of the occupants are expected to
(W/m2), clothing (clo) and relative humidity were selected 0, 60, 0.5 express dissatisfaction due to draft.

Table 8
Calculations of PMVNV and PPDNV at 0.6 m above each floor in the occupied area of each floor for seated activity.

Z (m) X¼1m X¼3m X¼5m

 
T ( C) V (m/s) PMVNV PPDNV T ( C) V (m/s) PMVNV PPDNV T ( C) V (m/s) PMVNV PPDNV
(a)
1 32.9 0.16 0.50 11 33.0 0.16 0.36 8 33.3 0.14 0.36 8
3 32.6 0.14 0.46 9.5 32.9 0.21 0.34 7.9 33.0 0.11 0.46 9.5
5 32.9 0.10 0.40 8.2 33.2 0.10 0.46 9.5 33.2 0.04 0.44 9.3

(b)
1 32.5 0.10 0.50 11 32.7 0.13 0.43 8.3 33.0 0.07 0.48 9.7
3 32.3 0.11 0.49 10.9 32.7 0.18 0.36 7.9 32.7 0.10 0.51 10.6
5 32.4 0.10 0.50 11 32.6 0.15 0.40 8.2 33.0 0.12 0.42 8.2

(c)
1 31.7 0.04 0.68 14.3 31.8 0.06 0.65 14 32.3 0.02 0.59 13.8
3 31.4 0.04 0.72 16.2 31.8 0.06 0.65 14 31.0 0.19 0.70 16.1
5 31.6 0.04 0.68 14.3 31.8 0.06 0.65 14 32.9 0.02 0.58 13.6

(d)
1 33.9 0.15 0.22 5.8 33.7 0.14 0.27 6.5 34.0 0.04 0.32 7.1
3 33.9 0.16 0.20 5.7 33.3 0.25 0.20 5.7 33.9 0.10 0.34 7.2
5 33.8 0.18 0.16 5.3 32.9 0.32 0.21 5.9 33.6 0.06 0.38 7.9
 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372 371

at the ankles, chest and head for a standing person at the height of
0.1 m, 1.1 and 1.5 were calculated. It was found that temperature
difference is less than 2  C, which corresponds to a PD of less than
5% [29]. This demonstrates that vertical air temperature differences
are moderate and shouldn’t cause major comfort issues for the
occupants.

5. Conclusions

This research establishes a reliable and effective methodology

for evaluating the solar-assisted buoyancy-driven ventilation
performance in an atrium building at the initial design stage.
Various atrium configurations and parametric studies were carried
out to examine the effect of each design parameter on the thermal
conditions in the building. Furthermore the thermal comfort
conditions in the building were evaluated in terms of well-known
thermal comfort indices. The following main conclusions were
drawn from this study.

 It was seen that the design curves [30] are very useful in the
design of an atrium building especially in establishing the sizes
of inlets and outlets in order to have equal ventilation flow
rates in each storey of the atrium building.
 From the analysis of the results obtained for various geometries
of the atrium space in the building, it was found that an atrium
integrated with a solar chimney on the roof resulted in
improved buoyancy-driven natural ventilation.
 The main geometrical and solar input parameters were
considered in terms of four output parameters: airflow rates,
temperature distribution, thermal comfort indices (PMVNV and
PPDNV) for the naturally ventilated building. From the results it
was found that an atrium space integrated with a taller solar
chimney with greater glass solar absorptance and solar trans-
mittance resulted in greater natural ventilation enhancement
which is in agreement with the literature.
 It was noted that changes in atrium depth, width and chimney
width have no significant influence on the buoyancy-driven
natural ventilation.
 The intensity of the solar heat flux is the most determinant
factor for inducing the buoyancy-driven ventilation airflows in
Fig. 9. Temperature top velocity middle turbulence intensity bottom contours along the atrium building. The air flow rate was found to rise with the
the horizontal planes at the height of 1.1 m from each floor in the occupied areas in the increase in solar heat flux. The glazing type (glazing properties
building considered. and shading) also affects the volume flow rate. An increase in
the values of the glazing properties (transmissivity and
absorptivity) was found to cause an increase in the induced
4.4.3. Vertical air temperature difference
airflow rates and corresponding changes in the temperature
Another cause of thermal discomfort is the vertical air temper-
distributions in the building.
ature differences that occupants are experiencing. For all the cases
 This study showed that the parameters selected govern the
considered, the temperature difference between the temperatures
indoor airflow patterns and the indoor comfort levels, e.g., the
results obtained for thermal comfort indices in response to less
solar intensity (at 18:00 h), less chimney height (2 m), less
absorptivity (0.075) and less transmissivity (0.16) for seated
activity rated the indoor environmental conditions at all floors
slightly cool acceptable. In response to other input parameters
the thermal conditions were rated as neutral comfortable
on the ground floor, the first floor and the atrium floor while on
the second floor the thermal conditions were rated as slightly
cool acceptable.
 Thermal comfort conditions in Case-5 selected as reference
geometry in the parametric study were evaluated in detail in
terms of thermal comfort indices (PMVNV, PPDNV and PD)
referring to both skin-bulk flow-net heat balance and air
draughts for seated activity. It was concluded that the thermal
Fig. 10. PD contours along the horizontal planes at the height of 1.1 m from each floor conditions developed in the building as a result of the use of
in the occupied areas in the building considered. buoyancy-driven ventilation are neutral comfortable on the
372 S. Hussain, P.H. Oosthuizen / Applied Thermal Engineering 40 (2012) 358e372

ground floor, the first floor and the atrium floor while on the [13] A. Tahir, Y. Osman, Investigating the potential use of natural ventilation in
new building design, Energy and Buildings 38 (2005) 959e963.
second floor they are slightly cool acceptable.
[14] S. Walker, H. Manchanda, L. Spindler, Building performance: analysis of
 The results of this study can be considered at the initial design naturally ventilated UK Office Building, in: Proceedings of RoomVent (2004)
stage of an energy-efficient atrium building in order to obtain Coimbra, Portugal.
a comfortable indoor thermal environment. Based on this [15] P.C. Liu, H.T. Lin, J.H. Chou, Evaluation of buoyancy-driven ventilation in
atrium buildings using computational fluid dynamics and reduced-scale air
study, future research will include wind speed and other model, Building and Environment 44 (2009) 1970e1979.
climatic factors. [16] M.J. Cook, Y. Ji, G.R. Hun, CFD modeling of natural ventilation: combined
wind and buoyancy forces, International Journal of Ventilation 1 (2003)
169e180.
Overall, the proposed methodology provides a useful procedure [17] Z. Zhai, Q. Chen, Sensitivity analysis and application guides for integrated
to quantify architectural-design effects on indoor thermal comfort. building energy and CFD simulation, Energy and Buildings 38 (9) (2006)
Probably, more accurate results could have been obtained by per- 1060e1068.
[18] M.J. Cook, K.L. Lomas, Buoyancy-driven displacement ventilation flows:
forming either unsteady RANS or LES simulations. However, due to evaluation of two eddy viscosity models for prediction, Building Services
the demand of a relatively large database, more computer intensive Engineering Research and Technology 19 (1998) 15e21.
CFD simulations would lead to prohibitive computational costs for [19] C.A. Rundle, M.F. Lightstone, P.H. Oosthuizen, P. Karava, E. Mouriki, Validation
of computational fluid dynamics simulations for atria geometries, Building
practical engineering applications, and the steady-state approach is
and Environment 46 (2011) 1343e1353.
considered at present adequate for this study. It should be [20] C.L. Joseph, C.H. Sam, Sensitivity analysis of energy performance of office
mentioned that the methodology introduced is general and it can buildings, Building and Environment 31 (1996) 27e39.
[21] A. Atif, D. Claridge, Atrium buildings, thermal performance and climatic
meet thermal comfort estimations produced by either turbulence
factors, ASHRAE Transactions 101 (1) (1995) 454e460.
modelling or experimental measurements. [22] M. Kavgic, D. Mumovic, Z. Stevanovic, A. Young, Analysis of thermal comfort
and indoor air quality in a mechanically ventilated theatre, Energy and
Acknowledgements Buildings 40 (2008) 1334e1343.
[23] R. Li, Adrian, Thermal comfort and environmental modeling in atrium build-
ings, in: PLEA2006 e 23rd Conference on passive and low energy architecture,
This work was supported by the Natural Sciences and Engi- Geneva, Switzerland, 6e8 September 2006.
neering Research Council of Canada (NSERC). [24] K.W.D. Cheong, E. Djunaedy, Y.L. Chua, K.W. Tham, S.C. Sekhar, N.H. Wong,
M.B. Ullah, Thermal comfort of an air-conditioned lecture theatre in the
tropics, Building and Environment 38 (2003) 63e73.
References [25] ASHRAE, ASHRAE Handbook Fundamentals, American Society of Heating,
Refrigerating and Air- Conditioning Engineers, Atlanta, USA, 1997.
[1] A. Moser, F. Off, A. Schalin, X. Yuan, Numerical modeling of heat transfer by [26] URL: www.healthyheating.com.
radiation and convection in an atrium with thermal inertia, ASHRAE Trans- [27] URL: http://www.eat.lth.se/fileadmin/eat/Termisk_miljoe/PMV-PPD.html.
actions 101 (2) (1995) 1136e1143. [28] P.O. Fanger, J. Toftum, Extension of the PMV model to non-air-
[2] N. Takahachi, E. Iwata, M. Funatsu, Simulation and evaluation of the envi- conditioned buildings in warm climates, Energy and Buildings 34 (6)
ronment of a large-scale atrium, in: Proceedings of Room Air Convection and (2002) 533e536.
Ventilation Effectiveness, Tokyo (1992), pp. 511e515. [29] ISO 7730, Moderate Thermal Environments e Determination of the PMV and
[3] B. Zhao, X. Li, Q. Yan, A simplified system for indoor airflow simulation, PPD Indices and Specifications of the Conditions for Thermal Comfort, 2nd ed.,
Building and Environment 38 (2003) 543e552. International Standards Organisation, Geneva, Ref no ISO 7730:1994 (E).
[4] Yiqun Pan, Gang Wu, Fujan Yang, Zhizhong Huang, CFD and daylight simu- [30] S. Hussain, P.H. Oosthuizen, An evaluation of turbulence models for the
lation calibrated with site measurement for waiting hall of Shanghai south numerical study of flow and temperature distribution in atria, in: 18th
railway station, in: Third National Conference of IBPSA Berkeley, California, Annual Conference of the CFD Society of Canada, London Ontario, 17e19
July 30eAugust 1, 2008. May 2010.
[5] P.H. Oosthuizen, M. Lightstone, Numerical analysis of the flow and tempera- [31] S. Hussain, P.H. Oosthuizen, Validation of numerical study of flow and
ture distributions in an atrium, in: Proceedings of the Int. Conference on temperature distribution in an atrium space, in: 19th Annual Conference of
Computational Methods for Energy Engineering and Environment-ICCM3E, the CFD Society of Canada Montreal, Canada, 27e29 April 2011.
Sousse, 20e22 November 2009. [32] S. Hussain, P.H. Oosthuizen, Numerical study of an atrium integrated with
[6] Y. Ji, M.J. Cook, V. Hanby, CFD modelling of natural displacement ventilation in hybrid ventilation system, in: 23rd Canadian Congress of Applied Mechanics,
an enclosure connected to an atrium, Building and Environment 42 (2007) Vancouver, BC, Canada, 5e9 June 2011.
1158e1172. [33] S. Hussain, P.H. Oosthuizen, Validation of numerical modeling of conditions in
[7] Y. Ji, M.J. Cook, Numerical studies of displacement natural ventilation in multi- an atrium space with a hybrid ventilation system, Building and Environment
storey buildings connected to an atrium, Building Services Engineering (2012). doi:10.1016/j.buildenv.2011.12.016.
Research and Technology 28 (3) (2007) 207e222. [34] S. Hussain, P.H. Oosthuizen, A. Kalendar, Evaluation of various turbulence
[8] K.H. Lee, R.K. Strand, Enhancement of natural of natural ventilation in build- models for the prediction of the airflow and temperature distributions in
ings using a thermal chimney, Energy and Buildings 41 (2009) 615e621. atria, Energy and Buildings (2012). doi:10.1016/j.enbuild.2012.01.004.
[9] G. Guohui, Simulation of buoyancy induced flow in open cavities for natural [35] J.M. Holford, G.R. Hunt, Fundamental atrium design for natural ventilation,
ventilation, Energy and Buildings 38 (2006) 410e420. Building and Environment 38 (2003) 409e426.
[10] T. Josef, H. Vitaly, H. Meir, Airflow and heat flux through the vertical opening [36] J.A. Palyvos, A survey of wind convection coefficient correlations for building
of buoyancy-induced naturally ventilated enclosures, Energy and Buildings 40 envelope energy systems, modeling, Applied and Thermal Engineering 28
(2008) 637e646. (2008) 801e810.
[11] J.L. Kevin, Architectural design of an advanced naturally ventilated building [37] A.F. Mills, Heat Transfer, second ed. Prentice Hall, New Jersey, 1999, pp.
form, Energy and Buildings 39 (2007) 166e181. 570e572.
[12] P.F. Linden, P. Cooper, Multiple sources of buoyancy in a naturally ventilated [38] R. Bassiouny, S.A. Koura, Analytical and numerical study of solar chimney use
enclosure, Journal of Fluid Mechanics 311 (1996) 177e192. for room natural ventilation, Energy and Buildings 40 (2008) 865e873.