Sei sulla pagina 1di 15

International Journal of Heat and Mass Transfer 54 (2011) 1933

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer


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

Numerical simulation and comparison with experiment of natural convection


between two oors of a building model via a stairwell
M.R. Mokhtarzadeh-Dehghan
School of Engineering and Design, Brunel University, Uxbridge, Middlesex UB8 3PH, UK

a r t i c l e

i n f o

Article history:
Received 12 May 2010
Accepted 8 September 2010
Available online 2 November 2010
Keywords:
Stairwell
Natural convection
Modelling
LES

a b s t r a c t
The paper presents a numerical study of three-dimensional buoyancy-driven ow in a half-scale model of
a two-oor building model. The model consists of an upper compartment and a lower compartment with
a stairway connecting the two oors. The model forms a closed system, with no inlet or outlet. The ow is
driven by a single heat source placed in the lower compartment. The study is linked closely to a previously published experimental study by the present author, which provided the details of the geometry
and the boundary conditions as well as data for comparison with the present numerical results. The
numerical method is large eddy simulation with the dynamic kinetic energy transport subgrid model.
Radiation exchange is modelled using the discrete ordinates (DO) radiation model. The thermal boundary
conditions on the model walls are set as heat ux. It is shown that the air temperature level is sensitive to
the initial conditions for temperature, but air velocity is unaffected. In order to study this effect further,
with the aid of the ke model, the measured wall temperatures are set as boundary conditions, which
removes the dependency on initial temperature. For the cases studied, comparisons are made between
the measured and computed wall temperatures, wall heat uxes, air temperature and air velocity. There
is a general agreement between the two results.
2010 Elsevier Ltd. All rights reserved.

1. Introduction
There are numerous practical situations involving natural convection in a uid e.g. [14]. Fundamentals of natural convection
ows may be studied by dening very simple problems; a basic
example being a uid-eld cavity of certain height-to-width ratio
with two vertical walls kept at different temperatures. Another approach to study is to investigate the ows in the real environment,
or dening experimental models, full-scale or small-scale, which
represent the real situation closely. A further approach is by mathematical modelling using numerical techniques. The progress in rapid generation of complex geometries and meshing of the ow
domain, together with numerous modelling techniques, has meant
that the use of numerical modelling is now widespread. However,
the question of validity of the results produced by a numerical
model remains. The term numerical or mathematical model referred to here embodies the geometry representing the actual
problem, a set of differential equations for the uid ow and heat
transfer, the boundary conditions, plus all other aspects such as the
grid, modelling of turbulence and details of discretization and solution techniques. It is usual practice to validate a numerical model
in a particular computer code using simple test cases. These test

E-mail address: reza.mokhtarzadeh@brunel.ac.uk


0017-9310/$ - see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijheatmasstransfer.2010.09.067

cases are usually more aligned to the simplifying assumptions employed within the numerical model. As an example, a typical general purpose computer model may rely on using relationships
derived from studying two-dimensional zero pressure gradient
turbulent boundary layer ows in order to model the near-wall
ow, which may depart signicantly from the ows which occur
in practice.
The ow between different zones of a building has been the
subject of study by numerous investigators. Among these are a
small number of studies on buoyancy-driven ows in stairwells
connecting compartments at different levels of a building. The ow
is induced by density differences caused by, for example, a source
of heat such as re. The previous experimental and numerical studies of buoyancy-driven ows, more closely related to the present
work, in terms of practical applications and the numerical techniques adapted, are those of Peppes et al. [5,6], Jiang and Chen
[7], Qin et al. [8], Sun et al. [9] and Ergin [10]. Peppes et al. [5,6]
studied buoyancy-driven ows in real buildings using large eddy
simulation (LES). They studied a naturally ventilated building of
6.3 m height comprising two oors connected via a stairwell. Further studies were carried out in a full-scale real building of height
13.0 m, comprising three oors and connecting stairwells. These
works also included numerical simulation based on Reynolds Averaged NavierStokes (RANS) of the transient ow using RNG ke
model. Jiang and Chen [7] studied natural convection between

20

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

Nomenclature
A
CP
DT
g
Gr
h
I
k
ksgs
Nh, Nu
p
q
qc
qr
qin
qout
Q
Q_ l
r, s
Pr
Re
St
Sij
t
T

area of horizontal opening (m2)


specic heat at constant pressure (kJ/kg K)
differential temperature (K)
gravitational acceleration (m/s2)
Grashof number, Gr = gb DT A h/m2
height of the model (m)
radiation intensity (W/m2)
turbulence energy (m2/s2)
subgrid-scale kinetic energy (m2/s2)
control angles in radiation model (rad)
ltered pressure (N/m2)
heat transfer rate through wall by conduction (W)
heat transfer rate by convection (W)
heat transfer rate by radiation (W)
incoming radiation heat ux (W/m2)
outgoing radiation heat ux (W/m2)
total nominal heat input (W)
wall heat loss (W/m2)
position and direction vectors (m)
Prandtl number
Reynolds number, Re = VA0.5/m
Stanton number, St = Q/qCP TavA(gh)0.5
rate of strain tensor (1/s)
time (s)
air temperature (C)

two compartments connected via a simulated window. One compartment was a test chamber and the other compartment was
the outer laboratory environment, both located on the same level.
Both measurements and numerical simulations were reported. For
the numerical simulation, they used the standard ke model, and
LES. Their results showed better agreement between the predicted
and measured air velocities and temperatures for LES compared
with RANS. Qin et al. [8] carried out a LES of the ow in a two-storey building model with two stairways. The ow was induced by a
re source placed on the lower oor. They validated their techniques by simulating the experimental study of Ergin-zkan
et al. [11] and compared their predicted results with the experimental data. Their simulations indicated the formation of distinct
layers of smoke and air ow in the stairway. Sun et al. [9] used
CFD to predict smoke movement in a six-storey stairwell, induced
as a result of a re in an adjacent compartment. They also used LES,
indicating an increasing trend in the use of this technique in largescale three-dimensional buoyancy-driven ows. The experimental
data of Ergin-zkan et al. [11] was also used [12] for numerical
simulation using the standard ke model. The case presented in
Ref. [12] formed an open case, having two openings, one in the
lower compartment allowing for air to enter at a velocity of
0.7 m/s, and the other in the upper compartment used as an outlet.
The results showed correct ow patterns as well as reasonable
agreement between the measured and computed velocity proles
at the mid-section of the opening between the two oors. The air
temperature, however, were higher than the measured values. Ergin [10] presented a further study of the ow in a one-half scale
stairwell model described previously in Ergin-zkan et al. [11].
The study of Ergin [10] completed a series of works e.g. [11,13]
which were conducted on a model which had a rather simplied
geometry of a two-oor building with a connecting stairwell and
a heat source placed on the lower compartment. Because of the
specic design of this model, the induced ow had a two-dimensional character. The experimental model was then extended to
form a one-half scale model of more realistic geometry [14].

T
Tav
Tw
Twav
u, v, w

u
V
xi
y+
X1

ltered temperature (C)


average temperature in horizontal opening (C)
local wall temperature (C)
wall average temperature (C)
velocity components (m/s)
ltered velocity component (m/s)
air velocity (m/s), volume of computational cell (m3)
coordinate system
non-dimensional distance from wall
Distance along the length of horizontal opening (m)

Greek symbols
b
coefcient of thermal expansion (1/K)
e
energy dissipation rate (m2/s3); emissivity
h, u
polar and azimuthal angle (rad)
l
dynamic viscosity (kg/m s)
m
kinematic viscosity (m2/s)
mt
kinematic eddy viscosity (m2/s)
q
density (kg/m3)
rt
turbulent Prantdl number
r
Stefan-Boltzmann constant (W/m2 K4)
Abbreviations
TBC
temperature boundary condition
HFBC
heat ux boundary condition

The experimental model presented in Ref. [14] consisted of a


lower compartment and an upper compartment connected via a
stairway. A heater placed in the lower compartment induced a
buoyancy-driven three-dimensional airow within the model.
The model was a closed system without an inlet or outlet. In the
absence of ow visualisation, the pattern of the ow was deduced
on the basis of similar previous work [12] and variations in the
measured temperature distributions on the walls of the model. It
was suggested that the experimental model could be used as a test
case for numerical simulation, with the measured heat losses from
the walls of the model as the boundary conditions. Having suggested such a possible test case, provided the motivation for the
author to carry out a numerical simulation of the ow and rene
the model. The contribution of the present paper is in the numerical investigation of the ow and direct comparison with the
experimental data. In doing so, the mathematical model is dened
in such a way that it can be readily used by other investigators.
Previously reported wall heat uxes [14] are supplemented by
additional data for walls temperatures so that both heat ux and
temperature boundary conditions may be implemented, or be used
for comparison with the predicted results. It should be noted that
the case investigated in this paper is different from the case presented in Ref. [12], which, as was noted earlier, was a simpler model having a two-dimensional character, and from the studies
reported in Refs. [59], because of the very different geometry
and ow conditions involved in these studies.

2. Problem under investigation


The set up of the numerical simulation is selected in accordance
with the stairwell model as described in Ref. [14]. A brief description of the experiment is therefore rst provided. Fig. 1 together
with Tables 1 and 2 provide the geometry, dimensions and denition of each wall of the stairwell model, which is the same as the
one described in Ref. [14], except for the omission of the space

21

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

F
Z X1

Ceiling
Back

Outlet
Front

G
O

N
M
Partition

R1

Floor

J
J

K
Inlet
Q

W1
B

A1 W
S

A1
A

B
A2

T1

K
T

Q1
P

Z1

Fig. 1. Schematic diagram of the stairwell model.

Table 1
Dimensions of the walls (mm).
AB
AC
BS
AK, BJ
AP, KL
PQ, LM
MN, LO
OG

1220
2386
3460
1344
940
608
1320
1200

RS, HI
TU
UW
AA1, BW1
BT1
AA2
t

612
580
650
100
180
50
18

Table 2
Denition of the walls. Walls shown in italic are omitted from the numerical model.
AKLP
PLO
POGR1
KGFC
BSIJ
JIED
CFED
APQ1B

Front lower 1
Front lower 2
Front lower 3
Front upper
Back lower
Back upper
Ceiling
Floor 1

QRSQ1
PQRR1
AKJB
KCDJ
RHIS
RHGR1
GFEI
ONHG

Floor 2
Floor 3
Inlet lower
Inlet upper
Outlet lower 1
Outlet lower 2
Outlet upper
Landing

below the stairway and landing, as shown in Table 2. There is a


horizontal opening between the two oors. The heat source was
a typical panel heater (radiator) used to heat rooms in buildings.
There is no inlet or outlet. The experimental stairwell model was
placed in a large room so that heat losses took place freely from
all the walls of the model. The model was raised from the oor level and was placed on a platform structure. Heat inputs of 300, 450,
600 and 750 W by the heater produced three-dimensional recirculating ows with measured velocities less than 0.2 m/s in the horizontal opening. The measured temperature of air was in the range
of 2842 C. The characteristics dimensionless numbers, based on
the dimensions of the horizontal opening and the measured air
velocity and temperature in the horizontal opening were
5997 < Re < 7363, 3.1  104 < St < 7.7  104, 2  10+8 < Gr < 3.68
 10+8. The ow is considered to have a laminar nature in most
part of the space, but transitional or turbulent in the areas in the
vicinity and above the heater. The problem also involves heat
transfer by radiation. The inclusion of the Reynolds number follows
the conclusion of Reynolds [15] and Reynolds et al. [16] that the
stairwell ows, characterised by circulating air ow between two
oors via a throat area, fall near the regime of transition and the
Reynolds number plays a role, for example in relation to the loss
coefcient. Reynolds [15] dened Re based on the reticulating vol-

ume ow rate and the throat area. It was also stated that, for
designing a stairwell model, it was necessary to match the Reynolds number for the model and the prototype. Reynolds et al.
[16] argued that, while Re in the range of 38007000 were obtained in a particular experiment, the Re for the resistance-generating processes is lower, about 1000, due to smaller-scale
processes adjacent to the walls.
The measured heat uxes from the walls of the stairwell, or wall
temperatures, and the heat input from the heater can be used as
the boundary conditions, whereas the measured air temperature
and velocity in the horizontal opening can be used for comparison
with the predicted results. Furthermore, if heat uxes are used as
the boundary conditions, the measured wall temperatures can also
be used for comparison, and vice versa, if temperatures are set as
boundary conditions. The results presented here are for the nominal heat inputs of 300 and 600 W, referred to as Case 1 and Case 2,
respectively. It should be noted that the actual measured total heat
loss was less than these nominal values [14] and therefore the actual rather than the nominal values were used. In doing so, the
space below the stairway and landing (Fig. 1) was omitted in the
present work, because this space formed a separate zone from
the main compartments of the model and had a small contribution
of less than 2.5% to the total heat loss. The exclusion of this space
meant that the associated heat loss had to be subtracted from the
total heat input, in order to have a balance between the heat input
set on the heater and heat output set on the walls. Having the measured heat losses on all surfaces, the heat uxes were calculated
based on the surface areas of each wall. In order to clarify the
method adopted, an example is provided here. In Ref. [14] for a
nominal heat input of 300 W from the heater, the actual measured
total heat loss reported was 275.2 W from which 7.4 W was associated with the space below the stairway. The difference of
267.8 W for the heater surface area of 0.816 m2 gives a heat ux
of 328.2 W/m2. Table 3 provides the values of heat ux set as the
boundary conditions. This table also gives the areas as calculated
by the CFD code. There are small differences with the actual values
used and also with the values given in Ref. [14] because of rounding the gures. An assumption implicit in the values given in Table
3 is the uniformity of heat ux per unit area over the walls as dened in Fig. 1. In the experiment, the measured wall temperatures
corresponded to smaller cells on each wall, but incorporating the
heat uxes from each cell as boundary condition was considered
not useful from a practical point of view, as in practice such detailed information is not normally available. It was also possible
to set the measured wall temperatures as the boundary conditions.

22

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

Table 3
Heat ux set on the stairwell walls.
Wall

Case 1
Q_ W=m2

Case 2
Q_ W=m2

Area (m2)

Inlet

Lower wall
Upper wall

17.81
6.16

34.48
13.82

1.66
1.25

Outlet

Lower wall
Upper wall

8.88
6.54

18.38
12.91

0.83
1.25

Back

Lower wall
Upper wall

12.9
8.16

26.93
15.72

4.71
3.54

Front

Lower wall 1
Lower wall 2
Upper wall

26.95
9.77
6.18

57.35
21.87
15.01

1.28
0.96
3.54

Floor

Floor 1
Floor 2

20.78
10.05

38.36
14.34

1.14
1.54

4.79
328.51

12.96
683.13

4.22
0.82

Ceiling
Heater

These are shown in Table 4. It should be noted that these temperatures, as in the case of heat uxes, are the averaged values over
the walls. As will be discussed later, although some results were
obtained using temperature boundary conditions, this approach
was not adopted in the LES simulations. The reason is that, it
was thought that setting the heat ux as the boundary condition
at least ensured the correct heat loss from each wall.
Ideally, it is desirable to obtain the wall temperatures or heat
uxes as part of the solution. Depending on the problem, it may
be possible to set the boundary conditions on the external walls
[17,18]. The resulting conjugate heat transfer problem would provide the internal wall temperatures. In the present study, the thickness of the walls were such that the same level of variations were
measured on the external walls and therefore the method would
simply add further calculations for heat transfer through the walls
and for radiation from the external walls to the outside atmosphere. A better solution would be to place the boundaries of the
computational domain in the surrounding atmosphere, where the
room temperature can be set. In this case, both internal and external wall temperatures would be calculated as part of the solution.
This method would be more difcult to implement for the present
case due to the complexity of the geometry and extra computing
resources required.
3. Mathematical model
The uid ow within the stairwell model is three dimensional,
incompressible and non-isothermal. The mathematical model consists of the governing equations of continuity, momentum and energy. The uid ow is driven solely by buoyancy forces, generated
Table 4
Average measured wall temperatures at two different heat inputs.
Wall

Case 1
Twav (C)

Case 2
Twav (C)

as a result of the heat transfer from the heater to the air. The calculation of buoyancy forces was achieved by making the air density to
be variable as a function of temperature, using the ideal gas law
with the pressure set at 101325 N/m2. Part of the ow rising from
the immediate vicinity of the heater is expected to reach the partition just above the heater and is diverted towards the interior of the
lower compartment. Other part of the rising ow nds its way to
the opening area on top of the stairway and enters the upper compartment. The ow is expected to be transitional or turbulent in this
early stage of the development, but becomes more laminar when it
slows down, as it reaches the regions farther away from the heat
source. The ow therefore cannot be modelled as laminar or fully
turbulent. The complex nature of the geometry with regard to the
position and orientation of the walls relative to each other and to
the ow is expected to make the modelling of the near-wall ow
difcult. For example, the ow adjacent to the oor is a slowly moving ow and tends to rise, whereas on the ceiling the ow is partly
impinging on the surface and partly moving strongly in parallel to
the wall with a tendency to ow downward in the regions close
to the side walls. The nature of the ow on the front and inlet walls
is also complex. Additionally, it may be said that the large-scale
eddy structures in the early development of the unsteady ow
above the heater, and also in the shear layer in the horizontal opening have an important role in determining the overall velocity and
temperature distributions. The prevailing complex ow conditions
therefore suggest an approach based on large eddy simulation. In
large eddy simulation, large eddies of size greater than a lter size
are resolved whereas the contribution of the smaller eddies are
modelled. This approach is based on the derivation of a set of ltered Navier-Stokes equations which govern the dynamics of large
eddies. Such derivation leads to the appearance of new terms, the
calculation of which requires modelling in order to achieve a closed
system of equations. Through a so-called subgrid-scale model, the
effects of the smaller subgrid-scale eddies on the larger, resolved
eddies are taken into account.
The results presented in this paper were obtained using large
eddy simulation as implemented in the ANSYS12 Fluent CFD code
[19]. The governing equations for the resolved quantities are written as

oui
0;
oxi



oui oui uj
1 op
o
ou
os


m i  ij g i ;

oxj
ot
q oxi oxj oxj
oxj
!
oT oui T
o
m oT
oq
 i;

ot
oxi
oxi Pr oxi
oxi

1
2
3

where sij and qi are unknown terms, given by

sij ui uj  ui uj ;

qi ui T  ui T:

The subgrid-scale models are introduced to calculate these unknown terms as part of the solution of the governing equations.
The subgrid-scale stress sij is obtained using the dynamic kinetic
energy transport (DKET) model [19]. This is an eddy viscosity model, where the kinematic eddy viscosity, mt, is obtained from:

Inlet

Lower wall
Upper wall

30.1
28.0

35.6
31.4

Outlet

Lower wall
Upper wall

27.7
27.7

31.9
31.1

Back

Lower wall
Upper wall

28.2
27.7

32.6
30.9

Front

Lower wall 1
Lower wall 2
Upper wall

31.4
28.1
27.8

38.6
32.1
30.9

p
where Df 3 V and V is the volume the computational cell. The
subgrid-scale kinetic energy ksfg is obtained from:

Floor

Floor 1
Floor 2

30.4
26.8

37.1
30.7

28.1

31.9

 32


oksgs ouj ksgs
oui
ksgs
o mt oksgs
sij
 Ce

oxj rk oxj
ot
oxj
oxj
Df

Ceiling

mt C k k1=2
sgs Df ;

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

23

where

ksgs


1 2
u  u2k :
2 k

The parameters Ck and Ce are obtained dynamically and re is set


to 1.0. The term sij is then obtained from the rate-of-strain tensor
for the resolved scale Sij using:

2
3

sij  ksgs dij 2mt Sij ;

where

Sij



1 oui ouj
:

2 oxj oxi

10

The subgrid-scale heat ux qi is obtained from:

mt oT
;
rt oxi

qi 

11

where the turbulent Prandtl number rt is obtained dynamically.


The boundary conditions for the governing equations are the
zero velocity on the wall and heat ux on each wall as dened in
Table 3. The calculation of these heat uxes were based on the
measurements of internal and external wall temperatures [14]. In
the experimental investigation [14] the wall temperatures were
measured after a long period of time, allowed for the establishment
of a steady state ow condition within the experimental model.
After achieving this condition, the variations in the measured wall
temperatures were small, typically <1%. Therefore, constant values
were set at the walls as boundary conditions.
The positions at which these temperatures were measured are
shown in Fig. 2(a)(f). In the experiments [14], the values of the
heat ux (or rate of heat loss per unit area, thus the negative sign
in Table 3) were obtained by calculating the heat transfer by conduction through the wall. Designating this heat ux as q, it can be
set equal to the sum of the heat ux to the air by convection, qc,
and the radiative heat ux, qr. The value of qr was obtained as part
of the numerical simulation using the discrete ordinates (DO) radiation model as implemented in the FLUENT code [19]. This model
uses the radiative transfer equation for the total radiation intensity, I(r, s), where r and s are the position vector and direction vector, respectively, over the three-dimensional space. Each octant of
the angular space is discretized into Nh  NU control angles, where
N is the number for discretization, and the angles h and U are the
polar and azimuthal angles, respectively. For the 3D space, the
radiative transfer equation is then solved 8 Nh  NU times. In the
present study Nh and NU were set equal to 6.0. The radiation calculation was updated every 10 ow iterations. The walls were taken
as grey diffuse walls. The rate of incoming radiative heat ux, qin is
obtained from [19]:

qin

Iin~
s ~
ndX

12

~
s:~
n>0

where Iin is the radiation intensity, ~


s is the direction vector, ~
n is the
normal vector, and X is the hemispherical solid angle. The outgoing
radiative heat ux is given by

qout 1  ew qin ew rT 4W ;

13

where ew is the wall emissivity. For the medium air, the absorption
and scattering coefcients were set to zero, the refractive index was
set to 1.0. In the experiments [14] the front and back walls were
made of Perspex and the other walls were made of wood, painted
black. The surface of the heater was painted cream. The emissivities
of the wooden, Perspex and heater walls were set as 0.9, 0.9 and 0.7
[10], respectively.
For LES, applying the no-slip condition (zero velocity) at the
wall and, in the absence of any near-wall ow modelling, the wall

Fig. 2. Stairwell walls and measurement positions: (a) front, (b) back, (c) ceiling, (d)
oor, (e) inlet, (f) outlet.

shear stress is obtained from the laminar stressstrain relationship. Similarly, the Fouriers law of conduction is used at the wall.

24

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

For the RANS based on the ke model, a so-called enhanced wall


treatment was adopted [19]. This method resolves the ow in
the near-wall region affected by viscosity but it requires a ne grid
resolution.
The partition between the lower and upper compartment was
treated as a solid wall allowing heat transfer by conduction
through it.
4. Other computational details
The pressure has no equation of its own. The method to derive
the pressure was SIMPLEC. This iterative method makes use of a
so-called pressure correction equation, the derivation of which is
based on the continuity equation. The main parameter obtained
by solving this equation is the pressurecorrection, which when
added to a guessed value of pressure gives an improved value for
the pressure. Starting from an initial guess, the pressure is therefore improved step by step as the iteration process progresses.
In LES, the discretization scheme for the convection terms of the
momentum equations was the bounded central differencing. The
discretization scheme for the thermal energy equation and subgrid
kinetic energy equation was second order upwind scheme. In the
ke model, a second order upwind scheme was used for the equations of momentum, k, and thermal energy, whereas 1st order

scheme was used for e. A rst order upwind was used for the DO
model in both LES and the ke model. As part of the discretization
of the momentum equation and implementation of a pressure
interpolation scheme at a cell face, the body force weighted method was invoked. This method provides improvements over the socalled standard pressure interpolation schemes in buoyancy-driven ows with large body forces.
The LES calculations are transient in nature for which a solution
based on second order implicit formulation was adopted. The time
step was kept constant, equal to 0.02 s. Suitability of the time step
may be checked with reference to the Courant number, which is
the ratio of the time step to the time a particle takes to travel a distance equal to a cell length. Inspection of the converged solution
showed that 94% of all cells had a Courant number less than 1.0,
and 99% less than 3.0. The results presented here were obtained
after 420 s, i.e. 21000 time steps. Taking an air velocity of
0.05 m/s over a circulation path of 7.2 m, the total computation
time corresponds to about 3 ow-through times. In a real life situation, as in experiments [14], for a given heat input on the heater
and initial wall temperatures being at, say, ambient temperature,
it takes many hours to achieve an established ow and certain wall
temperatures. But in the present case, it is expected that, since
experimental wall thermal conditions are already set on the walls,
a much shorter time is needed to achieve an established ow

Fig. 3. Computational grid: (a) front, oor, outlet; (b) stairs and stairwell wall, heater.

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

pattern. The local values of ow parameters such as velocity and


temperature may still vary about an average value. This is due to
the dynamic unsteady nature of the ow in the rising and spreading plume rather than any numerical instability. A further 20 s was
used to obtain statistical results.
The convergence was checked using the criterion that, for each
equation, the scaled sum of all the cells residuals was reduced to
values below a set limit. In the present simulations, the scaled
residuals fell below 3.0  105 for mass conservation, 6.0  107
for the velocity components, 4  108 for the energy equation,
1.5  106 for subgrid turbulence energy and 1.1  106 for the
radiative transfer equation. For the ke solutions, higher values
for the limits were accepted, e.g. about 2.0  104 for mass conservation, 5.0  105 for velocity and 2.0  104 for k and e. The overall balance for total heat transfer rate and radiation heat transfer
rate for LES and ke simulations with heat ux boundary conditions were achieved to a high degree. The worst cases were for
the ke simulations with temperature boundary conditions, but
the overall balance for total heat transfer rate was still less than
0.5%.
Apart from the air density referred to above, other physical
properties involved in the calculations are laminar dynamic viscosity, specic heat at constant pressure and laminar heat conductivity. These properties were allowed to change as a function of air
temperature by entering ve values of the property at ve temperatures over the range of 20100 C, and allowing the computer
code to interpolate.
The initialization of the problem at the start of the iteration process was as follows. The u, v and w-components of the velocity
were set to 0.0, 0.0 and 0.1 m/s, respectively. The initial temperature was chosen based on the range of average temperatures obtained on the walls as seen in Table 4. For the lower heat input
(Case 1) the initial value of temperature was set to 29.85 C
(303 K), which is close to the highest temperature achieved on
the inlet wall. Similarly, for the higher input (Case 2), the initial
temperature was set to 34.85 C. As will be discussed later, the setting of the initial temperature was found to affect the nal temperature level within the model when heat ux boundary conditions
were set.
The strategy in choosing the grid was to adopt a ne grid so that
important regions of the ow are well resolved. In a large eddy
simulation model, the size of the grid cells determine the range
of scales which are resolved and the range of scales which are
modelled. The areas of the model which were considered more
important and had a greater inuence on the ow were the regions
next to the heater and the stairwell walls, and the opening between the two oors. It was aimed to place the rst grid line next
to the wall at about 12 mm from the wall. However, due to the
complexity of the geometry, this was not achieved for all the walls,
such as the stairs. The grid had a total of 1,640,030 cells. The grid is
much ner than the ones used by other investigators in similar
work. Qin et al. [8], in their study of re-induced ow through a
full-scale two-storey building model using large eddy simulation,
had a grid of about 158,000 cells and placed the rst grid lines at
a nominal distance of y+ = 5. Jiang and Chen [7] in their study of
buoyancy-driven ow in a large-scale model of a building used a
grid of 700,000 cells. Because of large dimensions involved in these
studies, the distances between the nodes should have been much
greater than the distances used in the present study. Details of
the present grid are as follows.
The grid was structured comprising rectangular cuboid cells.
Fig. 3(a) shows the cells distributions on the stairwell walls with
the back, ceiling and stairwell wall removed. The distribution of
grid cells on the heater and stairway are shown in Fig. 3(b) and
(c). The nature of the grid with respect to the cell aspect ratio
may be noted from Fig. 3. The areas which appear darker indicate

25

high concentration of the grid lines, causing high aspect ratios.


These were caused by the necessity to use ne grids near the walls,
in particular for the heater surfaces, and due to a relatively small
heater thickness and a small gap between the heater and the front
wall. It should be noted that in most areas concerned, air ows parallel to the surface. It may be said, however, that a more advanced
grid could be set up or a larger number of cells could be used to improve the aspect ratio in these regions, but the results shown later
indicate that the present grid has produced acceptable results comparable with experiment.
The total volume of the computational domain was 8.52 m3.
From 1,640,030 cells, 1,594,590 cells were located in the uid
and 45,440 cells were placed in the solid partition. The total number of grid divisions was 174  95  109 in the x, y and z-directions, respectively. The grid divisions on the heater were
40  65  5. The throat area had a distribution of 60  55. Four
grid cells were placed within the thickness of the partition. The
distribution on a typical stair was 5  55  5. The rst grid line
next to the heater front wall, oor, inlet, back and stairwell walls
were located at a distance of less than 1 mm from the wall. For
the front and ceiling walls and the heater back wall, this distance
was between 1 and 3 mm. For the outlet, stairs and the patrician
walls the distance was about 10 mm and on the stairs about
20 mm. The calculated values of the non-dimensional parameter
y+ on the inlet, back, oor were <1, ceiling <2, front <4, heater <3,
outlet <8, stairs <15, landing <10 and stairwell wall <2. The cell
area ranged from a minimum of 1.11  107 m2 to a maximum
of 6.31  103 m 2. The cell volume was in the range of 9.89
 1011 to 3.4  104
It was necessary to use parallel processing for the LES simulations. Here an example of the processing time needed is provided.

Fig. 4. Histograms of air temperature and air velocity over the computational
domain. The vertical axis shows the percentage of total elements (Case 2).

26

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

Starting the simulation from time t = 0.0, the air motion started
mainly from the heater and then gradually spread to other parts
of the stairwell model as t increased. Convergence was achieved
at each time step. The time, t, taken to achieve convergence at each

time step varied, being much greater at the initial stages of computation than towards the nal time of 420 s. Using a Streamline parallel processing cluster in the Linux 64-bit environment, the
computational grid was partitioned into 16 parts which ran on 4

Fig. 5. Histograms of air temperature on the stairwell walls. The vertical axis shows the percentage of the total elements in the computational domain (Case 2). (a) Inlet, (b)
outlet, (c) front, (d) back, (e) ceiling, (f) oor, (g) heater.

27

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

nodes, each with 2 processors and 2 cores per processor, running at


1.8 GHz. The processing time required for one time step (0.02 s)
around the total time of 420 s for Case 1 was about 50 s.
5. Results and discussion
The results were obtained for two heat inputs set on the heater,
referred to as Case 1 and Case 2 (Table 3). These correspond to the
nominal heat inputs of 300 and 600 W in the experiments [14],
respectively.
5.1. Overall assessment of the results
The main parameters of interest are the air velocity and temperature in the horizontal opening between the two oors, and the
main stairwell wall temperatures, because they can be directly
compared with the experimental data. However, before these results are presented in detail, an overall scrutiny of the results is
rst presented for typical results obtained for LES of Case 2 with
heat ux boundary conditions.
Fig. 4(a) shows the histograms of the air temperature within the
computational domain. The air temperature is in the range of 21
57 C. It will become clear from the results which will be presented
later that the highest temperatures, as expected, relate to the regions near the heater, and the lowest ones to the regions in the
far corners of the stairwell model. Since in Case 2 the initial temperature was set at 34.85 C over the entire ow domain, the results indicate that the simulation allows a readjustment of the
temperature, such that in some areas the air temperature has decreased and in some areas it has increased, compared with the initial temperature. However, the room air temperature (outside of
the stairwell model) in the experiments for this heat input was
about 27 C. The air temperature is expected to be higher than this
value everywhere within the stairwell model. Therefore, the imposition of heat ux on the walls has forced the air temperature to go
below what was attained experimentally in the remote part of the
model.
Fig. 4(b) shows the histogram of predicted air velocity over the
full domain. The maximum calculated air velocity is about
0.88 m/s, but for the vast majority of the ow the velocity is below
0.1 m/s. As will be shown later the predicted values are realistic. A
statistical analysis of velocity uctuations for the period of 420
440 s showed that 75% of all the elements had RMS (root mean
square) values below 0.05 m/s, less than 1% had values greater than
0.15 m/s, and the higher values were about 0.2 m/s.

The histograms of the temperatures on the stairwell walls and


the heater surface are shown in Fig. 5(a)(g), for Case 2. Only a
small number of elements have a temperature below 27 C, which
as was noted earlier was the room temperature outside of the
stairwell model. These occur at far ends on the walls. The gures
are also useful in indicating the degree of variability of temperature (or heat ux) on these walls, which is an indication of whether
setting a uniform boundary condition over a large section of a wall,
in a typical application such as this, is justied. It may be said that
it is only on the ceiling where the degree of uniformity allows this
condition to be reasonably set. On the heater the temperature ranged from about 67 to 127 C. On the front of the heater, high temperatures were located at the top of the heater reaching a
maximum temperature of about 107 C, but most parts of the surface are between 92 and 107 C. In the experiments, the temperature on the centre of the heater front surface was measured at
about 100 C. In the study of Ergin [10] the calculated temperature
was also about 100 C. On the back of the heater, the temperature
showed greater variations, with the upper half being in the range
of 114124 C. It should be noted that this level of agreement
has been obtained with a simplied model of the heater. Although
the total heat ux set on the heater as the boundary condition was
based on the experimental value, the surface area differed from the
actual one. This is because the heater used in the experiments had
a corrugated surface with a round base, compared with the heater
with at surfaces and a uniform width assumed in the simulation.
Reproducing the actual geometry or a similar one would have major implications from the point of view of geometry construction
and gridding. The simplication made is therefore based on taking
a practical approach. It was not envisaged that the ow pattern
would be signicantly affected. The results shown later indicate
that this was in fact the case as the predicted ow pattern agrees
with the pattern obtained experimentally.
The above scrutiny of the range and order of magnitude of the
results indicate that acceptable values are obtained for both air
velocity and wall temperatures, which are generally in comparable
range with the experiment and independent calculations of Ergin
[10].
As was stated above, the total rate of heat transfer per unit area,
q, set on each wall as the boundary condition (Table 3) is the sum
of qc, by convection, and qr, by radiation. Table 5 shows the calculated values of qc and qr. A negative value of qc indicates heat transfer from the air to the wall. Therefore, apart from the front lower
wall 1 and oor 1 which show positive values, all other walls
shown in Table 5 receive heat by convection. The differing results
for these two walls can be attributed to their positions, that is,

Table 5
Heat transfer rate as boundary condition (q) and calculated (qc, qr) heat transfer rates using LES.
Wall

Case 1

Case 2

qr (W) Cal.

qc (W) Cal.

q (W) HFBC

qr (W) Cal.

qc (W) Cal.

q (W) HFBC

Inlet

Lower wall
Upper wall

21.8
3.4

7.8
4.3

29.6
7.7

44.5
6.6

12.8
10.7

57.3
17.3

Outlet

Lower wall
Upper wall

1.2
2.0

6.2
6.2

7.4
8.2

3.6
3.2

11.7
12.9

15.3
16.1

Back

Lower wall
Upper wall

34.6
14.0

26.1
14.9

60.7
28.9

76.4
24.3

50.5
31.4

126.9
55.7

Front

Lower wall 1
Lower wall 2
Upper wall

46.3
3.7
6.7

11.8
5.1
15.2

34.5
8.8
21.9

99.9
9.2
16.3

26.5
10.5
36.9

73.4
19.7
53.2

Floor

Floor 1
Floor 2

26.0
15.4

2.2
0.1

23.8
15.5

49.8
22.7

5.9
0.6

43.9
22.1

2.6
136.6

17.6
130.9

20.2
267.5

12.7
264.4

42.0
291.8

54.7
556.2

Ceiling
Heater

28

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

Fig. 6. Schematic diagram of ow pattern.

being close to the heater and being affected by radiation from it.
This can also be noted from the high values of qr for these walls
and also the wall temperatures on the cells 1, 2 (front wall) and
11, 12 (oor) in Fig. 7a and f, respectively. For front lower 1 and
oor 1, inspection of the local temperature differences, between local wall temperatures and local air temperatures at 1 mm distance
from the wall, showed variations in both values and sign. However,
on the whole, the effect of positive temperature differences was
greater, leading to the positive values of qc as given in Table 5.
For example, comparing the wall temperature on the front lower
1 which is located behind the heater with the air temperature
showed temperature differences in the range of 3.3 and 6.6 C,
with a cumulative average of 1.16 C.
In the arrangement of Ergin [10], the wall behind the heater had
a similar position to the front lower wall 1 in the present study,
and showed the same behaviour. But this is not the case for the
wall in Ergins model which has a similar position to oor 1 in
the present model. In Ergins case, the results showed that heat
was transferred to this part of the oor by convection. Negative
values of qr indicates that all walls, except the heater, also receive
heat transfer by radiation. The results indicate that radiation exchange between the walls plays a major role. The results of Ergin
showed that the main mode of heat transfer to the enclosure is
by convection. The present results show smaller contribution by
convection, which could be attributed to generally lower air velocities achieved here, caused by the larger domain of the present
model. From the total heat input set on the heater, about 51% is
transferred by radiation. This differs from the calculations of Ergin
where a value of 36.3% was reported. The results for the horizontal
walls, namely, oor and ceiling shows that the main mode of heat
transfer from the ceiling is by convection. This may be attributed to

the impingement of the rising plume against the wall and then
diversion of a strong ow along the wall. In the opposite, convection plays only a small role on the horizontal oor because of
low velocities involved. Except for the outlet wall, radiation is
more important in the lower compartment of the stairwell than
the upper compartment. The opposite is true in the upper compartment, where convection is more important.
In the discussion of the experimental results [14] the expected
ow pattern in the stairwell model was speculated, based on the
distribution of the wall temperatures and the air velocity and temperature proles in the opening between the two oors. Fig. 6
shows a sketch of the ow pattern based on the present predicted
results, which is in general agreement with the one proposed in
Ref. [14]. The predicted results, however, indicated a more complex
ow especially in the opening area. The air rising from the main
two surfaces of the heater tends to be diverted more towards the
back wall and the proportion of the air which enters the opening
area is concentrated in the region close to the front wall. The
downow is clearly from the landing area on top of the stairway,
and the downward ow along the stairs is maintained down to
the midway along the stairs, beyond which air tends to rise. Because of the opening next to the stairs, part of the downow in
the stairway spills over into the lower compartment. There is also
ow of air entering the opening between the two oors from the
underneath of the partition. The rising plume of air, upon reaching
the ceiling is diverted towards the inlet and outlet walls forming
two large recirculation zones in the upper compartment. Inspection of the results showed that very slow ow, generally recirculating, was established in the region between the back wall and the
stairway side wall. Complex ow was noticed in the region between the foot of the stairway and the back wall, with air tends

29

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933
80

60
55

70

50
60
Tw (oC)

Tw (oC)

45
40
`

35

50
`

40

30
30

25

lower
(a) front - case
1
0
2

upper
4

8
Cell number

10

12

14

lower
(g) front - case 2

20

16

40

40

36

36

32
`

28

24

8
Cell number

14

16

upper

lower
(h) back - case 2

20
0

10
12
Cell number

14

16

18

20

50

45

45

40

40
Tw (oC)

50

35

10
12
Cell number

14

16

18

20

35

30

30

25

25

upper

lower

upper

lower
(i) inlet - case 2

(c) inlet - case 1


20

20
0

8
Cell number

10

12

14

16

0
40

36

36

32

32

Tw (oC)

40

28

24

lower

24

upper

8
Cell number

10

12

14

16

28

lower

(d) outlet - case 1

upper

(j) outlet - case 2

20

20
0

10

12

Cell number

10

12

Cell number

40

40

36

36

32

32

Tw (oC)

Tw (oC)

12

24

20

28

24

28

24
(k) ceiling - case 2

(e) ceiling - case 1


20

20
0

8
Cell number

10

12

14

16

50

50

45

45

40

40

Tw (oC)

Tw (oC)

10

28

(b) back - case 1

Tw (oC)

upper

lower

Tw (oC)

upper
upper

32

Tw (oC)

Tw (oC)

20

35

30

8
Cell number

10

12

14

16

35

30

25

25
(f) floor - case 1

(l ) floor - case 2

20

20
0

8
Cell number

10

12

14

8
Cell number

10

12

14

Fig. 7. Wall temperatures on the walls of stairwell model. s Experiment, D 300 s, e 360 s, h 420 s LES, x ke.

to ow towards the inlet wall. Downward ow of air along the


walls was observed on the outlet, inlet and back walls due to the
recirculating nature of the ow. In the opening between the two

oors, although the rising air was mainly on the heater side and
the downow was mainly on the landing side on top of the stairway, there were small pockets of rising or falling air elsewhere in

30

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

this region. On the whole, the ow pattern appears to be acceptable and realistic.
5.2. Comparison with experimental data
Having assessed the general validity of the results, direct comparisons with the experimental data is now presented. Fig. 7 provides comparisons of the stairwell walls temperatures with the
predicted values for both heat inputs, at the same positions where
the temperatures were measured. The cell numbers referred to in
Fig. 7 correspond to the numbers used in Fig. 2. The predicted values are given at 300, 360 and 420 s time interval. Full agreement
between the results is not sought here because of the simplications made in the boundary conditions. By setting heat ux as
boundary condition, it is possible to obtain a different temperature
distribution, yet the overall heat balance is satised. Here, therefore, we are contented with achieving agreements between overall
trends and orders of magnitude. Fig. 8 provides comparisons for air
temperature and air velocity. The discussion rst focuses on the
LES results with heat ux boundary conditions. The results obtained using the ke model (also shown in Fig. 7) will be discussed
later.
5.3. Wall temperatures
On the front wall (Fig. 7a,g), signicant over-prediction is
noticeable for cells 1 and 2, where the former being opposite to
the heater and the latter just above it. The differences between
the values for these two cells are 70% and 33% (for Case 2, higher
heat input) and 50% and 15% (for Case 1, lower heat input), respectively. The percentage difference is much lower for all other points
and fall below 3% for cell 11 located just above the opening between the two oors, where the rising plume is located. The variation between two consecutive time steps, i.e. 360 s compared to
300 and 420 s compared to 360 s, is in the range of 1.52.5% for
Case 1 and 4.56% for Case 2.
Fig. 7(b) and (h) shows the results for the back wall. Two distinct levels of agreement are evident. For cells 110, where they

0.20

V (m/s)

0.15
0.10
0.05
0.00

(a)
0

0.3

0.6

0.9
X1 (m)

1.2

1.5

45

T ( oC)

40

5.4. Air temperature and velocity

35
30
25

are located far from the heater and the air ow is relatively slow
and mixing is weak, higher disagreement has resulted, reaching
maximum values of 3.5% and 10% for Case 1 and Case 2, respectively. For cells 1120, which are located in the region where there
is greater air movement and activity, the percentage difference is
lower and reaches a maximum of 2.5%. The variation of temperature between time steps as stated above is in the range of 2%.
Fig. 7(c) and (i) shows the results for the inlet wall. Better agreement is obtained on the part of the wall which is located in the
upper compartment where the maximum difference is 2%. In the
lower compartment the differences reach a maximum of 6% and
11% for Case 1 and Case 2, respectively. The highest percentage difference correspond to cells 7 and 8 which are located in the lower
parts of the wall adjacent to the back wall. The variations in the
wall temperature between time steps are 2% and 4% for Case 1
and Case 2, respectively.
Fig. 7(d) and (j) shows the results for the outlet wall. Greater
differences are noticeable in the lower compartment, in particular
for positions 1 and 4, which are closest to the far end of the oor.
The percentage differences fall between 3.5% to 5% and 9% to
1.5% for the two cases, respectively. In the upper compartment
the values are in the range of 12.5% for both cases. In the lower
compartment, the predicted values tend to be lower than the measured values, whereas the opposite is mostly true for the upper
compartment. The changes in wall temperatures between different
time steps are in the range of 3%.
Fig. 7(e) and (k) shows the results for the ceiling. For Case 1, the
predicted values are lower than the measured values, reaching a
maximum of about 5% for position 8, which is just above the middle of the horizontal opening and is in the region where the rising
plume impinges on the ceiling. For the higher heat input, the variations are between 2.5% and 4.5%, where the latter percentage
value occurs at position 9 which is also located in the same region
as referred to above, but is closer to the front wall. The maximum
variations in the predicted temperatures between two consecutive
time steps are in the range of 25%, which also occur in the same
region.
The results for the oor are shown in Fig. 7(f) and (l). The predicted results are always lower than the measured values. It appears that greater mixing due to a stronger ow takes place in
the experiment than in the simulation. By imposing the experimental values of heat ux as the boundary condition, the real heat
loss has been achieved, but with lower temperature levels in the
uid and on the wall. The variations between time steps are small,
being in the range of 0.71.3%.
On the whole the general trends for the wall temperatures are
well predicted, but as expected quantitative differences exist.
However, except for the front wall, the differences between predicted and measured values are less than 11%. Greater differences
relate to the areas close to the heater, where the effects of both
convection and radiation are strong, as well as areas far from the
main activity, such as the lower end of the outlet wall, where very
low velocities are predicted. Another area where larger differences
are noticeable is on the ceiling above the opening between the two
oors. This is where the rising plume impinges on the ceiling.

(b)
0

0.3

0.6

X1(m)

0.9

1.2

1.5

Fig. 8. Air velocity and temperature in the opening between the two oors. (a)
Velocity, (b) temperature; d, - - -, Case 1; s, , Case 2; d, s Experiment.

Fig. 8 shows a comparison between the measured and predicted


air temperature and velocity in the opening between the two
oors. The experimental proles were obtained in the mid-section
of the opening (along ZZ1 in Fig. 1). For the predicted results, it was
considered likely that the experimental conditions in the mid-section are achieved in the vicinity of the mid-line rather than at the
exact location where the experimental results were obtained.
Therefore the following procedure was followed. First statistical

31

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

5.5. Effect of initial temperature and the case of ke model


In the preliminary present LES calculations, for the case of heat
ux boundary condition (HFBC), it was noted that there was a
dependency of the temperature eld to the initial air temperature
set within the stairwell model, whereas the velocity eld was not
affected. For temperature boundary condition (TBC), however, both

the temperature and velocity elds were independent of the initial


temperature. The initial temperature which is set uniformly everywhere at the start of computation serves as an initial guess and the
temperature eld is free to change and take the correct distribution
in the nal converged solution. With HFBC, however, a converged
solution can be obtained with different level of temperature, set
by the initial air temperature. With TBC, since the walls temperatures are xed, the initial air temperature has not such an inuence
on the nal solution. It should be noted that in both cases, for any
converged solution and at any time step, the heat input from the
heater is balanced by the heat output from the walls, therefore
the internal energy of the system remains constant. Also, the pressure differentials determining the velocity eld are not affected. In
order to demonstrate the effect of initial temperature, additional
simulations were carried out using the ke model.
The wall temperatures were referred to in relation to Fig. 7. This
gure also shows the results obtained using the ke model with
the same initial temperatures, for Case 1 and Case 2, respectively.
The results for the ke model show the same general pattern but
overall they have deviated from the experimental values more
than the LES results, except for the oor (Fig. 7(f) and (l)). Other
exceptions are in the regions far from the centre of ow activity
where LES showed difculty, i.e. far corners in the lower compartment, near the outlet wall (Fig. 7(d) and (j)) and in the region adjacent to the lower inlet and lower back walls (Fig. 7(h) and (i)).
These differences indicate that although the general ow pattern
of ow moving between the two oors are similar, different details
in the ow pattern have been established. This deduction may be
also noticed from the air velocity and air temperature in the horizontal opening as shown in Figs. 9 and 10. These gures also show
the results obtained using HFBC with different initial temperatures. Fig. 9 shows smooth and mostly linear variations of temperature predicted in the case of the ke model. The proles of air
velocity (Fig. 10) show more distinct peaks and troughs, placed
at different locations in the horizontal opening. The effect of initial
temperature is clearly seen in these gures. Changes in the initial
temperature by 2.0 C have produced shifting of the air tempera45

T ( oC)

40
35
30
25

(a)
0

0.3

0.6

0.9

1.2

1.5

0.9
X1(m)

1.2

1.5

X1(m)

45
40
T ( oC)

averages over 20 s from 420 to 440 s were obtained. Next, the proles of velocity and temperature were obtained along ve equallyspaced lines within a band of 0.2 m in the middle of the opening
and then averaged. In other words, the proles shown in Fig. 8
are an average of ve different proles.
For the lower heat input the velocity is mainly underpredicted
(Fig. 8(a)). In the experiments, changes in the air velocity by about
0.05 m/s was possible; this amounts to 30% for a velocity of
0.15 m/s. On the whole, the magnitude and trends obtained by the
simulation can be considered as satisfactory. Fig. 8(b) shows the corresponding air temperature. The main features of the experimental
proles are (i) the region of relatively higher temperatures in the upow region (X1 < 0.6 m), (ii) the region of lower temperatures in the
downow (X1 > 0.9 m), and a region around X1 = 0.75 m where a
peak is noticeable. In the interpretation of the experimental data
[14] the peak was associated with the rising hot plume. In the simulation, the peak value is closer to the start of the opening area.
Fig. 6 shows that, in the simulation, the rising plume was placed
closer to the front wall than the middle of the opening where these
results are plotted. In this respect, therefore, there is a disagreement
with experiment in relation to the location of the plume.
Further realization of what is achievable in this type of ow in
terms of agreement between measured and computed values may
be obtained with reference to the previous work by other investigators. The experimental data in the half-scale stairwell model of
Ergin-zkan et al. [11] was used for comparison with the numerical simulation using the standard ke model by MokhtarzadehDehghan et al. [12]. The computed velocities along a line drawn
perpendicular, and half-way along the stairway, reached maximum
values of about 0.3 m/s. The computed values were generally higher showing differences of about 0.050.1 m/s with the measurement. Jiang and Chen [7] used LES to study natural convection in
a building model with a large opening simulating a window. The
air velocities were mostly less than 0.1 m/s. In their comparisons
of computed and measured velocity proles similar differences to
the present study were reported. Better agreement was obtained
for the temperature distribution. As part of the validation of their
work on re-induced ow using large eddy simulation, Qin et al.
[8] simulated the half-scale stairwell model of Ergin et al. [11].
They compared the simulated velocity distribution with those from
the experiments along two lines, one along a vertical line at the
start of the stairway and another along a horizontal line at the
end of the stairway. The measured values ranged from 0 to
0.25 m/s and the differences of 0.05 to 0.1 m/s between the predicted and measured values were common. No values for temperature distribution were reported. Sun et al. [9] studied smoke
movement in a real six-storey stairwell, driven by re on the
ground oor. They reported average measured velocities at different heights, reaching a maximum value of about 1.0 m/s. Comparison with the predicted results obtained using LES showed correct
trends with differences of mostly 0.10.2 m/s.
Considering the present LES results as whole, it may be concluded that the simulation has been reasonably successful in
obtaining the correct picture of the overall ow pattern, the overall
trends and orders of magnitudes of the wall temperatures, and air
velocity and temperature. The degree of agreement between prediction and experiment has been consistent with those reported
by other investigators in similar works.

35
30
25

(b)
0

0.3

0.6

Fig. 9. Air temperature in the opening between the two oors. (a) Case 1, (b) Case 2.
, LES, HFBC;
, ke, HFBC, initial temperature, (a) 31.85 C, (b) 36.85 C;
,
ke, HFBC, initial temperature, (a) 29.85 C, (b) 34.85 C;
, ke, HFBC, initial
temperature, (a) 27.85 C, (b) 32.85 C;
, ke, TBC, initial temperature, (a) 27.85,
29.85, 31.85 C, (b) 32.85, 34.85, 36.85 C.

32

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

0.30

V (m/s)

0.25
0.20
0.15
0.10
0.05

(a)

0.00
0

0.3

0.6

0.3

0.6
0.9
X1 (m)

X1 (m)

0.9

1.2

1.5

0.30
0.25
V (m/s)

0.20
0.15
0.10
0.05

(b)

0.00
0

1.2

1.5

Fig. 10. Air velocity in the opening between the two oors. (a) Case 1, (b) Case 2; ,
LES, HFBC;
, ke, HFBC, initial temperature;
, ke, TBC, initial temperature, (a) 27.85 C, 29.85 C, 31.85 C, (b) 32.85 C, 34.85 C, 36.85 C.

ture prole by the same amount, whereas the velocity proles remain unaffected (Fig. 10). The results for the stairwell wall temperatures (not presented here) also showed the same shift. Inspecting
the ow pattern showed more orderly ow behaviour than that
obtained using LES. Upon reaching the opening area, the air rose
much closer to the entrance to the horizontal opening and moved
straight up to the ceiling, which was then diverted to form large
clearly established recirculation zones. Such zones were also more
distinguishable and orderly in the lower compartment. The upow
and downow in the horizontal opening were also more distinguishable causing the variation in the prole seen in Fig. 10. On
the whole, it was concluded that the LES ow eld was more realistic than the one obtained using the ke model. The proles with
TBC are also shown in Figs. 9 and 10. As can be seen the proles are

Fig. 11. Heat transfer rate on stairewell walls. (a) Case 1, (b) Case 2;
,
Calculated using temperature boundary condition;
, Measured; On x-axis: 1:
inlet lower; 2: inlet upper; 3: outlet lower; 4: outlet upper; 5: back lower; 6: back
upper; 7: front lower 1; 8: front lower 2; 9: front upper 1; 10: oor 1; 11: oor 2;
12: ceiling.

similar to those with HFBC but noticeable differences exist. In this


case, there is no dependency on initial temperature. This nding
indicates that for natural convection in a closed system, setting
the wall boundary conditions based solely on heat ux has implications for the prediction of the temperature eld. The problem
is that for a practical ow application none of these conditions
are usually known in advance of the solution. When heat ux
boundary conditions were set, measured wall temperatures were
compared with the predicted ones in Fig. 7. Similarly, when temperature boundary conditions were set the measured wall heat
uxes (Table 5) can be compared with the predicted heat uxes given in Table 6 and shown in Fig. 11. The main differences relate to
the walls close to the heater, namely, inlet lower, oor 1 and front
lower 1, arising from the differences in the proportion of radiative
and convection heat transfer rates.

Table 6
Calculated heat transfer rate using temperature boundary conditions with the ke model.
Wall

Case 1
TBC

Case 2
TBC

qr (W)

qc (W)

q (W)

qr (W)

qc (W)

q (W)

Inlet

Lower wall
Upper wall

12.6
2.4

5.2
4.1

17.8
6.5

32.8
6.3

10.5
9.0

43.3
15.3

Outlet

Lower wall
Upper wall

2.1
2.3

4.1
4.4

6.2
6.7

3.4
3.9

8.3
12.0

11.7
15.9

Back

Lower wall
Upper wall

37.4
10.0

23.8
13.2

61.2
23.2

76.7
22.5

44.7
31.5

121.4
54.0

Front

Lower wall 1
Lower wall 2
Upper wall

59.3
4.6
7.5

3.5
5.8
16.1

62.8
10.4
23.6

124.0
9.2
22.7

2.1
10.7
40.0

126.1
19.9
62.7

Floor

Floor 1
Floor 2

11.6
14.8

2.7
4.5

8.9
19.3

21.4
22.5

8.0
7.5

13.4
30.0

3.9
146.2

15.4
121.2

19.3
267.4

5.1
308.3

34.9
247.9

40.0
556.2

Ceiling
Heater

M.R. Mokhtarzadeh-Dehghan / International Journal of Heat and Mass Transfer 54 (2011) 1933

33

6. Conclusions

References

The study aimed to investigate a previously published experimental data by the present author as a possible test case for
numerical simulation of a highly three-dimensional natural convection ow. The results showed a good level of consistency between different simulation results and the experimental data,
which gave supported to the validity of the experimental data for
this purpose, as well as the numerical methods used. The results
indicate that the case under investigation may provide a challenging test case for CFD codes under development. For such a closed
system, the solution may be sensitive to factors such as type of
thermal boundary conditions, near-wall treatment and turbulence
modelling, and therefore different solutions may be obtained. The
inclusion of radiation exchange between the walls and ability to
calculate the proportion of heat transfer by convection for such a
complex geometry may also prove challenging.
Because of the nature of the ow LES provided a better option
than the mostly common approach utilizing the ke model, but
at the expense of signicant demand on computing resources.
The results obtained using LES were more realistic, but the results
obtained using the ke model still proved to be useful and,
although displaying somewhat different ow details, it produced
the overall features and predicted correct order of magnitudes.
This was in spite of the use of near-wall modelling.
It was found that the use of heat ux boundary conditions on
the walls resulted in dependency of the temperature eld on the
initial air temperature, but the velocity eld was independent of
this parameter. This aspect of the simulation requires further
investigation in order to reduce this effect, as it has practical implications. Setting temperature boundary conditions on the walls produced the same temperature eld and thus such a dependency was
not present.

[1] P.F. Linden, The uid mechanics of natural ventilation, Annu. Rev. Fluid Mech.
31 (1999) 201238.
[2] M. Fordham, Natural ventilation, Renew. Energ. 19 (2000) 1737.
[3] H.B. Awbi, A. Hatton, Mixed convection from heated room surfaces, Energ.
Build. 32 (2000) 153166.
[4] G. Ziskind, V. Dubovsky, R. Letan, Ventilation by natural convection of a onestory building, Energ. Build. 34 (2002) 91102.
[5] A.A. Peppes, M. Santamouris, D.N. Asimakopoulos, Buoyancy-driven ow
through a stairwell, Build. Environ. 36 (2001) 167180.
[6] A.A. Peppes, M. Santamouris, D.N. Asimakopoulos, Experimental and numerical
study of buoyancy-driven stairwell ow in a three storey building, Build.
Environ. 37 (2002) 497506.
[7] Y. Jiang, Q. Chen, Buoyancy-driven single-sided natural ventilation in buildings
with large openings 46 (2003) 973988.
[8] T.X. Qin, Y.C. Guo, C.K. Chan, K.S. Lau, W.Y. Lin, Numerical simulation
of re-induced ow through a stairwell, Build. Environ. 40 (2005)
183194.
[9] X.Q. Sun, L.H. Hu, Y.Z. Li, R. Huo, W.K. Chow, N.K. Fong, Gigi C.H. Lui, K.Y. Li,
Studies on smoke movement in stairwell induced by an adjacent compartment
re, Appl. Therm. Eng. 29 (13) (2009) 27572765.
[10] S. Ergin, Surface radiation with conduction and natural convection in a twooor enclosure, Energ. Build. 32 (2000) 5770.
[11] S. Ergin-zkan, M.R. Mokhtarzadeh-Dehghan, A.J. Reynolds, Experimental
study of natural convection between two compartments of a stairwell, Int. J.
Heat Mass Transfer 38 (12) (1995) 21592168.
[12] M.R. Mokhtarzadeh-Dehghan, S. Ergin-zkan, A.J. Reynolds, Natural
convection between two compartments of a stairwell-numerical
prediction and comparison with experiment, Numer. Heat Transfer 27
(1995) 117.
[13] A.S. Zohrabian, M.R. Mokhtarzadeh-Dehghan, A.J. Reynolds, B.S.T. Marriot, An
experimental study of buoyancy-driven ow in a half-scale stairwell model,
Build. Environ. 24 (1989) 141148.
[14] M.R. Mokhtarzadeh-Dehghan, Natural convection between two oors of a
building via a horizontal opening-measurements in a one-half scale model, Int.
J. Heat Mass Transfer 50 (2007) 31413151.
[15] A.J. Reynolds, The scaling of ows of energy and mass through stairwells,
Build. Environ. 21 (34) (1986) 149153.
[16] A.J. Reynolds, M.R. Mokhtarzadeh-Dehghan, A.S. Zohrabian, The modelling of
stairwell ows, Build. Environ. 23 (1) (1988) 6366.
[17] A. Liaqat, A.C. Baytas, Conjugate natural convection in a square enclosure
containing volumetric sources, Int. J. Heat Mass Transfer 44 (2001) 3273
3328.
[18] L. Valencia, J. Pallares, I. Cuesta, F.X. Grau, Turbulent RayleighBnard
convection of water in cubical cavities: a numerical and experimental study,
Int. J. Heat Mass Transfer 44 (2007) 32033215.
[19] ANSYS FLUENT 12.0 Theory Guide, release 12.0 @ ANSYS. Inc. 200901-23.

Acknowledgement
The simulations reported in this paper were carried out using
ANSYS FLUENT ow modelling software under an academic license.

Potrebbero piacerti anche