Available online at www.sciencedirect.

com

ScienceDirect
Solar Energy 99 (2014) 110
www.elsevier.com/locate/solener

Fluid dynamic and heat transfer parameters in an urban canyon

S. Bottillo , A. De Lieto Vollaro, G. Galli, A. Vallati
Sapienza University of Rome DIAEE Via Eudossiana 18, 00184 Rome, Italy
Received 28 June 2013; received in revised form 22 October 2013; accepted 24 October 2013
Available online 22 November 2013
Communicated by: Associate Editor Matheos Santamouris

Abstract
A microclimatic analysis in a typical urban conguration, has been carried out. Using a CFD method, a N-S oriented urban street
canyon, with a given H/W ratio, has been examined. The standard ke turbulence model has been used to simulate a three-dimensional
ow eld and to calculate the thermo-uid dynamics parameters that characterize the street canyon. The aim of this study is to investigate the eect of solar radiation on the ow eld and thermal parameters within the canyon. A comparison between transient and stationary simulations has been performed to evaluate the importance of considering the thermal inertia eects in an urban street canyon
study. The dynamic characteristics of the 3D ow in the canyon have been compared with other numerical simulations and experimental
results. Furthermore a thermo-uid dynamic analysis of natural convection eects on the heat transfer coecient and turbulent kinetic
energy, has been carried out.
2013 Elsevier Ltd. All rights reserved.
Keywords: Urban microclimate; Urban canyon; CFD; Solar radiation

1. Introduction
The landscape of dense urban areas can be described by
units of street delimited by two continuous rows of buildings to form a canyon. This geometry is often described
by a single parameter, the canyon aspect ratio (H/W),
which is dened as the ratio of the building height (H) to
the width between buildings (W). As to the incoming solar
radiation and the heating of canyon surfaces, the orientation of the canyon relative to the solar path is also critical
in determining the timing and extent to which surfaces
receive direct sunlight. Several studies have been performed
on dierent street canyons (Takebayashi and Moriyama,
2012; Bozonnet et al., 2005; Lei et al., 2012; Xie et al.,
2007). An experimental validation of a 3D numerical simulation has been performed by Assimakopoulos et al.
(2006); they performed tests using a numerical model on
Corresponding author. Tel.: +39 06 44 58 56 64; fax: +39 06 48 80 120.

E-mail address: simone.bottillo@uniroma1.it (S. Bottillo).

0038-092X/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.solener.2013.10.031

a grid of buildings. By numerical tests characterized by a

2D spatial domain and with assigned surfaces temperatures, Lei et al. (2012) studied the impact of ground heating
on the ow elds in a street canyons, Xie et al. (2007) studied the eects of facades and ground heating on the pollutant dispersion, Saneinejad et al. (2011) investigated on the
heat transfer coecient in a street canyon, simulated as a
cavity, using the low-Reynolds number modeling; they
found a strong inuence of thermal eect on the ow eld.
Allegrini et al. (2012a) analyzed the convective heat transfer at building facades in several urban congurations,
using the adaptive wall function approach developed by
Defraeye et al. (2011) and Allegrini et al. (2012b); they concluded that the AWF provides more accurate heat transfer
analysis in urban CFD studies. Oerle et al. (2007) used
wind and temperature measurements to examine the thermal structure within a street canyon. They found that
buoyancy eects were not seen to have as large an impact
on the measured ow eld as has been shown in the numerical experiments. Kovar-Panskus et al. (2002) performed a

S. Bottillo et al. / Solar Energy 99 (2014) 110

wind tunnel study of the inuence of wall-heating on the

ow regime in a simulated canyon. They found little evidence of thermal eects, except in a very thin layer near
the heated wall, as Louka et al. (2002) noted from an experimental campaign in Nantes, France. Most of numerical
studies have been performed on innitely long street canyons and the prevailing wind direction has been assumed
perpendicular to them, so that the spatial domain has been
simplied from 3-D to 2-D. In this study, simulations have
been performed on a 3-D domain, investigating the impacts
of solar heating and ambient wind speed on the ow elds
and surfaces temperatures in a street canyon with a xed
H/W and L/W ratios. It has been chosen to study a street
canyon isolated from the urban environment, as Blocken
et al. (2007) and Allegrini et al. (2012a) did, to evaluate
how the buildings conguration aects a spatially homogeneous dynamic and thermal eld. Using the commercial
CFD code Ansys-Fluent, a series of numerical tests were
performed to point out the dierences between transient
and steady simulations and to investigate the eects of thermal eld on the surface temperature on the heat transfer
coecient and air velocity and turbulence.
2. CFD numerical model
In this section, the parameters of the computational
model and the boundary conditions are outlined.
The simulations have been performed with the commercial CFD code Ansys Fluent 14.0, 3D double precision,
pressure based version and the steady RANS equations
have been solved in combination with the standard ke
model. The governing equations can be expressed as
follows:
Momentum equation:
uj

p l @ 2 ui
@ui
1 @
@



u0i u0j fi :
q @xi q @xi @xj @xj
@xj

Continuity equation:
@ui
0:
@xi
Heat conservation equation:


@T
@
@T
ui

KT
0;
@xi @xi
@xi

where ui is the average speed of air ow; u0i u0j is the Reynolds stress; q is the air density; l is the molecular viscosity;
fi is the thermal-induced buoyant force; T is the potential
temperature; KT is the heat diusivity. The standard ke
model has been used to solve the turbulence problem.
The turbulence kinetic energy, k, and its rate of dissipation,
e, are obtained from the following transport equations:

 
@
@
@
lt @k
qk
qkui
l
Gk Gb  qe
@t
@xi
@xj
rk @xj
4

and
@
@
@
qe
qeui
@t
@xi
@xj


 
l @e
l t
re @xj

e
e2
C 1e Gk C 3e Gb  C 2e q
k
k

where Gk is the generation of turbulence kinetic energy due

to the mean velocity gradients; Gb is the generation of turbulence kinetic energy due to buoyancy; C1e, C2e and C3e constants and the KT and lt expressions are reported in the
standard ke model of Ansys Fluent 14.0, 2011; rk and re
are the turbulent Prandtl numbers for k and e, respectively.
To evaluate the impact of thermal eects, the incompressible
ideal gas module has been used for air density.
The simulated urban canyon has the following characteristics: it has an aspect ratio H/W = 1 and L/W = 5,
the orientation is NS, the buildings width and height are
20 m, the street width is 20 m and the street length is
100 m. Based on the best practice guidelines by Franke
et al. (2007) and Tominaga et al. (2008), the dimensions
of the computational domain have been chosen in relation
to the buildings height H (Fig. 1): the distance between the
side walls of the buildings and the north, east and south
planes is 5H = 100 m, instead the west plane is
15H = 300 m from the westerly building. The distance
between the roofs of the buildings and the upper plane is
5H = 100 m. The buildings dimensions determine the different domain extension behind the built area. When the
ow direction is transversal to the canyon length, the
obstacle size is maximum and the ow re-development
requires a distance of 15H from the buildings to the
outow bounds. Instead when the ow is longitudinal to
the canyon direction, the obstacle size is minimum and
the distance behind the built area is 5H. Those dimensions
allow to set the zero static pressure on the outlet plane and
zero gradients of all variables at the top and lateral sides of
the domain. The domain dimension over the buildings have
been chosen to take into account the blockage ratio,
dened as the ratio of the area blocked by the buildings
to the total cross-section area. This parameter depends
on the obstacles size and wind direction: when the building
obstacle area is minimum, the blockage ratio assumes the
value of 2%, instead when the wind impacts transversally
to the canyon direction, it assumes the maximum value
of 5.5%. To simulate the soil inuence, the computational
domain has been extended 5 m below the ground level.
The soil has been simulated setting the following
parameters: density = 1000 kg/m3; specic heat = 1000 J/
kgK; thermal conductivity = 2 W/mK; temperature
at -5 m = 288 K; emissivity = 0.9; solar radiation absorptivity (direct visible and infrared) = 0.8. The building walls
have: density = 1000 kg/m3; specic heat = 1000 J/kg K;
thermal conductivity = 0.15 W/m K; thickness = 0.30 m;
internal air temperature = 299 K; emissivity = 0.9; solar
radiation absorptivity (direct visible and near infrared) = 0.8. The resulting domain size is: 17.25  106 m3.

S. Bottillo et al. / Solar Energy 99 (2014) 110

Fig. 1. Three-dimensional domain view and points and planes of interest.

The temperature of surfaces has been obtained as result of

the heat transfers, setting up: the solar load module (longitude: 9.18, latitude: 45.47, UTC: +1), the temperature of
undisturbed air (303 K), the temperature of the internal
air of the buildings (299 K). To ensure an high quality of
the computational grid, it is fully structured and the shape
of the cells has been chosen hexahedral (Fig. 2): the normal
vector of a cell surface is parallel to the line connecting the
midpoints of neighboring cells. To simulate ow elds, in
the area of interest, 40 cells per cube root of the building
volume has been used and 20 cells per building separation
(Franke et al., 2007). For the vertical resolution of the canyon 20 cells have been used. Furthermore, the grid has been
arranged so that the evaluation height (1.55 m) and the
pedestrian wind speed at 1.52 m height is located higher
than the 3rd grid from ground surface. According to the
study of Ramponi and Blocken (2012), the velocity prole
has been set giving a uniform velocity magnitude of 2 m/s
at the velocity inlet boundary, the turbulence intensity at
10%. The aerodynamic roughness value z0 has been set in
relation to the roughness parameters in the ground surface
boundary conditions (Ramponi and Blocken, 2012): the
sand-grain roughness height ks and the roughness constant
Cs. Setting ks = 1m and Cs = 0.5, the resulting z0 (Eq. (6))
is 0.05 m (Blocken et al., 2007; Ansys Fluent Users Guide,
2011), that has been considered an appropriate value to
represent the roughness of the outer region (Blocken
et al., 2007; Norris and Richards, 2010). A zero roughness
height has been used for the building surfaces.

ks

9:793z0
Cs

As the ow approaches the built area the velocity inlet

prole looks fully-developed before reaching the buildings,
as it can be seen in Fig. 3, and it can be represented by Eq.
(7), where u* is the friction velocity, k is the Von Karman
constant (0.4) and z is the height coordinate.


u
z z0
uz ln
7
k
z0
The friction velocity value has been obtained by the correlation with the calculated value of the turbulent kinetic
energy (k) at the rst node above the ground, as shown
in Eq. (8) (Franke et al., 2007).
u k 0:5 C 0:25
l

where Cl = 0.09.
In Fig. 4 is shown the comparison between the velocity
inlet prole before reaching the urban canyon and the prole calculated with the logarithmic law (Eq. (7)).
3. Validation of the CFD model by wind tunnel experiment
data
The validation of the mathematical model used in our
study has been carried out through the comparison with
the wind tunnel experiment performed by Uehara et al.
(2000). In order to describe the roughness elements of the
urban environment, 44 rows of blocks with a size of

Fig. 2. Views of the computational grid.

S. Bottillo et al. / Solar Energy 99 (2014) 110

Fig. 3. Wind velocity inlet proles at dierent distances from the velocity inlet.

4. Results
4.1. Comparison between steady and transient simulation

Fig. 4. Comparison between the simulated wind velocity inlet prole and
the prole calculated with the logarithmic law (Eq. (7)).

100  100  50 mm3 have been used. Further, to represent

street canyons, 14 rows of blocks with a size of
100  100  100 mm3 have been set. The measured data
have been carried out in the space between the 5th and
the 6th of the 14 rows. The experiment has been characterized by a ow direction transversal to rows of blocks and,
at the measurement location, the ow appeared totally canalized. A numerical validation test has been performed on
our street canyon model, which has been scaled by the
reduction factor 1/200 and where the wind direction has
been considered totally transversal to the canyon direction.
The air temperature Ta was set as 20 C (293 K) and the
ground temperature Tf was set as 79 C (352 K), the inow
wind speed u0 was set as 1.5 m/s in order to reproduce the
conditions of the experimental tests. The comparison
between the data obtained with the wind tunnel experiments and the numerical test are illustrated in Fig. 5: the
vertical prole of normalized horizontal velocity u/u0 and
T T f
, evaluated on a central vertical line within the street
T a T f
canyon. As in Lei et al. (2012) study, it can be observed
that there is a good agreement between the mathematical
model and the wind tunnel experiment within the canyon.
Instead the numerical tests overestimate the wind velocity
and the air temperature above the canyon.

A comparison between the results obtained by the

steady simulation and the transient one has been carried
out. The steady simulation has been performed on 26 June
at 14:00, instead the transient simulation has been started
on 18 June at 00:00 and it has been stopped 9 days after,
on 26 June at 24:00. The ground temperature has been chosen as parameter of comparison between the transient and
steady simulation, to highlight the thermal inertia eects,
because the ground is the element with higher thermal
capacity. The results of the steady simulation have been
compared with the last day results of the transient one.
In Fig. 6 are shown the ground temperature values of the
three points of interest (Point A and B within the canyon,
placed at 1 m away from building, respectively from the
easterly facade and from the westerly one; Point Ext, external to the canyon, as shown in Fig. 1). As it can been
noticed, the temperatures of the various points chosen to
study the urban canyon in transient case are very similar
to the stationary one. This does not mean that the temperature of the soil inner layers are the same for both simulations, but we are interested on the surface temperature and
to study its inuence on the ow eld. The ground temperature of the external point is the same for both simulations,
instead, it can be noticed a dierence of 3 for the point A
and a smaller one for point B. Those dierences are due to
the ground thermal inertia which determines a lower temperature value as the point passes from shadow to sun
exposition and viceversa. The same consideration with an
opposite sign can be applied to the point B, which enter
into shadow shortly before 14:00. It can be concluded that
the stationary case for a thermo-uid dynamic simulation
of a street canyon is representative of the physical phenomena of the maximum solar load. For this reason, further
analysis will be performed on the stationary case, which
is much lighter as to the computational time. The analysis
of hourly trends of ground temperatures shows that it is
higher within the canyon than outside, during the hours
of maximum solar radiation, while, during the rest of the
day, their values are signicantly lower due to the eect
of shadows.

S. Bottillo et al. / Solar Energy 99 (2014) 110

Fig. 5. Comparison between the simulated data and the observed data by Uehara et al. (2000). (a) u/u0 and (b) (T  Tf)/(Ta  Tf).

Fig. 6. Comparison between transient and steady simulation: hourly trends of ground temperature of the three points of interest and the values calculated
at 14:00 with the steady simulation.

4.2. Impact of thermal eects

In order to evaluate the impact of thermal eects on our
street canyon model, a comparison between two simulations has been performed. The simulations have been performed in stationary case on 26 June at 11:00. In the rst
simulation the natural convection has been excluded,
instead in the second one it has been considered. The ow
regime impacting on the built area is described by a veloc-

ity inlet magnitude of 2 m/s and a wind direction of 45N,

so that the hottest facade is the windward one. The resulting average Richardson number (Ri) for the windward facade is 2.9 and for the leeward one is 1.0. The three
dimensional eect of the ow is evident in the formation
of a spiral ow, produced by the combination of the downward vertical vortex and the longitudinal component of the
wind velocity, as reported in other 3D simulations (Assimakopoulos et al., 2006; Santamouris et al., 1999); they

S. Bottillo et al. / Solar Energy 99 (2014) 110

found that the wind eld in urban areas is quite complex

and the simulated wind speed intensities can be totally different from the measured data. Fig. 7 shows the velocity
vectors on three vertical planes of interest for both simulations. The gure shows the XZ velocity vectors for both
simulations and the XY velocity vectors for the simulation
with natural convection activated. On the XY plane, the
ow pattern does not change signicantly for the two simulations; Fig. 7(d) shows that the ow within the canyon is
transported from the north opening to the south one. The
North and South planes are placed at 10 m from the
respective openings, instead the Central plane is in the middle of the canyon. The facade of the easterly building, in
shadow, is on the right side of the gures, instead the
facade of the westerly one, exposed to the sun radiation
is on the left side. When the natural convection is deactivated, the ow pattern changes signicantly from the north
opening to the Central plane. Instead, from the Central

plane to the south opening, the ow pattern remains basically constant. As it can be seen in Fig. 7(a1) and (d), in the
North plane the formation of a double vortex can be
observed. The vortices are generated by geometrical
discontinuities (the roof and the vertical corner of the easterly building) and they have two dierent rotation axis. The
upper vortex has its axis parallel to the canyon direction
(Fig. 7(a1)), and the lower has a vertical one (Fig. 7(d)).
From Central plane to the South plane, the aerodynamic
vortex coming from the roof is fully-formed and it occupies
all the space between buildings (Fig. 7(b1) and (c1)). In the
South plane (Fig. 7(c1)) the mass ow rate coming from the
roof is zero and the vortex is transported from the easterly
facade to the westerly one by the longitudinal ow. The
ow pattern of the North plane when the natural convection is activated (Fig. 7(a2)), is double-vortex in structure
as in the rst simulation. In the Central and South plane
(Fig. 7(b2) and (c2)) the ow pattern is divided in two

Fig. 7. XZ velocity vectors on the North, Central and South planes, respectively (a), (b) and (c). No natural convection simulation (a1), (b1) and (c1);
simulation with natural convection activated (a2), (b2) and (c2). XY velocity vectors at 10 m height with natural convection activated (d).

S. Bottillo et al. / Solar Energy 99 (2014) 110

counter-rotating vortex: the upper one due to the geometrical discontinuity and the second one, in the low corner
near the hot wall, due to the buoyancy eect. The convective vortex appears less developed than the simulations carried out by Lei et al. (2012) and Xie et al. (2007), probably
because of the 3D nature of our simulation. The impact of
thermal eects have been studied also through the analysis
of velocity and turbulent kinetic energy near the canyon
surfaces on the Central plane, for both simulations. For
the building facades a vertical line near the westerly one
exposed to the sun radiation, from ground level to building
roof, has been considered. For the ground surface instead,
an horizontal line, from the easterly facade to the westerly
one, has been taken. All those lines of interest are placed at
0.40 m from the respective surfaces. In Fig. 8 are shown:
the vertical trends of Z velocity component and turbulent
kinetic energy near the sun exposed facade, wall temperature and heat transfer coecient on the facade itself; velocity magnitude and turbulent kinetic energy above the
ground, temperature and heat transfer coecient on the
ground are shown in Fig. 8(d1) and (d2). The Z velocity
component along a vertical line at 0.40 m from the sunexposed wall (Fig. 8(a1) and (a2)) without natural convection is negative (downward) and it is aected only by the
aerodynamic vortex, instead, when the natural convection
is activated, it is aected by buoyancy forces and it is positive, from the ground to the half height of the building. On
the central plane, the buoyancy eect is maximum at 4 m
height from the ground level and it extends up to 0.60 m
distance from the hot wall. It has been noticed that the
temperatures of the facade and of the ground that are
not exposed to the direct radiation, are 6 higher than
the air temperature, but it does not seem to have relevant
eects on the ow eld. On the horizontal line the velocity
magnitude is lower when the thermal eects are excluded.
The turbulent kinetic energy is higher near all the surfaces
when the natural convection is activated and it strongly
aects the heat transfer coecient trends. In particular,
Fig. 8(b1), (b2), (d1) and (d2) shows that, activating the
natural convection, the turbulent kinetic energy increases
four times and the heat transfer coecient doubles, from
ground level to half height of the building. The natural
convection eect determines also an increase of air circulation within the canyon; as it can be seen in Fig. 8(c1) and
(c2), when the natural convection module is activated, the
velocity magnitude reaches the value of 2 m/s, instead
when it is deactivated, the maximum velocity magnitude
value is 1.5 m/s. As it can be seen in Fig. 8(a1), (a2), (c1)
and (c2), the surfaces temperatures are several degrees
lower when natural convection is activated. In order to
evaluate the impact of 3D eects on the heat transfer coefcient, we have performed a 2D simulation and a 3D simulation characterized by a transversal wind direction,
excluding the natural convection. The results show that,
when the domain is tridimensional and the wind direction
is transversal to the canyon axis, the heat transfer coecient values are very similar to the 2D simulation. The

2D results of the heat transfer coecient (Fig. 9) are congruent with the values reported in Saneinejad et al.
(2011), except for the upper part of the windward facade,
where our simulation shows higher values; this dierence
of heat transfer coecient is due to the shape of our 2D
canyon model; Saneinejad et al. (2011) simulate the street
canyon as a cavity in the ground, instead we simulate it
as two buildings over the ground level. When the domain
is 3D and the wind direction is 45N, the heat transfer coefcient values without buoyancy eects, shown in
Fig. 8(b1), and the values obtained with buoyancy eects,
Fig. 8(b2), are higher than the ones found in our 2D
(Fig. 9) and 3D transversal simulation, and the ones
reported by Saneinejad et al. (2011) and Allegrini et al.
(2012a). Those dierences are probably due to the 3D
eects and to the longitudinal speed component within
the canyon that, when the ambient wind speed is 2 m/s
and the direction is 45N, is more than two times higher
than the value of 0.5 m/s of the 2D simulations, reported
in Saneinejad et al. (2011). Allegrini et al. (2012b) showed
that, when the Ri > 1, the standard wall function overestimates the value of heat transfer coecient; our results
show that even when the natural convection is not considered, the hc values are much higher than the 2D simulations. Natural convection increases remarkably (as shown
in Fig. 8) and the values seem to be similar to the measured
ones on building facades (Defraeye et al., 2010). In this
study, the calculation of the heat transfer coecient (hc)
is carried out at numerical level by the usual relations of
the standard ke model and it is strongly related to air thermal conductivity (kT):
hc

dT
dy

kT
T

The air thermal conductivity is given by the following

expression:
kT qcp K T

10

where the thermal turbulent diusivity (KT) is related to the

kinematic turbulent viscosity (tT) through the turbulent
Prandtl number:
tT
PrT
11
KT
Which is taken equal to 0.85.
Thus, the kinematic turbulent viscosity is related to the
turbulent kinetic energy (k) and to turbulent dissipation
rate (e) according to the following relationship:
tT c l

k2
e

12

where cl is 0.09.
The turbulent dissipation rate e is expressed by the following equation (Tominaga et al., 2008):
e

cl3=4 k 3=2
l

13

S. Bottillo et al. / Solar Energy 99 (2014) 110

Fig. 8. Impact of thermal eects on a vertical central plane within the canyon, comparison between parameter evaluated without natural convection
(subscript 1) and with natural convection (subscript 2): Z velocity component along a vertical line at 0.40 m from the windward facade and its temperature
on a parallel line on the facade itself (a1 and a2); turbulent kinetic energy and heat transfer coecient along the same lines (b1 and b2); velocity magnitude
along an horizontal line at 0.40 m from the ground surface and it is temperature (c1 and c2), turbulent kinetic energy and heat transfer coecient along the
same lines (d1 and d2).

S. Bottillo et al. / Solar Energy 99 (2014) 110

related to the square root of turbulent kinetic, as shown

in Fig. 8 and in Eq. (14), that in turn depends on the local
dierence between the surfaces and air temperature.
5. Conclusions

Fig. 9. Heat transfer coecient on the windward (WW) facade and on the
leeward (LW) one for the 2D simulation of the street canyon.

where l is the mixing length.

So the correlation between the heat transfer coecient
and turbulent kinetic energy is:
hc

dT
qcp c1=4
l l dy 1=2
k
PrT T

14

In literature there are several empirical relationships

between heat transfer coecient over the external surfaces
of the buildings (hc) and air velocity (Defraeye et al., 2010),
but a few authors (Oleson et al., 2008; Masson, 2000) consider the eect of turbulent kinetic energy on hc. In those
studies the heat transfer coecient is related to the friction
velocity value, through the correlations reported in Mascart et al. (1995), so that the heat transfer coecient
depends on the atmospheric stability considering average
values on all the surfaces and for every wind speed directions. In our analysis, instead, the heat transfer coecient
is related to the square root of turbulent kinetic, as shown
in Fig. 8 and in Eq. (14). It has been noticed a direct correlation of the heat transfer coecient and the local thermal stratication: for example, the thermal eects on the
heat transfer coecient on the leeward wall in shadow
are less marked than on the sun-exposed wall, but an
increase of the average heat transfer coecient can be
observed (from 7.1 W/m2 K without thermal eects to
10.5 W/m2 K with natural convection module activated).
It seems that the buoyancy induced vortex aects the overall canyon ow eld. These variations of heat exchange
processes among the various surfaces of the canyon, can
have an eect on the surface temperatures, as seen in
Fig. 8, and then a direct inuence on thermal comfort indices like the mean radiant temperature (De Lieto Vollaro
et al., 2013), in particular the heat transfer coecient is

In this study a numerical simulation method has been

used to investigate the physical phenomena that characterize a street canyon with a certain H/W ratio, N-S oriented,
during a summer day. A fully 3D model has been simulated, considering the eects of solar radiation and radiative exchange on canyon air and surfaces temperature. In
the rst part, a comparison between a steady and a transient simulation has been carried out, and it can be concluded that the stationary case is representative of
thermo-uid dynamics parameters within the canyon during the hottest hours of the day. In the second part of
the study the impact of thermal eects on the ow eld
has been analyzed and the dierence between a 2D simulation and the fully tridimensional case has been shown.
Even when the ow eld is characterized by a strong velocity component parallel to the canyon direction, it has been
noticed that the buoyancy forces determine a thermal
induced vortex near the hottest facade. The obtained
results are coherent with the studies on numerical simulations performed by Lei et al. (2012) and Xie et al. (2007),
even if their simulation has been carried out on 2D
domains and the thermal eects are more signicant. The
importance of considering a 3D domain has been shown,
because the ow eld within the canyon changes signicantly as the air passes through the canyon itself and the
heat transfer coecient values are remarkable higher than
the 2D and the 3D transversal case. It can be seen that the
wind direction strongly aects the thermal processes within
the canyon. So that further analysis must be performed
considering not only dierent wind velocity intensities but
also dierent directions. Besides, buoyancy forces aect
the ow eld, determining thermal induced vortices and a
higher value of turbulent kinetic energy, aecting the heat
transfer coecient and consequently the surface temperatures. Taking into account that the standard ke seems to
overestimate the production of turbulent kinetic energy
(Assimakopoulos et al., 2006; Lakehal, 1998; Meroney
et al., 1999), further numerical analysis and measurement
campaigns must be performed to investigate the correlation
between the heat transfer coecient and the characteristics
(magnitude and direction) of ow eld in an urban canyon,
in order to determine the local heat exchange processes and
their eects on the thermal comfort in outdoor spaces and
on the buildings energy needs for cooling and heating.
References
Allegrini, J., Dorer, V., Carmeliet, J., 2012a. Analysis of convective heat
transfer at building facades in street canyons and its inuence on the
predictions of space cooling demand in buildings. Journal of Wind
Engineering and Industrial Aerodynamics 104106, 464473.

10

S. Bottillo et al. / Solar Energy 99 (2014) 110

Allegrini, J., Dorer, V., Defraeye, T., Carmeliet, J., 2012b. An adaptive
temperature wall function for mixed convective ows at exterior
surfaces of buildings in street canyons. Building and Environment 49,
5566.
Ansys Fluent version 14.0.0, 2011. Users Guide.
Assimakopoulos, V.D., Georgakis, C., Santamouris, M., 2006. Experimental validation of a computational uid dynamics code to predict
the wind speed in street canyons for passive cooling purposes. Solar
Energy 80, 423434.
Blocken, B., Stathopoulos, T., Carmeliet, J., 2007. CFD simulation of the
atmospheric boundary layer: wall function problems. Atmospheric
Environment 41, 238252.
Bozonnet, E., Belarbi, R., Allard, F., 2005. Modeling solar eects on the
heat and mass transfer in a street canyon, a simplied approach. Solar
Energy 79, 1024.
Defraeye, T., Blocken, B., Carmeliet, J., 2010. Convective heat transfer
coecients for exterior building surfaces: existing correlations and
CFD modeling. Energy Conversion and Management.
Defraeye, T., Blocken, B., Carmeliet, J., 2011. An adjusted temperature
wall function for turbulent forced convective heat transfer for blu
bodies in the atmospheric boundary layer. Building and Environment
46 (11), 21302141.
De Lieto Vollaro, R., Vallati, A., Bottillo, S., 2013. Dierent methods to
estimate the mean radiant temperature in an urban canyon. Advanced
Materials Research 650, 647651.
Franke, J., Hellsten, A., Schlunzen, H., Carissimo, B., 2007. Best practice
guideline for the CFD simulation of ows in the urban environment.
COST Action 732.
KovarPanskus, A., Moulinneuf, L., Savory, E., Abdelqari, A., Sini, J.F.,
Rosant, J.M., Robins, A., Toy, N., 2002. A wind tunnel investigation
of the inuence of solar-induced wall-heating on the ow regime within
a simulated urban street canyon. Water, Air, and Soil Pollution: Focus
2, 555571.
Lakehal, D., 1998. Application of the ke model to ow over a building
placed in dierent roughness sublayers. Journal of Wind Engineering
and Industrial Aerodynamics 73, 5977.
Lei, L., Lin, Y., Li-Jie, Z., Yin, J., 2012. Numerical study on the impact of
ground heating and ambient wind speed on ow elds in street
canyons. Advances in Atmospheric Sciences 29, 12271237.
Louka, P., Vachon, G., Sini, J.F., Mestayer, P.G., Rosant, J.M., 2002.
Thermal eects on the airow in a street canyon Nantes99
experimental results and model simulations. Water, Air, Soil Pollution:
Focus 2, 351364.
Mascart, P., Noilhan, J., Giordani, H., 1995. A modied parameterization
of ux-prole relationship in the surface layer using dierent rough-

ness length values for heat and momentum. Boundary-Layer Meteorology 72, 331344.
Masson, V., 2000. A physically-based scheme for the urban energy budget
in atmospheric models. Boundary-Layer Meteorology 94, 357397.
Meroney, R.N., Leitl, B.M., Rafailidis, S., Schatzmann, M., 1999. Windtunnel and numerical modeling of ow and dispersion about several
building shapes. Journal of Wind Engineering and Industrial Aerodynamics 81, 333345.
Oerle, B., Eliasson, I., Grimmond, C.S.B., Holmer, B., 2007. Surface
heating in relation to air temperature, wind and turbulence in an urban
street canyon. Boundary-Layer Meteorology 122, 273292.
Oleson, K.W., Bonan, G.B., Feddema, J., Vertenstein, M., Grimmond,
C.S.B., 2008. An urban parameterization for a global climate model.
Part I: Formulation and evaluation for two cities. Journal of Applied
Meteorology and Climatology 47, 10381060.
Ramponi, R., Blocken, B., 2012. CFD simulation of cross-ventilation for
a generic isolated building: impact of computational parameters.
Building and Environment 53, 3448.
Richards, P.J., Norris, S.E., 2010. Appropriate boundary conditions for
computational wind engineering models revisited. Journal of Wind
Engineering and Industrial Aerodynamics 99, 257266.
Saneinejad, S., Moonen, P., Defraeye, T., Carmeliet, J., 2011. Analysis of
convective heat and mass transfer at the vertical walls of a street
canyon. Journal of Wind Engineering and Industrial Aerodynamics
99, 424433.
Santamouris, M., Papanikolaou, N., Koronakis, I., Livada, I., Asimokopoulos, D., 1999. Thermal and air ow characteristics in a deep
pedestrian canyon under hot weather conditions. Atmospheric Environment 33, 45034521.
Takebayashi, H., Moriyama, M., 2012. Relationships between the
properties of an urban street canyon and its radiant environment:
introduction of appropriate urban heat island mitigation technologies.
Solar Energy 86, 22552262.
Tominaga, Y., Mochidab, A., Yoshiec, R., Kataokad, H., Nozue, T.,
Yoshikawaf, M., Shirasawac, T., 2008. AIJ guidelines for practical
applications of CFD to pedestrian wind environment around buildings. Journal of Wind Engineering and Industrial Aerodynamics 96,
17491761.
Uehara, K., Murakami, S., Oikawa, S., Wakamatsu, S., 2000. Wind
tunnel experiments on how thermal stratication aects how in and
above urban street canyons. Atmospheric Environment 34, 15531562.
Xie, X., Liu, C.H., Leung, D.Y.C., 2007. Impact of building facades and
ground heating on wind ow and pollutant transport in street canyons.
Atmospheric Environment 41, 90309049.