Building and Environment 43 (2008) 21122123

Thermal design tool for outdoor spaces based on heat balance

simulation using a 3D-CAD system
Takashi Asawa
a,
, Akira Hoyano
b
, Kazuaki Nakaohkubo
b
a
A & A Co., Ltd., 2-3-15 Kanda-Surugadai, Chiyoda-ku, Tokyo 101 0062, Japan
b
Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, 4259 Nagatsuda-cho, Midori-ku, Yokohama 226 8502, Japan
Received 26 September 2007; received in revised form 14 December 2007; accepted 14 December 2007
Abstract
This paper focuses on the development of a thermal design tool for use in planning outdoor spaces by combining a heat balance
simulation for urban surfaces, including buildings, the ground and greenery, with a 3D-CAD system that can be run on a personal
computer. The newly developed tool is constructed by improving the previous simulation model, which uses the geographic information
system (GIS) for the input data. The simulation algorithm is constructed so as to predict the surface temperature distribution of urban
blocks while taking into account the actual design of the outdoor space using the 3D-CAD system. A method of multi-tracing simulation
to calculate the sky view factor and radiative heat transfer is established. The optimal mesh size is examined for the tool so as to provide
detailed spatial geometry within a suitable calculation time. The simulation model is integrated with an all-purpose 3D-CAD software,
and the pre-processing method are constructed for practical use. The results obtained by applying this simulation tool to an area of
detached houses reveals that the tool is able to evaluate the effects of building shape, materials, and tree shade on the surface temperature
distribution, as well as the MRT and HIP, which are evaluation indices of the outdoor thermal environment.
r 2007 Elsevier Ltd. All rights reserved.
Keywords: Thermal environment; Environmental design; Numerical simulation; Surface temperature; 3D-CAD; Outdoor space
1. Introduction
In metropolitan areas of Japan, the heat island
phenomenon is a serious problem during the summer
months. Under such circumstances, the central government
of Japan adopted the Heat Island Countermeasure Scheme
in March of 2004. In the near future, it will be required for
developers and architectural designers to estimate the
potential effects of a proposed design on the urban thermal
environment at both the planning and design stages.
In recent years, the availability of numerical simulation
tools for evaluating the urban thermal environment that
can be implemented using personal computers has in-
creased gradually. For example, ENVI-met [1,2], provided
as freeware on a web, is a three-dimensional microclimate
model, based on computational uid dynamics (CFD),
designed to simulate microscale interactions between urban
surfaces, vegetation and the atmosphere in an urban
environment. This tool has been applied to a lot of
numerical studies with regard to the analysis of urban
microclimates [3,4]. AUSSSM TOOL [5], which has a
graphical user interface (GUI) to provide user support,
quantitatively estimates and evaluates the main causes of
the urban heat island (UHI) based on the multi-layer urban
canopy model.
Although an all-purpose 3D-CAD (Computer Aided
Design) software tool would be quite useful as a pre-post
processor for such simulation tools in practical use for
architectural designers and developers, at present, no
thermal design tools that use 3D-CAD systems are
available to aid such specialists in evaluating the urban
thermal environment. With regard to building simulation
tools, such as building energy simulation, thermal simula-
tion, and daylighting simulation, CAD has been applied as
the modeling data for these tools [68].
In the present study, we develop a thermal design tool
for use in planning outdoor spaces by combining a heat
ARTICLE IN PRESS
www.elsevier.com/locate/buildenv
0360-1323/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.buildenv.2007.12.007

Corresponding author.
E-mail address: asawa@aanda.co.jp (T. Asawa).
balance simulation for urban surfaces, including buildings,
the ground and greenery, with a 3D-CAD system
implemented on a personal computer.
2. Outline of simulation tool development that incorporates
a 3D-CAD system
The surface temperatures of buildings and the ground
are the primary factors that inuence increases in air
temperature and the formation of an outdoor thermal
radiant eld. In previous studies, the authors investigated
the relationship between land cover types and their
respective surface temperatures using airborne remote
sensing [9,10]. For microscale urban environments, the
authors developed a spherical thermography recording
system to quantitatively evaluate the relationship between
spatial geometry, including the materials of the spatial
components, and the thermal environment in outdoor
spaces [11]. Our ndings indicated the importance of
considering the outdoor surface temperature distribution
when planning outdoor and urban blocks, particularly in
hot and humid climates such as that of Japan in summer.
In order to extend the predictive potential of these
ndings, we developed a heat balance simulation model
that can predict the surface temperature distribution of an
urban area using Geographic Information System (GIS)
data [12]. In the previous simulation model, the building
geometry is simply modeled by entering the building height
information into the 2D-plan data derived from GIS data.
The present study improves the previous simulation
model, so as to predict the surface temperature distribution
of urban blocks, taking into account the actual design of
the spaces, including buildings, the ground and greenery.
The fundamental algorithm of this simulation model was
constructed in the previous study [12] by the authors, using
measured data obtained by airborne remote sensing, so
that the present paper focuses on the following two
important aspects. One is the simulation algorithm to
realize the heat balance calculation in an outdoor space
having a detailed spatial geometry, and the other is an
integration method of the heat balance simulation with a
3D-CAD system (Fig. 1) for use as a practical tool. The
prediction of local air temperature and humidity will be
considered in a future study, whereas the present study
deals with the prediction of the surface temperature
distribution.
Based on the prospect that this tool will be applied to
architectural design and urban block development projects,
the subject spatial scale of this simulation tool is set to
ARTICLE IN PRESS
Nomenclature
R
S
net solar radiation (W/m
2
)
R
L
net long wave radiation (W/m
2
)
R
La
atmospheric radiation (W/m
2
)
R
Lw
long wave radiation from surrounding build-
ings and ground (W/m
2
)
Q
H
sensible heat ux (W/m
2
)
Q
E
latent heat ux (W/m
2
)
Q
G
conductive heat ux (W/m
2
)
T temperature (K)
a
su
solar absorptivity
y incidence angle of direct solar radiation (rad)
I
DR
amount of direct solar radiation (W/m
2
)
F shape modulus (sky: sky factor)
I
SR
amount of sky solar radiation (W/m
2
)
I
RR
amount of reected solar radiation (W/m
2
)
e long wave emittance
s Stefan Boltzmann constant (W/(m
2
K
4
))
a,b constant on Brunts formula
e water vapor pressure near the ground (Pa)
a
c
convective heat transfer coefcient (W/(m
2
K))
l evaporation heat (J/kg)
b evaporation efciency ()
k mass transfer coefcient (kg/(m
2
s(kg/kg
0
)))
X absolute humidity (kg/kg
0
)
l heat conductivity (W/(mK))
Cr volumetric specic heat (W/(mK))
c cloud amount (0pcp10)
m coefcient of cloud altitude
a
o
over-all heat transfer coefcient (W/(m
2
K))
T
sr
surface temperature for indoor side (K)
T
r
room air temperature (K)
Subscripts
s surface
a atmosphere
w surrounding buildings and ground
n total number of objects that emit long wave
radiation
Fig. 1. Example of 3D-CAD model (a residential area).
T. Asawa et al. / Building and Environment 43 (2008) 21122123 2113
range in size from a single building to an ordinary urban
block.
3. Development of the simulation algorithmheat balance
calculation method reproducing spatial geometry
The heat balance calculation for predicting the
surface temperature of each mesh is performed as described
below. In the developed calculation method, the heat
balance on the external surfaces is simulated using a
high-resolution voxel mesh model generated by
3D-CAD. The calculation point of heat balance is
set for the center of each voxel mesh having
normal direction of the surface. The applied mesh
size is determined in Section 4.1. Fig. 2 shows a
schematic illustration of the mesh models and the
heat balance calculation method. Fig. 3 shows a
ow chart depicting the various components and
processes involved in this tool. The fundamental formula
of the heat balance calculation is shown in the following
formula (1):
/Heat balance formula on each surfaceS
R
S
R
L
Q
H
Q
E
Q
G
0
R
S
a
su
cos y I
DR
F
sky
I
SR
I
RR

R
L
F
sky
R
La
R
Lw

s
sT
4
s
F
sky
R
La

s
sT
4
s
R
Lw
F
w

s
sT
4
s
,
F
sky
F
w
F
sky

X
n
i1
F
i
1,
R
La

s
sT
4
s

s
sT
4
a
a b

e
p

s
sT
4
s
clear sky,
R
La

s
sT
4
s

s
sT
4
a
a b

e
p

s
sT
4
s

1:0 1 mc=10 cloudy sky,

R
Lw

s
X
n
i1
F
i

wi
sT
4
wi
,
Q
H
a
c
T
a
T
s
,
Q
E
lbkX
s
X
a
,
Q
G
l
qT
qx
. 1
/One dimensional heat conduction (inside of each
surface)S
qT
qt

l
Cr
q
2
T
qx
2
. (2)
/Boundary condition for indoor insideS
l
qT
qx
a
o
T
sr
T
r
. (3)
3.1. Direct solar radiation
The quantity of direct solar radiation I
DR
received by a
mesh is simulated by the ray-tracing method. If the ray
tracing is not interrupted by obstructions in the calculation
area, then the quantity of direct solar radiation that the
mesh surface receives is allotted to the mesh. The each
mesh has three-dimensional direction data using polar
coordinates (Fig. 4); the received quantity of direct solar
radiation is estimated, taking into account the solar
absorptivity and the three-dimensional direction of the
surface.
3.2. Sky solar radiation
The quantity of sky solar radiation I
SR
received by a
mesh is estimated by multiplying the sky view factor
of the mesh by the quantity of sky solar radiation for
unobstructed sky derived from weather condition data.
ARTICLE IN PRESS
Sky solar radiation
Atmospheric
radiation
Sky view factor
of each mesh
Solar radiation
(direct, sky, reflected)
Reflected solar
radiation
Long wave radiation
Convective
heat transfer
Direct solar
radiation
Building
Tree
Solar radiation reflection (specular, diffused)
Solar
transmission
Heat
conduction
Fig. 2. Schematic diagram of the mesh model and heat balance calculation method.
T. Asawa et al. / Building and Environment 43 (2008) 21122123 2114
The sky view factor is calculated by the multi-tracing
simulation from the mesh toward multiple hemispherical
directions. The tracing direction is established so that the
tracing density (interval) comes to have the same form
factor (Fig. 5, Formula (4)). The sky view factor is
estimated by counting the number of tracers reaching the
upper boundary surface. The relationship between the
number of tracers and the calculation accuracy is
investigated in Section 4.2.
m

p 1
s
,
ni p 2 i 1,
ARTICLE IN PRESS
Making urban block model using 3D-CAD software
Spatial Component Database
Material Database
Transforming 3D-CAD model into 3D-Mesh model for calculation
Direct solar radiation
Reflected solar
radiation
(Specular, Diffused)
Sky solar radiation
Atmospheric radiation
Convective heat
transfer
Calculation of surface
temperature
Display of surface temperature
distribution on 3D-CAD
MRT distribution at
a height of 1.5m
HIP
(Heat Island Potential)
Long wave Radiation
Vertical quantity of
total solar radiation
Relative humidity
Wind velocity
Sky view factor
Cloud amount
Building: building structure, member, material,
surface color and so on
Tree: species, solar transmittance and so on
Ground: ground cover, materialsand so on
Physical properties of material:
heat conductivity,volumetric specific heat,
solar reflectance, longwave emittance
Normal direction of the mesh
surface, Solar absorptivity,
The time,Longitude, Latitude
Separation of direct and
diffuse components of
radiation
Direct solar radiation
incident on a normal surface
Sky solar radiation incident
upon a horizontal surface
Brunts formula
Atmospheric radiation
(unobstructed sky)
Longwave emittance of
each mesh surface
Convective heat
transfer coefficient
Convective heat transfer
coefficient for indoor side,
indoor air temperature
Air temperature
1-dimentional heat
conduction
Jurgess formula
Longwave emittance
INPUT
SOLVER
OUTPUT
Heat balance calculation for each mesh
Heat balance on
each surface
Weather Condition
Database
Heat Transfer Calculation
Model Database
rooftop lawn, ivy
covered wall,
louver and so on
Building, Tree, Ground and so on
receiving and emitting
..
Fig. 3. The ow chart of this simulation tool.
T. Asawa et al. / Building and Environment 43 (2008) 21122123 2115
fi cos
1
2 i 1
2 m

,
Dyi 2p

ni, 4
where N is total number of tracer, m is division number for
elevation angle (round off), n[i] is division number for
horizontal angle (round off), f[i] is levation of the tracing
for i division (rad), and Dy[i] is horizontal division angle for
tracing (rad).
3.3. Reected solar radiation
The reected solar radiation I
RR
simulated by this tool
includes both specular reection and isotropic diffuse
reection. Both of these reection factors consider the rst
reection. Although specular reection is an important
factor with respect to the outdoor heat balance, particu-
larly in an outdoor space enclosed by glassed buildings, few
tools introduce this factor into the heat balance simulation
process.
The specular reective radiation is calculated such that
the tracing simulation toward the direction of the specular
reection is implemented, and the quantity of reected
solar radiation is allotted to the mesh that the tracing
reaches. Diffuse reective radiation is estimated under the
assumption of isotropic diffuse reection, following Lam-
berts cosine law, and the quantity of the diffuse reective
radiation that a mesh receives is calculated by performing
the multi-tracing simulations toward surrounding meshes.
The method used for the multi-tracing simulation is the
same as that for the estimation of the sky view factor. In
this tracing process, if a tracing hits a mesh having a diffuse
reection surface, then the quantity of reected radiation
of the mesh is obtained. This tracing is implemented for
multiple directions in order to estimate the total amount of
the received radiation value from the surroundings.
The reected solar radiation after rst reection is all
absorbed in the next meshes the reected tracers reach. The
effects of neglecting the second and more reections on the
heat balance of received solar radiation are approximately
5% in general urban block models. In the next study, the
reection model that can calculate second and more
reections will be introduced in case this tool is applied
to the urban block model having high solar reectance
surfaces.
3.4. Atmospheric radiation
The quantity of atmospheric radiation R
La
received by a
mesh is also estimated by the sky view factor and the
quantity of atmospheric radiation calculated by Brunts
formula for unobstructed sky.
3.5. Long wave radiation from the surroundings
This system involves a CAD model having detailed
spatial geometry and so requires a large number of meshes
for calculation. Difculties arise when calculating the long
wave radiation and surface temperature distribution
simultaneously, because even advanced computers do not
have sufcient memory to calculate the overall heat
balance. Therefore, a computer performance-based algo-
rithm is adopted for the calculation of long wave radiation
from the surroundings R
Lw
and surface temperature [12].
(i) The heat balance of each mesh is simulated by
assuming that the surface temperature of the sur-
roundings is equal to the air temperature derived from
weather condition data;
(ii) Long wave radiation received from the surroundings is
simulated based on these approximate surface tem-
peratures;
(iii) The heat balance of each mesh is then calculated again
using the results of step (ii), and the corrected surface
temperature, which shows the effect of long wave
radiation, is then obtained.
In this calculation, the multi-tracing simulation toward
surrounding meshes is implemented in order to determine
meshes, which are used to estimate the radiant ux from
the surroundings. The method of the tracing simulation is
the same as those used for the estimation of the sky view
factor and diffuse reective radiation. This calculation
deals with all surrounding subjects as perfect black body
with respect to long wave radiation. The effect of this
ARTICLE IN PRESS
i
i=1
i=m
[i]
[i]
Fig. 5. Establishment of tracing directions.
2
1
3
3
4
x
y
z
Normal direction
Horizontal angle
Vertical angle
A
B
Plane type B [deg.]
1. Vertical
2. Slope
3. Upward
4. Downward
0 ~360
0 ~360
A [deg.]
0
0 ~90
90
-90
Fig. 4. 3D-directions for each surface.
T. Asawa et al. / Building and Environment 43 (2008) 21122123 2116
assumption on the heat balance of long wave radiation is
within approximately 3% in general urban block models.
3.6. Convective heat transfer (sensible heat ux)
Convective heat transfer (sensible heat ux) Q
H
is
calculated under the assumption that there is no distribu-
tion of air temperature and wind velocity in the subject
urban canopy. This approximation is established under the
low-wind-velocity condition. The computer performance-
based calculation for surface temperature prediction, which
excludes CFD simulation, is realized by this approxima-
tion. The convective heat transfer coefcient is estimated
by Ju rges formula.
3.7. Latent heat ux
Lewis relation with convective heat transfer coefcient is
used for the estimation of mass transfer coefcient, k.
Evaporation efciency b is derived from experiments for
each wet surface, including water retentive pavement.
3.8. Unsteady static heat conduction
One-dimensional heat conduction (Formula (2)) in each
spatial component is simulated, using the above-mentioned
heat balance data ( Q
G
), as the boundary condition for
external surface. The boundary conditions for inner side of
the surface are the indoor air temperature for the building
(Formula (3)) and the underground temperature for the
ground. The backward-difference method is used for the
calculation of the unsteady static heat conduction. One
running simulation runs for ve days in order to obtain a
periodic steady-state solution with initial conditions of
periodic weather data. The calculation results of surface
temperature for the 5th day are used as the output.
3.9. Tree calculation model
The tree shape is modeled by the 3D-CAD, and the
crown is lled with meshes containing solar transmittance
data, in order to simulate the inuences of solar permeating
position and distance, in the crown, on the amount of solar
transmission (Fig. 6). The solar transmittance data per
mesh are determined by the formula shown in Fig. 6. The
measurement result of solar radiation under each sort of
tree can be used for the estimation of the value of solar
transmission, S
TR
. In the present study, the tree model does
not deal with the distribution of the solar transmittance
data in the crown. The surface temperature of a trees
crown is calculated by an empirical formula that is derived
using the parameters of solar radiation, air temperature,
and wind velocity [13].
3.10. Weather condition data
The weather condition data used for this calculation
consist of the vertical quantity of total solar radiation, air
temperature, relative humidity, wind velocity, and cloud
amount. Separation of direct and diffuse components of
solar radiation is implemented for the total solar radiation
derived from observed weather data. Bouguers formula
for direct solar radiation and Nagatas formula for sky
solar radiation are applied. Public observation data,
measured by a weather station or Stevenson screen type
apparatus, for example, can be used for this simulation.
3.11. Calculation results of heat balance and evaluation
indices
Fig. 7 shows the calculation results of heat uxes on
building walls, roof, and ground. The characteristics of
each heat ux factor on respective surfaces can be
simulated [14].
This simulation tool evaluates the impact of a proposed
building or urban block on the outdoor thermal environ-
ment by two indices based on surface temperature. One
index is the mean radiant temperature (MRT) for
evaluating the thermal radiant eld, and the other is the
heat island potential (HIP) [12], which indicates the total
sensible heat from the entire surface of the urban block
being analyzed. HIP is expressed as the measure of a
temperature, and is calculated by the following:
HIP
R
all_surfaces
T
s
T
a
dS
A
, (5)
where HIP is heat island potential (1C), T
s
is surface
temperature of each mesh (1C), T
a
is air temperature
ARTICLE IN PRESS
Solar transmittance per mesh: =
M
S
TR
/ 100
M : Number
of meshes
Solar transmission S
TR
%
at culmination altitude (hour)
Direct solar radiation 100%
Fig. 6. Tree mesh model.
T. Asawa et al. / Building and Environment 43 (2008) 21122123 2117
(derived from weather condition data) (1C), A is plane area
of the urban block (m
2
), and dS is area of each mesh (m
2
).
4. Establishment of the simulation parameters and the
system
4.1. Optimum mesh size investigation
This section describes the results obtained from the mesh
size optimization experiments. A small mesh size is
necessary in order to accurately reproduce detailed spatial
components, such as, eaves, verandas, and the shape of tree
crown. On the other hand, a small mesh size signicantly
increases the calculation load. Therefore, mesh size
optimization was conducted.
A building with a small and complex spatial geometry is
appropriate for the examination of the mesh dependence,
so that a detached house having verandas and eaves was
chosen as the building model for the present examination.
The net solar radiation quantity was calculated for
several different mesh size models. Figs. 8 and 9 shows the
relationship between the applied mesh size and the
calculation time. The mesh size was varied between 0.1
and 1.0 m. This gure shows that detailed spatial forms,
such as eaves and a wing wall, cannot be sufciently
reproduced when a 1.0 m mesh is applied. The calculation
time for the 0.2 m mesh is approximately the same as that
for the 0.5 m mesh. Although the 0.1 m mesh results in an
accurate reproduction of the building details, the calcula-
tion time is long. Therefore, the 0.2 m mesh is suitable for
use in the simulation of subjects having detailed spatial
ARTICLE IN PRESS
-200
-100
0
100
200
300
400
0 2 4 6 8 10 12 14 16 18 20 22
-400
-200
0
200
400
600
800
H
e
a
t

f
l
u
x

[
W
/
m
2
]
-400
-200
0
200
400
600
800
H
e
a
t

f
l
u
x

[
W
/
m
2
]
R
S
R
L
Q
H
Q
G
H
e
a
t

f
l
u
x

[
W
/
m
2
]
-200
-100
0
100
200
300
400
H
e
a
t

f
l
u
x

[
W
/
m
2
]
-200
-100
0
100
200
300
400
H
e
a
t

f
l
u
x

[
W
/
m
2
]
-200
-100
0
100
200
300
400
H
e
a
t

f
l
u
x

[
W
/
m
2
]
Time
0 2 4 6 8 10 12 14 16 18 20 22
Time
0 2 4 6 8 10 12 14 16 18 20 22
Time
0 2 4 6 8 10 12 14 16 18 20 22
Time
0 2 4 6 8 10 12 14 16 18 20 22
Time
0 2 4 6 8 10 12 14 16 18 20 22
Time
R
S
R
L
Q
H
Q
G
R
S
R
L
Q
H
Q
G
R
S
R
L
Q
H
Q
G
R
S
R
L
Q
H
Q
G
R
S
R
L
Q
H
Q
G
Fig. 7. Simulation results of heat uxes on each surface (single building model, H: 5 m, W: 5 m, wall: concrete wall with no thermal insulating material,
ground: concrete paving, the measurement point is the center of each surface). (a) Southern wall; (b) northern wall; (c) eastern wall; (d) western wall;
(e) at roof; (f) ground.
T. Asawa et al. / Building and Environment 43 (2008) 21122123 2118
geometry. This size allows for fast simulation while
accurately reproducing detailed outdoor spatial geometry.
Next, Fig. 10 shows the relationship between the mesh
size and the simulation results for the evaluation index,
HIP, which is an index of the mean surface temperature of
the subject urban block (Formula (5)). The mesh size is
varied between 0.1 and 1.0 m, and the 0.1 m mesh is used
for the comparison standard. The urban block model used
in this examination is an apartment housing area in a
substantial urban district in Tokyo (Fig. 9). These results
show that the differences in HIP among the examined mesh
sizes are 0.5 1C or less. It is conrmed that the mesh size
does not signicantly inuence the simulation results of
HIP when a mesh size of less than 1 m is used. Based on the
results of this section, if an application requires a surface
temperature output with a high-resolution distribution,
then the optimum mesh size is 0.2 m, as mentioned above.
However, the larger mesh size is applicable for the
evaluation of HIP index. The same results were obtained
by examinations using other urban blocks.
4.2. Optimum number of tracers in the multi-tracing
simulation
The increase in the number of tracers in the multi-tracing
simulation used in the estimation of the sky view factor and
ARTICLE IN PRESS
0
100
200
300
400
500
600
700
800
C
a
l
c
u
l
a
t
i
o
n

t
i
m
e

[
s
e
c
]
0.1 0.2 0.5 1.0
Mesh size [m]
3D-CAD model
Fig. 8. Relationship between the applied mesh size and the calculation time.
20 40m 0
NN
Fig. 9. Apartment housing model.
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0 2 4 6 8 10 12 14 16 18 20 22
D
e
f
f
e
r
e
n
c
e

o
f

H
I
P

[

C
]
100 - 200mm
100 - 500mm
100 - 700mm
100 - 1000mm
Time [hour]
Fig. 10. Difference of HIP for each mesh size (difference between 100 mm
mesh and each mesh size).
T. Asawa et al. / Building and Environment 43 (2008) 21122123 2119
radiative heat transfer causes the growth of the calculation
load, so that the optimum number should be determined
for practical use. In this section, we examine the relation-
ship between the number of tracers in the multi-tracing
simulation and the calculation accuracy using the sky view
factor, which is signicant with respect to outdoor heat
balance, as the evaluation standard index.
The urban block model used in this examination is the
same as that shown in Fig. 9. The root mean square
error (RMSE) index is used for this investigation of
calculation accuracy (Formula (6)). The maximum number
of tracers for calculating the RMSE, as the standard, is
over 63,000.
RMSE

P
N
n1
sf_e sf
63275

2
N
s
, (6)
where sf
63275
is sky view factor using the maximum
number of tracer (63,275), sf_e is sky view factor
using each tracer number, N is number of meshes,
RMSE is the calculation error index (root mean square
error).
The relationship between the number of tracers and the
RMSE index of the sky view factor for the whole
of the ground is shown in Fig. 11. As the number
of tracers increases, the value of the RMSE
decreases sharply. However, the difference in the
RMSE is small when using over approximately 500 tracers.
When 524 tracers are applied, the RMSE value is
less than 1%. The image in Fig. 11 shows the simulation
result for the sky view factor distribution on the ground
using 524 tracers. A clear distribution of the sky view
factor is calculated. This result conrms that for this multi-
tracing simulation the suitable number of tracers is more
than 500. The number is practical in terms of the
calculation load.
5. Integration of the simulation model with the 3D-CAD
system
5.1. Pre-processing method using the 3D-CAD system
In order to develop a practical simulation tool for
widespread application, the input and pre-processing
method of this tool are developed using the features of
the 3D-CAD system (VectorWorks) and the GUI. The 3D-
CAD software package includes features for making 3D-
object surface models, and to perform effective texture
rendering onto objects of complex geometry, including the
building envelop.
Buildings and trees are drawn using 3D objects (Fig. 1).
The drawing line style is selected as either single line or
double line containing the building member section.
Ground cover is drawn using 2D objects. Component
materials and building members, such as walls, roofs, and
veranda, can be input or selected by using various dialog
boxes and databases.
The components of each database are shown in Fig. 3.
The following data are stored in the Spatial Component
Database: (a) buildings: building structures, building
members, and component materials; (b) trees: species and
solar transmittance; and (c) ground: ground cover and its
inner component.
The parameters for the heat transfer calculation,
including the physical properties of materials and the heat
transfer calculation models, are then automatically selected
and determined from the aforementioned parameters and
the following databases. The physical properties of
materials are stored in a Material Database. Objects
that require specic modeling of heat transfer, for example,
rooftop lawns and water-permeable pavement, are also
included in the Heat Transfer Calculation Model Data-
base.
5.2. Method of generating the 3D-mesh model for
calculation
The CAD models generated by the above process are
then transformed into a 3Dvoxel mesh model that
includes the calculation parameters required for heat
transfer analysis. Fig. 12 illustrates a schematic diagram
of the conversion method. The 3D-CAD model is sectioned
horizontally at a certain height, and a horizontal section of
the CAD model is then generated. The 2D gure data of
the section are then transformed into mesh data. This
process is repeated automatically from the bottom to the
top of the CAD model at an interval of the mesh size.
Consequently, the 3Dvoxel mesh model is completed. The
calculation point is set for all of the meshes, and the
calculation parameters (component materials, normal
direction of the surface, etc.) input during the aforemen-
tioned process are automatically stored in the 3D-mesh
model. Tree mesh models are also generated by the same
process.
ARTICLE IN PRESS
0
0.5
1
1.5
2
2.5
15 30 45 60 75 90 0
76
150
524
250
1124
5006
10539
The number of the tracer
Sky View Factor distribution
Calculation time [min]
R
M
S
E

o
f

S
k
y

V
i
e
w

F
a
c
t
o
r

[
%
]
Fig. 11. Relationship between the number of tracer and RMSE of sky
view factor.
T. Asawa et al. / Building and Environment 43 (2008) 21122123 2120
In the post-process of this system, the mesh data of the
surface temperature calculation results are converted into
surface texture data, and the textures are visually projected
onto the 3D-CAD model generated in the pre-process.
6. Simulation tool application
In order to conrm the applicability of this simulation
tool for the purpose of thermal design of outdoor spaces,
this tool was applied to an area containing detached houses
surrounded by numerous trees with leafy canopies.
A site containing ve two-story houses in the Kanto region
was chosen so that the effect of tall trees on controlling the
thermal environment could be examined and the applicability
of the tool to microscale thermal environments could be
evaluated. The weather conditions used in the calculation
were for a summer day with clear skies.
The calculation was performed for three different
scenarios, and CAD models for each case are depicted in
Fig. 13. In the rst scenario, there are no trees in the area
(CASE 1). CASE 2 is a reproduction of the present
condition of the subject area, which contains numerous
trees. CASE 2 and CASE 3 differ with respect to the
number and position of trees. CASE 3 includes several
trees that are taller than the nearby houses. The
morphology of the trees is modeled on the Zelkova genus.
The solar transmittance data of the tree crown is derived
from the measurement of the amount of solar transmission
under the trees in the residential area modeled here. The
calculation time for the residential area considered in
this section was approximately 1 h (Pentium 4: 2 GHz,
RAM: 1 Gb).
Fig. 14 shows a birds-eye view of the surface
temperature distribution depicted in the 3D-CAD
model, as well as a perspective depiction of surface
temperature distributions (12:00). The differences in
surface temperature among respective surfaces, including
slope roofs, walls, and wooden decks are clearly illustrated.
ARTICLE IN PRESS
3D-CAD
model
1. The 3D-CAD model is
sectioned horizontally at
a certain height.
2. The 2D figure data of the section
is then transformed into mesh data.
4. The 3D-voxel mesh model is completed.
3. This process is
repeated automatically
from the bottom to the
top of the CAD model.
Horizontal section
3D-mesh
model
Fig. 12. Schematic diagram of the mesh model generation.
<CASE1: With no trees> <CASE3: With many tall trees> <CASE2: The present condition>
N
Fig. 13. CAD models of the residential area for calculation (three cases).
T. Asawa et al. / Building and Environment 43 (2008) 21122123 2121
In CASE 1, which has no trees, the surface temperatures of
roofs and the ground are observed to exceed 50 1C
due to direct exposure to solar radiation. Conversely,
in CASE2 and CASE 3, the effect of shade from
the tall trees results in numerous surfaces having tempera-
tures that are approximately equivalent to the ambient air
temperature.
Fig. 15 depicts the MRT distribution at a height of 1.5 m
for CASE 2. The MRT under the tall trees is 32 1C, which
is approximately 10 1C lower than that observed in areas
exposed to solar radiation.
Fig. 16 shows the diurnal variation of the HIP. This
gure indicates that the HIP drops markedly due to the
presence of tall trees. The difference in the HIP between
CASE 1 and CASE 3 at noon is approximately 15 1C,
which is approximately the same as the difference
estimated between asphalt-paved ground and lawn ground,
also illustrated in Fig. 16.
These results reveal that the tool is able to simulate the
effects of building shapes, materials and trees shade on the
surface temperature distribution, and the MRT and HIP
indexes. Therefore, this simulation tool has been demon-
strated to be useful in evaluating the impact of a proposed
design on the urban thermal environment by using the
results of surface temperature.
ARTICLE IN PRESS
CASE1 CASE2 CASE3
CASE1 CASE2 CASE3
MRT40C
MRT36C
MRT32C
25 35 45 55 65[C] 25 35 45 55 65[C]
25 35 45 55 65[C]
Air temp. Air temp. Air temp.
N
Fig. 14. Simulation results of the surface temperature of the residential area (clear sky day in summer, 12:00).
30 32 34 36 38 40
Mean Radiant Temperature [C ]
CASE2 12:00
Air temp.:32.3 C
N
Fig. 15. MRT distribution at a height of 1.5 m.
T. Asawa et al. / Building and Environment 43 (2008) 21122123 2122
7. Conclusion
This paper details the development of a simulation tool
that will enable the user to predict the surface temperature
distribution of buildings and outdoor spaces using a 3D-
CAD software as a thermal design tool. In order to allow
for prediction taking into account detailed outdoor spatial
geometry and to allow the system to be put into practical
use, the following features were examined and incorporated
into this simulation tool: a heat balance simulation
algorithm with high resolution mesh model, optimum
mesh size, a method of multi-tracing simulation for sky
view factor and radiative heat transfer, an input and pre-
processing method that uses a 3D-CAD system.
The results of the application of this simulation
tool in an area of detached houses conrms that the
simulation tool is able to simulate the effects of spatial
geometry and materials existing in the outdoor space,
including the effect of shading by tall trees, on the surface
temperatures, and evaluation indexes of outdoor thermal
environments.
The examination using a broader range of examples of
urban areas will be conducted in a future study. Future
research will focus on the introduction of a latent heat
calculation model to evaluate the inuence of rainfall and
evaporative cooling in detail. In addition, this system will
be combined with CFD model in order to integrate the
prediction of the air ow and air temperature distribution
into this system.
Acknowledgments
This study was supported in part by a Grant-in-Aid for
Scientic Research (B), no.14350313 (2002, 2003) from the
Japan Society for the Promotion of Science, and by a
Grant-in-Aid (2003) from the New Energy and Industrial
Technology Development Organization (NEDO).
References
[1] Bruse M, Fleer H. Simulating surfaceplantair interactions inside
urban environments with a three-dimensional numerical model.
Environmental Modelling and Software 1998;13:37384.
[2] ENVI-met web homepage /http://envi-met.com/S.
[3] Toudert FA, Mayer H. Numerical study on the effects of aspect ratio
and orientation of an urban street canyon on outdoor thermal
comfort in hot and dry climate. Building and Environment
2006;41:94108.
[4] Wong NH, et al. Environmental study of the impact of greenery in an
institutional campus in the tropics. Building and Environment
2007;42:294970.
[5] Tanimoto J, Hagishima A, Chimklai P. An approach for coupled
simulation of building thermal effects and urban climatology. Energy
and Buildings 2004;36:78193.
[6] Bazjanac V. Building energy performance simulation as part of
interoperable software environments. Building and Environment
2004;39:87983.
[7] Yan D, Jiang Y. An overview of an integrated building simulation
tooldesigners simulation toolkit (DEST). In: Proceedings of ninth
international IBPSA conference, 2005. p. 1393400.
[8] Still M. Development of base efciencies in building environment
simulation for building simulation 2003 conference. In: Proceedings
of eighth international IBPSA conference, 2003. p. 12516.
[9] Hoyano A, Koga K, Yamashita J. Analysis for sensible heat ux in
hillside development area by using side looking airborne MSS data.
Journal of Architecture and Planning (Transactions of AIJ)
1994;462:4958.
[10] Hoyano A, Iino A, Koga K, Sano H. Quantication of heat island
potential of hillside development areas using side-looking airborne
MSS data and its relationship to land cover conditions. Journal of
Architecture and Planning (Transactions of AIJ) 1995;471:2937.
[11] Asano K, Hoyano A. Application of a new spherical thermography
technique to monitoring outdoor long wave radiation elds.
Proceeding of SPIE, the Society of Photo-Optical Instrumentation
Engineers 1998;3436:31724.
[12] Iino A, Hoyano A. Development of a method to predict the heat
island potential using remote sensing and GIS data. Energy and
Building 1996;23:199205.
[13] Shimokawa K, Katayama T, Hayashi T, Tanimoto J, Ka H, Ikezawa
N. Prediction of thermal environment on streets with roadside trees
part 5. A study on thermal characteristics of the foliage crown by
measurements. Summaries of Technical Papers of Annual Meeting
(Environmental Engineering), Architectural Institute of Japan
1996;8:1312.
[14] Nakamura Y, Hiraoka H, Nishimura H. Field investigation on
surface temperature and surface heat uxes in an urban cavity space.
Journal of Architecture and Planning (Transactions of AIJ)
1986;364:5765.
ARTICLE IN PRESS
-5
0
5
10
15
20
25
30
0 12 15 18 21
CASE1 :No trees
CASE2 :Present
condition
Asphalt paved
Lawn
H
e
a
t

I
s
l
a
n
d

P
o
t
e
n
t
i
a
l

H
I
P

[

C
]
ground
ground
Time
3 6 9
CASE3 :Tall trees
Fig. 16. Diurnal variation of HIP (clear sky day in summer).
T. Asawa et al. / Building and Environment 43 (2008) 21122123 2123