Académique Documents
Professionnel Documents
Culture Documents
S
b
I
p
f
b
S
j d
I
M
j
j S
F
d
i
T
N
i
i S
F MRT +
=
+
=
= (1)
The first term on the right hand side of equation (1) sums the product of the view factor from the
subject to each surface (F
Si
) and the surface temperature (T
i
) to the fourth power for all surfaces
in the vicinity of the subject. This term can be simplified by assuming that all surroundings are at
ambient temperature, which results in a view factor of 1, and allows us to use the dry bulb tem-
perature provided in meteorological data.
The contribution of the diffuse radiation and direct radiation in the second and third terms re-
spectively are calculated in a similar manner. In each case, the incoming diffuse and direct radia-
tion (I
d
and I
b
respectively) are divided by the Stephan Boltzmann constant () and multiplied by
the surface absorption coefficient over the surface emissivity (/), the latter of which depends on
the properties of both the source and receiver. In this work, published values for / between the
sun and human caucasian skin (Incropera and DeWitt, 2002) are used since knowledge of cloth-
ing color is not known. Because this model was created for indoor environments, it is formulated
to allow for more than one source of diffuse radiation (i.e. multiple windows), which are summed
with separate view factors and diffuse radiation values. In an external environment, we can as-
sume a single source (the sky), where the view factor is that which is predicted using Ecotect,
and the method in section 2.2. Likewise, the view factor for the direct radiation is either one or
zero based on the assumption of a singular reception point as shown in the previous section.
The intensity of direct and diffuse radiation depends on the orientation of the absorbing sur-
face. By approximating the shape of the human body with a cylinder, we can predict the average
intensity of solar radiation on the surface of the body. For direct radiation, the intensity on the
surface in W/m
2
is found using Lamberts cosine law (equation 2), where I
b,N
is the direct normal
radiation, defined as the radiation intensity on a surface directly facing the sun, and is the angle
of incidence to the surface. The zenith and altitude angles are thus used to find the direct radia-
tion I
b
, on the top and sides of the cylinder respectively:
) cos(
,
N b
I
b
I = (2)
For diffuse radiation, meteorological data typically provides intensity values for horizontal
surfaces. Stephenson (1965) and Threlkeld (1963) showed that the clear-sky diffuse irradiance on
a vertical surface can be expressed as a fraction Y of the horizontal diffuse irradiance according
to equation (3):
)
2
cos 313 . 0 cos 437 . 0 55 . 0 , 45 . 0 max( + + = Y (3)
where is the angle of incidence from the sun to the normal vector of the surface. The average
value for Y is calculated for the perimeter of a full cylinder for a given solar altitude. Once both
the direct and diffuse radiation are known on the top and sides of the cylinder, the total radiation
is area weighted over the full area of the cylinder to produce the second and third terms in equa-
tion (1). The final result is a measure of MRT for a given ambient temperature, solar altitude, di-
rect normal radiation, and diffuse horizontal radiation.
The Fifth International Symposium on Computational Wind Engineering (CWE2010)
Chapel Hill, North Carolina, USA May 23-27, 2010
3 PREDICTING AIR VELOCITY
3.1 Computational Fluid Dynamics
Computational Fluid Dynamics (CFD) is a modeling tool that can be used to predict a velocity
field in an outdoor domain for a given wind condition. This method is employed here to predict
point velocity for each study point in the current work. The challenge in predicting wind for an
entire year is the large variation seen in both wind direction and wind speed. In most wind stu-
dies, prevailing winds are analyzed to provide insight into the most common wind events, how-
ever for this work, a full statistical measure of comfort is desired, and as such all wind events
should be approximated to some degree. In order to achieve this, CFD simulations from multiple
wind directions around the compass are conducted. At its highest resolution, the method outlined
here has included 36 simulations from wind directions at 10 increments, to ensure that all poten-
tial wind directions are represented by a simulation within 5 of the recorded hourly mean value.
Solutions have been generated using the Openfoam CFD code, and Reynolds Averaged Navi-
er Stokes turbulence modeling. The domain inlet conditions are defined using boundary layer
profiles as defined by Richard and Hoxey (1993), which assume a constant shear stress with
height. Velocity, turbulent kinetic energy, and turbulent dissipation profiles are defined based on
the atmospheric boundary layer friction velocity and the aerodynamic roughness length. For all
simulated wind directions, the geostropic velocity is set to a constant value calculated based on
the approximate average velocity measured at 10m, and scaled up to 600m based on ESDU (En-
gineering Science Data Unit) profiles corresponding to the roughness of the terrain surrounding
the measurement station.
Convergence is checked by probing velocities at all study locations and iterating until no sig-
nificant variation is seen. Velocity, pressure and turbulence factors see a minimum drop in resi-
duals of five orders of magnitude. Grid independence is not attainable due to the massive mesh
sizes required to represent developments and all of their surroundings. In most studies, the initial
mesh size is on the order of 5 millions cells. To achieve one level of refinement would require 40
million cells, which goes beyond the computational capacity available at RWDI.
CFD results generated for the full outdoor domain are expressed as a velocity ratio to that
measured at the meteorological station for each wind direction simulated. Using the same point
locations for the analysis performed in section 2, lookup tables are generated for each point by in-
terpolating the steady state CFD results. These lookup tables provide a fast method of producing
airspeed predictions at the study points for each combination of wind direction and wind speed
provided in the meteorological data.
3.2 Wind tunnel measurements
Velocity ratios for each wind direction can also be predicted using a boundary layer wind tunnel
and a scale model of the development and its surroundings. All wind tunnel tests conducted at
RWDI follow the requirements described in the ASCE manual of practice no. 67 on wind tunnel
studies of buildings and other structures (ASCE 1999), where artificial roughness is added to the
upstream portion of the tunnel to produce scaled profiles of velocity and turbulence at the inlet to
the study area. Results are produced for multiple wind directions by rotating the physical model
with respect to the incoming air flow. A constant geostrophic velocity is used for each wind di-
rection. Irwin sensors (Irwin, 1981) are placed at the same study points as outlined in section 2 to
produce mean velocities near the ground. While gusts speeds are typically used to evaluate wind
comfort due to mechanical forces, only the mean values are required for thermal comfort predic-
tions (Soligo et. al., 1998). Lookup tables for velocity ratios are produced at each point similar to
the previous section. For this work, CFD results are preferred because of their ability to produce
site-wide contours of wind speed, rather than discrete point measurements.
The Fifth International Symposium on Computational Wind Engineering (CWE2010)
Chapel Hill, North Carolina, USA May 23-27, 2010
4 RESULTS
The method of calculating SPMV* from the four environmental inputs is an iterative process
which is outlined in the work by Gagge et. al. (1986). An automated process is used to produce
the four environmental inputs from the meteorological data and the methods outlined above, fol-
lowed by the iterative calculation of SPMV* for each hour of the year. The process is repeated
for multiple study points, giving both a time and space dependent solution to thermal comfort.
With the exception of the CFD simulations, the calculations are simplified to the point where all
SPMV* measurements for the 21 points for the full year can be produced in less than one hour.
Post processing focuses on the time and type of use of the study area. For example, the given
case study is assumed to be a college campus, which houses students from fall to spring. Based
on past experience by RWDI consultants, it was assumed that a reasonable percentage of people
in the Abu Dhabi area will find an SPMV* rating of 2.5 or less acceptable for outdoor environ-
ments. There are no existing code values specifying a thermal comfort threshold, so the SPMV*
rating of 2.5 is based on solely on personal experiences, and surveys taken with Abu Dhabi resi-
dents by RWDI. Table 1 lists a subset of the results predicting the frequency of acceptable ther-
mal conditions for the southeast points as labeled in figure 1 for the fall and spring seasons.
Table 1. Frequency of comfort (% of total hours)
Study Point Spring Fall
Morn. Aft. Morn. Aft.
1 96 % 93% 85% 80%
2 98% 77% 96% 56%
3 97% 87% 87% 67%
4 96% 79% 80% 48%
5 100% 95% 99% 70%
A spatial view of comfort in the outdoor space is created by binning results by frequency of
comfort at the various study points. Figure 3 shows a sample result produced for fall afternoons
in the southeast region of the study area. The results are primarily used to provide guidance on
what building design features and surroundings contribute to the frequency of acceptable outdoor
conditions, and where mitigation may be required. In the example provided here, the highest fre-
quency of comfort is predicted in the narrower passageways where shading occurs often, and
wind can readily flow through the space as seen in points 1 and 5. The unfavorable conditions
seen above are a result of significantly reduced shading, and stagnant air in spaces where wind
throughways are not available as seen in points 2 and 4. The moderate locations are typically cha-
racterized by a balance between the two factors, as in point 3 where shading is poor, but condi-
tions are moderated due to the open flow passage through the center of the courtyard.
Figure 3. Frequency of thermal comfort tolerability for all afternoon hours in fall months
Predicted Comfort Frequency
20% - 30%
30% - 40%
40% - 50%
50% - 60%
60% - 70%
70% - 80%
80% - 90%
90% - 100%
The Fifth International Symposium on Computational Wind Engineering (CWE2010)
Chapel Hill, North Carolina, USA May 23-27, 2010
5 CONCLUSIONS
Multiphysical CFD models can be used to calculate the dry bulb temperature, air speed, humidi-
ty, and mean radiant temperature of an outdoor environment under a given set of conditions.
These four inputs provide the necessary information required to produce thermal comfort predic-
tions throughout an outdoor space. Such models serve as an effective method to evaluate condi-
tions at a fixed time, such as a worst case scenario. Because outdoor spaces can be subject to ex-
treme environmental fluctuations over the year, a small set of results may not show when or how
often a given space will meet comfort criteria. Repeating coupled CFD models to achieve predic-
tions for a full year would be impractical based on current computational capabilities.
The work described here provides methods and assumptions which simplify the interdepen-
dence of the environmental factors such that rapid predictions of thermal comfort can be pro-
duced while still maintaining a level of dependence on the local surroundings. The key simplifi-
cation is that of the mean radiant temperature, which is assumed to depend only on the dry bulb
temperature, solar radiation, and the location of the sun. This decoupling of the MRT from the air
velocities and surface radiation exchanges decreases the amount of information required from
CFD simulations to simple cold flow predictions of wind velocity ratios. With the exception of
the CFD simulations, which may take on the order of a few weeks, a full annual prediction of
thermal comfort for the 21 points presented here can be generated in less than one hour.
By providing annual statistical measures of temperature, humidity, air speed, and mean radiant
temperature, one of many thermal comfort indices can be used to create statistics which show the
frequency of comfort at various outdoor locations. This is of high value to masterplan designers
who can quickly produce preliminary estimates of which building design features may contribute
to increased frequency of thermal comfort.
6 REFERENCES
ASCE, 1999, Manual of practice for wind tunnel studies of buildings and structures. Manual No. 67., American So-
ciety of Civil Engineers.
ASHRAE, 2009, Fundamentals handbook. American Society of Heating, Refrigerating and Air-Conditioning Engi-
neers, Inc. Atlanta.
Bouyer, J., Vinvent, J., Delpech, P., and Carre, S., 2007. Thermal comfort assessment in semi-outdoor environments:
Application to comfort study in stadia. Journal of Wind Engineering and Industrial Aerodynamics 95, 963-976.
Fanger, P., 1970. Thermal comfort - analysis and applications in environmental engineering. McGraw-Hill Book
Co., New York.
Gagge, A.R., Fobelets, A.P., and Berglund, L.G., 1986. A standard predictive index of human response to the ther-
mal environment. ASHRAE Transaction-92, pt. 2B.
Incropera, F.P., DeWitt, D.P., 2004, Fundamentals of heat and mass transfer. John Wiley & Sons Inc., New York.
Irwin, H.P.A.H., 1981. A simple omnidirectional sensor for wind tunnel studies of pedestrian level winds. Journal of
Wind Engineering and Industrial Aerodynamics, 7, 219-239
Marion, W., 1995. New typical meteorological years and solar radiation data manual. PV Radiometric Workshop
Proceedings. Vail, Colorado.
Richards, P.J., and Hoxey, R.P., 1993. Appropriate boundary conditions for computational wind engineering models
using the k-e turbulence model. Journal of Wind Engineering and Industrial Aerodynamics, 46 & 47, 145-153.
Soligo, M.J., Irwin, P.A., Williams, C.J., Schuyler, G.D., 1998. A comprehensive assessment of pedestrian comfort
including thermal effects. Journal of Wind Engineering and Industrial Aerodynamics, 77 & 78, 753-756.
Stephenson, D.G., 1965. Equations for solar heat gain through windows. Solar Energy 9(2), 81-86.
Threlkeld, J.L., 1963. Solar irradiation of surfaces on clear days. ASHRAE Transactions-69, 24-36.
Wu, H., 1994 Pedestiran-level wind environment around buildings, Ph.D. dissertation, Concordia University, Mon-
treal, Canada.
Scudo, G., Dessi, V. Rogora, A., Evaluation of radiant conditions in urban spaces. Designing open spaces in the ur-
ban environment: a bioclimatic approach. CRES, 2004. Athens.