Professional Documents
Culture Documents
net/publication/257227002
A new parametric equation for the wind pressure coefficient for low-rise
buildings
CITATIONS READS
58 776
2 authors, including:
Ralph T Muehleisen
Argonne National Laboratory
109 PUBLICATIONS 480 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Ralph T Muehleisen on 22 January 2019.
Abstract
The building wind pressure coefficient (Cp) is an important quantity which is used in many
fields of building engineering including heating and cooling load calculations, ventilation design
and structural design. Cp is a dimensionless quantity that represents the proportionality between
the wind velocity and the pressure generated on the surface of the building. Values for Cp can
be obtained from full scale building tests, wind tunnel tests, or, more commonly, from parametric
equations derived from tests. The purpose of this paper is to analyze a set of wind tunnel tests
and present a new set of surface averaged wind pressure coefficient values for low-rise buildings
and a new parametric equation determined from a curve fit to the surface averaged data. The
resulting equations are compared to another popular low-rise parametric equation and another
popular wind pressure coefficient database. The new parametric equation is found to fit both
1. Introduction
Wind is a structural load on buildings and so it is very important in structural load analysis.
For simplified and analytical wind load calculation procedures, estimations of the wind pressure
coefficients, Cp, on the façade of the building are required. Most wind load building codes [1, 2]
suggest approximate mean pressure coefficients for common building shapes. In addition to wind
load calculations, wind pressure coefficients are also required in the estimation of air infiltration
into buildings [3]. Air infiltration is highly influenced by the indoor-outdoor pressure difference
which is the sum of three components: pressure difference due to the wind, pressure difference
due to stack effect, and the pressurization of the ventilation system. To estimate the wind
There are three primary methods which are usually used to estimate Cp: full-scale building
tests, wind tunnel tests, and parametric equations derived from experiments. Fully accurate
determination of the Cp for a particular building can be obtained only from full-scale tests [4, 5]
or wind tunnel tests [6, 7]. However, these kinds of tests are difficult, costly and require
significant time and expertise. For this reason, full-scale and wind-tunnel scale tests are usually
only used for very complex high-rise buildings or for the development of wind pressure
coefficient databases. The most common method for the prediction of Cp on low-rise buildings is
the use of parametric equations derived from measurements which can provide good and reliable
The most popular parametric equation for low rise buildings is probably that of Swami and
Chandra [10, 11]. Other authors extended this work to include the effects of nearby building
shielding [12, 13]. When shielding is taken into account, the number of variables can grow to be
quite large and developing good models becomes more difficult. Grosso’s parametric model
attempted to deal with all the possible variables, but the results were so complex that the author
himself stated that the method used was probably more interesting than the results obtained [14].
A review of several equations and methodologies was presented by Costola, Blocken and Hensen
[15]. For low-rise buildings, which constitute the overwhelming majority of all building
designed, what is needed is a set of simple to understand and implement parametric equations
that more accurately predict the performance of isolated buildings than the Swami and Chandra
equations . The effects of shielding by other buildings can then be included with simple
correction factors like the ones developed for the Swami and Chandra equation.
2. Background
Swami and Chandra developed equations for both low-rise and high-rise buildings using
step-wise regression analysis to a number of other previously published studies of wind pressure
coefficients, with the low-rise equation being a surfaced average normalized pressure coefficient.
The normalized pressure coefficient is the ratio of the actual average surface Cp to the Cp at zero
incidence angle, Cp(0). Cp(0) must be estimated by some other means, although Swami and
Chandra suggest that Cp(0)=0.6 can probably be used for all low rise buildings. The Swami
Cp Cp 0 ln 1.248 0.703sin 1.175sin 2 0.131sin 3 2G 0.769cos 0.07G 2 sin 2 0.717 cos 2 (1)
2 2 2 2
where is the incident angle of wind measured from the surface normal, and G=ln(S) is the
natural log of the side ratio S which is the ratio of building length to width. The overall
correlation coefficient for the regression was 0.797, indicating a fair, but not great, fit to the data.
It should be noted however that since Swami and Chandra were fitting to an amalgam of data,
rather than a single database, a correlation coefficient of 0.797 is actually quite good. Several
papers have incorrectly republished this equation as a simple expansion in terms of powers of
sin(/2) with the same coefficients seen above. The third, fifth, and seventh terms of the
equation can indeed be rewritten in terms of powers of sin(/2) using trigonometric identities,
but the but fourth cannot and, even if the other terms were rewritten, the coefficients would not
be the same.
One of the most popular databases of wind pressure coefficients is that published by the Air
Infiltration and Ventilation Centre (AIVC) [16]. The AIVC database is a combination of
different studies and presents Cp in tables as a single surface-averaged value for each face of the
rectangular and square buildings for wind directions from 0° to 315° in 45° increments. Tables of
data are available for exposed, semi-sheltered (surrounding obstacles with half of the building
height) and sheltered buildings (surrounding obstacles with the same height as the building) with
no more than three stories since this is the typical limit between low-rise and high-rise buildings.
The publically available part of the database is quite limited both in the number of side ratios and
The Tokyo Polytechnic University has recently released a large and very detailed database of
wind tunnel tests of a wide variety of low-rise buildings [17] as part of the Wind Effects on
Buildings and Urban Environment, the 21st Century Center of Excellence Program. The
database, developed primarily for structural engineering purposes, provides Cp values for a
larger range of side ratios, heights, and wind directions than the AIVC database. A variety of
data are available including surface averaged pressure coefficients, local pressure coefficient
contours, and the individual microphone time history pressure measurements. The surface
averaged pressure coefficient data are presented as plots of mean surface pressure as a function
of incident wind angle from 0° to 90° in 15° increments for all five building surfaces (four sides
and the roof), along with plots of the angle dependent RMS, maximum and minimum values.
With modern instrumentation, the Tokyo group has been able to make very detailed
measurements with a minimum of disturbance of the airflow, so we believe that this database
probably represents the most accurate set of wind tunnel tests ever published and we have thus
decided to use this database as the basis of our new parametric equation.
3. Methodology
Our new parametric equation presented was developed through curve fits to the low rise data
from the Tokyo database. The database included surface averaged wind pressure coefficient
graphs for each of the four sides of low rise buildings with depth-to-breadth short side ratios
D/B=1/1, 1.5/1, and 2.5/1 and depth-to-height ratios of D/H=1/1, 2/1, and 4/1 for wind angles of
0° to 90° in 15° increments. The geometry of the building as used in the Tokyo database and our
H B
B
D H
Wind D
Figure 1: Building geometry identifying the building sides 1-4, the building dimensions D, B, and H, and the wind
angle .
With this geometry, the wind angle is measured from the normal to side 1. Although the
model was only rotated 90, the symmetry of the building and measurements of surface pressure
on all 4 sides allows determination of the wind pressure coefficient for angles from 0° to 180°.
Since measurements were taken on all four sides, there is also some redundant data that can be
averaged for improved accuracy. For example, the wind pressure coefficient on side 2 and side 4
for a wind angle of =0° should be identical since the wind would be grazing along the surface
of both sides 2 and 4. Similarly, the pressure coefficients on side 1 and side 3 at a wind angle of
=90° should also be identical. We averaged the values of data that should have been identical.
Also note that by using all the measurements on side 2 and 4 one can obtain the results that
would have been obtained if the wind tunnel model would have been rotated and measured
again. This means that each measurement provides data for two different side ratios: data for
side ratio D/B comes from sides 1 and 3 and data for side ratio B/D comes from side 2 and 4,
meaning the measurements on three buildings, yields results for five different side ratios. The
resulting wind pressure coefficients obtained from side 1 and 3 measurements are shown in
Table 1 and the coefficients obtained from side 2 and 4 measurements are shown in Table 2. The
spatially averaged data from Tables 1 and 2 were then averaged over all height ratios to give
low-rise wind tunnel database measurements on sides 1 and 3. The wind angle is measured
from the normal to side 1. D is the length of sides 2 and 4 and B is the length of sides 1 and 3.
D/B D/H 0° 15° 30° 45° 60° 75° 90° 105° 120° 135° 150° 165° 180°
1.0 1.0 0.64 0.60 0.51 0.34 0.06 -0.31 -0.63 -0.71 -0.67 -0.55 -0.45 -0.36 -0.31
1.0 1.3 0.63 0.61 0.50 0.34 0.08 -0.28 -0.59 -0.71 -0.67 -0.56 -0.43 -0.34 -0.28
1.0 2.0 0.62 0.65 0.51 0.35 0.12 -0.18 -0.49 -0.66 -0.66 -0.51 -0.39 -0.32 -0.25
1.0 4.0 0.60 0.56 0.46 0.30 0.10 -0.16 -0.39 -0.54 -0.56 -0.47 -0.36 -0.30 -0.26
1.5 1.5 0.62 0.60 0.50 0.29 -0.06 -0.46 -0.71 -0.73 -0.63 -0.51 -0.45 -0.33 -0.25
1.5 2.0 0.61 0.60 0.48 0.26 -0.01 -0.36 -0.68 -0.73 -0.60 -0.49 -0.43 -0.31 -0.25
1.5 3.0 0.60 0.60 0.49 0.27 0.02 -0.26 -0.55 -0.65 -0.55 -0.43 -0.38 -0.29 -0.25
1.5 6.0 0.57 0.54 0.46 0.28 0.06 -0.20 -0.41 -0.52 -0.47 -0.39 -0.35 -0.30 -0.26
2.5 2.5 0.64 0.60 0.45 0.20 -0.18 -0.62 -0.83 -0.73 -0.58 -0.51 -0.41 -0.30 -0.22
2.5 3.3 0.66 0.62 0.48 0.21 -0.15 -0.53 -0.76 -0.75 -0.58 -0.48 -0.39 -0.28 -0.21
2.5 5.0 0.63 0.62 0.49 0.24 -0.07 -0.42 -0.70 -0.75 -0.58 -0.47 -0.37 -0.25 -0.22
2.5 10.0 0.61 0.57 0.46 0.26 0.00 -0.29 -0.51 -0.60 -0.49 -0.43 -0.39 -0.29 -0.22
Table 2: Cp as a function of wind angle , Side Ratio D/B, and Height Ratio D/H from Tokyo
low-rise wind tunnel database measurements on sides 2 and 4. The wind angle is measured
from the normal to side 2. D is the length of sides 2 and 4 and B is the length of sides 1 and 3.
D/B D/H 0° 15° 30° 45° 60° 75° 90° 105° 120° 135° 150° 165° 180°
0.4 1.0 0.68 0.66 0.58 0.44 0.19 -0.07 -0.34 -0.57 -0.70 -0.70 -0.60 -0.47 -0.40
0.4 1.3 0.61 0.61 0.54 0.41 0.18 -0.06 -0.29 -0.49 -0.65 -0.68 -0.59 -0.46 -0.38
0.4 2.0 0.60 0.59 0.51 0.41 0.20 -0.02 -0.25 -0.36 -0.52 -0.58 -0.53 -0.41 -0.35
0.4 4.0 0.58 0.57 0.50 0.37 0.18 -0.02 -0.19 -0.27 -0.36 -0.41 -0.40 -0.34 -0.33
0.67 1.0 0.63 0.63 0.56 0.40 0.15 -0.19 -0.51 -0.68 -0.71 -0.64 -0.50 -0.38 -0.31
0.67 1.3 0.63 0.59 0.52 0.37 0.14 -0.15 -0.47 -0.61 -0.71 -0.65 -0.48 -0.35 -0.31
0.67 2.0 0.60 0.60 0.52 0.40 0.16 -0.13 -0.39 -0.54 -0.62 -0.61 -0.42 -0.30 -0.28
0.67 4.0 0.58 0.57 0.49 0.35 0.14 -0.09 -0.30 -0.40 -0.45 -0.45 -0.35 -0.27 -0.25
1.0 1.0 0.64 0.62 0.56 0.37 0.07 -0.32 -0.70 -0.73 -0.68 -0.57 -0.44 -0.34 -0.26
1.0 1.3 0.66 0.66 0.55 0.38 0.09 -0.26 -0.62 -0.72 -0.68 -0.58 -0.43 -0.33 -0.26
1.0 2.0 0.69 0.65 0.55 0.37 0.13 -0.19 -0.49 -0.68 -0.66 -0.53 -0.37 -0.29 -0.23
1.0 4.0 0.63 0.60 0.49 0.34 0.16 -0.10 -0.38 -0.52 -0.54 -0.45 -0.34 -0.30 -0.26
Table 3: Height averaged Cp as a function of wind angle and Side Ratio D/B from Tokyo
Database.
D/B 0° 15° 30° 45° 60° 75° 90° 105° 120° 135° 150° 165° 180°
0.4 0.62 0.61 0.53 0.41 0.19 -0.05 -0.27 -0.43 -0.56 -0.59 -0.53 -0.42 -0.36
0.67 0.61 0.60 0.52 0.38 0.15 -0.14 -0.42 -0.55 -0.62 -0.59 -0.44 -0.33 -0.29
1 0.64 0.62 0.52 0.35 0.10 -0.22 -0.54 -0.66 -0.64 -0.53 -0.40 -0.32 -0.26
1.5 0.60 0.58 0.49 0.27 0.00 -0.32 -0.59 -0.66 -0.56 -0.45 -0.40 -0.31 -0.25
2.5 0.63 0.60 0.47 0.23 -0.10 -0.46 -0.70 -0.71 -0.56 -0.47 -0.39 -0.28 -0.22
The generation of height averaged Cp allows the Tokyo data to be compared with other
databases and to generate parametric equations that give height averaged and spatially averaged
Cp. The Swami and Chandra equation, Eqn. (1), and the height averaged Tokyo data of Table 3
are plotted in Fig. (2) for side ratios of S=0.4, 1.0, 1.5, and 2.0. As one can see, the Swami and
Chandra equations are not a particularly good fit to the Tokyo data, although the general trends
of the data are captured by the equation. The discrepancy between the Swami and Chandra
equation and such an accurate database is another reason to try to develop a new parametric
equation.
0.5
Cp
-0.5
0 50 100 150
Wind Angle
Figure 2: Plot of Swami and Chandra equation, Eqn. (1), and height averaged Cp from the Tokyo Database. Eqn. (1)
for S=0.4: S=1.0: , S=1.5 – –, S=2.5 - - ; Tokyo data for S=0.4: , S=1.0-: , S=1.5: , S=2.5: .
4. Results
The data of Table 3 were analyzed using the Systat TableCurve3d software package [18]
which implements a robust non-linear curve fitting routine. TableCurve3d uses a Levenburg-
Marquardt algorithm and ranks the fits according to a variety of fitting statistics including the
coefficient-of-fit R2, the F-statistic, the minimum mean square error, or other minimized error
functions. The Tokyo database values were fitted using standard least squares minimization and
where G=ln(S) and S=D/B is the side ratio, is the angle of wind incidence in degrees, and the
coefficients ai and bi are found in Table 4. The regression statistics are excellent with a
Table 4: Coefficients ai and bi for Eqn. (2) and the standardized fit error.
a0 6.12E-01 1.44E-02
a1 -1.78E-01 1.49E-02
a2 -1.15E-02 3.73E-04
a3 3.28E-05 2.56E-06
a4 1.67E-03 2.42E-04
b1 -3.12E-01 2.48E-02
b3 9.82E-05 5.20E-06
b4 2.15E-03 3.58E-04
A three-dimensional surface plot of Eqn. (2) is shown in Fig. 3 and line plots for select
0.4
1
0.2
0.5
0
Cp
0
-0.2
-0.5
-0.4
-1 3
0
2 -0.6
50
100 1
150 D/B
0 atio S=
Side R
Wind Angle
Figure 3: Three dimensional plot of the new parametric equation, Eqn (2), for spatially averaged Cp in low rise
buildings as a function of side ratio S and wind angle .
0.6 S=0.25
0.4 S=0.5
S=1.0
0.2 S=1.5
S=2.5
Cp
-0.2
-0.4
-0.6
0 20 40 60 80 100 120 140 160 180
Wind Angle
Figure 4: Plot of Eqn. 2 for S=0.25: , S=0.5: – –, S=1.0: , S=1.5: – – , S=2.5: – –.
A plot of Eqn (2) along with the spatially averaged Cp values from the Tokyo database for
S = 0.4, 1, 1.5, and 2.0 is shown in Fig. 5. The improved quality of the fit of the new equation
Cp
0
-0.5
0 50 100 150
Wind Angle
Figure 5: Plot of the new parametric equation, Eqn. (2), and surfaced averaged Cp from the Tokyo Database. Eqn.
(2) for S=0.4: S=1.0: , S=1.5 – –, S=2.5 - - Tokyo Cp data for S=0.4: , S=1.0-: , S=1.5: , S=2.5: .
The predictions of the equation are compared to the height averaged Tokyo low-rise
database, predictions from the Swami and Chandra low-rise equation and the AIVC database at
wind angles =0°, 45°, 90°, 135°, and 180° for a side ratio of S=1.0 in Table 5 and are plotted
in Figure 6.
Table 5: Comparison Between evaluation of Cp using Eqn. (1) , Swami and Chandra’s Equation,
and the values from the Tokyo and AIVC databases for side ratios S=D/B=1.0.
Cp
0
-0.5
0 50 100 150
Wind Angle
Figure 6 : Comparison of Cp values for S=1.0. Swami and Chandra, Eqn. (1): , new parametric Eqn. (2): ,
Tokyo surfaced average data: , and AIVC data: .
5. Discussion
The new rational equation, Eqn. (1), is a smooth function in both and G=ln(S). The
equation has a clear limit of Cp (a0+ a1G)/(1+ b1G) as 0 but does break down as S0.
However, it should be noted that S0 is not a valid geometry, indeed the practical lower limit of
S is probably about 0.1. With an R2=0.992, the new parametric equation is clearly a very good
representation of the height averaged Tokyo database which can also be seen in Fig 4. Fig. 3, the
three dimensional plot of Eqn. (2) can be used for rough prediction of Cp for a wide variety of S
and while Fig. 4 can be used for more accurate determination of Cp for a wide range of but
limited values of S. Fig. 5 shows that both equations fit the Tokyo database measurements well
for S=1.0; indeed the Swami and Chandra appear to be a better fit for 90 < < 165. This might
explain the continued popularity of the Swami and Chandra equations because S=1.0 is such
commonly used side ratio. However, the new parametric equations seem be a better match the
AIVC database than the Swami and Chandra equation for S=1.0.
6. Example
As an example of how this new equation might be used and how much of a difference the
new equation can make, consider finding the wind pressure coefficients on all four sides of a
low-rise building with a breadth of 20 m and a depth of 30 m and wind incident at an angle of 30
degree from normal on one of the 20 m sides. Referring to Fig. 1, we find Cp for side 1 by using
D/B=30/20=3/2 and =30; we find Cp for side 3 using D/B=3/2 and =150; we find Cp for
side 2 using D/B=20/30=2/3 and =120; and we find Cp for side 4 using D/B=2/3 and =60.
The resulting Cp directly from the Tokyo database (Table 3), the Swami and Chandra equation
and the new parametric equation are shown in Table 6. While both the new parametric equation
and the Swami and Chandra equation match the Tokyo database for side 1, the Swami and
Chandra equation is significantly different from the Tokyo database for sides 2, 3, and 4. Using
the Tokyo database as a reference, we can see that the new parametric equation is far more
accurate than Swami and Chandra for the typical use of determining wind pressure coefficients
of a low-rise building.
Table 6: Example of computing Cp for a 3020 m low-rise building with wind incident at a 30 angle from the
normal of Side 1. The new parametric equation values and Swami and Chandra equation are compared to the measured
values of Tokyo database.
The purpose of this work was to analyze the Tokyo wind pressure coefficient database to
create a new set of spatially averaged wind pressure coefficients for low-rise buildings and to
develop a new predictive equation based on the new coefficients. The spatially averaged wind
pressure coefficients with side ratios of S=0.4, S=0.667, S=1.0, S=1.5 and S=2.0 over a range of
incident wind energy angles of 0° to 180° were averaged for several height ratios. These data
were used to generate a new rational polynomial parametric equation for the prediction of
spatially-averaged wind pressure coefficients for low-rise buildings. The new parametric
equation fit the Tokyo database values with a goodness-of-fit R2=0.992. Compared to the popular
equation of Swami and Chandra, the new equation provides a better match to both the Tokyo
database coefficients and to the AIVC database coefficients and is easier to calculate by hand or
with a spreadsheet than the Swami and Chandra equations. Used in a typical example, the new
parametric equation is also shown to be more accurate than the Swami and Chandra equation.
The new parametric equation should be useful to both structural and HVAC engineers and easy
[1] A.S. 7-05:, Minimum Design Loads for Buildings and Other Structures, American
[2] EN 1991-1-4:2005, Eurocode 1: Actions on structures - Part 1-4: General actions - Wind
[3] ASHRAE, Ventilation and Infiltration, in: 2009 ASHRAE Handbook: Fundamentals,
2009.
[4] J.D. Ginger, C.W. Letchford, Net pressures on a low-rise full-scale building, Journal of
pressures and response accelerations of a high-rise building, Journal of Wind Engineering and
[7] M. Hussain, B. Lee, A wind tunnel study of the mean pressure forces acting on large
groups of low-rise buildings, Journal of Wind Engineering and Industrial Aerodynamics, 6 (3-4)
(1980) 207-225.
[8] G. Walton, Airflow and multiroom thermal analysis, ASHRAE Transactions, 88 (1982)
78-91.
[9] I.S. Walker, D.J. Wilson, Practical methods for improving estimates of natural ventilation
rates, in: The Role of Ventilation - 15th AIVC Conference, Buxton, UK, 1994, pp. 517-517.
[10] M.V. Swami, S. Chandra, Procedures for calculating natural ventilation airflow rates in
[11] M.V. Swami, S. Chandra, Correlations for pressure distribution on buildings and
[12] T. Sawachi, E. Maruta, Y. Takahashi, K. Sato, Wind Pressure Coefficients for Different
[14] M. Grosso, Wind pressure distribution around buildings: a parametrical model, Energy
building energy simulation and airflow network programs, Building and Environment, 44 (10)
(2009) 2027-2036.
[16] M.L. Orme, N. Leksmono, AIVC Guide 5: Ventilation Modelling Data Guide, AIVC,
2002.
Accessed 04-Mar-2011,
http://www.wind.arch.t-kougei.ac.jp/info_center/windpressure/lowrise/mainpage.html.