Professional Documents
Culture Documents
www.elsevier.com/locate/engstruct
Received 28 June 2003; received in revised form 1 March 2004; accepted 8 March 2004
Abstract
The mean interference effects between two and among three tall buildings are studied by a series of wind tunnel tests. Both the
shielding and channeling effects are discussed to understand the complexity of the multiple-building effects. The results show that
the upstream interfering buildings cause certain shielding effect by decreasing the mean wind load on the downstream principal
building. For buildings of the same height, the shielding effect increases and, therefore, the interference factor (IF) decreases, with
the increase of the breadth of the interfering buildings. However, due to the channeling effects, two adjacent interfering buildings
can significantly enhance the mean wind load on the principal building. In addition, the variation of the shielding effect is found
to be significant when the heights of interfering buildings range from 50% to 125% of the height of the principal building. How-
ever, higher interfering buildings may cause stronger channeling effects.
# 2004 Elsevier Ltd. All rights reserved.
Keywords: Tall buildings; Mean wind loads; Wind tunnel test; Interference effects; Channeling effects
Table 2
Comparison of the IFs of twin-building configuration
Interfering building at (x, y) (5b, 1.5b) (5b, 2.5b) (5b, 4b) (8b, 0) (8b, 1.5b) (5b, 2.5b) (8b, 4b)
Khanduri et al. [9] 0.78 0.90 1.0 0.57 0.74 0.86 0.99
Huang [11] 0.74 0.98 1.03 0.63 0.71 0.93 1.00
Present study 0.73 0.96 1.03a 0.57 0.64 0.88 1.02a
a
Denotes extrapolation, see Fig. 3.
rain, the more significant the shielding effects of the gory B. It can be seen that the minimum interference
upstream building. In addition, interference factors of factor is close to zero at smaller building spacing,
zero are seen in the figure at about 2b, 3b and 4b for which is slightly different from that of two-building
the three types of terrains, respectively. The position of configurations where IF equals to 0.2. In general, the
the zero value in exposure category B with a ¼ 0:16 is shielding effects are dominated by the nearer upstream
at a center-to-center spacing of 3b, that is almost the interfering building, especially at smaller spacing. Simi-
same as that of the result observed by Sakamoto and lar trends can be found for this configuration in
Haniu [1] under the similar terrain condition of open exposure category D.
terrain.
For the results of other building arrangements, 3.2.2. Side-by-side arrangement
Khanduri et al. [9] gave the IFs caused by an upstream Table 3 gives the interference factors of the three
building at the region of [2b–8b, 0–4b] in open terrain buildings in side-by-side configuration, where yA and
by means of synthesizing the results given by Taniike yB denote respectively the across-wind center-to-center
and Inaoka [10] and Saunders and Melbourne [6]. spacing, i.e. the across-wind coordinates defined in
Huang [11] also conducted some similar experiments in Fig. 3, of the two interfering buildings and the princi-
the same terrain of exposure category B. Table 2 lists pal building. It can be seen from the table that when
the above-mentioned results, together with the corre- the two interfering buildings are located at the same
sponding results that are predicted by the well-trained side of the principal building, they still can produce
neural network with the test data from the present shielding effects on the principal building, resulting in
study. The results show the good consistency and the an IF of 0.94. However, adverse effects of IF > 1 can
efficiency of the ANN-based method presented in this be found for most of the arrangements and the
paper. Of course, differences still can be found in the maximum IF is found to be 1.10 when the two inter-
table due to the use of different terrain categories and fering buildings are located at yA ¼ 3:2b and
building aspect ratios in different studies. yB ¼ 3:2b, respectively. This indicates that the two
Fig. 5 presents the IF contours for the two equal size symmetrically located interfering buildings can increase
buildings in uniform flow and exposure categories B 10% or even more wind load on the middle principal
and D. Clearly, the shielding effects vary with the building, that is to say, the channeling effect in this
degree of roughness of the upstream terrain. One can case is more significant than that of the two buildings
see that each contour has a negative region corre- in side-by-side arrangement.
sponding to the negative IF, which means that the
principal building is subjected to a reverse wind drag 3.2.3. Staggered arrangement
force. The negative IF region increases with the For the interference effects of three buildings, four
smoothness of the upstream terrain. variables (i.e. two x-coordinates and two y-coordinates
of the two interfering buildings) are included in each of
3.2. Results for configurations of three equal size the configurations and the results are very difficult to
buildings be expressed with simple contours. In this study, a sub-
stitute scheme is used to analyze the multi-variable test
3.2.1. Tandem arrangement results by fixing one interfering building (model A) at a
Compared with the mean interference effects of two certain position and varying the spacing between the
tandem-arranged buildings, the mean interference other interfering building (model B) and the principal
effects of the three tandem-arranged buildings (an building. An example is shown in Fig. 7 where the
arrangement of three buildings placed on behind the building model A is fixed at (6.1b, 2.4b).
other in the along-wind direction, i.e. yA ¼ 0 and yB ¼ In order to compare the interference factors of the
0 for the two interfering buildings, see Fig. 3) are more three-building configuration with those of the two-
significant than those of other kinds of arrangements. building configuration, model A is considered as an
Fig. 6 gives the distribution of the interference factors additional interfering building to the two-building con-
for three tandem-arranged buildings in exposure cate- figuration where the interference effects have been
Z.N. Xie, M. Gu / Engineering Structures 26 (2004) 1173–1183 1177
Fig. 5. IF contours for the configuration of two equal size buildings in different upstream terrains. (a) Uniform flow; (b) exposure category B; (c)
exposure category D.
shown in Fig. 5b. It can be found from the comparison centage of the positions of the corresponding inter-
results in Figs. 7 and 5b that the introduction of model ference factor over the whole test positions of the
A increases the most significant shielding region of configurations. From this figure, one can see that p is
IF 0:4 in the two-building configuration, with the 35% when IF is about equal to 1.0 for the two-building
maximum longitudinal spacing changing from 5b to configuration, but only about 13% for the three-build-
about 6b. The slight shielding region of 0:8 < IF 1 is ing configuration. In general, for different levels of
also broadened with the introduction of model A. IF 0:9, the value of p of three-building configuration
These results indicate that the shielding effects are is greater than that of two-building configuration.
enhanced in the three-building configuration. These results once again indicate that the shielding
However, the above-mentioned distribution in Fig. 7 effects of three-building configuration are more signifi-
is just a local description of the interference effects of cant than two-building configuration. However, due to
three-building configurations and cannot give the com- the channeling effect, the IF is found to be about 1.1
plete information of the interference effects for the con- for 2% of the complete set of interfering building
figuration. Statistics analysis for a thorough description arrangements, as shown in Fig. 8. That means that
of the interference effects is therefore needed and the there may be static amplification due to the existence
results are shown in Fig. 8, where p represents the per- of two nearby interfering buildings.
1178 Z.N. Xie, M. Gu / Engineering Structures 26 (2004) 1173–1183
Fig. 6. Interference factors of the three equal size and tandem-arranged buildings in exposure category B.
3.2.4. Channeling effect ings is located side-by-side with the principal building
The channeling effect was mentioned in ASCE 7-98 and the other one is arranged upstream, the inter-
[12] and other literature [13], but it has not been dis- ference factor can be 1.04, the same as that of the
cussed in any detail in the previous studies. The reason two-building configuration; and the region where
may partly be that this kind of effect is insignificant, IF 1:04 in exposure category B is much larger than
compared with the above-mentioned shielding effects. that in category D. It can be concluded that the chan-
However, the maximum IF is 1.04 in the present test in neling effect could be more significant in the smoother
the two-building configuration when the interfering terrain.
building is located side-by-side at (0, 3.2b). For the
configuration of three equal size buildings, the 3.3. Effects of breadth ratio
maximum IF can increase up to 1.10 when the two
interfering buildings are located at (0, 3.2b). To investigate the effects of the breadth ratio (here-
Fig. 9 presents the IF distributions from the present after referred to as Br) of across-section of the inter-
test for the two interfering buildings at y ¼ 3:2b in fering buildings to the principal building on the
exposure categories B and D. The most significant interference effects, five types of interfering building
interfering positions in the two categories of terrain are models with different breadths are tested. These inter-
found to be the same. From Fig. 9, one can see that fering models have the same height as the principal
two side-by-side upstream interfering buildings produce building model but with different breadths. The
almost no static amplification effect on the principal breadth ratios adopted in the test are 0.5, 0.75, 1.0, 1.5
building. Only when one of the two interfering build- and 2.0. The results are discussed in the following.
Fig. 7. Variations of the interference factor vs. the relative positions of interfering building model B for interfering building model A fixed at
(6.1b, 2.4b) in exposure category B.
the interfering buildings located at (0, 3.2b), i.e. in while the regions of 0:5 IF 0:9 remain unchanged
side-by-side arrangement with the principal building. It relatively.
might be anticipated that a parabolic relationship exists The increase of the building size could also enhance
between IF and Br. A maximum value of IF ¼ 1:16 is the adverse static amplification on the principal build-
recorded when Br ¼ 2, as shown in Fig. 10. This indi- ing when the two interfering buildings are located at
cates that the interfering building with Br ¼ 2 can some critical locations. For the five types of breadths
increase 16% mean wind load on the principal building of interfering buildings, the critical position for both of
when the center-to-center spacing of the two buildings the two interfering buildings are found to be about
is 3.2b. (0, 3.2b) in the present test grid region shown in
Fig. 3. The corresponding maximum interference fac-
3.3.2. Three-building configuration tors for different Br in exposure category B are listed in
More interfering buildings generally produce more Table 4. The table shows that the maximum IF increa-
ses with Br, and a maximum value of 1.195 is found for
significant shielding effects, and the shielding effects
the interfering buildings of Br ¼ 2. This indicate that
increase with the increase of Br. The statistical proper-
the two symmetrically located larger sized interfering
ties for the interferences effects of five types of inter-
buildings of Br ¼ 2 can increase 20% wind load on the
fering buildings are shown in Fig. 11. From this figure,
middle principal building.
one can see that the most notable shielding region
of IF 0:4 increases quickly with the increase of Br 3.4. Effects of height ratio
interfering buildings becomes even less significant in Fig. 14. Interference effects of different height ratios while two
the higher turbulence of exposure category D. The upstream interfering buildings are located at (6.1b, 1.6b).
results indicate that the effects of interfering building
with a height less than 0.5h can be neglected. Only with the increase of Hr. It can be also found from
when the heights of the interfering buildings are greater Fig. 14 that the shielding effects in smoother terrain of
or equal to 0.75h does shielding effects become notable. exposure category B are more significant than those in
In contrast to the significant change in interference category D.
factors for interfering buildings with Hr ¼ 0:5; 0:75 Based on the above results, it can be summarized
and 1:0, factors for interfering buildings with Hr ¼ that the sensitive height of interfering buildings for the
1:0; 1:25 and 1:5 vary only marginally. However, as mean interference effects are in the range from 0.5h to
indicated in Fig. 13, a slight difference for IF ¼ 0:7 1.25h, while the interference effects remain almost the
between the configuration of Hr ¼ 1 and that of Hr > same for higher interfering buildings. However, higher
1 in exposure category B is found. It shows that the interfering buildings cause stronger channeling effect,
shielding effects of the two interfering buildings with and the static amplification may increase with the
Hr ¼ 1 are greater than those of the two interfering increase of the height of the interfering buildings.
buildings with Hr > 1. Fig. 14 presents the variations A maximum value of 1.13 of IF is recorded when the
of the IF with respect to the height ratios of the inter- two interfering buildings with Hr ¼ 1:5 are located at
fering buildings when the two interfering buildings are (0, 1.6b) in terrain category D.
located at (6.1b, 1.6b) and (6.1b, 1.6b), respectively.
From this figure, one can see that the interference
factors decrease rapidly with the increase of Hr in the 3.5. Simplification of the results for three-building
range from 0.5 to 1.0, but for interfering buildings of in arbitrary configurations
Hr 1, the interference factors increase marginally Since four variables, i.e. two x-coordinates and two
y-coordinates of the two interfering buildings, are
involved in the analysis of the interference effects of
three-building configurations, the interference factors
cannot be simply expressed in a single contour as in the
two-building cases. The problem is how to deduce a
relatively simple and yet precise enough representation
method for practical applications from the complex
data from the wind tunnel tests. A reduced interference
factor (RIF) contour for three-building configuration is
thus proposed by synthesizing the effects over the
whole test domain.
Let PA(x,y) and PB(x,y) be the location coordinates
of two interfering buildings; the interference factor can
Fig. 13. Comparison of the distribution of the interference effects of then be expressed as
different height ratio configurations (three-building configurations,
exposure category B). IF ¼ f ðPA ;PB Þ; PA ;PB 2 X ð3Þ
1182 Z.N. Xie, M. Gu / Engineering Structures 26 (2004) 1173–1183
where X denotes the whole position domain of the Fig. 15 also shows the shielding effects in exposure
interfering buildings in the test. Simplifying the above category B are stronger than those in category D.
four-variable problem to a lower two-variable one, the Meanwhile, stronger channeling effects can also be seen
so-called RIF can be expressed as in exposure category B, that is to say, the maximum
static amplification in exposure category B is more ser-
RIF ¼ gðPA Þ ¼ max f ðPA ;PB Þ ð4Þ ious than that in category D.
PB 2X
An example is shown here to explain briefly how to
From Eq. (4), the reduced interference factor, RIF, apply Eqs. (4) and (5) in practical use. For two inter-
can easily be expressed by a simple contour in the simi- fering buildings A and B located at PA ð4b; bÞ and PB
lar way of the two-building configuration. From ð9b; 2bÞ in exposure category B, one can obtain two
Eq. (4), it can also be seen that a RIF value is always RIFs of gðPA Þ ¼ 0:81 and gðPB Þ ¼ 0:87 by interpolat-
greater than or equal to the corresponding IF. In order ing from the distribution of the RIFs shown in Fig. 15a.
Then according to Eq. (5), the interference factor for
to make the estimated interference factor in terms of
Eq. (4) close to the real one, i.e., IF, the interference this configuration is IF ¼ minð0:81;0:87Þ ¼ 0:81.
factor for practical purpose may be determined by the
following equation
4. Concluding remarks
IF ¼ minðgðPA Þ;gðPB ÞÞ ð5Þ The mean interference effects between two and
among three buildings with different configurations
Of course, the calculated result from Eq. (5) will also
have been studied by a series of detailed wind tunnel
be greater than or equal to that of Eq. (3), i.e. IF IF. tests. A good agreement between the current study and
However, from the practical point of view, interference the existing results in two-building configurations is
factors defined by Eq. (5) are conservative. found, which ensures the reliability of the results and
Based on the above definitions, Fig. 15 shows the conclusions proposed in the present study. For inter-
distributions of the RIF for the configuration of three ference effects of three-building configurations, the
identical buildings in exposure categories B and D, interference factor is simplified to an easier expressed
respectively. Distributions of only half of the region are RIF to simplify the experiment results. The main
drawn in the figure due to the symmetry of the RIF. results are summarized as follows.
Fig. 15. The RIFs of three equal size building configuration. (a) Exposure category B; (b) exposure category D.
Z.N. Xie, M. Gu / Engineering Structures 26 (2004) 1173–1183 1183
1. Generally, the effects of the upstream building(s) and the Science Foundation of Guangdong Province
show shielding effects and the corresponding mean (010455). They are gratefully acknowledged.
interference factors are less than 1.0. But the static
amplifications, due to the channeling effects, could
also lead to an increase of 10% of the mean wind References
load on the principal building when the two equal
size interfering buildings are located at (0, 3.2b), [1] Sakamoto H, Haniu H. Aerodynamic forces acting on two
or, in other words, the three buildings are arranged square prisms placed vertically in a turbulent boundary layer.
side-by-side. The observed maximum increase of the Journal of Wind Engineering and Industrial Aerodynamics
mean wind load can be 20% depending on the sec- 1988;31:41–66.
[2] Taniike Y. Interference mechanism for enhanced wind forces on
tion size and spacing of the buildings. neighbouring tall buildings. Journal of Wind Engineering and
2. The interference effects are sensitive to the breadth Industrial Aerodynamics 1992;41:1073–83.
of the interfering buildings. Larger upstream build- [3] English EC, Fricke FR. Interference index and its prediction
ings could produce more shielding effects on the using a neural network analysis of wind-tunnel data. Journal of
principal building and, meanwhile, side-by-side Wind Engineering and Industrial Aerodynamics 1999;83:567–75.
[4] Khanduri AC, Stathopoulos T, Bédard C. Generalization of
interfering buildings with larger size can produce wind-induced interference effects for two buildings. Wind and
more serious channeling effect on the principal Structures, An International Journal 2000;3(4):255–66.
building. [5] Kwok KCS. Aerodynamics of the tall buildings, a state of the
3. The height of the interfering buildings could also art in wind engineering. Proceedings of the Ninth International
affect the wind load on the principal building. The Conference on Wind Engineering, New Delhi, India. 1995,
p. 180–204.
results show that interference from lower interfering [6] Saunders JW, Melbourne WH. Buffeting effects of upwind build-
buildings with Hr 0:5 is negligible while the sensi- ings. Proceedings of the Fifth International Conference on Wind
tive height ratio of interfering buildings is in the Engineering, Fort Collins CO. Oxford: Pergamon Press; 1980,
range from 0.5 to 1.25. For higher buildings, the p. 593–605.
shielding effect is constant. However, higher inter- [7] Chinese Load code for design of building structures, GB50009-
2001. Beijing: Architectural Industry Press of China; 2002.
fering building may cause stronger channeling effects [8] English EC. Shielding factors for paired rectangular prisms: an
and the static amplifications will increase with the analysis of along-wind mean response data from several sources.
increase of the heights of the interfering buildings. Proceedings of the Seventh US National Conference on
Wind Engineering. Los Angeles, CA: University of California;
Due to the complex of the problem, the above dis- 1993, p. 193–201.
[9] Khanduri AC, Stathopoulos T, Bédard C. Wind-induced inter-
cussions and conclusions on the effects of the geometry ference effects on buildings—a review of the state-of-the-art.
of interfering buildings are still in the qualitative level. Engineering Structures 1998;20(7):617–30.
More tests and efforts are therefore needed to improve [10] Taniike Y, Inaoka H. Aeroelastic behaviour of tall building in
the understanding in this area. wakes. Journal of Wind Engineering and Industrial Aerody-
namics 1988;28(1):317–27.
[11] Huang P. Wind-induced interference effects on tall buildings.
Ph.D. Thesis. Tongji University, China, 2001.
Acknowledgements [12] Minimum design loads for buildings and other structures (ASCE
7-98). American Society of Civil Engineers; 1998.
This research is jointly supported by the National [13] Blackmore PA. Effect of flow channeling on gable wall pressures.
Science Foundation (50321003), the Foundation for Journal of Wind Engineering and Industrial Aerodynamics
University Key Teachers by the Ministry of Education, 1991;38:311–23.