You are on page 1of 9

ICBEST 2014

June 09-12, 2014, Aachen, Germany

Wind interference between two high-rise building models:


On the influence of shielding, channelling and buffeting on peak pressures

A.J. Bronkhorst1,2, C.P.W. Geurts1,2, C.A. van Bentum1 and B. Blocken2


1
TNO, Delft, the Netherlands
2
Eindhoven University of Technology, Department of the Built Environment, Eindhoven, the
Netherlands
Corresponding author: A.J. Bronkhorst, okke.bronkhorst@tno.nl

Abstract

The influence of interference between two high-rise building models on the minimum peak pressures was
investigated in an atmospheric boundary layer wind tunnel. Pressure measurements were performed on a square
model with an aspect ratio of 1 to 4. The influence of an interfering model with the same height as well as half
the height was investigated. A square full-height model resulted in minimum peak pressures on the facades 50%
larger than observed in the isolated configuration. This increase was not observed for a circular full-height model.
Both a square and circular half-height model resulted in increases of 30% on the facades at the same height as
the roof of the interfering model. This paper gives a detailed discussion of these increases in terms of the wind
behaviour and types of interference (i.e. channelling, buffeting and shielding).

1 Introduction

Wind loads in urban areas are difficult to determine without the help of wind tunnel measurements.
The influence of adjacent buildings on the local wind behaviour, also known as wind interference, is
hard to predict and can often not be neglected. Parametric studies on interference between two high-
rise buildings, e.g. Bailey and Kwok (1985), Khanduri (1997), and Kim et al. (2011), have identified
several configurations with adverse and beneficial influences on overall and local wind loads. How
these influences on the wind loads change with different parameter settings, such as angle of incidence,
building shape and separation distance, is less clear. For example, it is known that large peak pressure
coefficients are observed in the side-by-side arrangement with the wind parallel to the passage, e.g.
Kim et al. (2011). However, the range of wind angles where this increase in peak pressure is observed
and the influence of the interfering building shape are not well established. Bronkhorst et al. (2014)
have identified two configurations, i.e. (X, Y) = (1.5B, 0) and (X, Y) = (1.5B, 1.5B), where local
increases of more than 30% were observed in the minimum peak pressure coefficient. This paper
presents a more detailed analysis of the minimum peak pressure coefficient in these configurations.
The influence of angle of incidence, separation distance and interfering building shape are assessed,
and the observed changes in the peak pressure are discussed in terms of the related flow behaviour and
types of interference.

2 Background

Wind behaviour and resulting effects in configurations affected by interference are often associated
with three types of interference (Mara et al., 2012): (i) shielding or sheltering, (ii) channelling, and (iii)
buffeting. Figure 1 gives a schematic impression of the three interference types. Shielding, illustrated
in Figure 1(a), takes place when one building is positioned in the wake of another building.
Bronkhorst et al. Wind interference between two high-rise building models 2/9

Figure 1. Schematic illustrations of the interference types: (a) shielding, (b) buffeting, and (c) channelling.

The reduction in mean wind speed results in smaller overall and local wind loads. This type of
interference is observed in tandem arrangements, and slightly staggered arrangements. For the tandem
configuration, English (1993) developed a mathematical expression for the overall wind load, based
on a statistical analysis of several experimental studies. This expression relates the reduction in the
overall mean along-wind load, in terms of a shielding factor, to the separation distance, and the height
and width of the buildings. Although reasonably accurate, this expression was never implemented in
any building code, because the beneficial influence due to shielding can disappear over the lifetime of
the shielded building; the nearby building providing the shielding might be demolished.
In case of interference, buffeting is often characterized as the influence of turbulent fluctuations,
induced by one building, on another building (Saunders and Melbourne, 1979). The fluctuations can
be relatively large scale, e.g. due to the transverse motion of the wake, which can result in resonant
buffeting when its frequency is in tune with the natural frequency of the reference building (Bailey and
Kwok, 1985). However, the small-scale fluctuations in the shear layer affecting the aerodynamics of
the downwind building, illustrated in Figure 1(b), are also considered buffeting. Buffeting can have a
large influence on the dynamic behaviour of a high-rise building; increases in the dynamic response
with a factor of 3.2 to 4.4 have been reported by Bailey and Kwok (1985). These dynamic wind effects
are mainly related to large scale buffeting, i.e. resonant buffeting. Increases in the local peak load can
be a result of both large as well as small scale buffeting, depending on the position on the building
(Kim et al, 2011). Buffeting is typically observed in tandem and staggered configurations, but can also
be encountered in side-by-side arrangements (Stathopoulos, 1984).
Channelling is observed when wind is directed through a passage between buildings (Blocken et al.,
2007). For high-rise buildings, this channelling can result in a local increase in mean velocity. This
velocity increase is responsible for enhanced minimum pressures on the building facades in the
passage (e.g. Lam et al., 2008). Channelling especially has a large influence on the wind loads in side-
by-side configurations, but can be observed in any type of two building configuration.

3 Experiment Set-up

Wind tunnel experiments were performed in the open circuit atmospheric boundary layer wind tunnel
of TNO in the Netherlands. A detailed description of the boundary layer characteristics is given in
Bronkhorst et al. (2014).
Figure 2(a) indicates the configurations that were investigated for this study. The black dots specify
the measurements also discussed in Bronkhorst et al. (2014), the white dots specify the additional
measurements that will be presented in this study. The width and height of the reference model are
specified in Figure 2(b), H = 0.48 m and B = 0.12 m. Pressure taps were distributed as illustrated in
Figure 2(b), with 38 pressure taps on each façade face and 20 pressure taps on the roof face. Taps are
designated by face f (which can be A, B, C, D or E, E being the roof), row i and column j. For example,
the pressure tap on face A, in the second row and the fifth column is specified as tap A25.
The interfering models that were investigated had a square and circular plan with a width of 0.12 m,
and had the same and half the height of the reference model. These models will be referred to as the
full-height and the half-height interfering models.
Bronkhorst et al. Wind interference between two high-rise building models 3/9

Figure 2. (a) The investigated positions of the interfering models. (b) The pressure tap locations on the facades
and roof of the reference building model, and (c) pictures of the reference building model with the 4 investigated
interfering models.

Pressures were sampled with a rate of 400 Hz for a period of 20.48 s. The dynamic pressure (qH)
was determined with a pitot-static tube positioned at roof height (H = 0.48 m), 2.6 m in front of the
model and 0.7 m to the side. The mean velocity at this height was approximately 14.8 m/s, which
corresponds with a width-based Reynolds number of ReB = UHB/ν ≈ 1.1 x 105.

The pressure coefficient time series were divided in 32 intervals, cp,i(t), with a period ti = 0.64 s.
Minimum peak pressure coefficients were determined with:

c p ,min ( fij, ) 
1 N
 
 min c p ,i ( t )
N i 1
(1)

Figure 3 shows the histogram of the 32 minima obtained at pressure tap B51, and the probability
density function of the Gumbel distribution fitted on these 32 minima. The mean minimum pressure
coefficient obtained with equation (1) is also indicated. Mean and standard deviation pressure
coefficient were determined with:
c p ,mean ( fij , ) 
1 N

 mean c p ,i ( t )
N i 1

1 N

c p ,std ( fij , )   std c p ,i ( t )
N i 1
 (2)
Bronkhorst et al. Wind interference between two high-rise building models 4/9

Figure 3. Probability density function of the selected minimum peak pressure coefficients of each interval cp,i(t)
and a fitted Gumbel distribution for tap B51, the mean minimum peak pressure coefficient obtained through
averaging over the 32 selected minimum pressure coefficients is indicated with the cross.

4 Results and Discussion

This section presents results of the minimum peak pressure coefficient, and the mean and standard
deviation pressure coefficient. Changes in the pressure coefficient due to interference are discussed in
terms of magnitude. For example, a pressure coefficient of -2.5 is considered larger than a coefficient
of -1.5, and a change from -1.5 to -2.5 is regarded as an increase. The influence of the full-height and
half-height interfering models are discussed in separate paragraphs. Mean and standard deviation
pressure coefficients are investigated, to obtain a better understanding of the flow behaviour
responsible for the increase in minimum peak pressure.

4.1 Influence of full-height models


Figure 4 shows the minimum peak pressure coefficient distribution on face A of the reference model
for 5 angles of incidence between 60° and 120°. In the isolated configuration at θ = 90°, illustrated in
Figure 4(a), the peak pressure coefficients on face A vary between -1.0 and -1.8. The largest minimum
values for θ = 90°, i.e. cp,min = -1.8, are observed at taps A23 and A47. The value at tap A23 is related
to the separation bubble. The relatively large turbulence intensity at this height, i.e. Iu(H/3) = 0.2,
decreases the size of the separation bubble and intensifies the negative pressure inside. The large
minimum peak pressure coefficient at tap A47 is related to reattachment. The transverse motion of the
reattachment point and the small scale fluctuations inside the shear layer (Bronkhorst et al., 2013) are
responsible for large pressure peaks at this location.
The largest minimum peak pressure coefficient is observed at θ = 75°, with a minimum value in the
top left corner, i.e. cp,min(A61, 75°) = -2.4. The position of this peak value suggests it is caused by an
end effect, induced by the presence of the roof. This suggestion is supported by flow visualisation
studies performed by Kramer and Gerhardt (1991) and Tryggeson and Lyberg (2010). Near the top
corner of the side face, they found a strong upward curvature of the flow, which feeds the roof edge
vortex. We assume that this flow curvature has a strong influence on the separation bubble in the
corner region of the side face.
Bronkhorst et al. Wind interference between two high-rise building models 5/9

Figure 4. Minimum peak pressure coefficient contours on face A at 5 angles of incidence, for (a) the isolated
configuration, (b) the interfering square full-height model, and (c) the interfering circular full-height model.

Figure 4(b) shows the minimum peak pressure distribution on the reference model with an
interfering square full-height model at (X/B, Y/B) = (1.5, 0). The largest minimum pressures are
observed at θ = 90°: cp,min(A42, 90°) = -2.4 and cp,min(A61, 90°) = -3.0. The increase at tap A42 with
respect to the isolated configuration can be attributed to channelling. PIV measurements, described in
de Jong et al. (2012), were performed on the same models at the same height (2/3H). De Jong et al.
showed that channelling of the flow resulted in a 40% increase in the mean velocity near the entrance
of the passage and an increase of 80% in velocity fluctuations near the model surface in the passage.
The largest minimum peak pressure coefficient is, similar to the isolated configuration, observed in
the top corner, i.e. cp,min(A61, 90°) = -3.0. However, under the influence of the interfering square full-
height model it was not determined at an angle of incidence of 75°, but at 90°. The large minimum
peak pressure coefficient is still related to the curvature of the flow near the corner. Blocken et al.
(2007) showed for a side-by-side configuration that a large portion of the flow exits the passage at the
top, which indicates flow curvature near the top. They suggest the responsible mechanisms are wind-
blocking by the pressure field in front of the buildings, and passage resistance, in case of small
distances between the buildings. The pressure distribution at θ = 120° is quite similar to the
distribution observed at θ = 90° in the isolated configuration. This shows that the square full-height
Bronkhorst et al. Wind interference between two high-rise building models 6/9

model not only results in larger minimum peak pressures, but also increases the range of wind angles
at which large minimum peak pressures are observed.
Figure 4(c) shows the minimum peak pressure distribution on the reference model with a circular
full-height interfering model at (X/B, Y/B) = (1.5, 0). The circular shape of the interfering model
reduces the magnitude of the minimum peak pressures on the reference model at θ = 75° and θ = 90°.
The largest peak pressure coefficient is cp,min(A61, 75°) = -2.0, which is smaller than the largest value
observed in the isolated configuration. This shows that a circular full-height interfering model at (X/B,
Y/B) = (1.5, 0) has, from a structural point of view, a positive effect on the minimum peak pressure
distribution on face A of the reference model. For θ = 90°, the largest minimum is observed at tap A42:
cp,min(A42, 90°) = -1.8. This is the same position where the largest value was observed for the square
interfering model, however, the magnitude is 0.6 units smaller.
These findings show that the flow behaviour between two square full-height models is different
than that between a square and circular full-height model. While flow separation and vortex shedding
on the reference model are intensified by a square full-height model, the combination with a circular
model reduces the intensity of these flow processes. However, when translating these findings to full-
scale, the influence of the circular shape should be considered with due care. Experiments on circular
cylinders by e.g. Dryden and Hill (1930) and Roshko (1961) have shown that the flow characteristics,
specifically the position of separation, can still change at very high Reynolds numbers (ReB = 106 –
107). A more detailed study of the flow mechanism responsible for the beneficial influence of the
circular cylinder should make clear whether a similar Reynolds dependency can be expected.

Figure 5. Absolute values of the minimum peak and mean pressure coefficient, and the standard deviation
pressure coefficient versus the distance between the models determined at pressure tap (a) A61 and (b) A42.
Bronkhorst et al. Wind interference between two high-rise building models 7/9

The minimum peak, mean, and standard deviation pressure coefficient at pressure tap A61 and tap
A42 are illustrated in Figure 5(a) and 5(b) for configurations with an interfering square full-height
model at increasing distance in side-by-side arrangement, i.e. θ = 90° and increasing X/B. The isolated
configuration is presented as an interfering model at infinite distance.
With decreasing separation distance an increase is observed in minimum peak pressure coefficient
at both pressure taps. This increase is especially large for distances X/B < 3. At tap A61, illustrated in
Figure 5(a), the increase in minimum peak pressure for X/B between 1.5 and 3 is due to an increase
both in mean as well as standard deviation pressure coefficient. At tap A42, the increase in the same
X/B range is mostly a mean effect, i.e. especially the mean pressure becomes larger in this range. For
X/B < 1.5, the increase in minimum peak pressure coefficient at tap A42 is mostly due to an increase
in the standard deviation pressure coefficient, as the slope of the mean pressure coefficient drops off.
The opposite is observed at pressure tap A61, where the standard deviation drops off for X/B < 1.5, but
the mean pressure coefficient still shows a large increase.
These findings show that the increase in minimum peak pressure at different locations is the result
of different types of wind behaviour, in terms of mean and fluctuating properties. This indicates that
flow processes, such as separation, reattachment, and blocking have very different influences on the
mean and fluctuating flow characteristics at different locations. Furthermore, their intensity and range
of influence are very much dependent on configuration parameters, such as the distance between the
buildings.

4.2 Influence half-height models


Figure 6(a) shows the minimum peak pressure coefficient distribution on face B for the isolated
configuration at θ = 330°-30°. The distributions in Figure 6(a) are similar to the distributions shown in
Figure 4(a) for face A at θ = 60°-120°. The same flow processes discussed in paragraph 4.1, are
responsible for the minimum peak pressure observed in Figure 6(a).
Figure 6(b) and (c) show the influence of a square and circular half-height model on face B of the
reference model. The largest adverse influence is observed at a wind angle of 0° on the windward side
at tap B32. Both the square and circular half-height model result in cp,min(B32, 0°) = -2.3, at half the
height of the reference model. This is the same location as the roof of the interfering model, which
suggests the increase is related to flow processes induced by the roof. Kim et al. (2011) determined
similar increases at the same location in this configuration. In addition, they assessed the influence of
an interfering model with a height of 0.7H, and found increases on the reference model at this height.
They also indicate that the location with enhanced peak pressures is related with the height of the
reference model.
Based on these observations two flow processes related to buffeting (Saunders and Melbourne,
1979) are most likely responsible for the enhanced peak pressures:
(i) The increase in turbulent fluctuations due to the shear layer introduced by the roof of the
interfering model;
(ii) The downwards deflection of wind by the windward face of the reference model, which
due to the roof of the interfering model results in an increased volume flow around the
corners of the reference models.
The results of Kim et al. (2011) indicate that the adverse influence on the minimum peak pressures
near the roof of the interfering model is slightly smaller for the interfering model of 0.7H than for the
half-height model. This decline in influence with larger building height suggests the second process is
particularly important, as a smaller amount of air is deflected downward.
The influence of shielding by the half-height model can also be observed. Figure 6 shows that at a
wind angle of 345° both square and circular half-height model result in minimum peak pressure
coefficients close to 0, for heights below 0.5H. In the isolated configuration pressure coefficients
between -0.5 and -1.5 can be observed in the same region.
Bronkhorst et al. Wind interference between two high-rise building models 8/9

Figure 6. Minimum peak pressure coefficient contours on face B at 5 angles of incidence, for (a) the isolated
configuration, (b) the square half-height interfering model, and (c) the circular half-height interfering model.

5 Conclusion

A wind tunnel study was performed to determine the influence of interference on the minimum peak
pressure coefficients on a high-rise building model. The influence of a square and circular model of
half and the same height as the reference model were investigated. Large increases of up to 50%, due
to the square full-height model, were observed in the side-by-side arrangement on the passage facade
face of the reference model. It was shown that this increase, related to channelling, can be observed for
distances between the buildings smaller than 3B. The circular full-height model did not result in an
increase in minimum peak pressures. The processes of flow separation and reattachment are not
intensified by the circular cylinder, compared to the square cylinder. A more detailed investigation
into the flow mechanism responsible for the beneficial influence should clarify whether a similar
behaviour can be expected between full-scale high-rise buildings.
The half-height models resulted in an increase of 30% on the side face at half of the reference
model height, when the two models are positioned in a tandem arrangement. Flow behaviour related to
Bronkhorst et al. Wind interference between two high-rise building models 9/9

buffeting, i.e. the turbulence fluctuations introduced by the roof through the shear layer and the
deflection of the buffeted downward wind flow by the reference model, is responsible for this increase.

References

[1] Bailey, P.A., Kwok, K.C.S. (1985). Interference excitation of twin tall buildings. J. Wind Eng.
Ind. Aerodyn., 21, 323-338.
[2] Blocken, B., Carmeliet, J., Stathopoulos, T. (2007). CFD evaluation of wind speeds conditions in
passages between parallel buildings – effect of wall-function roughness modifications for the
atmospheric boundary layer flow. J. Wind Eng. Ind. Aerodyn., 95, 941-962.
[3] Bronkhorst, A.J., Geurts, C.P.W., van Bentum, C.A., Blocken, B. (2014). Wind interference
between two high-rise building models: Interference factors for minimum peak pressures on
facades and roof. Int. Conf. on Build. Env. and Tech., Aachen, submitted.

[4] Bronkhorst, A.J., Geurts, C.P.W., van Bentum, C.A., Blocken, B. (2013). Spectral analysis of
pressures measured on two high-rise building models in side-by-side arrangement. 6th Eur. Afr.
Conf. on Wind Eng., Cambridge.

[5] Dryden, H.L., Hill, G.C. (1930). Wind pressure on circular cylinders and chimneys. Bur. Stand. J.
Res., Washington, 5, 653-693.
[6] De Jong, A., Bronkhorst, A.J., Geurts, C.P.W., van Bentum, C.A. (2012). PIV and pressure
measurements on two high-rise models in tandem arrangement. 7th Int. Coll. on Bluff Body Aero.
and Appl., China.
[7] English, E.C. (1993). Shielding factors for paired rectangular prisms: an analysis of along-wind
mean response data from several sources. Proc. 7th US Nat. Conf. Wind Eng., Los Angeles, CA,
193-201.
[8] Khanduri, A.C. (1997). Wind-induced interference effects on buildings: integrating experimental
and computerized approaches. PhD thesis, Concordia University.
[9] Kim, W., Tamura, Y., Yoshida, A. (2011). Interference effects on local peak pressures between
two buildings. J. Wind Eng. Ind. Aerodyn., 99, 584-600.

[10] Kramer, C., Gerhardt, H.J. (1991). Wind pressures on roofs of very low and very large industrial
buildings. J. Wind Eng. Ind. Aerodyn., 38, 285-295.
[11] Lam, K.M., Leung, M.Y., Zhao, J.G. (2008). Interference effects on wind loading of a row of
closely spaced tall buildings. J. Wind Eng. Ind. Aerodyn., 96, 562-583.
[12] Mara, T.G., Ho, T.C.E., Isyumov, N. (2012). Interference and influence of nearby buildings: A
discussion of the design approach. Proc. Struct. Cong., ASCE, 798-808.

[13] Roshko, A. (1961). Experiments on the flow past a circular cylinder at very high Reynolds
number. J. Fluid Mech., 10, 3, 345-356.
[14] Saunders, J.W., Melbourne, W.H. (1979). Buffeting effects of upstream buildings. Proc. 5th Int.
Conf. Wind Eng., Fort Collins, 593-606.
[15] Stathopoulos, T. (1984). Adverse wind loads on low buildings due to buffeting. J. Struct. Eng.,
110, 2374-2392.
[16] Tryggeson, H., Lyberg, M.D. (2010). Stationary vortices attached to flat roofs. J. Wind Eng. Ind.
Aerodyn., 98, 47-54.

You might also like