Sie sind auf Seite 1von 11

JOURNALOF

; vJn'~
ELSEVIER

Journal of Wind Engineering


and Industrial Aerodynamics72 (1997) 187-197

The Silsoe Structures Building:


Comparisons of pressures measured at full scale
and in two wind tunnels
G . M . R i c h a r d s o n , R.P. H o x e y * , A.P. R o b e r t s o n , J.L. S h o r t
Silsoe Research Institute, Wrest Park, Silsoe, Bedford MK45 4HS, UK

Abstract
Two independent sets of full-scale wind pressure measurements have been completed on the
Silsoe Structures Building in both its configurations (featuring a conventional sharp eaves and
a modern curved eaves). Results are compared with those from two 1 : 100 scale measurement
programmes conducted in two leading wind tunnels on models of the building with both the
sharp and curved eaves. These comparisons highlight the importance of implementing certain
wind tunnel procedures in order to obtain good predictions of full-scale mean pressure
coefficients. However, areas of high negative pressure still tend to be under-estimated in
wind-tunnel measurements and this is considered to be associated with a Reynolds number
effect.

1. Introduction
The Silsoe Structures Building (SSB) [1] was constructed during 1986/87 specifically to undertake full-scale wind pressure measurements. The building was constructed with an optional eaves geometry offering either a traditional sharp eaves or
a curved eaves of 635 mm radius. Results of wind pressures generated by both
arrangements are presented. The 24 m long by 12.9 m span by 5.3 m ridge height
building with a 10 duo-pitch roof is located on a flat, exposed, open-country site at
Silsoe Research Institute (SRI). Measurements [2] show that the profile is well
represented by a log-law with a roughness length (Zo) of 0.01 m, and a frictional
velocity compared with the free-stream velocity at 10 m (u,/ulo) of 0.06. A brief
abstract of the pressure measuring procedure is presented in Section 2; a more
detailed account is given in Ref. [1].

* Corresponding author. E-mail: roger.hoxey@bbsrc.ac.uk.


0167-6105/97/$17.00 1997 Elsevier Science B.V. All rights reserved.
PII S01 67-61 05 (97)00274-2

188

G.M. Richardson el al.:d. Wind Eng. hul. Aerodvn. 72 (1997) 187-197

An objective of this work was to provide reliable full-scale data for comparison with
wind-tunnel measurements and computational fluid dynamics solutions. Two windtunnel studies were conducted, one at the University of Western Ontario and one
at the Building Research Establishment; these are described in Section 2. Comparisons between the full-scale and wind-tunnel results are restricted in this report to
comparing mean pressure coefficients (Section 4) since the mean can be determined
with higher statistical confidence; if agreement cannot be obtained, explanations
should be sought before examining peak or extreme values. The methods used to
analyse the full-scale data were presented in Ref. [3]; these methods can also be
applied to wind-tunnel measurements provided that a simultaneous record is made of
wind speed and direction immediately upstream of the model. Extensive analysis of
the data has also been completed exploring mean, RMS and peak values, and
probability and spectral densities. These analyses were presented and discussed in
Ref. [3].

2. Experimental details
2.1. Full-scale measurements

Two independent sets of full-scale pressure measurements have been made on the
SSB. The first set of measurements (SRI measurements) are described in detail in
Ref. [1]. Simple tapping holes were used on the side of the building and static
pressure probes [4] on the roof; the probes overcame the problem of rain-water
penetration and local effects caused by the profiled cladding. In total, 77 pressure
tapping points were installed on the building and measurements were made using
a sequence controller which sampled two pressures at one time. A dwell time of 4 rain
was used for each pair of tapping points, and 4 rain mean values were computed.
A complete definition of pressures over the building surface required many hours of
recording.
A second independent set of instrumentation was installed on the building for work
sponsored by the Building Research Establishment (BRE measurements). In this
experiment, 32 pressure tapping points were installed, each with a pressure transducer
mounted locally on the inside of the building. Simultaneous measurements were made
from the 32 pressure tapping points together with wind velocity sensed by a threecomponent sonic-anemometer. Records of I h duration were made, and these were
partitioned into six t0 min records for analysis.
In both experiments, all pressures were measured relative to a static pressure sensed
by a probe which was positioned at the ridge height of the building 20 m upstream of
the building where the influence of the pressure field associated with the building was
small. Pressure transducers were automatically zeroed using a solenoid valve to
equalise the pressure across the transducer, and were calibrated by using solenoid
valves to apply a common wind dynamic pressure to all transducers. This procedure
ensured a pressure measurement accuracy of better than _+ 1 N/m 2. For the SRI
measurements, plastic tubes up to 25 m in length conveyed the pressures from the

G.M. Richardson et al./J. Wind Eng. Ind. Aerodyn. 72 (1997) 187-197

189

tapping points to the transducers and this restricted the frequency response to 3 dB
down at 2.5 Hz. Pneumatic filters were installed to damp unwanted resonance effects.
The analogue outputs from the pressure transducers were digitised at a rate of 2.5 Hz,
processed and discarded after digitisation. For the BRE measurements, a 0.5 m length
of transmission tubing was used and there was no measurable attenuation up to
50 Hz. The sampling frequency was 5 Hz. The records were stored on digital magnetic
tape and are available for further analysis.

2.2. University of Western Ontario (UWO): (1988) 1:100 scale measurements


The boundary layer in wind tunnel No. 1 at U W O was developed over a distance of
about 20 m as the air was sucked down the tunnel. The working section was
approximately 2.5 m wide by 2.0 m high. Three large spires near the inlet were used to
increase the boundary layer thickness and to increase the turbulence scales. Roughness elements consisted of sets of staggered 25 mm cubes positioned to within 2 m of
the model and randomly spaced 7 mm high machine nuts distributed over the area
around the model.
The velocity profile at the centre of the turntable was defined using two linearised
single hot-wire anemometers. This profile was later used to correct the pressure
coefficients which were referenced to free-stream pressures recorded above the generated boundary layer. The limitations of this procedure have been presented by
Richardson et al. [5].
The 1:100 scale models were manufactured by the BRE who kindly loaned the
models for this comparative study. The tappings were 1.0 mm internal diameter holes.
The pneumatic system was essentially linear in its response to about 90 Hz. The
Statham 2.5 psi transducers provided digitized data accurate to within 0.044).08 of
typical wind dynamic pressure (q) at model ridge height.

2.3. BuiMing research establishment (BRE) : (1992) 1:100 scale measurements


The boundary layer in the BRE wind tunnel No. 3 was developed over a distance of
about 15 m as the air was blown down to the 2 m wide by 1.5 m high working section.
Initially, the air was passed through a 0.5 m square grid and over a sharp-toothed
barrier of 485 mm total height. Downstream of this was 14 m of 10 mm high surface
roughness elements. The final inner boundary layer was developed over the smooth
turntable by the addition of three concentric 10 mm high permeable barriers of looped
interlinked chain.
The velocity profile at the centre of the model position was measured using a
TSI Model 8451-51M-B air velocity transducer mounted horizontally at a constant reference height of 100ram and a vertically traversed Dantec Type 55
single hot-wire probe. The BRE 1:100 scale models were used, as in the U W O tests,
but this time with BRE tubing and sensors. The pneumatic system gave a unity
response to 100 Hz. The differential pressure transducers were Scanivalve model type
PDCR24D.

190

G.M. Richardson el al./J. ~Tnd Eng. Ind. Aerodyn. 72 (1997) 187-197

3. Data recording and analysis


The full-scale measurements were processed to give mean, RMS, maximum and
minimum values of surface pressures and similar statistics for the wind dynamic
pressure, together with the mean and RMS of the wind direction. A limited number of
additional measurements were made with both sets of instrumentation in which the
pressures were recorded at a higher digitisation rate of 20 Hz for spectral analysis. For
the SRI measurements, the pressure transducers were then positioned locally so that
tube lengths were reduced to less than 6 m, giving an improved frequency response of
3 dB down at 12 Hz.
The two model-scale data sets required transformation for comparison with the
full-scale data. In the case of the UWO data, considerable work was conducted to
interpret the data sets [5]. Following a pilot experiment [6], the BRE wind-tunnel
tests included simultaneous measurement of surface pressure and free-stream wind
dynamic pressure. This enabled sensor calibration and coefficient evaluation to be
made in the same manner as at full scale. Two corrections were applied: the first for
the static pressure from the pitot-static probe which in turbulent flow under-read by
4.2%, and the second for the 15% underestimation of ~ due to the long tube lengths
used for the reference pitot probe [7].
Pressure coefficients were obtained by non-dimensionalising measured pressures by
the free-stream wind dynamic pressure measured approximately 4 building heights
upstream at the ridge height of the building.

4. Comparison of results
4.1. Scale fi~ctors
A typical average reference velocity at building ridge height in the Silsoe full-scale
measurements was 10m/s. The UWO ridge height velocity was 6.9m/s, and
in the BRE wind tunnel it was 9 m/s. The reference height spectra in both wind
tunnels were at a 1:100 frequency scaling, in keeping with the linear scaling of the
models.

4.2. Profiles
A comparison of the longitudinal velocity {u) profiles from full scale and both wind
tunnels is given in Fig. 1. The line drawn for a full-scale surface roughness height (z0)
of 10 mm is a good fit to the full-scale and BRE wind-tunnel data, but the UWO data
show only limited agreement between heights of 3 and 15 m. The UWO measurements indicate a significant reduction in surface roughness close to the model
position.
The comparison of turbulence intensities in Fig. 2 also shows the BRE data to be in
closer agreement with the full-scale data. A comparison of reference height velocity
spectra was made by Richardson et al. [6,7]. The most significant differences in the

G.M. Richardson et al./J. Wind Eng. Ind. Aerodyn. 72 (1997) 187-197

191

Heightz(m)
100.00~

~o

~g (._..~z,zo)

=.,~==

I 0.00

1.00

0.10.

0,01

0.0

Io

,.,_,:.~e,,e r..=,..a,.s C,.) I

BRI~ model-1~.aJedate (1992)

0.5

1.0
Velocity ratio

1.5

2.0

Uz
iilo

Fig. I. Verticalprofilesof longitudinal velocity.

Height z(m)
20-

[]

Full-scala20 rnins records (1988 data)"


BRE model-scale66s records (1992 data)
UWO model-scale60s records (1988 data)

in

15.

i-i

10

[]

t3

Do

AA
0

o.t

0:2

q~

oia
Localturbulence intensity

o:4

o'.5

RMS z
uz

Fig. 2. Verticalprofilesof longitudinal turbulence intensity.

modelled boundary layers occur close to the ground, a region from where streamlines
originate which influence flow around the building, including those that stagnate and
those that separate from its sharp edges.

G.M. Richardson el al. ,,d. Wind Eng. Ind. derodvn. 72 (1997) 187 197

192

i/

-o.2 ~

" ~-_o.2

----_2_~

....

<"C)

Fig. 3. ('p distribution on curved ea~es SSB: transverse wind.


j

............. -'o

~.

",..

o.~

6"

II

-0.4

-~k) /

--0"2~ - . .

\ ~ ff---------~--

F'ig. 4. C v d i s t r i b u t i o n on s h a r p eaves SSB: transverse wind.

G.M. Richardson et al./J. Wind Eng. Ind. Aerodyn. 72 (1997) 187-197

193

O.C

,~. * ' ; ~

-0.24

11=

-0.5

,,

/ol

-0.75

0"

Cp _ 1.~

..Q

-1.24
-1.4
-1.75
-2-00

45

1"1

gO

135

180

225

270

315

SRI

BRE

360

WIND DIRECTION ( 0 )
Fig. 5. Full-scale data comparison: sharp eaved building, near-eaves roof tap.

4.3. Overall surface pressure distribution: transverse wind


Figs. 3 and 4 show the mean pressure coefficient (Cp) distributions for the full-scale
building in its curved and sharp eaved configurations. The significant difference
between these distributions occurs at the windward eaves where the suctions over the
windward roof are greater for the sharp eaves. This is due to the difference in the flow,
which was illustrated [8] through flow visualisation. Observations of the sharp eaves
flow separation in the U W O wind tunnel showed it to be less severe than at full scale,
and this is reflected in the reduced suctions recorded in this region in both wind
tunnels (as demonstrated by comparing Figs. 5 and 6 for the transverse wind direction
given by 0 = 180).

4.4. Specific tapping points


Fig. 5 shows the typically good agreement between the mean pressure coefficients
obtained from the two sets of full-scale measurements. This example is for a roof
tapping at the mid-length of the building, 1 m up from the windward eaves of the
sharp eaved building. The reduced scatter in the BRE data, evident near the 180 wind
direction, reflects the longer time average period of 10 min used to obtain these data
compared with 4 min used to obtain the SRI data. At the 270 and 90 directions there
is a suggestion that the SRI data are affected by inappropriate positioning of the
reference mast in relation to these approach wind directions which produced a bias in
the reference static pressure (data were not originally expected to be collected at these
directions). Fig. 6 compares the two wind-tunnel data sets for the same tapping and

G.M. Richardson et al. ,,d. Wind Eng. Ind. Aerodyn. 72 (1997) 187 197

194

0.0
-0.25
-0.5
-0.75
Cp

-1.0
-1.25

"4

70

-0

--t

-1.5
-1.TS ~ - -

........

+. . . . - - i
I

-2.0
45

g0

135

180

225

UWO
* BRE

......

270

315

360

WIND DIRECTION ( 0 )
Fig. 6. M o d e l - s c a l e d a t a c o m p a r i s o n : s h a r p e a v e d building, n e a r - e a v e s r o o f tap.

shows better than expected agreement considering the difference in flow characteristics in the lower profile. Comparing Figs. 5 and 6 shows that the high suctions which
result from flow separation at the windward sharp eaves for a transverse (0 = 180)
wind direction are significantly underestimated at model scale compared with those
measured at full scale, Previous comparisons made by Richardson [7] for this same
tap but with the curved eaves configuration in place show close agreement between
full-scale and model suctions
the explanation for this is that the curved eaves
produce a contrasting attached flow regime, as demonstrated by full-scale flow
visualisation, for which much closer model-scale predictions are attainable.
Figs. 7 and 8 relate to a sensor on a side wall near the gable end. Fig. 7 again shows
good agreement between the full-scale data sets, in this case for a side-wall tap at 3 m
height located 0.5 m from the gable end of the sharp eaved building. In contrast, Fig. 8
illustrates wind-tunnel differences possibly attributable to the boundary layer differences affecting the flow around the building. The higher maximum Cp values from the
BRE wind-tunnel indicates the stagnation of a streamline from a region higher in the
boundary layer. The difference between the Cv values for wind directions 270-315' is
possibly indicative of the higher free-stream turbulence intensity close to the ground
in the BRE wind-tunnel resulting in a more negative separation pressure. Li et al. [9]
indicate that this is due to a reduction in the size of the separation bubble. A comparison of Figs. 7 and 8 shows close agreement between full-scale and model-scale mean
Cp values. The maximum and minimum full-scale C~ values fall within the bounds of
the two sets of model-scale results. In this comparison, the suction produced in the
wind tunnel for the best boundary layer simulation has reproduced the minimum
full-scale suction accurately.

G.M. Richardson et al./J. Wind Eng. Ind. Aerodyn. 72 (1997) 187-197

195

1.0
0.75

0.2!

Cp

):

0.5

'/!

/1

0"

0.1

1i

-0.25

~f

-0.5

-0.75
-I.C

45

90

135

180

225

270

315

360

WIND DIRECTION ( 0 )
Fig. 7. Full-scale data comparison: sharp eaved building, side-wall tap near the gable end.

5. Discussion
The differences between the minimum mean Cpmeasured at full and model scale as
illustrated in Figs. 5 and 6 have also been observed by Okada et al. [10] when making
similar comparisons of data from the Texas Tech Building (TTB). Cochran et al. [11]
investigated the effect of departing from geometrically scaled tapping point areas for
the TTB, but they concluded that correct similitude is only necessary when the scale of
the flow is comparable to the tap size. Thus, it can be concluded that this effect does
not contribute to the differences in mean Cpbetween full- and model-scale results on
the SSB. Cochran et al. [11] also suggested that viscous stresses in the small modelled
vortices would be disproportionately large (a vortex Reynolds number effect). Cook
has made a similar comment recorded in the discussion of Richardson et al. [7].
Thus, as far as mean Cp's are concerned, wind tunnels may be used to estimate
wind loads over the majority of the surface of a building. However, regions
of separated and conical-vortex flow are likely to produce underestimations of
surface pressures at model scale. Furthermore, the streamlines influencing these
flow regions originate from close to the ground, just where it is most difficult to
sustain the shear-flow characteristics accurately in a wind tunnel. This problem
is likely to be accentuated with long buildings as the flow becomes more twodimensional and hence the separated flow will originate from even lower streamlines
in the profile as more of the approach flow goes over rather than around the
obstruction.

G.M. Richardson et al..'J. Wind Eng. Ind. Aerodyn. 72 (1997) 187 197

196

1.0

0.75

'

~-\,]

.....
I

0.5

/r

Cp

'i,zl

tF

0.25

!:
I"

0.0

-0.25

,--...

",

!
,t,

F
t

-05

~rj

tu

"

wit '

BRE

-0.7~

- 1..00

45

90

225
270
WIND DIRECTION ( 0 )
135

180

315

360

Vig. 8. M o d e l - s c a l e d a t a c o m p a r i s o n : s h a r p e a r e d building, side-wall t a p n e a r the g a b l e end.

The significance of the present full-scale data and comparisons with wind-tunnel
measurements is that more meaningful model-scale data would result if the full-scale
practice of simultaneously measuring surface and near-field reference pressures was
adopted in the wind tunnel. This would necessitate adequate sampling rates and
frequency responses and correction of the underestimation of static pressure from the
pitot-static probe in turbulent boundary layers.

6. Closing remarks

The Silsoe Structures Building has been the focal point of an interactive programme
of comparisons of wind-load data. The curved eaves feature has highlighted how small
architectural changes can have significant effects on load distribution. More
reliable wind-tunnel techniques have resulted from improvements to boundary layer
generation and from the adoption of full-scale measurement practices. The use of
remote reference pressures can result in reliable mean Cp data from wind tunnels.
However, statistics related to transient flow characteristics are more meaningful if
simultaneous local reference pressures and direction data are used in normalising the
surface pressure data. The future reliability of wind-tunnel data depends on the
incorporation of the lessons learnt, into codes of laboratory practice. There remains
evidence, however, that at model scale viscous damping attenuates the magnitude
of the pressure field in separated and vortex flows, and hence correction needs
to be applied for Reynolds number effects in these regions - work is continuing
on this.

G.M. Richardson et al./J. Wind Eng. Ind. Aerodyn. 72 (1997) 18~197

197

Acknowledgements
This programme of work was funded by the UK Ministry of Agriculture, Fisheries
and Food and the Building Research Establishment. Financial assistance was also
received from the Natural Sciences and Engineering Research Council of Canada
through operating grants made available to D. Surry and A.G. Davenport at the
University of Western Ontario.

References
[1] A.P. Robertson, A.G. Glass, The Silsoe structures building its design, instrumentation and research
facilities, Div. Note DN 1482, AFRC Institute of Engineering Research, Silsoe, UK, 1988, pp 1-59.
[2] R.P. Hoxey, P.J. Richards, Structure of the atmospheric boundary layer below 25 m and implications
to wind loading on low-rise buildings, J. Wind Eng. Ind. Aerodyn. 41-44 (1992) 317-327.
[3] R.P. Hoxey, P.J. Richards, G.M. Richardson, A.P. Robertson, J.L. Short, The Silsoe structures
building: the completed experiment Part 2, Proc. 9th Int. Conf. on Wind Engineering, New Delhi.
India, 9-13 January 1995.
[4] P. Moran, R.P. Hoxey, A probe for sensing static pressure in two-dimensional flow, J. Phys. E 12
(1979) 752 753.
[5] G.M. Richardson, D. Surry, The Silsoe building: comparison between full-scale and wind-tunnel data,
J. Wind Eng. Ind. Aerodyn. 5l (1994) 157 176.
[6] G.M. Richardson, D. Surry, The Silsoe building: a comparison of pressure coefficients and spectra at
model and full-scale, J. Wind Eng. Ind. Aerodyn. 41-44 (1992) 1653-1664.
[7] G.M. Richardson, P.A. Blackmore, The Silsoe structures building: comparison of 1:100 model-scale
data with full-scale data, Proc. 1st European and African Conf. on Wind Engineering, Guernsey,
20-24 September 1993, pp. 173 183.
[8] A.P. Robertson, Effect of eaves detail on wind pressures over an industrial building, J. Wind Eng. Ind.
Aerodyn. 38 (1991) 325-333.
[9] Q.S. Li, W.H. Melborne, An experimental investigation of the effects of free-stream turbulence on
streamwise surface pressures in separated and reattaching flows, Proc. 3rd Asia-Pacific Symp. on
Wind Engineering, Vol. II, Hong Kong, 13-15 December 1993, pp. 631-636.
[10] H. Okada, Y.-C. Ha, Comparison of wind tunnel and full-scale pressure measurement tests on the
Texas Teeh Building, J. Wind Eng. Ind. Aerodyn. 41-44 (1992) 1601-1612.
[11] L.S. Cochran, M.L. Levitan, J.E. Cermak, B.B. Yeatts, Geometric similitude applied to model and
full-scale pressure tap sizes, Proc. 3rd Asia-Pacific Symp. Wind Engineering, Vol. II, Hong Kong,
13 15 December 1993, pp. 917 922.

Das könnte Ihnen auch gefallen