Sei sulla pagina 1di 4

Full-scale wind pressures on a permeable roof of a low-rise building

B. Parmentier
Belgian Building Research Institute, Limelette, BE








1 INTRODUCTION

After experiencing huge wind storms during the
90s (1990-1995) a lot of research programmes
were launched worldwide to investigate the
reasons why so much roofs encountered severe
damages like truss-wall failures or more
frequently roof elements removals. This last issue
seems to be less important for the security aspects
but gives the insurance companies some
nightmares when seeing the bill. Actually, it is
more the number of damage occurrences during a
storm event than the expenses by occurrence that
can drastically increase the (repairing) costs.

In order to make a design of the roof elements
(and their fixings) the actual effects of the wind
must be known. The wind effects can be
determined according to different national
standards but these will be replaced by Eurocode 1
[1] in the forthcoming years.

In Eurocode 1 (EC1) the determination of the
pressures on a roof of a low-rise building is
obtained by using the pressure coefficients (CP)
associated with the building shape, the peak
velocity pressure and the permeability of the
different skins for roofs with more than one skin.
These factors are included in the equation (Eq.1).

W = (CP
e
-CP
i
).q
p
(z
e
) (1)

The European standard proposes to use a value of
Cp
i
(in the thin air layer just under the permeable
roof elements) depending on the nature of the
external pressure :
CP
i
=2/3.CP
e
if CP
e
<0 (2)
CP
i
=1.3.CP
e
if CP
e
>0 (3)

Let us call the factor = CP
i
/CP
e
the transmission
factor. So, the Eqs. (2) and (3) give :


mean
= CP
i,mean
/CP
e,mean
= 2/3 if CP
e,mean
<0 (4)


mean
= CP
i,mean
/CP
e,mean
= 1/3 if CP
e,mean
>0 (5)


The idea of the standard was that the frontal
permeability of the tiles could reduce de pressure
acting on them by equating the pressures above
and just under this external skin.

Despite the fact that full-scale measurements are
more and more used worldwide to validate theory
or to compare with model-scale tests an important
lack of knowledge exists concerning the values of
CP
i
. The research programme that will be
presented in this paper aimed to control the
validity of the assumptions given above
concerning the values of the pressure coefficients
and the transmission factor.

2 FULL-SCALE FACILITY

The facility that was used during the experiments
was described by Parmentier et al [4][2]. It
consists of an experimental house (L-B-H=10-5-
5.2 m) with a duo-pitched roof of 30 made of
tiles. The house was erected on the site of BBRI
in Limelette. It was build on a rotary base to
allow the study different angles of attack of the
wind (A.O.A.) whatever its direction. A
meteorological mast with 2 anemometers (5,2m
and 12m height) is positioned 30m upstream of
the building to minimize the effect of the flow
distortion associated with the house.






30
5 m
10
5.2 m
3.7 m
Figure 1 Experimental house in Limelette.
An acquisition frequency of 20Hz during 15
minutes was chosen to be out of the spectral gap
of the wind and to be sure all the information
associated with the pressure signal is recorded.

Measuring internal (in the air layer just under the
tiles) and external pressures is performed by 96
transducers while 3 transducers are used to record
the wind speeds and the wind direction at the
meteorological mast. A last channel is used to
record the position of the house from the North.
The rotation of the house is automated in function
of the targeted AOA (automatic increment of 5
after each run) to maximize the number of
recorded runs with different AOAs.

3 RESULTS

Results obtained from October 2000 until
November 2001 are discussed hereafter. A time
history of the wind speed at the ridge height (V
5
with an amplification factor of 15) and the
pressure of four different locations on the roof are
presented in Figure 2.

The longitudinal turbulence intensity I
v
is given in
Figure 3 as function of the wind direction. The
presence of high trees in the quadrant [200 220]
explains high values of I
v
(50%) while for other
wind directions from North the value of I
v
reaches
the interval [20% 30%] for high mean wind
speeds.

Wind x 15
Pressures
Figure 2 Time history for different pressure
taps and the wind speed at 5,2m.
0
0.1
0.2
0.3
0.4
0.5
0.6
150 170 190 210 230 250 270 290 310 330
Winddirection[]
T
u
r
b
u
l
e
n
c
e

I
n
t
e
n
s
i
t
y


[
/
]
Vm>3m/s
Vm>5m/s
Vm>6.5m/s

Figure 3 Longitudinal turbulence intensity as
function of the wind direction (from North).
The values of the mean pressure coefficient in
function of the AOA is given in Figure 4 and in
Figure 5 for the external pressure and the net
pressure (CP
e
-CP
i
), respectively. The results in
terms of pressure coefficients as function of the
position on the roof is given in Figure 6.

As it can be seen from Figure 4 the peak of the
external mean pressure coefficient is observed for
an AOA of about 25 and the values recorded
have reached -2 while EC1 proposes -1.5.

The first conclusions of the research are that some
places on the roof (zones of reattachment of the
incoming wind flow) show a good relation
between internal [even channel] and external
pressures [odd channel] while zones of high
variation (upwind verges) do not exhibit a good
correlation for critical AOAs between 45
(Figure 7 for downwind eave region and Figure 8
for upwind verge region).

Figure 4 External Mean pressure coefficient
vs. AOA.

Figure 5 Mean Differential pressure
coefficient vs. AOA.

Figure 6 Spatial distribution of the external
pressure coefficients on the roof (AOA=30).

The most critical mean pressure difference
coefficient that was recorded reached -1.7
(
mean
=0.15) while the minimum value given by
EC 1 is -0.5 (
mean
=0.66) for roofs with 30 pitch.
Consequently the existing standard leads to large
non conservative design values by allowing for a
too large reduction associated with the internal
pressure under the roof elements.

-150 -100 -50 0 50 100 150
-2
-1.5
-1
-0.5
0
0.5
1
Angle of attack from Ridge []
CPi 15
CPe 16

Figure 7 Mean Pressure coefficient vs. AOA.

-150 -100 -50 0 50 100 150
-2
-1.5
-1
-0.5
0
0.5
1
Angle of attack from Ridge []
CPi 5
CPe 6

Figure 8 Mean Pressure coefficient vs. AOA.
Another aspect that can be dealt with is the control
of the theoretical point of view of the Eurocode 1
when using the peak-factor approach. The first
step can be the check of the peak-factor parameter
g which is given by the following relation :

P C
CP CP
g

=
min
(6)

An averaging time of 1s is taken for the
calculation of the minimal pressure coefficient
(CP
e,min
). Eurocode 1 estimates the value of g at
3.5 [3]. The analysis of the full-scale data can
provide a reliable check of this important
parameter. This is shown in Figure 9. On this
figure the value of g (g
max
= most critical value for
all AOAs) is plotted against a limit (CP
e,mean,lim
)
which is taken to provide a sufficient level of
significance for the data used for the analysis. It
means that for a CP
e,mean,lim
of -1.5 that only values
of CP
e,mean
that were above this limit were taken
into account for the determination of the
maximum peak factor.

Figure 9 Peak factor max for external
pressure as function of a CP
e
minimum for
different tap locations (near verge).
The results of Figure 9 indicate that the peak
factor is very sensitive to the limit of the CP
e,mean

used for the analysis of taps in the vortex regions
(all the taps on that figure) and that for high values
of CP
e,mean
the peak factor is under the reference
value of 3.5. For lower values of CP
e,mean
(e.g. -1)
the peak factor can reach 6.3. If the pairs
(CP
e,mean
;g) are compared as input parameters of
Eq. (7) for the extreme cases of Figure 9 it can be
seen that the results are quiet similar (difference <
10% for I
v
=30%) :
For CP
e
=-1.75 and g=3.25 CP
e,min
= -5.16
For CP
e
=-1.00 and g=6.30 CP
e,min
= -4.78

Hence, it is important to keep in mind that the
worst external minimal pressure coefficients are
not always related to the worst external mean
pressure coefficients as it is thought in the peak-
factor approach.

CP gI CP
u
) 2 1 (
min
+ = (7)

A more detailed analysis can be completed by
checking the theory of the peak-factor approach
itself by plotting the calculated minimal pressure
coefficient as function of the (real) observed
minimal pressure coefficient as it has been made
by Uemastu et al [5]. The first one is calculated
by Eq. (7) while the other is the full-scale
observation. First results have shown that the
theoretical estimation was not on the safe side for
some runs.

Some differences (in terms of CPs) can be
explained by the fact that CPs in EC1 are for
areas of 1m and with a known probability of
excedence (0.02). The data presented in this paper
were not treated with probabilistic analysis. This
will be taken into account for the final analysis
when sufficient data will be obtained.

4 LITERATURE

[1] CEN, prEN 1991-1-4 (Eurocode 1) : Actions
on structures Part 1.4 : General actions
Wind actions, Draft DOC N316, Brussels,
August 2001.

[2] Parmentier, Mthodologie pour la conception
technique de lenveloppe du btiment : Effet
du vent sur les parois externes semi-
permables , Rapport annuel Biennale 2000-
2002, CSTC-WTCB-BBRI, Brussels, 1999.
(in french)

[3] Geurts, Transparency of pressure
coefficients, Proceedings of the 3th European
and African Conference on Wind Engineering
(EACWE), Eindhoven, 2001.

[4] Parmentier et al, Net pressures on the roof of
a low-rise building Full-scale experiments ,
Proceedings of the 3th European and African
Conference on Wind Engineering (EACWE),
Eindhoven, 2001.

[5] Uematstu & Isyumov, Peak gust pressures
acting on the roof and wall edges of low-rise
buildings, JWEIA 77&78, 1998.

Potrebbero piacerti anche