Beruflich Dokumente
Kultur Dokumente
Abstract
Th~s paper presents results of a fundamental, experimental study on wind loading of porous facade systems The important parameters - relatwe building dimensions, porosity, and gap width - have been vaned systematically m a model scale study Both time averaged and fluctuatmg pressures have been measured Therefore, wind loads using the quasi-static approach and the peak pressure approach may be derived The validity of the model scale investigation has been checked by full scale experiments
1. Introduction
Wind load data for the external surfaces of buildings (walls, roofs) as presented In building codes and standards are based on wind tunnel measurements on models with smooth, impermeable surfaces Those data have been traditionally used to calculate design wind loads for porous cladding systems like loose-laid pavers on roofs and shingles on facades. The use of porous surfaces has greatly increased over the past years in particular for the rehabilitation of roofs and facades. To provide additional thermal insulation, the insulation is usually fixed to the outside of the original wall The Insulation layer has to be protected against adverse weather conditions, in particular against rain The rain screen itself consists usually of a porous outer sheeting allowing for extraction of condensed moisture Typical systems are the pressure equilibration rain screen (PER) commonly used in N o r t h America and the back-vented ram screens (BVR) commonly used in Western Europe The wind load on the protection layer is due to the difference between external pressure and Internal pressure in the gap between the porous sheeting and Impermeable building wall Another recent application with increasing importance is the so-called convective facade Convective facades are used on medium and high-rise office buildings to avoid air conditioning. A second glass envelope is added resulting in some air space outside
* Corresponding author 0167-6105/94/$07 00 1994 Elsevier Science B V All nghts reserved SSDI 0 1 6 7 - 6 1 0 5 ( 9 4 ) 0 0 0 8 3 - P
38
HI
the office rooms During winter time this a~r space will be heated by solar radmt~on hke a greenhouse During summer time excessive heating of the rooms may be avoided by adding sunshades m the air space. 1 e outside the office windows Sufficient ventalataon of fresh air and m case of fire of smoke has to be ensured by the flow in the a~r space Whereas sufficient anformatlon concerning the external pressure distribution is available, the internal pressure has been only very httle investigated Results of experimental research are presented in Refs [1-3] and from theoretical considerations in Refs [4,5] This paper will present the major results of a fundamental, experimental study on wind loading of porous facade systems The important parameters have been varied systematically m a model scale study The validity of the model scale mvestagataon has been checked by full scale experiments
39
-I
-Cp Fig 1 Flow situation
a length of 2 m, the fetch length is approximately 6 m and the m a x i m u m velocity Umax = 23 m/s. The boundary layer is produced according to the method of Cook [7]. Three boundary layers have been used in the present tests, having profile exponents o% = 0 1, 0.2 and 0.3 The boundary layer scale is approximately 1"350 Models with relative dimensions h/a = 0.5, l, 1.5, 2 and 4 and b/a = 1, 2 and 4 with a constant width a = 100 m m have been investigated All models were sharp edged The pressure distribution was measured for various flow directions with flow direction perpendicular to the small side of the building defined as 0t = 0. The flow direction was varied in steps of A0t = 10 The same models were used to determine the external and internal pressures To measure the internal pressures a porous wall was added where the porosity was obtained by regularly spaced holes. Relative permeabihtles e -- 0.5%, 0.75% and 1% (based on the area of the building side under consideration) have been Investigated The gap flow resistance was varied by varying the gap width between the impermeable building wall and the porous facade (s/a = 0 0025, 0005 and 0.01) Typical full scale gap widths for BVR systems and convective facade systems are In the range s/a = 0.001 to 0 01 Each model was equipped with 180 pressure taps The pressure tap locations were determined from preliminary experiments and chosen in such a way that the expected locations of the m a x i m u m and minimum pressures were Included The data were sampled with a data acquisition system consisting of three scanivalves with internal
40
pressure transducers and a PC A 100 Hz low pass filter was used Each of the scanivalves was connected to the wind tunnel Prandtl-tube to measure the dynamic pressure The reference pressure for the transducer was the static pressure port of the Prandtl-tube at the roof hezght of the investigated building Each pressure tap was scanned w~th a frequency of 500 Hz From the accumulated data the time averaged pressures and the rms values of the pressure fluctuations were calculated and nondlmensIonahsed with the wind stagnaUon pressure at the bmldmg roof height To ensure a good frequency response the tube lengths between the scanner and the pressure taps had been opUmlsed following the recommendations by Cook [8] and Knoch [9] The m a x i m u m gain for frequencies up to 100 Hz Is below 1 18 To ensure the similarity of the net pressures m the full scale and the model scale expertments the raUo of through-flow resistance and gap flow resistance have to be the same m both cases For the through-flow resistance the s~mdarlty ~s saUsfied tf the pressure loss (Ap)-volume flow (I?) relauon,
Ap = CV",
(1)
for model scale and full scale facades is equivalent, I e exhibits the same values C and exponent n The range of those characteristic values has been checked by Gerhardt and K r a m e r [2], the model facades have been manufactured accordingly The similarity c o n d m o n for the gap flow resistance IS more complex The flow in the air space is governed both by the gap flow and the outflow (in the critical areas of large external suction) through the porous cladding The situation IS sketched m Fig 2 (left facing flow) The gap flow may be treated as the flow in the inlet section of a twodimensional channel Here, the pressure losses are mainly due to the establishment of the boundary layer The pressure loss at the channel entrance is approximately 2 16 times the stagnation pressure of the gap flow for the laminar model scale flow [10] For the full scale situation with turbulent flow the pressure loss is only approximately 1 4 times the stagnation pressure Thus, equivalent flow s~tuat~ons m model scale and full scale, t e the same gap flow resistance, may be achieved by adjusting the gap
panel
41
width The model scale gap width should be larger than the full scale gap width by a factor of x/2 16/1 4 = 1.24 If addmonal pressure losses occur at full scale due to the support structure (e g battens, see right facing flow in Fig 2) of the porous cladding system, the correction factor for the gap width will decrease. As a first estimate the gap width in model scale tests should be about 10% larger than in the full scale situation
4. Results
The influence of the wall permeabihty, the gap width and the relative building dimensions will be discussed for open country exposure flow The following results are for the lowest locally measured, time averaged pressure coefficients (maximum suction) for the flow angle ~ = 10 The angle ~ -- 0 is defined for the flow parallel to the long sides of the building. For most building dimensions investigated the flow angle = 10 led to the largest time averaged suctions occurring on the short side at about 15% to 20% of the building width downstream of the windward edge and at about 90% of the bmlding height. For the discussion of the various influence parameters the constant flow angle ~ = 10 has been chosen
-14 12.
- 1.0
~p,ex
-08
......~ 8 = 1%
. I ! ! --E=05% I
= 075%
~a
42
Increases slightly w~th decreasing wall permeability For the relatively large gap w~dth
s/a = 001 the gap flow resistance is obviously small Therefore, the through-flow
resistance does not alter the gap pressure appreciably
-1.4
~'p,ex
Cpdnt
--0.8
-0.6
~Og~
--0.2.
0.0
h/a
4.
Fig 4 Influence of the gap flow resistance on the net pressure coefficient
43
]
i
-LO Q.-O 8 to
_o,
b/a=2 b/a=1
-O4 -.41.,7.
0.0 0
2
h/a
44
-60
--
~::001
s/a=O0025 + *
Cp,ex
~
k
C~p=03
~p=02
O~p= 0 1
CP,net . . . . . . . . . . . . . . . . . . . .
^
Cp,lnt
2 h/a
4 5 Correlatton between external and mternal pressures The evaluation of the correlation ume histories of external and mternal pressures measured simultaneously at the same budding locauon for the critical flow condmon (~t = 10 ) are being analysed at present The first results show that relatwely high correlauon coefficients (p ~ 0.9) are observed when the gap flow resistance is relatwely high (s/a < 0.0025). For small gap flow resistance, i e. large gap width (s/a = 0 01), the correlation coefficients are smaller (p ~ 0.5) when the gap between the windward face and wmd parallel side is sealed Negative correlataon coefficients may occur lmmedmtely downstream of the leading edge when the vertical gap is not sealed and the gap flow resistance Is relatively small (s/a = 0 01) Detailed reformation of the correlation wdl be presented shortly in the Ph D thesis by Janser [12]
45
met. station --
I /"-
" <
~ Oll
ca 7m
12 ~351/''~
//-If
/..IN. ." /
/ /
section A-A
between the rain screen and the bmldlng wall was open at the top and the bottom and sealed at the vertical budding edge and around the windows. A 1"150 scale model of the test building and its surroundings was budt for the wind tunnel study The porous facade was simulated by a sheet metal model facade having an appropriate number of 0.8 m m diameter holes The number of holes was chosen to achieve the same effective porosity as for the full scale facade The &stance between the model scale ram screen and the building wall was calculated to achieve the same gap flow resistance as for the full scale facade taking into account the pressure losses due to the vertical battens This reqmred a gap width of approximately 1/150 of the full scale width, l e 0 5 m m 5 2 Data acquisition The pressure differences across six facade panels were measured The locations of the pressure transducers are sketched m Fig 7. Measurements were taken without and
46
with alrUght sealing of the gap between the ram screen and the verucal edge at the corner of the building The dlfferentml pressure Ap.et was measured using surface probes and a manometer with a range of 1 k P a with an accuracy of _+ 1 Pa The analogue signal was dlglused with an A/D converter Wind velocity and wind direction were measured with a cup anemometer and a vane, respectively, on a mobile meteorological station fixed to the roof, see Fig 7 The signals (pressure differential, wind speed, wind direction) were scanned at a frequency of 676 Hz for a period of 3 02 s If the pre-set conditions (threshold velocity, wind direction range, stability of wind direction) are satisfied, the ume histories are converted into wind velocity, wind direction and pressure dlfferentml averaged over the measuring period of 3 02 s In addition, the rms value of the pressure fluctuations is calculated Finally, the pressures are non-dimenslonallsed with the wind stagnation pressure to give pressure coefficients (averaged over 3 02 s) and rms-pressure coefficients The pressure d~strlbutlon for the model scale study was obtained for 45 pressure tap locations In addiuon, the wind velocity and wind direction at the equivalent position of the mobile meteorological stauon were obtained for various undisturbed flow condlUons Six of the model scale pressure taps (nos 6, 11, 12. 23, 31 and 35) coincided with the six full scale pressure tap locations Standard pressure measurement techniques including a scanlvalve were used for the model studies
5 3 Results
A typical result of the full scale test IS shown in Fig 8 For pressure tap location 31 (distance from the vertmal edge of approximately 2 m) the coefficient of the net
02
01
_...~a,~
--
~- - 0 1
Io -02
w~thout edge sealing -03 -04
-0 5 , I , I , I , I , I ~ I , I t
45
go
3.35
180
225
270
315
360
flow d,rection
Fig 8 Net pressure coefficient for tap 31 with and without edge seahng
47
-=_-ol o
(3. ~0 - 0 2 -0 3
-0 4
"
"
-0 5 -0 6 -0 7
-0 8 I , I ~ I ~ I ~ I ,
with e d g e s e a h n g
I , I
45
90
135
180
225
270
315
360
flow direction
05 O4 O3 O2 full - scale - test i i i i i i i
~ o =. -o1
,~i. -02 -03 -0 4 -05 -06 -0 7 -08 I 45 , I 90 , I 135 , I 180 , wind tunnel expenment
. . . . . . . .
flow direction
Flg 9 Companson between full scale test and wmd tunnel experiment (tap 31) with (upper) and without (lower) edge seahng
pressure ts plotted versus wind direction Flow &rectlon north (0t = 0 ) is defined as perpen&cular to the side including the pressure taps 23, 31 and 35. The pressure differential across the rain screen is given with and without seahng of the verUcal edge For the wind direcUon range considered, the net pressure, and thus the wind load, without edge sealing is sigmficantly larger than for the case with edge seahng. As pointed out earher [1,2], the edge seahng prevents the horizontal pressure eqmhbratlon from one budding side to the other Thus, the
48
pressure equthbrates m a m l y across the facade panels leadmg to very low net wind loads Ftg 9 gtves for the same pressure tap locatton the compartson between full scale and model scale tests The upper dmgram shows the results wtth edge seahng and the lower dmgram without The agreement between full scale and model scale tests is sattsfactory It ts better for the case w~th verttcal seahng Wtthout vemcal seahng the gap flow becomes more intense and the net pressure coefficients larger One reason for the poorer agreement between full scale and model scale results could be a minor difference in gap flow resistance between those two cases For the full scale measurements pressure tap location 31 is close to a wmdow, see Ftg 7 The gap flow m the wcmlty of the wmdow ts not well defined and the influence of the window has not been modelled for the wmd tunnel test
References
[11 H J Gerhardt and C Kramer, Wind loads on wind permeable building facades, J Wind Eng lnd Aerodyn 11 (1983) 1-20 [2] H J Gerhardt and C Kramer. Wmdkrafte an hinterlufteten Fassaden, Betonwerk + FertlgtellTechmk, Heft 1/1985, pp 46-53 [3] H J Gerhardt and F Janser, Wind loads on wind permeable facade systems, in Proc 1st IAWE European and African Regional Conf on Wind Engineering, ed N J Cook (Thomas Telford, London, 1993) [41 H J, Gerhardt, C Kramer and K K Bofah, Wind loading on loosely laid pavers and insulation boards for fiat roofs, J Wind Eng Ind Aerod~n 36 11990) 309-318 [5] N J Cook. The designer's guide to wind loading of building structures, Part 2 Static structures {Butterworths, London, 1990) [6] H J Gerhardt, Wind loads on roofing systems roof membranes tiles and loose-laid slabs, in Proc 4th Canadian Workshop on Wind Engineering, Toronto, November 1984 [7] N J Cook, Simulation techniques for short test section wind tunnels Roughness, barrier and mixing device methods, in Wind tunnel modelhng for civil engineering applications, ed T A Rheinhold (Cambridge Uni~ Press Cambridge, 1982) [8] N J Cook, Manufacture and calibration of restrictors and averaging manifolds for the measurement of fluctuating pressures, BRE Note No 54:80 [9] M Knoch, Entwicklung einer Methode zur amphtudenrichtigen Messung schnell schwankender Drucke mlttels pneumatlscher MeBstellenumschalter, Studienarbett RWTH Aachen, Fakultat fur Maschmenwesen (1980) [10] B E c k , Techmsche Stromungslehre (Springer, Berlin) [11] A G Davenport, The application of statlstlcal concepts to the wind loading of structures, Proc Inst Civil Eng 19 (1961) S44%472 [12] F Janser, Windbeanspruchung belufteter Aul3enwande Ph D thesis TU Berlin, to be submitted in July 1994