You are on page 1of 11

Journal of Wind Engineering

and Industrial Aerodynamics 83 (1999) 455}465

Wind loads on porous structures


P.J. Richards!,*, M. Robinson",#
!Department of Mechanical Engineering, University of Auckland, Private Bag 92019, Auckland, New Zealand
"Robinson Design Ltd, Private Bag, Waimauku, NW Auckland 1250, New Zealand
#Ultra Span Inc, 15135, 86th Ave, Surrey, BC, Canada V3S 4T8

Abstract

The wind loads on porous structures depend strongly on the resistance to through #ow,
which may be characterised by the loss coe$cient. It is shown that for round wire mesh screens
the loss coe$cient is related to the porosity (b). For other structures the loss coe$cient is
a function of the porosity and the construction. It is therefore suggested that it is useful to use an
e!ective porosity (b ), which is the porosity of a round wire mesh screen with the same loss
%
coe$cient. It is shown that loads on porous structures are less than those on solid structures by
a factor (1!b ). When porous structures are at an angle to the wind, the e!ective loss
%
coe$cient is reduced by a cos2(h) factor, where h is the angle between the wind and a normal to
the surface. As a consequence the corresponding e!ective porosity increases and the loads
decrease. These concepts are shown to match results from a number of sources. ( 1999
Elsevier Science Ltd. All rights reserved.

Keywords: Wind loads; Porous strutures; Round wire mesh

1. Introduction

Porous structures are widely used for a variety of applications including windbreak
fences, bird canopies for horticultural crops, shade houses and hail shelters. Although
the wind loads on such structures are similar to those on solid structures the #ow
through the structure modi"es the pressure distribution and therefore requires consid-
eration. In general the wind loads on planar porous surfaces are a!ected by their
resistance to through #ow and the overall geometry of the structure: the size, shape
and angle to the wind. A number of these aspects will be discussed in the subsequent
sections.

* Corresponding author. Tel.: #64-9-3737599; fax: #64-9-3737479.


E-mail address: pj.richards@auckland.ac.nz (P.J. Richards)

0167-6105/99/$ - see front matter ( 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 6 7 - 6 1 0 5 ( 9 9 ) 0 0 0 9 3 - 8
456 P.J. Richards, M. Robinson / J. Wind Eng. Ind. Aerodyn. 83 (1999) 455}465

2. Resistance to through 6ow

The resistance of any porous structure to #ow through the structure may be
characterised by a loss coe$cient k, which relates the pressure drop across the
structure to the volume #ow rate per unit area through the structure in the form
*p"k0.5o<2 (1)
T
where o is the air density (+1.2 kg/m3) and < is the volume #ow rate of air per unit
T
area through the structure in m/s. The loss coe$cient itself is a function of the porosity
b of the structure, the type of construction and the Reynolds number (Re) of the
through #ow, such that
k"k(b,Re, construction) (2)
with
Area open to through flow
b" (3)
Total area
and
o< d
Re" T , (4)
k
where d is a characteristic dimension of the elements making up the structure and k is
the kinematic viscosity (+1.8]10~5 Ns/m2) for air.

2.1. The ewect of porosity on the loss coezcient

Fig. 1 shows data derived from Annand [1] and Hoerner [2] for round wire mesh
screens at high Reynolds numbers ('2000). Also shown is the correlation given by
ESDU item 72009 [3], which "ts most of the data shown but does not behave at low
and high porosities in the manner expected. Hoerner [2], (quoting Borda [4]) suggests
that at low porosities it is expected that
kN(1!b)2/b2, (5)
whereas at high porosities he suggests that
kNC (1!b)/b2, (6)
$
where C is the sectional free #ow drag coe$cient of the elements of the structure. The
$
division by b2 in Eq. (6) is included to allow for the increased velocity around an
element due to other elements. However, this formulation includes an allowance for
the blockage of the element onto itself, which would occur even if the element was in
isolation, and hence Eq. (6) tends to overcompensate. For a rectangular mesh, such as
that illustrated in Fig. 2, analysis of the geometry suggests that a better allowance for
this increase gives
kNC (1!b)/(1!0.75(1!b))2. (7)
$
P.J. Richards, M. Robinson / J. Wind Eng. Ind. Aerodyn. 83 (1999) 455}465 457

Fig. 1. Loss coe$cients for round wire mesh screens.

Fig. 2. Mesh geometry.

Although Eqs. (6) and (7) give similar values for k with porosities greater than 0.8, it is
found that Eq. (6) overestimates the size of k in the range 0.5(b(0.8 whereas Eq. (7)
matches the measured values more closely.
For cylindrical elements at Reynolds numbers in the range of 103}104 the sectional
free #ow drag coe$cient C +1.0. Eqs. (5) and (7) may be combined into a single
$
equation
k(b,Re'2000,round)"((1!b)8/b8#(1!b)4/(1!0.75(1!b))8)0.25 (8)
458 P.J. Richards, M. Robinson / J. Wind Eng. Ind. Aerodyn. 83 (1999) 455}465

Fig. 3. The variation of loss coe$cient with Reynolds number.

which, as illustrated in Fig. 1, also "ts the data reasonably well and has the logical
limits of kNR as bN0 and kN0 as bN1.

2.2. The ewect of reynolds number on the loss coezcient

With round wire mesh screens it is generally found that the loss coe$cient is
reasonably constant at Reynolds numbers greater than 2000 but increases at lower
Reynolds numbers. It may be expected that this low Reynolds number variation
would be similar to that of a circular cylinder. Mills [5] gives a correlation for the drag
coe$cient of a cylinder in cross #ow, which is valid for 1(Re(104. This takes the
form

C "1#10/Re2@3. (9)
$
Both Annand [1] and ESDU [3] give information on the variation of the loss
coe$cient of round wire screens with Reynolds numbers less than 2000. In either case,
as illustrated in Fig. 3, the trends are reasonably modelled by a function similar to
Eq. (9), such that

k(b,Re)/k(b,Re'2000)"1.0#14.5/Re0.75. (10)

In practice, many porous structures are constructed from woven or knitted fabrics
with element sizes between 0.25 and 1 mm. As a result typical Reynolds numbers are
P.J. Richards, M. Robinson / J. Wind Eng. Ind. Aerodyn. 83 (1999) 455}465 459

of the order of a few hundred. In many design situations it may be impractical to take
into account Reynolds number e!ects, in which case it may be reasonable to use
a single Reynolds number of say 300 and k(b)"1.2k(b,'2000). This should be
conservative in most cases since the highest wind loads will occur at the highest wind
speeds and hence the highest Reynolds numbers and lowest loss coe$cients.

2.3. The ewect of construction on the loss coezcient

The data presented in Sections 2.1 and 2.2 are strictly valid for round wire meshes
but is also useful in order to de"ne an e!ective porosity b for other types of
%
construction. In dealing with round wire screens it is reasonable to assume that the
minimum #ow area occurs at the point of greatest restriction in the geometry.
However, if a mesh is constructed from #at webs or sharp-edged elements then the
#ow will form a vena contracta and so the minimum #ow area is smaller than that
estimated from the structure. With such structures, we have found it useful to quote an
e!ective porosity. The e!ective porosity is de"ned as being equal to the porosity of
a round wire mesh screen with the same loss coe$cient. For porous structures made
from #at webs the e!ective porosity, as suggested by Morgan [6], may be about 2/3 of
the geometric porosity b. For example, the windbreak material Paraweb, as used by
Richardson [7], has a geometric porosity of b"0.475 but since it is constructed from
50 mm-wide #at webbing strips, may be expected to have an e!ective porosity
b "0.317 and hence from Eq. (8), k"4.8. Experimental measurements in the wind
%
tunnel at Auckland showed the value of k to be 5.3. With structures made from slats or
other elements with a depth comparable to their width it is suggested that b +0.75b.
%
Although the e!ective porosity of a structure may be estimated, it is usually better to
measure the loss coe$cient and hence deduce the e!ective porosity from Eq. (8) or
Fig. 1.
The Reynolds number e!ects on meshes made from #at webs or other structures is
likely to be slightly di!erent from that given in Section 2.2. However, Hoerner [2]
shows that both circular and square plates, normal to the #ow, have a drag coe$cient
at Re"10 about three times that at Re"104, which is in line with Fig. 3, and so we
might expect a similar Reynolds number dependency with meshes constructed from
#at webs or sharp-edged elements. Hence Eq. (10) may be a reasonable approximation
with most styles of construction.

3. Wind loads on rectangular planar porous surfaces

The wind loads experienced by rectangular planar porous surfaces depend not only
on the porosity but also on factors such as aspect ratio, orientation to the wind and
the e!ects of the ground, both in terms of the interaction between the surface and the
ground and the e!ect of the ground roughness on the turbulence of the wind. In many
situations it is di$cult to isolate any particular e!ect, however a number of trends are
discernable.
460 P.J. Richards, M. Robinson / J. Wind Eng. Ind. Aerodyn. 83 (1999) 455}465

3.1. The ewect of porosity on wind loads for surfaces normal to the wind

One of the few porous structures for which both full-scale and wind-tunnel experi-
mental data is available is a windbreak fence. Fig. 4 shows the drag coe$cient data
for winds normal to fences with length to height ratios greater than 9, derived from
Refs. [7}14], plotted against the e!ective porosity which was estimated from
the details of the structure. Also shown in the graph is a curve based on Taylor's
momentum theory (valid for k(3(b'0.4)), which as quoted by Cook [15]
gives
C "k/(1#k/4)2. (11)
D
Another theoretical approach is to assume that with high porosities most of the
#ow approaching a normal surface will pass through it and so
C Nk (12)
D
which together with Eq. (7) gives
C NC (1!b)/(1!0.75(1!b))2. (13)
D $
This equation, with C "1.0, is also shown in Fig. 4 and is very similar to Eq. (11)
$
at high-porosities but increases more rapidly for porosities less than 0.7. Although
there is little data at high-porosities to con"rm these trends it may be noted that
both of these high-porosity theories suggest a near-linear variation of drag coe$cient
with e!ective porosity. Extending this linear behaviour to lower porosities gives
a curve
C "C (Solid)](1!b ) (14)
D D %

Fig. 4. Drag coe$cients for windbreak fences (Wind normal to the fence).
P.J. Richards, M. Robinson / J. Wind Eng. Ind. Aerodyn. 83 (1999) 455}465 461

Fig. 5. The reduction in drag coe$cient of porous fabrics perpendicular to the onset #ow with e!ective
porosity. Wind tunnel data.

which passes through the porous fence measurements and intersects the axis at
C (Solid)+1.5. Which is in the midst of the fairly scattered solid wall measurements.
D
AS1170.2-1989 [16] suggests that the wind load on free standing walls, when the
wind is normal to the wall, is given by
C "1.2](1!b2), (15)
D
where b is the geometric porosity. Eq. (15) is also a reasonable "t to the porous fence
data if b +b but is signi"cantly higher than Eqs. (11) or (13) at high porosities. The
%
solid fence limit of 1.2 in Eq. (15) is supported by full-scale and wind-tunnel measure-
ments such as those in Refs. [12,14] but is substantially lower than those given in
Ref. [10].
Support for Eq. (14), which suggests that the wind loads reduce in proportion to
(1!b ), has been obtained from measurements in the University of Auckland wind
%
tunnels. A range of eight porous fabrics were tested in a 0.3 m]0.3 m closed section
wind tunnel to determine their loss coe$cient and then the samples were mounted on
a 0.3 m]1.5 m frame which was located in a 1.5 m]1.5 m slotted walls wind tunnel.
The roof and #oor of the tunnel adjacent to the 0.3 m-long ends were solid and so the
tests were e!ectively two-dimensional. Forces were measured parallel and perpendicu-
lar to the frame for a range of angles to the onset #ow. In order to provide data for
a solid object the frame was covered with plastic "lm. Fig. 5 shows the ratio of porous
to solid drag coe$cients plotted against e!ective porosity for the situation where the
fabrics were perpendicular to the #ow. The values for the e!ective porosity were
obtained by inverting Eq. (8). Although there is considerable scatter in this data the
general trend is in line with Eq. (14).
462 P.J. Richards, M. Robinson / J. Wind Eng. Ind. Aerodyn. 83 (1999) 455}465

3.2. The ewect of angle to the wind on wind loads

When a planar porous structure is not perpendicular to the wind the forces acting
on the structure may be split into a normal force due to the pressure di!erence and
a frictional drag force parallel to the plane of the structure. The frictional drag force
may be accounted for by using a friction drag force coe$cient in the range of 0.01}0.04
depending on the surface roughness [16]. This coe$cient applies when the wind is
parallel to the plane of the structure and may be reduced by a factor sin(h) at other
times [14]. The angle h is the angle between the onset wind and a normal to the
structure.
The normal force coe$cient C , which equals the drag coe$cient when h"03,
n
reduces as the angle h increases. Fig. 6 shows the results obtained from the
University of Auckland tests which were described in the previous section. With the
exception of the two most solid fabrics the behaviours are all very similar. This
behaviour may be reasonably matched by an extension to the theory used to derive
Eq. (11) which gives
k cos 2(h)
C (h)" (16)
n (1#k cos 2(h)/4)2
and hence

A B
C (h) 4#k 2
n "cos 2(h) (17)
C (Normal) 4#k cos 2(h)
n

Fig. 6. The variation of C with wind angle, high aspect ratio structures.
n
P.J. Richards, M. Robinson / J. Wind Eng. Ind. Aerodyn. 83 (1999) 455}465 463

Fig. 7. The variation of C with angle from normal, low aspect ratio structures.
n

Shown in Fig. 6 is Eq. (17) with k"2.0. It may be noted that if Eq. (16) is compared
to Eq. (11) then it appears that when the plate is at an angle h the e!ective loss
coe$cient is k cos2(h).
Both Seginer [9] and Richardson [7] have made experimental observations of the
variation of the drag force on slatted porous windbreak fences (k+5) with azimuth
angle. Each of them concluded that the data was matched by

C "C (Normal)] cos(h) (18)


D D
which also matches the general shape of the low to medium k data band in Fig. 6.
It is clear from Fig. 6 that with high solidity a change in behaviour is observed at
high angles. As might be expected, at high angles the solid plate induces circulation
and the normal force is proportional to the usual angle of attack. It does appear that
even relatively low levels of porosity do a!ect this circulation.
Similar e!ects are apparent in data obtained by Letchford et al. [17], who have
wind-tunnel tested 1 : 50 scale models of various porous and solid canopy roofs with
a nominal plan of 15 m]15 m. The two porous structures tested were made from
perforated plates with geometric porosities b"11% and 23%. Loss coe$cient
measurements on these two plates gave k"340 and 33, respectively. Monoslope roof
data was obtained at roof pitches of 73, 153 and 273, with the wind approaching from
both directions. Fig. 7 shows Letchford's results along with data given by Hoerner [2]
for solid square plates. The solid monoslope roof data of Letchford is very similar to
Hoerner's square plate data but shows slight di!erences for the two wind directions.
In each case a slightly larger normal force was obtained when the wind approached
from the low end of the roof. These di!erences are probably due to interactions with
464 P.J. Richards, M. Robinson / J. Wind Eng. Ind. Aerodyn. 83 (1999) 455}465

the ground plain, which was 5 m (full-scale equivalent) below the roof. With the
porous roof data there is a progressive reduction in the normal force with increased
porosity.

4. Conclusions

The wind loads on porous structures depends strongly on the resistance to through
#ow, which may be characterised by the loss coe$cient. It has been shown that for
round wire mesh screens the loss coe$cient is related to the porosity (b). For other
structures the loss coe$cient is a function of the porosity and the construction. It is
therefore suggested that it is useful to use an e!ective porosity (b ), which is the
%
porosity of a round wire mesh screen with the same loss coe$cient. It is shown
that loads on porous structures are less than those on solid structures by a factor
(1!b ). When porous structures are at an angle to the wind the e!ective loss
%
coe$cient is reduced by a cos 2(h) factor, where h is the angle between the wind and
a normal to the surface. As a consequence, the corresponding e!ective porosity
increases and the loads decrease. These concepts have been shown to match results
from a number of sources.

Acknowledgements

The assistance of Ultra Span Inc. in sponsoring some of the work reported in this
paper is gratefully acknowledged.

References

[1] W.J.D. Annand, The resistance to air #ow of wire gauzes, J. Roy. Aeronaut. Soc. 57 (1953) 141}146.
[2] S.F. Hoerner, Fluid Dynamic Drag, published by the author, 1965.
[3] ESDU72009, Engineering Science Data Unit, London, 1972.
[4] Borda, Experience sur la resistance des #uides, Mem. de l' Academie Royale des Science, Paris, 1763.
[5] A.F. Mills, Heat transfer, Irwin, Homewood, IL, 1992.
[6] P.G. Morgan, Flow through screens of low solidity, J. Roy. Aeronaut Soc. 66 (1962) 54}56.
[7] G.M. Richardson, A permeable windbreak: its micro-environment and its e!ect on structural loads,
J. Agric. Eng. Res. 38 (1987) 65}76.
[8] D. Painter, Measured porosity, drag and shelter: polymesh `Lenoa shelter cloth, F.A.O (Agric. Eng.)
Conference, Lincoln College, New Zealand, 1982.
[9] I. Seginer, Flow around windbreaks in oblique wind, Boundary-Layer Meteorology 9 (1975)
133}141.
[10] A.P. Robertson, R.P. Hoxey, P.J. Richards, Design code, full-scale and numerical data for wind loads
on free-standing walls, J. Wind Eng. Ind. Aerodyn. 57 (1995) 203}214.
[11] L.J. Hagen, E.L. Skidmore, Windbreak drag as in#uenced by porosity, Trans. ASAE 14 (4) (1971)
464}465.
[12] A.P. Robertson, R.P. Hoxey, P.J. Richards, W.A. Fergusson, Full-scale measurements and computa-
tional prediction of wind loads on free-standing walls, J. Wind Eng. Ind. Aerodyn. 67&68 (1997)
639}646.
P.J. Richards, M. Robinson / J. Wind Eng. Ind. Aerodyn. 83 (1999) 455}465 465

[13] E.F. Kay, Aerodynamic design of arti"cial windbreaks, ME Thesis, University of Auckland, NZ, 1984.
[14] J.D. Holmes, Pressure and drag on surface-mounted rectangular plates and walls, ninth Australasian
Fluid Mechanics Conference, Auckland, NZ, 8}12 December 1988.
[15] N.J. Cook, The designer's guide to wind loading of building structures, Part 2, 1985.
[16] Standards Australia, SAA Loading Code, Part 2: Wind loads, Australian Standard AS1170.2-1989.
[17] C.W. Letchford, A. Row, A. Vitale, J. Wolbers, Mean wind loads on porous canopy roofs, IV
Asia-Paci"c Symposium on Wind Engineering, Surfers Paradise, Australia, July 1997.

You might also like