Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

The Resistance to Air Flow of Wire Gauzes

by

W. J. D. ANNAND, B.Sc. A.F.R.Ae.S.


Downloaded from https://www.cambridge.org/core. Tufts Univ, on 05 Oct 2017 at 06:23:58, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S036839310013007X

(Rolls-Royce Ltd.)

SUMMARY:—Wire gauzes are used in air duct systems and wind tunnels for many
purposes. Methods of estimating the pressure loss through a gauze currently available
are often found in practice to be unsatisfactory. In this paper data obtained by several
experimenters are correlated. It is shown that the effect of Reynolds number on
pressure drop is similar for a large range of gauze porosities. An empirical method
of estimation of the pressure drop is given which can be used with confidence for gauzes
of porosity (defined as the proportion of area not blocked by the wires) from 0-27 to 1-00,
and over a range of Reynolds number, based on approach velocity
and wire diameter, from 20 to 600.

1. Introduction For all these applications it is desirable to be able to


Wire gauzes are used in air duct systems and estimate accurately the pressure drop which will be pro-
wind tunnels for a variety of important duties. Some duced by a given gauze—particularly so when the gauze
examples are: — is to be used to reduce the non-uniformity of a stream,
since in this case17' a certain definite value of the ratio
(/) to catch stones or other injurious material, as of pressure drop to approach dynamic head (\pV2) is
in engine air intakes. necessary for the most complete removal of the irregu-
(ii) to produce uniformity of the velocity distribu- larities. Nevertheless experience shows that the empiri-
tion across a duct, as in many wind tunnels cal formulae usually employed are often considerably
and other test apparatus. Non-uniformities in error. Generally speaking, these have been based on
are reduced because the gauze presents greater very limited data, and it is the purpose of the present
resistance to flow where the velocity is higher. paper to review and correlate the available information
Refs. 2 and 7 present some theoretical con- and to derive an empirical relation which can be used
siderations and experiments on this application. with confidence throughout a wide range of conditions.
The use of gauzes in diffusing ducts presents
special features in that the effect of the gauze
penetrates upstream, reducing the causes of Notation
non-uniformity as well as the non-uniformity a coefficient in the formula for the estimation
itself. Extensive experiments are reported in of k (see Section 5)
Ref. 3. d mean wire diameter
(Hi) to reduce the turbulence of a stream. Much D1,D3 " depths of weave " in the two directions (see
use is made of fine gauzes for this purpose in Fig. 1)
low-turbulence wind tunnels. The components
of turbulence are modified in the contraction *1» *2 pitches in the two directions of the weave
on entering the gauze and, provided that the (see Fig. 1)
gauze wires are fine and the porosity suitably k pressure drop coefficient = pressure
chosen, the overall turbulence some distance drop/ipK 2
downstream can be considerably reduced below R Reynolds number =Kd/v
that in the approaching stream. Experiments 13 porosity
on this effect are reported (for example) in V approach velocity
Ref. 2, and Ref. 7 gives some theoretical con-
siderations. V kinematic viscosity

(iv) oppositely, to produce artificially high turbu-


lence. For this purpose, coarse grids of rela- 2. Definitions
tively large diameter rods are used; the large The main dimensions describing the geometry of a
eddies in the rod wakes presumably produce gauze are the mean wire diameter d, and the pitches /,
by their breakdown the turbulence required. and l2 in the two directions of the weave (see Fig. 1).
(v) to introduce known pressure drops into experi- From these can be calculated the porosity /?, defined as
mental systems. the area of the holes in a unit area of gauze, as projected
orthogonally onto a parallel plane. They are related by
Originally received June 1952; revised January 1953 P= [l-(d/l1)][l-(dll.z)].
VOL. 57 MARCH 1953 141
142 VOL. 57 JOURNAL OF THE ROYAL AERONAUTICAL SOCIETY MARCH 1953

Two other dimensions characterising the gauzes are


the " depths of weave " D1 and D2 in the two directions.
Fig. 1 makes clear the meaning of this term. Only one
investigator has measured these dimensions for the
gauzes tested. Although they might be expected to have
some influence on the flow pattern through the gauze,
the indications are that the effect on pressure loss is very
Downloaded from https://www.cambridge.org/core. Tufts Univ, on 05 Oct 2017 at 06:23:58, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S036839310013007X

small for gauzes of normal manufacture.


Any one of the gauze dimensions can be used to
form a Reynolds number characterising the flow. The -JD 2
diameter d has been chosen, giving a Reynolds number
R = Vdh, where V is the approach velocity and v the FIGURE 1. Diagram showing characteristic dimensions
of a gauze.
kinematic viscosity. It will be shown later that this
choice gives good correlation.
The pressure drop is defined as the loss in stagnation tappings about 5 cm. upstream and downstream of the
pressure from a plane immediately upstream of the gauze in a tube of 45 cm. diameter. Their results are
gauze to one sufficiently far downstream for the wakes presented only in graphical form, and it is difficult to
of the wires to have become indistinguishable by mix- pick off the individual test measurements accurately.
ing—that is, about 30 wire diameters downstream, to For this reason only the results for four gauzes have
judge from certain measurements on a coarse grid re- been made use of for analysis (unfortunately, not includ-
ported in Ref. 4. In practice the pressure drop is usually ing the gauze of greatest blockage; the dimensions
measured by the difference in static pressure between quoted for this are inconsistent).
planes some distance upstream and downstream of a
gauze sample in a duct of constant section. (b) Taylor and Davies"1' tested four gauzes, measur-
The pressure drop coefficient k is defined as the ratio ing the loss of static pressure between tappings about
of this loss to the approach dynamic head \pV2. 2 in. upstream and downstream of the gauze in a tube
of about 3 2 in. diameter. Their results are tabulated
and the measuring accuracy is good.
(c) Simmons and Cowdrey(4) tested eight gauzes,
3. Sources of Data measuring the loss of static pressure between tappings
Test results have been drawn from four sources, in 6 in. upstream and downstream of the gauze in a 6 in.
all of which each gauze was tested at a number square duct. Their results are tabulated and the
of air speeds. measuring accuracy is very good.
(a) Eckert and Pfluger(5) tested 16 commercial (d) Schubauer, Spangenberg and Klebanoff(2) tested
gauzes, measuring the loss of static pressure between 10 gauzes, using a rather different arrangement from

FIGURE 2. Representative selection of


o experimental points showing
correlation.
W. J. D. ANNAND THE RESISTANCE TO AIR FLOW OF WIRE GAUZES 143

any of the foregoing. The gauzes were fastened across 30


the end of a square duct and air was blown through
them into the open atmosphere. The pressure drop was
measured as the difference in pressure measured by
static probes placed centrally upstream and downstream
y io'
of the gauze. Here again the results are given only u.
graphically on a rather small scale, and the accuracy UJ
O
Downloaded from https://www.cambridge.org/core. Tufts Univ, on 05 Oct 2017 at 06:23:58, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S036839310013007X

o
with which points could be picked off is rather low.
O 0-5
Results for six of the gauzes were used in the analysis.
The measurements include the pressure drop due to
0.J! *—-
wall friction over different lengths of ducting. It is
impossible to estimate this loss accurately because the 0-2' ' 1 1 '
presence of the gauze completely alters the velocity IO 20 3-0 4-0 SO 60
profile near the wall, both upstream and downstream. LOG R

However, it is improbable that the correction to the FIGURE 3. Drag coefficient of a circular cylinder.
measured pressure drop coefficients would exceed
003, so that the omission of the correction is not of against log R for two gauzes from each source. In the
importance. complete plot, individual points would be indistinguish-
The dimensions of the gauzes included in the analysis able near the mean line; of 178 points representing tests
are given in Table I. Most are of square mesh or nearly on 22 gauzes, only 16 fall away from the mean line by
so; the greatest deviation from squareness is /2 = 0-80/1. more than 3 per cent, of k, and of these 16 all but five
are points picked off from the graphs of Refs. 5 and 2.
This degree of correlation is quite striking, and in
4. Variation of Pressure Drop Coefficient some respects was unexpected. Some variation of the
with Reynolds Number shape of the Reynolds number curve with ;8 was
The first step in the analysis was to plot all the expected, because the variation of k with R probably
pressure drop coefficients logarithmically against Rey- arises partly from variation of the discharge coefficient
nolds number based on the wire diameter. It was soon of the orifices formed by the meshes, and this variation
noticed that the forms of the curves obtained for the would be expected to depend on the contraction ratio of
various gauzes were very similar, and on investigation these orifices—that is, on 1//3. Further, it would be
it was found that identical curves, shifted only along the expected that at very high /3 the variation of the loss
log k axis, could be fitted to all; or alternatively, by coefficient with R would approach that of the drag
shifting each set of points parallel to the log k axis, all coefficient of an infinite circular cylinder; this curve is
the points could be brought onto a single curve. This plotted in Fig. 2 for comparison and it will be seen to
is illustrated in Fig. 2 where (log k - log kR=im) is plotted differ somewhat from the curve for k. Nevertheless, no
significant alteration of the curve shape with fi could
be detected over the range covered (0-21 < / 3 < 0 - 8 1 ) .
TABLE I
Again, in Ref. 2 observations are reported on the
DETAILS OF GAUZES TESTED
eddy formation behind several gauzes. It was found
Wire
that at very low Reynolds number regular eddy forma-
Pitches Weave depths Poro-
diam. sity tion behind the wires of a gauze did not occur; at higher
d: in. /j .- in. / 2 : in. hlh DJd DJd Reynolds number eddies were cast off with regular
00059 00125
periodicity. These two states presumably correspond to
Ref. 5 0-277
00055 00154 0-413 the observed flow pattern changes round isolated circu-
00098 00547 0-672 lar cylinders for which, at low Reynolds numbers, two
00118 00856 0-743 vortices form just behind the cylinder and remain there,
Ref. 6 00087 00251 0-0251 100 1-25 2-61 0-429 while at higher Reynolds numbers vortices are cast off
0013 00417 00417 100 1-53 2-23 0-475 alternately from opposite sides, forming the well-known
0020 0-0850 00850 100 1-94 2-15 0-585 " Karman trail." It was found in the tests of Ref. 2,
0-024 0126 0121 0-96 1-89 205 0-649 made on six gauzes, that the " eddy critical" Reynolds
Ref. 4 00044 0-0103 00099 0-96 0-318 number, at which the change-over from one regime to
00053 00126 00125 0-99 0-333 the other took place, varied directly with /?. This corre-
00068 00167 00167 100 0-351 lation would be expected to apply also to the variation
00067 00192 00166 0-86 0-389
00107 00333 00318 0-95
of k with R—that is, the curves of log k would be ex-
0-451
00075 0-0250 00249 100 0-490 pected to be similarly positioned when plotted against
00072 00417 0-0333 0-80 0-649 log (Rlfi) instead of log R. However, plotting shows
0-0147 00869 00832 0-96 0-682 that this is not the case (see Fig. 2).
Ref. 2 00135 0-025 0-212
00170 0050 0-436
The overall variation of loss coefficient over the
00055 00185 0-494 range of Reynolds number covered is 2 5 : 1 . By analogy
0007 0025 0-519 with the drag coefficient curve for an infinite cylinder,
00075 0042 0-672 the log k curve may be expected to continue to rise as
0-025 0-250 0-810 log R is further reduced, and it may also be expected
144 VOL. 57 JOURNAL OF THE ROYAL AERONAUTICAL SOCIETY MARCH 1953

TABLE III
M E A S U R E D V A L U E S O F k AT R = 100

\ Source P k 0-71 ( l - / ? 2 ) / ^

Ref. 5 0-277 8-22 8-53


0-413 406 3-46
0-672 0-81 0-86
Downloaded from https://www.cambridge.org/core. Tufts Univ, on 05 Oct 2017 at 06:23:58, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S036839310013007X

0-743 0-57 0-58


Ref. 6 0429 3 31 3-15
^—---. 0-475 2-45 2-44
-o-'' 0-585 1-47 1-36
0-649 0-79 0-98
Ref. 4 0-318 610 6-32
FIGURE 4. Possible extension of curve of Fig. 2 to higher 0-333 5-66 5-68
Reynolds numbers. 0-351 512 505
0-389 412 3-99
0-451 2-99 2-78
that large variations may occur at considerably higher 0-490 2-24 2-24
Reynolds numbers, despite the fact that the curve flat- 0-649 0-79 0-98
tens out at the upper end of the range covered. In Fig. 0-682 0-81 0 81
3, the cylinder drag curve is plotted over a wider range Ref. 2 0-212 13-21
of Reynolds number, and it will be seen that after 0-436 2-74
flattening out in the range of R between 100 and 250, 0-494 1 71
the curve drops again at about 350, then rises, levels off, 0-519 1-45
0672 0-72
and finally drops sharply at R ^ 2 x 105. Some idea of 0-810 0-38
the actual trend of the gauze pressure drop curve at
Reynolds number up to about 4,000 may be obtained
from three measurements made by Simmons and likely explanation is that the method of measurement
Cowdrey'4' on a grid of rods 0-25 in. diameter at 0-9 in. of the pressure on the downstream side of the gauze was
pitch. The observed values of k are given in Table II. at fault. The gauze under test was supported on flanges
attached to the duct outlet; it is known that a jet issuing
from a duct with a flanged outlet into still air produces
TABLE II
a small depression across the outlet, and it is probable
M E A S U R E M E N T S ON A C O A R S E G R I D O F 0 2 5 I N . that this depression is not uniform across the section.
DIAMETER RODS* 4 ', (/3 = 0-522) These results therefore were excluded from the investi-
gation of the variation of k with (3.
A ir speed V
ft. 1 sec. R k Some assistance in deciding on the form of the rela-
tion between k and (3 to be used as a basis for analysis
10 1310 1-34
20 2620 1-45
can be obtained from simple theoretical ideas. Consider
30 3930 1-50 a stream filament which just passes through a single
mesh. It suffers first a contraction down to the " throat"
of the mesh, during which its mean velocity is increased
In Fig. 4, log k - log k,t=100 is plotted against log J?; the to VI'/?; if it is assumed that losses during this process
method given in the next Section was used to estimate are negligible and that the stream breaks away from the
A. at R = 100. A curve somewhat similar to the cylinder wires at this section and then expands freely to the
drag curve can be used to link these points to the original area, the stagnation pressure loss during this
mean curve established up to i? = 600, as shown by process being that corresponding to a sudden expansion
the dotted line. through the area ratio (1 / /3), it is easily shown that the
loss coefficient k is equal to (1 -f3)2/ /32. In practice, the
flow pattern will not be as simple as this. There will be
5. Variation of Pressure Drop Coefficient friction losses at the wire surface, with the formation of
with Porosity boundary layers which will increase the effective con-
traction, and the point of breakaway probably moves
Using the general curve of Fig. 2 to extrapolate
round the wire as Reynolds number varies. However,
where necessary, it is possible to pick off for each gauze
it seems reasonable to investigate first the suitability of
the value of k corresponding to a given Reynolds num-
ber. Table III gives the values so obtained at R = 100.
Preliminary plotting showed that the values of k so TABLE IV
obtained from the results of Ref. 2 were consistently ISOLATED MEASUREMENTS FROM REF l
lower than the others. It seems probable that this is
a genuine difference arising from the different experi- k corrected
mental arrangement used. Absence of constraint on the p TestK k to 7? = 100 0-71 ( l - ^ 2 ) / / ? 3
downstream side of the gauze may have affected the 0-372 102 4-67 4-68 4-41
flow pattern, but it seems unlikely that this could have 0-469 165 210 2-37 2-52
had much effect near the centre of the gauze; a more 0-554 252 1-20 1-47 1 60
W. J. D. A N N A N D THE RESISTANCE TO AIR FLOW OF WIRE GAUZES 145

TABLE V
] i
VALUES O F a F O R V A R I O U S V A L U E S O F R \ ! i

R 20 25 30 40 50 60 70 80 90
a 1 30 118 109 0-97 089 084 080 0-76 0-73
Downloaded from https://www.cambridge.org/core. Tufts Univ, on 05 Oct 2017 at 06:23:58, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S036839310013007X

100 200 300 400 500 600


0-71 0-60 057 0-56 0-56 0-56

the slightly more general formula k = a(l+b/3 + c/32)l/32. —


If the formula is to be usable for gauzes of high /?, it 14 1-6 2 0 22 24 2 6 2ff
must give the value k = 0 for /3=10, corresponding to
the empty duct. The numerator must therefore FIGURE 6. Variation of multiplying factor a with
contain a factor (1 - /?), and it may be written Reynolds number.
k = a (1 - j8)(1 - c/3)/(32. By plotting k/32/(l - /?) against
/3, it was found that the best mean curve of this form is
Some indication that the formula can be used down
given by
to /3 = 0-21 is given by the fact that a similar formula
0-68(l-j3)(l + l-ljS) with a lower coefficient fits the results of Ref. 2 satis-
factorily down to this value.
Calculation showed, however, that over the range of the
tests (/? <( 0-27) the values of k given by this formula are 6. Working Method of Estimation of k
not significantly different from those given by the
simpler equation The working formula for the estimation of the
pressure drop coefficient is then
0-71 (1-/8 2 )
*= a(l-/32)

In Tables III and IV and Fig. 5, this relation is com-


pared with the measurements, including some isolated where a depends on the Reynolds number based on wire
measurements from Ref. 1 not used in deriving the diameter, as shown in Fig. 6 and Table V.
empirical formula. Agreement is quite good. The To assist in practical application, the nomogram
curve falls slightly high at /? less than 0 35 and slightly given in Fig. 7 has been prepared. From this chart the
low for /J between 035 and 0 4 5 . Better agreement multiplying factor a can be obtained directly from the
could probably be obtained by fitting a more elaborate values of the wire diameter, air speed, air temperature
equation containing higher powers of /3, but it has not and air pressure. The method of use is indicated by the
seemed to the author that the accuracy of the experi- dotted lines; join diameter and air speed and find inter-
mental data would justify this elaboration. section on reference line I; join air temperature and
pressure and find intersection on reference line II; join
the two intersection points and read off a. The point on
reference line II corresponding to I.C.A.N. standard sea
level conditions is ringed.
It should, perhaps, be emphasised once more that
the curve of variation of a with Reynolds number
B REFERENCE 5 should not be extrapolated outside the range of the test
X REFERENCE 6 data. The formula can be used with confidence for /?
O REFERENCE 4
not less than 027 and is probably adequate down to
A REFERENCE 1
/3 = O-20; below this value, there is no evidence as to the
adequacy of the formula, and testing would be advisable.
One other point concerning practical application
should be remembered. That is, that accurate estima-
tion of (i and of k at low values of ft calls for very
accurate measurement of the gauze dimensions. Table
VI gives the percentage error in k produced by a one

TABLE VI
8 * S L ^
E F F E C T O N ESTIMATE O F k O F O N E P E R C E N T . E R R O R
OF W I R E D I A M E T E R

/8 0-2 0-3 0 4 0-5 0-6 0 7 01


FIGURE 5. k plotted against 0 at R = 100.
Curve drawn is Jt = 0-71 (l-/? 2 )//? 2 . Percentage error of estimated A: 5 2 36 28 22 1-8 1 5 13
146 VOL. 57 JOURNAL OF T H E ROYAL AERONAUTICAL SOCIETY MARCH 1953

400-
-0-2

300-
DOTTED LINE SHOWS METHOD OF USE ;
FOR TEMPERATURE = I 5 ° C , PRESSURE' 6 LB./lN. 2 ,
Downloaded from https://www.cambridge.org/core. Tufts Univ, on 05 Oct 2017 at 06:23:58, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S036839310013007X

OS6 200- SPEE0=4O F T . / S E C , DIAMETER = 0 0 0 7 IN.,


-Ol
THUS GIVING a. = 0-83
-O 09
-O 08
-O 07
058
IOO -006
90 -OOS
0-60- 80 —
70 -004
60 —
P — -80
SO — -O 03
UJ 60
0-70- a.
40- ->
K — -40
a -002 <
rr
O m \l 3 in — -20
iu 30 — — n
O 080- o. z — 5 O
(/> — 4
< \ UJ
m
_i
S £- — + 2 0
< a. — + 4 0
-..—
• • * —

0-90- 20 \ UJ < — + 60
rr
-O Ol 3
IO
IOO- -O 009 m —
LU
•O 0 0 8 rr
Q_
1IO- •O 0 0 7 — ?o
a:
I-20-
IO — •0006 < JO
I 30- 9 -O OOS 40
8 — SO
I 40- x
— n - n n4i
7- "°°° STANDARD
6- SEA LEVEL
IOO
-0003 TEMPERATURE
AND PRESSURE
FIGURE 7. Nomogram for determination of multiplying factor a

per cent, error in measurement of the wire diameter d 2. SCHUBAUER, G. B., SPANGENBERG, W. G. and KLEBANOFF,
at various values of j3. At /3 = 0 3 , for example, one per P. S. (1950). Aerodynamic Characteristics of Damping
Screens. N.A.C.A. T.N. 2001, January 1950.
cent, error in d produces 3 6 per cent, error in k. Now
one per cent, error in d may, for the wire sizes commonly 3. SCHUBAUER, G. B. and SPANGENBERG, W. G. (1947). The
Effect of Screens in Wide-angle Diffusers. N.A.C.A. T.N.
used in gauzes, be less than 00001 in., so that the need 1610, June 1947.
for accuracy is evident.
4. SIMMONS, L. F. G. and COWDREY, C. F. (1945). Measure-
Provided that the limitations are observed and the ments of the Aerodynamic Forces Acting on Porous
necessary accuracy of gauze measurement obtained, the Screens. R. & M. 2276, August 1945.
method, according to a rough statistical estimate, should 5. ECKERT, B. and PFLUGER, F. (1941). Resistance Coefficients
give the pressure drop within 20 per cent, on 95 per cent, of Commercial Round Wire Grids (translated from Luft-
of occasions. fahrtforschung, Vol. 18, No. 4). N.A.C.A. T.M. 1003,
January 1942.
6. TAYLOR, G. I. and DAVIES, R. M. (1944). The Aerodynamics
of Porous Sheets. R. & M. 2237, April 1944.
REFERENCES 7. TAYLOR, G. I. and BATCHELOR, G. K. (1949). The Effect of
1. COLLAR, A. R. (1939). The Effect of a Gauze on the Wire Gauze on Small Disturbances in a Uniform Stream.
Velocity Distribution in a Uniform Duct. R. & M. 1867, Quarterly Journal of Mechanics and Applied Mathematics,
February 1939. 2, 1949, 1.

You might also like