Experimental studies of flow through

single gauzes
C.-C. Su and C.-C. Huang
Mechanical Engineering Department, National Taiwan University, Taipei, Taiwan, Republic of China

This article describes the experimental studies on the effect of f l o w Mach number and
Reynolds number, the open area ratio of the gauze, and the ratio of specific heats of the
fluid on the pressure drop of the f l o w through wire screens. The gauzes tested were of
ten different values of open area ratio. Air and helium were used as the test fluid. The
results s h o w that the pressure loss coefficient increases w i t h Mach number and that the
increase is sharp at choking. In the incompressible f l o w regime, the pressure loss coefficient
decreases w i t h increasing Reynolds number and open area ratio. However, the effect of
the ratio of specific heats is negligible. Empirical relations for the pressure loss coefficient
were derived based on these test results.

K e y w o r d s : pressure drop; single gauze; open area ratio; compressible f l o w

Introduction Experimentation
Single gauze has been used as a filter, to control turbulence, Eighteen single gauzes were tested. The specifications of these
and to create or eliminate large-scale velocity or pressure gauzes are listed in Table 1. The material of these gauzes was
nonuniformities. Due to the high ratio between their surface stainless steel except for gauze A1, which was copper. The wire
area and volume, finely woven gauzes are also suitable as the diameter and the aperture of the gauzes ranged from 0.036 to
packing material of the regenerator of the Stirling cycle machine. 0.16 mm and 0.05 to 0.63 mm, respectively. The open area ratio,
Unfortunately, no studies seem to have shed light on how the a, of the gauzes ranged from 0.31 to 0.64 and was divided into
flow characteristics of single gauze combine to produce the ten groups. Each group had the same value of a, but might
properties of the matrix of stacked gauzes used as regener- contain gauzes of different wire diameters and apertures, as
ators. A detailed study of the flow through single gauzes is shown in the table.
therefore necessary. The test circuit for the investigation of the pressure drop
The first experimental work on the pressure drop of the flow across single gauzes is shown in Figure 1. Detailed description
through single gauzes can be traced back to Adler. 1 The of the test system and the experimental procedure can be
variations of the pressure loss coefficient of the flow, f (defined found in the work of H u a n g ) ° The design working pressure
as the pressure drop across the gauze divided by the upstream of the system was 10 bar at room temperature. The test section
kinetic energy), versus the upstream Mach number, M, were shown was made of nylon 6/6. Single gauzes were clamped
plotted for different open area ratios of the gauze, a. Later on, between the two gauze holders of the test section. Tappings
the Reynolds number, Re, was pointed out to be another
governing factor of the pressure drop characteristics. 2 ~ Pinker
and Herbert 5 have tried to separate the individual effect of M , Table 1 Specifications of gauzes tested
Re, and ¢r on f. However, Roach 6'7 pointed out that the
influence of Re and M appeared to have been confused. The Aperture Wire diameter Mesh number Open area
studies were thus concentrated on the case of incompressible Gauze /, mm d, mm m, inch -1 ratio, a
flow. Su s also pointed out that Pinker and Herbert did not
A1 0.14 0.11 100
successfully separate the effects of these three governing factors, A2 0.14 0.11 101 0.31
since they used the irrelevant upstream Reynolds number. A3 0.05 0.04 280
Furthermore, based on the flow model of a one-dimensional B 0.071 0.05 210 0.34
(I-D) core outside the boundary layer developed along the wire Cl 0.2 0.125 78
surface, Su s found that the specific heat ratio, ~,, of the fluid C2 0.16 0.10 98
may also be an influencing factor. The deficiency in assuming C3 0.15 0.10 100 0.38
that the core is I-D was corrected later by Su, Hsieh, and C4 0.08 0.05 200
Chiu. 9 The results also show that the pressure-drop character- C5 0.062 0.04 250
istics are influenced by the four governing factors, namely, M, D1 0.066 0.036 250
D2 0.064 0.036 255 0.41
Re, a, and 7. Although Su's work s contains some experimental E 0.18 0.09 94 0.44
data of pressure drop across single gauzes, the range of a F1 0.23 0.10 78 0.48
investigated is quite narrow, i.e., from 0.365 to 0.379. The work F2 0.09 0.04 195
still to be solved is thus the experimental verification of these G 0.36 0.14 51 0.51
simulations with a convincingly broad range of a. H 0.32 0.11 59 0.64
I 0.51 0.16 38 0.57
J 0.63 0.16 32 0.64
Address reprint requests to Dr. Su at the Mechanical Engineering
Department, National Taiwan University, Taipei, Taiwan, R.O.C. Note: The material was stainless steel, except for gauze A1, which
Received 24 October 1 990; accepted 26 February 1991 was copper.

© 1991 Butterworth-Heinemann
Int. J. Heat and Fluid Flow, Vol. 12, No. 3, September 1991 273
 Flow through single gauzes: C.-C. Su and C.-C. Huang

~ to atmosphere single gauze was monitored by the differential pressure gauge

of 2 bar capacity. In addition, another differential pressure
rotameter gauge was used to measure the pressure drop in the downstream
conduit. The pressure immediately after the gauze could then
be obtained through extrapolation using measured pressure

ve
I P
i GDPGPG drop in the downstream conduit. However, the pressure im-
mediately before the gauze was estimated through measured
upstream pressure and the correlation for the pressure drop of
laminar flow in a circular duct. The distance between the
upstream pressure tapping and the section where the single
potentiometer thermocouples~ J gauze was located was 9 ram, while that between the two
tappings for the downstream differential pressure gauge was
12 ram. However, the first tapping behind the single gauze was
18 mm behind the gauze to avoid measuring the misleading
s w i t c h ' i c e water~ ~ ~
~-gQuzeholder pressure hump induced by the jet observed by Pinker and
Herbert. 5
single gouze---~fi
~/FT]
lter~ ~ "honeycomb" The absolute temperature at the upstream side of the gauze,
Tup, was measured with an ordinary thermocouple, while the
II .~LJf-uPstream conduit temperature difference between the upstream and downstream
sides was monitored with a differential thermocouple, as shown
conpressedgas in Figure 1. Both thermocouples were of copper/constantan
type and were calibrated beforehand.
Figure 1 Schematicof the testsystem(DPG,differentialpressure Two rotameters were used to monitor the volume flow rate
gauge; PG, pressure gauge) of the working fluid. Depending on the required flow rate for
a given flow condition, these meters were used independently.
were provided as shown for temperature and pressure measure- The capacity of these meters was 3.3 x 10- 3 m3/s (200 liter/min)
ments. The inside diameter of the assembly was 3.0 mm and and 16.7 x 10 -3 m3/s (1000 liter/min). These rotameters were
that of the conduits at the upstream and downstream sides was designed for measuring air flow rate. In the test with helium
20 mm each. A filter consisting of six layers of gauze C5 of 250 as the working fluid, the actual flow rate was estimated through
mesh per inch was clamped between the upstream gauze holder the square root of the density ratio between air and helium.
and conduit to prevent foreign materials from entering the test Before the test, the single gauzes were cleaned with acetone
section. This filter was cleaned each time the test gauze was in an ultrasonic cleaner. Compressed air and helium from a
changed. Since a pressure hump behind the single gauze had gas bottle were passed through the test section of the apparatus.
been observed by Pinker and Herbert, 5 the downstream tapping Pressure, temperature, and volume flow rate were then read
for the pressure measurement was located 18 mm behind the through the meters described above.
single gauze to avoid misleading results. The entrance and the
exit of the gauze holders were rounded for the purpose of
reducing the turbulence induced by the abrupt change in flow
area. In order to make the upstream flow uniform, the upstream Results and discussion
conduit contained a ring designed according to the criteria
suggested by Loehrke and Nagib. 11 This ring was 20 mm in Effect o f the M a c h number
length and was stacked with stainless steel tube of 0.63 mm Figure 2 shows the variation of the pressure loss coefficient, f,
O D and 0.33 mm ID. with upstream Mach number, M, for five values of open area
Four pressure gauges were used in the test. Two of them ratio, ~. These values of ~r cover the entire range of ~r
were of differential type, while others were of absolute type. investigated. Results of other values of a lie between those
These gauges had been calibrated with water and mercury shown in this figure. Note that f is defined as the ratio between
manometers. The maximum error in reading was within + 2.0% the difference in upstream and downstream pressures and the
of the capacity of the gauges; therefore, no correction was taken upstream kinetic energy of the fluid, i.e.,
in pressure readings. The arrangement of these pressure gauges
is shown in Figure 1. The upstream pressure was monitored f :-Pup-Pd" (1)
by the ordinary gauge, while the pressure difference across the (pu2/2 )~p

Notation Greek symbols

Ratio of specific heats
d Wire diameter = Re/2000
f Pressure loss coefficient p Density
l Aperture tr Open area ratio
M Upstream Mach number tr, i, Sinusoidal open area ratio (Equation (3))
M* Upstream choking Mach number
m Mesh number Subscripts
P Pressure dn Downstream conditions
Re Reynolds number inc Incompressible conditions
T Temperature isen Isentropic conditions
u Velocity up Upstream conditions

274 Int. J. Heat and Fluid Flow, Vol. 12, No. 3, September 1991
 Flow through single gauzes: C.-C. Su and C.-C. Huang

18 where g = nmd/2. By using %i,, the agreement between their

0"
test results and the isentropic choking Mach number, Misen,* is
o 0.31 n I"1 improved, as can be seen from Figure 3. However, if cr,i. were
16 " 0.38 0 used, M* of the present work would be much lower than Misen, *
o 0.44 " as shown in Figure 3. Since the results of Adler' also show that
no adjustment in open area ratio is necessary, the results of the
v 0.51 I"1 present work are considered more reliable.
14
• 0.64 o " The variation of f with M has the same tendency as those
obtained by the above-mentioned earlier works. Based on the
n & experimental results of the present work, an empirical relation-
12
ship between f and M is thus derived as
( M* "~°.16
f 10
[] A
o
8
f =fi,c\~] (4)

o The form of this expression is the same as that suggested by

Pinker and Herbert, 5 Su, s and Su, Hsieh, and Chiu. 9 The only
distinction is that the value of the exponent of the expression
is different for different works. The exponent in Equation 4 is
0.16, while those suggested by Pinker and Herbert, 5 Su, s and
Su, Hsieh, and Chiu 9 are 1/7, 1/5, and 0.176, respectively. The
difference in the general variation of the curve is not great, as
can be seen from Figure 4---in fact, more important is the
4 difference in the value of fi,c, which is defined as f at M = 0
and is obtained through the extrapolation of the measured
values.
Figure 4 shows that for a given Re, the values of f of the
present work are consistently lower than those of the experi-
mental results of Su. a A small contamination of the single gauze
i due to the accumulation of irrelevant materials on it may induce
0.03 009 0.15 021 027 033 039 a high value o f f , as pointed out by Su. a In the present work,
care has been taken to use a filter consisting of six layers of
M screen of 250 mesh per inch between the upstream gauze holder
Figure 2 f versus M for different values of a and conduit. The filter is considered to be able to effectively
reduce such a contamination. The results are therefore more

and that a is defined as 1.0

block area • present results ((7)
o'=1 o present results (0"sin)
flow area O. 9
• Adler 1
= (2) v Pinker & H e r b e r t S ( O )
0.~
• Pinker & HerbertS(Osin)
It can be seen from Figure 2 that for each group of a, f o Su 8
increases with M and the increase is sharp at some value of M, 03
2-D results s
indicating that choking has occurred. The choking Mach M*
number, M*, was suggested by Adler t and Pinker and Herbert 5 0.6
to be constant for a given a. However, Figure 2 shows that M*
varies somewhat for each a. In fact, the higher the Reynolds
number, Re, is, the lower M* tends to be. Nevertheless, the 0.5
variation is within about 5% of M* for each ~r. The choking
Mach number may therefore be taken as constant for each
0.4
given value of a. o
Figure 3 shows M* versus or. The discrepancy in M* between ~00 o []
the present measured results and those predicted by the 0.3
0
isentropic relationship between Mach number and free-flow B
area without considering the effect of the boundary layer is
relatively small. Physically, there must be a boundary layer 0.2
developing along the wire surface. However, the weaving of the
wire may make the free-flow area different somewhat from that 0.1
calculated using Equation 2. This weaving effect had been L-isentropic
pointed out by Pinker and Herbert. s Due to the high values
of M*, as shown in Figure 3, Pinker and Herbert suggested a 0
sinusoidal open area ratio, a,~,, defined as 0 0.1 0.2 03 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0". O',i.
a,i.=l-~{(l+292,(;-cos-'o)+30(1-02)
'/2} (3,
Figure 3 M* versus a and a,,.

Int. J. Heat and Fluid Flow, Vol. 12, No. 3, September 1991 275
Flow through single gauzes: C.-C. Su and C.-C. Huang

Re the present test results, the empirical expression for fi,c is

2O
• 680 "~ present results fi ~ /'1 + 18.7~
• 1380J nc=Jinc'Re=2OOO~ ` i~9.7~- ) 15)
18
o 670~ Sua
16 v 1340./ where fi.~.Re= 2000 is the incompressible pressure loss coefficient,
f -- -- 880"~ 2 - D results 9 fi.¢, at Re= 2000 and ~ = Re/2000. Equation 5 is also similar
14 1360 J o~ to those suggested by the above-mentioned earlier works.
540"~ 1-D results 8 o°v • However, some difference exists, as can be seen from Figure 5:
12 one 1340) o ~ • the values of fi.¢ of the present work are consistently higher
than both results of flow simulations. It seems that the results
10 v &l of the 2-D model are closer to the test results than those of the
v
1-D model. The present test work is therefore a justification of
8
the simulation of the 2-D model, even though the agreement
6 between measured and simulated results is not very satisfactory.

4
Effect of the open area ratio
2 Although Figure 5 is plotted for gauze C (a = 0.38), Equation 5
0 = I , I a i a I i I i I applies to other values of a. Now fi,c.Re=2000 must be deter-
0 0.04 0.08 0.12 0.16 0.20 0.24 mined before using Equation 5. Figure 6 shows the variation
of fi.c,Re=2000 with a. Based on the data shown in Figure 6,
M an empirical expression for fi.c,Re = 2000 is derived as
Figure 4 f versus M for cr= 0.38
log 1o fi,~.Re= 2000= -- 2.31 a + 1.46 (6)
The curves drawn with Equation 6 and the empirical equation
based on the 1-D model 8 intersect at about a=0.33, as shown
in Figure 6. At higher a, the values offi.c.se = 2000 of the present
results are consistently higher. The 2-D flow model also
demonstrates higher fi.¢,Ro= 2000 than the 1-D model. Although
6 the range of cr of the 2-D simulation is very narrow (0.36<
a<0.38), its results seem to support the present work. Figure 6
fine o e~- Equation (5) also shows the results of Adler I and Pinker and Herbert, 5 which
o are for reference only, since the value of Re in their works is
not known.

• o/x
o
o • present results
o
2 - D resul ts s
present results
10 o o 1-D results 8
• 2-D results 9
° ~ a ~ v Pinker & Herbents
o 1-D results 8
o Adler 1

0 = i ,I .... I , , I .... I , ,i .... I fine

10 100 1000 10000
Equation (6)
Re vo~
Figure 5 f*nc versus Re
go°
reliable. On the other hand, the values of f of the present
experimental work are higher than those obtained by both
\
simulations involving a one-dimensional (l-D) and two-dimen- 0
sional (2-D) core in the flow through single gauzes. Since the
flow through gauzes is very complicated due to the weaving of
the wire composing them, the present test results are therefore O
more trustworthy.
O
.1 I I I I I I I I I
Effect of the Reynolds number 0 0.1 0.2 03 0.4 0.5 0.6 0.7 0.8 0.9

The evaluation of f using Equation 4 requires knowing fine (7

beforehand. Figure 5 shows the effect of Re on f~.¢. Based on Figure 6 f~.cversus ~r for Re = 2000

276 Int. J. Heat and Fluid Flow, Vol. 12, No. 3, September 1991
 Flow through single gauzes."C.-C. Su and C.-C. Huang

14 i.e., Equations 4, 5, and 6. These expressions can be combined

into a single equation as
o h e l i u m (7"=1.67) z~ o,o
zx a i r ('1"=1.4) f= \ ~ f \ ~ f (8)
12 /.,,
0 ,,'k
z~ Diagrams showing the variation of f with two of the three
& governing factors, namely, M, Re, and a, for a given value of
the other factor can then be plotted with Equation 8. The results
10
A 0.6.
o&
o h e l i u m (Y=1.67)
z~ a i r (~'=1.4) o
0.5
8,,
z~P,/Pup

0.4

0.3 O~

2
0 o.i6 0.2

M
o~ ~
Figure 7 f versus
0.1
Effect of the specific heat ratio
In the simulation of the flow through single gauzes, the 1-D
and 2-D models 6'7 both indicate that the momentum thickness
of the boundary layer on the wire surface is a function of M, 0
Re, a, and 7. Pressure loss characteristics are therefore expected 003 0.09 0.15 0.18 021
to be affected by variation of 7. However, the effect of ~ on f
is negligibly small, as can be seen from Figure 7, in which f is M
plotted against M for two values of 7, i.e., 1.4 and 1.67 for air Figure 8 AP/Pu.versus
and helium, respectively. The results of the above-mentioned
simulations also show this tendency. In addition, the experi-
mental investigation by Su a on packed gauzes with air, helium,
and argon as the test fluids also supports this observation. The
suspicion that different fluids might have different pressure loss UD~
characteristics due to 7 is therefore removed.
However, AP/Pup,a practical parameter that can be measured O
easily, is affected by the variation of 7- Figure 8 shows that for
given M, Re, and a, AP/Pupof helium (~ = 1.67) is higher than
that of air (7 = 1.4). The difference is introduced by the ratio
between the definitions o f f and AP/Pup, namely,
AP/p.p_ Ap/p,p
f AP/(pu2/2).p
= 2t'YMz (7)
Equation 7 shows that for a given 7, (AP/P,p)/f is not constant
but varies with M 2. Therefore, the difference in AP/Pupfor Os
.~S
different ~, increases with M, and the difference is small at low
M, as shown in Figure 8.
%
It is clear, then, that f is a function of M, Re, and a. A 3-D
diagram showing pressure loss characteristics may be desired. • ~'
Figure 9 shows the variation of f with M and Re for a = 0.48
(gauze F). Since the surface shown is plotted with experimental
results, some ripples appear. To be practical, the 3-D diagram Figure 9 Three-dimensional diagram showing measured f versus
can be plotted with the empirical expressions discussed above, M and Re for cr=0.48

Int. J. Heat and Fluid Flow, Vol. 12, No. 3, September 1991 277
Flow through single gauzes: C.-C. Su and C.-C. Huang

will have the shape shown in Figure 9. (Note that Figure 9 is of the Republic of China for financial support of this work
plotted with experimental data and thus appears to have some under contract NSC77-0401-E002-21.
ripples.)

References

Conclusions 1 Adler,A. A. Variation with Mach number of static and total

pressures through various screens. NACA Wartime Report, No.
The pressure loss characteristics of the flow through single L5f28, 1946
gauzes were investigated experimentally. The resultant data 2 Annand, W. J. D. The resistance to air flow of wire gauzes. J.
R. Aero. Sci., 1953, 57, 141-146
were compared with those obtained with the simulation of the 3 Wieghardt, K. E. G. On the resistance of screens. Aero. Q.,
flow modeled as a core outside a boundary layer developed 1953, 4, 186-192
along the surface of the wire composing the gauze. The following 4 Cornell, W. G. Losses in flow normal to plane screens. Trans.
conclusions are drawn from this study: A.S.M.E., 1958, 80, 791-799
5 Pinker, R. A. and Herbert, M. V. Pressure loss associated with
1. The pressure loss coefficient, f, is a function of M, Re, and compressible flow through square-mesh wire gauzes. J. Mech.
tr. Although the specific heat ratio, y, may also be a governing Eng. Sci., 1967, 9, 11-23
factor, as indicated by flow simulations, the effect of ~ on f 6 Roach, P. E. The generation of nearly isotropic turbulence
is negligible. However, the ratio between the pressure drop downstream of streamwise tube bundles. Int. J. Heat Fluid
across single gauzes and the upstream pressure can be Flow, 1986, 7(2), 117-125
affected by ~. 7 Roach,P. E. The generation of nearly isotropic turbulence by
2. Depending on the upstream Mach number, choking may means of grids. Int. J. Heat Fluid Flow, 1987, 8(2), 82-92
occur within the aperture of the gauze. 8 Su, C.-C. An Enquiry into the Mechanism of the Pressure Drop
in the Regenerator o f the Stirling Cycle Machine. Ph.D. Thesis,
3. Before choking occurs, f increases with M. The increase is
Engineering Department, Cambridge University, Cambridge,
sharp at choking. England, 1986
4. For the flow in the incompressible regime, f increases with 9 Su, C.-C., Hsieh, S.-S., and Chiu, M.-C. A simulation of
decreasing Re. flow-through wire gauzes. Proc. 17th lASTED Int. Symp.,
5. Increasing tr has the effect of decreasing f. Lugano, Switzerland, 1989, 234-237
6. The variation o f f with M, Re, and a can be obtained through 10 Huang, C.-C. Experimental Studies of the Pressure Drop of
an empirical equation, i.e. Equation 8. the Flow through Wire Gauzes used in the Regenerator of the
Stifling Cycle Machine. Masters Thesis, Mechanical Engineering
Department, National Taiwan University, 1989
11 Loehrke, R. I. and Nagib, H. M. Control of free-stream
Acknowledgments turbulence by means of honeycombs: a balance between sup-
pression and generation. Trans. A.S.M.E. J. Fluids Eng., 1976,
The authors would like to thank the National Science Council 98, 342-353

278 Int. J. Heat and Fluid Flow, Vol. 12, No. 3, September 1991