International Journal of Heat and Fluid Flow 65 (2017) 60–72

Contents lists available at ScienceDirect

International Journal of Heat and Fluid Flow

journal homepage: www.elsevier.com/locate/ijhff

Pressure drop measurements for woven metal mesh screens used in

electrical safety switchgears
W. Bussière a,∗, D. Rochette a, S. Clain b, P. André a, J.B. Renard c
a
Université Clermont Auvergne, CNRS/IN2P3, LPC, Clermont-Ferrand F-63000, France
b
Centro de Matematica (CMAT), Campus de Gualtar, Universidade do Minho, Braga 4710 – 057, Portugal
c
LPC2E CNRS, 3A Avenue de la Recherche Scientifique, Orléans Cedex 2 45071, France

a r t i c l e i n f o a b s t r a c t

Article history: The paper deals with an experimental investigation of the pressure drops and friction factors induced by
Received 4 May 2016 sets of metal woven screens in the case of an incompressible fluid flow, namely water flow. These woven
Revised 25 January 2017
screens are of metallic type and are used in electrical safety devices, especially at the end of exhaust duct
Accepted 18 February 2017
of low voltage circuit breakers. In a first step, the whole of the set-up is used to check some approxima-
tions dedicated to pressure drops due to sphere beds around 275 μm and 375 μm as diameters. Results
Keywords: are discussed in terms of Darcyan and non-Darcyan permeabilities compared with published data and
Pressure drop are analyzed in terms of the Blake-type friction factor. In the second step, the pressure drops are mea-
Ergun law sured for stacks composed of eight different woven screens (of plain dutch type) formed with millimetric
Drag coefficient
wires. Various formulations of the pressure drops and friction factors published elsewhere are tested with
Friction factor
a special care dedicated to the choice of the geometry for the flow pattern (hydraulic diameter, spher-
Metal woven screens
Porous media ical diameter, cylindrical diameter) and to the consideration of laminar and turbulent contribution. We
then give the formulation that characterizes the fluid flowing through stacks of woven screens used in
electrical safety applications.
© 2017 Elsevier Inc. All rights reserved.

1. Introduction
Almost all the studies dealing with woven metal screens essen-
tially refer to topics such as high efficiency heat exchangers, en-
ergy storage units, regenerators, electronics coolers, catalytic reac-
tors, filtering industry, gas bearing supply systems, gas-liquid two
phase flow applications ((Wu et al., 2005; Belforte et al., 2009;
Kolodziej et al., 2009; Mahjoob and Vafai, 2008) and references
herein). But up to our knowledge, there is no published work with
experimental measurements of the drag coefficients dealing with
stacks of woven metal screens used for electrical safety purposes
which is the aim of the present work.
Electrical safety practices are of crucial importance in elec-
trical devices industry and have to follow the rules im-
posed by standards from IEC (www.schneider-electric.com 2007;
IEC 2003) and specific safety guidelines for example summarized
in (Schneider Electric–Square D 2003). In a recent past, filter tech-
nologies have been widely studied in order to improve exist-
∗ nition of an electric arc (typically between two metallic contacts
Corresponding author at: Laboratoire de Physique de Clermont, LPC CNRS
UMR6533, Campus Universitaire des Cézeaux, 4 Avenue Blaise Pascal, TSA 60026,
CS 60026, Aubière Cedex 63178, France.
E-mail address: william.bussiere@uca.fr (W. Bussière).
ing solutions and their associated performance especially for the
Medium Voltage (MV) electrical safety (Morel and Rival, 1990). We
recall that among the whole of the recommendations in relation
with the current study, the filter technologies have to ensure an
efficient working of the electrical switchgear and to help the clear-
ing of the very high energy level linked to the internal fault arc
clearance, with a specific capacity to be proof against very high
pressure and shock waves on one hand, and the brutal flow out
of hot gases at the exhaust (up to around 10 kK in the center of
the MV cell) on the other hand. Filters are thus designed to limit
the temperature, the velocity and the pressure of the gases at the
exhaust.
As a consequence innovative solutions have been carried out in
order to ensure the mandatory level of electrical safety. These im-
provements and the general rules are depicted in various patents,
among them we refer to Morel and Rival (1990), Rival et al.
(1999) and Faber et al. (2003).
Whatever the voltage range considered, MV or LV (Low Volt-
age), the electrical safety switchgears are designed to support the
clearance of the electric fault. This is done by means of the ig-
or electrodes or splitter plates) and hence of the consecutive arc
plasma which has to be considered as a time dependent resistive

http://dx.doi.org/10.1016/j.ijheatfluidflow.2017.02.008
0142-727X/© 2017 Elsevier Inc. All rights reserved.
 W. Bussière et al. / International Journal of Heat and Fluid Flow 65 (2017) 60–72 61

So it is not realistic to recreate and use such media to obtain re-

Nomenclature liable measurements for Darcyan and non-Darcyan permeabilities
(short duration, high gradients difficult to reproduce for systematic
a specific surface area (m −1 ) measurements, destructive hot power source). In a previous work
A, B coefficients in Ergun equation (-) (Bussière et al., 2006) we have done some measurements with air
A1 , A2 coefficients in Blake-type equation (-) as the test fluid.
C1 , C2 coefficients in Ergun-Bird equation (-) Among the possible risks that have to be taken into account,
CW wall correction factor in inertial term (-) the protection against arc flash is of high interest both for the
d1 warp wire diameter (m) safety of the maintenance staff and for the vicinity of the elec-
d2 weft wire diameter (m) trical apparatus. In the case of MV cells, many studies have
dp particle diameter (m) shown the significant role played by porous filters (Besnard, 2007;
D diameter of a single screen (m) Rochette et al., 2007; Rochette et al., 2008; Rochette et al., 2010;
DC cylindrical diameter model (m) Rochette et al., 2011) which are mainly of mineral kind and com-
Dh hydraulic diameter (m) posed of a deformable granular medium fixed by metallic screens
Dp , Deq particle equivalent diameter (m) (whose mean granulometry and porosity are carefully checked). In
DS spherical diameter model (m) these latter works, we have studied the specific influence of the
f, F friction factor (-) porous filters on fundamental physical properties playing an im-
k1 viscous constant of the Darcy–Forchheimer portant role during the MV switchgears work namely the pressure,
equation (m2 ) or Darcyan permeability term or temperature and velocity distributions in the MV cell and at the
permeability coefficient exhaust of the MV cell.
k2 inertial constant of the Darcy–Forchheimer More recently the filter technologies have been adapted to the
equation (m) or non-Darcyan permeability LV switchgears to take advantage of the improvements obtained
term or inertial coefficient in the MV range (Faber et al., 2003). The filter technology is dif-
L length of the porous medium (m) ferent in so far as the dispersed granular medium is not used
Le effective flow length (m) any more. The filter consists in one or more woven metal mesh
M wall correction factor (-) screens defined by different wire diameters, mesh sizes and porosi-
P pressure drop (Pa) ties which are placed on an optimal location at the gas exhaust of
Re, NRe Reynolds number (-) the switchgear. Obviously, as for the MV range, filters used in LV
T water temperature (°C) switchgears influence the main physical properties which charac-
v, w0 , U, Um measured water velocity or superficial water terize the gas at the exhaust, namely the pressure, the temperature
velocity (m s−1 ) and the velocity.
we effective or intersticial velocity (m s−1 ), we = The aim of the current work is to assess the pressure drops
w0 /ε and/or the friction factors of such filters in order to depict the
influence of filters formed by different stacked metal screens and
Greek symbols
to provide numerical values for modeling tools. In Section 2 the
ɛ porosity (-)
features of spherical particles and of the woven metal screens are
θ angle of the fluid flow (Kolodziej and Lojew-
given, and the experimental set-up is described. Section 3 is ded-
ska, 2009b)
icated to a brief review about the formulations of the Ergun ap-
μ water dynamic viscosity (Pa s)
proximation for spherical particles in order to validate the whole
ρ water density (kg m−3 )
of the experimental set-up by comparison with other published
τ tortuosity (-)
data. In Section 4 we report the pressure drop measurements ob-
ψ drag coefficient in the flow around approach
tained for woven metal screens. Experimental results are analyzed
(Innocentini et al., 20 0 0)
by means of modified Ergun law in order to assess specific param-
Subscripts eters such as the tortuosity, the angle between the flow inside the
D Darcy porous material and the flow direction induced by the porous ma-
F Fanning terial thickness. We give the pressure drop values and/or friction
k kinetic factors for the stacks of woven metal screens, these latters being
Wu for the three coefficients α , β and γ defined in the analyzed by means of different equations, one of which is specifi-
approximation of Wu et al. (2005) cally established for the studied screens.

2. Experimental set-up and procedure

component in the whole electric circuit. Due to the proximity of
the arc chamber walls (typically in polymer materials) the char-
acteristics and the properties (composition, electrical and thermal
conductivities, radiation, viscosity) of the arc plasma continuously
evolve due to the polymer wall ablation. They also depend on the
very high gradients in temperature and pressure of the arc cham-
ber. During the circuit breaker working some liquid (fused metal
droplets) and gaseous (hot vapors resulting from ablated walls)
materials are produced. Thus filters have to be designed to avoid
or at least to significantly limit the flow of these fluids at the ex-
haust duct of the circuit breaker as it is illustrated in Rochette et al.
(2015). One may easily understand that an arc plasma is a transient
phenomenon whose properties variations are very important: the
arc plasma chamber can have temperatures higher than 10 kK after
few milliseconds, pressures up to some MPa (Rochette et al., 2015).
2.1. Sphere beds features
Two sets of silica spheres provided by Sartorius and quoted
B. Braun Glasperlen 0.25–0.30 mm and 0.25–0.50 mm – respectively
called dp = 275 μm and dp = 375 μm in the paper – are used to
check and validate the experimental set-up on the one side, and
to measure the pressure drop formulations dedicated to spheres
on the other side. The features of the spheres and the porosity of
the sphere beds (mean values and standard deviations) are given
in Table 1(a). More details about the measurement and the defini-
tion of the aspect ratio and compactness are given in Bussière et al.
(2006). The relative discrepancy between the porosities obtained
for the two beds is ≈ 3% and has no specific meaning in regards to
experimental uncertainties. Experimental porosity values are 0.416
62 W. Bussière et al. / International Journal of Heat and Fluid Flow 65 (2017) 60–72

Table 1
Parameters of (a) the silica spherical particles and of (b) the woven metal mesh screens († , from (Woven Wire Cloth 2004, http://www.haverboecker.com March
2016)).

Particle diameter, dp (μm) Granulometric range (μm) Porosity, ɛ Shape factor (Bussière et al., 2006)
Aspect ratio Compactness

275 [250..300] 0.416 1.12 ± 0.04 0.93 ± 0.02

375 [250..500] 0.403 1.06 ± 0.04 1.01 ± 0.01

N° Mesh† Warp wire Weft wire Single screen Porosity† , ɛ Specific surface Diameter of single Hydraulic diameter,
diameter† , d1 (mm) diameter† , d2 (mm) thickness† (mm) area† , a (m−1 ) screen, D (mm) Dh from Eq. (1) (mm)

1 3.5 × 20 1.6 1.25 3.15 0.68 1272 40 2.14

2 5 × 32 1.25 0.9 2.59 0.67 1900 40 1.41
3 8 × 48 0.9 0.56 1.54 0.66 3464 40 0.762
4 11 × 56 0.71 0.47 1.44 0.64 4106 40 0.623
5 12 × 64 0.63 0.42 1.28 0.65 4222 40 0.616
6 21 × 85 0.45 0.3 0.92 0.61 7464 40 0.327
7 24 × 110 0.355 0.25 0.76 0.63 7760 40 0.325
8 30 × 150 0.224 0.18 0.51 0.65 9167 40 0.284

and 0.403 respectively for 275 μm and 375 μm. These values are
close to 0.395 – relative deviations are respectively ∼ 5% and ∼
2% – which show that the arrangement obtained after packing
is rather close to an orthorhombic one. The test cell diameter to
particle diameter ratios D/dp are respectively ∼145 and ∼107. The
granulometric analysis given in Fig. 1(a) and performed with a spe-
cific method described for example in Renard et al. (2002) shows
a larger granulometric interval for each set than the one quoted by
the supplier.

2.2. Woven metal screens features

A typical sketch of the woven metal screens used in the study

is shown in Fig. 1(b). The main features of these screens are given
in Table 1(b) for the eight different screens tested. The weft wires
are slightly thinner than the warp wires, the wire diameters are
of the millimetric domain ranging from 0.18 mm to 1.6 mm. Wo-
ven thickness varies from 0.51 mm to 3.15 mm which is linked to
the wire diameters used. The porosity ɛ is rather similar for each
metal screen and varies in the interval 0.61 to 0.68. Due to the
small variation, the porosity may not be considered as a relevant
parameter to distinguish each weave. On the contrary the specific
surface area a is a more relevant discriminating characteristic. For
woven screens numbered from 1 to 8, the specific area increases
from 1272 m−1 to 9167 m−1 . As a result the hydraulic diameter Dh
decreases respectively from 2.14 mm down to 0.284 mm, the defi-
nition being (Kolodziej et al., 2009):
4ε
Dh = (1)
a
Other formulations can be used respectively in the case of
spherical model (subscript S) (Wu et al., 2005) and in the case
of infinite cylinder model (subscript C) (Kolodziej and Lojew-
ska, 2009b):
1−ε
DS = 6 (2)
a
and
1−ε
DC = 4 (3)
a

2.3. Experimental set-up

The experimental set-up is displayed in Fig. 2(a) with the main
components. The water flow is created by means of a hydraulic
pump from DAB (model KV40) with an operating temperature
range up to 110 °C, a maximum working pressure equal to 25 bars.
The water flow is measured with a 8700 magnetic flowmeter from
Fig. 1. Porous media studied. (a) Granulometric distributions of the quoted 0.25–
0.30 mm (dp = 275 μm) and 0.25–0.50 mm (dp = 375 μm) silica spheres measured re-
spectively with 265 and 384 particles. (b) Schematic diagram of the woven metal
mesh screens (known as single plain dutch weave from Woven Wire Cloth (2004),
also called plain dutch as quoted in Armour and Cannon (1968)).
 W. Bussière et al. / International Journal of Heat and Fluid Flow 65 (2017) 60–72
Fig. 2. Schematic diagrams of (a) the experimental set-up and (b) the test cell with
pressure transducers (Pi ).
Rosemount (8705 flowtube equipped with a 8732 transmitter) al-
lowing flow measurements from 0.01 m s−1 to 10 m s−1 . The incer-
tainty is quoted at ± 0.5% for [0.3–10] m s−1 and ± 0.0015 m s−1 for
flow speed inferior to 0.3 m s−1 . The resolution is 0.1 L min−1 . The
volume of the water tank is 500 L and enables to work in closed
flow system. The temperature of the water flow is continuously
measured by means of a K-type thermocouple. From this measure-
ment the water density and the dynamic viscosity are deduced
from data quoted in Handbook of Chemistry and Physics (2012),
Huber et al. (2009), Kadoya et al. (1963) and Kaviany (1999). The
measurement is necessary to assess true values of viscosity and
density on one hand, and to take into account the influence of the
water heating due to the pump working on the other hand.
The test cell is presented in Fig. 2(b) with the major features.
The inner diameter is 40 mm. The test cell is equipped with four
static pressure transducers Kistler K-line calibrated up to 2 MPa
with output voltage set to 10 V ± 0.5%. The output voltage is mea-
sured with a 3014B Tektronix oscilloscope. Porous media or filters
(woven metal mesh screens) can be placed at the entrance and at
the exit of the test cell. A dedicated void of length L is reserved
for dispersed porous media such as silica sphere beds. Two metal-
lic screens of known mesh are used to hold particles when a dis-
persed porous medium is studied. In the case of dispersed porous
medium, the experimental porosity is assessed from the measure-
ment of the mass difference of the test cell before and after filling
and compacting.
In an initial stage, the permeability of dispersed porous media
is studied using the two quoted silica sphere beds. The objective
is first to validate the whole set-up using well-defined beds. Sec-
ondly we check the permeability formulations for spherical parti-
63
cles already published in the literature. The second step consists in
assessing the permeability of various stacks of woven metal mesh
screens that are currently used for electrical safety purposes.
3. Experimental results for sphere beds
3.1. Flow regimes
Prior to define the permeability formulations one has to won-
der about the type of the flow regime in regard to the Reynolds
number. The experimental Reynolds number is given in Fig. 3(a) in
function of the flow channel diameter D, the particle diameter dp
and is defined by:
ρvD
ReD = (4)
μ
and
ρvd p
Red p = (5)
μ
For the two sphere beds the ReD values are very similar (vary-
ing from ∼103 to 104 ), the discrepancies are insignificant and
linked to experimental uncertainties (density and dynamic viscos-
ity dependence with temperature). On the contrary, the Red p val-
ues are slightly higher for 375 μm (from ∼101 to 102 ) compared to
275 μm (from ∼101 to 5 × 101 ). From Kaviany (1999), the flow can
be considered as inertial-flow regime: this means that one has a
steady nonlinear laminar flow characterized by 1–10 < Red p < 150
where the inertial force tends to dominate over the viscous forces.
3.2. Porous medium permeability formulation
The pressure drop measurements with uncertainties are given
in Fig. 3(a) for the two sphere beds. For a given experimental ve-
locity value, higher pressure drops are observed for 275 μm which
is coherent with the lower corresponding hydraulic diameter.
The numerous studies dedicated to permeability assessments
published in the literature can be roughly summarized from the
type of porous medium point of view: sphere beds (Bussière et al.,
2006; Macdonald et al., 1979; Eisfeld and Schnitzlein, 2001;
Pope et al., 2011; Choi et al., 2008), foams (metallic (Mahjoob and
Vafai, 2008) or mineral (Moreira and Coury, 2004) or polymeric
(Schmid et al., 1999)), woven metal mesh screens (Wu et al., 2005;
Belforte et al., 2009; Kolodziej et al., 2009; Kolodziej and Lojew-
ska, 20 09b; Nika, 20 08; Kolodziej and Lojewska, 2009a) are the
main porous media quoted.
Whatever the porous medium, the starting point is the defini-
tion of the permeability given by the Darcy equation for an incom-
pressible fluid (Rohsenow et al., 1998) (the pressure drop is lin-
ear with the flow velocity, viscous term) improved by the Forch-
heimer equation (addition of the quadratic contribution, inertial
term) (Multiphase Flow Handbook 2006):
P 1 1
= μv + ρv2 = αμv + βρv2 (6)
L k1 k2
where α is the Darcy coefficient (m−2 ) with α = 1/k1 and β is the
Forchheimer coefficient (m−1 ) with β = 1/k2 . Eq. (6) describes the
pressure drop per unit length of fluid during the flow through a
porous medium. It includes the viscous contribution – character-
ized by the parameter k1 , also commonly known as the Darcyan
parameter, to take into account the linear dependency with the
flow rate. It also depends on the inertial contribution, described
by the parameter k2 , also commonly known as the non-Darcyan
parameter, to take into account the quadratic dependence with the
flow rate, and assumes a constant value for the viscosity and the
density of the flowing fluid.
64 W. Bussière et al. / International Journal of Heat and Fluid Flow 65 (2017) 60–72

Fig. 3. For sphere beds. (a) Pressure drops P/L (Pa/m) versus experimental velocity v for silica sphere beds with 275 μm and 375 μm particle diameters with best fit using
Eq. (7) and correlation coefficient. (b) Fanning friction factor versus Reynolds number from Eq. (8.2) with best fit.

Table 2
Various A and B values of Eq. (7) obtained for spherical particles. Data refering to Macdonald et al. (1979) correspond to the mean value and the standard deviation
(s.d.) when expressed. For quartz particles from Sartorius, dp = 275 μm corresponds to 0.25–0.30 mm and dp = 375 μm corresponds to 0.25–0.50 mm.

Ref. A B ɛ dp (μm) D/dp Red p Material Remarks about beads or

particles

Macdonald et al. (1979)a 132.7 1.291 0.366 to Quoted dp Red p /(1 − ε ) Glass Spherical and smooth glass
(±14% s.d.) (±16.5% 0.640 values report from ∼10−2 to particles, beads of narrow
s.d.) from ∼ 7.9 μm 104 size distribution
to 3.7 mm
Macdonald et al. (1979)b 321c 1.298 0.369 to Marble Spherical and smooth
(±11.5% 0.415 marble mixtures, uniform
s.d.) size or mixed in known
proportions
Bussière et al. (2006)d 273.5 10.54 0.38 275 ∼145 ∼2 to ∼50 Quartz Spherical and smooth
quartz particles
Bussière et al. (2006) d
241.3 2.077 0.41 375 ∼107 ∼8 to ∼65 Quartz Spherical and smooth
quartz particles
Bussière et al. (2006)d 409.6 0.454 0.37 250..315 ∼142 ∼8 to ∼110 Silica sand Silica sand particles with
aspect ratio ∼ 1.36
Bussière et al. (2006)d 317.6 0.464 0.36 355..400 ∼106 ∼10 to ∼142 Silica sand Silica sand particles with
aspect ratio ∼ 1.33
This worke 311.2 3.267 0.416 275 ∼145 ∼8 to ∼50 Quartz Spherical and smooth
quartz particles
This work e
279.9 2.417 0.403 375 ∼107 ∼10 to ∼95 Quartz Spherical and smooth
quartz particles
a
Reference therein Macdonald et al. (1979): Gupte AR, Dissertation, Karisruhe, 1970 and Rumpf H, Gupte AR, Chem. Ing. Tech. 43 (1971) 367
b
Reference therein Macdonald et al. (1979): Dudgeon CR, Houille Blanche, 21 (1966) 785
c
s.d. is not given because the uncertainty is too high
d
Working fluid is air at room temperature
e
Working fluid is water whose temperature is measured at each experiment

More information about the determination of k1 and k2 coeffi-
cients in the case of air flow can be found in our previous work
(Bussière et al., 2006). We give in Table 2 these values expressed
under the form of the coefficients A and B (see Section 3.3 for def-
inition) and the values obtained in the case of the current study
with the same method but with water flow.
3.3. Expressions and results for sphere beds
3.3.1. Pressure drops
For the pressure drop we refer to the Ergun equation
(Ergun, 1952) and to the modified Ergun equation as quoted by
Macdonald et al. (1979) where the original numerical values are
 W. Bussière et al. / International Journal of Heat and Fluid Flow 65 (2017) 60–72 65

Table 3
Comparison of k1 and k2 coefficients obtained for air flow (Bussière et al., 2006) and water flow with results from
Ergun (1952).

dp (μm) k1 (m2 ) k2 (m)

Air flow Water flow From Ergun (1952) Air flow Water flow From Ergun (1952)
−11 −11 −11 −6 −5
275 1.6 × 10 5.13 × 10 7.8 × 10 0.33 × 10 1.04 × 10 5.2 × 10−6
375 4.6 × 10−11 9.26 × 10−11 19 × 10−11 3.1 × 10−5 1.71 × 10−5 6.3 × 10−4
50 0 0 3.0 × 10−8 1.30 × 10−8 2.5 × 10−8 4.8 × 10−4 4.31 × 10−4 8.1 × 10−3

substituted by the coefficients A and B:
P (1 − ε ) μ
2
1−ε ρ 2
=A v+B 3 v (7)
L ε 3 D2p ε Dp
with A = 150 and B = 1.75 the original values of the two coeffi-
cients given by Ergun (1952). Dp is the equivalent or effective di-
ameter of particles defined by using a sphere model, i.e. D p = DS =
6(1 − ε )/a from Eq. (2). In the case of a bed formed by identical
spherical particles then Dp is exactly the diameter of the spherical
particle, thus D p = d p in Eq. (7) (Kaviany, 1999).
Experimental and calculated pressure drops are given in
Fig. 3(a) for 275 and 375 μm particles. For both particle diame-
ters, approximations are obtained with satisfactory correlation co-
efficients and we obtain respectively (A; B) = (311.2;3.267) with
R2 = 0.9994 for 275 μm and (A; B) = (279.9;2.417) with R2 = 0.9999
for 375 μm. Comparisons with published values obtained with
air flow and with formulations resulting from Ergun (1952) and
Rohsenow et al. (1998) are given in Tables 2 and 3.
3.3.2. Friction factors
Kolodziej et al. (2012) highlighted that there are two main ap-
proaches to describe the pressure drop of a fluid flowing through
stacked wire gauzes. In the approach flow through, the pressure
drop for the total bed length is given under the form of the Fan-
ning friction factor and Reynolds number is defined from the hy-
draulic diameter. The wire gauze is considered as a set or a bun-
dle of given model tubes. In the second approach flow around, the
pressure drop is given under the form of the drag coefficient ψ
calculated from the pressure drop for one gauze and from the frac-
tional opening of the gauze (different from porosity), the Reynolds
number being defined with the wire diameter. This latter is disre-
garded as the current work is concerned by stacks of woven metal
screens.
Eq. (7) can be expressed under the form of the Blake-type fric-
tion factor following Ergun (1952) by:
P D p ε 3 1−ε
fk = = 150 + 1.75
L ρUm2 1 − ε NReD p
with the Reynolds number defined by: NReD = D p ρUm /μ, where
p the coefficient A, we notice that our results are consistent with
Um is the superficial fluid velocity measured at average pressure.
The friction factor fk defined by (8.1) refers to the kinetic energy
losses, that is to say, the ratio of the total energy losses to the
quadratic term representing the kinetic losses. Another formula-
tion was established referring to the ratio of the pressure loss to
the viscous energy losses ( fv = 150 + 1.75 × NReD /(1 − ε )) but it is
p
not used in the current work. Following the modified Ergun equa-
tion of Macdonald et al. (1979), an equivalent formulation is given
by:
ε3 1−ε
Fk  =A +B
1−ε NRe 
ρV D
where Fk  1ε−ε = LP , NRe  = 0μ eq with Deq ≡ Dp and V0 ≡
3 Deq ε 3
ρV02 1−ε
w0 . The values for the two constants A and B are prescribed to be
valid whatever the type of flow. Some restrictions are given in Nika
(2008) for turbulent flows with A = 300 and B = 3.5 to take into ac-
count higher energy losses. Macdonald et al. (1979) have studied a
wide range of experimental data – various materials and shapes
such as glass spherical particles, cylindrical fibers, spherical mar-
bles either of uniform size or under the form of known mixtures
– in order to assess the two coefficients A and B of Eq. (8.2). Mean
values and standard deviations shown in Table 2 are calculated by
considering A and B as constant – even if a small dependence with
ɛ is observed – for a given experimental configuration. Finally Mac-
Donald et al. recommend two sets of values: (A, B) = (180,1.8) for
smooth particles and (A, B) = (180,4.0) for roughest particles as a
good approximation for engineering purposes.
Results are plotted in Fig. 3(b) with fitted approximations:
(A; B) = (306.0;3.379) with R2 = 0.9987 for 275 μm and (A;
B) = (280.2;2.413) with R2 = 0.9985 for 375 μm. Of course these val-
ues are identical to those obtained in Section 3.3.1.
Values obtained with air as working fluid have to be considered
apart from other data corresponding to water as working fluid.
All the results are given in Table 3 in terms of k1 and k2 coeffi-
cients. The starting point of Eqs. (6)–(8.2) is the equation of Darcy
hence incremented by Forchheimer contribution with the quadratic
fluid velocity term depending on density. Eqs. (6)–(8.2) or other
formulations (such as the semiheuristic momentum equation sug-
gested by Brinkman) have been established for incompressible
fluid (Rohsenow et al., 1998) which is not a valid assumption for
air or other gases. For example Innocentini et al. (20 0 0) clearly
show the discrepancies between the results obtained by means of
the Forchheimer equation for incompressible fluids or the Forch-
heimer equation for compressible fluids.
3.4. Conclusion for sphere beds
In Fig. 4(a) and 4(b), A and B values obtained with water
flow are plotted versus the tube-to-particle diameter ratio (D/dp )
and compared with the values quoted by Eisfeld and Schnitzlein
(2001) for which the D/dp ratio ranges from around 6 to 60. In
(8.1)
Eisfeld and Schnitzlein (2001) additional results are given for D/dp
< 10 for which A values are significantly higher than 500. For
higher quoted values (the highest value around 750 being dis-
regarded) with D/dp ∼ 145 and 107 respectively for 275 μm and
375 μm. For the coefficient B, the value obtained for dp = 375 μm is
similar to published data except for dp = 275 μm where the value
is significantly higher. Eisfeld and Schnitzlein (2001) has already
shown that an opposite trend is observed for B in comparison to A,
namely a slight increase of B when increasing the D/dp ratio. Such
a conclusion is reinforced by our measurements carried out with
D/dp ∼ 107 and 145.
To further the analysis, the influence of the wall proximity
(8.2) is studied by means of the formulation given by Choi et al.
(2008) where the pressure drop is given by the relation:
P (1 − ε )2 μM2 1 − ε ρ MCW 2
=A v+B 3 v (9)
L ε3 D2p ε Dp
66 W. Bussière et al. / International Journal of Heat and Fluid Flow 65 (2017) 60–72

1000 4
From Eisfeld and Schnitzlein (2001)
Andersson (1963) Redp = 0.07 .. 1315
Bernard and Wilhelm (1950) Redp = 2 .. 2324
Ergun and Orning (1949) Redp = 0.4 .. 30
Jeschar (1964) Redp = 1 .. 16884
750 This work
3
dp = 275 µm water
dp = 375 µm water

500 2

B
A

250 1

0 0
0 25 50 75 100 125 150 0 25 50 75 100 125 150
D/dp D/dp
(a) (b)
6
6x10 1.00
dp = 275 µm
Choi et al. (2008)
with 311.2 and 3.267
6
5x10 Cw
0.99

6
4x10
0.98
ΔP/L (Pa/m)

6
3x10
Cw

0.97
6
2x10

6 0.96
1x10

0 0.95
0.0 0.1 0.2
-1
v (m.s )
(c)
Fig. 4. Comparison of the (a) A Andersson (1963), Ergun and Orning (1949), Jeschar (1964), Bernard and Wilhelm (1950) and (b) B coefficients obtained in the current work
for water with data published in Eisfeld and Schnitzlein (2001) (and references therein) for spherical particles. (c) Influence of the wall effect depicted by the formulation
given by Choi et al. (2008) Eq. (9) (CW is the wall correction factor).

where M and CW are two factors defined for taking into account
the effect of the proximity of the wall (see Choi et al., 2008)
for detailed explanations of the model and original reference). For
dp = 275 μm, calculations provide M ≈ 1.008 while CW is plotted in
Fig. 4(c) from which we deduce that CW = 0.9750 +/- 0.0032 (s.d.)
for the velocity range we have considered, i.e. CW is very close to
unity for D/dp ∼ 145. Thus, our measurements can be considered
as weakly dependent on the wall influence.
Experimental results and tests of various approximations dedi-
cated to spherical particle beds have shown that the whole of the
experimental set-up is designed to perform realistic measurements
of pressure drops. Section 4 is dedicated to the study of pressure
drops and formulations in the case of woven metal screens used
for electrical safety purposes.
4. Experimental results for woven metal screens
Pressure drops due to woven metal screens are studied from
the friction factor point of view. One of the major difficulties for
 W. Bussière et al. / International Journal of Heat and Fluid Flow 65 (2017) 60–72 67

Table 4
Coefficients A1 and A2 deduced from Eqs. (10.1) and (10.2)
for the woven metal screens (30 measurement points for
each).

N° A1 A2 R2

1 – – –
2 – – –
3 583 0.613 0.93
4 637 0.153 0.96
5 396 0.572 0.97
6 422 0.976 0.98
7 653 0.268 0.98
8 – – –

combination of these two effects could explain the incoherents

trends obtained for the three stacks. Considering the screens 3 to
7, the best approximation is given by:

Fig. 5. Evolution of the Fanning friction factor f for stacks of woven metal screens
using the hydraulic diameter Dh . The curve corresponds to the best fit following
Eq. (10.2) (screens 1, 2 and 8 being disregarded) with A1 = 534 and A2 = 0.505.
this kind of porous media lies in the definition of the characteris-
tic length depicting the fluid flow through the void parts of the
screens. In the following subsections, we test different formula-
tions or geometrical models, that is to say the hydraulic diameter,
the spherical diameter and the cylindrical diameter. As the litera-
ture dealing with friction factor formulations is abundant, we aim
to select and test those dedicated to screens or stacks of screens
corresponding to nearly millimetric order wires screens. In the cur-
rent work, the woven metal screens are built using wire diame-
ters (weft and warp wire diameters being different) ranging from
0.18 mm to 1.6 mm and measurements of pressure drop are carried
out for stacks of these screens (typically five layers).
4.1. Fanning friction factor defined with the hydraulic diameter
As a starting point, the measurements are given in terms of the
Fanning friction factor following the Darcy–Weisbach equation as
recalled by Kolodziej and Lojewska, 2009a,b:
P 2ρw2e
= f (10.1)
L Dh
where the hydraulic diameter is defined by Dh = 4ε /a and the pro-
posed equation is:
A1
f = + A2 (10.2)
Re
with Re = ReDh . From Eq. (10.1) f is plotted in Fig. 5 as a func-
tion of the Reynolds number ReDh = ρ we Dh /μ and compared with
Eq. (10.2) to obtain the coefficients A1 and A2 given in Table 4
for the studied metal mesh screens. Results are not displayed for
the screens 1, 2 and 8 because the trends observed for these
screens deviate significantly from Eq. (10.2) and incoherent val-
ues are obtained for the two coefficients. We recall that in this
study measurements are carried out for stacks of identical screens
and not for only one screen with given and known characteristics.
The intermediate gap between two consecutive screens will intro-
duce more or less deviations from these given characteristics and
the value of the hydraulic diameter can become doubtful due to
variation of ɛ. Also it has been shown for example by Wu et al.
(2005) that metal screens of different woven types (plain square,
fill twill, fourdriner, plain dutch, twilled dutch) have different fric-
tion coefficients depending on the arrangement of the wires. The
534
f = + 0.505 (10.3)
ReDh
with a correlation coefficient R2 = 0.9248. Thus, we can conclude
that the Fanning friction factor, which is by definition the ratio of
the wall shear stress (τ W ) to the volumic density of kinetic energy
(1/2ρ v2 ) (Rohsenow et al., 1998), can not be approximated by a
relation like Eq. (10.2) with the Reynolds number defined from the
hydraulic diameter. Although Eq. (10.1) includes both the specific
surface area and the porosity in the definition by means especially
of the hydraulic diameter, results show that the approximation is
not valid for the studied woven metal screens.
4.2. Fanning friction factor defined with the spherical diameter
The second formulation is established considering a spherical
model to depict the fluid flow in the void space. Namely the spher-
ical diameter is used to write the friction factor and the Reynolds
number. This formulation has been widely used up to now in vari-
ous types of studies. Measurements are given in terms of the Fan-
ning friction factor following the Ergun equation (Ergun, 1952):
P D p ε 3
fk = (11.1)
L ρU 2 1 − ε
which is plotted versus the modified Reynolds number defined
D p ρU
by: 1Re
−ε = μ with the spherical diameter Dp defined above in
−ε
Section 3.3.1. The equation for approximation is fk = A 1Re + B and
we recall:
1−ε
fk = 150 + 1.75 (11.2)
Re
for low Reynolds number (Ergun, 1952), while:
1−ε
fk = 300 + 3.5 (11.3)
Re
for high Reynolds number (Nika, 2008). Experimental results and
approximations are plotted in Fig. 6(a) and (b) for the set of woven
screens stacks. Approximation computed from Eq. (11.2) shows a
good agreement for Re/(1 − ε ) > 102 excepted for the screens 1, 2
and 8, but it is far from experiments for Re/(1 − ε ) < 102 . Approx-
imation from Eq. (11.3) has to be disregarded both for the screens
and the Reynolds number range studied. As a result we conclude
that none of these approximations can be used with our screens
stacks.
In order to depict a wider range of Reynolds number on one
side, and to formulate a representative approximation for each
68 W. Bussière et al. / International Journal of Heat and Fluid Flow 65 (2017) 60–72
Fig. 6. (a) Evolution of the Fanning friction factor fk for stacks of woven metal
screens using the spherical diameter Dp Eqs. (11.1)–(11.5). Measurements are com-
pared with the following approximations: the modified Fanning friction factors from
Ergun (1952) for low (curve 1) and high (curve 2) Reynolds numbers; the empirical
equation from Jones et al. (herein (Wu et al., 2005)) (curve 3) and the empirical
equation from Wu (Wu et al., 2005) in the case of plain square type (curve 4). (b)
Empirical equation deduced for our measurements from Eq. (11.4) and spherical di-
ameter Dp following Wu et al. (2005) with prediction bands (−30% and + 30%).
type of metal screens on the other side, Wu et al. (2005) defined
an empirical equation given by:
1−ε
 1 − ε γW u
fk = αW u + βW u (11.4)
Re Re
where the coefficients α Wu , β Wu and γ Wu have to be determined
from experimental pressure drops for each type of metal screens
(plain square, plain dutch, full twill, twilled dutch types). The coeffi-
cient values obtained by Wu for plain square type are α Wu = 250,
β Wu = 1.69 and γ Wu = 0.071, and the curve is shown in Fig. 6(a).
The agreement with our experimental data is better but still not
satisfactory especially because of the trends for screens 1, 2 and 8.
The fitting coefficients α Wu , β Wu and γ Wu have been obtained for
all the experimental results disregarding the screens 1, 2 and 8, the
Fig. 7. Plot of Ergun-Bird equation following Eq. (12) with various values for C1 and
C2 coefficients. Are also given two approximations obtained from the measurements
only in the high Reynolds number domain (Re > 150): with coefficient C2 fixed at
3.5 and without any constraints.
fitted approximation being:
1−ε
 1 − ε 0.071
fk = 561 + 0.883 × (11.5)
Re Re
with the correlation coefficient R2 = 0.9225. The prediction bands
are plotted for ±30%. As a result we conclude that an approxima-
tion such as the one defined by Eq. (11.4) including the spheri-
cal diameter is not able to depict the pressure drop of the screens
studied in this work even if results can be considered as satisfac-
tory for screens numbered 3 to 7. From these two last sections we
also conclude that results obtained using Dh and D p = DS lead to
similar unsatisfactory trends for the studied screens.
4.3. Fanning friction factor defined with the cylindrical diameter
In the two last previous sections, we have shown that the ap-
proximations defined by means of the hydraulic and the spherical
diameters fail in the attempt to depict the pressure drops due to
the stacks of woven metal screens. In this section the following
points will be studied: the hypothesis of cylindrical diameters, the
definition of tortuosity, the effective velocity taking into account
the influence of the tortuosity.
As a first step before taking into account these latter points, we
have to identify the weaknesses of the formulations. We start from
the classical Ergun equations as cited by Bird (hence written EB),
Stewart, Lightfoot and given in Kolodziej and Lojewska (2009b):
P μw0 (1 − ε )2 ρ w20 1 − ε
= C1 + C (12)
L D2p ε3 2
D p ε3
where Dp is the equivalent spherical diameter, with original values
C1 = 72 and C2 = 1.75 hence amended by C1 = 150 and C2 = 1.75,
and by C1 = 300 and C2 = 3.5 for turbulent flows. These approx-
imations are plotted in Fig. 7 in the case of the stack with the
screen numbered 7. We see that none of the EB equations with
typical values for the coefficients C1 and C2 fit with the measure-
ments. The two following approximations are performed only for
the points for which the modified Reynolds number is higher than
150: if C2 is set to 3.5 the resulting approximation shows a very
poor correlation coefficient (R2 = 0.6597); if C1 and C2 are free of
any constraints, then a rather acceptable correlation coefficient is
obtained but the values of the two coefficients are far from classi-
cal values and they have no meaning (respectively 1092 and 4.8).
 W. Bussière et al. / International Journal of Heat and Fluid Flow 65 (2017) 60–72 69

Table 5 Table 6
Coefficients C1 and C2 obtained for the stacks by means of Experimental values of the angle θ deduced from the application of
Eq. (12) but with the cylindrical diameter instead of the Eq. (13.1) following Kolodziej and Lojewska (2009b) and Eq. (14) from
spherical diameter. Kolodziej et al. (2012).

N° C1 C2 R2 N° τ θ from Eq. (13.1) θ from Eq. (14)

1 2474 1.42 0.58 1 1.160 84.0 74.7
2 1603 3.09 0.54 2 1.165 85.5 78.3
3 893 1.90 0.92 3 1.170 82.0 67.9
4 661 1.42 0.96 4 1.180 79.4 57.9
5 416 2.02 0.87 5 1.175 79.7 60.2
6 412 3.42 0.88 6 1.195 82.7 66.0
7 485 3.20 0.93 7 1.185 82.4 67.1
8 381 3.98 0.98 8 1.175 81.9 66.9

In a second stage, the spherical diameter is replaced in

Eq. (12) by the cylindrical diameter defined by DC = 4 1−a ε result-
ing in the same equation with Dc equal to dw in Kolodziej et al.
(2012). The results for the constants C1 and C2 are given in Table 5.
These results are obtained disregarding the experimental points for
low Reynolds number (<150) as previously done. Since the spheri-
cal case, results do not show reliable values for the two constants.
Plots are not shown but the fitted curves are very similar to those
obtained for example in Fig. 7 for the woven metal screen 7.
Whatever the formulation tested up to now, we observe that
there are two main trends depending on the range of the Reynolds
number, namely low or high Reynolds number with a transition
value around 150. These two trends have been already observed
for example in Kolodziej and Lojewska (2009b) by Kolodziej et al.
In the current experiments all the Reynolds numbers ReDh , ReD p
and ReDc are lower than ∼ 4 × 103 and for most of the studied
woven screens they are smaller to 103 for the velocity investigated
range. So firstly, one has to pay great attention to the transient
domain between low and high Reynolds numbers which should be
described more accurately, and secondly, the spherical diameter Dp
has to be substituted with the cylindrical diameter DC .
We now discuss these two key points. For the depiction of the
transient domain, the improvement suggested by Kolodziej et al.
refers to previous studies (Fowler and Hertel, 1940; Harris et al.,
2001), concerned by porous media consisting of fibers from var-
ious materials. For such porous media, the path of the flow can
not be approximated by the classical effective velocity given by
we = w0 /ε but one has to take into account θ – the angle between
the direction of the flow inside the porous medium and the di-
rection given by the thickness of the crossed porous medium –
and the tortuosity. Following Kolodziej and Lojewska (2009b) and
Harris et al. (2001) the effective flow length and the porous thick-
ness are linked by: Le = Lτ /cos(θ ), and the effective flow velocity
is defined by: we = (w0 /ε ) · (τ /cos(θ ) ). These improvements are
of interest for the current study. Different stacks of woven screens
were tested, where woven screens have different porosities and
different diameters for the weaved wires.
At the end, the resulting pressure drop is the sum of the lami-
nar and turbulent contributions and the pressure drop is given by
(see Kolodziej and Lojewska (2009) for the full meaning and for-
mulation of each term) as:
P ρ w20 1 − ε τ 3
= 4( fapp + ft )
L 2DC ε 3 cos3 (θ )
with:
dw ρ w e Dh
χ+ = , Ree =
Dh Ree μ factor is inferior to 10%.
1.25 3.44
+ 16 − √
3.44 4χ + χ +
fapp Ree =  +
χ + 1 + 0.0 0 021/χ +2
0.079
ft = (13.4)
Re0e .25
where fapp and ft are respectively the friction factors for the lam-
inar and turbulent domains, DC is used in place of dw the wire
diameter, as proposed in Kolodziej and Lojewska (2009b). The tor-
tuosity is calculated from τ = 1 + (1 − ε )/2 which corresponds to
a geometric model of a cubic cell with one sphere inside as re-
ferred in Harris et al. (2001). The parameter χ + is a dimension-
less quantity characterizing the length of the elementary chan-
nel for the flow within a mesh of the screen; Ree is the effective
Reynolds number defined from the hydraulic diameter; fapp is de-
duced from Shah and London (1978) and ft is the Blasius formula
for smooth pipes in the turbulent domain with coefficients given
in Kolodziej and Lojewska (2009b), Kolodziej et al. (2012), Yan and
Lin (1999) and Ouyang and Aziz (1996). The experimental pressure
drops are fitted by means of Eq. (13.1), the fitting parameter be-
ing θ whose values are given in Table 6 together with the tortu-
osity deduced from the porosity of woven metal screens. The ex-
perimental pressure drop is compared with the approximation in
Fig. 8(a) for θ = 82.4° obtained for the woven screen 7, and for <
θ > = 82.4° the mean angle value obtained for the whole of the
studied woven screens. Satisfactory agreement is reported between
experiment and approximation for which discrepancies are inferior
to 10% at the most, except for both smaller and higher values of
measured velocities.
Other formulations have been tested for the turbulent friction
factor. In Ouyang and Aziz (1996) one finds an exhaustive sec-
tion dealing with the coefficients of the formulation first pub-
lished by Blasius. The relation is given under the form f = r Re−t
with various values for the coefficients r and t deduced from
the bibliography. In particular, the power law has been modi-
fied with t ranging from 0.079 down to 0.003678, the explana-
tion from Ouyang and Aziz (1996) arguing the various Reynolds
number domains and the very simple form of the approxima-
tion which may fail in representing real situations. Among the
power laws and considering Reynolds number range, we have
tested those with r = 0.316 and t = 0.25 from Fang et al. (2011) (for
smooth pipes with Re < 2 × 104 ), and r = 0.02118 and t = 0.1461
from Ouyang and Aziz (1996) (for smooth pipes up to Re ≤
2 × 106 ). The values obtained for θ show small differences ∼ 5% at
the most. This result is consistent with two considerations. Firstly
(13.1)
the Reynolds number range studied in the current work since one
has ReDh < 4 × 103 << 2 × 104 which appears as a high limit for the
use of t = 0.25 (Fang et al., 2011). Secondly, for all the studied wo-
ven metal screens, the turbulent contribution in the total friction
(13.2)
The discussion about the influence of tortuosity is improved by
Kolodziej et al. (2012), concerning the effective flow length and
(13.3) the effective flow velocity, where the formulations are respectively
amended by: Le = L/cos(θ ) and we = w0 τ /ε . The resulting pressure
70 W. Bussière et al. / International Journal of Heat and Fluid Flow 65 (2017) 60–72
Fig. 8. (a) Experimental pressure drop for the woven metal screen 7 (●) with the
approximation (Kolodziej and Lojewska, 2009b) from Eq. (13.1) calculated for the
fitted angle θ = 82.4° (red line) and for θ = 82.2° (blue line) the mean value calcu-
lated for the whole set of screens. (b) The same as (a) but with the approximation
from Eq. (14) (Kolodziej et al., 2012). (c) The same as (b) but with the tortuosity
value τ ≡ Q = 1.28. (For interpretation of the references to colour in this figure leg-
end, the reader is referred to the web version of this article.)
drop is then:
P ρ w20 1 − ε τ 2
= 4 ( f l + ft ) (14)
L 2DC ε 3 cos (θ )
where fl ≡fapp , τ is unchanged and we still use DC instead of dw as
in the original equation in Kolodziej et al. (2012). The results for θ
are given in Table 6 and the approximation for the woven screen
7 is shown Fig. 8(b) for θ = 67.1° The approximation deduced from
Eq. (14) is closer to the experimental points than it is with the use
of Eq. (13.1) but the trend is very similar.
One of the key point is the definition of the tortuosity used
in Eqs. (13.1) and (14). Many expressions can be found in the lit-
erature dealing with experimental formulations of the tortuosity
obtained for various materials and especially for woven screens.
Among them we first refer to Chinda and Brault (2012) for packed
cylindrical particles porous medium. A rather good agreement is
found between the depicted tortuosity model and the formula-
tion from Mauret and Renaud (1997) given by τ = 1 + 1.55 ln(1/ε )
called capillary model, under the condition that the porosity is not
too low, roughly higher than 0.5, depending on the geometry of the
cylindrical particles from the examination of the results. We have
tested this tortuosity formulation in Eq. (14), the tortuosity in-
creasing from 1.37 to 2.52 for the screens numbered 1 to 8. Results
(not detailed in the paper) show that it is not possible to fit our ex-
perimental pressure drops with some screens, which is mainly due
to the high value of tortuosity. We also refer to Armour and Can-
non (1968) and Fischer and Gerstmann (2013) for woven screens.
In Fischer and Gerstmann (2013), for woven screens of the dutch
twilled type (DTW) with 40 μm and 70 μm wire diameters, a Q tor-
tuosity factor of 1.3 is prescribed by means of formulations given
in Armour and Cannon (1968). In Armour and Cannon (1968) the
value Q = 1 is set for plain weaves and Q = 1.28 for dutch weaves.
The estimation deriving from Eq. (14) is applied in Fig. 8(c) with
Q = 1.28. Once again, the approximated trend is very similar to pre-
vious results.
In Fig. 9(a) we draw comparison with the experimental pres-
sure drops for the whole of the tested woven screens with the
calculated pressure drops obtained by means of Eq. (14) and τ =
1 + (1 − ε )/2 (results obtained with τ = 1 + 1.55 ln(1/ε ) are not
shown since they are very similar). We observe that there is a good
agreement for all the screens except for the woven screen 8. The
Fanning friction factors f = fl + ft are plotted in Fig. 9(b) for all
the woven screens together with the fitting curve (except woven
screen n°8) defined by f = 7.357 × Re−0.596 + 0.04351. This formu-
lation is chosen because the turbulent contribution is significantly
higher than the laminar contribution in the resulting Fanning fric-
tion factor f.
5. Conclusion
To improve the efficiency of LV circuit breakers and to com-
ply with IEC guidelines, researchers and engineers have to improve
the modeling tools used to simulate physical parameters and their
evolutions during electrical safety apparatus operation. One of the
key point studied in this work is the drag coefficients of woven
metal screens when submitted to the flow of a fluid. In fact these
screens are used as filters in LV circuit breakers and they play var-
ious important roles when an electric fault occurs. Firstly from the
thermal point of view they significantly reduce the temperature,
ionization degree, pressure, velocity, energy of the exhaust gases
consecutive to the arc plasma quenching. Secondly from the me-
chanical point of view, they tend to avoid the ejection of molten
metal droplets resulting from the electrical contacts erosion and
they strongly damp the shock wave linked to the significant energy
release by the default current. To increase the efficiency of the LV
circuit breakers including such filters we have built an experimen-
 W. Bussière et al. / International Journal of Heat and Fluid Flow 65 (2017) 60–72
Fig. 9. (a) Calculated versus experimental pressure drops for the whole of the wo-
ven screens. Approximation is deduced from Eq. (14) with τ = 1 + (1 − ε )/2, and
the prediction bands are given for 90%. (b) Fanning friction factors f = f l + ft de-
duced from the same approximation for the whole of the tested woven screens.
tal set-up designed to assess the pressure drops and drag coeffi-
cients of woven metal screens or stacks of those. From the mea-
surements given in terms of friction factors we have studied vari-
ous models to define the correct equation by taking into account
cylindrical diameters, angle between the direction of the flow in-
side the porous medium and the direction given by the thickness
of the crossed porous medium, and specific contribution of lami-
nar and turbulent domains. This equation is essential in LV circuit
breakers modeling for new design developments.
Acknowledgment
Authors would like to thank Mr. Michael Rips from Haver &
Boecker for having provided us with the parameters of the screens,
and Mr. Jean-Paul Gonnet, Charles Besnard and Marc Rival from
Schneider Electric for financial support and helpful discussions.
References
Andersson, K.E.R., 1963. Pressure Drop in Packed Beds. Trans. of the Royal Inst. of
Technol., Stockholm Sweden, N°201, cited after [27].
Armour, J.C., Cannon, J.N., 1968. Fluid flow through woven screens. AIChE J. 14 (3),
415–420.
Belforte, G., Raparelli, T., Viktorov, V., Trivella, A., 2009. Metal woven wire cloth
feeding system for gas bearings. Tribol. Int. 42, 600–608.
71
Bernard, R.A., Wilhelm, R.H., 1950. Turbulent diffusion in fixed beds of packed
solids. Chem. Eng. Progress 46, 233–244 cited after [27].
Besnard, C., 2007. La simulation numérique pour la conception de cellules sécurisées
en cas d’arc interne. In: SEE (Ed.), Proceedings of the Third European Conference
on HV and MV Substation Equipment, S6-1. Lyon, p. 3.
Bussière, W., Rochette, D., Latchimy, T., Velleaud, G., André, P., 2006. Measurement
of Darcy and Forchheimer coefficients for silica sand beads. High Temp. Mater.
Processes 10, 55–78.
Chinda, P., Brault, P., 2012. Tortuosity model of a three-dimensional stacked cylin-
drical particles porous medium for using with the thin film solid oxide fuel cell
electrodes. TIJSAT 17 (1), 42–53.
Choi, Y.S., Kim, S.J., Kim, D., 2008. A semi-empirical correlation for pressure drop in
packed beds of spherical particles. Transp. Porous Med. 75, 133–149.
Eisfeld, B., Schnitzlein, K., 2001. The influence of confining walls on the pressure
drop in packed beds. Chem. Eng. Sci. 56, 4321–4329.
Ergun, S., 1952. Fluid flow through packed columns. Chem. Eng. Prog. 48 (2), 89–94.
Ergun, S., Orning, A.A., 1949. Fluid flow through randomly packed columns and flu-
idized beds. Ind. Eng. Chem. 41, 1179–1184 cited after [27].
T.R. Faber, A.A. Maulandi, M. Rival, D.R. Pearson, 2003 US Patent No. 0048586.
Fang, X., Xu, Y., Zhou, Z., 2011. New correlations of single-phase friction factor for
turbulent pipe flow and evaluation of existing single-phase friction factor cor-
relations. Nuclear Eng. Des. 241, 897–902.
Fischer, A., Gerstmann, J., 2013. Flow Resistance of Metallic Screens in liquid,
Gaseous and Cryogenic Flow. EUCASS, p. 12.
Fowler, J.L., Hertel, K.L., 1940. Flow of a gas through porous media. J. Appl. Phys. 11,
496–502.
Handbook of Chemistry and Physics (2012), 92nd Ed. CRC Press.
Harris, D.K., Cahela, D.R., Tatarchuk, B.J., 2001. Wet layup and sintering of metal-con-
taining microfibrous composites for chemical processing opportunities. Compos-
ites 32, 1117–1126.
http://www.haverboecker.com, 2016, March Documentation Haver & Boecker, In-
formation – Toiles métalliques (Terminologie, types de tissage, et formes de
mailles), P 17F 327 072002 1 Fe.
Huber, M.L., Perkins, R.A., Laesecke, A., Friend, D.G., Assael, M.J., Metaxa, I.N., Vo-
gel, E., Mares, R., Miyagawa, K., 2009. New international formulation for the
viscosity of H2 O. J. Phys. Chem. Ref. Data 38 (2), 101–125.
IEC 62271-200 High voltage switchgear and controlgear – AC metal-enclosed
switchgear and controlgear for rated voltages above 1 kV and up to and includ-
ing 52 kV, first ed. (2003)
Innocentini, M.D.M., Pardo, A.R.F., Pandolfelli, V.C., 20 0 0. Influence of air compress-
ibility on the permeability evaluation of refractory castables. J. Am. Ceram. Soc.
83 6, 1536–1538.
Jeschar, R., 1964. Druckverlust in Mehrkornschüttungen aus Kugeln. Archiv Eisen-
hüttenwesen 35, 91–108 cited after [27].
Kadoya, K., Matsunaga, N., Nagashima, A., 1963. Viscosity and thermal conductivity
of dry air in the gaseous phase. J. Phys. Chem. Ref. Data 14 4, 1985.
Kaviany, M., 1999. Principles of heat transfer in porous media. Mechanical Engineer-
ing Series, 2nd ed. Springer Verlag.
Kolodziej, A., Jaroszynski, M., Janus, B., Kleszcz, T., Lojewska, J., Lojewski, T., 2009.
An experimental study of the pressure drop in fluid flows through wire gauzes.
Chem. Eng. Comm. 196, 932–949.
Kolodziej, A., Lojewska, J., 2009a. Flow resistance of wire gauzes. AIChE J. 55 (1),
264–267.
Kolodziej, A., Lojewska, J., 2009b. Experimental and modelling study on flow resis-
tance of wire gauzes. Chem. Eng. Process. 48, 816–822.
Kolodziej, A., Lojewska, J., Jaroszynski, M., Gancarczyk, A., Jodlowski, P., 2012. Heat
transfer and flow resistance for stacked wire gauzes: experiments and mod-
elling. Int. J. Heat Fluid Flow 33, 101–108.
Macdonald, I.F., El-Sayed, M.S., Mow, K., Dullien, F.A.L., 1979. Flow through porous
media – The Ergun equation revisited. Ind. Eng. Chem. Fundam. 18 (3), 199–208.
Mahjoob, S., Vafai, K., 2008. A synthesis of fluid and thermal transport models for
metal foam heat exchangers. Int. J. Heat Mass Transfer 51, 3701–3711.
Mauret, E., Renaud, M., 1997. Transport Phenomena in Multi-Particle Systems II: Pro-
posed New Model Based on Flow around Submerged Objects for Sphere and
Fiber Beds Transition between the Capillary and Particulate Representations.
Chem. Eng. Sci. 52, 1819–1834.
Moreira, E.A., Coury, J.R., 2004. The influence of structural parameters on the per-
meability of ceramic foams. Braz. J. Chem. Eng. 21 (1), 23–33.
Morel, R., Rival, M., 1990 Eur. Patent No. 0437151B1.
Multiphase Flow Handbook, (2006), ed. by C.T. Crowe, CRC, Part 10-1.
Nika, P., 2008 Convertisseurs thermoacoustiques – Moteurs et générateurs (in
french) Techniques de l’ingénieur Doc. BE 8061 1–25.
Ouyang, L.B., Aziz, K., 1996. Steady-state gas flow in pipes. J. Pet. Sci. Eng. 14,
137–158.
Pope, K., Naterer, G.F., Wang, Z., 2011. Pressure drop of packed bed vertical flow for
multiphase hydrogen production. Int. J. Hydrogen Energy 36, 11338–11344.
Renard, J.B., Worms, J.C., Lemaire, T., Hadamcik, E., Huret, N., 2002. Light scattering
by dust particles in microgravity : polarization and brightness imaging with the
new version of the PROGRA2 instrument. Appl. Optics 41 (4), 609–618.
Rival, M., Kilindjian, C., Clery, Y., Bonente, S., 1999 US Patent No. 5889249.
Rochette, D., Clain, S., André, P., Bussière, W., Gentils, F., 2007. Two-dimensional
modelling of internal arc effects in an enclosed MV cell provided with a pro-
tection porous filter. J. Phys. D 40, 3137–3144.
Rochette, D., Clain, S., Bussière, W., André, P., Besnard, C., 2010. Porous filter op-
timization to improve the safety of the medium-voltage electrical installations
during an internal arc fault. IEEE Trans. Power Delivery 25 (4), 2464–2470.
72 W. Bussière et al. / International Journal of Heat and Fluid Flow 65 (2017) 60–72
Rochette, D., Clain, S., Gentils, F., 2008. Numerical investigations on the pressure
wave absorption and the gas cooling interacting in a porous filter, during an
internal arc faults in a medium-voltage cell. IEEE. Trans. Power Delivery 23 (1),
203–212.
Rochette, D., Clain, S., Gentils, F., Wild, J., Bussière, W., 2011. Numerical simulation
of the porous filter properties for the internal arc mollifying effects. Elec. Power
Syst. Res. 81, 66–73.
Rochette, D., Clain, S., Gonnet, J.P., Bussière, W., 2015. Numerical study of the impact
of filters located in the exhaust duct of a low-voltage circuit breaker. IEEE Trans.
Comp. Packag. Manuf. Technol. 5 (1), 49–56.
Rohsenow, W.M., Hartnett, J.P., Cho, Y.I., 1998. Handbook of Heat Transfer. Mc-
Graw-Hill, p. 9. Chapters 4.
Schmid, R., Stuff, R., Klein, U.K.A., Jamjoon, F.A., Al-Suwaiyan, A., 1999. Low
free-stream turbulence in test sections through packed beds and fibrous mats.
Exp. Fluids 26, 451–459.
Schneider Electric–Square D, Arc flash protection with Masterpact NW and NT cir-
cuit breakers, Data Bulletin 0613DB0202R603 (2003).
Shah, R.K., London, A.L, 1978. Laminar Flow Forced Convection in Ducts. Academic
Press, NY p. 97 and 98.
Woven Wire Cloth, 2004. Reference Book, 4th edition Haver & Bocker.
Wu, W.T., Liu, J.F., Li, W.J., Hsieh, W.H., 2005. Measurement and correlation of hy-
draulic resistance of flow through woven metal screens. Int. J. Heat Mass Trans-
fer 48, 3008–3017 Herein references: D.P. Jones, H. Krier, Gas Flow Resistance
Measurements Through Packed Beds at High Reynolds Numbers, J. Fluids Eng.
105 2 (1983) 168-172.
www.schneider-electric.com, (2007) Présentation de la norme CEI 62271-202 (in
French).
Yan, Y.Y., Lin, T.F, 1999. Condensation heat transfer and pressure drop of refrigerant
R-134a in a smal pipe. Int. J. Heat Mass Transfer 42, 697–708.