See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/235779014

Influence of cleaning parameters on pulse-jet filter bags performances

Article  in  Filtration · January 2004

CITATIONS READS
7 1,952

6 authors, including:

Xavier Simon Dominique Thomas

INRS University of Lorraine
59 PUBLICATIONS   423 CITATIONS    158 PUBLICATIONS   1,754 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Separation of metallic ultra-fine particles by granular bed View project

study of pollutant transfer through nuclear containment View project

All content following this page was uploaded by Xavier Simon on 16 August 2017.

The user has requested enhancement of the downloaded file.

 [To Index]

INFLUENCE OF CLEANING PARAMETERS

ON PULSE - JET FILTER BAGS PERFORMANCES

X. Simon * (1,2), D. Thomas (1), D. Bémer (2), S. Callé (1), R. Régnier (2), P. Contal (1)

Aerosol Filtration Laboratory

(1) Chemical Engineering Sciences Laboratory

1, rue Grandville , BP 451, F-54001 NANCY Cédex, France
* Xavier.Simon@ensic.inpl-nancy.fr
(2) Institut National de Recherche et de Sécurité
avenue de Bourgogne, BP 27, F-54501 VANDOEUVRE-lès-NANCY Cédex, France

ABSTRACT

The aim of this study is to characterize pressure drop and acceleration profiles along a filter
bag subjected to different conditions of pulse-jet cleaning. A flow of clean air is imposed on
the 24 bags of a pilot-scale dust collector. Pulse-jet cleanings can be achieved either on the
single instrumented filt er bag, or on the 24 bags simultaneously. Various parameters (initial
bag tension, filtration velocity, compressed air tank pressure, impulsive flow duration and
geometry of the nozzle or tube injector) were investigated and their influences on the cleaning
performances of a filter bag were appreciated and compared. Results permit to better
understand the effects of pulse-jet cleaning and to identify the areas where cleaning seems to
be most effective. Choice and combination of those parameters: compressed air pressure,
geometry of injector nozzle and number of bags cleaned simultaneously, seem to be crucial
stages for the optimization of bags cleaning in the pulse-jet filters.

INTRODUCTION AND PREVIOUS WORKS

Pulse-jet cleaning dust collectors remain one of the most employed dust removal techniques
in industry in order to clean dust-laden gas before its throwing out or its recycling into
workplaces. Filter media used in industrial air cleaning are subjected to clogging and cleaning
cycles. In operation, the filtration of solid aerosol causes deposits to build up on the filter
media, resulting in an increase in pressure drop. Clogging presents two modes : depth-
clogging (irreversible) characterized by particles collected within the filter media and
clogging of the filter surface (reversible) due to the appearance of a layer of particles called
dust cake. The objective of cleaning is to take off particles collected on the filter surface in
order to decrease the pressure drop. Thus, the filter being regenerated, a new cycle of
filtration can then start [1].

An instantaneous pressure increase during pulse-jet cleaning leads to an important and rapid
expansion of filter bag and a great amount of particles, collected in the dust cake, can be
removed from the filter media. Cleaning will be effective if the detachment force, created
during movement and deformation of the bag, is higher than the global adhesion force that
retains the dust cake to the filter media. Actually, the increase of pressure causes the filter
fabric to expand outwards. This motion does not remove the cake since the inertia force will
press the cake against the fabric. Eventually the fabric can no longer be expanded outwards
and hoop stresses develop which oppose the outward motion and the fabric rapidly
decelerates. If this deceleration is sufficiently strong, the inertia of the cake mass will cause it
to keep moving and to break free from the fabric surface. The deceleration of the filter media
constitutes a key parameter because it is this final motion that cleans the filter [2, 3].

Some authors established profiles of acceleration and/or pressure drop along a bag filter [4-
12]. The results show that the acceleration caused by the bag motion dominates the cleaning
force at the bag top, and the reverse airflow caused by the overpressure dominates the
cleaning force at the bag bottom. The top and the bottom of a pulse-jet cleaning filter bag
seem to be the two areas which are cleaned best, but the mechanisms involved to dislodge the
cake are neve rtheless different. Reverse airflow seems to play an important role in the
cleaning of filter bags, mainly in their bottom part where it is considered to be the only
cleaning mechanism.
In spite of some differences between the experimental tests (equipments, filter bag
characteristics, filtration and cleaning conditions), tendencies appear in overpressure and
acceleration values created by a pulse-jet cleaning [2–10, 13–16, 18, 19]. Generally, the
orders of magnitude encountered in filter bag cleaning studies are :
- 1000 Pa for the overpressure in the bag ;
- from 50 to 300 g for the filter media acceleration.

Cleaning efficiency depends on a large number of parameters. Among all these parameters,
the most important are those associated with pulse-jet and filtration factors. According to the
literature, the parameters which mainly influence cleaning intensity and, thus, cleaning
effectiveness, are : the size of the injector tube and its distance to the top of the bag, the size
and initial pressure of compressed air tank [6–10, 14, 15, 17, 19, 21], the pulse duration [8]
and the filtration velocity [4, 6, 15, 19, 20].

Many previous works have already been carried out to better understand the influences of
pulse-jet cleaning parameters of a filter bag. Lots of information can be found there but, two
major tendencies can however be provided. Firstly, the results almost exclusively come from
pilot-scale test devices that involve few bags (more often, only one) and consequently, the
surrounding of the studied bag is very different from the one of a real dust collector. So, those
results may sometimes show a lack of representativeness, in comparison to industrial plant
reality. Moreover, the influence of the number of bags cleaned together has never been
investigated. Secondly, the influences of all parameters have never been studied on only one
fitting and/or with similar operating conditions for filtration and cleaning factors. Thus,
confronted with all these different, sometimes contradictory, experimental data, the
comparison of results becomes delicate.

The aims of this study are to investigate the effects of pulse-jet cleaning and the influences of
some parameters on the cleaning performances of one filter bag or of a group of bags. With
this intention, pressure drop and acceleration of the filter media have been measured in four
points along a bag and several cleaning conditions have been experimented.
 EXPERIMENTAL SETUP AND PROCEDURE

Tests are carried out on a laboratory dust collector with industrial size. It is instrumented with
24 cylindrical filter bags, allocated among five lines. Each bag is 1.5 m long and has a
diameter of 0.13 m. Thus, the dust collector presents a total filtration surface of 14.7 m2 if the
24 bags are connected (between 100 and 300 m 3.h-1.m-2). The experimental setup is
schematized below (Figure 1) :
FRC
M 1 HEPA filter
2 Aerosol generator
3 Protection against fire
4 Upstream sampling line
(12) 5 Dust hopper
(10)
(11) 6 Compressed air tank
7 Compressed air
8 Protection against fire
9 Downstream sampling line
(8) PR
10 Ultrasonic flowmeter
MR TR
11 Bypass
(9) 12 Fan

∆ Pv
(4) ∆P

PR PCV

AS
(3)
(1)
(7)
(2) (6)

AS

(5)

Figure 1 : Experimental setup.

Each bag has its own injector nozzle. Cleaning sequences are programmed using an
automaton. Thus, it is possible to select and clean one or several isolated bag(s), any group of
bags (between 1 and 24 bags) or several groups of bags (between 1 and 24 groups),
successively or simultaneously. This cleaning flexibility is necessary in order to study the
influence of bags on their neighbors. In addition, the pulse duration, t d , initiated by opening
electrical valves is adjustable and can vary from 100 to 1000 ms. The compressed air tank
pressure is also adjustable (P r = 3, 5 or 7 bar).

Five different injector nozzles have been tested :

- Position 1 : injector with nozzle of diameter d, at 0.08 m above the top of the bag
(d = 5, 10 or 15 mm)
- Position 2 : injector emerging at the top of the bag (d = 10 mm)
- Position 3 : injector being introduced into the highest first quarter of the bag (d =10 mm).

Each bag is also equipped with a Venturi flowmeter. They are connected above the bags, in
the downstream clean air cap, and linked with differential pressure transducers (Keller PD-
41/8885.D-0,05). The Venturi measures the air flow, that seeps through the bag ; it is a part of
the total flow measured by the ultrasonic flowmeter (Panametrics XGM868). So, the filtration
velocity, Uf, can be precisely calculated for each bag. Three velocities have been studied : Uf
= 2, 5 or 8 cm.s-1. The aerosol generator (PALAS BEG 1000S) is disconnected and the results
are obtained with clean air.
The pressure drop, ∆Pi = Pe − Pint, i (P e : external pressure, Pint,i : pressure inside the bag at
point i), is measured at four heights along the bag (A) (Figure 2) with differentia l pressure
transducers (Keller PD-41/8885.D -0,05). The sampling frequency is 100 Hz. The numerical
values, that will be presented, are the average of three tests (satisfactory reproducibility).

The filter media acceleration, a, is measured at four points along the bag (B) (Figure 2) with
low mass accelerometer (Brüel & Kjaer 4393 V), first stuck on the media. Accelerometers are
supplied with charge amplifier (two Brüel & Kjaer 2635, two KISTLER 5015A), that also
allow some adjustments (filtering, integration). The recording of acceleration’s data and their
post-processing are made with a specific acquisition device (LMS Instruments). The sampling
frequency is 2048 Hz. Numerical values which will be presented are the average of 30
acceleration measurements, divided in three tests of 10 measurements each (As we can see,
the reproducibility is lower than for the pressure drop measurements).

Figure 2 : Diagram of the experimental setup.

Filter tension is obtained by using a torque wrench on the bottom eleme nt of the bag. Two
states of tension have been tested : a functioning where no tension was imposed on the bags
and another where bags were strongly tightened (0.1 m.kg). The tested bags do not have
support cages but it exists five rings along the bag, which divide the bag into four parts and
force it to remain stiff in order to prevent collapse (bags A and B, Figure 2).

A common filter media employed in industrial dust collectors has been tested. It is a
thermobonded polyester needlefelt filter. The physical parameters of this media are shown in
Table 1 :
 Reference FiltraSud T12T56320
Surface weight G (g.m -2) 400 ± 5
Thickness Z (mm) 1.22 ± 0.10
Packing density α f 0.24 ± 0.02
Fiber diameter (Davies) d f (µm) 20.3 ± 3.4
Mean fiber diameter (µm) 20
Table 1 : Filter characteristics.

Observations of the media filter used, with a scanning electron microscope, are given on
Figure 3.

Figure 3 : Observations of the filter media surface.

RESULTS AND DISCUSSION

Description of pulse-jet cleaning consequences – Influences of the number of bags

cleaned together

The results for all 24 bags working together are presented. Cleanings are released for the 24
bags together or for only one bag. Some examples of pressure drop and acceleration traces
along the bag during the compressed air impulse are presented on the Figure 4.

The profiles traces of pressure drop and acceleration along a bag are globally similar whether
one bag or 24 bags are cleaned.

The pressure drop is brutally increased near the top of the bag (closer to the injector nozzle)
by the compressed air impulse, the bag being sucked towards its inside. This suction only
occurs in the top 40 centimeters of the bag. It is due to a depression inside the top of the bag,
undoubtedly caused by a “Venturi effect” near the compressed air exhaust. Then, a dominant
pressure inversion occurs and P e and Pint,2 become closer in the second quarter of the bag. The
rest of the bag is only subjected to a positive pressure pulse and the fabric moves outwards
due to the excess of pressure inside relative to the outside. So ∆P 3 and ∆P 4 are negative and
the pressure drop increases in the bottom of the bag (certainly due to the reflection of the
compressed air pulse).
∆P (Pa) a (g) ∆P (Pa) a (g)
200 200
2500 2500
2000
1500
∆P1 150
100
a1
ZOOM
2000
1500
∆P 1 150
100
a1
ZOOM
1000 50
200
150
1000 50 200
150
500 0
100 500 0 100
0 50 0 50
0 -50
-500 -50
-50 -500 0
-50
-1000 -100 -100 -1000 -100 -100
-1500 -150
-150
0,03 0,04 0,05 0,06
-1500 -150 -150
0,04 0,05 0,06 0,07
200 200
2500 2500
2000 ∆P2 150 a2 2000 ∆P 2 150 a2
1500 100 1500 100
1000 50 1000 50
500 0 500 0
0 0
-50 -50
-500 -500
-1000 -100 -1000 -100
-1500 -150 -1500 -150
200 200
2500 2500
2000 ∆P3 150 a3 2000 ∆P 3 150 a3
1500 100 1500 100
1000 50 1000 50
500 0 500 0
0 0
-500 -50 -50
-500
-1000 -100
-1000 -100
-1500 -150 -1500 -150

2500 200 200

2500
2000 ∆P 4 150 a4 2000 ∆ P4 150 a4
1500 100 1500 100
1000 50 1000 50
500 500
0 0 0 0
-500 -50 -500 -50
-1000 -100 -1000 -100
-1500 -1500
-150 -150
00 0,2
0.2 0,4
0.4 0,6
0.6 0,8
0.8 0 0,1 0,2 0,3 0,4 000 0,2 0,4 0,6 0,8
0 0,2 0,4 0,6 0,8 0 0.2
0,2 0.4
0,4 0.6
0,6 0.8
0,8 t (s) 0.2
0,2 0.4
0,4 0.6
0,6 0.8
0,8 0
0 0,1
0.2
0,2
0,2
0.4
0,4
0,3
0.6
0,6
0,4
0.8
0,8 t (s)
Cleaning of 1 bag Cleaning of 24 bags
Figure 4 : Pressure drop and acceleration traces along the bag, during a cleaning impulse
Uf = 5 cm.s-1 – P r = 7 bar – td = 100 ms – d = 10 mm (nozzle position 1) – no tension.
The maximum accelerations (negative and positive) are recorded at the top of the bag. This
indicates that the filter media is exceptionally stressed close to the injection point of the
compressed air pulse. Firstly, a negative acceleration, towards the inside of the bag, is
observed and then it occurs a bigger positive acceleration of the filter media towards the
outside until its maximum expansion. In theory, the only resistance to the outward
acceleration is the inertia of the fabric and the cake. During this motion, the fabric is
accelerating the cake outwards ; consequently the cake is being pressed against the fabric and
therefore does not leave it. When the fabric becomes taut, a hoop stress is set up which
opposes the outward motion and there is a tensile stress between the cake and the fabric which
leads to the cleaning. Regarding the acceleration values and visual observations of the media
motion, it can be predicted that the pulse-jet cleaning will certainly be efficient at the top of
the bag. The rest of the bag is only subjected to little accelerations, which do not lead to a
rapid motion of the bag. At the bottom of the bag, the accelerations and bag motion do not
dominate the cleaning force and the eventual dust cake release cannot be the result of this
cleaning mechanism.

Those considerations match up with works of other authors. But nevertheless, influences of
the number of bags cleaned together has never been investigated yet, in spite of the fact that
this parameter noticeably modifies the studied profiles along a bag. Actually, the maximum
values of curves peaks, for the cleaning of 24 bags, are always lower than for only one bag
(Table 3). The duration of pressure drop peaks for 24 bags are also shorter than for just one
bag (0.28 s against 0.42). But, it has not yet been possible to establish if those differences are
the result of a physical interference between bags or if they are the effect of a decrease of the
compressed air tank pressure which would not be able to feed 24 bags together without
loosing cleaning power.
Influences of the initial compressed air tank pressure, of the injector nozzle geometry
and position towards the top of the bag

Effects of the initial compressed air tank pressure on the peak values of pressure drop and
acceleration are gathered in Table 2 :
Cleaning of 1 bag Cleaning of 24 bags
Position Pr = 7 bar Pr = 5 bar Pr = 3 bar Pr = 7 bar Pr = 5 bar Pr = 3 bar
∆P (Pa) a - (g) a + (g) ∆ P (Pa) a - (g) a + (g) ∆P (Pa) a - (g) a + (g) ∆P (Pa) a - (g) a + (g) ∆P (Pa) a - (g) a + (g) ∆P (Pa) a - (g) a + (g)

1 2300 -112 168 1780 -89.5 121.9 920 -43.4 78.7 1910 -62 126 1345 -38.1 98 690 -12.1 36.9
2 -550 -16 27 -700 -9.4 13.8 -500 -8 7.9 -400 -9 6.5 -370 -5.6 6.7 -200 -3.7 3.8
3 -1060 -12.5 12 -920 -6.8 9.1 -670 -5.6 5.2 -400 -6 5 -300 -5.4 6.6 -180 -3.4 4.2
4 -1300 -7 15 -1020 -6 7.1 -900 -5.3 5.6 -600 -7 7.5 -500 -4.8 7.4 -350 -2.8 3.1

Table 2 : Peak values of pressure drop (∆P), of negative accelerations (a -) and of positive
accelerations (a +) along the bag according to initial compressed air tank pressure
(Uf = 5 cm.s -1, td = 100 ms, d = 10 mm (nozzle position 1), no tension).

The more Pr decreases, the smaller the maximum pressure drop and acceleration values
become. Whatever the tank pressure, traces of pressure drop and acceleration keep the same
profile along the bag. In all cases, the pressure peak or the acceleration peak is highest at the
top of the bag. The pressure drop decreases towards the le ngth of the bag, where a rising
overpressure appears. Close to the bottom of the bag, the peak pressure (absolute value)
increases due to the reflection of the compressed air pulse. The acceleration values rapidly fall
towards the length of the bag. The filter media is not subjected to fast motion in the bottom
half of the bag and acceleration values are small and very close. Those considerations are
especially evident at high value of tank pressure.

Various injector nozzles have also been tested under similar cleaning conditions and results
are gathered in Table 3 :
Cleaning of 1 bag Cleaning of 24 bags
Nozzle position 1 position 2 position 3 position 1 position 2 position 3
position
d = 5 mm d = 10 mm d = 15 mm d = 10 mm d = 10 mm d = 5 mm d = 10 mm d = 15 mm d = 10 mm d = 10 mm
Point of ∆P a - (g) ∆P a - (g) ∆P a - (g) ∆P a - (g) ∆P a - (g) ∆P a - (g) ∆P a - (g) ∆P a - (g) ∆P a - (g) ∆P a - (g)
measure (Pa) a + (g) (Pa) a + (g) (Pa) a + (g) (Pa) a + (g) (Pa) a + (g) (Pa) a + (g) (Pa) a + (g) (Pa) a + (g) (Pa) a + (g) (Pa) a + (g)
-32.8 -112 -148.8 -104.3 -66.7 -21 -62 -64.4 -105.3 -59.5
1 1040 2300 1100 3100 >5000 1040 1910 1000 2770 2300
47.5 168 198.7 83.9 33.6 44.9 126 127.3 79.6 28.2
-5.5 -16 -29.1 -77.5 -168.8 -5 -16 -16.3 -77
2 -350 -550 -1010 -960 1160 -180 -400 -9 6.5 -450 -320 1160
6.3 27 37.5 92.4 171.2 5.7 21.2 18.6 69
-3.2 -12.5 -13.7 -17.3 -31.8 -3 -6 -8.4 -10.4 -8.3
3 -490 -1060 -1260 -1015 -850 -230 -400 -620 -400 -500
3.4 12 13.9 18.9 36.5 3.4 5 9.1 12 10
-3.4 -7 -11.8 -9.9 -10.2 -3.5 -7 -8.5 -8.3 -6.8
4 -620 -1300 -1480 -1240 -1000 -540 -600 -830 -600 -700
4.7 15 14.6 20.6 10.5 3.8 7.5 9.4 18.8 7.9

Table 3 : Peak values of pressure drop (∆P), of negative accelerations (a -) and of positive
accelerations (a +) along the bag according to several injector nozzles
(U f = 5 cm.s -1, td = 100 ms, Pr = 7 bar, no tension).
 ∆P (Pa) ∆P (Pa) ∆P (Pa)
2500 2500 2500
2000 2000 2000
1500 1500 1500
1000 1000 1000
500 500 500
∆ P1 0
-500
0
-500
0
-500
-1000 -1000 -1000
-1500 -1500 -1500
2500 2500 2500
2000 2000 2000
1500 1500 1500
1000 1000 1000
500 500 500
∆ P2 0
-500
0
-500
0
-500
-1000 -1000 -1000
-1500 -1500 -1500
2500 2500 2500
2000 2000 2000
1500 1500 1500
1000 1000 1000
500 500 500
∆ P3 0
-500
0
-500
0
-500
-1000 -1000 -1000
-1500 -1500 -1500

2500 2500 2500

2000 2000 2000
1500 1500 1500
1000 1000 1000
500 500 500
∆ P4 0 0 0
-500 -500 -500
-1000 -1000 -1000
-1500 -1500 -1500
00 0,2
0.2 0,4
0.4 0,6
0.6 0,8
0.8 00 0,2
0.2 0,4
0.4 0,6
0.6 0,8
0.8 00 0,2
0.2 0,4
0.4 0,6
0.6 0,8
0.8

t (s) t (s) t (s)

d = 5 mm d = 10 mm d = 15 mm
Figure 5 : Pressure drop traces along the bag, during a cleaning impulse
Uf = 5 cm.s-1 – P r = 7 bar – td = 100 ms – nozzle position 1 – no tension.

∆P (Pa) ∆P (Pa) ∆P (Pa)

4500 4500 4500
3500 3500 3500
2500 2500 2500
1500 1500 1500
∆P1 500
-500
500
-500
500
-500
-1500 -1500 -1500

4500 4500 4500

3500 3500 3500
2500 2500 2500
1500 1500 1500
∆P2 500
-500
500
-500
500
-500
-1500 -1500 -1500

4500 4500 4500

3500 3500 3500
2500 2500 2500
1500 1500 1500
∆P3 500
-500
500
-500
500
-500
-1500 -1500 -1500

4500 4500 4500

3500 3500 3500
2500 2500 2500
1500 1500 1500
∆P4 500 500 500
-500 -500 -500
-1500 -1500 -1500
00 0,2
0.2 0,4
0.4 0,6
0.6 0,8
0.8 0
0 0,2
0.2 0,4
0.4 0,6
0.6 0,8
0.8 00 0,2
0.2 0,4
0.4 0,6
0.6 0,8
0.8

t (s ) t (s) t (s)
Nozzle position 1 Nozzle position 2 Nozzle position 3
Figure 6 : Pressure drop traces along the bag, during a cleaning impulse
Uf = 5 cm.s-1 – P r = 7 bar – td = 100 ms – d = 10 mm – no tension.
 Figure 7 : Acceleration traces along the bag, during a cleaning impulse
Uf = 5 cm.s-1 – P r = 7 bar – td = 100 ms – d = 10 mm – no tension.

The initial tank pressur e (P r) and the nozzle diameter (d) are generally regarded as the most
influent parameters for the pulse-jet cleaning intensity. When Pr or d increase, all cleaning
indices (overpressure, acceleration, reverse airflow) significantly increase [6 – 10, 19, 21]
(Table 3 and Figure 5). In this way, the cleaning effect is better and residual pressure drop
after cleaning becomes lower (increase of the duration of a clogging-cleaning cycle).
However, there always exists an optimum in the combination of Pr and d, which leads to
maximum cleaning performance. A too large nozzle diameter or inadequate tank pressure
cause a decrease in cleaning force and the energy will be wasted. Many previous investigators
[5, 7, 14, 17] also pointed out that a critical cleaning efficiency exists for different indices of
cleaning intensity, such as the peak of pulse overpressure, the average pulse overpressure
inside the bag and the fabric acceleration. If the index of cleaning intensity exceeds the
critical value, the cleaning efficiency improves only slightly while the dust emission will
increase and the filter media would get prematurely damaged. In an industrial plant, Pr is
about 5 to 6 bar. The most current opening of injector nozzle is the 10 mm diameter.

When the injector nozzle of diameter d = 10 mm precisely emerges at the top of the bag
(position 2) or is introduced into the higher quarter of the bag (position 3), the profiles change
(Figure 6). When only one bag is cleaned, the depression recorded in ∆P 1 is, logically,
accentuated as the cause of the depression (the air pulse-jet) approaches to the point of
measurement. In the same way, for the three other points of measurement, one notes that
fronts of ∆P and a shift forward to the bottom, from a distance approximately equal to the
distance to which the point of compressed air impulse is descended. So, the filter media show
great accelerations not only in the top quarter of the bag, but also on the whole top half of the
bag (Figure 7). The acceleratio n and pressure drop profiles are also modified when 24 bags
are cleaned (Table 3). Nevertheless, the bags seem to influence each other very little
compared to the one bag operation. Therefore, comments remain globally unchanged.
Contrary to Rothwell’s results [15] which show that inserting the injector into the filter
element degrades the attack of the overpressure pulse and leads to lower cleaning efficiency,
our results point out that inserting the compressed air injector into the filter increase the area
which is subjected to important accelerations of the filter media (Figure 7).

The injector nozzles in position 1 lead to important accelerations of the filter media at the top
of the bag (Table 3 and Figure 7). The acceleration of the media is highest at the bag top and
decreases with the bag length. Cleaning of the dust cake by mechanical movements of the
filter media will only be really effective in the top area because in the remainder of the bag,
accelerations are weak and do not lead to strong movements of the media. Results also show
that the larger the diameter d, the bigger the acceleration of the media ; that is understandable
because a greater air volume is then released. In the same way, overpressures in the bottom of
the bag (being able, in theory, to lead to a reverse airflow cleaning), are as more important as
the diameter of the injector nozzle is large. The pulse-jet cleaning intensity produced
simultaneously on the 24 bags seems less favourable to an effective cleaning, compared to the
same cleaning on a single bag.

Influences of initial bag tension, of the filtration velocity and of duration of pulse-jet
cleaning

Experiments did not reveal significant influences on the profiles of acceleration and pressure
drop for the following paramete rs : tension, filtration velocity (U f) and pulse-jet duration (t d).
The peaks of values and the shape of traces for each position of measurement remain the same
whether the filter bags are slightly or strongly tightened, whether the filtration velocity is
equal to 2, 5 or 8 cm.s-1 , and finally, whether the cleaning duration is n × 100 ms, n ranging
between 1 and 10.
 CONCLUSIONS AND PERSPECTIVES

The changes of acceleration and pressure drop along a bag were recorded on a pilot-scale dust
collector with industrial dimensions. Those results made it possible to establish and
understand the variations of the profiles of pressure drop and acceleration of the filter media
according to the distance between the point of injection of compressed air and the point of
measurement. Typical numerical values of pulse-jet cleanings were determined : depression
from 900 to 3000 Pa at the top of the bag, overpressures from 500 to 1500 Pa at the bottom of
the bag and accelerations from 50 to 200 g close to the injector nozzle. Experiments also
show that the pulse-jet cleaning is not effective on the whole height of the bag. If there is an
efficient cleaning at the bottom of the bag, it can only occur from a reverse airflow action or
from a gentle shake action but not from a rapid motion of the filter media. The initial
compressed air tank pressure and the geometry of the injector nozzle appear as principal
parameters, that influence pulse-jet consequences the most significantly. On the other hand,
tension of bags, filtration velocity and duration of cleaning are parameters that have less
influence and do not cause significant changes in the values of acceleration and pressure drop.
Lastly, influence of the number of bags simultaneously cleaned is very important but will
require complementary work to be completely and correctly interpreted. The following stage
is a similar study with dust generation and formation of a particle cake on the bags.

REFERENCES

[1] Callé S., Contal P., Thomas D., Bémer D., Leclerc D. ; Description of the clogging and
cleaning cycles of filter media ; Powder Tchnology, Vol. 123, pp. 40 – 52 ; 2002.
[2] Dennis R., Wilder J.E., Harmon D.L. ; Predicting pressure loss for pulse jet filters ;
Journal of the air pollution control association, Vol. 31, n° 9, pp. 987 – 992 ; 1981.
[3] Allen R.W.K., Goyder H.G.D., Morris K. ; Modelling media movement during cleaning of
pulse-jet fabric filters ; Chemical engineering research and design ; Vol. 77, n°3 ; pp. 223 –
230 ; 1999.
[4] Klingel R., Loffler F. ; Dust collection and cleaning efficiency of a pulse-jet fabric ;
Filtration and Separation, Vol. 20, n°3, pp. 205 – 208 ; 1983.
[5] Humphries W., Madden J.J. ; Fabric filtration for coal- fired boilers : dust dislodgement in
pulse jet filters ; Filtration and Separation, Vol. 20, n°1, pp. 40 – 44 ; 1983.
[6] Sievert J., Loffler F. ; Dust dislodgement in pulse -jet fabric filters ; First World Congress
on particle technology, Part IV ; Nurnburg, West Germany, pp. 111 – 126 ; 16 – 18 Avril
1986.
[7] De Ravin M., Humphries W., Postle R. ; A model for the performance of a pulse jet filter ;
Filtration and Separation ; Mai / Juin 1988.
[8] Lu H.C., Tsai C.J. ; Numerical and experimental study of cleaning process of a pulse-jet
fabric filtration system ; Environmental science and technology, Vol. 30, n°11, pp. 3243 –
3249 ; 1996.
[9] Lu H.C., Tsai C.J. ; A pilot-scale study of the design and operation parameters of a pulse-
jet baghouse ; Aerosol science and technology, Vol. 29, n°6, pp. 510 – 524 ; 1998.
 [10] Lu H.C., Tsai C.J. ; Influence of design and operation parameters on bag-cleaning
performance of pulse -jet baghouse ; Journal of environmental engineering (New York), Vol.
125, n°6, pp. 583 – 591 ; 1999.
[11] Dennis R., Wilder J.E. ; Fabric filter cleaning studies ; National Environmental Research
Center, Report EPA 650/2-75-009 ; 1975.
[12] Ellenbecker M.J., Leith D. ; Dust removal from non-woven-fabric-cleaning methods
need to be improved ; Filtration and Separation, Vol. 18, pp. 41 - 45 ; juillet 1981.
[13] Dennis R., Klemm H.A. ; Modeling concepts for pulse jet filtration ; Journal of the Air
Pollution Control Association, Vol. 30, n° 1, pp. 38 – 43, janvier 1980.
[14] Morris W.J. ; Cleaning mechanisms in pulse jet fabric filters ; Filtration and Separation,
Vol. 21, n°1, pp. 50 – 54 ; 1984.
[15] Rothwell E. ; The design and operating characteristics of reverse-jet assemblies for pulse
jet dust collectors ; Fourth World Filtration Congress, Part I, Ostend, Belgium, pp. 2.43 – 2.53
; 22 – 25 Avril 1986.
[16] Sievert J., Loffler F. ; Dust cake release from non woven fabrics ; Filtration and
Separation ; Novembre / Décembre 1987.
[17] Sievert J., Loffler F. ; Fabric cleaning in pulse-jet filter ; Chem. Eng. Process, Vol. 26,
pp. 179 – 183 ; 1989.
[18] Bustard C.J., Cushing K.M., Chang R.L. ; The potential of pulse-jet bag-houses for
utility boilers. II : Performance of pulse-jet fabric filter pilot plants ; Journal of the air and
waste management association, Vol. 42, n°9, pp. 1240 – 1249 ; 1992.
[19] Tsai C.J., Tsai M.L., Lu H.C. ; Effect of filtration velocity and filtration pressure drop on
the bag-cleaning performance of pulse-jet baghouse ; Separation Science and Technology,
Vol. 35, n°2, pp. 211 – 226 ; 2000.
[20] Leith D.H., First M.W., Feldmann H. ; Performance of a pulse-jet filter at high filtration
velocity – II. Filter cake redéposition ; Journal of Aerosol Pollution Control Association, Vol.
27, pp. 636 - 642 ; 1977.
[21] Morris K. ; Expert system control of fabric filters ; Filtration and Separation, Vol. 28,
n°4, pp. 275 – 280 ; 1991.

View publication stats