Filtration Characteristics of Fine Particulate Matters in a PTFE / Glass Composite Bag Filter

Bag houses are often used to control particulate matters and recover valuable resources in various industries. A bag filter is the most important component in a bag house system, and thus it is important to develop the best-quality filter media and determine the optimum operating conditions of a bag house system. This study focused on particle penetrations under an operating condition of a bag house system, and investigated the relationship between dust clogging and dust penetration as a function of pressure drops across the filter medium. The results showed a minimum collection efficiency of 80% for 0.45 μm particles at the initial stage of filtration, although this quickly recovered and remained over 99.99% with a pressure drop greater than 20 mmH2O with a newly developed porous filter (air permeability = 5.78 × 10 m). Furthermore, the recovery time depended strongly on particle size. While it was inversely proportional to the particle size at the initial stage of filtration, it showed no difference for particles larger than 0.725 μm due to its uniform dust cake as filter cleaning proceeded.


INTRODUCTION
Recent interests in health risks of PM 2.5 have led to various relevant studies into air quality management and control in many countries (Dockery et al., 1993;Tucker, 2000).The most widely used collection facility to remove fine particulate matters in the industrial field is a bag house composed of many cylindrical filter bags.It can obtain high collection efficiency without interrupting filtration in a continuous process by real-time filter regeneration.While the particulate matters are filtered, a pulse-jet is injected across the filter media generally from inside to outside.Even if a bag house system using pulse-jet cleaning is widely used, the particle collection mechanisms are still questionable.We can minimize the decrease of collection efficiency by understanding the collection mechanisms.
Generally, conventional collection mechanisms such as inertial force, interception, diffusion, due to a particle-fiber interaction are dominant at the initial stage of filtering particulate matters with fresh bag filters.However, the mechanisms are very complex during a continuous filtering process, because dust cakes are periodically formed on and detached from the filter surface.
So far, most of studies regarding bag house systems have focused on how to effectively dislodge dust cakes from the filter surface and understanding the filter cleaning mechanisms (Hata et al., 2003;Simon et al., 2007;Saleem and Krammer, 2007).
There have only been a few studies regarding particle penetration through filter media.Leith and First (1977) showed that "straight through" was a dominant mechanism in particle penetration at the initial stage of fibrous filtration, and "seepage" and "pinhole plug" mechanisms became dominant afterward.These mechanisms can lead to a sharp decrease following a slight increase in particle penetration (Bach and Schmidt, 2007).Moreover, Sievert (1988) also mentioned that emissions from a continuously operating bag house system peaked immediately after filter cleaning and rapidly decreased with dust loading.This trend appeared on other bag filter media such as rigid ceramics and needle felts (Umhauer et al., 2000;Binnig et al., 2005).
However, there have been no studies concerning the effect of particle size on particle penetration in cake filtration, even if the collection mechanisms are usually strongly dependent upon particle size.
Therefore, we made an effort to clarify the filtration mechanisms during the entire operating time in a bag house system by investigating particle penetration as a function of particle size.This work can improve the filtration performance of bag house systems in industrial fields.

Test Filter
The PTFE/Glass composite bag filter (PG-filter) used in this work was prepared by a foam coating of an emulsiontype PTFE (DuPont, TE-3893, 60.7 wt% solid content) on a Teflon B coated glass fiber mat.Detailed preparation of the test filter was described in an earlier article (Park et al., 2010).As shown in Fig. 1, the completed PG-filter was composed of three-dimensional porous structures.
Air permeability (K), which is one of the basic properties of porous filter media, was evaluated based on a classic Darcy's equation, where, ∆P is the pressure drop across the filter, L is the filter thickness, u is the filtration velocity and μ is the air viscosity (1.81 × 10 −5 Pa•s at 20°C).Therefore, air permeability (K) can be defined as (μ•u•L)/∆P, which is influenced by filter characteristics such as flow pattern, filter porosity and pore size (Lee, 1981).Thus, K value of each filter depending on the filter properties is a very important design factor to control filter performance (Seville, 1997).

Experimental Setup
A simple experimental rig for filter testing was built as depicted in Fig. 2. It was a typical filter chamber widely used as a part of air pollution control devices (APCDs) in industrial fields.Four bag-type filters (110 mm in diameter, 260 mm in length) were installed in a bag house system, which consisted of a dust generation part with a dust feeder, a dust collection part (bag-house), a filter cleaning part (compressor ~ pulse-jet) and a blower to produce the air flow across the device.The house was made of a transparent acryl to allow observation of the inside.The test device was operated at the filtration velocity of 3 m/min and the introduced dust concentration of 5.7 g/m 3 .High-pressure jet air (4 kg f /cm 2 ), which was delivered through a blow tube was periodically injected by two solenoid valves for filter cleaning.The jet nozzles were installed at the lower face of blow tube toward a filter bag in order to efficiently supply the jet air.Test dust, coal fly ash, was procured from a local coal-fired power plant.A dust feeder with a screw of 23 mm in diameter was used to supply dried test particles in the system.The constant feedrate of 6.1 g/min was obtained at the operating condition of a screw rotating speed of 3 rpm.Furthermore, negative pressure was always maintained in the bag house system, resulting in quick dispersed feeding.
The present work focused on fine particulate matters, especially PM 2.5 .Particle number concentrations were measured by a particle counter (Dust spectrometer 1.108, Grimm) following a diluter.The introduced dust concentration was measured downstream of the chamber (2 m away from the chamber exit to obtain a good mixture of particles) without installation of bag filter media.To equalize the introduced dust concentration in the chamber when we measured the filtration efficiency, the feedrate and flowrate were maintained constant throughout filtration.The test dust contained 70,000 #/cm 3 for 0.5 μm with the highest frequency, and the lowest with 13,000 #/cm 3 for 3 μm.

Air Permeability of Test Filter
Fig. 3 shows the measured pressure drop in relation to   The estimated permeability of the fresh filter was quite high, 5.78 × 10 −11 m 2 , in comparison to a commercial laminated membrane filter (1 × 10 −12 m 2 ) which is frequently applied to high temperature filtration in many industrial processes.
The PG-filter with 0.9 mm in total thickness has a highly porous PTFE layer with 0.1 mm in thickness, while the commercial membrane filter with 0.8 mm in total thickness has very thin laminated film less than 1.0 μm in thickness.
The pressure drops of tested filters are determined by mainly PTFE films.Therefore, highly porous structure of the PTFE foam coated film may lead to such a high permeability despite of its thick layer.

Filter Cleaning by Back-Flushing
Periodic filter cleaning by pulse-jet injection of highly pressurized air affects the filter life time and the overall filtration performance.It was programmed that jet air generated from a compressor should be injected into the individual filter bags upon reaching the maximum allowable pressure drop of 100 mmH 2 O.While the jet air passes counter-currently to the flue gas stream, the particulate matters deposited on outer surface of the filter bags could be detached and dropped down in gravity.Fig. 4 shows the variation of pressure drop of the PG filter and the commercial laminated membrane filter during dust filtration with jet pulsing.The injection interval of the jet air became shorter as filtration proceeded.The residual pressure drop of the PG filter immediately after back flushing remained quite low, under 20 mmH 2 O, comparing to that of the commercial laminated membrane filter, even though it showed a slight increase due to the infiltration of fine particles.Only five injections in the PG filter were carried out during filtration for 7 hours operation, while seven injections in the commercial laminated membrane filter were done.The frequency of air injection and the quantity of residual pressure drop may determine the filter cleaning efficiency.
Fewer automatic injections during a certain period imply more effective dust collection across the filter medium.Maintaining a consistent value of residual pressure drop would indicate that the cleaning is working very well.
Although the test filter was composed of two heterogeneous layers, most of the introduced particulate matters could be collected on the upper layer of PTFE rather than inside the pore channels (see the details in Park et al., 2010).Therefore, operation pressure resistance would depend mainly on the filter surface.That is, the overall pressure drop increases with dust loading on the filter surface and the residual pressure drop depends on the attractive force between a fiber and a particle.Even if the PG filter is relatively thick, it has high filter cleaning efficiency due to the low surface free energy of the filter medium (Park et al., 2011).The de-dusting behavior is more clearly examined in next section.

Particle Penetration during Filtration
Particle penetration during the entire filtration process is shown in Fig. 5.At the initial stage of filtration, approximately 10 5 particles per cubic centimeter of flow passed through the filter media; however, particle penetration drastically decreased over time, resulting from cake filtration.However, particle penetration was quickly increased again by jet flushing.In a continuous filtration process, the introduced particulate matters would clog the open pores in the filter and form secondary dust filter layers, leading to overall surface filtration.Such a structural transformation aids absolute filtration with only a little particle penetration.
Fig. 6 shows fractional collection efficiencies of the filter media as a function of particle size in overall operating conditions.One of the most important parameters in filtration is collection efficiency.In this work, fractional collection efficiency was determined by analyzing particle size distributions upstream and downstream of the filter.Fractional collection efficiency, η was calculated by upstream particle concentration, N i and downstream particle concentration, N o of the filter.
1 100 As can be seen in the graph, back-flushing obviously affected fine particle penetration.However, as particles were loaded on the filter surface, the penetration was remarkably diminished from after the first flushing.Due to particle blocking of open channels as well as the additional filter layer of dust cake on the filter surface, penetration by large particles over 0.725 μm was seldom observed.It took no longer than 10 minutes to reach a stabilized filtration condition after each flushing.Nevertheless, the penetration amount for the short initial period immediately after flushing cannot be ignored.This period corresponds to about 5 to 10 percent of the entire filtration period.This inevitable phenomenon in the conventional filtration system must be taken into series consideration.
Although, the prepared filter in this work achieved quite a high efficiency of fine dust collection, very tiny particles are still apt to pass through it even after the stabilization of filtration, and such a tendency would remain until filter replacement.Particles over 0.9 μm could be collected with high efficiency about 10 minutes after back-flushing.This implies distinguishable discreted dust collection amongst overall PM 2.5 .While national government regulates PM 2.5 for atmospheric dust, the emission control should focus more on PM 1.0 in particular being applied by filter facilities.Fig. 7 shows the potential collection mechanisms in a pulse-jet bag house system.The main cause of increasing efficiency over time is that the travelling fine particles are clogged on the filter surface and other fine particles collided into the deposited particles.Eventually, the morphology formed by deposited particles contributes to making long   paths which effectively capture the fine particles by continuous dust filtration.However, the deposited particles are thrown off the surface of the filter by pulse-jet cleaning at the maximum allowable pressure drop.The dispersed particles are appropriately rearranged on the filter surface, and the uniformity of the dust cake changes, which results in increased collection efficiency for submicrometer particles and a slight increase of pressure drop.

CONCLUSIONS
This experimental study investigated the filtration and cleaning characteristics of a bag filter system with a continuous filtration process using a PTFE/Glass composite filter prepared by a foam coating technology(air permeability = 5.78 × 10 −11 m 2 ).It was found that residual pressure drop after filter cleaning was maintained at low pressure drop below 20 mmH 2 O, and surface filtration was a dominant filtration mechanism of the PTFE/Glass composite filter.However, the fractional collection efficiency of the test filter was under 90% at the initial stage and immediately after filter cleaning, even though it recovered within 20 minutes and maintained stable fractional collection efficiency for particles in the size range of 0.45 μm to 2.5 μm.We also found that particle penetration strongly depended upon particle size.As particle size decreased, the collection efficiency of particles decreased due to lower impaction effect during filtration with a fresh filter.However, we observed no collection efficiency difference for particles over 0.725 μm after filter cleaning for the third time.The obtained results indicate that the uniformity of dust cake changed with repeated dust cake formation on the filter surface under the same filtration conditions.

* DISCLAIMER
The measured collection performance was confined to the tested filter and the filter does not represent the products of GE energy.

Fig. 1 .
Fig. 1.SEM image of the prepared filter surface.

Fig. 3 .
Fig. 3. Pressure drops across test filter media as a function of filtration velocity.

Fig. 5 .
Fig. 5.The change in total number concentrations of penetrated particles through the PG-filter.

Fig. 7 .
Fig. 7. Illustration of collection mechanisms in a pulse-jet bag house system.